UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structural studies of the regulation of gelsolin by small ligands Urosev, Dunja 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_2007-319475.pdf [ 17.16MB ]
JSON: 831-1.0060214.json
JSON-LD: 831-1.0060214-ld.json
RDF/XML (Pretty): 831-1.0060214-rdf.xml
RDF/JSON: 831-1.0060214-rdf.json
Turtle: 831-1.0060214-turtle.txt
N-Triples: 831-1.0060214-rdf-ntriples.txt
Original Record: 831-1.0060214-source.json
Full Text

Full Text

S T R U C T U R A L STUDIES OF T H E R E G U L A T I O N OF G E L S O L I N B Y S M A L L L I G A N D S by D U N J A U R O S E V M . S c , Moscow State University named after M . V . Lomonosov, 1999 M . S c , University of British Columbia, 2003 A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF T H E R E Q U I R E M E N T S F O R T H E D E G R E E OF D O C T O R OF P H I L O S O P H Y in T H E F A C U L T Y OF G R A D U A T E STUDIES (Chemistry) T H E U N I V E R S I T Y OF BRITISH C O L U M B I A September 2007 © Dunja Urosev, 2007 Abstract Gelsolin regulates the dynamic rearrangement of the actin cytoskeleton by severing actin filaments (F-actin) and capping the newly generated barbed filament ends. Calcium ions and phosphatidylinositol 4,5-bisphosphate (PIP2), in turn, control these-activities. Ca -binding within gelsolin primes the protein for binding actin filaments and, during the process, causes drastic rearrangement of its six domains (G1-G6). The significance of specific Ca 2 + -binding events in this activation process is slowly emerging. A structural basis for the ability of PIP2 to inhibit ab initio interactions of gelsolin with -actin filaments, as well as to induce dissociation of gelsolin already bound to F-actin, remains largely unknown. The reported binding of A T P to gelsolin has been suggested to modulate gelsolin-actin interactions. Association of gelsolin with both of these phosphate-rich molecules is sensitive to the presence of calcium ions. In the projects described in this thesis, I report results of X-ray crystallography c and computational molecular docking experiments to investigate aspects of the regulation of gelsolin by A T P , PIP2 and calcium ions. Successful introduction of A T P into crystals of inactive gelsolin identify for the first time detailed features of its molecular interactions with gelsolin. Computations confirm the binding of A T P to the observed site to be strong and specific. Demonstration that the ATP-binding site spans both the N - and C-terminal halves of the protein explains the decreased affinity of gelsolin for this ligand in the presence of calcium ions, which induce separation of the halves as part of the activation process. Computational docking experiments suggest residual affinity of activated gelsolin for A T P to be retained at the G2-G3 interface with actin. We propose a model for PIP 2-binding in the same surface-exposed pocket in gelsolin that is associated -with binding A T P phosphates. The model concurs with both the previously reported binding of PIP2 in this vicinity on gelsolin and the higher affinity of gelsolin for PIP2 than for A T P . The model, together with the structure of the Gl-G3/ac t in complex, provide insight into the roles of putative PIP 2-binding sites in both the N - and C-terminal halves of gelsolin. Lastly, exchangeability of metal ions in crystals of Gl-G3/ac t in reflects the transient nature of Ca -binding in G2 and helps to explain the loss of local structural stability in a gelsolin mutant that experiences enhanced susceptibility to proteolysis. 11 Table of contents Abstract ., i i Table of contents i i i List of tables vi List of figures vi i List of abbreviations ix Acknowledgments xi i i C H A P T E R I Introduction 1 A . Actin 1 1. Characteristics 1 2. The structure of an actin monomer 2 3. Properties and structure of actin filaments 3 B . Gelsolin 8 1. Properties and function 8 2. The structure of inactive gelsolin 11 3. Ca 2 + -binding in gelsolin 14 4. Mechanism of gelsolin activation 16 5. Models of F-actin severing and capping by gelsolin 19 6. Inhibition of gelsolin 23 7. The binding of A T P to gelsolin 26 8. Gelsolin in blood 28 9. Finnish-type familial amyloidosis (FAF) 29 10. Gelsolin and actin-associated proteins 31 11. Gelsolin and apoptosis 32 12. Gelsolin in cancer 33 C. Further insight into the regulation of gelsolin 34 C H A P T E R II Materials and Methods 36 A . Protein Purification 36 i n 1. Horse plasma gelsolin preparation 36 2. Preparation of rabbit skeletal muscle powder 37 3. Act in purification 37 4. Preparation and purification of gelsolin/actin complexes 38 4.1 Complexes of one gelsolin bound to two ATP-actin monomers (GA2ATP) or to two ADP-act in monomers ( G A 2 A D P ) 38 4.2 Gelsolin complex with longitudinally crosslinked actin monomers (GA X ) . . . . 39 B . X-ray crystallography 40 1. Protein crystals 41 1.1 The crystallization process , 42 1.2 Cryoprotection of crystals 44 1.3 Crystallization of gelsolin/actin complexes 45 2. Crystal soaking 47 2.1 Gelsolin crystal soaks in A T P solutions 48 2.2 T b 3 + soaks of G A 2 (Gl-G3/actin) crystals 49 2.3 Gelsolin crystal soaks in solutions of L P S 50 3. Single-crystal X-ray diffraction 51 3.1 Collection and processing of X-ray diffraction data 53 3.2 Molecular replacement and structure refinement 55 C. Computational molecular docking 57 1. Docking of gelsolin ligands to inactive gelsolin and Gl-G3/actin w i t h M o l D o c k 57 C H A P T E R III Results 60 A . Interactions of gelsolin with A T P 60 1. The structure of inactive gelsolin bound to A T P 60 2. In silico investigation of A T P binding to inactive gelsolin 64 3. Consequences of Ca 2 +-activation of gelsolin for the ATP-binding site 67 B . Computational docking of PIP 2 to gelsolin 75 C. Metal ion exchange in the N-terminal half of gelsolin 81 1. Exchange of calcium ions in Gl-G3/actin crystals with T b 3 + 81 iv 2. Back-exchange of T b 3 + ions present in Gl-G3/act in with calcium ions 83 3. Implications of transient calcium-binding in G2 for F A F 86 D. Towards the structures o f gelsolin/actin complexes 88 1. Purification, crystallization and X-ray diffraction analysis of gelsolin bound to different actin species ( G A 2 A T P / G A 2 A D P / G A X ) 88 2. Effects of sodium malonate on protein-protein interfaces in crystals of inactive gelsolin 92 E . Conclusions 95 Bibl iography 97 v List of tables Table 1. Crystallographic data and refinement statistics for the ATP-gelsolin structure 61 Table 2. E s c o r e values for docking ATP and ADP onto actin and gelsolin 66 Table 3. E s c o r e values for docking a rigid ATP structure onto G l - G3/actin 72 Table 4 . E s c o r e values for various phosphate-bearing ligands docked to gelsolin 76 Table 5. Crystallographic data and refinement statistics for terbium-soaked Gl-G3/actin crystals 86 vi List of figures Figure 1. Major actin monomer conformations 3 Figure 2. Four actin monomers arranged according to the Holmes filament model 5 Figure 3. In silico superposition of some F-actin-binding proteins (WH2, G1-G3, G4-G6, arp2/3 complex, B n i l p F H 2 ) onto a single actin monomer 6 Figure 4. Schematic representation of actin life cycle 8 2_|_ Figure 5. Individual gelsolin domains excised from the Ca - free structure of horse plasma gelsolin 11 Figure 6. The two gelsolin halves excised from the Ca 2 +-free structure of horse plasma gelsolin 12 Figure 7. A ribbon representation of the structure of equine plasma gelsolin determined in the absence of bound calcium ions 13 Figure 8. Types of Ca 2 + -binding sites in gelsolin 15 Figure 9. The structures of A ) activated G1-G3 bound to actin, B) activated G4-G6 and C) activated G4-G6 bound to actin 16 Figure 10. Conformational changes in gelsolin A ) C-terminal (G4-G6) and B) N -terminal (G1-G3) halves upon binding C a 2 + 17 Figure 11. Schematic representation of gelsolin activation by Ca ions '. 19 Figure 12. A model for an actin filament barbed end capped by gelsolin : 21 Figure 13. The sequence of events during the severing of F-actin by fully activated gelsolin 22 Figure 14. Model of Ca 2 +-free G 1 - G 3 interacting with actin 23 Figure 15. Structures of phosphate rich regulators of gelsolin 24 Figure 16. Regions in G2 implicated in F A F 31 Figure 17. Act in elution profile 38 Figure 18. Crystallization phase diagram 44 Figure 19. Hanging drop vapor diffusion setup 46 Figure 20. Solvent channels in crystals of inactive gelsolin 48 vi i Figure 21. Crystals of inactive gelsolin 49 Figure 22. X-ray diffractometer 54 Figure 23. The structure of ATP-gelsol in 62 Figure 24. Gelsolin residues involved in binding A T P 63 Figure 25. Comparison of crystallographically observed and docked A T P orientations in the A T P - binding site of actin and gelsolin 65 Figure 26. The absence of an ATP-binding site within the Ca -activated C-terminal half of gelsolin 68 Figure 27. A residual ATP-binding site in the Ca 2 +-activated N-terminal half of gelsolin 71 Figure 28. A T P docked into the G2-G3/actin cavity 72 Figure 29. A model for the binding of PIP2 on the surface of inactive gelsolin 77 Figure 30. Regions of domains G l and G2 involved in binding PIP 2 80 Figure 31. Substitution of terbium ions f o r C a Z T i nGl -G3 /ac t i n 82 Figure 32. Backwashing of Tb 3 +-soaked crystals of Gl-G3/act in with calcium ions 85 Figure 33. Region in the N-terminal half of gelsolin involved in F A F 87 Figure 34. Purification and crystallization of G A 2 A T P 89 Figure 35. Packing of Gl-G3/ac t in within P6s22 crystals 90 Figure 36. Formation and crystallization of G A X 91 Figure 37. Effects of sodium malonate on the conformation of the A - A ' loop in gelsolin 94 v i i i List of abbreviations A B P s A c t i n b ind ing proteins 2 A D F A c t i n dep loymer i z ing factor 7 A D P Adenos ine diphosphate 2 A m i n o acids: A l a ( A ) alanine 2 A r g (R) arginine 29 A s n (N) asparagine 9 A s p (D) aspartic ac id 9 C y s (C) cysteine 2 G l u (E) glutamic ac id 2 G i n (Q) glutamine 30 G l y (G) g lyc ine 32 H i s ( H ) his t idine 63 He (I) i soleucine 77 L e u ( L ) leucine 60 L y s ( K ) lys ine 2 Pro (P) prol ine 2 Trp ( W ) tryptophan 60 T y r ( Y ) tyrosine 9 A M P Adenos ine monophosphate 27 A P - a c t i n A c t i n monomer w i t h two point mutations, A 2 0 4 E and P 2 4 3 K 2 A p 3 A Diadenosine 5 \ 5 " ' - P l , P 3 - t r i p h o s p h a t e 27 A p 4 A Diadenosine 5 ' ,5" ' -P l ,P4- te t raphospha te 27 A p n A Diadenosine oligophosphates 27 A T P Adenos ine triphosphate 2 A U F S Absorbance units fu l l scale . . .• 38 B n i l p F G 2 B n i l f o rmin h o m o l o g y doma in 2 6 B P M Benzophenone-male imide 39 ix Buffer A 2 m M Tr i s -HCl , 0.2 m M C a C l 2 , 0.2 m M A T P , 1 m M D T T , pH 7.6 37 Buffer A ' 2 m M Tr i s -HCl , 0.2 m M C a C l 2 , 5 m M A D P , 1 m M D T T , pH 7.6 38 C E M - S S T-cell line 27 C 1 , C3 Domains 1 and 3 of CapG 83 D B P Vitamin D binding protein 28 D E A E 2-diethylamino-ethyl functional group 36 di-C4-PIP 2 P IP 2 containing two aliphatic chains each composed of f o u r - C H 2 - repeats 76 di-C8-PIP 2 PIP 2 containing two aliphatic chains each composed of e i g h t - C H 2 - repeats 75 D M F Dimethylformamide 39 D N A Deoxyribonucleic acid 57 DNase I Deoxyribonuclease 1 2 D T T Dithiothreitol 37 Escore Docking scoring function 58 E D T A Ethylene diamine tetraacetic acid 36 E G T A Ethylene glycol tetraacetic acid 20 F c Calculated structure factor amplitudes 56 F 0 Observed structure factor amplitudes 56 F-actin Filamentous actin 1 F A F Familial amyloidosis (Finnish type) 29 G A Complex of one gelsolin with one actin molecule 20 G A 2 Complex of one gelsolin with two actin molecules 20 G A 2 A D P Complex of one gelsolin with two ADP-act in molecules 35 G A 2 A T P Complex of one gelsolin with two ATP-actin molecules 35 G A 3 Complex of one gelsolin with three actin molecules 20 G A X Complex of gelsolin with longitudinally crosslinked actin monomers 39 G-actin Monomeric actin 1 G D P Guanosine diphosphate 26 G T P Guanosine triphosphate 26 G1-G6 Gelsolin domains 1 through 6 1 x H E P E S 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid 46 H I V Human immunodeficiency virus 33 H P L C High performance liquid chromatography 36 H1 Long a-he l ix present in each gelsolin domain 11 IP 2 Inositol 1,4-bisphosphate or inositol 4,5-bisphosphate 75 IP 3 Inositol 1,4,5-triphosphate 75 Kd Disscoiation constant 14 L P A Lysophosphatidic acid 10 L P S Lipopolysaccharides 10 M A D Multiple anomalous dispersion 52 M I R Multiple isomorphous replacement 52 M P D 2-methyl-2,4-pentanediol 45 M R Molecular replacement 52 P A F Platelet activating factor 28 P A G E Polyacrylamide gel electrophoresis 40 P D B Protein data bank 40 P E G Polyethyleneglycol 42 p G S N Plasma gelsolin 9 PIP 2 or PI(4,5)P2 Phosphatidylinositol 4,5-bisphosphate 23 PI(3,4)P2 Phosphatidylinositol 3,4-bisphosphate 23 PI(3,4,5)P3 Phosphatidylinositol 3,4,5-trisphosphate 23 p P D M N,N'-p-phenylenedimaleimide 39 PPIs Polyphosphoinositides 10 Rmerge i| h — {Ih)\l (-4)], where hi corresponds to the /th observation of an intensity for reflection of index h whose average intensity is (//,) 61 R M S Root-mean-square 61 sAms Saturated ammonium sulphate 48 SDS Sodium dodecyl sulphate 40 T M R Tetramethyl-rhodamine-5 maleimide 2 Tris 2-amino-2-(hydroxymethyl)-1,3-propanediol 36 v/v Volume/volume 46 x i w/v Weight/volume W H 2 Wiscott-Aldrich syndrome protein homology 2 xu Acknowledgments It is a pleasure to thank the many people who made this thesis possible. I am gratefully acknowledging my supervisor, Dr. Les Burtnick, for the invaluable guidance, constructive criticism and support throughout the PhD program, and helpful comments, English grammar corrections and overall revision of the written thesis. I would like to thank Dr. Suzana Straus, a member of my PhD supervisory committee for accepting to be the thesis reader under somewhat constrained time conditions, as well as for suggested corrections. For part of the crystallographic data collection and analysis I thank Drs. Les Burtnick and Robert Robinson. 1 am very grateful to Dr. Gunnar Olovsson who taught me how to use the X-ray facility in the Center for Blood Research, to collect crystal X-ray diffraction data and analyze them. I had the pleasure to supervise and work with many students who have done various projects in the Burtnick research group. Work beneficial to this thesis was that of Nadya Ogloff at the early stages of search for crystallization conditions for gelsolin/actin complexes and Blair Gage in the later period during actin crosslinking experiments. A l l the members of the Burtnick research group deserve a lot of thanks for interesting and fun times during research, especially Hui Wang, for great help in the lab on numerous occasions. I appreciate very much the encouragement I have received both directly or indirectly from my closest friends. I cannot end without thanking Sava for his patience and understanding throughout the whole thesis process and my family, especially my parents who long before taught me right what matters in life. That among many other things undoubtfully helped me be at this particular scientific stage. xm CHAPTER I Introduction Gelsolin, a protein comprised of six domains (G1-G6), controls the dynamic assembly and disassembly of the actin cytoskeleton and clears the circulation of potentially dangerous actin filaments released following cell death (reviewed by Kwiatkowski et al. 1999; Sun et al. 1999). The actin cytoskeleton is a network of actin filaments, which are polar linear polymers of actin monomers that preferentially assemble at one end (barbed end) and dissociate at the other (pointed end). Remodeling of the actin cytoskeleton is necessary to carry out a variety of cellular processes (reviewed by dos Remedios et al. 2003) and gelsolin's contribution to such rearrangements involves severing of actin filaments and capping of their newly created barbed ends to block annealing of filamentous fragments. A. Actin 1. Characteristics Actin is one of the most abundant proteins in eukaryotic cells and is an essential component of their cellular cytoskeleton. It exists in dynamic equilibrium between two principle forms, a monomeric (G-actin) and a filamentous (F-actin) form. Constant and diverse remodeling of actin filaments is required for a number of cellular processes, such as migration, endocytosis, cytokinesis, cell morphogenesis, apoptosis, signal transduction and others (reviewed by dos Remedios et al. 2003). One of the most important roles of the actin cytoskeleton is in cellular locomotion through extension of pseudopods. In unicellular organisms, this movement is essential for acquiring nutrients. In higher eukaryotes, fibroblasts are dependent on it for proper wound healing, platelets for the clotting process and immune cells for chemotactic movement (reviewed by Pollard et al. 2000). Act in is a major cytoplasmic protein but is found in the nucleus as well . It is proposed to be involved in the transcription process at several levels. The main hypothesis suggests that actin acts as a conformational switch, to control the assembly and activity of the chromatin remodeling apparatus and the remodeling of transcriptional 1 machines. Recently, the involvement of actin in coordination of extracellular signals and transcription activity has been explored widely (reviewed by Percipalle and Visa 2006). 2. The structure of an actin monomer Actin is a 42 kDa globular protein. Its structure in isolated, unmodified form has not yet been solved due to its tendency to oligomerize at the high protein concentrations and in chemical environments employed in protein crystallization trials. However, its basic structure has been inferred from the structures of actin complexes with a plethora of actin-binding proteins (ABPs) , such as DNase I (Kabsch et al. 1990), profilin (Schutt et al. 1993), gelsolin domain 1 (McLaughlin et al. 1993), gelsolin C-terminal half - domains G4-G6 (Robinson et al. 1999; Choe et al. 2002) and gelsolin N-terminal half - domains G1-G3 (Burtnick et al. 2004). Recently the crystal structures of modified actin monomers, rendered non-polymerizable, have been solved. These include actin chemically modified with tetramethylrhodamine (TMR) at Cys374 in subdomain 1 (Otterbain et al. 2001) and actin mutated at A204E and P243K in subdomain 4 (AP-actin) (Rould et al. 2006). General features of the inferred and non-polymerizable actin structures include two pairs of subdomains, 1&2 and 3&4, with a nucleotide, A T P or A D P , bound in a cleft in the center of the molecule. The nucleotide phosphate groups participate in coordinating 2"T" 2~F a divalent metal ion, usually M g or Ca , in a high affinity metal ion binding site. Both nucleotide and divalent cation affinities reside in the nanomolar range. Overall dimensions of the monomer are 55 x 55 x 35 A. Three major conformational representatives can be distinguished among the observed monomer forms. A s reviewed in Aguda et al. (2005), most actin structures fall into a closed state ATP conformation category (Figure 1A). The conformation of a subdomain 2 loop (D-loop) and slight variations in the angle between the two halves of actin appear as minor differences among these closed-form structures. The closed state ADP-actin conformation (Figure 1C) includes differences in the nucleotide cleft region and D-loop (Otterbein et al. 2001; Rould et al. 2006) and folding of the D-loop into a helix in the case of the TMR-modif ied actin structure (Otterbein et al. 2001). In the ATP open form (Figure IB) , subdomain 2 is rotated away from subdomain 4, which results in an open 2 nucleotide-binding cleft that facilitates nucleotide exchange. The only example of this classification is a profilin:actin complex. The conformation of an actin monomer/subunit within the context of an actin filament, i.e., once it is in contact with neighbouring actin subunits, is expected to be different from that o f actin in true isolation (reviewed by Aguda et al. 2005). For example, experimental data of Rould et al. (2006) and Otterbein et al. (2001) suggest that full expression of nucleotide-dependent changes upon A T P hydrolysis, e.g., D-loop ordering into a helix, requires stabilization via actin-actin contacts. It is emerging that differences between closed A T P and A D P conformations, realization o f which might be dependent on the state o f polymerization of actin, are fundamental to filament dynamics and lifecycle, and may affect filament interactions with a multitude of actin associated proteins (Zheng et al. 2007). Closed A T P Open A T P Closed A D P Figure 1. M a j o r actin monomer conformations A) An example of the closed form of the actin-ATP monomer (Protein Data Base (PDB) code 2BTF). The nucleotide (ATP) is shown as a ball-and-stick representation and the metal ion as a green sphere. The subdomains are coloured yellow (1,2) and blue (3,4). B) The open form of the actin-ATP monomer (PDB code 1HLU). C ) Closed structure of the actin monomer with ADP bound in the cleft (PDB code 1J6Z) [from Aguda et al. 2005]. 3. Properties and structure of actin filaments Act in filaments are linear polymers, composed of two staggered parallel rows of non-covalently linked actin monomers, twisted into a right-handed helix of 7-8 nm in diameter (reviewed by dos Remedios et al. 2003). They normally are assembled from ATP-containing monomers. The suggested role of A T P is in prevention of incorrect 3 binding of the incoming monomers to the growing filament (Becker 2006), hence ensuring stability of "young", highly dynamic filaments (reviewed by Pollard and Borisy 2003). A T P hydrolysis is not required for polymerization and takes place only once newly attached actin monomers are shifted towards the opposite end of the filament as a result of further filament growth. A s a result of A T P hydrolysis, three conformationally different regions can be distinguished within a mature filament. The fast growing (barbed or plus) end, where preferential assembly takes place, consists of ATP-bound actin subunits. The middle (and major) section is composed of A D P - P i actin subunits, and the slow growing (pointed or minus) end contains ADP-act in units. Inorganic phosphate release leads to conformational changes within the ADP-act in subunits that are associated with general destabilization of this segment of the filament. Such instability is a marker for depolymerization. Directional filament growth, with preferential monomer association at one end and dominant monomer dissociation at the other, called treadmilling, is driven by irreversible hydrolysis of A T P and higher A T P over A D P concentrations in the cell (reviewed by Pollard et al. 2000; Steinmetz et al. 1997). It is obvious that the major physiological role of actin ATPase activity is the utilization of the energy released from A T P hydrolysis for mechanical force generation through the directional polymerization of actin filaments. It is not only important as a factor in controlling the interactions of actin with itself, but with a host of actin associated proteins as well . Several F-actin binding proteins preferentially interact with A T P or A D P filament segments, which is crucial for the regulation of filament dynamics (reviewed by Pollard et al. 2000). The atomic resolution structure of filamentous actin remains unknown, despite the fundamental importance of its knowledge. However, using low resolution structural data from aligned filaments, together with the atomic resolution structure of monomelic actin, several models of the filament have been proposed. The first and most accepted is the Holmes fiber-diffraction model, published over 15 years ago (Holmes et al. 1990) (Figure 2 A and B) . In the electron micrographic reconstruction model, actin monomers have an open nucleotide cleft region (Belmont et al. 1999), in contrast to the closed one in the Holmes model. Third, a fundamentally different model is based on the observed actin-4 actin crystal contacts in crystals of the profilin:actin complex, which puts subdomains 1 and 2 on the central filament axis (Schutt et al. 1993). Figure 2. F o u r act in monomers ar ranged according to the Holmes fi lament model . Actin subunits shown in yellow and purple belong to one strand and monomers colored green and blue represent the second strand of the actin filament (Holmes et al. 1990). A) Subdomains 3 and 4 of each actin subunit are facing the filament axis, while subdomains 1 and 2 are located on the outside of the filament. B) Image A rotated by 90 degrees around the filament axis (images were generated from coordinates for the Holmes filament model (Holmes et al. 1990), with the use of UCSF Chimera software). Recently accumulated data on filament binding proteins in their respective complexes with an actin monomer strongly affirm the Holmes model (reviewed by Aguda et al. 2005). The interactions of these proteins with actin monomers are expected to reflect those with F-actin, even after allowing for minor conformational changes due to incorporation into filaments. Superposition of such proteins, e.g., the N and C-terminal halves o f gelsolin, W H 2 (Wiscott-Aldrich syndrome protein homology 2), Arp2/3, and B n i l formin homology domain 2, onto a single actin monomer was used to reassess filament models in silico (Figure 3). The figure reveals how extensively one surface of an actin monomer is utilized by filament binding proteins (Figure 3A and B , shown in red), leaving a free surface (Figure 3B) to form actin-actin contacts. This result is compatible with two filament models: the original Holmes closed-cleft (Holmes et al. 1990) and the Holmes-like open cleft (Belmont et al. 1999). Additional analysis of filament binding protein interactions with actin monomers favors the closed nucleotide cleft conformation, i.e. the original Holmes model (reviewed by Aguda et al. 2005). Figure 3. In silico superposi t ion of some F-act in-binding proteins ( W H 2 , G 1 - G 3 , G 4 - G 6 , arp2/3 complex, B n i l p F H 2 ) onto a single actin monomer. F-actin-binding proteins are shown in red. Actin subdomains 2 and 4 are indicated. A) actin monomer is in the same orientation as in Figure 1. B) Image A rotated to a position looking down the nucleotide cleft, towards a barbed end [from Aguda et al. 2005]. Nonetheless, definitive experimental proof, i.e. a high resolution structure, of the Holmes model is lacking. Therefore, much research effort is being focused on finding "the holy grail" of the actin field: a high-resolution structure of F-actin. 6 4. Actin life cycle Physiological cellular concentrations of M g C h (1-2 m M ) and KC1 (75-150 m M ) are high enough to promote virtually complete conversion of G into F-actin. However, in vivo, roughly the same amounts of actin are present in monomeric and filamentous forms (reviewed by Steinmetz et al. 1997). Furthermore, spontaneous polymerization rates determined for pure actin are not fast enough to produce the rapid filament dynamics observed in vivo. Regulation of monomer-polymer transitions in cells is governed by a plethora of G and F-actin binding proteins (reviewed by Ono 2007) (Figure 4). Nucleation of de novo filaments is attributed to the Arp2/3 and formin families. ATP-actin normally is sequestered by thymosin 04, to prevent spontaneous nucleation into filaments. Profilin is able to exchange ATP-act in interactions with thymosin 04 and to feed the ATP-monomers to the above-mentioned nucleators, or to existing barbed ends for further filament elongation. F-actin capping and severing proteins, such as gelsolin, are involved in the control of filament length/growth, as well as in the formation of actin-oligomers necessary for rapid redirection of filament growth. Filament branching is facilitated by the Arp2/3 complex (Figure 4). Alongside gelsolin, the ADF/cof i l in group of proteins is a crucial contributor to the overall regulation of the filaments. These proteins preferentially bind to ADP-act in regions of F-actin, near the pointed ends, and cause monomer dissociation that results in filament depolymerization. Profilin then catalyzes A D P for A T P exchange on these dissociated actin monomers that are subsequently stored in the thymosin pM/ATP-actin pool, completing an actin recycling pathway (reviewed by Pollard et al. 2000, reviewed by Ono 2007). Activities of these proteins are modulated in response to numerous cell signaling events. 7 membrane piofi l i i i f c ' O selsohn A T P A D P tiWl i t ^  j • i t -oC SE\TRING C i » U N C A P P I N G B R A N C H I N G \ / 0— thymosin ^* M ^ I DEBR.ANCHING / BUFFERING EXCHANGE • n f i b i i D E P O L \ T M E R I Z A T I O N Figure 4. Schemat ic representat ion of act in l ife cycle. Ge lso l i n : actin filament-binding, severing and capping protein (blue caps), Arp2 /3 complex: responsibe for nucleation and branching of filaments (green), C o f i l i n : actin filament-depolymerizing protein (rose stars), P r o f i l i n : actin monomer nucleotide exchanger (yellow ovals), P - thymosin: actin monomer-buffering protein (black clamps). Ac t in subunits within the filament, bound to A T P are represented in red and those bound to A D P in light blue (figure was adapted from Pollard et al. 2 0 0 0 ) . B. Gelsol in 1. Properties and function Gelsolin is the founding member of one superfamily of actin-binding proteins that controls actin dynamics in eukaryotes by severing actin filaments and capping filament ends, both intra- and extracellularly (reviewed by Sun et al. 1999; Sillaci et al. 2004). Additionally, it is involved in the control of cell morphology, growth and apoptosis (reviewed by Sun et al. 1999; Kwiatkowski 1999). Severing involves disruption of non-covalent (hydrophobic and electrostatic) interactions between actin units in an actin filament, after which gelsolin remains bound at the newly generated barbed end to 8 produce a capped filament to which monomers can no longer add. Gelsolin is indirectly involved in the nucleation of filament growth, through creation of free barbed ends as a result of their uncapping (reviewed by Sillaci et al. 2004). Gelsolin was first discovered in rabbit lung macrophages (Yin and Stossel 1979). Three gelsolin isoforms have been characterized and all arise by alternative splicing from the same gene. Cytoplasmic gelsolin is expressed in a wide variety of cell types. Its typical intracellular concentration is around 5 p M (reviewed by dos Remedios et al. 2003). This form facilitates rapid rearrangement of the cytoskeleton to enable fast motility that is crucial in cell types involved in stress responses such as hemostasis, inflammation, and wound healing (Witke et al. 1995; reviewed by dos Remedios et al. 2003). The importance of gelsolin's involvement in cell motility has been observed in many cases (Cunningham et al. 1991): locomotion rates are dependent on gelsolin expression levels, motility of cells treated with specific gelsolin inactivating peptides is reduced, and severe impairment of fibroblast movement occurs in gelsolin-knockout mice. Gelsolin-3 is a second cellular form of this protein. It is 11-amino acids longer at the N-terminus than the cytoplasmic form and is primarily localized in oligodendrocytes of brain tissue (Vouyiouklis and Brophy 1997). The only secreted form of the protein is represented by plasma gelsolin (pGSN). It differs from the original cytoplasmic variant by a 25 amino acid N-terminal extension. p G S N plays an important role in the blood stream, acting as an actin filament scavenger, clearing the circulation of potentially dangerous actin filaments (Vasconcellos and Lind 1993). A critical decrease in plasma gelsolin levels has been detected in many cases of tissue injuries (Suhler et al. 1997). A beneficial effect of therapeutic treatment with exogenous p G S N has been demonstrated through increased survival rates of mice subjected to lethal doses of endotoxin (Lee et al. 2007). A hereditary point mutation of A s p l 8 7 to Tyr or Asn in plasma gelsolin initiates several proteolytic events (Chen et al. 2001) that lead to amyloid fibril formation. Their accumulation outside cells affects' a range of dermatological, neurological and ophthalmologic disorders (Maury et al. 1990). The gelsolin superfamily includes vi l l in , adseverin, capG, advillin, supervillin and flightless I. V i l l i n is primarily localized in the intestinal epithelial cells and, in addition to severing, is able to bundle actin filaments (Ferrary et al. 1999). CapG is a capping, but 9 not a severing, protein and is found in both the cytoplasm and the nucleus (reviewed by Sillaci et al. 2004). Adseverin has been associated with control of cortical F-actin dynamics in the process of cellular secretion (Trifaro et al. 2000). In vivo, micromolar Ca ion concentrations are responsible for gelsolin's ability to bind, sever and cap F-actin (Yin and Stossel 1979; reviewed by Kwiatkowski 1999; Sun et al. 1999). Multiple Ca ion-binding events lead to drastic conformational changes that expose actin-binding sites within gelsolin (Pope et al. 1997; Kiselar et al. 2003; Burtnick et al. 2004; Choe et al. 2002). Artificial phosphorylation (Takiguchi et al. 2000) and low pH (Lagarrigue et al. 2003) enable gelsolin activation in vitro. Interactions of gelsolin with F-actin are inhibited by polyphosphoinositides (PPIs), which also induce dissociation of gelsolin bound to actin-filaments (Janmey and Stossel 1987). Recently, it was reported that lipid ligands, such as lysophosphatidic acid ( L P A ) and bacterial endotoxin (LPS), once bound to gelsolin, cause similar effects to those of PPIs (Meerschaert et al. 1998; Bucki et al. 2005). A T P was identified as a specific ligand to gelsolin (Yamamoto et al. 1989; Ito et al. 1990; Kambe et al. 1992), but the physiological role of this interaction remains uncertain. Protein binding partners of gelsolin include tropomyosin (Koepf and Burtnick 1992) and calponin (Ferjani et al. 2006) . Complex interactions between these proteins and gelsolin suggest that under certain cell conditions, they might take on a role as gelsolin regulators (Ferjani et al. 2007) . In addition to its primary role in actin filament remodeling, gelsolin has some particular roles in control of apoptosis and cancer progression. A lot of seemingly contradictory data have been accumulated on gelsolin suppressor and activator roles in cancer (Asch et al. 1996; Van den Abbele et al. 2007). However, experimental data (Rao et al. 2002; Shieh et al. 2006) support a dual role for gelsolin in carcinogenesis, dependant on the cancer stage. Analogously, pro and anti-apoptotic activities of gelsolin have been reported (reviewed by Sillaci et al. 2004). The N-terminal half of gelsolin, generated during apoptosis, contributes to uncontrolled cytoskeletal dismantling that leads to cell death (Kothakoa et al. 1997), while the full length and C- terminal half of gelsolin inhibit other apoptotic events, such as mitochondrial membrane potential loss and cytochrome c release (Koya et al. 2000; Kusano et al. 2000). 10 2. The structure of inactive gelsolin Gelsolin consists of six 120-130-amino acid domains that are similar in sequence and structure, but possess distinctive properties (Figure 5) (Vandekerckhove et al. 1990; Koepf et al. 1998; Burtnick et al. 1997). The structure of each domain is characterized by a central five or six stranded P-sheet that is sandwiched between a long a-helix (HI) running approximately parallel to the strands and a shorter a-helix (H2) almost perpendicular to the strands (Figure 5). The domain pairs, G I and G4, G2 and G5, G3 and G6, each have distinct structural characteristics that are not shared with the other domains (Burtnick et al. 1997). Based on sequence similarity (Kwiatkowski et al. 1986; Koepf et al. 1998), these domains arose as a result of gene triplication that was followed by a duplication event. Figure 5. Ind iv idua l gelsolin domains excised f rom the C a - free structure of horse plasma gelsolin (Burtnick et al. 1997). Common P-sheets and a-helices are identified in domain 5. The same coloring scheme will be used in subsequent figures [from Choe et al. 2002]. 1 1 G-actin-binding sites have been identified in domains G l and G4 (Kwiatkowski et al. 1985). G l binding to G-actin proceeds independently of Ca with a fCj ~ 5 p M , in contrast to the Ca 2 +-dependent and weaker binding (Kd ~ 1.8 p M ) of G4 (Bryan 1988; Pope et al. 1995). A s part of the larger construct (G4-G6), G4 affinity towards actin is increased to K<i ~ 25 n M . G l and G4 competition for the same binding site on actin (Pope et al. 1991) was confirmed through structural studies (McLaughlin et al. 1993; Choe et al. 2002; Burtnick et al. 2004). The extended C and D strands (Figure 5), unique to G l and G4 have been implicated in severing activity (Burtnick et al. 1997). In the context of intact gelsolin interactions with F-actin, these domains are expected to bind two different actin protomers. A n F-actin binding site is located in G2 (Bryan 1988; Way et al. 1992) and establishment of G2/actin interactions is suggested to be the first event in the initiation of severing. G5 , structurally very closely related to G2, does not share its functional characteristics. G3 and G6 do not possess primary actin-binding sites. However, as integral components of G1-G3 and G4-G6, respectively, they do make secondary contacts with actin (Figure 9 A and C) (Choe et al. 2002; Urosev 2003; Burtnick et al. 2004). A distinctive feature of both G3 and G6 in the inactive gelsolin structure is a kinked long helix, required to avoid steric clashes in the formation of the interdomain P-sheets from G l to G3 and G4 to G6, respectively (Figures 5 and 6). Figure 6. The two gelsolin halves excised f rom the C a -free s tructure of horse plasma gelsolin [from Burtnick et al. 1997]. 12 In the absence o f Ca ions, these six domains cluster into two sets, G1-G3 (N-terminal half of gelsolin) and G4-G6 (C-terminal half of gelsolin), o f highly similar pseudoquaternary structure (Figure 6). Dominant interactions within each half are between G1 and G3 and, analogously, between G4 and G6, via a continuous 10-stranded p sheet. Within the gelsolin superfamily, the C-terminal helical extension of G6 is an attribute unique to gelsolin (Figure 7). In the inactive conformation, the two gelsolin halves are packed together and "locked" through interactions between the C-terminal helix and the long helix of G2 (Figure 7) (Burtnick et al.1997). In this conformation, all actin-binding sites are obscured. Disruption of the G2/G6 interactions is suggested to be 2_|_ the first event in the process of Ca activation (Choe et al. 2002; Burtnick et al. 2004). Gelsolin may have evolved the unique tail latch mechanism to achieve finely tuned regulation of severing. Figure 7. A r ibbon representation of the structure of equine p lasma gelsolin determined in the absence of bound c a l c i u m ions. The overall molecular dimensions are approximately 85 x 55 x 35 A [from Burtnick et al. 1997]. 13 3. Ca -binding in gelsolin 9-F The presence of Ca ions is necessary for the binding and severing of actin filaments by gelsolin (Pope et al. 1991). Gelsolin activation is a multistep process involving many calcium ion-binding events that result in the transition of the molecule from a compact structure to a more open form in which three primary actin-binding sites become exposed (Pope et al. 1997; Choe et al. 2002; Kiselar at al. 2003; Burtnick et al. 2004). Activation proceeds through release of three identifiable latches: the tail latch (G2-G6) (Figure 7), and the G 1 - G 3 and G 4 - G 6 latches (Figure 6) (Robinson et al. 1999; Choe et al. 2002). The C-terminal helix imposes Ca dependency on gelsolin interactions with actin (Hellweg et al. 1993; Kwiatkowski et al. 1989; L i n et al. 2000). Tryptophan fluorescence (Kinosian et al. 1998; Gremm and Wegner 1999; L i n et al. 2000), equilibrium dialysis (Pope et al. 1995; Weeds et al. 1986) and dynamic light scattering (Pope et al. 1997) have been used to examine the binding of C a 2 + to gelsolin and its associated structural consequences. According to numerous studies, somewhere 2_|_ from two to eight Ca ions are implicated in the activation process. Two sites with IQ -0.2-0.7 p M have been proposed to reside in G6 and the N-terminal half, most likely in G2. The next highest affinity site, with Kd - 2 p M , has been associated with the C-terminal half (Pope et al. 1995; Pope et al. 1997; Kinosian et al. 1998; L i n et al. 2000). Ditsch and Wegner (1995) identified two lower affinity sites, in a range of ICj -25 p M -200 p M , to be associated with actin-binding events, in agreement with a reported 100 p M C a 2 + requirement for actin-binding (Gremm and Wegner, 1999; Al len and Janmey 1994). Radiolytic protein footpririting experiments (Kiselar et al. 2003) suggested that the tail latch opening and some conformational change within the C-terminal half occur at submicromolar Ca concentrations, and that the transition into a final activated structure is related to occupation of lower affinity sites in both the N and C-terminal halves of the protein. Protein crystallography has identified two types of C a 2 + ion-binding sites and a total of eight C a 2 + ions associated with gelsolin. Type-1 Ca 2 +sites are located at a gelsolin interface with actin, and engage domains G l and G4. Ca coordination in both cases is achieved in an analogous manner and involves G l u l 6 7 from actin and conserved Asp 14 residues, Asp 109 in G I and Asp487 in G4 (Figure 8A) (McLaughlin et al. 1993; Robinson et al. 1999; Choe et al. 2002). These sites moderate the affinity of activated gelsolin for actin. Type-2 C a 2 + sites are contained within gelsolin exclusively and are responsible for conformational changes that activate gelsolin. They have a dual purpose: to facilitate the observed disruption of Ca 2 +-free gelsolin; and subsequently, to stabilize the activated form (Choe et al. 2002). This is possible due to the involvement of interdomain linkers in the Ca coordination spheres at Type-2 sites (Figure 8B). One conserved Type-2 site has been identified in each gelsolin domain (Figure 8B) (McLaughlin et al. 1993; Kazmirski et al. 2002; Choe et al. 2002; Burtnick et al. 2004). Figure 8. Types of C a 2 + - b i n d i n g sites in gelsolin. Domain GI is used as an example: A) Type-1 Ca 2 +binding site: C a 2 + (golden sphere) coordination is completed by Glul67 of actin, a conserved Asp residue in GI (Asp 109) and two carbonyl oxygens further down the chain. B ) Type-2 Ca 2 +binding: C a 2 + (black sphere) is coordinated by a conserved glutamate residue (Glu97) in the long (HI) helix, a conserved aspartic residue prior to strand C and a carbonyl oxygen from the residue prior to the aspartate. Variations among the sites in different domains reside in the 4 t h coordinating residue, which is located either within that domain or belongs to the interdomain linker region [from Choe et al. 2002]. Type-1 C a b ind ing site Type-2 C a 2 + b ind ing site 15 4. Mechanism of gelsolin activation The active conformation of intact gelsolin has not been determined. However, crystallographic structures of activated Gl-G3/act in (Burtnick et al. 2004), and of activated G4-G6 by itself and in complex with an actin monomer are available (Choe et al. 2002; Robinson et al. 1999; Narayan et al. 2003) (Figure 9). A B C Figure 9. The structures of A) activated G 1 - G 3 bound to act in , B ) activated G 4 - G 6 and C ) activated G 4 - G 6 bound to act in. Actin monomer (in gray) subdomains are labeled 1-4. Type-1 C a 2 4 ions are shown as golden spheres and Type-2 C a 2 + ions are represented as black spheres. Actin-bound ATP and its associated C a 2 + ion are located in the middle of each actin monomer [from Burtnick et al. 2004; Narayan et al. 2003]. These reveal that the major changes upon activation of both N and C-terminal halves of gelsolin do not involve substantial intradomain rearrangements. The idea of interdomain movement being more important is supported by comparing G l in the Gl /ac t in structure (McLaughlin et al. 1993) with G l in Ca 2 +-free gelsolin (Burtnick et al. 1997), then by comparing individual domains in inactive gelsolin and its activated fragments. The interdomain linker peptides permit repositioning of individual domains with respect to each other (Figure 10A and B) . The most dramatic event during activation is disruption of the extended |3-sheets in G1-G3 and G4-G6. In the C-terminal half, this outcome is followed by rotation o f G6 away from G4 to establish new contacts with G5, resulting in 16 exposure of the actin-binding surface on G 4 (Figure 1 OA). Structural changes in the N-terminal half of gelsolin follow a very different path, reflecting the fact that G 2 can bind to the side of an actin filament, while G 5 does not. An extensive G 2 - G 3 interface is formed, while G I extends away from G 2 . This is achieved through displacement of the A' strand (residues 1 3 7 - 1 4 1 ) from the edge of the core P-sheet of G 2 and extension of the G 1 - G 2 linker. The new 3 0 A long linker enables G I and G 2 each to bind to a distinct site on the same actin monomer (Figure 9 A and 1 0 B ) . F i g u r e l O . Conformat iona l changes in gelsolin A ) C- te rmina l ( G 4 - G 6 ) and B) N- t e rmina l ( G 1 - G 3 ) halves upon b ind ing C a 2 + . Spheres colored black represent Type-2 C a 2 + ions and gold spheres represent Type-1 C a 2 + ions (Urosev 2003; Burtnick et al. 2004). When compared, the structures of calcium-bound G 4 - G 6 are essentially identical in the presence and absence of actin (Figure 9B and C). It follows that Ca2+-binding is solely responsible for the C-terminal half adopting a shape compatible with actin-binding 17 (Narayan et al. 2003; Choe et al. 2002). Despite the similarities in the location and number of Ca 2 + -binding sites in the two halves of gelsolin, the involvement of actin in attaining the observed conformation of G1-G3 cannot be excluded, primarily due to the important differences in the properties of G2 and G5. Nonetheless, the observed conformational changes, enabled through additional assistance from actin or not, are expected to reflect the general activation process in this half of gelsolin within the intact protein. When taken together, both Type-1 and five out of six Type-2 Ca 2 + -binding sites are occupied in the activated Gl-G3/act in and G4-G6/actin structures (Figures 9 and 10). A Type-2 C a 2 + ion is absent in G2 of Gl-G3/act in , although all o f the residues necessary for ligation are present (Choe et al. 2002). C a 2 + is able to bind to this isolated domain in solution (Huff et al. 2003) and the site is occupied by C d 2 + in the structure of isolated G2 (Kazmirski et al. 2002). It still remains to be revealed how occupation of this site in gelsolin comes into play. For example, it is possible that Ca 2 + -binding at this site is a transient event, required for disruption of the calcium-free structure, but then the C a 2 + is released once the final activated structure is achieved (Burtnick et al. 2004). Based on numerous functional and structural Ca 2 + -binding studies mentioned in the previous section, the following C a 2 + activation scheme for gelsolin was proposed. It involves all six Type-2 C a 2 + ions. Level A : A t approximately 0.2 p M C a 2 + concentrations (Pope et al. 1995; Kinosian et al. 1998), binding proceeds at the Type-2 site within G6 (with possibly simultaneous occupation of the Type-2 site in G2). Tail-latch opening is initiated (Figure 11A). Level B : This latch is fully released at 2 p M C a 2 + , with binding in G4 (Pope et al. 1995; Kinosian et al. 1998), which is accompanied by some domain repositioning in G4-G6 (Figure 1 IB) . Level C: At approximately 100 p M C a 2 + , binding to G5 leads to attainment of a fully opened G4-G6 (Kiselar et al. 2003). Simultaneously, binding proceeds in G3, which opens the G1-G3 latch and allows the G2 high affinity site with K j ~0 .7pM (Chen et al. 2001) to be occupied (alternative scenario to occupation of this site at Level A ) . G1-G3 is driven towards its final activated state, which subsequently results in Ca 2 + release from G2 (Figure 11C). Level D : At elevated C a 2 + concentrations, ~1 m M (Zapun et al. 2000), the G l site is occupied, yielding a final protein conformation capable of F-actin severing with full potential (Figure 1 ID). 18 100 uM 1 mM Figure 11. Schematic representation of gelsolin activation by C a 2 + ions. C a 2 + free gelsolin domains are depicted as hexagons and calcium-bound domains are shown as ovals. Ca2+"concentrations are indicated for each step [modified from Burtnick et al. 2004]. In vitro, low p H (6.0-6.5) activates gelsolin in the absence o f calcium (Lamb et al. 1993; Lagariggue et al, 2003). The latter authors propose almost the reverse direction of conformational changes when compared to that above for C a 2 + activation, a result of straining the interactions within the inactive conformation, which results in opening of the tail latch as the final event (Lagariggue et al, 2003). pH changes are known to occur in cells in response to certain signaling events. At least one of the other major effectors o f actin filament stability, the ADF/cof i l in family of proteins, is demonstrated to be pH-dependent (reviewed by dos Remedios et al. 2003). The possibility o f such gelsolin regulation in vivo requires further research. Activation of plasma gelsolin can, in addition, be reversibly regulated through artificial phosphorylation of a single site in the N-terminal extension (not present in the cytoplasmic form) that makes the molecule able to sever and cap actin filaments. Activation could be the result of charge repulsion between the acidic C-terminal end and the phosphorylated N-terminal end (Takiguchi et al, 2000). 5. Models of F-actin severing and capping by gelsolin A s described earlier, gelsolin possesses three primary actin-binding sites, residing in G l , G2 and G4 (Kwiatkowski et al. 1985; Bryan 1988; Way et al. 1992; Pope et al. 1991), and two secondary ones, in G3 and G6 (Choe et al. 2002; Urosev 2003; Burtnick et al. 2004). Once activated, gelsolin proceeds to bind both actin monomers and, at a lower rate, actin filaments, with approximately nanomolar affinities (Selve and Wegner 1986; Schoepper and Wegner 1991; Schoepper and Wegner 1992). It interacts faster and with higher affinity with A D P - over ATP-F-actin. In vitro, activated gelsolin has been 19 observed to form complexes that involve direct contact with up to three actin monomers ( G A , G A 2 and GA3) (Edgar 1990). Binding to two actin monomers proceeds through G I and G4, which, in the context of filament binding, reflect the interaction with actin protomers belonging to two different strands (McGough et al. 1998). Electron micrographs of gelsolin-decorated actin filaments suggest that the G3-G4 linker extends across the two filament strands (McGough et al. 1998). Further association of gelsolin with a third actin monomer is enabled through the F-actin binding site in G2. On addition 2_j_ of the Ca -chelator, E G T A , actin monomers held in a complex with the C-terminal half (G4-G6) of gelsolin (Bryan and Kurth 1984) dissociate from both quaternary (GA3) and ternary (GA2) protein complexes (Harris et al. 1988), leaving a stable G A complex. The minimal F-actin severing and capping unit is composed of domain G I and the G1-G2 linker (Way et al. 1992; Kwiatkowski et al. 1985). For this " G I plus", as well as for the G1-G3, filament-binding and severing activities are independent of C a 2 + ions. On the other hand, G4-G6 requires Ca ions to bind monomeric actin and is not able to sever filaments. In the context of intact gelsolin, Ca 2 + -binding to G4-G6 is responsible for internal cooperativity between the two halves that gives high potency to gelsolin's severing function (Selden et al. 1998). G2-G6 was identified as a minimal construct capable of actin filament nucleating activity, which requires at least two actin-binding sites, supposedly those in G2 and G4 (Way et al. 1989). The smaller entity composed of residues 150-406 (G2-G3 and 80% of the G 3 - G 4 linker) is efficiently able to cap actin filaments (Sun et al.1994). The observed binding of G1-G3 to an actin monomer (Urosev 2003; Burtnick et al. 2004) dictates that actin subdomains 1 and 2 should face outward from the core of an actin filament, so that G2 is able to initiate gelsolin's filament-binding and severing activities by anchoring gelsolin to the side of the filament. Such an actin subunit orientation is in agreement with the Holmes filament model (Holmes et al. 1990). G l -G3/actin (Urosev 2003; Burtnick et al. 2004) and G4-G6/actin (Choe et al. 2002), superimposed onto the terminal actin protomers of the two ADP-filament strands of the Holmes model, suggest what a gelsolin cap might look like (Figure 12A and B) . Inspection of this model reveals the close proximity of G2 to a second actin subunit, the penultimate one in the same strand to which G I is bound. Comparison reveals striking 20 similarities in the orientation of the G2 long helix with respect to the second actin and that of GI with respect to the first (Figure 12 A ) . This is in agreement with earlier findings that each o f these two helices is the location of a major actin-binding determinant (van Troys et al. 1996). Also , the polypeptide that links the two halves of gelsolin is sufficiently long to span the 63 A gap across the actin filament. LINKER Figure 12. A model for an act in filament barbed end capped by gelsolin. A ) Gl-G3/actin (Urosev 2003; Burtnick et al. 2004) and G4-G6/actin (Choe et al. 2002), overlaid onto the barbed end of the A D P -F-actin model (four actin subunits are drawn in blue and gray). The linker is long enough to span the gap of 63 A between domains G3 and G4, and is modeled in purple. B ) A view of this model looking directly at the capped barbed end [from Burtnick et al. 2004]. Overall, the model accounts nicely for gelsolin's abilities to bind, sever and cap a filament, with G2 making contact at an appropriate position to explain subsequent severing through binding by GI and G4, and yielding an obscured surface at the newly created barbed end. 21 Fully activated gelsolin is proposed to act in the following manner (Figure 13). Upon release of the tail latch, the G2-G3 structural unit binds to the side of the filament, followed by the G1-G2 and G3-G4 linkers wrapping over the surface of the filament to direct G I and G4 toward their binding sites. Next, combined disruption by G I and G4 of the interactions between actin subunits causes sufficient steric strain in the filament to dissociate the two actins below the ones to which GI and G4 are bound. The result is a gelsolin-capped, newly formed barbed end (Burtnick et al. 2004). BINDS SITES SEVERING Figure 13. The sequence of events du r ing the severing of F-act in by fully activated gelsolin. A c t i n subunits are shown in cyan [from Burtnick et al. 2004]. 2_|_ Typical intracellular Ca concentrations are in the low micromolar range, while millimolar concentrations are present in plasma. Since full activation of gelsolin requires roughly 1 m M Ca (Kiselar et al. 2003), a variation on the severing mechanism outlined above is expected to occur intracellularly. The lack of steric clashes when Ca 2 +-free G1-G3 is positioned on actin by overlaying its G2 domain onto the G2 of the activated Gl-G3/act in structure (Figure 14) suggests that we should not exclude the possibility of the inactive form G1-G3 initiating binding to an actin filament. Once bound, further activation to open the G1-G3 latch, as required for directing GI towards its binding site, might proceed with assistance from actin (Burtnick et al. 2004). 22 G3 3 I Figure 14 M o d e l o f C a 2 + -free G 1 - G 3 interact ing with actin. G1-G3, excised from the Ca 2 t-free gelsolin structure (Burtnick et al. 1997), is positioned on actin by overlaying its G2 domain onto the G2 position found in the Gl-G3/actin structure [from Burtnick et al. 2004]. 6. Inhibition of gelsolin Polyphosphoinositides (PPIs) are important signal transduction molecules. In particular, phosphatidylinositol 4,5-bisphosphate (commonly abbreviated in the literature as PIP2, although PI(4,5)P2 would be more accurate) (Figure 15 A ) , has been demonstrated to modulate many actin regulatory proteins, including gelsolin (Janmey and Stossel 1987), capping protein (CapG), and the actin monomer-binding proteins cofilin, profilin and others (reviewed by Logan et al. 2006). The hydrophilic head of a PPI is characterized by a large inositol ring with phosphate moieties that carry negative charges at physiological pH. Gelsolin also binds PI(3,4)P 2 and PI(3,4,5)P 3 (Chellaiah et al. 1998). PIP 2-binding sites on gelsolin are located within residues 132-140, 161-172 and 621-634 (Janmey et al. 1992; Y u et al. 1992; Cunningham et al. 2001; Feng et al. 2001). Residues 132-140 and 161-172 form integral parts of domains GI and G2, respectively, in both the active and inactive forms of the protein. The third PIP2-binding site involves residues 621-634 in domain G6. Each of these peptide regions contains a positively charged K x K K consensus sequence. Ca + ions increase the affinity of phosphoinositide binding to full length gelsolin (Lin et al. 1997), but decrease their binding to the isolated C-terminal half (Feng et al. 2001). The authors postulate that, as the cytosolic C a 2 + concentration rises 23 during stimulation, gelsolin severs filaments. PIP2 dissociates it from the filament ends, creating the conditions necessary for rapid amplification of filament growth (Lin et al. 1997). Mechanistically, this is in agreement with the requirement for the exposure of PIP2-binding sites in the N-terminal half of gelsolin that are obscured when the Ca concentration is low (Burtnick et al. 1997). Gelsolin partitioning between the plasma 2_|_ membrane, cytosol and actin filament ends, therefore, depends on the [Ca ] / [PIP2] ratio. Phospholipid binding by gelsolin has been shown to modulate phospholipase C activity in vivo (Sun et al. 1997), as well as to inhibit caspase-3, one of the effectors of apoptosis (Azuma et al. 2000). It is interesting that PIP2, once bound to gelsolin, promotes gelsolin phosphorylation at a site within G4 (DeCorte et al. 1997; De Corte et al. 1999). The functional role of this modification is not yet known, but an analogous event is observed with profilin, another actin-binding protein. Figure 15. Structures of phosphate r i ch regulators of gelsolin. A ) P I P 2 with inositol ring shown in blue, B ) L P A and C ) L P S with l ipid A represented in blue. 24 Lysophosphatidic acid ( L P A ) is the simplest representative of a glycerophospholipid (Mintzer et al. 2006) (Figure 15 B) . It usually is generated enzymatically from membrane precursors and then secreted. It was shown in vitro that L P A is as effective as PIP 2 in dissociating gelsolin-actin complexes (Meerschaert et al. 1998). L P A is known primarily for its hormone-like and mitogenic activities, and for activation of its own G-protein coupled receptor (Goetzl et al. 2000). Goetzl and coworkers (2000), along with Meerschaert et al. (1998), identified L P A as a competitive inhibitor of PIP2. It also was found that extracellular L P A binds to plasma gelsolin with higher affinity (Ka ~ 6 nM) than to a previously identified abundant LPA-carrier and modulator, serum albumin (ICj -360 nM) (Goetzl et al. 2000). Experimental results suggest that gelsolin's role with respect to L P A is concentration dependent. A t normal plasma concentrations, gelsolin might act as a high affinity L P A carrier, protecting it from biodegradation. A t concentrations of 10% or less of normal gelsolin levels, as can be observed in the extracellular fluids of injured tissues, gelsolin may play the role of L P A transporter and may augment its activating effects on certain cell types. Recently, a third group of lipid ligands of gelsolin was discovered. Lipopolysaccharides (LPS) are a major component of Gram-negative bacteria outer membrane cell walls that can induce toxic shock (Figure 15 C). Such LPS can bind gelsolin with even higher affinity than L P A or PIP2, and compete with them for an overlapping binding site. L P S inactivates the actin-severing function of gelsolin (Bucki et al. 2005). A s in the case of L P A , gelsolin might have a role in controlling LPS-mediated effects on cells, such as induction of transcription of various cytokines (Bucki et al 2005; Lee et al. 2007) and inhibition of thrombin activity (Bucki and Pastore 2006). 25 7. The binding of A T P to gelsolin After PPIs and Ca , A T P was identified as a major gelsolin ligand. Yamamoto et al. (1989) identified it as a biospecific eluant of plasma gelsolin bound to matrix-immobilized Cibacron Blue F 3 G A in the presence of 1 m M E G T A (a calcium ion chelator) (Ito et al. 1990; Yamamoto et al. 1990). A t physiological salt conditions, affinities, represented by Kd values from 0.28 p M (Kambe et al. 1992) to 32 p M (Gremm and Wegner 1999), have been determined, with the variation likely arising from the different buffer compositions and methods employed. Nucleotides such as G T P , G D P and A D P were able to elute gelsolin from the above mentioned affinity column (Ito et al. 1990), but only G T P showed significant association with gelsolin (at six-fold higher Kd than for A T P ) (Kambe et al. 1992). N o A T P hydrolytic activity in gelsolin was detected (Kambe et al. 1992; Yamamoto et al. 1990). Under millimolar M g 2 + ion conditions, the affinity of gelsolin for A T P decreases ten-fold, while in the presence of more than 10 p M C a 2 + , almost no binding is detectable. On the other hand, A T P interferes with C a 2 + -activation of gelsolin (Kambe et al. 1992), but enhances the cooperativity of Ca 2 + -binding (Gremm and Wegner 1999). These findings suggest that the substantial decrease in A T P -binding ability of gelsolin in the presence of C a 2 + is a result of binding site disruption 2_[_ caused by the binding of Ca (Kambe et al. 1992). Within the binary complex of gelsolin and actin, A T P bound to actin monomers is not exchangeable, while in the ternary complex of gelsolin and two actin monomers, one of the two A T P molecules is exchangeable, but at a decreased rate compared to that for a free actin monomer (Tellam et al. 1986). A s mentioned earlier, adenine nucleotides have been shown to affect the binding of gelsolin to monomeric and filamentous actin. Binding to G-actin proceeds more rapidly and with higher affinity in the presence o f A D P relative to A T P (Laham et al. 1993). Seemingly in contrast, the rate of actin binding to gelsolin to form a 1:1 complex (GA) was reported to be enhanced in the presence of A T P (Gremm and Wegner 1999). These data suggest that gelsolin-ATP interactions, in some way, play a regulatory role in the formation or stabilization of these protein complexes. Perhaps, there exists a second, relatively low affinity site, either on actin, gelsolin or at a gelsolin/actin interface (Laham et al. 1993). Furthermore, gelsolin selectively severs actin-ADP filaments, but not the ones containing bound A D P - P i (Allen et al. 1996). Also , 26 G4-G6 displays a preference for ADP-containing actin, while G1-G3 binds to A T P - and ADP-G-ac t in with similar affinities. However, A T P , but not A D P , concentrations in the range of 500 p M inhibit the binding of G1-G3 to actin monomers (Laham et al. 1995). Gelsolin has been identified as a diadenosine 5',5"'-P1,P3-triphosphate ( A p 3 A ) binding protein as well , with an estimated Kd of 0.3 p M (Vartanian 2003). Diadenosine oligophosphates ( A p n A ) are composed of two adenosine moieties linked with 2-6 phosphodiester linkages through 5 ' - 5 ' " sugar positions and are considered a new class of signaling molecule (second messengers) in eukaryotes. They are synthesized by aminoacyl-tRNA synthetases through enzyme-bound aminoacyl adenylate and A T P ( A p 4 A formation), and from adenylate and A D P in the case of A p 3 A . A p n A s are metabolized by hydrolases into A T P , A D P or A M P . The number of phosphate residues in A p n A s seems to define the cell status and be important for differential modulation of signal transduction (reviewed by Kisselev et al. 1998). Drastically different ratios of A p 3 A / A p 4 A in cultured cells are associated with cell differentiation and apoptosis (Vartanian et al. 1997; reviewed by Kisselev et al. 1998). Cells undergoing apoptosis have significantly reduced levels of A p 3 A , but elevated A p 4 A , while differentiating cells are characterized by increased A p 3 A content and unchanged A p 4 A (Vartanian et al. 1997). Tumor suppressor activity of the protein Fhit is assumed to be associated with involvement of its Ap 3 A-complex in cytokine signaling pathways that control cell proliferation (reviewed by Kisselev et al. 1998). Contrary to A T P binding, A p 3 A binding to gelsolin is insensitive to the presence of divalent cations and is not affected by a 100-fold excess of A T P , A D P , A M P or A p 4 A (Vartanian 2003). The cell morphology changes when A p 3 A is added to C E M - S S cells, suggestive of a correlation between cytoskeletal disruption and A p 3 A effects on gelsolin activation. The discovery of a strong interaction between gelsolin and A T P suggests that this nucleotide and some of its derivatives, such as A p 3 A , might be involved in the regulation of one or more of the many functions of gelsolin. 27 8. Gelsolin in blood Plasma gelsolin (pGSN) is the only circulating actin-severing protein (Witke et al. 1995). It is present in blood at 200-300 pg/ml (Y in et al. 1984). One proposed role for plasma gelsolin is that, together with plasma vitamin-D binding protein (DBP), it functions as an extracellular actin scavenger system, responsible for removal of actin released by tissue injury. The ionic strength of blood plasma favors formation of actin filament aggregates. B y severing F-actin, plasma gelsolin promotes polymer disassembly, which leads to reduced blood viscosity and prevention of microcirculation impediments and of actin-induced platelet aggregation (Harris et al. 1980; Vasconcellos and Lind. 1993). Free monomeric actin released from filament ends is then sequestered by D B P and cleared from the circulation by the liver (Herrmannsdoerfer et al. 1993). In humans, plasma gelsolin depletion is associated with a variety of tissue injuries, such as acute liver failure, myocardial infarction, myonecrosis and sepsis (Suhler et al. 1997). These conditions all involve actin release into the circulatory system. Depletion of actin-scavenger capacity in the presence of continued actin release precedes the poor outcome in situations of severe organ damage (Lee et al. 2007). A second hypothesis for p G S N ' s role in blood is related to its binding to bioactive (inflammatory) mediators such as L P A (Meerschaert et al. 1998), L P S (Bucki et al. 2005), A p 3 A (Vartanian 2003) and platelet activating factor (PAF) (Osborn et al. 2007). Plasma gelsolin is proposed to control mediator-induced cell activation that leads to an inflammatory response by acting as their buffering agent. Osborn et al. (2007) have shown in vitro that plasma gelsolin, in combination with serum albumin, can inhibit L P A - and PAF-induced platelet activation. A s mentioned, following tissue injury or bacterial infection, actin released into the plasma depletes gelsolin. A t the same time, injury elicits the production of inflammatory mediators that, unbuffered by p G S N , can lead to severe secondary injury or even death. p G S N was also shown to be effective in reversing the LPS-mediated inhibition of thrombin activity (Bucki and Pastore 2006), further expanding the diversity of roles for gelsolin in hemostasis. Cytoplasmic gelsolin is involved in various aspects of blood cell function as well . For instance, platelets have an important role in hemostasis. A s a result of injury to blood 28 vessel walls, platelets become exposed to and adhere to collagen fibers. The resultant platelet activation involves formation of filopodia and cell spreading through the extension of lamellae. Lamellae are composed of orthogonal arrays of short actin filaments. Collagen-induced activation of platelets requires gelsolin for actin assembly in the formation of lamellae. A similar situation ensues on thrombin activation of platelets (Falet et al. 2000). Gelsolin's role in the platelet storage limitation problem is slowly emerging. Platelet storage for the purposes of blood transfusion is limited to only five days at room temperature, due to unacceptable amounts of bacterial growth. Refrigerated storage at 4° C would abrogate the bacterial growth problem but also would result in rapid decrease in viability of these platelets, due to the cold-induced activation of the platelets that leads to their aggregation. Platelet cooling has been found to lead to gelsolin activation, which would influence any actin remodeling response (Hoffmeister et al. 2001). These findings have potential importance in the search for better platelet storage conditions. 9. Finnish-type familial amyloidosis (FAF) Amylo id diseases stem from the endoproteolytic cleavage of a precursor protein, which results in generation of amyloidogenic peptides. These peptides have a tendency to aggregate and form deposition plaques in the extracellular matrix of tissues. F A F is caused by a hereditary mutation of Asp 187 in gelsolin to Asn or Tyr, which leads to two successive proteolytic events. First, the resulting exposure of the A r g l 72-Ala l 73 peptide bond leads to attack at this site by furin during transport of p G S N through the trans-Golgi network (Chen et al. 2001) (Figure 16). Then, various unidentified matrix metalloproteinases further process the cleaved gelsolin to yield a 68 kDa segment, which is degraded into 5 and 8 kDa amyloidogenic peptides (Page et al. 2005). These generate plaque deposits in tissues such as skin, and in arterial, neurologic and ophthalmologic ones (Maury et al. 1990). Several hypotheses have been outlined to explain the effects of this mutation on the susceptibility to proteolysis. After the structure of inactive gelsolin was solved (Burtnick et al. 1997), it was proposed that G2 within the D187N(Y) mutant might be degraded due to the loss of important A s p l 8 7 contribution to the network of interactions 29 that stabilize the central P-sheet of G2. N M R studies of isolated human G2 imply the existence of a localized conformational change that leads to exposure of the cleavage site in the mutant form (Kazmirski et al. 2000), while investigation of a slightly longer G2 fragment (134-266) demonstrated the similar stability of F A F and wi ld type G2 (reviewed by Robinson et al. 2001). The confirmation of the existence of a conserved Type-2 Ca -binding site in isolated G2 (Kazmirski et al. 2002) and the identification of the side chain of Asp 187 as a Ca 2 +-coordinator, as inferred by sequence and structural analysis (Burtnick et al. 1997), led to a new suggestion that the activated form, rather than the inactive one, might be the target for proteases. The inability of both D187N and E209Q (another residue involved in the coordination of C a 2 + ) mutants of isolated G2 to bind Ca was demonstrated and emerged as a determining factor in the domain's overall instability and susceptibility to proteolysis (Huff et al. 2003; reviewed by Page et al. 2004). Further investigation of the activated Gl-G3/ac t in structure (Figure 9A) (Urosev 2003; Burtnick et al. 2004), as reported in the Results section of this thesis, helps explain the relationship between the inability to bind C a 2 + and susceptibility to proteolysis. 30 Figure 16. Regions i n G 2 impl ica ted in F A F . G2, as excised from the structure of Ca2+-free plasma gelsolin (Burtnick et al. 1997). Aspl87 is the mutation site in FAF and the bond attacked by furin is located in strand B (purple) [from Burtnick et al. 2004]. 10. Gelsolin and actin-associated proteins Besides actin, gelsolin binds two other common muscle proteins, tropomyosin (Koepf and Burtnick 1992) and calponin (Ferjani et al. 2006) . Tropomyosin binds laterally to actin filaments. In muscle cells, it plays an important part in the regulation of actomyosin contraction and. in non-muscle cells, it appears to be involved in the stabilization of actin filaments and modulation o f the actin binding properties of certain actin-binding proteins. Calponin, a smooth muscle protein, is also implicated in actomyosin regulation. Furthermore, tropomyosin and calponin bind to each other (Childs et al. 1992). Tropomyosin binding to gelsolin inhibits actin filament-severing, possibly due to overlap of actin and tropomyosin binding sites on gelsolin. Both are located in gelsolin domain G 2 (Maciver et al. 2000). Calponin and gelsolin interact at two sites. One involves G I and another, which is Ca 2 +-sensitive, is located in G 4 - G 6 (Ferjani et al. 2006; Ferjani et al. 2007) . Unlike the case with tropomyosin, calponin binding does not affect the severing activity of gelsolin but does inhibit its filament-nucleation activity (Ferjani et al. 2006) . Preservation of its severing ability is in agreement with the more 31 recent findings that a gelsolin complex with two actin monomers still interacts with calponin (Ferjani et al. 2007). It is possible that tropomyosin and calponin might yet be identified as in vivo protein regulators of gelsolin. A l l three proteins are, to a certain extent, associated with regulation of signalling pathways (reviewed by Sillaci et al. 2004; Maciver et al. 2000; Ferjani et al. 2006). 11. Gelsolin and apoptosis Apoptosis is initiated through a protease cascade that includes the critical effector, caspase-3. During apoptosis, G 1 - G 3 is generated from cytoplasmic gelsolin via the actions of caspase-3, -7 and -9. Cleavage takes place between residues Asp352 and Gly353 (Kothakota et al. 1997; Azuma et al. 2000). Free from the regulatory C-terminal half, pro-apoptotic G1-G3 participates in the preparation of a cell for death through uncontrolled dismantling of its actin cytoskeleton. A s reported earlier in the thesis, G l -G3 is able to sever actin filaments in the absence of calcium ions (Selden et al. 1998; Kothakota et al. 1997). The structure of isolated G1-G3 in the absence of C a 2 + is unknown. Due to the high Ca 2 +concentration requirement for release of the G1-G3 latch (-100 pM) (Kiselar et al. 2003), it has been proposed that isolated calcium-free G 1 - G 3 adopts a structure similar to that observed for G 1 - G 3 in calcium-free whole gelsolin (Burtnick et al. 2004). A s indicated by the model in Figure 14, isolated calcium-free G l -G3 is expected to be able to initiate contacts with actin filaments. In contrast, intact gelsolin and G4-G6 exhibit anti-apoptotic activity by blocking voltage-dependant anion channels, inhibiting mitochondrial membrane potential loss and cytochrome c release (Koya et al. 2000; Kusano et al. 2000). Sakurai and Utsumi (2006) reported that during apoptosis, posttranslational N-myristoylation of the caspase-generated C-terminal half of gelsolin (G4-G6) also takes place. The mechanism by which N-myristoylation affects the anti-apoptotic properties of G4-G6 remains unknown, but unlike unmodified G4-G6, N-myristoylated G4-G6 is not localized at the mitochondrial membrane, but evenly distributed in the cytoplasm. Also , formation of a gelsolin/PIP2/caspase complex leads to inactivation of certain caspases and subsequent inhibition of apoptosis (Azuma et al. 2000). Recently, 32 gelsolin domain G5 was shown to inhibit apoptosis by blocking the binding of an H I V protein responsible for inducing apoptosis in T-cells, with its target being a mitochondrial channel protein (Qiao and M c M i l l a n 2007). It is evident that gelsolin plays a multifaceted role in apoptosis and that the balance between the opposing activities of gelsolin has a significant effect on the onset and progression of programmed cell death. 12. Gelsolin in cancer Alterations in cell motility/migration during metastasis that involve actin filaments are important in the progression of cancer. Hence, a number of actin-binding proteins that regulate the integrity of actin filaments have been studied with respect to their involvement in cancer. Several studies indicate that gelsolin acts as a tumor suppressor (Asch et al. 1996; Tanaka et al. 2005), while others point towards its role as an activator (Van den Abbeele et al. 2007; Yang et al. 2006; Langbein et al. 2006; Thompson et al. 2007). Down regulation of gelsolin in several types of cancer cells significantly reduces their invasive properties as well as cell aggregation (Van den Abbeele et al. 2007). The importance not only of proliferation, but of motility markers, in assessing tumor progression and prognosis, is emerging from the works of Yang et al. (2006). In patients with non-small cell lung cancer, higher expression of gelsolin was significantly associated with poorer prognosis, indicating that gelsolin has prognostic significance as a motility marker. On the other hand, partial or total loss of gelsolin was detected in breast cancers of humans and rodents (Asch et al. 1996). Insufficient gelsolin may be responsible for impaired regulation of the actin filament cytoskeleton as well as modulations of transduction pathways, for example, ones involving PIP 2 . Such aberrations might lead to changes in cancer cell growth. These findings suggest a role for gelsolin as a suppressor of oncogenic changes, but need further investigation (Asch et al. 1996). The seemingly contradictory findings of roles for gelsolin as an activator and as a suppressor might arise because of the different cancers investigated, as well as from the studies having been conducted at different stages of cancer development (Van den 3 3 Abbeele et al. 2007). Rao et al. (2002) and Shieh et al. (2006) present experimental data to support the dual role of gelsolin in carcinogenesis through the existence of biphasic gelsolin expression patterns, with low expression in premalignant lesions, followed by-high expression in invasive tumors. Limited gelsolin staining was reported in oral precancerous lesions, but staining increased in primary and metastatic oral cell squamous carcinoma lesions (Shieh et al. 2006). Gelsolin expression was decreased in premalignant and malignant urothelial carcinomas. With the increase in tumor grade and stage, gelsolin levels gradually increased as well (Rao et al. 2002). Such observations likely reflect a balance of the multiple cellular functions performed by gelsolin in tumor invasion and cell growth, the underlying mechanisms of which still need to be elucidated. C. Further insight into the regulation of gelsolin Gelsolin functions are regulated by calcium ions and phosphatidylinositol 4,5-bisphosphate (PIP 2) (Yin and Stossel 1979; Janmey and Stossel 1987). In addition, reports of strong interactions between gelsolin and A T P led to suggestions that A T P might be involved in some of the multiple functions of gelsolin (Yamamoto et al. 1989; Kambe et al. 1992). The complexity of the regulation of gelsolin by these and other ligands prompted us to seek further understanding in this area. In what follows, I report on the advances we have made. Firstly, we sought to clarify the interactions between A T P and gelsolin. Here we report the structure of ATP-soaked crystals of calcium-free plasma gelsolin. A T P interacts with both of the semi-independent halves of the protein (the N-terminal half, G1-G3, and the C-terminal half, G4-G6) and binding involves some of the same residues that comprise part of a region previously identified to bind PIP 2 , an inhibitor of gelsolin-actin interactions. With the help of molecular docking software, we further investigated aspects of the involvement of A T P in regulation of the association of gelsolin with actin, and, in part, this led us to propose a model for PIP 2-binding on the surface of inactive gelsolin. Numerous Ca -binding events occur during activation of gelsolin to expose its actin-binding sites and enhance subsequent interactions with actin filaments. We probed 34 the exchangeability of C a 2 + ions with T b 3 + ions, and vice versa, in Ca 2 + -binding sites present in the N-terminal half of gelsolin in order to assess their importance in the stabilization of activated gelsolin and its interactions with actin. The results also help to clarify a controversy over the state of occupancy of a Ca 2 + -binding site in G2 in activated gelsolin (Burtnick et al. 2004; Kazmirski et al. 2002). Finally, part of this thesis provides an account of the progress towards experimental verification of a model that we had proposed previously for an actin filament capped at its barbed end by gelsolin (Burtnick et al. 2004). The following gelsolin/actin complexes were investigated: gelsolin bound to two ATP-act in monomers ( G A 2 A T P ) , gelsolin bound to two ADP-act in monomers ( G A 2 A D P ) , and gelsolin bound to a cross-linked actin species. Advances in their formation, purification, crystallization and X-ray diffraction analysis are presented. Some of the results reported in this thesis have been published in the scientific literature.1 Chumnarnsilpa, S., Loonchanta, A . , Xue B . , Choe, H . , Urosev, D . , Wang, H . , Lindberg, U . , Burtnick, L . D . , and Robinson, R . C . (2006). Calc ium ion exchange in crystalline gelsolin. J. M o l . B i o l . 357, 773-782 Urosev, D . , M a , Q. , Tan, A . L . , Robinson, R . C , and Burtnick, L . D . (2006). The structure o f gelsolin bound to A T P . J. M o l . B i o l . 357, 765-772 Burtnick, L . D . , Urosev, D . , Irobi, E . , Narayan, K . , and Robinson, R . C . (2004). Structure o f the N -terminal half o f gelsolin bound to actin: roles in severing, apoptosis and F A F . E M B O J. 23, 2713-2722 35 CHAPTER II Materials and Methods A. Protein Purification 1. Horse plasma gelsolin preparation The procedure is based on that of Bryan (1988) for the isolation of human plasma gelsolin and was described previously (Urosev 2003). The method uses sequential treatment of plasma with anion exchange media in the presence and then absence of C a 2 + ions, followed by a final affinity chromatography step. Minor changes to this procedure were implemented at the second chromatographic step where, instead of H P L C anion exchange chromatography, conventional chromatography (at 4 °C) with D E A E Sepharose Fast Flow media (GE Healthcare) was used. Following batch treatment, the sample was not concentrated but rather applied directly onto a D E A E Sepharose column (255 mm x 50 mm) that was equilibrated against 25 m M Tr i s -HCl , 1 m M E D T A , 50 m M N a C l , p H 7.8 (equilibration buffer), at a flow rate 2 mL/min. Subsequently, the column was washed with 1.5 L of equilibration buffer and gelsolin was eluted with a linear gradient of N a C l , in the concentration range of 50-350 m M in equilibration buffer (total volume of 1.5 L) . Elution was monitored by U V absorbance at 280 nm. Gelsolin was removed from the column at approximately 2/3 of the way through the gradient. These fractions were concentrated under nitrogen gas pressure using an Amicon ultrafiltration cell (30 k D a cut off Mil l ipore membrane) to a volume of 50-75 mL. Concentrated fractions were dialyzed against 2 L o f 25 m M Tr i s -HCl , 1 m M E D T A , p H 8.0, over two days, at 4 °C, changing daily. Dialyzed gelsolin was subjected to a final step of affinity chromatography (Urosev 2003). The final yield was about 20-25 mg of gelsolin from 500 m L of plasma. 36 2. Preparation of rabbit skeletal muscle powder Rabbit back and leg muscles (1kg) (PelFreez Biologicals) was minced in a blender with 3 L of cold 0.3 M KC1, 0.1 M K H 2 P 0 4 , 0.05 M K 2 H P 0 4 , 0.2 m M A T P , p H 6.5. This mixture was centrifuged at 7000 R P M (the same speed was used throughout the procedure) for 1 hour in a J L A 8.1 rotor (Beckman), in an Avanti J-20 X P centrifuge (Beckman), to remove myosin. The sediment was washed with 3-4 L of cold water, collected by centrifugation and homogenized in 4 L of cold 0.01M N a H C 0 3 , p H 8.0. Hemoglobin and myoglobin were removed upon centrifugation of this homogenate. The sediment was washed with cold water and repelleted. The pellet, rich in muscle thin filaments, was dehydrated by washing twice with 3-4 L of cold 95% ethanol, followed by 2 additional washes with 3-4 L of acetone. The residue was spread out on filter paper and left to dry in a fumehood for 2-3 hours. The resultant stringy white powder was packaged in 10 g portions and stored at -20 °C for up to several months. The yield was in the range of 70 g per kg of muscle. 3. Actin purification Act in was purified from the above acetone-dried powder, according to a procedure based on the method of Spudich and Watt (1971) (Urosev 2003). In this method, the ability of actin to polymerize was utilized to devise a rather quick and simple purification procedure. A s a last step in the procedure, after conversion of F-actin into G -actin and removal of residual F-actin through ultracentrifugation, predominantly monomeric actin was gel filtered through Sephacryl S-300 (97.5 x 2.5 cm) (GE Healthcare) at 2 mL/min in buffer A (2 m M Tr i s -HCl , 0.2 m M C a C l 2 , 0.2 m M A T P and 1 m M D T T , p H 7.6). Act in eluted as a sharp peak (Figure 17) and fractions corresponding to the last two thirds of the peak, representing monomeric protein, were collected and used to form complexes with gelsolin. The approximate yield was 6 mg/g of muscle powder. 37 Elution time of monomeric actm time Figure 17. A c t i n elution profi le from Sephacryl S-300 gel filtration column in 2 mM Tris-HCl, 0.2 mM CaCl 2 , 0.2 mM ATP and 1 mM DTT, pH 7.6, at a rate of 2 mL/min at 4 °C. The absorbance of the eluant was monitored at 280 nm and recorded, simultaneously, at sensitivities of AUFS 0.5 (shorter peak) and 1 (taller peak). The region of the peak corresponding to monomeric actin is indicated. 4. Preparation and purification of gelsolin/actin complexes 4.1 Complexes of one gelsolin bound to two ATP-actin monomers (GA2A T P) or to two ADP-actin monomers ( G A 2 A D P ) Purified actin bound to A T P , as described in the section above, was dialyzed against 1 L of 2 m M Tr i s -HCl , 0.2 m M C a C l 2 , 1 m M D T T and 5 m M A D P , pH 7.6 buffer (buffer A ' ) over three days, at 4 °C with daily buffer changes, to convert it into the A D P -actin form. Gelsolin, in 25 m M Tr i s -HCl , pH 8.0, 1 m M E D T A , was activated in the presence o f 2 m M C a C l 2 at room temperature, for 5 min. Then ADP-act in (or ATP-actin) in buffer A ' (or A ) was added to the activated gelsolin at a molar ratio of 2:1 such that the final solution contained equal volumes of 25 m M Tr i s -HCl , pH 8.0, 1 m M E D T A and 38 buffer A ' (or A ) . The resulting solution was incubated overnight at 4 °C to allow for complex ( G A 2 ) formation. Gel filtration chromatography was used to purify the complex. The G A 2 solution was applied to a Sephacryl S-300 column (97.5 x 2.5 cm) at 4 °C or room temperature in 2 m M Tris- HC1, 0.2 m M CaCb , 1 m M D T T , p H 7.6 buffer at a rate of 2 mL/min. Elution of the complex was followed by U V absorbance at 280 nm. G A 2 A U 1 usually eluted as a much sharper single peak than G A 2 A T P . The fractions corresponding to the rise of the peak, very top, and descent were kept in three separate pools and concentrated using Amicon centrifugal concentrators to -10-20 mg/mL each. The concentration of G A 2 A D P or G A 2 A r p was approximated spectrophotometrically using an absorption coefficient at 280 nm of 1.25 m L trig"1 cm" 1. 4.2 Gelsolin complex with longitudinally crosslinked actin monomers (GAX) The procedure used for actin crosslinking is based on the method of Doi (1992). Act in was treated with a four-fold molar excess of benzophenone-maleimide (BPM) (stock solution dissolved in dimethylformamide (DMF)) for 6 hours in the dark at 4 °C with constant shaking. Covalent modification with B P M proceeds at Cys374, located in subdomain 1 (Figure 2A). The reaction was stopped with 5 m M D T T , and the sample was dialyzed against 1-2 L o f buffer A without D T T overnight at 4 °C. Varying amounts of unlabeled actin, usually 1/3 or 1/4 of that used in the B P M labeling step, was added to the dialyzed BPM-labeled actin sample and polymerized in the presence of 150 m M KC1 and 2 m M M g C l 2 for 3-4 hours at room temperature or overnight at 4 °C. Crosslinking was performed by addition of a four-fold molar excess of a bifunctional reagent, N,N ' -p-phenylenedimaleimide (pPDM) (stock solution dissolved in D M F ) , to the polymerized BPM-actin/actin mixture and incubation on a shaker for 1 hour at room temperature. p P D M forms a crosslink between Cys374 and Lysl91 of subdomain 4 (Figure 2A) (Elzinga and Phelan 1984). The linker length allows crosslinking only between actin monomers belonging to the same strand. The reaction was terminated with 5 m M D T T and F-actin was pelleted at 35000 R P M for 3 hours at 4 °C in an Optima L90 ultracentrifuge (Beckman). The pellet was homogenized in buffer A without D T T and dialyzed against 2 L of the same buffer overnight at 4 °C. Under these conditions the 39 majority of crosslinked product was represented by trimers, as indicated by S D S - P A G E , with a significant amount of actin monomer still present. We decided to proceed, despite this outcome, since interest was present not only in formation of a gelsolin complex with an actin dimer, but with higher order actin forms as well . Modifications to the labeling conditions and the amounts of actin used in the two steps did not improve the yield of trimers. Gel filtration on Sephacryl S-300 was used in an attempt to separate actin trimers from monomers, but, due to the propensity of actin to form higher order structures, this approach was not very efficient, resulting in only a partial reduction of actin monomer/trimer ratio. Gelsolin was activated in the presence of C a 2 + ions in an analogous manner as for formation of GA2. It was then added to a mixture of crosslinked actin trimers and monomers in a two-fold molar excess and incubated overnight at 4 °C. Partial separation of the various complexes formed was achieved by gel filtration through Sephacryl S-300 in'2 m M Tr i s -HCl , 0.2 m M C a C l 2 , 0.2 m M A T P , 1 m M D T T , p H 7.6 buffer. B. X-ray crystallography Biomolecular X-ray crystallography is an analytical technique that utilizes the diffraction pattern generated upon the X-ray bombardment of a single crystal, usually protein or nucleic acid in nature, to solve atomic structures. Single crystal X-ray diffraction dates back to the beginning of the 20 t h century when the first experiments were carried out on salt crystals. Ha l f a century later, Perutz and Kendrew (1958) solved the first protein structure, that of sperm whale myoglobin. Presently, over 35,000 protein structures have been solved by X-ray crystallography and deposited in the Protein Data Bank (PDB) (Berman et al. 2000). 40 1. Protein crystals A s implied from the name of the technique, it is completely dependant on the possession of the crystalline form of the macromolecule of interest. Crystals are solids that are three dimensional repeats of a lattice motif. This motif is composed of asymmetric units that are related by symmetry operations. The asymmetric unit is the minimal unit that does not possess any internal symmetry. Hence, it may be represented by one or more whole protein molecules, or their subunits or domains. The unit cell, the smallest repeatable set of lattice points (represented by lattice motifs) is a parallelepiped in shape, characterized by a set of vectors with lengths a, b and c and associated angles a , (3 and y. A crystal lattice type is characterized by the angles defining the unit cell, its relative vector lengths, and its lattice motif (at locations other than at the cell edges). Lattice type and lattice motif symmetry represent a space group. Overall, the space group and unit cell parameters dictate crystal morphology and diffraction pattern (van Holde et al. 1998). Crystalline, ordered forms of macromolecules are necessary to gain the crucial orientational information that would otherwise be lost i f amorphous samples were to be used. A second dependency on crystals stems from the weak X-ray scattering properties of electrons. Hence, in order to generate distinguishable intensity diffraction data, repeating units of scatterers are necessary to amplify the signal. The size and conformational dynamics of protein molecules limit the number of interactions between them in a crystal. These factors are often responsible for crystallization being a bottleneck of the crystallographic technique. Even when success is achieved in obtaining a crystal, these protein-protein interactions central to a crystal's quality may be insufficient (due to the small relative ratio of interactions to molecular mass) to attain the same degree of 3D order as in small molecule crystals. This results in rather modest diffraction by protein crystals (the majority diffract up to 1.5-3 A resolution (PDB)). The presence of large solvent channels in protein crystals and their high solvent content accounts for observations that the protein structure found in crystals is virtually unchanged from that in solution. In addition, a number of biochemical features are preserved, such as ligand binding Kd's, enzymatic characteristics, etc. 41 1.1 The crystallization process Crystallization experiments used to be based mainly on trial and error, but much advancement has been made in recent years towards understanding the fundamental principles of crystallization. Hence, rational approaches to crystallization are developing (reviewed by Chayen 2004; McPherson 2004). These are based on knowledge of the protein phase diagram, adequate/informed manipulation for nucleation and growth, utilization of different screening strategies, etc. Modern crystallization is going to such lengths as performing microgravity experiments in search for mechanistic answers (McPherson 1993), researching the development of a universal nucleant (Chayen et al. 2006) , developing techniques for detection of nucleation events, and inventing instrumentation for controlling/separating crystal nucleation and growth (Shim et al. 2007) . Crystallization proceeds by two inseparable steps: nucleation and growth. Both depend on reaching a supersaturated state of the protein that, ideally, leads to formation of partially ordered intermediates. These transition into minimal stable ordered arrays, nuclei, and then growth continues. Protein supersaturation is a non-equilibrium state that, through formation of solid phase (desirably crystalline), moves to a saturated equilibrium state (reviewed by McPherson 2004). The kinetics involved in achieving supersaturation and the subsequent approach to equilibrium influences whether a crystalline or an amorphous solid forms. In order to achieve supersaturation, a successful method is to reduce the chemical potential of water, e.g. through addition of salts or polymers, or by controlled evaporation. Reduction in protein solubility can also be achieved through alteration of protein properties, e.g. through change in p H in such a way as to promote interactions between individual protein molecules. Chemical and conformational purity of the protein sample are also important for successful crystallization. Different precipitating agents include salts, organic solvents, and long and short chain polymers. Salts dehydrate protein molecules through competition for water. Multivalent anions such as phosphates and sulphates are most efficient. Organic solvents reduce the dielectric constant of the medium, which promotes interactions between protein molecules. Polymers, such as polyethyleneglycols (PEGs), produce exclusion 42 effects, since they acquire random conformations and leave limited space for macromolecules, hence forcing them into a solid state. Therefore, longer polymers do have a larger propensity to force proteins out of solution. Auxil iary substances, such as metal ions, antioxidants, and protein ligands/substrates, are often beneficial in favoring crystallization over precipitate formation and for improvement of crystal quality. There are a number of methods that are utilized for protein crystallization, which explore the crystallization phase diagram, shown in Figure 18, in different ways. The least exploratory is microbatch under oil (Figure 18, path I), where the starting concentrations of the precipitant and protein in the set-up almost do not change over time. The most common technique employed is known as the hanging (or sitting) drop vapor diffusion method (Figure 18, path II), which is based on evaporation of the solvent, allowing for a moderate increase in protein and precipitant concentrations with time. Capillary counterdiffusion, an emerging method, has possibly the biggest potential for phase diagram exploration. In such experiments the concentration of protein does not change, but precipitant concentration in the protein "compartment" rises from none through a gradient to supersaturation with time. This allows for partial separation of nucleation and growth stages, therefore increasing the likelihood of obtaining large crystals (Figure 18, path III) (Garcia-Ruiz 2003). 43 '1'... Supers aturation .3 o I Precipitation zone Undersaturation Supersolubility § Metasiable c u r v e Solubility curve > [Precipitant] Figure 18. Crys ta l l i za t ion phase d iagram. Major supersaturation zones are shown, with emphasis on the metastable zone that promotes the formation of large diffracting crystals, once nuclei have formed. Different exploration of the phase diagram space is achieved in the following crystallization methods: I: microbatch under oil, II: hanging/sitting drop vapor diffusion, III: capillary counter diffusion [adapted from Chayen 2004] 1.2 Cryoprotection of crystals X-ray data collection for protein crystals is typically performed at cryogenic temperatures in order to avoid/reduce radiation damage by ionizing X-rays. This also reduces the noise in reflections caused by thermal motions of atoms. Such a procedure increases crystal lifetimes 100-1000-fold, which enables the collection of a full diffraction data set from a single crystal. Typically during data collection, the crystal is exposed to a constant cold nitrogen stream at temperatures around 100 K . At such low temperatures, in order to avoid ice formation that would overwhelm the protein diffraction pattern and possibly destroy the crystal due to volume expansion, conditions are required such that the solution present in channels and surrounding the crystal is vitrified. Therefore, crystals are typically soaked briefly (from a few seconds to a few minutes) in the mother liquor with cryoprotectant present and then plunged into liquid 44 nitrogen or flash-cooled directly in the cold nitrogen stream. Cfyoprotectants are used to depress the freezing point of water and low temperatures (77 K or 100 K ) further slow down the process of ice formation. Successful soaking times and cryoprotectant concentrations depend on the size of the crystal and the nature of the mother liquor. Penetrative cryoprotectants are relatively small molecules, such as glycerol, 2-methyl-2,4-pentanediol (MPD) and P E G 400, that easily diffuse into solvent channels. Non-penetrative cryoprotectants, such as primarily higher molecular weight PEGs (PEG 6000/8000/10000) and oils, are used to mainly exchange water at the crystal surface. A negative aspect of cryoprotectant use is that they often increase mosaicity within the crystal, which results in a deterioration of diffraction quality. High pressure crystal cooling with helium results in high density amorphous ice formation and is emerging as a promising alternative method of crystal cryopreservation (K im et al. 2005). It does not require the use of cryoprotectants and does not affect crystal resolution as much. 1.3 Crystallization of gelsolin/actin complexes The main technique employed for the crystallization of gelsolin/actin complexes was the hanging drop vapor diffusion method (Figure 19). To achieve a gradual decrease in water content of the protein solution and approach to supersaturation, direct addition of a precipitating agent to the protein solution has to be avoided since it normally leads to the formation of a precipitate. The hanging drop vapor diffusion method relies on the use of a closed container, in which a concentrated solution of precipitant is separated from a more dilute one, in which the protein remains soluble. Namely, a small drop of protein sample (usually 2 or 3 pi) is mixed with a precipitant solution (usually of equal volume) on a plastic cap, which subsequently is inverted and screwed into place over the reservoir containing the precipitant solution. The initial precipitant concentration in the droplet is less than that in the reservoir. A s a result, water, as the only volatile component, wi l l slowly travel through the vapor phase to the reservoir solution. This leads to equilibration between the protein solution and the solution in the reservoir, during which the protein becomes concentrated. On reaching protein supersaturation, this may lead to protein 45 crystallization. Twenty-four-reservoir plates (Qiagen) were used for this experimental set-up. Figure 19. Hang ing drop vapor diffusion setup Initial attempts to crystallize the G A 2 complex were based on the known conditions for crystallization of Gl-G3/act in (Urosev 2003). The experiments resulted in the formation of crystals, the contents of which were analyzed by SDS gel electrophoresis. The presence of intact gelsolin and actin in the expected stoichiometric ratios was confirmed. However X-ray diffraction analysis of these crystals revealed only the Gl-G3/act in part of the complex. Such crystals grew at 4 °C from a protein sample at 10 mg/mL mixed with a reservoir solution of 100 m M sodium acetate, p H ranging from 4.5 to 4.8, 2% P E G 8000 (w/v), 2 m M C a C l 2 , at a 1:1 ratio (v/v) (total drop volume of 2 or 4 pi). The following cryoprotectant solutions were found suitable for these crystals: (1) 25% glycerol, 10% P E G 8000, 5 m M C a C l 2 , 400 m M N a C l , 100 m M sodium acetate, p H 4.6, and (2) 15 % P E G 8000, 10 % P E G 400, 5 m M C a C l 2 , 100 m M sodium acetate, p H 4.6, as well as (3) 8% trehalose, 20% glycerol, 2% P E G 8000, 5 m M C a C l 2 , 100 m M sodium acetate, pH 4.6. , G A X crystals grew at 4 °C from two different well solutions: (1) 2-4% P E G 6000, 2 m M C a C l 2 , 100 m M H E P E S , pH 6.0, and (2) 4-6% P E G 3350, 2 m M C a C l 2 , 100 m M Protein + 'Precipitant H 2 O 46 H E P E S , pH 6.0, each mixed with 10-20 mg/mL protein sample. Control protein drops set with G A 2 A T P did not yield any crystals. A l l protein complex crystals grew to less than 50 pm in maximal dimension and became visible under the microscope after anywhere from two or three weeks to two months. 2. Crystal soaking Common approaches employed to obtain the structure of a biomolecule bound to a small ligand are cocrystallization and soaking (reviewed by Danley 2006). Cocrystallization entails formation of a complex between the ligand and protein in solution and subsequent crystallization. In the soaking method, the ligand of interest is incubated in a bathing solution with the preformed apocrystal. Protein crystals are loosely packed, typically containing 30-80% solvent/water, which may be classified into bulk solvent and bound water interacting with protein molecules. Inner crystal packing/lattice interactions determine the size and shape of solvent channels (Figure 20). These are usually big enough (10-100 A) to provide good access for small ligands to the protein molecules. If the conformational changes induced upon ligand-binding are not compatible with the crystal packing interactions, or i f the binding simply can not occur due to the binding site being obscured because the relevant surface participates in the packing interactions, one would have to resort to cocrystallization as an alternative. Along with the mentioned lattice-related limitations, other factors that influence the success of soaking are protein-ligand binding affinity, ligand solubility in the mother liquor, and factors in the chemical environment, such as p H and ionic strength and composition. 47 Figure 20. Solvent channels in crystals of inactive gelsolin. The asymmetric unit, composed of two gelsolin molecules, is shown in yellow. The extended structure of the crystal, limited to a sphere of 100 A in radius, was generated using the symmetry operators for space group (P42|2). Symmetry molecules are shown in violet. The hollow regions represent space filled with solvent (generated with COOT). 2.1 Gelsolin crystal soaks in A T P solutions Crystals o f inactive gelsolin were grown as previously (Burtnick et al. 1997) in 100 m M Tr i s -HCl , p H 8.0 or 7.5, 0 .5- ImM E D T A , and a range of 30-34% saturated ammonium sulphate (sAms) at 18 °C or room temperature (Figure 21). 48 Figure 21. Crys ta l s o f inactive gelsolin. Grown from a 10 mg/mL protein sample with a reservoir solution of 34% sAms, 100 mM Tris-HCl, 0.5 mM EDTA, pH 8.0, at 18 °C, using hanging drop vapor diffusion. The space group for the gelsolin crystals is P42[2 with cell dimensions of a=b=169.4 A and c=154.2 A (Burtnick et al. 1997). Ammonium sulphate screens protein charges and therefore can interfere with the binding of specific charged ligands. P E G 400 was identified as a suitable substitute medium for gelsolin crystals. Hence, it was used to remove the majority of ammonium sulphate from the crystal lattice. This allowed for successful A T P binding. Crystals were washed and cryoprotected in a medium containing 30% (v/v) P E G 400, 100 m M Tris-H C l , I m M E D T A , pH 8.0, then soaked in that medium containing 5 m M A T P for 20 minutes. Subsequently, they were flash cooled by immersion in liquid nitrogen. 2.2 T b 3 + soaks of G A 2 (Gl-G3/actin) crystals Lanthanides such as T b 3 + have been widely used to probe Ca 2 + -binding sites in proteins because of their similar radii and coordination properties to C a 2 + . Unlike C a 2 + ions, T b 3 + ions are spectrally active (Horrocks 1993). Due to their "heavy" status, they perturb the diffraction pattern of the protein crystal they are bound to, allowing their 49 positions to be easily identified. Tb -bound protein crystals may also be used to solve the phase problem in crystallography. Two types of crystals grew at 4 °C from a 10 mg/mL G A 2 complex solution, with a reservoir solution composed of 100 m M sodium acetate, pH 4.6, 2% P E G 8000 and 2 m M C a C l 2 (hanging drop vapor diffusion). The smaller volume unit cell crystals, space group P3121 contained only a Gl-G3/act in complex, as discovered previously (Urosev, 2003). The larger volume unit cell crystals, with space group P6s22, were large enough to accommodate G A 2 , but X-ray analysis revealed only Gl-G3/act in . Both types of crystal were cryoprotected in a solution of mother liquor with 20 % glycerol and subsequently incubated for 24 hours in the same solution in the presence of 0.2 m M Tb(NC>3)3 (medium 1) (P6 522 crystals) or 1 m M T b ( N 0 3 ) 3 (P3i21 crystals), at 4 °C, to allow for C a 2 + / T b 3 + exchange. Crystals with the space group P6s22 were sequentially backsoaked in medium 1 with increasing C a 2 + concentrations of 0.1 m M , 1 m M and 10 m M , over a duration of 12 hours. A l l crystals were flash frozen in liquid nitrogen prior to data collection. 2.3 Gelsolin crystal soaks in solutions of LPS L P S contains phosphate moieties. Since sulphate ions are specifically prone to occupying phosphate binding sites within proteins, it was necessary to exchange the ammonium sulphate in gelsolin crystals with a medium that would allow ligand binding. Use of P E G 400, though successful with the batch of gelsolin crystals used for A T P -soaking experiments, often resulted in crystal deterioration. Further investigation identified salts of small organic acids as alternate medium-exhange candidates. These salts, in particular, sodium malonate, are very appealing due to their low degree of protein charge screening, to their cryoprotective nature, and to their positive effects on the physical stability of crystals (McPherson 2001). Incubation of gelsolin crystals in sodium malonate, citrate or tartrate solutions of the same ionic strength as that of the ammonium sulphate solution that gave rise to crystals did not lead to any visual deterioration of the crystals. Namely, 1.35 M sodium malonate, 0.6 M sodium citrate and 1.35 M sodium tartrate, each in 100 m M Tr i s -HCl , 1 m M E D T A , p H 7.5 or 8.0, were 50 used. Following three rounds of crystal bathing in one of the salt solutions, each for a duration close to 20 minutes at room temperature, crystals were soaked for additional periods of time (30 minutes, 2-3 hours, or overnight) in solutions additionally containing 1.5 mg/mL L P S . Crystals were subsequently cryoprotected in the presence of higher concentrations of sodium malonate (2.7 M ) , or, in the cases of the other two salts, with additional glycerol at 25 % (v/v). 3. Single-crystal X-ray diffraction X-rays are electromagnetic radiation with wavelengths between about 0.01 A and 100 A. They can be generated by bombardment of a metal target with high-energy (10 -100 keV) electrons that knock out inner shell electrons from the metal. Emission of X-ray radiation occurs when electrons transition from outer to inner shells to replace the dislodged electrons. A common metal target is Cu , which has strong K a X-ray emission at 1.5418 A. Another source of X-rays is synchrotron radiation. High-energy electrons steered in a circular storage ring by a series of electromagnets give rise to bursts of X-ray radiation each time there is a change in the direction of their linear path. X-ray beams produced are typically more than 1000-fold more intense than those generated by normal laboratory X-ray sources. This allows for more rapid data collection and obtaining improved diffraction data from small (20 pm) (Cusack et al. 1998) or weakly diffracting crystals. Otherwise, a useful crystal size for data collection is 0.1-0.3 mm in each dimension. X-rays are of the same order of magnitude in wavelength as the distances between atoms in a molecule (~1 A) or crystal and therefore are appropriate to determine the overall molecular shape and solve the three-dimensional arrangement of atoms in a protein. X-rays are primarily scattered as they pass through a crystal. These scattered waves interfere constructively only in some specific directions to produce a characteristic 2D X-ray diffraction pattern. Diffracted X-rays act as if they were reflected from a family of planes that pass through groups of atoms within the crystal. Constructive interference occurs when the distance traveled by X-rays reflected off these planes differs by an integral number (n) of half-wavelengths (Bragg's law). In order to collect a full data set at 51 fixed X, the crystal needs to be rotated to observe diffraction from all such planes in the crystal. Diffracted X-rays emerge from the crystal at different angles and have different intensities. Each diffracted X-ray generates a spot where it intersects the X-ray detector. These 2D images are merged computationally according to the crystal rotation into a 3D diffraction image. Diffraction spots represent lattice points of the so-called reciprocal lattice, parameters of which are used to obtain space group and unit cell characteristics of the crystal. Intensities of diffraction spots depend upon the structural organization of the macromolecule (van Holde et al. 1998), but cannot alone be used to determine that organization. A diffracted wave is described by a complex number, called a structure factor. Each structure factor contains information relating to both the amplitude and the phase of the wave. Scattering from every atom within the protein contributes to each structure factor, but to varying extents. Each point in the electron density map contains contributions from all of the reflections. Therefore, information from all structure factors, which requires both wave parameters, is necessary to best determine the electron density at each location within the unit cell, i.e. to determine all types of atoms and their positions in space. While diffraction wave amplitudes can be derived from diffraction intensities, the phase information cannot be experimentally obtained simply from those same intensities. Two common phasing techniques are multiple isomorphous replacement (MIR) and multiple anomalous dispersion ( M A D ) . Changes in the scattering amplitudes influenced by the presence of heavy atoms in M I R and by use of different X-ray wavelengths in M A D are used to obtain the phases (Taylor 2003). However, i f the structure of a chemically and structurally homologous molecule is known, the method of molecular replacement (MR) (used in the present work) can be implemented to solve the phase problem (Taylor 2003). A model built with M R is further refined so as to match as closely as possible the values of structure factors calculated from the model to those of experimental structure factors. Convergence of these structure factor values within a satisfactory limit allows a refined model to be presented as the solution to the structure (van Holde et al. 1998). 52 3.1 Collection and processing of X-ray diffraction data Diffraction data for ATP-bound gelsolin crystals and Tb 3 + /Ca 2 + -soaked G A 2 crystals, were collected at 100 K at the M A X - L a b Synchrotron Radiation Facility, Lund University, by Drs. Les Burtnick and Robert Robinson. Once access to the in-house X -ray (Cu-Ka) diffractometer in the laboratory of Dr. Natalie Strynadka was gained, testing of various crystals and full data set collection, such as for G A 2 crystals, were performed there. Once soaked in cryoprotectant, a crystal is harvested with a cryo-loop, flash cooled in liquid nitrogen, and mounted on a goniometer head (Figure 22). The goniometer allows for accurate positioning and rotation of the crystal within the X-ray beam. A n Oxford Cryostream 700-series model was utilized to provide constant cooling under a nitrogen stream (Figure 22). The X-ray beam (1.5418 A wavelength) was generated by a Rigaku M M H 0 0 7 H F generator and focused through 0.6/0.5 mm slits. Once a crystal was centered in the beam, with the help of a camera focused on the crystal, data collection of the first image was performed with X-ray detector control software M A R 3 4 5 D T B . The intensities of the reflections were recorded with an X-ray image plate detector (MAR345) (Figure 22). 53 X-ray image plate detector MAR345 Figure 22. X-ray diffractometer (in the lab of Dr. Natalie Strynadka). The X-ray generator is not shown, but all other relevant components are indicated in the figure. A single diffraction image accounts for only a fraction of the lattice points in the reciprocal lattice, but provides enough information to decide whether the particular crystal is suitable for collection of a full data set. A n y decision is based on the resolution limits of the crystal, its mosaicity, the shape and resolution o f diffraction spots, and their average intensity to noise ratio. Mosaicity is a measure of the smallest angle through which the crystal can rotate while a reflection is still observed. In order to process diffraction data properly, this parameter should ideally be below 1°. A complete data set consists of reflections that arise, ideally, from all lattice points in the reciprocal lattice, with typically >90% completion necessary to solve a crystal structure. The STRA TEGY option in the program M O S F L M (Leslie 1992) utilizes such parameters as estimated internal crystal symmetry (based on a single diffraction image), crystal orientation, X-ray wavelength and crystal-to-detector distance, to determine which range of crystal rotations is optimal for collecting a full data set. Throughout this range, diffraction images are collected in the oscillation mode. The 54 oscillation range is usually selected as a fraction of crystal mosaicity, chosen to avoid overlap of reflections, and was around 0.5 0 for most of the crystals analyzed in this work. The H K L 2 0 0 0 software package (Otwinowski and Minor 1997) was used to index the reflections in diffraction images, integrate their diffraction maxima, scale the intensities of partially recorded reflections appearing in separate images, and then merge the data from all diffraction frames. D E N Z O initially was used to find which reflection corresponds to which position in reciprocal space, i . e. to index the reflections, in the first diffraction image. This was performed with the same parameters utilized in STRATEGY after being subjected to refinement cycles, until the predictions of reflection locations (based on the current indexing) were matched with the observed diffraction pattern. Visualization of the diffraction space and refinement process was enabled through X D I S P L A Y F . A byproduct of indexing is that the unit cell parameters, along with the space group, are determined. The refined parameters determined from indexing the first frame were used to process all the images. This was followed by integration of diffraction maxima for all reflections. Wi th S C A L E P A C K , partially recorded reflections over all frames in the dataset were tracked and their intensities were corrected so as to have consistent intensity. This is important, since the relative intensities of the peaks are part of the information from which the macromolecular structure is determined. Scaling was followed by merging of all the images containing numerous reflections into one file with the information on the intensities of each spot. Finally, global refinement of crystal parameters, using the entire data set, was preformed to obtain their precise values. 3.2 Molecular replacement and structure refinement A s mentioned, phases of the diffracted waves are as important as their amplitudes in solving a protein structure. Amplitudes are proportional to the square root of diffraction intensities, obtained during data processing. In all cases in this thesis, structural information on the components of the investigated structure was available and, hence, molecular replacement was the method of choice to obtain the phase information. In the case of crystals of G A 2 complexes and ligand-soaked crystals of inactive gelsolin, the Gl-G3/act in and Ca 2 +-free gelsolin structures, respectively, were used as search 55 models. M O L R E P from the C C P 4 software package (CCP4 1994) searched for the optimal orientation and position of the molecules within the examined crystal lattice. First, the number of protein molecules in the asymmetric unit was assessed, based on the fact that solvent typically occupies 50% of the crystal volume. Then, rotation and translation functions were employed to yield an initial model, with estimated phases of diffracted waves that do not yet account for the observed (experimental) structure factors. Therefore, this step was followed by structure refinement cycles with R E F M A C 5 . The goal of this final stage was to find the atomic coordinates that give the best fit o f the amplitudes calculated from the current model |Fc | to the observed structure factor amplitudes |Fo| . During the automated refinement process, additional manual aid to the refinement through correction of the fitted polypeptide chain was enabled by examining the weighted difference electron density maps to spot the regions of positive (unaccounted for) and negative (falsely accounted for) electron densities. Inspection of electron density maps was performed using visualization software C O O T (Emsley and Cowtan 2004). The crystallographic R factor, R c r y s t , was one measure of the deviation between the measured and calculated structure factor amplitudes: ^ ^ a l l reflections I FQ F C | y a l l r e f l e c t i o n s I FQ | A subset of reflections (5-10%) was excluded from refinement, which allowed the progress of refinement to be additionally assessed through Rfree. It is calculated in an analogous manner to R c r yst, but only for this subset of reflections that was not used in the structure refinement. Rfree provides a less biased measure of the refinement process. Successful refinement results in the decrease and close convergence of these two R factors to values of less than 0.2, and allows the refined model to be presented as a solution to the structure. P R O C H E C K was used to assess the correctness of all the chemical parameters, such as bond lengths and angles, within the polypeptide chain of the final model. R c r y s t — 56 C. Computational molecular docking Computational molecular docking is a technique for predicting how two molecules interact with each other. Interactions which are usually studied using this approach are: protein-protein, protein-ligand or protein-DNA. In the protein-ligand case, docking entails identification of a low-energy binding mode of a ligand within the binding pocket of a protein of known structure. First, a search algorithm is used to find potential/acceptable ligand conformations and orientations in the binding pocket. Then a "scoring function" is utilized to evaluate the binding energies that indicate the likelihood that the given ligand orientation and conformation represent a favorable binding interaction. Scoring functions are representatives of potential energy of the given system that include covalent and non-covalent energy terms. Docking competence is useful in the areas of structure-based drug design and drug screening in pharmaceutical research. MolDock (Thomsen and Christensen 2006) was used to search for the potential PIP2 and A T P binding sites in gelsolin and Gl-G3/act in . The docking accuracy of MolDock is comparable to that of other widely used docking programs (Thomsen and Christensen 2006). 1. Docking of gelsolin ligands to inactive gelsolin and Gl-G3/actin with MolDock A complete docking procedure consisted of three steps: in silico preparation of proteins and ligands, cavity search/detection, and the actual docking run. Proteins and ligands subjected to docking were prepared in the following way: coordinates were extracted from P D B files, appropriate bond orders and atom hybridizations were assigned, hydrogens that were missing in P D B files were attached according to the geometry governed by the hybridization of their bonding atoms (energy minimization was not performed), and finally, partial charges on atoms at p H 7.0 were allotted. Protein molecules were kept completely rigid or allowed limited localized side chain flexibility. Usually, all the ligand bonds were kept rigid. A second step involved a grid-based search for identifying cavities (usually up to five) in the protein molecule as potential ligand binding sites. Finally, the potential cavities were assessed and ranked by performing 57 d o c k i n g s imulat ions. Because o f the stochastic nature o f the d o c k i n g search a lgor i thm, m u l t i p l e runs ( i f the l igand bonds were kept r i g i d , ten runs were sufficient) for each l igand-protein setup were performed to ensure convergence to the lowest-energy solut ion. T h e general hypothesis is that l o w e r energy scores, a l l reported i n arbitrary units , represent better protein-l igand b i n d i n g . T h e d o c k i n g scor ing funct ion, Escore, is defined as: Eq. #1 zr — p 4- F score inter inira Eq.#2 !€ iigatidfe l igand X A[l - cos(m-6 - 60)] + Echsh flexible bonds Eq.#3 hgand/e proteinl E?L?(rtJ) + 332.0—; 4 / V E i n t r a is the internal energy o f the l igand. T h e first term, EPLP , is a piecewise l inear potential that accounts for steric (van der W a a l s ) and hydrogen b o n d i n g contributions, the second is the torsional energy term (where 9 is the torsional angle o f the b o n d and 9o is 0 2 3 3 3 for sp -sp and n for sp -sp bonds), and E c i a s h penalizes the score i f the distance between two atoms is less than 2 A. E i n t e r is the l igand-protein interaction energy that is s u m m e d over a l l non-hydrogen atoms i n the l igand and the protein and takes hydrogen b o n d direct ional i ty into account. T h e second term i n E i n t e r describes electrostatic interactions between charged atoms and this contr ibut ion has a theoretical ly predictable magnitude. l t is a C o u l o m b potential w i t h a distance-dependent dielectr ic constant g i v e n b y D(r) = 4r. A l l other terms w i t h i n E i n t r a and E i n t e r are parametrized f r o m experimental values, so the. total energy does not necessari ly correlate w i t h the true l igand b i n d i n g aff inity values. 58 Crystal structures of ATP-actin from Gl-G3/ac t in (PDB ID: 1RGI) and of the here-reported ATP-gelsolin (PDB ID: 2FGH) were used as controls for the docking program search algorithm and scoring function. 59 CHAPTER III Results A. Interactions of gelsolin with A T P 1. The structure of inactive gelsolin bound to A T P Crystals of inactive gelsolin (Figure 21) were washed successfully with 30% (v/v) P E G 400, 100 m M Tr i s -HCl , 1 m M E D T A , pH 8.0 and then soaked in the same medium with 5 m M A T P . The crystals contain two gelsolin molecules per asymmetric unit and a difference electron density map allowed localization of two A T P molecules in that unit, one per gelsolin, in identical sites (Figure 23 B) . Unit cell parameters of the soaked crystals are, as expected, characteristic of those for inactive gelsolin crystals (Burtnick et al. 1997): P42,2, a=b=167.3 A, c=149.9 A and a=(3=y=90°. The structure of A T P -gelsolin was solved to 2.8 A resolution (Table 1). Other data collection and refinement statistics are provided in Table 1. Coordinates for the refined model of gelsolin with A T P bound have been deposited in the Protein Data Bank with accession code 2 F G H (Urosev et al. 2006). A T P binds across the two pseudosymmetrical halves of gelsolin in a positively charged pocket on its surface (Figure 23 A and C). The amino acid residues involved in contacts with A T P belong to domains G3 through G6, inclusive. The phosphate moieties of A T P interact with basic residues belonging mainly to the C-terminal domains (Arg629, Arg542, Lys503) (Figure 24 A and C) , while both charged and hydrophobic interactions are involved in stabilization of the adenine portion of the molecule (Figure 24 A and B) . This stabilization is provided mostly by domain G3 through residues Asp371, Arg370, Trp369, Arg363, and Leu380. These amino acids also comprise part of a POVbind ing sequence in gelsolin (Feng et al. 2001), the implications of which are discussed in the next section of this chapter. The structure of ATP-gelsolin reveals the basis for Ca 2 +-sensitivity of the interaction of gelsolin with A T P . Activation of gelsolin by C a is accompanied by conformational rearrangements of the domains with respect to one another. In particular, the two halves of activated gelsolin are thought to be able to act independently of each other, subject only to being tethered together through the G3-G4 linker peptide (Choe et al. 2002; Burtnick et al. 2004). Separation of the two halves of gelsolin would disrupt the 60 Data collection statistics Wavelength (A) 0.957 Resolution range (A) 20.0-2.8 (2.9-2.8)* Space group P42,2 Unit cell dimensions (a,b,c: A)/ (a,P,y: deg.) a=b=167.3, c=149.9/ a=P=y=90° Total reflections 166,794 (17,165) Unique reflections 52,123 (4768) Redundancy 3.2 (3.6) Completeness (%) 98.6 (97.0) Average I/S 26.7 (6.2) R m e r g e : S*[Z , | /* r - ( /*) | / ( /* ) ] (%) 7.4(19.3) Refinement statistics R f a c t o r ( % ) 24.2 (37.2) R f r e e ( % ) 26.5 (38.9) Total number of non-H atoms ( A T P and water) 11,378 (62,50) R M S deviation from ideal bond length (A) 0.011 R M S deviation from ideal bond angles (deg.) 1.36 Average B-factor (including A T P ) (A2) 59.6(27.8) Molecule consists o f residues 27-258, 267-755 * Subsequent numbers in parentheses correspond to data collected in the resolution range between 2.8 A and 2.9 A. Table 1. Crys ta l lograph ic data and refinement statistics for A T P - g e l s o l i n structure. 61 c I Figure 23. The structure of A T P - g e l s o l i n . A ) Gelsolin is shown in a ribbon representation and A T P is represented as a ball-and-stick model. B) A T P fit in to the difference electron density map calculated using data for inactive gelsolin ( P D B ID 1 DON) and ATP-soaked gelsolin crystals. C ) Electrostatic surface map o f gelsolin, with bound A T P shown in ball-and-stick representation (molecular orientation is as in A ) . Surface colored blue is positively charged, while surface shown in red is negatively charged. The adenine ring o f A T P is buried in this representation. 62 Figure 24. Gelsolin residues involved in binding ATP. A) The ATP-binding site in ATP-gelsolin. The molecular orientation is similar to that in Figure 23 A. B) An enhanced view of the adenine region of A T P as it is bound to gelsolin. The molecular orientation is rotated 90° around a vertical axis relative to that in A . C) A n enhanced view of the ribose and phosphate regions of ATP as they are bound to gelsolin. The molecular orientation is rotated around the horizontal-axis relative to that in A. [taken from Figure 2 of Urosev et al. 2006] 63 ATP-binding site found in inactive gelsolin. Likewise, the stabilization of the compact structure of inactive gelsolin by having A T P bound helps to explain the observed decrease in affinity of nucleotide-bound gelsolin for Ca ions (Kambe et al. 1992). 2. In silico investigation of A T P binding to inactive gelsolin Extraction of the coordinates of A T P from two known structures, that of actin in a complex with the N-terminal half of gelsolin (PDB ID 1RGI) and that of gelsolin bound to A T P (PDB ID 2FGH) , provided two similar rigid structural skeletons (Figure 25 A ) for use in computational docking experiments. When each rigid skeleton was used to search for a binding site on the protein coordinates associated with its parent P D B file, complete success in regenerating the entire P D B files was achieved (Figure 25 B and C). This provided a baseline test for performance of the docking program. The E S C O re values for the two cases, -226 for ATP-actin, and -175 for ATP-gelsolin (Table 2), correlate well with the relative affinities of these proteins for A T P , nanomolar in case of actin (Neidl and Engel 1979) and micromolar in case of gelsolin (Kambe et al. 1992; Gremm and Wegner 1999). The contribution of the presence of a C a 2 + ion to the affinity of ATP-binding in actin has been demonstrated experimentally (Kinosian et al. 1993). In agreement, in computational docking experiments, we observed the E S C O re o f - 2 2 6 to rise to -175 upon removal of Ca from the ATP-binding site of actin (Table 2). We next used the rigid A T P skeletons to search for binding sites in each other's apoprotein coordinates. Docking of A T P , using the A T P coordinates from P D B file 1RGI, in the ATP-binding site of gelsolin yielded E s c o r e = -171. This is comparable to the score when A T P coordinates were taken from P D B file 2 F G H . In contrast, when the latter version of the A T P coordinates was docked to the ATP-binding site of actin, the Escore rose significantly to -137 (Table 2). This may be caused by inefficient C a 2 + -coordination, since superimposition of A T P molecules from the two different coordinate sets, from the ATP-gelsolin structure and from that of Gl-G3/act in (Figure 25 A ) , reveals an incomplete overlap. Removal of this A T P ' s phosphate group to generate A D P , which then was docked into actin, gave the same E s c o r e = -137. 64 Figure 25. C o m p a r i s o n of crys ta l lographical ly observed and docked A T P orientations i n the A T P -b ind ing site of actin and gelsolin. A ) ATP molecules with coordinates from PDB files 2 F G H and 1 R G I superimposed. B) overlap of crystallographically observed ( 1 R G I ) and docked A T P orientations in the ATP-binding site of actin. C ) overlap of ATP orientations docked and observed in the ATP-binding site of gelsolin ( 2 F G H ) . 65 \ . B i n d i n g ^ \ site A T P / A D P ATP-b ind ing cavity in actin o f 1RGI Other identified cavities in actin o f 1RGI ATP-b ind ing cavity in gelsolin o f 2 F G H Other identified cavities in gelsolin o f 2 F G H A T P from 2 F G H -137 -98 -175 -107 A T P from 1RGI + C a 2 + -226 -110 -171 -116 A T P from 1RGI - C a 2 + -175 -98 A D P created from 2 F G H A T P -137 -105 -145 -95 A D P created from 1RGI A T P / A D P from 1J6Z -195/-194 -115 -158/-160 -121 Table 2. E s c o r e values for docking A T P and A D P onto actin and gelsolin. E S C O R E values correspond to the best scored ligand orientation in the ATP-b ind ing sites o f actin (protein coordinates from P D B ID 1RGI) and gelsolin (protein coordinates from P D B ID 2 F G H ) . In the case o f other potential pockets identified by the docking program, these values represent the lowest energies obtained for a set o f cavities. E S C O R E values are in arbitrary units. Docking of A D P , with coordinates taken from the ADP-ac t in -TMR structure (PDB ID 1J6Z) (Otterbein et al. 2001) or A D P created by deleting the last phosphate group from the 1RGI version of A T P , into the ATP-binding sites of both actin and gelsolin resulted in higher E s c o r e values compared to those for A T P docking. In the case of docking to actin, E S C Ore = -195, and to gelsolin, E S C Ore = -160 (Table 2). The observed rise in ESCore correlates with A D P binding to actin being approximately 200-fold weaker than for A T P (Neidl and Engel 1979), and with A D P exhibiting far less affinity for gelsolin than does A T P (Kambe et al. 1992). 66 For docking A T P or A D P to other sites selected by the software, E s c o r e values were calculated in the range of -100 arbitrary units, considerably poorer than for the bona fide sites discussed above. This supports the existence of a single strong A T P -binding site in each of actin and inactive gelsolin. A set of in silico calculations of the enthalpic interaction energy for A T P binding to actin and to gelsolin was performed by our collaborator, A . Tan, using Sybyl 7.0. This software utilizes a somewhat different force field-based scoring function than MolDock . The enthalpic energy for A T P binding to gelsolin was determined to be -176 kJ/mol, with approximately 2/3 of the contribution to this value provided by the interactions of the nucleotide with the C-terminal half of gelsolin. The estimated enthalpic energy for A D P interaction with gelsolin, performed through subtraction of the contribution of the A T P y-phosphate to the binding, was -109 kJ/mol. These interaction energies are comparable to the ones estimated by the same software for the adenine nucleotides binding to actin, in the range from -112 to -175 kJ/mol for the binding of A T P to actin, and -115 kJ/mol for the interaction between A D P and actin. The trend of A T P and A D P binding energies in actin and gelsolin estimated with Sybyl 7.0 is in agreement with that obtained with MolDock, which further supports the hypothesis that A T P is a strong and specific ligand for gelsolin. 3. Consequences of Ca 2+-activation of gelsolin for the ATP-binding site Inspection of the Ca -activated structures of G1-G3 (in the presence of actin) (Urosev 2003; Burtnick et al. 2004) and G4-G6 (Narayan et al. 2003; Choe et al. 2002) reveals the cluster of amino acids that forms the ATP-binding site in inactive gelsolin (Figures 26 and 27) to be dispersed. The binding site per se no longer exists in the activated form of the protein. Comparison of the structure of inactive G4-G6 with A T P bound, excised from the ATP-gelsolin data, with that of activated G4-G6 (PDB ID 1P8X) (Figure 26 A and C) shows that the amino acid residues that constitute the positively charged cluster responsible for binding the phosphate region of A T P become buried at the novel G5-G6 interface created during activation. Additionally, the positively charged region on the 67 Figure 26. The absence of an A T P - b i n d i n g site wi th in the C a -activated C - t e r m i n a l ha l f of gelsolin. A ) G4-G6 with ATP, excised from inactive ATP-gelsolin structure in Figure 23 A. B ) Electrostatic surface of inactive G4-G6, with bound ATP, oriented as in A. C ) G4-G6 in its active form (Narayan et al. 2003). G4 is oriented as in A. D) Electrostatic surface of activated G4-G6, oriented as in C. [taken from Figure 3 of Urosev et al. 2006] 68 surface of activated G4-G6 required for binding the phosphate tail of A T P is absent (Figure 26 D). Hence, the activated C-terminal half of gelsolin would not be expected to bind A T P . Indeed, further docking experiments could not suggest sites for binding of A T P to activated G4-G6, either in isolation or in its actin bound form, that were of affinity comparable to that found for A T P in inactive gelsolin ( E s c o r e = -175; Table 2). Rigid-body docking of A T P into cavities identified within activated G4-G6 and at its interface with actin yielded E s c o r e values, in the range from -119 to -52 . Al lowing flexibility in the local protein side chains did not change these values. In contrast, a modeling search for residual ATP-binding ability in activated G l -G3/actin arrived at a different conclusion (Figure 27). The model in Figure 26 C and D was generated by superimposing the ATP-bound G3 portion of inactive ATP-gelsolin (Figure 27 A and B) onto G3 in the Gl-G3/act in structure (PDB ID 1RGI). No steric clashes occur and the G3 residues that are the main participants in coordination of the adenine ring in inactive ATP-gelsolin retain their relative positions. In addition, the proximity of A T P in this model to residues Lys218 and Arg221, both active players in the tail latch of inactive gelsolin and in interactions of activated gelsolin with actin, identifies them as candidates for interaction with the phosphate tail o f A T P in a slightly modified model. These features, together with the proximity of A T P to actin in the model (Figure 27 D), offer an explanation for the reported residual affinity of activated gelsolin for A T P (Gremm and Wegner 1999). MolDock was used to evaluate and improve on the model presented in Figure 27. Two possible ATP-binding cavities at the interface of G1-G3 with actin, and one within G1-G3, were identified. Rigid-body docking of A T P in a cavity located between G2, G3 and actin resulted in an E s c o r e = -153 (pose 1; Figure 28 B) , while a cavity located at the GI-actin interface scored lower, E S C O r e = -128 (pose 2). Upon allowing protein side chain flexibility in the binding regions corresponding to pose 1 and pose 2, the E s c o r e for pose 1 dropped to -167 (pose 1 ' ; Figure 28 A ) , while E s c o r e for pose 2 stayed the same. Docking runs performed in the third identified cavity gave E s c o r e = -102 (Table 3), regardless of whether side change flexibility was permitted. Both poses 1 and 1' preserve some of the interactions of the adenine ring with G3 residues that were observed in the structure of inactive ATP-gelsolin (Figure 24), while the phosphate moieties interact 69 mainly with amino acid residues from actin (Figure 28 A and B) . Furthermore, the comparable E s c c , r e values for pose 1' (-167; Table 3) and for ATP-binding in inactive gelsolin (-175; Table 2) add support for the idea that A T P can bind near the interface between G2-G3 and actin, thought to be the first site of contact between activated gelsolin and F-actin (Burtnick et al. 2004). Perhaps the binding of A T P can moderate the docking of the N-terminal half of activated gelsolin to actin. 70 Figure 27. A residual A T P - b i n d i n g site in the C a -activated N- t e rmina l ha l f o f gelsolin. A ) G1-G3 with ATP, excised from inactive ATP-gelsolin structure in Figure 23 A . B ) Electrostatic surface of inactive G1-G3 with ATP bound, oriented as in A. C ) A ribbon representation of the structure of activated G1-G3 bound to actin (Urosev 2003; Burtnick et al. 2004). The complex is positioned such that G3 lies in a similar orientation to that in A. ATP is modeled onto this complex through superposition of G3 from A. D) Electrostatic surface of the Gl-G3/actin complex, oriented as in C, with modeled ATP. Argl68 and Argl69, in G2, are key participants in a PIP2-binding site, [taken from Figure 4 of Urosev et al. 2006] 71 System Cav i ty Rigid Gl-G3/actin Gl-G3/actin with side chain flexibility G2-G3/actin -153 -167 Gl/actin -128 -126 Located within G1-G3 -98 -102 Table 3. E s c o r e values for dock ing a r i g id A T P structure onto G l - G 3 / a c t i n . Reported E s c o r e values are for the best scored ligand orientation within each cavity. A L y » S 1 ! Gin215 B Figure 28. A T P docked into the G2-G3 /ac t i n cavity. A ) A close-up view of the ATP orientation corresponding to E s c o r e = -167 (pose 1'), obtained by permitting protein side chain flexibility during docking. B ) A close-up view of the ATP orientation corresponding to E s c o r e = -153 (pose 1), obtained by keeping the protein rigid during docking. C ) Electrostatic surface of Gl-G3/actin, oriented approximately as in Figure 27. ATP is situated as in pose 1' in the positively charged pocket of the G2-G3/actin cavity. In A and B , actin residues involved in interactions with ATP are depicted in grey and gelsolin residues are shown in yellow. ATP, coordinates for which were taken from PDB file 1RGI, was kept rigid during the docking simulations. 72 4. The physiological relevance of interactions between A T P and gelsolin The physiological relevance of the binding of A T P to gelsolin remains obscure. Biochemical evidence suggests that a fall in cellular A T P levels w i l l lower the calcium threshold for activation, leading to an increase in gelsolin-severed and capped actin filaments. Local variations in cellular A T P concentration have been proposed to occur at times of massive actin polymerization, such as during cell movement (Carlier et al. 1993). Under such conditions, the contribution of gelsolin to the recycling of actin filaments may be enhanced. More severe cellular A T P depletion, due to starvation, hypoxia or mitochondrial dysfunction, results in accumulation of less dynamic A D P - F -actin-rich structures that form further aggregates in an unregulated manner (Gourlay and Ayscough 2005). These aggregates are speculated to reduce energy expenditure and are correlated with increased sensitivity of a cell to apoptosis. Modulation of gelsolin by A T P and its preference for binding to ADP-F-act in over ATP-F-act in, indicate a potential involvement in apoptosis beyond that so far portrayed (reviewed in the introduction). Interactions between G1-G3, actin and A T P are complex and seemingly contradictory. On one hand, biochemical data show that A T P enhances the interactions of monomeric actin with G1-G3 within whole gelsolin (Gremm and Wegner 1999) while, on the other, A T P exerts a destabilizing effect on interactions between isolated G1-G3 and actin (Laham et al. 1995). Also , A T P dislodges gelsolin from actin filaments under conditions where G1-G3, but not G4-G6, may be expected to bind to actin (Bearer 1995). A possible explanation for such discrepancies is that, under the experimental conditions used, gelsolin binds to one actin monomer while isolated G1-G3 binds to two actin units (Way et al. 1992; Burtnick et al. 2004). The second monomer in the latter case would be bound above the first (Figure 12) and, hence, A T P might interfere directly through competition with actin for the filament binding site on G2. Another possibility is that the second actin monomer induces a conformational change in the first that is incompatible with the structure of ATP-bound G1-G3 (Aguda et al. 2005). Thus, any effect of A T P on gelsolin-actin interactions might depend on whether binding o f gelsolin proceeds to G - or F-actin. 73 Our docking results indicate a potential affinity for A T P at the interface between G2-G3 and actin. The existence of a low affinity nucleotide-binding site at a gelsolin-actin interface at physiological nucleotide concentrations (low micromolar in plasma; low millimolar within cells) has been speculated before in order to explain the observed effects of nucleotides on the interaction between gelsolin and actin (Laham et al. 1995). Formation of gelsolin/actin complexes in the presence of millimolar A T P indicates one direction for future investigation of ATP-binding to activated gelsolin. Furthermore, the ATP-gelsolin structure reported here provides as basis for the design of mutations of gelsolin to explore the effects of A T P on gelsolin-actin interactions. Positively charged, Arg629, Arg542 and Lys503 are crucial in establishing affinity of A T P for inactive gelsolin, through interactions with its y-phosphate group. Therefore, substitutions of these amino acid residues with ones that are negatively charged, e.g. glutamate, or with hydrophobic ones, such as leucine are expected to abolish the binding of A T P in this binding site. Alternatively, stabilization of the adenine portion of the molecule, largely influenced by Trp369, can be compromised by tryptophan mutation to e.g. tyrosine, a polar amino acid of similar size. 74 B. Computational docking of PIP2 to gelsolin High levels of PIP2 are associated with actin polymerization, whereas low levels block assembly or promote actin severing activity. PIP2 is able to release gelsolin both from actin monomers and from capped filaments (Janmey and Stossel 1987) and it inhibits ab initio binding of gelsolin to actin. One of its three identified binding sites on gelsolin includes amino acid residues 621-634 in the C-terminal half of the protein (Feng et al. 2001). This site includes some residues associated with binding A T P to gelsolin (Figure 23). Both of these ligands bind more strongly to G4-G6, within the context of the whole protein, in the absence of Ca than in its presence. A T P has been reported to dissociate gelsolin from actin filaments, but, unlike PIP 2 , only in specific low C a 2 + ion conditions (Bearer et al. 1995). These similarities between the two phosphate-rich regulators of gelsolin led us to investigate the positively charged pocket of gelsolin associated with A T P phosphate binding as a possible docking site for PIP 2 . The negatively charged inositol phosphate portion of PIP 2 fits into this positively charged pocket with E s c o r e = —171/—176 (MolDock) (Table 4), and an estimated binding energy o f - 1 5 8 kJ/mol (Sybyl 7.0; Urosev et al. 2006; Figure 29 A ) . These values are comparable to the ones estimated for the binding of A T P to gelsolin (Table 1). A s with A D P and A T P at this site (Table 1), phosphate moieties, here from the inositol ring, are major determinants of the overall affinity for binding, in this case, of the "head region" of PIP2. The E S C ore for the docked inositol bisphosphate (IP2) decreases from —133/—146 to -171 /—176 upon incorporation of an additional phosphate group to represent inositol triphosphate ( IP3) , the PIP2 head region (Table 4). Amino acid residues Arg542, Lys503, Arg370 and Arg625, which are involved in binding A T P , are positioned to accommodate this inositol phosphate portion of PIP2 in a well docked model (Figure 29 B) . Rigid docking of di-C8-PIP2, in the ATP-binding cavity of gelsolin yielded an E s c o r e = -158 (MolDock). The presence of two rather long aliphatic chains (eight - C H 2 -repeats in each) implies a great number of possible conformations for the overall molecule. Since the large number of flexible bonds increases the complexity of the docking search, to best assess the affinity of PIP2 for the ATP-binding pocket, a model for binding PIP2 was built up incrementally. Upon incorporation of the polar stem of the 75 tail region of PIP2 (Figure 29 C), its polar interactions with R714 and R550 brought the energy of interaction to -232 kJ /mol (Sybyl 7.0). \ L i g a n d C a v i t y \ IP 3 (1N4K7 I H F A ) Flexible d i -C 4 - P I P 2 /rigid d i -C8-PIP 2 IP 2 (1 I9Z) / (1HG2) L P S inner core ATP-b ind ing cavity in 2 F G H -171/-176 -233/-158 -133/-146 -173 Other cavities in 2 F G H -107 -80 -86 -130* * A cavity located between domains G2 and G 6 , and containing residues from the N-terminal half o f gelsolin that are involved in binding P I P 2 , gave E s c o r e = —198. Table 4. E s c o r e values for var ious phosphate-bearing l igands docked to gelsolin. I P 3 coordinates were obtained from P D B files ID 1N4K and I H F A . 1P 2 from coordinates in P D B file 1I9Z corresponds to inositol 1,4-bisphosphate, while IP 2 , from coordinates in P D B file 1HG2 corresponds to inositol 4,5-bisphosphate. Ligands were kept rigid, except in the case of d i -C4-PIP 2 . A s previously, in the case of multiple potential pockets identified by the docking program, only the most negative E s c o r e is presented. 76 A B C Figure 29. A model for the b ind ing of P I P 2 on the surface of inactive gelsolin. A ) A portion o f the electrostatic surface o f gelsolin from Figure 23 C with docked PIP 2 . B ) Interactions between gelsolin and the inositol phosphate region o f P IP 2 . C ) Contacts between gelsolin and the tail portion o f P IP 2 , with the orientation rotated approximately 90° around the vertical axis with respect to that in A . [taken from Figure 5 of Urosev et al. 2006] Further stabilization to -241 kJ/mol (Sybyl 7.0) and E s c o r e = -233 (MolDock) is achieved upon including the hydrophobic interactions between the first few atoms of an aliphatic chain lying near gelsolin residues Ile554, Ile555 and Ala560 (Figure 29 C). The interactions of the polar PIP2 head with Arg370 explain how a token presence from the N-terminal half in this predominantly C-terminal site can increase the affinity of the site for PIP2 (Feng et al. 2001). Docking calculations predict stronger binding of PIP 2 over A T P at this site, in agreement with the observed higher affinity of gelsolin for PIP 2 (Feng et al. 2001) over A T P (Gremm and Wegner 1999; Yamamoto et al. 1990). The model is also in agreement with the suggestion that PIP2-binding to gelsolin occurs at surface-exposed sites so as to permit formation of a compatible lipid/protein/water interface on incorporation of the amphiphilic phosphoinositide (Feng et al. 2001). The results of this modeling study support the findings that the diacylglyceryl moiety plays as important a part in the overall binding of PIP 2 as does the polar head (Lin et al. 1997; Feng at al. 2001). Furthermore, the binding energy calculated for docking PI(3,4,5)P3, another common phosphoinositide encountered in cells, into this model was -240 kJ/mol, consistent with it having a reported binding affinity comparable to that of PIP2 (Feng et al. 2001). Hence, even though it is not clear where the rest o f the aliphatic chain would be 77 located in an extension of our model, the model does represent a plausible PfPVbinding pocket in the C-terminal region of gelsolin. Soaking of inactive gelsolin crystals in the presence of polyphosphoinositides, as well as their cocrystallization with gelsolin, are presently being investigated to test this model. L P S , a more complex molecule than either PIP 2 or L P A , shares the feature o f having multiple negative charges. These are mainly located in the inner oligosaccharide core of this molecule that has been shown to participate in providing resistance for a bacterial cell to polycationic peptides and hydrophobic antimicrobial compounds. The minimal composition of L P S is represented by a lipid A moiety and an inner oligosaccharide core of varying sugar content. Reportedly, L P S has a higher affinity for gelsolin than does PIP2 (Bucki et al. 2005). This led us to perform docking calculations in which inner core epitopes of L P S were fit to the ATP-binding pocket o f gelsolin. A disaccharide (PDB ID 1Q9R) yielded E s c o r e = -108; a trisaccharide (PDB ID 1Q9Q) gave Escore = -144; and a pentasaccharide containing phosphate groups (PDB ID 1Q9W) gave EScore = -173 (Table 4). Furthermore, an E S C O re = -198 was achieved when this pentasaccharide was docked to the pocket that includes part o f the N-terminal PIP 2 -binding sequence of gelsolin, which is exposed in the presence of C a 2 + (L in et al. 1997). These results suggest phosphorylated pentasaccharides to be good candidates for examination of the mode of interaction between gelsolin and L P S . The Gl-G3/act in structure (Urosev 2003; Burtnick et al. 2004) enabled further investigation into the mechanism of P l lVinduced dissociation of gelsolin from actin filaments. One of the two PHVbind ing sites located in the N-terminal half o f gelsolin (GI residues 132-140 and G2 residues 161-172; Y u et al. 1992; L i n et al. 1997) overlaps an F-actin binding site located in G2 (residues 168-170; Puius et al. 2000). In the inactive gelsolin conformation, these regions are buried inside the molecule where they are a part of the extended (3-sheet between these two domains (Figure 30 A ) . Hence, the regulatory effects due to the binding of PIP2 to these N-terminal binding sites may only come into play once gelsolin is activated and bound to actin filaments. Therefore, PIP2 may induce release of gelsolin from an actin filament by direct competition for the actin-binding residues (Burtnick et al. 1997). Support for this idea is found in ADF/cof i l i n family 78 proteins, which also bind to PIP2 at a site that overlaps with their actin-binding site (Moriyama et al. 1992). In analogy to how phospholipase Cp2 presents a K x x x K x K K motif that directs the enzyme to PIP2-rich surfaces (Simoes et al. 1995), each PIP 2-binding region in the N -terminal half gelsolin contains a K x K K sequence (Figure 30). These lysine residues are exposed in the activated actin-bound structure of G 1 - G 3 (Figure 30 B) and may govern initial contacts with PIP 2 . Upon binding to PIP2, it is likely that G l and G2 would undergo PIP2-induced inter- rather than intra-domain conformational changes, facilitated by a conformational change in the G1-G2 linker. The G 1 - G 2 linker (residues 141-160) is largely hydrophobic (eight hydrophobic residues) and it separates the two mentioned N -terminal PIP2-binding regions. It follows two distinctly different paths prior to and after calcium activation (Figure 30), and thus represents a strong candidate to be involved in PIP2-induced interdomain conformational change. This linker makes very close contacts with actin in the G l -G3 /ac t i n complex, creating a 410 A 2 interaction interface. Through competition for this surface, PIP2 may dislodge the type-2 calcium ion from the Gl /ac t in interface and release gelsolin from actin (Weeds et al. 1995; Choe et al. 2002), possibly by internalizing the hydrophobic residues from the G 1 - G 2 linker and positioning the basic K x K K residues at the surface. In conclusion, molecular modeling and the Gl-G3/act in structure support the hypothesis that determinants of the initial interaction between gelsolin and PIP2 are expected to be different in the presence or absence of C a 2 + (Lin et al. 1997). It is possible that a PnVbased inhibitory mechanism involves a combined competitive approach and that, once dissociated, gelsolin may refold on the face of a P H V r i c h membrane, exchanging its GI/G2-PIP2 interactions (Figure 30) for G4-G6-PIP2 interactions, as modeled in Figure 28. Furthermore, release of C a 2 + in the relevant cellular region may lead to dissociation of PIP2 from gelsolin, in accordance with the Ca 2 +-sensitivity of the binding of PIP 2 to the C-terminal half of gelsolin. 79 Figure 30. Regions of domains G I and G 2 involved in b ind ing P I P 2 . A) G1-G2 excised from the Ca + -free structure of gelsolin (Burtnick et al. 1997). GI and G2 are colored pink, PIP 2-binding regions 132-140 and 161-172 are colored red and green, respectively. K x K K motifs with lysine residues drawn are labeled K x K K l and K x K K 2 . The G1-G2 linker (residues 141-160) is colored orange. B ) Activated GI and G2 bound to actin, as observed in the Gl-G3/actin structure (1RG1). Type-2 calcium ion in GI is represented as a gray sphere. The same coloring scheme is preserved as in A. [taken from Figure 3 of Burtnick et al. 2004] SO C. Metal ion exchange in the N-terminal half of gelsolin Isolated G1-G3 is capable of severing actin filaments independently of C a 2 + . A s there is no tail latch to open in this case, severing likely starts with G2 binding to the side of F-actin. This would be followed by opening of the G1-G3 latch in an actin-assisted manner, permitting the binding of G l in its normal way, which results in severing (Burtnick et al. 2004). Despite the absence of strict regulation by C a 2 + , its presence does affect some aspects of the structure and function of G1-G3, as implied by an observed increase in tryptophan fluorescence intensity, a decrease in proteolytic susceptibility, and an enhanced severing activity when C a 2 + is present (Pope et al. 1991; Selden et al. 1998). Also , the affinity of isolated G l for actin drastically increases in the presence of calcium ions and C a 2 + associated with its Type-2 site has been reported to be non-exchangeable in the presence of E G T A , a calcium ion chelator (Weeds et al. 1995). We probed the roles of the calcium ion-binding sites within the N-terminal half of gelsolin by varying the free metal ion environment surrounding Gl-G3/act in crystals. 1. Exchange of calcium ions in Gl-G3/actin crystals with T b 3 + Both crystalline forms of Gl-G3/act in , with space groups P3i21 and P6 5 22, initially grown in the presence of calcium ions, were soaked in solutions containing, 1.0 and 0.2 m M Tb(NOs)3 respectively. Anomalous diffraction by bound terbium ions enabled identification of their positions even at a relatively low overall resolution of crystal diffraction (3.5 A ) . Data collection and refinement statistics for the P3i21 crystals are provided in Table 5. In both types of crystal, T b 3 + ions replaced C a 2 + ions in the Type-2 sites of G l and G3, in the Type-1 site in G l , and at the location in actin associated with ATP-bound C a 2 + (Figure 31). Terbium ions also were observed to occupy the previously vacant Type-2 Ca binding site of G2 in these Gl-G3/ac t in crystals (Burtnick et al. 2004). 81 Figure 31. Subst i tut ion of t e rb ium ions for C a in Gl -G3/ac t in . The terbium anomalous difference electron density map, derived from terbium-soaked P3,21 crystals of Gl-G3/actin, is superimposed onto a representation of the polypeptide backbone structure of Gl-G3/actin (PDB ID 1RGI). Terbium ions, the positions of which are shown by positive density in this map, are found to occupy a single Type-1 and three Type-2 C a 2 + ion-binding sites on gelsolin, and a site at the bottom of the nucleotide cleft in actin. The metal ion-binding sites in gelsolin are denoted as Gl -2 , G2-2, G3-2 and G l - 1 . [taken from Figure 7 of Chumnarnsilpa et al. 2006] The absence of a metal ion in the Type-2 C a -binding site of G2 in the structure of Gl-G3/act in was unexpected when that structure was solved (Urosev 2003), as all other Type-2 sites in the activated N - and C-terminal halves of gelsolin were filled (Choe et al. 2002; Narayan et al. 2003). Analysis of diffraction data from crystals grown from solutions of the G A 2 complex, with space groups P6s22 and unit cell dimensions of approximately a=b=146 A, c=388 A, confirmed the vacancy of the Type-2 site in G2. A n 82 analogous situation with respect to occupancy of metal ion-binding sites is observed in the structure of an activated mutant of the protein, CapG, upon soaking CapG crystals in the presence of 10 m M E U C I 3 . CapG is a three-domain (C1-C3, respectively) member of a gelsolin family. It can cap, but not sever, actin filaments. The mutant form in question had part of its amino acid sequence, corresponding to the long helix of its first domain and the C1-C2 linker region, substituted with the analogous sequences from gelsolin in an attempt to introduce severing activity. The activated structure of this CapG mutant resembles that of activated G1-G3, as observed in G l -G3 / a c t i n structure. Its Type-2 calcium ion binding sites in domains C I and C 3 are filled with E u 3 + , while the analogous site in domain C2 is unoccupied (Zhang et al. 2006). In the case of gelsolin, occupancy of this site in a crystallographic structure has been observed only in the case of isolated G2 in the presence of Cd ions (Kazmirski et al. 2002). Similar to our findings that terbium ions can substitute for C a 2 + in Gl -G3 / a c t in , it 2_|_ is possible to replace the Ca ions bound into crystals of the activated C-terminal half of gelsolin ( G 4 - G 6 ) by soaking them in solutions that contain 1.0 m M Tb(NOs)3 (Chumnarnsilpa et al. 2006). A l l Type-2 metal ion-binding sites were found to be occupied with terbium ions. Interestingly, the Type-1 site in this actin-free structure that in the case of binding C a 2 + w o u l d normally be completed by actin (Choe et al. 2002), was occupied by T b 3 + . Such an observation suggests an involvement of Type-1 Ca 2 + -binding, not only in moderating the binding of gelsolin to actin, but possibly in the activation process for gelsolin as well . Crystallization of the isolated, activated N-terminal half of gelsolin would enable testing of whether its Type-1 Ca 2 + -binding site in G I can be occupied in the absence of actin, in analogy to the G 4 Type-1 site in the C-terminal half of the protein. 2. Back-exchange of Tb ions present in Gl-G3/actin with calcium ions P6522 crystals initially soaked in 0.2 m M T b 3 + were back-soaked in solutions of increasing concentration of up to 10 m M C a C l 2 . These crystals diffracted to a resolution of only 6.8 A (Table 5), but anomalous diffraction from the bound terbium ions enabled identification of the positions of any such ions in the G l -G3 / a c t i n structure that were not 83 displaced by Ca . This soaking procedure did not result in ion exchange at the Type-2 site in G2, but did result in loss of T b 3 + from the Type-2 sites in G l and G3 (Figure 32). These results confirm the existence of a valid Type-2 metal ion-binding site in G2, at least in the context of activated G1-G3 in Gl-G3/act in , and suggest that its occupancy may be governed by the conformational stability of gelsolin. Namely, the Ca -binding event in G2 might only be transient, existing on the path from an inactive to an active conformation of gelsolin only until the stable structure has been attained. Type-2 site occupancy by C d 2 + in isolated G2 (Kazmirski et al. 2002) may not have a physiological meaning. Although domains are semi-independent units, their stability is often conditioned by interactions with other domains. Interestingly, the C a 2 + in the Type-2 site in G l was previously regarded as non-exchangeable (Weeds et al. 1995). Our data suggest it to be readily exchangeable with T b 3 + , both in forward and reverse directions. Hence, in the context of Gl-G3/act in , the G l Type-2 metal ion-binding site operates in a standard equilibrium fashion. In contrast, 2+ while the Ca in the Type-1 metal-ion binding site sandwiched between G l and actin is exchangeable with a terbium ion in the context of the Gl-G3/act in complex (Figure 31), backwashing of terbium-soaked crystals with 10 m M CaCl2 does not readily displace this Tb (Figure 32). This is consistent with calcium-binding sites often displaying stronger affinity for terbium than for calcium ions, largely due to their greater charge (Evans 1983). In the case of G4-G6 crystals backwashed in the presence of excess calcium ions, all terbium ions were removed (Chumnarnsilpa et al. 2006). So, the lack of T b 3 + ion exchangeability in the Type-1 metal ion-binding site of G l , in the presence of actin, may reflect the importance of calcium ion-binding at this site for strong Gl-ac t in interactions during filament severing. 84 Figure 32. Backwash ing of T b -soaked crystals of G l - G 3 / a c t i n w i t h c a l c i u m ions. A terbium anomalous difference electron density map is superimposed onto the Gl-G3/actin structure. P6522 crystals initially soaked in 0.2 mM Tb(N0 3 ) 3 were backwashed with the a solution 10 m M CaCl 2 . Along with the Tb 3 + ion associated with ATP, those in Type-2 sites in GI and G3 were displaced by C a 2 + . The Type-2 site in G2, which was empty in the structure of Gl-G3/actin (PDB ID 1RGI) and the Type-1 site sandwiched between actin and G1, remained occupied by terbium ions, [taken from Figure 7 of Chumnarnsilpa et al. 2006] 8 5 G l - G 3 / a c t i n (Figure 30) G-G3/actin (Figure 31) Wavelength ( A ) 0.970 0.974 Resolution range ( A) 30.0-3.50 (3.69-3.5)* 30.0-6.80 (7.17-6.80)* Space group P3,21 P6 5 22 Unit cell (a,b,c: A) (a,p,y: deg.) a=b=144.6, c=128.3 a=p=90, y=120 a=b=146.3,c=388.8 a=p=90,Y=120 Total reflections 54, 943 (4035) 41,945 (6261) Unique reflections 17,781 (1659) 4492 (659) Redundancy 3.1 (2.4) 9.3 (9.5) Completeness (%) 88.9 (58.1) 96.0(100.0) Average 1/8 15.7(3.3) 14.7(5.2) R m e r g e : Z * [ Z ,| 4 - (Ih)\l (/*)] (%) 6.2 (26.3) 13.2 (40.3) R f a c t o r ( % ) 32.4 42.0 Table 5. Crys ta l lograph ic data and refinement statistics for terbium-soaked G l - G 3 / a c t i n crystals (Figures 30 and 31). Numbers in parentheses refer to data collected in the resolution range indicated with an asterix. 3. Implications of transient calcium-binding in G2 for F A F Mutation of A s p l 8 7 in gelsolin leads to accumulation of amyloid fibrils. In calcium-free gelsolin, the furin cleavage site (Chen et al. 2001) located between A r g l 7 2 and A l a l 7 3 lies protected in the B strand of the core P-sheet of G2, stabilized by a salt bridge between A s p l 8 7 and L y s l 6 6 (Figure 16) (Burtnick et al. 1997). On activation of gelsolin, this ionic interaction is abolished to allow L y s l 6 6 to interact with Glu263 from the G 2 - G 3 linker, which makes additional contacts with the A - B loop (Figure 33). 86 Figure 33. Region in the N- te rmina l ha l f of gelsolin involved in F A F . G2 and G3 are excised from the G l - G3/actin complex structure (1RG1), with the A - B loop shown in yellow and the B strand, containing the furin cleavage site, indicated in purple. Residue Lysl66 is shown forming a salt bridge to Glu263. [taken from Figure 6 of Burtnick et al. 2004] The net result is placement of the furin-sensitive site at the center of the G 2 - G 3 interface, shielding it from proteolytic attack. We hypothesize that transient binding of a calcium ion to the Type-2 site in G2 mediates conversion between the inactive and activated states of gelsolin. The F A F mutation site, Asp 187, has been shown to coordinate this ion and stabilize isolated G2 (Kazmirski et al. 2002; Huff et al. 2003). Mutation of A s p l 8 7 to Asn or Tyr prevents G2 from transiently binding the calcium ion and, therefore, is expected to alter the kinetics of interchange between the two protective states. So, the molecule w i l l spend more time, possibly becoming stranded, between conformations, rendering the cleavage site available for proteolysis. 87 D. Towards the structures of gelsolin/actin complexes 1. Purification, crystallization and X-ray diffraction analysis of gelsolin bound to different actin species ( G A 2 A T P / G A 2 A D P / G A X ) Our original procedure for preparation of the G A 2 A T P c o m p l e x (Urosev 2003) led to proteolytic cleavage of gelsolin during the weeks required for nucleation of crystal growth (Urosev 2003). To remove contaminant proteases, we added a final gel filtration step to our procedure for preparing actin. A n added benefit of this step is removal of aggregates, usually oligomeric chains, of actin prior to formation of complexes with gelsolin. The modified procedure resulted in complexes of intact gelsolin bound to two actin monomers (Figure 34 A and B) . Two types of crystals formed in solutions of the G A 2 A T P complex under conditions similar to those used for crystallization of G l -G3/actin: 2 % P E G 8000, 100 m M sodium acetate, 2 m M C a C l 2 , pH 4.6. Both types of crystals revealed ordered forms of only Gl-G3/act in . The crystal type characterized by the smaller volume unit cell, with a space group P3i21 and approximate unit cell dimensions a = b = 144 A , c = 128 A , had been observed previously (Urosev, 2003). The larger volume unit cell crystals, with space group P6 5 22, and approximate unit cell dimensions of a = b = 146 A , c = 388 A , were a novel crystalline form (Figure 34 C). The volume of such a unit cell is sufficient to accommodate Gl-G3/act in , which packs in a regular crystalline lattice, and the G4-G6/actin, acting as a solute component in the intervening solvent spaces (Figure 35). This suggestion is supported by the lack of detectable proteolysis during nucleation and growth of crystals of the complexes. Attempts to impose greater rigidity onto the complex by the crosslinking of protein molecules within the crystals, performed in the presence of gradually introduced 0.5 and 1% glutaraldehyde in the well solution, did not result in any additional electron density being observed. Alternative crosslinking approaches, such as crosslinking of the complex prior to crystallization, are being examined by another member of the Burtnick research group. 88 Figure 34. Pur i f ica t ion and crysta l l izat ion of G A 2 A T P . A ) Elution o f G A 2 A T P from a Sephacryl S-300 gel filtration column (97.5 x 2.5 cm) with 2 m M T r i s - H C l , 0.2 m M C a C l 2 , 1 m M D T T , p H 7.6 buffer at a rate of 2 mL/min . M i n o r amounts o f F-actin that form during complex formation were removed (shoulder peak). B) SDS-gel electrophoresis o f G A 2 A T P present in the fractions constituting the main peak. C ) P6 5 22 crystal grown by hanging drop vapor diffusion from a 10 mg/mL protein sample mixed in a 1:1 ratio (v/v) with well buffer: 2 % P E G 8000, 100 m M sodium acetate, 2 m M C a C l 2 , p H 4.6. 89 Figure 35. Pack ing of G l - G 3 / a c t i n wi th in P6 5 22 crystals. The asymmetric unit, consisting of G l -G3/actin, is shown in yellow. The extended structure of the crystal, up to 100 A across in this representation, was generated using the symmetry operators governed by the space group. Symmetry-generated molecules are shown in violet. Empty regions within the sphere represent space in the crystal occupied by solution, likely containing G4-G6/actin (image was generated using COOT). A multitude of crystallization trials were performed in pursuit of suitable conditions for obtaining crystals with a high degree of order in both gelsolin halves. One search, centred on polyethyleneglycols of different molecular weights as precipitating agents and various inorganic and organic salts as additives, did not yield any new type of G A 2 A T P crystal. Recently, a different strategy has been followed, based on small organic acids and their salts, e.g., malonate, formate and citrate, being the primary factor in promoting crystallization. Their usefulness in crystallization has been attributed to indirect stabilization of protein structure, or, more directly, through their incorporation at protein-protein interfaces (Larson et al. 2007). The long G3-G4 linker, in an extended form when activated gelsolin is bound to two actin monomers, lacking specific contacts with actin may explain the conformational dynamics within G A 2 . To test the importance of this linker for integrity o f the G A 2 complex, students of our collaborator, Dr. Robert Robinson, have prepared and are forwarding to us gelsolin constructs with different lengths of G3-G4 linker to crystallize alone and in complexes with actin. 90 Interactions of the N - and C-terminal halves of gelsolin with an ATP-act in monomer have been observed (Urosev 2003; Burtnick et al. 2004; Choe et al. 2002). However, gelsolin preferentially binds to ADP-F-act in over ATP-F-act in, and conformational differences in the nucleotide cleft region and D-loop of actin have been reported for A T P - and ADP-bound forms of the protein (reviewed by Aguda et al. 2005). Investigation of the basis for such a preference can be performed through crystallographic analysis of a complex formed with gelsolin and ADP-act in monomers. Therefore, G-actin was loaded with A D P by extensive dialysis against solutions containing millimolar concentrations of A D P and brought together in a complex with gelsolin to form GA2 A D P . Interestingly, crystals did not grow in the conditions that reproducibly yielded crystals of G A 2 A T P , suggestive of at least a subtle conformational difference between the two nucleotide forms of the complex. Screening of small organic molecule cocktails as precipitants is being conducted in our laboratory in order to advance this project. Yet another approach taken to produce stabile gelsolin/actin complexes consisted of resorting to formation of a complex of gelsolin with a crosslinked actin species, mainly represented by actin trimers (Figure 36 A ) . These allowed for the formation of higher order structures that can accommodate additional G2 contacts with actin beyond those involved in the gelsolin complex with a non-covalent actin dimer ( G A 2 ) . Such complexes (GA X ) were partially purified (Figure 36 B) . Fractions containing the least amounts of contaminating actin monomers were crystallized successfully in 4 % P E G 3350, 100 m M H E P E S , p H 6.0, 2 m M C a C l 2 (Figure 36 C). Crystals also were observed to grow in 2 % P E G 6000, 100 m M H E P E S , pH 6.0, 2 m M C a C l 2 . These crystals diffracted to a resolution of 6 A, which was insufficient to elucidate the structure of the complex. 91 Figure 36. Fo rma t ion and crysta l l izat ion of G A , . A) SDS-PAGE of crosslinked actin trimers (MW ~126 kDa) that were subjected to gel filtration on a Sephacryl S-300 column in 2 mM Tris-HCl, 0.2 mM CaCl 2 , 0.2 mM ATP, 1 mM DTT, pH 7.6 buffer. B) SDS-PAGE of G A X subjected to purification in the same way as actin trimers. Only partial separation is achieved and fractions corresponding to the indicated enclosed area were used in crystallization trials. C) G A X crystal grown from a 10 mg/mL protein sample mixed in a 1:1 (v/v) ratio (total volume 2 ul) with 4 % PEG 3350, 100 mM HEPES, pH 6.0, 2 mM C a C l 2 at 4 degrees, using the hanging drop vapor diffusion method. 2. Effects of sodium malonate on protein-protein interfaces in crystals of inactive gelsolin In our soaking experiments, sodium malonate media were identified as suitable for exchange of the ammonium sulphate media present in crystals of inactive gelsolin. The crystals preserved their features and diffracted to 3 A resolution. Crystals were soaked in the presence of 1.35 M sodium malonate, 100 m M Tr i s -HCl , p H 8.0, 1 m M E D T A for a period of approximately one hour. The same solution served as a cryoprotectant. Attempts to incorporate ligands, such as L P S , into such gelsolin crystals proved unsuccessful. However, X-ray diffraction analysis of these malonate-soaked crystals revealed a conformational change in the A - A ' loop at the gelsolin-gelsolin interfaces within the asymmetric units, bringing it closer to the core of domain G2 (Figure 37 A) . G l u l 5 6 in this loop region is a suitable substitute for A s p l 9 2 in interactions with Arg225 located in the long helix of G2 (Figure 37 B) . Similar minor 92 modifications at the surfaces of protein structures have proven crucial for successful crystallization (Larson et al. 2007). Hence, sodium malonate is a good candidate to be explored in greater depth as a precipitating agent in the crystallization of various gelsolin-based complexes. Malonate was not detected in the electron density map, which is in line with the supposed indirect nature of its influence on protein loop interactions with various parts of a protein (Larson et al. 2007). 93 Figure 37. Effects o f sodium malonate on the conformation of the A - A ' loop i n gelsolin. A ) Ribbon representations of the proteins in the asymmetric units of crystals of inactive gelsolin, one grown in the presence of sodium malonate and the second in ammonium sulphate (PDB file 1 DON), are superimposed. One molecule in the asymmetric unit is colored green and the other is shown in violet. Differences occur primarily in the A - A ' loop, as indicated by arrows: full arrows indicate the loop conformation in the presence of ammonium sulphate, while broken arrows point to the loop conformation in the presence of sodium malonate. B ) Superimposed G2 domains from the inactive gelsolin structures observed in the presence of sodium malonate (in blue, denoted with *) and ammonium sulphate (from PDB file 1 DON, in green), and from the activated Gl-G3/actin structure (PDB ID 1RGI, in purple). 94 E . Conclusions This thesis presents a study on interactions of gelsolin with its ligands, A T P , PIP 2 and calcium ions. A T P was successfully soaked into crystals of inactive gelsolin and the molecular details of its interactions revealed. In silico investigation of the docking of A T P to gelsolin supports the hypothesis that A T P is a strong and specific ligand for the observed site. Many previously reported observations on ATP-binding to gelsolin can be explained by the fact that the binding site spans both the N - and C-terminal halves of the protein. The separation of the two halves during Ca 2 + -induced activation clarifies the observed decrease in affinity of gelsolin for A T P in the presence of these ions. Computational docking experiments suggest that some residual affinity of activated gelsolin for A T P may be retained at the interface between G2-G3 and actin. Through such an interaction, the binding of A T P may moderate the anchoring of the N-terminal half of activated gelsolin to actin filaments, the first event in the filament severing process. The binding of PIP2 has been modeled at a surface-exposed site on gelsolin that is associated with binding A T P phosphate groups. Some of the amino acid residues that participate in these interactions have been previously reported to be associated with a PIP 2-binding region in the C-terminal half of the protein. The computed binding energy for such a model reflects the measured higher affinity of gelsolin for PIP 2 than for A T P . Inhibition by PIP2 of gelsolin binding to actin is expected to occur through its interactions with this binding site in the C-terminal half of gelsolin in its Ca 2 +-free form. The regulatory effects due to the binding of PIP2 at its two putative binding sites in the N -terminal portion of gelsolin may only come into play once gelsolin is activated and bound to actin filaments. The uncapping of gelsolin from actin filament ends may proceed through direct competition of PIP2 for the actin-binding site in G2. Furthermore, the G l -G2 linker may facilitate the exchange of (G1-G2)-PIP2 for (G4-G6)-PIP2 interactions modeled in this thesis, during the transition from active to inactive gelsolin. A novel crystalline form, with a larger unit cell, of a complex of intact gelsolin bound to two actin monomers is documented in the thesis. Crystallographic analysis revealed only Gl-G3/act in , although the unit cell volume and lack of detectable 95 proteolysis suggest that G4-G6/actin may be present in a form that lacks crystalline order in the solvent channels within these crystals. The absence of a calcium ion in the G2 Type-2 metal ion-binding site in such crystals, and the lack of exchangeability of a terbium ion at this site when Tb 3 +-soaked crystals were back-soaked in excess of C a 2 + , may be explained by transient binding of C a 2 + in G2. This particular Ca 2 + -binding event may help drive the conformational changes that accompany the activation process up until the stable structure of activated gelsolin has been attained. Proteolytic susceptibility of F A F gelsolin can be attributed to the absence of a complete metal ion coordination sphere in the mutant G2, which leads to a dysfunction in this particular Ca 2 + -binding event. 96 Bibliography Aguda, A . H . , Burtnick, L . D . , and Robinson, R . C . (2005). The state of the filament. E M B O Rep. 6, 220-226 Al len , P .G. , and Janmey, P . A . (1994). Gelsolin displaces phalloidin from actin filaments. A new fluorescence method shows that both C a 2 + and M g 2 + affect the rate at which gelsolin severs F-actin. J. B i o l . Chem. 269, 32916 -32923 Al len , P .G. , Laham, L . E . , Way, M . , and Janmey, P .A . (1996). Binding of phosphate, aluminum fluoride, or beryllium fluoride to F-actin inhibits severing by gelsolin. J. B i o l . Chem. 271, 4665-4670 Asch, H . L . , Head, K . , Dong, Y . , Natoli, F., Winston, J.S., Connolly, J .L. , and Asch, B . B . (1996). Widespread loss of gelsolin in breast cancers of humans, mice, and rats. Cancer Res. 56, 4841^1845 Azuma, T., Koths, K . , Flanagan, L . , and Kwiatkowski, D . (2000). Gelsolin in complex with phosphatidylinositol 4, 5-bisphosphate inhibits caspase-3 and -9 to retard apoptotic progression. J. B i o l . Chem. 275,3761-3766 Bearer, E . L . (1995). Cytoskeletal domains in the activated platelet. Cel l . Mo t i l . Cytoskel. 30, 50-66 Becker, E .W. (2006). The roles of A T P in the dynamics of the actin filaments of the cytoskeleton. B io l . Chem. 387, 401-406 Belmont, L . D . , Orlova, A . , Drubin, D . G . , and Egelman, E . H . (1999). A change in actin conformation associated with filament instability after Pi release. Proc. Natl . Acad. Sci. U S A 96, 29-34 Berman, H . M . , Westbrook, J., Feng, Z . , Gil l i land, G . , Bhat, T . N . , Weissig H . , Shindyalov I.N., and Bourne P.E. (2000). The Protein Data Bank. Nucleic Acids Res. 28, 235-242 Bryan, J. (1988). Gelsolin has three actin-binding sites. J. Cel l . B i o l . 106, 1553-1562 Bryan, J., and Kurth, M . C . (1984). Actin-gelsolin interactions. Evidence for two actin-binding sites. J. B i o l . Chem. 259, 7480-7487 Bucki , R., Georges, P . C , Espinassous, Q., Funaki, M . , Pastore, J.J., Chaby, R., and Janmey P .A . (2005). Inactivation of endotoxin by human plasma gelsolin. Biochemistry 44, 9590-9597 Bucki , R., and Pastore, J.J. (2006). Bacterial endotoxin as inhibitor of the enzymatic activity of human thrombin. Eur. J. Haematol. 76, 510-515 97 Burtnick, L . D . , Koepf, E . K . , Grimes, J., Jones, E . Y . , Stuart, D.I., McLaughl in , P.J., and Robinson, R . C . (1997). The crystal structure of plasma gelsolin: implications for actin severing, capping, and nucleation. Cel l 90, 661-670 Burtnick, L . D . , Urosev, D. , Irobi, E. , Narayan, K . , and Robinson, R . C . (2004). Structure of the N-terminal half of gelsolin bound to actin: roles in severing, apoptosis and F A F . E M B O J. 23, 2713-2722 Carlier, M . F . , Jean, C , Rieger, K . J . , Lenfant, M . , and Pantaloni, D . (1993). Modulation of the interaction between G-actin and thymosin B4 by the A T P / A D P ratio: possible implication in the regulation of actin dynamics. Proc. Natl Acad. Sci. U S A 90, 5034-5038 Chayen, N . E . (2004). Turning protein crystallization from an art into a science. Curr. Opin. Struct. B i o l . 14, 577-583 Chayen, N . E . , Saridakis, E. , and Sear, R.P. (2006). Experiment and theory for heterogeneous nucleation of protein crystals in a porous medium. Proc. Natl. Acad. Sci. U S A 103, 597-601 Chellaiah, M . , Fitzgerald, C , Alvarez, U . , and Ffruska, K . (1998). c-Src is required for stimulation of gelsolin-associated phosphatidylinositol 3-kinase. J. B i o l . Chem. 273, 11908-11916 Chen, C D . , Huff, M . E . , Matteson, J., Page, L . , Phillips, R., Kel ly , J.W., and Balch, W . E . (2001). Furin initiates gelsolin familial amyloidosis in the Golgi through a defect in C a 2 + stabilization. E M B O J. 20, 6277- 6287 Childs, T. J., Watson, M . H . , Novy, R .E . , L i n , J.J., and Mak, A . S . (1992). Calponin and tropomyosin interactions. Biochim. Biophys. Acta 1121, 41—46 Choe, H . , Burtnick, L . D . , Mejillano, M . , Y i n , H . L . , Robinson, R . C , and Choe, S. (2002). The calcium activation of gelsolin: insights from the 3 A structure of the G4-G6/actin complex. J. M o l . B i o l . 324, 691-702 Chumnarnsilpa, S., Loonchanta, A . , Xue B . , Choe, H . , Urosev, D. , Wang, H . , Lindberg, U . , Burtnick, L . D . , and Robinson, R .C . (2006). Calcium ion exchange in crystalline gelsolin. J. M o l . B io l . 357, 773-782 Collaborative Computational Project, Number 4. (1994). The C C P 4 suite: programs for protein crystallography. Acta Crystallogr. D B io l Crystallogr. 50, 760-763 Cunningham, C C , Stossel, T.P., and Kwiatkowski, D.J . (1991). Enhanced motility in NIH3T3 fibroblasts that overexpress gelsolin. Science 251, 1233-1236 98 Cunningham, C C , Vegners, R., Bucki , R., Funaki, M . , Korde, N . , Hartwig, J .H. , Stossel, T.P., and Janmey, P . A . (2001). Cel l permeant polyphsophoinositide-binding peptides that block cell motility and actin assembly. J. B i o l . Chem. 276, 43390^13399 Cusack, S., Belrhali, H . , Bram, A . , Burghammer, M . , Perrakis, A . , and Riekel, C. (1998). Small is beautiful: protein micro-crystallography. Nat. Struct. B i o l . 5 Suppl, 634-637 Danley, D . E . (2006) Crystallization to obtain protein-ligand complexes for structure-aided drug design. Acta Crystallogr. D B i o l Crystallogr. 62, 569-575 De Corte, V . , Demol, H . , Goethals, M . , Van Damme, J., Gettemans, J., and Vandekerchkove, J. (1999). Identification of Tyr438 as the major in vitro c-Src phosphorylation site in human gelsolin: a mass spectrometric approach. Protein Sci. 8, 234-241 De Corte, V . , Gettemans, J., and Vandekerchkove, J. (1997). Phosphatidylinositol 4,5-bisphosphate specifically stimulates PP60(c-src) catalyzed phosphorylation of gelsolin and related actin-binding proteins. F E B S Lett. 401, 191-196 Ditsch, A . , and Wegner, A . (1995). Two low-affinity Ca 2 + -binding sites of gelsolin that regulate association with actin. Eur. J. Biochem. 229, 512-516 Doi , Y . (1992). Interaction of gelsolin with covalently cross-linked actin dimer. Biochemistry 31, 10061-10069 Dos Remedios, C . G . , Chhabra, D. , Kekic , M . , Dedova, I .V., Tsubakihara, M . , Berry, D . A . , and Nosworthy, N . J . (2003). Act in binding proteins: regulation of cytoskeletal microfilaments. Physiol. Rev. 83, 433-473 Edgar, A . J. (1990). Gel electrophoresis of native gelsolin and gelsolin-actin complexes. J. Muscle Res. Cel l Mot i l . 11, 323-330 Elzinga, M . , and Phelan, J.J. (1984). F-actin is intermolecularly crosslinked by N,N ' -p -phenylenedimaleimide through lysine-191 and cysteine-374. Proc. Natl . Acad. Sci . U S A 81, 6599-6602 Emsley, P., and Cowtan, K . (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr. D B io l .Crystallogr. 60, 2126-2132 Evans, C. H . (1983). Interesting and useful biochemical properties of lanthanides. Trends Biochem. Sci. 8, 445^149 Falet, H . , Barkalow, K . L . , Pivniouk, V . I . , Barnes, M . J . , Geha, R.S. , and Hartwig, J .H. (2000). Roles of SLP-76, phosphoinositide 3-kinase, and gelsolin in the platelet shape changes initiated by the collagen receptor GPVI/FcRg-chain complex. Blood 96, 3786-3792 99 Feng, L . , Mejillano, M . , Y i n , H . L . , Chen, J. and Prestwich, G .D . (2001). Full-contact domain labeling: identification of a novel phosphoinositide binding site on gelsolin that requires the complete protein. Biochemistry 40, 904-913 Ferjani, I., Fattoum, A . , Maciver, S.K., Benistant, C , Chahinian, A . , Mana, M . , Benyamin, Y . , and Roustan, C. (2006). A direct interaction with calponin inhibits the actin-nucleating activity of gelsolin. Biochem. J. 396, 461-468 Ferjani, I., Fattoum, A . , Maciver, S.K., Manai, M . , Benyamin, Y . , and Roustan, C. (2007). Two distinct sites of interaction form the calponin: gelsolin complex and two calcium switches control its activity. Biochim. Biophys. Acta 1774, 952-958 Ferrary, E., Cohen-Tannoudji, M . , Pehau-Arnaudet, G. , Lapillonne, A . , Amman, R., and Ruiz, T. (1999). In vivo, v i l l in is required for Ca 2 +-dependent F-actin disruption in intestinal brush borders. J. Cell B io l . 146, 819-830 Garcia-Ruiz, J . M . (2003). Counterdiffusion methods for macromolecular crystallization. Methods Enzymol. 368, 130-154 Goetzl, E.J . , Lee, H . , Azuma, T., Stossel, T.P., Turck, C .W. , and Karliner, J.S. (2000). Gelsolin binding and cellular presentation of lysophosphatidic acid. J. B i o l . Chem. 275, 14573-14578 Gourlay, C .W. and Ayscough, K . R . (2005). The actin cytoskeleton in ageing and apoptosis. F E M S Yeast Res. 5, 1193-1198 Gremm, D. and Wegner, A . (1999). Co-operative binding of C a 2 + ions to the regulatory binding sites of gelsolin. Eur.J. Biochem. 262, 330-334 Harris, H . E . (1988). The binary complex of pig plasma gelsolin with Mg 2 + -G-ac t in in A T P and A D P . F E B S Lett. 233, 359-362 Harris, H .E . , Bamburg, J.R. and Weeds, A . G . (1980). Act in filament disassembly in blood plasma. F E B S Lett. 121, 175-177 Hellweg, T., Hinssen, H . , and Eimer, W . (1993). The Ca 2 + -induced conformational change of gelsolin is located in the carboxyl-terminal half of the molecule. Biophys. J. 65, 799-805 Herrmannsdoerfer, A . J . , Heeb, P.J., Feustel, P.J., Estes, J.E., Keenan, C.J. , Minnear, F . L . , Selden, L . , Giunta, C , Flor, J.R., and Blumenstock, F . A . (1993). Vascular clearance and organ uptake of G - and F-actin in the rat. A m . J. Physiol. Gastrointest. Liver Physiol. 265, 1071-1081 100 Hoffmeister, K . M . , Falet, H . , Toker, A . , Barkalow, K . L . , Stossel, T.P., and Hartwig, J .H. (2001). Mechanisms of cold-induced platelet actin assembly. J. B io l . Chem. 276, 24751— 24759 Holmes, K . C , Popp, D. , Gebhard, W. , and Kabsch, W . (1990). Atomic model of the actin filament. Nature 347, 44-49 Horrocks, W . D . , Jr. (1993). Luminescence spectroscopy. Methods Enzymol. , 226, 495-538 Huff, M . E . , Page, L . J . , Balch, W . E . , and Kel ly , J.W. (2003). Gelsolin domain 2 C a 2 + affinity determines susceptibility to furin proteolysis and familial amyloidosis of finnish type. J .Mol . B i o l . 334, 119-127 Ito, H . , Yamamoto, H . , Kimura, Y . , Kambe, H . , Okochi, T. and Kishimoto, S. (1990). Affinity chromatography of human plasma gelsolin with polyphosphate compounds on immobilized Cibacron Blue F 3 G A . J. Chromatog. 526, 397^106 Janmey, P .A . , Lamb, J., Al len , P .G. , and Matsudaira, P.T. (1992). Phosphoinositide-binding peptides derived from the sequences of gelsolin and vi l l in . J. B i o l . Chem. 267, 11818-11823 Janmey, P .A . and Stossel, T.P. (1987). Modulation of gelsolin function by phosphatidylinositol 4, 5-bisphosphate. Nature, 325, 362-364 Kabsch, W. , Mannherz, H . G . , Suck, D. , Pai, E.F. , and Holmes, K . C (1990). Atomic structure of the actin:DNase I complex. Nature 347, 37-44 Kambe, H . , Ito, H . , Kimura, Y . , Okochi, T., Yamamoto H . , Hashimoto, T. and Tagawa, K . (1992). Human plasma gelsolin reversibly binds M g - A T P in C a 2 + sensitive manner. J. Biochem. (Tokyo) 111, 722-725 Kazmirski , S.L., Howard, M . J . , Isaacson, R . L . , and Fersht, A . R . (2000). Elucidating the mechanism of familial amyloidosis- Finnish type: N M R studies of human gelsolin domain 2. Proc. Natl. Acad. Sci. U S A 97, 10706-10711 Kazmirski , S., Isaacson, R . L . , A n , C , Buckle, A . , Johnson, C M . , Daggett, V . , and Fersht, A . R . (2002). Loss of a metal-binding site in gelsolin leads to familial amyloidosis-Finnish type. Nat. Struct. B i o l . 9, 112-116 K i m , C . U . , Kapfer, R., and Gruner, S . M . (2005). High pressure cooling of protein crystals without cryoprotectants. Acta Crystallogr. D B i o l Crystallogr. 61, 881-890 Kinosian, H.J . , Newman, J., Lincoln, B . , Selden, L . A . , Gershman, L . C , and Estes, J.E. (1998). Ca regulation of gelsolin activity: binding and severing o f F-actin. Biophys. J. 75,3101-3109 101 Kinosian, H.J . , Selden, L . A . , Estes, J.E., and Gershman, L . C . (1993). Nucleotide binding to actin. Cation dependence of nucleotide dissociation and exchange rates. J. B i o l . Chem. 268, 8683-8691 Kiselar, J .G., Janmey, P .A. , A lmo , S.C., and Chance, M . R . (2003). Visualizing the C a 2 + -dependent activation of gelsolin by using synchrotron footprinting. Proc. Natl. Acad. Sci. U S A 100, 3942-3947 Kisselev, L . L . , Justesenc, J., Wolfson, A . D . , and Frolova, L . Y u . (1998). Diadenosine oligophosphates ( A p n A ) , a novel class of signalling molecules? F E B S Lett. 427, 157-163 Koepf, E . K . , and Burtnick, L . D . (1992). Interaction of plasma gelsolin with tropomyosin. F E B S Lett. 309, 56-58 Koepf, E . K . , Hewitt, J., V o , H . , Macgillivray, R.T., and Burtnick, L . D . (1998). Equus caballus gelsolin - c D N A sequence and protein structural implications. Eur. J. Biochem. 251,613-621 Kothakota, S., Azuma, T., Reinhard, C , Kl ippel , A . , Tang, J., Chu, K . , McGarry T.J., Kirschner, M . W . , Koths, K . , Kwiatkowski D.J . , and Will iams, L .T . (1997). Caspase-3-generated fragment of gelsolin: effector of morphological change in apoptosis. Science 278, 294 - 298 Koya, R . C , Fujita, H . , Shimizu, S., Ohtsu, M . , Takimoto, M . , Tsujimoto, Y . , and Kuzumaki, N . (2000). Gelsolin inhibits apoptosis by blocking mitochondrial membrane potential loss and cytochrome c release. J. B io l . Chem. 275, 15343-15349 Kusano, H . , Shimizu, S., Koya, R . C , Fujita, H . , Kamada, S., Kuzumaki, N . , and Tsujimoto, Y . (2000). Human gelsolin prevents apoptosis by inhibiting apoptotic mitochondrial changes via closing V D A C Oncogene 19, 4807-4814 Kwiatkowski, D.J . (1999). Functions of gelsolin: motility, signaling, apoptosis, cancer. Curr. Opin. Cel l B i o l . 11, 103-108 Kwiatkowski, D.J . , Janmey, P .A . , Mole , J.E., and Y i n , H . L . (1985). Isolation and properties of two actin-binding domains in gelsolin. J. B i o l . Chem. 260, 15232-15238 Kwiatkowski, D.J. , Janmey, P .A. , and Y i n , H . L . (1989). Identification of critical functional and regulatory domains in gelsolin. J. Cel l B i o l . 108, 1717-1726 Kwiatkowski, D.J . , Stossel, T.P., Orkin, S.H., Mole , J.E., Colten, H.R., and Y i n , H . L . (1986). Plasma and cytoplasmic gelsolins are encoded by a single gene and contain a duplicated actin-binding domain. Nature 323, 455-458 102 Lagarrigue, E. , Ternent, D. , Maciver, S.K., Fattoum, A . , Benyamin, Y . , and Roustan, C. (2003). The activation of gelsolin by low pH. The calcium latch is sensitive to calcium but not pH. Eur. J. Biochem. 270, 4105^1112 Laham, L . E . , Lamb, J .A. , Al len , P .G. , and Janmey, P .A . (1993). Selective binding of gelsolin to actin monomers containing A D P . J. B io l . Chem. 268, 14202-14207 Laham, L . E . , Way, M . , Y i n , H . L . , and Janmey, P .A . (1995). Identification of two sites in gelsolin with different sensitivities to adenine nucleotides. Eur. J. Biochem. 234, 1-7 Lamb, J .A., Al len , P .G. , Tuan, B . Y . , and Janmey, P .A . (1993). Modulation of gelsolin function. Activation at low pH overrides C a 2 + requirement. J. B i o l . Chem. 268, 8999-9004 Langbein, S., Lehmann, J., Harder, A . , Steidler, A . , Michel , M . S . , Aiken, P., and Badawi, J .K. (2006). Protein profiling of bladder cancer using the 2 D - P A G E and S E L D I - T O F - M S technique. Technol. Cancer Res. Treat. 5, 67-72 Larson, S.B., Day, J.S., Cudney, R., and McPherson, A . (2007). A novel strategy for the crystallization of proteins: X-ray diffraction validation. Acta Crystallogr. D B io l . Crystallogr. 63,310-318 Lee, P.S., Waxman, A . B . , Cotich, K . L . , Chung, S.W., Perrella, M . A . , and Stossel, T.P. (2007). Plasma gelsolin is a marker and therapeutic agent in animal sepsis. Crit. Care Med. 35, 849-855 Leslie, A . G . W . (1992). Recent changes to the M O S F L M package for processing film and image plate data. Jnt C C P 4 / E S F - E A M C B Newsl. Protein Crystallogr. 26 L i n , K . M . , Mejillano, M . , and Y i n , H . L . (2000). C a 2 + regulation of gelsolin by its C-terminal tail. J .Biol . Chem. 275, 27746-27752 L i n , K . M . , Wenegieme, E . , L u , P. J., Chen, C.S., and Y i n , H . L . (1997). Gelsolin binding to phosphatidylinositol 4,5-bisphosphate is modulated by calcium and p H . J. B i o l . Chem. 272, 20443-20450 Logan, M . R . and Mandate, C A . (2006). Regulation of the actin cytoskeleton by PIP2 in cytokinesis. B io l . Cel l 98, 377-388 Maciver, S.K., Ternent, D . , and McLaughl in P.J. (2000). Domain 2 of gelsolin binds directly to tropomyosin. F E B S Lett. 473, 71-75 Maury, C P . , Kere, J., Tolvanen, R., and de la Chapelle, A . (1990). Finnish hereditary amyloidosis is caused by a single nucleotide substitution in the gelsolin gene. F E B S Lett. 276, 75-77 103 McGough, A . , Chiu, W. , and Way, M . (1998). Determination of the gelsolin binding site on F-actin: implications for severing and capping. Biophys. J. 74, 764-772 McLaughlin, P.J., Gooch, J.T., Mannherz, H . G . , and Weeds, A . G . (1993). Structure of gelsolin segment 1-actin complex and the mechanism of filament severing. Nature 364, 685-692 McPherson, A . (1993). Effects of a microgravity environment on the crystallization of biological macromolecules. Microgravity Sci. Technol. 6, 101-109 McPherson, A . (2001). A comparison of salts for the crystallization of macromolecules. Protein Sci. 10,418-422 McPherson, A . (2004). Introduction to protein crystallization. Methods 34, 254-265 Meerschaert, K . , DeCorte, V . , DeVi l le , Y . , Vanderkerckhove, K . , and Gettemans, J. (1998). Gelsolin and functionally similar actin-binding proteins are regulated by lysophosphatidic acid E M B O J. 17, 5923-5932 Mintzer, E. , Sargsyan, PL, and Bittman R. (2006). Lysophosphatidic acid and lipopolysaccharide bind to the PIP2-binding domain of gelsolin. Biochim. Biophys. Acta 1758, 85-89 Moriyama, K . , Yonezawa, N . , Sakai, H . , Yahara, I., and Nishida, E . (1992). Mutational analysis of an actin-binding site of cofilin and characterization of chimeric proteins between cofilin and destrin. J. B i o l . Chem. 267, 7240-7244 Narayan, K . , Chumnarnsilpa, S., Choe, H . , Irobi, E. , Urosev, D. , Lindberg, U . , Schutt C. E. , Burtnick, L . D . and Robinson R . C . (2003). Activation in isolation: exposure of the actin-binding site in the C-terminal half of gelsolin does not require actin. F E B S Lett. 552, 82-85 Neidl , C . and Engel, J. (1979). Exchange o f A D P , A T P and 1: N6-ethenoadenosine 5'-triphosphate at G-actin. Equilibrium and kinetics. Eur. J. Biochem. 101, 163-169 Ono, S. (2007). Mechanism of depolymerization and severing of actin filaments and its significance in cytoskeletal dynamics. Int. Rev. Cytol . 258, 1-82 Osborn, T . M . , Dahlgren, C , Hartwig, J .H. , and Stossel, T.P. (2007).Modifications of cellular responses to lysophosphatidic acid and platelet-activating factor by plasma gelsolin. A m . J. Physiol. Cel l Physiol. 292, 1323-1330 Otterbein, L .R . , Graceffa, P. and Dominguez, R. (2001). The crystal structure of uncomplexed actin in the A D P state. Science 293, 708-711 104 Otwinowski, Z . , and Minor, W . (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, Macromolecular Crystallography, part A , 307-326 Page, L . J . , Huff, M . E . , Kel ly , J.W., and Balch, W . E . (2004). C a 2 + binding protects against gelsolin amyloidosis. Biophys. Biochem. Res. Comun. 322, 1105-1110 Page, L . J . , Suk, J .Y. , Huff, M . E . , L i m , H.J. , Venable, J., Yates, J., Kel ly , J.W., and Balch, W . E . (2005). Metalloendoprotease cleavage triggers gelsolin amyloidogenesis. E M B O J. 24,4124-4132 Percipalle, P., and Visa , N . (2006). Molecular functions of nuclear actin in transcription. J. Cell B io l . 172, 967-971 Pollard, T.D. , Blanchoin, L . , and Mull ins, R . D . (2000). Molecular mechanisms controlling actin filament dynamics in nonmuscle cells. Annu. Rev. Biophys. Biomol . Struct. 29, 545-576 Pollard, T.D. , and Borisy, G . G . (2003). Cellular motility driven by assembly and disassembly of actin filaments. Cel l 112, 453-465 Pope, B.J . , Gooch, J.T., and Weeds, A . G . (1997). Probing the effects of calcium on gelsolin. Biochemistry 36, 15848 -15855 Pope, B . , Maciver, S., and Weeds, A . G . (1995). Localization of the calcium sensitive actin monomer binding site in gelsolin to segment 4 and identification of calcium binding sites. Biochemistry 34, 1583-1588 Pope, B . , Way, M . , and Weeds, A . G . (1991). Two of the three actin-binding domains of gelsolin bind to the same subdomain of actin. Implications of capping and severing mechanisms. F E B S Lett. 280, 70-74 Puius, Y . A . , Fedorov, E . V . , Eichinger, L . , Schleicher, M . , and A l m o , S.C. (2000). Mapping the functional surface of domain 2 in the gelsolin superfamily. Biochemistry 39, 5322-5331 Qiao, H . , and M c M i l l a n , J.R. (2007). Gelsolin segment 5 inhibits HIV-induced T-cell apoptosis via Vpr-binding to V D A C . F E B S Lett. 581,535-540 Rao, J., Seligson, D. , Visapaa, H . , Horvath, S., Eeva, M . , Michel , K . , Pantuck, A . , Belldegrun, A . , and Palotie, A . (2002). Tissue microarray analysis of cytoskeletal actin-associated biomarkers gelsolin and E-cadherin in urothelial carcinoma. Cancer 95, 1247-1257 105 Robinson, R . C , Choe, S. and Burtick, L . D . (2001). The disintegration of a molecule: The role of gelsolin in F A F , familial amyloidosis (Finnish type). Proc. Natl . Acad. Sci . U S A 98,2117-2118 Robinson, R . C , Mejillano, M . , Le, V . P . , Burtnick, L . D . , Y i n , H . L . and Choe, S. (1999). Domain movement in gelsolin: a calcium-activated switch. Science 286, 1939-1942 Rould, M . A . , Wan, Q., Joel, P .B. , Lowey, S. and Trybus K . M . (2006). Crystal structures of expressed non-polymerizable monomeric actin in the A D P and A T P states. J. B i o l . Chem. 281,31909-31919 Sakurai, N . , and Utsumi, T. (2006). Posttranslational N-myristoylation is required for the anti-apoptotic activity of human tGelsolin, the C-terminal caspase cleavage product of human gelsolin. J. B io l . Chem. 281, 14288-14295 Schoepper, B . and Wegner, A . (1991). Rate constants and equilibrium constants for binding of actin to the 1: 1 gelsolin-actin complex. Eur. J. Biochem. 202, 1127-1131 Schoepper, B . and Wegner, A . (1992). Gelsolin binds to polymeric actin at a low rate. J. B io l . Chem. 267, 13924-13927 Schutt, C .E . , Mysl ik , J . C , Rozycki , M . D . , Goonesekere, N.C.and Lindberg, U . (1993). The structure of crystalline profilin-P-actin. Nature 365, 810-816 Selden, L . A . , Kinosian, H.J . , Newman, J., Lincoln, B . , Hurwitz, C , Gershman, L . C , and Estes, J.E. (1998). Severing of F-actin by the amino-terminal half of gelsolin suggests internal cooperativity in gelsolin. Biophys. J. 75, 3092-3100 Selve, N . , and Wegner, A . (1986). Rate constants and equilibrium constants for binding of the gelsolin-actin complex to the barbed ends of actin filaments in the presence and absence of calcium. Eur. J. Biochem. 160, 379-387 Shieh, D . B . , Chen, I.W., Wei , T . Y . , Shao, C . Y . , Chang, H.J . , Chung, C . H . , Wong, T . Y . , and Jin, Y . T . (2006). Tissue expression of gelsolin in oral carcinogenesis progression and its clinicopathological implications. Oral Oncol. 42, 599-606 Shim, J .U. , Cristobal, G. , Link, D.R. , Thorsen, T., Jia, Y . , Piattelli, K . , and Fraden, S. (2007). Control and measurement of the phase behavior of aqueous solutions using microfluidics. J. A m . Chem. Soc. 129, 8825-8835 Silacci, P., Mazzolai, L . , Gauci, C , Stergiopulos, N . , Y i n , H . L . , and Hayoz, D. (2004). Gelsolin superfamily proteins: key regulators of cellular functions. Cell M o l . Life Sci. 61, 2614-2623 106 Simoes, A . P . , Reed, J., Schnabel, P., Camps, M . , and Gierschik, P. (1995). Characterization of putative polyphosphoinositide binding motifs from phospholipase C beta 2. Biochemistry 34, 5113-5119 Spudich, J .A. , and Watt, S. (1971). The regulation of rabbit skeletal muscle contraction. I. Biochemical studies of the interaction of the tropomyosin-troponin complex with actin and the proteolytic fragments of myosin. J. B io l . Chem. 246, 4866-4871 Steinmetz, M . O . , Stoffler, D. , Hoenger, A . , Bremer, A . , and Aebi , U . (1997). Act in: From cell biology to atomic detail. J. Sruct. B io l . 119, 295-320 Suhler, E., Lin, W., Yin , H.L. and Lee, W . M . (1997). Decreased plasma gelsolin concentrations in acute liver failure, myocardial infarction, septic shock, and myonecrosis. Crit. Care Med. 25, 594-598 Sun, H . , L i n , K . M . , and Y i n , H . L . (1997). Gelsolin modulates phospholipase C activity in vivo through phospholipid binding. J. Cel l B i o l . 138, 811-820 Sun, H.Q. , Wooten, D .C . , Janmey, P .A . , and Y i n , H . L . (1994). The actin side-binding domain of gelsolin also caps actin filaments. Implications for actin filament severing. J. B io l . Chem. 269, 9473-9479 Sun, H.Q. , Yamamoto, M . , Mejillano, M . , and Y i n , H . L . (1999). Gelsolin, a multifunctional actin regulatory protein. J. B i o l . Chem. 274, 33179-33182 Takiguchi, K . , Yamashiro-Matsumura, S. and Matsumura, F. (2000). Artificial phosphorylation removes gelsolin's dependence on calcium. Cel l Struct. Funct. 25, 57-65 Tanaka, H . , Shirkoohi, R., Nakagawa, K . , Qiao, H . , Fujita, H . , Okada, F. , Hamada, J.I., Kuzumaki, S., Takimoto, M . , and Kuzumaki, N . (2005). s i R N A gelsolin knockdown induces epithelial- mesenchymal transition with a cadherin switch in human mammary epithelial cells. Int. J. Cancer 118, 1680-1691 Taylor, G . (2003). The phase problem. Acta Crystallogr. D B i o l . Crystallogr. 59, 1881-1890 Tellam, R . L . (1986). Gelsolin inhibits nucleotide exchange from actin. Biochemistry 25, 5799-5804 Thompson, C C , Ashcroft, F.J. , Patel, S., Saraga, G. , Vimalachandran, D. , Prime, W. , Campbell, F., Dodson, A . , Jenkins, R .E . , Lemoine, N .R . , Crnogorac-Jurcevic, T., Y i n , H . L . , and Costello, E . (2007). Pancreatic cancer cells overexpress gelsolin family capping proteins which contribute to their cell motility. Gut 56, 95- 106 Thomsen, R., and Christensen, M . H . (2006). MolDock: a new technique for high-accuracy molecular docking. J. Med. Chem. 49, 3315-3321 107 Trifaro, J . M . , Rose, S.D., and Marcu, M . G . (2000). Scinderin, a Ca -dependent actin filament severing protein that controls cortical actin network dynamics during secretion. Neurochem. Res. 25, 133-144 Urosev, D. (2003). Towards a crystallographic structure for the N-terminal half of gelsolin bound to actin. M.Sc . thesis, U B C Urosev, D. , M a , Q., Tan, A . L . , Robinson, R . C , and Burtnick, L . D . (2006). The structure of gelsolin bound to A T P . J. M o l . B i o l . 357, 765-772 Vandekerckhove, J. (1990). Act in binding proteins. Curr. Opin. Cel l B i o l . 2, 41-50. Van den Abbeele, A . , De Corte, V . , Van Impe, K . , Bruyneel, E. , Boucherie, C , Bracke, M . , Vandekerckhove, J., and Gettemans, J. (2007). Downregulation of gelsolin family proteins counteracts cancer cell invasion in vitro. Cancer Lett., in press. van Holde, K . E . , Johnson, W . C , and Ho, P.S. (1998). Principals of Physical Biochemistry. Prentice Hal l , Ch.6, 242-296 Van Troys, M . , Dewitte, D. , Goethals, M . , Vandekerckhove, J. and Ampe, C. (1996). Evidence for an actin binding helix in gelsolin segment 2; have homologous sequences in segments 1 and 2 of gelsolin evolved to divergent actin binding functions? F E B S Lett. 397, 191-196 Vartanian, A . A . (2003). Gelsolin and plasminogen activator inhibitor-1 are Ap3A-binding proteins. Ital. J. Biochem. 52, 9-16 Vartanian, A . , Prudovsky, I., Suzuki, H . , Dal Pra, I., and Kisselev, L . (1997). Opposite effects of cell differentiation and apoptosis on AP3A/AP4A ratio in human cell cultures. F E B S Lett. 415, 160-162 Vasconcellos, C . A . , and Lind, S.E. (1993). Coordinated inhibition of actin-induced platelet aggregation by plasma gelsolin and vitamin D-binding protein. Blood 82, 3648-3657 Vouyiouklis, D . A . , and Brophy, P.J. (1997). A novel gelsolin isoform expressed by oligodendrocytes in the central nervous system. J. Neurochem. 69, 995-1005 Way, M . , Gooch, J., Pope, B . , and Weeds, A . G . (1989). Expression of human plasma gelsolin in Escherichia coli and dissection of actin binding sites by segmental deletion mutagenesis. J. Cel l B i o l . 109, 593- 605 Way, M . , Pope, B . , and Weeds, A . G . (1992). Evidence for functional homology in the F-actin binding domains of gelsolin and alpha-actinin: implications for the requirements of severing and capping. J. Cel l B i o l . 119, 835-842 108 Weeds, A . G . , Gooch, J., McLaughlin, P., Pope, B . , Bengtsdotter, M . , and Karlsson, R. (1995). Identification of the trapped calcium in the gelsolin segment 1-actin complex: implications for the role of calcium in the control of gelsolin activity. F E B S Lett. 360, 227-230 Weeds, A . G . , Gooch, J., Pope, B . , and Harris, H .E . (1986). Preparation and characterization of pig plasma and platelet gelsolins. Eur. J. Biochem. 161, 69-76 Witke, W. , Sharpe, A . H . , Hartwig, J .H. , Azuma, T., Stossel, T.P., and Kwiatkowski, D.J . (1995). Hemostatic, inflammatory, and fibroblast responses are blunted in mice lacking gelsolin. Cel l 81, 41-51 Yamamoto, H . , Ito, H . , Nakamura, H . , Hayashi, E. , Kishimoto, S., Hashimoto, T. and Tagawa, K . (1990). Human plasma gelsolin binds adenosine triphosphate. J. Biochem. (Tokyo), 108, 505-506 Yamamoto, H . , Terabayashi, M . , Egawa, T., Hayashi, E. , Nakamura, H . and Kishimoto, S. (1989). Affinity separation of human plasma gelsolin on Aff i -Gel Blue. J. Biochem. (Tokyo), 105, 799-802 Yang, J., Ramnath, N . , Moysich, K . B . , Asch, H . L . , Swede, H . , Al rawi , S.J., Huberman, J., Geradts, J., Brooks, J.S., and Tan, D . (2006). Prognostic significance of M C M 2 , Ki-67 and gelsolin in non-small cell lung cancer. B M C Cancer 6, 203 Y i n , H . L . , Kwiatkowski, D.J . , Mole , J.E., and Cole, F.S. (1984). Structure and biosynthesis of cytoplasmic and secreted variants of gelsolin. J. B i o l . Chem. 259, 5271-5276 Y i n , H . L . and Stossel, T.P. (1979). Control of cytoplasmic actin gel-sol transformation by gelsolin, a calcium-dependent regulatory protein. Nature 281, 583-586 Y u , F . X . , Sun, H.Q. , Janmey, P .A. , and Y i n , H . L . (1992). Identification of a polyphosphoinositide binding sequence in an actin monomer-binding domain of gelsolin. J. B io l . Chem. 267, 14616-14621 Zapun, A . , Grammatyka, S., Deral, G . , and Vernet, T. (2000). Calcium-dependent conformational stability of modules 1 and 2 of human gelsolin. Biochem. J. 350, 873-881 Zhang, Y . , Vorobiev, S. M . , Gibson, B . G . , Hao, B . , Sidhu, G . S., Mishra, V . S., Yarmola, E . G. , Bubb, M . R . , A lmo , S.C,. and Southwick, F. S (2006). A CapG gain-of-function mutant reveals critical structural and functional determinants for actin filament severing. E M B O J. 25, 4458-4467 Zheng, X . , Diraviyam, K , and Sept, D. (2007). Nucleotide effects on the structure and dynamics of actin. Biophys. J., in press. 109 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items