Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards a crystallographic structure for the N-terminal half of gelsolin bound to actin Urosev, Dunja 2003

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

Item Metadata


831-ubc_2003-0453.pdf [ 3.43MB ]
JSON: 831-1.0060212.json
JSON-LD: 831-1.0060212-ld.json
RDF/XML (Pretty): 831-1.0060212-rdf.xml
RDF/JSON: 831-1.0060212-rdf.json
Turtle: 831-1.0060212-turtle.txt
N-Triples: 831-1.0060212-rdf-ntriples.txt
Original Record: 831-1.0060212-source.json
Full Text

Full Text

TOWARDS A C R Y S T A L L O G R A P H I C STRUCTURE FOR THE N - T E R M I N A L H A L F OF GELSOLIN B O U N D TO A C T I N By DUNJA UROSEV M . S c , Moscow State University named after M . V . Lomonosov, 1999  A THESIS SUBMITTED IN PARTIAL F U L F I L M E N T OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Chemistry) We accept this thesis as conforming to the required standard  THE U N I V E R S f i Y ^ F B R j f l S H July 2003 © DunjaUrosev, 2003  COLUMBIA  In  presenting  degree freely  at  this  the  available  copying  of  department publication  of  in  partial  fulfilment  University  of  British  Columbia,  f o r reference  this or  thesis  thesis by  this  a n d study.  f o r scholarly  his thesis  or for  her  of  financial  &H(fr1<STKY  T h e U n i v e r s i t y o f British Vancouver, Canada  Date  DE-6  (2/88)  I  I further  purposes  Columbia  gain  the agree  shall  requirements that  agree  may be  representatives.  permission.  Department  of  It  is  that  the  Library  permission  granted  by  understood  not be  for  allowed  an  advanced  shall for  the that  without  make  it  extensive  head  of  my  copying  or  my  written  Abstract Gelsolin belongs to a family of proteins that participates in the reorganization of cytoskeletal actin, a shared requirement of such processes as cell movement, cell division and apoptosis. Its functions include actin filament nucleation, severing and capping. Gelsolin consists of six structurally analogous domains (G1-G6) that, in the Ca -free, 2+  inactive state, are packed in such way that none of the actin-binding sites are accessible. Binding of Ca  2+  ions to gelsolin causes large changes in the relative positions and  orientations of the six domains, resulting in exposure of several actin-binding surfaces, and subsequent binding and severing of actin filaments. Filament end-binding sites have been identified previously in domains Gl and G4, while a distinctfilamentside-binding site has been attributed to G2. This thesis describes protein preparation and crystallization experiments that led to the structure at 3 A resolution of the N-terminal half of gelsolin (G1-G3) bound to one actin molecule in the presence of Ca  2+  ions. The structure reveals for the first time the  details of how G2 and G3 interact with the same actin to which Gl is attached. As expected, the changes in the relative orientation of domains within the N-terminal half of gelsolin on activation are large. Previously unidentified contacts between G3 and actin are observed. The existence of a Ca  2+  in the type-2 binding site in activated G3, which  was inferred by sequence comparison within gelsolin domains, is confirmed. In addition, docking the structure into existing molecular models for an actin filament permits proposal of a self-consistent mechanism for how intact gelsolin is activated, binds to the side of an actin filament, severs and then caps one of the newly cut filament ends.  ii  Table of contents Abstract  ii  Table of contents  iii  List of abbreviations  v  List of figures  vii  Acknowledgments  ix  CHAPTER I Introduction  1  A. Actin  1  1. Actin monomer  1  2. Actin filaments  2  B. Gelsolin  5  1. Gelsolin characteristics and properties  5  2. Gelsolin structure  8 13  3. Actin and Ca binding sites in gelsolin 2+  4. Changes in gelsolin on activation by Ca  2+  CHAPTER II Materials and Methods A. Protein purification  20  23 23  1. Preparation of horse plasma gelsolin  23  2. Actin purification  24  3. Formation and purification of the complex of one gelsolin with two actins  25  B. X-ray crystallography  26  1. Introduction  26  2. Crystallization of GA  2  32  3. Data collection  .'  33  4. Protein structure figure preparation  34  CHAPTER III Results and Discussion A. Purification of gelsolin  35 35  iii  B. Purification of the G A 2 complex  36  C. Structure of the N-terminal half of gelsolin bound to actin  36  1. Gl-G3/actin  36  2. Changes in the relative orientations of domains within G 1 - G 3 upon activation.. 40 3. Structural changes within domains caused by binding Ca 4.  Comparison of the activation processes in G 1 - G 3 and  5 . Further insight into gelsolin function  and actin  G4-G6  45 47  48  D. Summary  50  51  Bibliography  IV  List of abbreviations ABPs  Actin binding proteins  ADF  Actin deploymerizing factor  ADP  Adenosine diphosphate  Amino acids: Ala  alanine  Arg  arginine  Asn  asparagine  Asp  aspartic acid  Cys  cysteine  Glu  glutamic acid  Gly  glycine  His  histidine  Lys  lysine  Met  methionine  Pro  proline  Tyr  tyrosine  Val  valine  ATP  Adenosine triphosphate  AUFS  Absorbance units full scale  Buffer A  2 m M Tris-HCI, 0.2 m M CaCl , 0.2 m M ATP, 1 m M DTT, pH 7.6  Buffer 1  25 m M Tris-HCI, pH 7.8, 1 m M EDTA, 0.1 m M N a N  Buffer 2  25 m M Tris-HCI, pH 8.0, 1 m M E D T A  DEAE  2-diethylamino-ethyl functional group  DMSO  Dimethyl sulfoxide  DNase 1  Deoxyribonuclease I  DTT  Dithiothreitol  EDTA  Ethylene diamine tetraacetic acid  F-actin  Filamentous actin  FAF  Familial amyloidosis (Finnish type)  2  N  v  3  GA  2  complex of one gelsolin with 2 actin molecules  G-actin  Monomeric actin  G1 +  Gelsolin residues 25-160  G1-G6  gelsolin domains 1 through 6  HI  long helix present in each gelsolin domain  HPLC  High performance liquid chromatography  LPA  Lysophosphatidic acid  NMR  Nuclear magnetic resonance  PAGE  Polyacrylamide gel electrophoresis  PDB  Protein data bank  PEG  Polyethyleneglycol  PIP  Phosphatidylinositol 4,5 bisphosphate  2  PPIs  Polyphosphoinositides  RNA, mRNA  Ribonucleic acid, messenger ribonucleic acid  SDS  Sodium dodecyl sulphate  Tris  2-amino-2-(hydroxymethyl)-l,3-propanediol  v/v  volume/volume  w/v  weight/volume  vi  List of figures Figure 1. A) Atomic structure (resolved to 1.4 A) of an actin monomer with ADP bound at the center of the molecule and Ca bound at four 2+  sites and B) Structure of an actin monomer in a complex with DNase 1  2  Figure 2. Schematic representation of F-actin  3  Figure 3. Holmes model for F-actin  3  Figure 4. A ribbon representation of the structure of equine plasma gelsolin determined in the absence of bound calcium ions  9  Figure 5. A topological diagram following the polypeptide chain of equine plasma gelsolin through the structure of the Ca -free form  10  Figure 6. Identification of (3-strands and oc-helices in gelsolin domain G5  10  Figure 7. Individual domains excised from the Ca -free structure of equine 2+  plasma gelsolin  11  Figure 8. The two halves of gelsolin excised from the Ca -free 2+  structure and oriented to highlight the similar packing pattern of the component domains  12  Figure 9. Location of type-1 ( I ) and type-2 (II) Ca -binding sites in 2+  the Ca -free conformation of gelsolin 2+  14  Figure 10. A) An actin monomer in complex with gelsolin domain Gl and B) An actin monomer in complex with gelsolin domains G4-G6 Figure 11. Type-1 Ca -binding sites in Gl and G4 2+  14 15  Figure 12. Type-2 metal ion-binding sites in G l , G2, G4, G5, G6 and domain 2 of severin  16  Figure 13. Model of the complex of F-actin and gelsolin domains 1 and 2  17  Figure 14. 2.6 A resolution structure of a complex formed between actin and Gl with an extended linker from the start of G2  vii  18  Figure 15. G3 and G6 in the structure of inactive gelsolin act as latches to sterically hinder the approach of actin to the long helices of Gl and G4, respectively  19  Figure 16. Transition from G4-6 in the absence of Ca to its active 2+  conformation bound to actin  20  Figure 17. G4-G6 excised from the structure of Ca -free gelsolin and 2+  from the structure of G4-G6 bound to actin in the presence of Ca  2+  Figure 18. The fourteen Bravais lattices  21 27  Figure 19. Schematic representation of two-dimensional crystallization phase diagram  28  Figure 20. Typical diffraction pattern  30  Figure 21. Peptide sequence and Ca ion fitted into the corresponding part 2+  of an electron density map  31  Figure 22. Hanging drop crystallization set up  32  Figure 23. Elution of GA from a Bio-Rad Sephacryl S300 size exclusion column 36 2  Figure 24. SDS-PAGE analysis of crystals grown in the identical conditions to those subjected to diffraction experiments  38  Figure 25. Structure at 3.0 A resolution of G1-G3 bound to actin in the presence of Ca ions  39  2+  Figure 26. Comparison of the conformation of inactive Gl -G3 (A) with that in the presence of Ca and actin (B)  41  2+  Figure 27. Space filling representation of the Gl-G3/actin complex  42  Figure 28. Model for the capped barbed-end of F-actin  43  2_|_  Figure 29. A) Type-2 Ca -binding site observed in Gl in the context of G1-G3. B) Peptide backbone conformation around the predicted type-2 site in G2 in the context of Gl -G3. C) Ca bound at the 2+  type-2 site in G3, observed for the first time Figure 30. Actin contacts with the G1-G2 linker  44 45  Figure 31. Rearrangement of the kinked long helix of G3 as a result of binding Ca by G1-G3 2+  Figure 32. Comparison of the activated structures of the two halves of gelsolin  vm  46 48  Acknowledgments I would like to thank my supervisor, Dr. Les Burtnick, for the invaluable guidance throughout the thesis work. For the crystallographic data analysis I thank Drs. Les Burtnick and Robert Robinson. I am also very grateful to Dr. Eve Teh for occasional gel electrophoresis work. I appreciate very much the support I received from my family, friends and colleagues.  ix  )  CHAPTER I Introduction  A. Actin 1. Actin monomer  Actin is a very abundant protein, present in all eukaryotic cells and comprising as much as 20% of the total protein content in certain muscle cells. It is one of the most highly conserved proteins in the cell, and is expressed in three isoforms (reviewed by dos Remedios et al. 2003). The soluble, monomeric form of actin, G-actin, is a globular 42kDa protein comprised of four subdomains (numbered 1,2,3 and 4 in Figure 1). It binds one nucleotide molecule, usually A T P or A D P , in a hydrophilic pocket between subdomains  1 and 3. Phosphate  groups of the bound nucleotide participate in  coordinating a divalent metal ion (generally either M g  2 +  or Ca ) in a high affinity metal 2+  ion-binding site (Figure 1). In addition, lower affinity metal ion-binding sites might have a role in inducing conformational changes that facilitate polymerization of actin. Due to its tendency toward spontaneous self-assembly in solvent conditions generally employed to crystallize proteins, the crystallographic structure of G-actin up until recently (Otterbein et al. 2001) had been reported only as deduced from a complex with an actin-binding protein [e.g., DNase I (Kabsch et al. 1990) (Figure IB), profilin (Schutt et al. 1993), gelsolin domain G l (McLaughlin et al. 1993), and gelsolin domains G4-G6 (Robinson et al. 1999, Choe et al. 2002); reviewed by dos Remedios et al. 2003]. Otterbein et al. (2001) succeeded in crystallizing isolated G-actin bound to A D P after modification at Cys374 to block oligomerization. The overall dimensions of the monomer are 55 x 55 x 35 A .  1  B  A  Figure 1. A) Atomic structure (resolved to 1.4 A) of an actin monomer with ADP bound at the center of the molecule and C a  2 +  bound at four sites (represented as red spheres) (Otterbein et al. 2001). B)  Structure of an actin monomer (gray) in a complex with DNase I (red) (Kabsch et al. 1990) [from dos Remedios et al. 2003]. The most significant distinction between the isolated actin monomer and the one in the complex is a structured a-helix in subdomain 2 replaces the DNase I-binding loop.  2. Actin filaments  The biological functions of actin are linked to its ability to polymerize through non-covalent interactions into an extended filamentous form (F-actin). F-actin filaments are linear polymers, which can be considered as two staggered parallel rows of actins twisted into a right-handed helix having a diameter of 7-8 nm (reviewed by dos Remedios et al. 2003) (Figure 2). There are a total of thirteen monomer units in the two strands between crossovers within the helix and the distance between crossovers (the half-pitch of the helix) is 36 nm (Holmes at al. 1990). At one end of the filament, preferential assembly takes place (barbed end) and at the other, preferential disassembly occurs (pointed end). Thus, filamentous actin is a polar structure, which turns out to be crucial for muscle contraction and cell motility.  2  Figure 2. Schematic representation of F-actin: two staggered rows of actins twisted into a righthanded helix. Each actin monomer is represented as a sphere. (  A crystallographic structure for F-actin, the more important functional form of actin, does not yet exist. The most widely accepted model for the structure of F-actin was constructed (Holmes et al. 1990) by fitting individual monomer molecules, with appropriate rotation of the subdomains, into a low resolution F-actin model determined using electron microscopic data from ordered arrays of filaments (Figure 3). F-actin possesses a certain degree of flexibility, which probably plays a role in allowing different actin-binding proteins to make contact and effect their functions.  Figure 3. Holmes model for F-actin (Holmes et al. 1990). Actin monomers colored blue and yellow belong to one strand and those represented in cyan and red belong to the other.  3  The polymerization of actin is a spontaneous, fully reversible process (reviewed by dos Remedios et al. 2003). There are four stages in the polymerization process: activation, nucleation, elongation and filament annealing. The average steady state filament in physiological conditions in vitro is approximately 3 um long. In vivo, actin filaments are in a continuous state of assembly/disassembly, so there is always a small concentration of monomers in thefilamentpopulation and its proportion depends on the solvent conditions and presence of actin-binding proteins (ABPs). In the absence of ABPs, the extent of polymerization is determined by ionic strength, pH and temperature. Specifically, elevated temperatures (up to 40 °C), high KCl (more than 50 mM), high Mg  2+  concentrations (above 0.5 mM), and slightly acidic pH values (6 - 6.5), all promote  actin polymerization. Hydrolysis of ATP, which occurs within each monomer during Factin formation, is not required for the polymerization process, but is essential for the proper function of F-actin. Most likely this is related to the stability of the formed filament (reviewed by dos Remedios et al. 2003). The subunits within the polymer are held together in a non-covalent manner, predominantly by hydrophobic interactions. The interactions are stronger between monomers arranged longitudinally within the same strand in the two-start helical representation of actin, than between actins next to each other in different strands. Along with actin filaments being the principal component of thinfilamentsin muscle, they also form two and three-dimensional networks that are an integral part of the cellular cytoskeleton. The actin cytoskeleton is very dynamic and is remodeled and rearranged in response to a variety of signals that are relayed through actin-binding proteins. Such rearrangement is necessary to carry out a variety of cellular processes, such as cell movement, endocytosis, exocytosis, cell division, regulation of ion transport, apoptosis, and many others (reviewed by dos Remedios et al. 2003). The number of ABPs is very large and versatile. It includes G- and F-actin binding proteins, filament severing, capping, depolymerizing and crosslinking proteins, motor proteins that use actin filaments as rails, etc. (reviewed by dos Remedios et al. 2003)  4  B. Gelsolin 1. Gelsolin characteristics and properties Gelsolin gives its name to one family of ABPs found in eukaryotes. These proteins control actin filament assembly and disassembly, thus regulating the architecture and motility of cells. While gelsolin was found not to be essential for organism survival (reviewed by Sun et al. 1999), most likely due to the expression of other gelsolin family proteins that perform similar functions, rapid cell movement of certain dynamic cells specifically requires the presence of gelsolin (reviewed by Sun et al. 1999). Yin and Stossel first discovered the protein in rabbit lung macrophages in 1979 (Yin and Stossel 1979). In vitro, gelsolin activities include actinfilamentsevering, capping and nucleation. Nucleation manifests itself in acceleration of the initial rate of salt-induced actin polymerization from a pool of actin monomers. Severing involves disrupting the noncovalent interactions between actin units in an actin filament, cutting thefilamentin two. It proceeds through lateral binding of gelsolin to an actin filament, with subsequent intercalation between actin filament subunits. Once severing has occurred, gelsolin remains bound at the newly generated barbed end to produce a cappedfilamentto which monomers can no longer add and with which the pointed ends of other filaments can no longer anneal. Once bound to F-actin, gelsolin severs with close to 100% efficiency (reviewed by Sun et al. 1999). There are two forms of gelsolin, cytoplasmic and secreted, which are structurally and functionally very similar. A single gelsolin gene encodes both forms. With alternative transcription initiation and selective RNA processing, two distinct mRNA messages that code for the different forms are produced. Both protein isoforms consist of a single polypeptide chain. Secreted gelsolin, exemplified by plasma gelsolin, a 755amino acid protein, approximately 83 kDa in molar mass, differs from its cytoplasmic counterpart in possessing a 25-residue amino-termmal extension and a single disulfide bond between residues  Cysl88  and Cys201 (Kwiatkowski et al. 1986; Koepf et al. 1998).  The role of the N-terminal extension is not known. The disulfide bond forms in the oxidizing conditions found extracellularly and may contribute to the stability of the  5  second domain of gelsolin in its activated conformation (Burtnick et al. 1997; Zapun et al. 2000). Cytoplasmic gelsolin, together with other actin regulating proteins, enables efficient actin filament assembly/disassembly. After severing, gelsolin remains attached to the barbed end of the filament as a cap that blocks reannealing of filamentous fragments. Prevention of rapid growth at barbed ends, while disassembly proceeds at pointed ends, results in the rapid net filament disassembly. Dissociation of gelsolin from filaments capped in this way, e.g., after exposure to certain polyphosphoinositides, can generate numerous polymerization-competent ends in a short time. These act as nuclei for rapid growth of new filaments. In this way, gelsolin may facilitate rapid building of cytoskeleton in a new direction (reviewed by dos Remedios et al. 2003). Plasma gelsolin plays an important role in the blood stream, clearing the circulation of potentially dangerous actin filaments. When actin is introduced into blood plasma, upon cell injury or death, the ionic conditions of plasma favor formation of elongated actinfilaments.The presence of such polymers, in the absence of an actin scavenger  system,  would cause an increase  in blood viscosity  and impede  microcirculation. By performing its severing activity, gelsolin breaks filaments into oligomers that have negligible viscosimetric effect. Subsequently, the second protein of this scavenger system, plasma vitamin D-binding protein, associates with monomers dissociating from the newly formed gelsolin-capped oligomers, forming stoichiometric complexes that are cleared from the circulation by the liver (Herrmannsdoerfer et al. 1993). The actin binding and severing functions of gelsolin are stimulated by micromolar concentrations of Ca  2+  (reviewed by Kwiatkowski 1999; Sun et al. 1999). Certain  phospholipids, such as lysophosphatidic acid (LPA) and polyphosphoinositides (PPIs), inhibit gelsolin's severing ability and cause dissociation of gelsolin molecules from the capped ends of actinfilaments(reviewed by Kwiatkowski 1999; Sun et al. 1999). PPI-gelsolin interactions may not only be involved in the regulation of gelsolin, but may modulate lipid signaling events that constitute an important part of the signaling system in cells (reviewed by Kwiatkowski 1999).  6  Other findings infer cytoplasmic gelsolin's involvement in processes such as apoptosis (Kofhakota et al. 1997), cancer and ion channel regulation (reviewed by Kwiatkowski 1999). In apoptotic cells, gelsolin is cleaved by caspase-3, the core effector protease activated in mitochondrial apoptotic pathways, after which gelsolin no longer requires Ca  2+  to sever actin filaments. This causes uncontrolled gelsolin actions and  results in dismantling of the membrane cytoskeleton, leading to cell death (Kothakota et al. 1997). Gelsolin expression level is a marker for cell malignancies. It is down regulated in the majority of tumors during carcinogenesis (reviewed by Kwiatkowski 1999), meaning that potential regulatory pathways to cause cell death could not be executable through gelsolin. However, the exact relation between gelsolin, apoptosis and cancer has yet to be clarified, especially as some of the data in the literature appears contradictory. Beside actin, gelsolin also binds tropomyosin (Koepf and Burtnick 1992). Tropomyosin binds laterally to actin filaments and in muscle cells plays an important part in the regulation of actomyosin contraction. Binding of tropomyosin to gelsolin inhibits gelsolin's severing activity, possibly because of the proximity of the binding sites for these two proteins on gelsolin, which are both located in gelsolin's second domain (Maciver et al. 2000). Thus, in certain cell conditions, tropomyosin might take on a role as a regulator of gelsolin. Specifically, this kind of regulation might be possible in cancer cells, since gene expression alterations for both proteins were correlated with oncogenic change (Maciver et al. 2000). A point mutation in gelsolin was found to be the cause of familial amyloidosis (Finnish type), or FAF, which manifests itself by neurological, ophthalmologic and dermatological symptoms (Maury et al. 1990). In this disorder, residue Aspl87 in the second domain of gelsolin is replaced by Asn or Tyr, which results in exposure of a novel proteolytic site, the first in a cascade of proteolytic actions that lead to the formation of an amyloid-generating fragment (gelsolin residues 173-243), that deposits in the form of amyloid plaques in tissues of FAF sufferers (reviewed by Kwiatkowski 1999). This mutation affects only the secreted, but not the cytoplasmic form (Kangas et al. 1996). Eukaryotic organisms possess another group of ABPs that sever actin filaments, but do not cap them, and play an important role in the increased rates of both polymerization and depolymerization. These are single domain proteins (15-19 kDa) that  7  belong to the ADF/cofilin family. Even though sequence analogy is extremely low between this family and gelsolin one, the overall fold of the single domain of ADF/cofilin proteins closely resembles that displayed by each of the six domains of gelsolin. Phosphorylation and PPI binding both prevent interaction of ADF/cofilin family members with actin (reviewed by dos Remedios et al. 2003). The gelsolin and ADF/cofilin families most likely evolved from a common ancestor that early on in evolution diverged to create two differently regulated classes of actin severing proteins (Choe et al. 2002).  2. G e l s o l i n s t r u c t u r e  Among other members of the same protein family, such as adseverin, advilin, human flightless homologue I and supervillin, gelsolin is the most potent actin filament severing protein (reviewed by Sun et al. 1999; Lin et al. 2000). Such functional ability arises from its complex structure. Gelsolin is comprised of six structurally similar domains, Gl through G6, each consisting of approximately 120-130 amino acid residues. In the absence of Ca  2+  ions, the domains pack into a compact globular arrangement  (Figure 4) (Burtnick et al. 1997). Each domain consists of a 5 or 6 stranded (3-sheet at its core, each strand identified by a letter A through E, sandwiched between a 3.5-4.5 turn a-helix (HI) that runs approximately parallel to the strands in the sheet and a 1-2 turn ochelix (H2) roughly perpendicular to the (3-sheet (Figures 4-7). Comparison of the structures presented, first, within either column and, then, across the columns of Figure 7 clearly supports the theory that gelsolin arose as a result of triplication of the gene for a primordial single-domain protein, followed by duplication of the gene for the resultant three-domain protein. Evidence of the latter gene duplication event is clear when the grouping of domains from the two distinct halves of Ca -free gelsolin, G1-G3 and G42+  G6, are compared. (Figures 5 and 8) (Burtnick et al. 1997).  8  Figure 4. A ribbon representation of the structure of equine plasma gelsolin determined in the absence of bound calcium ions (Burtnick et al. 1997). The six domains of gelsolin are colored, respectively: G l , red; G2, light green, G3, yellow; G4, pink; G5, dark green; G6, orange. This color scheme will be used in several future figures. The overall molecular dimensions are approximately 85 x 55 x 35 A.  9  Figure 5. A topological diagram following the polypeptide chain of equine plasma gelsolin through the structure of the Ca -free form (Burtnick et al. 1997). 2+  Figure 6. Identification of (3-strands and a-helices in gelsolin domain G5 (Choe et al. 2002).  10  Figure 7. Individual domains excised from the Ca~ -free structure of equine plasma gelsolin +  (Burtnick etal. 1997).  1 1  Figure 8. The two halves of gelsolin excised from the Ca'^-free structure and oriented to highlight the similar packing pattern of the component domains (Burtnick et al. 1997).  Gelsolin possesses a C-terminal extension (following the G6 domain) that terminates in a short a-helix (Figures 4 and 5). This particular structural attribute makes gelsolin unique among the members of its family. This helix has a vital role in regulation of gelsolin's actin-binding activity (Burtnick et al. 1997; Choe et al. 2002; Lin et al. 2000). It, by reaching back from the C-terminus of the molecule to interact with the long helix of G2, constitutes part of the most obvious of three latches that lock inactive gelsolin into a conformation in which none of its actin binding sites are accessible. The pairs of domains in each row of Figure 7 possess some unique structural features that influence their different functions. Gl and G4 possess extended C and D strands, with an intervening C-D loop. These features have been implicated in severing activity (Burtnick et al. 1997). G2 and G5 are identified by an additional strand, A', in their core p-sheets, and a long A - A loop. F-actin and PIP2-binding sites are located in this region of G2, but have not been identified in G5, which, over the course of evolution, has lost these sites. A possible explanation for such an outcome is that surplus F-actin  12  binding (to G5) would compromise the efficiency of the PIP2 regulatory mechanism in that two, rather than one, PIP2 would be necessary to release one gelsolin molecule from an actin filament (Choe et al. 2002). A structural distinction of G3 and G6 is a kinked HI helix. These kinks in G3 and G6 avoid clashes with HI of Gl and G4, respectively, in the Ca -free form of gelsolin (Burtnick et al. 1997). 2+  3. Actin and C a  2+  binding sites in gelsolin  Proteolysis and deletion mutagenesis studies on gelsolin identified actin-binding domains in the protein (Kwiatkowski et al. 1985; Bryan and Hwo 1986; Way et al. 1989). Two of these domains, Gl and G4, are called G-actin or filament end-binding domains. The third, G2, is called an F-actin or filament side-binding domain. Full activation of gelsolin is achieved upon binding Ca , but there is controversy as to how many Ca ions 2+  2+  are involved, somewhere between three (Pope et al. 1997, Lin et al. 2000) and eight (Choe et al. 2002) (Figure 9). Unfortunately, a structure of the fully activated form of gelsolin, either in isolation or in a complex with actin, is not yet available. To infer what the mechanism of activation entails, researchers have explored structures of activated fragments of gelsolin, both in isolation and bound to actin [Gl/actin (McLaughlin et al. 1993), G4-G-6/actin (Robinson et al. 1999), activated G4-G6 (Narayan et al. 2003; Kolappan et al. 2003), Glplus/actin (Irobi et al. 2003)] (Figure 10).  13  Figure 9. Location of type-1 (I) and type-2 (II) Ca~ -binding sites (gold spheres and gray spheres, +  respectively) in the Ca -free conformation of gelsolin (Choe et al. 2002). 2+  A  B  Figure 10. A) An actin monomer (gray) in complex with gelsolin domain G l (red) (McLaughlin et al. 1993); B) An actin monomer (gray) in complex with gelsolin domains G4-6 (Choe et al. 2002) [from dos Remedios et al. 2003],  14  It is clear that there are several Ca -binding sites on gelsolin. These can be categorized into two groups (Robinson et al. 1999). A type-1 site involves a Ca  2+  at an  interface between a gelsolin domain and actin. Gl and G4 display such sites (Figure 11). These sites directly moderate the strength of interaction between actin and gelsolin. However, it is not required that they befilledin order for gelsolin to achieve an activated conformation (Kolappan et al. 2003; Narayan et al. 2003).  Figure 11. Type-1 Ca" -binding sites in G l and G4. Ca +  (shown as gold spheres) is coordinated  by a conserved Asp residue (Asp 109 and Asp487, respectively) and two carbonyl oxygen atoms further along the polypeptide chain, and Glul67 of actin (Choe et al. 2002).  Type-2 sites are wholly contained within gelsolin. Occupied type-2 sites have been identified (Figure 12) in Gl (McLaughlin et al. 1993; Irobi et al. 2003), in each of G4, G5 and G6 (Choe et al. 2002; Narayan et al. 2003), and in isolated G2 (Kazmirski et al. 2000). In addition, sequence alignment infers a type-2 site to exist in G3 (Choe et al. 2002). In total, then, six type-2 Ca -binding sites, one associated with each of the six 2+  domains of gelsolin, have been identified or predicted. These sites are thought to regulate shifting and repositioning of domains relative to each other during the activation process in order to fully expose the actin-binding sites on G l , G2 and G4.  15  Figure 12. Type-2 metal ion-binding sites in G l , G2, G4, G5, G6 and domain 2 of severin (D2). Ca  2 +  ions are shown as gray spheres in G l , G4, G5 and G6. The gray sphere in G2 represents a C d  2 +  ion.  The green sphere in severin D2 is a Na" ion (from Choe et al. 2002).  G2 possesses a distinct F-actin binding site. Unlike G l and G4, the interaction between G2 and actin is not mediated by a type-1 C a . 2+  Rather, Arg221 replaces the  Ca -coordinating Asp from HI that is characteristic of a type-1 site (Figure 11). The 2+  positively charged side chain of Arg221 may occupy in activated gelsolin a position analogous to that occupied by the type-1 C a  2+  of G l or G4, and still permit G2 to interact  with Glul67 of actin, enabling G2 to bind to the same region on actin that interacts with G l or G4 (Choe et al. 2002). In agreement with this proposal, residues of the long a-helix (HI) 207-223 (Van Troys et al. 1996; Puius et al. 2000) and those belonging to the A - A loop (especially 168-170) are thought to be involved in binding to F-actin (Puius et al. 2000; Renoult et al. 2001; Irobi et al. 2003). McGough et al. (1998) showed by the means of electron cryomicroscopy that gelsolin fragments G1-G3 and G2-G6 bridge two longitudinally associated actin monomers in a filament. Furthermore, they proposed that the G2-F-actin interface is centered axially at subdomain 3 and radially between subdomain 3 and 1 of the upper monomer of the bridged pair (McGough et al. 1998). Topologically, the side-binding surfaces of G2, including certain residues from HI and the A strand, are similar to the filament end-binding surface of G l (Puius at al. 2000).  16  These results, when combined, allow proposal of a self-consistent model for interaction of Gl and G2 with F-actin (Figure 13).  wcorw)  Interface  Figure 13. Model of the complex of F-actin and gelsolin domains 1 and 2. The Gl/actin structure (McLaughlin et al. 1993) is positioned at the barbed end of a Lorenz (Lorenz et al. 1993) model of F-actin. G2, modeled on consideration of G2 in inactive gelsolin (Burtnick et al. 1997) and activated, isolated severin domain D2 (Puius et al. 2000) (Figure 11), is fitted to a Gl-binding site as shown. A 30 A-gap between G l and G2 in the model is postulated to be spanned by an extended G1-G2 linker that requires some unfolding of the N-terminal portion of G2. [From Puius et al. 2000].  Recently, a structure at 2.6 A resolution of a complex between actin and human G1+ (gelsolin residues 25-160) became available (Figure 14) (Irobi et al. 2003). In this activated and actin-bound form of G1+, the G1-G2 linker peptide extends up the surface of actin and points to a position suggested to be the principal binding region for G2 in the  17  model presented in Figure 13, giving additional support to the model presented in Figure 13.  Figure 14. 2.6 A resolution structure of a complex formed between actin (cyan) and G l (in red) with an extended linker from the start of G2 (shown in yellow). A green star indicates roughly the position at which G2 would be found if it contacted the next actin monomer toward the pointed end of the same strand. A dark gray sphere represents C a  2 +  bound at the type-2 site on G l , and a yellow sphere, C d  at the type-1 site in the structure. A T P and a C d  2 +  2 +  bound  ion are shown in the nucleotide cleft in actin. (Irobi et al.  2003).  G2 contacting the side of F-actin is thought to be the initial event in the severing process (reviewed by Sun et al. 1999). It provides an anchor for the whole molecule, and allows G l to insert itself between actin units as the next step in severing. Phosphoinositide-binding regions have been identified in gelsolin domains G2 and G6 (Yu et al. 1992; Feng et al. 2001). Sites responsible for PPI binding include residues 135-142 (PI) and 150-169 (P2) on G2, and 620-634 ( P 3 ) on G6. PPIs are phospholipids characterized by large inositol headgroups and negative charges carried by phosphate moieties on the inositol ring. In vivo, gelsolin is most likely regulated by PIP2, the most common polyphosphoinositide in cells. All of the relevant peptide regions in G2 (PI and P2) and G6 ( P 3 ) contain a stretch of basic residues spaced almost identically, which is thought to be involved in binding negatively charged phosphate groups of the PPIs. F-actin and PPI binding sites PI and P2 partly overlap. Therefore, displacement of  18  actin by competition for the same site on gelsolin could represent a possible PPI inhibitory mechanism. Although G6 makes contact with actin in the structure of the G4-G6/actin complex (Robinson et al. 1999), biochemical observation of such a contact, or of one that involves G3 with actin, is lacking. These domains appear to have lost their abilities to bind at the part of the surface of actin that has been shown to bind Gl and G4, and which is thought to bind G2. In the regulatory process, G3 and G6 seem to have taken on the role of latches that hold the individual N- and C-terminal halves of gelsolin in a conformation that blocks actin binding, even after the tail latch has been released (Burtnick et al. 1997; Robinson et al. 1999). In inactive gelsolin, Gl and G3 are in such close contact, as are G4 and G6, that the actin binding surfaces on Gl and G4 are sterically hindered from contact with actin (Figure 15). Activation of these half-gelsolins 2+  by the binding of Ca  ions evolved to expose the relevant regions of gelsolin.  Figure 15. G3 and G6 in the structure of inactive gelsolin act as latches to sterically hinder the approach of actin to the long helices of G1 and G4, respectively (Robinson et al. 1999).  G6, in addition to being involved in hindering the approach of actin to HI on G4, also functions together with the C-terminal tail latch to contact and lock G2 in a position where it is masked from contact with actin. Interaction of Asp670 of G6 with Argl68 and Argl69 of G2 assists in formation of this latch. (Figures 4, 5 and 9) (reviewed by Sun et al. 1999; Lueck et al. 2000; Burtnick et al. 1997; Pope et al. 1997). Consideration of the portion of Figure 9 in which the sites labeled IIG2 and IIG6 are situated suggests that  19  Ca -binding at the type-2 sites of G2 and G6 might cooperatively open the tail latch, 2+  with further binding events triggering release of the G1-G3 and G4-G6 latches (Choe et al. 2002).  4. Changes in gelsolin on activation by Ca  As noted, a structure of activated intact gelsolin, either in isolation or in a complex with actin, is not yet available. However the active C-terminal half of gelsolin (G4-6) has been crystallized in complex with actin (Robinson et al. 1999) (Figure 16) and in an actin-free state (Kolappan et al. 2003; Narayan et al. 2003). Interestingly, the structure of activated G4-G6 does not depend on the presence of actin or on occupancy of the type-1 Ca -binding site. 2+  Figure 16. Transition from G4-G6 in the absence of C a  2 +  (Burtnick et al. 1997) to its active  conformation bound to actin (Choe et al. 2002). Silver spheres represent C a  2 +  bound at type-2 binding sites  and Ca * bound at the type-1 site is presented as a gold sphere [from Choe et al. 2002]. 2  Binding G4-G6 does not significantly influence the structure of actin. The major changes in G4-G6 on activation do not involve significant restructuring within individual domains (with the relatively minor exception of the straightening out of the kinked HI helix of G6). Rather, it is large-scale relative motions among the domains themselves that  20  effect exposure of the actin-binding site on G4, with linker peptides serving as molecular pivots (Figure 17).  Figure 17. G4-G6 excised from the structure of Ca~ -free gelsolin, on the left (Burtnick et al. +  1997), and from the structure of G4-G6 bound to actin in the presence of C a  2 +  (Robinson et al. 1999) on the  right. Note the approximate 90° rotation of G6 relative to G4 both in the plane of the page and perpendicular to it, and the approximate 40 A translation of G6 away from its position in the inactive state. The two C a  2 +  ions bound to G4 are shown as black spheres, [from Robinson et al. 1999]  During this process, the relative positions of G4 and G5 change only slightly, and the final position of G5 is completely away from actin (Figures 16 and 17). Despite the similarities in structures between the N - and C-terminal halves of gelsolin in the inactive molecule, it is not reasonable to use the activated G4-G6 structure as a template upon which to build a predicted structure for the activated N-terminal half of gelsolin (Robinson et al. 1999). G2, according to a large body of evidence, unlike G5, must contact actin. Experimental evidence already discussed (Figures 13 and 14) suggests that a part of the N-terminal portion of G2, perhaps the A ' strand and part of the A ' - A loop,  21  unfolds and extends to permit translation of G2 away from Gl in order to bind to F-actin as indicated in the model shown in Figure 13. In this work, we set about to crystallize a complex that involved intact, activated gelsolin together with two actin monomers. This would have provided a complete picture of the end result of severing and capping of an actin filament by gelsolin. This goal remains elusive, but we did succeed in obtaining the crystal structure of the activated Nterminal half of gelsolin bound to actin. As expected, large relative shifts among the domains are required to expose actin-binding sites on Gl and G2. The structure is dramatically different than that of the G4-G6/actin complex and presents a major advance in understanding the molecular mechanism by which gelsolin attaches to, severs and caps an actin filament.  22  CHAPTER II Materials and Methods A. Protein purification 1. Preparation of horse plasma gelsolin  A modified version of the method of Bryan (Bryan 1988) for the isolation of human plasma gelsolin was used to purify horse gelsolin. Horse serum (Pel-Freez Biologicals, Rogers, AR) was treated sequentially with an anion-exchange medium in the presence and then the absence of free C a . Typically, 500 ml of frozen serum was 2+  thawed under a cold stream of water, in the presence of protease inhibitors (50 ui of stock solutions of leupeptin and pepstatin at 2 mg/ml in H 0 and DMSO, respectively, were 2  added to achieve a final concentration of each inhibitor of 2 x 10" mg/ml). The serum 4  then was dialyzed against 3 changes of 4 L of 25 m M Tris-HCl, pH 7.5, 0.5 m M CaCl  2  buffer at 4 °C, over 3 days. The dialyzed serum was centrifuged at 10,000 rpm (Sorvall GSA rotor and RC-5B centrifuge) for 50 minutes at 4 °C to eliminate precipitated matter. Initial fractionation of serum proteins was achieved by batch anion exchange treatment of the supernatant after centrifugation. First, NaCl was added to the solution to 35 m M . Next, the solution was mixed with 1.5 - 2 L of settled volume of DEAE-Sephadex A-50 (Pharmacia) anion exchange resin that had been equilibrated against 25 m M Tris-HCl, pH 7.5, 0.5 m M CaCl , and 50 m M NaCl. The mixture was kept at 4 °C for 1.5 hours, 2  with gentle stirring every 15 minutes. In these buffer conditions, with 0.5 m M CaCl  2  present, gelsolin does not bind to DEAE-Sephadex A-50. The resultant slurry was filtered using Whatman #3 filter paper. The majority of the filtration was allowed to proceed by gravity. When the layer of solution above the settled gel had settled into the gel, vacuum was applied to extract additional protein solution. E D T A was added to the filtrate to 10 m M , after which the pH was adjusted to 7.8 using concentrated HC1. This solution then was concentrated by ultrafiltration using an YM-30 membrane (Amicon) to a volume of 50 ml. The concentrate was filtered through 0.22 um pore size Millipore filters and applied to an HPLC anion exchange column (150 x 20 mm, BioSep DEAE-P, Phenomenex) at room temperature by multiple injections of 2-ml volumes. The flow rate  23  through the column was 1.5 ml/min. A series of steps in NaCl concentration in the elution buffer were used to elute gelsolin. First, the column was washed with the equilibration buffer, 25 m M Tris-HCI, pH 7.8, I m M EDTA, 0.1 m M N a N (buffer 1) until the U V 3  absorbance of the eluant at 280 nm reached baseline. Then, proteins forming weaker interactions than gelsolin with the ion exchanger, were eluted with 44 m M NaCl in buffer 1. Gelsolin, along with some minor contaminants, was eluted with 60 m M NaCl in buffer 1. During this stage of the chromatography, fractions (approximately 5 ml) were collected manually. Gelsolin was localized in these fractions by polyacrylamide gel electrophoresis (PAGE) in the presence of sodium dodecyl sulfate (SDS) and 2-mercaptoefhanol. A final chromatographic step to remove contaminants involved affinity chromatography with Affi-Gel Blue (BioRad Laboratories). After dialysis against 25 m M Tris-HCI, pH 8.0, 1 m M E D T A to remove NaCl, the protein was applied at 1 ml/min to an Affi-Gel Blue column (12-15 ml bed volume in a 20-ml syringe) that has been equilibrated with 25 m M Tris-HCI, pH 8.0, 1 m M E D T A (buffer 2). After a two-bed volume wash with equilibration buffer, gelsolin was eluted with 1.5 m M ATP in buffer 2. Fractions, 6 ml in volume, were collected manually. Fluorescence was used as a protein detection method. The fluorescence intensity of each fraction was checked using a Perkin-Elmer luminescence spectrometer (model LS-5B) using an excitation wavelength of 280 nm and an emission wavelength of 340 nm. Appropriate fractions were collected and dialyzed against 25 mM Tris-HCI, pH 8.0, 1 m M EDTA, to remove ATP. The gelsolin concentration of the final dialyzed solution was determined by measuring its absorbance at 280 nm, using an absorption coefficient of 1.4 ml mg" cm" , as determined in this 1  1  laboratory (Ruiz Silva and Burtnick 1990).  2. Actin purification  Actin was purified from acetone-dried powder prepared from rabbit skeletal muscle according to a procedure based on the method of Spudich and Watt 1971. Typically, 2.5 g of rabbit muscle powder was extracted with 40 ml of 2 m M Tris-HCI, 0.2 m M CaCl , 0.2 m M ATP, 1 m M DTT, pH 7.6 (buffer A) for 30 minutes, on ice, with 2  constant stirring. The extract was filtered through a double layer of cheesecloth and  24  Whatmann #1 filter paper, using vacuum. The solid residue was washed with an additional 30 ml of buffer A andfilteredthe same way. The combined extracts were clarified by centrifugation at 35,000 rpm in a Beckman rotor 45Ti for 1 hour at 4 °C using a Beckman Model L3-50 ultracentrifuge, after which the pellet was discarded. G-actin in the supernatant was polymerized by addition of KCl and MgCl to a 50 mM and 2 mM, 2  respectively, and allowing the solution to stand overnight at 4 °C. To promote dissociation of actin binding proteins, such as troponin and tropomyosin, from actin filaments, KCl was added to 0.8 M. The solution was stirred for 2.5 hours at 4 °C, with subsequent centrifugation for 3 hours at 35,000 rpm (80,000 x g) in a Beckman rotor 42.1. This produces a translucent pellet of F-actin, leaving contaminating proteins in the supernatant. The translucent pellet was resuspended in 5 ml of buffer A (using a Potter homogenizer), and dialyzed against three changes of 1 L each of buffer A over 3 days to achieve reconversion of the protein to G-actin. Traces of polymerized actin were removed from the G-actin solution by ultracentrifugation at 80,000 x g for 3 hours. The concentration of G-actin in solution was determined spectrophotometrically (PerkinElmer lambda 4B UV/Vis spectrophotometer), using an extinction coefficient at 290 nm of 0.63 ml mg^cm" . 1  3. Formation and purification of the complex of one gelsolin with two actins  Gelsolin, in 25 mM Tris-HCI, pH 8.0, 1 mM EDTA was incubated for 5 min in the presence of 2 mM CaCl at room temperature. Then actin in buffer A was added to 2  the activated gelsolin at a molar ratio of 2:1 such that the final solution contained equal volumes of 25 mM Tris-HCI, pH 8.0, 1 mM EDTA and buffer A. The resulting solution was incubated overnight at 4 °C to allow for complex (GA ) formation. Gel filtration 2  chromatography was used to purify the complex. The GA solution was applied to a Bio2  Rad Sephacryl S300 column (90 x 2.5 cm) at room temperature and eluted with 25 mM Tris-HCI, pH 8.0, 2 mM CaCl at a rate of 1.2 ml/min. Elution of the complex was 2  followed by UV absorbance at 280 nm. GA either eluted as a single peak or its elution 2  was preceded by that of F-actin, which formed when a small excess amount of actin was present during complex formation. The fractions containing G A were pooled and 2  25  /  concentrated using Millipore and Biosep centrifugal concentrators, to a concentration of 10 mg/ml. The concentration of GA was approximated spectrophotometrically using an 2  absorption coefficient at 280 nm of 1.25 ml mg^cm" . 1  B. X-ray crystallography 1. Introduction Knowledge of the structure of a macromolecule is a major step toward understanding how it works. X-ray crystallography has been and continues to be the most productive technique for obtaining macromolecular structures at atomic resolution. Since the closest distance between two atoms in space is the length of a covalent bond, approximately 1 A (0.1 nm), x-ray radiation with a wavelength range of 0.1-10 nm is appropriate to determine the overall molecular shape and solve the three-dimensional arrangement of atoms in a protein. To observe diffraction, an ordered, repeated arrangement of macromolecules is necessary. Accordingly, the first step towards a structure is obtaining high quality single crystals, followed by diffraction pattern collection, and subsequent conversion of raw data into an electron density map of the macromolecule, with structure refinement as the final step (van Holde et al. 1998).  1 a. Crystal characteristics Crystals are solids that are exact repeats of a symmetric motif. Each symmetric system can be reduced to a level that is not symmetric. Rotational or screw operators applied to this asymmetric unit, generate a lattice motif (a symmetric motif). In the crystal, the lattice motif is translated in three dimensions to form a regular and repeated array called the crystal lattice, in which each lattice motif defines the edges of a unit cell, the smallest repeatable set of lattice points in the crystal. It contains all the atoms of the lattice motif. Since the crystal is composed of repeating unit cells in three dimensions, the shape of a unit cell is constrained to a being a parallelepiped. Each unit cell is defined by a set of vectors a, b, c, with lengths a, b and c, respectively, and with the angles a between b and c, [3 between a and c, and y between a and b. Lattice points are necessarily 26  found at the comers of the unit cell, but are not restricted to them. They may be found at the centers of the faces, or at the center of the unit cell, as shown in the Figure 2. There are only 14 unique types of crystal lattice, the Bravais lattices (Figure 18). A crystal's lattice type, along with the symmetry of its unit cell, defines a space group. There have been a total of 65 different space groups identified for naturally occurring biological macromolecules (van Holde et al. 1998). Crystal morphology is characterized by space group and unit cell parameters, and these dictate how the crystal will diffract x-rays.  mm  a: - -  I clinic.:  fit  sMonccljiisiJKS::  •  •I'-!:  1  i • I'.'l'. i •  -  IKBIIiBMi  r  4_i I'  I'.-l! I g n f j a '  ,—| j or  :  Willi  c  a •= i? = <"0": •> = ir.i"i ; =  r,:  lllMli  PllS.llDt.IWrl/il  ,  Iscmernc  ilBlil  Ifisiiilii  Figure 18. The fourteen Bravais lattices (  27  / b. Protein crystallization  There are a number of factors important for crystallizing a protein. A necessary step in growing protein crystals is to bring the concentration of the macromolecule in a solution well above its intrinsic solubility, i.e., to supersaturation, at which nucleation of a minimal crystal lattice can take place. It has been estimated that, at minimum, four unit cells need to associate to form a stable nucleation lattice. After such a seed is formed, there is a high probability of single molecules adding at the crystal surfaces, even at concentrations below those required for nucleation. Thermodynamically, the force that drives crystallization results from the nonequilibrium nature of a supersaturated solution. The system will return spontaneously to an equilibrium state, where crystalline and soluble protein forms may coexist. A twodimensional phase diagram shows what happens as the concentrations of precipitant and protein increase in a sample (Figure 19).  Concentration of precipitating agent Figure 19. Schematic representation of two-dimensional crystallization phase diagram (  Equilibrium requires a combination of concentrations of protein and precipitant that makes the chemical potential of each component in the system the same in the crystalline and the solution phases. This can be achieved with the desired outcome of crystal growth, but the more general end result is an amorphous precipitate. To tilt the  2 8  odds more in favor of crystallization, it seems that a slow approach both to achieving supersaturation and subsequent relaxation to equilibrium is best. Protein supersaturation is achieved by the use of precipitating agents such as salts that result in highly hydrated ions (e.g., ammonium sulfate), long chain polar molecules (e.g., polyethyleneglycols), or water-miscible organic solvents (e.g., isopropanol), that reduce the water content of a protein's hydration shell and, therefore, promote transfer of protein from solution to solid. Simply adding such agents to a protein solution generally leads to precipitation. However, in a closed container, separation of a concentrated solution of precipitant from a more dilute one, in which the protein remains soluble, results in slow transfer of volatile components through the vapor phase between solutions in an attempt to reach equilibrium. This vapor diffusion method often involves transfer only of water, generally the only volatile component of either solution. Water then migrates from the protein solution to the concentrated solution of precipitant, with the result being a slow increase in concentration of the protein solution, often to a state of supersaturation. Another important prerequisite for successful growth of crystals is purity of the macromolecule. For most proteins, more than 95% chemical purity is needed in order to produce crystals. Not less significant is the macromolecule's structural purity. By this, it is meant that the presence of a homogeneous population of conformations of the molecule is as important. It is difficult to find conditions at which the protein is going to crystallize in a well-controlled manner, despite knowledge of subsets of conditions under which the macromolecule is insoluble and of ones that favor stable conformation. Therefore, a large number of different solutions have to be screened to find the specific one that will satisfy the solubility and conformational requirements (van Holde et al., 1998). Typical dimensions of suitable crystals for data collection are 0.1-1 mm in each direction in space. With the use of more intense x-rays, such as are available at synchrotron sources, the minimum dimension can be reduced to 0.02 mm.  29  lc. X-ray diffraction and electron density maps  The equivalence in energy of the quantum of Roentgen radiation and that of the electrons in the atomic orbitals leads to primarily scattering of the incoming X-rays by electrons. If Bragg's law conditions are met, the reinforcement of the scattered radiation will occur and result in diffraction. Bragg's law says that a constructive interference from a set of parallel planes in the crystal will occur when the distance traveled by reflected X rays off these planes differs by an integral number (n) of wavelengths (van Holde et al. 1998). Diffracted X-rays emerge from the crystal at different angles and have different intensities. Each diffracted x-ray makes a spot-reflection, where it intersects the X-ray film (detector). These spots represent an occurrence of constructive interference of reflected waves (van Holde et al. 1998). In order to get a complete data set at a fixed X, the crystal needs to be rotated, so that every possible set of planes can satisfy Bragg's law and yield diffraction. Recorded diffraction images (sets of reflections) from each rotation combined constitute a complete data set-diffraction pattern (Figure 20). The positions of diffraction maxima (reflections) in the diffraction pattern are directly related to the shape and size of the crystal unit cell. The intensities on the other hand contain information needed to determine molecular structure (van Holde et al. 1998).  Figure  20.  Typical  diffraction  pattern (  lennart/more-info/xray.php).  Each reflection in the diffraction pattern represents a diffracted wave described by a complex number, called a structure factor. Each structure factor is a result of scattering from every atom within the protein; therefore the information from all structural factors is  30  required to determine all types of atoms and their position in space. A structure factor contains two independent physical parameters, the amplitude and the phase of the diffracted wave, both of which are necessary to solve the macromolecule's structure. The electron density at the particular position in the asymmetric unit is described by the specific set of amplitudes and phases. Due to device limitations, it is not possible to identify phases of diffracted waves. There are several methods for solving the phase problem. In molecular replacement method, used in this work, a known structure of a chemically and structurally closely related molecule is used to derive phase information for the unknown one. The amino acid sequence of the protein is fitted into a generated electron density map of the asymmetric unit (Figure 21). This initial structure is then refined, until the values of experimental structure factors and the ones calculated from the current structure model are as close as possible. The refined model can be presented as a solution to the structure if it satisfies the certain limit to the difference between these values (van Holde etal. 1998).  Figure 21.  Peptide sequence and Car ion (black sphere) fitted into the corresponding part of an +  electron density map (Choe et al. 2002).  / d. Resolution Refining a structure to atomic resolution means that the individual positions of covalently bonded atoms can be distinguished. Most protein crystals diffract to a  31  resolution between 1.8 and 3 A . High resolution macromolecular structures are those resolved to 1.2 A or better. At the resolution of 3 A , a protein crystal structure provides information on the location of the polypeptide backbone, the orientations of amino acid side chains (which can suggest the locations of salt bridges and hydrogen bonds), the degree of solvent accessibility, and identifies regions of flexibility and/or alternative structural conformations.  2. Crystallization of G A  2  GA was crystallized by the hanging drop vapor diffusion method (Figure 22). 2  Using the hanging drop technique, a small drop of protein sample, mixed with a precipitant solution, is placed on a plastic cap, which is then inverted over the reservoir solution and screwed into place. Nextal Biotechnologies Inc. plates containing 24 reservoirs were used in the experimental set up. The initial precipitant concentration in the droplet is less than that in the reservoir. Over time the crystallization solution will draw water from the droplet through the vapor phase such that equilibration will occur between the protein solution and that in the reservoir. During the equilibration process, the protein becomes concentrated, reaching supersaturation, which may lead to protein crystallization.  reservoir ^yj-jiiriji  Figure 22. Hanging drop crystallization set up (  32  Crystals were grown at 4 °C from a protein sample at 10 mg/ml mixed with a reservoir solution of 100 mM Na acetate, pH 4.7, 2% PEG 8000 (w/v), 2 mM CaCl , at a 2  1:1 ratio (v/v) (total drop volume of 2 or 4 u.1). These conditions were determined to be optimal after screening several variables, including pH, temperature, molecular size of PEG, and the concentrations of buffer components, PEG, and CaCl . Crystallization trays 2  were monitored regularly using a binocular microscope, and crystals generally became visible after two to three weeks. The crystals appeared dodecahedral in shape, and were less than 50 um in maximal dimension.  3. Data collection Prior to data collection, crystals were flash cooled to a liquid nitrogen temperature (77 K) in the presence of a cryoprotectant solution containing 100 mM Na Acetate buffer, pH 4.7, 25 % (v/v) anhydrous glycerol, 10% (w/v) PEG 8000, 400 mM NaCl, and 5 mM CaCl . 2  Small nylon fiber loops glued to metal pins (these and other crystal mounting apparatus were from Hampton Research, Inc.), were used for crystal transfer from the mother liquor to cryoprotectant solution. Observed through a microscope, a single crystal was trapped in the loop, and transferred into a well in a crystal screening plate that contained cryoprotectant. After an incubation time of several minutes in the cryoprotectant solution, the crystal was again picked up in a cryoloop, immersed in liquid nitrogen, placed into a Hampton magnetic cryo-cap storage vial, and transferred to a cane in a cryogenic transport dewar (MVE model SC 4/2v) filled with liquid nitrogen. Prior to transport to a synchrotron facility, liquid nitrogen was poured out of the dewar. The primary purpose of flash cooling of crystals is to reduce radiation damage to the crystal during exposure to an x-ray beam. Principally, this damage is due to creation of free radicals within the crystal, which is mostly solvent, and reaction of such free radicals with the ordered protein molecules under study.  Disorder induced by such  reactions leads to decreased quality diffraction, eventually to an unusable level. By lowering the temperature to that of liquid nitrogen, free radical diffusion is significantly  33  slowed down and therefore damage to the crystal during the data collection period may almost be eliminated. Cryoprotectant solutions are used in the process of flash cooling. These suppress crystallization of water present in the crystal system. Ice formation would degrade protein diffraction by overwhelming certain regions of the diffraction pattern, as well as by damaging the protein crystal lattice. The cryoprotectant in the solution modifies its physicochemical properties in such a way that a vitreous, glass-like state is attained for the solvent at liquid nitrogen temperatures. Common cryoprotectant agents used for freezing biocrystals include glycerol, various length polyethylene glycols, sucrose etc. Diffraction data were collected by Drs. L. Burtnick and R. Robinson at beamlines ID-29 and 14-4 of the European Synchrotron Radiation Facility in Grenoble, France. At this facility, high-energy electrons are steered in a circular storage ring by a series of electromagnets. At each change in the direction of the linear path of such electrons a burst of x-ray radiation is emitted, which can be collimated and directed down beamlines tangential to the main electron storage ring. Such beams are typically 1000-fold or more intense than those generated by normal laboratory x-ray sources. This allows for more rapid data collection and for collection of improved diffraction data from small or weakly diffracting crystals. Data were analyzed at Uppsala University by Drs. R. Robinson and L. Burtnick using conventional protein crystallographic software.  4. Protein structure figure preparation  Several of the ribbon representations of proteins were prepared by Dr. R. Robinson using the program MOLSCRIPT (Kraulis 1991). The remainder were created by myself with the use of R A S M O L software.  34  CHAPTER III Results and Discussion  A. Purification of gelsolin Gelsolin isolation from horse serum, as performed routinely in this laboratory, required three chromatographic steps. The first step, batch anion exchange treatment at pH 7.5 in the presence of 1 m M CaCb, aimed to eliminate serum albumin, which constitutes 60% of the total protein in blood plasma, as well as other proteins carrying a negative charge under the conditions of the treatment. Inactive gelsolin bears a negative charge at pH 7.5, but binding C a  2+  alters its surface charge properties sufficiently so that  it does not bind to the anion exchange resin. Following batch treatment, a divalent cation-chelating agent, EDTA, was added to the batch filtrate in order to restore gelsolin to its inactive condition, bearing a negative charge. The next step, when this work was in its early stages, used to involve conventional low-pressure anion exchange chromatography at pH 7.8. During the course of the project, modification of this step to use an HPLC-based anion exchange column facilitated the method and enabled good yields from less starting material. HPLC has several advantages over the open-column chromatography, such as decreased times of separation and improved resolution. With the majority of negatively charged proteins having been removed by batch treatment, and with the positively charged ones simply flowing through the column at this pH, gelsolin remains as the major protein bound to the column at the relatively low ionic strength employed. Elution with increased salt concentration, however, didn't yield sufficiently pure gelsolin for crystallization purposes. Therefore, further purification was necessary to achieve homogeneity both in terms of protein identity and of protein structure. A final step of treatment, with an affinity matrix, was retained for the purposes of enhancing the chances of crystallization of the product. Gelsolin has a high affinity for the dye Cibacron Blue F3GA, which is commercially available attached to an inert agarose support matrix. Addition of ATP at  35  1.5 mM to the equilibration buffer released crystallization-competent gelsolin from the matrix (Yamamoto et al. 1989).  B. Purification of the GA complex 2  The G A 2 complex, when subjected to size exclusion chromatography, generally eluted as a single peak (Figure 23). When small amounts of excess actin were present during formation of the complex, elution of a small F-actin peak preceded that of G A 2 .  Figure 23. Elution of G A from a Bio-Rad Sephacryl S300 size exclusion column (90 x 2.5 cm) in 2  25 m M Tris-HCI, 2 m M CaCl2, pH 8.0 at a rate of 1.2 ml/min. The absorbance of the eluant was monitored at 280 nm and recorded at sensitivities of 0.2 AUFS and 0.1 AUFS using a two-pen chart recorder. Elution time increases from right to left in this figure.  C. Structure of the N-terminal half of gelsolin bound to actin 1. Gl-G3/actin Diffraction data from our crystals were subjected to rigid body molecular replacement analysis using as models the coordinates for activated gelsolin G4-G6 bound to actin (PDB ID 1H1V) (Choe et al., 2002), G l bound to actin (PDB ID 1EQY) (McLaughlin et al. 1993), isolated G2 (PDB ID 1KCQ) (Kazmirski et al. 2002), and G3 excised from the structure of inactive gelsolin (PDB ID 1D0N) (Burtnick et al. 1997).  36  Such analysis using all of the models failed to yield an acceptable solution. However, analysis using various combinations of the models in the search yielded a single acceptable solution. This identified the proteins in the crystals to be actin and the Nterminal half of gelsolin, i.e., domains G1-G3. This outcome is likely the result of proteolytic digestion of gelsolin during the crystallization process, followed by 2"i*  crystallization of the observed complex. It is known that the presence of Ca  ions  induces gelsolin susceptibility to proteolytic attack. For example, the primary proteolytic cleavage site for a-chymotrypsin on gelsolin is located between Met406 and Ala407 in the G3-G4 linker region. This cleavage site is located in the middle of a seven-residue peptide segment that is not solvent-accessible in the conformation of Ca -free gelsolin. Ca - binding to gelsolin causes a large shift in the relative orientation of the two halves, 2+  leading to the more "open" state of the molecule. This exposes the G3-G4 linker to proteolytic attack (Koepf et al. 1998). In the GA complex, the G3-G4 linker connects the 2  two halves of gelsolin to two different actin monomers, which would belong to different strands in F-actin. It is expected to be extended and exposed to any proteolytic enzymes in the solvent. SDS-PAGE analysis of crystals grown in conditions identical to those subjected to diffraction experiments showed a dominant protein band at approximately 40 kDa, the zone to which actin and each of the two halves of gelsolin would migrate, with negligible signs of an 83 kDa band that would correspond to intact gelsolin (Figure 24). Thus, it is evident that proteolysis occurred during the 2-3 weeks typically required for crystal growth to be evident. This result is very similar to one reported from a crystallographic study of Ca -activated cloned mouse gelsolin (Kolappan et al. 2003). The authors had 2+  purified intact gelsolin, but during the period required for crystal growth, proteolysis had occurred and the crystals they thought would contain whole gelsolin actually contained only the C-terminal half of the protein. In our case, fitting amino acid sequence [residues 27 through 371 in the horse plasma gelsolin sequence (Koepf et al. 1998)] to electron density unambiguously demonstrates that our crystals are of the N-terminal half of gelsolin bound to a single actin monomer.  37  Figure 24. SDS-PAGE analysis of crystals grown in the identical conditions to those subjected to diffraction experiments. Lane 1: 80 kDa protein marker. Lane 2: 45, 66 and 97 kDa protein markers. Lane 3: protein content of crystals removed from crystallization solution. In Lane 3, the major species present have a molecular mass of about 40 kDa, and negligible 83 kDa gelsolin is evident.  Gl-G3/actin complex crystals (space group P3i21; unit cell: primitive hexagonal with a = b = 145.25 A , c = 129.95 A , a = (3 = 90°, y= 120°) diffracted to 3.0 A resolution. The structure contains one ATP molecule bound in the nucleotide-binding pocket of actin 2+  and four Ca ions, one of them associated with ATP in the actin monomer, one at the type-1 binding site at the interface between Gl and actin, and one each in the type-2 binding sites of Gl and G3. At 3.0 A resolution, we are unable to claim with confidence that the type-2 site on G2 is occupied (Figure 25).  38  Figure 25. Structure at 3.0 A resolution of G1-G3 bound to actin in the presence of C a Actin is colored cyan, G l is red, G2 is green, G3 is yellow, type-2 and type-1 C a and gold spheres, respectively, A T P is colored orange and its associated C a  2 +  2 +  2 +  ions.  ions are shown as black  is painted violet.  The structure of Gl-G3/actin reveals the conformational changes that must be undergone in the N-terminal half of the molecule on activation by C a  2+  to expose its  actin-binding sites. Its orientation on actin, the involvement of G2 and G3 with the same actin unit that binds G l , and occupation of the type-2 Ca -binding site on G3 are all 2+  being reported for the first time. These direct, detailed observations also provide insight  39  into the mode of binding of G2 to F-actin and suggest a model for the overall process of activation of gelsolin, its binding to the side of an actin filament, and its severing and capping activities.  2. Changes in the relative orientations of domains within G1-G3 upon activation  The largest and most obvious change occurring upon activation of the N-terminal half of gelsolin is the reorientation of the individual domains with respect to each other. The 10-stranded (3-sheet that runs continuously through the cores of Gl and G3 in resting gelsolin is severed and new inter-domain interfaces are established (Figure 26). Of most interest, because it is the initial step in gelsolin's actions on actin, is exposure of the Factin-binding site on G2. This involves release and extension of the edge strand, A', from the core helix of G2, to bridge a 30 A gap between G2 and G l . Concomitantly, the actinbinding site on Gl is exposed. The center of mass of Gl is translated more than 30 A away from its resting position relative to G2, a motion that would have required significant rotational reorientation in the horizontal and vertical directions as well. The positioning of G2 observed on subdomain 2 of actin is such that if another actin unit were docked one unit further along toward the pointed end of same strand within the Holmes model for an actinfilament(Holmes et. al. 1990), its long helix would lie in the groove between subdomains 1 and 3 of that new actin unit. Thus, G2 would bind to the same region of actin that is known to bind Gl and G4, and through a similar set of contacts, as modeled by Puius et al. (2000) (Figures 27 and 13). However, a small gap is present between G2 and model actin unit to which it is expected to bind (Figure 28). It is possible that the position of G2, or of the actin subdomain that binds it, might undergo minor conformational changes if the actin monomer in question were actually present. However, an alternative is that the gap may reveal a flaw in the Holmes model for F-actin [discussed in the review by dos Remedios et al. (2003)]. Puius et al. (2000) used the Lorenz model for F-actin (Lorenz et al. 1993), a modification of the Holmes model in which the major difference is in the position of actin subdomain 2, which is shifted somewhat towards thefilamentaxis in the Lorenz structure. A dramatically different model for an actinfilamentwas proposed by Schutt et  40  al. (1995). Their approach was to start with the ribbon-like structure of the profilin/actin complex and transform it into a helical F-actin that resembles those seen in electron micrographs. The resultant model has actin subdomains 3 and 4 situated on the outer surface of the filament, completely opposite their position in the Holmes and Lorenz model. As G2 and G3 interact with actin subdomains 1 and 2, it would be appropriate to have these facing outward from the F-actin filament. Clearly, our structure for G l G3/actin favors the Holmes over the Schutt-Lindberg model. (The Lorenz model has fallen out of favor in recent literature.)  Figure 26. Comparison of the conformation of inactive G1-G3 (A) with that in the presence of Ca  2 +  and actin (B). The orientations of G l are preserved in the two panels.  41  Figure 27. Space filling representation of the Gl-G3/actin complex.  G3 has not been reported previously to bind actin. Clearly, it does in the G l G3/actin complex. In addition, several new interactions are created between G2 and G3 in the process (Figure 26B). The new location of G3 must be considered when constructing a bridge to the second half of gelsolin, G4-G6, which is thought to reach across the F-actin filament axis and contact the actin monomer that will become the barbed-end of the second strand and thus complete severing and capping of the filament (Figure 28). The polypeptide that links Asp371 on G3 to Met412 onG4 is sufficiently long to span the gap across the actin filament in the model.  42  Figure 28. Model for the capped barbed-end of F-actin. The three actin units (cyan, blue and dark gray) are positioned according to the Holmes model (Holmes et al. 1990). The cyan-colored actin with the bound G1-G3 (red, light green, and yellow, respectively) and the blue-colored actin with the bound G4-G6 (pink, dark green, and orange, respectively) are the terminal units in a two-stranded representation of Factin. If more actins were added, the filament would extend upward, in the direction of the pointed end of the filament. Only gelsolin-bound C a  2+  ions are shown (gold spheres if type-1, gray if type-2). The model is  rotated approximately 90° to the left with respect to the view of Gl-G3/actin shown in Figure 25.  The recent demonstration that the structure of G4-G6 in the presence of Ca , but 2+  in the absence of actin (Figure 32A), is fully activated (Kolappan et al. 2003; Narayan et al. 2003) suggests that the changes observed on activation of G1-G3 may also require only the presence of Ca  2+  ions in one or more of the type-2 metal ion-binding sites within  the N-terminal half of gelsolin. Despite the similarities in the location and number of the Ca -binding sites in the two halves of gelsolin, the involvement of actin in attaining the 2+  observed conformation of G1-G3 cannot be excluded, primarily due to the important differences in the properties of G2 and G5.  43  There is not sufficient information available to assign unambiguous consequences to each individual C a ' - binding event. Nonetheless, some suggestions can be made in light of the new structure. Ca -binding at the type-2 site of G l involves a residue located 2+  in the G1-G2 linker region (Figure 29A and 12). As previously suggested (Burtnick et al. 1997; Robinson et al. 1999), this binding could cause reorientation of the linker region and contribute to the establishment of new relative positions of G l and G2. Similarly, the type-2 metal ion-binding site in G2 (Figure 12), having a coordination residue located in the G2-G3 linker (residue 259) (Kazmirski et al. 2002), might contribute to G3 attaining its final position in the active N-terminal half. Unfortunately, the quality of the electron density map for G1-G3 in this region does not allow for an unambiguous statement of the occupancy of the site (Figure 29B) and, therefore, of its role in activation. The structure of Gl-G3/actin in the presence of C a  2+  ions confirms the prediction  of the existence of a type-2 site in G3 (Choe et al. 2002). C a  2+  is observed in the  analogous location to that in other gelsolin type-2 sites (Figure 29C).  A  B  C  Figure 29. A) Type-2 Ca~ -binding site observed in G l i n the context of G1-G3. B) Peptide +  backbone conformation around the predicted type-2 site in G2 in the context of G1-G3. The resolution of the data was not sufficient to be certain of its state of occupancy. C) C a  2 +  bound at the type-2 site in G3,  observed for the first time.  The characteristics of the Gl/actin interface have been well documented (McLaughlin et al. 1993). For the first time, it is possible to identify the contacts formed  44  by G2 and G3 with the same actin to which G l is bound (Figure 25). As noted previously (Irobi et al. 2003), the extended G1-G2 linker forms multiple contacts with actin (Figure 30). The G1-G2 linker in G1-G3 forms these same interactions, as well as having G l bound in the same position as observed when G1+ binds to actin (Figures 14 and 25).  Figure 30. Actin contacts with the G1-G2 linker (from Irobi et al., 2003). The stereo pair of images stresses the strong polar interactions between chain segment Lysl50-Asnl55 of gelsolin (yellow) and segment Gly23-Val30 of actin (cyan). For clarity, only side chains that participate in polar interactions are depicted. Gelsolin residues (in bold type) interact with the following actin residues (in italics), starting from the bottom of the figure and moving upward: Lysl50 (N), Gly23 (O); Lysl50 (NZ), Asp24(ODl); Lysl50 (O), Asp25 (N); Hisl51 (NDl), Asp25 (OD2); Prol54 (N), Pro27 (O); Asnl55 (N), Arg28 (O); Asnl55 (OD1), Val30 (N); Asnl55 (ND2), Val30 (O).  3 . Structural changes within domains caused by binding of C a Preservation  of major  structural  2+  and actin  attributes of individual  domains  upon  transformation into their Ca -activated forms is apparent from comparison of each 2+  domain in inactive gelsolin (Burtnick et al. 1997) with various Ca" -activated structures now available, Gl/actin (McLaughlin et al. 1993), Gl+/actin (Irobi et al. 2003), isolated G2 (Kazmirski et al. 2002), G4-G6/actin (Robinson et al. 1999; Choe et al. 2002), isolated G4-G6 (Kolappan et al. 2003; Narayan et al. 2003).  45  Circular dichroism studies of intact gelsolin in solution (Koepf and Burtnick, 1996) suggest that there is little change in secondary structure, e.g. a-helical or |3-sheet content, upon addition of Ca  2+  to the solutions. NMR studies of isolated domains of other  members of the gelsolin family, villin (Markus et al. 1994) and severin (Schnuckel et al. 1995), also confirm that binding Ca  2+  does not result in rearrangement of the folding of  the polypeptide segments within those domains. The most obvious intra-domain change upon Ca -activation is the rearrangement of the long helix of G3 such that its axis is more closely parallel to the direction of the strands in the core (3-sheet. No longer is a kink in the helix evident, and it appears to be shortened by approximately one full turn overall (Figure 31). As discussed in the introduction section, to achieve the compact folded structure of intact Ca -free gelsolin, 2+  the long helices of G3 and G6 must bend away from a straight path in order to avoid steric clashes with the long helices of Gl and G4, respectively. Straightening of the helices in activated G1-G3 (shown here) and G4-G6 (Robinson et al. 1999) releases strain and, together with newly formed interfaces with other domains or with actin, contributes to stabilization of the active relative to the inactive conformation.  A  B  Figure 31. Rearrangement of the kinked long helix of G3 as a result of binding Ca" by G1-G3. A) +  G3 excised from the structure of inactive gelsolin (Burtnick et al. 1997). B) G3 excised from the G1 -G3/actin structure.  46  A second intra-domain change is evident in G2. In activated GI-G3, the G1-G2 linker region makes numerous contacts along the surface of actin and bridges a 30 A gap between Gl and G2 (Figure 25). In inactive gelsolin, this segment of polypeptide chain constitutes the edge strand (A') of the core (3-sheet of G2 and is structurally part of G2. This change, unlike the straightening of the long helices discussed above, is not mimicked on activation of the second half of gelsolin. It is at this level that the symmetry of structure and action breaks down between the first and second halves of gelsolin, and it is this difference that enables G2 to be unique among the six domains in being able to bind filamentous actin.  4. Comparison of the activation processes in G1-G3 and G4-G6  The N-terminal (G1-G3) and C-terminal (G4-G6) halves of gelsolin undergo different activation paths (Figure 32). Due to the fact that G2 binds to the side of actin filaments and G5 doesn't, it is not surprising that the activated conformation of G1-G3 is quite different from that of G4-G6 (Robinson et al., 1999). Following the tearing of the extended (3-sheets within both G1-G3 and G4-G6, dramatically different relative locations of G2 and G5 in their corresponding activated halves of gelsolin are established. G5 retains practically the same contacts with G4, while G2 moves away from Gl (Figure 32). Along with the mentioned resemblances in straightening and realignment of the kinked helices in both G3 and G6, an interesting observation that concerns both domains is that the structures in complex with actin show them making contacts with actin that had not been detected previously. Formation of such new interfaces, as well as those made by G2 with G3 and by G5 with G6, provides a way to stabilize the activated conformations of the two halves of gelsolin.  47  Figure 32. Comparison of the activated structures of the two halves of gelsolin. A) Activated G4-G6 (Narayan et al. 2003). B) Activated G1-G3 from the present work, positioned such that G l has a similar orientation to G4 from part A.  5. Further insight into gelsolin function  Knowledge acquired in the course of this work contributes to understanding how gelsolin participates in the regulation of actin. In resting gelsolin, a series of latches helps lock in a "closed" conformation that hides its actin-binding surfaces. Binding C a  2+  ions  releases the tail latch that holds together G6 and G2 and allows an "open" conformation of gelsolin in which two semi-independent halves are tethered together by an extended polypeptide linker. Additional domain-movement within each half of gelsolin takes place upon occupation of supplementary type-2 Ca -binding sites within them to fully expose 2+  the actin-binding regions. Following full activation of gelsolin, first in the sequence of events that lead to severing activity is binding of G2 to two longitudinally associated  48  actin monomers belonging to the same strand within F-actin. G2, then, plays the role of an anchor for the whole gelsolin molecule on actin. Next, the G1-G2 linker attaches along the lower actin protomer, in the direction of the barbed end of the filament. This directs G l towards its interaction site between subdomains 1 and 3 of the same actin protomer and forces detachment of the actin that had previously blocked that location. At this point, with one of the two strands of the filament already disrupted, the activated Cterminal half of gelsolin, primed for binding actin (Narayan et al. 2003; Kollapan et al. 2003), most likely attaches at the nearest monomer in the remaining intact strand, completing the severing action and leaving the newly created barbed end capped (Figure 28).  49  D. Summary  Activation for the N-terminal half of gelsolin by binding C a  2+  and actin produces  drastic changes in the relative orientations of the component domains when compared to their positions in inactive gelsolin. The active conformation of G 1 - G 3 also is substantially different from that of activated G 4 - G 6 , although some similarities exist as well. Rearrangements within G1-G3 during the process of activation prime G2 and G l for binding to actin. G2 binds to two actin protomers longitudinally arranged within Factin, one of which also binds both G l and G 3 . Existence of a type-2 C a  2+  bound in  activated G 3 , inferred from sequence analogies within the six gelsolin domains, is now experimentally confirmed. Future work clearly should be directed toward growth of crystals from an intact G A 2 or G A 3 complex. This might be achieved by additional modifications to the purification protocols to avoid or inactivate contaminant proteolytic enzymes, or to add enzyme inhibitors to the crystallization solutions. In parallel with this goal, it would be useful to try to crystallize either intact gelsolin or its N-terminal half in the presence of Ca  2+  ions, but in the absence of actin. A structure for either would answer whether fully  activated gelsolin can be obtained in the absence of actin. A structure of the former would answer questions about the crucial G 3 - G 4 polypeptide linker. In a different direction, rather than concentrating on structures that tell us how gelsolin binds to actin, it would be of interest to learn the molecular mechanisms that lead to dissociation of complexes once they have been formed, a process now referred to as "uncapping". For example, knowledge of how polyphosphoinositides, such as P I P 2 , bind to gelsolin would be a step toward understanding the regulated release of bound actin from otherwise stable complexes with actin.  50  Bibliography 1. Bryan, J. 1988. Gelsolin has three actin-binding sites. J. Cell Biol. 106(5): 15531562. 2. Bryan, J., and Hwo, S. 1986. Definition of an N-terminal actin-binding domain and a C-terminal Ca2+ regulatory domain in human brevin. J. Cell Biol. 102(4): 1439-1446. 3. Burtnick, L.D., Koepf, E.K., Grimes, J., Jones, E.Y., Stuart, D.I., McLaughlin, P J . , and Robinson, R.C. 1997. The crystal structure of plasma gelsolin: implications for actin severing, capping, and nucleation. Cell 90(4): 661-670. 4. 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. Am. J. Physiol. Gastrointest. Liver Physiol. 265, G1071-G1081. 5. Holmes, K G , Popp, D., Gebhard, W., and Kabsch, W. 1990. Atomic model of the actin filament. Nature 347(6288): 44-49. 6. Choe, H., Burtnick, L.D., Mejillano, M . , Yin, H.L., Robinson, R . C , and Choe, S. 2002. The calcium activation of gelsolin: insights from the 3A structure of the G4-G6/actin complex. J. Mol. Biol. 324(4): 691-702. 7. Dos Remedios, C.G., Chhabra, D., Kekic, M . , Dedova, I.V., Tsubakihara, M . , Berry, D.A., and Nosworthy, N.J. 2003. Actin binding proteins: Regulation of cytoskeletal microfilaments. Physiol. Rev. 83(2): 433-473. 8. Feng, L , Mejillano, M . , Y i n , H.L., Chen, J., and Prestwich, G.D. 2001. Fullcontact domain labeling: identification of a novel phosphoinositide binding site on gelsolin that requires the complete protein. Biochemistry 40(4): 904-913. 9. Irobi, E., Burtnick, L.D., Urosev, D., Narayan, K., and Robinson, R.C. 2003. From the first to the second domain of gelsolin: A common path on the surface of actin? FEBS Lett., in press. 10. Kabsch, W., Mannherz, H.G., Suck D., Pai, E.F., and Holmes, K.C. 1990. Atomic structure of the actin:DNase 1 complex. Nature 347(6288): 37-44.  51  11. Kangas, H . , Paunio, T., Kalkkinen, N . , Jalanko, A . , and Peltonen, L. 1996. In vitro expression analysis shows that the secretory form of gelsolin is the sole source of amyloid in gelsolin-related amyloidosis. Hum. Mol. Genet. 5(9): 12371243. 12. Kazmirski, S.L., 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. Biol. 9(2): 112-116. 13. Koepf, E.K., and Burtnick, L.D. 1992. Interaction of plasma gelsolin with tropomyosin. FEBS Lett. 309(1): 56-58. 14. Koepf, E.K., and Burtnick, L.D. 1996. Multiple pathways for denaturation of horse plasma gelsolin. Biochem. Cell Biol. 74: 101-107. 15. Koepf, E.K., Hewitt, J., Vo, H . , Macgillivray, R.T., and Burtnick, L.D. 1998. Equus caballus gelsolin—cDNA sequence and protein structural implications. Eur. J. Biochem. 251(3): 613-621. 16. Kolappan, S., Gooch, J.T., Weeds, A.G., and McLaughlin, P.J. 2003. Gelsolin domains 4-6 in active, actin-free conformation identifies sites of regulatory calcium ions. J. Mo.I Biol. 329(1): 85-92. 17. Kothakota, S., Azuma, T., Reinhard, C., Klippel, A., Tang, J., Chu, K., McGarry, T.J., Kirschner, M.W., Koths, K., Kwiatkowski, D.J., and Williams, L.T. 1997. Caspase-3-generated fragment of gelsolin: Effector of morphological change in apoptosis. Science 278: 294-298. 18. Kraulis, P.J. 1991. MOLSCRIPT: A program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24:946-950. 19. Kwiatkowski, D J . 1999. Functions of gelsolin: motility, signaling, apoptosis, cancer. Curr. Opin. Cell Biol. 11(1): 103-108. 20. Kwiatkowski, D.J., Janmey, P.A., Mole, J.E., and Yin, H.L. 1985. Isolation and properties of two actin-binding domains in gelsolin. J. Biol. Chem. 260(28): 15232-15238. 21. Kwiatkowski, D.J., Stossel, T.P., Orkin, S.H., Mole, J.E., Colten, H.R., and Yin, H.L. 1986. Plasma and cytoplasmic gelsolins are encoded by a single gene and contain a duplicated actin-binding domain. Nature 323(6087): 455-458.  52  22. Lin, K . M . , Mejillano, M . , and Yin, H.L. 2000. Ca  regulation of gelsolin by its  C-terminal tail. J. Biol. Chem. 275(36): 27746-27752. 23. Lorenz, M . , Popp, D., and Holmes, K . C 1993. Refinement of the F-actin model against X-ray fiber diffraction data by the use of a directed mutation algorithm. J. Mol. Biol. 234(3): 826-836. 24. Lueck, A . , Y i n , H . L , Kwiatkowski, D.J., and Allen, P.G. 2000. Calcium regulation of gelsolin and adseverin: a natural test of the helix latch hypothesis. Biochemistry 39(18): 5274-5279. 25. Maciver, S.K., Ternent, D., and McLaughlin, P J . 2000. Domain 2 of gelsolin binds directly to tropomyosin. FEBSLett. 473(1): 71-75. 26. Markus, M.A., Nakayama, T., Matsudaira, P., and Wagner, G. 1994. Solution structure of villin 14T, a domain conserved among actin-severing proteins. Prot. Sci. 3: 70-81. 27. 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. FEBS Lett. 276(1-2): 75-77. 28. 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 (2 Pt 1): 764-772. 29. 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(6439): 685-692. 30. 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. FEBS Lett., in press.  31. Otterbein, L , Graceffa, P., and Dominguez, R. 2001. The crystal structure of uncomplexed actin in the A D P state. Science 293: 708-711. 32. Pope, B.J., Gooch, J.T., and Weeds, A.G. 1997. Probing the effects of calcium on gelsolin. Biochemistry 36: 5848-15855.  53  33. Puius, Y . A . , Fedorov, E.V., Eichinger. L., Schleicher, M . , and Almo, S.C. 2000. Mapping the functional surface of domain 2 in the gelsolin superfamily. Biochemisty 39(18): 5322-5331. 34. Renoult, C , Blondin, L., Fattoum, A., Ternent, D., Maciver, S.K., Raynaud, F., Benyamin, Y . , and Roustan, C. 2001. Binding of gelsolin domain 2 to actin. A n actin interface distinct from that of gelsolin domain 1 and from ADF/cofilin. Eur. J. Biochem. 268(23): 6165-6175. 35. Robinson, R.C., Mejillano, M . , Le, V.P., Burtnick. L.D., Yin, H.L., and Choe, S.. 1999. Domain movement  in gelsolin: a calcium-activated switch. Science  286(5446): 1939-1942. 36. Ruiz Silva, B.E., and Burtnick, L.D. 1990. Characterization of horse plasma gelsolin. Biochem. Cell Biol. 68(4): 796-800. 37. Schnuchel, A . , Wiltscheck, R., Eichinger, L., Schleicher, M . , and Holak, T.A. 1995. Structure of severin domain 2 in solution. J. Mol. Biol. 247(1): 21-27. 38. Schutt, C.E., Myslik, J . C , Rozycki, M.D., Goonesekere, N . C , and Lindberg, U. 1993. The structure of crystalline profilin-beta-actin. Nature 365(6449): 810-816. 39. Schutt, C.E., Rozycki, M.D., Chik, J.K., and Lindberg, U . 1995. Structural studies on the ribbon-to-helix transition in profilin: actin crystals. Biophys. J. 68(4 Suppl): 12S-17S; discussion 17S-18S. 40. 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. Biol. Chem. 246(15): 4866-4871. 41. Sun, H Q . , Yamamoto, M . , Mejillano, M . , and Y i n , H.L. 1999. Gelsolin, a multifunctional actin regulatory protein. J. Biol. Chem. 274(47): 33179-33182. 42. van Holde, K . E . , Johnson, W . C , and Ho, P.S. 1998. Principals of physical biochemistry. Prentice Hall, Ch.6: 242-296. 43. Van Troys, M., Dewitte, D., Goethals, M., Vandekerckhove, ]. 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? FEBS Lett. 397(2-3): 191-196.  54  44. 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. Cell Biol. 109(1989): 593-605. 45. Yamamoto, H . , Terabayashi, M , Egawa, T., Hayashi, E., Nakamura, H., and Kishimoto, S. 1989. Affinity separation of human plasma gelsolin on Affi-Gel Blue../. Biochem. (Tokyo), 105(5): 799-802. 46. Yin, H.L., and Stossel, T.P. 1979. Control of cytoplasmic actin gel-sol transformation by gelsolin, a calcium-dependent regulatory protein. Nature 281(5732): 583-586. 47. Yu, 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. Biol. Chem. 267(21): 14616-14621. 48. 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 Pt 3:873-881.  Internet sources: 1. Birkbeck  College,  Walshaw,  J.  April  1995.  The  cytoskeleton.  <> (16 June, 2003) 2. Bravais lattices image <> (16 June, 2003) 3. Chirgadze, D. July 2001. Protein crystallization in action. <> (16 June 2003) 4. Hampton research  corp.  Hanging drop vapor diffusion crystallization.  • < 01/CG101 HDC.pdf > (16 June, 2003) 5. Virtue newsletter. June 2001. Protein crystal structure determination: X-ray diffraction and data analysis. <hthp://> (June 16, 2003)  55  


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