UBC Faculty Research and Publications

Selective depletion of microglial progranulin in mice is not sufficient to cause neuronal ceroid lipofuscinosis… Petkau, Terri L; Kosior, Natalia; de Asis, Kathleen; Connolly, Colúm; Leavitt, Blair R Nov 17, 2017

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

Item Metadata


52383-12974_2017_Article_1000.pdf [ 1.61MB ]
JSON: 52383-1.0361812.json
JSON-LD: 52383-1.0361812-ld.json
RDF/XML (Pretty): 52383-1.0361812-rdf.xml
RDF/JSON: 52383-1.0361812-rdf.json
Turtle: 52383-1.0361812-turtle.txt
N-Triples: 52383-1.0361812-rdf-ntriples.txt
Original Record: 52383-1.0361812-source.json
Full Text

Full Text

RESEARCH Open AccessSelective depletion of microglialprogranulin in mice is not sufficient tocause neuronal ceroid lipofuscinosis orneuroinflammationTerri L. Petkau1, Natalia Kosior1, Kathleen de Asis1, Colúm Connolly1 and Blair R. Leavitt1,2,3*AbstractBackground: Progranulin deficiency due to heterozygous null mutations in the GRN gene are a common cause offamilial frontotemporal lobar degeneration (FTLD), while homozygous loss-of-function GRN mutations are thoughtto be a rare cause of neuronal ceroid lipofuscinosis (NCL). Aged progranulin-knockout (Grn-null) mice display highlyexaggerated lipofuscinosis, microgliosis, and astrogliosis, as well as mild cell loss in specific brain regions. In thebrain, progranulin is predominantly expressed in neurons and microglia, and previously, we demonstrated thatneuronal-specific depletion of progranulin does not recapitulate the neuropathological phenotype of Grn-null mice.In this study, we evaluated whether selective depletion of progranulin expression in myeloid-lineage cells, includingmicroglia, causes NCL-like neuropathology or neuroinflammation in mice.Methods: We generated mice with progranulin depleted in myeloid-lineage cells by crossing mice homozygous fora floxed progranulin allele to mice expressing Cre recombinase under control of the LyzM promotor (Lyz-cKO).Results: Progranulin expression was reduced by approximately 50–70% in isolated microglia compared to WT levels.Lyz-cKO mice aged to 12 months did not display any increase in lipofuscin deposition, microgliosis, or astrogliosis in thefour brain regions examined, though increases were observed for many of these measures in Grn-null animals. Toevaluate the functional effect of reduced progranulin expression in isolated microglia, primary cultures were stimulatedwith controlled standard endotoxin and cytokine release was measured. While Grn-null microglia display ahyper-inflammatory phenotype, Lyz-cKO and WT microglia secreted similar levels of inflammatory cytokines.Conclusion: We conclude that progranulin expression from either microglia or neurons is sufficient to preventthe development of NCL-like neuropathology in mice. Furthermore, microglia that are deficient for progranulinexpression but isolated from a progranulin-rich environment have a normal inflammatory profile. Our resultssuggest that progranulin acts, at least partly, in a non-cell autonomous manner in the brain.Keywords: Frontotemporal lobar degeneration, Neuronal ceroid lipofuscinosis, Progranulin, Conditional knockoutmice, Neuropathology, Lysozyme promotor, Microglia* Correspondence: bleavitt@cmmt.ubc.ca1Centre for Molecular Medicine and Therapeutics, Department of MedicalGenetics, University of British Columbia, and Children’s and Women’sHospital, 980 West 28th Avenue, Vancouver, BC V5Z 4H4, Canada2Division of Neurology, Department of Medicine, University of BritishColumbia Hospital, S 192 - 2211 Wesbrook Mall, Vancouver, BC V6T 2B5,CanadaFull list of author information is available at the end of the article© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.Petkau et al. Journal of Neuroinflammation  (2017) 14:225 DOI 10.1186/s12974-017-1000-9BackgroundLoss-of-function mutations in the progranulin (GRN)gene cause neurological disease in patients, typically fron-totemporal lobar degeneration (FTLD) for heterozygous-null mutations [1, 2], and neuronal ceroid lipofuscinosis(NCL) in the rare case of homozygous-null mutations [3].Neuropathological analysis of patients with GRN-dependent FTLD reveals neuronal cell loss primarilyaffecting the frontal and temporal lobes of the brain, in-creased microgliosis in affected brain regions, and TDP-43 pathology [4]. Typical neuropathological features inNCL include early and robust microgliosis, astrogliosis,and lipofuscinosis in the thalamus, which spreads to otherbrain regions and is ultimately followed by extensive neur-onal cell loss [5]. NCL neuropathology is not yet con-firmed in GRN-null patients [3] but inferred based on theconsistency of clinicopathological features shared by GRNmutation carriers and patients with other genetic causesof NCL and the consistency of neuropathological featuresseen in other forms of NCL, their respective mousemodels, and Grn-null mice (reviewed in [6]).Progranulin, a secreted glycoprotein with ubiquitousexpression and pleiotropic actions in the body [7], isexpressed in most neuronal populations and in microgliain the brain [8] and plays a role in lysosome biology [9].Mice constitutively null for the homologous murine pro-granulin gene (Grn) display robust neuropathologicalfeatures consistent with NCL, including microgliosis,astrocytosis, and exaggerated deposition of NCL-likeautofluorescent pigment and/or lipofuscin, occurringearliest and most notably in the thalamus and laterbecoming widespread throughout the brain [10–15]. Be-havioral changes in Grn-null mice are modest and some-what inconsistent, though social dominance deficits[16–18] and obsessive compulsive-like (OCD-like) be-haviors are consistently observed [18].The relative contributions of neuron-derived andmicroglia-derived progranulin to behavioral and neuro-pathological phenotypes are a recently emerging area ofinvestigation. A deficit in social dominance is the onlyreported behavioral change in heterozygous Grn-nullmice [17], a phenotype that is recapitulated in neuron-specific Grn knockout mice [16]. Notably, the neuro-pathological features that robustly define Grn-null miceare not observed in heterozygous Grn-null nor neuronal-specific Grn knockout mice [16, 19].With respect to microglia, increased self-grooming, anOCD-like behavior that is partially regulated by the inflam-matory cytokine TNFα, was recapitulated in mice with pro-granulin knocked down specifically in microglia [18].Importantly, this study also observed deficits in socialbehavior that were not present in microglia-specific Grn-knockout mice, reaffirming that behavioral phenotypes canarise due to progranulin deficiency in a single cell type.In this study, we use mice with selective depletion ofprogranulin in myeloid-lineage cells, including microglia,to evaluate the NCL-like neuropathological phenotypesobserved in constitutive Grn-null mice. We find no evi-dence of neuropathological changes in myeloid-specificGrn-targeted mice despite robust neuropathology inGrn-null mice. The overall reduction of progranulin ex-pression in the brain in myeloid-specific Grn-targetedmice was moderate, which led us to examine geneexpression changes specific to microgliosis and lyso-somal dysfunction, as well as the inflammatory pheno-type of primary microglia cultures in response to astimulus, in myeloid-specific Grn-targeted mice. In allcases, we measured robust changes in Grn-null mice butnot in myeloid-specific Grn-targeted mice. We provideevidence that some progranulin-dependent phenotypesare non-cell autonomous, adding an additional level ofcomplexity to progranulin biology in the brain.ResultsProgranulin expression is decreased in Lyz-cKO mousemicrogliaProgranulin is primarily expressed in both neurons andmicroglia in the brain [8]. Although cellular levels ofmicroglial progranulin expression are relatively higherthan neuronal expression levels [8, 20], microglia ac-count for a much smaller proportion of cells in thebrain. We measured overall brain levels of Grn mRNAand protein and found that progranulin is reduced byapproximately 20–30% in Lyz-cKO mice (Fig. 1a, b).This moderate reduction in overall levels was ex-pected; however, to verify more robust knockdown inthe cell type of interest, we isolated microglia fromthe brains of 3-month-old mice by flow cytometryand again measured Grn mRNA and protein levels. Inisolated microglia, progranulin expression was re-duced by approximately 70% compared to Ctrl at themRNA level and approximately 50% of Ctrl at theprotein level (Fig. 1c, d).Neuropathology in Lyz-cKO mice does not replicate thatof GrnKO miceComplete loss of progranulin expression in the braincauses neuropathology in mice that mimics that of NCL;namely, exaggerated deposition of autofluorescent stor-age material and lipofuscin, as well as increased micro-gliosis and astrogliosis, which occur throughout thebrain but are most prominent in the thalamus [12]. Wequantified the amount of autofluorescence in four differ-ent brain regions as a surrogate for NCL-like storagematerial/lipofuscin accumulation. Autofluorescent ma-terial is detectable in the thalamus of Ctrl animals and issignificantly increased in GrnKO animals (Fig. 2a). Inthe thalamus of Lyz-cKO mice, the level ofPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 2 of 11autofluorescence was comparable to that of Ctrl mice.Similar results were seen in the CA3 region of thehippocampus (Fig. 2b) and striatum (Fig. 2d). In the cor-tex, the level of autofluorescence showed a similar trend,though the increase in GrnKO did not reach statisticalsignificance (Fig. 2c).Microgliosis was assessed by quantifying Iba1 immu-noreactivity in the same four brain regions. As expected,the thalamus showed a significant increase in Iba1 im-munoreactivity in GrnKO mice compared to Ctrl con-trols, but the level of Iba1 staining in Lyz-cKO animalswas similar to that of Ctrl mice (Fig. 3a). In the CA3 re-gion of the hippocampus, Iba1 immunoreactivity was in-creased relative to Ctrl in both Lyz-cKO and GrnKOmice (Fig. 3b). In the cortex, no significant differences inIba1 staining were observed (Fig. 3c), while in the stri-atum, we observed a significant increase in Iba1 stainingin GrnKO animals but not in Lyz-cKO animals com-pared to Ctrl mice (Fig. 3d).Finally, we quantified GFAP immunoreactivity toevaluate astrocytosis in the brain. In both the thalamus(Fig. 4a) and cortex (Fig. 4c), GFAP immunoreactivitywas significantly increased in GrnKO mice compared toCtrl, while in Lyz-cKO mice, there was no increase com-pared to Ctrl mice. No difference in GFAP staining wasobserved between the three genotypes in the CA3 regionof the hippocampus (Fig. 4b). In the striatum, there wassignificantly increased GFAP staining in GrnKO micecompared to Lyz-cKO mice (Fig. 4d), though the in-crease was not significantly different from Ctrl mice,where the quantity of GFAP staining was highly variable.Overall, the neuropathological phenotype of GrnKOmice is consistent and robust, with increased depositionof lipofuscin/NCL-like storage material, increasedmicrogliosis, and increased astrogliosis all being particu-larly apparent in the thalamus at 12 months of age. Wepreviously showed that these neuropathological changesare not present when progranulin levels are knockeddown in neuronal cells [19], and the present data nowshow that selective depletion of progranulin in microgliais also not sufficient to recapitulate this phenotype.Changes in gene expression in the brains of aged GrnKOmice are not present in Lyz-cKO miceBecause the depletion of progranulin in the brain in Lyz-cKO animals was not complete, we sought to evaluateadditional phenotypic changes related to microglial func-tion using a more sensitive method in older mice. Tothis end, we evaluated the expression of a panel of sixcell-type specific markers and lysosomal proteins in thethalamus of 18-month-old mice by quantitative RT-PCR.We observed significantly increased expression of CD68,a lysosomal protein used as a marker of activated micro-glia in the brain, in GrnKO mice compared to Ctrl mice,but not in Lyz-cKO mice (Fig. 5a). Lysosomal-associatedmembrane proteins 1 and 2 (Lamp1 and Lamp2) arelysosomal proteins which are both robustly expressed inthe brain [21]. Lamp2 expression was increased inGrnKO but not in Lyz-cKO mice, while Lamp1 expres-sion was not significantly different among the threegenotypes (Fig. 5b, c). Evaluation of CD11b expression,an integral membrane protein robustly expressed byCtrl Lyz-cKO GrnKO050100150Grn (% Ctrl)Whole brain proteinCtrl Het Lyz-cKO050100150Grn (% Ctrl)Isolated microglia proteinCtrl Lyz-cKO GrnKO050100150Cortex mRNAGrn expression (% Ctrl)Ctrl Het Lyz-cKO050100150Grn expression (% Ctrl)Isolated microglia mRNAa bc dFig. 1 Grn expression is significantly reduced in microglia in Lyz-cKO mice. a Grn measured in whole brain lysate by ELISA. N = 10 Ctrl, 6 Lyz-cKO,and 4 GrnKO mice. b Grn mRNA measured by qRT-PCR in cortex RNA samples. N = 4 Ctrl, 3 Lyz-cKO, and 2 GrnKO mice. c Grn measured by ELISAin adult microglia isolated by flow cytometry. Microglia from 4 mice per genotype were pooled and run as a single sample in duplicate. d Grn mRNA mea-sured by qRT-PCR in RNA extracted from adult microglia isolated by flow cytometry. N= 4 Ctrl, 4 Het, and 4 Lyz-cKO mice. Data represent mean ± SEMPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 3 of 11microglia, showed no differences between Ctrl, Lyz-cKO,and GrnKO mice (Fig. 5d), but Iba1 expression was sig-nificantly increased in GrnKO mice, but not Lyz-cKOmice, compared to Ctrl mice (Fig. 5e). GFAP mRNAexpression was more variable than other transcriptsmeasured, though we did observe an increase in GrnKOmice which was statistically significant compared to Lyz-cKO, though not Ctrl, mice (Fig. 5f ).Overall, our results demonstrate increased microgliosisin the thalamus in GrnKO mice that is not re-capitulated in Lyz-cKO mice. Increased expression ofIba1 mRNA was observed in Lyz-cKO brains, but giventhat other microglia markers did not show the sametrend, this may be an effect of Cre expression in micro-glia and not specific to progranulin knockdown.Selective depletion of progranulin in microglia does notrecapitulate the hyper-inflammatory phenotype observedin microglia isolated from GrnKO miceResults from both our previous study examiningneuronal-specific knockout mice [19] and from thecurrent study suggest that progranulin acts non-cellautonomously in the brain, that is, that progranulin pro-duced in a single cell type can suppress progranulin-dependent phenotypes in other cell types. However, inLyz-cKO microglia, progranulin knockdown was incom-plete, leaving open the possibility that the residual pro-granulin expression is sufficient to maintain normalmicroglia function. We therefore sought to (a) create amouse model where progranulin expression in microgliawas nearly absent and then (b) evaluate isolated primarymicroglia for a progranulin-dependent phenotype. Tothis end, we crossed Lyz-cKO mice to GrnKO mice toproduce Grnflox/KO; Lyz+/cre mice (hereon referred to asKO-Lyz-cKO mice). Primary microglia were culturedfrom early post-natal WT and WT-Cre (Ctrl), Het, Lyz-cKO, KO-Lyz-cKO, and GrnKO mice. Progranulin se-creted from cultured microglia after 24 h was easilydetectable by ELISA (Fig. 6a) and expectedly absent inGrnKO cultures. In Lyz-cKO cultures, progranulin ex-pression was reduced to less than 10% that of Ctrlmicroglia (Fig. 6a), while in KO-Lyz-cKO cultures,Fig. 2 Exaggerated lipofuscin deposition in GrnKO mice is not recapitulated in Lyz-cKO mice. Representative images of autofluorescence, a surro-gate for lipofuscin/NCL-like storage material, in the brains of 12-month-old WT, Lyz-cKO, and GrnKO mice are shown for the thalamus (a), CA3region of the hippocampus (b), cortex (c), and striatum (d). In each case, quantification of the images is given in the right-most panel. Data ispresented as the average integrated optical density (IOD) over the area measured in arbitrary units (a.u.). For each mouse, 4–6 images per regionwere measured and averaged to give a single value per animal per region. N = 5 Ctrl, 12 Lyz-cKO, and 4 GrnKO mice. Data represent mean ± SEM.p values were calculated using Tukey’s multiple comparison test after one-way ANOVAPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 4 of 11progranulin levels were not detectable above background(Fig. 6a). Progranulin mRNA, intracellular protein, andsecreted protein levels correlate with each other andaccurately reflect gene dosage in microglia derived fromheterozygous Grn-null mice (see Additional file 1: FigureS1), providing indirect evidence that recombination ofthe floxed Grn allele was near complete in this experi-ment. We then stimulated primary microglia cultureswith controlled standard endotoxin (CSE) and measuredIL-6 secretion into the media 24 h later. In unstimulatedcultures from all genotypes, IL-6 was undetectable in theconditioned media (data not shown). After CSE stimula-tion, IL-6 was reliably detected by ELISA in conditionedmedia. GrnKO primary microglia secrete significantlymore IL-6 than Ctrl cultures (Fig. 6b). Primary microgliagenerated from Het, Lyz-cKO, and KO-Lyz-cKO micesecreted similar levels of IL-6 as Ctrl microglia after CSEstimulation (Fig. 6b). Thus, despite near completeknockdown of progranulin expression, primary microgliaderived from Lyz-cKO or KO-Lyz-cKO mice do notrecapitulate the hyper-inflammatory phenotype observedin GrnKO microglia.DiscussionIn the present study, we show that reduced progranulinexpression in microglia is not sufficient to recapitulatethe NCL-like neuropathological features, nor the geneexpression changes present in the brains of Grn-nullmice. Furthermore, we show that the acute hyper-inflammatory phenotype of isolated Grn-null primarymicroglia is not recapitulated in progranulin-deficientmicroglia isolated from myeloid-specific knockout mice,despite near complete ablation of progranulin expressionin cultured microglia. These results, combined with ourpreviously published work showing that neuronal-specific knockdown of progranulin is also not sufficientto reproduce the neuropathological features of Grn-nullanimals, strongly support a non-cell autonomous rolefor progranulin in the brain.Some cell-autonomous functions for progranulin haverecently been reported. Deficits in social dominance arerecapitulated in neuronal-specific Grn-knockout mice[16], while increased self-grooming is recapitulated inmicroglia-specific Grn-knockout mice [18], suggestingthat each of these specific behaviors is dependent onFig. 3 Increased microgliosis present in GrnKO mice is not present in Lyz-cKO mice. Representative images of Iba1 immunoreactivity in the brainsof 12-month-old WT, Lyz-cKO, and GrnKO mice are shown for the thalamus (a), CA3 region of the hippocampus (b), cortex (c), and striatum (d). Ineach case, quantification of the images is given in the right-most panel. For each mouse, 4–6 images per region were measured and averaged togive a single value per animal per region. N = 5 Ctrl, 12 Lyz-cKO, and 4 GrnKO mice. Data represent mean ± SEM. p values were calculated usingTukey’s multiple comparison test after one-way ANOVAPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 5 of 11progranulin expression in the given cell type. For the ro-bust neuropathological phenotypes that characterizeaged Grn-null mice, it is clear that reduced progranulinexpression in neurons or in microglia is not sufficient torecapitulate the observed changes.Lipofuscin or NCL-like storage material accumulationoccurs primarily in post-mitotic neurons and is the re-sult of lysosomal dysfunction [22]. Since neuronal-specific Grn-knockout mice do not display exaggeratedlipofuscinosis [16, 19], it appears that extracellularly de-rived progranulin, presumably from microglia, acts toreplace neuron-derived progranulin and maintain lyso-somal function. Conversely, since microglia-specificknockdown of progranulin has no effect on lipofuscino-sis, it seems that neuron-derived progranulin is also suf-ficient to maintain normal lysosomal function. Thepossibility remains that an alternative cell type, eithercentral or peripheral, is responsible for the phenotype ofGrn-null mice or for producing progranulin in the brainwhen expression is knocked down in neurons or micro-glia. To test this hypothesis, a mouse model with specificdeletion of progranulin in both neurons and microgliamight be warranted. In addition, we cannot exclude thepossibility that other proteins are able to compensate forthe loss of progranulin in our model system.It remains unclear whether progranulin, similar to pro-saposin [23], can be derived via sorting at the trans-Golgi network (biosynthetic pathway) as well as fromthe extracellular space (endocytic pathway) in order tomaintain lysosomal function, or whether it acts strictlyvia the endocytic pathway, independent of the cell typeit is derived from. Also unclear is whether progranulin’srole in lysosomal biology is its only role in neurons andwhether or not lysosomal dysfunction is a driving patho-logical force in the development of FTLD. Accumulationof autofluorescent storage material in the tissues ofFTLD patients has recently been reported [24, 25], but adirect connection between storage material accumula-tion, lysosomal function, and the pathophysiology ofFTLD is not yet established. Lipofuscin accumulationmay occur with partial loss of progranulin expression[16, 24, 25], but the dramatic increase in accumulationFig. 4 Increased astrogliosis present in GrnKO mice is not present in Lyz-cKO mice. Representative images of GFAP immunoreactivity in the brainsof 12-month-old WT, Lyz-cKO, and GrnKO mice are shown for the thalamus (a), CA3 region of the hippocampus (b), cortex (c), and striatum (d). Ineach case, quantification of the images is given in the right-most panel. For each mouse, 4–6 images per region were measured and averaged togive a single value per animal per region. N = 5 Ctrl, 12 Lyz-cKO, and 4 GrnKO mice. Data represent mean ± SEM. p values were calculated usingTukey’s multiple comparison test after one-way ANOVAPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 6 of 11Ctrl Lyz-Cre GrnKO0. expression  (NF3)CD68Ctrl Lyz-Cre GrnKO0. expression  (NF3)LAMP1Ctrl Lyz-Cre GrnKO0. expression  (NF3)GFAPCtrl Lyz-Cre GrnKO0. expression  (NF3)LAMP2Ctrl Lyz-Cre GrnKO0. expression  (NF3)Iba1Ctrl Lyz-Cre GrnKO0. expression  (NF3)CD11bp < 0.001p < 0.01p < 0.05p < 0.01a bc de fFig. 5 Gene expression changes present in aged GrnKO mice are not recapitulated in Lyz-cKO mice. Evaluation of mRNA expression levels assessed byqRT-PCR of a CD68, b Lamp2, c Lamp1, d CD11b, e Iba1, and f GFAP in the thalamus of 18-month-old mice. N = 6 Ctrl, 8 Lyz-cKO, and 8 GrnKO mice.Data represent mean ± SEM. p values were calculated using Tukey’s multiple comparison test after one-way ANOVAWT/WT-CreHetLyz-cKOKO-Lyz-cKOGrnKO050100150GrnGrn (% WT)WT/WT-CreHetLyz-cKOKO-Lyz-cKOGrnKO050100150200250300350IL-6 (%WT)IL-6p < 0.001a bFig. 6 Isolated primary microglia from GrnKO mice display a hyper-inflammatory phenotype that is not present in Lyz-cKO mice. a Measurementof Grn by ELISA in the conditioned media from primary microglia cultures. b Measurement of the inflammatory cytokine IL-6 in the conditionedmedia of primary microglia cultures after 24 h of stimulation with controlled standard endotoxin (CSE). N = 4–9 wells per genotype derived fromtwo independent experiments. Genotypes are defined as follows: WT: Grn+/+, Lyz+/+; WT-Cre: Grn+/+, Lyzcre/cre; Het: Grn+/−; Lyz-cKO: Grnflox/flox,Lyzcre/cre; KO-Lyz-cKO: Grnflox/−, Lyzcre/cre; GrnKO: Grn−/−. Data represent mean ± SEM. p values were calculated using Tukey’s multiple comparisontest after one-way ANOVAPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 7 of 11in Grn-null mice and the dramatically different clinicalpresentation of GRN-null patients with NCL comparedto heterozygous GRN-null patients with FTLD do notsupport a direct dose dependence of these phenotypeson progranulin expression levels. The surprising obser-vation that progranulin expression is actually increasedin affected brain regions in FLTD patients with GRNmutations due to increased expression from activatedmicroglia [26] strongly supports the hypothesis that par-tial loss of GRN expression in neurons plays a cell-intrinsic role critical to sustained neuronal health.It is important to note that progranulin depletion incells of myeloid lineage, which include microglia, was in-complete in this study, and thus, there remains the pos-sibility that more complete cell type-specific knockdownof progranulin could lead to neuropathological changesin the brain. The LyzM promotor driving Cre recombin-ase expression produces incomplete recombination inmicroglia [27], similar to the level of recombination weobserved in this study (Fig. 1). Future studies might fur-ther evaluate the effect of reducing progranulin expres-sion in microglia using a promotor such as Cx3cr1driving Cre expression for more complete knockdown[27]. The Lyz-cKO mice used in this study still displayedreduced progranulin expression in microglia, and ourdata clearly show that this intervention is insufficient torecapitulate the NCL-like neuropathology that is charac-teristic of Grn-null mice.We achieved much more complete knockdown ofprogranulin in isolated primary microglia cultures, inparticular from KO-Lyz-cKO mice, and even in thisinstance, did not observe the same phenotype that isobserved in Grn-null microglia. The lack of a hyper-inflammatory phenotype in isolated microglia withselective depletion of Grn, despite a robust phenotypein constitutive Grn-null microglia, cannot be attrib-uted to the presence of secreted progranulin derivedfrom other cell types as no other cell types arepresent in this system. Instead, it may be that micro-glia that develop in a progranulin-rich environmenthave a normal response to an inflammatory stimulus.It has been reported that AAV-mediated over-expression of progranulin in Grn-null microgliareduced cytokine secretion after inflammatory stimu-lation [20], indicating a specific role for intrinsic pro-granulin expression in microglia in modulatingcytokine expression. Still, there may be a criticalperiod in microglia development when exposure tocirculating progranulin plays a role in shaping part ofthe inflammatory response mechanisms, and that afterthis critical period, exposure to progranulin is dis-pensable for normal cytokine expression and release.This hypothesis has not yet been tested but remains apotential area of future investigation.ConclusionWe have shown that knockdown of progranulin inmyeloid-lineage cells including microglia is not sufficientto recapitulate the neuropathological abnormalities, geneexpression changes, or the hyper-inflammatory profile ofisolated microglia that are present in constitutive Grn-null mice. This data, when combined with our previouswork and that of others, suggests that progranulin actsnon-cell autonomously in the brain in some instancesand adds a layer of complexity to understandingprogranulin-dependent phenotypes in the brain.MethodsMiceThe generation of “floxed” progranulin-targeted (Floxed)mice was previously described [28]. Mice expressing Crerecombinase knocked in to the Lys2 locus (referred to asthe LysMcre allele) were obtained from The Jackson La-boratory (B6.129P2-Lyz2tm1(cre)Ifo/J). HomozygousGrnflox/flox mice were crossed to Lyz2+/cre mice, andresultant pups heterozygous at both loci were then inter-crossed to produce Grnflox/flox; Lyz2cre/cre mice. For someexperiments, homozygous Grnflox/flox; Lyz2cre/cre micewere crossed to Grn−/− animals [28], and resultant off-spring heterozygous at all three loci were intercrossed toproduce littermates of mixed genotypes. Final experi-mental cohorts were comprised of mice from homozy-gous matings of Grnflox/flox; Lyz2cre/cre mice (referred toas Lyz-cKO); heterozygous crosses of Grnflox/−; Lyz2+/crex Grnflox/−; Lyz2+/cre mice, from which Lyz-cKO, Floxed,and Grn−/− with or without the Cre transgene (referredto as GrnKO) were selected as experimental animals;and Grn+/− x Grn+/− crosses.As we have previously shown that Grn+/+ (WT) andGrnflox/flox (Floxed) mice are not significantly different inprogranulin expression levels or neuropathology [19], WTand Floxed mice were grouped together and referred to ascontrol (Ctrl) mice. Grn+/− (Het) mice were included forreference when evaluating progranulin levels. For experi-ments using primary microglia cultures, mice homozygousfor the LysMCre allele but wild-type at the Grn locus (WT-Cre) were used as controls to exclude the possibility thatexpression of Cre in microglia alters cytokine release.Genotyping was performed on tail tip DNA at wean andconfirmed on a second DNA sample at sacrifice using pri-mer sequences given in Table 1.Mice were housed on ventilated racks in specificpathogen-free barrier facility with a 12-h light/darkcycle. Mice were group-housed with their littermates toa maximum of four mice per cage.Isolation of adult murine microgliaMice were sacrificed by CO2 inhalation followed by cer-vical dislocation. Whole brains were removed and brieflyPetkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 8 of 11washed in 1 mL of Hank’s buffered saline solution(HBSS) before being placed in Liberase (0.1 M HBSS,47.7 μL/mL reconstituted Liberase). After being rotatedfor 45 min at 37 °C, brains were mechanically homoge-nized using a P1000 pipette tip until tissue was dissoci-ated, then spun at 200g for 5 min at 18 °C. Supernatantwas removed, and homogenates were re-suspended in2 mL of re-suspension buffer (0.1 M HBSS, 0.5 μL/mLfiltered MgCl2) and filtered through a 70 μm cellstrainer. The sample tube and cell strainer were bothwashed with additional re-suspension buffer, afterwhich the combined homogenates were spun at 200gfor 5 min at 18 °C.Following homogenization, the supernatant was re-moved, samples were re-suspended in FACS buffer(0.1 M PBS, 1 mM EDTA, 1% BSA), 500 μL of Miltenyi®Myelin Removal Beads II were added to the solution,and samples were incubated at 4 °C for 15 min. Sampleswere then washed with FACS buffer and spun at 200gfor 10 min at 18 °C. Supernatants were removed andpellets were re-suspended in FACS buffer prior to mye-lin depletion using the AutoMACS (Miltenyi; Deplete_Sprogram). Following the automated magnetic separation,the negative fraction was collected and spun at 200g for5 min at 18 °C.Following myelin removal, samples were incubatedwith 30 μL of Miltenyi® CD11b magnetic beads in270 μL of FACS buffer for 15 min at 4 °C, after whichthey were washed with 2 mL of FACS buffer and spun at200g for 10 min at 18 °C. Cells were re-suspended in500 μL of FACS buffer prior to AutoMACS selectionusing the Possel_S program, after which the CD11b-positive portion was collected. The samples were spun at200g for 5 min at 18 °C and re-suspended in 100 μL ofFACS buffer for subsequent antibody staining.The cells were incubated with Ebioscience® CD11b-PE and Ebioscience® CD45-APC antibodies at a dilutionof 1:1000 for 15 min at 4 °C. An additional 150 μL ofFACS buffer was added along with Ebioscience® 7AADViability Dye at a dilution of 1:250. The sample wasthen subjected to flow cytometry sorting using a FAC-SAria machine, with an 85 nozzle and 45 psi setting,with CD45low CD11b+ cells isolated as the populationof interest.Protein extraction and quantification of Grn by ELISAWhole brain lysate was prepared from ten WT and/orFloxed mice, six Lyz-cKO mice, and four GrnKO miceat 3–4 months of age by homogenizing previously snap-frozen brains in a rotor-stator homogenizer for 30 s in1 mL of complete lysis buffer (50 mM Tris-HCl, 1%Triton-X, 150 mM NaCl, Halt phosphatase inhibitorcocktail (Thermo Fisher Scientific), Halt protease inhibi-tor cocktail (Thermo Fisher Scientific)). Total proteinwas assayed using Bradford reagent (BioRad).Microglia isolated by flow cytometry from four WT,four Lyz-cKO, and four Het mice at 3–4 months of agewere pooled and lysed in 100 μL of complete lysis buffer,then stored at − 80 °C until used.The quantity of Grn in whole brain lysate, sorted celllysate, or conditioned media was determined by anenzyme-linked immunosorbent assay (ELISA) using acommercially available kit (Mouse progranulin ELISA;Adipogen, Korea). Microglia supernatant samples werediluted 1:5; for whole brain lysate, 100 μg of protein wasused; for cell lysates from sorted microglia, the entiresample minus a small aliquot for protein quantificationwas loaded and results normalized to total protein, typ-ically 5–10μg. All samples were run in duplicate. TheELISA was conducted according to the manufacturer’sinstructions. Data represent the average per condition,Table 1 Primer sequences for genotyping and quantitativeRT-PCRGenotypingGene Forward primer sequence Reverse primersequenceGrn Common:5’-CGGAACACAGTGTCCAGATG-3’Intron 2:5’-ATCAACCAAAGGGTCTGTGC-3’Exon 5:5’-GTGGCAGAGTCAGGACATTCAAACT-3’Lys2 WT: 5’-TTACAGTCGGCCAGGCTGAC-3’Cre: 5’-CCCAGAAATGCCAGATTACG-3’Common:5’- CTTGGGCTGCCAGAATTTCTC-3’Quantitative RT-PCRGene Forward primer sequence Reverse primersequenceCD11b 5’-ATCCCCCTGCAAGACAGTGA-3’5’-AGCAGTCAGGCAGGGACATG-3’CD68 5’-GTGCTCATCGCCTTCTGCATCA-3’5’-GGCGCTCCTTGGTGGCTTAC-3’Gfap 5’-GAGTGGTATCGGTCTAAGTTTGCA-3’5’-CGATAGTCGTTAGCTTCGTGCTT-3’Grn 5’-CTGTAGTGCAGATGGGAAATCCTGCT-3’5’-GTGGCAGAGTCAGGACATTCAAACT-3’Iba1 5’-GTCCTTGAAGCGAATGCTGG-3’5’-CATTCTCAAGATGGCAGATC-3’Lamp1 5’-ACATCAGCCCAAATGACACA-3’5’-GGCTAGAGCTGGCATTCATC-3’Lamp2 5’-AGCACAGTATTTCCTGGTGCT-3’5’-CGACAGGAGTCAGGTTGTAAGTTAA-3’RplI13a 5’-GGAGGAGAAACGGAAGGAAAAG-3’5’-CCGTAACCTCAAGATCTCGTTCTT-3’Usf1 5’-CCTGTGGCGTGGCAGTCT-3’5’-TGCACGCCCACACTGTTT-3’Zfp91 5’-TCCTGTGGGCGACTCTTCAG-3’5’-TAATCCCTCTGGTCTGTATGATGCT-3’Petkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 9 of 11and all conditions that were compared directly were runon the same plate.RNA isolation and qPCRFor analysis of Grn mRNA expression, whole brain tis-sue (four Floxed, three Lyz-cKO, and two GrnKO mice)or microglia isolated from adult brain (four WT, fourLyz-cKO, four GrnKO) from mice 3–4 months of agewere collected; for analysis of cell-type specific and lyso-somal transcripts, the thalamus was micro-dissectedfrom six WT and/or Floxed, eight Lyz-cKO, and eightGrnKO mice at 18 months of age. Tissue samples wereimmediately frozen at − 80 °C. Samples were homoge-nized with a bead homogenizer in lysis buffer followedby total RNA extraction (PureLink RNA mini kit; Invi-trogen) performed according to the manufacturer’sinstructions. Reverse transcription of all samples wascarried out using the Superscript VILO kit (Invitrogen)according to the manufacturer’s instructions, using 1 μgof total RNA as input for cDNA synthesis. Followingthis, cDNA was diluted 1:10 in ddH2O for a total inputof 5 ng into the quantitative PCR reaction, done usingFastSybr (Applied Biosystems), and conducted on aStep-One ABI System (Applied Biosystems). Quantifica-tion of mRNA levels was accomplished using the stand-ard curve method, with amplification of target mRNAand control genes in separate wells. Each sample wasrun in duplicate. The relative amount of mRNA in eachwell was calculated as the ratio between target mRNAand a normalization factor created using three controlgenes (Usf1, RplI13a, and Zfp91) based on GeNorm [29].Values are presented as % WT and/or Floxed control.All primer sequences are provided in Table 1.ImmunohistochemistryImmunohistochemistry was performed as previouslydescribed [19]. Briefly, 25 μm floating sections wereplaced in net-well inserts and washed for 10 min inphosphate-buffered saline (PBS). After quenching, en-dogenous peroxidase activity was quenched with 3%H2O2 for 45 min, sections were blocked in 5% normalserum and 5% bovine serum albumin, followed by over-night incubation shaking at room temperature in pri-mary antibody diluted in 5% normal serum. After two15 min washes, secondary antibody diluted in 1% normalserum and PBS with 0.01% Triton X (PBS-T) wasapplied for 2 h shaking at room temperature. Sectionswere washed for 30 min in PBS before an amplificationstep was performed using an avidin–biotin–horserad-ish peroxidase complex kit (Vector Laboratories). Col-orimetric detection was achieved with the peroxidasesubstrate kit Vector DAB (Vector Laboratories) ac-cording to the manufacturer’s instructions. Sectionswere mounted by hand on onto glass slides (Fisher-brand Superfrost Plus) and dried overnight before be-ing dehydrated through a series of alcohols andxylene, and cover-slipped with DEPEX (Electron Mi-croscopy Sciences). Antibodies used were as follows:the microglia marker Iba1 (Wako; 1:2000, rabbit poly-clonal), the astrocyte marker GFAP (Sigma; 1:2000,mouse monoclonal), and appropriate biotinylated sec-ondary antibodies (Vector, 1:2000).Sections to be assessed for autofluorescence weremounted onto glass slides, washed in PBS-T for 30 min,and then stained with DAPI in PBS at 1:10,000 for5 min. Slides were washed in twice in PBS for 5 minprior to coverslipping.Image acquisition and analysisImages were acquired as previously described [12]. In-tegrated optical density measurements of signal inten-sity were acquired as previously described [8] toquantify autofluorescence, a surrogate for lipofuscindeposition, in 4 images per mouse taken from 5 WT/Floxed, 12 Lyz-cKO, and 4 GrnKO mice. For quantifi-cation of colorimetric stains (Iba1 and GFAP), athreshold was set that pseudo-colored stained areaswithin a defined region of interest in each image. Theaverage percent thresholded area for 4–6 images permouse taken from 5 WT/Floxed, 12 Lyz-cKO, and 4GrnKO mice is reported.Primary microglia isolation and stimulationPrimary microglia cultures were generated from postna-tal day 0–3 pups as previously described [30]. Cells werestimulated with controlled standard endotoxin (finalconcentration 100 ng/ml; Associates of Cape Cod, MA,USA) as previously described [30], and conditionedmedia was collected after 24 h and stored at − 20 °Cuntil used. The quantity of IL-6 in the conditionedmedia was measured using a commercially availableELISA (Ready-set-go IL6 ELISA, eBiosciences, SanDiego, USA) according to the manufacturer’s instruc-tions. Results are presented from two independentexperiments with 3–6 wells per genotype in eachexperiment.Statistical analysisAll statistical comparisons were performed as a one-way analysis of variance (ANOVA) with Tukey post-hocanalysis to compare individual means to control andcorrect for multiple comparisons (Prism 6, GraphpadSoftware Inc.). A p value less than 0.05 was consideredsignificant.Petkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 10 of 11Additional fileAdditional file 1: Figure S1. Progranulin mRNA, intracellular, andsecreted protein levels correlate in primary microglia cultures. (A)Progranulin mRNA levels were reduced by approximately 50% in Hetcultures compared to WT cultures and not detectable in GrnKO cultures.N = 2–3 wells/per genotype. (B) Progranulin protein quantified by ELISAon cell lysate and normalized to total protein per well shows acorresponding decrease of about 50% in Het cultures compared toWT cultures and negligible levels in GrnKO cultures. N = 6 wells/genotype. (C) Secreted progranulin detected by ELISA on conditionedmedia again shows progranulin reduced to approximately 50% in Hetcultures compared to WT cultures and not detectable in GrnKOcultures. N = 6 wells/genotype. (PDF 90 kb)AcknowledgementsNot applicableFundingThis work was supported by the Alzheimer Society of Canada (doctoral traineeaward to TP), and by the Canadian Institute for Health Research (BRL operatinggrant #97857).Availability of data and materialAll data generated and analyzed are included in this published manuscript.Authors’ contributionsTP and KA completed experiments to quantify progranulin levels andneuropathological analysis. NK performed flow cytometry. CC generatedprimary microglia cultures and performed all related experiments. TP andBRL conceived of the experiments. TP drafted the manuscript, and allauthors contributed to revisions and approved of the final version.Ethics approvalAll animal procedures were done with the approval of the Canadian Councilfor Animal Care and the University of British Columbia’s Animal CareCommittee.Consent for publicationAll authors consent to the publication of this manuscript.Competing interestsThe authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Centre for Molecular Medicine and Therapeutics, Department of MedicalGenetics, University of British Columbia, and Children’s and Women’sHospital, 980 West 28th Avenue, Vancouver, BC V5Z 4H4, Canada. 2Divisionof Neurology, Department of Medicine, University of British ColumbiaHospital, S 192 - 2211 Wesbrook Mall, Vancouver, BC V6T 2B5, Canada. 3BrainResearch Centre, University of British Columbia, Vancouver, BC V6T 1Z3,Canada.Received: 5 September 2017 Accepted: 9 November 2017References1. Baker M, et al. Mutations in progranulin cause tau-negative frontotemporaldementia linked to chromosome 17. Nature. 2006;442(7105):916–9.2. Cruts M, Kumar-Singh S, Van Broeckhoven C. Progranulin mutations inubiquitin-positive frontotemporal dementia linked to chromosome 17q21.Curr Alzheimer Res. 2006;3(5):485–91.3. Smith KR, et al. Strikingly different clinicopathological phenotypesdetermined by progranulin-mutation dosage. Am J Hum Genet. 2012;90(6):1102–7.4. Mackenzie IRA. The neuropathology and clinical phenotype of FTD withprogranulin mutations. Acta Neuropathol. 2007;114(1):49–54.5. Anderson GW, Goebel HH, Simonati A. Human pathology in NCL. BiochimBiophys Acta. 2013;1832(11):1807–26.6. Kollmann K, et al. Cell biology and function of neuronal ceroidlipofuscinosis-related proteins. Biochim Biophys Acta. 2013;1832(11):1866–81.7. Bateman A, Bennett HP. The granulin gene family: from cancer to dementia.BioEssays. 2009;31(11):1245–54.8. Petkau TL, et al. Progranulin expression in the developing and adult murinebrain. J Comp Neurol. 2010;518(19):3931–47.9. Kao AW, et al. Progranulin, lysosomal regulation and neurodegenerativedisease. Nat Rev Neurosci. 2017;18(6):325–33.10. Ahmed Z, et al. Accelerated lipofuscinosis and ubiquitination in granulinknockout mice suggest a role for progranulin in successful aging. Am JPathol. 2010;177(1):311–24.11. Ghoshal N, et al. Core features of frontotemporal dementia recapitulated inprogranulin knockout mice. Neurobiol Dis. 2012;45(1):395–408.12. Petkau TL, Hill A, Leavitt BR. Core neuropathological abnormalities inprogranulin-deficient mice are penetrant on multiple genetic backgrounds.Neuroscience. 2016;315:175–95.13. Petkau TL, et al. Synaptic dysfunction in progranulin-deficient mice.Neurobiol Dis. 2012;45(2):711–22.14. Wils H, et al. Cellular ageing, increased mortality and FTLD-TDP-associatedneuropathology in progranulin knockout mice. J Pathol. 2012;228(1):67–76.15. Yin F, et al. Behavioral deficits and progressive neuropathology inprogranulin-deficient mice: a mouse model of frontotemporal dementia.FASEB J. 2010;24(12):4639–47.16. Arrant AE, et al. Restoring neuronal progranulin reverses deficits in a mousemodel of frontotemporal dementia. Brain. 2017;140:1447.17. Filiano AJ, et al. Dissociation of frontotemporal dementia-related deficitsand neuroinflammation in progranulin haploinsufficient mice. J Neurosci.2013;33(12):5352–61.18. Krabbe G, et al. Microglial NFkappaB-TNFalpha hyperactivation inducesobsessive-compulsive behavior in mouse models of progranulin-deficientfrontotemporal dementia. Proc Natl Acad Sci U S A. 2017;114(19):5029–34.19. Petkau TL, Blanco J, Leavitt BR. Conditional loss of progranulin in neurons isnot sufficient to cause neuronal ceroid lipofuscinosis-like neuropathology inmice. Neurobiol Dis. 2017;106:14–22.20. Martens LH, et al. Progranulin deficiency promotes neuroinflammationand neuron loss following toxin-induced injury. J Clin Invest. 2012;122(11):3955–9.21. Furuta K, et al. Differential expression of the lysosome-associated membraneproteins in normal human tissues. Arch Biochem Biophys. 1999;365(1):75–82.22. Jalanko A, Braulke T. Neuronal ceroid lipofuscinoses. Biochim Biophys Acta.2009;1793(4):697–709.23. Zhou X, et al. Impaired prosaposin lysosomal trafficking in frontotemporallobar degeneration due to progranulin mutations. Nat Commun. 2017;8:15277.24. Terlizzi R, et al. Muscle ceroid lipofuscin-like deposits in a patient withcorticobasal syndrome due to a progranulin mutation. Mov Disord. 2017;32:1259.25. Ward ME, et al., Individuals with progranulin haploinsufficiency exhibitfeatures of neuronal ceroid lipofuscinosis. Sci Transl Med. 2017;9(385).26. Chen-Plotkin AS, et al. Brain progranulin expression in GRN-associatedfrontotemporal lobar degeneration. Acta Neuropathol. 2010;119(1):111–22.27. Goldmann T, et al. A new type of microglia gene targeting shows TAK1 tobe pivotal in CNS autoimmune inflammation. Nat Neurosci. 2013;16(11):1618–26.28. Petkau TL, et al. Sensitivity to neurotoxic stress is not increased inprogranulin-deficient mice. Neurobiol Aging. 2013;34:2548.29. Vandesompele J, et al. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes.Genome Biol. 2002;3(7):RESEARCH0034.30. Connolly C, et al. Enhanced immune response to MMP3 stimulation inmicroglia expressing mutant huntingtin. Neuroscience. 2016;325:74–88.Petkau et al. Journal of Neuroinflammation  (2017) 14:225 Page 11 of 11


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