Background
Loss-of-function mutations in the progranulin (
GRN) gene cause neurological disease in patients, typically frontotemporal 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 primarily affecting the frontal and temporal lobes of the brain, increased microgliosis in affected brain regions, and TDP-43 pathology [
4]. Typical neuropathological features in NCL include early and robust microgliosis, astrogliosis, and lipofuscinosis in the thalamus, which spreads to other brain regions and is ultimately followed by extensive neuronal cell loss [
5]. NCL neuropathology is not yet confirmed in
GRN-null patients [
3] but inferred based on the consistency of clinicopathological features shared by
GRN mutation carriers and patients with other genetic causes of NCL and the consistency of neuropathological features seen in other forms of NCL, their respective mouse models, and Grn-null mice (reviewed in [
6]).
Progranulin, a secreted glycoprotein with ubiquitous expression and pleiotropic actions in the body [
7], is expressed in most neuronal populations and in microglia in the brain [
8] and plays a role in lysosome biology [
9]. Mice constitutively null for the homologous murine progranulin gene (
Grn) display robust neuropathological features consistent with NCL, including microgliosis, astrocytosis, and exaggerated deposition of NCL-like autofluorescent pigment and/or lipofuscin, occurring earliest and most notably in the thalamus and later becoming widespread throughout the brain [
10‐
15]. Behavioral changes in Grn-null mice are modest and somewhat inconsistent, though social dominance deficits [
16‐
18] and obsessive compulsive-like (OCD-like) behaviors are consistently observed [
18].
The relative contributions of neuron-derived and microglia-derived progranulin to behavioral and neuropathological phenotypes are a recently emerging area of investigation. A deficit in social dominance is the only reported behavioral change in heterozygous Grn-null mice [
17], a phenotype that is recapitulated in neuron-specific Grn knockout mice [
16]. Notably, the neuropathological features that robustly define Grn-null mice are not observed in heterozygous Grn-null nor neuronal-specific Grn knockout mice [
16,
19].
With respect to microglia, increased self-grooming, an OCD-like behavior that is partially regulated by the inflammatory cytokine TNFα, was recapitulated in mice with progranulin knocked down specifically in microglia [
18]. Importantly, this study also observed deficits in social behavior that were not present in microglia-specific Grn-knockout mice, reaffirming that behavioral phenotypes can arise due to progranulin deficiency in a single cell type.
In this study, we use mice with selective depletion of progranulin in myeloid-lineage cells, including microglia, to evaluate the NCL-like neuropathological phenotypes observed in constitutive Grn-null mice. We find no evidence of neuropathological changes in myeloid-specific Grn-targeted mice despite robust neuropathology in Grn-null mice. The overall reduction of progranulin expression in the brain in myeloid-specific Grn-targeted mice was moderate, which led us to examine gene expression changes specific to microgliosis and lysosomal dysfunction, as well as the inflammatory phenotype of primary microglia cultures in response to a stimulus, in myeloid-specific Grn-targeted mice. In all cases, we measured robust changes in Grn-null mice but not in myeloid-specific Grn-targeted mice. We provide evidence that some progranulin-dependent phenotypes are non-cell autonomous, adding an additional level of complexity to progranulin biology in the brain.
Discussion
In the present study, we show that reduced progranulin expression in microglia is not sufficient to recapitulate the NCL-like neuropathological features, nor the gene expression changes present in the brains of Grn-null mice. Furthermore, we show that the acute hyper-inflammatory phenotype of isolated Grn-null primary microglia is not recapitulated in progranulin-deficient microglia isolated from myeloid-specific knockout mice, despite near complete ablation of progranulin expression in cultured microglia. These results, combined with our previously published work showing that neuronal-specific knockdown of progranulin is also not sufficient to reproduce the neuropathological features of Grn-null animals, strongly support a non-cell autonomous role for progranulin in the brain.
Some cell-autonomous functions for progranulin have recently been reported. Deficits in social dominance are recapitulated in neuronal-specific Grn-knockout mice [
16], while increased self-grooming is recapitulated in microglia-specific Grn-knockout mice [
18], suggesting that each of these specific behaviors is dependent on progranulin expression in the given cell type. For the robust neuropathological phenotypes that characterize aged Grn-null mice, it is clear that reduced progranulin expression in neurons or in microglia is not sufficient to recapitulate the observed changes.
Lipofuscin or NCL-like storage material accumulation occurs primarily in post-mitotic neurons and is the result of lysosomal dysfunction [
22]. Since neuronal-specific Grn-knockout mice do not display exaggerated lipofuscinosis [
16,
19], it appears that extracellularly derived progranulin, presumably from microglia, acts to replace neuron-derived progranulin and maintain lysosomal function. Conversely, since microglia-specific knockdown of progranulin has no effect on lipofuscinosis, it seems that neuron-derived progranulin is also sufficient to maintain normal lysosomal function. The possibility remains that an alternative cell type, either central or peripheral, is responsible for the phenotype of Grn-null mice or for producing progranulin in the brain when expression is knocked down in neurons or microglia. To test this hypothesis, a mouse model with specific deletion of progranulin in both neurons and microglia might be warranted. In addition, we cannot exclude the possibility that other proteins are able to compensate for the loss of progranulin in our model system.
It remains unclear whether progranulin, similar to prosaposin [
23], can be derived via sorting at the trans-Golgi network (biosynthetic pathway) as well as from the extracellular space (endocytic pathway) in order to maintain lysosomal function, or whether it acts strictly via the endocytic pathway, independent of the cell type it is derived from. Also unclear is whether progranulin’s role in lysosomal biology is its only role in neurons and whether or not lysosomal dysfunction is a driving pathological force in the development of FTLD. Accumulation of autofluorescent storage material in the tissues of FTLD patients has recently been reported [
24,
25], but a direct connection between storage material accumulation, lysosomal function, and the pathophysiology of FTLD is not yet established. Lipofuscin accumulation may occur with partial loss of progranulin expression [
16,
24,
25], but the dramatic increase in accumulation in Grn-null mice and the dramatically different clinical presentation of
GRN-null patients with NCL compared to heterozygous
GRN-null patients with FTLD do not support a direct dose dependence of these phenotypes on progranulin expression levels. The surprising observation that progranulin expression is actually increased in affected brain regions in FLTD patients with
GRN mutations due to increased expression from activated microglia [
26] strongly supports the hypothesis that partial loss of
GRN expression in neurons plays a cell-intrinsic role critical to sustained neuronal health.
It is important to note that progranulin depletion in cells of myeloid lineage, which include microglia, was incomplete in this study, and thus, there remains the possibility that more complete cell type-specific knockdown of progranulin could lead to neuropathological changes in the brain. The LyzM promotor driving Cre recombinase expression produces incomplete recombination in microglia [
27], similar to the level of recombination we observed in this study (Fig.
1). Future studies might further evaluate the effect of reducing progranulin expression in microglia using a promotor such as Cx3cr1 driving Cre expression for more complete knockdown [
27]. The Lyz-cKO mice used in this study still displayed reduced progranulin expression in microglia, and our data clearly show that this intervention is insufficient to recapitulate the NCL-like neuropathology that is characteristic of Grn-null mice.
We achieved much more complete knockdown of progranulin in isolated primary microglia cultures, in particular from KO-Lyz-cKO mice, and even in this instance, did not observe the same phenotype that is observed in Grn-null microglia. The lack of a hyper-inflammatory phenotype in isolated microglia with selective depletion of Grn, despite a robust phenotype in constitutive Grn-null microglia, cannot be attributed to the presence of secreted progranulin derived from other cell types as no other cell types are present in this system. Instead, it may be that microglia that develop in a progranulin-rich environment have a normal response to an inflammatory stimulus. It has been reported that AAV-mediated over-expression of progranulin in Grn-null microglia reduced cytokine secretion after inflammatory stimulation [
20], indicating a specific role for intrinsic progranulin expression in microglia in modulating cytokine expression. Still, there may be a critical period in microglia development when exposure to circulating progranulin plays a role in shaping part of the inflammatory response mechanisms, and that after this critical period, exposure to progranulin is dispensable for normal cytokine expression and release. This hypothesis has not yet been tested but remains a potential area of future investigation.
Methods
Mice
The generation of “floxed” progranulin-targeted (Floxed) mice was previously described [
28]. Mice expressing Cre recombinase knocked in to the Lys2 locus (referred to as the LysMcre allele) were obtained from The Jackson Laboratory (B6.129P2-
Lyz2
tm1(cre)Ifo
/J). Homozygous Grn
flox/flox mice were crossed to Lyz2
+/cre mice, and resultant pups heterozygous at both loci were then intercrossed to produce Grn
flox/flox; Lyz2
cre/cre mice. For some experiments, homozygous Grn
flox/flox; Lyz2
cre/cre mice were crossed to Grn
−/− animals [
28], and resultant offspring heterozygous at all three loci were intercrossed to produce littermates of mixed genotypes. Final experimental cohorts were comprised of mice from homozygous matings of Grn
flox/flox; Lyz2
cre/cre mice (referred to as Lyz-cKO); heterozygous crosses of Grn
flox/−; Lyz2
+/cre x Grn
flox/−; Lyz2
+/cre mice, from which Lyz-cKO, Floxed, and Grn
−/− with or without the Cre transgene (referred to as GrnKO) were selected as experimental animals; and Grn
+/− x Grn
+/− crosses.
As we have previously shown that Grn
+/+ (WT) and Grn
flox/flox (Floxed) mice are not significantly different in progranulin expression levels or neuropathology [
19], WT and Floxed mice were grouped together and referred to as control (Ctrl) mice. Grn
+/− (Het) mice were included for reference when evaluating progranulin levels. For experiments using primary microglia cultures, mice homozygous for the LysMCre allele but wild-type at the Grn locus (WT-Cre) were used as controls to exclude the possibility that expression of Cre in microglia alters cytokine release. Genotyping was performed on tail tip DNA at wean and confirmed on a second DNA sample at sacrifice using primer sequences given in Table
1.
Table 1
Primer sequences for genotyping and quantitative RT-PCR
Genotyping |
Gene | Forward primer sequence | Reverse primer sequence |
Grn | 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-PCR |
Gene | Forward primer sequence | Reverse primer sequence |
CD11b | 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’ |
Mice were housed on ventilated racks in specific pathogen-free barrier facility with a 12-h light/dark cycle. Mice were group-housed with their littermates to a maximum of four mice per cage.
Isolation of adult murine microglia
Mice were sacrificed by CO2 inhalation followed by cervical dislocation. Whole brains were removed and briefly washed 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 rotated for 45 min at 37 °C, brains were mechanically homogenized using a P1000 pipette tip until tissue was dissociated, then spun at 200g for 5 min at 18 °C. Supernatant was removed, and homogenates were re-suspended in 2 mL of re-suspension buffer (0.1 M HBSS, 0.5 μL/mL filtered MgCl2) and filtered through a 70 μm cell strainer. The sample tube and cell strainer were both washed with additional re-suspension buffer, after which the combined homogenates were spun at 200g for 5 min at 18 °C.
Following homogenization, the supernatant was removed, 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. Samples were then washed with FACS buffer and spun at 200g for 10 min at 18 °C. Supernatants were removed and pellets were re-suspended in FACS buffer prior to myelin depletion using the AutoMACS (Miltenyi; Deplete_S program). Following the automated magnetic separation, the negative fraction was collected and spun at 200g for 5 min at 18 °C.
Following myelin removal, samples were incubated with 30 μL of Miltenyi® CD11b magnetic beads in 270 μL of FACS buffer for 15 min at 4 °C, after which they were washed with 2 mL of FACS buffer and spun at 200g for 10 min at 18 °C. Cells were re-suspended in 500 μL of FACS buffer prior to AutoMACS selection using the Possel_S program, after which the CD11b-positive portion was collected. The samples were spun at 200g for 5 min at 18 °C and re-suspended in 100 μL of FACS buffer for subsequent antibody staining.
The cells were incubated with Ebioscience® CD11b-PE and Ebioscience® CD45-APC antibodies at a dilution of 1:1000 for 15 min at 4 °C. An additional 150 μL of FACS buffer was added along with Ebioscience® 7AAD Viability Dye at a dilution of 1:250. The sample was then subjected to flow cytometry sorting using a FACSAria machine, with an 85 nozzle and 45 psi setting, with CD45low CD11b+ cells isolated as the population of interest.
Protein extraction and quantification of Grn by ELISA
Whole brain lysate was prepared from ten WT and/or Floxed mice, six Lyz-cKO mice, and four GrnKO mice at 3–4 months of age by homogenizing previously snap-frozen brains in a rotor-stator homogenizer for 30 s in 1 mL of complete lysis buffer (50 mM Tris-HCl, 1% Triton-X, 150 mM NaCl, Halt phosphatase inhibitor cocktail (Thermo Fisher Scientific), Halt protease inhibitor cocktail (Thermo Fisher Scientific)). Total protein was 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 age were 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 cell lysate, or conditioned media was determined by an enzyme-linked immunosorbent assay (ELISA) using a commercially available kit (Mouse progranulin ELISA; Adipogen, Korea). Microglia supernatant samples were diluted 1:5; for whole brain lysate, 100 μg of protein was used; for cell lysates from sorted microglia, the entire sample minus a small aliquot for protein quantification was loaded and results normalized to total protein, typically 5–10μg. All samples were run in duplicate. The ELISA was conducted according to the manufacturer’s instructions. Data represent the average per condition, and all conditions that were compared directly were run on the same plate.
RNA isolation and qPCR
For analysis of
Grn mRNA expression, whole brain tissue (four Floxed, three Lyz-cKO, and two GrnKO mice) or microglia isolated from adult brain (four WT, four Lyz-cKO, four GrnKO) from mice 3–4 months of age were collected; for analysis of cell-type specific and lysosomal transcripts, the thalamus was micro-dissected from six WT and/or Floxed, eight Lyz-cKO, and eight GrnKO mice at 18 months of age. Tissue samples were immediately frozen at − 80 °C. Samples were homogenized with a bead homogenizer in lysis buffer followed by total RNA extraction (PureLink RNA mini kit; Invitrogen) performed according to the manufacturer’s instructions. Reverse transcription of all samples was carried out using the Superscript VILO kit (Invitrogen) according to the manufacturer’s instructions, using 1 μg of total RNA as input for cDNA synthesis. Following this, cDNA was diluted 1:10 in ddH
2O for a total input of 5 ng into the quantitative PCR reaction, done using FastSybr (Applied Biosystems), and conducted on a Step-One ABI System (Applied Biosystems). Quantification of mRNA levels was accomplished using the standard curve method, with amplification of target mRNA and control genes in separate wells. Each sample was run in duplicate. The relative amount of mRNA in each well was calculated as the ratio between target mRNA and a normalization factor created using three control genes (
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.
Immunohistochemistry
Immunohistochemistry was performed as previously described [
19]. Briefly, 25 μm floating sections were placed in net-well inserts and washed for 10 min in phosphate-buffered saline (PBS). After quenching, endogenous peroxidase activity was quenched with 3% H
2O
2 for 45 min, sections were blocked in 5% normal serum and 5% bovine serum albumin, followed by overnight incubation shaking at room temperature in primary antibody diluted in 5% normal serum. After two 15 min washes, secondary antibody diluted in 1% normal serum and PBS with 0.01% Triton X (PBS-T) was applied for 2 h shaking at room temperature. Sections were washed for 30 min in PBS before an amplification step was performed using an avidin–biotin–horseradish peroxidase complex kit (Vector Laboratories). Colorimetric detection was achieved with the peroxidase substrate kit Vector DAB (Vector Laboratories) according to the manufacturer’s instructions. Sections were mounted by hand on onto glass slides (Fisherbrand Superfrost Plus) and dried overnight before being dehydrated through a series of alcohols and xylene, and cover-slipped with DEPEX (Electron Microscopy Sciences). Antibodies used were as follows: the microglia marker Iba1 (Wako; 1:2000, rabbit polyclonal), the astrocyte marker GFAP (Sigma; 1:2000, mouse monoclonal), and appropriate biotinylated secondary antibodies (Vector, 1:2000).
Sections to be assessed for autofluorescence were mounted onto glass slides, washed in PBS-T for 30 min, and then stained with DAPI in PBS at 1:10,000 for 5 min. Slides were washed in twice in PBS for 5 min prior to coverslipping.
Image acquisition and analysis
Images were acquired as previously described [
12]. Integrated optical density measurements of signal intensity were acquired as previously described [
8] to quantify autofluorescence, a surrogate for lipofuscin deposition, in 4 images per mouse taken from 5 WT/Floxed, 12 Lyz-cKO, and 4 GrnKO mice. For quantification of colorimetric stains (Iba1 and GFAP), a threshold was set that pseudo-colored stained areas within a defined region of interest in each image. The average percent thresholded area for 4–6 images per mouse taken from 5 WT/Floxed, 12 Lyz-cKO, and 4 GrnKO mice is reported.
Primary microglia isolation and stimulation
Primary microglia cultures were generated from postnatal day 0–3 pups as previously described [
30]. Cells were stimulated with controlled standard endotoxin (final concentration 100 ng/ml; Associates of Cape Cod, MA, USA) as previously described [
30], and conditioned media was collected after 24 h and stored at − 20 °C until used. The quantity of IL-6 in the conditioned media was measured using a commercially available ELISA (Ready-set-go IL6 ELISA, eBiosciences, San Diego, USA) according to the manufacturer’s instructions. Results are presented from two independent experiments with 3–6 wells per genotype in each experiment.
Statistical analysis
All statistical comparisons were performed as a one-way analysis of variance (ANOVA) with Tukey post-hoc analysis to compare individual means to control and correct for multiple comparisons (Prism 6, Graphpad Software Inc.). A p value less than 0.05 was considered significant.