Department of Neuroscience, Mayo Clinic, Jacksonville, Florida, USA
Address correspondence and reprint requests to Dr. Rosa Rademakers, Department of Neuroscience, Mayo Clinic, Jacksonville, 4500 San Pablo Road, Jacksonville, FL 32224, USA. E-mail: email@example.com
Frontotemporal lobar degeneration (FTLD) is the second leading cause of dementia in individuals under age 65. In many patients, the predominant pathology includes neuronal cytoplasmic or intranuclear inclusions of ubiquitinated TAR DNA binding protein 43 (FTLD-TDP). Recently, a genome-wide association study identified the first FTLD-TDP genetic risk factor, in which variants in and around the TMEM106B gene (top SNP rs1990622) were significantly associated with FTLD-TDP risk. Intriguingly, the most significant association was in FTLD-TDP patients carrying progranulin (GRN) mutations. Here, we investigated to what extent the coding variant, rs3173615 (p.T185S) in linkage disequilibrium with rs1990622, affects progranulin protein (PGRN) biology and transmembrane protein 106 B (TMEM106B) regulation. First, we confirmed the association of TMEM106B variants with FTLD-TDP in a new cohort of GRN mutation carriers. We next generated and characterized a TMEM106B-specific antibody for investigation of this protein. Enzyme-linked immunoassay analysis of progranulin protein levels showed similar effects upon T185 and S185 TMEM106B over-expression. However, over-expression of T185 consistently led to higher TMEM106B protein levels than S185. Cycloheximide treatment experiments revealed that S185 degrades faster than T185 TMEM106B, potentially due to differences in N-glycosylation at residue N183. Together, our results provide a potential mechanism by which TMEM106B variants lead to differences in FTLD-TDP risk.
We studied the p.T185S TMEM106B genetic variant previously implicated in frontotemporal dementia with TAR DNA binding protein 43 pathology caused by progranulin mutations. Our cell culture studies provide evidence that the protective S185 isoform is degraded more rapidly than T185 TMEM106B, potentially due to differences in glycosylation. These findings suggest that low TMEM106B levels might protect against FTLD-TDP in these patients.
sodium dodecyl sulfate-polyacrylamide gel electrophoresis
single nucleotide polymorphisms
TAR DNA binding protein-43
Frontotemporal lobar degeneration (FTLD) is the second most frequent neurodegenerative disorder in individuals under the age of 65, accounting for 5–10% of dementia patients (Neary et al. 1998; Ratnavalli et al. 2002). FTLD patients often present with behavioral changes and personality dysfunction, corresponding to the pathology in frontal and temporal lobes (McKhann et al. 2001; Boxer and Miller 2005). Post-mortem analyses of FTLD brains determined that the most common pathological subtype involves intracellular inclusions of ubiquitinated TAR DNA binding protein 43 (TDP-43). In 2006, mutations causing a 50% loss in progranulin protein (PGRN) were found in the progranulin gene (GRN), accounting for nearly 20% of patients affected by this FTLD subtype (FTLD-TDP) (Baker et al. 2006; Cruts et al. 2006; Gass et al. 2006; Gijselinck et al. 2008). More recently, an expansion in a hexanucleotide repeat in the chromosome 9 open reading frame 72 gene was identified as an additional major genetic cause of FTLD-TDP (DeJesus-Hernandez et al. 2011; Renton et al. 2011).
Genetic studies have also served as a successful tool in the discovery of risk factors for FTLD-TDP. In 2010, Van Deerlin and colleagues performed a genome-wide association study in which single nucleotide polymorphisms (SNPs; top SNP rs1990622 T>C) located at the transmembrane protein 106 B gene locus (TMEM106B) on chromosome 7p21 were identified as FTLD-TDP risk factors (Van Deerlin et al. 2010). Moreover, the risk association of these SNPs was greatest in GRN mutation carriers (Finch et al. 2011; Van Deerlin et al. 2010), in which there was a highly significant decrease in the frequency of homozygous carriers of the rs1990622 minor allele, suggesting a protective effect of this allele in these patients. The exact relationship between TMEM106B polymorphisms and TMEM106B regulation and/or function, however, remains poorly understood. Initial studies showed a dose-dependent decrease in TMEM106B mRNA expression associated with the rs1990622 minor allele (Van Deerlin et al. 2010); however, this could not be confirmed in subsequent studies (Cruchaga et al. 2011; van der Zee et al. 2011). Also, even though individuals homozygous for the protective minor allele of rs1990622 showed higher plasma PGRN levels (Finch et al. 2011), only a minor absolute increase in PGRN was observed, suggesting that other disease mechanisms may be at play. Finally, we and others identified that rs1990622 is in perfect linkage disequilibrium with p.T185S (rs3173615), a coding variant in TMEM106B; however, thus far no studies have been reported to elucidate the functional consequence of p.T185S on TMEM106B.
At the time of these genetic discoveries, TMEM106B was a relatively uncharacterized transmembrane protein; however, recent publications have indicated that TMEM106B is a glycoprotein predominantly localized at the lysosomal membrane where it might interact with intracellular PGRN (Brady et al. 2013; Chen-Plotkin et al. 2012; Lang et al. 2012). Though these reports investigated the effect of TMEM106B expression on PGRN levels in vitro, the data are conflicting, and the effect of the coding variant p.T185S was not extensively studied. To further characterize TMEM106B and its role in FTLD-TDP, we investigated the risk (T185) and protective (S185) isoforms of TMEM106B and their effects on TMEM106B. First, we replicated the importance of TMEM106B polymorphisms on FTLD-TDP risk in GRN mutation carriers in a new cohort of patients. We further confirmed that TMEM106B is located in the lysosomes where it might interact with PGRN. Importantly, we showed that the protective (S185) TMEM106B isoform is consistently expressed at lower levels than the T185 TMEM106B isoform because of an increased rate of protein degradation, possibly resulting from changes in TMEM106B glycosylation. Thus, we provide the first insight into a functional difference between the risk (T185) and protective (S185) isoforms of TMEM106B.
Materials and methods
A total of 29 newly identified white patients with GRN mutations (14 females, 15 males) were included in the genetic association study. Patients were either identified by sequencing analysis or by low PGRN levels on a plasma PGRN ELISA test (Adipogen), and mutations included c.29T>C, c.90_91insCTGC, c.102delC, c.234_235delAG, c.328C>T, c.415T>C, c.592_593delAG, c.813_816delCACT, c.918C>A, c.933 + 1G>A, c.1072C>T, c.1252C>T, c.1428_1431delGGAT, and c.1477C>T. The mean age at diagnosis was 63.3 ± 10.5 years (range 44–76). Patients were ascertained from a total of five centers: Mayo Clinic Jacksonville (N = 5), Mayo Clinic Rochester (N = 1), University of British Columbia, Canada (N = 5), University of Western Ontario, Canada (N = 2), and IRCCS Istituto Centro San Giovanni di Dio Fatebenefratelli, Brescia, Italy (N = 13). An additional three patient samples were obtained from the Mayo Clinic Jacksonville Brain Bank. All patients agreed to be in the study and biological samples were obtained after informed consent with ethical committee approval from the respective institutions.
Genotyping and association analyses
Genotyping of rs1990622 and rs3173615 TMEM106B variants was performed using Taqman SNP genotyping assays (assay numbers C_11171598_10 and C_27465458_10, respectively; Life Technologies, Grand Island, NY, USA) on the 7900HT Fast Real Time PCR system. Genotype calls were made using the SDS v2.2 software (Life Technologies).
Construction and mutagenesis of cDNA clones
cDNA constructs encoding the V5-tagged p.T185S TMEM106B genes were constructed from heterozygous human cDNA. Briefly, total RNA was extracted from human frontal cortex brain tissue heterozygous for rs3173615 using the RNeasy Plus kit (Qiagen, Valencia, CA, USA). Reverse transcription was performed using the Superscript III system (Life Technologies), and cDNA was PCR-amplified with primers specific to TMEM106B to create BamHI and XhoI restriction sites, as well as the V5 tag. Amplified cDNA was digested with BamHI and XhoI restriction enzymes and cloned into the pAG3 and pAAV1 plasmids. The N183S mutation was induced independently in wild-type and mutant TMEM106B clones using the QuickChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA) and mutagenesis primers per the manufacturer's instructions. All clones were confirmed by sequencing analysis and transformed into Stbl3 competent E. Coli cells (Life Technologies), after which DNA for cell culture studies was obtained using the Nucleobond Xtra Maxi Plus EF kit (Clontech, Mountain View, CA, USA). Primer sequences are listed in Table S1.
A polyclonal rabbit TMEM106B antibody was generated through Thermo Fisher Pierce Protein Research (Rockford, IL, USA) against an N-terminal TMEM106B peptide comprised of amino acids 14:27 (KEDAYDGVTSENMR) . The following antibodies were used for western blotting experiments: rabbit polyclonal anti-TMEM106B (1 : 50 000 to 1 : 200 000), pre-immune rabbit serum (1 : 100 000), mouse monoclonal anti-V5 (1 : 200 000; Life Technologies), mouse monoclonal anti-lysosomal-associated membrane protein 1 (LAMP-1) (1 : 500; Santa Cruz Biotechnology, Dallas, TX, USA) and mouse monoclonal anti-GAPDH (1 : 500 000; Merdian Life Science, Memphis, TN, USA). Anti-mouse or anti-rabbit secondary horseradish peroxidase-conjugated antibodies (1 : 5000; Promega, Madison, WI, USA) were used for chemiluminescence. The following antibodies were used for immunofluorescence: rabbit polyclonal anti-TMEM106B (1 : 5000), mouse monoclonal anti-V5 (1 : 50 000; Life Technologies), mouse monoclonal anti-PGRN (1 : 100; R&D Systems, Minneapolis, MN, USA), and LAMP-1 (1 : 100; Santa Cruz Biotechnology). The anti-TMEM106B antibody and pre-immune rabbit serum were used as a concentration of 1 : 1000 for immunohistochemical analyses.
Cell culture transfection and western blotting
HeLa cells were maintained in Eagle's Minimum Essential Medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C, 5% CO2. One day prior to transfection, the cells were plated at 200 000 cells per well in a six-well culture dish. The next day, the cells were transfected with 2 μg of control plasmid DNA [tGFP- (Origene, Rockville, MD, USA) or empty pAAV1] or V5-tagged TMEM106B plasmid DNA in either the wild-type or mutant form using the Lipofectamine 2000 transfection reagent (Life Technologies). Cells were harvested in radioactive immunoprecipitation assay buffer (Boston BioProducts, Ashland, MA, USA) for western blot analysis. Cell lysates were diluted in an equivalent volume of 2X Novex Tris-glycine sodium-dodecyl sulfate (SDS) sample buffer (Life Technologies) supplemented with β-mercaptoethanol (to 5%) and denatured at 22°C for 30 min. Equal volumes were run on 10% SDS-polyacrylamide gels (Life Technologies), transferred to Immobilon membranes (Millipore, Billerica, MA, USA), and immunoblotted with the primary antibody. The next day, blots incubated with a horseradish peroxidase-conjugated secondary antibody and bands were detected by enhanced chemiluminescence using western Lightning Plus-ECL reagents (Perkin Elmer, Waltham, MA, USA). Peptide preabsorption experiments were performed for 2 h using 1 μg of the immunizing peptide before adding to the membrane. Western blot quantifications were normalized to GAPDH levels.
HeLa cells were plated and transfected on coverslips. Three days post-transfection, the cells were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS), and permeabilized in 0.3% Triton X-100 in PBS. Coverslips were then incubated in 10% bovine serum albumin in PBS at 22°C for 1 h before labeling with the primary antibody. The next day, the coverslips were labeled with the secondary antibody, incubated with Hoechst stain in PBS (1 : 5000; Life Technologies), and mounted on glass slides using Fluoromount-G (Southern Biotech, Birmingham, AL, USA). Images were acquired with a Zeiss LSM 510 META confocal laser-scanning microscope running either AIM or Zen2009 software (both from Carl Zeiss, Oberkochen, Germany) using a Plan-Apochromat 63X/1.4NA oil immersion objective at a scan zoom of 1.7 and optical slice depth of 0.7 μm. For colocalization analyses, TMEM106B and LAMP-1 weighted colocalization coefficients were obtained using the Zen2009 software (Carl Zeiss).
Tissue sections of the amygdala, hypothalamus, lentiform nucleus, neocortex, hippocampus, thalamus, midbrain, pons, medulla, and cerebellum were studied in 12 neurologically normal elderly individuals (8 females, 4 males) and 13 FTLD-TDP (4 female, 9 male) cases. Sections were cut at 5 μm thickness from formalin-fixed paraffin-embedded blocks, deparaffinized in xylene, rehydrated, and stained as previously described (Bieniek et al. 2013). Sections were processed with either pre-immune serum or the TMEM106B antibody and counter-stained with hematoxylin.
To determine PGRN expression levels in cell culture media and lysates, we used the Human Progranulin Quantikine ELISA kit (R&D Systems) following manufacturer's instructions. Samples were undiluted, performed in duplicate, and the provided human recombinant PGRN was used as a standard.
Cycloheximide was used at a concentration of 20 μg/mL to block protein synthesis. Lysosomal degradation was inhibited using 10 μM leupeptin and proteasomal degradation was obtained with 1 μM epoxomicin or 10 μM MG-132. For all drug treatments on T185- and S185-transfected cells, cells were harvested 0, 1, 3, 6, and 8 h after treatment for analysis by western blotting.
Transfected HeLa cells were incubated on ice for 15 min in lysis buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.5, 2 mM EDTA, 1% NP-40, protease inhibitor cocktail) and centrifuged at 3824 g, at 4°C for 5 min. The supernatant was transferred to clean tubes and 10 μg of protein per sample was denatured in 10X Glycoprotein Denaturing Buffer (New England Biolabs, Ipswich, MA, USA) at 22°C for 30 min. To perform deglycosylation, 500 units of recombinant endoglycosidase H (EndoH) (New England BioLabs) in 10X Reaction Buffer were used. Samples incubated in 10X Reaction Buffer without the presence of the enzyme served as a control. The samples with and without EndoH were incubated at 37°C for 1 h and the reaction was stopped with the addition of denaturing buffer 2X Sample Buffer (10% β-mercaptoethanol) at 22°C for 30 min. Deglycosylated TMEM106B was separated and analyzed using SDS-PAGE.
RNA isolation and real-time PCR
Total RNA was extracted from transfected HeLa cells using the RNeasy Plus Mini Kit (Qiagen), and its quality was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies). RNA samples were normalized to 50 ng/μL and using 300 ng as template, a reverse transcription reaction was performed using the Superscript III system (Life Technologies). Real-time quantitative PCR using an ABI7900 was performed in triplicate for each sample using gene expression probes for TMEM106B (Hs00998849_m1), GAPDH (Hs00266705_g1) and RPLP0 (Hs00420895_gh). Results were analyzed using SDS software version 2.2 (Life Technologies) and relative quantities of TMEM106B mRNA were determined.
For experiments in which only two groups were compared, significance was measured using a two-sample t-test. For analyses involving more than two groups, GraphPad Prism 5.04 (GraphPad Software, La Jolla, CA, USA) was utilized to perform a one-way anova followed by the Tukey's Multiple Comparison test. Using the same software, the drug time course experiments were analyzed by linear regression.
Genetic analysis of TMEM106B in a new cohort of GRN mutation carriers
To provide further support for TMEM106B SNPs in modifying the disease risk in GRN mutation carriers, we genotyped rs1990622 and rs3173615 (p.T185S) in a cohort of 29 newly identified symptomatic GRN mutation carriers. In this series, both SNPs were in complete linkage disequilibrium and only one patient was homozygous for the minor allele (rs1990622: TT n = 17, TC n = 11, CC n = 1). The frequency of homozygous carriers of the minor allele was significantly less (1/29, 3.4%) compared to what would be expected based on a previously genotyped control population (157/822, 19%; p = 0.03, OR = 0.15) (Finch et al. 2011). Interestingly, the patient carrying two copies of the minor allele is currently 71 years old with a diagnosis of mild cognitive impairment and had been identified as a GRN mutation carrier based on low PGRN plasma levels by ELISA (Ghidoni et al. 2012).
TMEM106B antibody characterization and protein localization
To better investigate the role of the TMEM106B protein in culture and in vivo, a rabbit polyclonal antibody raised against the N-terminus (aa 14–27) of the human TMEM106B protein was generated and tested both in HeLa cells transfected with V5-tagged TMEM106B and in human frontal cortex brain tissue (Fig. 1a). In both sample types, a predominant band was detected at approximately 40 kDa using the TMEM106B antibody, with a less prominent higher molecular weight species at approximately 70–90 kDa. Both the low- and high-molecular weight species in TMEM106B over-expressing HeLa cell lysates were also detected when blotted with a V5 antibody (Fig. 1b), and were no longer visible after preabsorption with the TMEM106B immunogen peptide (Fig. 1d). In brain samples, the higher molecular weight band appeared to be non-specific as it was also detected with the pre-immune rabbit serum, as well as after peptide incubation of our TMEM106B antibody. In contrast, the ~ 40 kDa species appeared specific in these tissue samples (Fig. 1c and d). Immunocytochemical experiments further supported the specificity of our antibody showing substantial overlap of V5 and TMEM106B antibody labeling in V5-tagged TMEM106B over-expressing HeLa cells upon confocal imaging (Fig. 1e).
Immunohistochemistry in paraffin sections of normal and disease brains with the TMEM106B antibody showed cytoplasmic punctate or granular immunoreactive structures in both neurons and non-neuronal cells, most notably microglia, in a pattern similar to that seen with immunohistochemistry for PGRN (Figure S1) (Ahmed et al. 2007). We did not observe differences in the distribution of neuronal TMEM106B in controls as compared with FTLD-TDP cases (Figure S2).
Previous reports have shown that TMEM106B is a type II transmembrane glycoprotein located within the late endosome and lysosomal compartments of the endomembrane system (Chen-Plotkin et al. 2012; Lang et al. 2012). Consistent with these studies, immunocytochemical analyses of cells transfected with either T185 or S185 TMEM106B showed punctate cytoplasmic labeling with our TMEM106B antibody, suggesting its localization to a subcellular compartment (Fig. 1f and g). Co-labeling of TMEM106B with a late endosomal and lysosomal marker, LAMP-1, showed overlap of these two antibodies, further suggesting that TMEM106B is, at least in part, located within the late endosomes or lysosomes in our cell culture system, with more colocalization of the T185 than S185 TMEM106B (T185 weighted colocalization coefficient 0.878 ± 0.010 as compared to S185 weighted colocalization coefficient 0.750 ± 0.019; p < 0.0005) (Fig. 1f and g).
To determine to what extent the TMEM106B isoforms colocalize with PGRN, cells were transfected with either T185 or S185 TMEM106B and labeled with TMEM106B and PGRN antibodies. We observed a striking overlap between PGRN and TMEM106B proteins, regardless of TMEM106B isoform (Fig. 2a and b). We also observed that cells transfected with T185 or S185 TMEM106B appeared to have more intense fluorescence with the PGRN antibody compared with non-transfected cells, indicating increased endogenous levels of PGRN in cells over-expressing TMEM106B (Fig. 2a and b). To more accurately determine whether T185 and/or S185 TMEM106B levels differently alter endogenous PGRN levels, we harvested T185- and S185-transfected HeLa cells 2 days after transfection and quantified both intra- and extracellular levels of PGRN using a PGRN ELISA. TMEM106B over-expression of either isoform caused a significant increase in PGRN levels in both intracellular and in the media as compared with control-transfected cells (Fig. 2c). These results confirm earlier observations published by Brady et al. and indicate that a direct effect on PGRN levels is unlikely to explain the risk associated with p.T185S in TMEM106B (Brady et al. 2013).
T185 TMEM106B is more highly expressed than S185 TMEM106B
Upon over-expression of T185 and S185 TMEM106B as part of our PGRN-related studies, we consistently observed a more prominent TMEM106B-immunoreactive band in T185-transfected cells (Fig. 2d). Upon quantification of TMEM106B protein levels, S185 expression was approximately 37% that of T185 expression both 2 and 3 days post transfection (p < 0.0001) (Fig. 3a and b). To determine whether these differences in TMEM106B isoform expression were because of changes at the RNA level, TMEM106B RNA levels were measured in T185- and S185-transfected cells at the same time points. Quantitative PCR analyses confirmed that TMEM106B RNA levels are not different in T185 versus S185 over-expressing cells at 2 or 3 days (Fig. 3c and d). To eliminate the possibility that differences between T185 and S185 protein expression are plasmid-specific, we further cloned the TMEM106B isoforms into a pAG3 mammalian expression plasmid and transfected these constructs into HeLa cells for 3 days. Consistent with our pAAV TMEM106B constructs, S185 TMEM106B protein was significantly less expressed than T185 TMEM106B (56% of T185 levels; p < 0.05), independent of RNA levels (Figure S3a, c, e). To ensure that these observations were not specific only to HeLa cells, we also transformed T185 and S185 TMEM106B into human embryonic kidney (HEK-293T) cells for 3 days, after which protein and RNA levels were quantified. Similar to HeLa cells, HEK-293T cells also expressed S185 TMEM106B significantly less than T185 (55% of T185 levels; p < 0.05), independent of RNA levels (Figure S3b, d, f). Taken together, these results suggest that a post-translational mechanism is responsible for the differences in protein levels between the risk (T185) and protective (S185) isoforms of TMEM106B.
TMEM106B isoforms differ in their rate of degradation
To determine whether T185 and S185 TMEM106B proteins are degraded at different rates, we treated T185- and S185-transfected cells with 20 μg/mL of cycloheximide to block protein synthesis. Western blotting of these lysates showed a gradual and marked decrease in TMEM106B levels of both isoforms with increasing time of cycloheximide treatment (Fig. 4a). We next quantified the changes in TMEM106B T185 or S185 levels over time and showed that the rate of degradation of the S185 isoform was significantly faster than the degradation rate of T185 TMEM106B (slopes are different; p = 0.016) (Fig. 4c). Based on our raw data, it would take approximately 8 h for 50% of T185 TMEM106B protein to be degraded as compared to only 2 h for S185 TMEM106B.
To rule out the possibility that different rates of protein synthesis of T185 and S185 TMEM106B might also contribute to their varying protein expression levels, we blocked TMEM106B degradation to observe the rate of TMEM106B intracellular accumulation. As it remains unclear as to what subcellular compartment is involved in TMEM106B's degradation, we first pharmacologically abrogated degradation in the proteasome using 10 μM MG132 and in the lysosome using 10 μM leupeptin for 8 h. Western blotting analysis of TMEM106B-transfected HeLa cells treated with MG132 did not show significant changes in TMEM106B expression after 8 h (Figure S4a). Similar results were obtained using a second proteasomal inhibitor, epoxomicin (data not shown). However, an increase in TMEM106B expression of both isoforms was observed upon treatment with leupeptin at 8 h (Figure S4b), indicating the lysosome as the predominant subcellular compartment involved in TMEM106B degradation. Thus, we performed a time course experiment in which TMEM106B-transfected cells were harvested 0–8 h after leupeptin treatment to observe the rate of TMEM106B protein synthesis after T185 or S185 over-expression. As indicated by western blotting, leupeptin treatment of TMEM106B-transfected cells caused a gradual increase in TMEM106B levels in both T185 and S185 over-expressing cells (Fig. 4b); however, the rate of protein synthesis of the T185 and S185 isoforms was not different (slopes are not different; p = 0.44) (Fig. 4d). Together these studies suggest that the differences observed between T185 and S185 protein expression is unlikely to be because of differences in the rate of protein synthesis, but more likely because of differences in the rate of their degradation.
Glycosylation at position N183 might be involved in TMEM106B isoform stability
In the recent report by Lang et al., TMEM106B was found to be a glycosylated protein with 5 N-glycosylation sites throughout the protein at amino acids N145, N151, N164, N183, and N256, each with an N-X-T/S consensus sequence (Lang et al. 2012). Simple glycan modifications are added to mature T185 TMEM106B at the first three N-glycosylation sites, whereas complex glycans were detected on the last two N-glycosylation sites of the protein. Proper protein glycosylation is critical for adequate protein folding and subsequent stability, localization, and function. Interestingly, the p.T185S variant in the TMEM106B protein is located within the glycosylation consensus sequence for the fourth N-glycosylation site. Therefore, p.T185S might contribute to differences in glycosylation at N183. To address this question, we first extended the time of SDS gel separation of lysates from HeLa cells over-expressing T185 and S185 TMEM106B isoforms; however, no obvious difference in their molecular weights was identified (data not shown). To more specifically determine whether N183 glycosylation differs between T185 and S185 TMEM106B, we subjected lysates from T185- or S185 over-expressing cells to digestion with EndoH. This enzyme is only capable of removing simple glycans, leaving complex modifications intact. Thus, if simple glycans are added to N183 of S185 TMEM106B, we expected to see a lower molecular weight product post EndoH digestion than with T185. EndoH treatment of both the T185 and S185 TMEM106B proteins however resulted in similar TMEM106B-immunoreactive bands (Fig. 5a), indicating that complex glycosylation is likely preserved at N183 in the S185 TMEM106B protein. EndoH digestion did not rule out the possibility that more subtle changes in complex gycosylation at N183 might account for differences in protein stability between T185 and S185 TMEM106B.
To further study the involvement of N183 in the stability difference observed between the TMEM106B isoforms, we introduced a N183S mutation in both T185 and S185 TMEM106B and compared expression levels of the mutants. As compared to non-mutated T185 and S185 TMEM106B, the N183S mutations caused a downward shift in the molecular weight of the TMEM106B protein (Fig. 5b), likely because of complete loss of glycosylation at amino acid 183 (Lang et al. 2012). Moreover, protein levels of the over-expressed N183S mutants resulted in a robust decrease in TMEM106B expression (Fig. 5b), underlining the significance of the N183 site for TMEM106B maturation. Importantly, quantification of mutant expression levels indicated that when the N183S mutation was introduced, TMEM106B RNA levels were still the same between T185 and S185 TMEM106B, but now the T185 TMEM106 protein levels were no longer significantly higher than that of S185 (Fig. 5c and d).
It has become increasingly clear that variants in or near the TMEM106B gene play a critical role in the risk of developing FTLD-TDP. Strong association was especially apparent in FTLD-TDP patients carrying GRN mutations, suggesting that TMEM106B might modify the disease through regulation of PGRN levels or function (Van Deerlin et al. 2010). In this study, we further confirmed the association of TMEM106B with FTLD-TDP in an additional 29 patients with loss-of-function GRN mutations. Only one patient (3.4% of patients) was homozygous for the minor protective allele of rs1990622. Consistent with our previous findings, the phenotype in this patient is relatively mild, with a diagnosis of mild cognitive impairment at the age of 71 years. In our laboratory, we previously identified 2 of 127 patients (1.6%) to be homozygous for the minor allele of rs1990622. Cumulatively, among unrelated probands of GRN mutation families, we identified only 3 of 156 (1.9%) patients to be homozygous for the minor allele of rs1990622 as compared to 157 of 822 (19.1%) control individuals studied to date (Van Deerlin et al. 2010; Finch et al. 2011, and this study). These genetic findings add to the growing body of evidence confirming TMEM106B as a disease risk factor and potential modifier in GRN-related FTLD-TDP.
To provide insight in the molecular mechanisms associated with TMEM106B variants, we focused our studies on the coding variant p.T185S, which is in complete linkage disequilibrium with rs1990622, as a potential functional variant implicated in FTLD-TDP risk. As the TMEM106B variants were associated most strongly with FTLD-TDP risk in GRN mutations carriers, TMEM106B might confer risk by directly affecting PGRN levels or function. Here, we show that both isoforms of TMEM106B (T185 and S185 TMEM106B) colocalize, in part, with lysosomal compartments. Immunofluorescence analyses of T185 and S185 TMEM106B also showed that both TMEM106B isoforms colocalize with PGRN. This is in agreement with two recent reports that also observed significant overlap of the TMEM106B and PGRN (Brady et al. 2013; Chen-Plotkin et al. 2012). Moreover, we found that PGRN levels were significantly increased as compared with control-transfected cells in the media and lysates of T185- and S185-transfected cells at 2 days. Even though the first characterization of TMEM106B did not reveal TMEM106B-induced changes in PGRN (Lang et al. 2012), we are now the third group to observe an increase in PGRN with TMEM106B over-expression (Brady et al. 2013; Chen-Plotkin et al. 2012). However, similar to findings by Brady et al. (Brady et al. 2013) both TMEM106B T185 and S185 isoforms similarly affected PGRN levels. These results, therefore, fail to explain the decrease in FTLD-TDP risk in people who are homozygous for the rs3173615 TMEM106B minor allele.
In previous reports discerning the association between TMEM106B and FTLD-TDP, attention has been drawn to the potential effect of the TMEM106B SNPs on TMEM106B mRNA expression levels (Van Deerlin et al. 2010; Brady et al. 2013; Chen-Plotkin et al. 2012). This began with the observation that individuals homozygous for the protective C-allele of rs199022 had lower TMEM106B RNA levels in brain tissue (Van Deerlin et al. 2010). However, this finding was performed using a small subset of samples and has not been replicated by other groups (Cruchaga et al. 2011; van der Zee et al. 2011). Although we cannot exclude that rs1990622 or variants in linkage disequilibrium with rs1990622 located in non-coding regulatory regions of TMEM106B might contribute to difference in TMEM106B expression levels in vivo, our study now provides strong evidence implicating p.T185S as a functional TMEM106B variant modulating TMEM106B protein levels. Using multiple cell lines and expression vectors, we consistently showed that the risk T185 TMEM106B isoform was expressed nearly two-fold greater than S185 TMEM106B. Subsequent analyses showed that the difference in expression resulted from a more rapid degradation of the S185 TMEM106B isoform in our cell culture system. Brady et al. did not report differences between these two TMEM106B isoforms; however, discrepancies might result from their use of a mouse cell line (Brady et al. 2013). In addition, the transfection levels were not reported for each isoform (Brady et al. 2013). Our findings support the hypothesis that higher TMEM106B protein levels are, at least in part, contributing to the risk differences between the T185 and S185 TMEM106B isoforms and provide the first variant-related difference in the post-translational regulation of TMEM106B.
The p.T185S coding variant of the TMEM106B protein is a part of the N-X-T/S glycosylation consensus sequence for N-glycosylation at position 183 (Lang et al. 2012). Protein glycosylation is a critical post-translational modification that enhances functional diversity as well as biological activity and expression level of a wide-range of glycoproteins. Even though a T or S residue at position 185 is expected to be sufficient for N-glycosylation at TMEM106B N183, we speculate that glycosylation may be different between the two isoforms with a slight change in the glycan composition and/or complexity at N183 in TMEM106B T185 compared with S185. In support of this hypothesis, it has been previously shown that T and S amino acids can affect N-glycosylation transfer rates in vitro, with less efficient interaction between glycotransferase enzymes in S-containing consensus sequences (Bause and Legler 1981). A previous study showed that complete loss of glycosylation at N183 results in retention of TMEM106B to the endoplasmic reticulum (ER) (Lang et al. 2012). Based on our results, cells over-expressing S185 TMEM106B still show TMEM106B localized to the lysosomes suggesting that more subtle changes in glycosylation may be at play. Subtle differences in S185 versus T185 TEM106B N183 glycosylation could explain why no gross differences in the molecular weights or EndoH digestion products of T185 versus S185 TMEM106B were observed. Confirming differences in the composition of complex N-glycans at TMEM106B amino acid 183 would require extensive mass spectrometry analyses and/or specific high performance liquid chromatography beyond the scope of this study. Thus, while our current data do not specifically confirm N183 glycosylation as the functional process involved in regulating T185 versus S185 TMEM106B expression levels, abnormal glycosylation of TMEM106B S185 could explain the enhanced degradation of this TMEM106B isoform. In line with this hypothesis, introduction of an artificial N183S glycosylation-defective mutant within the TMEM106B T185 and S185 isoforms ablated the observed differences in protein expression between these two isoforms.
In conclusion, our study is the first to demonstrate that the p.T185S coding variant, genetically associated with FTLD-TDP risk, acts as a functional variant to regulate TMEM106B protein levels, which we speculate are because of changes in glycosylation. As TMEM106B levels have been shown to be important for determining proper endolysosomal homeostasis, these results support a critical role for lysosomal dysfunction in the development of FTLD-TDP. We also confirmed an effect of TMEM106B expression on PGRN levels. However, in contrast to what would be expected from an FTLD risk factor, all published studies observed an increase (not decrease) in PGRN after over-expression of TMEM106B. As this increase in PGRN may be the consequence of lysosomal dysfunction it remains unclear how this finding has to be interpreted and whether it has any relevance with regards to FTLD-TDP disease risk. One possibility for the strong association of TMEM106B SNPs in patients with GRN mutations may be that these patients merely reflect a patient population vulnerable to additional genetic modifying factors such as TMEM106B. The recent observation of an effect of TMEM106B genotypes on cognition in amyotrophic lateral sclerosis patients and the presence of TDP-43 pathology in Alzheimer's disease patients, two neurodegenerative diseases in which patients are thought to have normal PGRN levels, is of interest in this regard (Vass et al. 2011; Rutherford et al. 2012).
Together, our findings support a critical role for p.T185S in FTLD-TDP risk by regulating TMEM106B protein levels providing a promising novel avenue for disease intervention in FTLD-TDP and related TDP-43 proteinopathies.
This study is supported by the Consortium for Frontotemporal Dementia Research (CFR), the Mayo Foundation and NIH grants R01 NS065782, R01 AG026251 and P50 AG016574. NET is funded by NIH grants R01AG032990 and P50AG016574. AMN is funded by the Association for Frontotemporal Degeneration (AFTD) Postdoctoral Fellowship. Dr. Hsiung is supported by a CIHR Clinical Genetics Investigatorship. The UBC frontotemporal dementia cohort studies are supported by CIHR grant (#179009) and PARF grant (C06-01). This study was further supported by grants Fondazione CARIPLO 2009-2633; Ricerca Corrente, Italian Ministry of Health and COEN015 from the Centres of Excellence in Neurodegeneration. The authors have no conflicts of interest to declare.