MG132 was obtained from Calbiochem (San Fransico, CA, USA). Monensin, kifunensin, cycloheximide, bafilomycin A1 were sourced from Sigma-Aldrich (St Louis, MO). All plasmids used for cloning were obtained from Cosmo Genetech Co. Ltd.

Human TREM2 construct was purchased from Sino Biological Inc., BDA, Beijing and pcDNA5-FRT/TO-HA was obtained from Dr. SW Kang Bio-Medical Institute of Technology. Wild-type human TREM2 (TREM2 wild-type) was inserted into Hind-III and Xho-I sites of pcDNA5-FRT/TO-HA. A point mutation in TREM2 was generated to substitute Arg with His at position 47 and thus produce the R47H variant. These modified TREM2 mutations of cDNAs were PCR-amplified and sub-cloned into the pcDNA5-FRT/TO-HA vector. All constructs was verified by DNA sequencing.

HeLa cells were maintained in DMEM (HyClone Laboratories, UT, USA) supplemented with 10% fetal bovine serum (HyClone Laboratories) and were incubated in 5% CO₂ at 37°C. All experiments involving transient transfection were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) in accordance with the manufacturer's instructions. Cells transfected with the wild-type and R47H variant of TREM2 were incubated in antibiotic-free DMEM media for 4 h, after which the media was replaced with DMEM supplemented with FBS.

Cell lysates were prepared in a buffer containing 1% SDS, 100 mM Tris-HCl, and pH 8.0 and were boiled for 5 min. To obtain Triton X-100-soluble and -insoluble fractions for the analysis of protein solubility, cells were incubated with Triton-lysis-buffer [1% Triton X-100, 20 mM HEPES pH7.5, 150 mM NaCl, protease inhibitor cocktail (Calbiochem)] for 15 min on a rocking platform as previously described (Kang et al., 2006). After centrifugation, the supernatant (Triton-soluble fraction) was collected. The Triton insoluble fraction (pellet) was resuspended in SDS-lysis-buffer (1% SDS, 100 mM Tris-HCl, and pH 8.0). Both fractions were reconstituted to equal volumes and analyzed by western blotting as described below.

To analyze the digitonin fraction, cells were incubated with 0.75% digitonin solution for 5 min on ice as previously described (Kang et al., 2006). Parts of the lysate were mixed with the sample buffer for the preparation of total cell sample and some of lysate was centrifuged at 800 g for 5 min. Supernatant (cytosol fraction) was collected and pellet were incubated with Triton-lysis-buffer (1% Triton X-100, 20 mM HEPES pH 7.5, 150 mM NaCl, protease inhibitor cocktail) for 5 min on ice. The pellet incubated with Triton-lysis buffer were centrifuged at 800 g for 5 min. Some of supernatant was collected for the membrane fraction and the remainder was centrifuged at 13,000 rpm for 30 min (the supernatant contained the soluble membrane fraction and the pellet the insoluble membrane fraction).

For western blotting, protein lysates from HeLa cells transfected with either the TREM2 wild-type or R47H mutant construct were fully solubilized with 1% SDS, and 100 mM Tris-HCl, pH 8.0. Protein concentrations were measured using the Bradford method. Equal amounts of protein were mixed with the sample buffer (62.5 mM Tris-HCl [pH 6.8], 1% [w/v] sodium dodecyl sulfate [SDS], 2.5% [v/v] glycerol, 0.5% [v/v] β-mercaptoethanol, and bromophenol blue), boiled at 100°C for 5 min, and stored at −20°C until use. Proteins were resolved by SDS-polyacrylamide gel electrophoresis at a constant voltage (100 V) and were subsequently transferred to polyvinylidene difluoride membranes (pore size, 0.2 mm; BioRad, CA) at 100 V for 1.5 h. After 1 h incubation in blocking PBST buffer (0.1% [v/v] Tween-20) containing 2% (w/v) bovine serum albumin and 2% (v/v) normal horse serum, blots were incubated with primary antibodies overnight at 4°C. Blots were then washed in PBST buffer, incubated with horseradish, peroxidase-conjugated, anti-IgG (1:5000; Pierce, Rockford, IL), and visualized using enhanced chemiluminescence reagents (Amersham, Arlington Heights, IL) and X-ray film. The primary antibodies used in the immunoblotting were rat anti-HA (1:5000; Roche, Mannheim, Germany), mouse anti-GRP78 (1:1000, BD Transduction Laboratories™, California, CA, USA), rabbit anti-phospho-eIF2α (Ser51) (1:1000, Cell Signaling Technology, Danvers, MA), rabbit anti-phospho PERK(Thr981) (1:100, Santa Cruz, CA), rabbit anti-PERK(H-300) (1:100, Santa Cruz, CA), and mouse anti-β-actin (1:10,000; Sigma-Aldrich).

Coomassie blue-stained protein bands were excised from the ge1, chopped into 1 mm³ pieces and washed repeatedly in 100 mM ammonium bicarbonate buffer for 10 min and then in acetonitrile for 10 min to enable enzyme penetration. The pieces were then dried thoroughly in a vacuum centrifuge and rehydrated in 200 mM ammonium bicarbonate containing 1 μL (or 500 U) of peptide N-glycosidase F (New England Biolabs). Enzymatic deglycosylation was then performed at 37°C for 16 h in the water bath. At the end of the digestion, the N-glycan supernatant was drawn out from the gel pieces and the extracted glycans were dried in a vacuum centrifuge.

MS analysis was performed in accordance with the optimized procedures published by Hua et al. (2013). Aqueous glycan solutions were injected by autosampler onto a chip-mounted nano-LC column (Agilent Technologies), consisting of a 9 × 0.075 mm i.d. enrichment column and a 43 × 0.075 mm i.d. analytical column. Both columns were packed with 5 μm porous graphitized carbon as the stationary phase. A rapid glycan elution gradient was delivered to the analytical column at 0.3 μL/min using solutions of (A) 3.0% acetonitrile and 0.1% formic acid (v/v) in water, and (B) 90.0% acetonitrile and 0.1% formic acid (v/v) in water, ramping from 0 to 44% B solution over 30 min. Remaining non-glycan compounds were flushed out with 100% B solution prior to re-equilibration. MS spectra were acquired via a TOF MS detector (model 6530, Agilent Technologies) in positive ionization mode over a mass range of m/z 500−2000 with an acquisition time of 1.5 s per spectrum. Following data acquisition, raw LC/MS data were processed using the Molecular Feature Extractor algorithm included in the MassHunter Qualitative Analysis software (Version B.6.00 SP1, Agilent Technologies). Using expected isotopic distribution and charge state information, extracted ion chromatograms were combined to create extracted compound chromatograms (ECCs) representing the summed signal from all ion species associated with a single compound (e.g., the doubly protonated ion, the triply protonated ion, and all associated isotopologues). Thus, each individual ECC peak could be taken to represent the total ion count associated with a single distinct compound. Each ECC peak was matched by accurate mass to a comprehensive library of all possible complex, hybrid, and high mannose glycan compositions based on known biosynthetic pathways and glycosylation patterns (Kronewitter et al., 2009). Deconvoluted masses of each ECC peak were compared against theoretical glycan masses using a mass error tolerance of 10 ppm. As samples originated from human cell lines, only glycan compositions containing hexose (Hex), N-acetylhexosamine (HexNAc), fucose (Fuc), and N-acetylneuraminic acid (NeuAc) were considered.

Western bands were measured and analyzed with Image J software. Data are presented as mean ± SEM from at least triplicate experiments. Statistical analyses were performed using GraphPad Prism 5.0 software using an unpaired t-test or two-way ANOVA; ^***P < 0.001; ^**P < 0.01; ^*P < 0.05.

To investigate whether, and how, the R47H variant of TREM2 reported in AD (Soragna et al., 2003; Jiang et al., 2013) affected the trafficking of TREM2, the R47H mutation was introduced into TREM2 and the variant protein was expressed in HeLa cells (Figure 1B and Supplementary Figure 1). Western blots confirmed that the expression patterns of wild-type TREM2 and R47H mutants differed as we previously observed (Park et al., 2015). The expected molecular weight of TREM2 is about 25 kDa. Both the wild-type and R47H variant of TREM2 showed a thick band of ~28 kDa (Figure 1B, c) with the N-glycosylated form detectable in the ER. The upper 30~40 kDa bands denote the glycosylated forms in the Golgi. The uppermost bands of R47H at about 40 kDa were more intense than those of the wild-type protein (Figure 1B, a), although the upper bands of R47H at about 30 kDa were at a lower intensity than wild-type (Figure 1B, b). the C-terminal fragment (CTF) of R47H at about 10 kDa was less intense than its wild-type counterpart (Figure 1B, d).

Since the upper bands for TREM2 on an immunoblot reflect its glycosylation status (Park et al., 2015), we suspected that the different expression patterns of wild-type TREM2 and its R47H variant (Figure 1B) are due to glycosylation changes. Hence, we treated cells with monensin (10 μM, 8 h) which blocks the transport from the medial to the trans cisternae of the Golgi stack and prevents terminal glycosylation (Griffiths et al., 1983). The highest western band (Figure 2A, a), which is more prominent for TREM2 R47H, is a monensin-sensitive glycosylated form but the lower band (Figure 2A, b), which is more prominent for the wild type protein, is resistant to monensin. Thus, the mobility difference between wild-type and R47H variant of TREM2 reflects the differences in their glycan structures.

To further analyze the glycosylation characteristics of TREM2, we treated cells with kifunensin, a mannosidase I inhibitor that prevents the trimming of mannose residues from high mannose N-glycans, thereby preventing the synthesis of complex N-glycans that are acceptors for α1–3-fucosyltransferases (Elbein et al., 1990; Yago et al., 2003). Kifunensine treatment decreased the intensity of the upper bands (Figure 2B, a and b), indicating that the higher TREM2 bands of ~30–45 kDa (Figure 2B, a and b) represent the addition of complex oligosaccharide chains to this protein in the Golgi and the thick, main band at about 28 kDa (Figure 1B, c) represents N-linked glycosylation with high mannose in the ER, further confirming our previous findings (Park et al., 2015).

To further investigate the glycosylation differences between the wild-type TREM2 and the R47H variant, the N-glycans released from these proteins (Figure 1B, a,b) were analyzed by nano-LC/MS. Figure 3A shows ECCs of the N-glycans found in wild-type TREM2 (top) and TREM2 R47H (bottom), respectively. Each peak in Figure 3A represents individual N-glycan species, color-coded according to biosynthetic class (high mannose, undecorated complex/hybrid, fucosylated complex/hybrid, sialylated complex/hybrid, or fucosylated-sialylated complex/hybrid). Approximately 100 N-glycan compound peaks with 38 distinct N-glycan compositions were identified in wild-type TREM2, whereas over 110 N-glycan compound peaks with over 45 distinct N-glycan compositions were identified in the TREM2 R47H variant. Casual inspection revealed that N-glycosylation of wild-type TREM2 is different from that of TREM2 R47H in terms of complex/hybrid N-glycans which were not detectable in wild-type TREM2 at the retention time of 22–28 min (red box). To more precisely examine the glycosylation differences between wild-type and TREM2 R47H, the glycans detected for both proteins were quantified using the chromatogram peak areas. Figure 3B shows a bar graph of the relative abundances associated with each glycan type in R47H vs. wild-type TREM2. Glycans were grouped into high mannose type, undecorated complex/hybrid, fucosylated complex/hybrid, sialylated complex/hybrid, and fucosylated-sialylated complex/hybrid type. The high mannose type glycans were found in high abundance in both wild-type and TREM2 R47H, accounting for ~80% of the total glycans. Overall, the non-high mannose type glycan, which is less abundant in TREM2, was elevated in TREM2 R47H. Moreover, we confirmed using a t-test that the levels of undecorated complex/hybrid type glycans (p = 0.012) and fucosylated-sialylated complex/hybrid type glycans (p = 0.021) had statistically significant differences between wild-type and TREM2 R47H.

To more precisely quantify the differences between wild-type and TREM2 R47H, the relative abundances of individual glycans were compared. The profiles of the N-glycan structures found in both wild-type and TREM2 R47H are shown in Figure 3C (undecorated complex/hybrid type glycans) and Figure 3D (fucosylated-sialylated complex/hybrid type glycans). Wild-type TREM2, in turn, was readily distinguishable from TREM2 R47H by elevated levels of glycans consisting of [Hex]₃₋₆[HexNAc]₃₋₅[Fuc]₁₋₂[NeuAc]_1. Interestingly, all glycans are biosynthetically highly related, each differing from the next by few monosaccharides.

To further investigate the biochemical differences between wild-type and TREM2 R47H, we tested the solubility of these proteins in triton-X100 as previously described (Kang et al., 2006). TREM2 R47H was increased in the insoluble fractions compared to the wild-type product (Figure 4A), suggesting the decreased solubility by R47H variant. We next treated cells with MG132, a proteasomal inhibitor, to block proteasomal degradation. TREM2 R47H was more elevated after MG132 treatment compared to wild-type (Figure 4B), indicating that this variant is unstable and degraded via the ER associated degradation (ERAD) pathway (Kroeger et al., 2009; D'Souza et al., 2013). We also treated cells with bafilomycin A1, a v-ATPase inhibitor, to block lysosomal degradation (Cho et al., 2014). TREM2 R47H was again more elevated after this treatment compared to wild-type (Figure 4C), again indicating that this variant is unstable and is largely degraded by lysosomes. The intracellular domain (ICD) of TREM2, which is generated after cleavage by secretases (Wunderlich et al., 2013; Kleinberger et al., 2014), was also increased in MG132- or bafilomycin A1-treated cells (Figures 4D,E).

The half-life of TREM2 was tested in transfected cells exposed to the protein synthesis inhibitor cycloheximide. In the wild-type TREM2 transfected cells, the exogenously expressed protein decreased very rapidly after cycloheximide treatment (Figures 5A,B), indicating that TREM2 is rapidly degraded under normal conditions. However, in the TREM2 R47H-transfected cells, the levels of the variant protein decreased more slowly after cycloheximide treatment compared to wild-type (Figures 5A,B), indicating a longer half-life. Interestingly, this feature was apparent in the ER forms of TREM2 R47H (Figure 5A, b and c, Figure 5B) (Park et al., 2015), but not so for the variant protein in the Golgi (Figure 5A, a, Figure 5B), suggesting that the R47H mutant has increased resistance to degradation in ER rather than to degradation after trafficking to the Golgi. Since the unglycosylated bands of TREM2 (Figure 5A, c) can reflect two forms, either in the ER or in the cytosol, we fractionated cells into cytosolic and membranous organelles, and further into soluble and insoluble membranous organelles (Figure 5C) as previously described (Kang et al., 2006). The results of this experiment indicated that the unglycosylated TREM2 form (Figure 5A, c) is in the ER membrane (Figure 5C, arrow) and not in the cytosol after retrotranslocation from the ER.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.