Evidence of an NgBR-NPC2 interaction
Analysis of the primary sequence of NgBR (Figure S1A) reveals a single C-terminal domain with homology to cis-isoprenyltransferase (cis-IPTase). cis-IPTase domains bind isoprenyl lipids and catalyze the condensation of isopentenyl diphosphate to farnesyl diphosphate to form long-chain polyprenyl diphosphates. Previous studies using partially purified NgBR showed that it lacks cis-IPTase activity (Miao et al., 2006
). In order to gain insights into NgBR function, we used the C-terminal domain of NgBR as bait and screened a human heart cDNA expression library in a yeast two-hybrid assay. After several rounds of screening and selection, unique clones were isolated and one of the strongest interactions was seen between the C-terminus of NgBR and NPC2 as measured by alpha-galactosidase secretion (Figures S1B, S1C). These data support the previously detected interaction between NPC2 and human dehydrodolichyl disphosphate synthase (DHDDS, also referred to as HDS, or human cis-isoprenyltransferase; hCIT) (Kharel et al., 2004
), the only other mammalian protein characterized thus far with a cis-IPTase domain (Shridas et al., 2003
). However, the functional relevance of the DHDDS-NPC2 interaction remains unknown.
The interaction between the cis-IPTase domain in NgBR and full length NPC2 was confirmed in mammalian cells by immunoprecipitation experiments. Transfection of NPC2-myc into CHO cells stably-expressing either vector alone (Vector), NgBR-HA (NgBR), or a deletion mutant of NgBR that lacks the C-terminal cis-IPTase domain (NgBRΔC) followed by immunoprecipitation of NgBR-HA demonstrates a clear interaction between NPC2 and NgBR that is absent in cells expressing NgBRΔC (). We also tested the idea that the previously described interaction between NgBR and the N-terminus of Nogo-B (AmNgB) might influence the NgBR-NPC2 interaction. Addition of recombinant AmNgB to CHO-NgBR cells had no effect on the interaction between NgBR and NPC2, suggesting that NgBR has a distinct subcellular role independent of AmNgB.
NgBR interacts with and stabilizes NPC2
Since NPC2 is a secreted glycoprotein, we examined NgBR topology using several approaches. Initially, we cloned the hNPC2 Asn58 glycosylation site in frame at the N- (‘N-glyco’) and C-termini (‘C-glyco’) of NgBR (Figure S2A). We also cloned this tag onto a well-characterized peripheral membrane protein, endothelial nitric oxide synthase (eNOS), as a negative control. Expression of the N- and C-glyco mutants of NgBR resulted in mobility shifts by SDS-PAGE that were sensitive to Endoglycosidase H (EndoH) treatment, suggesting that a proportion of these regions of NgBR are luminally oriented. Treatment of both WT and C-glyco eNOS did not show any appreciable differences in mobility (Figure S2B). We further addressed membrane topology of NgBR by limited proteolysis experiments. We cloned a myc epitope tag at the extreme N-terminus of NgBR and analyzed loss of this epitope after trypsin digestion (Figure S2C). In the absence of detergent, the luminal ER chaperone Grp94 is not cleaved to an appreciable extent by trypsin; however, when detergent is present, the membrane is permeabilized and Grp94 is accessible to protease. Analysis of myc and HA (N- and C-termini of NgBR, respectively) shows that a proportion of these epitopes are oriented luminally, as the accessibility of these epitopes to trypsin is enhanced in the presence of detergent. These data suggest that NgBR conforms to a topology different from that suggested by the initial bioinformatic analysis, in that both the N- and C-termini can be arranged with a luminal orientation.
Analysis of the primary sequence of NgBR using the SignalP server (Bendtsen et al, 2004
) results in the possibility of either a signal peptide (probability = 0.262) or a signal anchor (an uncleaved signal peptide; probability = 0.718) that consists of amino acids 1–23 from the N-terminus of NgBR. We took two approaches to distinguish between these possibilities. First, we utilized the myc-NgBR-HA construct described above to assess cleavage of the N-terminal 23 amino acids. If any portion of the NgBR N-terminus is cleaved by signal peptidase, then loss of the myc epitope should occur and thus be undetectable. SDS-PAGE analysis of myc-NgBR-HA showed that the myc epitope remains intact, and myc-NgBR-HA migrates with slightly slower mobility than NgBR-HA (Figure S2D), providing evidence of a potential signal anchor motif at the N-terminus of NgBR. In addition, we reasoned if NgBR contains a cleaved signal peptide, then deletion of the N-terminal 23 amino acids should result in a protein which migrates with the same mobility as WT NgBR by SDS-PAGE (Figure S2E). In fact, deletion of this N-terminal region of NgBR resulted in a protein which migrated faster than WT NgBR (Figure S2F), supporting the argument that this region remains intact in WT NgBR and is not a cleaved signal peptide. These data provide evidence that NgBR is not a canonical Type I transmembrane protein as originally predicted by bioinformatics analysis and is atypical in its membrane orientation. Collectively, these data support the conclusion that the C-terminal region of NgBR is necessary for the NgBR-NPC2 interaction in mammalian cells and confirm the yeast two-hybrid results.
While conducting the above co-immunoprecipitation experiments, we observed a significant and reproducible increase in ectopically-expressed NPC2 in CHO-NgBR cells relative to CHO-Vector or CHO-NgBRΔC cells. As the plasmid expressing NPC2 is under the control of a CMV promoter, a plausible explanation for the increase in NPC2 levels is that NgBR decreases the turnover of NPC2 protein. Indeed, increasing the concentration of NPC2-myc dose-dependently enhanced the levels of NPC2-myc in lysates from CHO-NgBR cells, but not in lysates from cells expressing either vector alone or NgBRΔC (). Moreover, levels of NPC2 secreted into the media of CHO-NgBR were greatly enhanced compared to CHO cells expressing vector alone (). In order to examine possible effects of NgBR on NPC2 protein stability, we performed pulse-chase experiments by expressing NPC2 in the absence or presence of NgBR. The t1/2 of NPC2 in the absence of NgBR was 3.5hrs, while the t1/2 of NPC2 in the presence of NgBR was 5.1hrs, reflecting an almost 50% increase in NPC2 stability ().
The preceding experiments were performed in a heterologous system in which both NgBR and NPC2 were ectopically expressed. In order to confirm these results in cells that express high endogenous levels of both NgBR and NPC2, we used HepG2 cells (a hepatocarcinoma cell line) and EA.hy926 cells (EA.hy; a hybridoma cell line created by fusion of HUVEC and A549) and employed an RNAi-mediated approach to decrease NgBR expression. siRNA directed against NgBR reduces NgBR levels by 80–90% () and the loss of NgBR reduces endogenous NPC2 expression in both cell lines (the loss of NgBR reduces NPC2 levels by 35.5% +/−and 49.3%+/−12.2 in HepG2 and EA.hy926 cells, respectively, n=5 independent experiments). In order to assess whether the loss of NgBR influenced the levels of another endosomal/lysosomal glycoprotein, we also probed for another soluble lysosomal glycoprotein, cathepsin D. Levels of cathepsin D remain unaffected by loss of NgBR, suggesting a specific effect of NgBR expression on NPC2 stability. In order to gain insight into the mechanism by which NgBR-regulates NPC2 stability, we tested the proteasomal dependence of NPC2 degradation. Treatment of cells with proteasome inhibitors, MG132 or lactacystin, prevented the loss of NPC2 following NgBR knockdown (). Collectively, these data suggest that NgBR expression is necessary for NPC2 stability.
The interaction between NgBR and NPC2 occurs in a pre-lysosomal compartment
In order to gain further insights into the interaction between NgBR and NPC2, we assessed NgBR and NPC2 colocalization by immunofluorescence. Detection of endogenous NPC2 by immunofluorescence is dependent on the usage of Bouin's fixative for reliable detection (Zhang et al., 2003
). Detection of NgBR is limited to HA-tagged NgBR using antibodies directed against the HA epitope in order to maintain compatibility with Bouin’s fixation. Endogenous NPC2 has been reported previously to localize to lysosomal, late endosomal, and trans-Golgi network (TGN) compartments (Berger et al., 2007
; Naureckiene et al., 2000
; Zhang et al., 2003
). NgBR-HA colocalizes with NPC2 both in punctuate structures and in a diffuse perinuclear compartment (). In agreement with previous reports, endogenous NPC2 colocalizes to a large extent with cathepsin D (). NgBR-HA, however, does not colocalize with cathepsin D, suggesting an alternate subcellular compartment in which the NgBR-NPC2 interaction takes place (). We tested a variety of organelle markers for colocalization with NgBR-HA, and determined that ER markers such as protein disulfide isomerase (PDI) and calnexin exhibit the highest degree of colocalization (). NPC2 contains a canonical signal peptide and is a soluble luminal glycoprotein, thus a fraction of NPC2 should be present within the ER lumen during its biosynthesis. In order to ascertain whether any degree of similar compartmentalization between NgBR-HA and NPC2 exists at steady-state, we co-labeled cells for NPC2 and PDI. NPC2 partially colocalizes with PDI, suggesting the ER is one compartment in which the NgBR and NPC2 interaction might occur ().
Evidence supporting localization of NgBR and NPC2 in a pre-lysosomal Compartment
We sought an independent, biochemical approach to more rigorously address the localization of NgBR in the context of these studies. Given the enrichment of NgBR in the ER compartment, we utilized a recently described approach to purify ER membranes from cultured cells using a continuous iodixanol gradient separation of ER and lysosomal membranes (Radhakrishnan et al., 2008
). As seen in , the ER marker calnexin predominantly localizes to the denser regions of the iodixanol gradient (designated as region (2) in ), whereas the lysosomal marker Cathepsin D is almost exclusively present in the floating membrane layer (designated as (1) in ). NPC2 again cofractionated with Cathepsin D as expected, however, the levels of endogenous NgBR were beyond detection due to dilution of the fractions. In order to assess the fractionation of NgBR, we ectopically expressed a FLAG-tagged NgBR cDNA and determined that NgBR shows a pattern of fractionation similar to calnexin () and the expression of NgBR enhanced the levels of endogenous NPC2 in regions 1 and 2 of the gradient. Since we could not detect NPC2 in region 2 of the fractions in the absence of ectopic NgBR expression (), yet could detect NPC2 in these regions of the gradient in the presence of NgBR, these data suggest that forced expression of NgBR may result in the appearance of NPC2 in the ER to a more appreciable extent. We tested this hypothesis by performing immunofluoresence in cells expressing either full-length NgBR or the C-terminal truncation mutant of NgBR that does not bind NPC2. In cells transfected with a control plasmid, NPC2 displays a punctate pattern that does not colocalize with the ER marker PDI (). In cells overexpressing WT NgBR, NPC2 exhibits a more reticular distribution coupled with the same punctate pattern seen in control transfected cells (). This distribution of NPC2 is dependent on the presence of the C-terminal region of NgBR, as cells expressing NgBRΔC display a pattern of NPC2 distribution similar to control (). These results are unlikely to be explained by non-specific effects of overexpression, as transfection of the protein palmitoyltransferase DHHC3 (Fernandez-Hernando et al, 2006
) also results in a pattern of NPC2 distribution similar to control plasmid transfected cells (). These data bolster the idea that NgBR functions to stabilize NPC2 during transit in the ER.
Figure 3 (A,B) ER membrane fractionation with iodixanol was performed as described by Radhakrishnan et al (see Supplementary Experimental Procedures). ER vs. lysosomal localization was assessed by fractionation of HEK-293T cells over an iodixanol gradient. Cells (more ...)
Our previous data showing that NPC2 protein levels are enhanced in the presence of NgBR allowed for further biochemical approaches to test this hypothesis. We tested the stage at which the NgBR-NPC2 interaction occurs using several approaches. We blocked NPC2 glycosylation with tunicamycin in the presence or absence of NgBR expression followed by cell lysis and immunoblotting for NPC2. Surprisingly, expression of NgBR increases NPC2 core protein levels in the presence of tunicamycin (), suggesting that NgBR functions to stabilize nascent NPC2. In order to further address the stage at which NgBR affects NPC2, we mutated Asn39 (of mature NPC2) to Gln (NPC2N58Q
has been shown previously to prevent proper targeting of NPC2 to the lysosome, presumably by interfering with a mannose-6-phosphate receptor (MPR)-dependent mechanism (Chikh et al., 2004
; Liou et al., 2006
levels are also higher in the presence of NgBR (), again suggesting a role for NgBR in stabilizing an extra-lysosomal form of NPC2. We next tested the possibility that NgBR might influence lysosomal NPC2 levels in a more direct manner. As has been shown previously, NPC2 is a secreted glycoprotein and conditioned medium from cells expressing NPC2 can functionally rescue cholesterol trafficking defects in cells deficient in endogenous NPC2 expression, providing strong evidence for internalization of exogenous NPC2 into the lysosomal compartment (Naureckiene et al., 2000
; Ko et al., 2003
). We collected and concentrated conditioned medium from CHO cells transfected with myc-tagged NPC2 and then incubated cells treated with Ctrl RNAi or NgBR RNAi for the indicated times (). Rather than result in loss of lysosomal NPC2 stability, downregulation of NgBR in this context seemed to slightly enhance NPC2 uptake. This effect was not seen in control experiments in which U18666A (a drug that induces cholesterol sequestration in late endosomes/lysosomes and mimics an NPC phenotype) was used as a control for accumulation in the lysosomal compartment (). This offers the interesting possibility that loss of NPC2 at the level of the ER may activate mechanisms to augment the uptake and storage of lysosomal NPC2. Finally, to test whether blockade of ER-Golgi trafficking with brefeldin A (BFA) would influence NPC2 stability, control or NgBR expressing CHO cells were treated with different concentrations of BFA. Interestingly, BFA further enhanced the stabilizing effect provided by NgBR (). Coupled with the immunofluorescence results, these data provide evidence of an interaction between NgBR and NPC2 occurring in a pre-lysosomal compartment.
Figure 4 (A) CHO cell lines (see ) were transfected with NPC2, incubated for 24 hours followed by treatment with 5ug/ml tunicamycin (TN) for 6 hours. (B) WT NPC2-myc or NPC2N58Qmyc was transfected into CHO cell lines and levels of NPC2-myc assessed by Western (more ...)
Evidence of a role for NgBR in intracellular cholesterol trafficking
Due to the accumulation of cholesterol that occurs with loss of NPC2 expression, we sought to determine whether a similar phenotype exists with loss of NgBR. In order to examine whether reduction of endogenous NgBR impacts cholesterol trafficking, HepG2 cells were treated with control siRNA (Ctrl RNAi), NgBR siRNA (NgBR RNAi) or U18666A (as a positive control to induce an NPC phenotype), stained with filipin and free cholesterol pools were examined by fluorescence microscopy. NgBR RNAi-treated cells exhibit increased free cholesterol levels as shown by the higher intensity of filipin fluorescence relative to Ctrl RNAi-treated cells (). These data suggest that NgBR influences cellular cholesterol trafficking and that the lack of NgBR expression results in accumulation of free cholesterol. In addition, we derived fibroblasts (MEFs) from WT and NgBR+/− ES cells and assessed free cholesterol levels by filipin staining. NgBR+/− MEFs exhibit an increased level of filipin fluorescence intensity relative to WT MEFs (), providing genetic support for a role for NgBR in intracellular cholesterol trafficking. To address the possibility that Nogo-B may regulate these NgBR-dependent defects in cholesterol trafficking, we stained Nogo-A/B double knockout fibroblasts with filipin. No appreciable differences in free cholesterol levels were observed in Nogo-A/B knockout cells (Figure S3), again suggesting that the NgBR-NPC2 interaction is independent of Nogo-A or –B.
Loss of NgBR induces free cholesterol accumulation in cells
The siRNA sequence used in the preceding experiments targets the 3' untranslated region (UTR) of the NgBR transcript, an approach that allows for the rescue of cells by expression of NgBR lacking the 3' UTR (Miao et al., 2006
). We exploited these sequence differences through introduction of an RNAi-resistant NgBR construct in cells treated with NgBR RNAi. Transduction of cells with adenoviral NgBR (Ad-NgBR) alleviates the increase in free cholesterol and the reduction in NPC2 levels seen after treatment with NgBR RNAi (). In order to address the mechanism by which this accumulation of cholesterol occurs, we treated cells with diminished NgBR with conditioned media collected from cells overexpressing secreted NPC2. Exogenous NPC2, but not mock transfected conditioned media nor media collected from cells expressing NPC2N58Q
, rescued the increase in free cholesterol seen with NgBR RNAi treatment (), suggesting that this effect is NPC2-dependent. Efforts to understand the basis of NPC disease have led to the discovery that the NPC phenotype is in part due to defects in sterol trafficking to sites of oxysterol biosynthesis (Frolov et al., 2003
; Zhang et al., 2008
). To address the question of whether NgBR might function in a similar manner, we incubated NgBR RNAi-treated cells with 25-hydroxycholesterol (25-HC), and again stained with filipin. Cholesterol accumulation seen after NgBR RNAi treatment was alleviated by incubation with 25-HC () as shown previously for cells deficient in NPC2 function by Ory and colleagues (Frolov et al., 2003
), lending further credence to the hypothesis that similar sterol trafficking defects occur with loss of NgBR or NPC2.
NgBR is important for proper maintenance of sterol sensing
Fibroblasts from Niemann-Pick Type C patients exhibit defects in sterol sensing as a result of the cholesterol accumulation that occurs upstream of the ER-resident sterol sensing machinery (Liscum and Faust, 1987
). In order to determine whether NgBR-dependent sterol trafficking defects lead to dysregulation of sterol sensing in a manner reminiscent of NPC2 disease, we reduced NgBR levels with RNAi and assessed sterol responsiveness. Initially, we assessed the relatively acute suppression of SREBP cleavage that occurs with addition of exogenous LDL to cells incubated in the absence of lipoproteins in the media. As seen in , induction of SREBP cleavage occurs when cells are cultured in the absence of lipoproteins (lipoprotein deficient serum, LPDS). This cleavage is partially suppressed when exogenous LDL is added to the medium (, +nLDL). However, knockdown of NgBR or treatment with U18666A abrogated this suppressive effect of LDL on SREBP cleavage. Seeking to determine the functional consequence of this dysregulation, HepG2 cells were treated with Ctrl RNAi or NgBR RNAi, the biosynthetic pool of cholesterol labeled using [14
C]-acetate and de novo
cholesterol biosynthesis measured. Knockdown of NgBR expression leads to an enhanced rate of cholesterol synthesis relative to Ctrl RNAi-treated cells (). Next, we assessed another index of sterol responsiveness by measuring the specific uptake of DiI-labeled LDL. EA.hy cells treated with Ctrl RNAi exhibit a dose-dependent decrease in LDL uptake when incubated for 48hrs in the presence of increasing concentrations of nLDL (data not shown). However, loss of NgBR expression leads to diminished sterol sensitivity, as DiI-LDL uptake is increased upon NgBR knockdown despite pre-incubation of cells with nLDL prior to DiI-LDL treatment (). Moreover, this increase in uptake is mirrored by an increase in cell surface binding of LDL (), suggesting that more LDL-R is present at the cell surface in the absence of NgBR expression. Indeed, HepG2 cells treated with NgBR siRNA exhibit increased levels of total and surface LDL-R (Figure S4B, and , respectively).
NgBR expression is necessary for proper regulation of cellular cholesterol Homeostasis
In order to determine the relative specificity of the effects described above, we performed rescue experiments with an NgBR cDNA construct lacking the RNAi target sequence. Treatment of HepG2 with NgBR RNAi leads to ~90% loss of NgBR expression, while RNAi against NgBR in cells stably expressing the coding sequence alone leads to loss of endogenous NgBR (E) but not transfected NgBR (R) (). Indeed, cells stably expressing transfected NgBR did not show any appreciable increase in LDL-R expression after siRNA treatment (Figure S4C, ; total and surface LDL-R, respectively), proving the specificity of the RNAi effects seen in prior experiments. We also treated cells with 25-HC as in the filipin staining experiments in order to further address the mechanistic basis for the increase in LDL-R levels. Incubation of EA.hy cells with 25-HC results in a dramatic decrease in both total (Figure S4D) and surface () LDL-R expression in cells treated with NgBR RNAi. Next, we performed analogous experiments in primary cultures of human skin fibroblasts since the above experiments were performed in immortalized cell lines. Fibroblasts treated with NgBR RNAi again exhibited increased intracellular cholesterol accumulation (Figure S5A) and diminished suppression of SREBP cleavage in the presence of LDL-cholesterol (Figure S5B), providing support for a role for NgBR in cholesterol homeostasis. Thus, the loss of NgBR leads to dysregulation of sterol homeostasis consistent with the accumulation of free cholesterol as occurs in NPC2 disease.
In summary, we have identified a function for NgBR as an NPC2 binding partner that regulates NPC2 stability/turnover. Our data suggest that the NgBR cis-IPTase homology domain is crucial for its interaction with NPC2 in yeast and heterologous mammalian expression systems. These data are particularly compelling in light of the previously described interaction between NPC2 and the cis-IPTase domain of DHDDS. Future efforts will seek to define a possible relationship between NPC2, NgBR and DHDDS.
Although it is well appreciated that the functional relationship between NPC1 and NPC2 is critical for cholesterol trafficking, our data provide compelling evidence for an unappreciated mode of NPC2 regulation via this newly described interaction. Interestingly, NgBR partially colocalizes with NPC2 in the ER and enhances the levels of nascent NPC2 protein. Our results are in agreement with numerous studies which have shown NPC2 to be primarily localized to a lysosomal compartment at steady-state; however, our data suggests that an additional level of regulation has evolved to ensure stability of NPC2 at its earliest steps of entry into the secretory pathway. The idea that a unique chaperone function may be conferred by non-canonical binding events within the ER is not unprecedented and has been suggested for other proteins (Ju et al., 2008
The effects of NgBR on NPC2 described herein appear to be independent of the only other described binding partner for NgBR, Nogo-B. Loss of NgBR genetically or using siRNA leads to cholesterol accumulation which, in turn, results in defects in ER sterol sensing, both of which are phenotypic hallmarks of mutations in NPC2. Thus, these data reveal insights into NPC2 trafficking, protein stability, and cholesterol homeostasis and raise the possibility that loss of function mutations in NgBR may promote defects in cholesterol metabolism.