Intramembrane glycine mediates multimerization of Insig-2, a requirement for sterol regulation in Chinese hamster ovary cells

MATERIALS AND METHODS

RESULTS

To identify amino acids in Insig-2 that are important for sterol-mediated effects on Scap and reductase, Insig-1-deficient SRD-14 cells were mutagenized with EMS, followed by selection for growth in medium containing 5% lipoprotein-deficient serum and 1 μg/ml 25-HC. This procedure yielded seven 25-HC-resistant colonies, one of which was isolated, expanded, and designated SRD-20. Figure 1A compares levels of nuclear SREBP-2 [the SREBP isoform that preferentially activates genes encoding cholesterol biosynthetic enzymes (32)] in wild-type CHO-7, Insig-1-deficient SRD-14, Insig-1- and Insig-2-deficient SRD-15, and SRD-20 cells. The cells were incubated in sterol-depleting medium containing lipoprotein-deficient serum, the reductase inhibitor compactin, and a low level of mevalonate that permits synthesis of essential nonsterol isoprenoids, but not cholesterol (33). Some of the dishes also received 1 μg/ml 25-HC. Following incubation for 16 h, the cells were harvested, lysed, and subjected to cell fractionation to generate membrane and nuclear extract fractions. Aliquots of each fraction were subjected to SDS-PAGE and immunoblotted with antibodies against Scap and SREBP-2. Processed forms of SREBP-2 accumulated in the nuclear extracts of sterol-depleted CHO-7, SRD-14, SRD-15, and SRD-20 cells (top panel, lanes 1, 3, 5, and 7, respectively). The addition of 25-HC blocked the appearance of nuclear SREBP-2 in CHO-7 and SRD-14 cells (lanes 2 and 4) but not in SRD-15 and SRD-20 cells (lanes 6 and 8, respectively). Suppression of SREBP processing in SRD-14 cells following extended sterol treatment (>16 h) is consistent with our previous findings (24). All of the cells produced similar amounts of Scap protein in membranes, regardless of the absence or presence of 25-HC (bottom panel).

The failure of SRD-20 cells to respond to the suppressive effects of 25-HC treatment suggested they may be deficient in Insig-2 as well as Insig-1. Thus, total RNA was isolated from SRD-14 and SRD-20 cells and subjected to reverse transcription reactions. PCR amplification of first-strand cDNA from SRD-14 cells produced a single 678 bp band for Insig-2 (Fig. 1B). A band of similar size was generated from SRD-20 first-strand cDNA, but an additional 368 bp band was also observed. Both forms of the Insig-2 cDNA from SRD-20 cells were subcloned, and 10–30 individual clones were subjected to DNA sequencing, which revealed the 678 bp fragment corresponds to full-length Insig-2 containing a G-to-A substitution that changes Gly-39 to Glu. It should be noted that the G-to-A substitution was not present in the Insig-2 cDNA from SRD-14 cells. As shown in Fig. 1C, Gly-39 is predicted to lie in the middle of the first of six transmembrane helices of Insig-2 (19). Sequence alignments revealed that Gly-39 is completely conserved in hamster, human, mouse, rat, xenopus, and zebrafish Insig-2 (Fig. 1D); it is also present in human Insig-1. DNA sequencing revealed that the 386 bp fragment encodes a splice variant of Insig-2 in which Exon 2 is fused to Exon 5, creating a frame shift. The resulting cDNA encodes amino acids 1–81 of Insig-2, followed by 22 novel amino acids and a stop codon; this fusion protein is likely nonfunctional.

In the experiment of Fig. 2A, B, CHO-7 cells were transfected with wild-type or mutant versions (G95E for Insig-1 and G39E for Insig-2) of pCMV-Insig-1/-2-Myc, expression plasmids that encode full-length human Insig-1 or Insig-2 followed by six copies of a c-Myc epitope. Following sterol depletion, the cells were incubated for 5 h in the absence or presence of 25-HC, after which they were harvested and subjected to cell fractionation. Immunoblots of the resulting membrane fractions with anti-Myc revealed dose-dependent expression of wild-type and mutant Insig-1; similar amounts of expression were observed in the absence and presence of 25-HC (Fig. 2A, compare lanes 2–7 with 8–13). We observed similar dose-dependent expression of wild-type Insig-2-Myc (Fig. 2B, lanes 2–7), but a different result was obtained for the G39E mutant. At low plasmid levels (0.03 μg), G39E Insig-2 was not detected in membranes (lanes 8 and 9). However, the mutant protein was visualized when 0.1 μg plasmid was transfected, but only when the cells were treated with 25-HC (compare lanes 10 and 11). At the highest level of plasmid (0.3 μg), G39E Insig-2 was detected in sterol-depleted cells and expression was marginally increased in the presence of 25-HC (lanes 12 and 13).

We next conducted an experiment to determine whether Scap mediates sterol-dependent expression of G39E Insig-2 (Fig. 2C). Scap-deficient SRD-13A cells (29) were transfected with low levels of G39E Insig-2 in the absence or presence of an expression plasmid encoding hamster Scap (pCMV-Scap). Following sterol depletion, some of the cells were treated with 25-HC or the proteasome inhibitor MG-132 and harvested for cell fractionation after 5 h. Immunoblotting revealed constant expression of wild-type Insig-2 that was not influenced by the presence of 25-HC or MG-132 (Fig. 2C, top panel, lanes 1–3); coexpression of Scap had little effect on this expression (top and bottom panels, lanes 7–9). In the absence of Scap, G39E Insig-2 was not detected in untreated cells or cells treated with 25-HC (lanes 4 and 5); however, the protein was readily visualized upon the addition of MG-132 (lane 6). With Scap coexpression, the mutant Insig protein was stabilized to a similar extent by the addition of 25-HC or MG-132 (lanes 10–12). These findings indicate that G39E Insig-2 undergoes proteasome-mediated degradation in sterol-depleted cells and becomes stabilized upon sterol repletion, which likely results from sterol-induced binding of the protein to Scap. This conclusion is supported by the coimmunoprecipitation experiment of Fig. 2D in which SRD-13A cells were transfected with Scap and either wild-type or G39E Insig-2. After sterol depletion for 16 h, the cells were then incubated in the absence or presence of 25-HC, after which detergent lysates were prepared and transfected Insig was immunoprecipitated with anti-Myc. Immunoblotting of the resulting pellet and supernatant fractions of the precipitation revealed that both wild-type and mutant Insig-2 form a sterol-induced complex with Scap (lanes 5 and 7, respectively). Similar results were obtained for wild-type and mutant versions of Insig-1 (Fig. 2E).

Considering that the Gly to Glu mutants of Insigs continue to bind Scap in the presence of 25-HC, we next designed an experiment to determine whether these mutants are capable of mediating sterol regulation of SREBP processing and reductase degradation. SRD-13A cells were transfected with wild-type or mutant Insig-1 and Insig-2, Scap, and pTK-HSV-SREBP-2, which encodes human SREBP-2 preceded by three copies of an HSV epitope. The cells were acutely depleted of sterols in medium containing LPDS, compactin, and hydroxypropyl cyclodextrin, a reagent that removes cholesterol from the plasma membrane of intact cells. The cells were then further incubated in the absence or presence of 25-HC for 5 h prior to harvesting and cell fractionation. In the absence of wild-type Insig-1 or Insig-2, we observed a band for processed, nuclear SREBP-2 that persisted in the presence of 25-HC (Fig. 3A, B, top panel, lanes 1 and 2). The coexpression of wild-type Insig-1 or Insig-2 led to the disappearance of this band upon 25-HC treatment (lanes 3 and 4); similar results were obtained with the Gly to Glu mutants of both Insigs (lanes 5 and 6). In contrast, G95E Insig-1 and G39E Insig-2 failed to mediate sterol-accelerated degradation of reductase as shown in Fig. 3C, D. CHO-7 cells were transfected with pCMV-HMG-Red-T7 (TM1-8), which encodes transmembrane helices 1–8 of hamster reductase (amino acids 1–348) followed by three copies of the T7 epitope; this region of reductase has been shown to be both necessary and sufficient for Insig-mediated degradation (6, 17). The membrane domain of reductase was completely refractory to sterol-induced degradation when transfected alone, owing to the saturation of endogenous Insigs (Fig. 3C, D, top panel, lanes 1 and 2). Coexpression of increasing amounts of wild-type, but not the Gly to Glu mutants of Insigs, restored sterol-accelerated degradation upon overexpressed reductase (compare lanes 3–6 with lanes 7–10 for Insig-1 and lanes 7–14 for Insig-2). The failure of the mutant Insigs to mediate reductase degradation is consistent with the coimmunoprecipitation experiment of Fig. 3E. CHO-7 cells transfected with the membrane domain of reductase together with either wild-type or G95E Insig-1 were subjected to a brief 25-HC treatment in the absence or presence of MG-132. Subsequently, the reductase membrane domain was immunoprecipitated from detergent lysates. Immunoblots of precipitated material revealed the presence of wild-type, but not G95E, Insig-1 in untreated and MG-132-treated cells (compare lanes 2 and 5 with 3 and 6).

Results presented thus far present an interesting conundrum in that G95E Insig-1 and G39E Insig-2 retain the capacity to mediate sterol regulation of Scap-SREBP, but not reductase. A recent study reported that purified Insigs bind to 25-HC as a dimer (34), which raises the question as to whether the proteins act as dimers (or multimers) in intact cells to regulate SREBP processing. Thus, we next conducted a coimmunoprecipitation experiment to probe for Insig multimerization in CHO-7 and SRD-15 cells, which provide the Insig-null background that is required to test our hypothesis (Fig. 4). For this purpose, the cells were transfected with various combinations of wild-type and mutant Insig-1 plasmids that encoded proteins fused to either T7 or Myc epitope tags. We used Insig-1 rather than Insig-2 to avoid effects that may be caused by Scap-dependent stabilization of mutant Insig-2 (see Fig. 2). Following transfection, detergent lysates were prepared and subjected to immunoprecipitation with anti-T7 antibody. Immunoblot analysis of the resulting pellet fractions with anti-T7 and anti-Myc antibodies revealed that the wild-type and mutant Insig-1-Myc coprecipitated with both version of Insig-1-T7 in CHO-7 cells (Fig. 5B, lanes 4–7). In SRD-15 cells, wild-type Insig-1-T7 precipitated both versions of the Myc-tagged protein (lanes 10 and 11). Similarly, the G95E Insig-1-T7 protein formed a complex with wild-type Insig-1-Myc (lane 12); however, it failed to associate with the G95E mutant (lane 13).

To correlate Insig-Insig coimmunoprecipitation shown in Fig. 4 with sterol-regulated processing of SREBP-2, we conducted the experiments shown in Fig. 5. The results of Fig. 5A show that as expected, wild-type forms of Insig-1 and Insig-2 conferred sterol regulation upon processing of overexpressed SREBP-2 in SRD-15 cells (top panel, compare lanes 1 and 2 with 3–6 for Insig-1 and lanes 11 and 12 with 13–16 for Insig-2). However, sterols failed to block processing of SREBP-2 in transfected SRD-15 cells when the Gly to Glu mutants of Insigs were coexpressed (lanes 7–10 for Insig-1 and 17–20 for Insig-2). Similar to the results in CHO-7 cells, the mutant forms of both Insigs did not mediate sterol-accelerated degradation of the reductase membrane domain in SRD-15 cells (data not shown). In the experiment of Fig. 5B, SRD-15 cells were transfected with pTK-HSV-SREBP-2 and either a low level (10 ng) of wild-type Insig-1 plasmid or a high level (0.2 μg) of the inactive mutant. Both forms of Insig-1 failed to mediate sterol suppression of nuclear SREBP-2 when transfected alone (top panel, lanes 3–6); however, coexpression of as little as 3 ng of wild-type Insig-1 together with the inactive mutant led to 25-HC-mediated inhibition of SREBP-2 processing (lanes 7 and 8). Similar effects were observed with 10 ng of wild-type Insig-1 (lanes 9 and 10). Considered together, the results of Figs. 4 and 5 support the hypothesis that Insig-1 functions as a multimer, the formation of which requires an intramembrane glycine residue at position 95.

DISCUSSION

Our previous studies show that Insig-1-deficient SRD-14 cells fail to grow indefinitely when cultured in the presence of 25-HC (24). We attributed this failure to residual sterol regulation, owing to the remaining Insig-2 in SRD-14 cells. This hypothesis was substantiated in subsequent experiments in which mutagenized SRD-14 cells were subjected to 25-HC selection, yielding SRD-15 cells that are deficient in Insig-2 as well as Insig-1 (25). In this study, we exploited the 25-HC sensitivity of SRD-14 cells to isolate a new line of mutant CHO cells designated SRD-20. Consistent with their growth phenotype, SRD-20 cells were fully resistant to 25-HC-mediated suppression of SREBP processing (Fig. 1A). DNA sequencing revealed that SRD-20 cells harbor a point mutation in one Insig-2 allele that produces a truncated, nonfunctional protein and another point mutation in the other Insig-2 allele that yields a glutamic acid for glycine-39 substitution. We failed to find mutations in any of the other genes known to participate in the SREBP pathway, including SREBP-1, SREBP-2, and Scap (data not shown). Gly-39 in Insig-2 corresponds to Gly-95 in Insig-1 and is predicted to lie in the middle of the first membrane-spanning segment of Insigs (Fig. 1C). Transfection experiments reveal that in sterol-depleted cells, G39E Insig-2 is rapidly degraded in a saturable fashion (Fig. 2B). This degradation is blocked by the addition of the proteasome inhibitor MG-132 or sterols (Fig. 2C). Sterol-dependent stabilization of G39E Insig-2 requires the presence of Scap, resembling that of wild-type Insig-1 when expressed at near endogenous levels (23). In sterol-deprived cells, Insig-1 binds the ubiquitin ligase gp78, which initiates Insig-1 ubiquitination and subsequent degradation (22). Sterols prevent degradation of Insig-1 by displacing gp78, which results from sterol-induced binding of Scap to Insig-1. In contrast, Insig-2 is a stable protein in both the absence and presence of sterols. The current data indicate that substitution of Glu for Gly-39 in Insig-2 destabilizes the protein, causing its recognition and subsequent ubiquitination by gp78. However, the mechanism by which G39E Insig-2 is recognized for ubiquitination by gp78 and whether this selection is similar to that of Insig-1 is presently unknown. It should be noted that destabilization by the Gly to Glu substitution does not destabilize Insig-1 (Fig. 2A), indicating that the effect is Insig-2 specific.

In wild-type CHO-7 cells, Scap-dependent stabilization of G39E Insig-2 correlated with the ability of the mutant protein to bind Scap and mediate sterol regulation of SREBP processing (Fig. 3A, B). These results contrast those obtained for reductase degradation in that G95E Insig-1 and G39E Insig-2 failed to mediate sterol-accelerated degradation of reductase and did not bind the enzyme in the presence of sterols (Fig. 3C–E). A plausible explanation for this discrepancy can be provided by the recent finding that purified Insig-2 binds 25-HC as a dimer (34). Thus, we postulated that when overexpressed, the mutant Insigs can associate with their endogenous equivalent to form active complexes capable of sterol-induced binding to Scap. When experiments were conduct in SRD-15 cells, both mutant Insigs failed to mediate sterol regulation of Scap and reductase (Fig. 4). However, sterol regulation of SREBP processing was restored in SRD-15 cells when substoichiometric amounts of wild-type Insig-1 were coexpressed with the inactive mutant (Fig. 5A). Coimmunoprecipitation experiments indicate that G95E Insig-1 can only form a multimeric complex in the presence of wild-type Insig-1 (Fig. 5B). These findings indicate that Insigs function as multimers to mediate sterol regulated processing of SREBPs and that one member of the complex must contain a glycine residue at position 39.

Although the combination of wild-type and mutant Insigs can bind Scap in the presence of sterols and prevent SREBP processing, the complex cannot bind reductase and accelerate its degradation. Although Insigs bind to both Scap and reductase, the current data suggest that formation of these sterol-dependent complexes occur through distinct sites on Insigs. It is notable that the Scap and reductase binding sites on Insigs exhibit some degree of overlap as indicated by the inhibition of reductase degradation by overexpression of the Scap sterol-sensing domain (6). However, complete resolution of this requires comparisons of detailed structural analyses of the sterol-dependent reductase-Insig and Scap-Insig complexes.