The UPR is an essential process by which cells respond to the accumulation of misfolded protein in the ER. This can occur either physiologically, as a result of acute secretory protein load, or as a result of pathological insults that may impede protein folding capacity. The basic ER stress response is well established involving an acute translational attenuation followed by the upregulation of a transcriptional profile that includes ER chaperone genes, ERAD components, protein transport genes, among others . However, it is probable that different cell types have a unique profile of upregulated genes that are dependent on a particular cells function and the secretory proteins it produces. The goal of the present study was to examine the ER stress response in the insulin producing pancreatic β-cell. To do this we developed a pancreatic β-cell culture model with inducible expression of a folding-deficient insulin fusion protein based on the proinsulin mutation found in the Akita mouse . The rationale behind this approach was that regulated expression of a single misfolded protein would allow for a temporal characterization of the UPR in β-cells. This may reflect the ER stress response to physiological conditions (acute insulin synthesis) and potentially a chronic response induced by pathological conditions occurring in type 2 diabetes (chronic insulin biosynthetic demands coupled with the effects of chronic free fatty acids, glucose and cytokines that induce ER stress).
Using the tetracycline/doxycycline-regulated expression system we created a double stable INS-1 cell clone with stable integration of the pTet-ON regulatory plasmid and the Ins2 (C96Y)-EGFP plasmid driven by the Tet/Dox-responsive promoter. The EGFP tag was used in order to facilitate selection of stable clones, as well as for sorting of the positive clones for cells with high expression of the Ins2 (C96Y)-EGFP fusion protein. In the clone selected for the studies basal expression of the mutant insulin is low, although detectable. Following doxycyline addition the mutant insulin is markedly induced by 24 h, reaching maximal expression by ~72 h and expression is maintained for several days. Expression of the mutant insulin fusion protein leads to a swollen ER lumen that is readily detectable in many cells and is indicative of misfolded protein accumulation and a stressed ER. This result is similar to the phenotype observed in the β-cells of Akita mice . As expected, expression of folding-deficient Ins 2 (C96Y)-EGFP fusion protein resulted in the induction of UPR pathways (PERK, IRE1 and ATF6).
To analyze the ER stress response in the cell line we performed microarray expression profiling and real time PCR analysis at various times following mutant protein expression. We chose time-points of Ins2 (C96Y)-EGFP expression when no significant apoptosis was observed (24 h) and longer expression time points (48 h and 5 days), when apoptotic cell death was detectable. After 24 h, 45 genes were induced, while the number of induced genes was substantially increased after 72 h (86) and 5 days (68). Interestingly, the number of downregulated genes was also increased at the longer incubation times (37 and 56 after 48 h and 5 d of doxycycline induction, respectively). This may reflect the observation that prolonged activation of IRE1α can result in relaxed specificity and cleavage of cellular mRNAs , resulting in the downregulation of some genes as a result of prolonged ER stress.
After 24 h of Ins2 (C96Y)-EGFP expression the most consistently induced genes were ER resident chaperones and co-chaperones (Grp78, Erdj4, P58
, Erdj3, Fkbp11) and an ERAD component (Herp). The J-domain containing co-chaperones (Erdj4, P58
) have been shown to be induced by ER stress in various cell types [20, 25–32]. Recently, P58IPK and ERdj3 have been found to interact with misfolded proteins in the ER prior to recruitment of the chaperone GRP78 [26, 29, 30, 32]. In addition, these proteins have also been implicated in targeting misfolded proteins for degradation via the ERAD system [20, 28]. Thus, expression of misfolded insulin causes the induction of chaperone proteins early in an effort to fold the mutant insulin. These chaperones are maintained at high levels throughout the time course of mutant insulin induction and some may assist in degradation of the misfolded molecule once folding is found to be futile (discussed below). The importance of P58IPK function in normal pancreatic β-cells in vivo is evidenced by increased pancreatic β-cell apoptosis and hyperglycemia in knock-out mice . However, the detailed molecular function of these co-chaperones in β-cells has not been examined to date.
Interestingly, although there are six mammalian DNAJ domain-containing proteins, we only found a subset to be induced by mutant insulin expression. Notable among those that were not upregulated is ERdj5. Recently, expression of a folding-deficient version of the secreted protein surfactant protein C in HEK293 cells caused the induction of various ER stress response genes, including Erdj4 and Erdj5, which were shown to be required for degradation of the folding-deficient surfactant protein C . Surfactant protein C, like proinsulin, contains disulfide bonds, but is not glycosylated. Thus, it is interesting that we do not observe induction of Erdj5 in β-cells expressing folding-deficient proinsulin. A recent report, however, has shown that ERdj5 can interact with EDEM proteins and is required for reducing disulfide bonds on misfolded, glycosylated proteins prior to retrotranslocation . EDEM genes were not detected in the microarray analysis, suggesting that proinsulin may not require EDEM or ERdj5 for its degradation.
Comparing the UPR response observed in our study to ER stress responses in other cell types shows that some of the early chaperone genes induced by mutant protein expression are not observed in other systems. For example, tunicamycin treatment of neuroblastoma cells  or expression of a retroviral protein in astrocytes, which induces ER stress , did not cause an upregulation of Erdj4, P58
, Erdj3, or Fkbp11. Importantly, all of these genes have been shown to be induced by ER stress caused by prolonged palmitate exposure in cultured mouse MIN6 cells . Saturated FFAs have been shown to induce ER stress in pancreatic β-cells and this may contribute to β-cell dysfunction [15, 36, 37]. In addition, a recent proteomic study examining islets from a diabetic insulin resistant mouse model has shown that the protein (and mRNA levels) of some of these (P58IPK , ERdj3, Fkbp11) are upregulated . Thus, the induction of these genes/proteins are likely important cell protective ER stress response proteins in β-cells and a detailed biochemical analysis of their function in mediating proinsulin folding and/or degradation is warranted.
In addition to ER chaperone genes and ERAD components, several other genes were found to be induced early as a result of Ins2 (C96Y)-EGFP expression, including Sdf2l1, Armet and Creld2. All three were abundantly upregulated at both early and late time points. The exact functions of the protein products of these genes in the ER stress response is largely undefined, although these proteins have also been reported to be induced by ER stress-inducing compounds such as tunicamycin in other cell types [39–41].
The appearance of prominent lower migrating bands recognized by anti-GFP antibodies indicates that the Ins2 (C96Y)-EGFP fusion protein is being degraded. Similar cleavage products have been detected with wild-type proinsulin tagged with GFP at the C-terminus and was suggested to occur from non-specific cleavage of the fusion protein at some point along the secretory pathway . In the case of the Ins2 (C96Y)-EGFP the cleavage products are not efficiently immunoprecipitated by anti-GFP antibodies and are not detected in ER fractions. Thus, the smaller migrating bands recognized by GFP antibodies are likely degradation products of proteolysis occurring in the cytoplasm. The full-length mutant insulin fusion protein appears to be inaccessible to the secretory pathway and is retained in the ER, which is consistent with a report showing that the C96Y mutation in proinsulin results in misfolding of the molecule and retention in the ER .
The Ins2 (C96Y)-EGFP fusion protein is likely degraded by the ubiquitin-dependent ERAD system. The ERAD-associated genes Herp and Sel1 were induced by mutant insulin expression, which is consistent with the finding that Sel1 is upregulated in Akita islets . Inhibiting the 26S proteasome resulted in an increase in the levels of Ins2 (C96Y)-EGFP and sensitized the cells to cell death. This was also the effect observed by reducing Herp expression using siRNA, a protein recently shown to be required for the degradation of a disulfide bond-containing mutant version of the kappa light chain . The process of Ins2 (C96Y)-EGFP degradation however, may be quite complex. If the proteasome was responsible for generating the degradation fragments observed (Figure 2) than inhibiting the proteasome would be expected to not affect the steady-state levels of these fragments; new fragments would be prevented from being generated from Ins2 (C96Y)-EGFP and fragments already present prevented from being degraded. However, most degradation fragments were actually increased by proteasome inhibition (Figure 8B), suggesting that other proteases may be responsible for generating the degradation fragments that are then targeted for ubiquitin-dependent degradation.
It should be pointed out that although expression of the Ins2 (C96Y)-EGFP fusion protein induces ER stress and degradation of the fusion protein, due to the presence of the EGFP tag the degradation of the fusion construct may not be identical to the degradation of untagged proinsulin C96Y. Furthermore, it is possible that the ER stress response elicited by untagged proinsulin C96Y is different compared to that of Ins2 (C96Y)-EGFP. These possibilities should be explored in future studies.
This model system also allowed us to examine potential gene expression changes associated with ER stress-induced cell death. An upregulation of the pro-apoptotic transcription factor Chop and one of its target genes (Trib3) was evident by 24 h of Ins2 (C96Y)-EGFP expression, which is indicative of PERK pathway activation and is likely to be a normal physiological response. At this time point no significant apoptosis is detected in the population. After 48 h of mutant insulin expression however, the levels of Chop and Trib3 were even higher, which coincided with detection of apoptosis in the population. Thus, a sustained increase in the levels of CHOP and TRIB3 proteins may tilt cells towards apoptosis induction. Pancreatic β-cells of CHOP-deficient heterozygous Akita mice are partially protected from cell death , although clearly there are other factors also involved. In the mutant insulin expressing cell line the most highly induced gene after 48 h was Trib3. Recently, the TRIB3 protein has been shown to be induced by the PERK-ATF4-CHOP pathway and has been implicated in mediating apoptosis [18, 19] and in hyperglycemia-induced pancreatic β-cell death . The mechanism of this effect, however, is unclear. Whether the TRIB3 protein is responsible for apoptosis induction and how much of the apoptosis is TRIB3-dependent in this model system requires future study.
Despite the abundant upregulation of Chop and Trib3 in response to mutant insulin expression, the level of apoptotic cells in the population remained rather low, even after long term expression of the mutant construct. This is likely because one of the main cell protective effects induced by the UPR, ERAD, is also upregulated. Inhibition of ERAD by inhibiting the cytosolic proteasome or inhibiting Herp upregulation sensitizes the cells to apoptosis, supporting the notion that the ERAD pathway may support cell survival by degradation of misfolded insulin.