Opposing Roles for ATF6α and ATF6β in Endoplasmic Reticulum Stress Response Gene Induction

The endoplasmic reticulum (ER) transmembrane proteins, ATF6α and ATF6β, are cleaved in response to ER stress, which can be induced by tunicamycin. The resulting N-terminal fragments of both ATF6 isoforms, which have conserved basic leucine-zipper and DNA binding domains but divergent transcriptional activation domains, translocate to the nucleus where they bind to ER stress-response elements (ERSE) in ER stress-response genes (ERSRG), such as GRP78. Although it is known that ATF6α is a potent activator of ERSRGs, the transcriptional potency and functions of ATF6β remain to be explored. Accordingly, N-terminal fragments of each ATF6 isoform (N-ATF6α and N-ATF6β) were overexpressed in HeLa cells and the effects on GRP78 induction were assessed. When expressed at similar levels, N-ATF6α conferred ∼200-fold greater GRP78 promoter activation than N-ATF6β. Because ER stress activates nuclear translocation of both ATF6α and β and because both bind to ERSEs, the effect of co-expressing them on GRP78 induction was assessed. Surprisingly, N-ATF6β inhibited N-ATF6α-mediated GRP78 promoter activation in a dominant-negative manner. Moreover, N-ATF6β inhibited TN-mediated GRP78 promoter activation, which requires endogenous ATF6α. ATF6 isoform-specific small inhibitory RNAs were used to show that, as expected, endogenous ATF6α was required for maximal ERSRG induction; however, endogenous ATF6β moderated ERSRG induction. These results indicate that compared with ATF6α, ATF6β is a very poor activator of ERSRG induction and it represses ATF6α-mediated ERSRG induction. Thus, ATF6β may serve as a transcriptional repressor functioning in part to regulate the strength and duration of ATF6α-mediated ERSRG activation during the ER stress response.

coded by a number of genes induced under such conditions are targeted to the ER where they act as chaperones to aid in folding and thus counteract the stress. A regulatory element located in many of these ER stress-response genes (ERSRG) is the ER stress-response element (ERSE), which is required for transcriptional induction.
ATF6␣, a member of the ATF/CREB family of transcription factors, is required for the maximal induction of numerous ERSGs (5,6). ATF6␣ is comprised of 670 amino acids and resides in the ER membrane (6). Upon ER stress, the cytosolic N-terminal portion of ATF6␣ (N-ATF6␣) comprising ϳ400 amino acids is released as a result of regulated intramembrane proteolysis (RIP) (7,8). N-ATF6␣ possesses a transcriptional activation domain (TAD), basic leucine-zipper (b-Zip) domain, DNA binding domain, and nuclear localization signals. N-ATF6␣ translocates to the nucleus where it combines with several other proteins to form an ERSE-binding complex that is responsible for the induction of ERSGs such as the ER chaperone, glucose-regulated protein 78 kDa (GRP78) (7)(8)(9). N-ATF6␣ is rapidly degraded in a proteasome-mediated process, and mutated inactive forms of N-ATF6␣ are slowly degraded (10). Thus, similar to several other potent transcription factors that exert rapid transient effects (11), the degradation of ATF6␣ upon transcriptional engagement apparently serves as a mechanism to rapidly turn off ERSG induction.
Another member of the ATF/CREB family of transcription factors (G13), which is also an ER membrane protein, was recently shown to be structurally homologous to ATF6␣ and to be cleaved during ER stress to generate an N-terminal fragment of approximately 400 amino acids (12). Because of its high degree of homology to ATF6␣, G13 has been named ATF6␤ (12). N-ATF6␣ and N-ATF6␤ possess highly conserved b-Zip domains and DNA binding domains. This conservation apparently allows N-ATF6␣ and ␤ to bind to ERSEs as homodimers or heterodimers (13); however, the functional significance of the binding of N-ATF6␤ to ERSEs, either alone or as a heterodimer with N-ATF6␣, is currently unknown.
Although N-ATF6␣ and ␤ have conserved b-Zip domains and DNA binding domains located near the center of each protein, the structure of the N-terminal region of N-ATF6␣, which possesses the TAD, differs markedly from that of N-ATF6␤. Notably, an 8 amino acid sequence located in the TAD of ATF6␣, which is required for maximal transcriptional activity (10), is absent from ATF6␤. This structural difference led us to hypothesize that compared with N-ATF6␣, N-ATF6␤ should exhibit much lower transcriptional activity (10). Moreover, if ATF6␤ is a poor transcriptional activator yet can bind to ERSEs with or in place of ATF6␣, it seems probable that ATF6␤ might serve as an endogenous repressor of the transcriptional induction effects of ATF6␣. This study was undertaken to test these hypotheses.

Plasmids
Cytomegalovirus-Galactosidase-Cytomegalovirus-␤-galactosidase, which codes for a galactosidase reporter driven by the cytomegalovirus promoter, was used to normalize for transfection efficiency.
N-ATF6␤ and N-ATF6␤-DN-N-ATF6␤ (3ϫ FLAG-ATF6␤-(1-392)) was prepared by PCR using HeLa cDNA as a template and PCR primers that introduced an XhoI site at amino acid 1 and a termination codon and EcoR1 site at amino acid 392. The resulting PCR product was cloned into 3ϫ FLAG-pcDNA3.1, which was described previously (10). N-ATF6␤-DN (3ϫ FLAG-ATF6␤-(116 -392)) was prepared by PCR using the FLAG-ATF6␤-(1-392) as a template and a PCR primer that introduced an XhoI site at amino acid 116 and the reverse sequencing primer. The resulting PCR product was excised with XhoI and EcoR1 and cloned into 3ϫ FLAG-pcDNA3.1.
GRP78-ERSE-Luc-This construct encodes an active ERSE from the human GRP78 gene driving SV40/luciferase in the vector pGL2-p (Promega) and has been described previously (11). GRP78-Promoter-Luc-This construct encodes the GRP78 promoter from Ϫ284 to ϩ7 driving luciferase. It was created by PCR using HeLa genomic DNA as a template and a sense primer that begins at base Ϫ284 and includes a KpnI restriction site and an antisense primer at ϩ341. The resulting PCR product was digested with KpnI and SalI, creating GRP78-Promoter (Ϫ284 to ϩ7), which was ligated into the vector pGL2 (Promega).
Luciferase-After cell lysis and centrifugation as described above, 100-l samples of cell lysate were combined with 100 l of luciferase buffer (the above described lysis buffer containing 45 mM MgSO 4 , 0.3 mM D-luciferin, and 3 mM ATP). An Optocompt II luminometer (MGM Instruments, Inc) was used to measure light emission of each sample for 10 s. Relative luciferase activities were determined by dividing luciferase values by ␤-galactosidase values. All of the values shown are the means of three cultures Ϯ S.E.
Immunoblotting-Cultures were extracted in a lysis buffer composed of 20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, 10 mg/ml leupeptin, and 10 mg/ml aprotinin. After clearing by centrifugation, the protein concentration of the lysate was determined, and after dilution with the appropriate amount of 2ϫ Laemmli buffer, equal amounts of protein from each sample were fractionated by 10% SDS-PAGE and then transferred to a polyvinylidene difluoride membrane. Membranes were then probed with a GRP78 antiserum (sc-1050, Santa Cruz Biotechnology) or with a FLAG antiserum, M2 antibody (F-3165, Sigma), or with antisera raised against human ATF6␣ or human ATF6␤ (a gift from Dr. K. Mori).
Pulse-Chase Labeling-HeLa cells were transfected with 30 g of the test expression vector as described above, and 3 ϫ 10 6 cells were plated on 60-mm dishes. Following a 48-h incubation in serum-containing medium, the cells were rinsed three times with warm Hanks' buffer (Invitrogen) and incubated for 2 h with 250 Ci of [ 35 S]methionine/ cysteine (Easytag Express protein-labeling mixture (PerkinElmer Life Sciences)) diluted in 2 ml of methionine/cysteine-free DMEM (Invitrogen) supplemented with 5% dialyzed fetal calf serum (Invitrogen). Following a 2-h incubation, this medium was removed and the cultures were washed twice with DMEM and then incubated with 3 ml of chase medium (DMEM containing 10% fetal calf serum, 2 mM unlabeled methionine, and 2 mM unlabeled cysteine). At the indicated times, the cells were rinsed three times with PBS and then scraped into 100 l of lysis buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 5 mM sodium pyrophosphate, 10 mM sodium glycerophosphate, 50 mM NaF, and 150 mM NaCl) containing 0.1% SDS. The lysate was then diluted with 400 l of lysis buffer without SDS to a final concen-tration of 0.02% SDS and then cleared by centrifugation. The supernatant was incubated 12-18 h at 4°C with 2 g of FLAG antiserum (F-3165, Sigma) followed by incubation with protein G-Sepharose beads and elution with SDS-PAGE sample buffer. Immunoprecipitated material was then resolved by SDS-PAGE (10% SDS gel) and dried down and exposed on a phosphorimaging screen.
Small Interfering RNAs-Small interfering (si)RNA was generated using a kit (Dicer siRNA generation kit, catalog number T510001) from Gene Therapy Systems, Inc. (San Diego, CA) according to the manufacturer's protocol. Forward and reverse PCR primers were designed such that they would generate PCR products between 500 and 1,000 bp in length. For human ATF6␣, the forward primer was 5Ј-GCCTTTATT-GCTTCCAGCAG-3Ј and the reverse primer was 5Ј-TCTTCGCTTTG-GACTAGGGAC-3Ј. For human ATF6␤, the forward primer was 5Ј-CCGATCTTCCCAGATCTTCAG-3Ј and the reverse primer was 5Ј-CACCCTGGATGAGGACAACTG-3Ј. The ATF6 sequences in each primer were flanked on the 5Ј ends with 20 nucleotides from the T7 RNA polymerase promoter. Human ATF6␣ or ␤-cDNAs were used as the templates for PCR. After PCR, the products (586 bp for ATF6␣; 622 bp for ATF6␤) were used as templates for in vitro transcription using T7 FIG. 3. N-ATF6␣ and N-ATF6␤ expression levels. Panel A, FLAG Western blot. HeLa cells were transfected with increasing amounts of expression constructs encoding FLAG-N-ATF6␣ or FLAG-N-ATF6␤. After 48 h, culture extracts were analyzed by SDS-PAGE and FLAG immunoblotting as described under "Materials and Methods." Panel B, FLAG immunoblot quantitation. HeLa cells were transfected, and extracts were assessed for FLAG-ATF6 expression as described above. Blots were analyzed densitometrically. All of the values (mean Ϯ S.E.) were normalized to the maximum expression level (n ϭ 3 cultures/plasmid level). * and † ϭ p Ͻ 0.05, different from all of the other values in the panel as determined using ANOVA followed by Newman-Keuls post hoc analysis. HeLa cells were transfected, and pulsechase labeling was performed as described above with the exception that the chase times were adjusted to better span the different half-lives exhibited by each ATF6 isoform. FLAG immunoprecipitation and SDS-PAGE were followed by densitometric analyses of FLAG-N-ATF6␣ and FLAG-N-ATF6␤. The results are shown as a percentage of the initial amount of label in each band. Each time point represents the mean Ϯ S.E. of three cultures.
RNA polymerase, which resulted in the formation of dsRNA ATF6related fragments. The resulting material was treated with DNase to remove the DNA template and RNase to remove any ssRNA. The resulting ATF6 dsRNA was then cleaved into small segments of RNA with the RNA Dicer enzyme and then transfected into HeLa cells using LipofectAMINE according to the manufacturer's protocol.
Immunocytofluorescence-Following transfection with the appropriate siRNA, HeLa cells were incubated in 10% fetal calf serum for 48 h. After washing twice with PBS, the cells were fixed with 4% paraformaldehyde, washed with PBS, permeabilized with 0.1% bovine serum albumin and 0.2% Triton X-100 in 1ϫ TBST, and blocked with 5% bovine serum albumin/TBST for 45 min. For detection of endogenous ATF6␣ or ␤, cells were incubated with primary antibody for 1 h at 37°C. Anti-ATF6␣ and anti-ATF6␤ antisera were gifts from Dr. Kazutoshi Mori. ATF6 antiserum binding was visualized with fluorescein isothiocyanate-conjugated secondary antiserum (1:500, Jackson Immuno-Research, Baltimore, MD).

RESULTS
To compare ATF6␣ and ␤-mediated ERSRG induction, we assessed the abilities of constructs encoding FLAG epitopetagged versions of the putatively active N-terminal fragments of each isoform (N-ATF6␣ and N-ATF6␤) to induce a luciferase reporter under the control of an isolated ERSE derived from the GRP78 promoter (6,14). Increasing the levels of the N-ATF6␣ construct in transfection experiments resulted in progressive increases in reporter expression as expected; however, the N-ATF6␤ construct appeared to have no effect on reporter activity, even at the highest concentration tested (Fig. 1A). Because reporter constructs regulated by isolated ERSE usually exhibit lower activity than native ERSE-containing promoters, we decided to examine the abilities of each construct to activate transcription from the native GRP78 promoter (6). Again, as expected, increasing levels of the N-ATF6␣ construct conferred a progressive increase in reporter expression amounting to a maximum of 48-fold over control (Fig. 1B). However, only at the highest concentration tested did the N-ATF6␤ construct exhibit apparent reporter induction, which did not reach statistical significance (Fig. 1B). A similar trend was observed when the relative abilities of N-ATF6␣ and ␤ to activate endogenous ERSRG expression were assessed. Although N-ATF6␣ conferred an approximate 4-fold increase in endogenous GRP78 protein levels, N-ATF6␤ conferred at most a 1.5-fold increase (Fig. 2, A and B).
To determine whether the differences between N-ATF6␣-and ␤-mediated reporter induction might be attributed to different transgene expression levels, FLAG immunoblot analyses were carried out. Surprisingly, at any given plasmid concentration, N-ATF6␣, which was transcriptionally more active, was always expressed at very low levels, whereas N-ATF6␤, which had essentially no transcriptional activity, was expressed at very high levels (Fig. 3A). Densitometric analyses demonstrated that at a given plasmid concentration, N-ATF6␤ was expressed at 10 -15fold higher levels than N-ATF6␣ (Fig. 3B). Based on the activity and expression results shown in Figs. 1 and 2, it was estimated that the specific transcriptional activity of N-ATF6␣ was approximately 200-fold greater than that of N-ATF6␤.
Our previous results showed that the expression levels and half-lives of various mutated forms of N-ATF6␣ were inversely related to their transcriptional activities, such that the higher the transcriptional activity, the lower the expression level and the shorter the half-life (10). Because N-ATF6␤ exhibited low activity yet was expressed at high levels, we compared its half-life to that of N-ATF6␣. The results of pulse-chase labeling experiments were consistent with this hypothesis showing that, after 3 h of chase incubation, Ͼ50% of the labeled N-ATF6␣ had disappeared (Fig. 4A, lanes 7 and 8), whereas very little of the labeled N-ATF6␤ had disappeared (Fig. 4A, lanes  15 and 16). Additional experiments showed that the half-lives of N-ATF6␣ and ␤ were ϳ2 and 5 h, respectively (Fig. 4B). Thus, the expression level and half-life of N-ATF6␤ mimic those of low activity N-ATF6␣ mutants.
Because the ER forms of ATF6␣ and ␤ are both cleaved in response to ER stress and because the N-terminal cleavage products both translocate to the nucleus and bind to ERSEs (13), we assessed the effects of co-expressing N-ATF6␣ and ␤ on GRP78 promoter activation. As expected, N-ATF6␤ alone exhibited very low levels of GRP78 promoter activation (Fig. 5A, bar 4) compared with N-ATF6␣ alone (Fig. 5A, bar 5). However, N-ATF6␤ inhibited N-ATF6␣-mediated GRP78 promoter activation in a dose-dependent manner (Fig. 5A, bars 6 -8). Thus, N-ATF6␤ exhibited a dominant-negative-like effect over N-ATF6␣. Truncated forms of N-ATF6␣ and N-ATF6␤ that are missing the putative N-terminal TADs have been shown to have DN effects on ERSRG induction (8,12). Because N-ATF6␤ appeared to behave similar to a DN, we compared its effects to N-ATF6␣-DN and N-ATF6␤-DN. In contrast to N-ATF6␤, neither N-ATF6␣-DN nor N-ATF6␤-DN was able to elicit any measurable transcriptional activation of the GRP78 promoter on its own (Fig. 5B, bar 4, and C, bar 4). However, in comparison to N-ATF6␤, both N-ATF6␣-DN and N-ATF6␤-DN effectively blocked N-ATF6␣-mediated GRP78 promoter stimulation (Fig. 5B, bars 6 -8, C, bars 6 -8). Thus, compared with N-ATF6␣, N-ATF6␤ exhibited a very low level of transcriptional activity and it acted in a dominant-interfering manner to repress N-ATF6␣-mediated gene induction.
The effects of ATF6␣ and ␤ on ER stress induced by TN were also assessed. Cultures transfected with empty vector exhibited low basal GRP78 promoter activity that was stimulated by Ͼ20-fold by TN as expected (Fig. 6, Con). Cultures that were transfected with N-ATF6␣ exhibited robust GRP78 promoter activation that was increased further upon TN treatment, also as expected (Fig. 6, ATF6␣). In the absence of TN, N-ATF6␤ conferred a very small induction of ERSRG. However, cultures transfected with N-ATF6␤ were unable to mount a full ER stress response following TN treatment, exhibiting 3-4-fold less GRP78 promoter activation than cultures transfected with empty vector (Fig. 6, ATF6␤). Although neither N-ATF6␣-DN nor N-ATF6␤-DN alone exhibited activity, it effectively blocked TN-mediated GRP78 promoter induction (Fig. 6, ATF6␣-DN  and ATF6␤-DN), as expected from their previous characterization as dominant-interfering proteins. Thus, N-ATF6␤ was an effective blocker of the endogenous ER stress-response machinery responsible for GRP78 promoter activation.
To examine the roles of endogenous ATF6␣ and ␤, siRNAs directed against each ATF6 isoform were developed. Immunocytofluorescence showed that cultures transfected with a control siRNA directed against GFP exhibited robust ATF6␣ and ␤ expression pattern consistent with localization to the rough ER and Golgi (Fig. 7, A and D). In contrast, cultures transfected with ATF6␣ siRNA exhibited considerably reduced ATF6␣ expression (Fig. 7, B versus A) but no effect on ATF6␤ expression (Fig. 7, E versus D). Moreover, cultures transfected with ATF6␤ siRNA showed no apparent effect on ATF6␣ expression (Fig. 7, C versus A) but reduced ATF6␤ expression (Fig. 7, F versus D).
Immunoblot analyses were carried out to further assess the effects of the siRNA reagents on endogenous ATF6␣ and ␤. As expected, cultures that were transfected with GFP siRNA expressed the highest levels of endogenous ATF6␣ and ␤ (Fig. 8A,  lane 1, and B, lane 1). Cultures transfected with ATF6␣ siRNA exhibited an approximate 82% reduction in ATF6␣ expression 2 (Fig. 8A, lane 2 versus 1) but no change in the levels of endogenous ATF6␤ (Fig. 8B, lane 2 versus 1). In contrast, cultures transfected with ATF6␤ siRNA exhibited a small 37% reduc- 2 The transfection efficiency in these experiments averaged 70 -80%. Thus, because there was an 82% reduction in the level of ATF6␣ in the culture extract, the reduction of ATF6␣ in those cells that were transfected was essentially complete .   FIG. 7. Effect of GFP, ATF6␣, or ATF6␤ siRNA on endogenous ATF6␣ and ATF6␤: immunocytofluorescence. HeLa cells were transfected with 25 ng of GFP, ATF6␣, and/or ATF6␤ siRNA as indicated. 48 h later, cultures were fixed, stained for ATF6␣ (panels A-C) or ATF6␤ (panels D-F), and then visualized by laser-scanning confocal microscopy as described under "Materials and Methods." tion in the level of endogenous ATF6␣ (Fig. 8A, lane 3 versus 1); however, they exhibited a 70% reduction in ATF6␤ expression (Fig. 8B, lane 3 versus 1). Finally, cultures transfected with both ATF6␣ and ␤ siRNA exhibited an 84% reduction in ATF6␣ expression (Fig. 8A, lane 4) and a 70% reduction in ATF6␤ expression (Fig. 8B, lane 4). Thus, the siRNAs were ATF6 isoform-selective and effected significant reductions in the expression levels of endogenous ATF6␣ and ␤ as assessed by immunocytofluorescence and immunoblotting.
To examine the effects of knocking down endogenous ATF6␣ and/or ␤ on the ER stress response, cultures were co-transfected with the various siRNAs and the GRP78 promoterdriven luciferase construct. When cultures were transfected with siRNA directed against GFP, they retained a robust ER stress response that was identical to cultures that did not receive siRNA (Fig. 9, compare None versus GFP) as expected. When cultures were transfected with ATF6␣ siRNA, they nearly completely lost the ability to respond to TN (Fig. 9, ATF6␣), also as expected. However, cultures transfected with ATF6␤ siRNA actually exhibited an enhanced ER stress response ( Fig. 9, ATF6␤), consistent with the hypothesis that endogenous ATF6␤ is a transcriptional repressor of ATF6␣. Lastly, cultures transfected with ATF6␣ and ␤ siRNA exhibited a similar ER stress response as those transfected with ATF6␣ siRNA alone (Fig. 9, ATF6␣ ϩ ATF6␤). The fact that there was a residual response in cultures treated with either ATF6␣ siRNA alone or with ATF6␣ and ATF6␤ siRNA could be the result of an incomplete knock down of ATF6␣. Alternatively, it could mean that a portion of the ER stress response is ATF6␣independent. Consistent with the latter possibility are studies suggesting that ERSRG induction can be mediated by the third member of the ATF6 family of b-ZIP proteins, XBP-1, which is also known to bind to ERSEs (15)(16)(17)(18)(19). Moreover, a recent study using fibroblasts from XBP-1 knock-out mice showed that XBP-1 is required for a full ER stress response (20). Lastly, HeLa cells transfected with both ATF6␣ and XBP-1 siRNA exhibited no residual ER stress response as measured by TNinducible GRP78 promoter-driven luciferase induction (data not shown).

DISCUSSION
The results of this study show that, compared with ATF6␣, ATF6␤ possesses very low specific transcriptional activity and, most likely by virtue of this characteristic, ATF6␤ can serve as a repressor of ATF6␣-mediated ERSRG induction. In support of our findings that ATF6␤ exhibits low specific transcriptional activity is a recent gene array study that failed to identify a single gene induced by ATF6␤ (20). Consistent with the repressor roles for ATF6␤ was our finding that knocking down ATF6␤ expression using siRNA actually increased TN-mediated ERSRG induction (Fig. 9).
It is of interest to consider the consequences of these apparently opposing actions of ATF6␣ and ␤ on the ER stress response. At least in part, the ratio of N-ATF6␣/N-ATF6␤ must play a role in determining the magnitude and duration of the ATF6-dependent component of ERSRG induction. To a first approximation, the N-ATF6␣/N-ATF6␤ ratio would be determined by the relative expression levels of the membrane forms of ATF6␣ and ␤ and the relative rates of cleavage of these forms following the onset of ER stress. Although these parameters remain to be determined in detail, several published studies may provide some insight. Immunoblot analyses suggest that in HeLa cells the levels of the membrane forms of ATF6␣ and ␤ are similar (12,13); however, the rates of cleavage appear to differ. In response to TN, the cleavage of ATF6␣ was shown to be maximal after approximately 3-4 h, whereas the cleavage of ATF6␤ did not reach a maximum until 8 h after TN treatment (12). Thus, the relative expression levels and times of generation are consistent with the hypothesis that N-ATF6␤ is generated after N-ATF6␣ and that it serves as an endogenous modulator of ATF6␣-mediated gene induction. Moreover, the half-lives of ATF6␣ and ␤ are also consistent with the view that ATF6␤, which is relatively long-lived, serves as an inhibitor of ATF6␣, which is extremely short-lived (Fig. 4).
Although the ATF6␣/␤ ratio and the rates of cleavage of each isoform are consistent with the hypothesis that N-ATF6␤ is a transcriptional repressor, it remains unclear why a specific inhibitor of N-ATF6␣-mediated gene induction is necessary. There are several other cases of transcription factor homologues that serve as repressors that ensure transient gene induction. For example, CREB and ICER are CREB/ATF family members that exert opposing effects on cAMP-dependent signaling (16). The b-Zip and DNA binding domains of CREB and ICER are nearly identical; however, ICER does not possess a functional TAD. Accordingly, ICER, which is generated shortly after CREB activation, serves as a transcriptional repressor of CREB-mediated signaling and, by doing so, it ensures the transient activation of CREB-mediated gene induction. Thus, similar to CREB and ICER, perhaps ATF6␣ and ␤ function together to ensure the transient activation of ATF6␣mediated gene induction during ER stress, and analogous to prolonged CREB activation, it may be that prolonged activation of ATF6␣ leads to effects that are not conducive to resolving the ER stress or promoting cell survival. Perhaps the inhibitory actions of ATF6␤ combined with the rapid rate at which N-ATF6␣ is degraded ensure that the effects of ATF6␣ are transient. Consistent with this possibility are previous studies showing that, in response to ER stress, ATF6␣ is activated rapidly but transiently (17). It has also been shown that if the ATF6␣-mediated events do not lead to resolution of the ER stress, there is a subsequent activation of XBP-1, which continues to induce some ER stress-response genes along with additional non-ATF6␣-inducible genes such as EDEM, a protease that degrades misfolded proteins (21,22).
Many questions remain to be addressed regarding the roles of ATF6␣ and ␤ during the ER stress response. Do ATF6␣ and ␤ work in concert to finely tune the extent and magnitude of ATF6-mediated gene induction analogous to CREB and ICER? Are the rates of ATF6␣ and ␤ generation and degradation regulated so they can be varied to suit the severity and nature of the ER stress? How do ATF6␣ and ␤ collaborate with XBP1 to regulate the temporal induction of ER stress-response genes? Future studies on the rates of ATF6␣ and ␤ cleavage in response to different stresses, as well as the mechanisms by which ␤ exerts its inhibitory effects on ␣, will be necessary to provide answers to these provocative questions.