A new role for the ER unfolded protein response mediator ATF6: Induction of a generalized antioxidant program

Optimal cellular health requires normal protein function which, in turn, is dependent on proteins “folding” into their appropriate three dimensional conformations. Roughly a third of cellular proteins undergo folding in the lumen of the endoplasmic reticulum (ER). However, under stressful conditions, unfolded proteins sometimes accumulate. This initiates a process termed the unfolded protein response (UPR)¹. One arm of the UPR is activated by the ER membrane protein ATF6, a fragment of which moves to the nucleus where it functions as a transcription factor to activate genes that mediate protein folding. Unexpectedly, Jin et al report in the current issue of Circulation Research a new function for ATF6: to transcriptionally activate the expression of a repertoire of antioxidant genes, many of which encode proteins that reside outside of the ER². Moreover, data are presented supporting a model in which this antioxidant program attenuates myocardial damage from ischemia/reperfusion.

Some of the information required for protein folding is intrinsic to its amino acid sequence, but chaperones are critical for most proteins to assume their appropriate conformation. Chaperones are a biochemically diverse group of proteins that reside in subcellular compartments where protein folding takes place (e.g. cytosol, mitochondria, ER). Essentially all proteins destined for various cellular membranes and most that will be secreted are directed to the ER by a signal sequence as they exit ribosomes in the cytoplasm. Folding in the ER is coordinated with post-translational modifications that include removal of the signal sequence, N-linked glycosylation, and disulfide bond formation³. These events require the integrated actions of several classes of chaperones, oxidoreductases (including protein disulfide isomerases), and prolyl peptidyl cis-trans isomerases.

When everything proceeds according to plan, high quality proteins are produced and the cell is happy. However, things can go awry. While genetic mutations in proteins can predispose them to abnormal folding, even more common are various cellular stresses, including hypoxia, reactive oxygen species, abnormalities in Ca²⁺ handling, and metabolic perturbations⁴. The first line of defense is designed to correct the situation by refolding the denatured proteins. This is accomplished by (a) temporarily putting the brakes on translation so as to decrease the load of additional proteins delivered to the ER; (b) augmenting the protein folding machinery; and (c) eliminating terminally misfolded proteins through ER-associated degradation (ERAD; which utilizes the proteasome) and autophagy (which utilizes the lysosome). If all of the above fail, the UPR may signal the cell to undergo apoptosis, although the precise determinants of this decision are poorly understood⁵.

Three intertwined pathways comprise the UPR¹. Each is initiated by an integral ER membrane protein that serves as a sensor for unfolded proteins in the ER lumen: (a) PERK (protein kinase RNA-like endoplasmic reticulum kinase); (b) IRE1 (inositol-requiring enzyme 1α); and (c) ATF6 (activating transcription factor 6). When ER protein folding is efficient, these three proteins are held inactive by interaction with the chaperone GRP78 (78 kDa glucose-regulated protein; also known as BiP). Although some evidence suggests that IRE1 and perhaps PERK can directly detect misfolded proteins, the canonical model is that activation of the UPR occurs when GRP78 is recruited away to interact with now exposed hydrophobic regions of misfolded proteins. This allows PERK to dimerize, autophosphorylate, and become an active kinase, which then phosphorylates the α subunit of eukaryotic initiation factor 2 (eIF2α). This results in attenuation of the translation of most mRNAs. However, phosphorylated eIF2α also upregulates translation of a subset of mRNAs that contain inhibitory open reading frames in their 5’ untranslated regions - one being ATF4 (activating transcription factor 4), which activates genes involved in the UPR. Release from GRP78 also promotes IRE1 dimerization and autophosphorylation, which activates a latent endonuclease function in this protein. IRE1 then splices out 26 nucleotides from the mRNA encoding XBP1 (X-box binding protein 1) in the cytoplasm, causing a frame shift that results in production of XBP1s. This protein functions as a transcription factor for genes encoding chaperones and ERAD components. Finally, dissociation of GRP78 allows ATF6 to translocate to the Golgi, where it is cleaved by two proteases that release its N-terminal half to move to the nucleus where it functions as a transcription factor that activates genes involved in protein folding.

While the UPR was identified over twenty-five years ago⁶, its pathogenic role in heart disease has been appreciated more recently. Of relevance to the current paper, all three arms of the UPR are activated during ischemia/reperfusion, and the UPR appears to attenuate the magnitude of myocardial damage in this context⁷. This has been demonstrated using experimental induction of the UPR with tunicamycin (inhibits N-linked protein glycosylation in the ER lumen) or thapsigargin (inhibits the sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase, thereby interfering with Ca²⁺ uptake into the ER)⁸. More specific approaches have included overexpression of XBP1⁹ or ATF6¹⁰, which reduce ischemia/reperfusion injury, and conversely this damage is exacerbated by XBP1 deletion⁹.

Jin et al have substantially extended their previous ATF6 overexpression studies to define the role of endogenous ATF6 in modulating ischemia/reperfusion injury and, in the process, have revealed an unexpected mechanism that contributes to the beneficial effects of this pathway. Using a combination of loss and gain of function approaches in cultured neonatal and adult cardiomyocytes, isolated perfused hearts, and intact mice, the authors demonstrated that endogenous levels of ATF6 are important for limiting ischemia/reperfusion-induced oxidative stress, cardiomyocyte death (likely necrosis), infarct size, and systolic dysfunction. They then went on to discover that, in addition to its known role in activating the transcription of genes encoding ER chaperones, ATF6 also activates the expression of multiple antioxidant genes. Of particular interest, some of the most highly induced antioxidants reside in non-ER locations, including catalase (peroxisomes) and Prdx5 (mitochondria and peroxisomes) – implying that ATF6 is inducing a generalized antioxidant program. They next explored the mechanism of induction and functional significance of catalase in more depth. First, not only is ATF6 sufficient to induce catalase, it appears to be required – both in response to generic ER stressors and ischemia/reperfusion. Second, chromatin immunoprecipitation data suggest that ATF6 acts directly on the catalase promoter. Finally, administration of catalase reversed the increments in cell death, myocardial tissue damage, and cardiac dysfunction resulting from ATF6 deletion. Thus, the activation of endogenous ATF6 limits cardiac damage during ischemia/reperfusion at least in part by inducing catalase.

Based on these data, the authors propose a two-phase model (Figure 1): First, hypoxia resulting from ischemia is the major stimulus for protein misfolding in the ER and subsequent activation of ATF6. This is plausible because oxygen is a necessary substrate in the series of redox reactions (see Fig 8E of Jin et al) that regenerate active (oxidized) protein disulfide isomerases¹¹ - critical for disulfide bond formation and protein folding. Their data showing that ATF6 is activated during ischemia are consistent with this notion. It remains possible, however, that other factors which operate during reperfusion (e.g. oxidative and metabolic stresses) also contribute to protein misfolding in the ER.

Figure 1. ATF6 induces an antioxidant program outside of the ER.

Ischemia impairs protein folding in the ER lumen in part because oxygen is required in the redox cascades that regenerate active protein disulfide isomerases. Other stresses induced by reperfusion may also contribute (see text; dotted line). Unfolded proteins in the ER lumen recruit the chaperone GRP78 away from PERK, IRE1, and ATF6, leading to their activation and initiation of the unfolded protein response. ATF6 translocates to the Golgi where it is cleaved by two proteases (scissors), which releases its cytoplasmic domain to move to the nucleus where it functions as a transcription factor. In addition to activating the expression of genes whose products mediate protein refolding (not shown), ATF6 activates the transcription of an antioxidant program. Some of the encoded proteins, such as catalase, reside in non-ER locations.

The second phase is that ATF6-dependent induction of the antioxidant gene program neutralizes ROS. But where is this ROS coming from? It is well accepted that a large portion of ROS during reperfusion originates in the mitochondria. Moreover, the induction of Prdx5 is consistent with ATF6 playing a role in neutralizing reactive species in this location, although this remains to be formally tested. But, ROS formation also takes place in other cellular compartments during reperfusion. In fact, protein refolding in the ER itself generates significant oxidative stress¹². Interestingly, ATF6-mediated induction of ER-localized antioxidants was not as robust as the induction of antioxidant proteins that reside in other locations. Whether this reflects the proteins that were studied or the general principle that ATF6 induces primarily a non-ER-localized antioxidant program will require further investigation.

An additional novel aspect of the antioxidant program reported by Jin et al is the proportion of ATF6-inducible genes that encode selenoproteins¹³, including Prdx5. Selenoproteins contain a 21^st amino acid, selenocysteine, in which a selenium atom is substituted for the sulfur in cysteine. This amino acid is specified by the UGA stop codon in combination with mechanisms that suppress polypeptide chain termination. Selenoproteins are uncommon (only 25 in the human genome), but appear to play important roles in antioxidant signaling. Their roles in heart disease remain to be fully defined.

In summary, this study provides the first evidence in any cell type linking the ER UPR mediator ATF6 with direct transcriptional induction of antioxidant proteins, many of which reside outside of the ER. Moreover, this mechanism plays an important role in attenuating myocardial ischemia/reperfusion injury. Additional studies will be needed to define potential roles of this pathway in normal biology and other disease states.