PerspectiveCell Biology

Sensing ER Stress

See allHide authors and affiliations

Science  30 Sep 2011:
Vol. 333, Issue 6051, pp. 1830-1831
DOI: 10.1126/science.1212840

The endoplasmic reticulum (ER) is sometimes described as the factory of the eukaryotic cell. Most lipids, glycans, and about a third of the proteome are synthesized in this organelle. As with any modern factory, quality control mechanisms are in place to sort and remove defective products from distribution from the ER. A key regulator of this microeconomy is the unfolded protein response (UPR) signaling pathway. On page 1891 in this issue, Gardner and Walter (1) describe a molecular mechanism by which the sensor protein Ire1 detects and communicates ER stress.

The Ire1 branch of the UPR is conserved among eukaryotes. Upon ER stress, Ire1 activates its cytosolic ribonuclease domain to splice a precursor mRNA, generating a mature message whose encoded protein activates stress-related genes. Metazoans have two additional mechanisms for sensing ER stress: the protein kinase RNA-like endoplasmic reticulum kinase (PERK) and activating transcription factor 6 (ATF6) (2, 3). PERK phosphorylates eukaryotic translation initiation factor 2α, which attenuates general translation (the ATF4 transcription factor remains active to express stress response genes). ATF6 moves from the ER to Golgi apparatus where it is cleaved, producing a transcription factor (p50 ATF6) that activates stress response genes. Although details of these signaling mechanisms are well established, how ER stress is “sensed” is less clear.

Direct engagement.

During ER stress, when the concentration of unfolded protein rises, Kar2/BiP is released from Ire1. Binding of the unfolded protein (for example, CPY*) to Ire1 drives a conformational change that favors Ire1 oligomerization and activation of the unfolded protein response. Specific peptide regions within the CPY* polypeptide (blue) bind to the luminal cleft formed by dimeric Ire1.


In yeast, the ER luminal chaperone protein Kar2 (BiP in mammals) binds to inactive monomeric Ire1, but not to activated dimers (4, 5). A titration model proposed that increasing the unfolded protein concentration competes away chaperone from Ire1. Concomitantly, steric inhibition is relieved to allow Ire1 dimerization and activation. However, the three-dimensional structure of the yeast Ire1 luminal domain showed similarity to peptide-binding domains of major histocompatibility complex proteins (6). Modeling revealed a cleft in the dimeric structure that could bind peptides. This, along with genetic evidence, suggested that direct binding of unfolded proteins might drive Ire1 dimerization. In addition, mutations at the predicted interface of higher-order oligomers of Ire1 inhibited activation, suggesting a functional role for forming larger complexes. However, the human Ire1α luminal domain structure proved to be incompatible with the direct unfolded protein-binding model (7). In vivo studies show that Ire1 mutants unable to bind Kar2/BiP fail to become activated in the absence of stress even if they dimerize (8). Furthermore, stress-induced cells reorganize Ire1 into clusters, a condition necessary for full activation (9, 10). Inhibition of unfolded protein aggregation in vitro with a purified Ire1 luminal domain suggests a capacity to bind unfolded proteins (10). Although the status of chaperone association alone is insufficient to determine the Ire1 activation state, Kar2/BiP binding does modulate the response (11).

Gardner and Walter provide the long-awaited evidence that yeast Ire1 senses ER stress by directly engaging unfolded proteins. The authors use CPY*, a misfolded variant of carboxypeptidase Y (CPY) that is widely used in the field of protein quality control (12). They show that CPY*, which triggers the UPR, binds directly to Ire1, whereas normal (folded) CPY does not. With a peptide array encompassing the entire polypeptide, the authors mapped putative Ire1 binding sites. The strongest interactions occurred with peptides representing the signal sequence and two broad regions in the carboxyl-terminal half of CPY. CPY fragments containing these sites could bind to Ire1 in vivo and also induced the UPR. Analysis of peptide sequences and relative binding affinities indicated a preference for certain basic and hydrophobic residues. Furthermore, mutations in the luminal cleft that impair activation in vivo also disrupt peptide binding in vitro. Notably, Ire1's preferences are similar to those of Kar2/BiP, suggesting that the two proteins might compete as a function of activation. However, because some peptides display different affinities for the two proteins and others bind exclusively to one protein, this scenario is unlikely.

The characterization of high-affinity peptides allowed Gardner and Walter to analyze a key tenet of activation, the formation of higher-order Ire1 oligomers. Measuring oligomerization through velocity sedimentation, the authors demonstrate a shift from peptide-free monomer, dimer, and tetramer forms in solution to high molecular weight complexes in the presence of peptide. These results suggest that peptide binding changes the conformation of Ire1 luminal domain interfaces to induce higher-order oligomers and a robust UPR (see the figure).

Still unresolved is the mechanism for activation of human Ire1α, whose structure remains at odds with the direct peptide-binding model. Gardner and Walter suggest that the reported structure of human Ire1α could represent the “closed” conformation in the absence of peptide, whereas the structure of yeast Ire1 represents the peptide-bound “open” conformation with oligomerization-competent interfaces. Whether the mechanism in humans is conserved or divergent awaits future studies. For now, we know that at least one Ire1 “sees” unfolded proteins through direct engagement.


View Abstract

Stay Connected to Science

Navigate This Article