Stress granules (SGs) are a prominently studied member of a family of ribonucleoprotein (RNP) granules evident at many phases of mRNA biology from transcription, to splicing, localization and translation [1], [2], [3], [4], [5], [6]. As higher order, multicomponent assemblies of proteins and mRNAs, SGs are a form of biomolecular condensate that has attracted considerable research interest and provides a leading example of liquid-liquid phase separation as a subcellular organization and regulatory principle [2], [7], [8], [9], [10]. As a field, the study of biomolecular condensates has leveraged principles of polymer physics and in particular the physicochemical basis of liquid-liquid phase separation to characterize the complex, dynamic behavior of RNP granules in non-equilibrium biological systems [9], [11], [12], [13], [14], [15]. Although compositionally distinct, SGs share with many RNP granules three common properties; they are non-membranous assemblies, contain RNA, largely in the form of untranslated mRNAs, and are enriched in RNA binding proteins (RBPs), many with intrinsically disordered and/or prion-like domains that support the multimerization behavior critical to macromolecular phase separation [2], [8], [16], [17], [18]. SGs can be distinguished from many RNP condensates in that, eponymously, their formation is elicited by diverse cell stress conditions and most prominently, but not exclusively, cell stress conditions that promote elongation factor 2α (eIF2α) phosphorylation and the consequent inhibition of the initiation stage of protein synthesis [4], [19], [20], [21], [22]. Notably, SGs biogenesis can also be elicited via inactivation of eIF4A, eIF4B, eIF4G, eIF4H, or PAPB as well as small molecule eIF4A inhibitor pateamine A, further validating the mechanistic coupling of protein synthesis initiation with SG biogenesis [23], [24], [25].

eIF2α phosphorylation is mediated by four kinases, GCN2, PKR, HRI, and PERK, each of which is responsive to distinct cellular stresses [26], [27]. GCN2 (general control non-derepressible 2) binds deacylated tRNAs and thus monitors steady state amino acid/charged tRNA levels [28], [29]. PKR (protein kinase R) is activated by binding to double-stranded RNA and serves a primary role in the cellular response to virus infection; PKR can also be activated by several cellular stressors, including serum starvation, and peroxide and arsenite treatment, via the protein activator PACT [28], [30], [31], [32]. HRI (heme-regulated kinase) is responsive to iron-heme levels and undergoes activation in response to sodium arsenite, an environmental toxin commonly used in studies of SG biology [33], [34]. PERK (PKR‐like ER kinase) is a resident ER transmembrane kinase with an ER lumenal unfolded protein sensing domain and a cytosolic eIF2α kinase domain [35], [36]. PERK activation occurs in response to the accumulation of unfolded proteins in the ER and can be elicited by a number of pharmacological disruptors of proteostasis including tunicamycin, an inhibitor of N-linked glycan biosynthesis which promotes nascent N-linked glycoprotein misfolding [37], [38], reducing agents such as dithiothreitol (DTT) or β-mercaptoethanol (BME), which prevent disulfide bond formation in the ER lumen and thereby disrupt protein folding [39], [40], and thapsigargin, which disrupts ER proteostasis via inhibition of the sarcoplasmic/endoplasmic reticulum Ca2+ ATPase SERCA [41], [42]. The mechanistic coupling of eIF2α phosphorylation to SG formation thus places SG biogenesis in the compendium of cellular responses to activation of the integrated stress response (ISR) (Fig. 1). In further support of protein synthesis initiation inhibition as the critical driver of SG biogenesis, pharmacological inhibition of eukaryotic initiation factor 4 A (eIF4A), an RNA helicase, by the compounds pateamine A or hippuristanol also stimulate SG formation [24]. Additionally, links between protein synthesis initiation inhibition and SG biogenesis are supported by the discovery that ISRIB (integrated stress response inhibitor), a small molecule inhibitor of the ISR that selectively binds and stabilizes active higher order oligomeric states of the dedicated eIF2 guanine nucleotide exchange factor eIF2B, blocks de novo SG formation and promotes the rapid dissolution of SGs when supplied to stressed cells [43]. Notably, ISRIB blocked sodium arsenite-induced SG formation and stimulated SG resolution when provided to cells pretreated with sodium arsenite or the pharmacological inducers of the unfolded protein response (UPR) tunicamycin and thapsigargin, which as noted above promote PERK activation [43]. The latter is of particular interest as the UPR selectively reports on the protein folding environment of the ER lumen.

A principal function of the UPR is to transiently suppress the translation of ER-targeted mRNAs, thereby reducing the flux of nascent polypeptides into the ER lumen. Intriguingly, RNA-seq analyses of purified SGs revealed that ER-targeted mRNAs are under-represented in the SG mRNA transcriptome, suggesting that localization to the ER compromises mRNA recruitment to SGs even when translation initiation is suppressed [44]. This finding may have a relatively simple technical explanation, where current methods for SG isolation are selective for soluble populations of SGs, as discussed by these authors [44]. The question of ER-targeted mRNA recruitment into SGs does however appear to be complex. For example, MDR1 transcripts, encoding the polytopic integral membrane protein multidrug resistance P-glycoprotein, were demonstrated to be refractory to SG recruitment [45]. Yet, disruption of the translation and ER-localization of MDR1 mRNAs, via insertion of a stable stem-loop in the 5′ UTR of a truncated MDR1 reporter, enabled SG recruitment of the (cytosolic) MDR1 reporter during arsenite stress [45]. As will be later discussed, recent evidence supports a role for the ER, perhaps fundamental, in SG biogenesis. Of particular importance, there is evidence, albeit limited, that mRNA recruitment into SGs may have a gene specificity element where ER-localization per se does not preclude mRNA recruitment into SGs but rather ER-localized mRNAs display variations in their sensitivity to SG recruitment [46].