Translational factor-related (TRAFAC) GTPases are one of the most ancient protein groups that existed already in the Last Universal Common Ancestor and includes today such fundamental components of any genetic system as, for example, the elongation factors Tu and G. It is within this primordial class of enzymes that, soon after Bacteria have diverged from Archaea, one specific branch has seen a particularly broad—and evolutionarily successful—expansion, spawning such functionally diverse proteins as septins, the iron transporter FeoB, the tRNA modification enzyme TrmE, the ribosomal large subunit (LSU) AFs YihA and EngA/Der, and finally Era (Leipe et al., 2002; Goto et al., 2013). The characteristic fusion of the N-terminal GTPase domain with the C-terminal KH domain created the unique “face” of the Era protein family and sealed its destiny to serve important roles in cellular RNA metabolism.

Somehow, Era imposed itself as a critical component of the SSU biogenesis, which must have made it indispensable already in early bacteria and ensured its widest spread and conservation in most extant bacterial phyla (Figure 1A). Indeed, congruent with its early origins, Era is often an essential protein (e.g., in E. coli, Salmonella enterica, Haemophilus influenzae, Synechococcus elongates—Inada et al., 1989; Takiff et al., 1989; Lerner and Inouye, 1991; Anderson et al., 1996; Akerley et al., 2002; Voshol et al., 2015). And even when it is not (e.g., in Staphylococcus aureus, Streptococcus pneumoniae, Mycobacterium tuberculosis, and some Bacillus subtilis strains), the loss of Era is associated with severe pleiotropic phenotypes (Minkovsky et al., 2002; Zalacain et al., 2003; DeJesus et al., 2017; Wood et al., 2019; Bennison et al., 2021). This unacceptably high fitness cost made the secondary loss of Era though not impossible, certainly very difficult. Among the known bacteria, only some extremely genome-reduced groups, such as Dependentiae, Chlamydia, and most Candidate Phyla Radiation (CPR), managed to overcome this “Era addiction” (Leipe et al., 2002; Gil et al., 2004). This trend is paralleled by simplification of their ribosomes, including the shortening of rRNAs and the disappearance of some r-proteins (Tsurumaki et al., 2022).

The next turn in the evolutionary saga of the Era family was the emergence of eukaryotes. While Archaea have never possessed such proteins, the arrival of bacterial endosymbionts—the ancestors of modern mitochondria and plastids—opened the way to acquire Era horizontally. This new evolutionary spread campaign turned out to be extremely successful. Besides obligate anaerobic eukaryotes with degenerated, genome-lacking mitochondria (Metamonada, Microsporidia), only higher fungi have lost Era, while essentially all other major eukaryotic clades (Discoba, the plastid-containing Diaphoretikes, most Amoebozoa and Opisthokonta, including animals) have kept it (Figure 1A). Although in all known Eukarya the era genes are now part of nuclear DNA, they encode exclusively mitochondrial or plastid-localized proteins that participate in the maturation of the corresponding organellar ribosomes (Panigrahi et al., 2009; Dennerlein et al., 2010; Uchiumi et al., 2010; Sun et al., 2019; Maiti et al., 2021; Méteignier et al., 2021; Valach et al., 2023). Plants typically have two Era homologues with nonoverlapping localizations in mitochondria and chloroplasts (Ingram et al., 1998; Suwastika et al., 2014; Cheng et al., 2018; Sun et al., 2019).

Intriguingly, while the origin of the plant mitochondrial Era proteins is transparent (their sequences clearly cluster with α-proteobacterial ones), the source of the chloroplast Era is more enigmatic. Instead of grouping with cyanobacterial Era proteins, they show significant similarity with those from the Bacteroidetes-Chlorobi group, suggesting that another horizontal transfer event was responsible for their emergence in the plant kingdom (Suwastika et al., 2014).

Given their critical roles in the organellar gene expression, eukaryotic Era proteins are typically essential for viability (Ingram et al., 1998; Gohda et al., 2003; Cheng et al., 2018), although there might be exceptions (Sun et al., 2019), and cell lines lacking mitochondrial Era, however extremely sick due to the lack of respiration, can be maintained on high-glucose media (Uchiumi et al., 2010; Xie et al., 2012).

Pointing at a deeply conserved biological role, Era homologues show high functional portability. Thus, S. enterica, Pseudomonas aeruginosa, Francisella tularensis, B. subtilis, Listeria monocytogenes, and Streptococcus mutans Era proteins complement E. coli era mutations, while human mitochondrial ERAL1 can replace its chicken counterpart (Pillutla et al., 1995; Anderson et al., 1996; Zuber et al., 1997; Powell et al., 1999; Minkovsky et al., 2002; Gohda et al., 2003; Auvray et al., 2007).

The genomic environment is often informative about the function and the transmission mode of genes, especially in prokaryotes. While the genomic neighborhood of era varies across Bacteria, the synteny tends to be relatively well preserved within individual microbial clades. This property suggests primarily vertical spread of this gene among Bacteria, as expected of a highly conserved and functionally critical protein. Indeed, published phylogenetic trees of Era closely follow the established phyletic relationships between grand taxa (Mittenhuber, 2001; Coleman et al., 2021; Moody et al., 2022).

Looking at era syntenies across multiple model species, one can single out a few genes that pop up time and again next to era or in its close proximity (Figure 1B). Thus, in many Terrabacteria, era is adjacent to genes encoding the ribosome assembly factor YbeY and its highly conserved partner YbeZ/PhoH-like, involved in the biogenesis of the platform and the 3′-domain of 16S rRNA (Liao et al., 2021; Andrews and Patrick, 2022). Three other genes associated with ribosome biogenesis and frequently found close to era are rnc, rsmE, and lepA. rnc encodes RNase III, which is involved in rRNA pre-processing at earliest stages of ribosome biogenesis (Dunn and Studier, 1973); the rnc gene precedes era in many species and is translationally coupled to the latter in many γ-proteobacteria (see the section “Post-transcriptional regulation in bacterial rnc-era operons”). RsmE is m³U1498 methyltransferase that methylates a conserved uridine in the 3′-minor domain of 16S rRNA (Basturea et al., 2006). LepA, another ancient TRAFAC GTPase also known as elongation factor 4, has been shown to participate in the SSU assembly, primarily at the level of the 3′-domain of 16S rRNA (Gibbs et al., 2017). Since the function of Era is intimately linked to the maturation of the platform and the 3′-domain of 16S rRNA, its connection with these proteins is functionally meaningful.

A few other neighbors associate Era more loosely with central genetic processes, such as replication (dnaG), recombination (recO), transcription (rpoD), translation (glyQS), and protein folding (dnaJ). Finally, some metabolic enzymes are unusually often encoded close to era; some of them are also frequent satellites of ybeY genes (cdd, dagK, pdxJ, acpS). The mechanistic meaning of these persistent syntenic associations with metabolic enzymes remains enigmatic. However, the majority of era neighbors belong to a group of most deeply conserved bacterial genes, highlighting the ancient origin and centrality of Era to the cellular metabolism (Gil et al., 2004).

Like other TRAFAC GTPases, Era has a ∼170 aa-long G-domain with a characteristic fold in which a 6-stranded β-sheet is surrounded by 5 α-helices in a highly conserved alternating pattern that brings together five diagnostic motifs involved in GTP binding and hydrolysis (Figure 2). All β-strands are parallel, except for β2, which is uniquely antiparallel in all TRAFAC GTPases (Leipe et al., 2002). The G1 motif (Walker A) connects the strand β1 and the helix α1 and has a consensus sequence GxxxxGK(S/T). It is involved in the binding of α- and β-phosphates of GTP/GDP and for this reason often referred to as ‘P-loop’ (P for ‘phosphate’; hence the name of the entire superfamily possessing this motif—‘P-loop NTPases’). The G2 motif, in the loop between the helix α1 and the strand β2, has a consensus sequence TTR containing an invariant Thr residue that binds a Mg²⁺ ion required for GTP hydrolysis (Wu et al., 1995). The G3 motif (Walker B) is found right after the strand β3 and follows the consensus DTPG; it participates in Mg²⁺ and γ-phosphate binding. Finally, the G4 and G5 motifs, located immediately after the strands β5 and β6 and having consensus sequences NKxD and SA, respectively, specifically recognize the guanine base of GTP/GDP (Bourne et al., 1991; Paduch et al., 2001).

Such an architecture of the active center explains why all Era proteins bind GTP and GDP with high affinity and specificity (Ahnn et al., 1986; Chen et al., 1999a). Interestingly, while dGTP can be bound quite tightly (i.e., the 2′-OH group of ribose is not important for the interaction), GMP or cGMP fail to associate with Era, likely because they form too few contacts with the G-motifs (Inada et al., 1989; Chen et al., 1990; Sullivan et al., 2000). ATP, UTP and CTP are not bound either since they are discriminated against by the G4 and the G5 motifs (Wu et al., 1995; Kawabata et al., 1997; Morimoto et al., 2002). Several studies showed that GDP binds to Era competitively and significantly more tightly than GTP. However, since both K _d s are in the low micromolar range (i.e., much less than the intracellular concentrations of GDP and GTP), Era most probably cannot operate as a GDP/GTP sensor, as sometimes speculated (Chen et al., 1990; Bourne et al., 1991; Wu et al., 1995; Sullivan et al., 2000). Instead, similar to other GTPases, it works as a molecular switch, and this property appears to be essential for its cellular function.

GTPases have become a paradigmatic example of how an enzymatically catalyzed chemical reaction can be coupled to mechanical movement. The resulting switching behavior, whereby the enzyme cycles between two different conformations, offers a simple and efficient mechanism driving forward a wide variety of biological processes, from translation to signal transduction (Bourne et al., 1991; Paduch et al., 2001). It is also a sensible way to rhythm the ribosome assembly. Indeed, several TRAFAC GTPases have been leveraged by bacteria, chloroplasts, and mitochondria to guide the biogenesis of both ribosomal subunits at specific assembly steps (Britton, 2009; Verstraeten et al., 2011; Goto et al., 2013; Maiti et al., 2021).

Structural studies by X-ray crystallography shed light on how Era uses GTP hydrolysis to change its shape (Tu et al., 2009). Free Era assumes one of two alternative conformations (Figure 3A). When it is bound to GTP (or its nonhydrolyzable analogue), all the G-motifs are involved in the interaction (Figures 2A,B), and the GTPase domain is rigid and closed. This is the ON-state of Era. However, upon GTP hydrolysis, the G2 and G3 motifs lose their ligands (Mg²⁺ and γ-phosphate), and the surrounding structural elements (so-called ‘switches I and II’) swing open. The resulting conformation is much looser and involves a significant movement of the adjacent loops and the helix α2. The apo-form of Era has essentially the same conformation (Chen et al., 1999b), i.e., both the apo- and the GDP-bound enzymes are in an inactive, OFF-state.

To make the cycling between the two conformations possible, a GTPase must be able to i) hydrolyze GTP to switch to the OFF-state and ii) exchange GDP for GTP to reset to the ON-state again (Figure 3A). GTP hydrolysis by Era occurs in a substrate-assisted manner: the γ-phosphate of GTP itself (activated by Mg²⁺) acts as a general base abstracting a proton from water; the resulting hydroxyl performs the nucleophilic attack on the γ-phosphate, with GDP fulfilling the role of the leaving group (Paduch et al., 2001; Tu et al., 2009). Similar to most switching GTPases, Era has poor intrinsic GTPase activity (Chen et al., 1990; Bourne et al., 1991; Wu et al., 1995; Chen et al., 1999a; Paduch et al., 2001). In fact, Era lacks an important Gln residue in the switch II that is responsible for aligning water for the nucleophilic attack in other small GTPases (such “incomplete” enzymes are called HAS-GTPases, for ‘hydrophobic amino acid substituted’—Mishra et al., 2005). Moreover, like all switching GTPases, Era also lacks a critical Arg residue required for the stabilization of the transition state. This residue is typically supplied in-trans by a GTPase activating protein (GAP) which, in the case of Era, remains unknown (Bourne et al., 1991; Paduch et al., 2001). However, like in many other TRAFAC GTPases, this role seems to be taken on by a potassium ion coordinated by two invariant Asn residues in the G1 motif and the so-called ‘K-loop’ imbedded in the switch I (Figure 2C). Addition of K⁺ (or similarly sized monocations) stimulates the GTPase activity of Era by an order of magnitude (Rafay et al., 2012; Shalaeva et al., 2018).

Switching GTPases also normally require a guanosine nucleotide-exchanging factor (GEF) to replace GDP with GTP and reset the GTPase to the ON-state (Bourne et al., 1991; Paduch et al., 2001). Strikingly, some biophysical data suggest that regardless its high affinity for both GDP and GTP, Era easily and rapidly exchanges them in solution. Both association and dissociation rates are so high that GDP can be replaced by GTP within seconds without the need of helper proteins. Therefore, the normally higher concentration of GTP in the cellular milieu must be sufficient a driver to enable unassisted resetting of Era into the ON-state (Sullivan et al., 2000). (See, however, the section “ERAL1 homeostasis in mammalian mitochondria” for a possible counterexample.)

The GTPase activity of Era is strictly required for its cellular function. Even apparently conservative changes in the G1 and G2 motifs or the switches (e.g., the K21R mutation in the G1 motif of E. coli Era) may not be tolerated and result in a lethal phenotype (Pillutla et al., 1995; Shimamoto and Inouye, 1996; Johnstone et al., 1999). Some milder mutations, that do not fully disrupt GTPase activity, produce viable but still severe phenotypes, such as heat and cold sensitivity, cell filamentation, significant growth delay, and inability to use certain carbon sources (Lerner et al., 1992; Lerner et al., 1995; Pillutla et al., 1996; Britton et al., 1997; Britton et al., 1998; Zhou et al., 2020). The N236I mutation in the G4-motif of human ERAL1 causes the Perrault syndrome (sensorineural deafness and ovarian dysgenesis) in humans (Chatzispyrou et al., 2017).

Downstream of the GTPase domain, all Era proteins obligatorily possess an extended KH domain (Figure 2). This rather compact entity of ∼120 aa includes a core region formed by two parallel β-strands (βB and βC) separated by the helix-turn-helix motif (αB and αC), characteristic of all KH domains. Two additional elements—the N-terminally situated helix αA and the antiparallel strand βA—permit to classify this KH domain as type II, which is common in bacteria and bacteria-derived organelles. Finally, the extra helix αD extends the canonical KH fold on the C-terminal side, which is characteristic of the Era family (Chen et al., 1999b; Johnstone et al., 1999; Nicastro et al., 2015).

KH domains are classic RNA-binding elements present in a wide variety of proteins, often in multiple copies. They typically recognize 4 consecutive nucleotides of single-stranded RNA using a characteristic GxxG motif within the helix-turn-helix (Figure 2). The specificity of recognition differs between proteins and is conferred by multiple residues of the core KH-fold (Nicastro et al., 2015). In Era, the KH domain is responsible for its general RNA-binding ability and the specific association with the 3′-minor domain of the SSU rRNA, more precisely, h45 and the downstream single-stranded tail at the 3′-end of the molecule (Johnstone et al., 1999; Meier et al., 2000; Akiyama et al., 2001; Hang et al., 2001; Gohda et al., 2003; Hang and Zhao, 2003; Sun et al., 2019). This interaction has been studied in detail in bacteria by X-ray crystallography (Tu et al., 2009; Tu et al., 2011). Era was found to interact with an extended single-stranded region right after h45. Using the GxxG motif, it recognizes the highly conserved GAUCA sequence (especially, the two adenines), while the helix αC, the strand βC and the variable loop between βA and βB additionally enable binding of the downstream anti-Shine-Dalgarno sequence CCUCC. Finally, the top side of the domain forms multiple contacts with the base of h45 and the universally conserved guanine of the GAUCA sequence (Figures 2A,B). The helix h45 and the adjacent GACGA site seem to be particularly important for human ERAL1, since the mammalian mitochondrial SSU rRNA lacks the anti-Shine-Dalgarno sequence (Dennerlein et al., 2010; Harper et al., 2023).

Like the GTPase domain, the KH domain is required for the cellular function of Era. Its removal or mutations in the helix-turn-helix motif or in the h45-interacting loops on the top of the domain show a severe loss-of-function phenotype in various species (Pillutla et al., 1995; Zuber et al., 1997; Johnstone et al., 1999; Zhao et al., 1999; Hang et al., 2001; Gohda et al., 2003; Auvray et al., 2007; Tu et al., 2011; Voshol et al., 2015; Sun et al., 2019).

The combination of a GTPase and an RNA-binding modules in one polypeptide begs the question whether Era may somehow coordinate their activities to time its intervention in the ribosome biogenesis pathway. Existing experimental evidence still struggles to provide a clear molecular mechanism for such a coordination. In this section, we will discuss disparate observations for and against the existence of a biologically relevant crosstalk between the two Era domains.

Fusions of Era with other domains are extremely rare (Figure 3B), which is probably explained by significant steric constraints imposed by the nascent SSU, the molecular environment where Era normally functions (see the next section). By far the most frequent fusion type features YbeY attached N-terminally to Era, further emphasizing a privileged relationship between these two ribosome biogenesis factors which likely collaborate during the platform assembly (Liao et al., 2021). This architecture is found in several members of the Clostridia group, some Selenomonadales, and Thermoleophilum album. Fusions with RNase III are observed in some Rickettsia species. Other recurrently found domain combinations (e.g., with CS and SGS domains in Peronosporales and with DUF916 in certain Streptomyces sp.) are currently difficult to interpret due to scarce data about their molecular functions.

First indication that Era is required for the SSU assembly comes from phenotypic analyses of Era-deficient bacteria. In E. coli and S. aureus, mutations, knockdown or knockout of era result in the depletion of 70S ribosomes with a concomitant accumulation of individual subunits, suggesting that the SSU and the LSU cannot assemble into a monosome. As a result, the cellular protein synthesis slows down, severely compromising growth and survival (Nashimoto et al., 1985; Inoue et al., 2003; Loh et al., 2007; Razi et al., 2019; Wood et al., 2019; Bennison et al., 2021).

As expected from the binding specificity of Era, the defect is on the SSU side. Upon Era depletion, 16S rRNA shows a gross processing defect, which is a typical phenotype of an impaired SSU assembly also observed upon deletion of other SSU biogenesis factors, such as RimM, RbfA, RsgA, and YbeY (Nashimoto et al., 1985; Bylund et al., 1998; Inoue et al., 2003; Himeno et al., 2004; Davies et al., 2010; Baumgardt et al., 2018; Trinquier et al., 2019; Wood et al., 2019). Indeed, a recent cryo-EM analysis of ribosomes from Era-depleted E. coli revealed the accumulation of severely under-assembled SSU particles. Many of them had immature body and failed to assemble the platform and the head regions altogether. Some other particles continued their biogenesis past this stage by taking parallel assembly routes and mostly managed to shape the head and the body. However, they still could not stably recruit the r-proteins bS1, bS21, and uS11 to the platform and showed only partial association and fragmented densities for some head constituents (uS7, uS9, uS13, uS19). Furthermore, the helices h23, h24, h44 and h45, forming the platform and the 3′-minor domain regions, remained largely immature (Razi et al., 2019). Therefore, the absence of Era dramatically perturbed the assembly, which strongly destabilized and functionally impaired the SSU.

Similar results were obtained for eukaryotic Era proteins. In human cells, both ERAL1 knockdown and the Perrault syndrome N236I mutation significantly destabilized the mitochondrial SSU, impacting the mitochondrial protein synthesis and respiration (Dennerlein et al., 2010; Uchiumi et al., 2010; Chatzispyrou et al., 2017). In rice, insertion into the WSL6 locus, encoding a chloroplast Era homologue, resulted in the disappearance of chloroplast ribosomes and the disruption of the organellar translation (Sun et al., 2019).

The profound impact of Era on the SSU assembly notwithstanding (Tamaru et al., 2018), its exact molecular function remains elusive. However, the sum of biochemical, genetic, and structural evidence now enables us to formulate better-framed hypotheses as to where and when it acts. The flagrant deficiency of the platform assembly and the interaction with h45 strongly implicate Era in the maturation of the central and 3′-minor domains (Tu et al., 2011; Razi et al., 2019). Indeed, when mixed with mature ribosomes, Era interacts in the cleft between the head and the platform of the SSU (Sharma et al., 2005; Razi et al., 2019), and a similar binding pattern has been observed in native mitochondrial SSU assembly intermediates (Harper et al., 2023). These latter structures not only confirmed the place of Era intervention but also hinted at the timing of its recruitment, helping us reconstitute the possible order of molecular events with respect to other factors implicated in the platform maturation in bacterial-type ribosomes. We will describe them in more detail in the following section.

First studies of Era binding to bacterial ribosomes were carried out in vitro with the use of a purified Era proteins and mature 30S subunits (Sayed et al., 1999; Sharma et al., 2005; Razi et al., 2019). These fully assembled SSUs were unlikely Era substrates as they are obviously very different from native SSU assembly intermediates (Mulder et al., 2010). Nevertheless, Era showed a strong affinity toward bS1-depleted 30S subunits (Sharma et al., 2005; Razi et al., 2019). Moreover, excess of Era prevented the reassociation of mature 30S and 50S subunits into 70S monosomes (Sharma et al., 2005). Finally, addition of a 5-fold excess of Era in the presence of GDP, GTP or GDPNP was sufficient to split 70S monosomes into individual subunits (Razi et al., 2019). These puzzling observations, quite unusual for an AF, begged several questions: i) how does the Era-SSU association occur? ii) what happens to mature ribosomes upon Era binding? iii) what is the biological significance of these phenomena?

An early cryo-EM study showed that upon coincubation of bS1-depleted 30S subunits from T. thermophilus with Era, a new bilobed density appeared in the cleft between the head and the platform (Sharma et al., 2005). Its position and overall shape are reminiscent of those of ERAL1 on mitochondrial SSU assembly intermediates (Harper et al., 2023), suggesting a similar binding mode. Era is wedged in such a way that the 50S subunit cannot join the 30S subunit anymore; it occupies the binding site of the r-protein bS1, which in bacteria is large, loosely bound and required for translation initiation (Sengupta et al., 2001). This coarse-grained structure rationalizes the 70S ribosome splitting and anti-association activities of Era in vitro and predicts that its binding to mature 30S subunits would interfere with translation. Unfortunately, the low resolution of this structure precludes reliable assignment of molecular interactions (Sharma et al., 2005).

More recently, an analogous experiment performed with E. coli Era and 30S subunits in the presence of GDPNP yielded a cryo-EM structure with a 3.9 Å resolution (Razi et al., 2019). Although the density between the head and the platform potentially corresponding to the bound Era was highly fragmented, making structural details of this interaction inaccessible, this structure turned out to be revealing with respect to the impact Era had on the mature 30S subunit. The head and the platform became more mobile; the helix h44, a key part of the decoding center, was virtually invisible, indicating its high flexibility. Additionally, the densities for tips of h23 and h24 were fragmented, and no densities could be observed for bS1, uS7, bS21, and the N-terminal part of uS13. Overall, the binding of Era appears to have ‘undone’ some of of the SSU assembly steps and critically destabilized its functionally essential regions, making it initiation-incompetent. In a way, Era reverted the SSU to a state more similar to its normal substrate encountered during the SSU biogenesis (Figures 4A,B).

The relevance of this striking phenomenon remains enigmatic. While a similar (albeit subtler) “unmaturation” activity was described for RbfA and RsgA, it required in all cases co-incubation of 30S subunits with an excess of the AF (Razi et al., 2017; 2019; Bikmullin et al., 2023). However, the in vivo abundance of Era in bacteria does not exceed ∼5% of that of ribosomes, making such conditions difficult to achieve inside the cell (Chen et al., 1990; Morimoto et al., 2002). The proposed hypotheses that Era may perform “unmaturation” of 30S subunits under particular stress conditions or help split hibernating ribosomes remain valid possibilities warranting experimental investigation (Razi et al., 2019).

Although eukaryotic Era homologues are only required for the assembly of mitochondrial and chloroplast ribosomes, which represent but a minor proportion of all cellular ribosomes, the essential nature of organellar translation for most aerobic eukaryotes makes them absolutely indispensable. Beyond a direct effect on the protein synthesis in the organelles, which was the focus of the previous sections of this review, defects in eukaryotic Era have far-reaching physiological consequences at the cellular and organismal levels. Below, we will briefly discuss these “secondary” phenotypes on the example of animal and plant Era homologues.

Many γ-proteobacteria tightly coordinate the expression of RNase III and Era, both of which are involved in early stages of the SSU biogenesis. This co-regulation occurs at transcriptional, post-transcriptional, and translational levels (Figure 5A). The two genes are co-transcribed from a shared σ⁷⁰ promoter to yield a polycistronic mRNA (Ahnn et al., 1986; Bardwell et al., 1989; Takiff et al., 1989; Matsunaga et al., 1996a; Powell et al., 1999). They are also translationally coupled: the stop-codon of the rnc ORF, encoding RNase III, overlaps the start-codon of the era cistron. The translation of era is apparently contingent on that of rnc through a re-initiation mechanism, which creates a simple means of stoichiometric control between the two proteins (Ahnn et al., 1986; Chen et al., 1990; Anderson et al., 1996; Powell et al., 1999). As an additional mechanism of co-translational control, both the ORFs have a suboptimal codon usage in E. coli, congruent with their relatively low level of production (Takiff et al., 1989; Chen et al., 1990). On top of this, like many globally acting RNA-binding proteins (Smirnov, 2022), RNase III is subject to autoregulation to avoid potentially toxic overproduction. It cleaves a long stem-loop structure in the 5′-UTR of the rnc-era mRNA, provoking its degradation (Matsunaga et al., 1996a; Matsunaga et al., 1996b). By this means, the expression of era is concomitantly downregulated by RNase III (Bardwell et al., 1989; Anderson et al., 1996).

A direct control of GTPase activity of E. coli Era by small molecules has been proposed early on. Owing to its role in carbohydrate metabolism (see the section “Era and cellular metabolism”), multiple central metabolites had been tested in vitro, and acetate and 3-phosphoglycerate were found to strongly stimulate GTP hydrolysis, whereas glyceraldehyde 3-phosphate acted as an inhibitor (Meier et al., 2000). The in vivo relevance of these biochemical observations is still unknown.

Much better studied and likely more physiologically important is the connection between Era and the alarmones (p)ppGpp at the heart of the stringent response (Figure 5B). pppGpp and ppGpp are synthesized in an ATP-dependent manner from GTP and GDP, respectively, by RelA-like enzymes broadly conserved in bacteria and plant chloroplasts. The activity of RelA is turned on in a tRNA/ribosome-dependent manner upon amino acid starvation and results in a rapid accumulation of both alarmones in the cell. Additional stresses (osmotic shock, starvation for fatty acids, phosphate, or iron) trigger (p)ppGpp production by a related enzyme, SpoT, which also has the ability to degrade the alarmones (Potrykus and Cashel, 2008; Irving et al., 2021). In E. coli, (p)ppGpp, eventually helped by the DksA protein, binds to RNA polymerase and differentially affects its activity on cellular promoters. Namely, (p)ppGpp fully represses the rRNA transcription to prevent the production of new ribosomes. By contrast, it activates amino acid operons to replenish their stocks (Paul et al., 2004; Paul et al., 2005). In Terrabacteria, the mechanism is slightly different: (p)ppGpp does not bind to RNAP, and the shear depletion of the cellular GTP pool caused by the alarmone synthesis directly abolishes the rRNA transcription (Krásný and Gourse, 2004; Kasai et al., 2006). All these measures allow the bacteria to optimize their resource management during starvation, avoid further stress, restore metabolic equilibrium, and ultimately resume growth (Potrykus and Cashel, 2008; Irving et al., 2021).

The stringent response has another important dimension: (p)ppGpp can interact with guanosine factor-dependent or -related proteins and directly modulate their activity. In diverse bacteria, the alarmones competitively inhibit enzymes of the GTP biosynthesis and salvage pathways (Kriel et al., 2012; Liu et al., 2015; Corrigan et al., 2016; Zhang et al., 2018; Irving et al., 2021). By contrast, (p)ppGpp potently stimulates its own producer RelA (Shyp et al., 2012). Both these events create a positive feedback mechanism switching the cell to the stringent response mode with a completely remodeled gene expression (Irving et al., 2021). Ribosome-associated GTPases were on the list as just too obvious targets, and several laboratories gathered compelling evidence that the stringent response--and (p)ppGpp in particular--directly impact on the ribosome biogenesis by controlling GTP-dependent AFs in various bacterial species. These results were recently comprehensively reviewed (Zegarra et al., 2023). Here, we will primarily focus on the relationship between Era and the alarmones.

Naturally affine for G-nucleotides with at least two phosphates, Era from E. coli and S. aureus binds (p)ppGpp with affinities close to those for GDP and GTP and using the same binding site (Corrigan et al., 2016; Zhang et al., 2018). Since the 3′-pyrophosphate moiety is not specifically recognized by other crystallized TRAFAC GTPases (Buglino et al., 2002; Pausch et al., 2018), the association constant must be largely determined by the status of the 5′-pyrophosphate. Thus, ppGpp and GDP bind Era similarly well (in the low-micromolar range) and significantly tighter than GTP and pppGpp (Corrigan et al., 2016; Zhang et al., 2018; Bennison et al., 2021; Zegarra et al., 2023). However, under optimal conditions, GTP is much more abundant in the cell than GDP and (p)ppGpp, meaning that Era is found primarily in a GTP-bound state.

However, during the stringent response, RelA generates up to 1–4 mM (p)ppGpp, which is now on the same foot as GTP, if not higher (Cashel, 1975; Varik et al., 2017). This creates conditions where the pool of Era (and other ribosome assembly GTPases, such as RsgA, RbgA, Der, ObgE) massively shifts to a (p)ppGpp-bound state (Corrigan et al., 2016; Bennison et al., 2021; Zegarra et al., 2023). This state appears to be unproductive: (p)ppGpp inhibits their GTPase activity (Corrigan et al., 2016; Pausch et al., 2018). Judging by the available structures of RsgA, RbgA and ObgE in complex with alarmones, (p)ppGpp blocks these AFs in an inactive conformation, likely because the 3′-pyrophosphate sterically interferes with the recruitment of the G2 motif (Buglino et al., 2002; Pausch et al., 2018; Bennison et al., 2021). Since the organization of the active center is very similar between ribosome-associated GTPases (Leipe et al., 2002; Goto et al., 2013), these findings suggests that (p)ppGpp-bound Era cannot operate its normal switching cycle during the SSU assembly either. Furthermore, (p)ppGpp considerably weakened the ribosome-binding ability of Era (Bennison et al., 2021). All these factors imply that upon stringent response, the ribosome assembly rapidly (within ∼2 min) stalls in part because Era and other GTPases cannot function anymore. The arrest of the SSU biogenesis is especially obvious, as it results in the accumulation of particles lacking the r-proteins uS2, uS3, uS14 and bS21 and containing unprocessed 16S rRNA (Trinquier et al., 2019; Wood et al., 2019; Zegarra et al., 2023). This immediate response, mediated by GTP depletion and a direct inhibitory action of the alarmones, permits the cell to halt the ribosome production much faster than the indirect mechanism involving transcriptional rRNA repression (Potrykus and Cashel, 2008; Irving et al., 2021; Zegarra et al., 2023).

Although this aspect has not been well studied, the (p)ppGpp binding by Era (and, consequently, its activity) might be regulated by associated proteins. In S. aureus, Era directly interacts with Rel_Sau, one of three (p)ppGpp synthases/hydrolases in this bacterium. Rel_Sau was found to mildly stimulate the GTPase activity of Era (Wood et al., 2019). In E. coli, SpoT interacts with the key partner of Era, YbeY, suggesting that all three proteins might be closely associated (Vercruysse et al., 2016). Interestingly, one study reported a direct interaction between Era and the nonspecific nucleotide pyrophosphatase MazG in E. coli. The significance of this complex is unknown; MazG does not influence the GTPase activity of Era (Zhang and Inouye, 2002). While MazG can degrade GTP, it cannot use pppGpp as substrate; however, MazG is under control of a (p)ppGpp-regulated promoter, and its overexpression significantly decreases (p)ppGpp levels during the stringent response (Aizenman et al., 1996; Gross et al., 2006).

In mammalian cells, ERAL1 is subject to both negative and positive regulation by at least two distinct mechanisms (Figure 5C). It appears to be a target of two main mitochondrial ATP-dependent proteases, CLPP and LONP1 (Szczepanowska et al., 2016; Key et al., 2019). The connection between ERAL1 and CLPP is especially intriguing, since CLPP is also associated with the Perrault syndrome (Jenkinson et al., 2013). Knockout of CLPP had a dramatic impact on the mitochondrial ribosome, with a particularly strong effect on the SSU. Free SSUs were found to accumulate at the expense of monosomes, resulting in moderately decreased mitochondrial translation. Indeed, an analysis of potential CLPP substrates identified several mitoribosomal proteins and a few other factors involved in the mitoribosome biogenesis and translation, including ERAL1. CLPP knockout resulted in a significant stabilization and accumulation of ERAL1, accompanied by its higher recruitment to the SSU. It was hypothesized that the abnormally high level of SSU-bound ERAL1 prevents the recruitment of the r-protein bS1m and the subunit joining, and that CLPP could provide a degradative ERAL1 removal mechanism to enable the SSU assembly to proceed further. Indeed, dampening ERAL1 levels in CLPP-deficient cells somewhat improved mitochondrial translation (Szczepanowska et al., 2016). However, the molecular mechanism behind the CLPP deficiency is likely different. In contrast to bacterial bS1, which is a large, multidomain protein loosely bound at the very last stage of assembly, mitochondrial bS1m is a small integral component of the SSU which is already installed in the State A and is too far from ERAL1 to produce any steric clashes (Figures 4A,B). Furthermore, proteolytic degradation as an AF ejection mechanism is unknown, and it is difficult to envision how the very bulky CLPP or its hexameric cofactor CLPX could access ERAL1 in the spatially constrained SSU environment. CLPP more likely degrades ERAL1 outside the SSU. In our opinion, it is more plausible that by targeting ERAL1 and other SSU proteins CLPP maintains the balance between the two mitoribosomal subunits, preventing abnormal SSU overproduction which may be disruptive for translation (Bruni et al., 2021). This model is supported by the finding that uS7m, which is a key ERAL1 partner on the nascent SSU (Figures 4A, B) and another protein linked to the Perrault syndrome (Kline et al., 2022), is also a CLPP substrate (Szczepanowska et al., 2016).

More recently, a positive mechanism of ERAL1 control has been proposed (Reyes et al., 2020). The putative GTP/GDP-exchange factor RCC1L/WBSCR16 localizes to human mitochondria, and its shortest isoform, RCC1L^V3, specifically interacts with the mitochondrial SSU. The RCCL1^V3 pulldown robustly co-purified the GTPases ERAL1, NOA1, and mS29/DAP3. RCCL1^V3 overexpression resulted in an increased recruitment of ERAL1 and NOA1 to the SSU and mildly impaired mitochondrial translation, echoing observations in CLPP-deficient cells (Szczepanowska et al., 2016). By contrast, RCCL1^V3 knockdown significantly reduced the overall abundance of ERAL1 and the mitochondrial SSU and severely impacted the protein synthesis in the organelle, like it happened in the Perrault syndrome patient with the ERAL1^N236I mutation (Chatzispyrou et al., 2017). It is currently unclear whether RCCL1^V3 is a genuine GEF for ERAL1 and/or other mitochondrial GTPases--such a mechanism would be an exciting twist in the biology of ribosome assembly-associated GTPases. But it does seem to be able to regulate the abundance and recruitment of ERAL1, likely at the level of the State A intermediate, where ERAL1, NOA1 and mS29 are all simultaneously present (Figure 4A).

The positive and negative regulatory mechanisms described in this section show that ERAL1 homeostasis is essential in mammalian cells as both ERAL1 deficiency and overproduction may be deleterious to mitochondrial translation, resulting in severe downstream effects, such as the Perrault syndrome.