Transcription regulates ribosome hibernation

Abstract

Most bacteria are quiescent, typically as a result of nutrient limitation. In order to minimize energy consumption during this potentially prolonged state, quiescent bacteria substantially attenuate protein synthesis, the most energetically costly cellular process. Ribosomes in quiescent bacteria are present as dimers of two 70S ribosomes. Dimerization is dependent on a single protein, Hibernation Promoting Factor (HPF), that binds the ribosome in the mRNA channel. This interaction indicates that dimers are inactive, suggesting that HPF inhibits translation. However, we observe that HPF does not significantly affect protein synthesis in vivo suggesting that dimerization is a consequence of inactivity, not the cause. The HPF-dimer interaction further implies that re-initiation of translation when the bacteria exit quiescence requires dimer resolution. We show that ribosome dimers quickly resolve in the presence of nutrients, and this resolution is dependent on transcription, indicating that mRNA synthesis is required for dimer resolution. Finally, we observe that ectopic HPF expression in growing cells where mRNA is abundant does not significantly effect protein synthesis despite stimulating dimer formation, suggesting that dimerization is dynamic. Thus, the extensive transcription that occurs in response to nutrient availability rapidly re-activates the translational apparatus of a quiescent cell and induces dimer resolution.

Introduction

Many microbes exist in a non-proliferating quiescent state that facilitates survival during the periods of nutrient limitation that characterize many environments (Lennon & Jones, 2011, Rittershaus et al., 2013). A major challenge facing quiescent cells is how to minimize energy consumption so as to maximize the limited available resources over a potentially extended time (Bergkessel et al., 2016). The most energy intensive metabolic process in a bacterial cell is protein synthesis, accounting for as much as ~70% of total energy consumption (Russell & Cook, 1995). Consistently, bacteria such as Bacillus subtilis substantially reduce protein synthesis when they exit exponential growth (Diez et al., 2020, Bernhardt et al., 2003). The ribosome is itself the most energetically costly macromolecular machine to synthesize, and it must be protected from degradation during this period of limited de novo ribosome synthesis. Importantly, this protection must be compatible with efficient re-initiation of protein synthesis when conditions improve (Remigi et al., 2019).

During active growth, cells contain 70S ribosomes engaged with mRNA, as well as free small (30S) and large (50S) ribosomal subunits (Green & Noller, 1997, Voorhees & Ramakrishnan, 2013). However, when bacteria enter stationary phase, two 70S ribosomes interact to form a 100S ribosome dimer as observed by ultracentrifugation of bacterial lysates on a sucrose gradient (Tissieres & Watson, 1958, Wada et al., 1990) or in situ by cryo-electron tomography (Ortiz et al., 2010). In phylogenetically diverse bacteria, 100S ribosome dimers are the predominant ribosome configuration in stationary phase (Ortiz et al., 2010, Prossliner et al., 2018) and ribosome dimerization is important for re-growth after stress and starvation (Puri et al., 2014, Akanuma et al., 2016) and for pathogenesis (Kline et al., 2015, McKay & Portnoy, 2015, Fazzino et al., 2015). Dimerization is mediated by a member of the ubiquitous Hibernation Promoting Factor (HPF) protein family that are typically expressed during post-exponential growth or stationary phase. Gamma-proteobacteria such as Escherichia coli encode a short form of HPF that, along with the small protein RMF, binds the ribosome to form the 100S dimer (Wada et al., 1990, Wada et al., 2000, Wada et al., 1995). Gram-positive bacteria, such as B. subtilis and Staphylococcus aureus, do not contain an RMF homolog, but instead encode a long form of HPF with two domains separated by a sequence with an as yet undetermined structure (Matzov et al., 2019). Loss of function mutations in HPF lead to increased ribosome degradation in species including Mycobacterium smegmatis (Trauner et al., 2012), S. aureus (Basu & Yap, 2016) and Pseudomonas aeruginosa. (Akiyama et al., 2017).

The C-terminal domain of each long-form HPF monomer mediates dimerization (Beckert et al., 2017, Franken et al., 2017). We recently demonstrated that in B. subtilis, the absence of HPF leads to loss of the essential ribosomal protein S2, which is the point of contact of the C-terminal domain (Feaga et al., 2020). Similar to HPF proteins from diverse species, the N-terminal domain of B. subtilis HPF binds the 70S ribosome at a site that overlaps both the A- and P- tRNA binding sites and the mRNA channel (Beckert et al., 2017). Therefore, HPF-bound ribosome dimers are incompatible with active translation, leading to the hypothesis that the role of HPF is to attenuate protein synthesis under physiological conditions (Prossliner et al., 2018).

Here, we show that contrary to the prediction of this model, dimerization is most likely a consequence of translation inactivity, not the cause. The structure of HPF-bound ribosome dimers implies that dimers must be quickly resolved in the presence of nutrient-replete conditions so that the cell can resume protein synthesis. We observe that dimers are indeed quickly resolved upon dilution of stationary phase cells into fresh medium. We further demonstrate that this resolution is blocked by the addition of an inhibitor of mRNA synthesis, suggesting that conversion of inactive dimers to active ribosomes is mediated by transcription. Finally, ectopic HPF expression in growing cells has no measurable effect on protein synthesis despite stimulating substantial dimer formation, indicating that dimers may be transient in vivo with the implication that dimerization is more dynamic than previously appreciated.

Results

HPF does not affect protein synthesis in vivo.

We compared protein synthesis in wild-type and Δhpf strains by monitoring incorporation of an alkyne-modified derivative of puromycin (OPP) that is subsequently linked to an azide-modified fluorescent dye by click chemistry (Liu et al., 2012). Puromycin is a tRNA analog that is covalently linked the C-terminus of nascent peptides as they are being synthesized by the ribosome. OPP-labeling enables the quantitation of protein synthesis in single cells under amino acid replete conditions (Diez et al., 2020). We labeled cultures of wild-type and Δhpf cells with OPP under conditions where B. subtilis HPF is expressed (Feaga et al., 2020, Drzewiecki et al., 1998). At the departure from exponential growth when cells enter transition phase (OD₆₀₀ = 2.5), dimers are observed in the wild-type (Fig. 1A), but the incorporation of OPP by wild-type or Δhpf cells was not significantly different between biological replicates of the experiment (p>0.1, unpaired, two-tailed t-test) (Fig. 1C). As expected, cells from a wild-type culture exhibited less protein synthesis at a stationary phase time point (OD₆₀₀ = 6.0) (Fig. 1B) as compared to the transition phase time point (Fig. 1A). However, the Δhpf strain also exhibited a similar decrease, indicating that the reduction in protein synthesis was not dependent on HPF. Three biological replicates of this experiment were performed on different days (Fig. S1A). The averages of these experiments are shown (Fig. 1C).

Fig. 1. HPF does not affect protein synthesis in vivo.

(Left) Wild-type or Δhpf single cells were assayed for protein synthesis during transition phase (OD₆₀₀~2.5) (A) and during stationary phase (OD₆₀₀~6.0) (B) by OPP incorporation. Three biological replicates were performed and a representative experiment (1079 cells) is shown, along with representative microscope images. (Right) An aliquot of cells was harvested in parallel to assay the presence of dimers by sucrose density gradient centrifugation. All biological replicates are included in Fig. S1. (C) Averages of these three biological replicates are shown. Not-significant (ns) indicates a p-value >0.1. Error bars indicate standard deviation.

OPP-incorporation measures abundance of nascent polypeptides in the P-site of the ribosome. 35S-methionine incorporation monitors peptidyl transfer. To confirm these measurements of protein synthesis by a different method, we grew cells to stationary phase in S7 minimal media (OD₆₀₀ of ~6.0), and then added ³⁵S-methionine to radiolabel newly synthesized proteins. Following cell lysis, proteins were resolved by SDS-PAGE and visualized by autoradiography (Fig. S1B). Again, we did not detect a significant difference in levels of protein synthesis in the wild-type parent in comparison with the Δhpf mutant strain (Fig. S1B) (unpaired two-tailed T-test). Thus, even though the abundance of ribosome dimers increased in the wild-type strain in stationary phase as compared to transition phase, correlating with a reduction in protein synthesis (Fig. 1A and B), these two different assays of protein synthesis suggest that dimer formation, per se, does not affect global protein synthesis.

Ribosome dimers resolve within minutes of entry into fresh media

One explanation for a lack of an effect of HPF-dependent ribosome dimer formation on protein synthesis in vivo is that 100S dimers are quickly resolved into translationally competent complexes free of HPF. Consistently, addition of HPF-bound ribosome dimers to a purified ribosome-free in vitro system stimulated protein synthesis within minutes (Fig 2A, (Feaga et al., 2020)) suggesting that 100S dimers are efficiently resolved in a purified system. Therefore, we also examined the dynamics of dimer resolution in vivo by growing B. subtilis to stationary phase and then transferring an aliquot of this culture into fresh media. After five minutes in fresh media with aeration, cells were harvested, lysed, and ribosomes from both the stationary phase culture and the newly diluted culture were analyzed. Ribosomes from the stationary phase culture were present in the dimer conformation, as expected, whereas ribosomes from cells that had been transferred into fresh media were present as 30S, 50S, and 70S assemblies, with very few ribosome dimers observed (Fig. 2B). This observation demonstrates that, similar to E. coli (Aiso et al., 2005), ribosome dimers in B. subtilis are rapidly resolved upon entry into fresh media.

Fig. 2. Ribosome dimers resolve within minutes of dilution into fresh media.

(A) Ribosomes purified from exponential phase or stationary phase cells were added to a ribosome-free PURE system. Exponential phase ribosomes were present in 30S, 50S and 70S configurations. Stationary phase ribosomes were present as dimers when added to the reaction. Production of DHFR-M2 was monitored by western blot using an anti-M2 antibody. A ribosome-free reaction is shown as a negative control. (B) Wild-type (JDB1772) cells were grown to stationary phase in LB (OD₆₀₀ ~6). An aliquot of this culture was diluted 1:4 in fresh LB medium and aerated at 37˚C for 5 min. The stationary phase culture (black, ‘before dilution’) and the diluted culture (blue, ‘after dilution’) were harvested, lysed, and resolved on a 10%−40% sucrose density gradient.

Inhibiting transcription inhibits resolution of ribosome dimers

What mechanism might account for the rapid resolution of ribosome dimers? HPF binds the 70S ribosome at a site overlapping the mRNA channel (Beckert et al., 2017 ) suggesting that mRNA and HPF are competitive for binding. Thus, mRNA produced when cells resume growth could outcompete HPF for ribosome binding, and therefore decrease ribosome dimerization. To test whether transcription is necessary for the resolution of ribosome dimers, we transferred a portion of a stationary phase culture to fresh media in the presence or absence of the transcription inhibitor rifampicin. After 5 minutes, the cultures were harvested, and lysates were resolved by sucrose density gradient centrifugation. When the culture was transferred to fresh media without rifampicin, dimers were resolved as observed previously. In contrast, when the culture was transferred to fresh media containing rifampicin, ribosome dimer abundance was similar to that observed in the parent stationary phase culture (Fig. 3A). This effect was not observed in a strain carrying a mutation in rpoB that renders the strain resistant to rifampicin (Fig. 3B), confirming that the observed effect on dimer abundance was dependent on the ability of rifampicin to directly inhibit transcription. To confirm that the effect of rifampicin was not due to the indirect reduction of protein synthesis that occurs when transcription is inhibited, we tested whether the translation inhibitor fusidic acid (Zhang et al., 2015) affects dimer resolution in vivo. Cultures were grown to stationary phase, and then diluted into fresh LB media containing rifampicin, fusidic acid, or no drug. In contrast to rifampicin, fusidic acid did not prevent the resolution of ribosome dimers (Fig. 3C), demonstrating that rifampicin was preventing dimer resolution by inhibiting the synthesis of mRNA and that protein synthesis, including that required for ribosome biogenesis, is not necessary.

Figure 3. Rifampicin treatment prevents resolution of ribosome dimers.

(A) Wild-type (JDB1772) cells were grown to stationary phase (OD₆₀₀ ~6) and diluted 1:4 in fresh media with or without 5 μg/mL rifampicin. (B) Rifampicin resistant cells (JDB4420) were grown to stationary phase (black, ‘before dilution’) and diluted 1:4 in fresh media (blue, ‘after dilution’) or with rifampicin. (red; ‘after dilution +rif’) (C) Wild-type (JDB1772) cells were grown to stationary phase and diluted 1:4 in fresh media containing no drug (blue, ‘after dilution’), rifampicin (red, ‘after dilution +rif’), or fusidic acid (green, ‘after dilution +fus’).

Ectopic expression of HPF does not attenuate protein synthesis during exponential phase

Together, our observations that dimers are quickly resolved under nutrient-replete conditions (Fig. 2B) and that dimer resolution depends on transcription (Fig. 3A) suggest that ectopic expression of HPF in transcriptionally active cells would not affect protein synthesis. To examine this possibility, we constructed a strain expressing HPF under the control of an inducible promoter. The ectopically expressed HPF was functional, as revealed by a 100S peak in density fractionated lysate derived from the induced culture (Fig. 4A and B). This strain contains an Δhpf deletion at the endogenous locus so only the effect of ectopically expressed HPF is assayed. Fewer dimers were produced if HPF was induced in early log phase (OD₆₀₀ 0.45) (Fig. 4A) than in late log phase (OD₆₀₀ 0.8) (Fig. 4B), consistent with our observation that rapidly growing cells produce fewer dimers (Fig. 3). We assayed OPP incorporation in cells induced in late log phase (Fig. 4B), since this condition produced the most ribosome dimers. We also determined that the ectopically expressed HPF was expressed to a level similar to that observed during stationary phase (Fig. S2A). As above, we also assayed protein synthesis using both ³⁵S-methionine incorporation (Fig. S2B). Both methods revealed that induction of HPF did not exert a statistically significant effect over biological replicates (Fig. 4B, S2B) (unpaired, two-tailed T-test). Thus, ectopic expression of HPF in exponential phase has no measurable effect on protein synthesis in these assays.

Figure 4. Ectopic expression of HPF does not significantly affect protein synthesis.

(A) Cells from strain JDB4228 (Δhpf amyE::P_hyperspank-hpf) were grown in LB to mid-log phase (OD₆₀₀ ~0.45) and induced with IPTG for 20 minutes. Cell lysate was harvested and resolved on a 10%−40% sucrose gradient. (B) Cells from strain JDB4228 were grown in LB to OD₆₀₀ ~0.8, induced with IPTG for 20 min and labeled with OPP. Cells were analyzed by fluorescence microscopy. A representative of two independent experiments is shown (836 cells measured). An aliquot of cells from this experiment grown in the presence (blue) or absence (orange) of IPTG were harvested and the abundance of ribosome dimers was assayed by sucrose density gradient centrifugation.

Discussion

Structures of 100S ribosome dimers show that HPF binding overlaps with the binding sites of essential translation factors (Matzov et al., 2019, Beckert et al., 2017, Polikanov et al., 2012, Franken et al., 2017). For example, the position of B. subtilis HPF on the 100S ribosome dimer overlaps the mRNA channel and the A and P tRNA binding sites (Beckert et al., 2017). HPF binding would sterically hinder these interactions and therefore it has been hypothesized that HPF inhibits translation. However, we find that wild-type and Δhpf strains exhibit similar protein synthesis activities (Fig. 1A, B) and ectopic expression of HPF does not inhibit protein synthesis even though it stimulates dimer formation (Fig. 4A, B). How can these results be reconciled with the structural data clearly suggesting that dimers are translationally inactive? Our observations that ribosome dimers are resolved quickly in vivo upon increased nutrient availability (Fig. 2B) and that in vitro translation is not affected by the addition of purified ribosome dimers, which are bound to HPF upon addition to the reaction (Fig. 2A), suggests that dimers in cells are rapidly converted into active 70S monomers free of HPF. Such behavior would reconcile our in vivo protein synthesis data with the 100S dimer structure, and suggests that dimerization is a consequence of translation inhibition. Our observation that inhibition of mRNA synthesis blocks this resolution (Fig. 3A) suggests that HPF binds the ribosome when the mRNA:HPF ratio is favorable. This situation is present under post-exponential conditions where HPF is expressed and transcription is attenuated (Planson et al., 2020). In cells that are transcriptionally active, HPF does not compete with mRNA sufficiently to inhibit translation. Thus, resolution of inactive 100S ribosome dimers to active 70S ribosomes upon nutrient availability is a direct consequence of the rapid increase in transcription that occurs during this transition (Planson et al., 2020).

Rapid dimer resolution in vivo implies that inactive (dimer) and active ribosome configurations (70S/50S/30S) exist in a dynamic equilibrium in cells, and that dimer disassembly is likely a passive process in B. subtilis. We propose that the relative stoichiometry of mRNA::HPF shifts this equilibrium either towards inactive dimers or toward active ribosomes. Several aspects of this speculative mechanism including its in vivo or in vitro kinetics and the specific pathway in vivo of dimer resolution are not clear. For example, does the 100S dimer initially resolve into two inactive 70S monomers or into a dimer of 30S subunits still bound by HPF? Given the absence of techniques to quantitatively analyze the dynamics of ribosome dimer assembly in vivo, answers to these questions await future technological development.

Presumably, competition between mRNA and HPF for binding to the mRNA channel would result in such an equilibrium. How might this competition take place? During initiation, the 30S Initiation Complex (30S IC) forms when the mRNA, the initiator N-formyl-methionyl-transfer RNA (fMet-tRNA^fMet) and the 30S subunit are brought together by the Initiation Factors IF1, IF2 and IF3 (Gualerzi & Pon, 2015). Thus, the HPF::mRNA stoichiometry would directly affect both initiation of translation and inversely ribosome dimerization. The E.coli pY protein, which is structurally homologous to the N-terminal domain of HPF, also binds the mRNA channel and pY competes with the binding of fMet-tRNA^fMet to the ribosome (Vila-Sanjurjo et al., 2004). Interestingly, Nsp1 of SARS-Cov2, which is important for down-regulating host protein synthesis during infection, similarly binds the mRNA channel of the eukaryotic ribosome. Nsp1 competes with RNA downstream of the start codon to bind the 40S subunit and the protein is unable to associate rapidly with 80S ribosomes assembled on an mRNA (Lapointe et al., 2020). Nsp1 also induces endonucleolytic cleavage of host mRNAs (Kamitani et al., 2006), a mechanism that would similarly favor Nsp1 binding as our experiments that used an inhibitor of transcription to favor HPF in the HPF:mRNA stoichiometry, thereby stabilizing dimers (Fig. 3).

If HPF binding is a consequence of the attenuation of transcription, the competition between HPF and mRNA implies that HPF is preferentially associating with ribosomes that are inactive. Consistently, we observed a 100S ribosome dimer peak in exponentially growing cells expressing HPF even though there was no measurable decrease in protein synthesis (Fig. 4A, B). This observation suggests that a substantial pool of non-translating ribosomes exists in actively growing cells as has been proposed (Remigi et al., 2019, Dai et al., 2016, Li et al., 2018). One way to increase the fraction of inactive ribosomes is by inhibiting translation initiation. We and others have shown that the alarmone (p)ppGpp, which is produced during post-exponential growth, inhibits IF2 (Milon et al., 2006, Vinogradova et al., 2020, Diez et al., 2020). Since IF2 mediates the isomerization of the 30S pre-initiation complex and promotes the subsequent association of the large 50S subunit (Caban et al., 2017), inhibition of IF2 by (p)ppGpp would likely increase the abundance free 30S subunits that are capable of association with HPF (Wishnia et al., 1975).

(p)ppGpp also regulates the expression of HPF in both B. subtilis and E.coli suggesting that HPF expression and dimer formation may be directly coupled to conditions that do not favor active translation such as high (p)ppGpp synthetic activity (Diez et al., 2020). In Gram-positive bacteria such as B. subtilis, (p)ppGpp synthesis is mediated by the dual function (p)ppGpp synthase/hydrolase RelA, which associates with the ribosome, and two non-ribosome associated synthases, SasA (RelP) and SasB (RelQ) that predominantly synthesize pGpp and ppGpp, respectively (Gaca et al., 2015, Irving & Corrigan, 2018, Yang et al., 2020). SasB is allosterically activated by pppGpp (Steinchen et al., 2015), and this allosteric activation is inhibited by mRNA (Beljantseva et al., 2017). Thus, increases in transcription would reduce synthesis of ppGpp, a known inhibitor of translation (Diez et al., 2020), and stimulate ribosome dimer resolution (Fig. 5). Although recent evidence suggests that transcription and translation in B. subtilis are not directly coupled during active growth (Johnson et al., 2020), our experiments are consistent with a functional coupling at least under nutrient limitation, as has been suggested for E. coli synthesizing (p)ppGpp (Chen & Fredrick, 2020). Thus, when transcription is attenuated and HPF is expressed during nutrient limitation, the equilibrium between HPF and mRNA would favor 100S dimer formation (Fig. 5A). In the presence of sufficient nutrients, increases in transcription would shift this equilibrium to favor 70S ribosomes and thereby stimulate translation (Fig. 5B).

Fig. 5. Transcriptional regulation of ribosome dimer formation during entry into and exit from quiescence.

Entry into quiescence (top): Following departure from exponential growth, the (p)ppGpp synthases RelA and SasB become activated and expressed, respectively, resulting in an increase in [ppGpp] (Nanamiya et al., 2008). ppGpp attenuates translation, by inhibiting IF2 (Diez et al., 2020), and attenuates transcription (Krasny & Gourse, 2004, Gourse et al., 2018) by causing a decrease in [GTP] (Anderson et al., 2020). Therefore, the rise in [ppGpp] results in increased abundance of 30S and 50S subunits that can form empty 70S ribosomes. In addition, ppGpp induces the expression of HPF (Tagami et al., 2012), which forms 100S ribosome dimers from the empty 70S ribosomes. Exit from quiescence (bottom): In the presence of sufficient nutrients [GTP] increases (Planson et al., 2020, Lopez, 1982), stimulating transcription (Krasny & Gourse, 2004, Gourse et al., 2018). mRNA production also inhibits SasB (Beljantseva et al., 2017), preventing its synthesis of ppGpp and thereby facilitating the rise in [GTP] that promotes further transcription. The increased transcription alters the mRNA:HPF stoichiometry, leading to the resolution of the 100S dimers to active 70S monosomes. This could be a consequence of shifting the equilibrium of monomer::dimer as the result of increased 30SIC formation in the presence of mRNA.

Several other candidate mechanisms have been proposed for dimer resolution. For example, the GTPase HflX dissociates 100S dimers in vitro (Basu & Yap, 2017). However, the effect of HflX is not necessary for dimer resolution as an in vitro system lacking HflX is capable of mediating the resolution of ribosome dimers (Fig. 2A) and strains lacking hflX do not exhibit substantially increased dimer abundance in vivo (Basu & Yap, 2017). Alternatively, in S. aureus, ribosome-recycling factor (RRF) and the GTPase EF-G, which together act to recycle 70S ribosomes in post-termination complex, also resolve 100S ribosomes in a GTP-dependent fashion (Basu et al., 2020). Interestingly, HPF is expressed in S. aureus during exponential growth (Basu & Yap, 2016, Ueta et al., 2010), so EF-G and RRF may resolve ribosome dimers that are subsequently prevented from re-forming because of the presence of mRNA. In contrast, B. subtilis EF-G does not appear to be required for dimer splitting since addition of fusidic acid did not prevent dimer resolution during re-growth (Fig. 3C). However, EF-G catalyzes both elongation and ribosome recycling during translation, and the inhibitory effect of fusidic acid on elongation is stronger (Zhang et al., 2015). Since fusidic acid stalls ribosomes on mRNA, the effect of fusidic acid on dimer recycling could be masked by an increase in stalled ribosomes. In either case, these results support a model in which mRNA is sufficient to block HPF binding and dimer formation.

Our experiments suggest that resolution of ribosome dimers involves a competition between HPF and mRNA for ribosome binding, presumably mediated by the N-terminal domain of HPF that contacts the mRNA channel of the ribosome. An unresolved question is the role of the C-terminal domain of HPF in dimer resolution. This domain mediates dimerization and interacts with ribosomal protein S2 (Kato et al., 2010, Beckert et al., 2017). Intriguingly, S2 is also a point of direct interaction between the ribosome and RNA polymerase ω subunit in the so-called expressome (Kohler et al., 2017, Demo et al., 2017, Webster et al., 2020). Alternatively, the transcription factor NusA may mediate this interaction, since it contacts both S2 and the RNAP β’ subunit in E. coli (Wang et al., 2020) and M. pneumoniae (O’Reilly et al., 2020). Thus, given our demonstration that transcription is necessary for dimer resolution (Fig. 3A), a subject for future research is investigating how direct interaction with RNAP affects ribosome dimer resolution.

Recent work from our laboratory demonstrates that the absence of HPF results in the loss of essential proteins S2 and S3 from the ribosome during incubation in stationary phase (Feaga et al., 2020). The HPF-dependent protection is due to dimerization mediated by the C-terminal domain as mutant forms of HPF that still bind the ribosome via their N-terminal domain do not prevent the loss of S2/S3 (Feaga et al., 2020). Based on experiments presented here, we propose that the function of the N-terminal domain is to act as a sensor for mRNA. Thus, resolution of ribosome dimers is coupled to the presence of mRNA, the ribosome substrate (Fig. 5B), and, in the absence of mRNA, HPF protects the ribosome from degradation. Altogether, these results suggest that Bacillus subtilis HPF is responsible for ribosome storage during conditions of nutrient limitation where mRNA synthesis is attenuated and that this mechanism can be quickly reversed in the presence of growth promoting conditions as signaled by active transcription.

Materials and Methods

Strains and media.

Strains were derived from B. subtilis 168 trpC2. To construct JDB4228 (Δhpf amyE::P_hyperspank-hpf) for exogenous expression of HPF, genomic DNA was amplified with primers HF152 (ATATAAGCTTAGGGAGGCGTTCTTTGATGAAC) and HF177 (CCGGGCTAGCTCATTATT CAGTCGGTTCAATTAAGCCAT) in order to amplify HPF and its native ribosome binding site. The resulting PCR product was cut with NheI and HindIII and cloned into the same sites of pDR111. pDR111-HPF was linearized and transformed into the Δhpf background and double cross-overs were selected on plates containing spectinomycin. Expression of inducible HPF was confirmed by western blot. Cells were grown in LB or S7 minimal media containing glycerol as a carbon source for radiolabeling experiments. For experiments with antibiotics, rifampicin was used at 5 μg/mL (10X MIC) and fusidic acid was used at 3.8 μg/mL (10X MIC).

Assay for translation by ³⁵S-methionine incorporation.

To assay wild-type or Δhpf cells in stationary phase, cells were grown in S7 minimal media to OD₆₀₀ ~7.0. To assay the effect of endogenous HPF induction cells were grown in S7 minimal media containing glycerol as the carbon source to OD₆₀₀ = 0.8 and induced with 1 mM IPTG for 20 min. Next, 1 mL of cells was transferred to a new tube and 4 μL of ³⁵S-methionine was added. Cells were incubated at 37˚C and aliquots removed after 10, 20, and 30 min, quenched with cold methionine, and placed on ice. Cells were lysed with 50 mM EDTA+ 1mg/mL lysozyme. Protein was resolved on by SDS-PAGE and viewed by autoradiogram.

Detection of protein synthesis in vivo using click chemistry.

As described (Atwal et al., 2020), 300 μL of cells were transferred to a 5 mL glass tube and 0.25 μL of 20 mM Click-IT OPP reagent (Thermofisher) was added. Cells were placed on a roller drum at 37˚C for 10 min to allow OPP incorporation, after which the reaction was quenched and cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS). Preparation of cells for microscopy proceeded with minor deviations from the manufacturer’s instructions as described (Diez et al., 2020). For analysis of WT vs. Δhpf cells, cells were grown to transition phase (OD₆₀₀~2.5) and stationary phase (OD₆₀₀~6.0) and labeled with the Click-IT reagent. Three biological replicates were performed on three separate days (Fig S1). A representative experiment is shown in both Figure 1 and Figure S1. For the HPF induction experiment, two biological replicates were performed on separate days, and a representative experiment is shown in Figure 4.

Microscopy and image analysis

Cells were spotted on agar pads containing 1% agarose in PBS and imaged with a Nikon 90i microscope using 100x phase contrast. Exposure times were 30 msec for phase contrast, and 20 msec for mCherry. Fluorescence intensity of ~500 single cells per experiment was determined using ImageJ and plotted with Prism.

Sucrose gradient density centrifugation

Harvested cells were lysed using a Fast Prep machine in ribosome gradient buffer (20 mM Tris-acetate [pH_4˚C 7.4], 60 mM ammonium chloride, 7.5 mM magnesium acetate, 6 mM β-mercaptoethanol, 0.5 mM EDTA). Lysate was cleared of debris by centrifugation at 4˚C at 20,000 rcf for 20 min. Ribosomes were purified by sucrose cushion (37.7% sucrose in ribosome gradient buffer). A260 for lysates or purified ribosomes was determined by Nanodrop, and 100 μL of normalized lysate or ribosomes (as indicated in figure legends) was loaded on a 10% to 40% sucrose gradient in ribosome gradient buffer. Gradients were run for 3 hours at 30,000 rpm in a Beckman SW41 rotor. Samples were collected with a BioComp gradient station and a BioComp TRiAX flow cell monitoring continuous absorbance at 260 nm.

In vitro translation

Ribosomes were purified from exponential (OD₆₀₀ = 0.9) and stationary phase (OD₆₀₀ ~6) cultures as described (Feaga et al., 2020). Ribosome concentrations were quantified by measuring A₂₆₀ of dissociated subunits and further confirmed by visualizing rRNA by agarose gel electrophoresis. Ribosomes were added to the ribosome-free PURExpress system (New England Biolabs) (Tuckey et al., 2014) to a final concentration of 100 nM. 125 ng of DNA template encoding M2-tagged DHFR (MW = ~20kDa) was added and the reaction incubated at 37˚C. The reaction was stopped after 10, 20 and 30 min by addition of SDS-PAGE sample buffer. Synthesis of the resulting M2-tagged DHFR was assayed by western blot using a commercial M2-antibody.

Western Blot

HPF expression was monitored by western blot using an aliquot of culture harvested after 20 min in the presence or absence of IPTG, or from wild-type cells grown to stationary phase (OD₆₀₀ ~6). Cultures were normalized by OD, pelleted, and resuspended in SDS-PAGE buffer. Lysate was resolved by SDS-PAGE and probed with anti-HPF antibody (Feaga et al., 2020) followed by HRP-conjugated anti-rabbit secondary (Sigma). Samples for western blot of in vitro translation reactions were resolved by SDS-PAGE and probed with HRP-conjugated anti-M2 antibody (1:20,000; Sigma).

Table 1.

Bacterial Strains and Plasmids

Strain	Relevant genotype	Description	Origin
JDB1772	trpC2	B. subtilis parent	(Gaidenko et al., 2002)
JDB4221	trpC2 ∆hpf::tet	hpf deletion	(Feaga et al., 2020)
JDB4228	trpC2∆hpf::tet Phyperspank-hpf	Ectopic expression of HPF	This study
JDB4420	trpC2 rpoB500	Rifampicin resistant	(Sonenshein et al., 1974)