Elongation factor P is required for EIIGlc translation in Corynebacterium glutamicum due to an essential polyproline motif

Translating ribosomes require elongation factor P (EF‐P) to incorporate consecutive prolines (XPPX) into nascent peptide chains. The proteome of Corynebacterium glutamicum ATCC 13032 contains a total of 1,468 XPPX motifs, many of which are found in proteins involved in primary and secondary metabolism. We show here that synthesis of EIIGlc, the glucose‐specific permease of the phosphoenolpyruvate (PEP): sugar phosphotransferase system (PTS) encoded by ptsG, is strongly dependent on EF‐P, as an efp deletion mutant grows poorly on glucose as sole carbon source. The amount of EIIGlc is strongly reduced in this mutant, which consequently results in a lower rate of glucose uptake. Strikingly, the XPPX motif is essential for the activity of EIIGlc, and substitution of the prolines leads to inactivation of the protein. Finally, translation of GntR2, a transcriptional activator of ptsG, is also dependent on EF‐P. However, its reduced amount in the efp mutant can be compensated for by other regulators. These results reveal for the first time a translational bottleneck involving production of the major glucose transporter EIIGlc, which has implications for future strain engineering strategies.

Transcriptional regulators, such as SugR, FruR, GntR1, and GntR2, form a complex network that up-and downregulates the expression of sugar transporters and metabolic pathways according to sugar availability (Engels and Wendisch, 2007;Frunzke et al., 2008;Gaigalat et al., 2007;Tanaka et al., 2008a;Tanaka et al., 2008b;Tanaka et al., 2014). The C. glutamicum gene ptsG (Cgl1360, cg1537) encodes the glucose-specific, membrane-bound EIIBCA component (EII Glc ) responsible for the majority of glucose uptake (Moon et al., 2005;Moon et al., 2007) (Figure 1). In an attempt to increase glucose uptake and growth rates, transcription of ptsG has been placed under the control of inducible promoters. However, the higher mRNA levels were only very weakly reflected in EII Glc copy numbers or increased glucose uptake rates (Frunzke et al., 2008;Krause et al., 2010;Lindner et al., 2013;Pfeifer et al., 2017;Tanaka et al., 2008a;Wang et al., 2014;Wang et al., 2018). It was also shown that the presence of the corresponding sugars increases the size of PTS clusters within the membrane without significant increases in copy number (Martins et al., 2019). These observations suggested that other regulatory mechanisms play a role in EII Glc production and glucose uptake capacity.
Ribosomes stall if certain polyproline motifs (XPPX) have to be incorporated into the polypeptide. In bacteria, elongation factor P (EF-P) evolved to overcome this translational obstacle (Doerfel et al., 2013;Ude et al., 2013). EF-P orthologs with the same function, named aIF-5A and eIF-5A, exist in all archaea and eukaryotes, respectively (Gutierrez et al., 2013;Prunetti et al., 2016). For their activation, EF-P and its eIF5A/aIF5A orthologs usually require the post-translational modification of a conserved amino acid residue located at the tip of a loop (Hummels et al., 2017;Lassak et al., 2015;Peil et al., 2012;Roy et al., 2011;Yanagisawa et al., 2010). We recently demonstrated that the EF-Ps of Actinobacteria-specifically, C. glutamicum, Streptomyces coelicolor, and Mycobacterium tuberculosis-alleviate ribosome stalling at polyproline motifs without the need for any activating posttranslational modification (Pinheiro et al., 2020).
In C. glutamicum, EF-P is required for the synthesis of many polyproline-containing enzymes of primary and secondary metabolism, as well as regulatory proteins (Pinheiro et al., 2020). Various studies have shown that polyproline motifs can play important functional roles in the catalytic center of enzymes, in the downregulation of the copy numbers of receptors, and in protein-protein interactions Starosta et al., 2014b;Ude et al., 2013). These motifs might also be important in protein folding and membrane insertion (Qi et al., 2018).
Here, we report that the translation of EII Glc in C. glutamicum is strongly dependent on EF-P, due to the presence of an essential polyproline motif at position 235/236. In a ∆efp mutant, very little EII Glc is produced and glucose uptake is correspondingly reduced, thus, essentially preventing the growth of cells on glucose as sole carbon source. These findings underline the fact that the translational level also needs to be considered in strain engineering.

| Elongation factor P is required for fast growth of C. glutamicum on glucose
Although virtually all diproline-containing motifs cause translational stalling, the duration of stalling is modulated by the amino acids located upstream and downstream of the arrest motif. Therefore, polyproline motifs can be classified as weak, moderate, or strong according to their ability to trigger ribosome stalling (Elgamal et al., 2014;Hersch et al., 2013;Starosta et al., 2014a;Woolstenhulme et al., 2015). This classification takes into consideration the translation initiation rate, the position of the motif within the peptide chain, and most importantly, the amino acid context up-(−2 and −1) and downstream (+1) of the diproline sequence itself. Evidence that there is also a hierarchy of pausing motifs in C. glutamicum originates from our previous proteome study in which the ∆efp mutant was compared with the parental wild-type strain (Pinheiro et al., 2020). We analyzed the amino acid sequences of the main proteins responsible for carbohydrate uptake, metabolism, and transcriptional regulation in C. glutamicum ATCC 13032 ( Figure 1, Table 1).
Among these sequences, moderate to strong stalling motifs were found in EII Glc , the permease of the glucose PTS, and GntR2, a global transcriptional regulator which, among other genes, stimulates the transcription of ptsG (Frunzke et al., 2008;Ikeda, 2012;Tanaka et al., 2014) (Figure 1, Table 1). Weak XPPX stalling motifs were found in EII Fru , PfkB, EII Scr , ScrB, GntP, and FruR ( Figure 1). To test whether EF-P is required for carbohydrate uptake and metabolism, we grew cells of the wild type and the efp deletion mutant in defined minimal medium in the presence of glucose, fructose, sucrose, or ribose as a sole carbon source. We observed major growth defect of the ∆efp mutant in a medium containing glucose as sole carbon source ( Figure 2e). The mutant exhibited slightly growth impairment relative to the wild type when grown on ribose, gluconate, fructose, or sucrose as sole carbon source ( Figure 2). Therefore, the severity of the growth defect correlates well with the expected stalling efficacies of the polyproline motifs observed in these transporters. After providing the efp gene in trans, growth of the mutant was indistinguishable from wild type under all conditions (Figure 2, gray symbols). As a further control, the efp mutant was grown in rich BHI medium, and no growth defect was observed ( Figure 2f).

| EF-P is required for translation of EII Glc
EII Glc has the strongest stalling XPPX motif found in any of the carbohydrate transporters identified in C. glutamicum. It was previously shown that fusion of the fluorescent mNeonGreen protein to the N-terminus of EII Glc (mNG-EII Glc ) has no detectable effect on the transporter's function (Martins et al., 2019). To investigate the impact of EF-P on the production of the transporter, we analyzed the fluorescence of cells expressing this chromosomally encoded mNG-EII Glc hybrid in efp + and efpgenetic backgrounds. The overall fluorescence of the efp − cells was significantly lower than that of the efp + control (Figure 3a,b). In addition, many of the efp − cells showed fluorescence values in the range of the background fluorescence of untagged C. glutamicum ATCC 13032 indicating that mNG-EII Glc production was strongly reduced in the mutant (Figure 3a,b). It should be noted that the foci in these cells are polyphosphate granules, which are well known when C. glutamicum is imaged at 488 nm excitation (Martins et al., 2019).
The positive effect of EF-P on EII Glc synthesis could also be demonstrated in vitro. In this experiment, we used an in vitro tran- , which allowed us to quantify the protein on Western blots using anti-Flag-Tag antibodies. In the presence of purified EF-P, the EII Glc translation rate was 2.3-fold faster than the control value without EF-P ( Figure 3c).
We then tested whether the lower EII Glc amount in the C. glutamicum Δefp mutant affects the rate of glucose uptake. The kinetics of EII Glc -mediated uptake of radiolabeled d-glucose-6-14 C were described previously, and yielded a K m of 14 μM and a V max of 35 ± 3 nmol min −1 mg −1 DW (Lindner et al., 2011). Under the same conditions, we obtained similar glucose uptake rates for the wild type, however, the uptake rate of the ∆efp mutant was significantly lower (Figure 3d). At saturating glucose concentrations, the uptake rate was determined to be 38 ± 4 nmol min −1 mg −1 DW for the wild type and 25 ± 4 nmol min −1 mg −1 DW for the Δefp mutant. As a negative control, we measured a glucose uptake rate of 0.07 ± 0.09 nmol min −1 mg −1 DW for the ΔptsG mutant. Taken together, these results reveal that the ∆efp mutant is still able to take up glucose, albeit at a markedly reduced rate, in agreement with the significantly reduced amount of the permease.

| Importance of the polyproline motif for EII Glc function
The requirement of EF-P for the translation of EII Glc raises the ques- We then used primers containing wobble codons for semi-random mutagenesis to construct a ptsG library in which the proline codons are replaced by random amino acid codons. We sequenced several of the resulting clones, which confirmed the diversity of sequences generated ( Figure 4d). Nevertheless, the introduction of the ptsG variant library into the ΔptsG ΔiolT1 ΔiolT2 strain allowed growth on glucose only when the wild-type sequence was expressed (Brühl, 2015). It is important to mention here that IolT1 and IolT2 are two inositol permeases that can function as glucose transporter and are able to suppress the growth retardation in the absence of glucose-PTS system (Ikeda et al., 2011;Lindner et al., 2011). Therefore, the additional deletions in iolT1 and iolT2 were introduced into the ΔptsG mutant of C. glutamicum to avoid formation of unwanted second-site suppressor mutations. Our finding suggests that the XPPX-motif is important for the enzyme's function.
The XPPX motif is located in the EIIC domain of the protein, which is responsible for the translocation of the carbohydrate  this question, we fused the gntR2 gene to a sequence encoding the fluorescent protein mCherry (C-terminal), and inserted this construct into the native locus in C. glutamicum. Then, we quantified the fluorescence of GntR2-mCherry in both efp + and efpstrains ( Figure 5a,b). The GntR2-mCherry level was 4.8 times lower in the efp − strain, confirming the dependency on EF-P. However, transcription of ptsG was not altered in the efp − strain (Figure 5c). The lack of an effect of the efp deletion could be related to the presence of two GntR-like regulators (GntR1 and GntR2) with redundant functions in C. glutamicum ATCC 13032. Only when both genes are deleted, growth defects are observed in glucose-containing medium, while each of the ΔgntR1 or ΔgntR2 mutants behaves like the wild type (Frunzke et al., 2008).

| D ISCUSS I ON
EII Glc is the major glucose transporter in C. glutamicum. Increasing the expression of ptsG, which codes for the glucose transporter, is a commonly used strategy to increase the glucose consumption rate in order to boost the production of amino acids and secondary metabolites (Krause et al., 2010;Lindner et al., 2013;Xu et al., 2016).
However, so far, the logarithmic increases in ptsG transcription levels caused by inducible promoters have been followed by only a small increase in EII Glc copy number and/or glucose uptake rate (Krause et al., 2010;Lindner et al., 2013;Wang et al., 2014;Wang et al., 2018). In this study, we demonstrate through a combination of bioinformatic analysis and phenotypic characterization of a C. glutamicum efp deletion mutant that EF-P is required for the translation of this carbohydrate transporter. In vivo, the deletion of efp resulted in a decrease in the content of EII Glc , lower glucose uptake rates, and impaired growth of cells on glucose as carbon source. The direct effect of purified EF-P on the translation of EII Glc was confirmed in vitro. Interestingly, the transcription of ptsG was not affected in the efp mutant, although one of its transcription activators GntR2 is dependent on EF-P. The fact that the reduction in the copy number of GntR2 in the efp mutant had little or no impact on ptsG transcription, could be explained by other regulators in C. glutamicum ATCC 13032 that compensate for its loss (Frunzke et al., 2008). These results indicate that regulation of ptsG expression is rather robust, whereas translation is impeded by periods of stalling, which can be alleviated by the activity of EF-P. This translational regulation might serve to prevent the overproduction of EII Glc molecules, thus protecting C.
glutamicum from the so-called phosphosugar stress . In E. coli protection against phosphosugar stress is provided by a complex regulatory network involving small RNA-initiated inhibition of ptsG translation, and Hfq-dependent ptsG mRNA degradation by RNase E (Maki et al., 2008;Morita et al., 2005). C. glutamicum does not possess an Hfq homolog (Kalinowski et al., 2003), and therefore, might use the polyproline-dependent stalling regulation instead.
EF-P was also found to be important for the synthesis of other carbohydrate transporters. The efp deletion mutant shows growth impairment depending on the available carbohydrates, and the severity of the growth defects is correlated with the efficacy of the polyproline stalling motifs that occur in the sequences that code for these proteins.
Polyproline motifs are frequently found in protein-protein interaction sites. In EII Glc , the EF-P-dependent polyproline motif is predicted to be located in a transmembrane domain of the EIIC subunit of EII Glc , which is involved in dimer formation. Avoiding a bottleneck in EII Glc production by replacing one or the two prolines of this motif completely inhibited growth of the corresponding C. glutamicum mutants on glucose as sole carbon source. An unbiased semi-random mutagenesis approach further confirmed that the consecutive prolines are essential for the function of EII Glc . Previous studies on the role of polyproline motifs in membrane-integrated proteins have focused on the E. coli acid stress receptor CadC (Ude et al., 2013) and the osmosensor EnvZ . CadC has a strong polyproline motif that fine-tunes its copy number (Ude et al., 2013).
A CadC variant in which the triproline motif is replaced by a pair of alanines is characterized by threefold higher copy number and a less sensitive stress response. On the contrary, the polyproline motif in EnvZ did not affect receptor copy number, but was found to be essential for dimerization and interaction with the modulator MzrA . Here, we propose that the polyproline motif in C. glutamicum EII Glc has a dual function: it is essential for permease activity, but also fine-tunes the amount of the transporter that is produced.
Overall, our study shows that EF-P plays an important role in the translation of carbohydrate transporters in C. glutamicum. Regulation at the translational level might be considered in the process of strain optimization, especially when aiming to increase carbohydrate uptake rates.

| Nucleotides, plasmids, and bacterial strain construction
DNA sequences, plasmids and strains used in this study are listed in Supplementary Tables S1-S3. Genomic DNA was purified from C.  Table S2 and identified by SacB counterselection. Codon replacements were introduced by overlap-extension PCR using mismatched primers (Ho et al., 1989). Primers containing random codons at positions 235 or 236 of ptsG were used to construct a ptsG library in the self-replicating vector pEKEx2.

| Growth conditions
Brain-heart infusion medium (BHI-Becton Dickinson) 37 g/L was used as the standard complex medium for growth of C. glutamicum.
Cells were also grown in chemically defined CGXII media supplemented with 2% (w/v) of the specified carbon source (Keilhauer et al., 1993). In general, cells were cultivated in 100 ml-baffled flasks filled with 25 ml with a starting OD 600 of 1 at 30°C on a rotatory shaker. Overnight cultures were done in BHI medium, and cells were washed in phosphate-buffered saline (PBS; pH 7.4) before inoculation. In some cases (Figure 4), growth experiments were done in 96-well plates incubated at 30°C and 220 rpm. Cells were freshly transformed to avoid formation of suppressor mutants. 300-ms or 60-ms exposure time, respectively. At least 300 cells per condition were analyzed. Digital images were analyzed using Fiji (Schindelin et al., 2012).

| RT-qPCR analysis
The RNA used for reverse transcription qPCR was isolated using the phenol-chloroform-isoamyl alcohol (PCI) protocol (Russell and Sambrook, 2001)  [v/v] glycerol), and lysed using a high-pressure system (Constant Systems). The cytosolic protein fraction was obtained by centrifugation of the lysate, and EF-P was purified on a Ni 2+ -nitrilotriacetic acid (NTA) resin (Qiagen). After washing with lysis buffer supplemented with 20 mM imidazole, bound EF-P was eluted with 200 mM imidazole in the same buffer.

| In vitro translation of EII Glc
The PURExpress In vitro Protein Synthesis Kit (NEB, E6800) was used for in vitro translation of EII Glc , with the following modifications. To avoid protein aggregation, the reaction mix was supplemented with 1 mM arginine (pH 7.0) and 1 mM β-mercaptoethanol. Purified posttranslationally modified E. coli EF-P was added at 1 µM concentration.
The same amount of lysis buffer was added to the negative control. loaded onto a SDS-polyacrylamide gel (Laemmli, 1970) to fractionate the proteins by size, and further analyzed after Western blotting.

| Western blot
The wet-transfer method was used to transfer the proteins from SDSpolyacrylamide gels to nitrocellulose membranes (Amersham, GE Healthcare Images were taken using the Odyssey CLx imaging system (LI-COR Biosciences).

| d-glucose-6-14 C uptake assay
All uptake measurements were performed as described earlier with minor modifications (Lindner et al., 2011). Cells were grown in BHI medium supplemented with 2% (w/v) glucose to OD 600 2, harvested by centrifugation, then washed three times in ice-cold CGXII medium (without carbon sources), resuspended to OD 600 = 2 and stored on ice until measurement. Prior to measurement, 2-mL aliquots of cell culture were incubated for 3 min at 30°C in a water bath. The reaction was started by the addition of 5, 50 or 500 µM d-glucose-6-14 C (59 mCi/nmol; Sigma, G9899). At 30-sec intervals, 200-µL samples were filtered through glass-fiber filters (Type F, Millipore) and washed twice with 2.5 ml of 100 mM LiCl. The radioactivity in the samples was determined using scintillation fluid (MP Biomedicals) and a scintillation counter (PerkinElmer).

| EII Glc tertiary-structure prediction
The amino acid sequence of the EIIC domain of EII Glc was uploaded to the Phyre2 platform using standard parameters (Kelley et al., 2015).

| Construction of the EIIC phylogenetic tree
The amino acid sequence of the EIIC domain of C. glutamicum EII Glc (Uniprot reference Q46072) corresponding to the amino acids from position 117 to 476 was downloaded and the 20.000 most similar sequences identified using Basic Local Alignment Search Tool (BLAST), excluding sequences from uncultured/environmental samples. The data set was reined by deleting partial sequences, hypothetical proteins, sequences that do not contain domain IIC and identical sequences. Data were retrieved against the Uniprot annotated database resulting in 4.218 non redundant sequences. The sequence of Bacillus cereus MalT was added for comparison. Protein sequences alignment was performed using MAFFT FFT-NS-2 method (Katoh et al., 2019). Maximal likelihood protein trees were constructed using the software tool RAxML-HPC v.8 (Stamatakis, 2014). Tree display was done using the software tool iTOL v3 (Letunic and Bork, 2016).

ACK N OWLED G M ENTS
This work was funded in part by a fellowship awarded to B.P. by the Brazilian Exchange Program "Science Without Borders". We thank the Deutsche Forschungsgemeinschaft for grants TRR174 (to K.J. and M.B.) and GRK2062 (to K.J.) to financially support this work. We are grateful to Dr. Oliver Goldbeck and Dr. Gerd Seibold who provided us with the strain C. glutamicum ΔptsG ΔiolT1 ΔiolT2. We thank Manuela Grafemeyer for providing strains C. glutamicum/pEKEX2