Discovery of Unannotated Small Open Reading Frames in Streptococcus pneumoniae D39 Involved in Quorum Sensing and Virulence Using Ribosome Profiling

A recent study used the pleuromutilin antibiotic retapamulin in combination with ribosome profiling (Ribo-RET) to identify the start codons of protein-coding genes in the E. coli genome and discovered 41 novel sORFs in E. coli (11). Retapamulin, a protein synthesis inhibitor, binds to the 50S subunit during ribosome assembly and traps the ribosome during translation initiation, providing an increased signal-to-noise ratio at translation start sites.

We applied Ribo-RET to identify translated sORFs in the S. pneumoniae D39 genome. A D39 capsule mutant was used for all Ribo-seq experiments, as the presence of capsule complicated the rapid isolation of cells and, hence, ribosomes by filtration or centrifugation for downstream processing. Prior to cell lysis, mid-exponential-phase cultures were treated with 62.5 ng mL⁻¹ retapamulin for 2.5 min, which corresponds to 100 times the MIC of the drug (see Fig. S1A in the supplemental material). The metabolic labeling experiments established that a 2.5-min treatment with retapamulin is sufficient to completely stop translation in the cell (Fig. S1B). The polysomes were isolated from the cells using the procedures described in Materials and Methods. However, we found that S. pneumoniae ribosomes tended to dissociate into subunits under conventional conditions of sucrose gradient centrifugation. To preserve ribosome integrity, we increased the concentration of MgCl₂ in the lysis buffer from 10 mM to 50 mM. This adjustment significantly stabilized the ribosomes and slightly diminished the activity of micrococcal nuclease (MNase) used for the preparation of the ribosome footprints (Fig. S1C). This resulted in a less-precise trimming of the footprints to the ribosome edge and a broader distribution of the footprint lengths. Notably, even after digestion with MNase, all sucrose gradient centrifugation profiles showed the presence of an additional peak whose sedimentation properties are consistent with either underdigested disomes or, as observed in S. aureus, hibernating ribosome pairs (Fig. S1D, E, G, and H) (30). Despite these potential complications, monomeric 70S:mRNA footprint complexes were isolated, and the Ribo-seq library was prepared for Illumina deep sequencing.

Ribo-seq data sets from untreated cultures revealed the translation of many annotated S. pneumoniae genes as well as the presence of ribosome footprints in some intergenic regions (Fig. 1A). As predicted, treatment of cells with retapamulin led to the accumulation of ribosomes at the start codons of genes (Fig. 1A). Metagene analysis showed that many ribosome footprints mapped to annotated start codons (Fig. S2B); however, a sizeable fraction of reads placed the ribosome as far as 20 nucleotides (nt) upstream from the annotated translation start sites (Fig. 1B and C). We observed in the raw sequencing data that the distribution of ribosome footprint read lengths ranged from 15 to 35 nucleotides, a surprise considering that studies performed on E. coli typically produce lengths of 28 nucleotides (Fig. 1D) (31). To determine whether the footprint size correlated with the ribosome’s location, read lengths were compared to start codon positioning, and it was found that the reads (27 to 35 nt) aligned best with annotated gene start sites (Fig. S2A). When only reads in this size range were used for mapping, two-thirds of the reads were situated at annotated start codons (Fig. 1C). Finally, to test the possibility that ribosome positioning was dependent on the initiation inhibitor used, we repeated the profiling experiment using lefamulin (Ribo-LEF), another pleuromutilin antibiotic reported to bind tightly to the ribosomal peptidyl transferase center (PTC) (32). The sequencing data sets generated from retapamulin- and lefamulin-treated samples produced outcomes that were nearly identical, reinforcing our confidence in the results of our ribosome profiling (Fig. S1E and H and Fig. S7).

Similar to Ribo-RET results obtained with E. coli (29), we also identified peaks of ribosomal footprints at putative internal start sites located within annotated ORFs. Such peaks may indicate instances of alternative initiation of translation or nested ORFs, as observed previously in E. coli (29) (Fig. S3).

Overall, the ribosome profiling results indicated that 85% of the annotated S. pneumoniae genome is actively translated under laboratory conditions. As such, antibiotic-assisted ribosome profiling provides a powerful tool to identify the S. pneumoniae translatome.

The S. pneumoniae D39 genome encodes ~2,700 intergenic coding sequences with the potential to encode small proteins 10 to 50 amino acids long. To identify true protein-coding regions, three independent sequencing data sets, two utilizing Ribo-RET and one utilizing Ribo-LEF, were used to map the translation initiation sites. Using a computational approach that mapped and quantified ribosome footprints, normalized to genome-wide sequence reads, we used the following criteria to define putative translation start sites of sORF candidates: at least 1 sequence read per million (rpm) mapped within 10 nucleotides of a theoretical sORF start codon (AUG, GUG, CUG, or UUG), and the respective full-length sORF did not overlap an annotated gene. By these criteria, we identified 117 (RET) and 103 (LEF) sORF candidates. In some instances of neighboring putative start codons, manual assessment of the coding region resulted in a refined list of 114 novel sORF candidates, designated rio (Ribo-seq-identified ORFs) (Tables 1 and 2).

The identified sORFs range in length from 5 to 43 amino acids, with the majority having an AUG start codon (Fig. 2A and B). Using tBLASTn analysis (33), we investigated the conservation of the sORFs among six clinically relevant S. pneumoniae serotypes (1, 4, 14, 19A, 23F, and 19F). We found that 49 sORFs were conserved in all six serotypes, with the remaining sORFs being conserved in at least one of the serotypes except for rio34, rio72, and rio89 (see Table 3). We also identified 10 sORFs encoding proteins containing putative signal peptides as determined by SignalP analysis (34) suggested their insertion into or translocation across the cytoplasmic membrane (Table 1). One of these, rio84, was found adjacent to an Rgg family member gene, and we hypothesized that it encodes the signaling peptide for a pheromone receptor QS system (see below). Nine sORFs are located within or overlap the previously annotated noncoding RNAs (ncRNAs), two of which were previously associated with fitness defects determined by transposon-insertion sequencing (TIS) (Tables 1 and 2) (1, 3).

To validate the Ribo-RET results and demonstrate sORF translation, 6 sORFs displaying the highest number of read counts at start codons (rio48, rio49, rio83, rio85, rio97, and rio106) and 4 sORFs located within documented ncRNAs (rio01, rio3, rio82, and rio56) were selected to be tagged with translation reporters (Table 1). A sequence encoding superfolder green fluorescent protein (sfGFP) (lacking its own start codon) was placed at the 3′ end of each selected sORF at its native chromosomal locus to generate in-frame translational fusions. If translated, the addition of sfGFP should increase the molecular weight of each sORF peptide by ~27 kDa. Cells containing the tagged constructs were cultured to mid-log phase in chemically defined medium (CDM) to mimic the conditions used in ribosome profiling experiments, and the expressed proteins were evaluated by Western blotting using an anti-GFP antibody. Of the 10 sfGFP-tagged constructs, 5 produced a strong band with the expected mobility on an SDS gel, verifying their translation (Fig. 2C). To demonstrate that sfGFP was not independently translated when placed in frame with sORFs, the start codon of the sORF rio3 fused to sfGFP was mutated from ATG to GGG. The production of the fusion protein was completely abolished, demonstrating not only the translation of the identified sORFs but also the accuracy of mapping its start codon by Ribo-RET/Ribo-LEF.

Unexpectedly, rio48::sfGFP, located immediately upstream of the gene encoding peptide release factor 2 (RF2), prfB, produced a strong band of ~70 kDa. In E. coli, the expression of RF2 is autoregulated by programmed frameshifting; RF2 deficiency stimulates a +1 frameshift resulting in the readthrough of the in-frame UGA stop codon and the translation of the full-size functional RF2 protein (35). In E. coli, previous studies have demonstrated that the frameshift mechanism exploits several key features of the prfB mRNA: a Shine-Dalgarno (SD) sequence 3 nucleotides upstream of the frameshift site (AGG GGG), the frameshift site (CUU UGA), and the context of the UGA stop codon flanked with a 3′ C (Fig. S4) (36–38). The short distance between the SD sequence and the frameshifting site creates tension destabilizing the interactions between the P-site and the anticodon of the ribosome, resulting in a +1 frameshift. Furthermore, the genetic context of the UGA stop codon in proximity to a C nucleotide has been demonstrated to be the least efficient termination signal (37–39). These key mRNA features are also conserved in rio48, suggesting that prfB in S. pneumoniae is regulated in a similar manner. Likewise, a +1 frameshift at the UGA stop codon of rio48 is in frame with downstream prfB, and therefore, programmed frameshifting during Rio48 translation could stimulate the expression of RF2. The rio48::sfGFP construct retains the UGA stop codon of rio48 after sfGFP and likely results in readthrough and the generation of the larger gene product corresponding to ~70 kDa seen on the immunoblot. Thus, in this instance, Ribo-RET likely identified the correct translation start site for prfB.

Previous studies identified and characterized RRNPP transcriptional regulators in streptococci and demonstrated their importance in regulating genes associated with virulence, immunosuppression, lysozyme resistance, and competence (40–43). Ribo-RET detected the presence of two sORFs encoding polypeptides of 33 amino acids (rio84) and 7 amino acids (rio83) in length that are adjacent to an Rgg-like transcriptional regulator (spv_1518, referred to as rgg1518 here) (Fig. 3A). The peptide encoded by rio84 has characteristics resembling those of other streptococcal pheromones, such as a positively charged N terminus and a Trp-X-Trp (WXW) motif at the C terminus (44), leading us to hypothesize that rio84 may encode the pheromone for Rgg1518 (Fig. 3A). To verify that Rgg1518 functions as a transcriptional regulator and to identify the genes under its regulation, transcriptome sequencing (RNA-seq) analysis was conducted to compare the gene expression of wild-type D39 to that of an isogenic deletion mutant, Δrgg1518 (Fig. 3B). The expression of the spv_1513-1517 operon located adjacent to rgg1518 and immediately downstream from rio83 was substantially decreased in the deletion mutant, a trend that we verified by quantitative real-time PCR (qRT-PCR) (Fig. 3C). The operon of genes spv_1513 to spv_1517 (hereafter spv_1513-1517) encodes proteins predicted to comprise an ABC transporter of an unknown substrate(s), suggesting that Rgg1518 could be a regulator of nutrient acquisition. A previous report found that the spv_1513-1517 operon was significantly upregulated when wild-type D39 bacteria were applied to A549 lung epithelial cells, suggesting a role during interactions with the host (25). We tested the impact that deleting the operon would have on adherence to or invasion of A549 cells but found no difference in attachment, internalization, or viability from the wild type, at least over short infection times (up to 4 h) (Fig. S5A and B). An independent recent report demonstrated that the presence of intact Rgg1518 is important for colonization of the murine nasopharynx by S. pneumoniae (23). To assess whether the spv_1513-1517 operon is responsible for this phenotype, we coinfected CD1 mice with 10⁵ CFU of wild-type D39 and 10⁵ CFU of the Δspv_1513-1517 mutant in the nasopharynx and determined the bacterial burden in the nasal passage over a span of 7 days. The Δspv_1513-1517 mutant decreased over time in comparison with wild-type S. pneumoniae; however, the difference was not statistically significant, suggesting that the conditions in our experiment were not conducive to show whether this operon plays a critical role in colonizing the murine nasal passage (Fig. S5C).

To assist in evaluating the potential contributions of Rgg1518, Rio83, and Rio84 to mediating cell-to-cell signaling, we constructed a luciferase-based transcription reporter using the promoter (P₁₅₁₇) identified by 5′ rapid amplification of cDNA ends (RACE) upstream of rio84 (Fig. 4A). The promoter-reporter construct was placed into an unlinked, neutral location in the chromosome of isogenic strains with deletions of rgg1518 or a combined deletion of its affiliated sORFs rio83 and rio84 (45). During growth in CDM, the wild-type reporter strain produced strong luminescence as the culture density increased, whereas the luminescence of the isogenic Δrgg1518 and Δrio83 Δrio84 mutants remained at low levels throughout the cultures’ growth (Fig. 4B; Fig. S6A). The expression of rio84 from a constitutive promoter (P_c-rio84) in the Δrio83 Δrio84 mutant background led to enhanced luciferase activity (Fig. 4B, yellow curve; Fig. S6A), indicating that the expression of rio84 in trans was sufficient to complement the Δrio83 Δrio84 mutant. These results support a model in which rio84 encodes a functional pheromone for Rgg1518, consistent with the results of a recent independent study (23). To identify the mature form of the pheromone, synthetic peptides of various lengths encompassing the C-terminal region of rio84 (C6, C8, and C12) were added to cultures. While the 6- and 8-amino-acid-long peptides were unable to stimulate transcription from the P₁₅₁₇ promoter, the C12 variant (IQLIWFETWFWG) efficiently induced the expression of P₁₅₁₇ in the wild-type or Δrio83 Δrio84 strain but not in the Δrgg1518 strain (Fig. 4C; Fig. S6B). Thus, the active form of the rio84 pheromone is likely confined within or is equivalent to this sequence.

RNA-seq results indicated that rio83 is downregulated in the absence of Rgg1518, suggesting that rio83 might be involved in the regulation of Rgg1518-based QS regulation. The translation of the rio83 sORF was validated by fusing it to sfGFP and was detected by Western blotting when cultures were treated with exogenous pheromone (Fig. 2C). The addition of the full-length synthetic rio83 peptide to cultures did not alter luciferase activity (Fig. 5A; Fig. S6C). Intriguingly, the reporter activity in a Δrio84 mutant, grown in the presence of C12, did not reach the level of luciferase activity seen in the wild-type or Δrio83 Δrio84 strain (Fig. 5B; Fig. S6D). Furthermore, complementing the Δrio83 Δrio84 strain with rio83 resulted in a complete loss of luminescence activity. These results suggest that rio83 serves as a negative regulator. However, the extent of its impact on the control of the putative ABC transporter (spv_1513-1517) remains unclear.

The Ribo-RET data set also identified the known signaling peptide (rio9) for the Rgg0112 transcriptional regulator (44) as well as additional sORFs found downstream of Rgg0112 (rio7, rio8, and rio10-15) (Table 1), which appear to be part of the Rgg0112 regulon based on our RNA-seq data comparing wild-type D39 to an rgg0112 mutant. Manual assessment of the Ribo-RET data set near other known Rgg-like transcriptional regulators identified sORFs that did not meet our initial search criteria (Table S3). rio119, found within the current annotation of srf-06 and partially overlapping Rgg144, encodes the previously characterized pheromone for Rgg0144 (21, 22). Additional sORFs (rio120, rio121, and rio122) were identified within the same locus, downstream of the Rgg0144 pheromone and overlapping the transcriptional regulator on the opposite strand. To date, the roles that these additional sORFs may play in the QS systems are unknown.

A previous TIS study identified a noncoding RNA, F52, in S. pneumoniae TIGR4 whose disruption negatively impacted the fitness of the pathogen in a mouse model of pneumonia (1). The S. pneumoniae reference strain D39 contains an ortholog of this ncRNA, which is referred to as srf-02 (3). One of the sORFs identified and confirmed in our experiments (Fig. 2C), rio3, overlaps the annotated ncRNA srf-02 (Fig. 6A). Given this overlap, we wondered if the fitness defect described in the TIS study might be attributable to a disruption of rio3 rather than the ncRNA. To test this hypothesis, the start codon of rio3 was mutated (ATG→GGG) to prevent the translation of the sORF. Separately, a deletion was generated (Δsrf-02^18–207) that extended through the srf-02 gene, which removed 188 3′-terminal nucleotides of the ncRNA while ensuring that the rio3 sORF remained intact. Neither mutant strain displayed a growth defect compared to the wild type in culture (Fig. 6B). In order to assess whether rio3 has an impact on nasopharyngeal colonization, 6-week-old BALB/c mice were infected intranasally with the wild type, the rio03^GGG mutant, the Δsrf-02^18–207 mutant, or the rio03^GGG complemented strain. The bacterial burdens in the nasal passage were enumerated at 24 h postinfection. Minimal differences were seen between the abilities of the wild-type, the Δsrf-02^18–207 mutant, and the rio03^GGG complemented strains to colonize the nasopharynx; however, the rio03^GGG mutant displayed a significant defect in colonizing the murine nasopharynx (Fig. 6C). These data indicate that the fitness defects attributed to srf-02 on the basis of the TIS experiments are instead related to disruption of the sORF rio3 identified in the S. pneumoniae genome by our Ribo-RET/Ribo-LEF approach.

All strains and plasmids used in this study are listed in Table S1 in the supplemental material. S. pneumoniae D39 was routinely grown on tryptic soy agar (TSA) supplemented with 5% sheep blood or cultured in Todd-Hewitt broth with 0.2% yeast (THY) and 0.5% Oxyrase (catalog number OB-0100; Oxyrase) or in a chemically defined medium (CDM) (50) supplemented with 1% glucose, 10% choline, and 0.5% Oxyrase at 37°C in an atmosphere of 5% CO₂. When appropriate, chloramphenicol (4 μg/mL), spectinomycin (150 μg/mL), kanamycin (200 μg/mL), erythromycin (0.3 μg/mL), or neomycin (20 μg/mL) was added to S. pneumoniae D39 cultures.

To generate competent S. pneumoniae D39 cells, wild-type D39 cells were grown in 7.5 mL THY supplemented with 0.013 N HCl and 0.05% glycine at 37°C in an atmosphere of 5% CO₂ to an optical density at 600 nm (OD₆₀₀) of 0.05 to 0.1. Cells were diluted into 1 mL THY to an OD₆₀₀ of 0.03; supplemented with a solution containing 10 mM NaOH, 0.2% bovine serum albumin (BSA), 1 mM CaCl₂, and 0.2 μg/mL competence-stimulating peptide 1 (CSP-1); and placed in a 37°C water bath for 14 min. Following incubation, ~850 ng of donor DNA was added, and cells were allowed to recover at 37°C with 5% CO₂ for 1 h, followed by plating onto TSA plates supplemented with 5% sheep blood and the appropriate antibiotic.

All S. pneumoniae D39 deletion mutants, listed in Table S1, were generated by transforming competent S. pneumoniae D39 cells with linear DNA containing upstream and downstream sequences that facilitate homologous recombination and were generated by Gibson assembly of PCR amplicons using the primers listed in Table S2. All strains were confirmed by sequencing the locations of the chromosome containing the relevant alterations. Specific constructs are described further here. To delete rgg1518 (strain IL20), a PCR-generated upstream flanking region (UFR) amplicon and a downstream flanking region (DFR) amplicon were joined with a chloramphenicol resistance cassette by Gibson assembly using NEBuilder HiFi DNA assembly master mix (New England BioLabs [NEB]). Strain IL40 (Δrio83 Δrio84::spec) was constructed by Gibson assembly using a spectinomycin resistance cassette. Strain IL108 contains a deletion of the noncoding RNA srf-02 (Δsrf-02^18–207::erm) without disrupting the overlapping sORF rio3; the UFR encompasses the first 17 nucleotides of srf-02. To generate the missense point mutations in strains IL91 (rio03^ATG-GGG-spec) and IL101 (rio83^ATG-GGG-spec), special oligonucleotides were designed to replace the start codon ATG with the glycine codon GGG. To generate strain IL91, two DNA fragments were generated using primer pairs ILp355/ILp356 and ILp354/KTp043, and overlapping PCR was performed to generate a PCR amplicon with the start codon mutation in rio3, which was subsequently used as the template to amplify the UFR for the construct. To generate strain IL101, overlapping PCR was performed as described above, using primer pairs ILp170/ILp161 and ILp169/ILp166.

To assess the expression levels of spv_1517, the intergenic region between spv_1517 and spv_1518 was amplified using a DNA template containing start codon mutations (GGG in place of ATG) in both rio83 and rio84. To attain the DNA amplicon containing the missense mutations, overlapping PCR was performed using primer pairs ILp166/ILp167 and ILp168/ILp161 for rio84 and primer pairs ILp170/ILp161 and ILp169/ILp166 for rio83. Overlapping PCR combined the two mutations on one DNA amplicon. The resulting linear piece of DNA was then used as a template to amplify the promoter region for spv_1517 using primer pair ILp161/ILp166. Using Gibson assembly, the upstream region of the bga locus was fused to the promoter fragment and linked to luxAB of pJC156, followed by P_c and the kanamycin resistance cassette from CP1296 and flanked downstream by 2,000 bp of the bgaA gene. The resulting reporter construct was transformed into wild-type D39, IL20 (Δrgg1518), IL40 (Δrio83 Δrio84::spec), and IL101 (rio83^ATG-GGG-spec).

To generate strain IL93 (Δrio83 Δrio84::spec bgaA::P₁₅₁₇^{rio83-GGG,rio84-GGG}-luxAB-P_c-kan-P_c-rio84), the luciferase reporter constitutively expressing rio84 driven by the P_c promoter and genomic DNA from strain IL81 (bgaA::P₁₅₁₇^{rio83-GGG,rio84-GGG}-luxAB-P_c-kan) were used as the template to amplify the reporter construct using primer pair ILp387/ILp264, which was then linked to the constitutive promoter P_c and fused to rio84 using Gibson assembly. This construct was transformed into wild-type D39 and IL40 (Δrio83 Δrio84::spec).

Start codon mutations in rio83 and rio84 were restored by transforming strains IL40 (Δrio83 Δrio84::spec) and IL101 (rio83^ATG-GGG-spec) with DNA fragments containing wild-type sequences, generating strains IL52 (Δrio83 Δrio84::spec bgaA::P₁₅₁₇^{rio83-ATG,rio84-ATG}-luxAB-P_c-kan), IL106 (Δrio83 Δrio84::spec bgaA::P₁₅₁₇^{rio83-GGG,rio84-ATG}-luxAB-P_c-kan), and IL113 (rio83^ATG-GGG-spec bgaA::P₁₅₁₇^{rio83-GGG,rio84-ATG}-luxAB-P_c-kan).

To generate the chromosomal superfolder GFP (sfGFP)-tagged constructs, we performed transformations using linear DNA amplicons as described above. Each construct fused sfGFP in frame in front of the stop codon, followed by a spectinomycin resistance cassette, and was flanked by UFR and DFR homologous sequences. Strain IL75 (D39 rio03^ATG-GGG-sfGFP) was constructed using strain IL91 (rio03^ATG-GGG-spec) as a template to amplify the missense mutation with primer pair ILp323/ILp324.

The sfGFP-tagged strains were grown in 10 mL CDM to an OD₆₀₀ of 0.4, and cells were collected at 4,000 × g for 10 min. Cell pellets were resuspended in 250 μL loading buffer (0.0625 M Tris [pH 8], 2% SDS, 10% glycerol, 5% 2-mercaptoethanol, 50 mg bromophenol blue) and lysed using a BioSpec bead beater for 10 min at maximum speed. Gel loading volumes of each sample were normalized by culture OD readings and resolved on a 12% SDS-PAGE gel at 150 V for 1.5 h. Gels were blotted onto 0.2-μm polyvinylidene difluoride (PVDF) membranes at 350 mA for 1.5 h, and the membranes were blocked overnight at 4°C with rocking in Tris-buffered saline plus 0.1% Tween (TBST) containing 5% BSA. Membranes were subsequently incubated for 1 h at room temperature, with rocking, with anti-sfGFP antibody (catalog number AE011; ABclonal) at a dilution of 1:3,000 in TBST plus 5% BSA. The membranes were then washed three times in TBST, followed by the addition of goat anti-rabbit IgG(H+L) (Thermo Fisher) at a dilution of 1:80,000 in TBST plus 5% BSA for 1 h with rocking at room temperature. The membranes were then washed three times, and sfGFP-tagged proteins were detected using the SuperSignal West Femto maximum-sensitivity substrate (catalog number 34094; Thermo Fisher). To prepare the working solution, equal volumes of the stable peroxide solution and the luminol-enhancer solution were mixed and incubated with the blot for 5 min, followed by exposure on a ProteinSimple FluorChem imaging system.

Synthetic peptides were purchased from ABclonal. All peptides were reconstituted in dimethyl sulfoxide (DMSO) at a final concentration of 10 mM and stored at −80°C. Peptide purity ranged from 50 to 80%.

Dilutions of the antibiotics retapamulin and lefamulin were prepared in CDM and loaded into a 96-well microtiter plate. D39 Δcps was grown in CDM to an OD₆₀₀ of 0.5 and diluted 10-fold to an OD₆₀₀ of 0.05 into the antibiotic-containing medium. Plates were incubated at 37°C in a microplate reader (Synergy 2; BioTek), and the OD was measured every 15 min over a span of 10 h.

Inhibition of translation by retapamulin and lefamulin was determined using metabolic labeling. All manipulations were performed at 37°C. D39 Δcps was inoculated from a starter culture (OD₆₀₀ of 1) into 6 mL and grown in CDM lacking methionine and containing 0.5% Oxyrase to an OD₆₀₀ of 0.5 at 37°C with 5% CO₂. Cells were diluted 10-fold into CDM without methionine and containing 0.5% Oxyrase and grown until the culture density reached an OD₆₀₀ of ~0.2, and three 350-μL aliquots were transferred to microcentrifuge tubes (two drug conditions and one control group). Retapamulin and lefamulin were individually added to Eppendorf tubes at a final concentration of 100× MIC. Prior to and immediately following the addition of antibiotics (0, 1, 2.5, 5, and 15 min), 28 μL of the culture was added to microcentrifuge tubes containing 0.3 μCi [³⁵S]l-methionine (specific activity of 1,175 Ci/mmol; MP Biomedicals) in 2 μL of CDM. After a 1-min incubation, 25 μL of the mixture was spotted onto Whatman 3MM paper discs prewetted with 7% trichloroacetic acid (TCA). The discs were boiled twice in 7% TCA for 5 min, soaked in 100% acetone for 2 min, and then air dried prior to being placed into a 5-mL scintillation cocktail and being read using a scintillation counter.

Ribosome profiling was conducted as previously described, with the following modifications (29, 61). D39 Δcps cells were grown to an OD₆₀₀ of 0.4 in 100 mL CDM supplemented with 0.5% Oxyrase at 37°C in an atmosphere of 5% CO₂. Retapamulin or lefamulin was added to individual 100-mL cultures at final concentrations of 100× MIC for 2.5 min. No antibiotic was added to the untreated control group. After 2.5 min, bacteria were harvested by centrifugation at 6,300 × g at 37°C for 4 min and flash-frozen in liquid nitrogen. Cells were cryo-lysed in 650 μL lysis buffer (20 mM Tris-HCl [pH 8.0], 50 mM MgCl₂, 100 mM NH₄Cl,5 mM CaCl₂) supplemented with 65 U RNase-free DNase I (catalog number 04716728001; Roche), 208 U SUPERase In RNase inhibitor (catalog number AM2694; Invitrogen), and 3 mM Guanosine 5′ –[b,g-imido] triphosphate trisodium salt hydrate (GMPPNP; catalog number G0635; Sigma-Aldrich). Pulverized cells were thawed at 30°C and spun at 20,000 × g for 10 min at 4°C to pellet insoluble debris. Clarified lysates were subjected to treatment with 450 U MNase (catalog number 10107921001; Roche), 120 U SUPERase In RNase inhibitor was added to the clarified lysates, and the reaction mixtures were incubated for 1 h at 25°C with shaking. The reaction mixtures were quenched with 5 mM EGTA, and the 70S monosome peak was isolated by sucrose gradient centrifugation (10 to 40% sucrose gradient) for 2 h 45 min at 39,000 × g. RNA was isolated by acid-phenol extraction and run on a 15% Tris-borate-EDTA (TBE)-urea polyacrylamide gel. RNA fragments ranging from 20 nucleotides to 38 nucleotides were excised, eluted, and used for library preparation as previously described, which included the addition of barcodes for multiplexing (31).

The ribosome footprint reads were analyzed as described previously (61). In brief, samples were demultiplexed, linker barcodes were removed, and 5 nucleotides from the 3′ end and 2 nucleotides from the 5′ end were removed as they were included in the library design (29, 31). The reads were aligned to the S. pneumoniae D39V (GenBank accession number CP027540.1) reference genome by Bowtie2 (v2.2.9) after discarding reads mapping to known tRNAs and rRNAs. Read lengths ranging from 28 to 34 nucleotides were included for the analysis; the first nucleotide of the P-site codon was assigned 15 nucleotides from the 3′ end of the read, as previously suggested (11).

Novel sORFs found within intergenic regions were identified based on the following criteria: a Ribo-RET peak of at least 1 sequence read per million (rpm) that mapped within 10 nucleotides of a theoretical sORF starting with AUG, GUG, CUG, or UUG and whose respective full-length sORF did not overlap an annotated gene. In some instances, multiple start codons were identified in the 10-nucleotide window; therefore, a manual approach was used to inspect each candidate relative to the Ribo-RET peak. The list of sORFs identified can be found in Table 1. The code used to analyze the data set can be found at https://github.com/ilaczk2/D39_ribosome_profile_MS.

Metagene analyses, to evaluate the positions of ribosomes at annotated genes with respect to the 5′ (start) and 3′ (stop) ends of genes, were performed according to a previously described protocol (62). Genes included in the analysis satisfied two criteria: a length of at least 200 nt and a read density of at least 0.005 rpm per nucleotide in all 5 samples (2 retapamulin, 1 lefamulin, and 2 controls). Coverage at each nucleotide position within a gene was normalized to the coverage density of the entire gene plus 50 nt of the flanking up- and downstream regions. The mean of these values was calculated and plotted for the windows around the start and stop codons.

Strains of interest were inoculated from flash-frozen starter cultures in CDM plus 0.5% Oxyrase at 37°C in an atmosphere of 5% CO₂ and reached exponential growth to an OD of ~0.5. Strains were then diluted 10-fold in CDM in a total volume of 150 μL in a 96-well white/clear-bottom plate (Sigma). When relevant, synthetic peptides were added to the wells at a final concentration of 10 μM. Dosing assays determined 10 μM to be the optimal concentration to induce the system. Decyl aldehyde (Sigma) was added to the spaces between the wells at a final concentration of 1% in mineral oil. The plate was covered and placed into the microplate reader (Synergy 2; BioTek) at 37°C with intermittent shaking. The luminescence (counts per second [CPS]) and optical density (OD₆₀₀) were measured every 15 min over a span of 10 h. Relative light units (RLU) were calculated by normalizing the CPS to the OD₆₀₀. Each assay was conducted in technical triplicates, and each figure shows results representative of data from at least three independent experiments (Fig. S6).

A549 lung epithelial cells (ATCC) were routinely cultured in Dulbecco’s modified Eagle’s medium (DMEM) plus 10% fetal bovine serum (FBS) without antibiotics at 37°C in an atmosphere of 5% CO₂. For the adhesion assay, A549 cells were seeded into a 24-well plate at 2 × 10⁵ cells/well. Following incubation overnight, each well was washed once with 1× phosphate-buffered saline (PBS). D39 Δcps and D39 Δcps Δspv_1513-1517::spec were grown in CDM plus 10% choline and 0.5% Oxyrase to an OD₆₀₀ of 0.5, washed in DMEM, and added to the cells at a multiplicity of infection (MOI) of 100:1 for 1 h at 37°C with 5% CO₂. Following incubation, cells were gently washed three times with 1× PBS to remove unbound bacteria, treated with 0.025% trypsin for 6 min at 37°C with 5% CO₂ to detach the cells, lysed with 0.1% saponin, and plated onto blood agar plates to determine bacterial CFU.

The initial steps for the intracellular survival assay were the same as the ones described above for the adhesion assay. To differentiate between internalized and external bacteria, epithelial cells were treated with 100 μg/mL gentamicin for 1, 3, and 4 h to kill the external bacteria. Following antibiotic treatment, the supernatant was aspirated, and cells were gently washed three times with 1× PBS. Cells were removed from the wells by the addition of 0.025% trypsin for 6 min and lysed with 0.1% saponin. The suspension was serially diluted and plated onto sheep blood agar plates to determine the bacterial burden.

Mice were housed at a biosafety level 2 facility and anesthetized with inhaled isoflurane (3%) when necessary. As shown in Fig. 6C, 6-week-old BALB/c mice were intranasally inoculated with wild-type D39, IL91 (rio03^ATG-GGG-spec), IL127 (rio03^ATG-GGG;rio03^GGG-ATG-kan), and IL108 (Δsrf-02^18–207::erm) at a dose of 1 × 10⁷ CFU/25 μL. A minimum of 10 mice were used for each bacterial inoculation. Mice were sacrificed 24 h after inoculation using carbon dioxide inhalation followed by cervical dislocation. The nasal passage of each mouse was isolated, homogenized in 500 μL 1× PBS, and plated onto blood agar plates to determine the bacterial CFU burden. As shown in Fig. S6C, 30 6-week-old CD1 mice were intranasally inoculated with bacterial suspensions containing 1:1 mixtures of wild-type D39 and IL97 (Δspv_1513-1517::spec) at a dose of 2 × 10⁵ CFU/20 μL. Ten mice were sacrificed either 24 h, 72 h, or 168 h after inoculation, followed by nasal passage isolation and plating onto blood agar plates to determine bacterial CFU.

The wild-type D39 and IL20 (Δrgg1518) strains were cultured in CDM supplemented with 10% choline plus 0.5% Oxyrase and grown to an OD₆₀₀ of 0.4 at 37°C with 5% CO₂. Three independent cultures of each strain were prepared. Cultures were harvested by centrifugation, supernatants were discarded, and cell pellets were suspended in 1 mL RNAlater (Ambion) and incubated at room temperature for 10 min. Following incubation, samples were centrifuged at 14,000 × g for 1 min, supernatants were discarded, and cell pellets were stored at −80°C. Total RNAs from wild-type D39, IL20 (Δrgg1518), and the Ribo-seq samples (retapamulin treated, lefamulin treated, and untreated) were extracted using the Ambion RiboPure RNA purification bacterial kit according to the manufacturer’s instructions and as previously described (63). Following the successful extraction of RNA, the Genome Research Core at the University of Illinois at Chicago (UIC) assessed RNA quality and quantity using the Tapestation 2200 system (Agilent), prepared the cDNA libraries, and processed samples on an Illumina HiSeq 4000 platform with 100-bp single reads. The raw sequencing data were analyzed by the Research Informatics Core at UIC.

cDNA was prepared from RNA using the Superscript III first-strand synthesis system (Thermo Fisher) according to the manufacturer’s instructions and as previously described (63). Total cDNA was diluted 1:10, and reaction mixtures were prepared using 1× Fast SYBR green master mix with the gene-specific primers listed in Table S2. qRT-PCR was performed using the CFX Connect real-time PCR detection system (Bio-Rad). All samples were run in biological and technical triplicates, and relative gene expression was determined using the 2^−ΔΔCT method.

5′ RACE was conducted as previously described (63). Total RNA was isolated as detailed in the section on RNA isolation and RNA sequencing above. cDNA synthesis of the spv_1517 transcript and template switching were performed using the NEB template-switching RT enzyme mix with primers specific for the spv_1517 operon (ILp151) and the template-switching oligonucleotide (TSO) (BRp311). The 5′ end of the spv_1517 transcript was amplified using IL151 primers and the TSO-specific primer BRp312 using the Q5 high-fidelity enzyme. The resulting PCR product was sequenced by Sanger sequencing.

sORF conservation was assessed as previously described (64). tBLASTn analysis was used to assess sORF conservation in six clinically relevant S. pneumoniae serotypes (1, 4, 14, 19A, 23F, and 19F). The amino acid sequence of each sORF was submitted to tBLASTn analysis. The following parameters were modified: the maximum number of target sequences was 250, the expected threshold was set to 100, and no low-complexity filter was used. The search was refined by selecting only sORFs that had 100% query coverage and ≥70% sequence identity. The sORFs highlighted in boldface type in Table 3 were too short for tBLASTn analysis; therefore, BLASTn was used to assess their conservation using the same parameters as the ones described above.

All raw ribosome profiling reads, RNA sequencing reads, and annotation files are available in the NCBI GEO database (https://www.ncbi.nlm.nih.gov/bioproject/857299; Accession #PRJNA857299). Analysis scripts are available on GitHub (https://github.com/ilaczk2/D39_ribosome_profile_MS).