Revisiting the regulation of the capsular polysaccharide biosynthesis gene cluster in Staphylococcus aureus

In Staphylococcus aureus, the capsular polysaccharide (CP) protects against phagocytosis, but also hinders adherence to endothelial cells and matrix proteins. Its biosynthesis is tightly controlled resulting in a heterogeneous phenotype within a population and CP being mainly detectable in non-growing cells. Capsular biosynthesis genes are encoded by a conserved capA-P operon whose expression is driven by an upstream promoter element (Pcap) in front of capA. The organization of Pcap is poorly understood, as is the interplay of different regulators that influence the early-Off/late-Heterogeneous cap transcription pattern. Here, we demonstrate that Pcap contains a main SigB-dependent promoter. The SigB consensus motif overlaps with a previously described inverted repeat that is crucial for cap expression. The essentiality of the inverted repeat is derived from this region acting as a SigB binding site rather than as an operator site for the proposed cap activators RbsR and MsaB. Furthermore, Pcap contains an extensive upstream region harboring a weak SigA-dependent promoter and binding sites for the cap repressors SaeR, CodY and Rot. We show that heterogeneous CP synthesis is determined by the combination of SigB activity and repressor binding to the upstream region. The direct SigB dependency and the upstream repressors are also sufficient to explain the temporal gene expression pattern at the transcriptional level. However, CP synthesis remains growth phase-dependent even when capA transcription is rendered constitutive, suggesting additional post-transcriptional regulatory circuits. Thus, the interference of multiple repressors with SigB-dependent promoter activity as well as post-transcriptional mechanisms ensure the appropriate regulation of CP synthesis. Importance The majority of bacterial pathogens produce an array of polysaccharides on their surface which are important virulence factors and thus serve as attractive vaccine candidates. However, the synthesis and assembly of these structures is highly variable and tightly regulated at various levels. In the human pathogen Staphylococcus aureus, the synthesis of the capsular polysaccharide (CP) is dependent on a complex regulatory network which ensures that CP is produced only in a fraction of stationary phase cells. Here, we determined main regulators that drive the peculiar CP expression pattern. We found that the interplay of the transcriptional repressors Sae, CodY and Rot with the alternative Sigma factor B is responsible for early-Off/late-Heterogeneous expression at the transcriptional level. The data also implicates post-transcriptional mechanisms that may act to avoid conflict in precursor usage by machineries involved in either synthesis of CP or other glycopolymers in growing bacterial cells.

P cap is poorly understood, as is the interplay of different regulators that influence the 23 early-Off/late-Heterogeneous cap transcription pattern. Here, we demonstrate that 24 P cap contains a main SigB-dependent promoter. The SigB consensus motif overlaps 25 with a previously described inverted repeat that is crucial for cap expression. The 26 essentiality of the inverted repeat is derived from this region acting as a SigB binding 27 site rather than as an operator site for the proposed cap activators RbsR and MsaB. 28 Furthermore, P cap contains an extensive upstream region harboring a weak SigA-29 dependent promoter and binding sites for the cap repressors SaeR, CodY and Rot. 30 We show that heterogeneous CP synthesis is determined by the combination of SigB 31 activity and repressor binding to the upstream region. The direct SigB dependency 32 and the upstream repressors are also sufficient to explain the temporal gene  defences (3,4). Therefore, it is being discussed as a target for immunotherapy and 63 as a vaccine candidate (5). 64 4 CP serotypes 5 and 8 are the two main CP serotypes produced by S. aureus strains 65 (6-9). Their structure is highly similar due to the closely related cap5 and cap8 gene 66 clusters. These allelic operons consist of 16 genes, cap5/8A to cap5/8P, whose gene 67 products are involved in CP biosynthesis, O-acetylation, transport and regulation (3,68 4, 10). The cap operon is thought to be mainly transcribed as a single large 17 kb 69 transcript driven by one principal promoter element (P cap ), located upstream of capA 70 (11,12). While cap gene clusters are extremely conserved across S. aureus 71 genomes, not all clinically relevant isolates produce CP and acapsular variants may 72 emerge during chronic infections. The loss of CP expression can typically be 73 explained by mutations in any of the cap genes essential for CP synthesis or in the 74 promoter region (13,14). For instance, acapsular strains from the USA300 lineage 75 were attributed to conserved mutations in the cap5 locus (15). However, this 76 assumption has been recently challenged by the finding that USA300 strains might 77 indeed produce CP during infection (16). In addition to mutations that abolish its 78 production, CP synthesis can also be switched off in response to environmental 79 conditions via a complex regulatory network. Extensive in vitro and in vivo analyses 80 have shown that cap expression is highly sensible to changes in nutrients, pH, CO 2 81 and oxygen availability (17)(18)(19)(20)(21)(22)(23)(24). Interestingly, CP synthesis was commonly found to 82 be strictly growth phase-dependent and detectable only in the post-exponential 83 growth phase (17,21,22,25,26). In addition, not all bacteria in a population are CP-84 positive as revealed by flow cytometry and immunofluorescence of in vitro and in vivo 85 grown bacteria (21,22,27,28). As only non-encapsulated cells are able to adhere to 86 endothelial cells (21), while CP protects bacterial cells from phagocytosis (29-31), it is 87 likely that CP heterogeneity provides better adaptability of the population as a whole. 88 So far the underlying regulatory mechanisms of this particular expression pattern 89 (early-Off/late-Heterogeneous) are only partially understood. In general, P cap activity correlates with CP synthesis, indicating that regulation occurs 91 predominantly at the transcriptional level (12,22,(32)(33)(34). Yet, the data to explain the 92 molecular mechanisms of cap regulation are puzzling. The identified transcriptional 93 start site (TSS) is not preceded by a classical Sigma factor consensus sequence 94 (12); instead, several inverted and direct repeats were identified further upstream, 95 amongst which a 10 bp inverted repeat (IR) was shown to be crucial for promoter 96 activity (12). It has been proposed that this IR functions as an operator site for the 97 cap activators RbsR and MsaB (35,36). RbsR also functions as a repressor of the 98 rbsUDK operon involved in ribose uptake. While the presence of ribose relieves 99 repression of the rbsUDK operon by RbsR, the presence and absence of ribose had 100 no effect on cap expression (35). MsaB is described as a transcriptional factor with 101 DNA binding capacity (36)  Rapid Amplification of cDNA Endings (5' RACE) ( Figure 1A). Most of the clones 145 (6/10) revealed a TSS at position -41 bp from the capA start codon, which is 146 preceded by a putative SigB consensus sequence. This SigB promoter consists of a 147 conserved SigB -35 motif (GTTTAA) and a -10 region harboring three mismatches 148 (ATGTAA versus GGGTAT) (51). Remarkably, the SigB -35 consensus sequence is 149 located within the IR that is crucial for P cap activity (12). The identified TSS confirmed 150 our RNA-Seq data and the TSS prediction of Prados et al. (49). In addition, 5' RACE 151 revealed one clone with a putative TSS at position -128 bp upstream of the capA 152 start codon, which was also predicted by Prados et al. (49). A conserved SigA   To further prove that SigB is directly involved in cap activation, the expression of the 181 full-length P cap promoter fusion was measured in wild-type and a sigB mutant over conditions, deletion of rbsR showed no effect on P cap activity ( Figure S1).

194
In the msa mutant, P cap promoter activity was lower than in the wild-type, supporting 195 the finding that MsaB/CspA contributes to cap activation ( Figure 3A). We 196 hypothesized that MsaB/CspA exerted its effect by modulating SigB activity. To test 197 this hypothesis, we used the double promoter fusion construct pCG742 to 198 simultaneously measure P cap and P asp23 activity. The P asp23 promoter is widely used 199 as marker for SigB activity (51,53,54) and was cloned in front of gpcer (52). We 200 found P cap and P asp23 activity to be highly correlated, in line with the assumption that Sae and Rot, additional repressive factors are acting on the P cap upstream region. 229 We further analyzed whether the repressors also affect the activity of the weak SigA-  Figure 5B). Binding is consistent with a putative 247 SaeR binding motif located between -79 bp and -94 bp upstream of the capA start 248 codon (56). Specific binding to the P cap upstream region was also found for Rot and 249 CodY ( Figure 5C and D). These findings are in line with the promoter activities 250 described above showing that Sae, Rot and CodY target the P cap upstream region 251 ( Figure 4A and B). Binding of the repressors to the downstream fragment (+10 to -77 252 from capA, see Figure 1A) is unspecific as band shifts are not eliminated by specific 253 unlabeled competitor ( Figure S3 A-D). However, it was previously found that CodY 254 can interact with a region further downstream (+160 to -152 from capA, see Figure   255 1A) reaching into the capA coding sequence (45). This region was so far not included 256 in our promoter activity assays or EMSAs. We could confirm CodY binding to this 257 region, when using a 3' extended downstream fragment (+160 to -77 from capA, see  (Figure 6). In addition, we generated a P cap mutant in which the 277 upstream promoter region containing the repressor binding sites was chromosomally 278 deleted (ΔP cap upstream, Figure 1C). Also in this strain CP production started earlier  However, throughout all experiments CP expression remained growth phase-288 dependent. We thought this could be due to the weak SigB promoter of P cap . 289 Therefore, we additionally altered the SigB -10 region to the conserved SigB -10 290 motif on the chromosome in the P cap upstream truncated strain (ΔP cap upstream, 291 strong SigB, Figure 1C). Together with constitutive sigB expression, this shifted the 292 onset of CP even further towards early growth phase. However, the majority of the 293 bacterial population still remained CP-negative in early logarithmic growth phase.  promoter. This is in line with a predicted SaeR binding site located between the two 391 promoters (56). We hypothesize that the repression of SigB-dependent promoter 392 activity occurs via steric interference, whereas the SigA-dependent promoter activity 393 might be repressed via a roadblock mechanism. Nonetheless, the molecular 394 mechanism for the long-distance effect of Rot and CodY on the SigB-dependent 395 promoter activity remains to be elucidated. One may speculate that secondary 396 structures of the promoter bring these two regulators in close proximity to the SigB 397 consensus motif, allowing them to interfere with SigB binding. It is well known that    piMAYcontrolfor and piMAYcontrolrev and sequencing, and then electroporated into 458 strain RN4220 and transduced into the target strains. Allelic exchange was 459 performed as described before (67)

493
For the generation of a cap mutant in USA300 JE2, pCG132 was transduced from 494 RN4220 pCG132 (33) into USA300 JE2 using phage Φ11.  T 2 was treated with MICROBExpress (Ambion) in order to remove rRNAs. After 527 treatment with Cap-Clip Phosphatase (Biozym) to remove pyrophosphate at the 5' 528 prime end of the native transcripts, a specific RNA 5' adapter (Table S3) was ligated 529 to the RNA. After phenol/chloroform extraction and ethanol precipitation, the RNA 530 was subjected to reverse transcription using oligonucleotide YFPCFPpolymorrev.

531
Nested PCR was performed using oligonucleotides Race2 and Racecapnestedrev 532 ( Table S3). The PCR amplicon was cloned into pCRII-TOPO (Invitrogen) following 533 the manufacturer's instructions. Single clones were analyzed via PCR using primers 534 Race2 and Racecapseqrev and the PCR products of 10 clones were sequenced with 535 primer Racecapseqrev. Infinite 200 microplate reader every half an hour with shaking at 37 °C. Absorbance 546 was measured at 600 ± 9 nm, gpVenus was excited at 505 ± 9 nm and emission was 547 measured at 535 ± 20 nm, and gpCerulean was excited at 434 ± 9 nm and emission normalized to gyrB. The genomic P cap variant is described in Figure 1C. Experiments 944 were performed in biological triplicates, error bars represent the standard deviation.

945
Statistical significance was determined by ordinary one-way ANOVA followed by 946 Tukey's multiple comparison test (ns: not significant, **: p<0.01). repaired) are repaired were grown to defined growth phase T 0 -T 4 and CP was 952 detected by immunofluorescence. In addition, the effect of P cap upstream truncation 953 and a strong SigB -10 consensus sequence (cap repaired, ΔP cap upstream, strong 954 SigB) was analyzed. This genomic P cap variant is described in Figure 1C.

955
Representative pictures of at least three independent cultures are shown.