EsaB is a core component of the Staphylococcus aureus Type VII secretion system

Type VII secretion systems (T7SS) are found in many bacteria and secrete proteins involved in virulence and bacterial competition. In Staphylococcus aureus the small ubiquitin-like EsaB protein has been previously implicated as having a regulatory role in the production of the EsxC substrate. Here we show that in the S. aureus RN6390 strain, EsaB does not genetically regulate production of any T7 substrates or components, but is indispensable for secretion activity. Consistent with EsaB being a core component of the T7SS, loss of either EsaB or EssC are associated with upregulation of a common set of iron acquisition genes. However, a further subset of genes were dysregulated only in the absence of EsaB. In addition, fractionation revealed that although an EsaB fusion to yellow fluorescent protein partially localised to the membrane, it was still membrane-localised when the T7SS was absent. Taken together our findings suggest that EsaB has T7SS-dependent and T7SS-independent roles in S. aureus.

EsxC substrate. Here we show that in the S. aureus RN6390 strain, EsaB does not genetically 23 regulate production of any T7 substrates or components, but is indispensable for secretion 24 activity. Consistent with EsaB being a core component of the T7SS, loss of either EsaB or 25 EssC are associated with upregulation of a common set of iron acquisition genes. However, 26 a further subset of genes were dysregulated only in the absence of EsaB. In addition, 27 fractionation revealed that although an EsaB fusion to yellow fluorescent protein partially 28 localised to the membrane, it was still membrane-localised when the T7SS was absent. Taken 29 together our findings suggest that EsaB has T7SS-dependent and T7SS-independent roles in 30 S. aureus. 31

INTRODUCTION
(12). Briefly, cells and supernatant were separated by a 10 min centrifugation step at 2770 g. 128 Cells were washed twice with PBS, adjusted to and OD 600 of 1 and digested using 50 µg/ml of 129 lysostaphin by incubation at 37°C for 30 min. Supernatants were filtered using a 0.22 µm filter 130 and TCA-precipitated in the presence of 50 µg/ml deoxycholate, as described. For S. aureus 131 subcellular fractionation, cells were grown to mid-log phase with shaking and treated as 132 previously described (12). Briefly, cells were harvested by centrifugation and resuspended in 133 TSM buffer (50 mM Tris-HCl pH 7.6, 0.5 M sucrose, 10 mM MgCl 2 ). Lysostaphin was added to 134 a final concentration of 50 µg ml −1 and cells were incubated at 37°C for 30 min to digest the 135 cell wall. At this point, protoplasts were sedimented to recover the cell wall (supernatant 136 fraction). Protoplasts were disrupted by sonication and the membrane was obtained after an 137 ultracentrifugation step at 227 000 g for 30 min and at 4°C. The supernatant was retained as 138 the cytoplasmic fraction. Samples were boiled for 10 min prior to separation in bis-Tris gels 139 and subsequent western blotting. 140 Polyclonal antisera were used at the following dilutions: α-EsxA 1:2500 (12), α-EsxB 1:1000 141  Anti-GFP antibody was obtained from Roche and used according to manufacturer's 143 instructions. 144

EsaB does not regulate the level of esxC transcripts in strain RN6390 146
A previous study has shown that a transposon insertion in the esaB gene results in an increase 147 in esxC transcripts in the Newman and USA300 strain backgrounds, and a concomitant 148 increase in the EsxC polypeptide, implicating it as a regulator (11). To investigate whether loss 149 of esaB by in-frame deletion affects the level of esxC mRNA in strain RN6390, we isolated 150 mRNA from the parental strain and the isogenic esaB mutant, prepared cDNA and undertook 151 reverse transcriptase PCR with primers covering either esxA (the first gene at the ess locus, 152 included as a negative control) or esxC ( Fig 1A). It can be seen (Fig 1C) that the level of 153 transcripts for each of these genes was qualitatively similar in the wild type and esaB 154

backgrounds. 155
To examine this quantitatively, we undertook RNA-Seq analysis on RNA prepared from three 156 biological repeats of the RN6390 and esaB strains grown aerobically in TSB to an OD 600 of 1. 157 Note that these experiments were performed at the same time as the RN6390 vs essC RNA-158 Seq analysis described in (15) and used the same RN6390 dataset. Fig 1D shows that the 159 level of esxC transcripts were indistinguishable between the wild type and esaB strains. 160 Analysis of the transcript levels of the other genes at the ess locus indicates that in general 161 they were also not significantly altered by the loss of esaB although there was a small increase 162 in the level of essB. We conclude that there is no evidence that esaB regulates the level of 163 esxC transcripts in RN6390. 164 We next examined the entire transcript profile of the esaB mutant to investigate the 165 transcriptional/post-transcriptional response to the loss of this small protein. We found 101 166 genes de-regulated in the esaB mutant compared to the parental strain (using a cut off of 167 logFC > 2 or < -2 and qvalue < 0.05, as applied previously (15)), Fig 2A. Of these, 43 were 168 upregulated by the loss of esaB whereas 58 were downregulated when esaB was absent -169 these genes are listed in Table 2. Interestingly, almost all of the genes that were differentially 170 regulated in the essC mutant (15) were also similarly regulated in the esaB strain (Fig 4B), 171 although there was a substantive subset of genes that were differentially expressed in the 172 esaB mutant but not the essC strain (Table 2). It can be seen that almost all of the iron 173 acquisition genes, including those for heme acquisition, staphyloferrin synthesis and uptake 174 and ferrichrome import were commonly upregulated by loss of either esaB or essC (Table 2). 175 Furthermore six of the eight downregulated genes from the essC strain were also down 176 regulated in the esaB strain (note that one of the two genes unaffected in the esaB dataset is 177 essC itself, which appears downregulated in the essC dataset because it has been deleted). 178 The finding that almost the entire subset of genes differentially regulated in the absence of 179 essC is also similarly altered by loss of esaB strongly suggests that EsaB is, like EssC, a core 180 component that is essential for activity of the secretion machinery in strain RN6390. 181 As mentioned above, a subset of transcripts were differentially expressed in the esaB but not 182 the essC strain. These include downregulated genes required for anaerobic nitrate respiration 183 (narGHJ/narK), some secreted proteases (sspA/B/C, aur), capsular polysaccharide synthesis 184 (capG/F/hysA), lactose metabolism (lacB/C/D) and antimicrobial peptide synthesis 185 (epiA/C/D/P). Many of these genes are under control of the essential two component 186 regulatory system AirSR (formerly YhcSR) (25-28). These findings suggest that EsaB may 187 have additional roles in the cell in addition to its requirement for T7 protein secretion. 188 189

EsaB is present at low amounts in cells of S. aureus RN6390 190
To explore the biological role of EsaB in T7 secretion, we firstly overproduced recombinant 191 EsaB with a cleavable His-tag in E. coli, and following cleavage of the tag the protein was 192 further purified by gel filtration chromatography (Fig 3A, B). The purified protein, which eluted 193 with an estimated molecular mass of approximately 12.8 kDa, is close to the expected size of 194 a monomer (9.1 kDa + 0.3 kDa retained following cleavage of the tag = 9.4 kDa). This is in 195 agreement with structural analysis of the B. subtilis EsaB homologue, YukD, which also 196 appears to be monomeric (20). 197 Polyclonal antisera were raised against purified EsaB and the antibody was affinity purified 198 against the EsaB antigen, before being used to detect the protein in whole cells of S. aureus. 199 purified EsaB indicated that the antibody was able to cross-react with as little as 25ng of 202 protein, which is equivalent to 1.6 x 10 11 EsaB molecules. Since the antibody was unable to 203 detect EsaB in whole cells from 9.6 x 10 8 colony forming units that were loaded onto the SDS 204 gel, we conclude that are less than 170 molecules of EsaB per cell. 205 Since we were unable to detect native EsaB in S. aureus cell extracts, we constructed a series 206 of tagged variants for which commercial antisera were available. To this end we introduced 207 His 6 , Myc, hemagglutinin (HA) and Strep epitopes onto the N-terminus of EsaB, and His 6 , Myc, 208 HA, mCherry or FLAG epitopes onto the C-terminus, but in each case were unable to detect 209 the tagged protein (not shown). We also introduced His 6 and His 9 epitopes into two predicted 210 loop regions internal to the EsaB sequence but again were unable to detect tagged EsaB (not 211 shown). The only tag we introduced that allowed detection of EsaB was a C-terminal yellow 212 fluorescent protein (YFP) tag. Fig 3D shows that basal production of either native (untagged) 213 EsaB or EsaB-YFP from plasmid vector pRAB11 was sufficient to restore secretion of the 214 T7SS extracellular protein EsxA and of substrates EsxB and EsxC to the culture supernatant. 215 Blotting the same cell samples for the presence of the YFP fusion protein (Fig 3E) showed 216 that it migrated at close to the predicted mass (37 kDa) of the EsaB fusion. There was no 217 evidence for degradation of the fusion protein even after prolonged exposure of the 218 immunoblot ( Fig 3E). We conclude that the YFP-tagged variant of EsaB probably retains 219 functionality. 220 EsaB is predicted to be a soluble cytoplasmic protein (10), and is known to share structural 223 homology with ubiquitin (20). Interestingly, a domain sharing the same fold is also associated 224 with the actinobacterial T7SS, being found at the cytoplasmic N-terminus of EccD (29), 225 indicating that ubiquitin-like proteins are essential features of all T7SSs. To determine the 226 subcellular location of EsaB-YFP, we blotted secreted and whole cell samples of the esaB 227 mutant strain producing plasmid-encoded EsaB-YFP with the YFP antiserum. Fig 4A shows  228 that EsaB-YFP was associated exclusively with the cellular fraction. 229 We next fractionated these cells to obtain cytoplasm, cell wall and membrane fractions. 230 Immunoblotting with antisera to control proteins known to localize to the cell membrane (SrtA) 231 and cytoplasm (TrxA) indicated that the fractionation had been largely successful, although 232 some SrtA was found in the cell wall fraction (Fig 4B). Blotting these same fractions for the 233 presence of EsaB-YFP showed that the protein localised to both the cytoplasm and membrane 234 fractions. Some degradation of the fusion protein was also noted in these experiments which 235 may result from the activation of proteases during fractionation. When unfused YFP was 236 produced in the wild type strain it did not localise to the membrane (Fig 4C), indicating that 237 membrane binding was unlikely to be mediated through the YFP portion of the fusion. 238 Next we tested whether EsaB-YFP localised to the membrane through interactions with 239 membrane components of the T7SS. To this end we repeated the fractionation in a strain 240 carrying a chromosomal deletion in all twelve genes at the ess locus ( Fig 1A). However this 241 did not alter the localization of EsaB-YFP, which was still detected in both cytoplasm and 242 membrane fractions (Fig 4B). It may be that EsaB-YFP localises to the membrane through 243 interaction with additional membrane proteins, consistent with additional, T7SS-independent 244 roles for EsaB suggested through the RNA-Seq analysis. Alternatively, we cannot rule out that 245 the membrane localization arises as an artifact of the C-terminal YFP tag, since this tag is 246 known to influence protein behaviour (e.g. (30)). 247 248

Mutagenesis of conserved hydrophilic and hydrophobic patches on EsaB 249
An alignment of EsaB homologues encoded across firmicutes ( Fig 5A) identifies a number of 250 highly conserved amino acids. Many of these are hydrophilic and fall on one face of the 251 predicted structure of EsaB including T8 (S. aureus numbering) which is highly conserved as 252 either threonine or serine, and the invariant K56. The presence of an invariant lysine is 253 particularly intriguing since there are a number of highly conserved lysine residues on the 254 structurally-related protein ubiquitin, that are used to assemble polyubiquitin chains (31). To 255 probe potential roles of these conserved residues we mutated each of T8, D10, L21, K30, 256 K52, K56, L66, G74 and D75 to alanine on plasmid-encoded EsaB and assessed whether the 257 variant EsaB proteins were able to restore T7 secretion activity to the esaB deletion strain. Ubiquitin has a conserved hydrophobic patch (Fig 6A, left) that forms a common site of 264 interaction with many different binding partners (32). Analysis of the predicted structure of 265 EsaB (Fig 6A, right) shows that there are some hydrophobic residues on the surface potentially 266 at positions approximating the hydrophobic patch region of ubiquitin. To assess whether these 267 hydrophobic residues may be involved in EsaB function, we firstly mutated V7, I44, I71 and 268 L77 to alanine residues. Fig 6B shows that these substitutions did not detectably affect T7 269 secretion indicating that the function of EsaB had not been compromised. We next substituted 270 each of these residues for a positively-charged lysine. This more drastic change of amino acid 271 side-chain was still tolerated at positions 44 and 77, but inactivated EsaB when substituted for 272 V7 or I71. 273 Finally we attempted to assess whether any of the inactivating substitutions E7K, T8A, T8R 274 or I71K altered the subcellular location of EsaB-YFP. However, when we introduced each of 275 these substitutions into EsaB-YFP we found that they destabilised the protein as it was almost 276 undetectable in whole cells (Fig 6D), precluding further analysis. We are therefore unable to 277 determine whether these substitutions directly alter EsaB function or have an indirect effect 278 by disrupting folding. 279

DISCUSSION 280
In this work we have investigated the role of EsaB in Type VII secretion. EsaB proteins are 281 conserved in firmicutes that produce the T7SS and are encoded at the same loci. Previous 282 work had implicated EsaB in the regulation of esxC transcripts (11), although this cannot be a 283 conserved role for EsaB proteins as they are found in all S. aureus strains, including the subset 284 that do not encode esxC (16). Here we show that EsaB does not regulate esxC in strain 285 RN6390, nor any of the other genes encoded at the ess locus. Instead, deletion of esaB is 286 associated with upregulation of genes involved in iron acquisition, mirroring the upregulation 287 of iron-acquisition genes seen when the core T7 component, EssC, is absent (15). This 288 supports the notion that EsaB is a core component of the secretion machinery in RN6390, and 289 in agreement with this, deletion of esaB prevented export of the T7-dependent extracellular 290 proteins EsxA, EsxB and EsxC. This conclusion is also in agreement with related studies in 291 B. subtilis, where the EsaB homologue YukD was shown to be essential for secretion of the 292 WXG100 protein YukE (17, 18). 293 The precise role of EsaB in T7 secretion is unclear. Structural analysis of B. subtilis YukD 294 shows that it shares a very similar fold to ubiquitin but that it lacks the ability to be conjugated 295 with other proteins (20). Interestingly, a ubiquitin-like domain is also associated with the 296 actinobacterial T7SS, being found at the cytoplasmic N-terminus of the polytopic EccD 297 membrane component (29), suggesting that EsaB-like components are essential features of 298 all T7SSs. Ubiquitin interacts directly with many different protein binding partners (32), and it 299 is therefore likely that EsaB interacts with one or more components of the T7SS, potentially 300 regulating activity. Post-translational regulation of the S. aureus T7SS has been suggested 301 because in some growth conditions the secretion machinery is present but there is no or very 302 little substrate secretion (12, 19). Other protein secretion systems are also post-translationally 303 regulated, for example the flagellar type III secretion system is regulated through interaction 304 of the FliI component with the second messenger cyclic di-GMP (33), and Type VI secretion 305 systems are regulated by phosphorylation (34). In this context, EsaB proteins contain a highly conserved threonine (or serine) residue close to their N-termini which we considered as a 307 potential site for phosphorylation. Intriguingly, substitution of EsaB T8 for alanine abolished 308 the function of EsaB, although introduction of either the phospho-mimetic glutamate at this 309 position or a positively charged lysine did not affect EsaB activity. 310 The low cellular levels of EsaB precluded further analysis of the native protein, but a C-terminal 311 fusion of EsaB with YFP partially localised to the cell membrane. We reasoned that binding of 312 EsaB-YFP to membranes was mediated through interaction with one or more of the T7SS 313 membrane proteins. However, some EsaB-YFP was still membrane associated when it was 314 analysed in a strain lacking all of the core T7 components, suggesting that it may interact with Overlap between up-and down-regulated genes in the esaB and essC datasets. 348