The streptococcal phase-variable Type I Restriction-Modification system SsuCC20p dictates the methylome of Streptococcus suis and impacts virulence

Phase-variable Type I Restriction Modification (RM) systems are epigenetic regulatory systems that have been identified in numerous human bacterial pathogens. We previously showed that an emerging zoonotic lineage of Streptococcus suis acquired a phase-variable Type I RM system named SsuCC20p. The SsuCC20p locus was present in the genome of disease-associated isolates from multiple streptococcal species. This indicates that it is not restricted to S. suis and can be acquired through horizontal gene transfer. We demonstrate that SsuCC20p phase-variability relies on a recombinase present within the locus. In vitro, only SsuCC20p is responsible for the genome methylation profiles that were detected in the representative zoonotic S. suis isolate 861160. In addition, we show that, contrary to previous observations, hsdS genes located downstream of the hsdM gene and the recombinase gene, can contribute to the SsuCC20p genome methylation profile. SsuCC20p locked mutants expressing a single hsdS each showed unique genome methylation profiles. The differential genome methylation of the distinct locked mutants caused phase dependent differences in global gene expression in a growth condition dependent manner. We observed significant differences in virulence between hsdS locked mutants in a zebrafish larvae infection model. These data indicate that the streptococcal phase-variable Type I RM system SsuCC20p can impact bacterial virulence via epigenetic regulation of gene expression and potentially contributes to the zoonotic potential of S. suis. Importance Phase-variation contributes to the virulence of bacterial pathogens as it allows a single strain to produce phenotypic diverse subpopulations. Phase-variable Restriction Modification (RM) systems are systems that allow for such phase-variation via epigenetic regulation of gene expression levels. The phase-variable RM system SsuCC20p was found in multiple streptococcal species and was acquired by an emerging zoonotic lineage of Streptococcus suis. We show that the phase-variability of SsuCC20p is dependent on a recombinase encoded within the SsuCC20p locus. We characterized the genome methylation profiles of the different phases of SsuCC20p and showed that the differential genome methylation within the phases causes differences in gene expression levels and virulence. Altogether, we show that the acquisition of a phase-variable RM system impacts virulence and can potentially contribute to the zoonotic potential of S. suis. Bacterial pathogens can increase their virulence through acquisition of mobile elements containing epigenetic regulatory systems such as RM systems.

systems that have been identified in numerous human bacterial pathogens. We previously 23 showed that an emerging zoonotic lineage of Streptococcus suis acquired a phase-variable 24 Type I RM system named SsuCC20p. The SsuCC20p locus was present in the genome of 25 disease-associated isolates from multiple streptococcal species. This indicates that it is not 26 restricted to S. suis and can be acquired through horizontal gene transfer. We demonstrate 27 that SsuCC20p phase-variability relies on a recombinase present within the locus. In vitro, 28 only SsuCC20p is responsible for the genome methylation profiles that were detected in the 29 representative zoonotic S. suis isolate 861160. In addition, we show that, contrary to 30 previous observations, hsdS genes located downstream of the hsdM gene and the 31 recombinase gene, can contribute to the SsuCC20p genome methylation profile. SsuCC20p 32 locked mutants expressing a single hsdS each showed unique genome methylation profiles.

33
The differential genome methylation of the distinct locked mutants caused phase dependent 34 differences in global gene expression in a growth condition dependent manner. We

40
Importance 41 Phase-variation contributes to the virulence of bacterial pathogens as it allows a single strain 42 to produce phenotypic diverse subpopulations. Phase-variable Restriction Modification (RM) 43 systems are systems that allow for such phase-variation via epigenetic regulation of gene 44 expression levels. The phase-variable RM system SsuCC20p was found in multiple Introduction 60 Streptococcus suis is an opportunistic bacterial pathogen in pigs and an emerging zoonotic 61 pathogen (1). Human infections can lead to meningitis, streptococcal toxic-shock like 62 syndrome and septicemia (2, 3). Human infections are linked to exposure to pigs, such as 63 (occupational) handling of pig (products) or consuming undercooked or raw pig products (2, 64 4). S. suis is classified into serotypes based on capsular polysaccharides (CPS) structure 65 and into sequence types, which in turn are clustered into clonal complexes, based on its 66 genomic background as assessed by multi-locus sequence typing (MLST). Most human 67 infections are caused by S. suis serotype 2 of clonal complex (CC) 1, although infections 68 with other serotypes (e.g. serotype 14) and genotypes (e.g. CC20) have also been reported 69 (3,5).

71
In the Netherlands, a unique zoonotic serotype 2 CC20 clade has been identified, which is 72 more closely related to the non-zoonotic but virulent serotype 9 CC16 clade than to the 73 zoonotic serotype 2 CC1 clade (5). Three major genomic differences in the accessory 74 genome between CC16 and CC20 have been postulated to contribute to zoonotic potential 75 of CC20 strains (5). These include a capsule switch through acquisition of a serotype 2 CPS 76 locus (i), acquisition of a 89k pathogenicity island, previously identified in Chinese zoonotic 77 outbreak isolates (ii), and acquisition of a 18.5kb prophage region with a complete Type I 78 Restriction Modification (RM) system with phase-variable specificity subunits named 79 SsuCC20p (iii)(5, 6).

81
Phase-variable RM systems can be found in many pathogenic bacteria and have been 82 shown to regulate bacterial virulence (7). Type I RM systems consist of three host specificity 83 determinants (hsd) genes encoding a specificity subunit (S), a modification subunit (M) and a 84 restriction subunit (R) (Fig. 1A)

92
recombination is mediated via inverted repeats (IRs), which in some systems is (partially) 93 mediated by a recombinase within the same locus (8-10). Unique hsdS alleles, when 94 expressed with an hsdM, give unique methylation profiles in the genome. The methylation of 95 the genome can affect gene expression by affecting the binding of regulatory proteins, such 96 as transcription factors, to regulatory sequences upstream of genes (11,12). In this way, the 97 phase-dependent methylation profiles can impact virulence, as was shown for Streptococcus 98 pneumoniae (13).

100
We characterized SsuCC20p and show that SsuCC20p is phase variable, expressed and 101 actively methylates the S. suis genome. In a zebrafish larvae infection model of bacterial 102 infection (14-23), isogenic mutants with a single hsdS allele and distinct genome 103 methylation profiles differed in virulence.

126
Allele D was only found within one Chinese ST17 isolate. All except one isolate were 127 associated with disease (supplemental material Table S2).

129
SsuCC20p is located on a prophage region and is therefore likely acquired via horizontal 130 gene transfer (6). To identify other species that carry the same Type I RM system, we  Table S3). All but one S. agalactiae strain had a functional hsdS comprised of two 136 TRDs and five strains had three instead of four TRDs within the locus. In these same five S.

137
agalactiae strains, the orientation of xerD was reverted compared to the rest of the locus and 138 these strains had lost the IRs identified in S. suis SsuCC20p. SsuCC20p was also identified 139 in five other streptococcal species, but with more genomic differences than found for the S.  Table S3; supplemental material Table S3). In 142 case the host health state was reported for the isolate carrying this type I RM system 143 (10/23), the isolate was associated with disease (supplemental material Table S3).

146
The presence of a site-specific recombinase (xerD) and TRDs flanked by IRs within a single 147 locus, combined with the presence of multiple hsdS alleles in different genome assemblies 148 suggest that SsuCC20p is phase-variable. Whilst four different alleles of SsuCC20p have 149 been identified in CC20 S. suis genomes, the presence of these different alleles within a 150 single isolate has not been demonstrated yet. We chose the zoonotic ST20 isolate 861160 151 as our model strain because a single contig assembly of its genome is available (25). We

162
SsuCC20p is expressed and differentially methylates the genome

163
In previously identified Type I RM systems, the hsdS allele that is not located directly 164 downstream of the hsdM gene is silent, because it is encoded on the opposite strand, lacks 165 a start codon, or lacks a promoter (27). In SsuCC20p, a complete and functional hsdS is

174
To determine if SsuCC20p methylates the genome, the methylome of wildtype 861160 175 grown in THY was compared with a ΔhsdS strain lacking each of the 4 TRDs and xerD using 176 PacBio HiFi sequencing. In the wildtype, two m6A methylation profiles were identified 177 (GC m6 AN 5 GTC/G m6 ACN 5 TGC and GC m6 AN 5 CTC/G m6 AGN 5 TGC) with the Type I RM system

201
In S. pneumoniae, the site-specific recombinase (creX) present in the Type I RM system 202 SpnIII is essential for the TRD recombination between the small IRs (15bp), but not for TRD 203 recombination between the larger IRs (85bp and 333bp) (9). The IRs within SsuCC20p are 204 14bp, therefore we expected that TRD recombination in SsuCC20p is mediated by xerD 205 present in the locus, which was tested in ΔxerD mutants in 861160. After three sequential  Table S4).

217
To test the predicted hsdS allele assignment to the observed methylation profiles in the WT 218 861160 strain, locked mutants (LM) were constructed in which the silent TRDs and xerD 219 were replaced by an erythromycin resistance cassette so that the LMs express only a single 220 hsdS allele (Fig. 5A). In all LMs, the expression of hsdM, hsdS, and hsdR was confirmed

231
The unique genome methylation profiles of the LMs can result in differentially expressed 232 genes, as shown for CPS biosynthesis genes in S. pneumoniae (13). Thus, we aimed to

250
The unique genome methylation profiles of the LMs could affect S. suis antimicrobial 251 resistance, growth or virulence (13). The LMs did not differ in growth rate in THY at 37 °C or 252 in antimicrobial susceptibility to several antibiotics (Fig. 6B, supplemental material Fig. S4A).

253
The potential difference in virulence of LMs was assessed in a zebrafish larvae infection 254 model, which has been used to study the virulence of S. suis and other Streptococcus 255 species before (14-16, 19, 29). In the zebrafish larvae infection model, LM-E was less 256 virulent causing a higher survival of zebrafish than LM-H or WT, LM-A did not significantly 257 differ from LM-E, LM-H or WT (Fig. 6C). This difference could not be attributed to difference 258 in growth rate at the zebrafish larvae incubation temperature (28 °C) (supplemental material

264
Here, we characterized the phase-variable Type I RM system SsuCC20p that is encoded on 265 a mobile element and is found in multiple streptococcal species including a zoonotic S. suis 7 lineage. The SsuCC20p hsdS allele dependent genome methylation impacted S. suis gene 267 expression and virulence in a zebrafish larvae infection model for bacterial pathogenesis.

269
In contrast to previously characterized phase-variable Type I RM systems, we demonstrate 270 that additional hsdS alleles further downstream of hsdM are not always silent. To the best of 271 our knowledge, in all phase-variable Type I RM systems described so far, only the hsdS 272 allele directly downstream of the hsdM is transcribed and involved in genome methylation 273 (10,13,26,27,30,31,34

299
Our query for SsuCC20p in S. suis isolates identified 4 hsdS alleles, but only 3 could be 300 detected within the 861160 isolate under the used experimental conditions. Strain YS12

301
(ST17), that has the hsdS D allele in its genome assembly, belongs to a different ST than 302 the strain used in our experiments (ST20

331
The impact of genome methylation on the transcriptome is dependent on culture conditions.