Peptidoglycan acetylation states determine the colony opacity phenotypes

Our previous study showed that the two-component system TCS11, consisting of the sensing kinase HK11 and response regulator RR11, promotes the HsdS_A1 genomic methylome and thereby the formation of O colony phase in S. pneumoniae [28]. While the parental strain (ST606) produced 80.3% O and 19.7% T colonies, the ∆hk11 mutant (TH11861) showed only 1.1% O colonies (Fig 1B), which agrees with the importance of TCS11 in the formation of O colonies [28]. However, the Δhk11 colony phenotype remained the same when it was reverted with the intact hk11 (hk11^rev*, TH11857; Fig 1B). This result indicated that other gene(s) beyond hk11 was involved in the colony phenotype. Whole genome sequencing of Δhk11^rev* (TH11857) revealed a single nucleotide replacement of the -7^th thymine residue by a guanine residue in the 5’ non-encoding region of the pgdA gene (myy0814) (Fig 1C). The pgdA encodes a cell wall NAG deacetylase (Fig 1A) [20]. Subsequent correction of this mutation (hk11^rev, TH13471) restored the O/T colony ratio to that of WT (Fig 1B). Additional genetic manipulations of pgdA confirmed that the deacetylase activity of PgdA is required for O colony phase (Fig 1D). The role of PgdA in the phase variation was further verified in serotype-6A (TH6671) and serotype-35B (TH6675) strains (Figs 1E and S1).

Pneumococcal PG is also acetylated at its C-6 hydroxyl of NAM by the O-acetyltransferase Adr (Fig 1A) [14]. So, we next investigated the impact of NAM O-acetylation on pneumococcal colony opacity by targeted mutagenesis. As shown in Fig 1F, the proportion of O colonies in Δadr mutant was reduced to 22.9% from 83.1% in WT. Revertant with the intact adr gene in the adr deletion background (adr^rev) led to an increase in the O colony ratio to that of WT. To ascertain the role of Adr’s enzymatic activity in pneumococcal colony opacity, we generated an adr mutant by replacing serine with alanine at position 438, a key amino acid for O-acetyltransferase activity [30]. Similar to the phenotype of the adr deletion mutant, the adr^S438A mutant showed significant attenuation on the capacity of forming O colonies, generating only 3.4% O and 96.6% T colonies (Fig 1F). The causal relationship between Adr O-acetyltransferase activity and colony opacity was also confirmed in serotype-6A (TH6671) and serotype-35B (TH6675) strains. The adr^S438A derivative of TH6671 produced a marginal level of O colonies (Figs 1G and S1). A similar degree of reduction in O colony was also observed in the adr^S438A mutant of TH6675. The T-dominant phenotype in the adr^S438A derivatives of TH6671 and TH6675 was reversed to that of parental strains with the intact adr. These findings demonstrated that Adr-catalyzed O-acetylation favors opaque colony phase.

NAM O-acetylation stabilizes the O phase gene configuration

Previous studies have shown that the O and T phase variation is controlled by PsrA-catalyzed DNA inversions in the cod locus (Fig 2A) [25,27]. We thus tested whether the PgdA- and Adr-catalyzed PG modifications impact colony opacity through the programmed DNA inversions by generating pgdA^D275N and adr^S438A mutants in a pneumococcal strain that carried a fixed hsdS_A1 allele (TH6552) because the active site point mutation of the invertase PsrA uniformly formed O colonies [25]. Similar to the colony phenotype of pgdA^D275N, the hsdS_A1-pgdA^D275N mutant (TH14570) completely lost the ability to produce any O colonies. This phenotype was fully converted to that of parental strain when reverted with the intact pgdA (hsdS_A1-pgdA^rev, TH13467) (Fig 2B). This result showed that PG N-deacetylation modulates pneumococcal colony opacity through a PsrA-independent mechanism. In sharp contrast, the absence of PsrA completely blocked Adr from affecting colony opacity. Compared with significant reduction of O colonies in the adr^S438A strain, the hsdS_A1-adr^S438A strain (TH14736) showed uniform production of O colonies as the parental strain (Fig 2B). This functional dependence of Adr on PsrA suggested that NAM O-acetylation controls colony phase by modulating the hsdS inversions.

To ascertain the relationship between hsdS inversions and PG modifications in modulating colony phase, we assessed the ratio of hsdS_A1-genotype bacteria in the enzymatic inactivation adr^S438A and pgdA^D275N mutants by detecting the relative mRNA abundance of hsdS_A1 since only the HsdS_A1 methylation contributes to the formation of O colonies [28]. Normalized by the abundance of hsdS_A1 mRNA in the hsdS_A1-fixed mutant, a similar level of hsdS_A1 mRNA was detected in the WT and pgdA^D275N, while the adr^S438A population exhibited a significant decrease in relative abundance of hsdS_A1 mRNA (Fig 2C). The reduced hsdS_A1 representation in the adr^S438A strain was restored to the WT level with the intact adr. However, the impact of Adr on hsdS inversions became undetectable with the dysfunction of PsrA (Fig 2D). These data showed that NAM O-acetylation shapes pneumococci toward the hsdS_A1 allelic configuration in the cod locus.

PsrA catalyzes DNA inversions by recognizing three pairs of inverted repeats in the coding regions of hsdS_A (IR1.1, IR2.1 and IR3.1), hsdS_B (IR1.2), and hsdS_C (IR2.2 and IR3.2) (Fig 2A) [25]. We characterized how NAM O-acetylation impacts hsdS inversions by detecting the forward and reverse orientation of the invertible sequences in the cod locus by quantitative PCR as described [26]. The hsdS_A1-fixed strain was used as a positive control to set the forward configurations for the IR1-, IR2- and IR3-bound sequences. Consistent with the O-dominant phenotype of WT strain, the IR1-, IR2-, and IR3-bound sequences were predominantly oriented in the forward direction (Fig 2E). While pgdA^D275N mutant showed a similar genotype as WT, adr^S438A mutant shifted the IR1- and IR2-bound sequences to the reverse direction.

We verified the impact of NAM O-acetylation on hsdS inversions in serotype-2 strain D39 and its adr derivative. As presented in S2 Fig, the adr^S438A mutant of D39 showed significant reduction in the proportion of hsdS_A1-carrying variant (S2A Fig) and the orientation shift of IR-bound sequences (S2B Fig). These experiments indicated that NAM O-acetylation broadly modulates hsdS inversions in S. pneumoniae.

NAM O-acetylation is required for the O phase methylome

Since the pneumococcal methylome catalyzed by the HsdS_A1 methyltransferase defines the O phase [26], we determined whether NAM O-acetylation affects the methylome by comparing DNA methylation status between WT and adr^S438A strains. The 6-A methylation (6-mA) states in the sequences recognized by the six HsdS_A allelic variants were determined by PacBio single molecule real-time (SMRT) sequencing (Fig 3A). Consistent with the phenotypic dominance of HsdS_A1-defined O phase in WT bacteria, nearly all the 2,060 loci of the HsdS_A1 motif (5’-CRA^m6AN₈CTT-3’) were methylated in the WT genome, whereas less than 50% of the HsdS_A2 (5’-CRA^m6AN₉TTC-3’) and HsdS_A3 (5’-CRA^m6AN₈CTG-3’) counterparts were methylated (Fig 3B and 3C). This result indicated that the vast majority of bacterial cells in the WT population express the hsdS_A1 allele in the cod locus as reported previously [25,28]. In striking contrast, no methylation was detected for any of the HsdS_A1 motifs in the adr^S438A genome, but the methylation rates for the HsdS_A2 and HsdS_A3 motifs were significantly increased. In particular, virtually all of the HsdS_A3 motifs (99.7%) became methylated in the adr^S438A genome (Fig 3B and 3C). However, PacBio sequencing revealed the WT level of HsdS_A1 methylation in the adr revertant (Fig 3B and 3C), unequivocally demonstrating a causal relationship between NAM O-acetylation and O phase methylome. In contrast with the phenotypic disconnection between hsdS inversions and PgdA-driven colony opacity, the pgdA^D275N strain showed loss of the HsdS_A1 methylome, as well as other HsdS_A and even Spn556I/III methylomes (S1 Table), indicating the crucial role of NAG N-deacetylation in genomic methylation activity rather than modulating hsdS inversions. These striking differences in the methylome of pgdA mutant were not due to potential differences in sequencing depth, since there was a comparable level of total reads among the WT, adr^S438A and pgdA^D275N strains (S2 Table). Taken together, the NAM O-acetylation by Adr is absolutely required for the HsdS_A1 methylome in S. pneumoniae.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The loss of HsdS_A1 methylome in the absence of NAM O-acetylation.

(A) Illustration of experimental design for detecting pneumococcal genomic methylation by PacBio sequencing. (B) Genomic methylation by HsdS_A1-6 MTases in WT, pgdA^D275N, adr^S438A and adr^rev strains. The methylated adenine nucleotides in each of six DNA motifs are highlighted in red. “# in genome” indicates the total copies of each methylation sequence in both strands of ST556 genome (accession CP003357.2). “% detected” and “# detected” represent the ratio and the number of loci detected by PacBio sequencing, respectively. (C) Methylation rates for HsdS_A1-3 recognition motifs in WT, pgdA^D275N, adr^S438A and adr^rev strains.

https://doi.org/10.1371/journal.ppat.1013286.g003

LytA synchronizes NAM O-acetylation with the O phase methylome

The connection between NAM O-acetylation and hsdS inversions may be explained by two possibilities: 1) NAM O-acetylation enriches hsdS_A1-genotype bacteria due to growth advantage of the O variant, and 2) the status of NAM O-acetylation is sensed and relayed to the intracellular milieu to regulate hsdS inversions by an unknown signaling pathway. The first possibility is unlikely since our previous study showed a similar growth pattern between isogenic variants carrying hsdS_A1 and other hsdS_A alleles [25]. Given the known role of NAM O-acetylation in resistance to the major autolysin LytA in S. pneumoniae [31], we tested the second possibility. We performed mutagenesis analysis of LytA, as well as other three PG hydrolases (LytB, LytC and CbpD), because these proteins also bind and cleave PG (Fig 4A) [9]. While deleting lytB, lytC or cbpD in the adr^S438A strain did not result in obvious impact on hsdS inversions, deleting lytA in adr^S438A led to the significant increase in hsdS_A1-carrying bacteria (Fig 4B). Reinstating the wildtype lytA sequence in the adr^S438A-ΔlytA mutant restored the low proportion of hsdS_A1 configuration, similar to the parental strain adr^S438A. In a similar fashion, adr^S438A-ΔlytA mutant produced 100% O colonies (Fig 4C). In contrast, the lytB, lytC and cbpD mutants displayed a similar proportion of T/O ratio as the parental strain. These data strongly suggested that LytA functionally links hsdS inversions to colony phase, in response to the change of NAM O-acetylation state.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The role of LytA in functionally linking decreased NAM O-acetylation with hsdS inversions and colony phase.

(A) Cell wall cleavage sites of pneumococcal PG hydrolases. MurNAc, N-acetylmuramic acid. GlcNAc, N-acetylglucosamine. (B to C) Relative abundance of the hsdS_A1 mRNA (B) and O/T colony phenotype (C) in lytA, lytB, lytC and cbpD mutants in the adr^S438A background. Significance between adr^S438A and other mutants is presented. (D) Illustration of the functional domains and amino acid residues of LytA. (E to F) Relative abundance of the hsdS_A1 mRNA (E) and O/T colony ratio (F) in different isogenic adr-lytA double mutants. Significance between adr^S438A and other mutants is presented.

https://doi.org/10.1371/journal.ppat.1013286.g004

LytA is a cell wall hydrolase of S. pneumoniae that cleaves the amide bond between NAM and L-alanine in a zinc-dependent manner [32], which is responsible for autolysis at the stationary phase, during nutrient depletion or when PG synthesis is inhibited by antibiotics [33,34]. The amidase activity of LytA is also known to contribute to T colony appearance [25]. To uncouple the O-acetylation-dependent and -independent impact of LytA on colony opacity, we constructed a lytA deletion mutant in WT strain. As reported previously [25], ΔlytA formed 100% O colonies (S3A and S3B Fig). However, ΔlytA had a similar level of hsdS_A1 mRNA with WT. This result indicates that, besides its direct impact on colony appearance by PG hydrolysis, LytA is able to modulate colony opacity in an O-acetylation-dependent manner.

LytA consists of an N-terminal catalytic amidase domain (AMI) and a C-terminal choline-binding domain (CBD) for protein anchoring to choline moiety of WTAs (Fig 4D). We assessed the specific contribution of the AMI and CBD domains to its function in modulating hsdS inversions in the O-acetylation-absent background. To our surprise, while the CBD-deficient strain (adr^S438A-ΔlytA^CBD) exhibited a similarly low proportion of hsdS_A1-carrying bacteria as the parental strain, the mutant without AMI (adr^S438A-ΔlytA^AMI) showed relatively higher representation of hsdS_A1 configuration (Fig 4E). By comparison, both the AMI and CBD mutants produced 100% O colonies, highlighting the dual functions of LytA (Fig 4F). This result suggested that the enzymatic domain of LytA is responsible for connecting the NAM O-acetylation status with the genetic configuration in the cod locus.

Several amino acid residues in the LytA amidase domain are responsible for the glycan-binding and catalytic, or zinc-binding activities (Fig 4D) [35,36]. We determined the roles of these activities in modulating hsdS inversions, in response to the loss of NAM O-acetylation, by selective change of representative residues. All the point mutants showed a comparable level of LytA as WT (S3C Fig). While mutating the two glycan-binding residues (33^rd serine and 41^st tyrosine) in adr^S438A abolished the dominance of the hsdS_A1 configuration in the absence of NAM O-acetylation, the loss of the catalysis (adr^S438A-lytA^E87A) and zinc-binding (adr^S438A-lytA^H133A) residues did not yield obvious impact on the hsdS_A1 dominance in the same strain background (Fig 4E). On the other hand, all the three mutants produced 100% O colonies (Fig 4F). This result showed that the glycan-binding activity, but not enzymatic function, of LytA is essential for modulating hsdS inversions, while the PG hydrolysis by LytA is required for T colony formation.

Multiple LytA-associated proteins contribute to NAM O-acetylation-dependent hsdS inversions

Considering that LytA functions in the extracellular milieu [34], it must engage its partner(s) to transmit the lack of NAM O-acetylation across the cell membrane to modulate hsdS inversions. Because the LytA glycan-binding activity is essential for modulating hsdS inversions in the absence of the NAM O-acetylation, we reasoned that WTA-anchored LytA is disengaged from PG when the C6-OH group of NAM is acetylated by the partner protein(s) due to the relatively higher affinity to LytA than PG; the lack of NAM O-acetylation makes PG more attractive to LytA than the partner protein(s); the LytA-less partner protein(s) activates the downstream signaling cascade to modulate hsdS inversions. We tested this hypothesis by comparing LytA-associated proteins via co-immunoprecipitation (Co-IP). Pneumococci expressing a Strep-tagged LytA were subjected to protein crosslinking with 1% formaldehyde before being lysed and incubated with biotin-conjugated beads to isolate proteins associated with LytA.

Mass spectrometry analysis identified 16 potential LytA-binding proteins under the NAM O-acetylation conditions (Fig 5A and S3 Table). Although these proteins were similarly expressed in two strains, they were more abundantly enriched in WT by Co-IP than in adr^S438A. Adr is also produced at the same level in two strains (S3D Fig). Some of these are intracellular proteins (e.g., ribosomal proteins and metabolic enzymes), which may interact with LytA before it is secreted or be contaminants. Based on the essential gene list of S. pneumoniae [37], we selectively characterized 7 genetically amendable hits. Deleting ptvB or pcpA in adr^S438A yielded a similar phenotype as the ΔlytA mutant. The adr^S438A-ΔptvB and adr^S438A-ΔpcpA strains showed significantly higher proportions of the hsdS_A1-positive bacteria (Fig 5B). By comparison, deleting the remaining 5 genes did not yield obvious effect.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The requirement of LytA-associated proteins for linking decreased NAM O-acetylation with the orientations of hsdS inversions.

(A) LytA-associated proteins enriched in the absence of NAM O-acetylation. The proteins pulled down from the lysates of ST606 and adr^S438A strains by LytA-coated beads are presented as the average of the peak area obtained from four biological repeats in two individual experiments. (B) Relative abundance of the hsdS_A1 mRNA in the mutants of LytA-associated proteins as in Fig 2C. Significance between adr^S438A and other mutants is presented. (C) The domain structures of PtvB and PcpA. T, transmembrane region. S, signal peptide. LRR, leucine-rich repeat region. CBD, choline-binding domain. (D to E) Relative abundance of the hsdS_A1 mRNA (D) and colony phenotypes (E) of ptvB mutants in WT and adr^S438A backgrounds. (F to G) Relative abundance of the hsdS_A1 mRNA (F) and colony phenotypes (G) of pcpA mutants in WT and adr^S438A backgrounds.

https://doi.org/10.1371/journal.ppat.1013286.g005

PtvB was identified for its involvement in pneumococcal tolerance to vancomycin [38], but its precise function remains unclear. It contains a predicted short N-terminal transmembrane domain and a large extracellular segment (Fig 5C). In agreement with the requirement of PtvB for configuring hsdS inversion in the absence of NAM O-acetylation, PtvB is necessary for the T dominant phenotype of the adr^S438A strain. The ΔptvB mutant lost the T dominant genotype (Fig 5D) and phenotype (Fig 5E) of adr^S438A. This change was fully restored in revertant with the intact ptvB (adr^S438A-ptvB^rev). In sharp contrast, the absence of PtvB did not affect hsdS inversions and colony phenotype of WT. These experiments verified that PtvB promotes the non-hsdS_A1 orientation of hsdS inversions only in the absence of NAM O-acetylation. However, we did not observe physical interaction between LytA and PtvB by bacterial adenylate cyclase-based two-hybrid (BATCH) experiment (S3E Fig), indicating that the functional linkage of the two proteins is bridged by an unknown partner. Furthermore, we tested potential collaboration of LytA and PtvB in promoting vancomycin tolerance based on the contribution of PtvB to pneumococcal tolerance to vancomycin [38]. Surprisingly, the loss of LytA itself made the wildtype bacteria more susceptible to vancomycin (S4 Fig), which made it difficult to interpret the functional connection between LytA and PtvB in vancomycin tolerance.

While PcpA is reported as a virulence factor [39,40], and a potential vaccine candidate [41,42], its physiological function remains to be defined. It contains a C-terminal leucine-rich repeat region and a N-terminal choline-binding domain (Fig 5C) [43]. Our additional experiment also confirmed that PcpA is involved in regulating hsdS inversions (Fig 5F). In the adr^S438A background, deleting pcpA led to the return of the T-dominant phenotype to the O-dominant level of WT, which was completely restored in pcpA revertant (Fig 5G). Surprisingly, unlike the deletion of ptvB, the absence of pcpA in WT background significantly affected hsdS inversions and colony opacity as well. The proportions of O colonies and hsdS_A1-carrying bacteria were significantly reduced in the ΔpcpA mutant (Fig 5F and 5G). This indicated that PcpA plays a more complex role than PtvB in regulating hsdS inversions.

PtvB modulates hsdS inversions by interacting with PtvC and DimA

Because PtvB is predicted to be primarily localized in the extracellular space with a transmembrane segment and a short intracellular tail of six amino acids, PtvB is unlikely to directly modulate hsdS inversions. We identified pneumococcal protein(s) that interacts with PtvB in the absence of NAM O-acetylation. The PtvB-associated proteins were enriched by incubating biotin-conjugated beads with the lysates of the Strep-PtvB expressing WT and adr^S438A strains. Mass spectrometry revealed 39 proteins that were selectively enriched in adr^S438A as compared with WT (Fig 6A and S4 Table). Since many of these proteins might be contaminants being accidentally crosslinked to PtvB or other proteins, we verified the functional involvement of 16 genetically amendable genes in hsdS inversion regulation by mutagenesis in adr^S438A. Only ΔhsdM and Δmyy1025 showed significant impact on hsdS inversion in adr^S438A as ΔptvB, in terms of hsdS_A1-carrying bacteria, whereas deleting the other 14 genes led to marginal or no effect (Fig 6B). The hsdM encodes the DNA methyltransferase subunit of the type I R-M system in the cod locus, and is located immediately upstream of hsdS_A [25]. Since the enrichment of HsdM could be explained by potential transcriptional upregulation of the hsdRMS_A operon [26], we chose to focus on myy1025.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Functional relationship of DimA and PtvC with PtvB.

(A) PtvB-associated proteins enriched by PtvB-coated beads from the lysates of ST606 and adr^S438A strains are presented as in Fig 5A. (B) Relative abundances of the hsdS_A1 mRNA in the mutants of PtvB-associated proteins are shown as in Fig 2C. Significance between adr^S438A and other mutants is presented. (C to D) Relative abundance of the hsdS_A1 mRNA (C) and colony phenotypes (D) of dimA mutants. (E) Detection of DimA interactions with PtvA, PtvB and PtvC by bacterial two-hybrid assay. Colonies on the MacConkey/maltose plates (upper panel) and β-galactosidase activity (lower panel) are shown for each reporter strain. PC, positive control (pKT25-zip and pUT18C-zip); NC, negative control (empty vectors pKT25 and pUT18C); BC, blank control without plasmid. Significance between NC and experimental groups is presented. (F) Detection of physical interactions among PtvA, PtvB and PtvC by bacterial two-hybrid assay. The data are shown as in E. (G to H) Relative abundance of the hsdS_A1 mRNA (G) and colony phenotypes (H) of ptvA, ptvB or ptvC mutants in adr^S438A background. Significance between adr^S438A and other mutants is presented. (I) The protein interaction model of LytA and PtvB to modulate hsdS inversions. The thickness of the double-headed arrows represents the interaction intensity.

https://doi.org/10.1371/journal.ppat.1013286.g006

The myy1025 encodes a cytoplasmic protein of 424 amino acids without any characterized or predicted function. We renamed it as DNA inversion modulator A, dimA. Further experiments showed that the impact of DimA on the hsdS configuration was reversed with the intact dimA in adr^S438A (Fig 6C). In addition, deleting dimA in WT did not yield significant change in the hsdS configuration. Consistent with its impact on hsdS inversions, deletion of dimA in adr^S438A ablated the O colony-dominant phenotype, but removing dimA in WT did not yield obvious impact on the colony phenotype (Fig 6D). These experiments showed that the PtvB-associated DimA exerts an essential role in linking NAM O-acetylation and DNA inversions in the cod locus.

We next characterized the physical interaction between PtvB and DimA using BATCH assay. The T18 fragment of adenylate cyclase was fused to DimA at either N- and C-terminus, and co-expressed with T25-tagged PtvB at its C terminus. While positive control displayed expected colony color and β-galactosidase activity, the DimA fusions did not display significant change in the presence of T25-PtvB (Fig 6E). This result suggested the lack of direct interaction between PtvB and DimA, which may be attributed to the extremely short cytoplasmic tail of PtvB.

PtvB is predicted to form a transmembrane complex with PtvA and PtvC, both of which are encoded by the same ptvABC operon (S5A Fig) [38]. Accordingly, the colony and β-galactosidase results demonstrated positive interactions of PtvB with PtvC but not with PtvA (Fig 6F). Since PtvC contains a relatively longer cytoplasmic domain (173 amino acids) than PtvB, we tested the likelihood that PtvC interacts with DimA. DimA with the N-terminal T18 showed positive colony color and significant level of β-galactosidase when being co-expressed with T25-tagged PtvC (Fig 6E). However, the C-terminal tagged DimA did not show obvious interaction with PtvC (S5B Fig). This result indicated that cytoplasmic protein DimA indirectly interacts with membrane-bound PtvB through its partner PtvC. Lastly, we verified the functional role of PtvC in regulating hsdS inversions by deleting ptvC in adr^S438A. The adr^S438A-ΔptvC mutant showed significant increase in hsdS_A1-carrying bacteria (Fig 6G) and O colonies (Fig 6H). These phenotype of adr^S438A-ΔptvC were restored to the parental level in ptvC revertant. It is obvious that the functional impact of ptvC deletion on hsdS inversions was much less pronounced than ptvB or dimA deletion. This result suggested that PtvC partially contributes to the PtvB-mediated regulation of hsdS inversions (Fig 6I).

The lack of NAM O-acetylation triggers the up-regulation of the invertase PsrA

To define how DimA modulates the PsrA-catalyzed inversions, we assessed its potential interaction with PsrA by BATCH assay. As shown in Fig 7A, the experiment did not show any significant direct interaction between these two proteins. This result indicated that DimA regulates hsdS inversions without direct interaction with PsrA.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Upregulation of psrA transcription in the absence of NAM O-acetylation.

(A) Detection of interactions between DimA and PsrA by bacterial two-hybrid assay. Colonies on the MacConkey/maltose plates (left panel) and β-galactosidase activity (right panel) are shown for each reporter strain as in Fig 6E. Significance between NC and experimental groups is presented. (B) Venn diagram of differentially expressed genes in WT and adr-lytA mutant relatively to adr^S438A. Genes were sorted by a cut-off value of fold change ≥1.5 and P_adj < 0.05. (C) Fold change of nine differentially expressed genes in the mRNA level of WT and adr-lytA mutant relatively to adr^S438A. (D) The transcription of psrA in the adr^S438A mutants was detected by qRT-PCR.

https://doi.org/10.1371/journal.ppat.1013286.g007

To further identify the functional connection between NAM O-acetylation and hsdS inversions, we determined pneumococcal transcriptomes in the presence (WT strain) or absence (adr^S438A) of NAM O-acetylation by RNA sequencing (RNA-seq). Pairwise comparison revealed 215 genes with at least 1.5-fold transcriptional change by the loss of the Adr acetyltransferase activity (Fig 7B and S5 Table). Likewise, RNA-seq comparison between adr^S438A and adr^S438A-lytA^S33Q-Y41A also identified 249 genes with at least 1.5-fold changes in transcription in the absence of the glycan-binding activity of LytA (Fig 7B and S6 Table). To simplify these complex data, we reasoned that the factor(s) linking NAM O-acetylation and hsdS inversions should be commonly affected by the loss of the acetyltransferase of Adr (WT vs. adr^S438A) and glycan-binding activity (adr^S438A vs. adr^S438A-lytA^S33Q-Y41A). Along this line, we identified nine genes whose transcription was significantly altered under the two conditions (Fig 7C). psrA and hsdS_C were the two genes with significant upregulation both in the lack of NAM O-acetylation and the glycan-binding activity of LytA. The psrA mRNA was increased by 1.7-fold in adr^S438A as compared with WT. To a greater extent, there were 5.5-fold more psrA transcripts in adr^S438A than that in adr^S438A-lytA^S33Q-Y41A. The expression of hsdS_C, which is located immediately upstream of psrA in the cod locus (Fig 2A), was up-regulated by at least 3-fold in adr^S438A. Additionally, we observed modest transcriptional down-regulation in seven genes, including hsdS_B. hsdS_B is located at the immediate downstream of psrA (Fig 2A), but is transcriptionally separated from psrA by a transcriptional terminator [25].

In the context of our previous observation that overexpression of psrA leads to enrichment of the transparent non-hsdS_A1 allelic configurations [29,44]. We next focused on verifying the expression of psrA under various strains by quantitative RT-PCR. As compared with WT, adr^S438A showed a modest but significant increase in psrA transcription (Fig 7D). In a consistent manner, psrA expression was reduced to the WT level in the absence of LytA glycan-binding activity. These data suggested that the LytA-mediated regulatory circuit modulates hsdS inversions at least in part by transcriptional upregulation of psrA. We also observed that the enhanced psrA transcription in adr^S438A was reduced to the WT level by removing dimA (Fig 7D), suggesting DimA somehow impacts the transcription of psrA in the absence of NAM O-acetylation. However, the linkage between DimA and the regulation of psrA remains to be defined. Together, these results suggested that, in response to the absence of NAM O-acetylation, LytA modulates epigenetic and cellular phases of S. pneumoniae at least in part by transcriptional upregulation of psrA through a multi-component signaling circuit.

Acetyl-CoA may be a linker between nutrient availability and NAM O-acetylation

Jones et al. have shown that NAM O-acetyltransferase A (OatA) of Staphylococcus aureus uses acetyl-CoA as the donor of the acetyl group to modify NAM [45]. Based on the high sequence identity between OatA and Adr, acetyl-CoA likely acts as the substrate of pneumococcal Adr. Acetyl-CoA of S. pneumoniae is primarily produced by pyruvate formate lyase encoded by pfl and pyruvate dehydrogenase complex (PDHC) (Fig 8A) [46,47]. PDHC is encoded by acoA, acoB, acoC and acoL (S6A Fig). We tested potential impact of acetyl-CoA availability on NAM O-acetylation in the absence of pyruvate formate lyase (Δpfl) or PDHC (ΔacoB). Our repeated attempts to construct a double mutant were unsuccessful, likely due to synthetic lethality. Both the ΔacoB and Δpfl strains showed significant reduction in acetyl-CoA (Fig 8B). In a consistent pattern, the level of NAM O-acetylation was also significantly reduced in ΔacoB and Δpfl, as compared with that in WT (35.7%) (Fig 8C). This result showed that cellular acetyl-CoA level greatly impacts NAM O-acetylation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. The impact of acetyl-CoA on NAM O-acetylation and colony phase.

(A) The synthesis of acetyl-CoA (AcCoA) and acetyl phosphate (Ac ~ P) from carbon sources. Pyruvate oxidase SpxB is responsible for the synthesis of Ac ~ P. The pyruvate dehydrogenase complex (PDHC) and pyruvate formate lyase (Pfl) are the functional enzymes to produce AcCoA under aerobic and anaerobic conditions, respectively. (B to E) The intracellular acetyl-CoA amount (B), NAM O-acetylation level (C), relative abundance of the hsdS_A1 mRNA (D), and colony ratios (E) of ΔacoB (TH17288) and Δpfl (TH17290) are shown as mean ± s.d. of 3-4 replicates in a representative experiment. (F) Growth curve of ST606 in TSB medium. (G to J) The intracellular acetyl-CoA amount (G), NAM O-acetylation level (H), the relative abundance of the hsdS_A1 mRNA (I), and colony ratio (J) of ST606 pneumococci at various time points cultured in TSB medium. The values at 3 h were used as references for statistical comparison. (K) Growth curve of ST606 in CDM with different concentrations of glucose. (L to M) The intracellular acetyl-CoA amount (L) and relative abundance of the hsdS_A1 mRNA (M) of ST606 bacteria collected at 1.5 h in CDM with different concentrations of glucose. The values at 20 g/L glucose were used as references for statistical comparison.

https://doi.org/10.1371/journal.ppat.1013286.g008

To test whether the cellular acetyl-CoA level can indirectly modulate hsdS inversions, we measured the ratio of hsdS_A1-carrying bacteria under the acetyl-CoA deficient conditions. The ΔacoB mutant showed significant reduction in the ratio of hsdS_A1-carrying bacteria as compared with WT (Fig 8D). To a lesser extent, the hsdS_A1-positive bacteria were also significantly decreased in Δpfl. Phenotypically, ΔacoB exhibited significant decrease in the proportion of O colonies, whereas the Δpfl did not show obvious phenotype (Fig 8E and S6A Fig). The more severe impact of acoB deletion on hsdS inversions agrees with the dominant role of PDHC in pneumococcal acetyl-CoA biosynthesis under aerobic conditions [46]. These results revealed that cellular acetyl-CoA availability substantially influence the extent of NAM O-acetylation, and indirectly modulates hsdS inversions.

As illustrated in Fig 8A, pyruvate is the main precursor of acetyl-CoA in S. pneumoniae [46,48]. It is thus reasonable to predict that the nutrient availability determines the level of cellular acetyl-CoA. To test this possibility, we cultured WT bacteria to the lag, exponential and stationary phases, and measured viable bacterial counts (colony forming unit, CFU/ml) (Fig 8F), and corresponding acetyl-CoA levels (Fig 8G) at various time points post inoculation. As compared with bacteria at the lag phase (hr 3), the cells at the logarithmic phase (4, 5.5 and 6.5 h) showed a maximal level of acetyl-CoA (Fig 8G). Cellular acetyl-CoA remained at a high level at the early death phase (8.5 h), and dropped at the late death phase (21 h). NAM O-acetylation was also found to change in a growth phase-dependent manner (Fig 8H). The growth phase-dependent dynamics of cellular acetyl-CoA and NAM O-acetylation indicates the correlation among nutrient availability, the level of cellular acetyl-CoA, and cell wall O-acetylation.

To determine the impact of growth phase on hsdS inversions, we quantified the relative proportion of hsdS_A1-carrying bacteria at different growth phases. Virtually bacteria possessed the hsdS_A1 allele in the cod locus at the early logarithmic phase (3 h) (Fig 8I). However, the proportional abundance was steadily decreased from 58.3% at 4 h to 47.0% at 21 h. At the phenotypic level, the proportion of O colonies gradually decrease from the mid logarithmic phase to the late dying period (Fig 8J). These data suggested that the extent of NAM O-acetylation reflects environmental conditions, including nutrient availability, reduced pH and accumulation of toxic metabolites.

We finally assessed the impact of carbon source on acetyl-CoA by growing pneumococci in a chemical defined medium (CDM) with various concentrations of glucose. Consistent with the glucose concentration-dependent growth (Fig 8K), there was a concentration-dependent reduction both in the level of acetyl-CoA abundance (Fig 8L) and the proportion of hsdS_A1-carrying bacteria (Fig 8M) at 1.5 h post inoculation, at which the impact of glucose concentration on bacterial growth became obvious. This change was not due to glucose-dependent change of adr transcription because the adr mRNA level did not show significant change under various glucose concentrations (S6B Fig). These data support the notion that cellular nutrient availability impacts the hsdS configurations.

Bacterial strains and cultivation

All the bacterial strains used in this study are summarized in S7 and S8 Tables. S. pneumoniae clinical isolate ST556 and its streptomycin-resistant derivative ST606 (ST556 rpsL1, containing a point mutation in ribosomal protein small subunit L) were used as the parental strains for mutant construction unless otherwise indicated [25]. E. coli strain DH5α for harboring specific plasmids and BL21(DE3) for producing recombinant proteins were bought from Solarbio company (Beijing, China). E. coli BTH101 was used as the reporter strain in the BATCH system [63]. Luria-Bertani (LB) broth was used for culturing E. coli strains. Pneumococci were cultured in a chemical-defined medium (CDM) with yeast extract (C + Y medium), tryptic soy broth (TSB), or on tryptic soy agar plate at 37ºC with 5% CO₂ [28]. CDM was prepared according to previous studies [62]. Before cultivation in CDM with different concentrations of glucose, pneumococci were incubated in C + Y broth to an optical density at 620 nm (OD₆₂₀) of 0.5, subsequently washed with PBS and resuspended in CDM to an initial OD₆₂₀ of 0.01. Appropriate antibiotics were added to the media when necessary.

Chemicals and reagents

All commercial culture media were purchased from BD (NJ, USA). All the premixed or ingredients of chemicals were purchased from Sigma (Shanghai, China) unless otherwise described. All reagents and commercial kits for molecular biology procedures were obtained from New England Biolabs (Beijing, China) unless otherwise described.

Bacterial mutagenesis

Pneumococcal mutants were constructed as described [28]. Markerless mutants were derived from streptomycin-resistant parental strains, including ST606 (556 rpsL1), TH6671 (P384 rpsL1), TH6675 (ST877 rpsL1), and TH6552 (the hsdS_A1-fixed strain in ST556 rpsL1 background) using JC1 (a modified Janus cassette) replacement method [28]. PgdA, Adr and LytA point mutants were established by in situ replacing JC1 sequence with the fusion PCR products of up- and down-stream sequences of target regions. In pgdA^D275N, the 823^rd G of pgdA was changed to A based on a previous study [64]; in adr^S438A, the 1,312^th T of adr gene was changed to A [30]; in lytA^E87A, the 260^th A of lytA was changed to C; in lytA^H133A, the 397^th C and 398^th A of lytA were changed to G; in lytA^S33Q-Y41A, 97^th T, 98^th C, 121^st T and 122^nd A for lytA were changed to C, A, G, and C, respectively [35]. The deletion mutants of myy0041, myy0606, myy0713, myy0734, myy1361, myy1406, myy1427, myy1585, and myy1950 were constructed by replacing their entire encoding regions with chloramphenicol resistance gene cat (amplified from the plasmid pIB166) [65]. The relevant plasmids, primers, and genetic manipulations are summarized in S9, S10 and S11 Tables, respectively.

Microscopic quantification of O and T colonies

The opacity of pneumococcal colonies (colony phase) was observed after incubation on catalase-TSA under 37˚C, 5% CO₂ for 17 h as described previously [28]. The number of O and T colonies in the central area on each plate (circling approximately 100 colonies) were quantified. The representative colonies of each strain on the catalase-TSA plate were photographed as the same time under a dissection microscope at magnification of 2 × 10 times [28].

RNA sequencing

RNA-seq was performed by Novogene Bioinformatics Technology (Tianjin, China) as described [28]. Significant difference of transcripts between ST606 and its derivatives was defined by a cut-off value of fold change ≥1.5 and P_adj < 0.05. Genes with less than 30 read counts were excluded.

Quantitative real-time reverse transcriptase PCR

The relative proportion of hsdS_A1-carrying bacteria in single populations of ST606 was determined by assessing the abundance of hsdS_A1 mRNA with quantitative real-time reverse transcriptase PCR (qRT-PCR) as described [28]. The relative proportions of bacteria carrying hsdS_A1, hsdS_A2, hsdS_A3, hsdS_A4, hsdS_A5, and hsdS_A6 in individual populations were determined by quantifying the abundance of their respective mRNAs using qRT-PCR. Primers used for qRT-PCR are listed in S12 Table.

Detection of hsdS gene configurations

The orientations of hsdS genes in the cod locus were determined by qPCR with the specific primer pairs (listed in S12 Table) targeting the three inverted repeats (IR1.1/1.2, IR2.1/2.2 and IR3.1/3.2) as described [26]. The relative abundance of each IR in different directions is presented as 2^-(∆CT). And the ratio between the relative abundance of forward and reverse sequence of each IR was calculated and described in percentage. The sum of the relative abundance of forward and reverse sequence of each IR was defined as 100%.

Genome sequencing

Genome sequence of the spontaneous mutant TH11857 (ST606 hk11^rev*) was determined by the next generation sequencing as described [66]. Genomic methylation was detected by single molecule real-time (SMRT) sequencing on PacBio RSII platform as described [28]. Genome sequencing and SMRT sequencing were performed by Novogene Bioinformatics Technology (Tianjin, China).

In vivo co-immunoprecipitation

The in vivo co-immunoprecipitation (Co-IP) was performed to identify proteins that potentially interact with LytA and PtvB as described, with minor modifications [67]. The following strains expressing Strep-tagged proteins in various strain backgrounds were constructed in ST606 (TH16167, Strep-lytA; TH17335, Strep-ptvB) and adr^S438A (TH16192, adr^S438A Strep-lytA; TH17336, adr^S438A Strep-ptvB) as described [28]. Strep-tag II was fused to its N-terminal of LytA and C-terminal of PtvB. These strains were cultured in TSB with 600 U/ml of catalase to an OD₆₂₀ of 0.6, and immediately cooled on ice. Bacteria were washed with buffer W (20 mM HEPES, pH 8.0, 100 mM NaCl) and subsequently treated with 1% formaldehyde to induce protein-protein cross-linking as described [68]. Tris-HCl (pH 8.0) was supplemented into bacterial suspension to the final concentration of 250 mM to terminate reaction, followed by wash using the buffer W. Bacteria were resuspended in pre-cooled lysis buffer (20 mM HEPES, pH 8.0, 100 mM NaCl, 1% Triton X-100) containing protease inhibitors, and was homogenized using the French Pressure Cell. The lysate was centrifuged at 4°C to remove cell debris. The supernatant was co-incubated with Strep-Tactin Sepharose 50% suspension (IBA, Germany); beads bound to target proteins were collected and washed by centrifugation and resuspension with buffer W. Bound proteins were eluted with buffer E (20 mM HEPES, pH 8.0, 100 mM NaCl, 5 mM desthiobiotin). The resulting proteins were detected by SDS-PAGE; protein bands were excised for liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis. The spectra from each run were searched against S. pneumoniae ST556 database using Proteome Discovery searching algorithm (v1.4) [69].

Bacterial adenylate cyclase-based two-hybrid assay

Protein interaction was assessed using BATCH system based on the interaction-mediated reconstitution of two complementary adenylate cyclase fragments as described [29]. Specifically, lytA was cloned into pKT25 and pKNT25 to generate plasmids encoding LytA with T25 fused to the N-terminal (T25-LytA) or C-terminal (LytA-T25). ptvB (lacking transmembrane region) and pcpA (lacking signal peptide) were cloned into pUT18C and pUT18 to generate vectors expressing ptvB and pcpA with T18. To study pairwise interactions among PtvA, PtvB and PtvC, ptvA and ptvB were cloned into pKT25, while ptvB, ptvC and ptvBC (ptvB and ptvC are co-transcribed) were inserted into pUT18C to establish vectors encoding PtvB, PtvC, and PtvBC with T18 fused to the N-terminal. To test pairwise interactions between DimA and PtvA, PtvB, and PtvC, ptvC gene was cloned into pKT25 to generate plasmid encoding PtvC with T25 fused to the N-terminal (T25-PtvC), while dimA was cloned into pUT18C to generate vectors encoding dimA with T18 fused to the N-terminal (T18-DimA) and C-terminal (DimA-T18). The reverse tagged pair was constructed similarly by cloning dimA into pKT25 to generate T25-DimA. To test pairwise interaction between PsrA and DimA, the psrA gene was cloned into pKT25 and pKNT25 to generate plasmid encoding PsrA with T25 fused to the N-terminal (T25-PsrA) or C- terminal (PsrA-T25), respectively.

pKT25-zip and pUT18C-zip respectively encoding the T25- and T18-fused leucine zipper (35-aa-long, derived from a yeast transcriptional activator protein GCN4) were used as the positive control [63,70]. Reporter E. coli BTH101 strain was inoculated on the MacConkey/maltose plate containing ampicillin (100 μg/ml), kanamycin (100 μg/ml), and IPTG (0.5 mM) for 4–8 days until the colonies of positive control became fuchsia. In addition to visual assessment, β-galactosidase activity of each reporter E. coli BTH101 strain was measured to quantify the functional complementation of T25 and T18 mediated by the interactions as described [70].

Western blotting

LytA and pyruvate oxidase SpxB were detected by Western blotting using rabbit anti-LytA and anti-SpxB antisera as described previously [71]. The density of each protein band was digitized with the ImageJ software (ImageJ 1.47v; National Institutes of Health) on the basis of its chemiluminescence intensity level. The relative protein abundance of LytA was calculated by normalizing the protein band density of LytA to that of SpxB.

Acetyl-CoA quantification

Acetyl-CoA was measured using Acetyl-CoA Content Assay Kit (Solarbio, China) according to the supplier’s instructions. Bacteria were harvested, washed twice with ice-cold PBS, and sonicated. After centrifugation, cell lysates were analyzed in a 96-well plate.

Vancomycin tolerance

Pneumococcal vancomycin tolerance was evaluated as described [38]. Briefly, bacteria were cultured to an OD₆₂₀ of 0.5 in THY medium. Culture aliquots were incubated in the presence or absence of 0.5 μg/ml vancomycin. Bacterial viability was measured by plating for CFUs at 3, 6, and 18 h post treatment.

Quantification of NAM and NAM O-acetylation

Cell wall materials were extracted according to the previous study [15]. The extent of NAM O-acetylation was assessed as described [72]. To quantify NAM O-acetylation, cell wall extracts were treated with 0.2 M NaOH to saponify the O-acetyl group. The produced acetate was further derivatized and quantified using Dionex Ultimate 3000 UPLC system coupled to a TSQ Quantiva Ultra triple-quadrupole mass spectrometer (Thermo Fisher, CA) (equipped with a heated electrospray ionization probe in negative ion mode) as described previously [73]. Data analysis and quantitation were performed by the software Xcalibur 3.0.63 (Thermo Fisher, CA). NAM in pneumococcal cell wall was measured as described previously [74].

Statistical analysis

All the original data were summarized in S13 Table and analyzed by GraphPad Prism. The ratio between O and T colonies and the proportion between forward and reverse IR-bound sequences in different pneumococcal strains were analyzed by two-sided Chi-square test, Yates’ continuity corrected Chi-square test, or Fisher’s exact test (by means). The difference of relative hsdS_A1 mRNA abundance, gene expression, and the relative activity of β-galactosidase were evaluated by two-tailed unpaired Student’s t test. Bacterial CFU values in vancomycin tolerance experiment were analyzed by two-way ANOVA. Differences with a P value of < 0.05 (*), < 0.01 (**), < 0.001(***) or < 0.0001 (****) are defined as statistically significant.