Gram-positive bacteria are surrounded by a thick cell wall consisting of a complex network of peptidoglycan with covalently attached glycopolymers that comprise a large family of structurally diverse molecules, including wall teichoic acid, mycobacterial arabinogalactans and capsular polysaccharides. From these, wall teichoic acid is perhaps the most widespread and best-studied molecule. This polyanionic, phosphate-rich glycopolymer is critical for functions including cell division, antibiotic resistance, metal ion homeostasis, phage-mediated horizontal gene transfer and protection of bacteria from host defense peptides and antimicrobial enzymes1,2. As such, these structures and their biosynthetic pathways are attractive targets for antibiotic development and vaccine design. Interestingly, many streptococci lack classical wall teichoic acid and instead express glycopolymers that are characterized by the presence of l-rhamnose (Rha)3. These structures comprise about 40–60% of the bacterial cell wall mass, and are historically used for serological grouping of streptococci3. The glycopolymers of two human streptococcal pathogens, S. pyogenes or group A Streptococcus (GAS) and S.mutans, are respectively referred to as group A carbohydrate (GAC) and serotype c carbohydrate (SCC). These glycopolymers share a characteristic (→3)α-Rha(1→2)α-Rha(1→) polyrhamnose backbone, but are serologically distinguished based on their specific glycosyl side-chain residues, that is N-acetyl-β-d-glucosamine (GlcNAc) in GAC4 and α-glucose (Glc) in SCC5. GAC and SCC play key roles in cell morphology and division6, resistance to certain cell wall-targeting antibiotics7, biofilm formation8 and pathogenesis of GAS and S.mutans9,10,11. Importantly for both pathogens, GAC and SCC have been evaluated as vaccine antigens. Immunization with GAC or SCC induces opsonophagocytic antibodies that enhance killing of GAS and S.mutans, respectively5,12,13. In addition, GAC has proven efficacious as a vaccine antigen through active immunization in mice12,13.

The GAC and SCC biosynthetic pathways are encoded by 12-gene clusters3,11, herein designated as gacABCDEFGHIJKL and sccABCDEFGHMNPQ (Fig. 1a), respectively. The first seven genes in both clusters are conserved in many streptococcal species, and they participate in polyrhamnose backbone synthesis and transport14. In GAS, gacI, gacJ, gacK and gacL encode the machinery to generate and add the GlcNAc side-chain to the polyrhamnose backbone11,15, whereas the genes required for Glc side-chain generation are not clearly identified in S.mutans. In addition to these streptococcal species, similar gene clusters are present in a wide variety of streptococcal, lactococcal and enterococcal species3.

Fig. 1: GacH homologs are required for hGIIA bactericidal activity against GAS and S.mutans.
figure 1

a, Schematic representation of GAC and SCC biosynthetic gene clusters. SCC biosynthesis encoding gene cluster smu.824–835 was renamed sccABCDEFGHMNPQ. Sequence identity (%) between encoded homologous proteins is indicated. Sequences of GAS 5005 and S.mutans UA159 were used for identity comparison. bd, Identification of gacH in transposon sequencing (Tn-seq) screen and validation for hGIIA resistance. Transposon gene locus tags of the 47 hGIIA-resistant mutants after exposure of Krmit mutant transposon library to lethal concentrations of hGIIA (b). Susceptibility of GAS 5448 and S.mutans to hGIIA concentration range following deletion of gacH in GAS 5448 (c) and the gacH-homologous gene sccH (d), respectively. Symbols and error bars represent the mean and s.d., respectively (n = 3 biologically independent replicates and each replicate represents three technical replicates). P values were calculated by two-way analysis of variance (ANOVA). Bonferroni multiple comparison test was used to statistically compare multiple groups. *P < 0.05; **P < 0.01; ***P < 0.001. The precise P values are listed in Supplementary Table 2.

In addition to the polyrhamnose biosynthesis genes, the GAC and SCC biosynthetic clusters contain further conserved genes of unknown function, gacH and sccH, respectively, which are annotated as putative glycerol phosphate (GroP) transferases. Recently, we employed the Krmit GAS transposon mutant library16 and identified gacI and gacH as genes that confer bacterial sensitivity to human group IIA-secreted phospholipase A2 (hGIIA)17, an important bactericidal protein of the innate immune system against Gram-positive pathogens18. Complementary to that study, we now identify gacH as the only valid hit when the Krmit library was exposed to a lethal concentration of hGIIA. Interestingly, gacH was also identified as a gene providing resistance to zinc toxicity, which is a mechanism deployed by neutrophils to kill GAS19. In pursuit of the underlying mechanism, we have characterized the function of GacH at the genetic, biochemical and structural level. Our study identifies a previously overlooked GroP modification on both GAC and SCC, and pinpoints GacH homologs as the enzymes responsible for the respective GroP modifications.


GacH and SccH confer sensitivity to hGIIA

We previously identified gacH in a GAS Tn-seq transposon library screen as a potential hGIIA susceptibility gene17. To identify additional resistant mutants, we exposed the Krmit GAS transposon library16 to a lethal concentration of hGIIA. PCR sequencing identified that 43% of the recovered mutants had a transposon insertion in gacH, and 26% in M5005_Spy_1390 (Fig. 1b and Supplementary Table 1). M5005_Spy_1390 was identified in the initial susceptibility screen as an artifact due to biased transposon insertions17 and was not investigated further. To validate our finding for gacH, we generated a gacH deletion mutant in a GAS serotype M1T1 clone 5448, creating 5448ΔgacH. Deletion of gacH rendered GAS resistant to hGIIA over the tested concentration range, and was reversed by complementation with gacH on an expression plasmid (5448ΔgacH:pgacH, Fig. 1c). This gacH-dependent hGIIA resistance was also observed in two different GAS backgrounds, 2221 (M1T1 clone strain) and 5005 (clinical covS mutant isolate of M1T1 strain) (Supplementary Fig. 1a,b), demonstrating that the effect is conserved across GAS strains of the M1T1 background and independent of CovRS status—a two-component system that regulates about 15% of the genes in this bacterium20.

To investigate whether hGIIA susceptibility was also influenced by gacH homologs in other streptococci, we deleted sccH in S.mutans (SMU) serotype c strain Xc, creating SMUΔsccH. SMUΔsccH was completely resistant to the tested hGIIA concentrations (Fig. 1d) and susceptibility was restored to wild-type levels by plasmid-expressed sccH. However, heterologous expression of gacH in SMUΔsccH did not restore the phenotype (Supplementary Fig. 1c), suggesting that the enzymes may target different substrates. Altogether, our data indicate that deletion of gacH homologs renders streptococci more resistant to hGIIA bactericidal activity and that GacH function is species-specific.

GacH and SccH provide protection from zinc toxicity

Recent evidence indicates that neutrophils deploy zinc poisoning as an antimicrobial strategy against GAS during phagocytosis19. To resist Zn2+ toxicity, GAS expresses a zinc efflux system encoded by czcD19. To search for additional Zn2+-resistance genes, we performed a Tn-seq screen of the GAS Krmit transposon library16 using two Zn2+ concentrations, 10 and 20 µM, selected based on growth inhibition analysis (Supplementary Fig. 2a). Genomic DNA for Tn-seq analysis was collected after T2 and T3 passages (Supplementary Fig. 2b). In addition to the expected importance of czcD, gacI and gacH transposon insertions were significantly reduced in the library (P < 0.05) after growth with 20 µM Zn2+ in both T2 and T3 passages compared to untreated controls, indicating that these genes provide resistance against Zn2+ toxicity (Fig. 2a–d).

Fig. 2: Deletion of gacI and gacH renders GAS susceptible to Zn2+.
figure 2

ad, Tn-seq volcano plots showing representation of czcD, gacH and gacI in GAS Krmit transposon library screens for Zn2+ tolerance. log2 fold change (log2 FC) in fitness was plotted against adjusted P values from Tn-seq analyses through an established pipeline using EdgeR and DEseq2 (n = 4 biologically independent replicates used for analysis)47,48,49. The outline of the experiment is shown in Supplementary Fig. 2b. Tn-seq screens of the transposon library were conducted using 10 µM Zn2+ at T2(a), 10 µM Zn2+ at T3 (b), 20 µM Zn2+ at T2 (c) and 20 µM Zn2+ at T3(d). e,f, Zn2+ sensitivity as tested by drop-test assay using strains GAS 5448 wild type, 5448ΔgacH and 5448ΔgacH:pgacH (e); and GAS 5448 wild type, 5448ΔgacI and 5448ΔgacI:gacI (f). 5448ΔczcD was included as a positive control in both panels. Each drop-test assay experiment was performed independently at least three times and yielded the same results.

To validate our findings, we grew 5448ΔgacH and 5448ΔgacI11 on solid rich medium supplied with different Zn2+ concentrations (Fig. 2e,f). Both mutants showed reduced growth in the medium supplied with 1.25 mM Zn2+, which was restored following complementation with the respective genes (Fig. 2e,f). Again, we checked for function conservation by extending our experiments to S.mutans. Indeed, SMUΔsccH was more sensitive to Zn2+ in comparison to the parental strain and the phenotype could be restored by sccH but not by gacH (Supplementary Fig. 3). Hence, our results provide strong evidence that GacH and SccH are important in protecting streptococci from Zn2+ toxicity.

Crystal structure reveals that GacH is a GroP transferase

GacH is predicted to contain 11 N-terminal transmembrane segments and an extracellular C-terminal catalytic domain (eGacH). To test the hypothesis that GacH is a GroP transferase, eGacH was expressed and purified from Escherichia coli. Its crystal structure was determined in apo form (Protein Data Bank (PDB) No. 5U9Z) at 2.0 Å resolution (Fig. 3a,b) and in complex with GroP (PDB No. 6DGM) at 1.49 Å resolution (Fig. 3c). The apo- and GroP-containing eGacH structures belong to different crystal forms, with two molecules in the asymmetric unit. Analysis of the dimer interface and other crystal contacts revealed that the dimer interface has the largest surface of all crystal contacts (1809 and 1894 Å2 in the two crystal forms). However, it is scored below the stable complex formation criteria and recombinant eGacH behaves as a monomer in solution. The structures of the apo- and GroP-bound eGacH monomers are very similar, with root mean square deviation of 0.3 Å for 380 superimposed Cα atoms, as well as between the non-crystallographic copies.

Fig. 3: Structure of eGacH.
figure 3

a, Predicted topology of GacH showing 11 transmembrane helices and structure of extracellular domain with the enzymatic active site oriented toward the cell membrane. b, Structure of apo eGacH viewed at the active site with the Mn2+ ion shown as a violet sphere. c, A close-up view of the active site GacH crystal structure in complex with sn-Gro-1-P.

The extracellular C-terminal catalytic domain has an α/β core structure that is characteristic for the sulfatase protein family, with the closest similarity to lipoteichoic acid (LTA) synthase LtaS21,22 (Supplementary Figs. 4a,b and 5) and LTA primase LtaP23 (Supplementary Table 3). LtaS and LtaP are GroP transferases that participate in biosynthesis of LTA, a crucial constituent of Gram-positive cell envelopes, consisting of a poly(GroP) backbone linked to a glycolipid membrane anchor24. The catalytic site of eGacH contains a Mn2+ ion coordinated by residues E488, T530, D711 and H712, equivalent to residues E255, T300, D475 and H476 of a C-terminal extracellular domain of LtaS (eLtaS) from Staphylococcus aureus (Fig. 3c and Supplementary Figs. 4c,d and 5). The structure of eGacH in complex with GroP revealed the position of the ligand in the active site with the phosphoryl group oriented towards the Mn2+ ion, and coordinated by residues G529, T530 and H650 (Fig. 3c). The glycerol 2- and 3-hydroxyl groups form hydrogen bonds with side-chains of residues R589, H580 and N586. The positions of GroP and coordinating residues are similar in eGacH and S. aureus eLtaS structures. For example, the glycerol moiety forms hydrogen bonds with residues H580 and R589 in GacH and equivalent residues H347 and R356 in S. aureus eLtaS (Fig. 3c and Supplementary Fig. 4c,d)21. Thus, the structure of eGacH in complex with GroP is consistent with the idea that GacH and LtaS use related catalytic mechanisms to transfer GroP to substrates.

To functionally assess the requirement of the catalytic residues, we examined the bactericidal activity of hGIIA in 5448ΔgacH and SMUΔsccH expressing catalytically inactive versions of gacH and sccH, in which the active site T530 and T533 codons were replaced by alanine, respectively (Supplementary Fig. 6). The non-functional gacH and sccH did not restore hGIIA susceptibility (Fig. 1c,d), indicating that the GroP transferase activity of the gacH and sccH gene products is required for the observed hGIIA-dependent phenotypes.

GacH cleaves phosphatidylglycerol to release GroP

Experimental evidence suggests that LtaS utilizes the GroP head group of the membrane lipid phosphatidylglycerol as donor for poly(GroP) backbone biosynthesis, liberating diacylglycerol24,25. To assess whether GacH also catalyzes the cleavage of phosphatidylglycerol to yield GroP for a transfer reaction, we performed an in vitro experiment employing the eGacH protein and a fluorescently labeled artificial substrate, NBD-phosphatidylglycerol. Incubation of eGacH with NBD-phosphatidylglycerol yielded a fluorescent product (Supplementary Fig. 7a) with the same mobility on silica gel thin-layer chromatography as NBD-diacylglycerol, which was obtained from NBD-phosphatidylglycerol by enzymatic cleavage with phospholipase C from Bacillus cereus. Furthermore, the eGacH product recovered from the silica gel plate and analyzed by liquid chromatography–mass spectrometry (LC–MS) yielded a spectrum consistent with NBD-diacylglycerol and identical to that of the phospholipase C product (Supplementary Fig. 7b–d). The formation of NBD-diacylglycerol by eGacH required the active-site residues, since the eGacH variant, eGacH-T530A, was not active in the assay (Supplementary Fig. 7a). These observations are consistent with the assignment of GacH as a GroP transferase.

GacH homologs decorate glycopolymers with GroP

Phylogenetic analysis of either the full-length or extracellular domains of GacH homologs and LtaS-related proteins revealed that these proteins fall into distinct clades of GroP transferases, suggesting that the proteins may transfer GroP to different substrates (Supplementary Fig. 8). To assess whether gacH homologs modify the respective streptococcal glycopolymers with GroP, we enzymatically released GAC and SCC from purified cell walls from GAS and S.mutans strains. Subsequently, the enriched polysaccharide preparations were analyzed for glycerol and phosphate. Hydrolysis with HCl released a large amount of glycerol from GAC and SCC isolated from wild-type bacteria (Fig. 4a,b and Supplementary Fig. 9a). Furthermore, we detected high levels of inorganic phosphate after incubation of these acid-treated samples with alkaline phosphatase (Fig. 4a,b and Supplementary Fig. 9a), which was not detected when intact GAC was treated with alkaline phosphatase (Supplementary Fig. 9b,c). This indicates that the phosphoryl moiety is present as a phosphodiester, consistent with its identification as GroP. In contrast to wild-type GAC and SCC, the glycopolymers isolated from 5448ΔgacH, 5005ΔgacH and SMUΔsccH contained a markedly reduced amount of glycerol and phosphate (Fig. 4a,b and Supplementary Fig. 9a), which was restored only by complementation with wild-type gacH, but not gacH-T530A, for GAS (Fig. 4a) or plasmid-expressed sccH for SMU (Fig. 4b).

Fig. 4: GacH and SccH modify their respective glycopolymers with sn-Gro-1-P.
figure 4

a,b, Analysis of glycerol and phosphate content in GAC and SCC isolated from GAS 5448 wild type, ΔgacI, ΔgacH and ΔgacH complemented with native gacH or a catalytically inactivated version of gacH (gacH-T530A) (a), and S.mutans wild type, ΔsccH and ΔsccH complemented with sccH or gacH (b). The concentrations of phosphate and glycerol are presented relative to the wild-type strain. Bars and error bars represent the average and s.d., respectively (n = 3 biologically independent samples). P values were calculated and adjusted by two-way ANOVA and Bonferroni’s multiple comparison test. c, Diethylaminoethanol (DEAE)-Sephacel elution profile of GAC isolated from ~90 mg of GAS cell wall. Fractions were analyzed for carbohydrate (•) and phosphate (O). df, Identification of the enantiomeric form of GroP associated with GAC. d, The GroP isomers were recovered from GAC following alkaline hydrolysis and separated by liquid chromatography as outlined in Methods. The elution positions corresponding to standard Gro-2-P and sn-Gro-1-P/sn-Gro-3-P are indicated by the arrows. LC–MS analysis identified two extracted ion chromatogram peaks for the molecular GroP ion mass/charge number (m/z) 171.004 [M-H], which eluted at 9.48 (e) and 9.89 min (f). Based on the accurate mass and retention times, these two peaks were assigned as Gro-2-P and sn-Gro-1-P/sn-Gro-3-P, respectively, by comparison with authentic chemical standards. Experiments depicted in cf were performed independently twice and yielded the same results.

In accordance with our functional data, expression of gacH did not restore the glycerol and phosphate levels in SCC of SMUΔsccH (Fig. 4b). This suggested that GroP modifications might involve the species-specific side-chains (Glc versus GlcNAc) rather than the identical polyrhamnose backbone. Indeed, the glycerol and phosphate contents in GAC isolated from the GlcNAc-deficient mutant, 5448ΔgacI, were significantly reduced (Fig. 4a). Importantly, analysis of GAS strains for total carbohydrate, phospholipid and phosphatidylglycerol contents established that deletion of gacH had no effect on these components (Supplementary Fig. 10a–c). Furthermore, the semi-quantitative immuno-dot-blot analysis of GAS strains with anti-GAC antibodies demonstrated that the absolute amount of GAC is not affected by gacH deletion (Supplementary Fig. 10d). Analysis of the glycosyl composition of purified cell walls demonstrated that the absence of GacH and SccH did not affect the Rha/GlcNAc and Rha/Glc ratios, respectively (Supplementary Fig. 10e,f).

To provide further evidence that GAC is modified with GroP, GAC samples were subjected to alkaline hydrolysis to release GroP, as described in ref. 26, and the hydrolysate was analyzed by LC–MS for high-molecular weight fragment ions arising from GroP. Compared to wild type, deletion of either gacH or gacI markedly reduced the levels of GroP in GAC (Supplementary Fig. 11). Complementation of 5448ΔgacH with native gacH, but not with inactive gacH-T530A, fully restored GroP levels (Supplementary Fig. 11). Thus, the differences in GroP content for the gacH and gacI deletion mutants are consistent with a role for GacH in modification of the GlcNAc side-chain of GAC with GroP.

To show that GroP is attached directly to GAC, the wild-type GAC was further purified by a combination of size-exclusion and ion-exchange chromatography (Fig. 4c and Supplementary Fig. 12a). The majority of the rhamnose- and phosphate-containing material was bound to the ion-exchange column and eluted as a single coincident peak (Fig. 4c). The GAC purified from 5005ΔgacH did not bind to the column (Supplementary Fig. 12b). Interestingly, the 5005ΔgacH GAC does appear to contain a small amount of phosphate that may have arisen from the phosphodiester bond linking GAC to peptidoglycan. Taken together, our data directly support the conclusion that GAC is modified with GroP donated by GacH.

GacH decorates GAC with sn-Gro-1-P

To assess which GroP enantiomer was incorporated in GAC, GroP liberated from purified GAC by alkaline hydrolysis was analyzed further (Supplementary Fig. 13). As described in detail in ref. 26, if GAC is modified by sn-Gro-1-P, alkaline hydrolysis of the phosphodiester bond should result in the formation of a mixture of sn-Gro-1-P and Gro-2-P, whereas modification by sn-Gro-3-P would yield a mixture of sn-Gro-3-P and Gro-2-P26. As expected, LC–MS analysis of GAC-derived GroP revealed the presence of two GroP isomers of approximately equal proportions, with LC retention times and major high-molecular weight ions consistent with standard sn-Gro-1-P/sn-Gro-3-P and Gro-2-P (Fig. 4d–f and Supplementary Fig. 14). The recovered GroP was characterized further by enzymatic analysis using an sn-Gro-3-P assay kit. Under reaction conditions in which sn-Gro-3-P standard produced a robust enzymatic signal, incubation with an equal amount of either sn-Gro-1-P or the unknown GroP resulted in negligible activity (Supplementary Fig. 15). When sn-Gro-3-P was mixed with an equal amount of either sn-Gro-1-P or the unknown mixture of GroP isomers, 85.8 and 90.0% of the activity detected with the standard sn-Gro-3-P alone was found, confirming that the negative result using the unknown mixture was not due to the presence of an inhibitory compound in GroP preparation. Taken together, our results indicate that GacH decorates GAC with sn-Gro-1-P, which is most probably derived from phosphatidylglycerol.

GroP is attached to the C6 hydroxyl group of GlcNAc

To unambiguously establish the presence and location of GroP in GAC, the glycopolymer was isolated from wild-type GAS and analyzed by nuclear magnetic resonance (NMR) (Fig. 5a, Supplementary Table 4 and Supplementary Figs. 16 and 17). The details of NMR analysis are described in Supplementary Notes. Wild type GAC is partially substituted by a GroP residue at O6 of the side-chain β-d-GlcpNAc residue; based on integration of the cross-peaks for the anomeric resonances in the 1H,13C-HSQC NMR spectrum, the GAC preparation carries GroP groups to ~25% of the GlcNAc residues. To validate the two-dimensional (2D) NMR results, a triple-resonance 1H,13C,31P NMR experiment based on through-bond 1JHC as well as 2JCP and 3JCP correlations27 was carried out. The three-dimensional (3D) NMR experiments revealed the 1H NMR chemical shifts of H5ʹ and the two H6ʹ protons of the β-d-GlcpNAc residue, as well as the two H1 protons and H2 of the Gro residue that all correlated to 13C nuclei (Fig. 5b). The 13C NMR chemical shifts of C5ʹ and C6ʹ of the β-d-GlcpNAc residue, as well as C1 and C2 of the Gro residue, all correlated to the 31P nucleus (Fig. 5b), and the above protons correlated to the 31P nucleus (Fig. 5b). Taking into consideration the GacH-mediated mechanism of GAC modification by GroP, as well as the biochemical experiments carried out herein, the substituent at O6 of β-d-GlcpNAc is an sn-Gro-1-P group (Fig. 5c).

Fig. 5: NMR analysis confirms the presence of GroP on the C6 GlcNAc hydroxymethyl group of GAC.
figure 5

a,b, Selected regions of NMR spectra of GAC. a, Multiplicity-edited 1H,13C-HSQC in which methylene groups have opposite phase and are shown in red (center box); 1H,13C-HSQC-TOCSY with an isotropic mixing time of 120 ms (left box); 1H,13C-HMBC with a mixing time of 90 ms (top box); 1H,31P-hetero-TOCSY with an isotropic mixing time of 80 ms (bottom box). b, 1H,13C-plane (center box); 13C, 31P-plane using a nominal nJCP value of 5 Hz (left box); 1H, 31P-plane (bottom box) of a through-bond 3D 1H,13C,13P NMR experiment. Cross-peaks are annotated as GIII corresponding to the GlcNAc residue, GIIIʹ being the GroP-substituted GlcNAc residue and Gro the glycerol residue. NMR chemical shifts of 1H (horizontal axis), 13C (left axis) and 31P (right axis and left box in b) are given in ppm. Experiments depicted in a,b were performed independently three times and yielded the same results. c, Schematic structure of the GAC repeating unit consisting of → 3)-α-l-Rhap-(1 → 2)[β-d-GlcpNAc6P(S)Gro-(1 → 3)]-α-l-Rhap-(1 → . d, The mechanism and roles of GroP cell wall modification in streptococci.


In Gram-positive bacteria, many peptidoglycan-attached glycopolymers contain negatively charged groups in the repeating units2. Previous detailed studies deduced the chemical structure of glycopolymers from GAS and S.mutans3,4,5. However, none identified anionic groups in these structures except for one study that reported the presence of glycerol and phosphate in GAC and proposed that this GroP is part of the phosphodiester linkage connecting GAC to peptidoglycan28. Similarly, other reports identified substantial concentrations of phosphate in the glycopolymers isolated from a number of streptococcal species29,30,31. Phosphate detection was either disregarded as contamination with LTA29 or further analyzed using 1H NMR or 13C NMR methods5,31,32 that do not directly detect phosphoryl moieties in polysaccharides. In our report, we unambiguously confirm that the glycopolymers of GAS and S.mutans are in fact polyanionic molecules through decoration of their respective glycan side-chains with GroP (Fig. 5c,d).

We identified and structurally characterized an alternative class of GroP transferase enzymes, represented by GacH, which modifies GAC with GroP in the human pathogen GAS. According to our phylogenetic analysis, GacH homologs are present in many streptococci (Supplementary Fig. 8), suggesting that these bacteria express glycopolymers with GroP-modified side-chains, as we have demonstrated here for S.mutans. GacH belongs to the alkaline phosphatase superfamily of which two GroP transferases involved in LTA synthesis, LtaS and LtaP, have been biochemically and structurally characterized21,22,23,33. LtaS and LtaP are membrane proteins that use the membrane lipid phosphatidylglycerol as the GroP donor for the transfer reaction25. Our structural analysis of GacH in complex with GroP indicates that the T530 residue participates in the formation of a GroP-enzyme intermediate similar to observations in LtaS, where the GroP molecule is complexed in the active site threonine residue which functions as a nucleophile in phosphatidylglycerol hydrolysis21,22,23. The importance of this residue was also confirmed functionally, since complementation of gacH mutant strains with a T530A gacH variant could not restore GroP content in GAC and hGIIA sensitivity to wild-type levels. The observations, that the extracellular domain of GacH cleaves phosphatidylglycerol and the GroP in GAC is the sn-Gro-1-P enantiomer, strongly suggest that GacH uses phosphatidylglycerol as its donor substrate for the transfer reaction, similar to LtaS (Fig. 5d).

In Gram-positive bacteria, the modification of teichoic acids with d-alanine provides resistance against antibiotics, cationic antimicrobial peptides and small bactericidal enzymes including hGIIA, and affects Mg2+ ion scavenging1,2,34. It has been assumed that incorporation of positively charged d-alanine into teichoic acids decreases negative bacterial surface charge resulting in reduced initial binding of cationic antimicrobial peptides to the bacterial surface due to ionic repulsion35,36. Our study demonstrates that addition of the negatively charged GroP group to glycopolymers protects streptococci from zinc toxicity but also renders bacteria more sensitive to hGIIA activity.

A large body of evidence indicates that phagocytic cells utilize Zn2+ intoxication to suppress the intracellular survival of bacteria37. Zinc is essential as a key catalytic or structural element for a wide variety of proteins and its concentration needs to be maintained at a specific level, which requires sophisticated systems for uptake and efflux of metal ions. Hence elevated levels of zinc in the cytosol result in cellular toxicity19, which for GAS is due to inhibition of central carbon metabolism38. One mechanism of microbial susceptibility to zinc toxicity is mediated by extracellular competition of Zn2+ for Mn2+ transport and thereby mediates toxicity by impairing acquisition of Mn2+, the essential nutrient metal39. Accordingly, the phenotypes of our GroP and GlcNAc side-chain-deficient mutants could be explained either by ‘trapping' of Zn2+ in the wild-type cell wall by GroP, or the increased Mn2+-binding capacity of GroP-modified bacterial cell wall. which has been proposed to act as the conduit for the trafficking of mono- and divalent cations to the membrane34.

Charge-dependent mechanisms probably underlie the increased hGIIA susceptibility of GAS and S.mutans expressing GroP-modified glycopolymers. hGIIA is a highly cationic enzyme that catalyzes the hydrolysis of bacterial phosphatidylglycerol40,41, ultimately leading to bacterial death through lysis. Traversal of this bactericidal enzyme through the Gram-positive cell wall to the plasma membrane is charge-dependent. Indeed, the absence of d-alanine modifications in teichoic acids severely compromises S.aureus survival when challenged with hGIIA41,42. Similarly, the GacH/SccH-dependent GroP modifications on glycopolymers are required for hGIIA to exert its bactericidal effect against GAS and S.mutans, respectively. We have previously demonstrated that loss of the entire GlcNAc GAC side-chain strongly hampers hGIIA trafficking through the GAS cell wall, with a minor contribution of reduced hGIIA binding to the cell surface17. Since GroP modifications were also lost in the GlcNAc side-chain-deficient mutant, 5448ΔgacI, described in this study, we now assume that the mechanisms of the hGIIA-dependent phenotype are similar in the gacI and gacH mutants.

Another very important aspect of our study is the identification of an alternative, potentially antigenic, epitope on the surface of streptococci. GAS is associated with numerous mild to life-threatening invasive diseases43 and is also causative of post-infectious sequelae, including rheumatic heart disease43. In particular, the invasive manifestations and post-infectious sequelae are difficult to treat with antibiotics and a GAS vaccine is urgently needed to combat these neglected diseases. The GAC is an attractive candidate for GAS vaccine development, due to its conserved expression in all GAS serotypes and the absence of the constitutive component of GAC, Rha, in humans12,13. However, it has been proposed that the GAC GlcNAc side-chain may elicit cross-reactive antibodies relevant to the pathogenesis of rheumatic fever and rheumatic heart disease44,45,46. Moreover, persistence of anti-GAC and anti-GlcNAc antibodies is associated with a poor prognosis in rheumatic heart disease45. These clinical associations and the lack of understanding of the pathogenesis of GAS post-infectious rheumatic heart disease have hampered progress in the development of GAC-based vaccines against GAS. However, the GAC GlcNAc decorated with GroP might be an attractive candidate for GAS vaccine development because GroP-modified GlcNAc represents a unique epitope that is absent from human tissues. Thus, our study has implications for the design of a safe and effective vaccine against this important human pathogen for which a vaccine is currently lacking.


Bacterial strains, growth conditions and media

All plasmids, strains and primers used in this study are listed in Supplementary Tables 5 and 6. GAS and S.mutans strains were grown in Todd–Hewitt broth supplemented with 1% yeast extract (THY) without aeration at 37 °C. S.mutans plates were grown with 5% CO2. For hGIIA-mediated killing experiments, S.mutans strains were grown in Todd–Hewitt broth without yeast extract and with 5% CO2. E. coli strains were grown in Lysogeny broth medium or on Lysogeny broth agar plates at 37 °C. When required, antibiotics were included at the following concentrations: ampicillin at 100 μg ml–1 for E. coli; streptomycin at 100 μg ml–1 for E. coli; erythromycin (Erm) at 500 μg ml–1 for E. coli, 5 μg ml–1 for GAS and 10 μg ml–1 for S.mutans; chloramphenicol (CAT) at 10 µg ml–1 for E. coli and 2 µg ml–1 for GAS and S.mutans; spectinomycin at 200 μg ml–1 for E. coli, 100 μg ml–1 for GAS and 500 μg ml–1 for S.mutans; and kanamycin at 300 µg ml–1 for GAS.

To identify genes providing resistance against Zn2+ toxicity, Roswell Park Memorial Institute (RPMI) 1640 (without glucose) (Gibco) was supplemented with guanine, adenine and uracil at a concentration of 25 µg ml–1 each, d-glucose at a concentration of 0.5% w/v and HEPES at 50 mM. Vitamins were provided by BME Vitamins 100× (Sigma, No. B6891).

Genetic manipulations

Plasmids were transformed into GAS and S.mutans by electroporation or natural transformation as described previously6. All constructs and mutants were confirmed by PCR and sequencing analysis (Eurofins MWG Operon and Macrogen).

Genetic manipulation of GAS: To construct 5005ΔgacH and 2221ΔgacH, 5005 chromosomal DNA was used as a template for amplification of two DNA fragments using primer pairs 5005-f/gacHdel-r and gacHdel-f/5005-r, which were fused and amplified using a PCR overlap method50 with primer pair 5005-f/5005-r to create the deletion of gacH. The PCR product was digested with BamHI and XhoI and ligated into BamHI/SalI-digested plasmid pBBL740, transformed into 5005 and 2221, and CAT-resistant colonies were selected on THY agar plates. Several potential double-crossover mutants were selected as previously described51.

To construct the plasmid for in cis complementation of 5005ΔgacH, 5005 chromosomal DNA was used as a template for amplification of gacH using the primer pair 5005-f/5005-r, which was cloned in pBBL740 through restriction-ligation using BamHI and XhoI. The plasmid was transformed into the 5005ΔgacH strain, and CAT-resistant colonies were selected on THY agar plates. Double-crossover mutants were selected as described above.

To construct 5448ΔgacH, GAS 5448 chromosomal DNA was used to amplify up- and downstream regions flanking gacH using primer pairs 5448-f/5448CAT-r and 5448CAT-f/5448-r. Primers 5448CAT-f and 5448CAT-r contain 25 base pair (bp) extensions complementary to the CAT resistance cassette. Up- and downstream were fused to the CAT cassette using 5448-f/5448-r, and cloned into pHY304 through digestion-ligation using XhoI and HindIII, yielding plasmid pHY304ΔgacH. After plasmid transformation into 5448, the double-crossover mutant 5448ΔgacH was selected as previously described15.

To complement 5448ΔgacH, gacH was amplified from 5448 chromosomal DNA using primer pair gacH-EcoRI-f/gacH-BglII-r, digested using EcoRI/BglII and ligated into EcoRI/BglII-digested pDCerm, yielding pgacH_erm. To make a catalytically inactive variant of gacH, the mutation T550A was introduced into pgacH_erm using Gibson Assembly site-directed mutagenesis with the primers gacH-T530A-F, gacH-T530A-R, repB-isoF and repB-isoR. The plasmids were transformed into 5448ΔgacH and selected for Erm resistance on THY agar plates. Transformation was confirmed by PCR, yielding strains 5448ΔgacH:pgacH and 5448ΔgacH:pgacH-T530A, respectively.

To construct SMUΔsccH, S.mutans Xc chromosomal DNA was used to amplify up- and downstream region flanking using primer pairs sccH-f/sccH-erm-r and sccH-erm-f /sccH-r. Primers sccH-erm-f and sccH-erm-r contained 25 bp extensions complementary to the Erm resistance cassette. Up- and downstream PCR fragments were mixed with the Erm cassette and amplified as a single PCR fragment using primer pair sccH-f/sccH-r. The sccH knockout construct was transformed into S.mutans as described previously6. Erm-resistant single colonies were picked and checked for deletion of sccH and integration of Erm cassette by PCR, resulting in SMUΔsccH. For complementation, sccH and gacH were amplified from S.mutans Xc and GAS 5448 chromosomal DNA, respectively, using primer pairs sccH-EcoRI-f/sccH-BglII-r and gacH-EcoRI-f/gacH-BglII-r. The PCR products were digested with EcoRI/BglII and ligated into EcoRI/BglII-digested pDC123 vector, yielding psccH and pgacH_cm, respectively. To make a catalytically inactive variant of sccH, the mutation T553A was introduced into psccH using Gibson Assembly site-directed mutagenesis with the primers sccH-T553A-F, sccH-T553-R, repB-isoF and repB-isoR. The plasmids were transformed into SMUΔsccH as previously described6. CAT-resistant single colonies were picked and checked for the presence of psccH or pgacH_cm by PCR, yielding strains SMUΔsccH:psccH, SMUΔsccH:psccH-T533A and SMUΔsccH:pgacH, respectively.

To create a vector for expression of eGacH in E. coli, the gene was amplified from 5005 chromosomal DNA using the primers gacH-NcoI-f and gacH-XhoI-r. The PCR product was digested with NcoI and XhoI, and ligated into NcoI/XhoI-digested pCDF-NT vector. The resultant plasmid, pCDF-GacH, contained gacH fused at the N terminus with a His-tag followed by a TEV protease recognition site. To produce a catalytically inactive variant of eGacH, the mutation T530A was introduced into pCDF-GacH using Gibson Assembly site-directed mutagenesis and the primers gacH-T530A-F, gacH-T530A-R, Str-isoF and Str-isoR.

Identification of hGIIA-resistant GAS transposon mutants

The GAS M1T1 5448 Krmit transposon mutant library16 was grown to mid-log phase (OD600 = 0.4). Colony-forming units (1 × 105) were incubated with 27.5 µg ml–1 recombinant hGIIA52 in triplicate for 1 h at 37 °C and plated on THY agar plates supplemented with kanamycin. The position of the transposon insertion of resistant colonies was determined as described previously53. hGIIA susceptibility experiments were performed as described previously17.

Determination of selective metal concentrations

To determine the target concentration of Zn2+, colonies of 5448 wild type and 5448ΔczcD19 were scraped from THY agar plates, resuspended and washed in phosphate buffered saline (PBS) to OD600 = 1 and used to inoculate freshly prepared modified RPMI (mRPMI) containing varying concentrations of Zn2+ to OD600 = 0.05 in a 96-well plate. Growth at 37 °C was monitored at OD595 every 15 min using the BMG Fluostar plate reader.

Tn-seq library screen for Zn2+ sensitivity

The 5448 Krmit Tn-seq library at T0 generation16 was thawed, inoculated into 150 ml prewarmed THY broth containing kanamycin and grown at 37 °C for 6 h. The culture (T1) was centrifuged at 4,000g for 15 min at 4 °C and the pellet resuspended in 32.5 ml saline. Freshly prepared mRPMI or mRPMI containing 10 µM or 20 µM Zn2+ was inoculated with 500 µl culture into 39.5 ml media, creating a 1:20-fold inoculation. These T2 cultures were then grown at 37 °C for exactly 6 h, at which point 2 ml of these cultures was inoculated again into 38 ml of freshly prepared mRPMI alone or mRPMI containing 10 µM or 20 µM Zn2+. The remaining 38 ml of T2 culture was harvested by centrifugation at 4,000g for 10 min at 4 °C and pellets stored at −20 °C for subsequent DNA extraction. Cultures were grown for additional 6 h, at which point T3 cultures were harvested by centrifugation at 4,000g for 10 min at 4 °C and pellets stored at −20 °C.

Tn-seq Krmit transposon insertion tags were prepared from the cell pellets as previously described16,47. After quality control with the Bioanalyzer instrument (Agilent), the libraries of Krmit insertion tags were sequenced (50 nt single-end reads) on an Illumina HiSeq 1500 in the Institute for Bioscience and Biotechnology Research Sequencing Core at the University of Maryland, College Park. Tn-seq read datasets were analyzed (quality, filtering, trimming, alignment, visualization) as previously described16,47 using the M1T1 5448 genome as reference for read alignments. The ratios of mutant abundance comparing the output to input mutant pools were calculated as a fold change for each GAS gene using the DEseq2 and EdgeR pipelines47,48,49.

Drop-test assays

Strains 5448 wild type, 5448ΔgacI, 5448ΔgacI:gacI, 5448ΔgacH, 5448ΔgacH:pgacH, S.mutans wild type, SMUΔsccH, SMUΔsccH:psccH and SMUΔsccH:pgacH were grown in THY to mid-exponential growth phase, adjusted to OD600 = 0.6 and serially diluted. Five microliters were spotted onto THY agar plates containing varying concentrations of Zn2+ (ZnSO4·7H2O). Plates were incubated at 37 °C overnight and photographed.

Protein expression and purification

To purify eGacH and eGacH-T530A, E. coli Rosetta(DE3) carrying the respective plasmid was grown in Lysogeny broth at 37 °C to OD600 = 0.4–0.6 and induced with 0.25 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 18 °C for approximately 16 h. Bacteria were lysed in 20 mM Tris-HCl pH 7.5, 300 mM NaCl by a microfluidizer cell disrupter. The soluble fractions were purified by nickel-nitrilotriacetic acid chromatography. The eluate was dialyzed into 20 mM Tris-HCl pH 7.5, 300 mM NaCl in the presence of TEV protease (1 mg 20 mg–1 protein) and reapplied to a nickel-nitrilotriacetic acid column equilibrated in 20 mM Tris-HCl pH 7.5, 300 mM NaCl to remove the cleaved His-tag and any uncleaved protein from the sample. The protein was further purified by size-exclusion chromatography on a Superdex 200 column in 20 mM HEPES pH 7.5, 100 mM NaCl.

To express seleno-methionine (Se-Met)-labeled eGacH, E. coli Rosetta(DE3) carrying eGacH was grown in Lysogeny broth at 37 °C to OD600 = 0.5. The bacteria were centrifuged and resuspended in M9 minimal media supplemented with Se-Met. Protein expression was induced with 0.25 mM IPTG, and the cultures were grown at 16 °C for approximately 16 h. Se-Met-labeled eGacH was purified as described above.

Crystallization, data collection and structure solution

The conditions for eGacH crystallization were initially screened using the JCSG Suites I–IV screens (Qiagen) at a protein concentration of 9 mg ml–1 by the hanging-drop vapor diffusion method. Crystals of Se-Met-substituted eGacH were grown in 0.1 M HEPES pH 7.5, 10% PEG8000 and 8% ethylene glycol. Crystals were transferred to crystallization solution supplemented with 20% ethylene glycol and flash-frozen in liquid nitrogen. The data were collected at APS 22-ID at a wavelength of 0.9793 Å. Crystals of GroP•eGacH complex were obtained using crystallization solution containing 0.2 M calcium acetate, 0.1 M MES pH 6.0 and 20% PEG8000. sn-Gro-1-P (Sigma Aldrich) was mixed with eGacH at 10 mM before crystallization. Initial crystals of the GroP•eGacH complex belonged to the same crystal form as apo GacH, but crystals of different morphology grew epitaxially after several days. These crystals displayed better diffraction and were used for structure determination of the GroP•eGacH complex. Crystals were cryoprotected in crystallization solution supplemented with 10 mM sn-glycerol-1-phosphate and 20% ethylene glycol and vitrified in liquid nitrogen. The data were collected at SSRL BL9–2 at a wavelength of 0.97946 Å.

All data were processed and scaled using XDS and XSCALE54. The structure of eGacH was solved by the Se single-wavelength anomalous diffraction method. Se atom positions were determined using the HySS module in PHENIX55. The structure was solved using AutoSol wizard in PHENIX55. The model was completed using Coot56 and refined using phenix.refine in PHENIX55. Ramachandran distribution analysis of the eGacH final structure with MolProbity57 indicates that 96.6 and 3.4% residues are in favored and allowed regions, respectively, with no outliers.

The structure of the GroP•eGacH complex was solved by molecular replacement using Phaser in PHENIX55 and the dimer of apo eGacH as a search model. The model was adjusted using Coot and refined using phenix.refine. Difference electron density corresponding to GroP molecules was readily identified after refinement. GroP molecules were modeled using Coot. The geometric restraints for GroP were generated using Grade Web Server ( (Global Phasing). The last several rounds of refinement were performed using 19 translation/libration/screw groups, which were identified by PHENIX55. Ramachandran distribution analysis of the GroP•eGacH final structure indicated that 97.2 and 2.8% residues are in favored and allowed regions, respectively, with no outliers. The structures were validated using Coot, MolProbity and wwPDB Validation Service ( Statistics for data collection, refinement and model quality are listed in Supplementary Table 7.

In vitro assay of eGacH enzymatic activity

Purification of 16:0–6:0 NBD-phosphatidylglycerol lipid (Avanti) was performed by preparative thin-layer chromatography as described previously25, dissolved in CH3OH and stored at −20 °C until use. The lipid was dried and dispersed in octyl-glucoside by sonication before addition of the remaining components. Reaction mixtures contained 0.05 M sodium succinate pH 6.3, 10 mM MnCl2, 0.05 M NaCl, 0.25% octyl-glucoside, 20 µg NBD-phosphatidylglycerol, ultrasonically dispersed in 0.5% octyl-glucoside (Branson 2200 bath sonicator) and either no enzyme, 20 µg eGacH or 20 µg eGacH-T530A in a total volume of 0.02 ml. Following incubation at 37 °C for 3 h, the reaction was stopped by the addition of 0.08 ml CHCl3/CH3OH (2:1) and analyzed for fluorescence on a BioRad ChemiDoc MP Imaging System using the fluorescein preset mode, as described previously25. The migration position of the NBD-diacylglycerol product was determined from the product of a separate reaction containing purified phospholipase C from B. cereus (Sigma Aldrich).

Isolation of cell wall

Cell wall was isolated from exponential phase cultures by the sodium dodecyl sulfate-boiling procedure and lyophilized as previously described15.

GAC purification

GAC was released from the cell wall by sequential digestion with mutanolysin (Sigma Aldrich) and recombinant PlyC amidase15, and partially purified by a combination of size-exclusion chromatography and ion-exchange chromatography. Mutanolysin digests contained 5 mg ml–1 of cell wall suspension in 0.1 M sodium acetate, pH 5.5, 2 mM CaCl2 and 5 U ml–1 mutanolysin. Following overnight incubation at 37 °C, GAC was separated from the cell wall by centrifugation at 13,000g, 10 min and precipitated from 80% acetone (−20 °C). The precipitate was sedimented (5,000g, 20 min), dried briefly under nitrogen gas and redissolved in 0.1 M Tris-HCl, pH 7.4 and digested with PlyC (50 µg ml–1) overnight at 37 °C. Following PlyC digestion, GAC was recovered by acetone precipitation as described above, redissolved in a small volume of 0.2 N acetic acid and chromatographed on a 25 ml column of BioGel P10 equilibrated in 0.2 N acetic acid. Fractions (1.5 ml) were collected and monitored for carbohydrate by the anthrone assay. Fractions containing GAC were combined, concentrated and desalted by spin column centrifugation (3,000 MW cutoff filter). GAC was loaded onto an 18 ml column of DEAE-Sephacel. The column was eluted with a 100 ml gradient of NaCl (0–1 M). Fractions were analyzed for carbohydrate by the anthrone assay and phosphate by the malachite green assay following digestion with 70% perchloric acid. Fractions containing peaks of carbohydrate were combined, concentrated by spin column (3,000 MW cutoff) and lyophilized.

Anthrone assay

Total carbohydrate content was determined by a minor modification of the anthrone procedure. Reactions contained 0.08 ml of aqueous sample and water and 0.32 ml anthrone reagent (0.2% anthrone in concentrated H2SO4). The samples were heated to 100 °C for 10 min, cooled in water (room temperature) and the absorbance at 580 nm was recorded. GAC concentration was estimated using an l-rhamnose standard curve.

Phosphate assay

Approximately 1.5 mg of GAS cell wall material was dissolved in 400 µl H2O and 8 µg ml–1 PlyC and incubated at 37 °C, rotating for approximately 16 h. Additional PlyC was added and incubated for a further 4–6 h. To liberate SCC from S.mutans cell walls, 1.5 mg of S.mutans cell wall material was incubated 24 h with 1.5 U ml–1 mutanolysin in 400 µl of 0.1 M sodium acetate, pH 5.5 and 2 mM CaCl2. The samples were incubated at 100 °C for 20 min and centrifuged for 5 min at maximum speed in a table top centrifuge. The supernatant was transferred to a new micro-centrifuge tube and incubated with 2 N HCl at 100 °C for 2 h. The samples were neutralized with NaOH in the presence of 62.5 mM HEPES pH 7.5. To 100 µl of acid hydrolyzed sample, 2 µl of 1 U µl–1 alkaline phosphatase (Thermo Fisher) and 10 µl 10× alkaline phosphatase buffer was added and incubated at 37 °C, rotating, overnight. Released phosphate was measured using the Pi ColorLock Gold kit (Innova Biosciences), according to the manufacturer’s protocol.

Total phosphate content was determined by the malachite green method following digestion with perchloric acid. Samples containing 10–80 µl were heated to 110 °C with 40 µl 70% perchloric acid (Fisher Scientific) in 13 × 100 borosilicate disposable culture tubes for 1 h. The reactions were diluted to 160 µl with water and 100 µl was transferred to a flat-bottom 96-well culture plate. Malachite green reagent (0.2 ml) was added and the absorbance at 620 nm was read after 10 min at room temperature. Malachite green reagent contained 1 vol. 4.2% ammonium molybdate tetrahydrate (by weight) in 4 M HCl, 3 vol. 0.045% malachite green (by weight) in water and 0.01% Tween 20.

Glycerol assay

Samples for glycerol measurement were prepared as described for the phosphate assay but were not digested with alkaline phosphatase. Instead, glycerol concentration was measured using the Glycerol Colorimetric assay kit (Cayman Chemical) according to the manufacturer’s protocol.

Glycosyl composition analysis

Glycosyl composition analysis of GAS and S.mutans cell wall samples was performed at the Complex Carbohydrate Research Center by combined gas chromatography/mass spectrometry of the per-O-trimethylsilyl derivatives of the monosaccharide methyl glycosides produced from the sample by acidic methanolysis as described previously15.

Dot-blot analysis of GAC

The semi-quantitative immuno-dot-blot analysis of GAC expressed by GAS strains was conducted essentially as described15 with the following minor modification: the PlyC-digested cell wall fractions were serially diluted before spotting to a nitrocellulose membrane.

Total phospholipid content analysis

GAS cells (50 ml) grown in THY broth to OD600 = 0.5 were centrifuged, washed with PBS, resuspended in 5 ml PBS and incubated with PlyC (100 µg ml–1) at 37 °C for 1 h. A total lipid extract was prepared by a modification of Bligh–Dyer extraction, freed of non-lipid contaminants by Folch partitioning, as described previously15, and redissolved in 1 ml CHCl3/CH3OH (2:1). An aliquot (20 µl) was analyzed for total phospholipid using the malachite green phosphate assay following perchloric acid digestion at 120 °C for 1 h, and the remainder was analyzed for phosphatidylglycerol.

Phosphatidylglycerol and NBD-diacylglycerol analysis

The analysis of phosphatidylglycerol and NBD-diacylglycerol was performed using an Ultimate 3000 ultra HPLC system coupled to a Thermo Q-Exactive Orbitrap mass spectrometer equipped with a heated electrospray ion source (Thermo Scientific). Lipid extracts were separated on a Waters ACQUITY BEH C8 column (2.1 × 100 mm, 1.7 µm) with the temperature maintained at 40 °C. The flow rate was 250 µl min–1, and the mobile phases consisted of 60:40 water/acetonitrile (A) and 90:10 isopropanol/acetonitrile (B), both containing 10 mM ammonium formate and 0.1% formic acid. The samples were eluted with a linear gradient from 32% B to 97% B over 25 min, maintained at 97% B for 4 min and re-equilibration with 32% B for 6 min. The sample injection volume was 5 µl. The mass spectrometer was operated in positive and negative ionization modes. The full-scan and fragment spectra were collected at a resolution of 70,000 and 17,500, respectively. Data analysis and lipid identification were performed using Xcalibur 4.0 and Lipidsearch 4.1 (Thermo Fisher).

Total carbohydrate content analysis

GAS cells (10 ml) grown in THY broth OD600 = 0.5 were centrifuged, washed twice with PBS and resuspended in 0.2 ml distilled water. The cell suspension was assayed for carbohydrate content by the anthrone assay as described above.

Analysis of GAC for GroP

GAC (prepared from ~1.5 mg of cell wall) was hydrolyzed in 0.1 ml 0.5 M NaOH (100 °C, 1 h) to release GroP, as described in ref. 26. Following alkaline treatment, the reaction was neutralized with acetic acid, supplemented with 5 nmol of citronellyl phosphate (as internal standard) and centrifuged on an Amicon Ultra Centrifugal Filter (0.5 ml, 3,000 nominal molecular weight limit). The filtrate was lyophilized and analyzed with a Q-Exactive mass spectrometer and an Ultimate 3000 ultra HPLC system (Thermo Fisher Scientific) using a silica-based SeQuant ZIC-pHILIC column (2.1 × 150 mm, 5 µm, Merck) with elution buffers consisting of (1) 20 mM (NH4)2CO3 with 0.1% NH4OH in H2O and (2) acetonitrile. The column temperature was maintained at 40 °C, and the flow rate was set to 150 µl min–1. MS detection was performed by electrospray ionization in negative ionization mode with source voltage maintained at 3.0 kV. The capillary temperature, sheath gas flow and auxiliary gas flow were set at 275 °C, 40 arb and 15 arb units, respectively. Full-scan MS spectra (mass range m/z 75–1,000) were acquired with resolution R = 70,000 and AGC target 1 × 106. Extracted ion chromatograms for GroP and citronellyl phosphate were obtained from the LC–MS chromatograms and used to estimate relative GroP content.

Identification of the stereochemistry of the GroP moiety of GAC

GroP was liberated from GAC by alkaline hydrolysis as described in ref. 26 and re-fractionated on BioGel P10. The bulk of the GAC elutes in the void volume and GroP elutes in the inclusion volume as identified by LC–MS. Column fractions containing GroP were combined, concentrated by rotary evaporation (30 °C, under reduced pressure) and desalted on BioGel P2. The stereochemistry of the GroP was determined by an enzymatic method using the Amplite Fluorimetric sn-Gro-3-P Assay Kit (AAT Bioquest) according to the manufacturer’s instructions.

NMR spectroscopy

The NMR spectra were recorded on a Bruker AVANCE III 700 MHz equipped with a 5 mm TCI Z-Gradient Cryoprobe (1H/13C/15N) and dual receivers, and a Bruker AVANCE II 600 MHz spectrometer equipped with a 5 mm TXI inverse Z-Gradient 1H/D-31P/13C. The 1H and 13C NMR chemical shift assignments of the polysaccharide material were carried out in D2O solution (99.96%) at 323.2 K unless otherwise stated. Chemical shifts are reported in parts per million using internal sodium 3-trimethylsilyl-(2,2,3,3-2H4)-propanoate (TSP, δH 0.00 ppm), external 1,4-dioxane in D2O (δC 67.40 ppm) and 2% H3PO4 in D2O (δP 0.00 ppm) as reference. The 1H,1H-TOCSY experiments (dipsi2ph) were recorded with mixing times of 10, 30, 60, 90 and 120 ms. The 1H,1H-NOESY experiments58 were collected with mixing times of 100 and 200 ms. Uniform and non-uniform sampling (50 and 25% NUS) were used for the multiplicity-edited 1H,13C-HSQC experiments59 employing an echo/anti-echo-TPPI gradient selection with and without decoupling during the acquisition. The 2D 1H,13C-HSQC-TOCSY experiments were acquired using MLEV17 for homonuclear Hartman–Hahn mixing, an echo/anti-echo-TPPI gradient selection with decoupling during acquisition and mixing times of 20, 40, 80 and 120 ms. The 2D 1H,31P-Hetero-TOCSY experiments60 were collected using a DIPSI2 sequence with mixing times of 10, 20, 30, 50 and 80 ms. The 2D 1H,31P-HMBC experiments were recorded using an echo/anti-echo gradient selection and mixing times of 25, 50 and 90 ms. The 3D 1H,13C,31P27 spectra were obtained using echo/anti-echo gradient selection and constant time in t2 with a nominal value of nJCP of 5 Hz and without multiplicity selection. The spectra were processed and analyzed using TopSpin 4.0.1 software (Bruker BioSpin).

Statistical analysis

Unless otherwise indicated, statistical analysis was carried out on pooled data from at least three independent biological repeats. A two-way ANOVA with Bonferroni multiple comparison test was used to compare multiple groups. A P value equal to or less that 0.05 was considered statistically significant.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.