Streptococcus pneumoniae, S. mitis, and S. oralis Produce a Phosphatidylglycerol-Dependent, ltaS-Independent Glycerophosphate-Linked Glycolipid

ABSTRACT Lipoteichoic acid (LTA) is a Gram-positive bacterial cell surface polymer that participates in host-microbe interactions. It was previously reported that the major human pathogen Streptococcus pneumoniae and the closely related oral commensals S. mitis and S. oralis produce type IV LTAs. Herein, using liquid chromatography/mass spectrometry-based lipidomic analysis, we found that in addition to type IV LTA biosynthetic precursors, S. mitis, S. oralis, and S. pneumoniae also produce glycerophosphate (Gro-P)-linked dihexosyl (DH)-diacylglycerol (DAG), which is a biosynthetic precursor of type I LTA. cdsA and pgsA mutants produce DHDAG but lack (Gro-P)-DHDAG, indicating that the Gro-P moiety is derived from phosphatidylglycerol (PG), whose biosynthesis requires these genes. S. mitis, but not S. pneumoniae or S. oralis, encodes an ortholog of the PG-dependent type I LTA synthase, ltaS. By heterologous expression analyses, we confirmed that S. mitis ltaS confers poly(Gro-P) synthesis in both Escherichia coli and Staphylococcus aureus and that S. mitis ltaS can rescue the growth defect of an S. aureus ltaS mutant. However, we do not detect a poly(Gro-P) polymer in S. mitis using an anti-type I LTA antibody. Moreover, Gro-P-linked DHDAG is still synthesized by an S. mitis ltaS mutant, demonstrating that S. mitis LtaS does not catalyze Gro-P transfer to DHDAG. Finally, an S. mitis ltaS mutant has increased sensitivity to human serum, demonstrating that ltaS confers a beneficial but currently undefined function in S. mitis. Overall, our results demonstrate that S. mitis, S. pneumoniae, and S. oralis produce a Gro-P-linked glycolipid via a PG-dependent, ltaS-independent mechanism. IMPORTANCE The cell wall is a critical structural component of bacterial cells that confers important physiological functions. For pathogens, it is a site of host-pathogen interactions. In this work, we analyze the glycolipids synthesized by the mitis group streptococcal species, S. pneumoniae, S. oralis, and S. mitis. We find that all produce the glycolipid, glycerophosphate (Gro-P)-linked dihexosyl (DH)-diacylglycerol (DAG), which is a precursor for the cell wall polymer type I lipoteichoic acid in other bacteria. We investigate whether the known enzyme for type I LTA synthesis, LtaS, plays a role in synthesizing this molecule in S. mitis. Our results indicate that a novel mechanism is responsible. Our results are significant because they identify a novel feature of S. pneumoniae, S. oralis, and S. mitis glycolipid biology.

The goal of our study was to determine whether S. mitis produces multiple types of LTAs and whether S. mitis ltaS mediates production of type I LTA, using the type strain ATCC 49456 as a model. We used normal-phase liquid chromatography (NPLC)-electrospray ionization/mass spectrometry (ESI/MS) to analyze membrane lipids in the mitis group streptococci. This technique is highly sensitive and specific and allows for the detection and characterization of LTA anchors and other LTA biosynthetic intermediates whose cellular levels are too low to be detected by conventional techniques such as thin-layer chromatography (TLC). We identified intermediates of type IV LTA synthesis in S. mitis, S. oralis, and S. pneumoniae. To our surprise, a type I-like LTA intermediate was observed not only in S. mitis, which encodes ltaS, but also in S. oralis and S. pneumoniae, which lack ltaS orthologs. Moreover, while S. mitis ATCC 49456 ltaS confers poly(Gro-P) synthesis when heterologously expressed in Escherichia coli and an S. aureus ltaS-deficient mutant, we confirm that S. mitis ATCC 49456 does not produce a polymer detectable by a type I LTA antibody. Importantly, ltaS contributes to S. mitis ATCC 49456 fitness, because deletion of ltaS impacted growth in human serum-supplemented medium. Overall, our results demonstrate that S. mitis, S. oralis, and S. pneumoniae synthesize intermediates of two structurally distinct lipid-anchored polymers, one type IV LTA, and one a Gro-P-containing polymer whose full structure remains to be determined.

RESULTS
Mitis group streptococci produce glycolipid intermediates of two structurally distinct LTAs. LTA is usually anchored to the membrane by a saccharide-linked diacylglycerol (DAG) glycolipid (23). Structure of the glycolipid anchor varies among different LTA types, bacterial species, and even culture conditions (38). In S. pneumoniae, the pseudopentasaccharide repeating units of type IV LTA are proposed to be assembled on an undecaprenyl pyrophosphate (C 55 -PP) anchor and then transferred to a glucosyl-DAG (Glc-DAG) anchor ( Fig. 1A) (25). In S. aureus, type I LTA is typically assembled on a diglucosyl-DAG (Glc 2 -DAG) anchor ( Fig. 1A) (39). Listeria monocytogenes also produces type I LTA, which is linked to a galactosyl-glucosyl-DAG (Gal-Glc-DAG) anchor (40). Thus, lipid profiling has the potential to identify LTA intermediates, thereby revealing possible types of LTAs produced by a bacterium. To perform lipidomic analysis of mitis group streptococci, total lipids were extracted from bacterial cultures with a modified acidic Bligh-Dyer method and analyzed with NPLC-ESI/MS (41). We analyzed the type strain of S. mitis (ATCC 49456, referred to as SM61 hereafter), S. oralis (ATCC 35037 and the endocarditis isolate 1647), two clinically isolated S. pneumoniae strains (D39 and TIGR4), and Streptococcus sp. strain 1643 (referred to as SM43 hereafter), a human endocarditis isolate that was clinically identified as S. mitis but shares higher genomic identity with S. oralis (Table 1) (18,42).
Three C 55 -PP-linked intermediates of type IV LTA biosynthesis were detected in all strains analyzed. Specifically, these intermediates are C 55  To confirm the possible monosaccharide identity of the DAG-linked sugars, in silico analyses were performed to identify orthologs of known glycolipid biosynthetic genes in the genomes of the tested strains. S. pneumoniae produces the glycolipid Gal-Glc-  (43), for which the biosynthetic genes have been partially identified. These genes can be separated into two major groups corresponding to the biosynthetic steps they are responsible for: (i) production of nucleotide-activated sugars and (ii) transferring of the activated sugar moieties to DAG (38). As shown in Table 2, these genes include the following: confirmed UDP glucose (UDP-Glc) production gene pgm (encoding a-phosphoglucomutase) and galU (encoding UTP:a-glucose-1-phosphate uridyltransferase) (44); Leloir pathway genes that are proposed to produce UDP galactose (UDP-Gal), specifically galK (encoding galactokinase) and galT2 (encoding galactose-1-phosphate uridylyltransferase 2) (45, 46); and glycosyltransferases encoded by genes Spr0982 and cpoA which sequentially transfer Glc and Gal residues to DAG, respectively (47,48). S. pneumoniae R6 is an avirulent and unencapsulated derivative of S. pneumoniae D39 (49). These two strains share the same glycolipid biosynthetic genes. Using S. pneumoniae R6 as reference, orthologs of Gal-Glc-DAG biosynthetic genes with $87% amino acid identity were identified in the genomes of SM61, S. oralis ATCC 35037, SM43, and S. pneumoniae TIGR4 (Table 2). This analysis suggests that the DHDAG detected in our experiments is likely to be Gal-Glc-DAG.
Biosynthesis of (Gro-P)-DHDAG requires phosphatidylglycerol in mitis group streptococci. In S. aureus, the Gro-P of type I LTA is produced from hydrolyzation of membrane PG (36), a process that is also required for Gro-P modification of streptococcal rhamnose-containing cell wall polysaccharides (50). To verify whether PG is the source of Gro-P for (Gro-P)-DHDAG biosynthesis in mitis group streptococci, we analyzed the lipid profiles of cdsA and pgsA mutants. The gene cdsA is required for the synthesis of CDP-DAG, which is then converted by PgsA to produce phosphatidylglycerophosphate (PGP), the immediate precursor of PG (Fig. 1A) (18,41). We previously reported that cdsA deletion mutants of S. mitis and S. oralis do not synthesize PG, nor does a pgsA deletion mutant of SM43 (18,41) (Fig. 2). Thus, lipid anchor profiles of    (Table 1). These results demonstrate that cdsA and pgsA, or more specifically the ability to synthesize PG, are required for the biosynthesis of (Gro-P)-DHDAG in SM61 and SM43. S. mitis, S. oralis, and S. pneumoniae cell extracts do not react with a type I LTA antibody. Currently, enzymes known to transfer Gro-P from PG for Gro-P polymer synthesis or Gro-P modification include the following: (i) S. aureus LtaS, the type I LTA synthase that produces poly(Gro-P) (36); (ii) L. monocytogenes LtaP, the type I LTA primase that has an overall structure and active site sequences that are very similar to those of LtaS, except that it links only the first Gro-P unit to the glycolipid anchor (35,40); and (iii) the recently identified streptococcal Gro-P transferase GacH that links Gro-P to cell wall-attached glycopolymers (50). Bioinformatic analyses predict no orthologs of either ltaP or gacH in the genomes of the mitis group streptococci assessed here, yet an ortholog of ltaS is present in S. mitis as previously reported (35).
If S. mitis ltaS functions the same as its ortholog in type I LTA-producing bacteria like S. aureus, polymers of Gro-P will be produced and may be detectable using an anti-type I LTA antibody. Western blot analysis using a previously described anti-type I LTA antibody was conducted for SM61, SM43, S. oralis ATCC 35037, and S. pneumoniae strains. No signal was detected from cell lysates of these strains (Fig. 3) or from cell lysates of SM61 that overexpress ltaS in trans from an anhydrotetracycline-inducible vector (Fig. S3). These results are in accordance with previous observations of no immunoluminescence detection of Gro-P polymers in SM61 (34). The validity of the antibody was confirmed by positive signals detected from cell lysates of Streptococcus agalactiae, Streptococcus pyogenes, and S. aureus, all three of which produce type I LTA (Fig. 3) (36,51,52). Interestingly, no signal was detected from cell lysate of Enterococcus faecalis OG1RF (Fig. 3), another bacterium known to produce type I LTA (53,54), which as reported previously is poorly recognized by the anti-type I LTA antibody (55).
S. mitis LtaS mediates production of poly(Gro-P) in an E. coli heterologous host. For the following analyses, the S. mitis type strain ATCC 49456 (SM61) was used as a model, and its ltaS ortholog (SM12261_RS03435) was renamed ltaS. We heterologously expressed S. mitis ltaS in E. coli to verify the function of the gene. This approach was previously used in studies of S. aureus ltaS (36). Plasmid pET-ltaS (Table 3) was constructed so that the expression of S. mitis ltaS could be induced with isopropyl-b-D-1thiogalactopyranoside (IPTG) in E. coli. As shown in Fig. 4A, with the addition of IPTG, detectable bands produced by anti-type I LTA antibody targeting were observed for E. coli (pET-ltaS), demonstrating that S. mitis ltaS is sufficient to mediate the production of poly(Gro-P). S. mitis ltaS complements an S. aureus ltaS mutant for type I LTA production. In S. aureus, LtaS is required for proper cell division and efficient cell growth at 37°C (36,  56). To further confirm the physiological function of S. mitis ltaS in Gram-positive cells, we expressed it in a previously reported S. aureus strain that has its native ltaS gene under the control of an IPTG-inducible promoter (strain ANG499). Without IPTG, ANG499 is deficient for type I LTA production and has a growth defect when cultured at 37°C (36,56). S. mitis ltaS was introduced into strain ANG499 by the plasmid pitetR-ltaS (Table 3), which has the S. mitis ltaS coding region under the control of the tetracycline-inducible promoter P xyl/tet . Addition of anhydrotetracycline (ATC) induces expression of S. mitis ltaS. Note that we included ATC in all experimental cultures described below, because we observed an ATC-dependent growth defect that confounded direct comparison of cultures grown in the presence or absence of ATC (Fig. S4).
As expected, strain ANG499 with the empty plasmid vector pitetR grew more slowly and reached a lower final optical density at 600 nm (OD 600 ) value when cultured without IPTG compared to with IPTG (Fig. 4B). As expected, type I LTA production by S. aureus LtaS was induced by IPTG, confirmed by Western blot analysis (Fig. 4C) and detection of type I LTA intermediates (Gro-P) 2 Fig. S5). Strikingly, the growth of ANG499 was also rescued by the expression of S. mitis ltaS from pitetR-ltaS (Fig. 4B), and type I LTA production was observed, as shown by Western blot  4C) and lipidomic analysis ( Fig. 5A and Fig. S5). These data demonstrate that S. mitis ltaS can complement the function of S. aureus ltaS and promote production of type I LTA in S. aureus. Surprisingly, (Gro-P)-Glc 2 -DAG ([M-H] 2 ion at m/z 1059.6 of Fig. 5B) was detected at comparable levels from all S. aureus cultures, including the natively ltaS-deficient strain in the absence of IPTG induction. Expression of S. aureus ltaS does not confer detectable type I LTA signals in S. mitis. To test whether S. aureus ltaS can mediate poly(Gro-P) production in S. mitis, S. aureus ltaS was introduced into S. mitis with the plasmid pitetR-SAltaS. Similar to pitetR-ltaS, pitetR-SAltaS encodes S. aureus ltaS under the control of the ATC-inducible promoter P xyl/tet . Type I LTA production was detected by Western blot analysis for E. coli (pitetR-SAltaS) induced with ATC (Fig. 6). However, no signals were observed for S. mitis (pitetR-SAltaS) induced with ATC. In addition, lipidomic analysis detected no further structure beyond a single Gro-P linked to DHDAG for S. mitis (pitetR-SAltaS) induced with ATC. IPTG was added to mid-log-phase bacterial cultures followed by another 30-min incubation at 37°C before cell pelleting. Three biological independent replicates were performed for each sample. (B) Growth curves of S. aureus ANG499 containing either pitetR or pitetR-ltaS grown in tryptic soy broth (TSB) with the addition of either 150 ng/ml anhydrotetracycline (ATC) only or 150 ng/ml ATC and 0.5 mM IPTG as indicated. Samples were grown in TSB with 0.5 mM IPTG overnight, followed by subculturing into fresh TSB with the indicated addition of induction reagents and incubated for 3 h. Then, another subculturing to an OD 600 of 0.1 with fresh media same as the previous incubation was performed. After the second subculture, OD 600 values were measured every hour and plotted. (C) Western blot detection of type I LTA from S. aureus ANG499 containing either pitetR or pitetR-ltaS. Samples were grown in the same way as described above for panel B, after the first subculturing and incubation, cells equal to 1 ml of OD 600 at 1.2 were harvested, followed by lysate preparation and immunodetection. Schematics of induction expression of chromosomal or plasmid-carried ltaS were shown in both panels A and C. Loading controls of both panels A and C were stained with Coomassie blue. Western blot band intensity in panel C was normalized to the loading control and the pitetR-ltaS sample. For panels B and C, four biological replicates were performed; averages of the sample values were plotted with the error bars depicting standard deviations. Statistical analyses were performed with one-way analysis of variance (ANOVA); significant difference was determined by P value of ,0.05. S. mitis lacking ltaS has increased serum susceptibility. To investigate functions of ltaS in S. mitis, ltaS was deleted and exchanged for the erythromycin resistance marker ermB, generating S. mitis DltaS. Of note, (Gro-P)-DHDAG was still detected in the S. mitis DltaS strain, demonstrating that LtaS is not required for the addition of the Gro-P unit to the DHDAG (Table 1).
Unlike S. aureus, which requires ltaS for efficient growth, deletion of ltaS in S. mitis does not confer a growth defect under laboratory culturing conditions. Specifically, when growing in Todd-Hewitt broth at 37°C, the doubling time of the DltaS strain is 39.8 (63.7) min, which is not significantly different from the 40.2 (63.5) min doubling time of wild-type S. mitis (Fig. S6). Considering that the growth deficiency of S. aureus lacking ltaS could be mitigated by culturing at a lower temperature (56), the growth of wild-type S. mitis and DltaS strains cultured at a higher temperature was measured to determine whether the ltaS mutant was compromised for temperature-related stresses. The temperature 42°C was chosen as a representative of fever. Both wild-type and DltaS strains exhibited slower growth at 42°C compared to 37°C; however, no significant difference in growth rate was observed between the strains (46.2 [63.0] and 47.4 [63.8] min doubling times for the wild-type and DltaS strains, respectively). Moreover, no difference in susceptibilities to antibiotics targeting peptidoglycan biosynthesis, membrane integrity, and protein synthesis were observed (see Table S1 in the supplemental material). Thus, under these laboratory culture conditions, ltaS is not essential for the growth of S. mitis.
In addition, a potential role for ltaS in host-microbe interactions was investigated. As an oral commensal, the environment S. mitis colonizes is exposed to human gingival crevicular fluid, which is an extrudant of serum with lower concentrations of complement (57). Moreover, when invading the bloodstream and causing bacteremia and infectious endocarditis, S. mitis is constantly exposed to blood. Thus, human serum is a useful medium component for laboratory reconstruction of the host growth conditions. Supplementation of human serum into chemically defined medium (CDM) promotes the growth of S. mitis compared to nonsupplemented CDM (Fig. 7). Deletion of ltaS does not confer a significant difference in growth in Todd-Hewitt broth or unsupplemented CDM but does result in a significant growth deficiency in human serumsupplemented CDM, and makes S. mitis more sensitive to the killing effect of complete serum (Fig. 7). In addition, bacterial colony forming units (CFU)/ml counts were significantly higher for the DltaS mutant cultured in heat-inactivated serum compared to the mutant cultured in complete serum (Fig. 7); this significant difference was not observed for the wild-type strain. These results suggest that although ltaS is not required for growth of S. mitis under laboratory conditions, it confers protection against heat-sensitive serum components. Further investigation is needed to elucidate such interactions.

DISCUSSION
In this work, we used NPLC-ESI/MS to analyze the glycolipid profiles of S. mitis, S. oralis, and S. pneumoniae strains. For all of the tested strains, biosynthetic intermediates of two structurally different LTAs were detected ( Fig. 1 and Table 1). First, consistent with literature, the biosynthetic intermediates of the type IV LTA were detected, which is in agreement with genomic analysis of the biosynthetic genes (26). The second distinct LTA is indicated by the detection of (Gro-P)-DHDAG, which is similar to type I LTA polymers and unexpected based on previous reports, and thus has been the focus of this study.
On the basis of the results of genomic analysis, we proposed that the newly identified (Gro-P)-DHDAG is structured as (Gro-P)-Gal-Glc-DAG. The glycolipid Gal-Glc-DAG has been reported as the dominant glycolipid species in S. pneumoniae, and our prediction is in accordance with this previous report (43). However, the full pathway for Gal-Glc-DAG synthesis has not been fully experimentally verified in the mitis group streptococci; the stereochemistry of the hexoses requires further confirmation with structural analysis, such as with nuclear magnetic resonance (NMR).
The PG-dependent (Gro-P)-DHDAG biosynthetic process in S. mitis was then investigated, which led to the main focus of this study, functional verification of S. mitis ltaS. Through heterologous expression, we confirmed that S. mitis ltaS could directly synthesize Gro-P polymers in both E. coli and S. aureus. However, it appeared that S. mitis LtaS functions somewhat differently from S. aureus LtaS, as the expression of S. mitis ltaS does not fully complement the growth deficiency and the amount of type I LTA produced ( Fig. 4B and C), which is not unexpected considering that S. mitis and S. aureus LtaS share only 38% sequence identity (26).
We did not detect a Gro-P polymer in wild-type S. mitis using Western blot analysis. Explanations as to why we could not detect the polymer include the following. (i) S. mitis does not produce the Gro-P polymer; instead, (Gro-P)-DHDAG is the complete and final product. (ii) A very small amount of the Gro-P polymer is produced under the culture conditions investigated here. (iii) Unique structural modifications on the Gro-P polymer hinder antibody recognition. (iv) LtaS acts on a different substrate than (Gro-P)-DHDAG in S. mitis. Further large-scale purification and structural analysis of the (Gro-P)-DHDAG-containing polymer produced by mitis group streptococci are required.
Interestingly, heterologous expression of S. aureus ltaS in S. mitis does not confer poly(Gro-P) production detectable by either Western blot or lipidomic analysis. It is possible that, as suggested above, unique structural modifications on the Gro-P polymer hinder antibody recognition or that S. aureus LtaS lacks the appropriate substrate  (5), 95% human serum (95) with 5% phosphate-buffered saline (PBS), Todd-Hewitt broth (THB), and 95% heat-inactivated human serum (HIS) with 5% PBS. The CFU/ml of cultures after 8-h incubation are shown. The CFU/ml of E. coli K-12 MG1655 grown in 95% human serum is below the detection limit (10 5 ; not shown in figure); the CFU/ml of E. coli cultured in HIS is shown. Each symbol represents the value for one biological independent repeat. Statistical analysis was performed with the Mann-Whitney method. P values are indicated above the line. Statistical significance was defined by P value of ,0.05, and significant P values are shown in red.
(s) in S. mitis to catalyze type I LTA synthesis, in which case further studies about the substrate recognition and binding activities of S. aureus and S. mitis LtaS are needed. Last but not the least, it is possible that the canonical LtaS enzymatic function of producing poly(Gro-P) is inhibited in S. mitis.
The findings that (Gro-P)-DHDAG is still present in S. mitis DltaS, as well as in S. oralis and S. pneumoniae, which are species that do not carry genes that encode any orthologs of ltaS, suggest the existence of an unknown PG-dependent Gro-P transferase in these species that is responsible for the synthesis of (Gro-P)-DHDAG. Unexpectedly, (Gro-P)-Glc 2 -DAG is also seen in S. aureus deficient for ltaS, suggesting that an unidentified Gro-P biosynthetic enzyme(s) or biological process(es) may exist in S. aureus as well, but this is more speculative.
In other Gram-positive pathogens that synthesize type I LTA, LtaS and its product, LTA, are essential for proper cell division (40,56,58,59). Inhibiting the function of LtaS is effective in extending the survival of S. aureus-infected mice (60) and sensitizing multidrug-resistant E. faecium to antibiotics (61). Though S. mitis ltaS is not essential for proper growth of the bacterium in normal laboratory media or for synthesizing (Gro-P)-DHDAG, it does provide some advantage to S. mitis when human serum is present in the culture media and protects against the heat-sensitive serum components.
In summary, we provide evidence that a type I-like LTA might coexist with type IV LTA in S. mitis, S. oralis, and S. pneumoniae and queried the role of ltaS in this process in a model S. mitis strain. To our knowledge, there is only one previous report which documents a bacterial species producing two structurally different LTAs, in Streptococcus suis, an invasive pathogen of pigs (62). Our lipidomic and genomic studies show that we have an incomplete understanding of glycolipids, LTAs, and LtaS function in mitis group streptococci and their potential roles in host-microbe interactions.

MATERIALS AND METHODS
Bacterial strains and growth conditions. Unless indicated otherwise, E. coli was grown in Luria-Bertani (LB) medium, Streptococcus strains were grown in Todd-Hewitt (TH) medium (BD Biosciences) with S. pneumoniae grown in TH medium supplemented with 0.5% yeast extract (BD Biosciences), and E. faecalis and S. aureus were grown in tryptic soy (TS) medium (BD Biosciences). All bacterial cultures were incubated at 37°C, unless otherwise noted. Streptococci were cultured with 5% CO 2 . Chemically defined medium (CDM) was made as previously described, with the addition of 0.5 mM choline (63). Human serum-supplemented medium was made by adding complete human serum (Sigma-Aldrich) to CDM to a final concentration of 5% (vol/vol). Antibiotic concentrations were as follows: kanamycin, 50 mg/ml in E. coli, 250 mg/ml in S. aureus, and 500 mg/ml in S. mitis; erythromycin, 50 mg/ml in E. coli. Transcription of genes controlled by the promoter P xyl/tet was induced with anhydrotetracycline (ATC) at a final concentration of 150 ng/ml. Isopropyl-b-D-1-thiogalactopyranoside (IPTG)-inducible expression was mediated by the addition of IPTG to a final concentration of 1 mM. Bacterial strains and plasmids used in this research are listed in Table 3.
Mutant generation. Deletion of cdsA (SM12261_RS08390) in S. mitis ATCC 49456 was conducted as previously described (65,66). Briefly, approximately 2-kb flanking regions on either side of cdsA were amplified using Phusion polymerase (Thermo Fisher). PCR products were digested with restriction enzyme XmaI (New England Biolabs) and ligated with T4 DNA ligase (New England Biolabs). Ligated products were amplified using primers 61cdsA_Up_F and 61cdsA_Dwn_R (see Table S2 in the supplemental material), followed by gel extraction with the QIAquick gel extraction kit (Qiagen) per the manufacturer's instruction. The linear construct was transformed into S. mitis via natural transformation as described previously (66). The DcdsA mutant was selected with 35mg/ml daptomycin and confirmed by Sanger sequencing (Massachusetts General Hospital DNA Core) of the PCR product of the cdsA deletion region.
Deletion of ltaS in S. mitis ATCC 49456 was conducted similarly with some slight modifications. Specifically, a 1-kb DNA fragment containing ermB was generated through PCR amplification using plasmid pMSP3535 as the template (67). Then, splicing by overlap extension PCR was performed to produce a 5-kb amplicon that sequentially contained a 2-kb fragment upstream of ltaS, a 1-kb ermB-containing fragment in reverse orientation, and a 2-kb fragment downstream of ltaS. The PCR product was analyzed on a 0.8% agarose gel and extracted using the QIAquick gel extraction kit (Qiagen) per the manufacturer's instruction. Transformation of the 5-kb amplicon into S. mitis was performed as described previously (66). The DltaS mutant was selected with 20 mg/ml erythromycin and confirmed with Illumina genome sequencing (UTD Genome Core Facility).
Plasmid construction. Plasmids used in this research are listed in Table 3 with description of their functions. All primers used in this research are listed in Table S2.
The shuttle plasmid pABG5 was used for heterologous gene expression in Gram-positive bacteria (68). Specifically, the DNA fragment containing the S. mitis ltaS coding region was amplified using primers LtaS_F and LtaS_R, and the pABG5 plasmid backbone was linearized through PCR using primers pABG5-5 and pABG5-3. Gibson assembly was conducted per the manufacturer's instructions (NEBuilder HiFi DNA assembly master mix; New England Biolabs), followed by transformation of the product into E. coli DH5a. The pABG5 with ltaS insert was further linearized with primers YW55 and YW56 and ligated with an 848bp DNA fragment via Gibson assembly, producing the plasmid pitetR-ltaS. The 848-bp fragment contained a tetracycline-controlled promoter P xyl/tet and the tetracycline repressor gene tetR in reverse orientation. Insertion of this 848-bp fragment immediately upstream of the ltaS coding region makes ltaS expression inducible by ATC addition. The sequence of the 848-bp fragment was obtained from plasmid pRMC2 in the Addgene sequence database (69,70), and the fragment was synthesized commercially (Integrated DNA Technologies). Induced production of the target gene ltaS was confirmed by Western blotting. The empty vector control pitetR was constructed via linearization of pitetR-ltaS with PCR using primers YW58 and YW59, followed by Gibson assembly for gap closure. The removal of the ltaS coding region was confirmed by Sanger sequencing (Massachusetts General Hospital DNA Core). Plasmid pitetR has an EcoRI site inserted after the P xyl/tet -controlled ribosomal binding site. ATC induction of S. aureus ltaS is mediated by plasmid pitetR-SAltaS. Specifically, an amplicon containing the S. aureus ltaS (SAV0719) coding region was obtained via PCR using primers YW72 and YW73, followed by Gibson assembly of this amplicon with linearized pitetR generated via EcoRI digestion. Successful insertion was confirmed with Sanger sequencing (Massachusetts General Hospital DNA Core), and the confirmed construct was transformed into E. coli DH5a for expression analysis. Plasmid pET-ltaS that mediates isopropyl-b-D-1-thiogalactopyranoside (IPTG)-inducible overexpression of ltaS was generated through insertion of the ltaS coding region immediately after the IPTG-inducible promoter region of pET-28a(1) (Novagen). Successful insertion was confirmed with Sanger sequencing (Massachusetts General Hospital DNA Core). The confirmed construct was transformed into E. coli BL21(DE3) pLys for expression analysis.
Antibiotic susceptibility testing. Antibiotic susceptibility testing was performed according to the bioMérieux Etest protocol with slight modifications. Specifically, a single colony of either the wild-type S. mitis ATCC 49456 or DltaS strain was selected from cation-adjusted Mueller-Hinton (MH) (BD Bacto) agar cultures, inoculated into 1 ml of MH broth, and incubated for 6 to 8 h at 37°C with 5% CO 2 . Then, 2 ml of fresh MH broth was added to the 1-ml culture, and incubation was resumed. After overnight incubation, the OD 600 of the cultures were measured, and samples having an OD 600 value of ,0.2 were excluded from the following experimental procedures. Cultures were spread onto prewarmed MH agar plates with sterile cotton-tipped applicators, and plates were air dried for 15 to 20 min inside a biosafety cabinet. Then, Etest strips (Etest by bioMérieux) prewarmed to room temperature were applied to the plates with aseptic technique. The plates were incubated overnight at 37°C with 5% CO 2 . The MIC was determined by the intersection of the zone of inhibition with the Etest strip. At least three biological independent replicates were performed for each antibiotic-strain combination.
Western blot analysis. Detection of type I LTA via Western blot analysis was performed as previously described (39,71).
For E. coli, single colonies were grown overnight in LB broth with appropriate antibiotics, followed by dilution to an OD 600 of 0.1 with fresh media into two replicates. For E. coli DH5a containing pitetR-SAltaS, ATC was added to one set of cultures to a final concentration of 150 ng/ml, followed by 3-h incubation at 37°C before cell harvest. For E. coli BL21(DE3) pLys containing pET-ltaS, diluted bacterial cultures were incubated at 37°C for 3 h, and then IPTG was added to one set of cultures to a 1 mM final concentration, followed by another 30-min incubation at 37°C before cell harvest. To harvest cells, culture densities were normalized to an OD 600 of 0.6, and 1 ml was pelleted, washed, resuspended in 100 ml of 2Â Laemmli sample buffer, and boiled for 15 min. Boiled samples were stored at -20°C prior to electrophoretic analysis.
For S. aureus, single colonies of each S. aureus strain were grown overnight in TS broth with 0.5 mM IPTG, 5 mg/ml erythromycin, and 250 mg/ml kanamycin, and then subcultured to an OD 600 of 0.1 into fresh TS broth containing 5 mg/ml erythromycin, 250 mg/ml kanamycin, and either 150 ng/ml ATC or 150 ng/ml ATC with 0.5 mM IPTG. After 3-h incubation, the OD 600 was measured, and cells equivalent to 1 ml of 1.2 OD 600 were pelleted. Cell pellets were washed and resuspended with 1 ml phosphate-buffered saline (PBS), followed by five cycles of bead-beating at 6.5 m/s for 45 s, with 5 min on ice between cycles (FastPrep-24; MP Biomedicals). After centrifugation at 200 Â g for 1 min, cell lysates were collected, followed by pelleting at 17,000 Â g for 10 min. The material was resuspended in 100 ml of 2Â Laemmli sample buffer (Bio-Rad) followed by boiling for 15 min in a heating block.
For streptococci and E. faecalis, unless indicated, OD 600 values of the overnight cultures were measured, followed by pelleting of cells equivalent to 1 ml of 1.2 OD 600 . Induction of ltaS overexpression in S. mitis was conducted similarly as in S. aureus. Specifically, overnight cultures of S. mitis containing pitetR-ltaS, pitetR, or pitetR-SAltaS were diluted to an OD 600 value of 0.1 into fresh TH broth with 150 ng/ml ATC. After 7-h incubation, cells equivalent to 1 ml of a culture with an OD 600 of 1.2 were harvested. All cell pellets were washed and resuspended with 1 ml PBS, then followed with the same cell disruption and lysate preparation processes as described above for S. aureus samples.
Separation of cell lysate materials are conducted through sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Specifically, 15 ml of each boiled sample was loaded onto a 15% SDS-PAG gel, followed by electrophoresis at consistent 100 voltage and subsequent polyvinylidene difluoride (PVDF) membrane transfer at consistent 350 mA. The blocking solution was PBS containing 0.05% (wt/ vol) Tween 20 and 10% (wt/vol) nonfat milk; antibody solutions were PBS with 0.05% (wt/vol) Tween 20 and 5% (wt/vol) nonfat milk. For S. aureus samples, 3 mg/ml human IgG (Sigma) was added to the blocking and antibody solutions to block the activity of protein A. Primary antibody targeting type I LTA (clone 55; Hycult Technology) and secondary antibody (horseradish peroxidase [HRP]-conjugated antimouse IgG; Cell Signaling) were used at dilutions of 1:2,500 and 1:5,000, respectively. After adding HRP substrate (Immobilon Western; Millipore) and shaking at room temperature for 3 min, chemiluminescence signals were detected with the ChemiDoc touch imaging system (Bio-Rad) with default chemiluminescence settings. Relative band intensity was analyzed with the Image Lab Software (Bio-Rad).
Lipidomic analysis. Extraction of total lipids from stationary-phase cells was performed by acidic Bligh-Dyer extraction as previously described (18). Specifically, cells were grown to stationary phase in at least 5 ml of medium, followed by collection and storage at -80°C until lipid extraction with the acidic Bligh-Dyer methods. The dried lipid extracts were dissolved in 100 ml of chloroform-methanol (2:1, vol/ vol). Typically, 10 ml of the dissolved solution were injected for LC/MS analysis. NPLC-ESI/MS of lipids was performed as previously described (41,72) using an Agilent 1200 quaternary LC system (Santa Clara, CA) coupled to a high-resolution TripleTOF5600 mass spectrometer (Sciex, Framingham, MA). An Ascentis Si high-performance liquid chromatography (HPLC) column (5 mm; 25 cm Â 2.1 mm; Sigma-Aldrich) was used. Mobile phase A consisted of chloroform-methanol-aqueous ammonium hydroxide (800:195:5, vol/vol/vol). Mobile phase B consisted of chloroform-methanol-water-aqueous ammonium hydroxide (600:340:50:5, vol/vol/vol/vol). Mobile phase C consisted of chloroform-methanol-water-aqueous ammonium hydroxide (450:450:95:5, vol/vol/vol/vol). The elution program was as follows: 100% mobile phase A was held isocratically for 2 min and then linearly increased to 100% mobile phase B for 14 min and held at 100% mobile phase B for 11 min. The LC gradient was then changed to 100% mobile phase C for 3 min and held at 100% mobile phase C for 3 min, and finally returned to 100% mobile phase A over 0.5 min and held at 100% mobile phase A for 5 min. Instrumental settings for negative ion ESI and MS/MS analysis of lipid species were as follows: ion spray voltage (IS) = 24,500 V; current gas (CUR) = 20 lb/in 2 (pressure); gas-1 (GS1) = 20 lb/in 2 ; declustering potential (DP) = 255 V; and focusing potential (FP) = 2150 V. The MS/MS analysis used nitrogen as the collision gas. Data acquisition and analysis were performed using the Analyst TF1.5 software (Sciex, Framingham, MA).
Serum survival test. Overnight cultures of S. mitis were pelleted and washed with PBS, followed by subculturing into different media to an OD 600 of 0.1. Cultures were incubated at 37°C with 5% CO 2 for 8 h. At t = 0 and t = 8 h of incubation, bacterial CFU were quantified by serial dilution and plating on TH agar. E. coli K-12 MG1655 was prepared in a similar way described above and subcultured into 95% complete human serum (Sigma-Aldrich) and 95% heat-inactivated human serum to confirm the presence and absence of bactericidal activity, respectively. E. coli CFU were quantified by serial dilution and plating on LB agar.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.