Glycosyltransferases within the psrP Locus Facilitate Pneumococcal Virulence

PsrP has previously been identified as a necessary virulence factor for many serotypes of S. pneumoniae and studied as a surface glycoprotein. Thus, studying the effects on virulence of each glycosyltransferase (GT) that builds the PsrP glycan is of high importance.

ABSTRACT

The pneumococcal serine-rich repeat protein (PsrP) is a high-molecular-weight, glycosylated adhesin that promotes the attachment of Streptococcus pneumoniae to host cells. PsrP, its associated glycosyltransferases (GTs), and dedicated secretion machinery are encoded in a 37-kb genomic island that is present in many invasive clinical isolates of S. pneumoniae. PsrP has been implicated in establishment of lung infection in murine models, although specific roles of the PsrP glycans in disease progression or bacterial physiology have not been elucidated. Moreover, enzymatic specificities of associated glycosyltransferases are yet to be fully characterized. We hypothesized that the glycosyltransferases that modify PsrP are critical for the adhesion properties and infectivity of S. pneumoniae. Here, we characterize the putative S. pneumoniae psrP locus glycosyltransferases responsible for PsrP glycosylation. We also begin to elucidate their roles in S. pneumoniae virulence. We show that four glycosyltransferases within the psrP locus are indispensable for S. pneumoniae biofilm formation, lung epithelial cell adherence, and establishment of lung infection in a mouse model of pneumococcal pneumonia.

IMPORTANCE PsrP has previously been identified as a necessary virulence factor for many serotypes of S. pneumoniae and studied as a surface glycoprotein. Thus, studying the effects on virulence of each glycosyltransferase (GT) that builds the PsrP glycan is of high importance. Our work elucidates the influence of GTs in vivo. We have identified at least four GTs that are required for lung infection, an indication that it is worthwhile to consider glycosylated PsrP as a candidate for serotype-independent pneumococcal vaccine design.

INTRODUCTION

Streptococcus pneumoniae is a Gram-positive bacterial pathogen that colonizes the host nasopharynx and upper respiratory tract. Although colonization in healthy individuals is asymptomatic, dissemination of the bacterium to otherwise sterile sites can prompt invasive pneumococcal diseases (IPDs) such as pneumonia (lungs), otitis media (middle ear), bacteremia (bloodstream), or meningitis (meninges) (1, 2). Interestingly, the Streptococcus genus has tremendous potential for glycosylating surface proteins that aid in invasion into deeper host tissues (3). S. pneumoniae is the leading cause of all community-acquired pneumonia cases (4). IPDs account for one million deaths in children less than 5 years of age each year (5). Despite vaccination programs and availability of antibiotic therapies, the mortality rate for pneumococcal pneumonia remains as high as 40% in infants and the elderly (6, 7).

The pneumococcal serine-rich repeat protein (PsrP) is an adhesin protein that mediates bacterial attachment to host as well as other bacterial cells (8). This large glycoprotein is encoded in a 37-kb genomic island, psrP-secY2A2 (Fig. 1A). The psrP island is present and conserved in many invasive clones of S. pneumoniae (9). These loci carry all of the genes necessary to glycosylate and export the PsrP to the bacterial surface (10). PsrP consists of a signal peptide and a short, glycosylated serine-rich repeat region, followed by a basic region composed mostly of lysine residues, a very long, glycosylated serine-rich repeat region, and a domain to anchor the protein to the cell wall peptidoglycan.

FIG 1.

(A) Genomic organization of the psrP-secY2A2 locus. Shaded arrows indicate putative transport genes, whereas white arrows indicate putative or experimentally determined glycosyltransferases (UniProt TIGR4 gene products SP_1772 [PsrP] to SP_1755). (B) Domain organizations of the glycosyltransferases encoded by the psrP locus, along with proposed functions for GTs based on in vitro transfer assays performed by Jiang et al. (12).

Previous studies have reported that gtfA catalyzes the first step of PsrP glycosylation with the addition of an N-acetylglucosamine (GlcNAc) residue (11, 12), Gtf3 catalyzes the addition of a glucose residue, and the third and fourth steps add glucose or galactose moieties to form 6 potential glycan modifications through in vitro transfer of nucleotide sugar donors to a PsrP-derived peptide by recombinant enzymes (12). These glycan structures are summarized in Fig. 1B along with domain architecture of the glycosyltransferases (GTs) used in this study. Previously, it was postulated that the glycosylation status of PsrP may be different in vivo (10).

Multiple reports have implicated PsrP in the establishment of lung infection in murine models (8–10, 13–15). As of yet, the specific roles of the PsrP glycans in disease progression and bacterial physiology are not completely understood. Moreover, the activities of psrP locus glycosyltransferases that build the mature, in vivo glycan and their potential role in bacterial virulence mechanisms have yet to be fully characterized. In this study, we began to investigate the S. pneumoniae psrP locus glycosyltransferases and elucidate their roles in S. pneumoniae virulence. We generated single-gene mutant strains for gtfA, gtf3, glyD, glyE, glyA, and glyF. We determined the enzymes necessary for S. pneumoniae biofilm formation, lung epithelial cell adherence, and establishment of pneumococcal pneumonia in a mouse intratracheal infection.

RESULTS

PsrP is expressed in ΔGT mutant strains of S. pneumoniae.

Glycan modifications on proteins often contribute to protein stability and solubility (16). To determine if deletion of PsrP-modifying glycosyltransferases influences PsrP surface localization, we probed wild-type (WT) or deletion strains with serum generated against the polybasic N terminus of the protein, termed 72N. In a whole-cell enzyme-linked immunosorbent assay (ELISA) (Fig. 2A), wild-type and GT mutant strains showed high 72N serum IgG binding, whereas the Omega (ΔpsrP-secY2A2) strain was not recognized by the antisera, indicating that the GT mutant strains display PsrP. Furthermore, cell lysates from wild-type and mutant strains were separated by SDS-PAGE and blotted with 72N serum. A high-molecular-weight protein band was reactive in all ΔGT strains but not the Omega strain (Fig. 2B); total protein load was normalized in each sample. These results were supported by flow cytometry on selected ΔGT strains (Fig. 2C), confirming similar levels of surface-localized PsrP. Proteomics of Δgtf3 extracts identified PsrP peptides SAVLEKTVEK, VTNDGSKLTFTYTVTYVNPK, and SASTSASASASTSASASASTSASE detected by mass spectrometry after trypsin and GluC digestion, whereas no PsrP-derived peptides were detected in the Omega strain (Fig. 2D; see also Tables S1 and S2 in the supplemental material). These data indicate that PsrP is expressed on the surface of ΔGT mutant strains, and defects in glycosylation do not drastically influence protein surface localization or secretion. Reverse transcription-PCR (RT-PCR) was performed using primers designed to amplify a fragment within the glycosyltransferase gene, and knockouts were confirmed by lack of amplification. Results confirmed that recombination did not result in significant polar effects on adjacent genes (Fig. 2E).

FIG 2.

(A to C) Whole-cell ELISA (A), Western blot (B), and flow cytometry (C) showing expression and surface localization of PsrP in ΔGT strains. Wild-type and mutant strains were probed with antisera raised against the N-terminal basic region of PsrP. (D) Proteomic identification of PsrP peptides present in the Δgtf3 strain but not in the psrP-deficient Omega strain. (E) RT-PCR of adjacent genes in the psrP locus in mutant strains to confirm knockout and demonstrate no polar affects from knockout procedure. RNA was isolated from mutant strains, and cDNA was prepared by reverse transcriptase. No reverse transcriptase was used as control to ensure no genomic DNA contamination in reactions (data not shown). Statistical analysis was performed using one-way ANOVA with a P value of <0.0001 and Sidak’s multiple-comparison test. ns, P > 0.05; **, P < 0.01; ****, P < 0.0001.

Glycosylation mediates adherence to lung epithelial cells.

A critical step in disease progression to pneumococcal pneumonia is adherence to lung epithelium (2). Because PsrP is an adhesion protein, we tested the wild-type and GT mutant strains to determine the role of PsrP glycosylation in an in vitro lung epithelial cell adherence assay using A549 cells. We observed a significant reduction in A549 binding in vitro for all of the mutant strains tested, with the exception of ΔglyD and ΔglyE (Fig. 3). Eighty-nine percent of the wild-type TIGR4 (The Institute for Genomic Research serotype 4) strain that was incubated on the confluent A549 monolayer bound, whereas only 52 to 56% of the most of the mutant strains bound. The binding for the ΔPsrP and Ω strains leads us to believe that this was not PsrP-mediated binding. Again, ΔglyD and ΔglyE mutants had results similar to those for the wild type, at 86% and 93%, respectively. These results demonstrate a critical role for extended glycosylation of PsrP in binding to lung epithelium and led us to postulate that ΔglyD and ΔglyE mutants do not have critical activities that promote this binding or may have redundant enzymatic function in glycan extension.

FIG 3.

A549 adherence of wild-type and mutant strains of S. pneumoniae. The ability of wild-type and mutant S. pneumoniae to bind to a confluent monolayer of A549 cells was examined. Results are displayed as percentage of total input. The assay was performed in duplicate. Statistical analysis was performed using one-way ANOVA with a P value of <0.0001 and Dunnett’s multiple-comparison test. ns, P > 0.05; ***, P < 0.001.

The PsrP locus glycosyltransferases are essential for biofilm formation.

PsrP has been reported as a major adhesin that promotes bacterial aggregation and biofilm formation (8, 10, 13). We hypothesized that glycosyltransferases orchestrating PsrP glycosylation could contribute to establishment of biofilms. We tested the capacity for the wild-type and mutant strains to form a biofilm in a 24-h assay on a polystyrene microtiter plate or glass coverslips for confocal microscopy. Biofilms were stained with crystal violet (Fig. 4A and B) or Syto 9 (Fig. 4C and D). The Ω strain, which is missing PsrP and all associated glycosyltransferases, failed to form a thick biofilm. Additionally, the ΔgtfA, ΔglyF, Δgtf3, ΔglyA, and ΔpsrP strains were significantly impaired in the ability to form biofilms (Fig. 4A and C). Chromosomal complementation of each of the four GT mutants that had a defect in biofilm formation (ΔgtfA, ΔglyF, Δgtf3, and ΔglyA) restored the biofilm forming ability (see Fig. S1 in the supplemental material). Interestingly, the ΔglyE and ΔglyD strains formed biofilms of intensities and thicknesses comparable to those of the wild-type strain, which grew to an average thickness of 27.2 μm (Fig. 4D). Our results suggest that GlyD and GlyE are dispensable in in vitro and in vivo virulence assays. This raises the possibility that these are redundant gene products or play similar roles to build a functional glycoform. Protein BLAST analysis reveals 38% identity across the length of the GlyE sequence and 58% positive amino acid homology (Fig. S2). We generated a double knockout of these enzymes and performed a biofilm formation assay followed by crystal violet staining. While the ΔglyD and ΔglyE strains were able to form biofilms similarly to the WT (Fig. 4E), the double mutant strain was not, indicating that there may be a redundant role for these enzymes in the glycosylation of PsrP to build a glycoform essential for biofilm formation.

FIG 4.

(A and B) Biofilm formation assay with crystal violet staining. Attachment of wild-type and mutant strains to the bottom of a 96-well polystyrene plate after 24 h was examined. Biofilm was stained using crystal violet. (A) Before dissolution of crystal violet, stained bacteria were viewed on a Leica light microscope with a 20× objective. (B) After dissolution of crystal violet, absorbance was read at 550 nm. (C and D) Biofilm formation assay with Syto 9 staining. Attachment of TIGR4 (WT) and mutant strains to borosilicate glass coverslips after 24 h was examined. (C) Biofilm was stained using Syto 9 and imaged by confocal microscopy on an Olympus FV1200 microscope. (D) Average thickness of the biofilms was calculated using the COMSTAT image analysis software plugin on ImageJ. (E) Gel electrophoresis displaying the glyD and glyE genes of wild-type and their absence in ΔglyE ΔglyD S. pneumoniae and biofilm formation assay with crystal violet staining. After dissolution of crystal violet, absorbance was read at 550 nm. Statistical analysis was performed using one-way ANOVA with a P value of <0.0001 and Sidak’s multiple-comparison test. ns, P > 0.05; ***, P < 0.001; ****, P < 0.0001.

Glycosyltransferases are indispensable for lung infection.

It is known that the PsrP locus of the TIGR4 strain of S. pneumoniae is required for lung infection in mice (17). However, the contribution of the PsrP-specific glycosyltransferases to in vivo glycosylation remains unclear. We performed an intratracheal infection of BALB/c mice with the wild-type and mutant strains to determine the roles of each GT in establishment of pneumococcal pneumonia. While one mouse infected with the PsrP locus-deficient Ω strain, and one mouse each from the groups infected with the ΔgtfA, ΔglyD ΔglyE, and ΔglyF strains, had measurable bacterial titers in the bronchoalveolar lavage fluid (BALF), surprisingly, only the wild-type, ΔglyE, and ΔglyD strains were able to consistently infect the lung tissues of these mice (Fig. 5). These results validate our previous observations in vivo, demonstrate a role for the extended glycosylation of PsrP in establishing infection, and further support the notion that GlyE and GlyD have redundant or compensatory biologically relevant enzymatic activities. No measurable bacterial loads were found in the blood of the mice in any groups.

FIG 5.

Intratracheal infection of wild-type and mutant strains of S. pneumoniae. Shown are bacterial titers in bronchoalveolar lavage fluid of wild-type and mutant S. pneumoniae-infected BALB/c mice. Mice were intratracheally infected with 10⁷ CFU of each strain. Statistical analysis was performed using the Kruskal-Wallis test with a P value of 0.0008 and Dunn’s multiple-comparison test. ns, P > 0.05; **, P < 0.01.

Glycosyltransferase sugar nucleotide specificities.

For our initial characterization of the putative and known GTs of the psrP locus, we tested each enzyme for its ability to hydrolyze UDP-sugars to determine sugar donor specificity in the absence of an acceptor substrate (Fig. 6A to E) (18). The reactions were assessed by detecting free UDP after hydrolysis using the UDP-Glo glycosyltransferase assay kit (Promega) and six common available ultrapure UDP-sugars (14). While each enzyme showed UDP-sugar hydrolysis, it is important to consider additional or preferred activities for other S. pneumoniae sugar nucleotide donors, such as dTDP-Rha, UDP-ManNAc, and UDP-FucNAc. The results confirmed the UDP-GlcNAc specificity of GtfA (Fig. 6A). Furthermore, The only two donor sugars hydrolyzed by these GTs were UDP-glucose and UDP-galactose. While GlyE, GlyD, and GlyA preferentially utilize UDP-galactose, Gtf3 shows specificity for UDP-glucose. Surprisingly, GlyF hydrolyzed UDP-glucose and UDP-galactose at very similar levels (Fig. 6B).

FIG 6.

Comparison of UDP-sugar donor specificities of psrP locus glycosyltransferases. Each enzyme was incubated with a 50 μM concentration of each UDP-sugar for 16 h at 37°C, and UDP release was detected by the UDP-Glo (Promega) assay. Data are presented as the ratio of the UDP detected from the reactions with enzyme to the reactions without enzyme. Statistical analysis was performed using one-way ANOVA and Sidak’s multiple-comparison test.

DISCUSSION

There is limited knowledge of PsrP glycosylation and biosynthetic pathways and the potential role of PsrP glycans in S. pneumoniae virulence and immunogenicity of PsrP (8, 10, 11, 17). However, there is an abundance of putative GTs with unknown in vivo properties. Efforts to dissect the enzymatic pathway for PsrP glycosylation have begun to examine specificity and sequence of transfer to PsrP (12), but delineating the oligosaccharide structures on the natively expressed protein has remained out of reach due to its extremely high propensity for glycosylation. Previous reports have implicated the pneumococcal serine-rich protein in bacterial aggregation, biofilm formation, and virulence capacity of S. pneumoniae. However, these studies have defined molecular interactions with the PsrP peptide backbone, specifically the N-terminal basic region of the protein (8, 17, 19). Specific roles of PsrP glycosylation, and contributions of PsrP-modifying glycosyltransferases, have not been elucidated. Moreover, enzymatic properties and specificities of these glycosyltransferases have yet to be fully characterized. The majority of pneumococcal pneumonia clinical isolates contain the psrP locus in the genome, and the downstream glycosyltransferases are surprisingly well conserved, suggesting their critical activities for S. pneumoniae pathogenesis. We hypothesized that many of the glycosyltransferases that modify PsrP are critical for the adhesion properties and infectivity of S. pneumoniae. In this study, using single-gene mutant strains for the putative glycosyltransferases within this locus, we determined the enzymes necessary for S. pneumoniae virulence properties in vitro and in vivo.

Previous reports have suggested that gtfA is important for PsrP stability (10). While our results also support this notion, we are able to detect PsrP with strong intensities in many ΔGT strains with infectivity defects. Previous work by Lizcano and colleagues has demonstrated the importance of PsrP in pneumococcal pathogenesis (13). In addition, it has been demonstrated that S. pneumoniae lung epithelium binding is mediated through the PsrP basic region binding to keratin 10 (19). It has been proposed that extended PsrP glycosylation could interact or aid in stabilization of the capsule or other surface components (13). Consistent with this hypothesis, there is a clear and demonstrable defect in pathogenesis of ΔGT S. pneumoniae strains. We can, therefore, consider that these defects result from altered or missing glycan modifications on PsrP, although direct evidence for glycosylation changes is not presented here.

Our results indicate that GlyD and GlyE may have redundant activities. Additionally, our preliminary UDP-sugar hydrolysis assays suggest that these enzymes both preferentially transfer galactose moieties. Future experiments will determine how these enzymes influence PsrP glycosylation.

Based on reduction in lung epithelial cell binding by most ΔGT strains, we propose the potential existence of additional S. pneumoniae receptors, such as specific surface lectins on lung epithelium that recognize one of the four hypothesized mature PsrP glycoforms contributing to this adherence, potentially the O-GlcNAc-Glc-Gal-Gal glycoform (12). Sequential in vitro glycosylation experiments by Jiang et al. implicate GtfA, Gtf3, either GlyE or GlyD, and GlyA in building this glycoform. Interestingly, those in vitro experiments were not able to identify a role for GlyF toward any proposed glycoforms, although our data suggest an essential role for this enzyme (12). This can be explained in two ways: (i) GlyF could transfer a unique sugar, contributing to the virulent glycoform, or (ii) GlyF acts on an underlying glycan structure that was not replicable in vitro. GlyD and GlyE contributions toward building multiple glycoforms in vitro, combined with their dispensability for lung infection, suggest that while these glycoforms may be possible, they may have a less critical role in virulence mechanisms of S. pneumoniae or have redundant enzymatic functions. There is tremendous conceivable microheterogeneity of glycosylated PsrP, as well as potentially complex and diverse interactions with other components of the S. pneumoniae surface that are critical modulators of biofilm formation and pneumococcal aggregation that will be elucidated in future studies. It was previously postulated that PsrP and its glycans could serve as a tent poles for stabilization of the CPS and other surface proteins (13). Such interactions would significantly influence biofilm formation. Future studies will investigate these interactions at a molecular level.

In conclusion, our results demonstrate that the majority of putative glycosyltransferases within the psrP locus are indispensable for S. pneumoniae biofilm formation, contribute to lung epithelial cell adherence, and support S. pneumoniae lung infection in a murine model of pneumococcal pneumonia. Future studies will utilize the ΔGT strains constructed here to analyze changes in glycosylation profile and determine specific transfer activities of each glycosyltransferase within the PsrP glycosylation pathway by direct glycomic analysis. Furthermore, PsrP glycans may be a promising vaccine target, as the immense size of the protein allows these epitopes to extend past the capsular polysaccharide shield.

MATERIALS AND METHODS

Bacterial strains and mutant construction.

Streptococcus pneumoniae type 4 (TIGR4 strain) and T4ΔpsrP-secY2A2 (Ω), generous gifts from Carlos Orihuela (University of Alabama at Birmingham), were cultured aerobically without shaking at 37°C on tryptic soy agar with 5% sheep blood (TSAB) or in Todd-Hewitt broth plus 0.5% yeast extract (THY; BD Biosciences). Glycosyltransferase knockout (ΔGT) strains were generated by allelic replacement with a “Sweet” Janus cassette (20). Briefly, DNA fragments flanking each gene were amplified and assembled up- and downstream of the cassette containing SacB and kanamycin resistance genes. ΔGT strains were selected and grown with kanamycin supplementation (200 μg/ml). For chromosomal complementation, DNA fragments flanking each gene were assembled up- and downstream of the gene of interest and a chloramphenicol resistance gene and selected and grown with chloramphenicol supplementation (50 μg/ml). Primers used for mutant construction can be found in Table S3.

Mice.

Five-week-old female BALB/c mice were obtained from Taconic Biosciences (Hudson, NY) and housed in the Central Animal Facility at the University of Georgia. Mice were kept in microisolator cages and handled under biosafety level 2 (BSL2) hoods. All mouse experiments were in compliance with the University of Georgia Institutional Animal Care and Use Committee under our approved animal use protocol, which adheres to the principles outlined in U.S. Government Principles for the Utilization and Care of Vertebrate Animals Used in Testing, Research, and Training (https://olaw.nih.gov/policies-laws/gov-principles.htm), the Animal Welfare Act, the Guide for the Care and Use of Laboratory Animals (21), and the AVMA Guidelines for the Euthanasia of Animals. (22).

RNA isolation and RT-PCR.

S. pneumoniae cultures were harvested at mid-log phase (optical density [OD] at 600 nm, 0.6) and RNA was purified using the E.N.Z.A. bacterial RNA kit, followed by TRIzol (Thermo Fisher Scientific) extraction of RNA from contaminating genomic DNA as described previously (23). RNA purity was assessed with NanoDrop, and 1 μg of RNA was used for reverse transcription using the iScript cDNA synthesis kit (Bio-Rad).

72N terminal serum production and whole-cell ELISA.

The coding region of the N terminus of PsrP from signal peptide through two SRR2 repeats (amino acids 1 to 320) was amplified from TIGR4 genomic DNA using primers with adapter attB sites used in Gateway cloning systems (Thermo Scientific). A BP Clonase reaction was performed to insert the gene into the pDONR221 vector. After sequence confirmation, an LR Clonase reaction was performed to insert the gene into the pET-Dest42 destination vector for the expression of a C-terminal His-tagged fusion protein. The vector was transformed, expressed, and purified as described above for S. pneumoniae glycosyltransferases. The purified protein was used to immunize 8-week-old BALB/c mice at 5 μg/dose with alum as an adjuvant. Seven days after booster immunization, serum was taken by tail vein bleeding.

ELISA plates (96 well; Nunc) were coated with 10⁷ CFU of paraformaldehyde-fixed S. pneumoniae strains in phosphate-buffered saline (PBS; pH 7.2) overnight at 4°C. Plates were washed 4 times with PBS plus 0.1% Tween (PBS-T) 20 using a BioTek 405/LS microplate washer. After 1 h of blocking at room temperature with 1% bovine serum albumin (BSA) in PBS, microplate wells were incubated for 2 h with a 1:2,000 dilution of 72N serum in PBS-T. Plates were washed and then incubated for 2 h at room temperature with a 1:2,000 dilution of goat anti-mouse IgG–alkaline phosphatase (Southern Biotech; number 1030-04) in PBS-T. After washing, plates were incubated for ∼30 min at 37°C with 2 mg/ml of phosphatase substrate (Sigma; S0942) in 1 M Tris–0.3 mM MgCl₂. Absorbance at 405 nm was measured on a BioTek synergy H1 microplate reader. Each experiment included two replicates and was repeated three times.

Western blotting.

Overnight cultures of WT and mutant S. pneumoniae strains were pelleted by centrifugation at 5,000 × g for 15 min and resuspended in PBS with 0.1% sodium dodecyl sulfate plus 0.1% sodium deoxycholate. After shaking at 200 rpm at 37°C for 30 min, the samples were centrifuged at 20,000 × g for 30 min. Cell lysates (supernatants) were normalized for protein concentration by NanoDrop (A280) and mixed with loading dye, and 10 μg of total protein was separated by SDS-PAGE on a NuPage 3 to 8% Tris-acetate gel (Invitrogen) at 150 V for 80 min. Proteins were transferred to a polyvinylidene difluoride (PVDF) membrane at 100 V for 50 min. The membrane was blocked with 3% BSA in TBS-T and probed with a 1:2,000 dilution of 72N serum and a 1:4,000 dilution of goat anti-mouse IgG–horseradish peroxidase (HRP) (Biolegend).

Flow cytometry.

A total of 10⁷ CFU of wild-type and selected S. pneumoniae mutant strains were resuspended and incubated overnight in a 50:1 solution of bacterial staining buffer (BSB; 1% BSA in PBS with 0.025% NaN₃) and 72N serum. Cells were washed and resuspended in BSB with a 1:200 dilution of Alexa Fluor 647–goat anti-mouse IgG antibody (BioLegend; number 405322) for 30 min. The bacteria were washed and fixed in BSB with 2% formaldehyde. The fluorescent signal was analyzed on CytoFLEX (Beckman Coulter, Hialeah, FL). Analysis was done with FlowJo v10.1 software (Tree Star, Inc., Ashland, OR).

Proteomics.

Twenty micrograms of total protein prepared as described above was resolved on a NuPage 3 to 8% Tris-acetate gel (Invitrogen). The gel was Coomassie stained, and the faintly stained high-molecular-weight band was excised and destained in 40 mM ammonium bicarbonate–50% acetonitrile. reduced by incubation with 10 mM dithiothreitol (DTT) for 1 h at 56°C, followed by carboxyamidomethylation with 55 mM iodoacetamide in the dark at room temperature for 45 min, and then digested with 1 μg each of sequencing-grade trypsin and GluC (Promega) in 40 mM ammonium bicarbonate overnight at 37°C. The resulting peptides were dried down in a SpeedVac and reconstituted in 0.1% formic acid. The peptides were resolved on an Acclaim PepMap rapid-separation liquid chromatograph (RSLC; C₁₈ column) and analyzed on a Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific). Selected precursors were fragmented using collision-induced dissociation (CID) set to 38%. The raw tandem mass spectrometry (MS/MS) spectra were searched against the uniport proteome database for Streptococcus pneumoniae TIGR4 with a precursor mass tolerance of 10 ppm and fragment tolerance of 0.3 Da. Static modification of +57.021 Da (carbamidomethylation of cysteine residues) and dynamic modification of +15.995 Da (oxidation of methionine residues) were allowed in the search parameters. Results were searched using Byonic software filtered at a 1% false-discovery rate for peptide assignments.

A549 cell adherence assay.

A549 cells were grown to confluence in 24-well plates, washed, and incubated with 10⁷ CFU/ml of wild-type and mutant S. pneumoniae for 2 h at 37°C. After 3 washes with sterile PBS, counts of adhered bacteria were determined by lysis with 0.1% Triton X-100 and plating dilutions (17). Each experiment included three replicates and was repeated two times.

Biofilm formation assay with crystal violet staining.

Equal CFU of mutant and WT S. pneumoniae were inoculated individually into wells of 96-well polystyrene plates containing THY broth in quadruplicate. Cultures were incubated at 37°C and evaluated for biofilm formation at 24 h. Supernatants containing nonadhered cells were discarded and attached biofilms were stained with crystal violet, dried, suspended in 30% acetic acid, and quantified by measuring the absorbance at 550 nm on a BioTek synergy H1 microplate reader (8, 24). For light microscopy, wells were visualized prior to suspension in 30% acetic acid on a Leica DM IL LED light microscope with a 20× lens objective. Each experiment included three replicates and was repeated two times.

Confocal microscopy of biofilms.

Equal CFU of mutant and WT S. pneumoniae were inoculated individually into wells of 24-well polystyrene plates with coverslips at the bottoms of the wells containing THY broth in quadruplicate. Cultures were incubated at 37°C and evaluated for biofilm formation at 24 h. Supernatants containing nonadhered cells were discarded and attached biofilms were stained with Syto 9 (Thermo Scientific) according to the manufacturer’s instructions. After gentle washing, coverslips were mounted on slides with Vectashield mounting medium (Vector Labs). Imaging was performed with an Olympus FV1200 microscope. Quantification of average biofilm thickness was performed using COMSTAT2 image analysis software in ImageJ (25). Each experiment included three replicates and was repeated two times.

Intratracheal infection.

Five-week-old BALB/c mice were anesthetized with a ketamine-acepromazine-xylazine mixture. After effective anesthesia, mice were hung by their incisors with their backs resting on a solid surface. Mice were inoculated intratracheally with 10⁷ CFU of wild-type TIGR4, PsrP locus-deficient, or GT mutant strains in 50 μl of PBS using a gel loading tip to intubate the trachea. Bacterial burden in the lungs was assessed 48 h after infection by obtaining bronchoalveolar lavage fluid at ∼1.5 ml and plating serial dilutions of BALF. The experiment was performed twice. Data are representative of those from one experiment.

Recombinant glycosyltransferase production.

Coding regions of GTs lacking a stop codon were obtained in Gateway clone sets from BEI Resources. An LR Clonase (Thermo Scientific) reaction was performed to insert the gene into the pET-Dest42 destination vector for the expression of a C-terminal His-tagged fusion protein in BL21 cells. Glycosyltransferases were produced as described previously, with minor modifications (24). Briefly, BL21(DE3) cells transformed with pET-DEST42-GT plasmids were grown in Luria broth supplemented with 100 μg/ml of ampicillin at 37°C, and cell density was monitored by absorbance at 600 nm. Once the OD at 600 nm reached 0.6, the cells were transferred into 25°C. Protein expression was induced by the addition of isopropyl-β-d-1-thiogalactopyranoside at a final concentration of 1 mM, and the culture was incubated with shaking for 8 h. Cells were harvested by centrifugation, resuspended in PBS (pH 7.2), and lysed by sonication. The lysate was clarified by centrifugation at 17,000 × g for 1 h at 4°C and passed through a 0.45-μm syringe filter. Enzymes were purified by nickel-nitrilotriacetic acid (Ni-NTA) resin at 4°C and eluted with 300 mM imidazole, and buffer was changed to PBS (pH 7.2). Protein concentration was determined by the bicinchoninic acid assay according to the instructions of the manufacturer. Purity was evaluated by Coomassie staining.

UDP-Glo glycosyltransferase hydrolysis assays.

Ultrapure UDP-Glc, UDP-GlcNAc, UDP-Gal, UDP-GalNAc, and UDP-GlcA were purchased from Promega Corporation. Ultrapure UDP-Xyl was a generous gift from Osman Sheikh from the Wells laboratory at the University of Georgia. The specificity of each recombinant enzyme’s sugar nucleotide hydrolysis activity was determined by incubation of 3 μg of >90% pure protein with a 50 μM concentration of a single UDP-sugar in the absence of an acceptor in 20-μl reaction mixtures containing 20 mM Tris (pH 7.4) and 10 mM MgCl₂ at 37°C for 16 h. Control samples were all reaction components, except enzymes. Hydrolysis reactions were stopped by the addition of the UDP-Glo detection reagent, and detection of free UDP was performed using the UDP-Glo glycosyltransferase assay per the manufacturer’s instructions. Each sugar nucleotide hydrolysis reaction was combined in a ratio of 1:1 (20 μl:20 μl) with the UDP-Glo detection reagent. Ten-microliter volumes of each mixture were placed in independent wells of a white, flat-bottom 384-well plate (Corning) and incubated at room temperature. After 1 h of incubation, relative luminescence units (RLU) were measured using a microplate reader (Synergy H1; BioTek).

Statistical analysis.

Statistical analysis was performed in GraphPad Prism 8 software. Group analysis was performed using one-way analysis of variance (ANOVA) or Kruskal-Wallis test. Statistical significance was assessed in comparison to the WT strain using the default multiple-comparison test in Prism. Specific multiple-comparison tests performed are provided in figure legends.