Massively parallel barcode sequencing revealed the interchangeability of capsule transporters in Streptococcus pneumoniae

Abstract

Multidrug/oligosaccharidyl-lipid/polysaccharide (MOP) family transporters are essential in glycan synthesis, flipping lipid-linked precursors across cell membranes. Yet, how they select their substrates remains enigmatic. Here, we investigate the substrate specificity of the MOP transporters in the capsular polysaccharide (CPS) synthesis pathway in Streptococcus pneumoniae. These capsule flippases collectively transport more than 100 types of capsule precursors. To determine whether they can substitute for one another, we developed a high-throughput approach to systematically examine nearly 6000 combinations of flippases and substrates. CPS flippases fall into three groups: relaxed, type-specific, and strictly specific. Cargo size and CPS acetylation affect transport, and we isolated additional gain-of-function flippase variants that can substitute for the peptidoglycan flippase YtgP (MurJ). We also showed that combining flippase variants in a single cassette allows various CPS precursors to be flipped, which may aid glycoengineering. This study reveals that MOP flippases exhibit broad specificity, shaping the evolution of glycan synthesis.

Fig. 1.. Diversity of the capsule flippases and their substrates.

(A) Structures of the CPS repeating units were retrieved from the carbohydrate structure database (52). Pairwise comparisons were performed by calculating their Tanimoto coefficient. Hierarchical clustering depicts the differences in their chemical structures. (B) Structural models of CpsJ generated by AlphaFold (47). Phylogenetic trees were constructed with MEGA (50) by calculating the root mean square deviation (RMSD) of the CpsJ models. (C) Synthesis of the serotype 34 capsule. The repeating unit is assembled on the lipid carrier undecaprenyl phosphate (Und-P) in the cytoplasm. Sugar residues are sequentially added by activities of the polyprenol phosphoglycosyl transferase (PGT) CpsE and glycosyltransferases (GTs). The completed repeat units are transported across the cell membrane by Cps34J before polymerization and ligation to the peptidoglycan. (D) The structural model of Cps34J is predicted by AlphaFold, showing two pseudosymmetrical bundles of transmembrane helices, which adopt an inward-facing configuration and are colored based on their sequence conservation. Variable and conserved regions are colored in cyan and magenta, respectively. (E) cps23BJ** alleles identified in this study. cps23BJ(P254S) was mutagenized and introduced into strain NUS1992 [P_Zn-ytgP ∆ytgP::P-erm] to identify mutant alleles that could substitute for the function of ytgP (the peptidoglycan lipid II flippase in S. pneumoniae). The P254S mutation is highlighted in green. Amino acid changes that relax the specificity of Cps23BJ are colored blue (30), while previously unidentified residues are shown in red. See also fig. S1.

Fig. 2.. Interchangeability of capsule transporters.

(A) A high-throughput approach to examine cross-complementation of capsule transporters. In total, 87 cpsJ alleles were cloned upstream of a DNA barcode, and the constructs were pooled and used to transform 1 of the 79 capsule-switch mutants (29). The resulting transformants are merodiploid with two copies of cpsJ genes. Next, the native cpsJ flippase was deleted (∆cpsJ), and the survivors were expected to carry a CpsJ variant that could replace the native cpsJ. The survivors were pooled, and their genomic DNA was purified. The barcodes of the ectopic cpsJ constructs were then amplified and sequenced (Bar-seq). The number of reads recovered for each barcode represents the fitness of the mutant carrying the corresponding cpsJ allele. Cross-complementation of the flippases is illustrated by a clustered heatmap generated using Clustvis (49). (B) The barcoded cpsJ constructs are functional. Strains NUS0353 [Serotype 19B isogenic capsule-switch mutant; CPS19B] and NUS2944 [CPS19B P-kan-cps19BJ-barcode] were grown in BHI broth at 37°C in 5% CO₂ until they reached the early exponential phase. Cultures were induced by competent stimulating peptides and transformed with the ∆cps19BJ cassette. Transformants were plated on blood agar plates and incubated overnight at 37°C in 5% CO₂ before imaging. (C) Immunostaining with and without anti-serogroup 19 antisera confirmed the presence of CPS in the ∆cps19BJ // cps19BJ-barcode mutant. A positive Quellung reaction was shown on the right with a plus sign. Scale bar, 2 μm. (D) Bar-seq experiments confirmed the cpsJ alleles that could complement cps2J (21). (E) Shown is a heatmap depicting the interchangeability of cpsJ alleles. The blue color represents the natural log ratios of the NGS read counts after the cpsJ allele at the native locus was deleted. Shown in (B) and (C) are representative images from three biological replicates. See also figs. S2 to S4.

Fig. 3.. Capsule transporters can be classified into three groups.

Hierarchical clustering done by Clustvis (49) on the heatmap shown in Fig. 2E revealed three classes of cpsJ alleles. Type-specific CpsJ transporters are specific to certain CPS precursors, likely recognizing some structural features like galactofuranose and ribofuranose residues at the nonreducing ends (red and orange boxes on the right). Strictly specific CpsJ flippases (as indicated by the arrows) exclusively transport their cognate precursors. Last, CpsJ flippases with relaxed specificity (“relaxed”), such as Cps23BJ** and Cps7AJ, can transport many types of cargo. The reducing ends are on the right, and the curved arrows on the left indicate the polymerization sites.

Fig. 4.. Substrate features that affect transport by CpsJ.

(A) The additional glucose residue in serotype 19C may impede substrate translocation by Cps33AJ and Cps43J. (B) The lack of a functional wchU (i.e., 19B) enabled complementation of cps19CJ by cps33AJ and cps43J. Shown are the results of Bar-seq. An increase in the ratios of the read counts in the libraries after and before cpsJ deletions suggests transportation of the indicated substrates by the ectopic cpsJ allele. The findings were validated by reconstructing the strains. Briefly, strains NUS4071 [CPS19B // P-cps33AJ], NUS4072 [CPS19B // P-cps43J], NUS4074 [CPS19C // P-cps33AJ], NUS4075 [CPS19C // P-cps43J], NUS4188 [CPS19C ∆wchU // P-cps33AJ], and NUS4189 [CPS19C ∆wchU P-cps43J] were grown in BHI broth at 37°C in 5% CO₂ until they reached the early exponential phase. Cultures were transformed with the ∆cps19CJ or ∆cps19BJ cassette as indicated. Transformants were plated on blood agar plates and incubated overnight at 37°C in 5% CO₂ before imaging. See also Table 1. Experiments were done twice with similar results. (C) Acetylation impedes translocation of the serotype 18F substrate by Cps15BJ. (D) Cps15BJ could complement Cps18CJ but not Cps18FJ. The findings were validated by reconstructing the strains. Briefly, strains NUS4203 [CPS18C // P-cps15BJ], NUS4204 [CPS18C // P-cps15FJ], NUS4205 [CPS18F // P-cps15BJ], and NUS4206 [CPS18F // P-cps15FJ] were grown in BHI broth at 37°C in 5% CO₂ until they reached the early exponential phase. Cultures were transformed with the ∆cps18FJ or ∆cps18CJ cassette as indicated. Transformants were plated on blood agar plates and incubated overnight at 37°C in 5% CO₂ before imaging. See also Table 2. Experiments were done twice with similar results. See also fig. S5.

Fig. 5.. Expression of flippases with relaxed specificity causes growth defects.

(A to C) Cells expressing Cps34J but not Cps14J or Cps28J exhibited growth defects. The cognate flippases are highlighted in blue. Dots below the orange line indicate a >2-fold decrease in abundance, and dots below the red line have a >5-fold decrease. (D) Cps34J is toxic in the serotype 14 background. Strains NUS0403, NUS4643, NUS4644, and NUS5047 were grown in BHI broth at 37°C in 5% CO₂ until the OD₆₀₀ was between 0.2 and 0.4. Cultures were normalized based on their optical density, serially diluted, and spotted on blood agar without (left) or with (right) added ZnCl₂ and MnCl₂. (E) Strains NUS0671, NUS4207, NUS4285, and NUS5048 were grown in BHI and spotted on blood agar plates as described in (D). (F) Strains NUS4643 and NUS4644 were grown to the early log phase and normalized to an OD₆₀₀ of 0.1. Growth was continued for an hour with or without the supplement of ZnCl₂ and MnCl₂ before imaging by phase contrast microscopy. Red arrows indicate defective cells. Scale bar, 2 μm. (G and I) The areas of the cells were quantified by MicrobeJ [n = 889 for (G) and 1689 for (I)]. P values were computed by Mann-Whitney U tests. n.s., not significant. (H) Strains NUS4207 and NUS4285 were grown and imaged as described in (F). Shown in (D), (E), (F), and (H) are representative images from three biological replicates. See also figs. S6 to S15.