Reagents were purchased from Sigma Aldrich (St. Louis, USA) or Thermo Fisher Scientific (Waltham, USA) unless otherwise specified. Factor 11c antiserum (Statens Serum Institut, Copenhagen, Denmark) was used to detect serotype 11A, 11E and 11D capsule (20). Hybridoma supernatants containing Hyp11AG2 and Hyp9VG2 murine monoclonal antibodies (mAb), were used for detection of serotype 11A and 9V capsule, respectively, as previously described (21, 22). Hanks’ buffered saline solution (HBSS) plus calcium buffer (HBC) was HBSS supplemented with 2.2 mM CaCl2 and 0.5% bovine serum albumin (BSA). Opsonization buffer (OBB) was HBBS containing Ca2+/Mg2+ supplemented with 0.1% gelatin and 5.2% fetal bovine serum. Cell expression media (CEM) was Dulbecco’s Modified Eagle’s Medium/Nutrient Mixture F12 supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 750 μg/mL of geneticin (Life Technologies, Carlsbad, USA).

Normal human serum (NHS) was obtained from a healthy adult volunteer donor according to a protocol approved by the University of Alabama at Birmingham Institutional Review Board (IRB-140618001). For a complement source lacking classical pathway activity, we used commercially available C1q-depleted human serum (Complement Technology, Inc., Texas, USA; A300), which we have shown to also be depleted of ficolin-2 (ΔC1f2 serum) (23). Affinity-purified recombinant ficolin-2 (rFicolin-2) or CEM supernatant containing 1.2mg/L rFicolin-2 (CEMf2), was obtained from huf2E, a Chinese Hamster Ovary K1 cell line transfected with the human FCN2 gene, as previously described (24). In inhibition experiments, preparations of ΔC1f2 serum with rFicolin-2 were preincubated on ice for 60 min alone or with the following final concentrations of inhibitors: 10 mg/L acetylated BSA (Sigma Aldrich, St. Louis, USA), 10 mg/L BSA (Sigma Aldrich, St. Louis, USA), 100 mg/L 11A capsule polysaccharide (American Type Culture Collection, Manassas, USA), 100 mg/L of cell wall polysaccharide/teichoic acid (Staten Serum Institut, Copenhagen Denmark), or 1:100 diluted silica clot activator (SCA) obtained through elution of a 10-ml plastic red-top tube (catalog no. 367820; BD Biosciences, Franklin Lakes, USA) with 1 ml of water, as previously described (23).

The strains used in this study are listed in Table 1 . Clinical isolate 4011-06 [obtained from the USA Centers for Disease Control and Prevention (CDC) as a “serotype 11A” specimen (19, 27)] contained the three strains MNY31, MNY32, and MNY33, which were identified and subcloned according to their distinct degrees of reactivity with Hyp11AG2 mAb (21). The clinical strains AP278, MNZ2293-A, and MNZ2293-C were characterized previously (21, 25). The laboratory and recombinant strains SPEC6B, MNZ272, JC03, JC04, AMB03, and TIGR-JS were described previously (19, 21, 28). Reference strains TIGR4, D39, and R36A were kindly provided by David Briles (Birmingham, USA). Unless otherwise noted, bacteria were cultured at 37°C and 5% CO2 on blood agar plates (BD Biosciences, Franklin Lakes, USA) or in Todd Hewitt broth or agar (BD Biosciences, Franklin Lakes, USA) with 0.5% yeast extract (THY). Freezer stocks were made from THY broth cultures (OD600 = 0.6) supplemented with 15% glycerol and stored at -80°C until needed.

Primers used in this study are listed in Table 1 . Plasmid pTEX2 was kindly provided by Reinhold Bruckner (Nürnberg, Germany) and contains a cargo region under control of Tet-On inducible promoter (29). Plasmid pONC6.8 was obtained by fusing pTEX2 linearized without its cargo (PCR amplified with primers 5964 and 3299) and a 1052 bp DNA fragment containing wcjE (PCR amplified with primers 5965 and 3300 using MNZ272 genomic DNA as template), using HiFi DNA assembly per manufacturer recommendations (New England Biolabs, Ipswich, USA); and then performing site-directed mutagenesis using primer 51092 and 3925 to replace the native wcjE TTG start codon with a TAAGGAGGCATTTGAATG ribosomal binding site (underlined) plus ATG start codon (bolded). Recombinant strains OC6.8 and FG02 were obtained by transforming pONC6.8 into the serotype 11E strain JC04 and serotype 9A strain JC02, respectively, with selection on THY agar containing 15μg/mL trimethoprim. This resulted in insertion of cargo DNA adjacent to the genomic bgaA site. Sequence of insert introduced 0C6.8 is available on NCBI under accession number MZ054181. Expression of wcjE was induced by incubating bacteria in THY broth containing 100µg/mL anhydrous tetracycline (Abcam, Cambridge, UK) at 37°C for 30 min.

Bacterial freezer stocks were thawed, washed twice, and resuspended to an OD₆₀₀ = 0.5 in HBC. Suspended bacteria were then mixed 1:1 with HBC containing either 10% NHS, 0.1% factor 11c antiserum, 1% Hyp11AG2 or Hyp9VG2 supernatant, or 20 µg/mL of biotin-conjugated Dolichos biflorus agglutinin (DBA, Vector Laboratories, Inc., Burlingame, USA), which detects the Forsmann antigen on pneumococcal LTA/TA (30), and then incubated on ice for 60 min. Bacteria were washed and resuspended in HBC containing fluourescein-labeled anti-ficolin-2 antibody, anti-rabbit antibody, anti-mouse IgG antibody, or streptavidin, respectively, and incubated in the dark on ice for 30 min. Bacteria were washed, suspended in 200µL of PBS with 5µM of DRAQ5, and incubated at room temperature for 30 minutes. Bacteria were washed and suspended in 20µL of 10% neutral buffered formalin solution (Sigma-Aldrich, St. Louis, USA). Bacteria were deposited and allowed to settle on a coverslip for 30 min at RT. After gently washing unbound bacteria with PBS, coverslips were placed onto 20µL of Fluoromount G (Southern Biotech, Birmingham, USA) deposited on a microscope slide, incubated at 4°C for 24 h, and sealed with nail polish. Specimens were visualized on an A1R SIM fluorescence microscope (Nikon, Tokyo, Japan) with 1x100 oil objective and 1.45 numerical aperture, and captured with a Hamamatsu SIM camera. 3D reconstitutions, deconvolution and gamma adjustments were performed using Element software (Nikon, Tokyo, Japan).

Gelatin veronal buffer (GVB, Sigma-Aldrich, St. Louis, USA) supplemented with 13.3% ΔC1f2 serum and 16.6% CEMf2 was preincubated with or without inhibitors on ice for 15 min. 75µL of prechilled mixture was added to 25µL of GVB containing 5 × 105 CFUs of bacteria in microplate wells. Plates were incubated for 30 min at 37°C in 5% CO2 with shaking and then placed on ice to stop complement deposition. After bacteria were washed with ice-cold GVB, surface bound C3 or C4b/C4c bound were stained using murine anti-human C3 mAb (ThermoFisher Scientific, Waltham, USA; LF-MA0132, clone 28A1) and PE-conjugated anti-mouse Ab (BD Biosciences, Franklin Lakes, USA), or FITC-conjugated murine anti-human C4 mAb (ThermoFisher Scientific, Walktham, USA), and detected by flow cytometry using BD Accuri C6 Plus (BD Biosciences, Franklin Lakes, USA) and FCS Express software (Pasadena, USA).

OPK assays were performed as previously described (17). Briefly, 7µg/mL of affinity-purified rFicolin-2 diluted in OBB was mixed 1:1:1 with OBB supplemented with 24% ΔC1f2 serum (normal or heat-inactivated), and with OBB alone or OBB with inhibitors. Following incubation on ice, 30µL of prechilled mixture was added to microwells containing 1x103 CFU bacteria suspended in 10µL of OBB. After 15 min incubation with shaking at 37°C, 40µL of OBB containing 4x105 HL60 cells (American Type Culture Collection, Manassas, USA, CCL-240) differentiated with DMF, was added to each well. Plates were incubated with shaking at 37°C for 30 min. Finally, 10uL from each microwell was spotted on THY agar plates, which were incubated overnight at 37°C with 5% CO2. CFUs were enumerated using ProtoCOL colony-counting software (Synbiosis, Cambridge, UK), in accordance with the well-characterized UAB OPK protocol (described at: http://www.vaccine.uab.edu. Accessed 2 July 2021).

Genomic DNA extracted from strains MNY31, MNY32, MNY33, MNZ2293-A and MNZ2293-C using Monarch Genomic DNA purification kit (New England Biolabs, Ipswitch, USA) served as templates to construct DNA libraries with Nextera XT DNA sample preparation kit (Illumina, San Diego, USA). Sequencing was performed by the UAB Heflin Center Genomics Core Lab using the MiSeq platform (Illumina, San Diego, USA). Reads had adapters removed with Trimmomatic v0.38 (31), were assembled into draft genomes using de-novo assembler Unicycler v0.4.7 (32). Raw reads and assembled contigs are available on NCBI under Bioproject PRJNA779043. Scaffolds.fasta files were used for downstream analysis.

For clonality analyses, 39 additional pneumococcal serotype 11A genome assemblies from prior studies ( Table S1 ) (33, 34), which represent strains obtained from 18 individuals, were obtained from NCBI. Species identity of the 39 NCBI genomes and five genomes from the current study, was confirmed using the ANIm method from pyANI v0.2.7 (35) and the cutoff of ANIm≥96% compared to a serotype 11A the reference genome (GenBank accession no. CP018838). Prokka v1.13.78 was run on all scaffold files to enumerate and identify open reading frames >500 bp in length. To identify core and accessory genomes (determined according to 95% nucleotide identity), we used.gff files produced by Prokka and Roary v3.13 (36) with the script “roary -e -n -p [number of genomes] -i 95 *.gff”. Using the “core_gene_alignment.fasta” as input, Snp-sites v2.4.0 (37) was used to remove indels and create multiFASTA alignment containing the core genome single nucleotide polymorphism (SNP) sites for each core genome. We calculated pairwise core genome SNP distance (from snp-sites multiFASTA alignment) and total genome average nucleotide identity (using pyANI results). The relatedness of strains obtained from the same patient served as references to empirically determine a cutoff defining clonal relationships between isolates, according to histogram visualization.

To clarify the role of the LP in innate immunity against pneumococcal disease, we evaluated ficolin-2’s interaction with pneumococcal surface structures using fluorescent microscopy. We examined three closely related serogroup 11 strains: MNY31, whose surface is extensively decorated with wcjE-mediated 11A PS (detected with Hyp11AG2 mAb); MNY32, an 11E strain lacking 11A PS despite comparable expression of serogroup 11 capsule polysaccharide (detected with factor serum 11c); and MNY33, an 11A variant (11Av) expressing low levels of 11A PS that localized to the bacterial septa ( Figure 1A ). As expected, when bacteria were incubated with 5% human serum, ficolin-2 diffusely bound MNY31 and did not bind MNY32 ( Figure 1A , fourth column). Notably, mirroring the focused distribution of 11A PS, ficolin-2 principally bound to the septa of MNY33. 11A PS and ficolin-2 ligands similarly colocalized to the septa of a different 11Av strain, AP278 ( Figure 2A ) (25).

We then evaluated whether wcjE mediates ficolin-2 binding through the modification of non-capsule surface structures by introducing an intact 11A wcjE allele under a tet promoter into multiple strains. As expected, exogenous wcjE expression in the 11E recombinant strain, OC6.8, was sufficient to restore 11A PS production and ficolin-2 binding ( Figure 2B ). In contrast, though exogenous wcjE expression was sufficient to mediate the synthesis of 9V PS (detected with Hyp9VG2 mAb) by a 9A recombinant strain, FG02, it did not result in ficolin-2 binding ( Figure 1A ). Lastly, confirming prior observations from flow cytometry and serum depletion assays (17, 18), we detected no binding of ficolin-2 to pneumococci expressing various non-serogroup 11 capsule types (i.e., TIGR4, D39, and SPEC6B, expressing serotypes 4, 2, and 6B capsule PS, respectively) or strains lacking capsule PS (i.e. TIGR-JS, R36A, and AMB03, with the latter being an 11A recombinant strain lacking essential capsule synthesis machinery but conserving an intact wcjE) ( Figure 2C ). This was despite all tested strains being bound by biotinylated Dolichos biflorus agglutinin (DBA), a lectin that binds the Forssmann antigen characteristically present in the non-reducing end of pneumococcal TA/LTA (30) ( Figure 2C ). Altogether, these findings confirm that ficolin-2 directly binds specific wcjE-dependent capsule PS structures.

Serotype switching from serotype 11A to 11E through mutational inactivation of wcjE results in evasion of ficolin-2-mediated complement deposition and subsequent antibody-independent opsonophagocytic killing (OPK) (17). To investigate whether the partial reduction of wcjE-mediated 11A PS displayed by 11Av pneumococci (21) also eludes this innate defense, we first examined complement deposition on MNY31, MNY32, and MNY33 bacteria incubated in human serum depleted of both ficolin-2 and C1q, the key recognition molecule of the classical pathway of complement (23). Upon adding recombinant human ficolin-2 (rFicolin-2), we observed substantial C4 and C3 deposition on the 11A strain MNY31, with only minimal C3 deposition on the 11Av strain MNY33 and negligible deposition of either C4 or C3 on the 11E strain MNY32 ( Figures 1B, C ). This complement deposition was eliminated by preincubating rFicolin-2 with purified 11A capsule PS, silica clot activator (SCA), or acetylated bovine serum albumin (acBSA), which are all established inhibitors of ficolin-2 binding (17, 38), but not with pneumococcal TA or non-acetylated BSA. Lastly, consistent with prior findings (17), we observed high degree of in vitro ficolin-2-dependent, C1q-independent OPK of MNY31 (<10% of survival), greatly reduced ficolin-2-mediated OPK of MNY33 (~75% survival), but no significant killing of MNY32 ( Figure 1D ). In all cases, OPK was abrogated with heat inactivation of serum or preincubating rFicolin2 with binding inhibitors ( Figure 1D ). This demonstrates that even partial loss of 11A capsule PS expression confers a survival benefit against ficolin-2-dependent OPK.

As a species, pneumococci rely on commensal colonization of the human nasopharynx (NP), a host niche where serum lectins like ficolin-2 have a minimal presence. While 11A is among the most epidemiologically prevalent pneumococcal serotypes identified in NP carriage studies, it demonstrates the lowest propensity among major serotypes to cause invasive pneumococcal disease (IPD) (17). In contrast, the 11E/11Av phenotypes are almost exclusively identified among isolates causing disease, especially IPD (26), and 11E/11Av isolates harbor a wide variety of wcjE mutations that are rarely shared between two strains (19, 21, 25, 26). The disproportionate epidemiology of 11E/11Av and lack of clonally propagated wcjE mutations is consistent with these pneumococci encountering an environment favoring the convergent, “dead-end” microevolutionary inactivation of wcjE later in the infection process.

Indeed, MNY31, MNY32, and MNY33, as well as the previously described 11A/11Av pair MNZ2293-A and MNZ2293-C (26), comprise groups of strains co-isolated from single blood cultures obtained in the USA and Israel, respectively. Whole genome sequencing and clonality analysis supports that these co-isolates originated from a single colonizing clone. The pairwise whole genome average nucleotide identity (wgANI) and core genome single nucleotide polymorphism distances (cgSNP) between MNY strains (wgANI=99.9972-99.9976% and cgSNP=4-6) and MNZ2293 strains (wgANI=99.9965% and cgSNP=8) were comparable to the values displayed by presumptive clonal strains belonging to serotype 11A lineages in prior studies ( Figure 3 ) (33, 34).

MNY31 and MNZ2293-A each contain an intact and putatively functional wcjE (21). MNY32 and MNY33 contain an ISSpn5 transposon inserted at base pair 501 and an IS1515 transposon inserted at base pair 969 of wcjE, respectively. These insertions resulted in a five (MNY32) or three (MNY33) base pair direct repeat sequence flanking the respective insertion sequences. MNZ2293-C contains a C502T nonsense mutation in wcjE, as previously reported (26). Thus, these co-isolated strains likely represent instances in which wcjE-deficient variants (MNY32/MNY33 and MNZ2293-C) were in the process of replacing their respective 11A precursors (MNY31 and MNZ2293-A) during IPD.