Identification of Zinc-Dependent Mechanisms Used by Group B Streptococcus To Overcome Calprotectin-Mediated Stress

Group B Streptococcus (GBS) asymptomatically colonizes the female reproductive tract but is a common causative agent of meningitis. GBS meningitis is characterized by extensive infiltration of neutrophils carrying high concentrations of calprotectin, a metal chelator. To persist within inflammatory sites and cause invasive disease, GBS must circumvent host starvation attempts. Here, we identified global requirements for GBS survival during calprotectin challenge, including known and putative systems involved in metal ion transport. We characterized the role of zinc import in tolerating calprotectin stress in vitro and in a mouse model of infection. We observed that a global zinc uptake mutant was less virulent than the parental GBS strain and found calprotectin knockout mice to be equally susceptible to infection by wild-type (WT) and mutant strains. These findings suggest that calprotectin production at the site of infection results in a zinc-limited environment and reveals the importance of GBS metal homeostasis to invasive disease.

granulocytic cells during active infection. Additional studies have shown that, in response to GBS, neutrophils elaborate extracellular traps decorated with lactoferrin (38) and that S100A9, a calprotectin subunit, is present in the blood and amniotic fluid during intrauterine GBS infection (39). These observations suggest that GBS experiences metal limitation during infection, but the mechanisms used by GBS to overcome nutritional immunity remain unknown.
Bacteria utilize a number of strategies to obtain zinc during infection, including direct uptake of the metal, the use of metallophores, and piracy from zinc-bound host proteins. While there is a myriad of strategies employed to obtain zinc, the AdcABC/ ZnuABC family of ATP-binding cassette transporters are present in most bacteria. Streptococcal pathogens S. pneumoniae and S. pyogenes harbor two zinc-binding proteins, AdcA and AdcAII/Lmb, whereas GBS is particularly distinct as it possesses three zinc-binding proteins, AdcA, AdcAII, and Lmb (40,41). AdcA and AdcAII/Lmb have been shown to utilize distinct mechanisms to bind zinc ions and shuttle them through the AdcBC transporter and are important for growth in zinc-restricted environments and infection (42)(43)(44)(45).
Here, we investigated GBS fitness during calprotectin stress using a newly constructed saturated transposon mutant library and targeted amplicon sequencing. We characterized the global requirements for GBS survival during nutritional immunity, identifying 258 mutants, 123 underrepresented and 135 overrepresented, that impact calprotectin sensitivity. We show here that characterized and putative metal transporters are important for calprotectin survival in vitro and that the zinc uptake machinery contributes to survival during calprotectin-induced starvation and invasive disease progression. These results provide insight into zinc-dependent mechanisms that GBS employs to evade the host immune response and nutritional immunity to successfully cause disease and establish a groundwork to study the comprehensive effects of chelation on multimetal transport in GBS.

RESULTS
GBS growth inhibited in the presence of calprotectin. Previous studies have shown calprotectin inhibits bacterial growth by limiting nutrient metal ions (7)(8)(9). To characterize the response of GBS to metal chelation by calprotectin, we assessed growth of GBS strains in the presence of purified calprotectin. Disease clinical isolates A909 (serotype Ia) (46), CJB111 (serotype V) (47), and COH1 (serotype III) (48) were incubated with increasing concentrations of purified calprotectin ranging from 0 to 480 g/ml, and growth was assessed by optical density and plating for viable bacteria. Although various levels of chelation sensitivity were detected, growth of all GBS strains (as measured by optical density at 600 nm [OD 600 ]) was significantly inhibited at high, but still physiologically relevant, calprotectin doses ( Fig. 1A to C). Similar patterns of growth inhibition were observed when growth was assessed by enumerating CFU following an 8-h incubation with calprotectin. Exposure to calprotectin at concentrations higher than 120 g/ml significantly inhibited growth of all GBS strains (Fig. 1D to F), while supplementation of zinc sulfate during calprotectin stress restored growth in all three strains (see Fig. S1A to C in the supplemental material). Additionally, we determined how a panel of 27 vaginal isolates collected from the vaginal tracts of pregnant women (49) survived in the presence of calprotectin. All strains were grown with or without 120 g/ml calprotectin, and data are displayed as percent inhibition in treated versus untreated controls for isolates belonging to five different capsular serotypes. We observed a mean percent inhibition across the isolates that ranged from 30% to 55%. When mean inhibition of each serotype was compared against the others, we observed that vaginal isolates belonging to serotype V exhibited the most variability between strains and were significantly more resistant to calprotectin-mediated chelation than serotype Ia isolates (Fig. 1G). Representative invasive isolates A909, CJB111, and COH1 were included in the isolate panel in their capsular serotype grouping and are denoted by the open shapes in Fig. 1G. These data suggest that while there is variation in levels of sensitivity across strains and serotypes, GBS is broadly sensitive to the antimicrobial activity of calprotectin.
Essential genes for GBS growth in calprotectin. To successfully colonize the host or survive within highly inflammatory environments during infection, GBS must cope with nutritional immunity and, specifically, metal limitation imposed by calprotectin. To identify factors that are important for responding to calprotectin-mediated chelation, we constructed a GBS saturated Krmit transposon (Tn) mutant library in the CJB111 strain background as described previously for group A Streptococcus (50). Analysis of the library revealed 68,857 unique insertion sites across the GBS genome (Fig. 2). To identify essential genes for GBS growth, we outgrew the Tn mutant library in Todd Hewitt broth with yeast extract (THY), modified RPMI medium (mRPMI), and mRPMI plus subinhibitory (60 g/ml) and inhibitory (480 g/ml) concentrations of calprotectin. We recovered CFU from these growth conditions, extracted genomic DNA, and prepared sequencing libraries as described in Materials and Methods. Transposon insertions were sequenced as previously described (51) with minor changes, and sequenced reads were mapped back to the GBS genome. Bayesian statistical analyses (52) identified, in the absence of calprotectin, 206 essential genes for growth in THY and 450 essential genes for growth in mRPMI ( Fig. 3A and B), with 153 essential genes common to both media conditions ( Fig. 3C; see also Table S1). The genes deemed essential for growth in THY and mRPMI were assigned clusters of orthologous groups of proteins (COGs) and were found to be involved primarily in translation, ribosomal structure, and biogenesis (23% in THY, 13% in mRPMI), replication, recombination, and repair (16% in THY, 13% in mRPMI), cell wall/membrane/envelope biogenesis (10% in THY, 10% in mRPMI), and carbohydrate transport and metabolism (10% in THY, 7% in mRPMI) ( Fig. 3D and E).
Bayesian analyses were then used to determine the essential genes for growth in the presence of calprotectin. These analyses compared essential genes from the base medium (mRPMI) and subinhibitory and inhibitory concentrations of calprotectin. From these analyses, we identified 40 genes that were essential specifically for growth in mRPMI, 30 genes that were essential only for growth in a subinhibitory dose of calprotectin, and 55 genes that were essential only for growth in an inhibitory calpro- , and COH1 (C) was assessed by measuring optical density (OD 600 ) following an 8-h incubation with recombinant calprotectin (0 to 480 g/ml) or by quantitating CFU (D to F). (G) Sensitivity was assessed by OD 600 across a panel of vaginal isolates (closed shapes) and invasive isolates (open shapes) following an 8-h incubation with 120 g/ml calprotectin. Data are displayed as percent growth inhibition compared to that of untreated isolate controls. All experiments were performed in technical triplicates (n ϭ 3), and data were averaged from three independent experiments. Significance for panels A to F was determined by Kruskal-Wallis with Dunn's multiple-comparison tests comparing treated samples to untreated controls. Significance for panel G was determined by one-way ANOVA with Tukey's multiple-comparison test. *, P Ͻ 0.05; **, P Ͻ 0.01; ***, P Ͻ 0.001; ****, P Ͻ 0.0001. tectin dose. The remaining 306 essential genes were deemed important for growth across all environments (mRPMI, subinhibitory calprotectin, and inhibitory calprotectin) ( Fig. 3F and G; Table S1).
Global impact of calprotectin on GBS fitness. To determine the global effect of calprotectin on GBS fitness, differential analyses were performed using DESeq2 com- The inner four circles present the results of Bayesian analysis of GBS gene essentiality under different growth conditions (THY, mRPMI, mRPMI plus subinhibitory calprotectin, and mRPMI plus inhibitory dose calprotectin, in order toward center; essential genes (red), nonessential genes (green), and excluded genes in either gray (too small for analysis) or black (inconclusive call). The center circle compiles the summary analysis of GBS genes under all four growth conditions, with essential genes under all conditions (red), nonessential genes under all conditions (yellow), and small/inconclusive genes (gray/black).
Global GBS Response to Calprotectin ® paring samples treated with subinhibitory (60 g/ml) or inhibitory (480 g/ml) calprotectin and untreated mRPMI controls. Genes found to be essential for growth in mRPMI (pink), subinhibitory calprotectin (blue), inhibitory calprotectin (purple), or across two treatments (gray) were excluded from fitness analyses ( Fig. 4A and B). We characterized the global impact of calprotectin stress on GBS growth and identified a total of 258 mutants in the output pool whose growth was significantly impacted by calprotectin. We identified 135 mutations that conferred a fitness advantage for GBS during calprotectin stress, with 94 mutants in subinhibitory-dose and 98 mutants in inhibitory-dose calprotectin ( Fig. 4A and B; Table S1), with 57 mutants common between the two calprotectin treatment groups. We also identified 123 mutants that resulted in a fitness defect during calprotectin stress; of those, 93 mutants were important in the subinhibitory dose of calprotectin and 76 mutants were important in the inhibitory levels of calprotectin ( Fig. 4A and B; Table S1), with 46 mutants observed in both calprotectintreated samples (Fig. 4C). Of the Tn mutants that were identified as underrepresented following treatment with calprotectin, COGs were identified for 76 of the mutations observed in subinhibitory treatment and 60 of the mutations observed in inhibitory calprotectin treatment. The most abundant COGs of known function were those involved in inorganic ion transport, amino acid and carbohydrate transport and metabolism, and defense ( Fig. 4D and E). Approximately 15% of the mutants underrepresented in both concentrations were grouped into the inorganic ion COG and were previously characterized or putative systems involved in metal ion uptake or efflux (Table 1). These systems involved in maintaining metal homeostasis were many of the . Essential genes are depicted as yellow (A) or pink (B), nonessential genes are shown in black, and inconclusive genes are shown in gray. The x axis is a linear representation of the A909 genome. EggNOG 5.0 was used to assign COGs to determine functions for essential genes for growth in THY (D) and mRPMI (E). Venn diagrams depict the essential genes for growth in mRPMI and THY (C) or mRPMI and subinhibitory (60 g/ml) and inhibitory (480 g/ml) calprotectin (F). (G) Linear map represents Bayesian analyses of essential genes for growth in mRPMI and subinhibitory and inhibitory calprotectin. most significantly underrepresented mutants as denoted in the volcano plots (shown in red) for each calprotectin concentration ( Fig. 4A and B).
To gain more insight into how the identified metal transport systems may contribute to homeostasis, all proteins of interest were clustered by ortholog against characterized metal transport machinery in the closely related Streptococcus pneumoniae TIGR4 genome. Of the three genes encoding zinc-binding proteins, we identified only adcA (SAK_0685) and adcAII (SAK_1898) as important for growth in calprotectin. The gene encoding their cognate ATPase, adcC (SAK_0218), and one of the genes encoding FIG 4 GBS genomic fitness screen in calprotectin. Volcano plots identify essential genes for growth in mRPMI and subinhibitory (60 g/ml) calprotectin (A) and inhibitory (480 g/ml) calprotectin (pink, blue/purple, and gray) (B) as well as nonessential genes (black) and genes involved in metal transport (red). (39). Essential genes (all but "nonessential" and "metal transport") were excluded from fitness analyses by DESeq2. (C) Venn diagram depicts the underrepresented mutants detected in low (60 g/ml) and high calprotectin (480 g/ml) and common genes important for growth under both conditions. COGs were assigned to genes that contribute to survival in subinhibitory dose (B) and inhibitory dose (C) of calprotectin using EggNOG 5.0. (F) Schematic of GBS metal importers that contribute to survival during calprotectin stress. Global GBS Response to Calprotectin ® a streptococcal histidine triad protein, shtII (SAK_1897), were also underrepresented in our Tn sequencing analyses (Fig. 4F). Underrepresentation of mutants in adcAII was specific to treatment with inhibitory levels of calprotectin, while mutants in the remaining zinc transport genes detected were defective in calprotectin survival independent of concentration (Table 1).
In addition to the zinc machinery, significant underrepresentation was detected for Tn insertions in the genes encoding the manganese/iron ABC transporter, mtsABC (SAK_1554-1556), and the gene encoding the manganese/iron natural resistanceassociated macrophage protein (NRAMP), mntH (SAK_0871) (53) (Fig. 4F). Ortholog clustering again confirmed the conservation of the GBS transporter MtsABC to the pneumococcal transporter PsaABC (53,54); however, other pathogenic streptococci, including S. pyogenes and S. pneumoniae, are devoid of manganese-and irondependent NRAMP transporters (55)(56)(57). Additionally, mutants in genes that comprise two other putative metal ion uptake systems were identified as underrepresented (Fig. 4F). nikD (SAK_1539) is an ATP-binding protein encoded within the operon nikABCDE (SAK_1538-1542) that encodes a putative but uncharacterized nickel ABC transport system, and the second underrepresented and uncharacterized ABC transport system is encoded by SAK_0252-0255; however, the substrate transported by this system remains unknown (Fig. 4F). The substrate-binding proteins of both putative metal transport systems belong to the NikA/DppA/OppA superfamily.
Calprotectin induces expression of zinc import machinery. Upon identifying mutants in genes encoding two zinc-binding proteins, adcA and adcAII, in our transposon sequencing analyses, we hypothesized that differential expression of genes involved in zinc acquisition might occur following exposure to calprotectin, a natural source of zinc limitation. Quantitative reverse transcriptase PCR (qRT-PCR) analysis of genes encoding the zinc-binding proteins adcA, adcAII, and lmb was performed following treatment with calprotectin or the cell membrane permeable chelator N,N,N=,N=tetrakis(2-pyridinylmethyl)-1,2-ethanediamine (TPEN). TPEN chelates zinc with an extremely high affinity (dissociation constant [K d ] ϭ 10 Ϫ15 M); however, it has been shown to bind other metal ions, including nickel (58), iron (59), and copper (60). Fold changes in gene expression were calculated by the comparative threshold cycle (ΔΔC T ) with gyrA serving as an internal control and were compared to expression observed in untreated controls. Expression of adcA was induced 4-fold following exposure to calprotectin and 10-fold following TPEN treatment, while expression of adcAII and lmb was more robustly upregulated, with 18-and 15-fold inductions, respectively, in response to calprotectin treatment and 320-and 227-fold inductions, respectively, after treatment with TPEN ( Fig. 5A to C). Expression of SAK_0514, ortholog to the cation diffusion facilitator encoded by czcD of S. pneumoniae, was also assessed by qRT-PCR to confirm that calprotectin and TPEN were inducing zinc-limited conditions. As expected, expression of czcD was downregulated following exposure to both (see Fig. S2A). To confirm what was previously described for GBS in zinc-limited chemically defined medium (40,41), we observed that mutants lacking individual zinc-binding proteins exhibited similar calprotectin sensitivity as the wild-type (WT) GBS strain, while growth of a triple ΔadcAΔadcAIIΔlmb mutant was reduced (Fig. S2B). An increased sensitivity to calprotectin was also observed in a triple ΔadcAΔadcAIIΔlmb mutant compared to that for the WT in both A909 and CJB111 strain backgrounds ( Fig. S2C and D). These significant differences, though subtle are consistent with previous results that demonstrate functional redundancy exists between zinc-binding proteins and suggest that additional transport systems could be contributing to survival during calprotectin stress.
Zinc homeostasis contributes to GBS virulence and meningitis. Our results thus far suggest that GBS utilizes zinc uptake machinery to cope with calprotectin stress in vitro; thus, we hypothesized that zinc homeostasis would contribute to GBS virulence. Using a murine model of GBS systemic infection, we infected mice (C57BL/6) intravenously with the A909 WT or the isogenic ΔadcAΔadcAIIΔlmb mutant strain. Infection with the WT GBS strain resulted in significantly higher mortality than with the mutant strain (Fig. 6A). By 36 h postinfection, 7/8 WT infected mice succumbed to infection. Conversely, only 3/8 mice challenged with the ΔadcAΔadcAIIΔlmb strain succumbed to infection by the experimental endpoint of 144 h (Fig. 6A). At the time of death or the experimental endpoint, blood and brains were harvested to determine bacterial load. Despite similar levels of bacterial CFU recovered from blood (Fig. 6B), a significantly higher bacterial burden was observed in brain tissue (Fig. 6C) of WT GBS-infected animals than in animals infected with the ΔadcAΔadcAIIΔlmb mutant strain. We further detected an increase in KC, a neutrophil chemokine, in brain homogenates of mice challenged with WT GBS compared to that in the ΔadcAΔadcAIIΔlmb mutant strain (Fig. 6D), suggesting a more robust infection and increased inflammation in WTinfected mice. Similar results were also observed in the GBS CJB111 background (see Fig. S3A to D) and during infection in another mouse (CD-1) background (Fig. S3E to G). . Recovered CFU were quantified from brain tissue homogenates (B and F) or blood (C and G). (D and H) Cytokine abundance was quantified from brain tissue homogenates by ELISA. Statistical analyses include log rank (Mantel-Cox) tests for panels A and E and unpaired Student's t tests for panels B to D and F to H. *, P Ͻ 0.05; **, P Ͻ 0.01; ns, not significant.
Global GBS Response to Calprotectin ® To determine the contribution of calprotectin specifically to GBS disease progression, we infected S100A9 Ϫ/Ϫ mice with WT and ΔadcAΔadcAIIΔlmb GBS. We observed that S100A9 Ϫ/Ϫ mice were equally susceptible to WT and mutant GBS (Fig. 6E) and had similar bacterial loads in the brain and blood (Fig. 6F and G) and levels of neutrophilic chemokine KC in brain tissue (Fig. 6H). Interestingly, S100A9 Ϫ/Ϫ mice were less susceptible to WT GBS infection than WT mice. Taken together, these data indicate that GBS zinc homeostasis is required for invasive disease, specifically, in the presence of host calprotectin.

DISCUSSION
GBS infections are known to result in increased immune cell influx and inflammation, specifically, neutrophilic infiltrate (35,61). As these are characteristic signs of bacterial meningitis, GBS would encounter high concentrations of granulocyte-derived calprotectin (62) during infection. Calprotectin makes up more than 50% of the neutrophil cytosol and has been proven to be an effective molecule at starving incoming pathogens of nutrient metal ions (7,63). However, despite this mechanism employed by the immune system to impede bacterial growth, GBS continues to cause life-threatening illnesses, suggesting that this bacterium possesses machinery to thwart host defenses and permit survival. Here, we have examined the global effect of calprotectin stress on GBS fitness using a newly developed mariner GBS transposon mutant library. We identified systems involved in zinc and manganese/iron homeostasis as well as putative metal-transport systems that were not previously described in GBS to be important for growth in the presence of calprotectin. Through mutagenesis and functional analyses, we determined that the Adc zinc acquisition system, comprising three zinc-binding proteins, promotes survival during calprotectin stress and contributes to systemic infection in vivo. Furthermore, the loss of calprotectin in vivo ablates the requirement of zinc homeostasis for GBS virulence. These data support the growing appreciation for the role of zinc uptake in bacterial pathogenesis and provide new insight into the mechanisms by which GBS resists nutritional immune challenge (Fig. 7).
In this study, we demonstrate, for the first time, the global impact of calprotectinmediated metal chelation on GBS fitness using transposon library screening. We observed that growth of both GBS disease and colonizing clinical isolates was inhibited by physiologically relevant concentrations of recombinant calprotectin. Serotype V isolates were significantly more resistant to chelation than serotype Ia isolates in vitro, but the basis for this requires further investigation. As our current understanding of GBS metal homeostasis is limited, we sought to characterize the global effects of calprotectin-mediated metal chelation on GBS fitness, utilizing a newly constructed Krmit transposon mutant library. This screen identified COG categories of GBS gene function during calprotectin stress, with the most abundant genes of known function involved in inorganic ion transport, amino acid and carbohydrate transport and metabolism, transcription, cell wall biogenesis, and defense mechanisms. Some of the most significant underrepresented factors identified were those involved in metal transport, including zinc/manganese/iron uptake and efflux. Our data also identified GBS essential genes for growth in rich media, including THY and mRPMI, and our results were consistent with a previous study using transposon sequencing of a different GBS strain, which identified essential genes in tRNA synthesis pathways, glycolysis, and nucleotide metabolism (64). We did, however, observe some differences. The transcriptional regulator ccpA was previously deemed part of the GBS essential genome (64), though in our analysis, ccpA was only essential in mRPMI. Similarly, the global nutritional regulator codY, which is essential for Streptococcus pneumoniae growth (65) and nonessential for GBS in rich medium (64), was found in our study to be nonessential for growth in THY but essential for GBS growth in mRPMI. Together, these data suggest that the essential genome of GBS is largely similar across strains, although this is dependent on the growth medium.
To survive in metal-limited environments, bacterial pathogens possess tightly regulated, high-affinity metal uptake systems, and many Gram-positive bacteria are known to use the ZnuABC/AdcABC zinc transport systems (66)(67)(68). In the case of pathogens such as Staphylococcus aureus and Bacillus anthracis, each possesses a single zincbinding protein, AdcA and ZnuA, respectively (69,70), and their associated ATP-binding cassette transporter permease and ATPase are AdcB/ZnuB and AdcC/ZnuC (40,71). The zinc uptake machinery of streptococcal pathogens S. pneumoniae and S. pyogenes possess two zinc-binding lipoproteins, AdcA and AdcAII/Lmb. GBS is particularly unique in that in addition to AdcA and AdcAII, it harbors a third zinc-binding protein, annotated as Lmb (72), encoded on a mobile element that is cotranscribed with a second streptococcal histidine triad protein, ShtII. Lmb is thought to have been acquired by horizontal gene transfer and shares homology with Lsp of Streptococcus pyogenes (73), but the direct origin remains unknown. Originally annotated as laminin-binding proteins, Lmb and Lsp were thought to contribute to adherence, though this interaction has been debated and may be species or strain dependent (44,74,75). The GBS zinc uptake machinery is encoded by four distinct operons and is under the regulation of the zinc-dependent AdcR repressor (40). AdcR is involved in maintaining intracellular zinc homeostasis and has been shown to be important for growth under zinc-limited conditions (40,41). Components of this system are also significantly upregulated during GBS murine vaginal colonization and following incubation with human blood (76,77). In addition to the ABC transporters, bacteria have evolved other mechanisms to maintain zinc homeostasis, examples include the metallophore staphylopine produced by Staphylococcus aureus that binds zinc ions and is imported by the CntABCDF machinery (70) and the TdfH transporter of Neisseria gonorrhoeae that binds calprotectin directly to hijack and secure zinc ions (78). Although, similar systems have not been described in GBS.
Our transposon mutant screen during calprotectin treatment identified loss-offunction mutations in the zinc-binding protein AdcA as the most significantly underrepresented metal mutation. Mutants in the AdcC subunit of the zinc-dependent ABC transporter, zinc-binding proteins AdcA and AdcAII, and the streptococcal histidine triad protein ShtII were also underrepresented in our screen. Interestingly, mutations in Lmb and Sht did not result in fitness defects in our transposon library screen, suggesting that in a competitive growth environment, loss of either AdcA or AdcAII reduced GBS fitness in the presence of calprotectin. However, in monoculture, while loss of all three solute-binding proteins sensitized GBS to calprotectin, their individual losses did not. Expression of adcAII and lmb were both induced to greater extent than adcA by Global GBS Response to Calprotectin ® both calprotectin and TPEN, and mutations in AdcAII were specifically observed in inhibitory calprotectin treatment. Collectively, these observations could indicate that the GBS zinc importers may uniquely contribute to resisting metal limitation or other aspects of infection, but further investigation is needed.
Additional systems of interest that were underrepresented in our calprotectin transposon library screen were mtsABC (SAK_1554 to 1556), mntH (SAK_0871), sczA (SAK_0515), czcD (SAK_0514), and cadD (SAK_2051). mtsABC encodes the manganese/ iron-dependent ABC transporter, and mtsA is a component of the core GBS genome and could be a conserved system for survival during calprotectin-mediated stress within the host (53,79). Additionally, mntH encodes the manganese/iron NRAMP and is known to be important for survival under acidic conditions similar to what GBS would encounter during inflammation or within the phagolysosome (57). Similar to what has been shown for zinc import, the mtsABC transporter was shown to be upregulated in human blood and during vaginal colonization (76,77). In the context of metal ion efflux, we identified mutations in sczA, czcD, and cadD, which all resulted in fitness defects when grown in media containing calprotectin. SczA is a zinc-dependent transcriptional activator of the cation diffusion facilitator protein CzcD. This system has been shown to contribute to bacterial survival during zinc toxicity, neutrophil and macrophage killing, and GAS virulence (80)(81)(82)(83). To date, SczA has been suggested to be an activator in GBS (80), but the functional roles of SczA and CzcD in metal efflux and GBS survival have not been described. Additionally, the cadDX operon in Streptococcus salivarius has been shown to have both cadmium-and zinc-inducible repression (84); thus, CadD may function similarly in GBS, but this warrants further investigation. Recently, a new highly virulent GBS sequence type (ST), ST485 of the clonal complex 103, has become increasingly common in China, specifically, with the frequency of isolation quickly climbing from 1% to 14% (85). These isolates have evolved from a genetic lineage capable of causing both human and bovine disease, and both the increase in virulence and rapid emergence of these isolates are thought to be due to the acquisition of the cadDX operon (85).
An additional strength of our study is the sensitivity of our screen to detect genes involved in overcoming various degrees of metal starvation. We identified 30 genes that were essential for GBS growth in a subinhibitory concentration of calprotectin (see Table S1 in the supplemental material), representing genes that are necessary for overcoming low-level metal sequestration but are nonessential for survival in extreme metal limitation. Genes of importance include those encoding two ribosomal proteins, key enzymes involved in glycolysis, two enzymes involved in folate metabolism, and a cobalt transporter ATPase. We also identified 55 uniquely essential genes for survival in inhibitory levels of calprotectin. These genes represent those that are important for growth when GBS encounters high degrees of starvation or the starvation of multiple metals. Systems of interest in these data were genes encoding four ribosomal proteins, six prophage-related proteins, phosphotransferase systems, and the stress response serine protease HtrA (Table S1). These differential findings are significant, as it is becoming increasingly appreciated that metal starvation/intoxication occurs across a gradient and that the maintenance of metal homeostasis is dynamic and requires numerous fine-tuned responses. These data are supported by previous studies that show streptococcal metabolism (86,87) to be dependent on metal ions and that ribosomal proteins serve as reservoirs for intracellular zinc during metal limitation (88).
As the relative contribution of zinc homeostasis to GBS virulence had not been previously characterized, we utilized a murine model of GBS systemic infection to compare WT and ΔadcAΔadcAIIΔlmb strains. These experiments demonstrated that the ΔadcAΔadcAIIΔlmb mutant was significantly attenuated compared to WT GBS in two different GBS strain backgrounds and in different mouse strains. These data further validate our in vitro results and confirm the importance of zinc uptake machinery to the pathogenesis of GBS infection. To determine the contribution of host calprotectin to the GBS disease process, we utilized a calprotectin knockout mouse strain (S100A9 Ϫ/Ϫ ) (89,90). In contrast to the phenotypes observed in WT mice, S100A9 Ϫ/Ϫ mice were equally susceptible to GBS WT and ΔadcAΔadcAIIΔlmb strains (Fig. 7A), suggesting that zinc transport machinery is expendable when the zinc-limiting pressure of calprotectin is absent. Furthermore, we observed that WT mice exhibited increased mortality due to GBS infection compared to that of S100A9 Ϫ/Ϫ mice. In our studies, nearly 90% of WT mice infected with WT GBS succumbed to illness by 48 h, whereas the S100A9 Ϫ/Ϫ mice infected with WT GBS did not reach 50% lethality until 90 h postinfection. Similar trends were recently observed in S100A9 Ϫ/Ϫ mice challenged with S. pyogenes (91). Additionally, these data are consistent with previous findings that suggest a role for calprotectin as an immunological alarmin in promoting inflammatory signaling (92)(93)(94). Studies to determine the specific role of calprotectin in inflammation during GBS disease progression warrant further investigation.
Here, we report the generation and utilization of a highly saturated GBS mariner transposon library to investigate bacterial response to calprotectin-mediated metal chelation. Genome-wide screening revealed numerous metabolic pathways and metal transport systems that may contribute to the ability of GBS to overcome calprotectin stress and nutritional immunity. Our results emphasize the importance of zinc transport to the development of GBS systemic infection, highlighting the significance of zinc homeostasis to disease progression. As zinc uptake machinery is highly conserved across streptococcal pathogens, they present a promising target for the development of novel antimicrobials.
Calprotectin growth assays. Briefly, GBS cultures grown overnight (for 18 h) in THB were diluted 1:50 into 100 l in a 96-well microtiter plate with 38% 3ϫ modified RPMI medium (mRPMI) ( (70,96). The range of recombinant calprotectin was previously established for assaying bacterial survival in vitro (17,97). At 8 h postinoculation, growth was assessed by measuring optical density (OD 600 ) and plating serial dilutions to quantitate CFU. In experiments where zinc was supplemented to counter calprotectin chelation, zinc sulfate was added at 50 M for A909 or 100 M for COH1 and CJB111 strains.
Generation of mariner (Krmit) mutant libraries in GBS for Tn sequencing. In vivo mariner transposition for random mutagenesis in GBS was accomplished using the pKrmit system originally developed for GAS (50), as previously described (50,98). Briefly, GBS CJB111 cells were transformed by electroporation with 300 g of pKrmit and outgrown in THY at 30°C (permissive temperature for pKrmit replication) for 4 h. Transformants were then selected by plating on THY agar containing 300 g/ml kanamycin (Km) 100 g/ml and spectinomycin at 30°C for 48 h. The presence of intact pKrmit was phenotypically tested as previously described (98), and proper GBS transformants were stored at Ϫ80°C. For Krmit transposition, an individual pKrmit-containing GBS freezer stock was used to inoculate 250 ml of THY containing Km and incubated overnight at 37°C (T 0 ) (nonpermissive temperature for pKrmit replication). The quality of Krmit transposition was tested as previously described (98,99): the complexity or randomness of Krmit mutant libraries was assessed by amplifying the insertion sites of random mutants by arbitrarily primed PCR (AP-PCR), Sanger DNA sequencing, and mapping onto the appropriate GBS genome. Percent randomness was determined by defining a ratio of unique insertions (by AP-PCR sequencing) among a tested population.
Transposon library screening. The GBS pooled Krmit library was grown overnight in THB with 300 g/ml kanamycin. Overnight culture was back diluted into 38% 3ϫ mRPMI with 62% calprotectin buffer and 0, 60, or 480 g/ml purified calprotectin. Samples were incubated at 37°C for 8 h. Following incubation, samples were centrifuged and resuspended in 200 l phosphate-buffered saline (PBS). The total volume was spread onto Todd-Hewitt agar (THA) plates and incubated overnight at 37°C. Bacterial growth from each treatment condition was collected and pooled into three samples, and genomic DNA was extracted using a Zymobiomics DNA miniprep kit (Zymo Research).
Transposon library sequencing. Library preparation and sequencing were performed as previously described (51) by the microarray and genomics core at the University of Colorado Anschutz Medical Campus. Briefly, genomic DNA was sheared to approximately 340-bp fragments and processed through the Ovation Ultralow V2 DNA-Seq library preparation kit (Tecan); 9 ng of each library was used as the template to enrich by PCR (16 cycles) for the transposon insertions using Krmit-specific (TCGTCGGCAG CGTCAGATGTGTATAAGAGACAGCCGGGGACTTATCATCCAACC) and Illumina P7 primers. The enriched PCR products were diluted 1:100, and 20 l was used as the template for an indexing PCR (9 cycles) using Global GBS Response to Calprotectin ® the TruSeq P5 indexing and P7 primers. Sequencing was performed using an Illumina NovaSeq 6000 in a 150-base paired-end format.
Bioinformatic analyses of Tn sequencing. Two annotated Streptococcus agalactiae genomes were used as the reference for these analyses, namely, CJB111 (AAJQ01000001) and A909 (NC_007432). The provided annotation is based on the A909 genome, as the annotated CJB111 genome has significantly lower coverage. Italicized directory names (ending in/) refer to directories within the git repository for this project available at https://github.com/abelew/sagalacticae_2019. Much of the postprocessing was handled by the hpgltools R package (100). Approximately 2.5 million reads of each raw library were queried for quality with Fastqc (101) before removing the mariner inverted terminal repeat (ITR) leading sequences with cutadapt (102). These libraries were aligned against the reference genomes with Bowtie (103, 104) using options to allow one mismatch (-v 1) and randomly assign multimatched reads to one of the possible matching positions (-M 1). The resulting alignments were converted to sorted/compressed binary alignments (105) and counted (106) against the reference genome coding DNA sequence (CDS) and intergenic regions. The essentiality software package (52) provides an opportunity to query statistically significant stretches of TAs that have no observed insertions to further inform its metric of essentiality. The insertion data were therefore converted into its expected format and passed to the version 1.21 of the implementation Python script. The resulting table provided a count of the number of insertions observed in each open reading frame (ORF), the number of observed TAs, the maximum length of the nonobserved sequence, the nucleotide span of this region, a call on whether each ORF is essential, and the posterior probability for each call. The default options were used, except multiple runs were performed with the minimum hit parameter set to 1, 2, 4, 8, 16, and 32. These operations were performed via CYOA (https://github.com/abelew/CYOA). In a separate invocation, the three replicates for the control, low concentration, and high concentration samples were concatenated into a single sample, and essentiality was run on the combined samples. The libraries were quantified with respect to relative coverage, similarity, and saturation and with respect to available TA insertion points. These tasks were performed using the hpgltools and the input text/wig files for the essentiality package. Thus, the essentiality input files were read into the R function "plot_saturation" and used to visualize the saturation of each library. This was done by taking the log 2 (hits ϩ 1) for each position and plotting them as a set of histograms. Comparison and normalization of control (input) and experimental (output) libraries were performed similarly to the essentials software package (107, 108) but using a combination of voom/ limma, EdgeR, DESeq2, EBSeq, and a statistically uninformed basic analysis instead of EdgeR. Pairwise Euclidean distances, Spearman correlation coefficients, and principal-component analyses were then used to visualize the similarities/differences between normalized libraries. Clustering of orthologous groups of proteins (COGs) were assigned using EggNOG 5.0.0 (109), and Venn diagrams were calculated and plotted using BxToolBox (BioInfoRx, Inc., Madison, WI).
Ortholog clustering. To evaluate if metal transport machinery involved in survival during calprotectin stress was orthologous to that in characterized systems, the closely related Streptococcus pneumoniae TIGR4 (GCF_000006885.1) and S. agalactiae A909 (GCF_000012705.1) were used as input for the program OrthoFinder v. 2.2.6 (110). Domain architecture of proteins in each orthogroup collected were evaluated using InterProScan v. 5.27-66.0 (111). Predicted operons in GBS were determined using the Database of prOkaryotic OpeRons (DOOR 2 ) (112)(113)(114).

Mouse model of GBS systemic infection.
All animal experiments were conducted under the approval of the Institutional Animal Care and Use Committee (number 00316) at the University of Colorado Anschutz Medical Campus and performed using accepted veterinary standards. We utilized a mouse model of systemic infection as previously described (28,36,37), in which female 8-week-old C57BL/6, C57BL/6 S100A9 Ϫ/Ϫ , and CD1 mice were injected intravenously with 3 ϫ 10 8 CFU of WT A909 or A909 ΔadcAΔadcAIIΔlmb or 1 ϫ 10 7 CFU CJB111 or CJB111 ΔadcAΔadcAIIΔlmb mutant strains. Mice were euthanized, and blood and brain tissues were collected. Tissue homogenates and blood were plated on THA to quantify GBS CFU burden.
Statistical analyses. Significance during calprotectin growth experiments was determined by the Kruskal-Wallis test with Dunn's multiple-comparison posttest with treated samples compared to untreated controls. Normality was confirmed for clinical isolate data by the Shapiro-Wilk test, and significance was determined by one-way analysis of variance (ANOVA) with Tukey's multiple-comparison posttest. Calprotectin growth assays using GBS WT and mutant strains was analyzed by the Kruskal-Wallis test with Dunnett's multiple-comparison test. Statistical differences in murine experiments were determined by the log rank Mantel-Cox test for survival and by an unpaired Student's t test. Statistical significance was accepted when P value was Ͻ␣, with an ␣ of 0.05. Data availability. Sequencing reads from the transposon sequencing analyses are available in the NCBI Sequence Read Archive (SRA) under the accession number PRJNA667098.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.