Identification of glyoxalase A in group B Streptococcus and its contribution to methylglyoxal tolerance and virulence

Authors: Madeline S. Akbari, Luke R. Joyce, Brady L. Spencer, Amanda Brady, Kevin S. McIver, Kelly S. Doran

Abstract

Group B Streptococcus (GBS) is a Gram-positive pathobiont that commonly colonizes the gastrointestinal and lower female genital tracts but can cause sepsis and pneumonia in newborns and is a leading cause of neonatal meningitis. Despite the resulting disease severity, the pathogenesis of GBS is not completely understood, especially during the early phases of infection. To investigate GBS factors necessary for bloodstream survival, we performed a transposon (Tn) mutant screen in our bacteremia infection model using a GBS mariner transposon mutant library previously developed by our group. We identified significantly underrepresented mutations in 623 genes that contribute to survival in the blood, including those encoding known virulence factors such as capsule, the β-hemolysin, and inorganic metal ion transport systems. Most of the underrepresented genes have not been previously characterized or studied in GBS, including gloA and gloB, which are homologs for genes involved in methylglyoxal (MG) detoxification. MG is a byproduct of glycolysis and a highly reactive toxic aldehyde that is elevated in immune cells during infection. Here, we observed MG sensitivity across multiple GBS isolates and confirmed that gloA contributes to MG tolerance and invasive GBS infection. We show specifically that gloA contributes to GBS survival in the presence of neutrophils and depleting neutrophils in mice abrogates the decreased survival and infection of the gloA mutant. The requirement of the glyoxalase pathway during GBS infection suggests that MG detoxification is important for bacterial survival during host-pathogen interactions.

IMPORTANCE

A transposon-mutant screen of group B Streptococcus (GBS) in a bacteremia mouse model of infection revealed virulence factors known to be important for GBS survival such as the capsule, β-hemolysin/cytolysin, and genes involved in metal homeostasis. Many uncharacterized factors were also identified including genes that are part of the metabolic pathway that breaks down methylglyoxal (MG). The glyoxalase pathway is the most ubiquitous metabolic pathway for MG breakdown and is only a two-step process using glyoxalase A (gloA) and B (gloB) enzymes. MG is a highly reactive byproduct of glycolysis and is made by most cells. Here, we show that in GBS, the first enzyme in the glyoxalase pathway, encoded by gloA, contributes to MG resistance and blood survival. We further demonstrate that GloA contributes to GBS survival against neutrophils in vitro and in vivo and, therefore, is an important virulence factor required for invasive infection.

INTRODUCTION

Streptococcus agalactiae (group B Streptococcus, GBS) is an opportunistic pathogen that commonly resides in the gastrointestinal and lower female genital tracts but can cause infection in newborns and is also increasingly associated with non-pregnant individuals, especially older adults and patients with diabetes (1–3). GBS asymptomatically colonizes the vaginal tract in up to 30% of people but can instigate complications during pregnancy and birth, such as preterm labor, and serious infections in newborns, such as sepsis, pneumonia, and meningitis (1, 4–6). Research into GBS intrauterine infection during pregnancy thus far indicates that GBS-activated inflammatory pathways ultimately result in preterm births (7). If GBS is vertically transferred to the neonate, the resulting infection is categorized as either early-onset disease (EOD, 0–6 days of life) or late-onset disease (LOD, 7–90 days of life) depending on the timing of symptom presentation (8). Neonatal meningitis caused by both EOD and LOD GBS disease requires a sustained level of bacteremia prior to the penetration into the brain and, even after treatment, frequently results in long-lasting neurological effects and long-term morbidity (4, 5). Although intrapartum antibiotic prophylaxis is administered to colonized pregnant women to prevent the detrimental effects of GBS infection, GBS isolates are increasing in resistance to second-line antibiotics over time (9), and intrapartum antibiotic prophylaxis is not effective in preventing LOD. Therefore, studying the GBS pathogenesis of meningitis, including bacteremia, is important for the development of novel treatments and therapeutics to prevent GBS infection and reduce morbidity and mortality.

Previous work has determined the GBS transcriptome as well as genes necessary for survival in human blood in vitro and for survival of the murine female reproductive tract (10–13). These data sets as well as other studies have shown that GBS possesses an arsenal of virulence factors that directly contribute to pathogenesis such as β-hemolysin/cytolysin, superoxide dismutase, capsule, adherence proteins, and metal transport systems (10, 14, 15). β-hemolysin/cytolysin (βH/C) and capsular polysaccharides are the most well-studied factors associated with GBS pathogenesis and are regulated by the well-known two-component system, CovR/S (15). βH/C is an ornithine rhamnolipid pigment synthesized by the cyl operon and has both hemolytic and cytolytic capabilities against a variety of host cells including red blood cells, neutrophils, macrophages, and epithelial cells (16–19). As a result, βH/C has been shown to contribute to GBS blood, lung, and brain infection (17, 20). Capsular polysaccharide is surface-associated and made up of different arrangements of monosaccharides that form capsular repeat units (21, 22). There are 10 known GBS capsular serotypes with serotype III being highly associated with neonatal infections, such as meningitis, and which is overrepresented in invasive isolates worldwide (9, 23, 24). Group B streptococcal capsular polysaccharide was first studied over 40 years ago and has been shown to help GBS evade host immune defenses by mimicking host antigens and blocking complement-mediated opsonophagocytic killing as well as to facilitate GBS biofilm formation (22, 25–27). Despite these studies, the contribution of GBS metabolism to colonization and infection in vivo has been a neglected area of study in the field (10, 15).

Here, we performed a transposon-mutant screen (Tn-sequencing) using a murine bacteremia model to discover additional genes necessary for GBS fitness in murine blood in vivo. GBS survival within the blood is an essential prerequisite to penetrating the blood-brain barrier and subsequent development of meningitis. Tn-sequencing allows for the identification of genes that may be continuously expressed but are essential in certain environments. Here, we identify that the glyoxalase pathway is required for GBS bloodstream survival. The glyoxalase pathway consists of two genes, gloA and gloB, and is involved in methylglyoxal detoxification (28, 29). Methylglyoxal (MG) is a toxic byproduct of normal cell metabolism (30), and we confirm that the first enzyme in the pathway, encoded by gloA, contributes to GBS MG detoxification and invasive infection. Furthermore, we found that gloA is necessary for GBS neutrophil survival, and depleting neutrophils rescue the gloA mutant in vivo.

RESULTS

Genome-wide analysis of GBS factors involved in bloodstream survival

To identify genes necessary for GBS survival in murine blood, we utilized a bacteremia model of infection with our previously described Tn mutant library in the CJB111 strain, a GBS isolate from a case of neonatal bacteremia without focus (17, 31–33). Briefly, mice were intravenously infected with the Tn mutant library and the infection was allowed to progress up to 27 h. The input Tn mutant library and libraries recovered from the blood were processed as described in Materials and Methods. To identify transposon insertion sites, sequenced reads were mapped to the GBS CJB111 genome, which identified 623 genes as significantly underrepresented (Padj < 0.05, log2FC < −1) and 95 genes as significantly overrepresented (Padj < 0.05, log2FC > 1) in the blood compared to the input library (Fig. 1A) (Table S1). The significant gene hits were equally distributed across the genome. Significant genes were then assigned clusters of orthologous groups of proteins (COGs). The number of significant gene hits in each COG was normalized to the total number of genes in each COG revealing sRNA, amino acid transport and metabolism, and inorganic ion transport and metabolism as the COGs containing the most underrepresented genes during GBS survival in the blood (Fig. 1B). We detected many classes of GBS virulence factors and genes known to contribute to GBS infection as significantly underrepresented (Table 1). Some of these genes are involved in hemolytic pigment biosynthesis, capsule biosynthesis, two-component regulatory systems, metal transport, glutamine transport, and purine metabolism. When we investigated other underrepresented genes that have not been previously characterized in GBS, we found homologous genes to glyoxalase A and B of the glyoxalase pathway to both be significantly underrepresented with fold changes of −18.38 and −25.63, respectively (Table 1). The glyoxalase pathway is a ubiquitous two-step process found across all kingdoms of life and is the primary mechanism of MG breakdown (Fig. 1C) (28). A highly reactive carbonyl byproduct of normal cell metabolism, most MG is primarily generated from glycolysis, but can also be produced from other metabolic pathways such as lipid and protein metabolism (34).

Methylglyoxal tolerance differs across GBS strains

GBS contains glyoxalase A and B homologs, also known as lactoylglutathione lyase (ID870_02260) and hydroxyacylglutathione hydrolase (ID870_06660), respectively. These are hypothesized to be involved in MG detoxification and therefore, tolerance. To begin to characterize this pathway in GBS, we grew several clinical GBS isolates, representing various capsule serotypes, in the presence of 1.0 mM MG in a modified chemically defined media (mCDM) (64) (Fig. 2). These concentrations of MG resulted in an observed increase in lag phase without a decrease in CFU, suggesting MG has bacteriostatic properties (Fig. S1A). CJB111 exhibited the highest sensitivity to 1.0 mM MG with the largest increase in lag phase quantified by the change in time to max OD600 at 8.00 h. All of the strains tested had significantly decreased time to max OD600 when compared to CJB111, and COH1 exhibited the lowest sensitivity with the smallest time to max OD600 at 1.83 h. Overall, different GBS isolates had varying degrees of resistance to MG, but resistance was not correlated with serotype or growth rate differences (Fig. S1B). To explore the strain differences in MG tolerance further, we selected three representative serotype strains commonly used by our group and others working on GBS pathogenesis with low to high CJB111 (V), A909 (Ia), and COH1 (III). First, we compared GloA amino acid sequences between these three strains and previously characterized GloA from Streptococcus pyogenes and Escherichia coli (Fig. S2A) (29, 65–67). We chose these strains since the GAS glyoxalase pathway has been investigated previously and it is within the same genus and the E. coli GloA has a solved crystal structure which is important for protein modeling. The CJB111 GloA is 63.5% identical to GAS GloA and 40.5% identical to E. coli GloA. Interestingly, the GloA from CJB111 and A909 are 100% identical while the COH1 GloA is only 99% identical due to a single amino acid change of an alanine to a serine (A45S) in a non-conserved region. To determine how common this variant was in GBS, we generated a phylogenetic tree using BlastP and FigTree to compare 57 GBS genomes and found that 12 out of 57 GloA proteins (21%) have the A45S change with another five having a different A45 variant (Fig. S2B). Proteins with the A45S variant also clustered together in the tree suggesting a common ancestral strain, however, only COH1 from the strains tested in Fig. 2 had this variant. To assess tertiary structure, a predicted protein model for GBS GloA was generated using AlphaFold2 and had extremely high confidence for most residues (Fig. S2C and D). The predicted structure was compared to the solved E. coli GloA structure (PDB 19FZ) and found to have highly similar topology and conserved metal binding residues (Fig. S2C). GloA was also modeled in its active form as a dimer to show predicted active sites (Fig. S2E Left). The A45S variant from COH1 GloA was included in the dimer and modeled to be next to the predicted active site (Fig. S2E Right). Lastly, using our selected representative strains, we investigated baseline transcription regulation of the glyoxalase pathway by comparing mid-log transcript levels for gloA and gloB using real-time quantitative polymerase chain reaction (RT-qPCR) and found that COH1 has a higher abundance of both gloA and gloB transcripts compared to CJB111 and A909 (Fig. S2F). Altogether, the predicted GloA protein in GBS contains conserved residues important for enzyme activity and metal binding but it is not clear if the common A45S amino acid change correlates with enzyme activity. In addition, transcript expression for gloA and gloB are increased in COH1 which could explain higher MG resistance.

Glyoxalase A contributes to methylglyoxal detoxification in GBS

To confirm our Tn-sequencing results we chose to study the first enzyme in the glyoxalase pathway, GloA. Using allelic exchange mutagenesis, we constructed a mutant in gloA (∆gloA) and a complemented strain (pgloA) in CJB111 as described in Materials and Methods. MG detoxification was then tested using these strains by MG quantification and growth curve analysis. First, to measure if MG accumulates in the ∆gloA strain, we measured MG concentrations using an enzyme-linked immunosorbent assay (ELISA) on lysed cell pellet samples for CJB111 WT, ∆gloA, and pgloA strains. The concentration of MG was normalized to the total protein concentration of each sample and found to be significantly increased in the ∆gloA mutant compared to the CJB111 WT and complemented strains (Fig. 3A). To determine if this accumulated MG in the ∆gloA mutant may be toxic/impact GBS growth, all strains were inoculated into mCDM with or without the addition of 0.5 mM MG. Indeed, a greater growth delay was observed for ∆gloA with a change in time to max OD600 at 4.92 h compared to 3.58 h for WT or 2.83 h for pgloA, which confirms gloA is involved in MG detoxification (Fig. 3B). Furthermore, the OD600 at 8 h was compared between strains and confirmed a significant decrease in ∆gloA growth compared to WT or the complemented strain following the addition of MG (Fig. 3C). To determine if the growth delay we observed with MG was due to the emerging resistant subpopulation, we grew CJB111 WT to mid-logarithmic phase with 0.5 mM MG and then inoculated fresh media with or without increasing MG (Fig. S3A). From this experiment, we still observed a growth delay with fresh MG indicating it is not a more resistant subpopulation emerging. As MG is primarily produced from glycolysis in cells (bacteria and host), we further investigated the impact of GloA on GBS growth in mCDM with increasing glucose concentrations. However, we did not observe a growth defect for the ∆gloA mutant compared to WT or complemented strain at low (biologically relevant blood glucose concentration), medium (ideal concentration for GBS growth), or high glucose concentrations tested (Fig. S3B). Additionally, upon assessment of general virulence characteristics, we also did not observe a difference in susceptibility to hydrogen peroxide or hemolytic activity between CJB111, ∆gloA, and pgloA (Fig. S3C and D). Taken together MG quantification and growth analysis suggest that GloA contributes to MG detoxification in GBS. Our results also suggest that, under the conditions tested, GBS may not produce enough MG from glucose metabolism to negatively impact its growth.

Glyoxalase A is necessary for GBS survival in vivo

To further confirm the Tn-sequencing results and determine if gloA is important during infection, we repeated our bacteremia model of infection by intravenously injecting mice with 1.5–2 × 10^7^ CFU CJB111 WT or the ∆gloA mutant and monitoring the infection for up to 72 h post-infection. Mice infected with the ∆gloA mutant exhibited significantly decreased mortality compared to those infected with WT, with greater than 75% surviving to the experiment endpoint (Fig. 4A). In order to monitor CFU burden over time, blood samples were taken from surviving mice at 24 and 48 h post-infection and the time of death (TOD) (Fig. 4B). Mice infected with ∆gloA had significantly decreased blood burdens as soon as 24 h post-infection and in tissue burdens at the time of death, indicating the mutant strain is not able to survive as well in the bloodstream and disseminate to other organs compared to WT CJB111 (Fig. 4B and C). It is important to note that all of the strains had similar blood burdens at 6 h post-infection (Fig. S4A) and that the ∆gloA-infected mice that succumbed to the infection before 72 h had the highest CFU counts in the blood at 24 h (Fig. 4B). Additionally, we infected mice with the complemented strain and confirmed that it was able to survive in the blood longer than ∆gloA (Fig. S4).

Importance of GBS glyoxalase A to neutrophil survival

Pro-inflammatory M1-type macrophages primarily use glycolysis to generate energy (68) and have been shown to produce aldehydes, such as MG, in response to infection (66, 69–71). However, neutrophils are the primary immune cell GBS encounters during acute infection (55), which also utilize glycolysis as their primary energy source (66, 68–72). Therefore, to determine if GBS gloA contributes to neutrophil survival we performed in vitro neutrophil killing assays using differentiated HL60-neutrophils with the CJB111 WT, ∆gloA, and pgloA strains. At 5 h post-infection, the ∆gloA mutant strain exhibited significantly decreased survival compared to WT or the complemented strains (Fig. 5A). This phenotype was independent of serum killing and cytotoxicity since serum by itself did not impact GBS survival over time and there were low levels and no differences in HL60 cytotoxicity between the strains at 5 h (Fig. S5A and B). We also confirmed that ∆gloA has decreased survival in the presence of primary murine bone marrow neutrophils (BMNs) (Fig. S5C). Next, we used cytochalasin D treatment to block phagocytosis of HL60-neutrophils which resulted in a significant increase in survival for CJB111 WT and pgloA strains (Fig. 5B). The ∆gloA mutant also had a minor increase in survival with cytochalasin D treatment but it was not significant when compared to the DMSO control and there was still a significant decrease in survival when compared to WT and complement indicating the mutant is susceptible to multiple mechanisms of neutrophil killing. To evaluate if this increased killing might be due to a general increase in neutrophil production of MG in response to GBS infection, we measured the accumulation of intracellular MG-modified proteins in HL60-neutrophils using flow cytometry. Consistent with the literature (30) we observed that all cells contained detectable MG-modified proteins. Interestingly, however, upon infection, we observed a significant increase in anti-MG geometric mean fluorescent intensity (MFI) compared to uninfected controls (Fig. 5C), indicating that infection increases intracellular MG within HL60s. This increase was also only observed in high glucose conditions (Fig. S5D), which is similar to what has been described for S. pyogenes where they reported that GloA contributed to survival against neutrophils with elevated glucose (65). To examine the contribution of neutrophils controlling GBS infection in vivo, we depleted neutrophils in mice prior to intravenous infection. Mice were injected intraperitoneally with anti-Ly6G or an isotype control 24 h (73) before intravenous infection with 1 × 10^7^ CFU CJB111 or ∆gloA. Upon assessing morbidity and mortality of these groups over 72 h post-infection, we observed that both CJB111 and ∆gloA-infected neutrophil-depleted mice exhibited a significant increase in mortality when compared to their non-depleted cohorts (Fig. 5D), although percent survival of neutrophil-depleted mice infected with ∆gloA remained higher than neutrophil-depleted mice infected with CJB111. At 12 h post-infection, we observed that neutrophil depletion abrogated the attenuated phenotype of the ∆gloA mutant in the blood, as the CJB111 and ∆gloA-infected, neutrophil-depleted mice did not differ in blood burdens (Fig. 5E). Furthermore, CJB111 and ∆gloA CFU burdens were significantly increased in the neutrophil-depleted mice compared to the non-depleted mice. Taken together these results show that the attenuation of the ∆gloA mutant can be partially rescued with neutrophil depletion and suggest that MG produced by other cell types may aid in the defense against GBS bloodstream infections.

DISCUSSION

GBS must be able to survive multiple host niches to cause invasive infections in neonates. Some of these environments include the vaginal tract, amniotic fluid, and blood (1, 74). Tn-sequencing is a powerful and common method for investigating bacterial genes necessary for survival and fitness in these different environments. Previously, an ex vivo Tn-sequencing was performed in human blood using a TnSeq library in the GBS A909 serotype Ia background (75). Their results found similar underrepresented genes compared to our in vivo data set such as genes involved in capsule biosynthesis, metal homeostasis, and arginine metabolism. Interestingly, they identified relA to be underrepresented, which encodes a GTP pyrophosphokinase and is a central regulator of the stringent response in GBS. They found that relA not only controls stringent response activation and the arginine deiminase pathway but also impacts βH/C production. While we did not observe relA in our data set, other putative stress response proteins like ytgB and asp1 were significantly underrepresented. Most notably, asp1 is annotated as an Asp^23^/Gls^24^ family envelope stress response protein and was found to be upregulated when GBS was incubated in human blood (13, 76) and downregulated after exposure to high glucose (77). We also observed that the two-component system (TCS) dltRS and a dlt gene were underrepresented, which are involved in modulating surface charge and contribute to cationic antimicrobial peptide resistance and decrease phagocytic killing (35, 54). Previous studies have shown that a dltA mutant exhibited decreased virulence in a murine model with significantly lower burdens in tissue and blood compared to WT GBS (35). Another TCS, bceRS/nsrRK, exhibited the highest negative fold change of the underrepresented TCS and has been shown to contribute to bacitracin (antibiotic), nisin (lantibiotic), cathelicidin/LL-37 (human antimicrobial peptide), and oxidative stress resistance (56, 57). Its role in GBS pathogenesis was demonstrated as a bceR mutant yielded decreased virulence during murine infection and decreased biofilm formation (56). Another top gene hit identified within the present study to be important for GBS blood survival is the C5a peptidase scpB. scpB is already known to be involved in complement evasion and fibronectin binding and is associated with neonatal isolates (36, 78). Overall, identifying these known virulence factors in our study demonstrates the validity of our in vivo Tn-sequencing screen to identify novel factors important for blood survival in mice and supports previous research in the streptococcal field. It also provides another resource for developing new hypotheses and research projects. For example, MG detoxification has not been previously characterized in prior GBS studies.

MG is a highly reactive electrophilic species (RES) and a byproduct of normal cell metabolism which can be spontaneously or enzymatically produced by all cells (28, 30) with up to 90% of cell MG estimated to come from glycolysis alone (79). Notably, MG is also a precursor to advanced glycation end products (AGEs) and is associated with many other human diseases like diabetes, cancer, and neurological disorders like Alzheimer’s disease (30, 80). The most well-known and ubiquitous pathway for MG detoxification is the glyoxalase pathway which consists of glyoxalase A (gloA) and glyoxalase B (gloB) enzymes. Recently, the glyoxalase pathway, especially gloA, in Listeria monocytogenes was found to contribute to intracellular survival in macrophages and during murine infection (66). In addition, the glyoxalase pathway in S. pyogenes was shown to be important for survival against neutrophils in a glucose and myeloperoxidase-dependent manner (65). The GBS gloA and gloB homologs were underrepresented in our Tn-sequencing data set, therefore, we hypothesized that GBS may encounter host-derived MG during bacteremia as a response to infection (65, 69, 81, 82). The first step in the glyoxalase pathway is mediated by GloA and, in this study, we have characterized its contribution to GBS MG tolerance in vitro and infection in vivo. We observed that a ΔgloA mutant exhibited decreased survival in the presence of HL60-neutrophils as well as primary murine neutrophils and that its virulence was largely restored in neutrophil-depleted mice. We also measured an increase in MG-modified proteins in HL60-neutrophils upon GBS infection which is likely from increased production of MG by the cells themselves. These results indicate that MG-mediated killing may constitute another important defense mechanism used by immune cells to kill invading bacteria.

We also observed that tolerance to MG varies across GBS isolates and is not definitively correlated to serotype, GloA amino acid sequence, or glyoxalase gene regulation. Of note, two of the serotype III strains, the most commonly isolated serotype from neonatal invasive infections (83), had the first and second highest overall tolerance to MG, but additional strains would need to be tested to determine if serotype III has higher average MG tolerance. Interestingly, COH1 (serotype III), which had the highest resistance to MG, was the only strain tested that had the A45S variant in GloA and higher baseline transcript levels of gloA and gloB compared to CJB111 and A909 strains. The A45S variant is also only 10 amino acids away from the K55 residue which is predicted to be involved in metal binding and could impact folding or metal coordination (67, 84, 85). However, other GBS strains tested also had high resistance to MG, like K79, which does not have the A45S variant. Therefore, it is unlikely that this amino acid change is the sole determinant of enzyme activity, but further investigation is needed to determine if GloA protein variants and regulation impact GBS MG tolerance. Relatedly, Streptococcus mutans was shown to be more tolerant to MG than most other commensal oral Streptococcal species and was also shown to outcompete Streptococcus sanguinis when MG was present in a competition experiment (86). From this previous study and our work shown here, we hypothesize that differences in GBS MG tolerance could be influenced by the environments they were found in, like the presence of other bacterial species or host differences.

Components of metal transport systems were also significantly underrepresented in the Tn-sequencing data set with top hits including the zinc import system adcAAIIBC and lmb, the manganese import system mtsABC, and the putative nickel import system nikABCDE (Table 1). These results suggest GBS requires trace metals to survive in the blood. Previously it has already been shown that zinc and manganese transporters are important for maintaining GBS metal homeostasis and contribute to vaginal colonization and female reproductive tract ascension as well as blood survival (13, 32, 76, 87). In addition, both zinc and manganese import systems are important in combating nutritional immunity, or the sequestration of nutrients by the host, mediated by a neutrophil-produced metal-binding protein called calprotectin (33, 46). The nickel transporter, however, has not been as well characterized. In our previous study, we attempted to measure nickel concentrations in a nikA mutant, but it was under the limit of detection in our samples. However, we did observe lower levels of copper indicating the system could be transporting more than one metal (47). There are only nine known enzymes present in archaea, bacteria, plants, and primitive eukaryotes that are nickel-dependent, with urease being the most notorious, however, GBS does not encode a urease enzyme (32, 47, 88, 89). Interestingly, another of the nine known nickel-dependent enzymes is GloA which was found to use nickel (Ni^2+^) as a cofactor in E. coli (34, 67, 85). Therefore, the importance of the nickel transporter in the blood could be due to increased GloA activity but the requirement for nickel in GBS remains to be elucidated.

MG is formed primarily from glycolysis but it can be produced, albeit to a lesser extent, from lipid, ketone, and protein metabolism (30, 34). MG is toxic to cells due to its electrophilic properties allowing it to react with different molecules, like DNA and protein, and effectively arrest growth (81). For example, MG has been shown to increase mutation rates in L. monocytogenes by binding DNA (66) and inhibits protein synthesis and modification by binding to guanine and arginine residues (90–93). It is important to note that in some bacteria, like E. coli and L. monocytogenes, MG can be formed directly from dihydroxyacetone phosphate during glycolysis by methylglyoxal synthase (mgsA); however, GBS, like other streptococci, do not have an MG synthase gene (65, 86). The lack of a synthase gene further supports our hypothesis that GBS encounters host-derived MG toxicity during infection. Since approximately 99% of cellular MG is thought to be already bound to molecules like DNA and protein it is difficult to quantify accurately; however, intracellular MG concentrations are consistently estimated below 10 µM and are known to be dependent on glutathione concentrations (30, 94, 95). In the serum of diabetic individuals, the concentration of MG and MG-derived AGEs is increased compared to healthy individuals, most likely due to increased glucose concentrations (95). MG is historically tied to diabetes because it is known to exacerbate diabetic complications like microvascular and kidney dysfunction and contribute to the progression of the disease (95). Our lab has shown that GBS is a common colonizer of infected diabetic wounds (96) and we show here the production of MG-modified proteins by HL60-neutrophils is dependent on glucose and GBS infection (Fig. 5C; Fig. S5D). Therefore, research into the role of the GBS glyoxalase pathway in the context of diabetic wound infection is a current area of study.

Lastly, the conversion of MG to D-lactate by the glyoxalase pathway was first described over 100 years ago and is the most ubiquitous and conserved process for MG detoxification across all kingdoms of life (28). MG detoxification has not been thoroughly studied in streptococci in the context of disease and has never been characterized in GBS. Thus far, studies focusing on S. pyogenes, S. mutans, and S. sanguinis have shown gloA to be the primary modulator of MG tolerance in vitro with gloB mutants having little effect (65, 86). This is in support of what was observed with L. monocytogenes, but not Salmonella where it was found that gloB was more important for Salmonella resistance to oxidative stress and killing by macrophages (66, 97). Here, we found gloA to be dispensable to GBS tolerance of H2O2 (Fig. S3C) but the contribution of gloB remains to be examined. Additional enzymes known to break down MG into acetol or lactaldehyde intermediates include aldose, aldehyde, and MG reductases. A putative aldo/keto reductase (yvgN) was significantly underrepresented in our data set (Table 1) but its contribution to GBS virulence requires further investigation (34).

In this study, we demonstrate, for the first time, the essential role of the glyoxalase pathway in GBS MG resistance and overall pathogenicity during bloodstream infection. We investigated the role of GloA in vitro and in vivo and confirmed it is important for growth in the presence of MG, survival against neutrophils, and during invasive infection. Our study also provides further evidence in support of the aldehyde hypothesis that MG detoxification is an important component for bacterial survival against host defenses; however, the role of the glyoxalase system in GBS survival in macrophages requires further investigation (69). Specifically, we show increased MG production by neutrophils in response to infection. Overall, research aimed at understanding metabolic mechanisms used by bacteria to survive in the blood and RES toxicity will be important for the development of new treatments and therapies for infection and will expand our knowledge about host-pathogen interactions.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions

See Table S2 for strains and primers used in this study. GBS strains were grown statically at 37°C in Todd-Hewitt Broth (THB) unless otherwise stated. Streptococcal chemically defined medium (64) was modified by omitting L-cysteine and adding 22 mM glucose unless otherwise stated. Escherichia coli strains for cloning were grown in LB at 30°C or 37°C with rotation at 250 rpm. Kanamycin and erythromycin (Sigma-Aldrich, St. Louis, MO) were supplemented to media at 50 µg/mL and 500 µg/mL, respectively, for E. coli. Kanamycin, spectinomycin, and erythromycin (Sigma-Aldrich, St. Louis, MO) were supplemented to media at 500 µg/mL, 100 µg/mL, and 5 µg/mL, respectively, for streptococcal strains.

Routine molecular biology techniques

All PCR reactions utilized Phusion or Q5 polymerase (Thermo Fisher, Waltham, MA). PCR products and restriction digest products were purified using a QIAquick PCR purification kit (Qiagen, Venlo, NL) per manufacturer protocols. Plasmids were extracted using the QIAprep miniprep kit or plasmid midi kit (Qiagen, Venlo, NL) per manufacturer protocols. Restriction enzyme digests utilized XmaI, EcoR1, and BamH1 (New England Biolabs, Ipswich, MA) for 2 h at 37°C in a thermocycler. Ligations utilized Quick ligase (New England Biolabs, Ipswich, MA) at room temperature for 5 min or Gibson Assembly Master Mix (New England Biolabs, Ipswich, MA) per manufacturer protocols. All plasmid constructs were sequence confirmed by Sanger sequencing (CU Anschutz Molecular Biology Core, Aurora, CO) or whole plasmid sequencing (Quantara Biosciences, Hayward, CA).

The mutant strains were generated as previously described (12, 32). Briefly, for the gloA mutant, genomic 5′ and 3′ regions flanking the gloA gene were amplified and fused with a spectinomycin cassette by FailSafe PCR (Lucigen, Middleton, WI). Fragments and pHY304 vector were digested with restriction enzymes and ligated using Quick Ligase. The ligation reaction product was transformed into chemically competent E. coli. pHY304 plasmids were purified from E. coli and electroporated into GBS CJB111 genetic background. Constructs were confirmed by PCR and sequencing. Complement strains were generated by amplifying the gloA gene in GBS and linearizing pABG5 by PCR. Products were ligated using Gibson assembly and then transformed into chemically competent E. coli. Plasmids were purified from E. coli and electroporated into GBS CJB111 ΔgloA genetic background. Primers used in the construction of strains are listed in Table S2. The mutant had no growth or hemolysis defects observed (Fig. S3B through D).

Study approval

Animal experiments were approved by the Institutional Animal Care and Use Committee at the University of Colorado Anschutz Medical Campus protocol #00316 and were performed using accepted veterinary standards. The University of Colorado Anschutz Medical Campus is AAALAC accredited, and its facilities meet and adhere to the standards in the “Guide for the Care and Use of Laboratory Animals.” All mice were purchased from Charles River Laboratories (CD1) and housed in pathogen-free, biosafety level-2 animal facilities.

In vivo transposon screening

Triplicate cultures of the pooled CJB111 pKrmit transposon library (32) were grown overnight at 37°C in THB with kanamycin at 300 µg/mL and back diluted to an OD600 0.4. Libraries were normalized to ~4 × 10^7^ CFU/100 µL and injected via tail-vein into 6–8-week-old CD-1 male mice using the established hematogenous infection model (98–101). Blood was collected by cardiac puncture between ~18 and 28 h post-infection. A 100 µL of input library and blood was plated in duplicated on CHROMagar Strep B with 300 µg/mL kanamycin and incubated overnight at 37°C to collect recovered transposon mutants. Bacterial growth from spread plates was collected and 3–4 mice per library were pooled together, genomic DNA was extracted using a ZymoBiomics DNA miniprep Kit (Zymo Research).

Transposon library sequencing

Libraries were prepared and sequenced at the University of Minnesota Genomics Center according to https://www.protocols.io/view/transposon-insertion-sequencing-tn-seq-library-pre-rm7vzn6d5vx1/v1. Briefly, genomic DNA was enzymatically fragmented, and adapters added using the NEB Ultra II FS kit (New England Biolabs), and ~50 ng of fragmented adapted gDNA was used as a template for enrichment by PCR (16 cycles) for the transposon insertions using mariner-specific (TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCGGGGACTTATCATCCAACC) and Illumina P7 primers. The enriched PCR products were diluted to 1 ng/uL and 10 uL was used as a template for an indexing PCR (nine cycles) using Nextera_R1 (iP5) and Nextera_R2 (iP7) primers. Sequencing was performed using a 150-base paired-end format on an Illumina NextSeq 2000 and Illumina NovaSeq 6000 system to generate ~40–60 million reads per library.

Tn-sequencing analysis

The R1 reads from both sequencing runs were concatenated and quality was assessed using FastQC (102) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were trimmed using Cutadapt (v 4.2) (103) with the following parameters; sequence length with a minimum of 12 bases, removal of flanking “N” bases, reads trimmed of 3′ “G” bases, and reads were trimmed with the reverse complemented mariner transposon sequence (ACTTATCAGCCAACCTGTTA). TRANSIT (v 3.2.7) (104) was used to align trimmed reads to the CJB111 genome (CP063198) and for analysis of transposon insertion sites. The Transit PreProcessor processed reads using default parameters with the Sassetti protocol, no primer sequence, and mapped to the genome sequence using Burrows-Wheeler Alignment (105). Insertion sites were normalized using the Total Trimmed Reads method in TRANSIT and analyzed using the resampling method to compare the insertion counts recovered in blood vs the input library using default parameters, with the addition of ignoring TA sites within 5% of the 5′ and 3′ end of the gene. Significance determined by Padj < 0.05 and log2FC < −1 or >1. All sequencing reads have been deposited into NCBI SRA under BioProject ID PRJNA1125445.

Murine model of bloodstream infection

We infected mice as previously described for CJB111 to cause invasive infection (98–101). Briefly, fifteen 8-week-old CD1 male mice were intravenously challenged with 1.5–2 × 10^7^ CFU GBS CJB111, ∆gloA, or pgloA. At 6, 12, 24, and/or 48 h post-infection, blood samples were taken by tail prick and plated on THA to quantify GBS CFU burden. Once mice reached a moribund state or 72 h post-infection mice were sacrificed, and blood was harvested by cardiac puncture, and lung and heart tissue were removed and homogenized in sterile phosphate buffered saline (PBS). All samples were plated on THA or CHROMagar to quantify GBS CFU burden. For neutrophil depletion, 11–13 mice per group were given 200 µg InVivoMAb anti-mouse Ly6G antibody (Bio X Cell, Lebanon, NH) or 200 µg InVivoMAb rat IgG2a isotype control diluted in InVivoPure pH 7.0 dilution buffer by intraperitoneal injection 24 h before infection.

GloA protein comparisons

GloA amino acid sequences were aligned using ClustalOmega (106) and the alignment figure was created using the ESPript Server (107) (https://espript.ibcp.fr). Protein IDs ABA45143.1 (GBS A909), QOW77196.1 (GBS CJB111), WP_001116201.1 (GBS COH1), WP_002985686.1 (GAS 5448), and P0AC81.1 (E. coli K-12). GBS GloA phylogenetic tree was generated using NCBI BlastP (108, 109) and visualized using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/). The protein ID QOW77196.1 (GBS CJB111) was used as the query against S. agalactiae and only proteins with a percent identity and query cover greater than 50% are shown. The dimeric structure of the S. agalactiae GloA was predicted using AlphaFold2 (110) as implemented in ColabFold (111). PyMOL (version 2.5.2, Schrödinger, LLC.) was used to create images of the predicted GloA structure and the E. coli glyoxalase I crystal structure RCSB PDB entry 19FZ (67, 112) (RCSB.org).

In vitro growth comparisons

Overnight cultures of Streptococcal strains were diluted 100 in mCDM with or without methylglyoxal (MG, Sigma-Aldrich M0252, St. Louis, MO) or hydrogen peroxide 3% wt/wt (VWR, Radnor, PA) at concentrations listed in figure legends in a 96-well plate. For strain/serotype growth comparisons with MG, the overnight cultures were first normalized to OD600 0.6 in PBS before being diluted to 100 in mCDM with or without MG. For testing the effects of MG pre-exposure on growth, overnight cultures were diluted 100 in mCDM with 0.5 mM MG and allowed to grow until the mid-logarithmic phase (OD600 0.4–0.6). Then the mid-logarithmic cultures were used to inoculate a new 96-well plate with fresh mCDM with or without MG. For all growth curves longer than 8 h the plate was covered with a Breathe-Easy gas permeable sealing membrane (USA Scientific, Ocala, FL) and then incubated at 37°C without shaking in a Tecan Infinite M Plex for up to 24 h with OD600 taken every 30 min. For growth curves with CFU/mL shown, the 96-well plate was incubated at 37°C without shaking and samples were taken every 2 h for dilution plating on THA. All growth curves were performed in biological triplicate.

ELISAs on culture pellets

Overnight cultures of Streptococcal strains were diluted 100 in mCDM and then grown for 4 h at 37°C. A 3 mL of each culture was pelleted, re-suspended in PBS, and then homogenized using 0.1 mm dia. Zirconia beads. Methylglyoxal concentration in culture samples was measured using an ELISA kit (Biomatik EKN53482, Kitchener, ON, CA) per manufacturer instructions. Protein concentration in each culture was quantified by Bradford assay, using bovine serum albumin (BSA) standard, and each group was performed in biological duplicate or triplicate.

RT-qPCR

Samples were made by centrifuging 1 mL aliquots of cultures grown to mid-log phase in mCDM and then re-suspending in 1 mL fresh mCDM and incubating for 30 min at 37°C. A 1 mL of RNAProtect Bacteria Reagent (Qiagen, Venlo, NL) was then added before centrifuging and washing pellets with ice-cold PBS. Sample RNA was prepped using the NucleoSpin RNA kit (Macherey-Nagel, Dueren, DE) and TURBO DNase treated (Invitrogen by Thermo Fisher, Waltham, MA) per manufacturer instructions. A 250 ng of RNA was made into cDNA for each sample using the qScript cDNA Synthesis Kit (Quantabio, Beverly, MA) per manufacturer instructions. cDNA was then diluted 20 in water and RT-qPCR run using PerfeCTa SYBR Green FastMix (Quantabio, Beverly, MA) per manufacturer instructions and glcK, gloA, and gloB qPCR primers (see Table S2). Each sample was run in technical duplicate for each gene. Each sample Cq value for glcK, gloA, and gloB was normalized to the total average CJB111 Cq value for each gene, respectively.

Hemolysis assay

Overnight cultures of Streptococcal strains were diluted 100 in THB and then grown to mid-log phase at 37°C. Cultures were then normalized to OD600 0.4 in PBS. A volume of 400 µL blood, 400 µL PBS, and 200 µL normalized culture was added to each sample microfuge tube. A volume of 400 µL blood and 600 µL sterile water was added to positive control tubes and 400 µL blood and 600 µL PBS was added to negative control tubes. Tubes were made in technical duplicate and incubated at 37°C with rotation. At 24 h, 100 µL aliquots were taken and centrifuged at 5,500 × g for 1 min. OD543 of supernatant was measured using a Tecan Infinite M Plex and %lysis was calculated by subtracting negative control from all samples and then dividing samples by positive controls.

HL60-neutrophil and primary bone marrow neutrophil-killing assays

HL60 cells were cultured in RPMI + 10% FBS, differentiated with 1.25% DMSO (Sigma-Aldrich, St. Louis, MO) for 4 days, and infected as previously described (113). Briefly, GBS strains were grown to mid-log, normalized in PBS, and then opsonized in 10% normal human serum or heat-killed (HK) serum in FBB buffer (0.5% BSA and 2.2 mM CaCl2 in HBSS) for 15 min in a 96-well plate. HL60 cells were diluted to the desired concentration in FBB buffer and then either pre-incubated with 20 µM cytochalasin D (Sigma-Aldrich C8273, St. Louis, MO) or DMSO (Sigma-Aldrich, St. Louis, MO) vehicle control for 15 min with rotation or immediately added to each well with opsonized bacteria to reach an MOI of 0.002. The plate was incubated at 37°C with shaking for 5 h. %Survival at 5 h was calculated by dividing CFU recovered from wells with GBS opsonized with normal serum by CFU recovered from wells with GBS opsonized with HK serum. CFU from control wells without HL60 cells were also quantified.

Murine neutrophils were isolated from the bone marrow of 8–12-week-old C57B6/J mice using an Anti-Ly-6G Microbeads kit (Miltenyi Biotec 130-120-337, Germany) per the manufacturer’s instructions. Neutrophil isolations yielded ≥95% purity and were used within 1 h of isolation. Murine BMNs were re-suspended in RPMI + 5% FBS and then adhered to TC-treated 96-well plates for 30 min prior to infection. Neutrophils were infected at an MOI of 0.01 and then synchronized by centrifugation (200 × g for 5 min). Neutrophils were incubated for 4 h in a cell culture incubator at 37°C with a constant rate of 5% CO2. At time points, supernatants were removed and neutrophils were washed with DPBS. Neutrophils were then incubated with 2% Saponin in PBS for 12 min and then dilution plated on THA to enumerate GBS CFU/well. To calculate %associated, the recovered CFU/well for all biological replicates of WT CJB111 at 4 h were averaged and then each biological replicate for each strain was normalized to the WT average.

Cytotoxicity assay

Cytotoxicity of HL60-neutrophils after infection with GBS strains was determined by LDH release using the CyQUANT LDH Cytotoxicity Assay (Invitrogen by Thermo Fisher C20301, Waltham, MA). Briefly, HL60-neutrophils were infected as described in the paragraph above, and then at 5 h, 100 µL of supernatant from each well was removed and spun down at 500 × g for 1 min to remove cells and debris. Then the kit was performed as per the manufacturer’s instructions.

Flow cytometry detection of methylglyoxal-modified proteins in HL60 cells

To determine the impact of infection on methylglyoxal levels in HL60-neutrophils, differentiated Hl60s were first re-suspended in fresh RPMI + 10% FBS with 0 or 20 mM glucose and allowed to equilibrate for 2 h. PBS or WT GBS at MOI 20 was added to 1 mL aliquots of 10^6^ differentiated HL60 cells and incubated at 37°C with rotation for 2.5 h and then harvested by centrifugation (500 × g). Cells were then stained using eBioscience Fixable Viability Dye eFluor 506 (Catalog # 65-0866-18) in PBS for 30 min at room temperature followed by anti-human Cd11b antibody conjugated to FITC (1:20 dilution; BD Biosciences 562793) in MACS buffer (1.25 g BSA, 0.185 g EDTA, 250 mL PBS) for 30 min at room temperature. Cells were fixed and permeabilized using the FoxP3 fixation/permeabilization kit (Thermo Fisher Scientific, Catalog # 00-5523-00) according to manufacturer’s instructions before staining for intracellular methylglyoxal using an anti-MG antibody conjugated to PE (clone 9E7; Cat # MA5-45812; recognizes methylglyoxal-modified proteins) or an IgG2a isotype control (Cat # MG2A04) at final concentrations of 0.67 ug/mL (30 min in permeabilization buffer at room temperature). Stained cells were run on a BD LSRFortessa (BD Biosciences) using the BD FacsDiva software (v9) and analyzed by BD FlowJo software (v10.8). Gating strategy was determined by fluorescence minus one control. Flow cytometric histograms display anti-MG staining of 2000 CD11b+ events per sample.

Statistical analysis

Statistical analysis was performed using Prism version 10.1 for Windows (GraphPad Software, San Diego, CA, USA) as described in figure legends.