Identification of Glyoxalase A in Group B Streptococcus and its contribution to methylglyoxal tolerance and virulence

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 blood stream 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 539 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 confirm 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.

Author Summary

A mutant screen of group B Streptococcus (GBS) in an invasive murine infection model revealed factors known to be important for GBS survival in the blood 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). MG is a highly reactive byproduct of glycolysis and is made by most cells. The glyoxalase pathway is the most common metabolic pathway for MG breakdown and is only a two-step process using glyoxalase A (gloA) and B (gloB) enzymes. 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 neutrophil survival in vitro and in vivo and, therefore, is an important virulence factor required for invasive infection. Research aimed at understanding metabolic stressors during infection is important for the development of new treatments and will expand our knowledge about host-pathogen interactions.

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 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 datasets 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 polysaccharide 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 penetrate the blood-brain barrier and subsequence 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 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 rescues 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 hours. 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 539 genes as significantly underrepresented (p_adj < 0.05, log₂FC < −1) and 95 genes as significantly overrepresented (p_adj < 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 were 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 methylglyoxal (MG) breakdown (Fig. 1C) [29]. 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]. To begin to characterize this pathway in GBS, we grew several clinical GBS isolates in the presence of MG in a modified chemically defined media (mCDM) [35]. Interestingly, different GBS isolates had varying degrees of resistance to MG with CJB111 exhibiting the highest sensitivity and 515 exhibiting the most resistance to MG (Fig. 1D).

Figure 1.

In vivo transposon mutant sequencing of GBS survival in the blood. (A) CIRCOS atlas representation of the GBS CJB111 genome is shown with base pair ruler on the outer ring. The inner ring in blue shows log₂FC (0 to −5/max) of significantly underrepresented genes (P_adj < 0.05). The next inner ring in pink shows log₂FC (0 to 5/max) of significantly overrepresented genes (P_adj < 0.05). The most inner ring in grey denotes genes with P_adj < 0.001. Underrepresented genes or operons of interest are labeled in the center (also listed in Table 1). (B) Clusters of orthologous genes (COG) assignments for significant gene hits normalized to the total number of GBS genes in each COG. The total number of significant genes in each COG are in parentheses. (C) Diagram of the glyoxalase pathway for methylglyoxal (MG) breakdown. (D) Percent growth of GBS isolates representing serotype Ia, Ib, III, and V and S. pyogenes 5448 (GAS) with 0.5 mM MG in mCDM at 6 hrs. Significance determined by (A&B) TRANSIT analysis and Trimmed Total Reads (TTR) normalization with P_adj < 0.05 and Log₂ Fold Change < −2 or > 2.

Table 1.

GBS virulence factors contribute to survival in blood.

Locus Tag	Gene Name	Description	Fold Change	Adj. P-value	Function	GBS Reference
Virulence Factors
ID870_00715	dltX	teichoic acid D-Ala incorporation-associated protein DltX	−21.26	0	Lipoteichoic acid biosynthesis	[36]
ID870_01745	scpB	segregation/condensation protein B	−27.67	0	Complement evasion and adhesion	[37]
ID870_02600	pil2a-bp	PI-2a subunit	−8.22	0	Adhesion	[38]
ID870_04170	esxAl	WXG100 family type VII secretion target	−4.99	0.00432	Type VII Secretion System	[12]
ID870_04190	esaB	EsaB/YukD family protein	−20.68	0.02043	Type VII Secretion System
ID870_05725	iagA	glycosyltransferase	−20.97	0	Lipid biosynthesis	[39]
ID870_05905	cylK	CylK protein	−7.11	0.00118	β-hemolysin biosynthesis	[40, 41]
ID870_05910	cylJ	CylJ protein	−2.73	0.03089	β-hemolysin biosynthesis
ID870_05915	cylH/I	beta-ketoacyl-[acyl-carrier-protein] synthase family protein	−4.23	0.0031	β-hemolysin biosynthesis
ID870_05920	cylF	aminomethyltransferase family protein	−12.91	0	β-hemolysin biosynthesis
ID870_06030	pill-bp	PI-1 major pilin	−7.01	0	Adhesion	[42]
ID870_07490	brpA	LCP family protein	−4.35	0	Biofilm regulation	[43]
ID870_09735	yfhO	YfhO family protein	−9.38	0.00063	Glycan biosynthesis	[44]
Capsule
ID870_02780	capA	CapA family protein	−5.24	0.0017	Capsule biosynthesis	[22, 45, 46]
ID870_03485	cps4A	LCP family protein	−6.15	0	Capsule regulation
ID870_03500	cpsD	tyrosine-protein kinase	−7.36	0	Capsule biosynthesis
ID870_03505	cpsE	sugar transferase	−9.85	0	Capsule biosynthesis
ID870_03510	cpsF	UDP-N-acetylglucosamine--LPS N-acetylglucosamine transferase	−26.72	0	Capsule biosynthesis
ID870_03515	cpsG	multidrug MFS transporter	−28.64	0	Capsule biosynthesis
ID870_03530	cpsN	glycosyltransferase family 2 protein	−11.31	0	Capsule biosynthesis
ID870_03535	cpsO	glycosyltransferase family 2 protein	−7.41	0	Capsule biosynthesis
ID870_03540	cpsJ	glycosyltransferase family 2 protein	−13.36	0	Capsule biosynthesis
ID870_03545	cpsK	glycosyltransferase family 52 protein	−28.25	0	Capsule biosynthesis
ID870_03550	cpsL	oligosaccharide flippase family protein	−22.16	0	Capsule biosynthesis
ID870_03555	neuB	N-acetylneuraminate synthase	−5.21	0	Capsule biosynthesis
ID870_03560	neuC	UDP-N-acetylglucosamine 2-epimerase (hydrolyzing)	−10.06	0	Capsule biosynthesis
ID870_03565	neuD	acetyltransferase	−19.97	0	Capsule biosynthesis
Metal Ion Transport
ID870_00250	Imb	zinc ABC transporter substrate-binding protein	−7.36	0.01419	Zinc uptake and adhesion	[47]
ID870_02010	mtsA	metal ABC transporter substrate-binding protein	−58.89	0	Manganese uptake	[32]
ID870_02020	mtsC	metal ABC transporter permease	−7.31	0	Manganese uptake
ID870_02070	nikA	nickel ABC transporter or nickel/metallophore periplasmic binding protein	−10.85	0	Nickel/copper uptake (putative)	[32, 48]
ID870_02660	fhuG	iron ABC transporter permease	−6.41	0	Siderophore-dependent iron uptake	[49]
ID870_02665	fhuB	iron ABC transporter permease	−6.50	0	Siderophore-dependent iron uptake
ID870_05550	mntH	Nramp family divalent metal transporter	−2.87	0.03162	Manganese uptake	[50]
ID870_07395	copZ	heavy-metal-associated domain-containing protein	−68.12	0.00559	Copper efflux	[51]
ID870_08645	adcC	metal ABC transporter ATP-binding protein	−32.67	0	Zinc uptake	[47]
Two-Component Systems (TCS)
ID870_00330	maeR	response regulator	−6.96	0	Malic acid metabolism regulation (TCS-15)	[52–54]
ID870_00705	dltR	response regulator transcription factor	−6.02	0	Lipoteichoic acid regulation (TCS-14)	[36, 52, 53, 55]
ID870_04460	ciaH	HAMP domain-containing histidine kinase	−8.63	0	Antimicrobial peptide resistance (TCS-10)	[52, 53, 56]
ID870_04495	bceS/nsrK	sensor histidine kinase	−10.41	0.0017	Antimicrobial resistance (TCS-9)	[52, 53, 57–59]
ID870_04500	bceR/nsrR	response regulator transcription factor	−20.11	0.00118	Antimicrobial resistance (TCS-9)
ID870_10420	cssS	HAMP domain-containing histidine kinase	−9.13	0	Two-component system (TCS-19)	[52, 53]
Metabolism
ID870_00585	purA	adenylosuccinate synthase	−15.03	0	Purine metabolism	[60]
ID870_02260	gloA	lactoylglutathione lyase	−18.38	0	Methylglyoxal detoxification
ID870_02270	yvgN	aldo/keto reductase	−4.00	0.00518	Aldehyde detoxification
ID870_02315	glnQ	amino acid ABC transporter ATP-binding protein	−5.28	0.00063	Glutamine transport	[61]
ID870_02320	glnP	ABC transporter substrate-binding protein/permease	−4.69	0	Glutamine transport
ID870_04545	guaA	glutamine-hydrolyzing GMP synthase	−12.55	0	Purine metabolism	[60, 62]
ID870_06660	gloB	MBL fold metallo-hydrolase	−25.63	0.04532	Methylglyoxal detoxification
ID870_08815	argH	argininosuccinate lyase	−4.79	0.00063	Arginine metabolism	[13, 63, 64]
ID870_09305	purB	adenylosuccinate lyase	−16.22	0	Purine metabolism	[60]
ID870_09420	purC	phosphoribosylaminoimidazole succinocarboxamide synthase	−27.67	0.00118	Purine metabolism	[60]
ID870_09800	guaB	IMP dehydrogenase	−38.05	0	Purine metabolism	[60, 62]
ID870_10075	argF	ornithine carbamoyltransferase	−13.36	0	Arginine metabolism	[13, 63, 64]
ID870_10080	arcC	carbamate kinase	−9.58	0.00264	Arginine metabolism

^*pvalue < 0.0001 is assigned pvalue of 0 by TRANSIT analysis

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, MG tolerance. To explore the strain differences in MG tolerance further, we selected three representative serotype strains commonly used in the lab with low to high resistance: CJB111 (V), A909 (Ia), and COH1 (III). First, we compared GloA and GloB amino acid sequences between these three strains and previously characterized proteins from Streptococcus pyogenes (GAS) and Escherichia coli (Fig. 2A&B) [28, 65–67]. We chose these strains since the GAS glyoxalase pathway has been investigated previously and it is within the same genus and there is a solved crystalized structure for GloA and GloB in E. coli which is important for protein modeling. The CJB111 GloA is 63.5% identical to GAS GloA and 40.5% identical to E. coli GloA while CJB111 GloB is 94.3% identical to GAS GloB and 25.7% identical to E. coli GloB. The GloA from A909 and CJB11 and GloB from all three GBS strains are 100% identical. Interestingly, 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 12 out of 57 GloA proteins (21%) have the A45S change with another 5 having a different A45 variant (Fig. 2C). 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. We also generated a phylogenetic tree for GloB to compare 106 GBS genomes but there were only 2 variants, Q55K and F134L, which occurred in at least 5 GloB proteins (5%) (Fig. 2D). To assess tertiary structure, predicted protein models for GBS GloA and GloB were generated using AlphaFold2 and had extremely high confidence for most residues (Fig. 2E&F, Fig. S1A&B). The predicted structures were compared to the solved E. coli GloA (PDB 1F9Z) and GloB (PDB 6RZ0) structures and found to have highly similar topology and conserved metal binding residues (Fig. 2A,B,E,F). GloA was also modeled in its active form as a dimer to show predicted active sites (Fig. S1C Left). The A45S variant from COH1 GloA was included in the dimer (Fig. S1C Right) and modeled to be next to the predicted active site. Also of note, there were 2 large regions on or near the C-terminus of the GloB protein in E. coli that were missing in GBS and GAS which contributes to the low % identity between the proteins and extra alpha-helices in the predicted structure. 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 RT-qPCR, and found that COH1 has higher abundance of both gloA and gloB transcripts compared to CJB111 and A909 (Fig. S1D). All together, the predicted GloA and GloB proteins in GBS contain conserved residues important for enzyme activity and metal binding but it is not clear if common amino acid changes correlate with enzyme activity. In addition, transcript expression for gloA and gloB are increased COH1 which could explain higher MG resistance.

Figure 2.

Characterization of GBS glyoxalase proteins. (A) Alignment of GloA and (B) GloB amino acid sequences from GBS CJB111, GBS A909, GBS COH1, S. pyogenes 5448, and E. coli K-12. Green stars indicate known or predicted metal binding sites and colored bar indicates confidence of structure prediction. (C) Phylogenetic tree for 57 GBS GloA proteins. Proteins/branches with a mutation in amino acid residue A45 are labeled and colored. Red indicates an A45S mutation, purple indicates an A45T mutation, and grey indicates a mutation that only occurred once. (D) Phylogenetic tree for 106 GBS GloB proteins. Proteins/branches with amino acid changes in Q55 and F134 are labeled and colored. (E) Superimposed tertiary structures for E. coli K-12 GloA (yellow) and predicted GBS A909/CJB111 GloA (cyan). The grey spheres are Nickel 2+ ions. (F) Superimposed tertiary structures for E. coli K-12 GloB (yellow) and predicted GBS A909/CJB111/COH1 GloB (cyan).

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 (ΔgloA) 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 ΔgloA strain, we measured MG concentrations using an ELISA on lysed cell pellet samples for CJB111 WT, ΔgloA, 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 ΔgloA 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 growth delay was observed for ΔgloA compared to WT with the addition of MG, which confirms it is involved in MG detoxification and also indicates that MG is bacteriostatic at this concentration (Fig. 3B). The OD₆₀₀ at 8 hrs was compared between strains and confirmed a significant decrease in ΔgloA growth compared to WT or the complemented strain following the addition of MG (Fig. 3C). As MG is primarily produced from glycolysis in all (bacteria and host) cells, 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 ΔgloA mutant compared to WT or complemented strain at low, medium (max growth for GBS in mCDM), or high glucose concentrations tested (Fig. S2A). Additionally, upon assessment of general virulence characteristics, we also did not observe a difference in susceptibility to hydrogen peroxide or hemolytic activity between CJB111, ΔgloA, and pgloA (Fig. S2 B&C). 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.

Figure 3.

The contribution of glyoxalase A to GBS methylglyoxal detoxification. (A) ELISA MG quantification of cell pellets for WT CJB111, ΔgloA, and pgloA strains after growth in mCDM for 4 hours. (B) Growth curve measured by OD₆₀₀ for WT CJB111, ΔgloA, and pgloA strains grown with or without 0.5 mM MG in mCDM. (C) Comparison of growth shown in (B) at 8hrs between strains. Significance determined by (A) One-way ANOVA or (C) 2way ANOVA with Fisher’s LSD multiple comparisons test with P < 0.05. * < 0.05, **** < 0.0001.

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 CJB111 WT or the ΔgloA mutant and monitoring the infection for up to 72 hours post-infection. Mice infected with the ΔgloA 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 hrs post-infection and at the TOD. Mice infected with ΔgloA had significantly decreased blood burdens at all time-points and in tissues 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&C). Of note, the ΔgloA infected mice that died before 72 hours had the highest CFU counts in the blood at 24 hours.

Figure 4.

Methylglyoxal detoxification is necessary for GBS infection. (A) Survival curve of mice tail-vein injected with 1.5–2 × 10⁷ CFU WT CJB111 or ΔgloA. (B) Recovered CFU counts from the blood of infected mice at 24 and 48 hours post-infection. (C) Recovered CFU counts from the blood, heart, and lungs of infected mice at time-of-death. In B&C, limit of detection is indicated with dotted line and mice that died before 72 hours are indicated as a square. Significance determined by (A) Log-rank (Mantel-Cox) test, (B) Mixed-effects analysis with uncorrected Fisher’s LSD test, or (C) multiple Mann-Whitney U tests with P < 0.05. * < 0.05, ** < 0.01.

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 [56], which also utilize glycolysis as their primary energy source [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, ΔgloA, and pgloA strains. At 5 hrs post-infection, the ΔgloA 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 hrs (Fig. S3A&B). We also confirmed that ΔgloA has decreased survival in the presence of primary murine bone marrow neutrophils (BMNs) (Fig. S3C). To evaluate if this increased killing might be due to increased 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. 5B), indicating that infection increases intracellular MG within HL60s. This increase was also only observed in high glucose conditions (Fig. S3D), which is similar to what has been described for S. pyogenes where they reported that GloA contributed to neutrophil survival 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 hrs [73] before intravenous infection with 1 × 10⁷ CFU CJB111 or ΔgloA. Upon assessing morbidity and mortality of these groups over 72 hours post-infection, we observed that both the CJB111 and ΔgloA-infected neutrophil depleted mice exhibited significant increases in mortality when compared to their non-depleted cohorts (Fig. 5C), although percent survival of neutrophil-depleted mice infected with ΔgloA remained higher than neutrophil-depleted mice infected with CJB111. At 12 hours post-infection, we observed that neutrophil depletion abrogated the attenuated phenotype of the ΔgloA mutant in the blood, as the CJB111 and ΔgloA-infected, neutrophil depleted mice did not differ in blood burdens. Further, CJB111 and ΔgloA CFU burdens were significantly increased in the neutrophil-depleted mice compared to the non-depleted mice (Fig. 5D). Taken together these results show that the attenuation of the ΔgloA 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.

Figure 5.

Methylglyoxal detoxification is necessary for GBS survival against neutrophils. (A) Survival of WT CJB111, ΔgloA, and pgloA strains after five hours of infection of HL60-neutrophils. (B) Flow cytometry quantification of intracellular MG-modified proteins in HL60-neutrophils with or without WT CJB111 infection. Left: Representative histogram displaying MG signal that includes isotype control, uninfected, and infected cells. Right: Geometric mean fluorescent intensity (MFI) quantification for MG in uninfected and infected samples. (C) Survival curve of normal or neutrophil depleted mice that were tail-vein injected with 10⁷ CFU WT or CJB111 ΔgloA. (D) Recovered CFU counts from the blood of infected mice 12 hours post-infection. Limit of detection is indicated with dotted line. Significance determined by (A&D) One-way ANOVA with Fisher’s LSD multiple comparisons test, (B) Unpaired Student t test, or (C) Log-rank (Mantel-Cox) test with P < 0.05. * < 0.05, *** < 0.001, **** < 0.0001.

Discussion

GBS must be able to survive multiple host niches to cause invasive infection 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 by Hooven et al. 2017 using a TnSeq library in the GBS A909 serotype Ia background [75]. Their results found similar underrepresented genes compared to our in vivo dataset 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 dataset, other putative stress response proteins like ytgB and asp1 were significantly underrepresented. Most notably, asp1 is annotated as an Asp²³/Gls²⁴ 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 [36, 55]. 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 [36]. 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 [57, 58]. Its role in GBS pathogenesis was demonstrated as a bceR mutant yielded decreased virulence during murine infection and decreased biofilm formation [57]. 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 [37, 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, methylglyoxal (MG) detoxification has not been previously characterized in prior GBS studies.

MG is a highly reactive electrophilic species (RES) and byproduct of normal cell metabolism which can be spontaneously or enzymatically produced by all cells [29, 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 L. 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 dataset; 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 gloA 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 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 second and third highest overall tolerance to MG but additional strains would need to be tested to determine if serotype III have higher average MG tolerance. Interestingly, COH1 (serotype III) had high resistance to MG and 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 515, 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, S. mutans was shown to be more tolerant to MG than most other commensal oral Streptococcal species and was also shown to outcompete S. 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 dataset 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, 47]. 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 [48]. There are only 9 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, 48, 88, 89]. Interestingly, another of the 9 known nickel-dependent enzymes is GloA which was found to use nickel (Ni²⁺) 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 still 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 (DHAP) during glycolysis by methylglyoxal synthase (mgsA); however, GBS, like other streptococci, do not have a 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 serum of diabetic individuals, the concentration of MG and MG-derived AGEs are 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. 5B, Fig. S3D). Therefore, research into the role of the GBS glyoxalase pathway in the context of diabetic wound infection is a current area of study.

Lastly, 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 [29]. 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. S2B) 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, an essential role of the glyoxalase pathway to GBS MG resistance and overall pathogenicity during blood stream 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 in that MG detoxification is an important component for bacterial survival against host defenses [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 treatment 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 THB, unless otherwise stated. Streptococcal chemically defined medium [35] was modified by omitting L-cysteine and adding 22mM 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 QIAquick PCR purification kit (Qiagen, Venlo, NL) per manufacturer protocols. Plasmids were extracted using 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 hours 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. S2).

Study approval.

Animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) 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 × 10⁷ 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 ~18–28 hours post infection. 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 were collected and 3–4 mice per library were pooled together, genomic DNA extracted using ZymoBiomics DNA miniprep Kit (Zymo Research).

Transposon library sequencing.

Libraries were prepared and sequenced at the University of Minnesota Genomics Center (UMGC) 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 1ng/ul and 10 ul was used as a template for an indexing PCR (9 cycles) using Nextera_R1 (iP5) and Nextera_R2 (iP7) primers. Sequencing was performed using 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 were 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 (BWA) [105]. Insertion sites were normalized using the Total Trimmed Reads (TTR) 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 p_adj < 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, 20 8-week-old CD1 male mice were intravenously challenged with 1.5–2 × 10⁷ CFU GBS CJB111 or ΔgloA. At 12, 24, and/or 48 h post-infection, blood samples were taken by tail prick and plated on THA to quantify GBS CFU burden. At time-of-death 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 PBS. All samples were plated on THA or CHROMagar to quantify GBS CFU burden. If it was noted that mice did not receive the full injection during infection and/or did not have CFU counts in the blood and tissues then they were removed from the dataset. For neutrophil depletion, mice were given 200μg InVivoMAb anti-mouse Ly6G antibody (Bio X Cell, Lebanon, NH) or 200μg InVivoMAb rat IgG2a isotype control diluted in InVivoPure pH 7.0 dilution buffer by intraperitoneal injection 24 h before infection.

Glyoxalase protein comparisons.

GloA and GloB amino acid sequences where aligned using ClustalOmega [106] and the alignment figure was created using the ESPript Server [107](https://espript.ibcp.fr). GloA protein IDs used: 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). GloB protein IDs used: ABA44385.1 (GBS A909), QOW76118.1 (GBS CJB111), WP_001115859.1 (GBS COH1), WP_011285457.1 (GAS 5448), and AAC73317.1 (E. coli K-12). GBS phylogenetic trees were generated using NCBI BlastP [108, 109] and visualized using FigTree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/). The protein IDs QOW77196.1 (GBS CJB111 GloA) and QOW76118.1 (GBS CJB111 GloB) were used as the query against S. agalactiae and only proteins with percent identity greater than 65% and query cover greater than 50% are shown. The dimeric structures of the S. agalactiae GloA and GloB was predicted using AlphaFold2 [110] as implemented in ColabFold [111]. For GloB, AlphaFold2 could not generate a predicted structure for the first 21 amino acids on the NH2-terminal segment so these were made invisible. PyMOL (version 2.5.2, Schrödinger, LLC.) was used to create images of the predicted GloA and GloB structures and the E. coli glyoxalase I RCSB PDB 19FZ and glyoxalase II RCSB PDB 6RZ0 [67, 112](RCSB.org).

In vitro growth comparisons.

Overnight cultures of Streptococcal strains were diluted 1:100 in mCDM with or without methylglyoxal (MG, Sigma-Aldrich M0252, St. Louis, MO) or hydrogen peroxide 3% w/w (VWR, Radnor, PA) at concentrations listed in figure legends in a 96-well plate. For strain/serotype comparisons with MG, the plate was incubated at 37°C without shaking and OD600 readings were taken at 0 and 6 hrs. Due to growth rate differences between strains, percent growth was calculated by dividing the average OD600 with MG by the average OD600 no MG control for each individual strain. For 24 hr growth curves with MG, 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 24 hrs with OD600 taken every 30 min. For growth curves with hydrogen peroxide, the plate was incubated at 37°C without shaking and culture aliquots were taken every 2 hrs for dilution plating on THA.

ELISAs on culture pellets.

Overnight cultures of Streptococcal strains were diluted 1:100 in mCDM and then grown for 4 h at 37°C. 3 mL of each culture was pelleted, re-suspended in 200 μL 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 manufacture instructions. BSA protein assay standard was used to quantify protein concentration in each culture sample.

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. 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 manufacture instructions. 250 ng of RNA was made into cDNA for each sample using the qScript cDNA Synthesis Kit (Quantabio, Beverly, MA) per manufacture instructions. cDNA was then diluted 1:20 in water and RT-qPCR run using PerfeCTa SYBR Green FastMix (Quantabio, Beverly, MA) per manufacture 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 1:100 in THB and then grown to mid-log phase at 37°C. Cultures were then normalized to OD600 0.4 in PBS. 400 μL blood, 400 μL PBS, and 200 μL normalized culture was added to each sample microfuge tube. 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 hrs, 100 μL aliquots were taken and centrifuged at 5500 × g for 1 min. OD543 of supernatant was measured using a Tecan Infinite M Plex and %lysis calculated by subtracting negative control from all samples and then dividing samples by positive controls.

HL60 culture and primary bone marrow neutrophil killing assay.

HL60 cells were cultured in RPMI + 10% FBS, differentiated with 1.25% DMSO 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 2.2mM CaCl₂ in HBSS) for 15 min in a 96-well plate. Then, HL60 cells were diluted to desired concentration in FBB buffer and added to each well to reach an MOI of 0.002. The plate was incubated at 37°C with shaking for 5 hrs. %Survival at 5 hrs was calculated by dividing CFU recovered from wells with GBS opsonized with normal serum by CFU recovered from wells with GBS opsonized with heat-killed (HK) serum. CFU from control wells without HL60 cells were also quantified. Murine neutrophils were isolated from 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 bone marrow neutrophils (BMNs) were re-suspended in RPMI + 5% FBS 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 hr 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 hrs, 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 per manufacture 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 20mM glucose and allowed to equilibrate for 2 hrs. PBS or WT GBS at MOI 20 was added to 1 mL aliquots of 10⁶ differentiated HL60 cells and incubated at 37°C with rotation for 2.5 hours 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 minutes at room temperature followed by anti-human CD11b antibody conjugated to FITC (1:20 dilution; BD Biosciences 562793) in MACS buffer (1.25g BSA, 0.185g EDTA, 250mL PBS) for 30 minutes 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 minutes 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 (FMO) controls. 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.

Supplementary Material

Supplement 1

Table S1. Complete in vivo blood Tn-sequencing dataset.

Supplement 2

Figure S1. GBS GloA and GloB predicted structures and variant analysis. idDT confidence scores across GBS (A) GloA and (B) GloB predicted structures. (C) Left: Predicted tertiary protein dimer for GBS A909/CJB111 GloA. Right: 180° horizontal rotation of the predicted tertiary protein dimer with the blue monomer containing GBS COH1 A45S mutation. Black arrows indicate predicted active site. (D) Baseline transcription of gloA and gloB genes grown to mid-log in mCDM and quantified by RT-qPCR. Gene transcript levels were normalized to the average CJB111 levels. Significance determined by 2way ANOVA with uncorrected Fisher’s LSD test, P < 0.05. * < 0.05, ** < 0.01.

Supplement 3

Figure S2. The impact of Glyoxalase A on GBS growth and virulence. (A) Growth curve measured by OD₆₀₀ for WT CJB111, ΔgloA, and pgloA strains grown in mCDM with 10, 22, or 50 mM glucose. (B) Growth curves for CJB111, ΔgloA, and pgloA strains in mCDM with 0.1% (29.4mM) hydrogen peroxide quantified by CFU enumeration. (C) Percent red blood cell lysis of human blood for CJB111, ΔgloA, and pgloA strains relative to positive lysis control at 24hrs post-inoculation.

Supplement 4

Figure S3. The impact of serum, primary cells, and glucose on neutrophil phenotypes. (A) Survival of CJB111, ΔgloA, and pgloA strains over time in the presence of serum without HL60-neutrophils. Percent survival was calculated by dividing CFU recovered from wells with GBS opsonized with normal serum by CFU recovered from wells with GBS opsonized with heat-killed (HK) serum. (B) Cytotoxicity of HL60-neutrophils incubated with CJB111, ΔgloA, and pgloA that had been opsonized with HK or normal serum after 5 hrs of infection. (C) Recovered CFU counts normalized to the WT average of WT CJB111, ΔgloA, and pgloA strains after 4 hr infection of primary bone marrow neutrophils (BMN). (D) Flow cytometry quantification of intracellular MG-modified proteins in HL60-neutrophils with or without WT CJB111 infection in 0mM glucose media. Left: Representative histogram displaying MG signal that includes isotype control, uninfected, and infected cells. Right: Geometric MFI quantification for MG in uninfected and infected samples. Significance determined by (C) one-way ANOVA with Uncorrected Fisher’s LSD test or (D) Unpaired Student t test, P < 0.05.

Supplement 5

Table S2. Primers and Strains