Public health data collected by PulseNet support a role for spv genes in cases of bacteremia caused by Salmonella Typhimurium and I 4,[5],12:i:-

ABSTRACT

Nontyphoidal Salmonella enterica (NTS) infections in humans typically present as self-limiting gastroenteritis, but a small percentage progress to extraintestinal illness, including bacteremia. The spvRABCD (Salmonella plasmid virulence) locus is a plasmid-localized regulon and is hypothesized to contribute to NTS bacteremia. We analyzed 10,544 serotype Typhimurium (I 4,[5],12:i:1,2) and 5,271 serotype I 4,[5],12:i:- (abbreviated as “I 4:i:-”) genomes from human case isolates that were collected and uploaded to PulseNet between 1 January 2019 and 11 May 2022. We examined the relationships between isolate spv status, infection site, patient age, sex, and seven-gene multilocus sequence typing (MLST) sequence type. Comparing the spv+ and spv− subpopulations of Typhimurium and I 4:i:- revealed respective 3.5- and 2.2-fold increases in the proportion of blood isolates among the spv+ subpopulation. Bivariate analysis revealed associations between spv+ status, patient age group, patient sex, and infection site for both serotypes. Generalized estimating equation models found increased odds of isolation of spv+ isolates from blood compared with stool for both serotypes (Typhimurium adjusted odds ratio [aOR]: 4.1 [95% CI: 3.4–5.1] P < 0.001; I 4:i:- aOR: 2.3 [95% CI: 1.6–3.2] P < 0.001) and increased odds of urine vs stool isolation for spv+ isolates for Typhimurium (aOR: 1.6 [95% CI: 1.3–2.0] P < 0.001). Patient age and sex were also predictive of blood and urine isolation, relative to stool. Additionally, MLST sequence types with higher rates of spv carriage showed higher rates of blood infections. In summary, our analysis strongly supports a role for the spv locus in bloodstream infection cases caused by Typhimurium and I 4:i:-.

IMPORTANCE

This study details a comprehensive analysis of 15,815 clinical isolates of Salmonella Typhimurium and I 4,[5],12:i:- collected through the PulseNet USA Network to investigate the role of the spvRABCD locus in nontyphoidal Salmonella infections, with a focus on its association with bloodstream infections. These findings demonstrate increased odds of bacteremia in strains that carry the spv genes in two prevalent nontyphoidal serotypes. This research builds on and confirms previous studies’ implications that these genes contribute to the risk of blood infections by utilizing large human data sets to directly connect bacterial genotype to a specific isolation site in a human. This work both enhances our understanding of the virulence mechanisms of Salmonella and underscores the importance of genomic surveillance in identifying strains that pose a heightened risk for invasive disease.

INTRODUCTION

Salmonella enterica is a gram-negative, rod-shaped bacterial pathogen often linked to foodborne illness. Human infections with S. enterica may present as enteric (or typhoid) fever, self-limiting gastroenteritis, or extraintestinal illness leading to bacteremia and focal infections. S. enterica serotypes are classified as either typhoidal or nontyphoidal, with the former limited to serotypes that cause enteric fever: Typhi, Sendai, and Paratyphi A, B (tartrate-negative), and C. Nontyphoidal Salmonella (NTS) serotypes normally cause only self-limiting gastroenteritis (1). Still, approximately 5% of NTS infections result in extraintestinal illness leading to bacteremia and focal infections, which is termed invasive NTS (iNTS) disease (1–3). Some host-adapted NTS serotypes, such as Dublin and Choleraesuis, as well as the genetically distinct Typhimurium ST313, cause iNTS disease at much higher rates (4, 5).

Despite its relatively low incidence in high-income countries, the threat and global public health burden posed by iNTS disease is significant, with an estimated 535,000 cases worldwide in 2017 (5, 6). This burden weighs most heavily on the populations of sub-Saharan Africa, especially young children (5–9), and consequently, most iNTS research has been conducted using data collected from this region. Many studies have attempted to determine both the host risk factors and the characteristics of the infecting strain that tend to increase the likelihood of iNTS infection. Among predisposing host factors are compromised immune status, extremes of age, malnutrition, malaria infection, other underlying illnesses (such as sickle cell anemia), and genetics (3, 4, 8, 10–17). Both Salmonella pathogenicity islands (SPI-1 and SPI-2), which together facilitate epithelial invasion and survival in the macrophage, are required for extraintestinal infection (18). In addition, several studies have found that serotypes that carry the spv genes (Salmonella plasmid virulence) are overrepresented among isolates recovered from blood samples (3, 19, 20). The previously mentioned host-adapted serotypes Dublin and Choleraesuis, as well as Enteritidis, Typhimurium, and I 4,[5],12:i:- (hereafter, abbreviated as “I 4:i:-”)—a monophasic serotype largely evolved from Typhimurium (I 4,[5],12:i:1,2) (21–26)—have a propensity to carry the spv genes (3, 18, 20, 27–29).

The spv regulon consists of five genes—spvRABCD—typically localized to an IncF virulence plasmid usually designated pSLT, which has been observed in subspecies I serotypes (30, 31). In some non-subspecies I serotypes, spv genes are sometimes localized to the chromosome (31). The functions of these genes have been assessed in a variety of animal models and/or in vitro human cell lines (32). The first gene, spvR, encodes a LysR-type transcriptional regulator and is required for virulence in a BALB/c mouse model and human monocyte-derived macrophages (13, 33–35). It is located upstream and is transcribed separately from the spvABCD genes, which form an operon (13, 36). During the stationary phase, SpvR acts in conjunction with the stress response sigma factor RpoS to activate transcription at its own promoter and that of spvA (33, 37, 38). In contrast, SpvA appears to negatively regulate the spvABCD operon, acting to fine-tune expression and counter the positive feedback loop of SpvR (39).

The last three genes carry out functions in virulence, primarily through interference with host immunity (40). SpvB, SpvC, and SpvD are all injected into host cells by the type III secretion systems encoded in SPI-1 and/or SPI-2 (41, 42). SpvB is required for virulence in chickens (Salmonella Gallinarum) (43) and human monocyte-derived macrophages (35). SpvB and SpvC are both required for virulence in mice (13, 44, 45). While spvD is not required for the invasion of Caco-2 human epithelial cells, it appears to strongly promote Salmonella reproduction within these cells (46). Moreover, recent work by the USDA identified spvD as one of eight genes associated with highly pathogenic Salmonella (HPS) and is included in their HPS multiplex PCR assay (47).

As previously mentioned, several studies have pointed to a role for the spv genes in iNTS disease due to the overrepresentation of spv-carrying serotypes among blood isolates. However, it is also true that most illness caused by these same serotypes—except for Dublin, Choleraesuis, and Typhimurium ST313—is gastrointestinal in nature, indicating that the presence of the spv locus is not itself predictive of invasive illness. It is possible that these genes act opportunistically or in combination with other environmental factors and/or host characteristics to enact iNTS disease. This locus does not appear to be required for gastrointestinal infection, as most serotypes are spv-negative. In addition, all NTS serotypes cause invasive infections at some frequency (48, 49), indicating that the presence of this locus alone is not sufficient to enact an invasive infection.

One of the earliest studies investigating the Salmonella virulence plasmid provided the first and possibly strongest support for a direct connection between the spv genes and human blood infections (19). This study utilized a 4-kb restriction fragment from the center of the region known to be responsible for virulence—later known to be the spv regulon—from Salmonella Dublin’s virulence plasmid. The fragment was hybridized to lysed colonies from clinical isolates of fecal, extraintestinal, or blood origin, and results showed that 76% of extraintestinal and blood isolates hybridized compared to 42% of fecal isolates.

Our goal in this study was to assess the relationship between iNTS disease in humans and carriage of the spv genes by the infecting isolate from a large data set derived from the PulseNet USA Network. PulseNet is the national molecular subtyping network for foodborne disease surveillance and collects WGS data from Salmonella case isolates and limited patient metadata (50–52). This report details our analysis of 15,815 human clinical isolates of serotypes Typhimurium and I 4:i-. We evaluated the relationships of isolate spv status and age and sex of the patient to the source site of the specimen (i.e., blood, urine, or stool). We also considered the isolates’ seven-gene multilocus sequence typing sequence type (MLST ST) in the type of infection caused and spv status. Finally, we examined the phylogeny of these isolates using both seven-gene MLST and core genome MLST (cgMLST) approaches.

MATERIALS AND METHODS

Study design

Data set export

A data set of all Salmonella specimen data submitted to PulseNet between 1 January 2019 and 11 May 2022 was exported for subsequent analysis. PulseNet switched to using whole-genome sequencing as its primary cluster detection method in 2019, and thus, this time period yielded the most complete surveillance data to date at study onset. The exported data included all available patient metadata and unique sequencing identification numbers linkable to the National Center for Biotechnology Information (NCBI) assigned to each isolate. All PulseNet data submitted to NCBI had to meet PulseNet quality metrics for Salmonella (53). The metadata ultimately utilized for our study included the isolate’s serotype, the site of infection from which the sample was taken (called the “source site”), the patient’s age and sex, and the location (U.S. state) of the public health laboratory that submitted the sample.

Serotype selection and determination

We chose to limit our analysis to serotypes Typhimurium and I 4:i:- only, as previous data analysis (unpublished) indicated that both serotypes have both robust spv-positive (“spv+”) and spv-negative (“spv−”) subpopulations, which would allow comparison and analysis within each serotype. Serotype was first determined using SeqSero1 as implemented in BioNumerics v7.6 (54). In cases where the SeqSero1-generated serotype disagreed with the information manually input by the submitting lab, we used SeqSero2 v1.2.1 (55, 56) to determine the serotype.

Identification of the spv status of isolates

Assembly FASTA files were obtained from NCBI GenBank using Entrez v15.5 (57). Using NCBI Blast+ v2.9.0 (57), we created a custom FASTA database for each serotype and queried them for the presence of the spv locus using nucleotide BLAST (BLASTn). We also included three positive control queries: invA, ttrA, and fur. Consistent with the methodology used by the Salmonella Reference Laboratory at CDC, gene targets of invA and ttrA were used as a verification check of Salmonella species identification metadata provided by NCBI at the time of analysis (58, 59). Supplemental data from a recent study also indicated that the third target, fur, is highly conserved across 810 strains randomly selected from 14 Salmonella enterica subspecies I serotypes, including Typhimurium and I 4:i:- (60). Our genes of interest included all five genes from the spvRABCD locus. Salmonella Typhimurium LT2 (NCBI accession NC_003197.2) served as the reference genome for the positive control queries, and the pSLT plasmid from Salmonella Typhimurium LT2 (NCBI accession NC_003277.2) served as the reference genome for the spv queries. Additional detail on the reference sequences is provided in Table S1.

We considered a hit to have ≥95% nucleotide identity and ≥60% cumulative coverage across contigs to their query gene. Genomes were required to meet these thresholds for at least two of the three positive control genes to be included in the analysis.

Source site reconciliation

Source site data in PulseNet were obtained from a free-form input by the submitting laboratory and required pre-processing and consolidation. Inputs indicating stool or a source from within the gastrointestinal tract were consolidated into “Stool/Intestinal” (also simply called “stool”). Inputs indicating blood or plasma sources were consolidated into “Blood”. Inputs indicating urine were consolidated into “Urine.”

Seven-gene MLST

We used mlst v2.19.0 with the “senterica” scheme (https://github.com/tseemann/mlst) to determine isolates’ seven-gene MLST sequence type using FASTA files as input. Minimum spanning trees were constructed using the seven-gene MLST profiles in Grapetree v1.5.0 (61) using the standard neighbor-joining (NJ) method.

cgMLST

Dendrograms were constructed using cgMLST profiles in BioNumerics v7.6.3 using the unweighted pair group method and arithmetic mean (UPGMA) algorithm and exported in Newick format for visualization in iTOL v6 (62). Isolate subsampling was carried out in JupyterLab v3.4.5 using Python v3.9.13, Pandas v1.4.3, and Random v3.

Statistical analyses

Patient ages were grouped into six categories, ranging from 0 to 75 years of age and older. Isolates with unknown patient sex were classified as missing. Isolates with missing or incomplete state of origin data were classified using their unique identifier, which contained their state information.

Overall association between three factors—patient age, patient sex, and spv status—and specimen isolation from blood vs stool and urine vs stool (source site) was examined using Pearson’s chi-square test. The association of these factors was also examined within each individual serotype. Additionally, descriptive analyses of the ST by serotype and source site were performed.

The relationship between source site and each of the three factors (patient age, patient sex, and spv status) was further assessed using a generalized estimating equation (GEE) with a binomial distribution and a logit function. This process was carried out twice to assess blood vs stool and urine vs stool, enabling dichotomous analysis of the source site. The GEE method was selected to account for any potential correlation between isolates collected from different states. Two additional GEE models were carried out for each of the two serotypes separately. Furthermore, two sensitivity analyses were executed by excluding (i) ST302 and ST313 and (ii) ST34 from the GEE models. All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).

RESULTS

Data set preparation and BLASTn analysis

The original data set exported from PulseNet included 11,747 Typhimurium and 5,833 I 4:i:- human isolates collected between the period of 1 January 2019 and 11 May 2022. We excluded 954 (8%) Typhimurium and 451 (8%) I 4:i:- from further analysis that either did not have assemblies available on NCBI GenBank or could not be unambiguously linked to their GenBank assemblies. We also excluded 226 (2%) Typhimurium and 104 (2%) I 4:i:- isolates whose source site was missing, vague, or not categorized as blood, urine, or stool. These exclusions brought our data set down to 10,567 Typhimurium and 5,278 I 4:i:-. All genomes met our requirements for carriage of two out of the three positive control genes, and all three controls were present in all but three isolates.

Following BLASTn analysis for the spv genes, we noted that spvB was the best-conserved gene among our samples by a very slight margin (Table 1). We considered this together with spvB’s superior virulence characterization in the literature and opted to label samples as spv+ if they met the identity and coverage thresholds for spvB alone. To ensure that samples were accurately labeled as spv−, we identified 23 Typhimurium and 7 I 4:i:- isolates that were positive for spvR, spvA, spvC, and/or spvD, but spvB was not found or did not meet the coverage threshold. These were excluded from the data set, bringing the final data set down to 10,544 Typhimurium and 5,271 I 4:i:- isolates (Table 2).

TABLE 1.

BLASTn hit counts and percentages for each spv gene and the whole spvRABCD regulon for Typhimurium and I 4,[5],12:i:-, PulseNet, USA, 1 January 2019–11 May 2022

Genes detected	Typhimurium (10,770)		I 4:i:- (5,375)
	# of isolates	% of isolates	# of isolates	% of isolates
spvR	5,224	49.54	1,885	35.76
spvA	5,231	49.61	1,893	35.91
spvB	5,244	49.73	1,894	35.93
spvC	5,185	49.17	1,842	34.95
spvD	5,035	47.75	1,784	33.85
spvRABCD	5,013	47.54	1,781	33.79

TABLE 2.

Descriptive statistics between all metadata variables and spv status, overall and by serotype Typhimurium and I 4,[5],12:i:-, PulseNet, USA, 1 January 2019–11 May 2022^a,b

	Both serotypes			Typhimurium			I 4,[5],12:i:-
	All 15,815	spv+ 7,138 (45)	spv− 8,677 (55)	All 10,544	spv+ 5,244 (50)	spv− 5,300 (50)	All 5,271	spv+ 1,894 (36)	spv− 3,377 (64)
Age, n (%)
0–5	4,001 (26)	1,738 (25)	2,263 (27)	2,779 (27)	1,271 (25)	1,508 (30)	1,222 (24)	467 (26)	755 (23)
6–19	2,447 (16)	1,077 (16)	1,370 (16)	1,662 (16)	792 (16)	870 (17)	785 (16)	285 (16)	500 (15)
20–45	3,291 (22)	1,413 (21)	1,878 (23)	2,104 (21)	1,058 (21)	1,046 (21)	1,187 (24)	355 (20)	832 (26)
46–59	2,037 (13)	908 (13)	1,129 (14)	1,313 (13)	672 (13)	641 (13)	724 (14)	236 (13)	488 (15)
60–74	2,270 (15)	1,107 (16)	1,163 (14)	1,520 (15)	816 (16)	704 (14)	750 (15)	291 (16)	459 (14)
75+	1,125 (7)	592 (9)	533 (6)	760 (7)	427 (8)	333 (7)	365 (7)	165 (9)	200 (6)
Missing	644	303	341	406	208	198	238	95	143
Sex, n (%)
Female	7,461 (50)	3,337 (50)	4,124 (50)	4,936 (49)	2,436 (49)	2,500 (49)	2,525 (51)	901 (51)	1,624 (51)
Male	7,532 (50)	3,381 (50)	4,151 (50)	5,069 (51)	2,504 (51)	2,565 (51)	2,463 (49)	877 (49)	1,586 (49)
Missing	822	420	402	539	304	235	283	116	167
Source site, n (%)
Blood	913 (6)	652 (9)	261 (3)	657 (6)	511 (10)	146 (3)	256 (5)	141 (7)	115 (3)
Urine	844 (5)	420 (6)	424 (5)	548 (5)	313 (6)	235 (4)	296 (6)	107 (6)	189 (6)
Stool	14,058 (89)	6,066 (85)	7,992 (92)	9,339 (89)	4,420 (84)	4,919 (93)	4,719 (90)	1,646 (87)	3,073 (91)

^a Values in parentheses indicate the percentage of a variable within its category for that column, missing data excluded.

^b Percentages may not add up to 100 due to rounding.

We noted that 95.6% of Typhimurium and 94.0% of I 4:i:- isolates that were spvB+ had the complete spvRABCD regulon, with spvD most likely to be undetected. A full breakdown of the spv regulon patterns present in our data set is shown in Fig. S1. Tables showing all isolates and their NCBI accessions, source sites, age groups, MLST sequence types, and BLASTn hits are provided in File S1.

Analysis of spv+ vs spv− subpopulations

When comparing the proportions of each source site (stool, blood, or urine) within the spv− and spv+ subpopulations, only the proportion of blood isolates increased substantially, constituting a 3.5-fold increase (2.8%–9.7%) among spv+ Typhimurium and a 2.2-fold increase (3.4%–7.4%) among spv+ I 4:i:- (Fig. 1). Proportions of urine and stool isolates appeared to vary only marginally from the spv− to spv+ subpopulations. Data tables showing raw number counts for each source site and all spv genes present for each site are available in Table S2.

Fig 1.

Comparisons of the percentages comprising each source site in the spv− and spv+ subpopulations of (A) Typhimurium and (B) I 4,[5],12:i:-. PulseNet, USA, 1 January 2019–11 May 2022.

Descriptive statistics and bivariate analysis

Table 2 summarizes descriptive statistics and bivariate analysis for the whole data set as well as for each serotype. Across all isolates, 6% (913/15,815) were cultured from blood. Blood isolates constituted 6% (657/10,544) of Typhimurium and 5% (256/5,271) of I 4:i:-, respectively. Overall, 45% of all isolates (7,138/15,815) were spv+. Typhimurium isolates were 50% spv+, and I 4:i:- isolates were 36% spv+. Across all data, 71% of blood isolates (652/913) were spv+, while only 50% (420/844) of urine isolates and 43% (6,066/14,058) of stool isolates were spv+. These proportions varied when considered within each serotype. For Typhimurium, the percentages of spv+ isolates among source types were 78% (511/657) for blood, 57% (313/548) for urine, and 47% (4,420/9,339) for stool. For I 4:i:-, the percentages of spv+ isolates among source types were 55% (141/256) for blood, 36% (107/296) for urine, and 35% (1,646/4,719) for stool.

Bivariate analyses were carried out for blood vs stool isolates (Table 3) and urine vs stool isolates (Table 4). For both the serotypes combined and held separate, these analyses indicated significant associations between patient age groups and isolates taken from each of blood and urine compared with stool (all P-values < 0.001). For Typhimurium isolates, patient sex was associated with each of blood and urine vs stool (P < 0.001), but for I 4:i:-, this was only the case for the urine vs stool comparison (P < 0.001). Finally, spv status was associated with blood vs stool in both Typhimurium and I 4:i:-, but only Typhimurium isolates showed a significant association for spv status in the urine vs stool comparison (P < 0.001).

TABLE 3.

Descriptive statistics and bivariate analysis comparing blood vs stool isolates, overall and by serotype Typhimurium and I 4,[5],12:i:-, PulseNet, USA, 1 January 2019–11 May 2022^a,b

Blood vs stool	Both serotypes				Typhimurium				I 4,[5],12:i:-
	All 14,971	Blood 913 (6)	Stool 14,058 (94)	χ2	All 9,996	Blood 657 (7)	Stool 9,339 (93)	χ2	All 4,975	Blood 256 (5)	Stool 4,719 (95)	χ2
Age, n (%)				<0.001				<0.001				<0.001
0–5	3,939 (27)	64 (7)	3,875 (29)		2,733 (28)	45 (7)	2,688 (30)		1,206 (25)	19 (8)	1,187 (26)
6–19	3,180 (22)	126 (14)	3,054 (23)		2,035 (21)	87 (14)	1,948 (22)		1,145 (24)	39 (16)	1,106 (25)
20–45	1,888 (13)	200 (23)	1,688 (13)		1,220 (13)	144 (23)	1,076 (12)		668 (14)	56 (22)	612 (14)
46–59	2,369 (16)	24 (3)	2,345 (17)		1,609 (17)	19 (3)	1,590 (18)		760 (16)	5 (2)	755 (17)
60–74	2,041 (14)	290 (33)	1,751 (13)		1,371 (14)	198 (31)	1,173 (13)		670 (14)	92 (37)	578 (13)
75+	942 (7)	182 (21)	760 (6)		642 (7)	142 (22)	550 (6)		300 (6)	40 (16)	260 (6)
Missing	612	27	585		386	22	364		226	5	221
Sex, n (%)				<0.001				<0.001				0.496
Female	6,839 (48)	358 (41)	6,481 (49)		4,553 (48)	247 (39)	4,306 (49)		2,286 (49)	111 (46)	2,175 (49)
Male	7,350 (52)	517 (59)	6,833 (51)		4,931 (52)	389 (61)	4,542 (51)		2,419 (51)	128 (54)	2,291 (51)
Missing	782	38	744		512	21	491		270	17	253
spv status, n (%)				<0.001				<0.001				<0.001
spv+	6,718 (45)	652 (71)	6,066 (43)		5,065 (51)	146 (22)	4,919 (53)		3,188 (64)	115 (45)	3,073 (65)
spv−	8,253 (55)	261 (29)	7,992 (57)		4,931 (49)	511 (78)	4,420 (47)		1,787 (36)	141 (55)	1,656 (35)

^a Values in parentheses indicate the percentage of a variable within its category for that column, missing data excluded.

^b Percentages may not add up to 100 due to rounding.

TABLE 4.

Descriptive statistics and bivariate analysis comparing urine vs stool isolates, overall and by serotype Typhimurium and I 4,[5],12:i:-, PulseNet, USA, 1 January 2019–11 May 2022^a,b

Urine vs stool	Both serotypes				Typhimurium				I 4,[5],12:i:-
	All	Urine	Stool	χ2	All	Urine	Stool	χ2	All	Urine	Stool	χ2
	14,902	844 (6)	14,058 (94)		9,887	548 (6)	9,339 (94)		5,015	296 (6)	4,719 (94)
Age, n (%)				<0.001				<0.001				<0.001
0–5	3,937 (28)	62 (8)	3,875 (29)		2,734 (29)	46 (9)	2,688 (30)		1,203 (25)	16 (6)	1,187 (26)
6–19	3,165 (22)	111 (14)	3,054 (23)		2,017 (21)	69 (13)	1,948 (22)		1,148 (24)	42 (15)	1,106 (25)
20–45	1,837 (13)	149 (18)	1,688 (13)		1,169 (12)	93 (18)	1,076 (12)		688 (14)	56 (20)	612 (14)
46–59	2,423 (17)	78 (10)	2,345 (17)		1,643 (17)	53 (10)	1,590 (12)		780 (16)	25 (9)	755 (17)
60–74	1,980 (14)	229 (28)	1,751 (13)		1,322 (14)	149 (28)	1,173 (13)		658 (14)	80 (28)	578 (13)
75+	941 (7)	181 (22)	760 (6)		616 (6)	116 (22)	500 (6)		325 (7)	65 (23)	260 (6)
Missing	619	34	585		386	22	364		233	12	221
Sex, n (%)				<0.001				<0.001				<0.001
Female	7,103 (50)	622 (77)	6,481 (49)		4,689 (50)	383 (74)	4,306 (49)		2,414 (51)	239 (84)	2,175 (49)
Male	7,015 (50)	182 (23)	6,833 (51)		4,680 (50)	138 (26)	4,542 (51)		2,335 (49)	44 (16)	2,291 (51)
Missing	784	40	744		518	27	491		266	13	253
spv status, n (%)				<0.001				<0.001				0.657
spv+	6,486 (44)	420 (50)	6,066 (43)		5,154 (52)	313 (57)	4,420 (47)		1,753 (35)	107 (36)	1,646 (35)
spv−	8,416 (55)	425 (50)	7,992 (57)		4,733 (48)	235 (43)	4,919 (53)		3,262 (65)	189 (64)	3,073 (65)

^a Values in parentheses indicate the percentage of a variable within its category for that column, missing data excluded.

^b Percentages may not add up to 100 due to rounding.

Multivariate analysis for the prediction of invasive source sites

We utilized GEE multivariate analysis to determine the significance of the fold increases for blood in spv+ subpopulations and ascertain whether any of our variables were associated with isolation from more invasive source sites: blood and urine, as opposed to stool. For these analyses, we calculated the ability of serotype, patient age, patient sex, and spv+ genotype to predict blood (Table 5) and urine (Table 6) isolation vs stool.

TABLE 5.

Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) comparing blood vs stool from GEE models, overall and by serotype, PulseNet, USA, 1 January 2019–11 May 2022^a,b

Blood vs stool	Both serotypes		By serotype
			Typhimurium		I 4,[5],12:i:-
	aOR	P-value	aOR	P-value	aOR	P-value
	(95% CI)		(95% CI)		(95% CI)
Serotype
Typhimurium	1.1 (0.9–1.4)	0.323
I 4,[5],12:i:-
Age
0–5	0.3 (0.2–0.5)	<0.001	0.3 (0.2–0.6)	<0.001	0.4 (0.2–0.7)	0.001
6–19	0.2 (0.1–0.4)	<0.001	0.2 (0.1–0.5)	<0.001	0.2 (0.1–0.4)	<0.001
20–45	Ref.		Ref.		Ref.
46–59	2.9 (2.1–3.8)	<0.001	3.1 (2.3–4.1)	<0.001	2.4 (1.3–4.5)	0.004
60–74	4.0 (3.1–5.1)	<0.001	4.0 (3.0–5.2)	<0.001	4.0 (2.3–7.0)	<0.001
75+	5.6 (4.4–7.2)	<0.001	6.7 (4.9–9.0)	<0.001	3.8 (2.2–6.7)	<0.001
Sex
Female	0.7 (0.6–0.8)	<0.001	0.6 (0.5–0.8)	<0.001	0.9 (0.7–1.1)	0.231
Male	Ref.		Ref.		Ref.
spv status
spv+	3.4 (2.9–4.0)	<0.001	4.1 (3.4–5.1)	<0.001	2.3 (1.6–3.2)	<0.001
spv−	Ref.		Ref.		Ref.

^a Variable groups are serotype, age, sex, and source site.

^b Bold values indicate significant P-values at P ≤ 0.05.

TABLE 6.

Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) comparing urine vs stool from GEE models, overall and by serotype, PulseNet, USA, 1 January 2019–11 May 2022^a,b

Urine vs stool	Both serotypes		By serotype
			Typhimurium		I 4,[5],12:i:-
	aOR	P-value	aOR	P-value	aOR	P-value
	(95% CI)		(95% CI)		(95% CI)
Serotype
Typhimurium	0.9 (0.8–1.1)	0.39
I 4,[5],12:i:-
Age
0–5	0.4 (0.3–0.6)	<0.001	0.5 (0.3–0.8)	0.005	0.3 (0.2–0.6)	<0.001
6–19	1.0 (0.7–1.3)	0.916	1.0 (0.7–1.5)	0.835	0.9 (0.6–1.4)	0.668
20–45	Ref.		Ref.		Ref.
46–59	2.3 (1.8–3.1)	<0.001	2.4 (1.7–3.5)	<0.001	2.2 (1.4–3.5)	0.001
60–74	3.6 (2.7–4.7)	<0.001	3.8 (2.6–5.7)	<0.001	3.2 (2.2–4.7)	<0.001
75+	6.9 (5.2–9.1)	<0.001	7.4 (4.9–11.2)	<0.001	6.4 (3.4–9.6)	<0.001
Sex
Female	3.6 (3.0–4.4)	<0.001	3.0 (2.4–3.8)	<0.001	5.6 (4.1–7.8)	<0.001
Male	Ref.		Ref.		Ref.
spv status
spv+	1.4 (1.2–1.7)	<0.001	1.6 (1.3–2.0)	<0.001	1.1 (0.8–1.4)	0.727
spv−	Ref.		Ref.		Ref.

^a Variable groups are serotype, age, sex, and source site.

^b Bold values indicate significant P-values at P ≤ 0.05.

For the blood vs stool analysis (Table 5), we found that spv+ genotype was associated with increased odds of a strain being isolated from blood as opposed to stool for both Typhimurium (adjusted odds ratio [aOR] = 4.1; 95% confidence interval [CI] 3.4–5.1, P < 0.001) and I 4:i:- (aOR = 2.3; 95% CI 1.6–3.2, P < 0.001). Age groups followed a similar pattern to the bivariate analysis, demonstrating a significant association with the isolation site even after adjusting for other factors. Typhimurium and I 4:i:- both showed the same trend of having increased odds of blood infection for the three age groups older than the reference group—the “20–45,” presumed immunocompetent age group—and decreased odds of blood infection (i.e., increased odds of gastrointestinal infection) for the two youngest age groups (Typhimurium: all P-values < 0.001; I 4:i:-: all P-values ≤ 0.004). For Typhimurium, the estimated odds of blood isolation were lowest in the “6–19” age group (aOR = 0.2; 95% CI 0.1–0.5, P < 0.001) and highest in the “75+” age group (aOR = 6.7; 95% CI 4.9–9.0, P < 0.001). For I 4:i:-, the estimated odds of blood isolation were lowest in the “6–19” age group (aOR = 0.2; 95% CI 0.1–0.4, P < 0.001) and highest in the “60–74” age group (aOR = 4.0; 95% CI 2.3–7.0, P < 0.001). Intriguingly, in Typhimurium only, we found decreased odds for blood isolation among patients reported as female compared with those reported as male (aOR = 0.6; 95% CI 0.5–0.8, P < 0.001). Serotype was not associated with blood infection (aOR = 1.1; 95% CI 0.9–1.4, P = 0.323) in the combined serotype model.

For the urine vs stool analysis (Table 6), we found that spv+ genotype was predictive of isolation from urine for Typhimurium only (aOR = 1.6; 95% CI 1.3–2.0, P < 0.001). The association of age group to urine as the source site showed similar trends to that of the blood vs stool analysis, but with some key differences. Again, for both Typhimurium and I 4:i:-, the three oldest age groups had higher odds of urine isolation, and the “0–5” age group had lower odds of urine isolation compared to the reference group (Typhimurium: all P-values ≤ 0.005; I 4:i:-: all P-values ≤ 0.001). However, the odds of the “6–19” age group having urinary infection were not significantly different from the reference group. Estimated odds of urine isolation for both serotypes range from the lowest in the “0–5” group (Typhimurium: aOR = 0.5, 95% CI 0.3–0.8, P = 0.005; I 4:i:-: aOR = 0.3, 95% CI 0.2–0.6, P < 0.001) and highest in the “75+” age group (Typhimurium: aOR = 7.4, 95% CI 4.9–11.2, P < 0.001; I 4:i:-: aOR = 6.4, 95% CI 3.4–9.6, P < 0.001). Furthermore, both serotypes showed increased odds of urine isolation from patients designated female (Typhimurium: aOR = 3.0, 95% CI 2.4–3.8, P < 0.001; I 4:i:-: aOR = 5.6, 95% CI 4.1–7.8, P < 0.001). Finally, as in the blood vs stool analysis, serotype was not associated with urine isolation.

Sensitivity analyses excluding ST313, ST302, and ST34

We conducted a sensitivity analysis excluding ST313 and ST302 due to the well-documented tendency of Typhimurium ST313 to cause bloodstream infections (4, 16, 63, 64), which may be due to other virulence factors in addition to the spv genes (65, 66). Typhimurium ST302 was recently noted to be a one-locus variant of ST313 (66), and phylogenetic analysis indicates that ST302 and ST313 are more closely related to each other than to other ST19 strains (67). The exclusion of these STs caused minor fluctuations in some of the aOR magnitudes but did not change the findings for either the blood or urine vs stool (Table S4). Of note, the odds of spv+ strains causing a blood infection decreased from 4.1 (95% CI 3.4–5.1, P < 0.001) to 3.6 (95% CI 2.9–4.5, P < 0.001) when these STs were excluded.

We conducted a second sensitivity analysis excluding ST34, which lacks the spv genes and was present among both serotypes in our data set, though as mentioned previously, ST34 comprised 3,088 (59%) of the I 4:i:- isolates and only 13 (0.1%) of Typhimurium (Table 7). We opted to reconduct the analysis without ST34 to determine whether this lineage was the source of these differences and observed some changes (Table S5). When ST34 was excluded, the Typhimurium serotype became weakly predictive of blood infection (aOR = 1.3, 95% CI 1.0–1.8, P = 0.032). For I 4:i:-, the odds of “46–59” were no longer significantly different from the reference age group for both blood and urine (blood: aOR = 2.1, 95% CI 0.9–4.8, P = 0.084; urine: aOR = 1.8, 95% CI 0.8–4.0, P = 0.181). Furthermore, there was a reduction in the odds for urine vs stool isolation from a patient designated female, relative to male, from 5.6 (95% CI 4.1–7.8, P < 0.001) to 3.7 (95% CI 2.2–6.3, P < 0.001) for I 4:i:-. As might be expected, excluding ST34 also increases the odds of bloodstream infection for spv+ strains in I 4:i:- from 2.3 (95% CI 1.6–3.2, P < 0.001) to 4.6 (95% CI 1.5–14.0, P = 0.008).

TABLE 7.

Frequencies of seven-gene MLST sequence type by spv status serotype for Typhimurium and I 4,[5],12:i:- both separately and combined, PulseNet, USA, 1 January 2019–11 May 2022^a,b

	Both serotypes			Typhimurium			I 4,[5],12:i:-
	All	spv+	spv−	All	spv+	spv−	All	spv+	spv−
	n = 15,815	n = 7,138	n = 8,677	n = 10,544	n = 5,244	n = 5,300	n = 5,271	n = 1,894	n = 3,377
ST19	11,406 (72.9)	6,545 (93.2)	4,861 (56.4)	9,544 (91.5)	4,702 (91.1)	4,842 (92.0)	1,862 (35.7)	1,843 (98.9)	19 (0.6)
ST2072	169 (1.1)	169 (2.4)		169 (1.6)	169 (3.3)
ST2076	41 (0.3)		41 (0.5)	41 (0.4)		41 (0.8)
ST213	96 (0.6)	7 (0.1)	89 (1.0)	96 (0.9)	7 (0.1)	89 (1.7)
ST2379	217 (1.4)		217 (2.5)				217 (4.2)		217 (6.5)
ST302	135 (0.9)	135 (1.9)		135 (1.3)	135 (2.6)
ST313	52 (0.3)	52 (0.7)		52 (0.5)	52 (1.0)
ST34	3,101 (19.8)		3,101 (36.0)	13 (0.1)		13 (0.2)	3,088 (59.2)		3,088 (92.2)
ST36	247 (1.2)		247 (2.9)	246 (2.4)		246 (4.7)	1 (0.02)		1 (0.03)
ST3631	31 (0.2)	31 (0.4)		31 (0.3)	31 (0.6)
ST3805	11 (0.1)	11 (0.2)					11 (0.2)	11 (0.6)
ST99	23 (0.1)	23 (0.3)		23 (0.2)	23 (0.4)
Other	113 (0.7)	51 (0.7)	62 (0.7)	80 (0.8)	42 (0.8)	38 (0.7)	33 (0.6)	9 (0.5)	24 (0.7)
Missing	173	114	59	114	83	31	59	31	28

^a Values within parentheses indicate the percentage comprising each ST within the column, excluding “Missing”.

^b Percentages may not add up to 100 due to rounding.

Seven-gene multi-locus sequence type analysis

Characterization of STs in data set

A seven-gene MLST ST was assigned to 98.9% of isolates, and the relationship between isolates’ ST and spv status was subsequently analyzed (Table 7). For the subsequent analyses, only STs with at least 10 isolates were considered. Most STs were either completely spv+ (ST2072, ST302, ST313, ST3631, ST3805, and ST99) or completely spv− (ST2706, ST2379, ST34, and ST36). Exceptions to this were ST19 at 57% (6,545/11,406) spv+ and ST213 at 7% (7/96) spv+. Additionally, most STs were exclusive to either Typhimurium (ST2072, ST2076, ST213, ST302, ST313, ST36, ST3631, and ST99) or I 4:i:- (ST2379 and ST3805). ST19, ST34, and ST36 were detected in both serotypes.

Across all isolates for both serotypes (Table 7), the majority (72%; 11,406/15,815) belonged to ST19. ST19 accounted for 92% (6,545/7,138) of all the spv+ isolates, though only 56% (4,861/8,677) of the spv− isolates. ST34 was the second most common ST at 20% (3,101/15,815) and constituted 36% of all spv− isolates. When broken down by serotype, 91% (9,544/10,544) of Typhimurium isolates were ST19, whereas in I 4:i:-, ST34 was the most common at 59% (3,088/5,271) and ST19 was the second most common at 35% (1,862/5,271). The spv+ subpopulations of both Typhimurium and I 4:i:- primarily comprise ST19 at 90% (4,702/5,224) and 97% (1,843/1,894), respectively. Intriguingly, while Typhimurium isolates of ST19 were split nearly 50/50 by spv status (49.3% spv+ and 50.7% spv−), I 4:i:- isolates of ST19 were 99% spv+. As for the spv− subpopulations of each serotype, ST19 accounted for 91% (4,842/5,300) of spv− Typhimurium, and ST34 accounted for 92% (3,088/3,377) of spv− I 4:i:-.

Frequencies of isolates’ source site varied by sequence type, although most isolates, regardless of source type, were ST19. A full breakdown of these frequencies is provided in Table S3.

Relationship between the spv status and blood infections within STs

Figure 2 provides a visualization of the relationships between the STs, their spv composition, and the percentage of blood isolates contained within them. STs are arranged by highest to lowest percentage of blood isolates from top to bottom, and bars are colored according to the percentage of spv+ isolates within the ST. Due to the nearly 50/50 split of spv+ and spv− ST19 in Typhimurium, it is shown with all isolates and broken down into its spv+ and spv− constituents. Except for Typhimurium ST34, we observe that most of the bars at the top (highest percentage of blood isolates) represent STs that are mostly spv+, while those toward the bottom (lowest percentage of blood isolates) are mostly spv−. This shows that the STs comprising the most blood isolates are also the ones with at or near 100% spv+ isolates, and conversely, STs with the least percentage of blood isolates also tend to be spv−.

Fig 2.

Blood infection percentages of STs and their corresponding spv+ percentage by serotype; PulseNet, USA, 1 January 2019–11 May 2022. Analysis of all STs for Typhimurium (Tm) and I 4,[5],12:i:-. ST19 in Typhimurium is shown with all isolates and split by spv status (“+” or “−” indicating spv+ and spv−). The percentage of blood infections within each ST is given on the x-axis, and STs are arranged from highest to lowest percentage of blood isolates from top to bottom. Bar colors indicate the percentage of spv+ isolates within each ST according to the key at the bottom.

Phylogenetic analyses

Seven-gene MLST tree analysis

We further visualized all isolates in NJ trees generated and visualized in Grapetree (61) using seven-gene MLST profiles with metadata for source site and spv status. The resulting NJ trees indicate that most of the I 4:i:- isolates are within the clade containing ST34 and ST2379, while the remainder are interspersed within largely Typhimurium clades (Fig. 3A). The presence of spv varies across and within clades, with a few apparent patterns. The clade beginning at the branch point labeled “A” is largely spv−, as is that containing ST34 and ST2379 (Fig. 3B). The only spv+ node, indicated by a box, within this clade is also the only Typhimurium node and is of ST4476. The clade containing ST2072 and that containing ST303 and ST313 are both primarily spv+.

Fig 3.

Neighbor-joining tree based on seven-gene MLST profiles of Typhimurium and I 4,[5],12:i:- isolates was colored by (A) serotype and (B) spv status. STs with at least 10 isolates are additionally labeled next to nodes. Node size scales with the number of isolates, but due to the large sizes of ST34 and ST19, the scale has been skewed toward smaller isolate sizes for better visualization. Numbers of isolates for each category are provided within the legend. Nodes with more than one type within them are pie graphs displaying the relative frequency of each type. Branch lengths have been scaled logarithmically. The letter “A” denotes a largely spv− clade, with the boxed node calling attention to the sole spv+ node in this clade (see text).

cgMLST tree analyses

To consider whether closely related isolates showed patterns in spv carriage and source site, we constructed a core genome MLST dendrogram for 500 stool and 500 blood isolates randomly selected from the Typhimurium ST19 subset (Fig. 4). Subsampling to a maximum of 1,000 isolates was deemed necessary to allow reasonable computational times and promote ease of tree interpretation. We chose to subsample within Typhimurium ST19, which represents 91.5% of Typhimurium isolates, to ensure that roughly 50% of our randomly selected isolates would be spv+ (Table 7). The resulting tree shows a noticeable pattern of spv+ status (light blue overlying clades) aligning more frequently with blood source site (dark blue in outer ring) than with stool source site (white in outer ring). In many cases, blood isolates cluster among closely related isolates rather than being evenly interspersed throughout the tree, which is particularly evident at the clade labeled “Node A.”

Fig 4.

cgMLST dendrogram for 1,000 randomly subsampled Typhimurium ST19 isolates (500 stool and 500 blood). Blue background overlying the tree branches inside the circle indicates spv+ status. Darker blue shading over the isolate labels in the outer ring indicates isolates from blood samples.

For a more detailed look at genetic relationships among the 913 blood isolates (657 Typhimurium and 256 I 4:i:-), we constructed a cgMLST tree, which is shown in Fig. 5. Intriguingly, the ST34 and ST2379 blood isolates, which constitute all the spv− I 4:i:- isolates, form a subclade (Node A) that diverges from a larger clade, including spv− Typhimurium ST19 and ST213. In contrast, the spv− Typhimurium blood isolates are distributed across several groups throughout the tree. Some are within the previously noted clade containing Node A, and a few appear in isolated clades or single branches of ST34 and ST36. However, most spv− Typhimurium blood isolates are ST19 and descend from clades of spv+ isolates, likely indicating loss of the pSLT plasmid at these branch points. We further noted a second instance of I 4:i:- appearing to descend from Typhimurium at Node B, this time from an spv+ clade of ST19 and with all descendants remaining spv+.

Fig 5.

cgMLST dendrogram for all 913 blood samples across both serotypes. Background color inside the circle indicates serotype (blue for Typhimurium and pink for I 4:i:-). Isolate labels forming the inner of the two outer rings indicate the MLST ST. The spv status is indicated by both the branch color (black, spv+; white, spv−) and the outermost ring color (red, spv+; white/invisible, spv−).

DISCUSSION

These analyses indicate that spv+ S. enterica strains are more likely to be associated with bloodstream infections than spv− strains. Our analyses strongly support an increased odds for blood isolates from humans to carry the spv genes relative to stool isolates. This was true for both serotype Typhimurium and serotype I 4:i:- isolates, though the effect was stronger in the former. Furthermore, our analysis revealed a Typhimurium-specific association between spv+ status and isolation from urine, an atypical site of infection for Salmonella in this study (Table 2; 6% across both serotypes), that may indicate predisposing factors in the patient (68).

Our work does not suggest that the spv genes specifically promote blood infection, as 85% of all spv+ isolates were derived from stool. Instead, this work supports the hypothesis that spv+ Salmonella strains act opportunistically with other factors to promote more invasive infections. Some of these factors, such as coinciding immune suppression or underlying disease, could not be assessed in this study. We attempted to examine extremes of age as a coinciding factor with spv+ status to predict more invasive infections through an additional analysis in which our model included an interaction term between spv+ status and each of the patient age and patient sex variables. The inclusion of these interaction terms did not enhance the model’s performance, so we elected to retain the simpler version of the model presented here. However, we note that the discarded model indicated that age and spv+ status do not interact to predict source site in our data set (data not shown).

While not spv-specific, our analysis further indicated reduced odds of bloodstream infection among the two youngest age groups and increased odds of bloodstream infection among the oldest three age groups (source site relative to stool and age groups relative to the “20–45” group) (Table 5). The increased propensity for bacteremia in all groups above 45 years of age is expected, especially for those over 65 (69). However, it is intriguing that the groups under age 25, especially the “0–5” group, are significantly more likely to have gastrointestinal infections over blood infections compared with the presumably immunocompetent “20–45” group. A possible explanation is that adults in this age group are underrepresented among stool isolates for behavioral reasons: adults may be more likely to wait out gastrointestinal symptoms without visiting a doctor (70), whereas children, especially very young children, are more likely to see a doctor due to parental intervention. For the urine vs stool analysis, the same pattern largely held except for the “6–19” group, which was indifferent to urine as a source site (Table 6). We also found increased odds for blood isolation from male patients compared with female patients, but the opposite was true for urine isolation (Table 4). These sex-specific associations align with previous findings that males are at higher risk for bacteremia (71), while females are more likely to contract urinary tract infections (72).

Our analysis also allowed a look at the variety of patterns that are present in the spv regulon (Fig. S1). We found that for serotypes Typhimurium and I 4:i:-, if a single spv gene is detected in a sample, the entire regulon is likely to be present. The 3′ end of the regulon was more likely to have an undetected gene or genes: spvD alone was most likely to be missing, and spvC and spvD together were second most likely to be missing. This absence may be due to low coverage, assembly error, and/or spvD’s low GC content, which can reduce sequencing accuracy in cases where the Nextera XT Library Preparation Kit was used (73).

Although not the primary focus of this work, we observe that the cgMLST tree shown in Fig. 5 indicates two points at which I 4:i:- arises from the Typhimurium clade. This concurs with the conclusions of other studies that indicate that I 4:i:- emerged from Typhimurium and that many lineages of I 4:i:- are monophasic versions of Typhimurium (see Introduction). We further note several instances in which the spv genes are lost, presumably due to the loss of pSLT. These instances appear more frequently in the cgMLST tree containing both stool and blood samples from Typhimurium ST19 (Fig. 4) than the one containing blood samples alone (Fig. 5). The pSLT plasmid encodes up to three mechanisms to reinforce its stability through bacterial generations (74), and the relatively frequent loss of pSLT seen here suggests that further analysis comparing lineages that retain pSLT with those that lose it could reveal conditions or other genetic backgrounds that affect pSLT maintenance.

Our analysis does include some limitations. As mentioned in Materials and Methods, we used the location of the public health laboratory that submitted the sample data to PulseNet as the U.S. state for GEE analysis. The patient from whom the sample was obtained may or may not have resided in the same state as this lab. We do not expect this to affect our findings or interpretation. Additionally, multiple isolates corresponding to a single patient may have originated in samples collected over the course of an infection. PulseNet recommends state labs consider an isolate a duplicate if it is within 6 months of the original isolate and is within three pairwise alleles by cgMLST. Labs are directed to avoid duplicate uploads, but if this occurs, PulseNet curators keep the upload that has an older isolation date if both have the same source or the entry with the more invasive source if isolates have different sources (i.e., blood or urine as opposed to stool). Therefore, we cannot use this data set to determine the exact frequency that isolates are recovered from individual source sites or track the progression of an infection in a single individual through multiple source sites. This practice of keeping the isolate from the most invasive source site aligns with the objectives of our study. It ensures that isolates from a single patient would not be counted multiple times in our analysis, meaning isolates from the more invasive source sites are retained and quantified. While remaining duplicate isolates may be present in our data set, the proportion of duplicates within the data set is likely to be too low to affect our analysis and statistical outcomes.

In summary, we have demonstrated a strong relationship between spv genes and the site of infection in humans, confirming they are important virulence factors. This is the largest study to date on this correlation. More work is needed to fully understand the many factors that intertwine to allow the spv locus to promote iNTS infections.

DATA AVAILABILITY

All NCBI accession numbers and metadata used in this study, as well as the seven-gene MLST allele calls, can be found in File S1.

ETHICS APPROVAL

This activity was reviewed by CDC, deemed research not involving human subjects, and was conducted consistent with applicable federal law and CDC policy. See, e.g., 45 C.F.R. part 46; 21 C.F.R. part 56; 42 U.S.C. §241(d), 5 U.S.C. §552a, 44 U.S.C. §3501 et seq.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.00504-26.

File S1. spectrum.00504-26-s0001.xlsx.

Contains isolate level data, NCBI accession ids, and additional detailed characterization information.

Supplemental figures and tables. spectrum.00504-26-s0002.ppt.

Figure S1 and Tables S1 to S5.

OPEN PEER REVIEW. reviewer-comments.pdf.

An accounting of the reviewer comments and feedback.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.