Genome Sequence of the Thermotolerant Foodborne Pathogen Salmonella enterica Serovar Senftenberg ATCC 43845 and Phylogenetic Analysis of Loci Encoding Increased Protein Quality Control Mechanisms

Thermal interventions are commonly used in the food industry as a means of mitigating pathogen contamination in food products. Concern over heat-resistant food contaminants has recently increased, with the identification of a conserved locus shown to confer heat resistance in disparate lineages of Gram-negative bacteria. Complete sequence analysis of a historical isolate of Salmonella enterica serovar Senftenberg, used in numerous studies because of its novel heat resistance, revealed that this important strain possesses two distinct copies of this conserved thermotolerance locus, residing on a multireplicon IncHI2/IncHI2A plasmid. Phylogenetic analysis of these loci in comparison with homologs identified in various bacterial genera provides an opportunity to examine the evolution and distribution of loci conferring resistance to environmental stressors, such as heat and desiccation.

ABSTRACT

Salmonella enterica subsp. enterica bacteria are important foodborne pathogens with major economic impact. Some isolates exhibit increased heat tolerance, a concern for food safety. Analysis of a finished-quality genome sequence of an isolate commonly used in heat resistance studies, S. enterica subsp. enterica serovar Senftenberg 775W (ATCC 43845), demonstrated an interesting observation that this strain contains not just one, but two horizontally acquired thermotolerance locus homologs. These two loci reside on a large 341.3-kbp plasmid that is similar to the well-studied IncHI2 R478 plasmid but lacks any antibiotic resistance genes found on R478 or other IncHI2 plasmids. As this historical Salmonella isolate has been in use since 1941, comparative analysis of the plasmid and of the thermotolerance loci contained on the plasmid will provide insight into the evolution of heat resistance loci as well as acquisition of resistance determinants in IncHI2 plasmids.

IMPORTANCE Thermal interventions are commonly used in the food industry as a means of mitigating pathogen contamination in food products. Concern over heat-resistant food contaminants has recently increased, with the identification of a conserved locus shown to confer heat resistance in disparate lineages of Gram-negative bacteria. Complete sequence analysis of a historical isolate of Salmonella enterica serovar Senftenberg, used in numerous studies because of its novel heat resistance, revealed that this important strain possesses two distinct copies of this conserved thermotolerance locus, residing on a multireplicon IncHI2/IncHI2A plasmid. Phylogenetic analysis of these loci in comparison with homologs identified in various bacterial genera provides an opportunity to examine the evolution and distribution of loci conferring resistance to environmental stressors, such as heat and desiccation.

INTRODUCTION

Heat-resistant bacterial contaminants have increasingly become a source of concern in the last several years especially with regard to food safety and human health. Although many members of the family Enterobacteriaceae exhibit variable heat resistance after heat shock, relatively few strains have been found to demonstrate an innate thermotolerant phenotype. Recently, however, a genetic locus found to confer heat resistance has been identified in certain strains of Cronobacter sakazakii, Escherichia coli, and Klebsiella species (1–3). Using a top-down proteomic approach, Williams et al. identified a protein biomarker sequence in Cronobacter correlating with thermotolerance to reverse engineer the DNA sequence of the protein that was homologous to a protein found in thermotolerant Methylobacillus flagellatus (4). Gajdosova and coworkers expanded upon this work and sequenced an 18-kbp region containing this open reading frame (ORF) and determined that the full locus is required for highest thermotolerance (1). Independently, Bojer et al. (3) isolated several Klebsiella pneumoniae strains over the course of a 2-year surveillance study of patients in a Danish hospital; this study demonstrated multiple antibiotic resistance phenotypes (3). These nosocomial isolates were associated with persistence in contaminating reusable endoscopes, and Bojer et al. (3) hypothesized that the adaptability of the isolates was due to their resiliency to harsh environmental stressors such as heat. Bojer et al. then identified a locus encoding small heat shock proteins and a novel Clp ATPase termed ClpK (3). This locus showed high similarity to the same chromosomal region from Methylobacillus that Williams et al. (4) identified and Gajdosova later sequenced (1). Subsequently, Bojer and coworkers reported that the clpK gene was cotransferred with plasmids encoding multiple drug resistance hosted by emerging K. pneumoniae clones associated with nosocomial outbreaks (5). More recently, Mercer et al. (2) and Dlusskaya et al. (6) sequenced an E. coli beef isolate that was able to survive in beef patties grilled to 71°C. Genomic comparisons of highly heat-resistant E. coli isolates resulted in the identification of a 14-kbp locus of heat resistance with greater than 99% identity to the aforementioned loci in Cronobacter and Klebsiella (2). Lee et al. expanded on the thermotolerance genetic module, noting that other ORFs on the loci are homologous to proteins involved in protein homeostasis and proposed that the loci be called transmissible loci for protein quality control (TLPQC) (7, 8). This locus appears to have been horizontally acquired in some bacterial lineages, as it is found not only in a number of clinically relevant members of the family Enterobacteriaceae but also more broadly in other Proteobacteria species (7).

A striking example of innate thermotolerance in the foodborne pathogen Salmonella enterica subsp. enterica has been reported for a particular strain of S. enterica serovar Senftenberg (9). This isolate, S. Senftenberg ATCC 43845 (originally referred to as strain 775W), has been used extensively in thermal resistance studies over the last several decades (9–12); it was first reported to be extremely thermotolerant in 1946 by Winter et al. (13), having been isolated previously by E. Beckler in 1941 from Chinese egg powder at the Massachusetts Department of Public Health (14). While dried egg products are not commonly sold in modern times, dried egg powder was used in rations in many Western countries, especially England, during the onset of World War II. Indeed, multiple reports of salmonellosis in the 1940s were associated with consumption of powdered eggs (15, 16).

Despite being used in numerous studies examining thermotolerance, the genetic determinants conferring heat resistance in S. Senftenberg ATCC 43845 have not been identified previously. Analysis of the complete genome of S. Senftenberg ATCC 43845 revealed the presence of two closely related loci with more than 98% identity to the thermotolerance loci previously studied in Cronobacter, Escherichia, Klebsiella, and Pseudomonas (1–3, 8) on a potentially conjugable IncHI2 plasmid. As the synteny of the genetic loci is highly conserved broadly across diverse members of the phylum Proteobacteria, here we designate the putative thermotolerance loci in S. Senftenberg TLPQC-1 and TLPQC-2 (Fig. 1) in keeping with the nomenclature proposed by Lee et al. (7). Comparative analysis of the S. Senftenberg TLPQCs, as well as other known homologs, provides a unifying framework for understanding the origin and evolution of loci conferring mechanisms of increased protein maintenance and reveals the prevalence and conservation of these loci within the Proteobacteria.

FIG 1 .

Genetic synteny is highly conserved between TLPQC (transmissible loci for protein quality control). The synteny of the TLPQC is highly conserved, suggesting an origin of horizontal gene transfer between several genera (Salmonella, Methylobacillus, Cronobacter, Escherichia, and Klebsiella). Despite reductions in the loci, the synteny of the loci is conserved in the various clades. Conserved ORFs with predicted functions are colored in the figure. The functions are indicated by the following letters (and colors): A, small heat shock proteins (dark blue); B, Clp protease (green); C, cardiolipin synthase (teal); D, FtsH protease (pink); E, YfdX family protein (magenta); F, thioredoxin (black); G, KefB glutathione-regulated potassium efflux pump (purple); H, Zn-dependent protease (red); and I, periplasmic serine protease (dark green). Letters within parentheses indicate that the gene is truncated. The E. coli AW 1.7 nucleotide accession number is NZ_LDYJ01000141.

RESULTS AND DISCUSSION

Genomic features of S. Senftenberg ATCC 43845.

S. Senftenberg ATCC 43845 has been used extensively in thermal resistance studies over the last several decades, having been identified as an extremely thermotolerant strain of Salmonella in 1946 by Winter et al. (13). A draft genome assembly consisting of 114 contigs (139-kb contig N50; accession number NZ_AOXX00000000) and a total length of 5,184,857 bases have previously been reported for this strain (17). Here we present the complete and closed, finished-quality genome sequence of S. Senftenberg ATCC 43845 based on long-read sequencing. The circularized chromosome consists of 4,920,660 bp with a chromosomal GC content of 52.2%, comparable to other finished Salmonella (Fig. 2). Annotation by the NCBI Prokaryotic Genome Annotation Pipeline (version 3.3) indicated the presence of 5,105 total genes, with 4,980 coding sequences (CDS) and 125 genes encoding RNAs (GenBank accession number CP016837). This includes 3,930 predicted proteins with homology to known proteins, in addition to 979 hypothetical or “miscellaneous” predicted proteins. In addition, the chromosome contains potential restriction-modification (RM) systems, including a DNA (cytosine-5)-methyltransferase motif, as well as phage integration-related N⁶-adenine methylation systems. Genome analysis using PHASTER (18) revealed the presence of two intact prophages and one questionable prophage (Fig. 2). Manual inspection revealed a Mu-like prophage, with up to 77% nucleotide identity but only 11% coverage of phage D108 by PHASTER at nucleotide positions 30,172 to 82,470. The Mu-like prophage is integrated downstream of mnmE (also known as thdF or trmE) which is a hot spot integration site for other mobile genetic elements in Salmonella (19). The other intact phage identified by PHASTER is a prophage related to the Haemophilus influenzae HP1 phage at positions 738,135 to 768,638. Manual inspection of the phage region identified as questionable by PHASTER analysis showed what appears instead to be an intact prophage related to the Salmonella phage vB_SosS_Oslo at nucleotides 1,549,983 to 1,599,675. No known virulence factors were associated with any of the prophages. A putative integrative conjugative element (ICE) identified by BLASTn is integrated into tRNA^Phe at nucleotide positions 4,324,610 to 4,433,343. BLAST analysis of the ICE showed no virulence factors; however, the element appears to encode an arsenic resistance locus and a copper homeostasis and silver resistance island (CHASRI) (20). Significantly, there were no predicted thermotolerance loci on the chromosome of the isolate revealed by BLAST analysis.

FIG 2 .

Genome map of S. Senftenberg ATCC 43845. The outer ring displays the mobile genetic elements found in S. Senftenberg ATCC 43845. The second and third rings depict predicted CDS on the forward and reverse strands, respectively. The innermost rings depict GC content and skew.

Features of the pSSE ATCC-43845 plasmid.

The finished assembly of S. Senftenberg ATCC 43845 revealed the presence of a large 341,373-bp circular plasmid, not previously described for this strain, and is here designated pSSE ATCC-43845. Combining the plasmid length with the chromosome takes the total genome complement of the strain to 5,262,033 bases, slightly exceeding that predicted by the short-read draft assembly (17). Annotation of this plasmid sequence via the same method as the chromosome predicts 336 genes on the plasmid, including two complete thermotolerance loci with distinct sequences between nucleotides 189,272 to 211,036 and 134,033 to 148,118 of GenBank accession number CP016838 (Fig. 3). These two loci are designated TLPQC-1 and TLPQC-2. Both thermotolerance islands have characteristic high average GC content (61%), contain the canonical clp protease gene, and are flanked by small heat shock proteins and predicted DNA-binding genes, in keeping with previously reported heat tolerance islands (Fig. 1) (1–3, 7). The GC content of the islands suggests that the origins of TLPQC-mediated heat resistance lie in Proteobacteria with higher average GC content than Salmonella species.

FIG 3 .

Map of the pSSE ATCC-43845 plasmid. The thermotolerance islands are indicated as orange regions on the map and demonstrate increased high GC content (black track). pSSE ATCC-43845 shares high homology with the backbone of the R478 plasmid (red tBLASTx). Base modification signals generated by BaseModFunctions v2.1.R and Circos are overlaid on the plasmid map. Qmod values are height proportional to the interpulse distance ratio for each of the modified bases on the positive strand (red outermost track) and negative strand (red innermost track). Blue hash marks correspond to the CGCAN₈TCNG motifs present on the plasmid.

BLASTn comparison of the entire plasmid sequence to microbial sequences in GenBank indicates that it is closely related to the lncHI2 R478 plasmid, sharing up to 98% identity with more than 61% coverage. In silico analysis of the plasmid sequence with PlasmidFinder (21) identified two replicon regions, IncHI2 and IncHI2A. Comparative analyses further revealed that the pSSE ATCC-43845 plasmid has a comparable GC content value of 48.0% to the GC content of R478 at 45.5% (Fig. 3) (22, 23). Further evidence that pSSE ATCC-43845 is a member of the IncHI2 plasmid family is found by comparing transposon insertion sites which have been used as markers to track closely related plasmids of the same family. In R478 and closely related IncHI2 plasmids with antibiotic resistance, Tn10 and the linked tetracycline resistance genes have been used for this purpose (24), and in pSSE ATCC-43845, the Tn10 transposition site (BFF41_25185 [Fig. 4]) was found to be intact. As with R478, several heavy metal resistance determinants are present on pSSE ATCC-43845, including a tellurite resistance island and another copy of CHASRI (23, 25). As illustrated in Fig. 4, the variable regions of R478 and pSSE ATCC-43845 are bracketed by the tellurite resistance operon and CHASRI. This variable region in R478 encodes mercury resistance and chloramphenicol (cat) and kanamycin (aphA) resistance, while in pSSE ATCC-43845, this region contains the two TLPQC loci. DNA sequence analyses showed that pSSE ATCC-43845 does not contain genes encoding drug resistance determinants. In keeping with this finding, antimicrobial sensitivity phenotyping (as evaluated by broth microdilution testing methods; TREK, Sensititre CMV3AGNF plates) showed S. Senftenberg ATCC-43845 to be pansusceptible to 15 antimicrobial agents tested on the panel. This finding is novel and suggests that pSSE ATCC-43845 is a relic of IncHI2 plasmids before the era of therapeutic antibiotic use, as present-day examples of IncHI2 plasmids have frequently been identified as carriers of antibiotic resistance genes (26–28).

FIG 4 .

BLASTN comparison of the R478 and pSSE ATCC-43845 plasmids. The backbone for R478 is highly conserved in pSSE ATCC-43845. The majority of differences between the two plasmids reside between the tellurite resistance island and CHASRI (copper homeostasis and silver resistance island). The lack of any known antibiotic resistance genes, such as aphA (kanamycin) or cat (chloramphenicol), on pSSE ATCC-43845 is also noted. The integration site for Tn10 with tetracycline resistance is in BFF41_25185 in pSSE ATCC-4384. Heavy metal resistance genes (purple), antibiotic resistance genes (teal), TLPQC genes (red), transposase genes (blue), CHASRI genes (silver), and rep (magenta) and the mercury resistance (mer), methylase (dam), tellurite resistance (ter), arsenic resistance (ars), and replicase (rep) genes are indicated.

Methylation analysis.

Sequence analysis using the Pacific Biosciences platform produces data that can detect base modification in the source DNA (29). Detecting these modifications is dependent on both the depth of sequence coverage and the magnitude of the effect of the modification on the kinetics of nucleotide incorporation. Nucleotide modification is predicted from the interpulse distance data (see Materials and Methods) with N⁶-adenine methylation producing a strong signal that is easily detected at low coverage, while N⁵-cytosine methylation demonstrates a weaker signal that requires deeper coverage. Methylation motif analysis of the S. Senftenberg sequence data and comparisons to established motifs in REBASE (30) showed adenine N⁶-methylation of the sites GATC, CAGAG, and ATGCAT (the adenine methylated is shown underlined), as observed previously with other Salmonella serovars (31). Additional motifs were identified in S. Senftenberg and are summarized in Table 1. Of particular interest was methylation of the CGCAN₈TCNG motif, as the presence of a cluster of three of these motifs was observed in each of the novel clp protease genes found in TLPQC-1 and TLPQC-2. This is illustrated in Fig. 3, where modification values (Qmod) indicating methylation of this site on the positive and negative strands are visualized as the outermost and innermost tracks, respectively, on pSSE ATCC-43845. Depiction of the methylation state of this site on the plasmid also reveals a skew of the presence of these motifs, which is likely due to the chimerism of the R478 backbone. The presence of these motifs in TLPQC-1 and -2 is noteworthy and hints at the possibility that this motif may play some regulatory role in the expression of genes contained in the TLPQC loci. Additional base modification motif maps of the plasmid are visualized in the supplemental figures (see Fig. S1 and S2 in the supplemental material).

TABLE 1 .

Base modification summary table

Motifa	MTase ORFb	No. of motifs in genomec	No. of motifs modifiedc	% modifiedc	Type/subtyped	Coveragee
CAGAG	BFF41_17760	6,607	6,576	99.5	III beta	82.9
ATGCAT	BFF41_02695	776	248	37.7//37.7	II beta	88.8
ACYN6GTTC	BFF41_00175	797//797	793//796	99.5//99.9	I gamma	80.6
CGCAN8TCNGf	BFF41_20045	849//849	847//838	99.8//98.7	I gamma	82.3
GAAN7RTACf	BFF41_21025	1,036//1,036	1,019//1,030	98.4//99.4	I gamma	82.0
GATCg	BFF41_02205	41,312	41,202	99.7//99.7	II	82.6

^aModified adenine bases are shown in bold type; cognate bases are shown in bold font.

^bMotifs are assigned to their respective putative methylases (methylase [MTase] ORFs).

^cNumbers separated by double slashes indicate the values for motifs detected on the positive and negative strand shown before and after the double slashes, respectively, if known.

^dThe type and subtype refer to the class of methylase.

^eCoverage indicates the depth of sequence at the minimum quality threshold set in the protocol.

^fMotifs determined to be unique to S. Senftenberg ATCC 43845, as determined by REBASE.

^gThe GATC motif cannot be assigned unambiguously, but BFF41_02205 was assigned as the most likely candidate.

FIG S1 

GATC methylation for pSSE ATCC-43845. The map of plasmid pSSE ATCC-43845 is shown, with the thermotolerance locus islands indicated as orange regions. Base modification signals generated by BaseModFunctions v2.1.R and Circos are overlaid on the plasmid map. Qmod values are height proportional to the interpulse distance ratio for each of the modified bases on the positive strand (red outmost track) and negative strand (red innermost track). Blue hash marks correspond to the GATC motifs present on the plasmid. Download FIG S1, PDF file, 2.2 MB ^{(2.3MB, pdf)} .

Copyright © 2017 Nguyen et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

FIG S2 

CAGAG methylation for pSSE ATCC-43845. The map of plasmid pSSE ATCC-43845 is shown, with the thermotolerance locus islands indicated as orange regions. Base modification signals generated by BaseModFunctions.v.2.1.R and Circos are overlaid on the plasmid map. Qmod values are height proportional to the interpulse distance ratio for each of the modified bases on the positive strand (red outmost track) and negative strand (red innermost track). Green hash marks correspond to the CAGAG motifs present on the plasmid. Download FIG S2, PDF file, 1.6 MB ^{(1.6MB, pdf)} .

Copyright © 2017 Nguyen et al.

This content is distributed under the terms of the Creative Commons Attribution 4.0 International license.

Comparison of TLPQC loci.

A phylogenetic analysis (see Materials and Methods) was performed to determine the relationship of the Salmonella heat tolerance islands to those in Klebsiella plasmids, Pseudomonas chromosomes, and other bacteria. Heat tolerance islands were identified by searching the literature (previous reports) or by searching for similarity of nucleotide sequence records in GenBank to representative TLPQC sequences by BLASTn (see Materials and Methods). A total of 91 heat tolerance islands, including 48 reported to be located on chromosomes, 42 reported to reside on plasmids, and one presumably on a plasmid (Table 2), as well as the two TLPQC islands from plasmid pSSE ATCC-43845, were included in the analysis. For each species, the thermotolerance island was manually curated from the reported sequence (Table 2). Previously, two clades (A and B) of heat tolerance islands, which were termed “locus of heat resistance” (LHR), were identified (2). We chose to use instead the nomenclature of TLPQC proposed and adopted by Lee et al. (7), since a number of ORFs within the LHR islands have been shown or predicted to encode protein quality control functions involved in surviving a range of stressors beyond heat, including antibiotic exposure, desiccation, and oxidative stress. However, we propose to amend the TLPQC classification by Lee et al. (7) and incorporate phylogenetic analysis from our work and Mercer et al. (2) for multiple clades (Fig. 5). Thus, the broader phylogeny incorporating the pSSE ATCC-43845 islands (Fig. 5) indicates three clades, TLPQC-1, -2, and -3, with the third clade comprised of a reduced TLPQC island primarily observed in plasmids of Klebsiella. There is high genetic conservation within clades and less genetic conservation between clades, and genetic synteny is maintained across multiple genera (Fig. 1 and 5). The two pSSE ATCC-43845 islands fall within clades 1 and 2, consistent with the names assigned here, and are no more closely related to other plasmid-borne islands than to chromosome-borne islands.

TABLE 2 .

Strain information for the sequence data used in the phylogenetic analysis of TLPQC loci in this study

Species	Strain or plasmid	Accession no.	TLPQCa	Location
Achromobacter xylosoxidans	FDAARGOS_162	CP014065	1	Chromosome
	MN001	CP012046	1	Chromosome
Alicycliphilus denitrificans	K601	CP002657	1	Chromosome
Burkholderia multivorans	ATCC 17616	CP000868	1	Chromosome
Burkholderia multivorans chromosome 1	ATCC BAA-247	CP009832	1	Chromosome
Cupriavidus gilardii chromosome 1	CR3	CP010516	1	Chromosome
Dechlorosoma suillum	PS	CP003153	1	Chromosome
Desulfovibrio alaskensis	G20	CP000112	1	Chromosome
Klebsiella pneumoniae	J1 plasmid 1	CP013712	1	Plasmid
Marinobacter aquaeolei	VT8	CP000514	1	Chromosome
Methylobacillus flagellatus	KT	CP000284	1	Chromosome
Pandoraea apista	AU2161	CP011501	1	Chromosome
	DSM 16535	CP013481	1	Chromosome
	TF80G25	CP011279	1	Chromosome
	TF81F4	CP010518	1	Chromosome
Pantoea sp.	PSNIH2	CP009866	1	Chromosome
Pseudomonas aeruginosa	ATCC 27853	CP015117	1	Chromosome
	F30658	CP008857	1	Chromosome
	F9670	CP008873	1	Chromosome
	PA38182	HG530068	1	Chromosome
	S04 90	CP011369	1	Chromosome
	S86968	CP008865	1	Chromosome
	T38079	CP008866	1	Chromosome
	W36662	CP008870	1	Chromosome
	W45909	CP008871	1	Chromosome
	W60856	CP008864	1	Chromosome
	Carb01 63	CP011317	1	Chromosome
Pseudomonas balearica	DSM 6083	CP007511	1	Chromosome
Pseudomonas pseudoalcaligenes	CECT 5344	HG916826	1	Chromosome
Pseudomonas resinovorans	NBRC 106553	AP013068	1	Chromosome
Pseudomonas sp.	VLB120	CP003961	1	Chromosome
Pseudomonas stutzeri	CGMCC 1.1803	CP002881	1	Chromosome
	DSM 4166	CP002622	1	Chromosome
Ralstonia mannitolilytica chromosome 1	SN82F48	CP010799	1	Chromosome
Ralstonia pickettii chromosome 1	12D	CP001644	1	Chromosome
	DTP0602	CP006667	1	Chromosome
Salmonella enterica Senftenberg	pSSE ATCC-43845	CP016838	1	Plasmid
Vibrio parahaemolyticus chromosome 1	UCM-V493	CP007004	1	Chromosome
Cronobacter malonaticus	LMG 23826	CP013940	2	Chromosome
	CMCC 45402	CP006731	2	Chromosome
Cronobacter sakazakii	ATCC 29544	CP011047	2	Chromosome
Enterobacter asburiae	CAV1043 pCAV1043-51	CP011587	2	Plasmid
Enterobacter cloacae	ECNIH5 pENT-22e	CP009855	2	Plasmid
	ECR091 pENT-4bd	CP008907	2	Plasmid
Escherichia coli	P12b	CP002291	2	Chromosome
Klebsiella oxytoca	pKO_JKo3_1 DNA	AP014952	2	Plasmid
Klebsiella pneumoniae	CAV1193 pCAV1193-258	CP013323	2	Plasmid
	CAV1344 pCAV1344-250	CP011623	2	Plasmid
	KPNIH39 pKPN-332	CP014763	2	Plasmid
	pKPN_CZ	JX424424	2	Plasmid
Klebsiella pneumoniae subsp. pneumoniae	KPNIH27 pKPN-262	CP007734	2	Plasmid
Obesumbacterium proteus	DSM 2777	CP014608	2	Chromosome
Salmonella enterica Havana	CFSAN024771	JWQI01000019.1	2	Unknown
Salmonella enterica Senftenberg	pSSE ATCC-43845	CP016838	2	Plasmid
Yersinia enterocolitica	(Type O:5) YE53/03	HF571988	2	Chromosome
	FORC-002	CP009456	2	Chromosome
Cronobacter sakazakii	SP291	CP004091	3	Chromosome
	NCTC 8155	CP012253	3	Chromosome
Enterobacter asburiae	CAV1043	CP011591	3	Chromosome
Enterobacter cloacae complex chromosome 1	35734	CP012162	3	Chromosome
Enterobacter cloacae subsp. cloacae	ATCC 13047	CP001918	3	Chromosome
Klebsiella pneumoniae	207M1D0 KPN207_p3	LT216439	3	Plasmid
	30660/NJST258_1 pNJST258N1	CP006927	3	Plasmid
	32192p	CP010574	3	Plasmid
	34618 p34618	CP010393	3	Plasmid
	500_1420 p500_1420	CP011981	3	Plasmid
	BK32179 pBK32179	JX430448	3	Plasmid
	CAV1392 pCAV1392-131	CP011577	3	Plasmid
	DMC1097 pDMC1097	CP011977	3	Plasmid
	JM45 p1	CP006657	3	Plasmid
	KP-1 pKP1-19	CP012884	3	Plasmid
	KpN01 pKpN01-SIL	CP012989	3	Plasmid
	KpN06 pKpN06-SIL	CP012994	3	Plasmid
	Kpn555 pKPN-d90	CP015132	3	Plasmid
	KPNIH36 pKPN-fff	CP014649	3	Plasmid
	O6CO7 pIT-6C07	LT009688	3	Plasmid
	pKN-LS6	JX442974	3	Plasmid
	pKp848CTX	LM994717	3	Plasmid
	Plasmid 1	CP015823	3	Plasmid
	PMK1 PMK1-A	CP008930	3	Plasmid
	pUUH239.2	CP002474	3	Plasmid
	ST15 pKP02022	KF719972	3	Plasmid
	ST23 pKP007	KF719971	3	Plasmid
	ST258 pKPN-IT	JN233704	3	Plasmid
	ST48 pKP09085	KF719970	3	Plasmid
	U25 PU25001	KT203286	3	Plasmid
	UHKPC07 pUHKPC07	CP011987	3	Plasmid
	UHKPC33 pUHKPC33	CP011990	3	Plasmid
Klebsiella pneumoniae subsp. pneumoniae	KPNIH10 pKPN-498	CP007729	3	Plasmid
	KPNIH24 pKPN-e44	CP008800	3	Plasmid
	KPNIH32 pKPN-a68	CP009777	3	Plasmid
	KPX pKPX-1	AP012055	3	Plasmid
	MGH 78578 pKPN3	CP000648	3	Plasmid

^aThe TLPQC-1, TLPQC-2, and TLPQC-3 loci are indicated by 1, 2, and 3, respectively, in the table.

FIG 5 .

Genomic comparison of thermotolerance islands. MrBayes phylogeny tree reveals three different clades for TLPQC-1 (red branches), TLPQC-2 (blue branches), and TLPQC-3 (green branches). Graphic representations of representatives indicated in bold along the tree are shown to the right in the inset to illustrate conservation, with gray connections between the loci indicating sequence homology. The thermotolerance islands of Proteobacteria show high degrees of conservation, indicating an origin of horizontal gene transfer. Very high genetic conservation is observed within the clade, with less conservation between clades. The truncated cardiolipin synthase and FtsH CDS is conserved across all members of TLPQC-2. Strains in the inset are as follows: Methylobacillus flagellatus (MF), S. Senftenberg TLPQC-1 (SS1), Cronobacter sakazakii (CS), S. Senftenberg TLPQC-2 (SS2), Klebsiella pneumoniae (KP), and Enterobacter asburiae (EA). The functions of the genes are indicated by the following letters (and colors): A, small heat shock proteins (dark blue); B, Clp protease (green); C, cardiolipin synthase (teal); D, FtsH protease (pink); E, YfdX family protein (magenta); F, thioredoxin (black); G, KefB glutathione-regulated potassium efflux pump (purple); H, conserved protein with GGDEF domain (yellow); I, Zn-dependent protease (red); J, periplasmic serine protease (dark green). CDS in orange indicate hypothetical proteins. Black stars indicate TLPQCs that are found within bioinformatically identified integrative conjugative elements.

Members of the TLPQC-1 clade are broadly distributed within the Proteobacteria (Fig. 5) with examples found in environmental isolates such as Methylobacillus flagellatus and Dechlorosoma suillum (32, 33), as well as within clinically relevant bacteria such as Pseudomonas, Pandorea, and Burkholderia (8). The majority of TLPQC-1 were found to be chromosomally located, with the exception of S. Senftenberg TLPQC-1 and another on plasmid 1 of Klebsiella pneumoniae J1 (GenBank accession number CP013712). Comparative analyses of these chromosomally bound TLPQCs suggest that they are inserted into larger mobile genetic elements (data not shown). TLPQC-1 genes in strains marked with black stars (Fig. 5) indicate that the loci are found within an integrative conjugative element (ICE) that is integrated into tRNA^Gly. Attachment repeats flanking the putative ICE were identified with the core att nucleotide sequence 5′ TGGAGCGGGCGATGGGAA. Members of the TLPQC-1 lineage contain the longest islands of the three lineages (~15-kbp average), with core genes predicted to encode a phage-like transcriptional regulator, two small heat shock protein variants, a novel Clp protease, cardiolipin synthase, the membrane-associated ATP-dependent protease FtsH, two variants of YfdX protein family with potential chaperone activity, thioredoxin, a glutathione regulated K⁺ efflux pump related to KefB, a phosphate starvation-inducible PsiE family protein, a Zn-dependent protease, and a periplasmic serine protease (7) (Fig. 5).

Although most of the ORFs in TLPQC-1 have not been experimentally verified, one of the small heat shock proteins, sHsp20c, was recently characterized and found to aggregate in a 24-mer that plays a role in ameliorating stress resistance as a protein chaperone in Pseudomonas (8). Additionally, the novel Clp protease, ClpG_G1, has been shown to confer thermal resistance in Cronobacter, Escherichia, Pseudomonas, and other Enterobacteriaceae (3, 8). The rest of the ORFs in TLPQC-1 have not been studied in depth; however, it has been shown that the full TLPQC is needed for the greatest thermal protection (1). The conservation of TLPQC and protein domains within the uncharacterized ORFs to known proteases, chaperones, and other quality control proteins suggests that these additional ORFs may contribute to survival in other adverse conditions such as osmotic shock or desiccation resistance.

The second lineage observed, TLPQC-2, is the best-studied TLPQC thus far, having first been identified in Cronobacter sakazakii contaminants of powdered infant formula (1). Islands in this lineage are shorter than TLPQC-1 (~14 kb) primarily because of the truncation of genes predicted to encode cardiolipin synthase (cls) and FtsH (hflB). Although these truncated ORFs are conserved across various TLPQC-2s, it is unknown whether they are transcribed and expressed or what function they might serve. TLPQC-2 islands were observed on both chromosomal and plasmid locations and were predominantly found in members of the Enterobacteriaceae, including Escherichia, Obesumbacterium, Klebsiella, Cronobacter, Salmonella, and Yersinia (Fig. 5) (1–3). Bacterial species that possess TLPQC-2 are of great interest to food industry and human health communities, as many are relevant to food safety such as Obesumbacterium proteus as brewery yeast contaminants (34) and C. sakazakii as infant formula contaminants (1). Given that a number of TLPQC-2 loci are found on plasmids, the potential exists for transmission and subsequent conferring of heat resistance to foodborne pathogens. In S. Senftenberg, the presence of the two TLPQCs on a multireplicon conjugative plasmid highlights the potentially mobile nature of these environmental stress resistance determinants.

The third lineage, TLPQC-3, is the shortest of the three loci (mean length of 6.5 kb), and composed only of the ORFs that encode a heat shock protein, the Clp protease, a truncated version of cardiolipin synthase, the Zn-dependent protease, and the periplasmic serine protease (Fig. 5). Found primarily within Klebsiella pneumoniae plasmids, this TLPQC has been associated with the emerging multidrug-resistant K. pneumoniae ST16 clone (5, 35). The Clp protease in TLPQC-3 has been characterized and was named ClpK, sharing 98.3% amino acid identity with the pSSE ATCC-43845 ClpG_G1 (3). The high degree of conservation of the core elements within these islands underscores their apparent importance to host fitness.

Conclusions.

We describe the presence of two broadly disseminated but uncommon TLPQCs present in the important pathogen Salmonella enterica. The discovery of two TLPQCs in S. Senftenberg ATCC 43845 underscores its uniqueness and suggests a likely mechanism for the unusually thermotolerant phenotype repeatedly observed for this strain (9–13). As this strain of S. Senftenberg has been maintained in collections since at least 1941, S. Senftenberg ATCC 43845 is also one of the oldest known isolates to harbor TLPQCs. The presence of TLPQC-1 and TLPQC-2 on a single plasmid raises questions about the selective pressures and genetic events that led to their evolution. As the original source of S. Senftenberg ATCC 43845 was powdered eggs, it is tempting to suggest a correlation between the acquisition of the TLPQC loci and exposure to conditions likely encountered in the production of powdered egg. Regardless of how it evolved, the complete genome sequence and methylome data of this unusually thermotolerant strain, as well as the phylogenetic analysis of these broadly distributed islands conferring enhanced mechanisms of protein maintenance, will facilitate future studies on the role of TLPQC loci in extreme heat and desiccation resistance within Gram-negative bacteria important to human health and food safety. In addition, the complete sequence of an IncHI2 R478-like plasmid isolated at the dawn of the era of therapeutic antibiotic use will aid in furthering our understanding of the evolution of resistance plasmids.

MATERIALS AND METHODS

DNA isolation and genome sequencing.

S. Senftenberg ATCC 43845 was acquired from ATCC and cultured statically at 37°C in Trypticase soy broth (Becton, Dickinson, Franklin Lakes, NJ) for 18 to 20 h. DNA was purified using Qiagen genomic-tip 100/G columns and the DNeasy blood and tissue DNA isolation kit (Qiagen, Valencia, CA) using the manufacturer’s recommended protocol. Sequencing libraries were prepared following the recommended Pacific Bioscience protocols. Single molecule real-time sequencing (SMRT) was performed using P6/C4 chemistry on a Pacific Bioscience (PacBio) RS II instrument (Pacific Biosciences, Menlo Park CA), resulting in average subreads of >7 kbp and mean coverage of 172.59×. The HGAP3 protocol in smrtanalysis v2.3 was used to assemble sequence reads and polish the contig. To circularize the contigs, a self/self dot plot of the contig sequences was generated in Geneious 9.1.5 (Biomatters Ltd., New Zealand) and used to identify duplicated sequence at the contig ends. The duplicated sequence was trimmed to generate a circularized sequence (36). OriFinder was used to determine the origin of replication and to reset base position 1 of the chromosome. The position of circularization was confirmed by mapping all the reads to the renumbered contig using the resequencing protocol in smrtanalysis v2.3, which also provided a second round of polishing via the included quiver routine. Genome and plasmid sequence data were annotated using the NCBI Prokaryotic Genome Annotation Pipeline and deposited into GenBank (GenBank accession numbers CP016837 and CP016838, respectively).

Genome analysis.

The sequences of the thermoresistance cluster originally identified in C. sakazakii (GenBank accession number FR714908) (1) and the thermotolerance loci in S. Senftenberg were used to identify related islands in the GenBank nr database using BLASTn. Genomes with strong similarity to these sequences were identified from 20 genera, and genome sequences were used to extract 91 additional thermotolerance islands by manual inspection to determine the island boundaries (Table 2). Sequence inspection and manipulation were performed in Geneious vR9.1.5. Manual inspection included BLAST comparisons of flanking mobile element genes, which revealed little homology between the different TLPQCs so these flanking sequences were trimmed and discarded to define the boundaries of the TLPQC. The extracted TLPQCs were aligned by MAFFT (37) in Geneious 9.1.5, and the resulting aligned sequences were analyzed by jModelTest 2 to determine the parameters for MrBayes (36, 38, 39). The best model for the full-aligned sequence of the TLPQC was determined to be TRN+I+G (Tamura-Nei model with invariant sites and discrete gamma distribution) using the Akaike information criterion (40). MrBayes 3.2.6, utilized as a Geneious plug-in, was used to calculate a Bayesian inference tree of the various TLPQCs. As the TRN+I+G model is not possible in MrBayes 3.2.6, the GTR+I+G model (generalized time reversible model with invariant sites and discrete gamma distribution) was used, as overparameterizeration gives substantially less bias than underparameterizeration (41), and 3,000,000 iterations were performed with these defined parameters. The first 25% of the iterations were discarded as burn-in, as the Markov chains have not reached stationary and may alter the final result. Visualization of the tracer output and examining the standard deviation of the split frequencies were used to assess the quality of the tree. As the standard deviation of the split frequencies reached only <0.04, the chains have truly not converged but were acceptable, given the extensive computing resources to run the calculations. Additional maximum likelihood trees generated by PhyML (42) gave further support for the Bayesian consensus tree (not shown). Representatives of TLPQCs derived from the Bayesian phylogenetic tree were extracted and visualized in an Easyfig BLASTn comparison (43). CGView Server was used to visualize chromosome and plasmid sequence data with default tBLASTx settings (44).

Base modification analysis.

Genome analysis of nucleotide base modifications were detected using the RS_Modification Motif_Analysis.1 protocol in smrtanalysis v2.3 (Pacific Biosciences, Menlo Park, CA) with the default threshold quality value (QV) of 30. The resulting motif_summary.csv files were uploaded to the restriction enzyme database (REBASE) and assessed for the presence of novel restriction modification systems (30). Additionally, kinetic Qmod values [defined by the log-transformed P value from the t test, −10log(P value)] with a threshold value of 55 were extracted and processed by BaseModFunctions v2.1.R (45) for visualization in Circos (46). The threshold value of 55 was based on the bimodal distribution of modification values observed in the kinetic detection map. Base modification data have been uploaded to GenBank.

Accession number(s).

Accession numbers for the strains used in this study can be found in Table 2.