Genome diversity of Shigella boydii

Abstract

Shigella boydii is one of the four Shigella species that causes disease worldwide; however, there are few published studies that examine the genomic variation of this species. This study compares genomes of 72 total isolates; 28 S. boydii from Bangladesh and The Gambia that were recently isolated as part of the Global Enteric Multicenter Study (GEMS), 14 historical S. boydii genomes in the public domain and 30 Escherichia coli and Shigella reference genomes that represent the genomic diversity of these pathogens. This comparative analysis of these 72 genomes identified that the S. boydii isolates separate into three phylogenomic clades, each with specific gene content. Each of the clades contains S. boydii isolates from geographic and temporally distant sources, indicating that the S. boydii isolates from the GEMS are representative of S. boydii. This study describes the genome sequences of a collection of novel S. boydii isolates and provides insight into the diversity of this species in comparison to the E. coli and other Shigella species.

This comparative genomic study identifies the diversity of Shigella boydii isolates when compared to reference isolates of closely related pathogens.

Shigella is a Gram-negative pathogen and the cause of shigellosis, a potentially deadly diarrheal disease whose symptoms range from mild intestinal discomfort to death depending on severity (Rasko et al. 2008; Sahl et al. 2015). Each year Shigella species cause 165 million cases of shigellosis with an estimated 1.1 million of those cases resulting in death (Kotloff et al. 2013). For 3 years, the Global Enteric Multicenter Study (GEMS) identified pathogens, such as Shigella and pathogenic Escherichia coli, believed to be a cause of moderate-to-severe diarrhea (MSD) in children aged 0–59 months in the endemic areas of sub-Saharan Africa and South Asia (Kotloff et al. 2013). GEMS was an age stratified, matched case-control study that demonstrated that Shigella were consistently in the top five of all cases of MSD in each of the age groups (Farag et al. 2013).

There are four species of Shigella: Shigella sonnei, S. flexneri, S. dysenteriae and Shigella boydii, each with their own global burdens and epidemiological profile (Livio et al. 2014). From the 1130 Shigella isolates collected during the 36 months of the GEMS, 5.4% (61/1130) were identified as S. boydii (Livio et al. 2014). While this is a proportionally small contribution to the overall observed cases of MSD compared to the other three Shigella species, S. boydii still makes up a significant component of the overall Shigella burden (Baker, Parkhill and Thomson 2015). By increasing the number and diversity of S. boydii genomes available to the scientific community, further functional studies can focused on this understudied and underreported pathogen.

Considering the burden of disease caused by Shigella, there are relatively few Shigella genomic studies (Wei et al. 2003; Yang et al. 2005); however, recent studies on S. sonnei (Holt et al. 2012, 2013) and S. flexneri (Connor et al. 2015) genomics have detailed the temporal and spatial virulence of these species. The 28 S. boydii isolates sequenced in this study represent all of the S. boydii isolates identified at the Bangladesh and The Gambia GEMS sites during the first 24 months. A total of 31 S. boydii isolates were identified at these two sites over the complete GEMS 36 month period (24 in Bangladesh and 7 in The Gambia; Livio et al. 2014), thus we have examined the majority of the isolates from these two sites. These 28 S. boydii genomes from GEMS were compared with 14 S. boydii isolates already in the public domain, labeled in this study as ‘historical isolates’ (Table 1). By presenting these genomes along with corresponding clinical and phylogenomic data, this study will begin to shed light on this pathogen and allow a deeper understanding of the role of this organism in the broader context of global Shigella infections.

Table 1.

Shigella boydii isolates examined.

Isolate			Shigella boydii	Number of	Total	GenBank	Short read
name	Origin	Year	phylogenetic clade	contigs	bp	accession	archive
3083-94	Arizona, USA	1994	2	6	4874659	NC_010658	NDa
SB_3594-74	Colorado, USA	1974	3	96	4634068	AFGC00000000	SRA020641.2
SB_965-58	Minnesota, USA	1958	1	96	5184598	AKNA00000000	SRA020850.2
248-1B	Chile	1995	3	166	4788006	AMKG00000000	NDa
SB_08_0009	British Columbia, Canada	2008	3	165	4864228	AMJZ00000000	NDa
SB_08_0280	Ontario, Canada	2008	2	124	4835559	AMKA00000000	NDa
SB_08_2671	Manitoba, Canada	2008	3	185	4817878	AMKB00000000	NDa
SB_08_2675	Alberta, Canada	2008	2	335	4832830	AMKC00000000	NDa
SB_08_6341	Ontario, Canada	2008	2	138	4800746	AMKD00000000	NDa
SB_09_0344	British Columbia, Canada	2008	2	174	4821210	AMKE00000000	NDa
SB_4444-74	Idaho, USA	1974	3	314	4976495	AKNB00000000	SRS270182
SB_5216-82	Bulgaria	1963	1	75	4882454	AFGE00000000	SRA020642.2
SB_S6614	Kenya	2005	3	479	4610666	AMJU00000000	NDa
SB_S7334	Kenya	2007	3	249	4711626	AMJX00000000	NDa
100705	The Gambia	2009	3	404	4361489	LSCP00000000	SRP072004
100706	The Gambia	2008	1	336	4475997	LPSY00000000	SRP072003
102252	The Gambia	2008	2	455	4380029	LPSX00000000	SRP072001
102265	The Gambia	2009	3	468	4547444	LPSW00000000	SRP072000
102309	The Gambia	2009	3	429	4384003	LPSV00000000	SRP072009
600080	Bangladesh	2008	1	358	4263951	LSCB00000000	SRP071936
600266	Bangladesh	2008	2	801	4158234	LPTT00000000	SRP071943
600375	Bangladesh	2008	2	473	4386200	LPTS00000000	SRP071984
600384	Bangladesh	2008	1	305	4367268	LPTR00000000	SRP071983
600657	Bangladesh	2008	2	467	4322758	LSCA00000000	SRP071982
600690	Bangladesh	2008	1	300	4742925	LPTQ00000000	SRP071994
600710	Bangladesh	2008	2	463	4333602	LPTP00000000	SRP071993
600746	Bangladesh	2009	1	386	4539801	LPTO00000000	SRP071992
601143	Bangladesh	2009	2	461	4291898	LPTN00000000	SRP071991
601276	Bangladesh	2009	2	472	4318093	LPTM00000000	SRP071985
601294	Bangladesh	2009	3	442	4353514	LPTL00000000	SRP071989
602068	Bangladesh	2009	3	432	4354983	LPTK00000000	SRP071988
602144	Bangladesh	2009	2	689	4246468	LPTJ00000000	SRP071986
602339_II	Bangladesh	2009	3	413	4474771	LPTI00000000	SRP071999
602385	Bangladesh	2009	2	553	4292491	LPTH00000000	SRP071998
602404	Bangladesh	2009	2	471	4297951	LPTG00000000	SRP071997
602573	Bangladesh	2009	3	545	4500753	LPTF00000000	SRP071996
602682	Bangladesh	2009	3	414	4484949	LPTE00000000	SRP071995
602988	Bangladesh	2009	2	897	4165605	LPTD00000000	SRP072008
603122	Bangladesh	2009	3	925	4329955	LPTC00000000	SRP072007
603150	Bangladesh	2009	3	416	4516839	LPTB00000000	SRP072006
603210	Bangladesh	2009	3	542	4474419	LPTA00000000	SRP072005
603233	Bangladesh	2009	1	495	4758784	LPSZ00000000	SRP072002

^aPrimary sequence was not deposited in the SRA.

In total, 28 S. boydii genomes were identified for sequencing and analysis from the 727 Shigella isolates obtained from Bangladesh and the Gambia during the GEMS (Kotloff et al. 2013; Livio et al. 2014). Multiple studies on the GEMS including the interpretation of the clinical, epidemiologic and microbial findings have been published (Farag et al. 2013; Lindsay et al. 2013; Baker, Parkhill and Thomson 2015; Sahl et al. 2015), and the genomic studies are now underway. In the current study, we have included 14 S. boydii genomes that had been published previously (Rasko et al. 2008; Sahl et al. 2015), including the first S. boydii genome that was in the public domain (strain BS512, (aka CDC 3083–94); serotype 18, Assembly Accession number NC_010658). For all GEMS isolates, genomic DNA was prepared as previously described with established methods (Sahl et al. 2011) taking precautions to minimize the passage number.

The genome sequence of each GEMS isolate was generated at the Institute for Genome Sciences, Genome Resource Center on the Illumina HiSeq2000 using paired-end libraries with 300 bp inserts. The draft genomes were assembled using Minimus (Sommer et al. 2007) to merge contigs generated using two different assemblers, Velvet assembly program (Zerbino and Birney 2008) (with kmer values determined using VelvetOptimiser v2.1.4 (http://bioinformatics.net.au/software.velvetoptimiser.shtml)), and the Edena v3 assembler (Hernandez et al. 2008). The metrics for the resulting assemblies corresponding GenBank accession and SRA numbers are presented in Table 1. For the S. boydi 3083–94, genomic DNA for sequencing was isolated from a stock culture and two Sanger sequencing libraries were constructed —a small insert library (4–5 kb) and a large insert library (10–12 kb) from which 20 656 and 47 348 reads were sequenced, respectively. The CDC 3083–94 genome was assembled as previously described (Rasko et al. 2008).

The 72 E. coli and Shigella genomes were aligned using Mugsy (Angiuoli and Salzberg 2011), and homologous blocks were concatenated using the bx-python toolkit (https://bitbucket.org/james_taylor/bx-python). The columns that contained one or more gaps were removed using Mothur (Schloss et al. 2009). These concatenated regions from each genome were used to construct a maximum-likelihood phylogeny with 100 bootstrap replicates using RAxML v7.2.8 (Stamatakis 2006) and visualized using FigTree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/) (Fig. 1). Additionally, the level of similarity of protein-encoding genes was compared between the 42 S. boydii genomes in this study using a large-scale BLAST score ratio (LS-BSR) analysis as previously described (Hazen et al. 2013; Sahl et al. 2013, 2014). Protein-encoding genes were predicted for each genome sequence using Prodigal (Hyatt et al. 2010). The genes were then combined into gene clusters using uclust (Edgar 2010), and the gene clusters were assigned using a stringent nucleotide identity threshold of ≥90%. The protein-encoding genes that were considered present, with significant similarity had BSR values ≥0.8, while those with BSR values <0.8 but ≥0.4 were considered to be present but divergent and <0.4 were considered absent. The predicted protein function of each gene cluster was determined using an ergatis-based (Orvis et al. 2010) in-house annotation pipeline (Galens et al. 2011).

Figure 1.

Phylogenomic tree containing GEMS S. boydii isolates (white circles), S. boydii isolates that are currently in public databases (gray circles) and a collection of reference E. coli and Shigella species genomes (black circles). Three clades of S. boydii isolates are identified by color with clade 1 in blue (broken into two smaller subclades as denoted by the shades of blue), clade 2 in red and clade 3 in green. The tree was inferred with Figtree 1.4.2 with Bootstrap support values from 100 replicates are shown at the black square nodes. Distance for the number of nucleotide changes is shown to be 0.0040 with corresponding bar length.

The 42 S. boydii genomes from the 28 GEMS isolates and 14 historical isolates represent a collection of isolates that are geographically and temporally distributed (Table 1). The average genome size of the 28 GEMS S. boydii isolates examined is 4397 328 bp (range 4158 234–4758 784 bp). The average number of contigs in this collection is 493 (range 300–801 contigs), and the average GC percent is 50.75% (range 50.36%–51.19%). These values are typical of previously generated Shigella species genomes (Holt et al. 2012, 2013; Sangal et al. 2013; Connor et al. 2015; Sahl et al. 2015).

The phylogenomic analysis of the genomes was completed using the Mugsy algorithm (Angiuoli and Salzberg 2011). The alignment utilized in this reference-independent comparison contains ∼3.0 Mb which was a greater amount of genomic content than previous Shigella comparisons with this method (Sahl et al. 2015), suggesting a high level of conservation within the isolates of this species. The inferred phylogeny also identifies the S. boydii genomes as being separated from any of the E. coli reference genomes (Fig. 1). This E. coli and Shigella separation has been identified previously (Rasko et al. 2008; Tenaillon et al. 2010; Sahl et al. 2015). Additionally, phylogenomic analysis demonstrated that the S. boydii are separated into three clades (labeled clades 1, 2 and 3 in Fig. 1), with clade 1 potentially able to be further subdivided into two additional subgroups (Fig. 1). This data suggest that S. boydii clade 1 potentially diverged from clades 2 and 3 at an earlier point in the development of the S. boydii species (Fig. 1). This pattern is similarly observed in a global Shigella analyses recently published by our group (Sahl et al. 2015); however, the number of S. boydii isolates in that analyses were limited compared to our current study.

The comparisons included in this study are the greatest number of S. boydii compared in any one study, 42 isolates. Interestingly, the S. boydii isolates do not segregate by any specific geographic location or date of isolation, suggesting that this study has captured the genomic diversity of this species of Shigella. Both spatially and temporally, the new isolates from GEMS are distributed alongside the historic S. boydii isolates and are distributed between the three S. boydii clades (Table 1). This indicates that this GEMS collection of isolates has captured the diversity of the S. boydii species.

Analysis of the gene content via LS-BSR (Sahl et al. 2014) identified total of 7355 gene clusters in the 28 GEMS Shigella genomes. Among those gene clusters, a core S. boydii genome of 2477 gene clusters that are present in all S. boydii genomes examined. Comparing the gene clusters of the 28 GEMS genomes to the 14 previously sequenced S. boydii genomes identifies a core genome of 2230 genes that were present with significant similarity in the 42 of the S. boydii genomes in this study.

When the phylogenomic data from Fig. 1 is combined with the LS-BSR data, protein-encoding genes that are clade specific were identified. Annotation of these specific regions also provided potential insight into the gene function for the unique genes in these three clades. Unique genes are present in all of the isolates in the clade of interest, but lacking in the isolates from the other clades. Shigella boydii clade 1 had 98 unique genes compared to S. boydii clades 2 and 3, which had only 4 and 12 unique genes, respectively (Table 2; Table S2, Supporting Information). The clade-specific S. boydii genes included inner membrane components for a transport system and zinc-binding proteins from clade 1, several phage component proteins from clade 2 and two integrase family proteins from clade 3 (Table S2, Supporting Information). There were also several hypothetical proteins that were identified as clade specific: 13 from clade 1, 4 from clade 2 and zero from clade 3. Why and how these unique genes arose solely in S. boydii clade one creates another reason S. boydii requires further functional analysis.

Table 2.

Gene prevalence in S. boydii clades.

		LS-BSR
		≥0.8a
		Unique gene clusters
Clades	Total isolates	Uniqueb	Prevalentb
Clade 1	8	98	128
Clade 2	18	12	38
Clade 3	16	4	56
Total	42	114	209

^aLS-BSR cutoff for gene clusters was greater or equal to 0.8 in selected clade and less than or equal to 0.4 in the other three clades

^bUnique clusters have 100% similarity in one clade and 0% in the other two clades: Prevalent gene clusters are present in >90% of one clade and present in <20% of the other two clades.

If the criteria for clade specificity are broadened to identify prevalent genes (i.e. genes present in >90% of isolates of one clade but in <20% of the isolates of the other two clades), these numbers increases to 128 genes in S. boydii clade 1, 56 genes in S. boydii clade 2 and 38 genes in S. boydii clade 3 (Table 2; Table S2, Supporting Information). The relatively larger increase observed in S. boydii clades 2 and 3 suggests that there is more overlap within the gene content of these clades when compared to the content of clade 1. This further suggests a distinct separation of the members of S. boydii clade 1 from other S. boydii. This increased gene repertoire contains transmembrane proteins in clade 1, phage-associated proteins in clade 2 (tail subunits, head-tail connectors and phage portal proteins) and a collection of phage and metabolism proteins in clade 3 (Table S2, Supporting Information). Similar approaches to the ones used have been utilized in the past to identify common features that are in use as PCR-based diagnostics for Shigella (Sahl et al. 2015) and E. coli (Hazen et al. 2013; Sahl et al. 2013, 2014).

Shigella is a pervasive human pathogen that causes life-threatening disease. Genomics of understudied pathogens like S. boydii will allow continued development in the fields of pathogen identification, phylogenetic categorization and potential functional characterization of the identified clade- and species-specific genomic regions.

SUPPLEMENTARY DATA

Supplementary Data.

FUNDING

This project was funded in part by federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services under contract number HHSN272200900009C, grant number 1U19AI090873 and startup funds from the State of Maryland.

Conflict of interest. None declared.