Whole genome sequencing identifies circulating Beijing-lineage Mycobacterium tuberculosis strains in Guatemala and an associated urban outbreak.

Limited data are available regarding the molecular epidemiology of Mycobacterium tuberculosis (Mtb) strains circulating in Guatemala. Beijing-lineage Mtb strains have gained prevalence worldwide and are associated with increased virulence and drug resistance, but there have been only a few cases reported in Central America. Here we report the first whole genome sequencing of Central American Beijing-lineage strains of Mtb. We find that multiple Beijing-lineage strains, derived from independent founding events, are currently circulating in Guatemala, but overall still represent a relatively small proportion of disease burden. Finally, we identify a specific Beijing-lineage outbreak centered on a poor neighborhood in Guatemala City.

1. Introduction

Tuberculosis (TB) poses a major challenge to global health. The causative agent, Mycobacterium tuberculosis (Mtb), was responsible for around 1.5 million deaths in 2013 [1]. The seven major lineages of Mtb have historically been associated with specific geographical regions, and substantial evidence exists that genetic differences between lineages influence disease presentation and outcome [2-6]. Beijing lineage strains, also known as Lineage 2 strains, have emerged as important drivers of global Mtb burden [7-9]. Notably, outside of East Asia, “modern” Beijing strains account for the majority of this burden [10]. These “modern” Beijing strains display elevated rates of drug-resistance, rapid progression of disease and increased transmission [11-16]. While there is still debate as to precisely when the Beijing lineage originated, it appears to have arisen in East Asia and, through successive waves of human migration, spread throughout this region and beyond [10,17]. Positive selection on virulence-associated genes as well as compensatory mutations negating the fitness-cost of drug-resistance are key factors in making Beijing the most successful contemporary lineage [17-20].

The emergence of “modern” Beijing strains to new regions has been particularly pronounced in Eastern Europe and Africa, regions in which Euro-American strains are also present [21-23]. South Africa serves as one striking example, where data suggest that Beijing strains have continually expanded since their arrival to account for an increasing proportion of total burden [20]. Beijing strains in South Africa have developed resistance to standard drug therapies and contributed to poor outcomes for patients [24,25].

In the Western Hemisphere, the emergence of Beijing strains has thus far been more complex. Beijing strains have spread across the United States, with notable outbreaks occurring in New York City, NY, and Houston, TX [26-28]. In contrast, South and Central America have heretofore been thought to be relatively isolated from this trend, and Euro-American/Lineage 4 strains predominate [29-36]. Current evidence suggests Beijing strains have failed to gain a foothold in South America outside of Peru, where a substantial number of Chinese and Japanese immigrants settled throughout the 19th and 20th centuries [30]. However, there is very little information on Mtb strain diversity in Central America. Thus far, the sparse evidence that does exist indicates that Beijing strains contribute only minimally to the burden of tuberculosis in the region [17,37].

Here we investigate the status of Beijing-lineage strains in Guatemala, the most populous country in Central America. We take a whole genome sequencing (WGS) approach to assess recent TB cases at an HIV clinic and associated public hospital. We combine our results with global genome sequences as well as epidemiologic patient data to assess patterns of Beijing-lineage Mtb in Guatemala.

2. Materials and methods

2.1. Patient population

The Clínica Familiar “Luis Angel García” (CFLAG) is an HIV-specialized clinic located at the Hospital General San Juan de Dios, one of Guatemala's two public teaching hospitals in Guatemala City. It provides comprehensive treatment and follow-up for people living with HIV for both outpatient and inpatient care and has provided medical care for more than 10,000 patients over the last thirty years. From 2010 to 2014, routine spoligotyping was performed at CFLAG on all Mtb isolates from CFLAG and the associated hospital during that period, derived from a total of 514 independent patients. Spoligotyping identified 11 Beijing-lineage strains; five of the 11 strains could be regrown for WGS.

2.2. DNA extraction and genotyping

Bacterial culturing, spoligotyping, and DNA extraction were performed according to standard practices (detailed in Supplemental Material) [38,39].

2.3. Sequencing, alignment and SNP calling

Using Illumina HiSeq 2500 and Illumina MiSeq platforms, we sequenced each strain at >500-fold coverage with a minimum read length of 50 base pairs. Additional sequence reads acquired from published WGS datasets also had minimum read lengths of 50 base pairs. We downloaded reads from the National Center for Biotechnology Information Sequence Read Archive and the European Nucleotide Archive, as described in [2,3,17,40]. We collected sequence reads for 24 globally extant strains of Mtb representing the seven major lineages [2,3,41], as well as 96 globally extant Beijing strains [17]. We aligned reads against the H37Rv reference genome (GenBank: AL123456.3) using BWA [42]. We called variants using SAMtools and filtered with VarScan for a minimum read depth of 10, a consensus quality score of 20, and a minimum variant frequency of 0.75 [43,44]. We discarded SNPs adjacent to indels and within repetitive regions of the genome. Additionally, as is standard practice, we discarded SNPs affecting genes commonly associated with drug-resistance to avoid homoplastic mutations among distantly related strains [10,17,45]. We visualized variants among the Guatemalan isolates associated with the outbreak with Circular Visualization for Microbial Genomes (URL: http://civi.cmbi.ru.nl/).

2.4. Nucleotide sequence accession numbers

The sequences determined in this study have been deposited in the NCBI Sequence Read Archive under accession numbers: GG-135-10 – , GG-120-10 – , GG-152-12 – , GG-131-11 – , GG-219-11 – , and GG-30-13 – . GG-219-11 and GG-30-13 were sequential isolates taken from the same patient 17 months apart.

2.5. Tree builds

All trees were based on genome-wide SNPs derived according to the parameters specified above. We constructed a superset of SNPs for each strain with reference alleles occupying sites for which no variants were detected using custom Perl scripts. These SNPs informed neighbor-joining and maximum-likelihood methods of phylogeny construction. We implemented neighbor-joining methodology with ClustalW2 using pairwise similarity scores of SNP supersets as a measure of genetic distance [46]. We generated a maximum-likelihood phylogeny with RAxML using a GTR model of nucleotide substitution [47]. For each method, 1000 bootstrap replicates provided support for nodes on the tree. The phylogenies derived from each method were congruent (compare Figure 2 to Figure S1). Trees were visualized with FigTree (see URL tree.bio.ed.ac.uk/software/figgtree).

Figure 2

Neighbor-joining phylogeny based on 6584 SNPs among 101 global Beijing isolates of Mtb. Guatemalan isolates are indicated in orange. Scale bar indicates substitutions per site.

3. Results

3.1. Spoligotyping of Mtb strains in Guatemala city

Euro-American strains of Mtb have historically been thought to prevail in Central America, with very little contribution from Beijing strains. In a collection of 514 patient isolates collected from 2010 to 2014 in by the Hospital San Juan de Dios and CFLAG, we identified 11 (2.1%) with Beijing-lineage spoligotypes. Isolates during this time period came from both HIV-positive and HIV-negative patients. Patients with Beijing-lineage strains included individuals living in the capital, Guatemala City, as well as two patients from rural departments (Table 1).

Table 1

Spoligotypes and patient risk factors for the 11 identified Beijing isolates from the Clínica Familiar “Luis Angel García” (CFLAG) and its associated public hospital.

Isolate	Spoligotype	Predicted lineage	WGS performed	Lineage by WGS	Risk factors	Residence	Year of Isolation
GG-120-10	000000000003770	Beijing	Yes	Beijing	HIV positive	Urban	2010
GG-135-10	000000000003770	Beijing	Yes	Beijing	HIV positive	Rural Department	2010
GG-22-10	000000000003771	Beijing	No	–	HIV negative	Urban	2010
GG-206-10	000000000003771	Beijing	No	–	HIV negative	Urban, near local jail	2010
GG-32-10	000000000003771	Beijing	No	–	HIV status unknown	Urban, near local jail	2010
GG-131-11	000000000003771	Beijing	Yes	Beijing	HIV positive	Urban, near local jail	2011
GG-219-11	000000000003770	Beijing	Yes	Beijing	HIV positive; history of incarceration	History of incarceration in local jail	2011
GG-152-12	000000000003770	Beijing	Yes	Beijing	HIV positive; history of incarceration	History of incarceration in local jail	2012
GG-10-14	000000000003771	Beijing	No	–	HIV positive; history of incarceration	Urban; history of incarceration in local jail	2014
GG-107-14	000000000003771	Beijing	No	–	HIV negative	Rural Department	2014
GG-109-14	000000000003771	Beijing	No	–	HIV negative	Urban	2014

Because no Beijing-lineage cases had been sequenced from Guatemala, we sought to further investigate the identity of these strains by WGS. Beyond identifying strain background, the resolution of WGS allows analysis of microevolution among closely related isolates, and offers insight into the transmission history of related strains [48]. We attempted to culture isolates retrospectively from all 11 patients, but only five of these isolates could be regrown in sufficient quantities for WGS (Table 1).

3.2. Phylogeny of Guatemalan Beijing-lineage strains

We used a sample of 24 globally extant strains of Mtb, representative of the seven major lineages, to initially place the Guatemalan Beijing strains in a broad phylogenetic context [3,40]. Neighbor-joining and maximum-likelihood methods of phylogeny construction based on 18,039 genome-wide SNPs resolved all 29 strains into the seven principal lineages of Mtb, and placed the Guatemalan isolates among known Beijing strains (see Methods) (Figure 1). To independently confirm the identity of these strains as Beijing-lineage, we searched for previously identified Beijing-specific variants. All five Guatemalan isolates harbored the Beijing lineage ACG- > AGG SNP in codon 20 of Rv2450c and the GGG-> AGG SNP in codon 176 of Rv2952 (Supplemental Table 1), con-firming their status as Beijing strains [41].

Figure 1

Neighbor-joining phylogeny based on 18,039 SNPs among 29 global strains of Mtb. The tree is rooted by the outgroup M. canetti. Branches are colored by lineage: Pink – Lineage 1/Indo-Oceanic; Blue – Lineage 2/East-Asian; Purple – Lineage 3/East-African-Indian; Red – Lineage 4/Euro-American; Brown – Lineage 5/West-Africa I; Green – Lineage 6/West Africa II; Yellow – Lineage 7/Ethiopia. Guatemalan isolates are marked with stars. Scale bar indicates substitutions per site.

To understand the relationship of the Guatemalan Beijing strains to other globally extant Beijing strains, we generated phylogenetic trees based on 6584 SNPs among a total of 101 Beijing strains (see Methods) [17]. The Guatemalan strains were distributed on three independent branches in all constructed trees (Figure 2 and Figure S1). Three isolates (GG-131-11, GG-219-11, and GG-152-12) clustered on a single branch of the tree near strains previously identified in China and Thailand. A fourth isolate, GG-120-10, fell on an independent branch of the phylogenetic tree, and is most closely related to strains identified in Nepal. Finally, one of the Guatemalan patient isolates, GG-135-10, appeared to represent a more basal sublineage, and clustered with the so-called “ancient” Beijing strains. Notably, none of the isolates came from individuals who reported travel outside of Guatemala.

We next assessed by SNP analysis whether the phylogenetic distance was consistent with multiple introductions of Beijing strains into the country. Approximately 500 SNPs separated GG-135-10 from the four other Guatemalan isolates, and GG-120-10 differed from GG-131-11, GG-219-11, and GG-152-12 by ~200 SNPs (Supplemental Table 1). The last three isolates were highly similar, differing by at most 31 SNPs, suggestive of a recently shared progenitor strain (detailed further below).

Phylogenetic analysis indicated GG-120-10, GG-131-11, GG-219-11 and GG-152-12 are evolutionary “modern” Beijing strains, while GG-135-10 is of an ancestral Beijing sublineage. Thus, we assessed each Guatemalan isolate for the insertion variant T1406760 + G in Rv1258c, as this variant is known to be a marker for the “modern” Beijing genotype [49]. As expected, four isolates displayed this polymorphism: GG-120-10, GG-131-11, GG-219-11, and GG-152-12. As additional confirmation, the four “modern” Beijing strains also harbored a GGA- > CGA mutation at codon position 58 in the DNA repair enzyme gene mutT2, yet GG-135-10 did not [8,50]. Together, these data reveal that both evolutionarily “modern” and “ancient” Beijing strains are present in Guatemala.

3.3. Epidemiologic analysis of a Beijing-lineage outbreak in Guatemala city

We noted that three Guatemalan Beijing-lineage isolates (GG-131-11, GG-219-11, and GG-152-12) demonstrated a tight clustering on one branch of the tree, indicating transmission of a single founder strain (Figure 2). 125 SNPs separate this clade from its nearest neighbor on the tree (8662-05) (Figure 3). Of these, 94 SNPs are shared among all three strains. 18 SNPs are unique to GG-131-11, and 13 unique to GG-219-11 and GG-152-12. The patients from whom GG-219-11 and GG-152-12 were isolated had spent time within the same city jail, but their incarceration periods had not overlapped. Isolates from these two patients were the most closely related strains we identified, varying from each other by only 1 SNP. One variant shared between these two strains includes a GAC- > AAC mutation in codon 94 of the DNA gyrase gene gyrA, a mutation reported to associate with fluoroquinolone-resistance [51,52].

Figure 3

Circular diagram of polymorphisms unique to the Guatemalan micro-outbreak strains plotted on the H37Rv reference genome. The outer wheel displays base pair coordinates. The second and third wheels represent positive- and negative-strand genes, respectively. Circles four through seven display genome positions of the 125 polymorphisms that separate the Guatemalan outbreak isolates from their nearest neighbor (8662-05) on the tree. The purple shaded wheel with yellow dots displays the 94 SNPs shared among all three isolates. The blue shaded wheel with orange dots displays the 18 SNPs specifically found in GG-131–11 (star denotes presence of two SNPs that are too close to resolve on the diagram). The red wheels display SNPs found in the jail-associated isolates GG-219-11 (outer, blue dots) and GG-152-12 (inner, green dots). The innermost circle represents GC-skew.

The closely related strain GG-131-11 differed from the jail isolates by 31 SNPs, and did not contain the gyrA variant. This patient lived in the same neighborhood as the jail where the patients had resided, but the third patient had no other known connection other than residence in the vicinity of the jail. Among the six strains identified as Beijing, but for which WGS could not be performed, two of them derived from HIV-negative patients who also lived in the same neighborhood as the jail (Table 1). In total, six of the 11 identified Beijing-lineage cases were centered on this same neighborhood.

Finally, the HIV-positive patient (Table 1) from whom strain GG-219-11 was isolated abandoned treatment completely and then returned to the clinic 17 months later, at which time a second isolate (GG-30-13) was obtained. A comparison of the sequential isolates revealed a single variant between the initial isolate and that taken after 17 months of untreated disease. These data reflect a relatively low mutation rate for this particular Beijing-lineage strain in the absence of antibiotic therapies and in the context of a compromised immune system.

4. Discussion

In this study, we use WGS to investigate Beijing Mtb strains circulating in Guatemala. The Beijing lineage, which has emerged globally, is considered generally more virulent than strains of other lineages, and has a higher frequency of drug-resistance [11-16]. Prior to this study, information on Beijing isolates in Guatemala has been sparse [5]. Here, we have for the first time conducted WGS on Beijing isolates from Central America for an in-depth characterization of the potential regional emergence of these strains.

We infer from WGS and phylogenetic analysis that multiple Beijing strains have been introduced into Guatemala and that these strains have transmitted within Guatemala itself; notably, none of the affected patients has any history of travel outside Guatemala, strongly suggesting in-country circulation of each strain. Patients infected with Beijing-lineage strains included both HIV-negative and HIV-positive individuals. Central and South America have traditionally been thought to harbor almost exclusively Euro-American Mtb strains [29-36]. Here we confirm the presence of circulating Beijing strains in Central America and link one set of strains to an outbreak in an urban setting.

We were able to connect patients through a combination of strain sequences and epidemiology. We identified an outbreak of one Beijing-lineage strain in a poor neighborhood in urban Guatemala centered on a city jail. Two of the cases were patients who had resided in the jail but not during overlapping time periods, suggesting the presence of other latently or undiagnosed actively infected inmates. Indeed, one of the unsequenced but spoligotype-identified Beijing strains (Table 1, strain GG-10-14) was from a patient who had spent time in the same jail in 2008. The Beijing-lineage outbreak strain does not represent the only Mtb strain within the jail. There were eight additional TB cases associated with the jail that, by spoligotyping, were identified as Euro-American/Lineage 4 strains (T1, X1, H2, LAM3, LAM9 families) (D.L.-B., N.M., E.A. and B.S., unpublished results).

Of the 11 Beijing isolates we identified, six were associated with a local jail or the neighborhood directly around it. Although we were unable to identify direct epidemiologic links between the neighborhood and jail-associated strains, the geographical and phylogenetic clustering is striking. Workers in the jail (guards and other staff), exiting inmates, and proximity to public bus lines within the neighborhood are possible links between the community and the jail. Consistent with previous results on outbreak strains of Mtb and rates of divergence, the clustering of these cases as well as substantial nucleotide divergence (31 SNPs) between the neighborhood and jail-associated strains suggests an outbreak that has lasted a number of years and the presence of other unidentified patients in the area [53-56]. The mutation rate of Mtb is estimated to be between 0.3 and 0.5 SNPs per year, and it has been proposed that a maximum of five SNPs separating isolates indicate ongoing transmission [54-56]. However, one study demonstrated an epidemiological link between strains that differed by 14 SNPs, and bursts of mutations have been shown to occur in clinical isolates of Mtb [57,58]. Together, these results suggest the founder strain arrived in Guatemala City and infected a citizen who entered the jail (either an employee or inmate), thereby establishing clonal transmission of this isolate within the jail's population. The strain's entry into the jail likely initiated its divergence from related isolates circulating in the broader community.

In spite of no known anti-tuberculosis therapy for these patients prior to diagnosis, we identified a mutation that arose in the isolates and was associated with resistance to fluoroquinolones [51]. This mutation may have arisen in a progenitor strain due to the use of over-the-counter antibiotics, readily available in Guatemala, or through prescription from another practitioner. Regardless, this finding highlights the issue of emergent fluoroquinolone-resistant Beijing-lineage strains in countries with easy over-the-counter access to antibiotics.

Overall, we describe multiple circulating Beijing strains that have been introduced independently into Guatemala, including an “ancient” Beijing strain. However, the overall contribution to disease burden in Guatemala City still appears relatively low, with 11 Beijing isolates out of 514 reported cases (2.1%), in line with reports from other Central and South American countries [29-37]. Guatemalan immigrants to the United States with active TB show similar rates of infection with Beijing strains (2% of 609 genotyped cases reported 2009–2013, as compared with 15% prevalence of Beijing strains among all TB cases reported in the US in the same period) (Smita Ghosh, Centers for Disease Control and Prevention, personal communication).

The emergence of the Beijing lineage in Guatemala has important implications for public health in Central America, where no WGS on Mtb had been performed previously. More broadly, this work helps increase our understanding of the global spread of the Beijing lineage, which presents particular challenges in combating TB.