Molecular Typing of “Candidatus Bartonella ancashi,” a New Human Pathogen Causing Verruga Peruana

The genus Bartonella consists of 29 recognized species and three subspecies of hemotropic Gram-negative bacteria that infect a wide range of mammals by transmission through various arthropod vectors, such as human body lice, fleas, and sandflies. Among them, Bartonella henselae, Bartonella quintana, and Bartonella bacilliformis are established as pathogens of human importance, causing cat-scratch fever, trench fever, and Carrion's disease, respectively, while an increasing number of other Bartonella species have recently been found to be infectious to humans, particularly immunocompromised persons (1, 2). Additionally, B. bacilliformis is transmitted by Lutzomyia sandflies and causes a biphasic syndrome (Carrion's disease) consisting of an acute phase, Oroya fever, and a chronic phase known as verruga peruana that is characterized by benign and persistent red-purple raised skin nodules (3). B. bacilliformis infections are seen only in the Andes mountain region of Peru, Ecuador, and Columbia (2,500 to 8,000 feet above sea level) (3, 4). Although B. bacilliformis was the first Bartonella species discovered, over 100 years ago, and is a pathogen causing significant morbidity and mortality, our understanding of this pathogen and the causes of Carrion's disease are rather limited (2, 4–6). Currently, B. bacilliformis is the only agent definitively identified to cause Carrion's disease, though Bartonella rochalimae caused an Oroya fever-like illness in a traveler returning from Peru (2, 4, 6). Here, we further characterized a new Bartonella pathogen causing verruga peruana (7) by using multilocus sequence typing (MLST) and multispacer sequence typing (MST).

The agent was initially isolated from a 3-year-old boy with verruga peruana living in Caraz, Peru (located in the Andes mountain range of Peru), who had been enrolled in a clinical trial to compare the efficacies of azithromycin and rifampin for the treatment of verruga peruana caused by B. bacilliformis. The child had 56 lesions, distributed mainly on the extremities, but did not have fevers, arthralgias, or malaise. He had no evidence of anemia at presentation, and his Giemsa blood smear was negative for intracellular organisms. Blood samples taken from the patient (clinical trial subject 20) were culture positive for Bartonella at the time of enrollment (20.00) and again at day 60 (20.60). Peripheral blood thin smears were prepared at the Ministry of Health (MOH) Hospital and later confirmed at the reference laboratory in Lima. Blood samples for culture were collected in sodium citrate tubes and transported to Lima, where they were cultured in sealed flasks, using a modified F-1 medium with liquid overlay of RPMI 1640 medium with 10% fetal bovine serum, and then were observed for 8 weeks at 28°C without additional CO₂. Confirmation of bacterial isolates as B. bacilliformis was done by PCR using primers for the citrate synthase gene (gltA).

The patient was treated with azithromycin on day 0 and day 7, which typically results in effective tissue levels of antibiotic for 2 weeks. His rash resolved fully, but he had another positive blood culture at day 60.

Pure bacterial isolates (from culture) were initially subjected to microbiology testing and PCR and Sanger sequencing of a 338-bp gltA fragment for identification of B. bacilliformis. The gltA sequences from both isolates were identical to each other but were found to be significantly different from those of B. bacilliformis and other known Bartonella species. Fragments of rrs (1,424 bp) and rpoB (RNA polymerase beta-subunit) (589 bp) were sequenced, and a BLAST query against the NCBI nonredundant DNA database (nr/nt) confirmed the sequence dissimilarity, indicating a probable emerging Bartonella human pathogen that was subsequently provisionally named “Candidatus Bartonella ancashi” (7).

To more fully characterize “Candidatus Bartonella ancashi,” the initial isolate (20.00) and the second isolate (20.60) were subjected to gene sequencing, MLST, and MST. MLST and MST techniques are used in the differentiation of Bartonella species and strains (5, 8, 9). MLST is useful for the detection of novel Bartonella species, while MST is useful for uncovering genetic differences between strains (5, 8, 9).

DNA extracted from both clinical isolates was subjected to PCR amplification and Sanger sequencing as previously described (Table 1) (1, 7, 10–14). Furthermore, the genomes of both isolates were fully sequenced and assembled using a Roche GS FLX Titanium sequencing system and assembly software GSAssembler, version 2.5.3 (Roche 454 Life Sciences, Branford, CT). Complete sequences for five housekeeping genes, gltA (1,341 bp), rpoB (4,149 bp), ftsZ (1,776 bp), groEL (1,644 bp), and ribC (riboflavin synthase, 642 bp), and for rrs (1,474 bp) and the 16S-23S intergenic spacer (ITS) (940 bp) were extracted and submitted to GenBank (ribC [KC886734], ftsZ [KC886735], gltA [KC886736], groEL [KC886737], rpoB [KC886738], rrs [KC886739]], and ITS [KC886740]). The sequences from both Sanger and 454 pyrosequencing are 100% identical to each other. Additionally, the gene sequences of “Candidatus Bartonella ancashi” isolates 20.00 and 20.60 were found to be 100% identical to one another at these loci. The full gene sequences of isolate 20.60 were then used for molecular typing, including MLST and MST. For these isolates, the MLST analysis included the concatenation of rrs (16S rRNA genes), gltA, rpoB, ftsZ (cell division protein), groEL (60-kDa heat shock protein), and ribC (riboflavin synthase) for phylogenetic assessment, while the MST analysis was conducted using the ITS region.

The nucleotide identities between the five housekeeping genes and rrs of “Candidatus Bartonella ancashi” and known Bartonella species are summarized in Table 2. The results show 90% or less sequence similarity between all the housekeeping genes from “Candidatus Bartonella ancashi” and those of known Bartonella species, while the rrs identity was 99% or less between the isolate and known Bartonella species. MLST phylogeny was assessed using concatenated sequences for these 6 loci of 21 Bartonella type strains and “Candidatus Bartonella ancashi” (Fig. 1). The analysis indicates that “Candidatus Bartonella ancashi” is a member of the Bartonella genus and most closely related to B. bacilliformis (Fig. 1). To further explore the extent of their relatedness, complete ITS sequences for “Candidatus Bartonella ancashi,” B. bacilliformis isolates, B. rochalimae, and Bartonella clarridgeiae were aligned for the MST analysis (Fig. 2) (1). ITS sequences are highly variable between strains of the same species and are able to provide more insight into the genetic diversity and relatedness of genotypes within a species (1). B. bacilliformis isolates cluster in close proximity to each other, with the exception of B. bacilliformis isolate LA6.3, which belongs to a highly divergent group of B. bacilliformis isolates, strain type 8 (5). “Candidatus Bartonella ancashi” is located on an isolated branch away from the B. bacilliformis isolates, including B. baciliformis isolate LA6.3, and away from B. rochalimae, providing more evidence that “Candidatus Bartonella ancashi” is unique.

In the majority of cases, a patient is given a diagnosis of Carrion's disease, caused by B. bacilliformis, based on clinical characteristics, blood smears, and microbiological culture, which are usually insufficient to distinguish novel Bartonella species causing either verruga peruana or Oroya fever from B. bacilliformis. Additionally, Bartonella has proved to be a difficult organism to culture from clinical samples. For these reasons, PCR and sequencing-based genotyping methods, such as single-gene sequencing and MLST/MST, need to be used for identifying new species of Bartonella from rodents, arthropods, and clinical specimens (9, 15, 16). MLST methods, which involve concatenating a predetermined number of gene sequences, generally include the housekeeping genes (ftsZ, gltA, groEL, ribC, and rpoB) and rrs for phylogenetic analysis (9, 15), while MST is based on the variable 16S/23S intergenic linker region and is valuable for genotyping strains of Bartonella species (8). Additionally, fragments of the gltA and rpoB genes are found to have the highest discriminating power for Bartonella species identification and classification (16). La Scola et al. proposed that a Bartonella agent be considered a new species if a 327-bp gltA gene fragment and an 825-bp rpoB gene fragment were <96% and <95.4% similar, respectively, to a known Bartonella species (12). The similarity ranges for the rpoB and gltA gene fragments for “Candidatus Bartonella ancashi” fall well below these values, thereby providing evidence that the isolates represent a novel Bartonella species.

Furthermore, very low sequence divergence of genetic loci for sequence types (ST) within a species was demonstrated in MLST studies, for instance, 0 to 0.4% for B. quintana and 0.3 to 1% for B. henselae (17, 18). The sequence data obtained from the gltA and the rpoB gene fragments for the isolates (20.00 and 20.60) show notably low identity and high divergence from B. baciliformis and other known Bartonella species (Table 2). Phylogenetic analyses using MLST and the more strain-specific ITS sequences indicate that “Candidatus Bartonella ancashi,” while most closely related to B. baciliformis by MLST analysis, is in fact highly dissimilar from B. baciliformis by MST analysis, thereby providing clear insight into the unique evolution of this new Bartonella agent.

The results from this study, along with the recent discovery of B. rochalimae and the genetic diversity of B. bacilliformis (up to 5.3% divergence), indicate the possibility of a diverse group of Bartonella species or novel B. bacilliformis-like agents able to cause human disease circulating in areas of Peru where Carrion's disease is endemic (5, 7). Further microbiological, microscopic, and functional characterization of the isolates and whole-genome comparative analyses with B. bacilliformis and related species will further our understanding of the taxonomy of the Bartonella genus and, more importantly, provide insights into the genomic structure variations and dynamics, mechanisms for the high host adaptability, and pathogenesis of these endemic and opportunistic pathogens.

We acknowledge and thank Yu Yang for her technical support.

The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of the Army, Department of Defense, or the U.S. Government. As employees of the U.S. Government, this work was prepared as part of our official duties and therefore, under Title 17 USC paragraph 105, copyright protection is not available.

This work was supported by the Global Emerging Infections Surveillance and Response System, a Division of the Armed Forces Health Surveillance Center (work unit number 0000188 M.0931.001.A0074 and grant number I0361_12_WR) and the Wellcome Trut (grant number 81828).

There are no conflicts of interest to declare.