We defined the frequent presence of spirochetes in feces of patients with cholera. A 3-step method was used to establish the distribution of spirochetes in rice-water stool: 1) the presence of spirochetes in rice-water stool was determined by using dark-field microscopy; 2) the diversity of Brachyspira spp. within a subset of these patients was quantified by 16S rDNA analysis; and 3) the genetic diversity within the most abundant species was determined by analysis of NADH oxidase (nox).

Rice-water stool samples were collected from symptomatic cholera patients (>15 years of age with no history of antimicrobial drug therapy) during the spring cholera outbreak of 2006 in Dhaka, Bangladesh, at the International Centre for Diarrheal Disease Research, Bangladesh, as part of a larger study (14). Samples were examined by dark-field microscopy for V. cholerae and other bacteria, and the presence of V. cholerae, lytic vibriophage, and ETEC was determined by using standard methods (14). Samples were preserved at 4°C in formalin, or cell pellets were resuspended in phosphate-buffered saline and 15% glycerol and stored at –80°C. Analysis showed that 36% (23/64) of samples that were positive for only V. cholerae also harbored spirochetes (Technical Appendix, panels A–F). Spirochetes were also found in 4/11 and 5/15 samples that harbored ETEC alone or both ETEC and V. cholerae, respectively. When we removed ETEC-positive samples as potential confounders, the presence of spirochetes was independent of lytic vibriophage in V. cholerae–positive stool samples (Technical Appendix, panel G), which is in contrast to the documented trend concerning non–V. cholerae bacteria in rice-water stool (14). The ratio of spirochetes to V. cholerae was ≈1 and independent of lytic vibriophage (Technical Appendix, panel H).

In the second step, 10 samples were chosen for molecular analysis on the basis of a large amount of spirochetes. Samples were heated (99°C for 10 min), and standard PCR for a diagnostic segment of the 16S rDNA gene (15) was performed by using genus Brachyspira–specific primers (5′-GTCTTAAGCATGCAAGTC-3′ and 5′-AACAGGCTAATAGGCCG-3′). Products were cloned and sequenced bidirectionally (GenBank accession nos. FJ599620–FJ599639). B. pilosicoli was the most common species, found in all 10 samples (Technical Appendix, panel I). B. aalborgi was the second most common species (7 samples). A second set of panspecific 16S rDNA degenerate primers (5′-GTTTGATYCTGGCTYAGARCKAACG-3′ and 5′-CCSSTACGGMTACCTTGTTACG-3′) confirmed the presence of Brachyspira spp. and suggested that spirochetes of other genera were not present. The added resolution of nox analysis also identified B. hyodysenteriae at lower abundance.

Culture, purification, and microscopy of B. pilosicoli from glycerol stocks was used to cross-validate the molecular approach. Standard fastidious anaerobe agar (FAA) supplemented with bovine blood (10%), spectinomycin (400 μg/mL), and polymyxin B (5 μg/mL) was determined empirically to be the optimal medium for isolation of spirochetes. Dilutions of glycerol stocks from each patient were plated on FAA agar; single colonies were best obtained with an FAA overlay. Plates were incubated for 21 days at 37°C in an atmosphere of 94% H₂ and 6% CO₂. Most colonies gave rise to sigmoidal spiral cell morphotypes similar to the morphotype of the American Type Culture Collection (ATCC) (Manassas, VA, USA) control strain for B. pilosicoli. Additional controls were ATCC B. hyodysenteriae, Helicobacter pylori, and Borrelia burgdorferi. Subsequent molecular analysis of patient isolates confirmed them as B. pilosicoli. Consistent with previous studies, B. aalborgi was not cultured from patient samples.

In the third step, 5 samples were analyzed by using nox sequence comparisons that yield higher phylogenetic resolution. DNA was extracted and a nox segment was amplified by using degenerate primers (5′-GCYGGHACATGGGCDGCAAAAAC-3′ and 5′-CAAATACRCAAATAGCRTTAG-3′). Products were cloned and sequenced bidirectionally (GenBank accession nos. FJ599589–FJ599619). B. pilosicoli was the most common nox sequence found. A phylogenetic tree (Figure) demonstrated that individual patients harbored clonal lines of B. pilosicoli (patients B, D, and E) or more diverse strains (patient A). Overall, B. pilosicoli strains found in cholera patients were extremely diverse relative to the known outgroup species, which indicates the potential for detection of new species related to intestinal spirochetosis.

It has been casually observed for a century that stool from cholera patients harbors spirochete-like bacteria. We now define the major agents present as B. pilosicoli and B. aalborgi. More than one third of the cholera patients had spirochetes in their stools at densities equal to those of V. cholerae. The pathophysiology of intestinal spirochetosis in this setting and its relevance to human health remain unknown. Epidemiologic analysis of intestinal spirochetosis has so far relied on retrospective studies of colonic tissue collected for reasons secondary to the disease (4). We recommend a community-based prospective study of stool samples from healthy persons and patients with diarrhea by using the techniques described herein; animal reservoirs should be identified as a potential point of control. In the context of V. cholerae infection, we hypothesize that spirochetes may be present before V. cholerae infection and exacerbate the already devastating clinical course of cholera.