Characterization of a novel Rochalimaea species, R. henselae sp. nov ...

Vol. 30, No. 2

JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 1992, p. 265-274

0095-1137/92/020265-10$02.OO/0 Copyright © 1992, American Society for Microbiology

Characterization of a Novel Rochalimaea Species, R. henselae sp. nov., Isolated from Blood of a Febrile, Human Immunodeficiency Virus-Positive Patient RUSSELL L. REGNERY,1* BURT E. ANDERSON,' JILL E. CLARRIDGE III,2 MARIA C. RODRIGUEZ-BARRADAS,3 DANA C. JONES,' AND JANICE H. CARR4 Division of Viral and Rickettsial Diseases' and Hospital Infections Program,4 National Center for Infectious Disease, Centers for Disease Control, Atlanta, Georgia 30333, and Baylor College of Medicine2 and Veterans Affairs Medical Center,3 Houston, Texas 77030 Received 18 July 1991/Accepted 24 October 1991

Isolation of a Rochalimaea-like organism from a febrile patient infected with human immunodeficiency virus was confirmed. Analysis of 16S rRNA gene sequences, together with polymerase chain reaction and restriction endonuclease length polymorphism analysis of a portion of the citrate synthase gene, demonstrated that the agent is closely related to members of the genus Rochalimaea and that the isolate is genotypically identical to the presumptive etiologic agent of bacillary angiomatosis. However, the same genotypic analyses readily differentiated the new isolate from isolates of other recognized Rochalimaea species as well as other genera of bacteria previously suggested as putative etiologic agents of baciliary angiomatosis and related syndromes. We propose that the novel species be referred to as Rochalimaea henselae sp. nov.

tion (PCR)-restriction endonuclease length polymorphism (RFLP) analysis of the citrate synthase gene from the novel isolate, were used to identify and characterize the agent. This report describes the clinical findings relating to the febrile patient from whom the organism was isolated on several occasions; describes the results of our genotypic analysis, which was critical for identification of the isolate and; and relates these results with those recently reported by other laboratories (13, 14, 18, 20). The isolate represents the type strain of a novel species, hereafter referred to as Rochalimaea henselae sp. nov. The possible relationships between this organism and various disease syndromes (in addition to fever and including BA and bacillary peliosis hepatis) are explored.

Until now, all isolates of human pathogens positively identified as belonging to the genus Rochalimaea have been limited to the etiologic agent associated with trench fever and have been referred to as Rochalimaea quintana. Trench fever was a very serious source of disease during World War I; morbidity from the disease was considered the most prevalent of all diseases among Allied troops in France (21). There have been estimates that 1,000,000 military personnel were infected during the World War I era (22). After recurring to a more limited extent during World War II, this disease largely disappeared from prominence and has only been sporadically reported in the literature, although a limited number of reports suggest that the agent is not restricted to wartime environments (26). One other recognized species of Rochalimaea, Rochalimaea vinsonii, has been isolated on one occasion from North American rodents but has not been associated with human disease (2). rRNA gene sequences, identified as being Rochalimaea spp.-like, have recently been associated with bacillary angiomatosis (BA) in California (18). Although Rochalimaea organisms were not isolated in that study, the rRNA gene sequence data strongly suggest that an association exists between BA and an organism similar to R. quintana. In 1990, a man infected with human immunodeficiency virus was admitted to the Veterans Affairs Medical Center in Houston for evaluation of fever. Two blood samples, which were submitted for fungal cultures, yielded a slow-growing organism that was recognized as being Rochalimaea spp.like (4). The isolation of a Rochalimaea-like agent, referred to hereafter as the Houston-i isolate, was confirmed from a third blood sample from this patient. Differences in the initial growth characteristics of this fastidious, slow-growing organism in comparison with those of established isolates of Rochalimaea type species were noted. Sequence analysis of the 16S ribosomal gene, as well as polymerase chain reac*

MATERIALS AND METHODS

Type cultures. Two Rochalimaea isolates, representing two different recognized species, were obtained from the American Type Culture Collection (ATCC; Beltsville, Md.). R. quintana ATCC VR-358 and R. vinsonii ATCC VR-152 were routinely cultivated at 35°C in a 5% carbon dioxide atmosphere on tryptic soy agar supplemented with 5% defibrinated sheep blood (26). Rickettsia prowazekii, isolate Breinl (ATCC VR-142), was cultivated in Vero cell cultures, and cytoplasmic extracts containing rickettsiae were made as described previously (16). Isolation and cultivation of the organism from the patient's blood. Blood from the patient was drawn either directly into a Wampole Isostat tube (Wampole Laboratories, Cranbury, N.J.) or simply into a Vacutainer tube containing EDTA (Becton Dickinson, Rutherford, N.J.); isolates were obtained from both starting preparations. The organism was reisolated from frozen (-85°C), EDTA-treated blood without a significant loss of titer. Primary isolations were made on commercial brain heart infusion agar (BHIA) containing 5% sheep blood (BBL, Becton Dickinson, Cockeysville, Md.), tryptic soy agar (TSA) supplemented with 5% sheep blood (BBL), and heart infusion agar (HIA) containing 5%

Corresponding author. 265

266

REGNERY ET AL.

rabbit blood (BBL). Cultures were maintained at 35°C in a humidified incubator containing 5% carbon dioxide. Bacteriological plates were routinely examined. As noted elsewhere in this report, the Houston-1 isolate was cultivated from blood at various times during the course of the patient's disease episode, including after relapse of fever following cessation of antibiotic therapy. Additional Rochalimaea isolates. Four Rochalimaea-like isolates, which were previously submitted to the Centers for Disease Control for microbial identification, were compared with the Houston-1 isolate and recognized Rochalimaea species. Two of these isolates were recovered from patients in Oklahoma (patients 1 and 3 [20]), one isolate originated in a patient who apparently acquired his illness in Arkansas (10), and a fourth isolate originated in San Diego County, California (courtesy of the state of California Department of Health Services). This last isolate currently represents one of the first Rochalimaea isolates, that we are aware of, that has been'obtained in recent years as well as one of the first isolates reported from a human immunodeficiency virusinfected individual (November 1986). Clinical biochemical analysis. Biochemical tests were performed by standard methods (9) and by using the RapID ANA II system (Innovative Diagnostic Systems, Inc., Atlanta, Ga.), which tests for the presence of preformed enzymes. Tests for motility included observation of growth characteristics in motility agar and direct observation of bacilli by dark-field microscopy. The presence of catalase was tested for by emulsifying a colony in hydrogen peroxide and checking for the presence of microscopic bubbles under a coverslip. The presence of oxidase was tested for by using tetramethyl-p-phenylenediamine. Microscopic analysis. Four-day-old cultures of the Houston-1 isolate were prepared for microscopy by flooding a blood agar plate containing Rochalimaea spp. with phosphate-buffered saline and then gently sweeping adherent colonies off the agar surface with a bacteriological loop. A small aliquot of this material was placed directly onto a clean microscope slide, fixed with heat, and stained with Gimenez stain. Other material was fixed with glutaraldehyde and prepared for electron microscopy. Briefly, the glutaraldehyde-fixed material was filtered onto a Nuclepore filter (pore size 0.2 ,um; Nuclepore Corp., Pleasanton, Calif.) and washed three times with Sorenson's buffer (pH 5.0). The filtered material was treated in 1% osmium tetroxide for 2 h and again washed three times with Sorenson's buffer. The specimens were dehydrated in a graded series of increasing concentrations of ethanol (30 to 100%). The dehydrated specimens were immersed in hexamethyldisilizane (Polysciences, Inc., Warrington, Pa.) for 2 h and then dried in a desiccator overnight. Finally, the specimens were placed on a stub, sputter coated with gold, and observed with a Philips (model 515) scanning electron microscope. Fatty acid analysis. Whole-cell fatty acid analysis was performed on R. henselae sp. nov. cultures incubated at 35°C in air and harvested after 4 days of growth on chocolate agar (BBL). Fatty acid methyl esters were chromatographed on a Hewlett-Packard series II 5890 gas chromatograph (12) and were identified by using a computer-assisted comparison of the retention time of the sample with that of a standard mixture (Microbial-ID, Newark, Del.). 16S rRNA gene sequence analysis. DNA for PCR amplification was extracted from pure cultures of R. quintana, R. vinsonii, and R. henselae Houston-1 by sodium dodecyl sulfate-proteinase K lysis followed by phenol-chloroform extraction as described previously (23). The resulting aque-

J. CLIN. MICROBIOL.

ous phase was concentrated by using a Centricon 30 concentrator (Amicon Corp., Danvers, Mass.) and was washed three times with 2 ml of TES (10 mM Tris [pH 8.0], 1 mM EDTA, 10 mM NaCl). PCR amplification was performed by using a thermal cycler and GeneAmp reagents (Perkin-Elmer Cetus, Norwalk, Conn.). Two pairs of "universal," degenerate primers known to amplify approximately 92% of the 16S rRNA gene, as two separate PCR products, from all eubacteria previously studied (30) were used to prime the PCR synthesis of products that were subsequently used for cloning and sequence analysis. The 5' end of each primer was modified to contain unique restriction endonuclease sites to facilitate cloning. Each sample was amplified for three cycles at 94°C for 1 min, 48°C for 2 min, and 66°C for 1 min and 30 s; this was followed by 27 cycles at 88°C for 1 min, 52°C for 2 min, and 68°C for 1 min and 30 s. The resulting PCR products were isolated from a 1.0% agarose gel and cloned into pUC19. Clones were sequenced by using double-stranded sequencing with T7 DNA polymerase (Sequenase; U.S. Biochemicals, Cleveland, Ohio). Each isolate was amplified, cloned, and sequenced at least twice to prevent the reading of PCR incorporation errors; if discrepancies were detected, a third, independent sequence was produced. Great care was taken not to introduce contaminating bacterial DNA into the PCRs by using the universal primers because of their broad range of amplification. Nucleotide sequence accession numbers. GenBank accession numbers for the respective 16S rRNA gene sequences are as follows: R. quintana, M73228; R. vinsonii, M73230; R.

henselae, M73229. PCR-RFLP analysis. RFLP analysis was applied to PCRamplified DNA, which was primed with nondegenerate oligonucleotides previously demonstrated to initiate the synthesis of PCR products of approximately 381 nucleotides from a portion of the rickettsial citrate synthase gene (16). Chromosomal DNA from R. prowazekii was used as a positive control for PCR synthesis and digestion; controls containing no DNA template were always included in PCR amplifications. Digested DNA fragments were separated and analyzed by using standard electrophoretic protocols and methods described previously (16). The numbers of comigrating DNA fragments that were observed between homologous PCR-RFLP digests of two or more isolates were counted. Data from the number of comigrating DNA fragments were used to derive estimates of sequence relatedness by the methods described by Upholt (24) and were subsequently used by others to estimate sequence divergence between related bacteria (3, 11, 16). RESULTS Case report. A previously asymptomatic, HIV antibodypositive, 40-year-old man was admitted with a 2-month history of daily fever, extreme fatigue, anorexia, and the loss of 10 kg of weight. He was a former intravenous drug user and had a history of extensive travel within the United States. During the previous 4 months he had lived in Corpus Christi, Tex., where he had close contact with cats, ferrets, and dogs as well as children who were presumably infested with scabies. He denied receiving any scratches or animal bites, except for the bites of fleas. On admission to the hospital his temperature was 39.4°C and he had a heart rate of 95 beats per minute and a blood pressure of 90/46 mm Hg. His physical examination was remarkable for the presence of thrush and a 2/6 holosystolic murmur, a palpable liver and spleen, and the presence of perirectal ulcers. His leukocyte

VOL. 30, 1992

CHARACTERIZATION OF A NOVEL ROCHALIMAEA SPECIES

count was 4,900/mm3 (80% segmented, 11% lymphocytes, 4% monocytes, 4% eosinophils, and 1% basophils), hemoglobin was 8.8 g/dl, hematocrit was 26.9%, total protein was 7 g/dl, and albumin was 3.2 g/dl. His CD4 count was 20 cells per mm3. The remainder of the laboratory evaluation was within normal limits. Chest roentgenogram revealed increased interstitial markings. Repeat routine bacterial cultures of urine and blood specimens were negative. Cultures of a perianal ulcer grew herpes simplex virus type 2, and the patient was started on acyclovir. A bone marrow biopsy specimen showed megakaryocytic hyperplasia; no organisms were identified. An echocardiogram revealed a thickened aortic valve. Abdominal computed tomography scan showed hepatosplenomegaly with no focal lesions. An isotope gallium citrate scan, lower gastrointestinal series, lymphangiography, computed tomography scan of the head, and cerebrospinal fluid examination were within normal limits. Study of bronchoalveolar lavage specimen failed to reveal Pneumocystis carinii, acid-fast bacilli, or fungi; viral cultures were negative, and there was no evidence of malignancy. The patient continued to have daily fevers of 38 to 40°C throughout this period. Five weeks after admission, blood cultures taken on the first and eighth days of hospitalization were reported positive for a Rochalimaea-like organism (4). With the presumptive diagnosis of trench fever, the patient was started on a 21-day course of doxycycline (100 mg, twice a day); after 48 h he defervesced. Blood, urine, bone marrow, and bronchoalveolar lavage fluid cultures remained negative for mycobacteria and fungi. Six weeks after the discontinuation of therapy, fever, anorexia, and malaise recurred. Blood cultures drawn at this time were again positive for a Rochalimaea-like organism, and treatment with doxycycline for 1 month (same dose as described above) was reinstituted, with an immediate and positive response. After a second relapse of fever, the patient completed 2 months of doxycycline (same dose as described above). Repeat cultures of blood specimens taken in the subsequent 6 months have been negative, and the symptoms associated with his initial infection have not recurred. Growth characteristics. Blood from the febrile patient, when cultured on either commercial BHIA-sheep blood, TSA-sheep blood, or HIA-rabbit blood, yielded characteristic colonies which were visible after 9 to 10 days of incubation. The approximate titer of colony-forming organisms in the patient's blood was 30/ml after recrudescence of fever following the second course of antibiotic therapy. Primary colonies were deeply invaginated (cauliflowerlike), firm, adherent, and tenaciously imbedded in the surface of the agar. All original individual colonies isolated from the patient's blood had similar morphologic and growth characteristics. Close inspection of subcultured plates revealed minute colony formation by 6 days after inoculation, although clear colony morphology was not evident at that time. After multiple passages of fresh colonies, the incubation time to colony visualization decreased substantially, and discrete colonies could be discerned after 3 to 4 days. The invaginated colony morphology became less pronounced after multiple, relatively rapid passages. Colony growth was not limited by incubation time, and colonies continued to grow progressively larger over a period of several weeks (Fig. 1). Several of these latter initial growth characteristics of the Houston-1 isolate were in contrast to those noted for the ATCC Rochalimaea type strains, which typically grew relatively rapidly without any delay in passaging, had shiny, smooth colonies, and were not similarly imbedded in the agar. Likewise, although the Rochalimaea

267

species isolates obtained from ATCC proliferated rapidly on the surfaces of cultured cells, initial Houston-1 isolate material did not produce a similar generalized infection when it was inoculated onto Vero cell monolayers, thus suggesting that cocultivation with eucaryotic cells is not the method of choice for primary isolation. The Houston-1 isolate, after additional laboratory passages on solid medium (and perhaps more analogous to the ATCC type strains in terms of more extensive passage history), was not retested for the ability to grow rapidly on eucaryotic monolayers. After reinoculation of the organism onto either chocolate agar or TSA-sheep blood, there was no growth in air at 22 or 42°C but good growth at 30 and 35°C. Colonies grew to a slightly larger size when they were incubated in CO2 (8%) at 35°C than when they were cultured without added CO2 at 35°C. Growth on subculture was also achieved on HIArabbit blood or TSA-sheep blood when the plates were incubated in candle jars, as described previously for other Rochalimaea isolates (20). No growth was observed on Sabouraud dextrose medium. Except for the production of peptidases, the Houston-1 isolate was biochemically inert when it was tested by typical clinical procedures. The RapID ANA II system, which is designed primarily for the clinical identification of anaerobic organisms by detection of specific preformed enzymes, has also been shown to be useful for the identification of difficult-to-identify aerobic organisms (19). The RapID ANA II system, when it was used for the analysis of the Houston-1 isolate, detected a limited number of enzyme-substrate cleavage reactions which included the cleavage of leucylglycine, glycine, proline, phenylalanine, arginine, and serine, resulting in an identification number 000671; a known microbe is not currently associated with this identification number; however, members of the genus Rochalimaea are not yet part of the commercial diagnostic data base (6). Negative clinical assays, in our hands, included those testing for catalase, urease, esculin hydrolysis, motility, nitrate reduction, and oxidase. Preliminary attempts to cultivate the Houston-i isolate in stationary, liquid media did not produce turbid suspensions of individual organisms; however, the blood agar platederived inoculum material appeared to act as foci for the growth of limited numbers of large cohesive aggregates. Reinoculation of agar-grown organisms into Bactec 660 6A or 7A bottles (Becton Dickinson) did not result in sufficient growth to change the growth index compared with that of the uninfected controls. Staining and morphologic characteristics. Rapidly proliferating organisms, which were obtained after several subpassages, stained readily with Gimenez histological stain. Organisms so stained appeared as small red bacilli, often slightly curved (Fig. 2). Organisms obtained from older, but still quite viable, colonies resisted uptake of Gimenez stain. The material which was successfully used for light microscopy was also prepared for and observed with a scanning electron microscope (Fig. 3). As with the Gimenez-stained material and the observations of growth habits noted during various culturing experiments, the organisms viewed with the scanning electron micrograph appeared to form cohesive aggregates, with relatively few organisms existing freely. The average size of the organisms visualized was approximately 2 p.m in length by 0.5 to 0.6 ,um in width. Evidence of flagella was not observed, although specialized procedures for flagellum fixation and preservation were not used prior to electron microscopy. The morphologies of the individual organisms were not studied during greatly different periods

268

REGNERY ET AL.

J. CLIN. MICROBIOL.

FIG. 1. Growth on blood agar. Fourth-passage, 24-day-old Houston-1 isolate colonies were cultured on sheep blood-TSA bacteriological plates.

of cultivation; however, all organisms observed within individual microscopic preparations (which presumably include the products of multiple generations) appeared to be relatively uniform in size (Fig. 2 and 3). Fatty acid analysis. The major fatty acids observed after whole-cell fatty acid analysis of the Houston-1 isolate were octadecenoic acid (C18:1, 54 to 56%), octadecanoic acid (C18:0, 18 to 20%), and hexadecanoic acid (C16:0, 17%). The absence of other detectable fatty acids excluded possible identification of almost all other bacteria except members of the genus Brucella. This fatty acid pattern was similar to that observed with R. quintana and other recent Rochalimaealike isolates (20). 16S rRNA gene sequence analysis. To determine the possible relationship of the Houston-i isolate to members of the genus Rochalimaea as well as other bacteria, the 16S rRNA genes from the Houston-1 isolate and R. vinsonii were sequenced. Universal primers allowed amplification of approximately 1,400 nucleotides of the rRNA gene sequence as two separate PCR products. The 767-bp products, corresponding to the 5' half of the 16S rRNA gene, that were produced by using primers EC11 and EC12 (modified versions of POmod and PC3mod primers [30]) were observed when the Houston-1 isolate (Fig. 4, lane F), R. quintana (Fig. 4, lane G), and R. vinsonii (Fig. 4, lane H) were amplified. No product was observed when these primers were used to amplify a negative control containing no DNA template (Fig. 4, lane I). Similarly, a 737-bp product, corre-

sponding to the 3' half of the 16S rRNA gene, that was produced with primers EC9 and EC10 (modified versions primers P3mod and PC5 used by Wilson et al. (30) was seen when the Houston-1 isolate (Fig. 4, lane B), R. quintana (Fig. 4, lane C), R. vinsonii (Fig. 4, lane D), and no DNA (Fig. 4, lane E) were used as templates for amplification. These PCR products were cloned and sequenced. The 16S rRNA gene sequences used for comparison and alignment were obtained by taking a consensus of three independent sequences for each cloned PCR product. The first and second sequences obtained for the Houston-1 isolate had three nucleotides in disagreement, and the first and second sequences for R. vinsonii had two ambiguities. In both cases, a third sequence agreed with one of the two previous sequences at these ambiguous positions and was taken as the consensus. The occasional disagreement among sequences was assumed to be the result of polymerasenucleotide incorporation errors and has been observed by other investigators who used this technique (18). The entire sequence was used for alignment by using the Gap program of the Genetics Computer Group (5). The sequence of the Houston-i isolate was compared with 16S rRNA gene sequences on file with GenBank and showed the greatest homology with R. quintana (98.7%) and lesser homologies with 16S rRNA gene sequences from more distantly related organisms (Table 1). In our laboratory, we sequenced the 16S rRNA gene from R. quintana (Fuller strain) and found it to differ slightly from the sequence reported previously (28).


VOL. 30, 1992

I

4

.4.

269

products obtained from different Rochalimaea isolates (Fig. 5). PCR-amplified products were digested with seven restriction endonucleases and subjected to polyacrylamide gel electrophoresis. Obvious differences were seen in many of the digest patterns of PCR-amplified citrate synthase sequences from the various isolates; PCR-RFLP analysis allowed for the rapid differentiation of isolate genotypes (Fig. 5). The numbers of DNA fragments produced by digestion of the PCR-amplified, citrate synthase-specific DNA with seven restriction endonucleases are tabulated in Fig. 6, together with the number of comigrating fragments. Estimates of sequence divergence derived by numerical analysis of the percentage of comigrating fragments illustrate that all of the isolates examined have substantial inferred citrate synthase sequence divergence (6 to 11%). PCR-RFLP analysis was also used to correlate the identity of the Houston-i isolate with those of other recent Rochalimaea-like isolates of human origin. The PCR-RFLP patterns obtained from these isolates were identical to comparable patterns derived from the Houston-i isolate and were readily differentiated from the patterns of other Rochalimaea species (Fig. 7). In addition, partial 16S rRNA gene sequences obtained from one of the isolates from Oklahoma (patient 1 [20]) were identical to the 16S rRNA gene sequences obtained from the Houston-i isolate (data not shown). DISCUSSION

lOptm FIG. 2. Light microscopy of Gimenez-stained Houston-i isolate. Rapidly growing colonies of the Houston-1 isolate were washed from the surface of an agar plate, and a small amount of the suspension was heat-fixed to a microscope slide and stained by standard Gimenez staining procedures.

Using our data, we found the 16S rRNA gene sequence from the Houston-1 isolate to be 98.7% related to R. quintana and 99.3% related to R. vinsonii. The R. quintana and R. vinsonji sequences were found to be 98.9% related. The partial 16S rRNA gene sequence determined by Relman et al. (18) for the putative etiologic agent of BA was found to be identical to the corresponding portion of the 16S rRNA gene sequence obtained from the Houston-i isolate of R. henselae sp. nov. (Table 1). PCR-RFLP analysis. RFLP analysis of specific genes, amplified by the PCR technique, has been shown to be useful for identifying rickettsial genotypes and species (16). Oligonucleotides that were previously demonstrated (16) to be suitable for priming PCR amplification of a portion of the citrate synthase genes from nearly all rickettsial species as well as from R. quintana were tested for their ability to prime DNA amplification from DNAs purified from the Houston-1 isolate and R. vinsonii. PCR products were readily produced by using conditions comparable to those reported previously (16). All three of the uncut Rochalimaea citrate synthase PCR products were slightly larger (approximately 400 bp) than those produced for members of the genus Rickettsia (approximately 381 bp). Variation was noted between the sizes of PCR-amplified citrate synthase

The growth characteristics of the freshly isolated Houston-i agent contrasted with those of well-established type species of Rochalimaea. With passaging, colony morphology and speed of growth of the novel agent began to more closely resemble those of other Rochalimaea type species. We have not yet fully investigated the influences of culture conditions and various passaging levels on the novel isolate. Rochalimaea isolates obtained in the Soviet Union after World War II, when cultivated on a solid medium, were reported to have yielded colonies described as cauliflowerlike that were deeply imbedded in the substrate and that formed large aggregates in liquid culture (15); these terms are essentially identical to those used here to describe the colonies of the initial isolate material of the Houston-1 agent. This same description of growth morphology of the Soviet isolates has been contrasted to the morphologic features noted by others who obtained original isolates of what has become identified as the R. quintana type species

(25).

Although R. henselae appears to be a fastidious and slow-growing organism, it can be cultivated by standard laboratory procedures. Relatively rapid growth (4 days between subculture) of the Houston-1 isolate was achieved by multiple passaging of fresh colonies shortly after they initially became visible. The original isolates were made on BHIA and TSA supplemented with 5% sheep blood (4) (see Materials and Methods). Subsequent reisolation from the patient's blood was also achieved by using HIA plates supplemented with 5% rabbit blood, as recently described by Slater et al. (20). It is noteworthy that semiautomated, clinical bacterial isolation procedures, which often rely on liquid medium-based assays, in the absence of exogenous gaseous carbon dioxide, may not be suitable for cultivation or detection of primary Rochalimaea isolates. Moreover, such cultures may not be maintained for an incubation

A

I pm_ :.

V

-W4

k

AA

I,,:

B-, ... B.

...

FIG. 3. Scanning electron microscopy of glutaraldehyde-fixed R. henselae Houston-1 isolate. Rapidly growing colonies of the Houston-i isolate were prepared as described in the legend to Fig. 2, fixed with glutaraldehyde, filtered onto a 0.2-,um-pore-size Nuclepore filter, and prepared for scanning electron microscopy. Although individual organisms were visible (A), most bacilli appeared to form large aggregates after suspension in phosphate-buffered saline and during glutaraldehyde fixation (B). 270


VOL. 30, 1992

1 2 3 4 5 6

1 2 3 4 5 6

271

1 2 3 4 5 6

500-

249

311-

151200

-

°140oo664842 40= 24 -

733 bp-

767 bp A

AB CD EF G H

I

FIG. 4. Agarose gel electrophoresis of PCR products obtained amplification with the universal primers EC9 and EC10 (lanes B through E) and EC11 and EC12 (lanes F through I). The sources of template DNA used for PCR amplification are as follows: lanes B and F, R. henselae Houston-1 isolate; lanes C and G, R. quintana; lanes D and H, R. vinsonii; lanes E and I, no DNA control. Ten-microliter aliquots of PCR-amplified products were electrophoresed through a 1.0% agarose gel, stained with ethidium bromide, and compared with molecular size standards (+X174 phage DNA cleaved with HaeIII), which are shown in lane A.

C

FIG. 5. Examples of PCR-amplified, restriction endonucleasedigested DNA fragments. (A) Undigested citrate synthase-amplified sequences; lanes 1 and 6, 4)X174 restriction fragment Hinfl digest (size standards); lane 2, Houston-1 isolate; lane 3, R. vinsonii; lane 4, R. quintana; lane 5, R. prowazekii. Citrate synthase-amplified sequences digested with restriction endonucleases TaqI (B) and MseI (C) (digests are arranged in the same order as in panel A). Numbers on the left are in base pairs.

upon

period sufficient to detect growth of a primary isolate. Experiments are under way to evaluate the systems that are most suitable for clinical detection. To date, the limited number of primary isolations of Rochalimaea species have all been accomplished, to the best of our knowledge, on solid media (4, 10, 15, 20, 25, 26). Evidence of Rochalimaea-like 16S ribosomal sequences, extracted from lesions of BA, has recently been used to establish a link between BA, found primarily among immunocompromised individuals, and an unidentified Rochalimaea agent. This analysis was based on approximately 400 bp of PCR-amplified 16S rRNA sequences (isolates were not obtained). This sequence (GenBank accession number M59459 [18]) and the sequence derived from pure cultures of R. henselae sp. nov., Houston-i isolate, are identical. These

completely homologous sequences are good evidence that the two causative agents are one and the same species. Similarly, the relative variations in 16S rRNA sequences reported to exist between the BA-associated agent and other species of microorganisms, as reported previously (18), are in complete concordance with the interspecies sequence variations observed in this study (Table 1). The variation between 16S rRNA gene sequences noted between the Houston-1 isolate and other type species of Rochalimaea (Table 1) suggests that the Houston-1 isolate represents a new species within the genus Rochalimaea. This conclusion is supported by the fact that the 0.7% 16S rRNA gene sequence divergence seen between the Houston-1 isolate and R. vinsonii is greater than the 0.5% divergence reported for R. prowazekii and Rickettsia typhi (27).

Rochal imaea

%

Speciesa BA-TFb .............................................

vinsonii Rochalimaea quintana .................................. Bartonella bacilliformis ................................. Brucella abortus ..................................... Rochalimaea

....................................

Homology with R. henselae Houston-1

100.0 99.3

98.7 95.6 94.0

Cat scratch fever agent (AFIP) .......................

87.9

Rickettsia rickettsii ....................................

84.9

Ehrlichia risticii .....................................

84.9

The entire 16S rRNA gene sequence (when available) was used for alignment, as described previously (5). The R. henselae Houston-1 isolate, R. vinsonii, and R. quintana sequences were determined in our laboratory; all other sequences were obtained from GenBank. AFIP, Armed Forces Institute of Pathology. b Partial 16S rRNA sequence from the putative etiologic agent of BA (18). a

Rochal imaea

Rochal imaea

Rickettsia

henselae

vinsonii

quintana

prowazekii

(19)

10/34

16/32

8/32

10.8

(15)

8/28

6/28

6.0

11.0

(13)

8/26

12.2

13.6

10.3

(13)

henselae

Rochalimaea

TABLE 1. Relatedness between the Houston-1 isolate 16S rRNA gene and various eubacteria

Rochal imaea

vinsoni i

Rocha 1 imaea

quintana

Rickettsia

prowazeki i I

FIG. 6. Comigrating DNA fragments and estimated sequence divergence. Citrate synthase-specific PCR-amplified DNA was cleaved with restriction endonucleases AluI, HhaI, Hinfl, MseI, RmaI, Sau3A, and TaqI. Numbers in parentheses (along the diagonal) indicate the total number of fragments used in analysis of each species. Fractions in the upper right sector indicate the number of comigrating DNA fragments for each pair of species divided by the number of fragments present for both species. Numbers in the lower left sector correspond to the estimated percentage of sequence divergences.

272

REGNERY ET AL.

A I

2 3 54 F6 7 8 9

J. CLIN. MICROBIOL. R 1

2 3 4 5

77 8 Q

FIG. 7. PCR-RFLP patterns for multiple R. henselae isolates. DNAs from various Rochalimaea isolates were amplified by using primers specific for citrate synthase sequences and were digested with either HhaI (A) or TaqI (B). 4X174 restriction fragment Hinfl digests were used as size standards (lanes 1 and 9). R. quintana (lane 2) and R. vinsonii (lane 3)-derived DNAs were included for comparison. Other isolates were Houston-1 (lane 4), two isolates from Oklahoma (lanes 5 and 6), one isolate from Arkansas (lane 7), and one isolate from California (lane 8). AluI, MseI, and Hinfl digests of comparable PCR products revealed no differences among the R. henselae isolates (data not shown).

These two species of Rickettsia are clearly distinct species among the order Rickettsiales, to which the genus Rochalimaea belongs. The postulated relationship between the agent responsible for BA and the agent responsible for cat scratch fever has been the object of considerable speculation (1, 8, 18). Similarly, questions have been raised regarding the relation of Bartonella bacilliformis to the etiologic agent of BA (13, 18); B. bacilliformis is known to cause a disease that is perhaps analogous to BA. The elucidation of the 16S ribosomal gene sequence for R. henselae, in combination with the recent publication of comparable sequences from both the cat scratch fever bacillus and B. bacilliformis (13), now permits a direct genotypic comparison of these organisms (Table 1). Although B. bacilliformis demonstrates considerable genotypic relatedness to members of the genus Rochalimaea, it is clear that R. henselae and other members of the genus Rochalimaea represent distinct species. R. henselae and the agent described as the cat scratch fever bacillus are more distantly related. The possible role that R. henselae plays in certain examples of clinically and histopathologically defined cat scratch fever awaits further analysis; examples of BA temporally associated with cat scratch wounds are well documented (7). PCR-RFLP analysis clearly differentiated the R. henselae sp. nov. genotype from that of either R. quintana or R. vinsonii. The relatively large differences noted between the numbers of comigrating fragments of the Houston-1 isolate and two Rochalimaea type species isolates led to citrate synthase sequence divergence estimates (6 to 11%) equaling or exceeding similar estimates for citrate synthase sequence divergence among recognized rickettsial species (e.g., 2 to 6%) (16). RFLP sequence divergence estimates made in this manner have maximal numerical significance for closely related sequences; numerical confidence levels decline with, more distantly related sequences (24). The numerical differences noted among citrate synthase gene sequences for the three Rochalimaea species tested in this study should be emphasized less than the observation that the sequences are evidently so dissimilar that quantification is imprecise.

Multiple restriction endonuclease digests of the citrate synthase-specific PCR products from Rochalimaea-like isolates from Oklahoma (two isolates), Arkansas (one isolate), and southern California (one isolate) demonstrated that all of the isolates studied are identical to one another and the Houston-1 isolate by the PCR-RFLP genotypic criteria used in this study (Fig. 7). These data document the genetic homogeneity of these isolates, which were first cultivated in different laboratories, and demonstrate that these isolates should be considered a single species. Likewise, these data document the widespread distribution of these infections. However, it should also be noted that an R. quintana isolate has been identified in Oklahoma (29) and 16S ribosomal gene sequences homologous to those of R. quintana have been documented in California (18); at least two species of Rochalimaea appear to be a responsible for contemporary human infections within the United States. Both 16S rRNA gene sequence analysis and PCR-RFLP analysis of the citrate synthase gene led to the conclusion that the Houston-1 isolate, while having similarities to species recognized as belonging to the genus Rochalimaea, is not the same species as either R. quintana or R. vinsonii. The newly proposed name for this agent, R. henselae sp. nov., recognizes the significant contribution of Diane M. Hensel, who has been instrumental in making several of the original isolations of this organism (20, 29). As noted above, what appears to be the same species has been isolated in multiple localities (4, 10, 20), and good evidence of infection has been detected by still other laboratories (14, 18). On the basis of limited samples and reports, it also seems clear that the disease spectrum of this organism may be variable and may include a syndrome of fever and bacteremia, as well as BA and bacillary peliosis hepatis (14, 17, 18, 20) (see Results). The natural history of the disease and methods of transmission are also under investigation. The results of these studies may suggest intervention methods for persons at risk of acquiring serious infections. In an effort to minimize taxonomic redundancy and confusion, the name for the novel species was arrived at by mutual agreement with other investigators who had an interest in describing the new species (29). Similarly, descriptive features for the novel species (see below) were compiled in collaboration with other investigators experienced in R. henselae isolation and characterization

(29).

Description of R. henselae sp. nov. R. henselae (hen'.sel. a.e. N. L. gen. n. henselae), is named in honor of Diane M. Hensel, a dedicated clinical microbiologist who has contributed much to the isolation and identification of isolates of this species (20, 29). The type strain is the Houston-1 isolate recovered from the blood of a febrile, 40-year-old human immunodeficiency virus-positive man. The R. henselae cell size, determined by scanning electron microscopy, is estimated to be approximately 1 to 2 ,um in length by 0.5 to 0.6 ,im in width; rods frequently appear to be slightly curved (see also reference 20). Rapidly growing organisms are gram negative and are readily stained with Gimenez stain. Optimal growth occurs on enriched medium with erythrocytes; R. henselae primary isolate growth has been achieved on BHIA and TSA supplemented with 5% sheep blood as well as HIA supplemented with 5% rabbit blood. Original colonies of R. henselae were observed after 9 to 15 days of incubation at 35°C in a 5% C02, humidified atmosphere or at 30°C in air; optimal growth does not occur anaerobically (see also reference 20). Subsequent passages of R. henselae demonstrated


VOL. 30, 1992

more rapid growth (e.g., 5 days until there was visible colony formation and 3 to 4 days until there was obvious microbial growth on plates). Cells are oxidase and catalase negative (or weakly catalase positive), nonreactive in carbohydrate utilization tests, and urease negative. Twitching motility has been reported (20) but was not observed in this study. The Houston-1 primary isolate colony morphologic features include durable, invaginated colony formation (cauliflowerlike) imbedded in the agar substrate. Later passages of the Houston-1 isolate displayed colony morphologic features similar to that of other Rochalimaea type species; smooth, shiny colonies were not deeply imbedded in agar. Organisms have a tendency to autoagglutinate when they are placed in liquid suspension, autoadhere when they are removed with a loop from culture plates, and adhere to glass or plastic as a film of cells in fluid media. The species nova is distinguished from other Rochalimaea species by distinctive 16S rRNA gene sequence, PCR-RFLP profiles derived from the citrate synthase gene, growth characteristics of the original isolate, and whole-cell DNA relatedness (29). Distinctive cellular fatty acid profiles can be used to help identify isolates as members of the genus Rochalimaea. When multiple isolates of R. henselae were analyzed for cellular fatty acid composition, very slightly higher percentages of C18 have been reported relative to those found for R. quintana (29). Strains are broadly susceptible to antimicrobial agents in vitro (20). To date, almost all R. henselae isolates have been originally cultivated from the blood of patients, many of whom were potentially immunocompromised (4, 10, 20, 29) (see Results). 0

ACKNOWLEDGMENTS

We thank Robert Weaver and Dannie Hollis for helping to identify and locate Rochalimaea-like isolates previously submitted to the Centers for Disease Control for identification. We also thank Bobbye Simon, Christine Rambo, and Holly Neville for technical assistance and Don Howard for photomicrographic help. We also thank Walter Bond, Dwane Rhoden, and Ken Wilson for helpful suggestions and James Olson and Richard Hamill for critical review of the manuscript. We are grateful to Thomas MacAdoo for advice regarding the latin species designations.

1.

2. 3.

4.

5. 6. 7.

REFERENCES Angritt, P., S. M. Tuur, A. M. Macher, K. J. Smith, C. S. Park, F. P. Hobin, and C. Myrie-Williams. 1988. Epithelioid angiomatosis in HIV infection: neoplasm or cat scratch disease? Lancet i:996. (Letter.) Baker, J. A. 1946. A rickettsial infection in Canadian voles. J. Exp. Med. 84:37-50. Clark-Curtiss, J. E., and G. P. Walsh. 1989. Conservation of genomic sequences among isolates of Mycobacterium leprae. J. Bacteriol. 171:4844 4851. Clarridge, J. E., B. B. Simon, and H. McGee. 1991. Unnamed gram-negative rod isolated from AIDS patient: description and changes on subculture, C-372, p. 404. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. American Society for Microbiology, Washington, D. C. Devereux, J., P. Haeberli, and 0. Smithies. 1984. A comprehensive set of sequence analysis programs for the VAX. Nucleic Acids Res. 12:387-395. Innovative Diagnostic Systems. 1989. RapID ANA II code compendium. Innovative Diagnostic Systems, Atlanta, Ga. Kemper, C. A., C. M. Lombard, S. C. Deresinski, and L. S. Tompkins. 1990. Visceral bacillary epithelioid angiomatosis: possible manifestations of disseminated cat scratch disease in the immunocompromised host: a report of two cases. Am. J. Med. 89:216-222.

273

8. LeBoit, P. E., T. G. Berger, B. M. Egbert, T. S. Benedict Yen, M. H. Stoler, T. A. Bonfiglio, J. A. Strauchen, C. K. English, and D. J. Wear. 1988. Epithelioid haemangioma-like vascular proliferation in AIDS: manifestation of cat scratch disease bacillus infection? Lancet i:960-963. 9. Lennette, E. H., A. Balows, W. J. Hausler, and H. J. Shadomy (ed.). 1985. Manual of clinical microbiology, 4th ed. American Society for Microbiology, Washington, D. C. 10. Lucey, D., M. Dolan, M. Garcia, S. Wegner, G. Morgan, J. Landgraf, C. K. English, T. L. Hadfield, and R. Almedia. 1990. Program Abstr. 30th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 566. 11. McFadden, J. J., P. D. Butcher, R. Chiodini, and J. HermonTaylor. 1987. Crohn's disease-isolated mycobacteria are identical to Mycobacterium paratuberculosis, as determined by DNA probes that distinguish between mycobacterial species. J. Clin. Microbiol. 25:796-801. 12. Miller, L., and T. Berger. 1985. Bacterial identification by gas chromatography of whole cell fatty acids. Hewlett-Packard application note 228-41. Hewlett-Packard, Avondale, Pa. 13. O'Connor, S. P., M. Dorsch, A. G. Steigerwalt, D. J. Brenner, and E. Stackebrandt. 1991. 16S rRNA sequences of Bartonella bacilliformis and cat scratch disease bacillus reveal phylogenetic relationships with the alpha-2 subgroup of the class Proteobacteria. J. Clin. Microbiol. 29:2144-2150. 14. Perkocha, L. A., S. M. Geaghan, T. S. Benedict Yen, S. L. Nishimura, S. P. Chan, R. Garcia-Kennedy, G. Honda, A. C. Stoloff, H. Z. Klein, R. L. Goldman, S. Van Meter, L. D. Ferrell, and P. E. LeBoit. 1990. Clinical and pathological features of bacillary peliosis hepatitis in association with human immunodeficiency virus infection. N. Engl. J. Med. 323:15811586. 15. Pshenichkov, A. V. 1957. Results of many years' combined study of trench fever. J. Microbiol. Epidemiol. Immunobiol. 28:1392-1397. 16. Regnery, R. L., C. E. Spruill, and B. D. Plikaytis. 1991. Genotypic identification of rickettsiae and estimation of intraspecies sequence divergence for portions of two rickettsial genes. J. Bacteriol. 173:1576-1589. 17. Relman, D. A., S. Falkow, P. E. LeBoit, L. A. Perkocha, K.-W. Min, D. F. Welch, and L. N. Slater. 1991. The organism causing bacillary angiomatosis, peliosis hepatis, and fever and bacteremia in immunocompromised patients. N. Engl. J. Med. 324: 1514. 18. Relman, D. A., J. S. Loutit, T. M. Schmidt, S. Falkow, and L. S. Tompkins. 1990. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N. Engl. J. Med. 323:1573-1580. 19. Shawar, R., J. Sepulveda, and J. E. Clarridge. 1990. Use of the RapID-ANA system and sodium polyanetholesulfonate disk susceptibility testing in identifying Haemophilus ducreyi. J. Clin. Microbiol. 28:108-111. 20. Slater, L. N., D. F. Welch, D. Hensel, and D. W. Coddy. 1990. A newly recognized fastidious gram-negative pathogen as a cause of fever and bacteremia. N. Engl. J. Med. 323:15871593. 21. Strong, R. P. (ed.) 1918. Trench fever: report of commission, Medical Research Committee, American Red Cross. Oxford University Press, Oxford. 22. Swift, H. G. 1920. Trench fever. Arch. Intern. Med. 26:76-98. 23. Tzianabos, T., B. E. Anderson, and J. E. McDade. 1989. Detection of Rickettsia rickettsii DNA in clinical specimens using polymerase chain reaction technology. J. Clin. Microbiol. 27:2866-2868. 24. Upholt, W. B. 1977. Estimation of DNA sequence divergence from comparison of restriction endonuclease digests. Nucleic Acids Res. 4:1257-1265. 25. Varela, G., J. W. Vinson, and C. Molina-Pasquel. 1969. Trench fever. II. Propagation of Rickettsia quintana on cell-free medium from the blood of two patients. Am. J. Trop. Med. Hyg. 18:708-712. 26. Vinson, J. W. 1966. In vitro cultivation of the rickettsial agent of trench fever. Bull. W.H.O. 35:155-164.

274

REGNERY ET AL.

27. Weisburg, W. G., M. E. Dobson, J. E. Samuel, G. A. Dasch, L. P. Mallavia, 0. Baca, L. Mandelco, J. E. Sechrest, E. Weiss, and C. R. Woese. 1989. Phylogenetic diversity of the rickettsiae. J. Bacteriol. 171:4202-4206. 28. Weisburg, W. G., C. R. Woese, M. E. Dobson, and E. Weiss. 1985. A common origin of rickettsiae and certain plant pathogens. Science 230:556-558.

J. CLIN. MICROBIOL. 29. Welch, D. F., D. A. Pickett, L. N. Slater, A. G. Steigerwalt, and D. J. Brenner. 1991. Rochalimaea henselae sp. nov., a cause of septicemia, bacillary angiomatosis, and parenchymal bacillary peliosis. J. Clin. Microbiol. 30:275-280. 30. Wilson, K. H., R. B. Blitchington, and R. C. Greene. 1990. Amplification of bacterial 16S ribosomal DNA with polymerase chain reaction. J. Clin. Microbiol. 28:1942-1946.