Mycobacterium liflandii Outbreak in a Research Colony of Xenopus (Silurana) tropicalis Frogs

Abstract

A research colony of Xenopus (Silurana) tropicalis frogs presented with nodular and ulcerative skin lesions. Additional consistent gross findings included splenomegaly with multiple tan-yellow nodular foci in the spleen and liver of diseased frogs. Copious acid-fast positive bacteria were present in touch impression smears of spleen, skin, and livers of diseased frogs. Histologically, necrotizing and granulomatous dermatitis, splenitis, and hepatitis with numerous acid-fast bacilli were consistently present, indicative of systemic mycobacteriosis. Infrequently, granulomatous inflammation was noted in the lungs, pancreas, coelomic membranes, and rarely reproductive organs. Ultrastructurally, both extracellular bacilli and intracellular bacilli within macrophages were identified. Frogs in the affected room were systematically depopulated, and control measures were initiated. Cultured mycobacteria from affected organs were identified and genetically characterized as Mycobacterium liflandii by polymerase chain reaction amplification of the enoyl reductase domain and specific variable numbers of tandem repeats. In recent years, M. liflandii has had a devastating impact on research frog colonies throughout the United States. This detailed report with ultrastructural description of M. liflandii aids in further understanding of this serious disease in frogs.

Xenopus species have been a popular laboratory amphibian model since the 1950s. The Western clawed frog, Xenopus (Silurana) tropicalis, has become increasingly popular over the traditionally used African clawed frog, Xenopus laevis. X. (Silurana) tropicalis is the only member of its species to have a diploid genome, which facilitates comparative genomic analysis. The production of transgenic frogs involves integration of DNA into sperm nuclei followed by transplantation into unfertilized eggs, which results in hundreds of transgenic, nonchimeric embryos that do not require additional breeding. The shorter generation time of 4 to 6 months for X. (Silurana) tropicalis in comparison to 1 to 2 years for X. laevis makes transgenic experiments even more efficient.² With the recent availability of the genomic sequence of X. (Silurana) tropicalis and its annotation, the importance of this species in research, specifically in the fields of comparative genetics and vertebrate developmental biology, is poised to increase even more in the future.¹⁸

Laboratory X. (Silurana) tropicalis are obtained from vendors, usually as wild caught, or they are bred in-house from an original wild-caught stock. Wild-caught frogs can potentially harbor various bacterial, viral, mycotic, or parasitic infectious agents, some of which may be subclinical.⁸ However, when frogs are subjected to a laboratory environment, these diseases may become clinically apparent and cause morbidity and mortality. A common clinical sequela to disease in Xenopus (Silurana) frogs is bloating caused by whole-body edema, which can be infectious or noninfectious in origin.⁸ An ubiquitously recognized infectious disease in Xenopus sp. is bacterial dermatosepticemia, also known as red leg syndrome, causing cutaneous erythema, edema, epidermal erosions/ulcers, dermal necrosis, anorexia, and coelomic effusions. Bacterial agents implicated in this disease include gram-negative bacteria such as Aeromonas hydrophila, other aeromonads, pseudomonads, and enterobacteria such as Citrobacter, Proteus, and Salmonella, as well as gram-positive bacteria such as Streptococcus and Staphylococcus.⁸ Other bacteria such as Flavobacterium meningosepticum have been associated with septicemic disease in a colony of X. laevis frogs. Clinical signs included ascites, anasarca, dyspnea, extreme lethargy, congestion of web vessels, petechial hemorrhages, and sudden death.¹⁶

Chlamydia (Chlamydophila) pneumoniae was diagnosed in a breeding colony of X. (Silurana) tropicalis frogs of which 91% died within 4 months of arrival from a vendor. Clinical signs included lethargy, sloughing of the skin, and bloating. The affected livers had lesions consistent with lymphohistiocytic hepatitis with individual hepatocyte necrosis, fibrosis, and numerous granular basophilic bodies resembling bacteria in the cytoplasm of hepatocytes and histiocytes. Ultrastructually, typical chlamydial forms (reticulate, elementary, and intermediate bodies) were identified within vacuoles in hepatocytes and macrophages. The disease was further complicated by a coinfection with a chytridiomycete fungus in a portion of the affected frogs.²⁷ The chytridiomycete fungus, named Batrachochytrium dendrobatidis, is now recognized as a significant cause of mortality and a decline in global amphibian populations, including various species of frogs.¹⁰ In an epizootic of chytridiomycosis involving 55 adult male X. (Silurana) tropicalis frogs maintained in a laboratory setting at the University of California, Berkeley, epidermal lesions induced by B. dendrobatidis were identified as the primary reason for nearly 80% mortality in the affected frogs. Clinically, all diseased frogs typically had anorexia, lethargy, dark skin pigmentation, and excess skin sloughing with a lack of slime layer. Histologically, lesions were consistent with severe hyperplastic and spongiotic dermatitis and colonization of the stratum corneum by large numbers of zoosporangia of B. dendrobatidis.²⁴

Cryptosporidium sp. causing emaciation and proliferative gastritis were identified in X. laevis frogs. The source of the infection was not identified.¹⁴ Xenopus (Silurana) sp. are also susceptible to infections with helminths such as the cutaneous capillarid Pseudocapillaroides xenopi. Members of a colony of X. laevis frogs exhibited wasting syndrome characterized by skin color changes, roughed texture of the skin due to loss of mucus, weight loss, and decreased production of oocytes by females.⁶ Noninfectious laboratory stressors can also induce clinical disease in Xenopus (Silurana) species mimicking those seen with infectious diseases. X. laevis frogs having received human chorionic gonadotropin (hCG) for ovarian stimulation developed subcutaneous edema, coelomic distention, and bloating. Internally, the frogs exhibited ascites and large quantities of free-floating oocytes in the coelonic cavity.¹³

Nontuberculous mycobacteria (NTM) are free-living, acid-fast bacterial species that are common in aquatic environments.²⁰ Mycobacterium marinum, a common fish pathogen,³ is responsible for intracellular granulomatous infections in fish and humans.²⁶ A genetic relative to M. marinum, Mycobacterium ulcerans, causes an extracellular and intracellular infection known as Buruli ulcer in humans. Sequencing of the M. marinum and M. ulcerans genome confirms M. ulcerans evolved from a common M. marinum progenitor.³⁷ These 2 species differ by a single base pair in the 16S rRNA gene²² in addition to a plasmid-encoded toxic polyketide-derived macrolide, mycolactone, and an insertion sequence, IS2404, both of which were believed to be absent from M. marinum.³⁶ However, this insertion sequence, as well as a unique mycolactone, has been recently identified in an unusual clade of M. marinum isolated from fish and frogs.^26,36,37

Multiple species of NTM have been isolated from frogs, including M. marinum,^9,25 Mycobacterium chelonae,¹⁵ Mycobacterium gordonae,²⁸ Mycobacterium szulgai,⁷ Mycobacterium xenopi,²⁹ and Mycobacterium liflandii.^12,34 M. liflandii was first isolated in a 2001 outbreak of mycobacteriosis in a colony of X. (Silurana) tropicalis frogs.³⁴ M. liflandii has since been responsible for devastating outbreaks in research Xenopus frog colonies across the United States and in a European colony of the tropical clawed frog, X. (Silurana) tropicalis.^12,32 M. liflandii is similar to M. ulcerans in that both contain an IS element, IS2404, used for strain typing and once thought unique to M. ulcerans.^22,31,36 Neither can grow at temperatures above 35°C, and although M. liflandii also encodes mycolactone, there are currently no reports of it causing disease in humans.²²

M. ulcerans infection in humans causes a necrotizing and ulcerative dermatitis and panniculitis with extension into the underlying muscle and bone. Systemic spread of infection is uncommon. In the initial stages, histological features include extensive tissue necrosis, large clusters of extracellular bacteria, and sparse inflammation (macrophages and neutrophils). The disease progresses to the organizing stage, which involves large numbers of macrophages, plasma cells, and lymphocytes at the margins of necrotic zones as well as fibroblasts and capillaries growing into these zones. During the final healing stage, only scant bacteria are present.^17,30 M. ulcerans was initially thought to be an extracellular pathogen due to the presence of large numbers of extracellular bacteria in areas of necrosis with sparse inflammatory cells, although increasing ultrastructural evidence suggests an intracellular growth phase within macrophages.^17,30,33 The related pathogen, M. liflandii, causes a spectrum of lesions in frogs that follows a general pattern of ulcerative to granulomatous dermatitis; visceral granulomas primarily involving the liver, spleen, kidneys, and ovaries; coelomitis; skeletal myositis; septicemia; and coelomic effusion in some cases. Large numbers of free and intrahistiocytic acid-fast bacilli with prominent necrosis and a variable population of inflammatory cells were often associated with the granulomatous lesions.^12,32,34

In this report, we describe an outbreak of M. liflandii in a research colony of X. (Silurana) tropicalis frogs housed in individual tanks within a single room. Although infection of various species of frogs with M. liflandii has been previously reported, this report thoroughly investigates M. liflandii infection in a colony of X. (Silurana) tropicalis frogs via clinical findings, gross examination, histopathology, culture, electron microscopy, and molecular characterization.

Materials and Methods

All Xenopus frogs evaluated as part of this report were originally purchased from a domestic vendor and maintained in an Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC)–accredited facility. The frogs were routinely propagated for developmental research studies involving hormone priming followed by manual expression of oocytes. All research protocols were compliant with the US Public Health Service (USPHS) Policy on Humane Care and Use of Laboratory Animals and approved by the Massachusetts Institute of Technology Committee on Animal Care. The frogs were housed in 3 rooms named A, B, and C for the purpose of this report. In room A, X. (Silurana) tropicalis frogs were divided and maintained in 7 individual static tanks and used for oocyte collection procedures. Frogs from room B were rederived from gentamicin-treated oocytes expressed from X. (Silurana) tropicalis frogs in room A. The X. (Silurana) tropicalis offspring were maintained in an interconnected automated recirculation system. In room C, X. laevis frogs were also maintained in an interconnected automated recirculation system in a manner similar to that of room B frogs. A water temperature range of 80.6 to 84.2°F was maintained in the X. (Silurana) tropicalis rooms by use of a space heater in room A and by means of an internal thermostat within the water recirculation system in room B. All frogs were housed in reverse-osmosis water pretreated with Instant Ocean salt (60 g/100 L) and neutral buffer (10 g/100 L). Water quality parameters included conductivity 1.4 ± 0.1 and pH 6.5 ± 0.3. In room A, growing frogs were fed every day and water changed 2 to 3 times per week, and adult frogs were fed every other day and water was changed twice per week. Frogs were fed Purina Trout Chow, and remaining food following feeding was siphoned from the tanks.

Frogs in various rooms were routinely monitored during the course of the outbreak. Water and pH recordings were periodically analyzed, and all visual inspections and recording of clinical signs were performed by clinical veterinarians (JFR, CE). The range of clinical signs evaluated included coelomic distension, presence of skin lesions, and/or mortality.

Frogs were periodically sampled and submitted for necropsy following euthanasia via immersion in 1% benzocaine for 45 minutes followed by decapitation. Representative sections of all major organs were collected and fixed in 10% neutral buffered formalin overnight. Samples were processed by routine paraffin embedding, sectioned at 4 µm, and stained by hematoxylin and eosin (H&E), Ziehl-Neelsen (ZN), and Fite’s acid-fast methods. Touch impression smears of various organs taken at necropsy were air dried, ethanol fixed, and stained by ZN acid-fast method. Fresh tissue samples were collected for bacterial culture and stored at −80° C in freeze media until further analysis. Representative samples of liver from 2 diseased frogs fixed initially in either 10% neutral buffered formalin or 1.5% glutaraldehyde were submitted for electron microscopy. The samples were refixed in 2.5% glutaraldehyde and 3% paraformaldehyde with 5% sucrose in 0.1M sodium cacodylate buffer (pH 7.4), washed with 0.1M sodium cacodylate buffer 3×, and post fixed in 1% OsO₄ in veronal-acetate buffer (Palade’s osmium, pH 6.0) for 1 hour. The samples were stained in block overnight with 0.5% uranyl acetate in veronal-acetate buffer (Kellenberger’s UA) and then dehydrated and embedded in Spurr’s resin. Sections were cut on a Leica Ultracut UCT microtome with a Diatome diamond knife at a thickness setting of 50 nm and stained with 2% uranyl acetate and lead citrate. The sections were examined using an FEI Tecnai Spirit at 80 kV.

Selected tissue samples of liver and spleen collected from frogs during necropsy were submitted to the Massachusetts State Diagnostic laboratory for culture and to the University of Tennessee for culture and species identification. At the University of Tennessee, diseased frog tissues were inoculated onto Lowenstein-Jensen (LJ) media. Media were incubated at 32°C with 5% CO₂. Bacterial colonies from affected frog tissue were observed on LJ media within 30 days of incubation. DNA was extracted from affected tissues and cultured isolates using methods previously described.³⁶ Briefly, tissues were lysed in 400 µl lysis solution (100 mM Tris, 50 mM NaCl, 1.33% SDS, 0.2 mg/ml RNase A) and 1 g of 1.0-mm glass beads. Following centrifugation, the supernatant was added to 5M potassium acetate and incubated at −20° C overnight. Following centrifugation, supernatants were added to 0.66M guanidine hydrochloride and 63.3% ethanol. DNA was purified using a MOBIO spin filter and 2 washes consisting of 500 µl wash solution (10 mM Tris, 1 mM EDTA, 50 mM NaCl, 67% ethanol) and 500 µl ethanol. DNA was eluted in 200 µl elution solution (10 mM Tris) and stored at −20° C until further analysis. Genetic characterization was performed by polymerase chain reaction (PCR) amplification on isolates from 3 samples (case No. 16 spleen, case No. 6 spleen, and case No.17 spleen) of the enoyl reductase (ER) domain found on the mycolactone plasmid found in M. ulcerans and M. liflandii and by probing for specific variable numbers of tandem repeats (VNTR). Primers and PCR amplification conditions used were as previously described for the ER domain and VNTR locus MIRU1, locus 6, ST1, and locus 19.³⁶

Ethanol-soluble lipids (ESLs) were extracted from samples as previously described³⁶ by first pelleting cells isolated from diseased frog tissue and then stirring cells in ethanol for approximately 2 hours. ESLs were separated from the pellet by centrifugation, transferred to a new glass tube, and dried under nitrogen. ESLs of 2 samples (case Nos. 16–17 spleens) were resuspended in ethanol and then ran using thin-layer chromatography (TLC) by spotting 5 µl of ESLs onto a silica gel plate. M. liflandii XL5 (5 µl) was used as a control. The plate was placed in a TLC chamber with ~75 ml 90:10:1 chloroform/methanol/water. The TLC was allowed to run until ~4 cm from the top of the plate. The plate was removed from the chamber, and the distance traveled by the solvent front was marked with a pencil. The plate was developed by dipping it into ceric sulfate/ammonium molybdate in 2M sulfuric acid followed by high heat application. Rf values were defined as the distance traveled by the samples (or control) divided by the distance traveled by the solvent.

Results

Clinical Signs and Disease Progression

In December 2008, the first observation of clinical disease with associated mortality was recorded in one of the tanks (#3) of room A. Multiple frogs in the tank exhibited nodular to ulcerative skin lesions. Three frogs with visible skin lesions were selected for detailed postmortem evaluation (case Nos. 1–3). Touch impression smears of spleen, liver, and skin were assessed by ZN acid-fast stain, and samples were submitted for bacterial culture. A preliminary diagnosis of nontuberculous (atypical) mycobacteriosis was made based on the presence of acid-fast bacilli in impression smears and histological sections, and the affected tank was depopulated. Approximately 40 to 45 days later, frogs with similar skin lesions were noted in 2 more tanks in room A (#2 and #7). Four frogs were necropsied from tank 2; 3 had visible skin lesions, and 1 frog was clinically normal. Over the next few weeks (70–80 days following the index cases), the disease eventually spread to all 7 tanks in the affected room. By mid-March 2009, all tanks in room A were depopulated and disinfected. During the course of the outbreak, a total of 20 frogs were selected from room A for detailed postmortem evaluation, of which 12 were individually identified to be clinically diseased, and the remaining 8 were classified as clinically normal. Clinical signs observed during the course of the outbreak in room A included poor body condition (1/20 examined frogs, case No. 8—only presenting sign), skin lesions (9/20 examined frogs), coelomic distension/generalized edema (2/20 examined frogs), and death.

Daily water quality logs were available only for 2 of the 7 tanks in the affected frog room. It was noted during the month of the initial outbreak that the temperature of the water in one of the affected X. (Silurana) tropicalis tanks ranged between 65.66 and 68.36°F. The disease outbreak was restricted to room A, and frogs housed in room B (rederived X. [Silurana] tropicalis) and room C (X. laevis) with automated recirculation systems have remained unaffected for more than a year since the first evidence of mycobacteriosis in frogs at our institution.

Gross Pathology

Frogs were considered positive for mycobacteriosis if they had one or more typical clinical, gross, or histological lesions coupled with the presence of acid-fast bacilli within lesions and/or a positive culture/PCR for Mycobacteria spp. Significant gross findings in diseased frogs included multifocal nodular and ulcerative dermal lesions (9/12 positive cases), splenomegaly with splenic granulomas (10/12 positive cases), and hepatic granulomas (9/12 positive cases). The skin lesions were multifocal, primarily involving the limbs, the dorsum, and ventral abdomen. Appearance ranged from discrete, raised, tan-red nodules (0.5–2.5 mm diameter; Fig. 1); punctate to rectangular, cavitating skin ulcers (0.5–5 mm diameter) with associated hyperemia (Fig. 2); to smooth, tan-yellow, glistening skin abrasions (Fig. 3). Splenomegaly (2–5 times the size of unaffected frogs) was the most consistent lesion in positive cases, and on sectioning, the splenic parenchyma had distinct, variably sized, multifocal to coalescing tan-yellow foci (Figs. 4, 5). Grossly, the affected livers had multiple discrete to coalescing, tan-white, variably sized (0.5–2 mm diameter), slightly raised to flat foci on both the superficial and cut surfaces (Fig. 5). Multifocal, slightly raised, white pinpoint foci were also noted in the lungs of 2 affected frogs. The disease process generally involved either one or few organs (skin, liver, spleen, lungs, etc.), but was occasionally disseminated (2/12 positive cases), involving skin and most viscera, including liver, spleen, lungs, pancreas, kidneys, peritoneum, and reproductive organs, with associated coelomic distension and effusion (Fig. 5). In one frog (case No. 8) from room A, poor body condition was the only gross finding. Two clinically normal rederived X. (Silurana) tropicalis frogs from the unaffected room B and one X. laevis frog from room C evaluated as part of the screening process did not exhibit any significant gross lesions.

Figure 1.

Skin; Xenopus (Silurana) tropicalis frog, case No. 1. Nodular, tanred, discrete foci (arrows) on skin of the dorsum.

Figure 2.

Skin; Xenopus (Silurana) tropicalis frog, case No. 2. Punctate to triangular, red, ulcerated skin lesions on the ventral abdomen and limbs.

Figure 3.

Skin; Xenopus (Silurana) tropicalis frog, case No. 2. Tan, glistening skin ulcer (arrow). Inset: Higher magnification of ulcer.

Figure 4.

Abdominal viscera; Xenopus (Silurana) tropicalis frog, case No. 3. Splenomegaly with tan discoloration.

Figure 5.

Abdominal viscera; Xenopus (Silurana) tropicalis frog, case No. 17. Disseminated disease tan-yellow foci (granulomas) in the spleen, liver, and ovary (star) with hepatosplenomegaly and coelomic effusion.

Cytology and Histopathology

Histologically, the skin, spleen, and liver of frogs with gross lesions were consistent with necrotizing and granulomatous (with or without ulceration) dermatitis (Figs. 6, 7), necrogranulomatous splenitis (Figs. 9, 10), and necro-granulomatous hepatitis, respectively (Fig. 14). These discrete, granulomatous foci were localized to the regions of grossly visible foci in various organs and were characterized by a prominent central core of extensive necrosis admixed with large numbers of extracellular bacilli that stained faintly by H&E (Fig. 10) and magenta red by the ZN acid-fast method (Figs. 8, 11) and Fite’s acid-fast stain (not shown). The necrotic cores, especially in the spleen, liver, and the skin (to a lesser extent), were encircled by low to moderate numbers of pale staining to vacuolated round to polygonal cells (foamy vacuolated histiocytes/epithelioid macrophages) admixed with small numbers of lymphocytes and granulocytes. Touch impression smears of the spleen (Figs. 12, 13) and, to a lesser extent, skin and liver contained large numbers of acid-fast extracellular and intracellular bacteria. Numerous intracellular bacteria were often present within vacuoles of foamy macrophages (Fig. 13). Multinucleated cells were not present. The nodular granulomas in the skin were paucicellular and extended from the subdermal adipose tissue into the overlying dermis with associated collagenolysis, lymphangitis, epidermal degeneration/necrosis, and mild reactive hyperplasia. The deeper margins of the cutaneous granulomas often had numerous engorged blood vessels and mild fibroplasia (Fig. 6). In skin samples with ulceration, the granulomatous inflammation was localized along the periphery of the ulcers with extension into the underlying musculature (myositis; not shown). Highest bacterial loads were present in the skin and spleen followed by liver with lesser numbers in other organs. In the lungs of a few diseased animals, the pulmonary serosa and the pulmonary interstitium were thickened by multifocal, often perivascular, variably sized aggregates of low to moderate numbers of foamy macrophages and few lymphocytes (Fig. 15). However, typical nodular granulomas were infrequent in the lungs. Rarely, intracellular acid-fast bacilli within circulating cells (monocytes) were also observed (not shown). In some (5/12) positive cases, the inflammatory process also extended into the pancreas with architectural effacement in the form of mild to moderate, multifocal to coalescing, necrogranulomatous pancreatitis (Fig. 16) and variable numbers of acid-fast bacilli. Necrogranulomatous coelomitis was also present in some disseminated cases. One female frog (case No. 17) had multifocal, necrogranulomatous salphingitis and oophoritis (Fig. 17) with intralesional acid-fast bacilli. Testicular involvement with the presence of acid-fast bacilli was observed in a diseased male frog (case No. 16). The kidneys of most diseased frogs and some clinically normal frogs frequently had small, multifocal, lymphohistiocytic, interstitial inflammatory aggregates with variable tubular degenerative changes. Acid-fast bacilli were usually not present in the kidneys except for low numbers in one frog with disseminated disease (case No. 17).

Figure 6.

Skin; Xenopus (Silurana) tropicalis frog, case No. 2, hematoxylin and eosin. A typical Mycobacterium liflandii–induced skin nodule with origin in the subdermal adipose tissue diagnosed as necrotizing and granulomatous dermatitis and lymphangitis with associated neovascularization along the ventral margins.

Figure 7.

Skin; Xenopus (Silurana) tropicalis frog, case No. 2, hematoxylin and eosin. Higher magnification from Figure 6 showing prominent necrosis and low cellularity.

Figure 9.

Spleen; Xenopus (Silurana) tropicalis frog, case No. 2, hematoxylin and eosin. Large, discrete granuloma with prominent necrotic core.

Figure 10.

Spleen; Xenopus (Silurana) tropicalis frog, case No. 2, hematoxylin and eosin. Higher magnification of Figure 9 showing the necrotic core with clusters of faintly staining bacteria (arrowhead) and a peripheralcellular zone (arrow) composed of inflammatory cells (foamy/vacuolated macrophages, granulocytes).

Figure 14.

Liver; Xenopus (Silurana) tropicalis frog, case No. 3, hematoxylin and eosin. Focal, discrete, necrogranulomatous lesion in liver with a central necrotic core bordered by a pale, cellular layer of inflammatory cells (macrophages).

Figure 8.

Skin; Xenopus (Silurana) tropicalis frog, case No. 2, Ziehl-Neelsen stain. Lowpower magnification showing dense clusters of acid-fast bacilli (arrows) within the necrotic core. Inset: Numerous intraand extracellular acid-fast bacilli.

Figure 11.

Spleen; Xenopus (Silurana) tropicalis frog, case No. 2, serial section of (e) showing numerous acid-fast positive bacilli (arrow head) within the necrotic core and relative paucity at the margins in the cellular zone.

Figure 12.

Spleen; Xenopus (Silurana) tropicalis frog, case No. 3, touch impression smear, Ziehl-Neelsen stain. Lowmagnification image showing prominent pink staining indicative of large numbers of acid-fast positive bacilli.

Figure 13.

Spleen; case No. 3, touch impression smear, Ziehl-Neelsen stain. Oil objective image showing both extracellular and intracellular acid-fast positive bacilli.

Figure 15.

Lung; Xenopus (Silurana) tropicalis frog, case No. 16, hematoxylin and eosin. Expansion and thickening of the pulmonary interstitium by disorganized aggregates of macrophages and granulocytes with minimal necrosis.

Figure 16.

Pancreas/coelom; Xenopus (Silurana) tropicalis frog, case No. 10, hematoxylin and eosin. Extensive necrosis with associated hemorrhage and inflammation in the pancreas with extension into adjacent coelomic space.

Figure 17.

Ovary; Xenopus (Silurana) tropicalis frog, case No. 17, hematoxylin and eosin. Expansion of the ovarian stroma by necrosis and associated moderate granulomatous inflammation with follicular atrophy/loss.

Clinically, normal frogs from affected tanks in room A (8/20 frogs), as well as from unaffected rooms B (2 X. [Silurana] tropicalis frogs) and C (1 X. laevis frog), did not have typical granulomatous lesions in all examined tissues, but occasionally, small aggregates of lymphohistiocytic inflammatory cells were present in a few organs. Acid-fast bacilli were not demonstrated in any of these 3 frogs. In addition, the livers of some diseased as well as nondiseased frogs had increased melanomacrophages with mild to moderate extramedullary hematopoiesis. Moderate lymphoid hyperplasia was also seen in the spleen of a few diseased frogs.

Ultrastructural Findings

Electron microscopic evaluation of the liver of one frog (case No. 17) with distinct nodular granulomas demonstrated numerous intracellular and extracellular bacteria within epithelioid macrophages/polygonal cells with associated activation and/ or cellular degenerative/necrotic changes (Fig. 18). These changes included cellular and membrane swelling, membrane blebbing, loss of cytoplasmic density, organelle disorganization, loss of Rough Endoplasmic Reticulum (RER), mitochondrial swelling, cytoplasmic vacuolation (membrane-bound phagolysosomes/lipid), presence of membranous whorls, nuclear swelling, chromatin margination and loss, and nuclear dissolution. On high magnification, the intracellular bacilli measured 0.25 to 0.5 mm in width and 1.5 to 2.5 mm in length and were present within phagosomal vacuoles or free in the cytosol (Figs. 19, 20). The cytosol of the bacilli contained a round to oval, central or subterminal, electron-dense body. In both longitudinal and cross sections, multiple bacilli were often noted within a single cell, usually inside a fused secondary lysosome (Fig. 20). In some bacilli, a distinct cell wall was discernible, especially in those undergoing degenerative changes, with contraction of the bacterial cytoplasm leaving a clear space beneath the cell wall (Fig. 20). In another acid-fast positive liver sample from a frog with no obvious nodule formation, bacteria were not identified by ultrastructural evaluation.

Figure 18.

Liver; Xenopus (Silurana) tropicalis frog, case No. 17, electron microscopy. Lowmagnification ultrastrucutural image of a segment of Mycobacterium liflandii granuloma with extracellular and intracellular bacteria within epithelioid cells (macrophages). Note cellular degenerative changes with cell swelling, membrane blebbing, vacuolation, and loss of cytoplasmic and membrane detail.

Figure 19.

Liver; Xenopus (Silurana) tropicalis frog, case No. 17, higher magnification image of Figure 18 showing multiple Mycobacterium liflandii organisms (arrowheads) within a macrophage in longitudinal sections mostly free in the host cell cytoplasm. Note the indistinct bacterial cell membrane with the cytosol of the bacilli containing a central to subterminal, prominent, round to oval electrondense material (arrow).

Figure 20.

Liver; Xenopus (Silurana) tropicalis frog, case No. 17. Another higher magnification image depicting multiple intracellular bacilli in both longitudinal and cross sections within phagolysosomal vacuoles (long arrow) and with formation of a membranous whorl (short arrow). Note the prominent bacterial cell wall in cross sections of some degenerate bacteria that have shrunken cytoplasm (arrowhead).

Mycobacterium liflandii Isolation and Molecular Characterization

In an initial analysis of samples performed at the Massachusetts state diagnostic laboratory, tissues from 2 frogs (case Nos. 2–3) tested positive for mycobacteria by fluorochrome staining, and the spleen of 1 frog (case No. 1) showed growth of mycobacterial colonies after 90 days; however, the cultures became nonviable, and further analysis was not performed. At the University of Tennessee, frog tissue samples selected from the affected room A were analyzed: case No. 3 (liver), case No. 6 (spleen), case No. 8 (spleen), case No. 14 (spleen), case No. 16 (spleen), and case No. 17 (liver and spleen). Splenic tissue from 1 frog (case No. 5) from the unaffected room B was also analyzed. Samples from frog case No. 8 (liver, spleen, and kidneys) sourced from a tank that was considered unaffected at the time of clinical evaluation were acid-fast negative (by both ZN and Fite’s method) on histological examination. Interestingly, all samples from affected room A, including the acid-fast negative liver from case No. 8, were found to be positive for M. liflandii by culture and PCR. The splenic sample from case No. 5 from the rederived X. (Silurana) tropicalis room B was negative for mycobacterial growth. Representative data shown from 3 isolates (case No. 6, case No. 16, and case No. 17) depict a positive ER signal (Fig. 21) as well as VNTR profiles matching that of M. liflandii control (Figs. 22–25). The ESL TLC profiles of 2 selected isolates (case Nos. 16–17; Fig. 26) showed that the Rf value for mycolactone E was 0.54 as compared to 0.56 of the M. liflandii control mycolactone E from the University of Tennessee. This minor difference of 0.2 between our isolates and the control was considered an artifact arising from loading the samples too close to the edge of the silica plate, creating an “edge or solvent effect.”

Figure 21.

Polymerase chain reaction (PCR) targeting the enoyl reductase (ER) domain of the mycolactone plasmid of Xenopus (Silurana) tropicalis frog mycobacterial isolates obtained by culturing tissue samples from diseased frogs. All isolates were ER positive.

Figure 22.

Polymerase chain reaction (PCR) targeting of MIRU1 locus of variable number of tandem repeats (VNTR) of selected isolates from Xenopus (Silurana) tropicalis (case Nos. 16, 6, and 17) in comparison to Mycobacterium ulcerans (MU), mycolactoneproducing Mycobacterium marinum isolates (MMDL, MMBB), and Mycobacterium liflandii (M. lif). All 3 frog isolates had a VNTR PCR profile matching that of M. liflandii.

Figure 25.

Polymerase chain reaction (PCR) targeting of locus 19 of variable number of tandem repeats (VNTR) of selected isolates from Xenopus (Silurana) tropicalis (case Nos. 16, 6, and 17) in comparison to Mycobacterium ulcerans (MU), mycolactoneproducing Mycobacterium marinum isolates (MMDL, MMBB), and Mycobacterium liflandii (M. lif). All 3 frog isolates had a VNTR PCR profile matching that of M. liflandii.

Figure 26.

Thinlayer chromatography of extracted ethanolsoluble lipids from Mycobacterium liflandii–infected frog tissue specimens. The Rf values of the diseased samples (case Nos. 16–17, left and middle lanes) were slightly lower than the positive M. liflandii control (right lane). However, the difference between the samples and the positive control is only .02. M. liflandii isolated from both samples (case Nos. 16–17) contained mycolactone E.

Control Measures

The affected room A was quarantined as soon as the first case was identified to prevent further disease transmission to the other Xenopus rooms, and the laboratory personnel were warned about the potential for zoonosis and the need for appropriate personal protective equipment while working with the frogs. The room was eventually depopulated and disinfected with vaporized hydrogen peroxide; tanks and related equipments were appropriately disposed of.

Discussion

The NTM mycobacterium, M. liflandii, has been increasingly recognized as a cause of disease outbreaks in different species of laboratory frogs at multiple research institutions throughout the world. Various descriptions of the disease, including this report, are similar in terms of clinical picture, morbidity, pathology, and molecular characteristics with some minor variations. In our outbreak, typical necro-granulomatous lesions were consistently observed in the spleen, skin, and liver with lesser degree of involvement of the pancreas, lungs, coelom, and reproductive organs. In the first described M. liflandii outbreak in X. (Silurana) tropicalis,³⁴ the skin lesions were mostly paucimicrobial, unlike visceral granulomas (liver, spleen, and kidneys). However, in this outbreak, distinct granulomas with a high intralesional bacterial load were consistently observed in the skin and spleen of diseased frogs. In addition, both nodular and ulcerative skin lesions were present in many affected frogs, which was indicative of a sequential progression from an initial deep, subdermal, necro-granulomatous, nodular lesion to an ulcerative phase, similar to Buruli ulcers in humans.¹⁷ In M. liflandii infection in a European colony of clawed frogs X. (Silurana) tropicalis, numerous acid-fast bacteria were frequently observed in the kidneys, ovaries, and oviductal lumen as well as the lumens of the stomach, gallbladder, and intestinal tract, but the liver, lungs, and tibia were unaffected and free of acid-fast bacilli.³² In our report, ovarian or testicular involvement with detectable acid-fast organisms was present in 2 frogs, thus raising the possibility of mycobacterial transmission by oocytes/sperm used in embryo transfer protocols.

Different species of mycobacteria tend to incite granulomatous inflammation with varying morphological patterns depending on the host species affected, host genetics, bacterial virulence factors, immune status of the host, dynamics of bacterial-host interplay, duration/phase of infection, coinfections if any, therapy/vaccination, and so on.^1,5,17,20,35 In general, inflammation caused by mycobacteria can lead to formation of typical discrete, well-organized granulomas (tuberculoid granulomas) as seen in infections caused by members of the Mtb complex (ie, Mycobacterium tuberculosis in humans) as well as a polar tuberculoid form of leprosy in humans caused by Mycobacterium leprae.^1,35 Tuberculoid granulomas are usually a result of strong cell-mediated immunity (CMI)—specifically, a form of delayed type hypersensitivity (DTH)—and morphologically may be simple or complex in organization, usually paucibacillary, with or without central necrosis (caseation), and composed predominantly of histiocytes, epithelioid macrophages, and multinucleated giant cells (Langhan’s type). The macrophages and giant cells are in turn encircled by variable numbers of lymphocytes, plasma cells, and a peripheral rim of fibroblasts. Lepromatous granulomas arise in the context of robust humoral immunity but poor CMI responses to mycobacterial antigens. They typically lack structural organization and are characterized by diffuse infiltration mostly of macrophages (foamy, granular, or vacuolated types) with numerous intracytoplasmic acid-fast bacilli (multibacillary), sparse lymphocytes, and rare to none giant cells.^1,35 NTM usually cause a range of granuloma patterns in humans and often are characterized by prominent necrosis and large numbers of bacilli. Multinucleated giant cells may be seen as in M. ulcerans granulomas of humans.^4,17,20,30 Similarly, M. liflandii granulomas in frogs, as observed in this report and earlier studies,^9,25,26,28 are usually characterized by a mixed granuloma pattern with some features of tuberculoid granuloma, such as the presence of discrete nodules and prominent central necrosis, but also share characteristics of the lepromatous type, as evidenced by the presence of large numbers of bacilli (extracellular and intracellular) and lack of giant cells. Lymphocytes, granulocytes, and fibroplasia may be variably observed in M. liflandii infection in frogs. The failure to induce giant cell formation during M. liflandii infection might be due to an inherent poor ability of activated amphibian macrophages to fuse and form multinucleated giant cells. Not surprisingly, this lack of giant cells is also a feature of granulomas of M. marinum and M. gordonae in various species of frogs.^5,25

The ultrastructural features of M. liflandii granulomas, as observed in this report by the presence of numerous extracellular bacilli within necrotic areas as well as intracellular bacilli within degenerate to viable infected cells (histiocytes/epithelioid macrophages), are suggestive of the prominent cytotoxic nature of the intracellular infection. Within epithelioid cells, multiple bacilli were seen inside vacuoles (bacterial phagosomes/phagolysosomes) and also free in cytoplasm indicative of active bacterial replication within phagosomes and its resultant rupture and release of bacteria into the cytosol. The presence of degenerate bacteria within some vacuoles of phagocytes is also suggestive of an ongoing concomitant though inefficient phagolyosomal activation/fusion mechanism to inhibit intracellular bacterial replication. In our study, although fine details of the bacterial cell wall were not visualized by electron microscopy, the general morphological appearance of M. liflandii as well as associated cytological alterations identified in this report closely resembles that of M. ulcerans in humans³⁰ and M. marinum in frogs.⁵

Mycolactone is the major virulence determinant of several species of mycobacterium,²³ including M. liflandii. It is encoded from a large megaplasmid containing 3 essential genes for mycolactone synthesis.¹⁹ The lactone core is encoded by mlsA1and mlsA2 genes.²² Three copies of an ER domain are present in mlsA1 and one in mlsA2. There are 4 to 8 copies of the ER domain/bacterial cell assuming a plasmid copy number of 1 to 2. Because the sequence is present in sufficient copy number to provide high sensitivity as well as approximate calculations of genome equivalents, ER sequence is commonly targeted for PCR analysis to identify mycolactone-producing species of mycobacteria. The fatty acid side chain, which is unique to a specific mycolactone, is encoded by the mlsB gene.¹⁹ The molecular structure of the fatty acid side chain determines the cytopathogenicity of a particular mycolactone. Mycolactone E, isolated from M. liflandii, was demonstrated as less cytopathic to murine L929 fibroblasts compared to mycolactone A/B. The difference in cytopathogenicity may be attributed to the absence of a hydroxyl group at C′12 in the fatty acid side chain of mycolactone A/B.²² Mycolactone A/B isolated from M. ulcerans is believed to be responsible for the extreme tissue necrosis and immunosuppression observed in patients with Buruli ulcer.²³ Injection of this toxin into the guinea pig induced significant tissue damage and an alteration of the immune response¹¹. Consistent with earlier studies,^12,23 the M. liflandii strain identified in this report also produced mycolactone E. The prominent necrosis observed within the granulomatous lesions in frogs with M. liflandii infection is attributed to the potent cytotoxic effect of mycolactone E.

The frogs in the affected room were purpose-bred animals purchased from a commercial source, but the vendor’s stocks were originally derived from wild-caught African frogs. It is possible that M. liflandii is subclinically endemic in all populations of frogs derived from wild-caught stocks, and the disease becomes clinically apparent when frogs are maintained under laboratory conditions. In all probability, the frogs in this report acquired the infection from parental frogs through oocytes or sperm. Infection likely remained subclinical until environmental stress to the host allowed the bacteria to evade the immune system, ultimately resulting in clinical signs. Affected frogs shedding large quantities of bacteria into the environment may have also contributed to the spread of the disease to any frogs with skin abrasions. Other Mycobacterium spp., such as M. marinum, have been shown to induce chronic nonlethal granulomatous disease in immunocompetent frogs and acute lethal infections in immunosuppressed frogs.²⁵ It should be noted that during the month of the outbreak, the average monthly recorded water temperature range in the affected tanks was considerably lower (65.66–68.36°F) than that recommended for keeping X. (Silurana) tropicalis in captivity (80.6– 84.2°F). Hence, low water temperature in conjunction with handling stress from oocyte collection procedures presumably induced an immunosuppressed state in these frogs. Interestingly, at the time of publication, the active disease has not been identified in the neighboring rooms with rederived X. (Silurana) tropicalis from gentamicin-treated oocytes or X. laevis colonies, which are maintained in more stable recirculation systems.

Screening of laboratory Xenopus for diseases is essential if M. liflandii and other Mycobacterium spp. are endemic in these populations. Screening of oocytes of newly imported frogs was proposed in the European M. liflandii outbreak of X. (Silurana) tropicalis frogs where the organism was identified in the ovarian tissue of all affected frogs.³² A combination treatment with rifampin-streptomycin for 8 weeks is the only drug regimen currently recommended by the World Health Organization for treatment of Buruli ulcer in humans.²¹ This drug regimen has not been adapted for treatment of frogs with mycobacterium outbreaks or for preventative treatment of eggs for rederivation.

We suggest implementing a sentinel frog program to identify endemic infections via necropsy, histopathology, culture, and PCR. Identification of M. liflandii and other mycobacterial infections within a facility is important for colony management and public health considerations. Infected frog colonies must be strictly quarantined to determine health status. Although M. liflandii, to our knowledge, has not been identified in humans, the similarity in growth characteristics, mycolactone production, and skin pathology to the human pathogen, M. ulcerans, raises speculation on its zoonotic potential. Laboratory personnel should adhere to strict guidelines for personal protection and use methods to minimize cross-contamination between colonies of frogs maintained in research aquaria.

Figure 23.

Polymerase chain reaction (PCR) targeting of locus 6 of variable number of tandem repeats (VNTR) of selected isolates from Xenopus (Silurana) tropicalis (case Nos. 16, 6, and 17) in comparison to Mycobacterium ulcerans (MU), mycolactoneproducing Mycobacterium marinum isolates (MMDL, MMBB), and Mycobacterium liflandii (M. lif). All 3 frog isolates had a VNTR PCR profile matching that of M. liflandii.

Figure 24.

Polymerase chain reaction (PCR) targeting of ST1 locus of variable number of tandem repeats (VNTR) of selected isolates from Xenopus (Silurana) tropicalis (case Nos. 16, 6, and 17) in comparison to Mycobacterium ulcerans (MU), mycolactoneproducing Mycobacterium marinum isolates (MMDL, MMBB), and Mycobacterium liflandii (M. lif). All 3 frog isolates had a VNTR PCR profile matching that of M. liflandii.