We characterize Brucella infection in a wild southern sea otter (Enhydra lutris nereis) with osteolytic lesions similar to those reported in other marine mammals and humans. This otter stranded twice along the central California coast, US over a 1-yr period and was handled extensively at two wildlife rehabilitation facilities, undergoing multiple surgeries and months of postsurgical care. Ultimately the otter was euthanized due to severe, progressive neurologic disease. Necropsy and postmortem radiographs revealed chronic, severe osteoarthritis spanning the proximal interphalangeal joint of the left hind fifth digit. Numerous coccobacilli within the joint were strongly positive on Brucella immunohistochemical labelling, and Brucella sp. was isolated in pure culture from this lesion. Sparse Brucella-immunopositive bacteria were also observed in the cytoplasm of a pulmonary vascular monocyte, and multifocal granulomas were observed in the spinal cord and liver on histopathology. Findings from biochemical characterization, 16S ribosomal DNA, and bp26 gene sequencing of the bacterial isolate were identical to those from marine-origin brucellae isolated from cetaceans and phocids. Prior reports document the zoonotic potential of the marine brucellae. Isolation of Brucella sp. from a stranded sea otter highlights the importance of wearing personal protective equipment when handling sea otters and other marine mammals as part of wildlife conservation and rehabilitation efforts.
Most well-characterized Brucella strains are associated with livestock, pets, and terrestrial wildlife, but diverse marine-origin brucellae are increasingly recognized. Since the first report of an atypical Brucella from a bottlenose dolphin (Tursiops truncatus; Ewalt et al. 1994), the number of known-infected marine hosts has grown, especially among cetaceans (dolphins, porpoises, whales) and pinnipeds (seals, sea lions, walruses; Foster et al. 1996; Ross et al. 1996; Jahans et al. 1997). Two species of marine Brucella are known, based on host, biochemical, and molecular criteria: Brucella ceti (cetacean-origin strains) and Brucella pinnipedialis (seal-origin strains; Bricker et al. 2000; Foster et al. 2007). However, precise characterization of the taxonomy, host range, and pathophysiology of marine brucellae awaits further study (Bricker et al. 2000; Moreno et al. 2002).
Serologic studies reveal low to moderate Brucella antibody prevalences in sea otters from California, Alaska (Hanni et al. 2003), and Russia (Goldstein et al. 2011). However, aside from a single report in a European river otter (Lutra lutra) (Foster et al. 1996), Brucella infection of estuarine or marine otters has not been confirmed, and associations with morbidity and mortality have not been described. Sea otters occupy a specialized niche in coastal ecosystems, acting as both keystone species and sentinels for anthropogenic pollution (Jessup et al. 2004). Infectious disease is a common cause of southern sea otter (Enhydra lutris nereis) mortality (Kreuder et al. 2003; Miller et al. 2010). Although shark-associated mortality and protozoal disease have received much attention in recent years (VanWormer et al. 2013; Tinker et al. 2016), few bacterial diseases have been well characterized. Here we describe the isolation and preliminary characterization of Brucella from a southern sea otter, and describe associated lesions and potential health risks for humans working in wildlife rehabilitation facilities.
MATERIALS AND METHODS
On 26 December 2001, a subadult female southern sea otter was found weak and emaciated (14 kg) on Pismo Beach, California, US and was transported to rehabilitation facility #1. Superficial lacerations were noted on the left flank and left hind flipper, with an exposed left hind fifth proximal interphalangeal joint. Swelling, crepitus, and reluctance to use the affected limb were suggestive of infection. A joint aspirate contained Gram-negative bacteria, but culture was not performed. A second subcutaneous abscess was identified and drained on the right hind limb. Based on a provisional diagnosis of shark bite with secondary infection, the wounds were repeatedly debrided, and the animal was treated with intramuscular enrofloxacin (Baytril, Bayer, Leverkuse, Germany) and penicillin (Dual-pen, AgriPharm, Westlake, Texas, USA) for 2 wk.
During hospitalization, occasional episodes of disorientation, tremors, unusual tameness, and reduced reactivity to external stimuli were noted. Serodiagnostic tests revealed elevated titers to Toxoplasma gondii and Sarcocystis neurona (Table 1), so the otter was placed on oral antiprotozoal medication (Diclazuril, Virbac, Fort Worth, Texas, USA). From December 2001 through March 2002 progressive clinical improvement was noted. The left hind interphalangeal joint wound eventually closed, but remained swollen, firm, and mildly warm, and the otter was reluctant to use this foot. All other lesions healed satisfactorily and as clinical condition improved, stereotypic behavior suggestive of stress due to captivity increased. Following intraperitoneal implantation of a very high frequency transmitter, the animal was released at Shell Beach, California on 21 March 2002.
The otter stranded again on 22 December 2002 at Oceano Dunes, California (10 km from the original release site). Examination at rehabilitation facility #2 revealed emaciation and generalized paresis and asymmetrical (left>right) hind limb paraparesis. The otter was unable to leave the water unaided or flex her spine to groom her abdominal fur, necessitating periodic grooming by staff. As nutritional condition improved, self-grooming increased, but the otter avoided grooming her dorsal lumbar region and resisted staff attempts to brush this area. Minimal use of the left hind flipper was noted throughout hospitalization. Several weeks poststranding, multiple alopecic, pink, raised plaques were noted on the patient's head. These were attributed to poor grooming or stereotypic rubbing. Also noted were tremors, intermittent left front limb rigidity, and paresis that worsened with stimulation. Diclazuril therapy was resumed, but the tremors worsened, and progressive, severe stereotypic circling was noted. Due to a poor prognosis, euthanasia was performed 53 d poststranding.
Necropsy, histopathology, and protozoal immunohistochemistry, culture, serology, and PCR
Necropsy, including radiographs, bacterial culture, and cryo-archiving of tissues and serum, was performed the following day. Tissue samples were formalin-fixed, paraffin-embedded, and 5-μm sections were stained with H&E for histopathology. Skin was not collected for microscopic examination. Upon receiving bacterial culture results (described in the upcoming text), the frozen-thawed left hind fifth proximal interphalangeal joint was collected, formalin-fixed, decalcified, and processed for microscopic examination. Fite's acid fast and Gomori methenamine silver stains were also performed on selected tissues using standard laboratory protocols.
During necropsy, brain tissue was collected aseptically to test for T. gondii and S. neurona infection via cell culture and PCR amplification of the B1 gene as described previously (Miller et al. 2002). Postmortem serum was evaluated with an indirect fluorescent antibody test employing polyclonal antisera directed against T. gondii and S. neurona, and immunohistochemical stains for both parasites were prepared from paraffin-embedded brain, spinal cord, and skeletal muscle, as described (Miller et al. 2002, 2008).
Brucella culture, biochemical characterization, biotyping, serology, and immunohistochemistry
Swabs inoculated from heart blood, gallbladder, and the left hind fifth proximal interphalangeal joint were held in Amies transport media (Copan Diagnostics Inc., Murrieta, California, USA), then plated onto sheep blood and MacConkey agar (Hardy Diagnostics, Santa Maria, California, USA) and streaked for isolation on the same day as the necropsy. Plates were incubated at 35 C in 5% CO2 for 4 d, and any with visible bacterial growth were submitted to the University of California, Davis Veterinary Medical Teaching Hospital for identification. Bacterial colonies were subcultured, Gram stained, tested for urease activity and CO2 dependence, and bacterial identity was confirmed with 16S ribosomal (r) DNA sequence analysis (Murray and Stackebrandt 1995).
Serum collected throughout each period of captive care and at necropsy was submitted to Agriculture Canada (Ottawa, Canada) to assess the presence and concentration of Brucella-reactive antibodies. A competitive enzyme-linked immunosorbent assay (cELISA) and a fluorescence polarization assay (FPA) were used to screen sera for antibodies to Brucella spp., as previously described (Nielsen et al. 1996a; Lucero et al. 1999). The cELISA and FPA do not require species specificity. The cELISA measures antibody capable of competing with a mouse monoclonal antibody specific for Brucella O-polysaccharide for antigen binding sites on the polystyrene plate. The amount of competition is measured using a goat antibody to Mouse immunoglobulin G labelled with enzyme. Less conjugate binding indicates higher antibody activity in the test sample. The FPA measures any antibody of any species capable of binding to a labelled antigen and thereby reducing the rotational rate of the labelled antigen. In both cases, the serologic tests were set up according to standards described by the World Organisation for Animal Health and were performed by the World Organisation for Animal Health Regional Reference Laboratory at the Canadian Food Inspection Agency, Nepean, Canada. Because species-specific reference control sera were not available, assay validity was established using bovine standard reference sera.
To screen for systemic brucellosis and clarify associations between Brucella sp. detection and observed lesions, major tissues including formalin-fixed brain, spinal cord, and the decalcified left interphalangeal joint were immunostained with antibodies directed against Brucella abortus at the Veterinary Services Laboratory (Fort Collins, Colorado). This assay and antibody has been demonstrated to label brucellae in several pinnipeds (Garner et al. 1997; J.C.R. unpubl.). Tissues (5 μm) were stained using a labeled streptavidin–biotin system employing polyclonal B. abortus antibody, as described by Rhyan et al. (1997), and were examined on a compound microscope.
Molecular characterization for Brucella
Molecular characterization of the Brucella isolate was performed at University of California, Davis using published protocols for amplification and sequencing of a 775-bp section of the 16S rDNA, and the genes coding for the Brucella 26 kDa protein (BP26) (Cloeckaert et al. 2000). The DNA sequences were analyzed using Chromas (Technelysium Pty Ltd., Tewantin, Queensland, Australia), GeneDoc (Nicholas et al. 1997), and Geneious 5.3.6 (Biomatters, Auckland, New Zealand) software.
Necropsy, histopathology, and protozoal immunohistochemistry, culture, serology, and PCR
The otter was in excellent nutritional condition (20.2 kg) following 8 wk of rehabilitation. Musculature was symmetrical and adequately developed, with no gross abnormalities of the vertebrae, intervertebral discs, brain, or spinal cord.
Postmortem radiographs and gross necropsy revealed a severe osteolytic lesion spanning the left hind fifth proximal interphalangeal joint, corresponding with a region of grossly apparent soft tissue swelling (Fig. 1A, B). Articular surfaces were irregular and roughened with erosion of hyaline cartilage, periarticular fibrosis, and minimal opaque tan joint fluid. Moderate diffuse lymphadenopathy was noted; affected lymph nodes were solid and tan with prominent cortical thickening.
Minimal orange-white mottling of the myocardium and mild hepatosplenomegaly was observed (possible euthanasia artifact). Approximately 100 acanthocephalan parasites (Profilicollis spp.) were deeply embedded in the wall of the distal duodenum and jejunum. Points of acanthocephalan attachment corresponded to 1–2 mm diameter, raised, yellow serosal nodules indicative of transmural parasite migration. The omentum was slightly thickened, red, and opaque, but no peritoneal fluid was observed.
Microscopic examination of the decalcified left hind fifth proximal interphalangeal joint revealed chronic granulomatous osteomyelitis and arthritis, with erosion of articular cartilage, exposure of underlying trabecular bone, and extensive periarticular fibrosis. Sparse inflammation was admixed with bacteria along articular surfaces, underlying marrow spaces, and Haversian canals.
Striking multifocal nodular granulomatous myelitis was also noted (Fig. 1D, E), although there was no gross or microscopic evidence of spinal compression or pressure necrosis. Spinal granulomas were sparsely distributed, well demarcated, large (50–100 μm), and were composed of dense aggregates of epithelioid macrophages and monocytes, with adjacent areas of moderate perilesional congestion and inflammation. Smaller granulomas were observed throughout the hepatic parenchyma. No gross or microscopic evidence of oophoritis or endometritis was noted in this immature female.
Also noted was mild multifocal nonsuppurative meningoencephalitis dominated by small lymphocytes, and rare large (≥100 μm2) cavitated lesions with scant perilesional nonsuppurative inflammation. Scant, amorphous basophilic crystalline material was scattered along the edge of some cavitated lesions (dystrophic mineralization), along with gitter cells containing granular blue-gray pigment (lipofuscin) and sparse glial cells. Rare thin-walled protozoal tissue cysts compatible with T. gondii were observed in brain tissue with no adjacent inflammation. Although immunohistochemistry was attempted, no parasite profiles were present in the recuts. No bacteria were observed in the brain, spinal cord or liver on H&E, acid fast, or Gomori methenamine silver stains.
Several 50–100 μm-long, thick-walled intracytoplasmic protozoal sarcocysts with prominent surface projections and fine internal septations encompassing thousands of tiny banana-shaped zoites (Sarcocystis spp.) were observed in skeletal myofibers. These were associated with mild myositis. These sarcocysts (an incidental finding) showed weakly positive labelling for S. neurona on immunohistochemistry.
Lymph nodes exhibited moderate follicular and paracortical lymphoid hyperplasia. Mild lymphocytic inflammation was observed in the myocardium, adrenal cortex, and renal cortices. Mild multifocal granulomatous omentitis, including rare foreign body giant cells surrounding mineralized debris, was attributed to degenerating peritoneal acanthocephalans.
Serum collected 4 d after the first stranding was weakly positive for T. gondii and S. neurona antibodies (Table 1). Testing of postmortem serum samples revealed increased reactivity to T. gondii and S. neurona. Tachyzoites were visible in cell monolayers following exposure to brain tissue collected at necropsy. The zoites were morphologically consistent with T. gondii and the cells were T. gondii-positive via B1 gene PCR. Brain and spinal cord were negative for S. neurona by histopathology, immunohistochemistry, cell culture, and PCR.
Brucella culture, biochemical characterization, biotyping, serology, and immunohistochemistry
All agar plates inoculated with heart blood and gallbladder samples were negative for bacteria 4 d postinoculation. No growth was apparent on a MacConkey plate inoculated with the left fifth interphalangeal joint swab after 4 d, so the plate was discarded. After 4 d of incubation, the interphalangeal joint sample that was plated on blood agar yielded numerous tiny monomorphic, nonhemolytic, pale gray to nonpigmented colonies of Gram-negative coccobacilli. This plate was submitted to the University of California, Davis Veterinary Medical Teaching Hospital where subculture confirmed growth of Gram-negative, urease-positive coccobacilli that required CO2, and 16S rDNA sequence analysis confirmed the presence of Brucella sp. The isolate (hereafter denoted as SSO-1) was submitted to the National Veterinary Services Laboratory. Cryopreserved lung and multiple lymph nodes were culture-negative for Brucella at the National Veterinary Services Laboratory.
Serum samples from both stranding episodes and from necropsy were assessed for Brucella antibodies using cELISA and FPA. Results from both tests indicated that Brucella seroconversion occurred between release from rehabilitation facility #1 and the second stranding (Table 1). Brucella titers did not vary appreciably throughout the second period of hospitalization, suggestive of chronic infection.
All tissues tested except for lung and the left hind interphalangeal joint samples were Brucella-immunonegative. In the lung, one mononuclear cell within a pulmonary vein contained a cytoplasmic cluster of positive-staining coccobacilli. The interphalangeal bones contained large numbers of B. abortus-immunopositive bacterial coccobacilli along the joint surface, in necrotic Haversian canals and in a small marrow space (Fig. 1C).
Molecular characterization of Brucella
The 775-bp partial 16S rDNA gene sequence from bacteria isolated from the left hind interphalangeal joint (GenBank accession DQ295026) was identical to all GenBank Brucella spp. sequences, confirming this isolate as Brucella sp. A 1,900-bp bp26 amplicon from SSO-1 contained an IS711 insertion downstream of the bp26 gene, consistent with marine Brucella strains (Cloeckaert et al. 2000).
Well documented as a cause of disease in terrestrial animals and humans (Morgan and Corbel 1984), bacteria of the genus Brucella also infect marine mammals, causing disease of varying severity. To our knowledge, this is the first report of Brucella infection in a sea otter. Infection appears to have caused, at a minimum, chronic granulomatous arthritis, based primarily on pathogenicity and host preference includes eight terrestrial species, each associated with particular hosts: B. abortus (cattle; Bos taurus), B. melitensis (sheep; Ovis spp. and goats; Capra spp.), B. suis (swine; Sus scrofa), B. ovis (sheep), B. canis (Canidae), B. neotomae (wood rats; Neotoma spp.; Morgan and Corbel 1984), B. microti (voles; Scholz et al. 2008), and B. inopinata (tree frogs; Fischer et al. 2012). Except for B. neotomae, all brucellae can exhibit significant host pathogenicity, causing placentitis, metritis, abortion, epididymitis, orchitis, discospondylitis, and myeloencephalitis (Morgan and Corbel 1984). Interspecies transmission is recognized, such as sharing of B. melitensis infection across livestock species (Kahler 2000). Marine-origin brucellae are more recent discoveries (Bricker et al. 2000). Two species are currently recognized: B. pinnipedialis (associated with seals, Phocidae; sea lions, Otaridae; and walruses, Odobenidae) and B. ceti (associated with Cetacea; porpoises, dolphins, and whales). Although an understanding of their evolutionary origins and host range is incomplete, the potential pathogenicity of marine brucellae for animals and humans is well recognized (Miller et al. 1999; Sohn et al. 2003; McDonald et al. 2006; Hernández-Mora et al. 2008).
The SSO-1 Brucella strain was isolated from a chronically infected joint, which was positive for Brucella on immunohistochemistry (Fig. 1C). Necropsy revealed chronic granulomatous osteoarthritis (Fig. 1A, B) and myelitis (Fig. 1D, E) that was distinct from other known sea otter inflammatory diseases, including coccidioidomycosis (Huckabone et al. 2015) and toxoplasmosis (Miller 2008). Disseminated granulomas are common in Brucella-infected animals and humans (Ceviker et al. 1989; Bingöl et al. 1999). Due to sparse bacterial loading, these lesions are commonly negative on culture and special stains (Gonzalez et al. 2002; Sohn et al. 2003). Immune-associated disease might contribute to lesion development and severity (Krishnan et al. 2005).
The relative contributions of concurrent brucellosis and toxoplasmosis with respect to observed progressive neurologic disease are unknown. Although the cavitated and partially mineralized brain lesions were typical of chronic toxoplasmosis in sea otters (Miller 2008), humans, and experimentally-exposed rodents (Stahl et al. 2004), and T. gondii-like tissue cysts were observed nearby, due to lesion chronicity the underlying cause could not be confirmed. Based on the severity of the spinal granulomas and their similarity to lesions in Brucella-infected animals and humans (Gonzalez et al. 2002; Sohn et al. 2003; Gonzalez-Barrientos et al. 2010), it is possible that both T. gondii and Brucella contributed to morbidity.
Brucella has not previously been isolated from sea otters, but prior serologic surveys have revealed low to moderate antibody prevalence in otters from California (6%), Alaska (3–8%), and Russia (28%) (Hanni et al. 2003; Goldstein et al. 2011). Brucella antibodies have also been detected in sympatric harbor seals, Steller sea lions (Eumetopias jubatus), bottlenose dolphins, and walruses (Odobenus rosmarus) (Nielsen et al. 1996b; Burek et al. 2005; Lambourn et al. 2013). Despite serologic evidence of exposure, infection was not previously confirmed in sea otters. This could be because infection is often transient, subclinical, and characterized by low bacterial burdens, as for many other host species (Moreno et al. 2002). Infection might also be underdiagnosed because brucellae are fastidious, slow-growing, and easily obscured by other bacteria (De Miguel et al. 2011). Systemic protozoal infections or other common infectious diseases of southern sea otters could also inhibit detection due to lesion overlap. Finally, sea otter spinal cords are not routinely examined microscopically, so the unique granulomas that appear to be associated with Brucella infection in this case could be under-recognized.
Isolate SSO-1 was confirmed as Brucella sp. based on 16S rDNA sequence and molecular and phenotypic features consistent with marine-origin brucellae, including presence of an extra copy of the IS711 gene at the bp26 locus (Cloeckaert et al. 2000). In common with B. pinnipedialis from pinnipeds and a river otter, growth of SSO-1 was CO2-dependent. Carbon dioxide dependence was the most accurate predictor of host origin among 102 pinniped and cetacean isolates. Although some exceptions have been noted, pinniped strains are generally CO2-dependent, and cetacean strains CO2-independent (Dawson et al. 2008).
Routes of Brucella exposure in humans and animals include ingestion, inhalation, conjunctival infection, transplacental transfer, and transcutaneous spread through traumatized skin (Carvalho Neta et al. 2010). Southern sea otters consume filter-feeding invertebrates that can concentrate fecal bacteria (Miller et al. 2010), raft in groups, groom extensively, can haul out on rough surfaces, and often wound conspecifics during territorial and breeding interactions. Thus, all known routes for Brucella infection are possible, with perhaps the exception of lungworms (Garner et al. 1997), given that pinniped lungworms are not known to parasitize sea otters.
Although marine brucellae often exhibit mild pathogenicity in host species, highly variable (and sometimes severe) pathogenicity is reported across hosts (Perrett et al. 2004). For example, experimentally infected cattle aborted (Rhyan et al. 2001), although sheep remained asymptomatic and lesion-free (Perrett et al. 2004). In humans, spinal and intracranial granulomas have been associated with seizures, progressive paresis, and paralysis (Ceviker et al. 1989; Bingöl et al. 1999; Sohn et al. 2003; Krishnan et al. 2005).
This report broadens the Brucella host range, and describes associated lesions in sea otters. Multifocal granulomas were similar to those described from other Brucella-infected animals and humans, but distinct from classical protozoal and fungal-associated inflammatory lesions of sea otters (Miller 2008; Huckabone et al. 2015). Our report also highlights potential health risks for persons rehabilitating or consuming infected marine animals. Three cases of brucellosis associated with infection by marine brucellae have been reported in humans; two Peruvian patients presented with severe, progressive neurologic disease (Sohn et al. 2003), and a New Zealand man suffered vertebral osteomyelitis (McDonald et al. 2006). Marine-origin brucellosis might be underdiagnosed in humans, because these bacteria are difficult to detect and are unlikely to be considered, except following laboratory exposure. At-risk human populations often receive minimal medical surveillance, and clinical signs can be nonspecific and easily confused with other infectious or immune-mediated diseases. Persons with higher occupational risk for marine brucellae exposure include veterinarians, animal rehabilitation personnel, laboratory technicians, and those engaged in subsistence harvest (Sohn et al. 2003; MacDonald et al. 2006; Sears et al. 2012). However, only one case of laboratory-acquired marine brucellosis has been reported in humans (Brew et al. 1999). All three naturally-acquired cases of marine-origin brucellosis in humans were speculatively associated with consumption of raw fish, not marine mammal contact, suggesting that consumption of undercooked seafood, especially fish, could pose a higher risk for human infection.
This research was supported in part through funding from the California Department of Fish and Wildlife (CDFW). We acknowledge the assistance of staff and volunteers from CDFW, The Marine Mammal Center (TMMC), the Monterey Bay Aquarium (MBA), and US Geological Survey–Biological Resources Division (USGS–BRD) for sea otter carcass recovery, sample collection, and data processing. Additional technical support was provided by Klaus Nielsen, Debbie Brownstein, Andrea Packham, Darla Ewalt, and Ann Melli.