Figure 1

Figure 1. Diaphragmatic view of diseased liver from an aborted ovine fetus (fetus B) with possible Francisellaceae infection, Uruguay. A) Myriad discrete, white to yellowish, round, coalescing foci are visible, ranging from...

Figure 2

Figure 2. Microscopic images of diseased liver from an aborted ovine (fetus B) with possible Francisellaceae infection, Uruguay. A) Hematoxylin and eosin stain of section of liver. Hepatic histoarchitecture in center of...

In July 2015, two (≈1%) of ≈200 pastured sheep on a family farm in Colonia, Uruguay, aborted at ≈4 months of gestation. Autopsies on twin aborted fetuses (A and B) showed similar gross lesions (Table 1), consisting of severe multifocal widespread necrotizing hepatitis (Figure 1), and moderate fibrinous peritonitis and pericarditis. Samples of liver, adrenal gland, spleen, lung, heart, kidney, and brain tissues of both fetuses were fixed in formalin, then processed, embedded in paraffin, microtome-sectioned, and stained with hematoxylin and eosin. Histopathologic examination revealed severe acute multifocal random fibrinonecrotizing neutrophilic and histiocytic hepatitis (Figure 2, panel A), multifocal necrotizing and neutrophilic myocarditis, multifocal neutrophilic bronchiolitis and alveolitis, and multifocal fibrinous splenic capsulitis.

We processed formalin-fixed paraffin-embedded sections of liver from fetus B for immunohistochemistry to detect Francisella antigen (Appendix). We used tissue from a squirrel with culture- and PCR-confirmed F. tularensis septicemia as a positive control. We performed antigen retrieval in a decloaking chamber by using Antigen Decloaker citrate buffer (Biocare Medical, https://biocare.net). We applied a specific mouse monoclonal IgG3 raised against F. tularensis lipopolysaccharide, F. tularensis LPS Monoclonal Antibody (T14) (Thermo Fisher Scientific, https://www.thermofisher.com) as the primary antibody at 1:1,000 dilution. We used Mouse-on-Farma HRP-Polymer (Biocare Medical) and 3-amino-9-ethylcarbazole (Thermo Fisher Scientific) for antigen detection. For negative controls, we replaced F. tularensis monoclonal antibody with normal mouse IgG for both ovine and squirrel reactions.

Immunohistochemistry revealed strong abundant intralesional granular immunoreactivity in the necrotic foci of the fetal ovine liver, which was largely intracytoplasmic in infiltrating neutrophils and macrophages (Figure 2, panel B). We observed immunoreactivity in the positive control squirrel tissue but not in the negative controls. We conducted ancillary testing to rule out other ovine abortifacients (Table 2).

We postfixed formalin-fixed sections of liver from fetus B in modified Karnovsky’s fixative, 1% osmium tetroxide, and 0.1 mol cacodylate buffer, then processed and embedded sections in resin for transmission electron microscopy. Despite suboptimal ultrastructural tissue preservation due to autolysis, intrahistiocytic and extracellular ≈0.7–1.7 μm gram-negative coccobacilli colocalized with the foci of necrotizing hepatitis.

The lack of historical reports of tularemia outside endemic areas of North America and Eurasia has been puzzling (6). Recently, tularemia emerged in Australia (3) and reemerged in the Northern Hemisphere (7). South America has been considered free of tularemia (7); a status that seems to be based solely on the lack of disease reporting. However, tularemia might have been undiagnosed because of limitations in disease surveillance systems in the region. No clinical disease caused by Francisella spp. in mammals in the Americas south of Mexico has been described. Our results raise concerns about the possible occurrence of tularemia in South America.

The abortifacient effects of F. tularensis in sheep have been described in the United States, and tularemia has been regarded as an overlooked syndrome in sheep (9). From a pathologic viewpoint, necrotic foci in the liver, spleen, or lungs in late term aborted ovine fetuses are characteristic of tularemia and should raise suspicion, although gross lesions can be absent even in cases with typical histologic inflammatory and necrotizing lesions (9). Contrary to most bacterial abortifacients of sheep (11), F. tularensis is not visible upon histopathologic examination of tissues stained with hematoxylin and eosin, Steiner silver, or Gram stains, even in tissues that have a high bacterial burden demonstrated by immunohistochemistry (9). The ultrastructural demonstration of intracellular gram-negative coccobacilli of the expected size in phagocytic and inflammatory cells in tissues with lesions, as in our case, aids in the diagnosis. Diagnostic investigation of any case of ovine abortion with fetal lesions indicating a bacterial etiology should include ancillary testing to identify F. tularensis and rule out other abortigenic pathogens (11).

The etiologic diagnosis in our case was reached by the immunohistochemical demonstration of abundant intralesional antigen by a specific monoclonal antibody raised against F. tularensis lipopolysaccharide. Immunohistochemistry has proven useful for identifying F. tularensis in diagnostic settings (9,12). F. tularensis lipopolysaccharide is a main specific antigen and virulence factor and differs from the lipopolysaccharide of other gram-negative bacteria (13). According to the manufacturer, the primary antibody we used for immunohistochemistry does not cross-react with F. tularensis subsp. novicida, Yersinia pestis, Y. pseudotuberculosis, Y. enterocolitica, Vibrio cholerae, Escherichia coli, Salmonella enterica serovar Typhimurium, Brucella abortus, B. suis, B. ovis, B. melitensis, or B. neotomae. We tested the immunohistochemistry in cases of abortion caused by Campylobacter jejuni and C. fetus but observed no cross-reactivity. Although cross-reaction with other members of Francisellaceae cannot be completely ruled out, F. tularensis is currently the only species of this family recognized as an ovine abortifacient. Definite species and subspecies identification requires bacterial isolation and DNA analysis, which we were unable to perform because the available specimens were unsuitable.

Sheep with tularemia have been implicated in disease transmission to sheep industry workers (10). In the case described here, the owners of the sheep lived on the farm and were in contact with the affected flock regularly; however, we do not know whether they had clinical signs consistent with tularemia.

The source of infection in this sheep remained unknown. However, F. tularensis has a broad animal reservoir, including arthropods, rodents, lagomorphs, and marsupials (6,14). Brown hares (Lepus europaeus), a species that plays a primary role in the ecology of tularemia in Europe (12), have been introduced to Uruguay and are frequently seen around the affected farm. In addition, F. tularensis can be transmitted by ticks, several of which, including Amblyomma spp., Haemophysalis spp., and Ixodes spp. ticks, are endemic in Uruguay. Of note, a gamma-proteobacterium related to Francisella-like organisms, but different from F. tularensis, was identified in Uruguay in Amblyomma triste ticks (15), the most prevalent tick species reported in human tick bites in the country.

We provide pathologic and immunohistochemical evidence of disease caused by a possible Francisellaceae member in sheep in Uruguay. Additional research is needed to isolate and speciate the pathogen and elucidate its regional epidemiology. Nonetheless, veterinarians, physicians, and public health officials should be aware of possible tularemia in South America.