Pulmonary adiaspiromycosis in armadillos killed by motor vehicle collisions in Brazil

Knowledge of infectious diseases in wildlife provides important information for preventing potential outbreaks of zoonotic diseases. Adiaspiromycosis is a neglected human disease caused by dimorphic Onygenales fungi. The disease is produced by the inflammatory response against growing adiaspores, leading to granulomatous pneumonia. In humans, adiaspiromycosis is relevant in immunosuppressed patients. In animals, it is associated with pneumonia in fossorial species. Given the potential role of armadillos in the epidemiology of adiaspiromycosis, in this study, we sought to investigate the occurrence and pathological features of adiaspiromycosis in roadkilled armadillos. In total, 54 armadillo carcasses were suitable for postmortem pathologic examinations between February 2017 and 2020. Adiaspores, associated with granulomatous lesions, were observed in ten six-banded (Euphractus sexcinctus) and two southern naked-tailed armadillos (Cabassous unicinctus). A previously uncharacterized Onygenales species was molecularly identified in two E. sexcinctus. In summary, herein we report 12 cases of pulmonary adiaspiromycosis (PA) in two species of free-living armadillos in Brazil. Both, the morphology of the fungus, as well as the histopathological findings (granulomatous inflammatory response to adiaspores) are consistent with PA; however, as the molecular identification differs from the reported species, the potential impact of this fungus for human PA is unknown, and we cannot rule out its impact on public health.

Surveillance of infectious diseases in wildlife is a valuable tool for the prevention and reduction of human disease outbreaks 1,2 . Armadillos (order Cingulata) play a role in the epidemiological cycle of some zoonoses (e.g., hanseniasis and paracoccidioidomycosis) 3,4 . Despite being protected species, armadillo meat is commonly consumed in several South American countries 5,6 . Direct contact between humans and armadillos could pose a transmission route for several pathogens, thus, health investigations on armadillos may be relevant to public health.
Adiaspiromycosis is a fungal disease caused by the dimorphic fungi Emmonsia crescens and Blastomyces parvus (formerly known as Emmonsia parva) (family Ajellomycetaceae, order Onygenales), considered saprophytes and commonly isolated from the soil 7 . In contrast to other mycoses, once their conidia enter the host, mainly through respiratory via, they become adiaconidia (adiaspores), precluding their replication but increasing their size 8 . The disease is caused by the host´s immune response against the growing adiaspores, leading to the formation of granulomas 9 . The lesions are mainly restricted to the lungs and occasionally regional lymph nodes, although ocular and systemic adiaspiromycosis have also been described 8,10 . These lesions have been reported in humans, wild fossorial mammals (rodents, moles, armadillos), some species of carnivores, deer, horses and anurans ("Supplementary Material").
The first description of the disease was made in wild rodents of the United States in the 1940s 11 . In humans, the first case was reported in France in the 1960s 12 . Adiaspiromycosis was widely studied in the 1970s, mainly in wild small mammals 13 . Nevertheless, reports have been intermittent during the last decades. While human cases are generally associated with immunosuppression, the clinico-pathological features in free-ranging animals remain largely unknown [14][15][16][17][18] . Human pulmonary adiaspiromycosis (PA) is generally an incidental finding in diagnostic images (X-rays and tomography) and can be a differential diagnosis of tuberculosis and/or  23 . Adiaspiromycosis has been also described in three other armadillo species: hairy (Chaetophractus villosus), pichi (Zaedyus pichiy) and seven-banded armadillo (Dasypus septemcinctus) from Argentina 24 .
Since there are reports of PA in Brazil, and the etiologic agent has already been reported in armadillos (preliminarily by molecular technique) 23 , we hypothesized a possible role of armadillos in the epidemiology of PA. Considering that the number of wild animals killed by MVC in Brazil is remarkable 25 , and the potential role of armadillos in the epidemiology of adiaspiromycosis, in this study, we sought to investigate the occurrence and pathological features of adiaspiromycosis in roadkilled armadillos. Here we report 12 cases of PA in two species of wild armadillos of Brazil: the six-banded armadillo (Euphractus sexcinctus) and the southern naked-tailed armadillo (Cabassous unicinctus).
Microscopically, all these armadillos had pulmonary adiaspores with varying degrees of localized inflammation, ranging from none to marked granulomatous response (Table 2). Morphologically, adiaspores presented a bi-or trilaminar wall comprising a thin and brightly eosinophilic outer layer, a thick pale eosinophilic mid layer, and an internal basophilic layer which surrounded a core of round basophilic granular material but did not exhibit endospores. Adiaspores averaged 91.1 μm in diameter (ranged from 10.9 to 839.5 μm) and were distributed multifocally within the alveolar and bronchial interstitium spaces. Adiaspores' wall components were highlighted by PAS, GMS and WS stains (Fig. 2). Inflammation severity correlated to adiaspores' size and varied from few reactive macrophages surrounding the adiaspores to severe nodular granulomatous inflammation with www.nature.com/scientificreports/ large numbers of epithelioid macrophages, neutrophils, lymphocytes, eosinophils, rare multinucleated giant cells (foreign body type), and fibroplasia. The granulomas could contain a viable adiaspores (up to 200 μm in diameter) or cores with empty adiaspores admixed with neutrophils and necrotic cell debris. Several cases (91.7%, 11/12) had hyperplasia of the lymphoid tissue associated with bronchi and bronchioles. This prompted histochemical investigation for potential Mycoplasma sp. and/or Filobacterium sp. (formerly CAR Bacillus) structures; however, these were not detected by WS stain. One armadillo had pulmonary coinfection with a trichurid nematode leading to marked bronchitis. Complete microscopic findings are listed in Table 3. No fungal structures and/or associated inflammatory response were observed in other tissues {skin, skeletal muscle, tongue, oropharynx, tonsil, salivary glands, esophagus, stomach, small and large intestines, liver, gallbladder, pancreas, larynx, trachea, lung, heart, great vessels, thymus, spleen, lymph nodes (mandibular, prescapular, mediastinal and mesenteric), kidney, urinary bladder, thyroid and parathyroid glands, adrenal glands, diaphragm, cerebrum, cerebellum, eye, mammary gland, testicle, ovary, uterus, epididymis and prostate} examined. Two 640-bp sequences (after excluding primers) of the fragment comprised between 18S rRNA gene and 26S rRNA gene were obtained from frozen lung samples of two six-banded armadillos (cases 10 and 12) with pulmonary intralesional adiaspores. The 26S rRNA was also amplified and sequenced in these two animals. These sequences are available through the GenBank database [MT258564, MT258563, MT258566 and MT258565]. No sequences were recovered from the DNA extractions from FFPE lung samples of the remaining ten armadillos with adiaspores for any of the two selected PCR protocols.
The ITS sequences obtained from two six-banded armadillos (case 10 and 12) were identical between them, and had the highest nucleotide (NT) identity (87.6%) to an uncultured organism amplified from the soil of the rhizosphere of tomato (Solanum lycopersicum) in Mexico (JN660517), likely E. crescens (Cordero-Ramirez et al. 2012), followed by 86.5% similarity with two E. crescens sequences from lung samples of mammals (AF038334, AF038337). The unique sequence type found in armadillos clusters separately in our ITS phylogram (Fig. 3).

Discussion
Armadillos are among the most common species recorded in Brazilian roadkill monitoring programs, in this study were registered a total of 2941 armadillo carcasses over a 3-year period. This is primarily because some species in this group are generalist with large distribution areas and relatively high densities [26][27][28] . Examination of roadkill wildlife is pivotal for a better understanding and monitoring of infectious diseases with zoonotic potential, particularly in synanthropic and peri-urban wild animals 1,29 . In this study, armadillos killed by MVC were regarded as an important source of information for the surveillance of PA. www.nature.com/scientificreports/ Adiaspiromycosis and emmonsiosis-like fungi are isolated from the soil 7 . Previous studies detailed that this order (Cingulata) presents a higher occurrence of infection by Emmonsia adiaspores when compared to other mammals (marsupials, rodents, lagomorphs and carnivores) 13 . It is reasonable to believe that the fossorial and digging habits of armadillos likely promote inhalation of soil adiaspores and subsequent lung lesions. However, we only found cases in two of the four species collected. However, the habits of D. novemcintus are less fossorial than E. sexcinctus and C. unicinctus is actually considered subterranean 30 . The low sample size of P. maximus (N = 2) may be why no cases were detected.
Microscopically, PA in man and animals is characterized by granulomatous pneumonia with intralesional adiaspores. The local inflammatory response includes histiocytes, neutrophils, epithelioid macrophages, lymphocytes, and occasionally multinucleated giant cells; and this is directly proportional to the adiaspores size. Adiaspores are usually bi-to trilaminar and their wall stains positive by PAS and GMS. E. crescens has a larger diameter (200-700 µm) compared to B. parvus (20-40 µm) 13,16,31 . These characteristics were observed in the armadillos analyzed in this study, however, even though the diameter of the adiaspores averaged 129.2 µm (Min = 10.9 µm; Max = 839.5 µm), being compatible with E. crescens, only 86.5% nt similarity for ITS region was obtained. The culture of these fungi was not performed due to their difficulty and the need of high biosecurity levels, fact that limited technically the morphologic characterization of the fungi 31 . www.nature.com/scientificreports/ The identities of our ITS sequences with the nearest ones is even lower than those reported for core genes of E. parva isolates (88.6%), a species proposed as polyphyletic, and that those proposed for E. crescens strains (91.8%) based on genome identity 32 . The D1-D2 sequences are highly similar to other Onygenales species. Based on the molecular results, we identified a hitherto undescribed Onygenales species in pulmonary lesions of two six-banded armadillos causing adiaspiromycosis. Interestingly, aside from E. crescens and B. parvus, potential novel fungal species have been described in wild mammals with adiaspiromycosis, including the report of a unique sequence type more related to Emmonsiellopsis in two northern hairy-nosed wombats (Lasiorhinus krefftii) a fossorial species from Australia 13 . The zoonotic potential of the fungus infecting armadillos is unknown. Our molecular results raised the question about which adiaspore-producing species are infecting humans in Brazil, in light of most of the human cases are diagnosed only based on morphology, without ancillary molecular testing. To this date, in a literature review, at least 13 reports of human PA are available in the country ("Supplementary Material"). Furthermore, superficial adiaspiromycosis was reported to cause granulomatous conjunctivitis in children in the Amazon basin, and histopathologic examination of ocular nodules identified adiaspores-like structures 10 . Future studies molecularly identifying the fungal species in humans are necessary to explore the epidemiology of the disease, including the use of armadillo species as indicator of potential exposure areas to humans.
In summary, we provide pathological and molecular evidence of a relatively poorly known fungal disease in wild armadillos. Both, the morphology of the fungus, as well as the histopathological findings (granulomatous inflammatory response to adiaspores) are consistent with PA; however, as molecular identification differs from the reported species, the potential impact of this fungus for human PA is unknown, and we cannot rule out its impact on public health. To the authors' knowledge this is the largest report of this disease in wild mammals from Brazil.

Methods
Road monitoring. Between February 2017 and February 2020, a periodic road monitoring was carried out to identify the number of armadillos killed by MVC in two states of Brazil, Mato Grosso do Sul (MS) and Sao Paulo (SP) (See 25 for details) 25 . Epidemiological and biological data [species, age class, sex, season (rainy or dry) and coordinates] were recorded from all carcasses. Roadkilled armadillos in good preservation status were selected for investigate the occurrence and pathological features of adiaspiromycosis by anatomopathological study 33 . Pulmonary adiaspiromycosis survey. Representative tissue samples of lungs, trachea and tracheobronchial lymph nodes were collected, and fixed in 10% neutral buffered formalin or frozen at − 20 °C or − 80 °C. These tissues were processed routinely, embedded on paraffin-wax, sectioned at 5 μm-thick and stained with hematoxylin and eosin (H&E) for routine microscopic analysis. Special stains periodic acid-Schiff (PAS), Grocott methenamine silver (GMS), Masson's trichrome (MS) and Warthin-Starry stain (WS) were applied in selected cases to better characterize the histopathologic findings. Microphotographs of the adiaspores were collected and subsequently measured using the Image J software 34 .

Data availability
All data generated or analyzed during this study are included in this published article.