Inoculation into mice has long been the classical method for isolating and identifying pathogenic fungi present in environmental samples such as soil. Many studies have been performed over several decades, mainly by intraperitoneal inoculation into albino, non-isogenic and non-immunocompromised mice, thereby producing knowledge on the geographic distribution, natural habitats and environmental microfoci of pathogenic fungi, especially Histoplasma and Coccidioides spp.
Due to its nature, the animal model works as a biological filter, selecting species or lineages thermo tolerant to 35 - 37°C with metabolic and genetic properties that permit their survival and multiplication in mammalian tissues. Usually, when suspected soil material is inoculated intraperitoneally, the saprobic microbiota composed of bacteria and fungi are blocked and eliminated by the immune system of the inoculated mice. In the presence of fungal agents of systemic mycoses, they may multiply and disseminate to regional lymph nodes and other organs like the lungs, liver, spleen, kidneys, skin and/or central nervous system. Spleen and liver were the organs that allowed the highest positivity for isolating Coccidioides spp. of the inoculated mice .
Coccidioides spp. isolates have been obtained from soil samples of known endemic areas. Usually, the positivity is very low when the samples are collected randomly, even in endemic areas; however, when sampling is directed to a specific suspected site related to cases of acute pulmonary coccidioidomycosis with a consistent epidemiological history of dust inhalation, the probability of obtaining positive samples increases significantly. In fact, such sites may harbor microfoci of Coccidioides spp. where they find suitable ecological conditions to multiply and reach high spore concentrations in restricted areas. These quantitative aspects have been demonstrated for Cryptococcusneoformans and C. gattii through plating onto selective Niger Seed agar (NSA) medium, which allows the concentration of viable fungal propagules to be estimated . Nevertheless, this plating method is not used to detect the agents of coccidioidomycosis due to its high biological risk and because other fast-growing saprobic fungi may lead to misidentification. Thus, there are no adequate tools for estimating the concentration of Coccidioides spp. elements in various substrata, natural habitats or environmental sources related to outbreaks of coccidioidomycosis, where high concentrations of the fungus may exist.
The low frequency of C. immitis isolation from soil samples may be due to seasonal variations or a non-homogeneous distribution in the soil. A study conducted in the US investigated environmental samples collected over eight years in the same endemic area detected the presence of C. immitis, ranging from 0 - 43% . Few environmental isolates of C. immitis and C. posadasii from endemic areas of Mexico and the United States are available for scientific purposes. Recent studies on the phylogeny and molecular epidemiology of Coccidioides spp. were based mainly on clinical isolates from different geographical regions [1, 9]. Therefore, environmental isolates of C. posadasii from semi-arid northeastern Brazil are of interest for these studies.
Regarding the environmental samples collected in and around two excavated armadillo (D. novemcinctus) burrows in Elesbão Veloso and Caridade do Piauí, we obtained positivity rates of 30% and 21.4%, respectively, using the mouse inoculation method. These rates seem very satisfactory when compared to literature data Greene et al. 2000 . The low number of soil samples collected in a specific contaminated habitat excavated during armadillo hunting may have contributed to these results. Moreover, it should be taken into consideration that only a small amount (1 g) from each soil sample was examined after suspending it in 50 mL of saline, from which only 0.5 mL was inoculated into each mouse. Thus, it is possible that viable propagules of Coccidioides spp. present in the sample were not inoculated, producing a false negative result. Beyond the quantitative aspect, the animal model is incapable of detecting lineages unable to grow at 37°C or present in numbers too low to invade and grow in mammalian tissues. On the other hand, propagules with low metabolic activity can remain in latency in soil. In fact, most aspects of the population structure of Coccidioides spp. in the environment remain unknown.
Curiously, during the investigation of the samples from Caridade do Piauí, the same method of animal inoculation permitted the simultaneous isolation of C. posadasii and Cryptococcus neoformans from one soil sample, while C. neoformans was isolated from another soil sample that was negative for C. posadasii. These findings demonstrate the complexity of the fungal microbiota in environmental habitats, such as in this case of D. novemcinctus. These habitats are not exclusive to armadillos, but they are shared with wild rodents, snakes, scorpions, birds and many insects. In the surroundings of this armadillo burrow, it was observed a resting site of Zenaida auriculata, a New World tropical dove endemic to South America, that appear periodically in this region. It is possible that Coccidioides spp., Cryptococcus neoformans and C. gattii, interacting together and/or with other living elements of the soil microbiota, as well as with several other hosts, may generate adaptations and select lineages of these pathogenic fungi. The demonstration of naturally acquired coccidioidomycosis in D. novemcinctus armadillos captured in Piauí reinforces the complexity of this subject . Nevertheless, there have been no investigations of naturally acquired coccidioidomycosis in other species of armadillos, or in other animals such as rodents, foxes, goats, horses, donkeys, cattle and other mammals.
Molecular biological techniques have been used to identify pathogenic fungi. Sandhu et al. (1995) analyzed 116 cultures of several human pathogenic fungi using the universal primers U1 and U2 to amplify the conserved 28S rDNA region of fungi, which was then hybridized with probes specific for each fungal species . Sixteen clinical isolates of C. immitis tested by this method demonstrated 100% positivity in identifying this species.
Another approach used for the identification of isolates of C. immitis is direct PCR using primers with nucleotide sequences based on the gene csa, which is a 19-kDa specific C. immitis antigen secreted in the growth phase of fungal cultures that generates a product of about 519 bp . In another study, Bezerra et al. (2006) obtained 100% positivity analyzing the DNA of 19 cultures of C. immitis: twelve clinical isolates from the state of Piauí and seven isolates preserved for 50-75 years in the culture collection of the Department of Mycology from the Instituto Oswaldo Cruz at FIOCRUZ in Rio de Janeiro .
Regarding the development of molecular methods for the detection of Coccidioides spp. directly in soil samples, obtaining an adequate DNA preparation represented a large challenge. Using mechanical agitation followed by direct cellular enzymatic lysis, we obtained DNA samples with a molecular weight concentrated above 1.5 kb, which were suitable for the amplification reactions by PCR. It should be mentioned that only recently adequate equipment and a Fast DNA SPIN kit for soil (QBIOgene, Carlsbad, CA, USA) allowed the attainment of this suitable DNA from soil samples.
In the present study, the primers designed to detect Coccidioides spp. 28S rDNA in soil took into consideration the low number of copies of the target DNA present in soil. This permitted the detection of Coccidioides spp. 28S rDNA in six isolates from the USA and two from Argentina, as well as in thirteen Brazilian isolates. The molecular detection of any of the Coccidioides species in soil or in clinical specimens is of equal importance.
Optimization of direct PCR with specific primers to detect C. immitis was first performed with DNA extracted from 21 lineages of Coccidioides spp. (eight from the Fungal Culture Collection at IOC/FIOCRUZ, and nine clinical and four environmental isolates from the state of Piauí preserved at the Laboratory of Mycology of IPEC/FIOCRUZ). In this way, we detected a 28S rDNA fragment with a product of nearly 375 bp in 19 out of the 21 isolates tested. However, applying a semi-nested PCR system to these DNA samples with a new pair of primers specific for Coccidioides spp., we detected bands of sizes compatible with the expected fragment in the DNA of all cultures tested. As a control, the DNA of 41 lineages of other human pathogenic fungi (S. schenckii, P. brasiliensis, H. capsulatum, A. niger, A. fumigatus, A. nidulans, B. dermatitidis, M. canis, T. rubrum, T. mentagrophytes, C. neoformans and C. gattii) were submitted to the same protocol, and all results were negative. The results were also negative when the protocol was applied to DNA from bacteria.
Our results indicate the high specificity of PCR with these primers and highlight the increased sensitivity, expected in nested PCR reactions using DNA obtained from soil samples. The next step was to optimize direct PCR with specific primers for detecting Coccidioides spp. in the DNA extracted from our 24 soil samples. The direct PCR method revealed the expected fragment only in 8 (33.3%) soil samples, but when the semi-nested system was used, all the soil samples were positive, thus confirming to be a very sensitive method for detecting Coccidioides spp. 28S rDNA. It is important to note that all of the positive soil samples were collected in and around armadillo burrows strongly suspected to be heavily contaminated because their disturbance caused acute cases of human and canine coccidioidomycosis. It is possible that these restricted sites harbor high concentrations of viable arthroconidia of C. immitis, which are easily detected by animal inoculation, as well as dormant or dead fungal elements with DNA partially preserved, which can only be detected by molecular tools. To evaluate these factors, it should be of interest to analyze soil samples collected in concentric circles from the center of the focus.
As controls for the PCR protocols applied to our soil samples from Piauí, we analyzed DNA extracted from soil samples collected in non-endemic areas of the cities of Goiânia (capital of the state of Goiás) and Brasília (Capital of Brazil), and none presented the 375-bp band, reinforcing our results. Thus, we believe it is important to note that the primer system RFA12 + P2 was able to identify both C. immitis and C. posadasii.
The molecular detection of Coccidioides spp. in suspected soil or in clinical specimens has obvious importance for epidemiological studies and laboratory diagnosis of coccidioidomycosis. Furthermore, molecular procedures such as PCR present substantial advantages, as they reduce the biological risk inherent in the classical techniques and reduce the time necessary to identify a suspected environmental focus or diagnose a clinical case to a few hours. On the other hand, this 28S rDNA marker is not able to distinguish C. immitis from C. posadasii in positive soil samples. However, other markers can be used to detect these specific species. Umeyama et al. (2006) describe species-specific primers for C. immitis based on the ITS1 and ITS4 region, and they were able to differentiate isolates of C. immitis and C. posadasii 
The methodology described in the present study was found to be a sensitive and specific tool for detecting Coccidioides spp. in soil. We believe that the RFA12 + P2 primer system will be useful for epidemiological investigations of clinical cases as well as for environmental studies to identify hazardous sites in Brazil and elsewhere.