Mycoplasma agassizii is a common cause of upper respiratory tract disease in Mojave desert tortoises (Gopherus agassizii). So far, only two strains of this bacterium have been sequenced, and very little is known about its patterns of genetic diversity. Understanding genetic variability of this pathogen is essential to implement conservation programs for their threatened, long-lived hosts. We used next generation sequencing to explore the genomic diversity of 86 cultured samples of M. agassizii collected from mostly healthy Mojave and Sonoran desert tortoises in 2011 and 2012. All samples with enough sequencing coverage exhibited a higher similarity to M. agassizii strain PS6T (collected in Las Vegas Valley, Nevada) than to strain 723 (collected in Sanibel Island, Florida). All eight genomes with a sequencing coverage over 2x were subjected to multiple analyses to detect single-nucleotide polymorphisms (SNPs). Strikingly, even though we detected 1373 SNPs between strains PS6T and 723, we did not detect any SNP between PS6T and our eight samples. Our whole genome analyses reveal that M. agassizii strain PS6T may be present across a wide geographic extent in healthy Mojave and Sonoran desert tortoises.
Citation: Luzuriaga-Neira A, Sandmeier FC, Weitzman CL, Tracy CR, Bauschlicher SN, Tillett RL, et al. (2021) Mycoplasma agassizii, an opportunistic pathogen of tortoises, shows very little genetic variation across the Mojave and Sonoran Deserts. PLoS ONE 16(2): e0245895. https://doi.org/10.1371/journal.pone.0245895
Editor: Chih-Horng Kuo, Academia Sinica, TAIWAN
Received: November 26, 2020; Accepted: January 10, 2021; Published: February 3, 2021
Copyright: © 2021 Luzuriaga-Neira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Sequencing data have been deposited to the NCBI BioProject database (accession number. PRJNA655797).
Funding: This work was supported by grant F19AC00070 from the US Fish and Wildlife Service, and partially supported by Nevada INBRE (funded by grant P20GM103440 from the National Institute of General Medical Sciences, National Institutes of Health). AL and DAP were also supported by a grant from the National Science Foundation (MCB 1818288). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Epidemiological theory offers a basic framework to ask questions about the evolution and genetic diversity of pathogens across the landscape. In particular, highly virulent pathogens in vertebrate hosts may show little genetic diversity due to a new, clonal strain sweeping through a population in an epidemic . When pathogens lead to chronic, non-fatal disease, they often are not subject to intense pressure from the immune system and therefore may acquire more genetic diversity by drift . Both of these models assume strong adaptive immune responses when the immune system is activated. However, ectothermic animals often do not show strong adaptive immunity, and genetic diversity of virulent, as well as chronic, pathogens may be the rule instead of the exception . Furthermore, different microbes may also have very different evolutionary patterns. For example, mycoplasmas (Mollicutes) include both pathogenic and nonpathogenic species, and evolution across the group is marked by reduction of the genome and horizontal gene transfer .
Epidemiological research has shown that evolution and genetic diversity may vary drastically among host–mycoplasma systems. For example, when Mycoplasma gallisepticum was first introduced to wild passerine populations, selection appeared to favor increasing virulence . In contrast, the human pathogen Mycoplasma pneumoniae shows fewer genetic differences among strains, despite a global distribution . Mycoplasmal diseases provide excellent systems for studying genetic diversity of pathogens, as they often exhibit naturally high mutation rates, likely due to a reduction of repair mechanisms for DNA [6,7].
Here, we investigate the genetic diversity of a mycoplasma, Mycoplasma agassizii, that causes upper respiratory tract disease (URTD) in Mojave desert tortoises (Gopherus agassizii) in the USA. While M. agassizii can occur at very high frequencies in some populations, it only causes disease when it reaches high loads within animals [8,9]. Tortoises can mount an adaptive immune response to M. agassizii, but it is not effective at clearing the pathogen and can result in recrudescence of the disease . Such patterns of chronic disease are not unique to tortoises and occur in many other vertebrate animals due to diseases caused by mycoplasmas . However, how tortoises reduce loads of silent infections is currently unknown, but likely involves innate immunity and is possibly mediated by innate B1 lymphocytes and natural antibodies . We have an incomplete understanding of the factors leading to increased infection intensity and disease, and whether disease is driven primarily by host immune function or by differences in the strains of M. agassizii. Conserved regions of the M. agassizii genome (16S rRNA, the 16-23S intergenic spacer region, and RNA polymerase ß subunit) showed some single nucleotide polymorphisms (SNPs), but very low diversity across the range of the Mojave desert .
One problem in understanding this disease in wild animals is that M. agassizii is hard to culture, and researchers currently use liquid cultures in SP4 broth, which can require multiple weeks after isolation from tortoises to grow to mid-log phase . In addition, isolation rates of the microbe from tortoises are inconsistent, and liquid culturing of mucosal samples that contain many different types of bacteria often results in some contamination (Sandmeier et al., unpublished data). However, during the course of a wide-ranging survey of disease in desert tortoises in 2010–2012 [9,13], we used a modified technique to attempt to culture M. agassizii from 106 animals, representing 12 populations of Mojave desert tortoises and 2 populations of Sonoran desert tortoises (Gopherus morafkai).
To quantify genetic differences among strains, we used a whole-genome next-generation sequencing (NGS) technique to compare wild samples to previously published genomes for two strains of M. agassizii (PS6T and 723) and one strain of the closely-related microbe, Mycoplasma testudineum (BH29T) [3,15]. PS6T was first collected in 1991 from a sick Mojave desert tortoise in the Las Vegas Valley, 723 was isolated from a sick gopher tortoise (Gopherus polyphemus) in Florida, and M. testudineum BH29T was isolated from a Mojave desert tortoise but does not seem to cause respiratory disease [9,16–19].
We hypothesized that strains would show genetic diversity, as predicted by epidemiological theory and tortoise immune defenses against M. agassizii. Most of the sampled animals that were positive for M. agassizii showed no signs of disease, and our study took place in a time without epizootic conditions [9,20]. Thus, we expected that the diversity among strains would be mostly due to geographic distance. We also predicted that strains collected in 2011–2012 would show some divergence from PS6T, due to evolution since PS6T was collected. Finally, we predicted that strains would be more closely related to PS6T than to 723. We also hypothesized that we would find genetic differences in strains found in tortoises with signs of disease compared to those without. Similarly, we expected that we would find genetic differences in strains isolated from tortoise populations with a higher prevalence of disease and M. agassizii than in populations with a low signature of disease (sensu Sandmeier et al. ).
Materials and methods
All institutional and national guidelines for the care and use of animals were followed and included permits: IACUC (00465 and 0554), Nevada Division of Wildlife (S33080), California Fish and Game (SC-007374), Utah Division of Wildlife (5COLL8886), U.S. Fish and Wildlife Service (TE030659) and National Park Service (MOJA-2012, ZION-2012-SCI-006), Arizona Game and Fish Department Permit (SP793245).
Sample collection and culture
We collected nasal lavage samples, as previously described , from free-ranging Mojave and Sonoran desert tortoises in 2011–2012. Lavage samples were filtered through a 0.45 μm filter into SP4 broth and placed on ice in the field. In a sterile environment, these samples were further filtered through a 0.1 μm filter and incubated at 30˚C in a portable incubator. We used the protocol described by ATCC (https://www.atcc.org/products/all/700616.aspx), and to our knowledge, other techniques have not been verified for this organism [14,17]. In 2011, culture samples were sent on ice to collaborators at the University of Nevada, Reno for which these eight samples originated [9,13,20].
DNA extraction and genomic sequencing
Genomic DNA was extracted using the Qiagen DNeasy Blood and Tissue protocol (Qiagen, Valencia, CA) for Gram-negative bacteria and eluted with ultrapure water. Extracted DNA was quantified using a Qiagen QIAxpert system and Picogreen assay. The presence of Mycoplasma spp. was verified with qPCR , with the assumption that either or both of M. agassizii and M. testudineum may have grown from the tortoise sample cultures. Genomic sequencing was performed using the Illumina Nextera XT DNA Library Preparation Kit (Illumina, Inc., San Diego, USA) with the Illumina NextSeq500 platform (150bp, paired-end) and up to 2 ng of DNA per sample at the Nevada Genomics Center (University of Nevada, Reno). Sequencing was achieved in a multiplex system, using dual index sequences from the Illumina Nextera XT Index kit (index 1, N701 to N715; index 2 S502 to S511). Raw sequences were deposited in the NCBI Sequence Read Archive (SRA) database under bioproject PRJNA655797.
Sequencing reads were demultiplexed, assigning each read to its respective culture sample and removing barcodes. Because we detected some contamination from other bacteria, we used the bbsplit.sh script from the BBtools program , with default parameters, to extract Mycoplasma-like reads. This software extracts the sequences that are similar to a set of reference genomes (identity ≥ 35%) using a local pairwise alignment method. Three reference mycoplasma genomes were used to extract Mycoplasma-like reads from the raw fastq files: two M. agassizii genomes (strains PS6T and 723, with NCBI accession numbers NQMN00000000 and NQNY00000000, respectively) and one M. testudineum genome (strain BH29T; NNCE00000000) [3,15]. The fastQC program v.0.11.8  was used to evaluate the quality of the filtered fastq files.
To determine which of the three reference genomes is the most closely related to each field culture, the filtered reads were aligned to each of the three genomes using the NCBI-magicblast v1.3.0 software. For most samples, most reads matched M. agassizii PS6T –the only cases in which most reads matched M. agassizii 723 or M. testudineum BH29T were cases with very few Mycoplasma-like reads. Thus, we used M. agassizii PS6T as reference in all subsequent analyses. To evaluate the accuracy of our method, we repeated our analyses using the reads previously generated in our labs for sequencing the reference M. agassizii and M. testudineum genomes [3,15].
For each sample, all filtered reads were aligned against the M. agassizii PS6T genome for SNP identification. The backtrack algorithm from the Burrows-Wheeler Aligner (BWA) , with default parameters, was used to generate sequence alignment files (SAM files). The Sequencing Alignment/MAP software SAMtools v. 1.4  was employed to generate, sort, index, and remove duplicates from the binary alignment files (BAM files). Only reads that mapped with a map quality score over 30 were retained. This map quality score threshold reduces the probability of mapping/aligning a random read to a reference genome to 0.001. For each alignment, basic statistics were calculated using the stats function from the SAMtools package. We estimated the genome coverage by averaging the per nucleotide position coverage (depth) by the PS6T across all genome nucleotides. We classified our cultured samples into four categories according to their sequencing coverage: high (> 40x, n = 1), medium (25–40x, n = 2), low (2–25x, n = 5) coverage genomes, and samples with a very low number of Mycoplasma-like reads or very low coverage (< 2x, n = 30). De novo assembly was also performed as a complementary method to detect the presence of variants between the studied samples; to that end we used the SPAdes genome assembler v.3.9 set to default parameters.
We used the Analysis of Next Generation Sequencing Data (ANGSD) software v. 0.9 for SNP calling. The methods incorporated in ANGSD are more suitable for low and medium coverage data, because they are based on genotype likelihoods avoiding uncertainty in SNP identification typical of poor sequencing data . We conducted different analyses to screen for genetic variation among our samples depending on their sequencing coverage. First, we performed pairwise comparisons between the M. agassizii PS6T genome and each of the high and medium coverage samples (n = 3). For this analysis, we used the SAMtools genotype likelihood model (GL 1) implemented in ANGSD, to identify SNPs only in sites with a high probability of being polymorphic (SNP_pval < 10−6). In a second analysis, we performed SNP calling from a pooled dataset containing all high and medium coverage samples (n = 3). For these first two analyses, we tested several depth filters (setMinDepth), considering only sites with coverages of at least 5, 10, 15, and 20x. Finally, we conducted a similar variant identification analysis pooling all high, medium, and low coverage samples (n = 8) and both the PS6T and 723 genomes. Pooling the data reduces the likelihood of sequencing errors being interpreted as SNPs. For this analysis, we used the same filters mentioned above but replaced the minimal depth filter per site by the minimal depth by sample (minIndDepth = 2). In addition, we considered only sites with more than three reads supporting a site in each sample. To avoid the false-positive identification caused by the low coverage sequencing, we only considered a variant as a SNP if it was present in more than 50% of the individuals (minInd > 5).
A maximum-likelihood phylogenetic tree was obtained using the polymorphic sites present among the high, medium, and low coverage samples and the PS6T and 723 reference genomes. The tree was computed using the Molecular Evolutionary Genetics Analysis (MEGA) X software , using default parameters.
For the eight genome samples with low to high coverage, we evaluated whether or not the samples were collected from animals with active URTD infections, as scored by clinical signs of fresh or dried mucus in and around the nares of the animals . We also determined whether there was a difference in disease or pathogen prevalence among the populations from which these eight samples originated [9,13,20]
Mycoplasma spp. presence in desert tortoise isolates
Mycoplasma agassizii is a small, aerobic, slow-growing species, requiring weeks-to-months of culturing for visible growth. Due to its difficulty to culture, attempts to grow M. agassizii from mucus samples containing many microbes may result in false-negatives or non-pure cultures. We collected and attempted to culture M. agassizii from 92 Mojave desert tortoises and 14 Sonoran desert tortoises in 2011 and 2012. We extracted DNA from 86 cultures that appeared to have replicated and reached mid-log growth at the expected timescale. This resulted in 75 DNA samples from Mojave desert tortoises and 11 DNA samples from Sonoran tortoises that were subjected to Illumina sequencing. These samples included 67 that tested as positive for M. agassizii using qPCR after culturing. After the resulting sequencing reads were filtered using the genomes of M. agassizii (strains 723 and PS6T) and M. testudineum (strain BH29T) as reference, we obtained 214 to 688,878 Mycoplasma-like reads per sample in 38 samples, 37 of which were qPCR positive for M. agassizii (Tables 1 and 2 and S1). The 48 remaining samples were discarded after the filtering step due to the lack of Mycoplasma-like reads, which we attribute to the difficulty of culturing mycoplasmas. Eight of the samples had over 21,000 Mycoplasma-like reads and were used in analyses to identify SNPs, as we detailed in Table 1.
For each sample, we counted the number of reads that perfectly matched (with 100% identity) each of the three reference genomes (Table 1). We verified the validity of this analysis by confirming that reads from the reference strains correctly matched their expected genomes: most reads resulting from sequencing of the M. agassizii PS6T strain (data from ref. ) matched the PS6T genome, most reads resulting from sequencing of the M. agassizii 723 strain (data from ref. ) matched the 723 genome, and most reads resulting from sequencing of the M. testudineum BH29T strain (data from ref. ) matched the BH29T genome (Table 1). The three reference strains/species also showed significant overlap among sequence reads (some reads perfectly matched two or all three genomes), indicating that there are conserved regions among these three genomes (Table 1), as expected from the fact that they are closely related [3,15]. Thus, some of the reads derived from sequencing one of the reference genomes matched not only that genome, but also one or the two other reference genomes.
For all of the 14 samples with more than 5,752 Mycoplasma-like reads (Tables 1 and S2), the numbers of reads matching the M. agassizii PS6T genome were higher than those matching the M. agassizii 723 genome, and much higher for either M. agassizii genome than for the M. testudineum BH29T genome (Table 1). This indicates that the field cultures are more closely related to M. agassizii PS6T than to the other strain of M. agassizii or to M. testudineum. In the 24 samples with fewer than 5,752 total Mycoplasma-like reads, we found that 6 had more reads matching M. agassizii 723 than reads matching M. agassizii PS6T, but all had fewer reads that matched M. testudineum.
Given that all higher-quality field culture sequencing datasets matched the M. agassizii PS6T genome better than the M. agassizii 723 or M. testudineum BH29T genomes (Table 1), all subsequent analyses were conducted using the PS6T genome as reference. As depicted in Fig 1 and Table 1, one sample had high coverage (61x), two samples had medium coverage (25x and 39x), five samples had low coverage (5x to 2x), and the remaining 30 samples had very low coverage (lower than 2x; Tables 2 and S1 and Fig 1). Low coverage samples still indicated similarity to M. agassizii PS6T and represent diverse tortoise populations across the Mojave and Sonoran Deserts (S1 Table and Fig 1).
Lack of genetic variation among Mycoplasma agassizii isolates from desert tortoises
We aligned all the reads resulting from the high (n = 1), medium (n = 2) and low (n = 5) coverage samples to the reference M. agassizii PS6T genome for SNP identification. The SNP calling for these eight samples covered between 54% and 99.9% of the PS6T genome used as reference. Pairwise comparisons between the reference M. agassizii PS6T genome and each of the high and medium coverage sampled genomes did not identify any SNPs. Similarly, analysis of a pooled dataset containing all high and medium coverage genomes (n = 3) did not reveal any SNPs. When we used more stringent criteria for SNP calling, but also included the 5 low coverage genomes (n = 8), we still did not identify any SNPs. The absence of SNPs was confirmed by visual inspection of the alignments using the Integrative Genomics Viewer program (IGV v2.3; ref. )–no potential SNP was supported by more than 2 reads. The lack of variation was confirmed for the three samples with the higher number of reads. To that end, we performed de novo assembling and compared the resulting contigs to the PS6T genome using the reordering function from the ABACAS algorithm . Comparison of the genomes of M. agassizii strains PS6T and 723 and our samples, however, resulted in 1373 SNPs, which resulted from differences specific to M. agassizii 723. A maximum likelihood phylogenetic tree of all our high, medium and low coverage samples and the two M. agassizii reference genomes is shown in Fig 2.
Table 3 depicts the disease status of these same eight samples and their respective populations, previously described in Weitzman et al. [13,20] and Sandmeier et al. . None of these tortoises exhibited signs of disease at the time of sampling. Within the Mojave Desert, these eight tortoise populations represent an average level of clinical disease from all 24 populations that were surveyed over this same time period (0–25% prevalence of disease; refs. [9,13]). In addition, they span almost the entire range of prevalence of M. agassizii also quantified in 24 populations across the range (0–90% prevalence of M. agassizii; refs. [9,13]).
To our knowledge, this is the first study that has successfully cultured and described M. agassizii from predominantly subclinical tortoises across the entire range of Mojave desert tortoises as well as some Sonoran desert tortoise locations. Despite a limited number of samples with deep coverage, we used conservative techniques to identify samples as identical or nearly identical to the type culture isolate PS6T. This was surprising, given the high genetic variation in most species of mycoplasmas . Disease in tortoises has been shown to be caused by both recrudescence and transmission, which would lead to predictions of evolution within the host, transmission of variable strains among individuals, and high genetic variation , a pattern seen in other chronic mycoplasmal diseases [30,31]. Similarly, at the time that PS6T was isolated from tortoises in the Las Vegas Valley, a variety of cultured strains were identified from different tortoises based on phenotypic characteristics . However, it is important to emphasize that we filtered all our field samples prior to culturing in an effort to reduce contamination. Brown et al.  showed that not all isolates cultured equally well after passage through filters. Below, we discuss two alternate interpretations of our data. (1) PS6T is indeed the most common genetic strain of M. agassizii in Mojave and possibly also in Sonoran desert tortoises. (2) PS6T cultures more readily than other strains, and we thus detected this strain in many of our sampled tortoises. We compare the patterns described with more well-characterized host–mycoplasma systems in agricultural animals and humans. Regardless of this unresolved pattern, we found that Mojave desert tortoises seem to differ in their endemic M. agassizii from strain 723 that originated in gopher tortoises . Finally, we suggest other tools for studying this disease given current limitations in reliable culturing techniques.3.1. Mycoplasma spp. presence in desert tortoise isolates.
Alternate explanations for patterns of little genetic variation
Eight samples from diverse locations (Fig 1) had >21,000 Mycoplasma-like sequencing reads, or a coverage greater than 2x (Tables 1 and S2). In these sequences we found that most reads matched PS6T and not 723 (Tables 1 and S2). Furthermore, most of the reads that did match 723 also matched PS6T, indicating that they fall within conserved regions of these two similar genomes . Strikingly, we identified no SNPs, even though Weitzman et al.  found a very low frequency of SNPs in conserved regions of the genome in uncultured samples from these same populations (nonetheless, the samples used by Weitzman et al. do not overlap with the ones used here). However, Weitzman et al.  conducted Sanger sequencing on 25–51 samples, representing 16S rRNA, the 16-23S intergenic spacer region, and RNA polymerase ß subunit. Therefore, our study corroborates previous results that even in ribosomal regions, M. agassizii has relatively little genetic variation–even though mycoplasmas generally have much higher variation in ribosomal DNA than is found in other organisms [6,7,32].
Because we observed a very low incidence of disease during our three-year survey of Mojave and Sonoran desert tortoises [9,20], we do not attribute the pattern of little genetic diversity to clonal spread of a virulent mycoplasmal strain (e.g., ref. ). Although PS6T was isolated from a wild tortoise with signs of URTD in Las Vegas in 1991 [17,34], it is of unknown virulence and is not the highly virulent strain that was used in an earlier infection experiment that led to rapid, consistent disease in all exposed tortoises . To our knowledge, that isolate has not been preserved or genetically sequenced, despite its crucial role in identifying M. agassizii as the etiological agent of URTD by the almost complete fulfillment of Koch’s postulates . Of course, current genetic technology did not exist at the time when outbreaks of disease were most severe (1989–1991) and strains were intentionally cultured from epizootic areas for experimental investigations of URTD’s etiology (reviewed in refs. [35,36]).
If PS6T is the most dominant strain of M. agassizii in the Mojave and Sonoran Deserts, that would suggest stabilizing selection or low mutation rates, in particular because there has been so little change over time (18 years since PS6T was originally collected) and space (the extent of the Mojave desert; Fig 1). Interestingly, other mycoplasmal diseases show more genetic variation but also often do not show evolutionary patterns related to geography, time, or disease status of individuals [30,37,38]. More similar to our own results, Xiao et al.  found high genetic similarity of samples of another chronic respiratory pathogen, M. pneumoniae, in human populations. Such patterns could occur in systems where low virulence or selection for a constant phenotype results in higher fitness for the pathogen. However, high similarity between M. pneumoniae genomes is associated to their geographical distribution and could also depend on the circulation period. For instance, a recent study detected up to 5% variability between genomes from different circulation times and sampling locations across South Korea . Low mutation rates have not been documented for other species of mycoplasmas, and the class Mollicutes shows some of the highest mutation rates of any group of prokaryotes . In these other species, high mutation rates have been attributed to loss of DNA-repair mechanisms [6,7]. If M. agassizii truly does have low mutation rates, it would be interesting to test whether selection pressure could have led to more stable DNA replication and repair in this species.
Indeed, M. agassizii does not seem to be very virulent during average or stable environmental conditions. For example, transmission of M. agassizii between tortoises is so slow that it requires 24–48 hours of close contact . In addition, the majority of tortoises that test positive for M. agassizii show no signs of disease, have very low microbial loads, and seem to be chronic carriers of the microbe [8,10]. Because tortoise adaptive immune responses are ineffective at clearing M. agassizii, such host responses are unlikely to act as a selective pressure for higher virulence or evasion of the host immune system [1,2]. In fact, it may lead to higher fitness of the pathogen to remain relatively avirulent and increase the likelihood of transmission to another host . However, PS6T was initially collected from a diseased individual, and it may represent a strain that can become more virulent or transmit more easily at higher loads [8,9]. Stabilizing selection or selection for small phenotypic changes may be part of maintaining chronic infections and the suspected long co-evolution between this host and its pathogen [2,12].
The alternate, not mutually-exclusive explanation is that multiple strains, including PS6T, are maintained in individual tortoises or populations, but we lacked the tools to identify them. Because our field samples had low initial cell numbers, we needed to first culture the samples before extracting and sequencing DNA. However, recent research using genetic techniques in the absence of culturing clinical samples shows that identification of multi-strain infections and a greater diversity of strains is becoming increasingly common [40,41]. Culturing microbial samples often biases results toward the fastest growing strain, thus leading to erroneous conclusions about the true natural diversity of strains [42,43]. Our use of two filtering steps and the culture conditions recommended specifically for PS6T may have caused us to select for this strain from our wild samples . While contamination of cultures could also explain recovery of the same genetic strain, samples were collected, cultured, and frozen at disparate and non-overlapping times. Similarly, they were not exposed to the cultures of the type strain PS6T.
General patterns and future directions
We found two other patterns in our dataset. First, we did not detect any sample that clearly identified more closely to 723 than to PS6T. Furthermore, the reads that did align with the strain originally isolated from a gopher tortoise from Sanibel Island, Florida (strain 723) were from regions of the genome that are conserved between the two isolates (Tables 1 and S2). This suggests that, similar to other mycoplasmas that infect multiple species, there is strain variation among host species [44–46]. However, we found no variation in M. agassizii from Mojave and Sonoran desert tortoises. Whether this is due to closer genetic similarity between the two desert tortoise species versus gopher tortoises, or due to geographic distance–with PS6T predominant in the West and 723 or another strain predominant in the East–remains to be determined.
The second pattern is that all of the eight best-quality sequencing datasets were obtained from tortoises without any signs of disease (Table 3). These tortoises spanned populations with moderate disease prevalence and wide variation of M. agassizii prevalence, representative of values found throughout the region (Table 3; refs. [9,13,20]). Thus, PS6T is found across the Mojave and Sonoran deserts, regardless of levels of higher or lower disease in distinct populations. If strains of M. agassizii are homogenously spread across the two deserts, then it seems unlikely that pathogen genetics predicts the variable patterns of disease observed in tortoises [35,36]. Given recent evidence of differences in Mojave tortoise immune parameters and their relationship to load of M. agassizii and prevalence of disease across populations, it is likely that the drivers of disease reside primarily in the host immune function, triggered when M. agassizii load is high [9,13,20]. Experimental studies are needed to tease apart these factors, and possibly relate them to different environmental conditions. For example, both host and pathogen factors have been shown to drive mycoplasmal disease in house finches. Differences in virulence of M. gallisepticum and not host immunity appear to drive prevalence of disease in some wild populations . Additionally, it has also been shown that host tolerance and resistance have evolved in response to pathogen pressure [4,48,49].
Finally, we suggest using additional, culture-independent genetic tools to answer the question of whether mixed strain infections are common in the respiratory tract of tortoises. The fact that researchers can extract only small quantities of this microbe from the nares of live animals is a characteristic that is shared with many other wildlife diseases . Similarly, the fastidious nature of M. agassizii in culture is also shared by many mycoplasmas and other microbes [11,51]. We suggest that we can use the published genomes of M. agassizii and M. testudineum to identify genetic regions with expected high rates of variability (e.g., refs. [3,15,44]). In particular, we may see differences in virulence genes and genes associated with growth rates across isolates . Like most other species of mycoplasma, M. agassizii has only one copy of 16S rRNA, and the qPCR for this gene can thus be used to quantify the number of cells within a sample [6,52]. If we then create a series of qPCR assays for potential virulence and growth factors in PS6T, we can test for a gain or loss of those genes per M. agassizii cell in specimens obtained from wild animals. Similar assays have been used to detect loss and gain of virulence genes in human and livestock diseases [53–55]. Gain and loss of genes may be common in mycoplasmas , and quantifying such changes will allow us to test whether tortoises have mixed-strain infections. Quantitative PCR is especially powerful in analyzing samples with relatively low quantities of DNA, but has only recently been used for quantifying disease and genetic differences in and among samples, (e.g., refs. [52,55]).
In conclusion, we verified that M. agassizii strain PS6T exists in subclinical Mojave and Sonoran desert tortoises. This pattern suggests that enzootic, not epizootic, disease conditions may be the norm across these populations, at least in recent times. However, we also suggest that culturing samples may have skewed our data and will create a similar problem in other research. We suggest that qPCR may be an alternate approach to explore true diversity of mycoplasmal infections within uncultured samples taken from both subclinical and diseased wild animals.
S1 Table. Results of qPCR reactions for cultured field samples using Mycoplasma agassizii primers and probe.
The authors are grateful to Kris Kruse from the Nevada Genomics Center for technical assistance, and to the Nevada Genomics Center for providing sequencing services for free. We also thank Sally DuPre, Hamid Mohammadpour, and Nichole Maloney for their assistance in culturing these samples, and Julia Allen for helpful discussions.
- 1. Dykhuizen D, Kalia A. The population structure of pathogenic bacteria. Evolution in Health and Disease, ed Stearns SC, Koella JC. 2007:185–98.
- 2. Sandmeier FC, Tracy RC. The metabolic pace-of-life model: incorporating ectothermic organisms into the theory of vertebrate ecoimmunology. Integr Comp Biol. 2014;54(3):387–95. Epub 2014/04/23. pmid:24760792.
- 3. Weitzman CL, Tillett RL, Sandmeier FC, Tracy CR, Alvarez-Ponce D. High quality draft genome sequence of Mycoplasma testudineum strain BH29T, isolated from the respiratory tract of a desert tortoise. Stand Genomic Sci. 2018;13:9. Epub 2018/04/11. pmid:29682170; PubMed Central PMCID: PMC5896096.
- 4. Tardy L, Giraudeau M, Hill GE, McGraw KJ, Bonneaud C. Contrasting evolution of virulence and replication rate in an emerging bacterial pathogen. Proc Natl Acad Sci U S A. 2019;116(34):16927–32. Epub 2019/08/01. pmid:31371501; PubMed Central PMCID: PMC6708350.
- 5. Xiao L, Ptacek T, Osborne JD, Crabb DM, Simmons WL, Lefkowitz EJ, et al. Comparative genome analysis of Mycoplasma pneumoniae. BMC genomics. 2015;16(1):610. pmid:26275904
- 6. Razin S, Yogev D, Naot Y. Molecular biology and pathogenicity of mycoplasmas. Microbiol Mol Biol Rev. 1998;62(4):1094–156. pmid:9841667; PubMed Central PMCID: PMC98941.
- 7. Rocha EP, Blanchard A. Genomic repeats, genome plasticity and the dynamics of Mycoplasma evolution. Nucleic Acids Res. 2002;30(9):2031–42. pmid:11972343; PubMed Central PMCID: PMC113839.
- 8. Aiello CM, Nussear KE, Esque TC, Emblidge PG, Sah P, Bansal S, et al. Host contact and shedding patterns clarify variation in pathogen exposure and transmission in threatened tortoise Gopherus agassizii: implications for disease modelling and management. J Anim Ecol. 2016;85(3):829–42. Epub 2016/04/10. pmid:26935482.
- 9. Sandmeier FC, Weitzman CL, Tracy CR. An ecoimmunological approach to disease in tortoises reveals the importance of lymphocytes. Ecosphere. 2018;9(9):e02427.
- 10. Sandmeier FC, Maloney KN, Tracy CR, Hyde D, Mohammadpour H, Marlow R, et al. Chronic disease in the Mojave desert tortoise: Host physiology and recrudescence obscure patterns of pathogen transmission. Ecol Evol. 2017;7(24):10616–29. Epub 2017/11/04. pmid:29299243; PubMed Central PMCID: PMC5743692.
- 11. Simecka JW. Mycoplasma diseases of animals. Mycoplasmas, molecular biology and pathogenesis. 1992:391–416.
- 12. Sandmeier FC, Tracy CR, Hagerty BE, DuPré S, Mohammadpour H, Hunter K. Mycoplasmal upper respiratory tract disease across the range of the threatened Mojave Desert tortoise: associations with thermal regime and natural antibodies. Ecohealth. 2013;10(1):63–71. Epub 2013/04/12. pmid:23579813.
- 13. Weitzman CL, Sandmeier FC, Tracy CR. Prevalence and diversity of the upper respiratory pathogen Mycoplasma agassizii in Mojave desert tortoises (Gopherus agassizii). Herpetologica. 2017;73(2):113–20.
- 14. Brown MB, Schumacher IM, Klein PA, Harris K, Correll T, Jacobson ER. Mycoplasma agassizii causes upper respiratory tract disease in the desert tortoise. Infect Immun. 1994;62(10):4580–6. pmid:7927724; PubMed Central PMCID: PMC303146.
- 15. Alvarez-Ponce D, Weitzman CL, Tillett RL, Sandmeier FC, Tracy CR. Mycoplasma agassizii strains PS6T and 723 isolated from Gopherus tortoises with upper respiratory tract disease. Stand Genomic Sci. 2018;13:12. Epub 2018/04/27. pmid:29725499; PubMed Central PMCID: PMC5921776.
- 16. Brown DR, Merritt JL, Jacobson ER, Klein PA, Tully JG, Brown MB. Mycoplasma testudineum sp. nov., from a desert tortoise (Gopherus agassizii) with upper respiratory tract disease. Int J Syst Evol Microbiol. 2004;54(Pt 5):1527–9. pmid:15388705.
- 17. Brown MB, Brown DR, Klein PA, McLaughlin GS, Schumacher IM, Jacobson ER, et al. Mycoplasma agassizii sp. nov., isolated from the upper respiratory tract of the desert tortoise (Gopherus agassizii) and the gopher tortoise (Gopherus polyphemus). Int J Syst Evol Microbiol. 2001;51(Pt 2):413–8. pmid:11321087.
- 18. Brown MB, McLaughlin GS, Klein PA, Crenshaw BC, Schumacher IM, Brown DR, et al. Upper respiratory tract disease in the gopher tortoise is caused by Mycoplasma agassizii. J Clin Microbiol. 1999;37(7):2262–9. pmid:10364595; PubMed Central PMCID: PMC85132.
- 19. Brown DR, Crenshaw BC, McLaughlin GS, Schumacher IM, McKenna CE, Klein PA, et al. Taxonomic analysis of the tortoise mycoplasmas Mycoplasma agassizii and Mycoplasma testudinis by 16S rRNA gene sequence comparison. Int J Syst Bacteriol. 1995;45(2):348–50. pmid:7537069.
- 20. Weitzman CL, Gov R, Sandmeier FC, Snyder SJ, Tracy CR. Co-infection does not predict disease signs in Gopherus tortoises. R Soc Open Sci. 2017;4(10):171003. Epub 2017/10/18. pmid:29134096; PubMed Central PMCID: PMC5666279.
- 21. Bushnell B. BBTools software package URL http://sourceforge.net/projects/bbmap. 2014.
- 22. Andrews S. FastQC: a quality control tool for high throughput sequence data URL: http://www.bioinformatics.babraham.ac.uk/projects/fastqc. 2010.
- 23. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60. Epub 2009/05/18. pmid:19451168; PubMed Central PMCID: PMC2705234.
- 24. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics. 2009;25(16):2078–9. Epub 2009/06/08. pmid:19505943; PubMed Central PMCID: PMC2723002.
- 25. Korneliussen TS, Albrechtsen A, Nielsen R. ANGSD: Analysis of Next Generation Sequencing Data. BMC Bioinformatics. 2014;15:356. Epub 2014/11/25. pmid:25420514; PubMed Central PMCID: PMC4248462.
- 26. Kumar S, Stecher G, Li M, Knyaz C, Tamura K. MEGA X: Molecular Evolutionary Genetics Analysis across Computing Platforms. Mol Biol Evol. 2018;35(6):1547–9. pmid:29722887; PubMed Central PMCID: PMC5967553.
- 27. Thorvaldsdóttir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV): high-performance genomics data visualization and exploration. Brief Bioinform. 2013;14(2):178–92. Epub 2012/04/19. pmid:22517427; PubMed Central PMCID: PMC3603213.
- 28. Assefa S, Keane TM, Otto TD, Newbold C, Berriman M. ABACAS: algorithm-based automatic contiguation of assembled sequences. Bioinformatics. 2009;25(15):1968–9. Epub 2009/06/03. pmid:19497936; PubMed Central PMCID: PMC2712343.
- 29. Woese CR, Stackebrandt E, Ludwig W. What are mycoplasmas: the relationship of tempo and mode in bacterial evolution. J Mol Evol. 1984;21(4):305–16. pmid:6085735.
- 30. Königsson MH, Bölske G, Johansson KE. Intraspecific variation in the 16S rRNA gene sequences of Mycoplasma agalactiae and Mycoplasma bovis strains. Vet Microbiol. 2002;85(3):209–20. pmid:11852188.
- 31. Marques LM, Buzinhani M, Guimaraes AM, Marques RC, Farias ST, Neto RL, et al. Intraspecific sequence variation in 16S rRNA gene of Ureaplasma diversum isolates. Vet Microbiol. 2011;152(1–2):205–11. Epub 2011/04/16. pmid:21601382.
- 32. Woese C, Stackebrandt E, Ludwig W. What are mycoplasmas: the relationship of tempo and mode in bacterial evolution. Journal of molecular evolution. 1985;21(4):305–16.
- 33. Becker CA, Thibault FM, Arcangioli MA, Tardy F. Loss of diversity within Mycoplasma bovis isolates collected in France from bovines with respiratory diseases over the last 35 years. Infect Genet Evol. 2015;33:118–26. Epub 2015/04/23. pmid:25913158.
- 34. Atcc.org. 2020. Mycoplasma agassizii. Brown et al. ATCC ® 700616™. [online] Available at: <https://www.atcc.org/Products/All/700616.aspx#history> [Accessed 24 November 2020].
- 35. Sandmeier FC, Tracy CR, Hunter K. Upper respiratory tract disease (URTD) as a threat to desert tortoise populations: A reevaluation. Biological Conservation. 2009;142(7):1255–68.
- 36. Jacobson ER, Brown MB, Wendland LD, Brown DR, Klein PA, Christopher MM, et al. Mycoplasmosis and upper respiratory tract disease of tortoises: a review and update. Vet J. 2014;201(3):257–64. Epub 2014/06/04. pmid:24951264.
- 37. Buim MR, Buzinhani M, Yamaguti M, Oliveira RC, Mettifogo E, Timenetsky J, et al. Intraspecific variation in 16S rRNA gene of Mycoplasma synoviae determined by DNA sequencing. Comp Immunol Microbiol Infect Dis. 2010;33(1):15–23. Epub 2008/09/04. pmid:18775565.
- 38. Fookes MC, Hadfield J, Harris S, Parmar S, Unemo M, Jensen JS, et al. Mycoplasma genitalium: whole genome sequence analysis, recombination and population structure. BMC Genomics. 2017;18(1):993. Epub 2017/12/28. pmid:29281972; PubMed Central PMCID: PMC5745988.
- 39. Lee JK, Seong MW, Shin D, Kim JI, Han MS, Yeon Y, et al. Comparative genomics of Mycoplasma pneumoniae isolated from children with pneumonia: South Korea, 2010–2016. BMC Genomics. 2019;20(1):910. Epub 2019/11/29. pmid:31783732; PubMed Central PMCID: PMC6884898.
- 40. Cummings LA, Kurosawa K, Hoogestraat DR, SenGupta DJ, Candra F, Doyle M, et al. Clinical Next Generation Sequencing Outperforms Standard Microbiological Culture for Characterizing Polymicrobial Samples. Clin Chem. 2016;62(11):1465–73. Epub 2016/09/13. pmid:27624135.
- 41. Farmanfarmaei G, Kargarpour Kamakoli M, Sadegh HR, Masoumi M, Abdolrahimi F, Fateh A, et al. Bias in detection of Mycobacterium tuberculosis polyclonal infection: Use clinical samples or cultures? Mol Cell Probes. 2017;33:1–3. Epub 2017/01/24. pmid:28130122.
- 42. Gorski L. Selective enrichment media bias the types of Salmonella enterica strains isolated from mixed strain cultures and complex enrichment broths. PLoS One. 2012;7(4):e34722. Epub 2012/04/04. pmid:22496847; PubMed Central PMCID: PMC3319605.
- 43. Plazzotta G, Cohen T, Colijn C. Magnitude and sources of bias in the detection of mixed strain M. tuberculosis infection. J Theor Biol. 2015;368:67–73. Epub 2014/12/29. pmid:25553967; PubMed Central PMCID: PMC7011203.
- 44. Delaney NF, Balenger S, Bonneaud C, Marx CJ, Hill GE, Ferguson-Noel N, et al. Ultrafast evolution and loss of CRISPRs following a host shift in a novel wildlife pathogen, Mycoplasma gallisepticum. PLoS Genet. 2012;8(2):e1002511. Epub 2012/02/09. pmid:22346765; PubMed Central PMCID: PMC3276549.
- 45. Tulman ER, Liao X, Szczepanek SM, Ley DH, Kutish GF, Geary SJ. Extensive variation in surface lipoprotein gene content and genomic changes associated with virulence during evolution of a novel North American house finch epizootic strain of Mycoplasma gallisepticum. Microbiology. 2012;158(Pt 8):2073–88. Epub 2012/05/24. pmid:22628486.
- 46. Register KB, Thole L, Rosenbush RF, Minion FC. Multilocus sequence typing of Mycoplasma bovis reveals host-specific genotypes in cattle versus bison. Vet Microbiol. 2015;175(1):92–8. Epub 2014/11/11. pmid:25433454.
- 47. Hawley DM, Dhondt KV, Dobson AP, Grodio JL, Hochachka WM, Ley DH, et al. Common garden experiment reveals pathogen isolate but no host genetic diversity effect on the dynamics of an emerging wildlife disease. J Evol Biol. 2010;23(8):1680–8. Epub 2010/06/17. pmid:20561136.
- 48. Bonneaud C, Balenger SL, Russell AF, Zhang J, Hill GE, Edwards SV. Rapid evolution of disease resistance is accompanied by functional changes in gene expression in a wild bird. Proc Natl Acad Sci U S A. 2011;108(19):7866–71. Epub 2011/04/27. pmid:21525409; PubMed Central PMCID: PMC3093480.
- 49. Adelman JS, Kirkpatrick L, Grodio JL, Hawley DM. House finch populations differ in early inflammatory signaling and pathogen tolerance at the peak of Mycoplasma gallisepticum infection. Am Nat. 2013;181(5):674–89. Epub 2013/03/14. pmid:23594550.
- 50. Wobeser GA. Essentials of disease in wild animals: John Wiley & Sons; 2013.
- 51. Snyder LA, Loman N, Pallen MJ, Penn CW. Next-generation sequencing-the promise and perils of charting the great microbial unknown. Microb Ecol. 2009;57(1):1–3. Epub 2008/11/18. pmid:19015912.
- 52. Braun J, Schrenzel M, Witte C, Gokool L, Burchell J, Rideout BA. Molecular methods to detect Mycoplasma spp. And Testudinid herpesvirus 2 in desert tortoises (Gopherus agassizii) and implications for disease management. J Wildl Dis. 2014;50(4):757–66. Epub 2014/08/14. pmid:25121400.
- 53. Gonzales TK, Kulow M, Park DJ, Kaspar CW, Anklam KS, Pertzborn KM, et al. A high-throughput open-array qPCR gene panel to identify, virulotype, and subtype O157 and non-O157 enterohemorrhagic Escherichia coli. Mol Cell Probes. 2011;25(5–6):222–30. Epub 2011/09/10. pmid:21925264.
- 54. Verstraete K, Van Coillie E, Werbrouck H, Van Weyenberg S, Herman L, Del-Favero J, et al. A qPCR assay to detect and quantify Shiga toxin-producing E. coli (STEC) in cattle and on farms: a potential predictive tool for STEC culture-positive farms. Toxins (Basel). 2014;6(4):1201–21. Epub 2014/03/27. pmid:24681714; PubMed Central PMCID: PMC4014729.
- 55. Hardy L, Jespers V, Van den Bulck M, Buyze J, Mwambarangwe L, Musengamana V, et al. The presence of the putative Gardnerella vaginalis sialidase A gene in vaginal specimens is associated with bacterial vaginosis biofilm. PLoS One. 2017;12(2):e0172522. Epub 2017/02/27. pmid:28241058; PubMed Central PMCID: PMC5328246.
- 56. Citti C, Dordet-Frisoni E, Nouvel LX, Kuo CH, Baranowski E. Horizontal Gene Transfers in Mycoplasmas (Mollicutes). Curr Issues Mol Biol. 2018;29:3–22. Epub 2018/04/12. pmid:29648541.