Introduction

Wildlife disease surveillance is important for multiple reasons: disease can negatively impact biodiversity, human health, domestic animal and livestock health, and economic stability [1]. Surveillance can aid not only with mitigation, but also with preventing the spread of disease to new areas. To successfully monitor and manage wildlife diseases, it is essential to have an accurate method of detection, otherwise there is a risk of inaccurately estimating factors such as prevalence, virulence, and transmission [2]. Imperfect diagnostic tests can also hinder our ability to predict spread [2] and result in the allocation of resources to inefficient management plans [3].

Lymphoproliferative disease virus (LPDV) is an avian retrovirus first identified during the 1970’s in Europe and Israel, and again in North America in 2009 [4]. While its entire geographic distribution has yet to be determined, to date it has been identified in Wild Turkeys (Meleagris gallopavo) along the East coast [5–7], as far West as Colorado [4, 8], as far south as Texas [9] and in several Canadian provinces: Manitoba, Ontario, and Quebec [10, 11]. Although previous outbreaks in Europe and Israel were in domestic turkeys, in the United States, natural infection has only been identified in Wild Turkeys [4]. Compared to their domestic counterparts, that experienced heightened levels of internal lesions and mortality [12], prior North American studies utilizing hunter-harvested turkeys have indicated that infection in wild birds is typically asymptomatic [4, 8, 10, 11, 13], although mortalities have been observed [4]. Clinical symptoms, when present, are typically non-specific, such as lethargy or anorexia [12]. To date, little is known of the transmission mechanism [4], population-level effects, and impacts on poults (chicks) [11].

Retroviruses function by integrating their RNA genome into host DNA, so as DNA replicates, the virus does as well [14, 15]. Viral RNA is transcribed into DNA via reverse transcriptase and is then integrated into the host’s genome. Organization of the LPDV provirus is as follows: 5’-LTR-gag-pro-pol-env-LTR-3’ [4, 16]. The complete proviral genome of the first LPDV case in North America (12/AR/2009) was 7,432 nucleotides long including LTRs [4]. Virus expression and replication first occur in bone marrow, then distribute to other lymphoid tissues [8, 16]. In a prior study by Thomas et al. [8] it was found that bone marrow was the most efficient tissue for diagnosing LPDV in turkeys obtained from hunter-harvest, compared to spleen and liver.

Diagnosis of LPDV is typically through polymerase chain reaction (PCR) utilizing DNA obtained from bone marrow, although whole blood and cloacal swabs have also been found to be a suitable alternative in live birds [13, 17]. When testing for LPDV, the partial p31/ partial capsid region of the gag gene is widely used among researchers [4, 8, 10, 11, 13]. However, this region may not consistently detect infection due to variability among viral strains across the distribution of turkeys (Renshaw R., Cornell University, personal communication). An exploratory examination of LPDV detection in samples collected from turkeys in both the eastern and western distribution of their range suggested that a different section of the LPDV provirus beginning in the U3 region of the Long Terminal Repeat and extending into the gag Matrix (henceforth referred to as LTR) may be more accurate at detecting infected birds (Renshaw R., Cornell University, personal communication). Long terminal repeats are an essential element of retroviruses, containing components essential for integration and transcription of the provirus [18], gene expression, and insertional mutagenesis [19]. Studies of other retroviruses have also found that PCR primers targeting different regions of the proviral genome had varying detection rates [20, 21].

With Wild Turkey populations suspected to be declining (as indicated by observation of declines in spring harvest numbers) across the United States due to unknown causes [22], some natural resource managers are interested in expanding our understanding of LPDV and its potential role in declines. To do so, it is important to use the best available diagnostic test. The goal of this study was to determine if there are differences in LPDV detection rates when testing the same set of samples with the gag and LTR sections of the provirus and whether using LTR leads to the detection of more LPDV positive Wild Turkeys in Iowa than are detected using gag.

Methods

Results

Extracted DNA ranged in concentration from 8.01ng/μL to 1275.26ng/μL prior to standardization. A total of 925 samples were of sufficient quality for testing and had unambiguous results with the gag gene. Of these, LPDV was undetected in 772 samples and 153 tested positive using the gag gene (S1 Data). Infection status appeared to be unrelated to DNA concentration, with samples of both low and high DNA concentrations testing LPDV positive using both gag and LTR. A subset (n = 417)– 45% of gag positive and 45% of gag negative samples–was randomly selected for paired testing with LTR (Table 1). The subset of samples that tested positive with LTR that were sequenced with LTR primers all most closely matched LPDV sequences in the National Center for Biotechnology Information (NCBI) BLAST database (pairwise identity and percent identical ranged from 93.4% to 99.0%), confirming detection of LPDV with LTR. Gag and LTR sequences were deposited in GenBank under the accession numbers OR026188- OR026260 and OR026261-OR026271, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Pairwise comparison of LPDV detected (+) and undetected (-) samples using the gag and LTR sections of the provirus.

https://doi.org/10.1371/journal.pone.0296856.t001

The positive percent agreement (PPA) was 97.1% (68/70) and the negative percent agreement (NPA) was 68.0% (236/347), indicating greater disagreement with LTR for samples testing gag negative than samples testing gag positive (Table 2). Cohen’s kappa (κ = 0.402) was significant (Z = 10.26, p < 0.0001), falling between the categories of minimal and weak agreement between the results from gag and LTR (Table 2). The McNemar mid-p test was also significant (OR = 55.5, p < 0.0001), indicating the number of gag positive/LTR negative and gag negative/LTR positive samples was disproportionate, and that there were significantly more samples that tested positive with LTR and negative with gag. (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Evaluation of LTR as a diagnostic test for LPDV infection against the standard (gag) using positive percent agreement (PPA), negative percent agreement (NPA), Cohen’s kappa, and McNemar’s odds ratio (OR).

https://doi.org/10.1371/journal.pone.0296856.t002

Discussion

This study was performed to examine differences in detection rates when testing Iowa Wild Turkeys for LPDV with the gag and LTR sections of the provirus, therefore assessing the utility of LTR for detection of LPDV. The data demonstrated that there was high agreement in results between gag and LTR in samples that were gag positive, with both testing positive in 97.1% of gag positive samples. However, there was a high level of disagreement for samples that were gag negative, with both testing negative in only 68.0% of gag negative samples. LTR resulted in the identification of an additional 111 samples as LPDV positive. Sequencing confirmed that LTR was detecting the presence of LPDV in turkeys. These results show that, at least in Wild Turkeys in Iowa, the LTR section of the LPDV provirus is able to detect infected individuals including those testing negative with the gag gene, and that testing with the gag gene results in false negatives. Although LTR did detect more positive animals than gag in this study, it did not detect all cases. One of the two gag positive but LTR undetected cases was sequenced with gag primers to confirm LPDV infection and that the disagreement was not due to lab error.

The high mutation rate in viruses can lead to large amounts of genetic variation [35, 36], which is exhibited in LPDV by the many different strains identified across studies based on the gag gene [4, 8, 10, 11, 37–39]. PCR is used to detect LPDV infections by using primers that bind to a specific region of the provirus that has been inserted into the turkey genome, but may fail to detect the virus even when it is present when there are high levels of variation that lead to mismatches between primers and the binding region of the sample [35]. Because of this, diagnostic methods that are successful in one geographic region may be less successful in another geographic region. As a virus spreads among individuals, mutations will occur, creating increasing amounts of diversity between nucleotide sequences as the infection spreads. The results of this study suggest that the LTR section may be a more conservative region of the LPDV provirus and thus less likely to contain genetic variation that would negatively affect primer binding success in comparison to the gag gene (Renshaw R., Cornell University, personal communication).

This study included only samples from Iowa Wild Turkeys, which represents a small part of the Wild Turkey range and the distribution of LPDV that has been documented to date [4–11]. Because of the geographic limitation of this study, it is not yet clear if LTR performs similarly well at detecting LPDV in Wild Turkeys in other geographic regions. Another limitation is that a relatively small number of gag positive samples were tested with LTR. A larger sample size might provide a more accurate estimate of LTR’s false negative rate. Further, while sequenced samples consistently matched LPDV sequences in the NCBI BLAST database and re-testing samples reinforced confidence in positive results, not all gag or LTR tested samples were sequenced, thus there is the possibility of false positives.

Future research could include testing for LPDV with LTR in additional regions covering the rest of the Wild Turkey’s range to determine if LTR performs similarly in other geographic regions, or if its ability to detect additional positive turkeys is restricted to the Midwest. Testing could also be conducted in subspecies other than Eastern Wild Turkeys. Other sections of the LPDV provirus or whole genome sequencing could also be considered for viability as a diagnostic test, as could other types of PCR depending on sample type (e.g., real-time, reverse-transcriptase). While nucleotide conservation was previously found to be similar across the four viral genes of LPDV (gag, pro, pol, env), pol was found to have the highest level of conservation (90.1%), compared to gag (88.3%), pro (88.3%), and env (86.6%) [4]. This might suggest pol as another viable option for testing. Further, a recent study of LPDV in Wild Turkeys in Texas employed quantitative PCR (qPCR) and the env gene for surveillance [9]. Universal primers have been developed for other viruses and could provide another possible option for future research [35].

Successfully monitoring and managing LPDV in Wild Turkey populations requires an accurate method of detection [2, 3]. Detection methods with a high rate of false negatives might result in an underestimation of prevalence and incorrectly identifying risk factors or areas of concern [2]. If LTR performs similarly in other geographic regions, it would suggest that LPDV prevalence might be higher than has previously been reported. By correctly identifying positive cases, additional risk factors may come to light, or previously identified variables may prove to be less significant than initially believed, allowing for better predictions. Identifying risk factors can help identify areas potentially at risk and identifying areas with high numbers of cases might suggest areas to target for future studies on LPDV. This in turn can lead to more efficient management and a better allocation of resources [3].

Supporting information

Acknowledgments

We would like to thank Randall Renshaw from Cornell University who shared his hypothesis about differences in LPDV detectability using gag vs LTR that led us to conduct this study and for providing samples for use as positive controls. We thank the Iowa Department of Natural Resources staff and hunters for their assistance in obtaining samples from harvested Wild Turkeys.