Distribution of Brugia malayi larvae and DNA in vector and non-vector mosquitoes: implications for molecular diagnostics

Background The purpose of this study was to extend prior studies of molecular detection of Brugia malayi DNA in vector (Aedes aegypti- Liverpool) and non-vector (Culex pipiens) mosquitoes at different times after ingestion of infected blood. Results Parasite DNA was detected over a two week time course in 96% of pooled thoraces of vector mosquitoes. In contrast, parasite DNA was detected in only 24% of thorax pools from non-vectors; parasite DNA was detected in 56% of midgut pools and 47% of abdomen pools from non-vectors. Parasite DNA was detected in vectors in the head immediately after the blood meal and after 14 days. Parasite DNA was also detected in feces and excreta of the vector and non-vector mosquitoes which could potentially confound results obtained with field samples. However, co-housing experiments failed to demonstrate transfer of parasite DNA from infected to non-infected mosquitoes. Parasites were also visualized in mosquito tissues by immunohistololgy using an antibody to the recombinant filarial antigen Bm14. Parasite larvae were detected consistently after mf ingestion in Ae. aegypti- Liverpool. Infectious L3s were seen in the head, thorax and abdomen of vector mosquitoes 14 days after Mf ingestion. In contrast, parasites were only detected by histology shortly after the blood meal in Cx. pipiens, and these were not labeled by the antibody. Conclusion This study provides new information on the distribution of filarial parasites and parasite DNA in vector and non-vector mosquitoes. This information should be useful for those involved in designing and interpreting molecular xenomonitoring studies.


Background
Human lymphatic filiarasis (LF) is caused by the mosquito-borne filarial nematodes Wuchereria bancrofti, Brugia malayi, and B. timori. These parasites are currently targeted for elimination by the Global Program for the Elimination of Lymphatic Filariasis (GPELF), and workers in this program have reported both achievements and future challenges to eliminating parasite transmission in endemic areas [1][2][3]. One important component of the elimination program is the ability to estimate infection prevalence and transmission rates, especially during mass drug administration (MDA), in order to accurately evalu-ate the progress towards the goal of LF transmission interruption [4]. Molecular detection assays provide sensitive and specific tools for identifying and distinguishing parasites in host populations. Molecular techniques commonly used to study LF infection, or exposure, in humans include the detection of parasite DNA, circulating filarial antigen, and filarial antibodies in blood samples [5]. Molecular techniques also have been applied to the detection of filarial worms in mosquitoes, and these primarily target parasite DNA [6][7][8][9].
The detection of parasite DNA in mosquito samples is a valuable tool for molecular xenomonitoring (MX), but this does not differentiate parasite developmental stages or distinguish whether the DNA is from living or dead parasites [10][11][12]. Recently, RNA-based assays have been developed to detect B. malayi and W. bancrofti in mosquitoes [13,14], including the distinction of B. malayi infected (a constitutive parasite transcript) and infective mosquitoes (a L3-specific transcript) [14]. However, RNAbased detection assays have not yet been tested in the field or incorporated into LF surveillance programs. Vector-parasite interactions influence the applicability and interpretation of molecular detection assays used in vector surveillance studies. There are several factors that should be carefully considered when using molecular techniques to investigate parasites within the mosquito intermediate host, including the (1) various life cycle stages and their tissue locations, (2) likelihood of parasite development to the infective stage, i.e., vector competence, and (3) limitations of the particular detection assay, i.e., ability to distinguish infection stages and living from dead parasites. The separation of mosquitoes into body regions has been used to circumvent the inability of some assays to distinguish infective-stage parasites. For example, Anopheles spp. have been divided into two body regions (head/thorax and abdomen) to provide better estimates of mosquitoes infected with Plasmodium sporozoites and/or pre-sporo-zoite stages [15][16][17] and the heads of blackflies have been removed (by mass dissection techniques) for the restricted, head-only, PCR assays targeting Onchocerca DNA, which is more likely to provide a better estimate of infective-stage parasites because other developmental stages generally reside outside of the head [18,19].
The studies conducted herein follow our previous work, which demonstrated that DNA-based diagnostics are unable to distinguish the developmental stage of LF parasites or whether parasites are living or dead in the mosquito [10]. Despite these limitations, there are benefits to using DNA-based assays over dissection to assess the persistence of filariasis in populations. Because filarial DNA is detectible for two weeks or longer following a microfilaremic blood meal in both vector and non-vector mosquitoes, all anthropophilic mosquitoes can be included in the screening of mosquitoes for parasite DNA to provide MX data [10]. Herein, we have further examined the persistence of filarial parasites and parasite DNA in mosquitoes; we used a combination of mosquito dissection, immunohistology and PCR assays to determine the location(s) of filarial worms and DNA in mosquitoes that are susceptible or refractory to filarial parasite development. These studies allowed us to assess the potential value of tissue specific assays (e.g., mosquito heads only) to estimate the prevalence of infective-stage larvae in mosquitoes; we also investigated the issue of direct mosquito to mosquito transfer of parasite DNA that could confound MX studies. Table 1 summarizes the recovery of parasites from dissected mosquitoes. In Ae. aegypti-LVP, 73.6% of the recovered B. malayi mf at 2 h post ingestion (PI) had successfully penetrated the midgut, with 49.4% located in the thorax. At 14 days post ingestion of microfilaremic blood (DPI), 80% of Ae. aegypti harbored L3s. In contrast, from Cx. pipiens only mf were recovered and 93% of them were found in the midgut lumen.

B. malayi DNA detection in pooled mosquito body regions
PCR results are summarized in Figure 1. Parasite DNA was detected in 74% of Ae. aegypti-LVP pooled body region samples (n = 300) and 36% of Cx. pipiens (n = 300) pooled body regions tested. These differences were highly significant (P < 0.0001). In Ae. aegypti-LVP, parasite DNA was detected in all four body regions with 43, 71, 88, and 96% of heads, midguts, abdomens, and thoraces (n = 75 for each) positive by qPCR, respectively. Parasite DNA also was detected in all Cx. pipiens body regions, with 17, 24, 47, and 56% of heads, thoraces, abdomens, and midguts (n = 75 for each) positive, respectively. The differences in the percentage of B. malayi DNA positive samples were significant between mosquito species in all body regions (heads, P = 0.001; thoraces, P < 0.0001; abdomens, P < 0.0001) except the midguts (P = 0.09). The detection of parasite DNA within certain mosquito body regions was positively or negatively correlated with time. Specifically, the detection of B. malayi DNA was negatively correlated with time in 'whole body' Cx. pipiens (r 2 = 0.93, P = 0.0075) and Cx. pipiens midguts (r 2 = 0.82, P = 0.034), and positively correlated with time in Ae. aegypti-LVP heads (r 2 = 0.77, P = 0.05).

B. malayi DNA detection in individual mosquito body regions
Individual mosquitoes that were separated into body regions for DNA detection assays were compared to results of the pooled mosquito body regions (Fig. 2). Following a microfilaremic blood meal of 191 mf/20 μl blood, there was no difference in DNA detection between pooled and individual Ae. aegypti-LVP body regions (Pvalues = 0.35-0.61) or between pooled and individual Cx. pipiens body regions (P-values = 0.78-1.0). Figure 3 summarizes the detection of B. malayi DNA in individual housed mosquitoes and their voided excreta and feces. All mosquitoes were positive for parasite DNA immediately (2 hr) after ingesting microfilaremic blood. From 1-4 DPI all Ae. aegypti-LVP tested positive for parasite DNA, and three of these (15%) mosquitoes had detectible B. malayi DNA in their feces. In contrast, 60% of Cx. pipiens were DNA negative at 4 DPI, but B. malayi DNA was detected in 100% of Cx. pipiens feces tested at 3-4 DPI.

Detection of parasite DNA in mosquito excreta and feces
Of the twenty samples of feces collected over the entire observation period of 1-4 DPI from each species, B. malayi DNA was detected in 15 and 65% of Ae. aegypti-LVP and Cx. pipiens fecal samples, respectively (P = 0.003).

Parasite DNA contamination of B. malayi positive and negative mosquitoes
In order to test the possibility that infected mosquitoes contaminate uninfected mosquitoes while they are together in the same trap or sampling tube, we housed uninfected Ae. aegypti-LVP together with Cx. pipiens that had fed on a microfilaremic gerbil. After 7 days, mosquitoes were collected and pooled by species. None of the 17 Ae. aegypti-LVP pools (with 10 mosquitoes per pool) were positive by real-time PCR. In contrast, 17 of the 21 Cx. pipiens pools (with 5 mosquitoes each) were positive. Most of these samples had relatively high Ct values indicating small amounts of B. malayi DNA, but 5 pools had higher Ct values ranging between 29 and 35. Although B. malayi DNA can be detected in feces of infected mosquitoes, feces did not cause false positive DNA signals from uninfected mosquitoes after co-housing.

Detection of Bm14 in mosquito-stage parasites by immunohistology
Immunohistology studies were performed to confirm the dissection results and to better document the fate of B.
Prevalence of B. malayi DNA in pooled samples of experi-mentally infected mosquitoes by body part (head, thorax, abdomen, and midgut) at different times post ingestion of microfilaremic blood malayi in vector and non-vector mosquitoes. In Ae. aegypti-LVP, unlabeled mf were detected within the midgut directly after the bloodmeal (Fig. 4A). Strong-labeling was observed with the Bm14 antibody after the larvae reached the thoracic muscles. Thus this antibody can be used to sensitively detect developing filarial larvae in vectors ( Fig.  4B-E). Strongly labeled L3s were observed in all body parts of Ae. aegypti-LVP at 14 DPI ( Fig. 4F-G). L3s were not confined to the head or the thoracic musculature; they were also seen in the abdomen, outside of the midgut (Fig. 4J).
Stretched, intrauterine mf in adult female B. malayi are usually labeled by the Bm14 antibody (Fig. 5A), but mf were not labeled in the midgut of Cx. pipiens directly after the blood meal. No larvae were detected in histological sections of Cx. pipiens at later times points (Fig. 5C, D). Dead and/or dying larvae in the thorax of Brugia refractory Ae. aegypti-RKF were not labeled by Bm14 (Fig. 5G, I, J). In contrast, developing larvae in Ae. aegypti-LVP were always strongly labeled at the same time points (Fig. 5F, H).
These results suggest that the anti-Bm14 antibody specifically detects viable and developing B. malayi larvae in vectors.

Discussion
The application of molecular assays to detect parasites within vectors is influenced by the vector-parasite interaction. In these studies, we analyzed the detection of filarial worms in susceptible and refractory mosquitoes by targeting parasite DNA or protein in molecular assays. The three mosquito strains examined have very different interactions with B. malayi that were documented in our previous paper on detection of parasite DNA from living and dead parasites within mosquitoes [10]. The follow-up studies discussed herein were designed to investigate: (1) the mosquito body region(s) containing the persistently detected parasite DNA within Cx. pipiens (in this mosquito species mf are seldom able to migrate out of the midgut lumen), and (2) the detection of parasite proteins as larvae develop in susceptible and refractory mosquitoes.
Comparison of B. malayi DNA prevalence in pooled and individually tested mosquito body parts Separation of mosquitoes into body regions (head, thorax, abdomen, and midgut) prior to PCR assays provided further details on the location of detectible parasite DNA within mosquitoes. In both vector Ae. aegypti-LVP and non-vector Cx. pipiens, the head region had the lowest parasite DNA detection rates (43 and 17% respectively) when all time points are considered. However, there was a striking difference at 14 days when nearly 100% of pools of Ae. aegypti-LVP heads and very few pools of Cx. pipiens heads contained parasite DNA. This finding correlates well with infectivity rates at that time point. Although DNA detection in mosquito heads might provide a better estimate of infectivity rates than DNA detection in whole mosquitoes (Fig. 2), we do not advocate separating heads for this purpose in field studies, because false positive and false negative infectivity signals are likely to be high with this approach. Because PCR can detect DNA from live or dead parasites from any developmental stage, heads could be falsely positive (without L3s being present) because of remnants of ingested mf (especially in mosquitoes with armed cibarial and/or pharyngeal pumps). Heads could provide a false negative infectivity signal when mosquitoes contain L3s in other body parts. Our results and prior studies [20] have shown that L3s are not restricted to the head.
B. malayi DNA also was not restricted to the midgut in Cx. pipiens. Although most mf do not leave the Cx. pipiens midgut lumen, mf are sometimes detected outside of the midgut by dissection (Table 1 and Erickson and Christensen, unpublished data). Although these few mf that do penetrate the midgut epithelium could be the source of parasite DNA outside of the Cx. pipiens midgut, an alternative explanation was also examined. In compatible vectors, the majority of mf penetrate the midgut epithelium very quickly (within 1.5 h) after ingestion [21], and those that are unable to penetrate remain within the midgut lumen and are destined for digestion and/or defecation. This is in agreement with our observation that at 2 h PI almost 50% of mf were recovered from the thorax. In parasite-vector pairs that limit or prevent mf penetration of the mosquito midgut (e.g., Brugia--Cx. pipiens and certain Wuchereria--Culex spp. interactions) [22][23][24][25], parasite material is likely to be shed in mosquito feces. To test the hypothesis that DNA detected outside of the midgut is due to parasite DNA on exterior mosquito surfaces, excreta and feces from Ae. aegypti and Cx. pipiens were examined for parasite DNA. B. malayi DNA was detected in voided materials of both species, but significantly more Cx. pipiens had detectible parasite DNA in their excreta and feces following a microfilaremic blood meal (Fig 3).
Although external contamination could cause false positive PCR results, our mosquito co-housing results suggest that cross-contamination is probably uncommon.
However, to prevent cross-contamination, the following steps are recommended for processing wild-caught mosquitoes: (1) Immediately kill mosquitoes to limit mosquito-mosquito contact, especially if blood engorged mosquitoes are present; (2) When killing mosquitoes, avoid methods of slow death in which the mosquito host is killed and parasites are not, because it is possible for L3s to escape the body of living, dead, and dying mosquitoes, especially if mosquitoes come in contact with aqueous fluids [26][27][28]; (3) Pool mosquitoes as soon as possible after collection; (4) When sorting mosquitoes into pools, tools such as forceps and brushes should not pierce the mosquito body; and (5) Mosquito sampling should be restricted to whole females (head, thorax and abdomen intact). This will reduce error that could be introduced by testing random body parts from unidentified individuals.
Immunohistochemistry provides an alternative approach to detecting and studying parasite migration and development in mosquitoes; several studies have used this approach with antibodies to Plasmodium circumsporozoite protein (CSP) [16,[29][30][31]. Most immunodiagnostic research on filariasis has focused on the parasite stages that occur within the vertebrate host [32,33], and this work has produced sensitive and specific diagnostic tests for Bancroftian and brugian filariasis [5,34,35]. In con-

Conclusion
Improved methods are needed for assessing changes in mosquito infection and infectivity rates in the context of LF control/elimination programs. As infection rates in humans and vectors decrease following MDA, increased numbers of mosquitoes must be tested to accurately estimate parasite prevalence [40]. This makes dissection impractical, and favors use of molecular detection assays with pooled mosquitoes. This study provides new information on the persistence of filarial worm DNA in nonvectors that has practical implications for MX studies regarding methods for processing field-caught mosquitoes and for interpreting MX data. Additional studies are needed to determine whether the presence of Bm14 antigen is a reliable marker for viable filarial worms in pooled mosquito samples. Although this study focused on B. malayi, the findings may be of interest to scientists and Immunohistological detection of B. malayi larvae in Ae. aegypti-LVP using polyclonal antibody to recombinant antigen Bm14

Mosquito maintenance and parasite exposures
Mosquitoes used for these studies were obtained from colonies of Aedes aegypti (black-eyed, Liverpool strain; LVP), Ae. aegypti (Rockefeller strain; RKF) and Culex pipiens pipiens (Iowa strain) maintained at the University of Wisconsin-Madison, as previously described [21,41]. These mosquitoes differ in their vector competence for B. malayi.
Ae. aegypti-LVP support the development of B. malayi from mf to L3s, but parasites do not develop in Cx. pipiens, because mf do not penetrate the midgut epithelium. In Ae. aegypti-RKF, mf penetrate the mosquito midgut and migrate into thoracic muscles where they fail to develop to L2s. This mosquito strain was only used for comparison in the immunohistology experiments. Four-to seven-dayold mosquitoes were sucrose starved ~14 h prior to blood feeding. Mosquitoes were exposed to B. malayi by blood feeding on microfilaremic cat blood in a water-jacketed membrane feeder fitted with a parafilm membrane [42].
Mosquitoes also were blood fed on uninfected gerbils (Meriones unguiculatus) to serve as parasite-negative, blood-fed controls. Engorged mosquitoes were sorted and maintained in the laboratory.
Laboratory animals were handled according to guidelines approved by the Animal Care Committee at the University of Wisconsin-Madison. The mf densities of B. malayiinfected cat blood obtained from the NIAID Filariasis Research Reagent Repository Center http://www.filariasis center.org used in these studies ranged from 24-191 mf/ 20 μl blood. ŵŐ rated into head, thorax, midgut, and abdomen, and each body region was teased apart and individually examined for parasites by microscopy as previously described [10].

Mosquito dissection
Ae. aegypti-LVP and Cx. pipiens were separated into body regions (head, thorax, midgut, and abdomen) and placed, separately or in pools of four, into 2.0 ml microcentrifuge tubes for parasite DNA detection. To create a pooled sample, four mosquitoes were separated into body regions, and the body regions were combined by type into tubes. For example, four mosquitoes were used to produce one pool of four heads, one pool of four thoraces, one pool of four abdomens, and one pool of four midguts. Five pooled samples were prepared at 2 h, 1, 3, 7, and 14 d PI; thus, a total of 20 mosquitoes were collected at each time point for pooled samples. In addition to creating pooled samples, individual mosquitoes were dissected into body regions as described above and then placed individually into tubes. Five individuals were dissected at each time point to create twenty samples: five tubes contained individual heads, five contained a single thorax, five contained an abdomen, and five contained a midgut. These samples were screened for B. malayi DNA to compare detection results between individuals and pooled samples. All samples were cataloged to track a given body region back to the particular mosquito or pool of mosquitoes, allowing the assay results from each body region to be combined into results for the 'whole mosquito.' For example, if at least one of the body region samples was positive; then the 'whole mosquito' was considered positive and only if all body regions tested negative; then the 'whole mosquito' was considered negative.

Collection of excreta and feces from individual mosquitoes
To examine possible sources of parasite DNA contamination, mosquitoes were individually housed to collect material voided by excretion (excreta processed from Malphigian tubules) and defecation (feces containing undigested material from the midgut) [43]. Immediately following blood engorgement, mosquitoes were individually housed in 2.0 ml microcentrifuge tubes. Mosquitoes were maintained for up to 4 DPI by providing 10% sucrose solution in drops on the mesh-screen top fitted on each tube. At 24 h intervals, mosquitoes and their voided matter were collected for DNA extraction. First, the mosquito was removed from the collection tube, transferred to a clean microcentrifuge tube and flash frozen in dry ice. Then, the collection tube containing the excreta and/or feces was labeled and also frozen. All samples were stored at -80°C until DNA extraction. Five mosquitoes of each species were collected at 2 h and 1-4 d PI, and voided material was sampled at 1-4 DPI. In addition, the appearance of a blood bolus was recorded for each mosquito to provide data on the presence or absence of blood meal remnants within the mosquito midgut. Mosquitoes that blood fed on uninfected gerbils also were housed individually and sampled at each time point for negative controls.

Co-housing studies to determine whether B. malayi DNA is transferred from infected to uninfected mosquitoes
Uninfected Ae aegypti were housed together with B. malayi-infected Cx. pipiens that had fed on two B. malayiinfected gerbils with microfilaremias of 68 and 146 mf/20 μl blood. Engorged Cx. pipiens were sorted from nonblood fed mosquitoes, and the engorged mosquitoes were mixed with non-blood fed Ae. aegypti. Four cartons were studied with each containing ~60 uninfected Ae. aegypti and ~40 B. malayi-infected Cx. pipiens. At 7 DPI, mosquitoes were collected for qPCR detection of parasite DNA. Cold anesthetized mosquitoes were sorted by species and placed in pools of five individuals for Cx. pipiens and pools of ten individuals for Ae. aegypti. Mosquitoes were flash frozen on dry ice and stored at -80°C.

DNA extraction and detection
Genomic DNA from pooled mosquitoes or body parts was extracted using a commercial column method as described previously [10]. Quantitative real-time PCR was performed using an MGB probe to detect a 120 bp fragment of the Brugia HhaI repeat [44]. In all real-time PCR assays, water was used as no-template negative control; DNA extracted from a pool of non-infected mosquitoes acted as extraction negative control, and 100 pg of DNA isolated from adult B. malayi was used as positive control.

Immunohistology
Five mosquitoes of each species were collected at 2 h, 1, 3, 7, and 14 d PI (i.e., the same time points as DNA detection assays) and stored in 80% ethanol at room temperature until embedding. Mosquitoes were embedded in paraffin, and B. malayi larvae were stained using the alkaline phosphatase anti-alkaline phosphatase method as described previously [45]. A polyclonal mouse antibody raised against recombinant Bm14 protein was used as the primary antibody for these studies [38].

Statistical Analysis
Data were graphed and analyzed with GraphPad Prism 5.0 http://www.graphpad.com. Fisher's exact tests, with two-tailed P-values, were used to compare parasite development and DNA detection between Ae. aegypti-LVP and Cx. pipiens. Pearson correlation tests were used to test for trends in parasite DNA detection over time. Statistical results were considered significant at P ≤ 0.05.