Several technical advances were required for a molecular L3 detection assay including a simple method for preserving parasite RNA in mosquitoes, a method for efficiently extracting RNA from parasites in pools of mosquitoes, identification of an L3-activated gene, and a method to sensitively detect that gene's transcript. We found that parasite RNA could be efficiently extracted from infected mosquitoes preserved in RNAlater (Ambion, Inc.) using a simple BB-grinding technique and a phenol-based extraction procedure. We then identified TC8100 as a L3-activated transcript and developed methods for its detection by conventional RT-PCR and qRT-PCR.
The B. malayi EST genome project provided an important starting point in our search for an L3-activated transcript with information on genes expressed in several parasite stages essential for our initial search. A L3 assay required an authentic L3-activated transcript with no expression in the early larval stages present in mosquitoes (Mf, L1 or L2). We carefully evaluated a panel of candidate genes; only the collagen gene TC8100 was confirmed to be a truly L3-activated transcript.
Time course studies were performed with RNA isolated from infected mosquitoes rather than with parasites isolated at different times after mosquito feeding. This approach proved the practical point that the target parasite sequence could be detected in infected mosquitoes preserved in RNAlater. The development of B. malayi
in mosquitoes is not completely synchronous. Testing multiple replicates of infected mosquitoes at daily intervals and using 2–3 infected mosquitoes in each pool permitted us to be certain that we were covering the entire developmental timeline of the parasites in mosquitoes. Previous studies have documented the developmental time course of B. malayi
in AeL mosquitoes reared under the laboratory conditions used in this study 
. Within 48 hours of a blood meal, the majority of ingested microfilariae exsheathe and reach the ‘sausage’ stage (L1). The first molt to the L2 stage takes place between days 3–6, and the second molt to the L3 stage begins as early as day 6 but peaks between days 9–11 PBM. Therefore, a transcript that is truly L3-activated would be expected to have no expression prior to day 6 PBM, with most samples beginning expression by day 9, 10 or 11 PBM.
The expression profile of TC8100 was consistent with the BmL3 developmental timeline; the majority of the samples tested showed expression beginning at day 9 PBM corresponding to the peak timing of the L3 molt. Only one sample showed expression of TC8100 at day 6 PBM, which is the earliest reported time of L3 development. In addition to being activated at the beginning of L3 development, it is important to note that a successful diagnostic target must also continue to be expressed throughout the lifespan of the L3 parasite so that a mosquito containing L3 of any age can be detected. Our data showed that TC8100 transcripts were present in early L3 and in later L3.
The constitutively expressed B. malayi gene tph-1 complemented TC8100 because it provided a useful marker to indicate the presence (and successful extraction) of filarial RNA in mosquito samples. The tph-1 primers also amplified a homologous target sequence in RNA from the closely related filarial species W. bancrofti but not in RNA from the more distantly related animal filarid D. immitis. Therefore, in settings where there is no overlap between Brugia and Wuchereria infections, the TC8100/tph-1 multiplex assay could be used to evaluate infection and infectivity rates simultaneously.
One consideration for the implementation of any new diagnostic technique is the practicality of using it as a surveillance tool. The storage of vectors in RNAlater eliminates any major limitations regarding mosquito collection. The mosquitoes can be stored for one day at ambient temperature and for at least several months at −20°C. Any laboratory that is already set-up to perform PCR would be able to use the conventional RT-PCR assay with no additional equipment investment. For the real-time assay, the investment of a real-time PCR machine would be necessary, but the cost of such instrumentation is dropping to levels as low as $16,000, making it a reasonable investment option. At this time, the cost of the RNA extraction and RT-PCR is approximately $5 per pool of mosquitoes, or $0.22 per mosquito. If the reactions are run in duplicate, the cost per pool would be approximately $6.30, or $0.25 per mosquito. This compares to the current cost of $5.00 per pool for the xenomonitoring DNA assay. The throughput for the conventional assay for a single technician would be estimated at 2,000 mosquitoes per week (80 pools with 25 mosquitoes per pool), while the real-time assay throughput is estimated at 4,000 mosquitoes per week (160 pools of 25 mosquitoes). The advantage to the real-time assay includes both a higher throughput level (reduced labor cost), as well as a reduction in potential contamination due to the elimination of post-PCR product handling. The real-time assay is a more cost efficient method, and thus, would be the recommended transmission surveillance tool wherever possible anticipating that the cost will decrease as time goes on.
A potential limitation of this L3 detection assay is that it cannot differentiate between B. malayi
and B. pahangi
(a zoonotic parasite) in mosquitoes. However, the same limitation applies to the traditional means of detecting L3 in mosquitoes, namely dissection with microscopy 
. Our results showed that the rbp-1
assay is specific for B. pahangi
. This test can be used to help clarify results obtained from mosquitoes collected in areas where the two Brugia
species are co-endemic. Only positive samples would need to be tested in this way, and the number of positive mosquito pools should become very small as infection rates fall in humans and mosquitoes following MDA.
One key limitation of this study is that our assays have not yet been tested with field-caught mosquitoes. Nevertheless, there have been many calls for the development of molecular tests for detection of filarial L3 in vectors 
. Our results serve as a proof of principle that L3-specific assays are feasible. Field studies are now needed to assess the practical value of such tests as tools for documenting the interruption of transmission in the context of filariasis elimination programs.
Finally, we need to note that the L3-detection assay based on TC8100 is specific to Brugia and does not detect L3 of W. bancrofti, the parasite responsible for the majority of the global burden of LF. Clearly, there is a need for the development of a diagnostic tool for the detection of WbL3 infective vectors as well. Surprisingly, we have been unable to identify the orthologue of this Brugia L3 activated gene, BmTC8100, in W. bancrofti. We are actively testing potential targets that can be used to detect W. bancrofti L3 in mosquitoes.