The qPCR method offers several advantages over conventional PCR. First, when analyzing large number of samples with conventional PCR, the post-PCR steps such as gel-electrophoresis and manual scoring of results are labor-intensive, delay availability of data, generate large quantities of waste, and increase risk of exposure to ethidium bromide, a mutagen. These concerns are eliminated when qPCR is used. Secondly, it may be desirable to quantify DNA template or copy number of the target gene in samples, which qPCR can accomplish. Thirdly, while qPCR is not always more sensitive than conventional PCR [26], many studies including this current one have shown that the former method can detect much lower DNA template concentrations compared to the latter [27–30]. Thus, sensitivity of detection is better, as was shown here.
Identification of arthropod blood meals by qPCR has been applied to sand flies, biting midges, kissing bugs, fleas and mosquitoes [22, 23, 31–35]. Most of these were SYBR green-based systems; only three were probe-based. Of the three probe-based qPCR, one was for identifying Australian mammals in Culex mosquito blood meals and did not include humans, pigs or dogs [23], another was for identifying blood-meal hosts of biting midges and included humans and pigs but not dogs [33] and another was for identifying flea blood meals and included humans and dogs but not pigs [35]. Notably, the human probe in the latter flea blood-meal study was tested here and found to cross-react with dog DNA. Surprisingly, a thorough Google Scholar search did not find a paper describing probe-based qPCR designed specifically for identifying mammalian hosts of Anopheles blood meals. The qPCR assay described in the present study utilized the new non-fluorescent quencher dye QSY (Catalog number: 4482777; Applied Biosystems, Foster City, CA, USA) and a PCR solution optimized for probe-based multiplex qPCR (TaqMan Multiplex Master Mix, Catalog number: 4461882; Applied Biosystems) to detect the common blood-meal hosts of PNG mosquitoes.
When evaluating the sensitivity of the qPCR assay from amplifications of 10-fold dilution series of target DNA samples, the lowest detectable concentration for human DNA (10-4 ng/ml) was ten-fold greater than for pigs and dogs which was 10-5 ng/ml. This difference could be attributed to the copy numbers of the target DNA sequences; the pig and dog probes target mitochondrial genes which exist in multiple copies per cell, whereas the human probe targets a single-copy nuclear DNA sequence. Several human probes targeting various mitochondrial gene locus were designed and tested (see Additional file 3: Table S3). However, they all exhibited non-specific amplification of the two non-human hosts in vitro despite appearing to be target specific by in silico test. Nevertheless, the detectable limit of human DNA concentration with the current probe (10-4 ng/ml) is sufficiently low for detecting mosquito blood meals. The lower detectable limit of qPCR compared to the conventional one, particularly dog, indicates a difference in the sensitivity of the two methods.
The results show that the blood-meal qPCR was more sensitive at detecting host DNA in mosquitoes (detection success rate of 89%) compared to the more commonly used conventional, multiplex PCR (detection success rate of 55%). It is possible that the 11% of mosquitoes whose blood-meal hosts were not identified by the qPCR could have fed on other host sources (e.g. chickens, cats). However, when subjected to two conventional PCR utilizing generic mammalian and avian primers, none showed a positive result, which was consistent with findings from our previous study [24]. Thus, the likelihood that host breadth was greater than the three hosts we targeted with probes here is low. A common observation in all of these unamplified blood meals was that they all contained traces of blood in their abdomens (< 0.3 ml), based on light microscopy examination of the mosquito abdomens 4–8 hours after they were collected. Given the non-nucleated status of mammalian red blood cells and disproportionately low ratio of white to red blood cells, the small volumes of blood meal were likely insufficient to yield a detectable concentration of DNA.
The trace blood assertion was further supported by showing that the qPCR-quantified host DNA concentration of single blood meals was significantly higher in those samples that were detected in the conventional PCR than those that were undetected for all three hosts. The mean DNA concentration for the detected samples was expected to be statistically the same between the three hosts. However, the result showed otherwise; pig DNA concentration was significantly four-folds higher than humans and two-folds less than dog (Fig. 2a). This heterogeneity in host DNA quantity could be caused by factors such as variation among the hosts in the number of white blood cells per unit volume of blood. However, a more plausible explanation is variation in the average quantity of blood mosquitoes obtain from each host as a result of variation in host sensitivity to mosquito bites. That is, humans are more sensitive than pigs followed by dogs to mosquito bites, causing them to quickly interrupt blood-feeding mosquitoes before they had time to acquire the maximum amount of blood. Despite the variation between the three hosts in their blood-meal DNA quantity, the mean DNA concentration for mosquitoes that were undetected in the conventional PCR was expected to be the same. The expectation is based on the reasoning that below a certain DNA concentration threshold, the conventional PCR primers for all three hosts become insensitive to DNA and fail to amplify. However, the result showed that although the mean DNA concentration was statistically the same for humans and pigs as predicted, it was five-folds higher for dog (Fig. 2a). This indicates that the dog conventional primers are less efficient and have higher sensitivity threshold than the other two hosts, which is consistent with its behavior observed in the amplification of DNA dilution series.
The conventional PCR did not detect mixed blood meals sensitively; most of the samples identified as mixed blood meals by qPCR were identified as single blood meals by the conventional method (Table 3). The inaccurate detection of mixed blood meals was attributed to low DNA quantity of one or more of the hosts in a mixed blood-meal sample and was supported by the result which showed statistically lower mean DNA concentration for undetected compared to detected hosts (Fig. 2b). Thus, a primary outcome of this study is the sensitivity of the probe-based, qPCR method to detect blood from different mammal species in the same blood meal. This finding indicates that a significant proportion of unidentified blood-meal sources in studies that used the conventional, multiplex PCR [24, 36–41] may, among other factors, be due to the sensitivity of this method. Furthermore, the proportion of mixed blood meals may be underestimated and single blood meals overestimated in some published studies. At the very least, such findings indicate interrupted blood feeding, an important variable contributing to transmission [42].