Mosquito composition in collection sites
There were variations in the mosquito compositions of Long Tengoa and Long Luping, which are only 40 km apart and with an altitude difference of 500 m, even though the numbers of each species collected were small. Previous studies have consistently shown that mosquito composition differs in collection sites with different ecological settings, even when the sites were only 1,500 m away [26,29]. Peak biting times of identical mosquito species were also reported to vary between entomological survey sites [32,60]. The fact that both vectors in Long Luping (An. balabacensis and An. barbirostris Clade VI) were not found in Long Tengoa implies that a different species of Anopheles mosquito could potentially be the vector in the latter site. However, one of the limitations of this study is the relatively short duration of collection periods and the small number of mosquitoes collected; a more extensive survey may well have found An. balabacensis and An. barbirostris in Long Tengoa. Similarly, only one An. latens was collected at each of the two sites in Lawas and it is possible that if more An. latens had been collected and analysed, one or more may have tested Plasmodium-positive. Anopheles latens would then have been incriminated as a vector for P. knowlesi in Lawas, just like it has been for the Kapit District in Central Sarawak[30]. Nevertheless, the detection of simian malaria parasites in the salivary glands of wild-caught An. balabacensis and An. barbirostris Clade VI calls for more detailed longitudinal studies on seasonal variation in composition and on the bionomics of the vectors in this area, including host preference.
Identification of Plasmodium
Sporozoites of P. knowlesi and other simian malaria parasites were identified in 9 An. balabacensis and 5 An. barbirostris Clade VI by nested PCR assays. The An. balabacensis and An. barbirostris Clade VI examined had sporozoite infection rates of 29.0 % and 27.8% respectively, which are exceedingly high compared to other vector studies in Malaysia. In the neighbouring Malaysian Borneo state of Sabah, An. balabacensis was identified as the vector for P. knowlesi with sporozoite rates ranging from 1.03 to 3.42% at three different sites [32]. Infection rates of <2.00 % were reported among vectors in central Sarawak and among An. cracens in Peninsular Malaysia [29,31,60]. The only other study known to have found a comparable sporozoite rate was a study done in Palawan, Philippines where 29.4% of the An. balabacensis examined had sprorozoites [61]. The high sporozoite rates of the vectors could be attributed to the site of collection as well which is consistent with the finding of the previous study involving An. latens in Kapit, Sarawak [60] where it was found that the sporozoite rate was highest in the forest, followed by in the farm at the forest-fringe, and in the long house. Long Luping is an abandoned army camp which is far away from any human settlement. It is potentially a foraging site for the macaques due to the banana trees that grow around its perimeter. No vector control activities have been carried out since the army camp was abandoned and macaques were sighted at the site during the entomological surveys, suggesting that macaques and mosquitoes could forage and breed, respectively, undisturbed in this site.
The high infection rates could also be due to the way sampling was conducted compared to other reported studies [4,30–32,62,63]. Firstly, the current method recovers both sides of the salivary glands without rupturing any to check for sporozoites, increasing the yield of extracted Plasmodium DNA. Secondly, the current method does not retain the head of the specimens which might contain inhibitors which would disrupt PCR assays. PCR inhibitors were previously found to be present in the heads of Culex pipiens and An. punctipennis which caused false negative results in detection of Wolbachia pipientis in samples tested [64]. The efficiency of the PCR assay was later restored when specimens were decapitated prior to DNA extraction. Although we had high sporozoite infection rates by nested PCR assays, we failed to generate the longer (1,500 bp – 2,000 bp) Plasmodium SSUrDNA amplicons from 5 samples that were malaria-positive by nested PCR assays. This could be caused by the very low number of sporozoites present in the salivary glands of the mosquitoes and the low sensitivity of the PCR assay to amplify long fragments from a sample with low template concentration.
Diversity and density of Plasmodium infection in vectors
In line with the discovery of at least 7 Plasmodium species infecting long-tailed macaques in Sarawak [65], it is unsurprising that multiple unidentified Plasmodium species were recovered from An. balabacensis in this study (Figure 2). However, the low quantity of DNA extracted from the salivary glands prevented the sequencing of another gene, such as the mitochondrial genome, which would have been necessary to determine whether these are indeed novel species of Plasmodium. It is highly likely that unidentified Plasmodium species co-infected the vectors when they fed on macaques which are known to host a diverse range of Plasmodium [1,4,65,66].
Accurate identification of Plasmodium by PCR might also be impeded when uncharacterised Plasmodium such as those found in sample LW45 (C10 and C13) could be detected by P. cynomolgi-specific primers (Figure 3). This demonstrates the need for the sequencing of a considerable genomic locus length for proper identification of the species of Plasmodium. On the other hand, the unsuccessful attempts to sequence P. fieldi from sample LW67, and P. cynomolgi from LW45, could be due to the low density of P. fieldi and P. cynomolgi respectively among the other Plasmodium co-infecting each of these mosquitoes. As the PCR amplification prior to cloning amplifies the SSUrRNA genes of all Plasmodium species indiscriminately, the scarcity of any species of Plasmodium DNA in the sample reduces the chance of its amplicon being produced during PCR amplification. The difficulty in obtaining sequencing data of genes of Plasmodium derived from vectors is probably the main reason why previous studies on vectors of knowlesi malaria have only used nested PCR assays [15,30,32,33,63] and for the lack of studies describing the diversity and density of Plasmodium infection in vectors. With an increasing number of zoonotic malaria infections in the world [67–69], epidemiological studies of these inadequately studied species within human populations that come into close contact with macaques during activities in the forest and forest-fringe will also be required to monitor potential host-switch events.
Molecular characterisation of vectors
While all the An. balabacensis identified in this study had formed a monophyletic clade with the other An. balabacensis from Sabah and Indonesia, the collected An. barbirostris Clade VI appeared to have formed its own clade with those from Peninsular Malaysia, distinct from the other members of the Barbirostris Subgroup. It is highly likely that the An. barbirostris Clade VI reported in this study is a previously uncharacterised, closely related species within the Barbirostris Group. In addition to being morphologically and phylogenetically similar to other members of the Barbirostris Group, the length of its ITSII region has further supported this hypothesis (Table 3). To the best of our knowledge, members of the Barbirostris Complex are the only Anopheles species with an ITSII sequence length of more than 1.5 kb, mainly due to its multiple internal repeats [35]. Despite the roles of both An. barbirostris s. l. and An. balabacensis in malaria transmission [70–74], the taxonomy of the latter species and its species group was studied more extensively than the former [35,38,40,75–77]. It is therefore likely that more cryptic species will be discovered in the Barbirostris Group than within the Leucosphyrus Group.
The Leucosphyrus Group has been long thought to be the only species group capable of transmitting P. knowlesi. Anopheles kochi from the Kochi Group was suspected as a vector due to its high susceptibility to P. knowlesi infection under experimental conditions and its simiophilic biting behaviour but the parasite was never recovered from any An. kochi collected in the natural environment [4,78,79]. DNA of P. knowlesi was recently detected from the carcasses of An. donaldi and An. sundaicus collected from Sabah (Malaysia) and the Nicobar and Andaman Islands (India), respectively [63,80]. Apart from these two, our study now reports An. barbirostris Clade VI, a species from an entirely different subgenus than the Leucosphyrus Group, to be a vector for P. knowlesi in a natural setting. This suggests that there are possibly more Anopheles species that could transmit the pathogen and maintain the high level of endemicity in the macaque populations.
Nevertheless, in Sarawak, more detailed studies need to be conducted on the bionomics of both vectors to provide data for the implementation of appropriate vector control. Species-specific molecular assays should also be designed and utilised in future vector incrimination studies for this species range in order to correctly identify malaria infective mosquitoes.
Implications for vector control in Sarawak
The incrimination of An. balabacensis and An. barbirostris Clade VI as novel vectors for P. knowlesi in Northern Sarawak calls for re-evaluation of current and future vector control methods in the state. Detailed studies first need to be undertaken to determine the feeding behaviour and host preference of these vectors. From the aspect of insecticide-based preventative measures, future insecticide resistance surveys should include both An. balabacensis and An. barbirostris Clade VI to ensure that the insecticide used would still be efficient in killing these vectors. As the main vector control method currently adopted by the Sarawak State Health Department is the regular spraying of residual insecticide on houses in malarious area, spraying could also be considered for uninhabited buildings like the army camp in Long Luping, where human presence is intermittent. Apart from insecticides, clustered regularly interspaced short palindromic repeat (CRISPR)-based gene drive has been recently suggested as one possible prospect for the control of P. knowlesi vector(s) [81]. Gene drive is a biotechnology method used to increase the spread of a genetic trait (e.g. mosquito sterility/mosquito immunity against Plasmodium infection) into the wild population. As gene drive spreads a certain genetic trait in a mosquito population by means of mating and inheritance, it is an extremely species-specific vector control method [82]. Southeast Asia however, might not be ready for this technology as primary vectors of P. knowlesi in many parts of Southeast Asia have yet to be determined.