First molecular identification of mosquito vectors of Dirofilaria immitis in continental Portugal

Background Canine dirofilariasis due to Dirofilaria immitis is known to be endemic in continental Portugal. However, information about the transmitting mosquito species is still scarce, with only Culex theileri identified to date, albeit with L1-2, through dissection. This study was carried out to investigate the potential vectors of Dirofilaria spp. in continental Portugal. Methods Mosquitoes were collected in three distinct seasons (Summer, Autumn and Spring), 2011–2013, in three districts. CDC traps and indoor resting collections were carried out in the vicinity of kennels. Mosquitoes were kept under controlled conditions for 7 days to allow the development of larval stages of Dirofilaria spp.. DNA extraction was performed separately for both head+thorax and abdomen in order to differentiate infective and infected specimens, respectively, in pools, grouped according to the species and collection site (1–40 specimen parts/pool), and examined by PCR using pan-filarial specific primers. Mosquito densities were compared using non-parametric tests. Dirofilaria development units (DDU) were estimated. Results In total, 9156 female mosquitoes, from 11 different species, were captured. Mosquito densities varied among the 3 districts, according to capture method, and were generally higher in the second year of collections. From 5866 specimens screened by PCR, 23 head+thorax and 41 abdomens pools, corresponding to 54 mosquitoes were found positive for D. immitis DNA. These belonged to 5 species: Culex (Cux) theileri (estimated rate of infection (ERI)=0.71%), Cx. (Cux) pipiens f. pipiens and f. molestus (ERI=0.5%), Anopheles (Ano) maculipennis s.l. (ERI=3.12%), including An. (Ano) atroparvus, Aedes (Och) caspius (ERI=3.73%) and Ae. (Och) detritus s.l. (ERI=4.39%). All but Cx. pipiens, had at least one infective specimen. No D. repens infected specimens were found. Infection rates were: 3.21% in Coimbra, 1.22% in Setúbal and 0.54% in Santarém. DDU were at least 117/year in the study period. Conclusions Culex theileri, Cx. pipiens, An. maculipennis s.l. An. atroparvus, Ae.caspius and Ae. detritus s.l. were identified as potential vectors of D. immitis in three districts of Portugal, from Spring to Autumn, in 5 of the 6 collection dates in 2011–2013. Implications for transmission, in the context of climate changes, and need for prophylactic measures, are discussed.

Although the natural hosts of Dirofilaria spp. are dogs and wild members of the genus Canis, canine dirofilariasis (CD) infections may occur in a variety of species, including cats, other wild mammals and humans [2,3]. Previously human dirofilariasis (HD) was considered a rare disease, but a recent increase in the number of CD and HD cases, particularly after 2000, has resulted in it being classified as an emerging zoonosis [4,5]. Recent accounts of autochthonous cases of CD have stemmed from Slovakia [6], Hungary [7], Poland [8], and of HD from Hungary [9], Poland [10], Ukraine [11], and seroreactivity prevalences ranging from 5%-27% amongst humans, have been recorded in Serbia [12].
Dirofilaria spp. are transmitted by several mosquito species belonging to a wide range of genera in different parts of the world, such as Culex, Aedes and Anopheles [5]. Vectors ingest microfilariae, while feeding on an infected host, which then cross the midgut wall and migrate to the Malpighian tubules (MT) where they develop from first to third stage larvae. Later, the L3 (infective larvae) migrate to the proboscis through which they slide while the mosquito is feeding on another host, becoming sexually mature within six months in the main pulmonary arteries and right ventricle [1]. Transmission of dirofilariasis is dependent upon the presence of sufficient numbers of infected and microfilaremic dogs, susceptible mosquitoes, and a suitable climate to allow extrinsic incubation of the parasite in the mosquito vector [13,14]. Environmental factors, namely climatic and ecological, may affect the life cycle parameters of both the mosquito vector and filarial parasites.
Canine dirofilariasis due to D. immitis is known to be endemic in continental Portugal. In 2009-2010 the overall sero-prevalence in Northern and Central Portugal was 2.1% for CD [23]. A recent survey, 2011-2013, in three districts of Centre-South, has revealed an overall parasitological prevalence rate of 15.1%, the highest in Setúbal (24.8%), followed by Coimbra (13.8%) and Santarém (13.2%) [24].
Despite these high prevalences, information about the transmitting mosquito species was still scarce in continental Portugal, with Cx. theileri as the only likely vector of Dirofilaria spp. [25]. In addition, high densities of mosquito populations, namely Cx. theileri, Cx. pipiens s.l. An. maculipennis s.l. and Ae.caspius, were recorded in the above mentioned areas [26]. Hence, the purpose of this study was to identify potential vectors of Dirofilaria spp. by using a polymerase chain reaction (PCR) with species specific primers on mosquito populations from those three districts of continental Portugal, Coimbra, Santarém and Setúbal, collected in the vicinity of kennels being surveyed for CD, in a multidisciplinary project, for a period of two consecutive years.

Sampling area
The research was concentrated on three districts of Portugal: Coimbra (Centre), Santarém, and Setúbal (Centre-South), located at the basins of rivers Mondego, Tejo and Sado, respectively ( Figure 1). These districts present different prevalences of dog infections, ecological and overall soil use, although they have in common the presence of the main rice culture areas in the country. The number of localities surveyed in each district was, Coimbra-four, Santarém-five, and Setúbal-four.
Daily temperature data of 2011, 2012 and 2013 from stations operated close to the collections sites, were obtained from "Instituto Português do Mar e da Atmosfera" [27]. Average minimum and maximum monthly temperatures and rainfall values, for the study period are depicted in Table 1.

Mosquito collection and identification
Mosquitoes were collected by CDC light traps baited with dry ice, between 5.00 p.m. and 7.00 a.m., for active adult mosquitoes and with mechanical aspirators in the early morning targeting indoor resting mosquitoes (IR). Collections were carried out in kennels (whose identities are confidential) or their vicinity, but also in suburban or rural areas in those districts. Collections were carried out from 2011 to 2013, in July, October-November and April-May corresponding to Summer, Autumn and Spring seasons. Mosquitoes were kept in the insectary under controlled conditions of temperature and humidity (25±2°C, 70±5% RH), a photoperiod of 12 h:12 h (light:dark) and fed 10% sucrose solution, for 7 days to allow bloodmeal digestion and eventual parasite development to the infective L3 stage [1], as done in other studies [22]. After this period, those specimens still alive were frozen until species identification was carried out according to the keys of Ribeiro & Ramos [28]. Mosquitoes that were dead at time of trap collection, or that died during the following seven day period, were also frozen, identified and screened for filarial infection.

Set of biological material for PCR analysis
Mosquito females were dissected into head+thorax and abdomen to discriminate between Dirofilaria spp. infective/infected status, respectively [29]. Specimens belonging to the same collection, species, in identical gonotrophic stage, that had been the same number of days in the insectary, were joined in pools of these body parts, ranging from 1 to 40 specimen parts. Specimens of Cx. pipiens s.l.
were individually analyzed, due to the sympatric existence of the two biological forms, pipiens and molestus of the sensu strictu species, in Portugal [30].

DNA isolation
Genomic DNA was extracted from samples using the CTAB (Cetyltrimethylammonium bromide) method, adapted from Stothard et al. [31] by grinding the mosquito samples in a buffer (100 mMTris, 1.4 M NaCl, 20 mM EDTA, 2% Hexadecyltrimethylammonium bromide (CTAB), 0.2% mercaptoethanol) and incubating with  proteinase K (Bioline) at 55°C for 90 min with agitation. Phenol/chloroform/isoamyl alcohol was used for further DNA purification. DNA was ethanol precipitated and pellet was suspended in TE buffer (pH 7.0).
For all PCR reactions described above, amplified products were separated on 1.5% agarose gel eletrophoresis and observed under UV light.

Sensitivity test of PCR
In order to determine the sensitivity of the PCR assay, two procedures were devised. Assays were carried out with DNA extracted from Cx. theileri female mosquitoes from IHMT colony, also separated into head+thorax and abdomen: i) to determine the minimum amount of parasite DNA that would be detected by the PCR assay, definite amounts of parasite DNA (10 ng, 5 ng, 1 ng, 0.1 ng, 10 pg and 1 pg) were mixed with 80 ng of mosquito DNA, and PCR reaction was performed in same conditions as described above. This showed that it was able to detect up to 10 pg of parasite DNA in 80 ng of mosquito DNA, either from head+thorax or abdomen; ii) it was also determined the sensitivity cut-off of detecting an infected mosquito in a pool of 40 mosquitoes. After the first individual specimen of Cx. theileri positive for D. immitis was detected, a sample from this pool with 80 ng/μl of total DNA, was diluted in uninfected Cx. theileri DNA at the same 80 ng/μl concentration.

DNA sequencing and analysis
Products from the first PCR described (panfilarial) were purified by QIAquick PCR Purification Kit (Qiagen) and sequenced by Macrogen. Sequences were edited and aligned using BioEdit [35], and compared to other similar sequences available in Genbank, as identified through BLAST [36].

Calculation of the infection rate of mosquitoes
The infection rate of mosquitoes were estimated by: i) Minimum infection rate (MIR), i.e. the number of positive mosquito pools/total number of mosquitoes in pools tested×1000, and ii) Estimated Rate of Infection (ERI) which is adjusted for pooled samples, by the formula: ERI=1−(1-x/m)1/k where x is the number of positive pools; m the number of examined pools and k the average number of specimens in each pool [37].

Ethical considerations
The study was approved by the Commission on Ethics of the Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa with reference 21-2013-TM, and all procedures were performed according to national and European legislation.

Mosquito data and statistical analysis
Mosquito densities are presented as the number of mosquitoes captured per trap-night for CDC collections, or as the number of mosquitoes collected per collectorhour for IR collections. The arithmetic mean and the standard deviation were calculated for densities per district for all collections of each type, and date. However, the median and interquartile interval (Q1-Q3) revealed to be most appropriate for this data.
Statistical analysis was carried out using the SPSS package version 20.0 for Windows [38]. Kolmogorov-Smirnov (Lilliefors modification) and Shapiro-Wilk tests were used to analyse data for normality, while Levene's test was used to test for homogeneity of variance. Due to the lack of normality of the data, large standard deviations and lack of homogeneity of variance, non-parametric tests were used to analyse mosquito densities [39]. Mann-Whitney (MW) and Kruskal-Wallis (KW) tests were used for comparing, respectively, mosquito densities between the two years, and mosquito densities among the three districts. In the latter case, whenever significant differences were found, multiple comparisons were performed using the Dunn-Bonferroni (DB) pairwise comparisons.
Differences in mosquito rates of infection among species and locations were compared using Chi-squared test and Fisher's exact test.

Estimation of Dirofilaria development units (DDU)
In order to determine the hypothetical period in which there was risk of heartworm disease transmission in the surveyed areas, Dirofilaria Development Units (DDU) were calculated. For each day in which the average temperature was >14°C, temperature at which there is no extrinsic development of the parasite, the difference between the average temperature and 14°C was calculated (i.e. for Tmean≥15, DDU=Tmean-14) [40]. The sum of DDUs in the 30 days following the first day with average temperature >14°C, designated as DDU 30 , was then calculated. When DDU 30 is ≥130, it is assumed that a mosquito that might have taken a blood meal on a microfilaremic host on that particular day, had the possibility of allowing the completion of the extrinsic cycle, hence becoming infective, admitting an average mosquito life span of 30 days [2,13,40], independently of temperatures lower than 14°C during that period [41]. With this data, a bar graph was plotted depicting the favourable days for the completion of the extrinsic cycle, and for the transmission of heartworm in the areas and time periods studied [42].

Mosquito species captured and relative abundance
In total, 9156 female mosquitoes were caught in the whole sampling period (July/2011-May 2013), representing 11 species from five different genera. Culex (Culex) theileri was the most frequent species (5812, 63. The district of Santarém showed the highest number of mosquitoes captured (7818, 85.4%), followed by Coimbra (679, 7.4%) and Setúbal (659, 7.2%). Relative frequencies of the mosquito species caught in the different districts are depicted in Figure 1. Culex theileri was the most abundant species in Santarém and Coimbra, followed by Cx. pipiens s.l.. In Setúbal, the most frequent species found were Cx. pipiens s.l., Cx. theileri, An. maculipennis s.l. and Ae. caspius.
Average mosquito densities, and respective relative frequencies, were estimated according to the collecting method ( Figure 2, Table 2). For CDC traps, total mosquito densities differed among the 3 districts (KW: 14.231, DF=2, P=0.001) for the joint collections of the sampling period. Santarém exhibited a higher mosquito density compared just to Coimbra.
Culex theileri revealed different densities in the three districts (KW: 8.548, DF=2, P=0.014), being relatively more abundant just in Santarém compared to Coimbra.
Anopheles maculipennis s.l. collected by CDC traps did not reveal differences among the three surveyed districts.
As for IR collections, these densities also differed among the 3 districts for the total of the collecting period (KW: 9.802, DF=2, P=0.007). Mosquito density in Setúbal was significantly higher just in relation to Coimbra.
Culex pipiens s.l. also differed among the 3 districts for the total of the collecting period (KW: 7.230, DF=2, P=0.027), being more abundant just in Setúbal compared to Coimbra.
Anopheles maculipennis s.l., Cx. theileri and Ae. caspius were not significantly different between the three districts, in IR collections.
As to mosquito densities on the two surveying years,

Molecular detection of D. immitis DNA in mosquitoes
For PCR analysis, we used 5866 adult female mosquitoes. In total, 1815 head+thorax pools and 1529 abdomen pools were screened using the pan-filarial primers. This difference is due to bloodfed or semigravid females that still contained undigested blood in the abdomen, in order to avoid contamination of Dirofilaria spp. DNA that might be in the blood meal, thus preventing assumption of an established mosquito infection. Dirofilaria immitis DNA was found in the four most frequent species, but also in Ae. detritus s.l. (Table 3, with respective values of MIR, ERI and 95% CI). Culex pipiens s.l. positive pools for D. immitis, were identified as Cx. pipiens s.s., 7 form pipiens and 1 form molestus, which was from Setúbal.
The distribution of positive mosquitoes over the three sampled districts, and their respective values of MIR, ERI and 95% CI are depicted in Table 4.
No Overall, mosquitoes with D. immitis DNA were found in all collecting dates, but November 2011, usually by both methods and in more than one district (Figure 2, Table 5).

Estimation of transmission risk of Dirofilaria spp. by mosquitoes
The calculation of the DDU 30 for the three studied districts showed that there was, at least, 152 days in 2011, 119 days in 2012 and 117 days in 2013 with suitable conditions for the completion of the extrinsic development of Dirofilaria spp., and consequently, for its transmission to the vertebrate host (Figure 3). Most of the infected mosquito pools detected in this work (red lines) are in agreement with the determined favourable development periods.

Discussion
To our knowledge, this is the first report of molecular evidence for natural infections of mosquitoes with D. immitis in continental Portugal. Despite known prevalence of canine dirofilariasis (CD), the knowledge of its natural and potential vectors in mainland Portugal was scarce, with a historical study considering Cx. theileri as a probable vector of Dirofilaria spp. [25]. We report the finding of An. maculipennis s.l., Ae. caspius, Ae. detritus s.l. and Cx. theileri as likely competent vectors of D. immitis, i.e. with DNA in head+thorax, and Cx. pipiens form pipiens and form molestus, as likely vectors, i.e.with DNA only in abdomens, but from mosquitoes without any traces of bloodmeal. In this work, only D. immitis was detected, in contrast with recent findings of D. repens in other Southern European countries such as Italy, albeit in much lower rates than D. immitis, in Cx. pipiens [16,17], or at similar rates as D. immitis in Cx. pipiens and Ae. albopictus [44]. On the other hand, this is not surprising as D. repens was not found in parallel animal surveys in the same districts of Portugal [24].
Culex theileri, Cx. pipiens s.l. and An. maculipennis s.l. have already been implicated as vectors in countries such as Spain [15,20], Italy [16,17], Turkey [18] and Iran [21]. Aedes caspius and Ae. detritus s.l. are here for the first time, to the best of our knowledge, implicated as natural competent vectors of D. immitis. Aedes caspius had been found positive for the whole mosquito [17], and Ae. detritus s.l. for the abdominal portion [45], hence requiring confirmation. This is also the first study in which both biological forms of Cx. pipiens s.s., form pipiens and form molestus have been found infected with D. immitis. In Portugal, there are to date, records of Cx. pipiens and Cx. torrentium as members of the Cx. pipiens complex [46]. Culex torrentium is rare and occurs only in northern and mountain areas of the country [46], therefore, none of the collected specimens could belong to this species. As to Culex quinquefasciatus, although it has not yet been recorded in Portugal, hybrids with Cx. pipiens have recently been found in Greece [47]. For this reason, and considering either the ongoing climatic changes and its consequences on species distribution, or the similar PCR results between form molestus of Cx. pipiens s.s. and Cx. quinquefasciatus, all specimens were treated as Cx. pipiens s.l.; the molecular identification being made only for positive specimens for D. immitis. Anopheles maculipennis s.l. has also been previously found infected with D. immitis [16]. However, An. atroparvus is the only member of this complex occurring south of the Montejunto-Estrela mountain range, and even to the north of this range a proportion of nine An. atroparvus to one An. maculipennis s.s. was found [46,48,49]. Thus, we can be confident that the positive An. maculipennis s.l. in the district of Setúbal are in fact An. atroparvus, hence becoming the first vector incrimination for this species.
Aedes detritus s.l. were not differentiated as the technique available at the time of this study would preclude the screening for dirofilarial DNA.
Infection rates were similar whether estimated as MIR or ERI, probably as a great proportion of our pools were of a single mosquito specimen. The species with the highest infection rates were Ae. caspius (3.7%), followed by An. maculipennis s.l. (3.1%), Cx. theileri (0.7%) and Cx. pipiens s.l. (0.5%). The highest infection rate was in fact recorded for Ae. detritus s.l.. However, infection rates based on small sample sizes, i.e. <1000, may not accurately represent the true infection rate in the population [50]. Whereas in the case of An. maculipennis s.l. and Ae. caspius the sample size is 400 and 270, respectively, with somewhat large 95% CIs, and therefore, infection rates should be interpreted with caution, in the case of Ae. detritus s.l., with a sample of 23 specimens and a much wider 95% CI, the very high infection rate has a reduced significance.
In Portugal, previous detection of Dirofilaria spp. L1 and L2 larvae in the MT of Cx. theileri, in the district of Setúbal, had yielded an infection rate of 4.76%, again, in a small sample (N=42) [25]. In islands of Macaronesia, Cx. theileri has been found at infection rates of 0.16% in the Canaries [20] and circa 0.9%-1.13% in Madeira [19], to which the values in this study are more approximate.
In continental Portugal there are 41 identified species of mosquitoes [46], however, An. maculipennis s.l., Cx. pipiens s.l., Cx. theileri and Ae. caspius are the most abundant and broadly distributed [26]. In the three districts surveyed in this work, these were also the species with highest densities. Total mosquito densities were lower in Coimbra, the northern most district, in agreement with previous surveys, namely for Cx. pipiens s.l., by both methods, as a sign of identical capture yield for this species [26]. Conversely, CDC trap catches were able to show different densities among districts, for Cx. theileri and Ae. caspius, as they are superior for targeting these species [26]. On the other hand, IR catches yielded higher numbers of An. maculipennis s.l., as IR tends to be a more adequate method to capture this species [51]. Nevertheless, in this study, a striking difference was reported in the mosquito abundance in the district of Setúbal, with much lower densities compared to previous works [26,52,53]. The reasons contributing to this may well be i) the location of the collecting sites close to kennels, as per the experimental design, and which in this district were in areas not favourable for mosquito breeding, as opposed to earlier works which included rice fields and wetlands; ii) relatively low number of collections, with only one set in peak breeding season; and iii) strong winds registered in some of the collecting dates, as noted in field collection registers.
Collections in the second year yielded higher densities, particularly for Cx. theileri and Cx. pipiens s.l.. This increase may be due to local environmental variables, particularly climatic. Although no significant differences were registered for the average temperatures, precipitation was higher in the three districts in the second sampling year, 2012/2013. Nevertheless, considering there were only three collecting moments per year, there is not enough data to draw conclusions on the seasonal dynamics of mosquitoes.
Infection rates in mosquitoes were not in agreement with prevalence rates of CD found in the same research project [24], despite having targeted mosquito collections to the vicinity of kennels. Highest and lowest infection rates for mosquitoes were registered in Coimbra and Santarém, respectively, which had similar prevalence rates of CD. On the other hand, Setúbal, which had the highest CD prevalence rate, registered an intermediate mosquito infection rate. There are many factors whose influence is still unknown (vector efficiency of each species, overall level of protection in the dog population by preventive therapy, local environmental conditions in studied areas, etc.).
During the two year survey, infected mosquitoes were found in five of the six collection dates, representing an almost continuous presence of infected vectors, particularly in Santarém. Furthermore, infected mosquitoes were found by both methods in most of the collecting dates and sites. It can be argued that infected mosquitoes in IR collections may have become infected in the hosts in the shelters, however, the CDC trap collected mosquitoes represent the mosquito population searching for hosts and are proof therefore of circulating infected vectors. Coimbra was the only district with the five infected species as Ae. detritus s.l. was only found infected there, while Santarém and Setúbal registered four infected species.
The calculation of the DDU 30   infective and infected mosquitoes detected were collected during these favourable periods. The few exceptions are probably due to their maintenance in the insectary for the seven day period, which proves highly important in such studies. Although a 7 day period at 25°C may not be enough for the completion of the extrinsic cycle, a compromise had to be taken to compensate for mosquito mortality and filarial DNA degradation, while allowing for complete digestion of bloodmeals.
Activity of these mosquitoes, whether infected, infective or neither, was found in the three time point collections, corresponding to Spring, Summer and Autumn. In the context of climate changes, particularly in Portugal, where temperature increases have reached 0.5°C/decade since 1970, more than twice higher than the global median temperature [54], and with future scenarios that may range between 3-5.8°C by 2040-2090, the activity period for mosquitoes, and hence mosquito-borne diseases are likely to increase [55]. This is further relevant as dirofilariasis is recognized as an expanding zoonosis, particularly in Europe [4,5]. Our results are in agreement with predictions of occurrence and seasonality of Dirofilaria spp., with peaks of infection in Summer, from June to September, even in countries of Northern Europe [2,13,14,42].

Conclusions
We have confirmed and reported new mosquito vectors of dirofilariasis in three districts of Portugal with high prevalence of CD. To our knowledge, the present study is the first PCR screening for Dirofilaria spp. in mosquitoes for continental Portugal. Our results confirm that not only Cx. theileri is capable of becoming infected with D. immitis, but also Ae. caspius, An. maculipennis s.l., An. atroparvus, Cx. pipiens of both bioforms pipiens and molestus and Ae. detritus s.l. can support the development of D. immitis, and with the exception of Cx. pipiens, to the L3 infective stage, based on the presence of filarial DNA in the head+thorax. Most of these results were in agreement with the prediction of 130 DDU 30 for the regions surveyed. The finding of infected and infective mosquitoes in the three districts and in the Spring-Autumn interval heightens the necessity for prophylactic protective measures to prevent transmission at least during this period. Further studies are necessary to ascertain whether transmission season is wider than the interval Spring-Autumn.