Untangling the transmission dynamics of primary and secondary vectors of Trypanosoma cruzi in Colombia: parasite infection, feeding sources and discrete typing units

Background Trypanosoma cruzi is the causative agent of Chagas disease. Due to its genetic diversity has been classified into six Discrete Typing Units (DTUs) in association with transmission cycles. In Colombia, natural T. cruzi infection has been detected in 15 triatomine species. There is scarce information regarding the infection rates, DTUs and feeding preferences of secondary vectors. Therefore, the aim of this study was to determine T. cruzi infection rates, parasite DTU, ecotopes, insect stages, geographical location and bug feeding preferences across six different triatomine species. Methods A total of 245 insects were collected in seven departments of Colombia. We conducted molecular detection and genotyping of T. cruzi with subsequent identification of food sources. The frequency of infection, DTUs, TcI genotypes and feeding sources were plotted across the six species studied. A logistic regression model risk was estimated with insects positive for T. cruzi according to demographic and eco-epidemiological characteristics. Results We collected 85 specimens of Panstrongylus geniculatus, 77 Rhodnius prolixus, 37 R. pallescens, 34 Triatoma maculata, 8 R. pictipes and 4 T. dimidiata. The overall T. cruzi infection rate was 61.2% and presented statistical associations with the departments Meta (OR: 2.65; 95% CI: 1.69–4.17) and Guajira (OR: 2.13; 95% CI: 1.16–3.94); peridomestic ecotope (OR: 2.52: 95% CI: 1.62–3.93); the vector species P. geniculatus (OR: 2.40; 95% CI: 1.51–3.82) and T. maculata (OR: 2.09; 95% CI: 1.02–4.29); females (OR: 2.05; 95% CI: 1.39–3.04) and feeding on opossum (OR: 3.15; 95% CI: 1.85–11.69) and human blood (OR: 1.55; 95% CI: 1.07–2.24). Regarding the DTUs, we observed TcI (67.3%), TcII (6.7%), TcIII (8.7%), TcIV (4.0%) and TcV (6.0%). Across the samples typed as TcI, we detected TcIDom (19%) and sylvatic TcI (75%). The frequencies of feeding sources were 59.4% (human blood); 11.2% (hen); 9.6% (bat); 5.6% (opossum); 5.1% (mouse); 4.1% (dog); 3.0% (rodent); 1.0% (armadillo); and 1.0% (cow). Conclusions New scenarios of T. cruzi transmission caused by secondary and sylvatic vectors are considered. The findings of sylvatic DTUs from bugs collected in domestic and peridomestic ecotopes confirms the emerging transmission scenarios in Colombia. Electronic supplementary material The online version of this article (doi:10.1186/s13071-016-1907-5) contains supplementary material, which is available to authorized users.


Background
Chagas disease caused by the protozoan parasite Trypanosoma cruzi, affects about six million people in Latin America. The main transmission mechanism is by insect vectors (stercoralian route). The insects responsible for vector transmission belong to the subfamily Triatominae (Hemiptera: Reduviidae), composed by approximately 140 species of 5 tribes [1]. The natural habitats of triatomines include palm trees, tree holes, cracks in rocks, small caves and other animal shelters [2]. The main vectors of T. cruzi in the Southern Cone countries are Triatoma infestans, Triatoma brasiliensis and Panstrogylus megistus; Rhodnius prolixus and Triatoma dimidiata in the Andean region and parts of Central America, and Triatoma dimidiata and T. barberi in Mexico [3]. A total of 26 species have been reported in Colombia; of these 15 have been shown to be naturally infected with T. cruzi [4]. Thus, R. prolixus and T. dimidiata are considered primary vectors, whereas P. geniculatus, T. maculata, R. pictipes and R. pallescens are considered secondary vectors, among others that have been found naturally infected with T. cruzi. In Colombia, there are approximately 436,000 people infected with T. cruzi, with an annual incidence of 5,250 vector-borne cases per population [5]. In addition, some studies revealed that species at greatest risk of transmission are R. prolixus, T. dimidiata, T. maculata and T. venosa [4]. However, vector control programs have focused on domiciled species as R. prolixus and T. dimidiata.
Trypanosoma cruzi transmission mostly occurs in three epidemiological cycles: sylvatic (enzootic), domestic and peridomestic, where the parasite circulates among triatomines, mammal reservoirs and human hosts. Around 180 sylvatic and synanthropic species of mammal have been described to date, which act as reservoirs of T. cruzi inhabiting places near human settlements [6]. Trypanosoma cruzi exhibits remarkable genetic diversity and has been classified by international consensus in six Discrete Typing Units (DTUs) (TcI-TcVI), plus a new genotype associated to anthropogenic bats (TcBat) [7]. The DTUs present associations with transmission cycles, geography, vector species and clinical manifestation to some extent [7][8][9][10][11][12]. The DTU with the broadest geographical distribution is TcI and due to its high genetic diversity has been divided into two genotypes, associated to domestic and sylvatic foci (TcIDom and Sylvatic TcI, respectively) [13][14][15][16].
Current molecular tools have improved the detection and genotyping of the parasite in the vectors as well as the identification of the vectors' feeding sources involved in the transmission of the parasite. However, in Colombia there is scarce information about the relationship between different vector species, T. cruzi infection and eco-epidemiological aspects of the transmission cycles. Control programs in this country have focused on domiciled vectors while epidemiological and scientific research has been restricted to the Caribbean region. Therefore, the study of vectors in other areas of the country with a holistic perspective that considers T. cruzi infection rates, genotyping and feeding sources will help to identify the hosts and understand the dynamics of parasite transmission. This information will be useful to generate vector control strategies and accurate surveillance of Chagas disease. Thus, the objective of this study was to apply this holistic perspective to six triatomine species collected in different transmission cycles from seven departments in Colombia.

Study area and collection of triatomines
A total of 245 specimens corresponding to six species (85 P. geniculatus, 77 R. prolixus, 37 R. pallescens, 34 T. maculata, 8 R. pictipes and 4 T. dimidiata) were collected in the departments Guajira, Antioquia, Cesar, Norte de Santander, Meta, Casanare and Huila ( Fig. 1; Additional file 1: Table S1). Insect capture in the sylvatic cycle was performed using two techniques: manual search and modified Noireau baited chicken traps in palms and distant to housing areas. Additionally, insects inside houses and in the peridomestic ecotope were collected. All insects were stored in separate jars, including a description of the site of collection and georeferenced using GPS. The insects were identified using taxonomic keys and stored in 100% ethanol until processing [17]. All of the specimens except T. dimidiata were collected from domestic, peridomestic and sylvatic cycles of transmission (Additional file 1: Table S1; Fig. 2).

Molecular characterization of blood sources
All the 245 specimens were submitted to identification of feeding preferences by PCR-HRM (Polymerase chain reaction-High resolution melting) as previously reported [23]. Also, the feeding preferences were corroborated by direct sequencing of cytb using the primers CytbFw (5′-CCC CTC AGA ATG ATA TTT GTC CTC A-3′) and CytbRv (5′-CCA TCC AAC ATC TCA GCA TGA TGA AA-3′). The resulting sequences were edited in MEGA 6.0 [24] and submitted to BLASTn similarity search. This methodology allows the detection of 14 host species involved in the epidemiological cycles of Chagas disease as previously reported [23].

Statistical analyses
We calculated the frequency of T. cruzi infection, DTUs, TcI genotypes and feeding preferences across species and ecotopes (transmission cycles). To establish the association between the variables, Chi-square test was implemented with Monte Carlo adjustment with 10,000 simulations and G-test including pairwise comparison. Gtest was not applied in cases where the contingency table had zeros (Additional file 2: Table S2). Additionally, using an unconditional logistic regression, without including the intercept, the risk for infection with T. cruzi was estimated (OR, 95% IC) according to the demographic and ecoepidemiological characteristics using EpiInfo V.3.5.4 software and R package version 3.3.1.1. Statistical significance was established with a P-value < 0.05.

Trypanosoma cruzi detection in six species of triatomines
The information regarding the frequency of infection with T. cruzi by geographical location, ecotope and insect stage is shown in Table 1. The overall T. cruzi infection rate was 61.2% (n = 150). The species with the highest percentage of infection with T. cruzi was P. geniculatus, followed by R. prolixus and T. maculata (Fig. 3a). In addition, one R. prolixus was positive for T. rangeli.
Finally, the species positive for T. cruzi and bloodfeeding on humans are shown in Table 3.

Discussion
Triatomine species collected in this study with the exception of R. prolixus and T. dimidiata are considered sylvatic and secondary vectors. Nevertheless, they were mainly collected in domestic ecotopes (Fig. 2), and there was an association between the ecotopes and species; and the main dietary source was human blood mainly in the domestic ecotope. Feeding sources with sylvatic reservoirs blood was the lowest in our dataset. These findings reflect the intrusion of vectors from sylvatic habitats to domestic habitats and their adaptation to the available feeding sources.
The high percentage of infection with T. cruzi and its relationship to the specific tested variables (peridomestic ecotope, feeding with human and opossum blood), together with the occurrence of sylvatic DTUs (sylvatic TcI and TcIII); P. geniculatus and T. maculata with high percentages of infection and feeding with human blood. They all suggest the existence of possible new transmission scenarios caused by intrusion of secondary vectors (mainly P. geniculatus and T. maculata in Meta and Guajira departments, respectively). The association of TcI genotypes with ecotopes, feeding sources and species is relevant given that the higher frequencies corresponded to sylvatic TcI, domestic ecotopes, feeding sources of humans, domestic and sylvatic animals. Therefore, since the statistically significant associations are the evidence of parasite population's movement from sylvatic to "domestic" populations with sylvatic strains is confirmed. This is of paramount relevance due to the impact of sylvatic T. cruzi in the acute phase and outbreaks of oral transmission. Additionally, statistically significant association between food sources (human and opossum) and T. cruzi infection reaffirms the potential of D. marsupialis as an important reservoir of the parasite [25,26]. Our results are in accordance with other studies conducted in the Caribbean region of Colombia, which showed that secondary vectors play an important role in the different epidemiological transmission cycles of T. cruzi. These studies have shown that the frequency of patients with positive serology in the presence of sylvatic vectors is similar to the frequency in the presence of domestic vectors [4,27]. Furthermore, our findings reinforce the role of R. prolixus as a domestic vector in Colombia given that among the triatomine species collected in this study, R. prolixus had the highest frequency of feeding with human blood and at the same time showed high rate of T. cruzi infection (33.8%). The frequency of R. prolixus specimens collected in domestic habitats was 98.4% (Fig. 3a). The feeding sources of R. prolixus were mainly humans and domestic animals (Fig. 4a) and the DTUs detected were primarily associated with domestic cycles (TcI and TcII), and mostly infected with TcIDom (Fig. 5). These findings must be a support for the vector control programs in the country. Mainly, because most of the efforts have been focused on domestic vectors such as R. prolixus and T. dimidiata and our findings evidenced the potential risk of T. cruzi transmission by secondary vectors. Therefore, monitoring and control strategies specifically designed for sylvatic vectors are required in Colombia [28].
Surprisingly, we detected TcV in specimens of P. geniculatus and R. pictipes (sylvatic vectors). This DTU has been reported in domestic cycles from southern Latin-American countries [7,29]. However, insects infected with TcV have been reported in domestic habitats in Colombia including some reports of human infections [10,30]. TcV associated with sylvatic cycles has been reported with a frequency of 3.5% [29]. Consistent with our findings of TcV, a recent study [31] reported the presence of TcV in Colombian isolates obtained from P. geniculatus, R. prolixus, T. venosa and armadillos, using high-resolution markers: MLST, MLMT and ten mitochondrial markers. Messenger et al. [31] have also shown that the Colombian TcV isolates are due to migration processes from Southern Cone countries and not to local hybridization processes.
Triatoma maculata has a wide geographical distribution in Colombia. We observed a high invasion of domestic ecotopes by this species (Fig. 2), consistent with other studies in Colombia, Brazil and Venezuela where this species even presents morphological and genetic changes across individuals collected in domestic ecotopes [26,27,[32][33][34]. Triatoma maculata has not been included in the vector control programs because its diet is mainly composed of bird blood [32,[35][36][37] and some studies have reported low frequency of infection in Brazil and Venezuela [36,38]. By contrast, herein the frequency of T. cruzi infection was 67.6% and the percentage of feeding with human blood was 75.0% with the presence of TcIDom in some specimens collected in peridomiciliary habitats. Recent studies in Colombia and Venezuela have revealed infection frequencies between 38.0 and 75.0% and the presence of "TcIb" genotype that is associated with the peridomestic cycle [26,27,33,34,39,40]. Regarding the DTUs herein detected, most of the specimens were infected with TcI and TcIII (sylvatic DTUs) and domestic specimens harbored TcIII suggesting how T. maculata can connect domestic and sylvatic transmision cycles (Fig. 4). Our results and previous reports highlight the relevance of T. maculata as a potential vector in Colombia suggesting the need to prioritize this species in vector control programs, and additionally, to be cautious about the potential risk of domiciliation that this species may have. Panstrongylus geniculatus is the most widely distributed in Latin America species of the genus Panstrongylus [41]. In Colombia this species has the widest geographical distribution and is recorded in 25 departments including the Department of Meta [4]. In this study, most of the specimens were collected from the domestic ecotope ( Fig. 2; Additional file 3: Table S3). Panstrongylus geniculatus is considered a sylvatic vector that inhabits in burrows primarily associated with armadillos, opossums, rodents and bats [41]. Accordingly, there are several reports in Latin America and Colombia demonstrating the intrusion of adult specimens of P. geniculatus in domestic habitats [42][43][44][45] and also findings of different nymphal stages in human dwellings mainly in Brazil, Venezuela and Colombia (Amalfi, Antioquia) [46][47][48][49][50]. Colonization in domestic habitats may be due to changes generated by housing construction and alteration of the ecosystems in the municipalities analyzed and/or by the attraction generated by the artificial light [51].
Our results showed P. geniculatus ranking first in T. cruzi infection ( Fig. 3a; Table 2; Additional file 3:  Table S3), similar to previous reports in Brazil, Venezuela   [40,46,48,49]. We detected five DTUs in this species (TcI-TcV) that is in accordance with previous reports in Brazil, Venezuela and the Colombian Orinoco, and showing the interesting permissivity of this species facilitating the transmission of a wide variety of DTUs [10,46,48,51]. In contrast with other reports, we report a wide range of feeding sources with mammals of different epidemiological cycles and human blood. These results demonstrate the strong adaptive ability to different food sources and possibly explaining the high rate of infection and variety of DTUs. Human and canine blood constitute an important food source for P. geniculatus and similar to that observed in Venezuela, where also ten outbreaks of T. cruzi oral transmission involving this species were reported [48,49,52]. A small quantity of P. geniculatus fed on sylvatic sources (opossums, rodents, bats and armadillos) explaining the detection of sylvatic DTUs (TcI sylvatic, TcIII and TcIV) across our dataset [2,41,53] (Fig. 5a). All these findings could be a signal of an intrusive process from sylvatic ecotopes to domestic ones facilitated by adaptation, occurring in parallel in Colombia, Venezuela and Brazil [42][43][44][45], even accompanied by morphological changes in insects [54,55]. This pattern of ecotope intrusion explains the incrimination of P. geniculatus in oral outbreaks in Colombia [28,56]; it is mandatory that surveillance strategies are deployed not only to avoid incrimination but also a possible domiciliation process. Another sylvatic species is R. pallescens, given that its habitats are tree palms of Attalea butyracea and the invasion in domestic ecotopes is seen through intrusion [2,[57][58][59][60]. Herein, insects were found in the three ecotopes and most specimens (both nymphs and adults) were collected in sylvatic habitats mainly in wine (A. butyracea) and oil (Elaeis guineensis) palm trees, while 21.6% (adults) were collected in domestic habitats. This is consistent with previous studies in Colombia [27,57,58,60,61] and Panama [40,58,[62][63][64][65], where the percentage of T. cruzi infection is similar to our results. Moreover, it was observed that the main feeding source was human blood across the three ecotopes. This might be due to agriculture at the collection sites, where palms are nearby to homes and this can facilitate contact between the vectors and human hosts. We report that R. pallescens also fed on different mammals such as D. marsupialis in the peridomicile habitat and bats, mice and dogs in the sylvatic ecotope, showing an interplay between peridomestic and sylvatic cycles [59, 62-64, 66, 67]. Finally, in the domestic ecotope the sylvatic TcI was only detected, showing that in fact, the specimens of R. pallescens might correspond to intrusions in homes from the sylvatic habitat. However, in contrast Cantillo et al. [58], detected "TcIb" in specimens of R. pallescens collected in domestic habitats, confirming that in Colombia there is a risk of R. pallescens intrusion supported by the high presence blood-feeding on humans and T. cruzi infection.

Conclusions
To the best of our knowledge, we conducted the first robust study sampling secondary vectors of T. cruzi in Colombia from different locations of the country. We used a broad variety of techniques to detect T. cruzi infection, DTUs, TcI genotypes and feeding sources that allowed us to understand the transmission dynamics in secondary vectors such as P. geniculatus, T. maculata and R. pallescens. Our findings reinforce the epidemiological relevance of these species and highlight the need to include them in the vector control programmes as well as in the entomological surveillance systems. Most of the secondary insects captured harbored sylvatic DTUs and fed on human blood highlighting the importance of these species. Our results demonstrate the need of the government to invest on the control of them in the light of their effort to interrupt T. cruzi transmission in Colombia.