| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
 |
ABSTRACT |
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
INTRODUCTION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The blood feeding behavior of Anopheles mosquitoes has been the subject of considerable research, because of its fundamental importance to the transmission of human malaria parasites. Of particular interest are the theoretical5,6 and empirical2,7–12 implications of non-random mosquito–host interactions. Heterogeneities in the frequency at which humans are bitten may be caused by variation in attraction and/or permissiveness of hosts to mosquitoes. Variation can be influenced by the race, age, size, and health of human hosts as well as by environmental factors or anthropogenic modifications of the environment. The complexities of non-random human exposure to mosquito bites can, therefore, have a substantial impact on observed patterns of malaria transmission and ultimately prediction of disease risk and disease prevention strategies.
Development of polymerase chain reaction (PCR)-based methodologies since the early 1990s—primarily motivated by forensic science—has improved the capacity of researchers to identify precisely the person from whom a mosquito imbibed its blood meal.10–13 In this study, we used a DNA fingerprinting approach developed for Aedes aegypti12 to examine interactions between two of the most important malaria vectors in sub-Saharan Africa (Anopheles funestus and An. gambiae) and humans in western Kenya, where both mosquito species and malaria transmission have been extensively studied.14–16 Our goal was to provide new details and build on existing knowledge of mosquito blood feeding behavior. Over a series of days, we made repeated morning, aspiration collections of mosquitoes resting in natural human habitations so that the specimens we examined represented natural feeding patterns. DNA profiling of blood meals was used to determine the exact person(s) each mosquito had bitten. To examine potential habitat effects, we compared feeding patterns in high-altitude and lower-altitude areas with distinct ecology and malaria epidemiology.16 The design of previous studies largely used experimental huts, artificial combinations of human hosts, or less specific methods of identifying the person bitten. This study had the following four objectives. First, we determined if there is a predictable pattern for the people who are bitten most often. If heterogeneities are consistent across different ecological settings and can be accounted for in an operationally and economically feasible way, details about mosquito–human interaction could significantly enhance application, evaluation, and success of disease prevention strategies.2,9 Second, we determined whether multiple blood meals during a single gonotrophic cycle17,18 are taken from different people frequently enough to be epidemiologically significant. A relative increase in biting rate is expected to increase the force of malaria transmission.6,19 Third, we determined the extent to which mosquitoes collected resting indoors imbibed blood only from people who slept in the collection house. It is important to validate the assumption that all mosquitoes resting inside houses bit residents of that household13 because this is the basis for some estimates of entomological risk (i.e., entomological inoculation rate). Fourth, we compared these patterns in mosquito–human interaction across two different habitats (lowland and highland sites) to determine whether patterns in blood feeding behavior differ in ecologically and epidemiologically distinct environments.16,20
|
MATERIALS AND METHODS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Mosquitoes were collected in houses based on preliminary assessments for presence of resting anophelines. In Kombewa during 6–13 November 2003, we collected mosquitoes from 12 houses in four compounds. Very few resting anophelines were observed in houses in Iguhu during November 2003, and therefore, no wild mosquitoes were collected from that site during that time period. During 21–30 June and 20–22 July 2004, anophelines were collected from 10 houses in three of the same Kombewa compounds that were sampled during November 2003. During 4–22 June 2004, mosquitoes were collected from eight houses in three compounds in Iguhu. Temperature, humidity, and rainfall data were recorded from each site using HOBO automated weather stations (Onset Computer Corp., Bourne, MA).
DNA detection. To determine the time interval between when a mosquito took a blood meal and when we could detect allelic profiles,21 we carried out a time series study. During 10–13 November 2003, An. gambiae s.s. from a laboratory colony (originating from Iguhu and maintained in colony since October 2002) were allowed to feed on a human arm (TWS). Within 1 hour of feeding, fully engorged females were divided equally and placed into one of two 30-cm2 screened cages. Both cages were held at ambient environmental conditions: one in a house near the field laboratory, which was the same altitude and environment as Kombewa, and the other in a house in Iguhu near homes where wild mosquitoes were collected during 2004. Hourly temperature data were recorded at both sites. Mosquitoes were provided with water but no food (i.e., no carbohydrate or additional blood). At 6-hour intervals for 2 days, five mosquitoes were removed from each cage, and within 1 hour, their abdomens were removed and ground in lysis buffer as described below.
Mosquito collection and processing. Mosquitoes resting on surfaces inside houses were collected by mouth aspiration between 8:00 AM and 1:00 PM, transferred to 0.5-L cardboard cartons with a mesh top, transported to the field laboratory where they were sedated with cold (4°C), and identified based on morphology under a dissecting microscope.22 Abdomens of engorged females were removed and ground in 400 µL of lysis buffer (1% SDS added to TE buffer). For each sample, a code for the collection house, collecting date, degree of engorgement, and color of homogenate after grinding was recorded in the field laboratory before refrigeration (4°C) and shipment to UC Davis for fingerprinting analysis.
Oral swabs. Allelic profiles of all people sleeping in study compounds were determined from cheek cells collected in oral swabs in compliance with approved human subjects protocols. The inner surface of each person’s mouth was rubbed with four sterile wooden applicator sticks, and cells were suspending in 400 µL of lysis buffer, which consistently yields 6–200 ng of human DNA.12,23
DNA fingerprinting. Methods for DNA fingerprinting were modifications of approaches described by Chow-Shaffer and others23 and DeBenedictis and others.12 DNA was extracted from blood meals and oral swabs with phenol-chloroform using Phase Lock tubes (5'-3' Corp., Boulder, CO), re-suspended in 50 µL ddH2O, and stored at –20°C. A slot blot (ACES Human DNA quantification kit; Gibco-BRL Life Technologies, Gaithersburg, MD) was used to determine the amount of template DNA needed for optimum PCR amplification. Alleles at the CTT (CSF1, TP0X, and TH01) and Silver-STR (D16S, D7S, and D13S) loci were amplified using a GenePrint triplex STR multiplier kit (Promega, Madison, WI) and a GeneAmp 2400 PCR System Thermal Cycler (Perkin-Elmer). Amplified allelic DNA was separated on 4% 19:1 acrylamide:bis denaturing gels and visualized by silver staining, and data were stored in Microsoft Excel spreadsheets. Allelic profiles of blood meals were matched with those of cheek swabs using the computer program BloodID, ver. 1.01. The program lists the identification codes of all people whose blood might have been in the blood meal, and then lists in order of group size the identification code(s) of the person or groups of people whose combined alleles account for all the alleles in the blood meal. A priori matching criteria are described by DeBenedictis and others.12
Species identification. Head and thoraces from a sample of specimens collected during each time period and at both locations were examined using rDNA-PCR markers to determine their species designation following the methods of Scott and others24 for An. gambiae s.l. or Koekemoer and others25 for An. funestus s.l. Only specimens that were morphologically identified as An. gambiae s.l. or An. funestus s.l. and whose blood meal was completely profiled were identified to species.
Research with human subjects. Human DNA samples were collected after informed consent was obtained from all adult participants, from parents or legal guardians of minors, and assent from minors in adherence to human subjects protocols approved by Human Subjects Institutional Review Boards at KEMRI (protocol 637), UC Davis (protocol 200210596), and SUNY Buffalo (protocol BIO0020701A). Similar approval was obtained for TWS to allow laboratory-reared mosquitoes to feed on his arm.
|
RESULTS |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
DNA detection.
Human DNA in experimentally engorged An. gambiae was amplified and visualized in most specimens up to 30–42 hours after feeding, depending on temperature (Figure 1
). In Kombewa, where mean minimum and maximum temperatures were 19–30°C, all blood meals were either completely or partially profiled after 30 hours of digestion (8.3 degree-days). At 36 hours, we failed to detect DNA in more than one half the blood meals. In Iguhu, mean minimum and maximum temperatures ranged from 16 to 28°C, and all blood meals were either completely or partially profiled after 42 hours (6.8 degree-days). At 48 hours, human DNA in more than one half of the blood meals was not detected.
|
Blood meal analyses. A total of 1,343 engorged wild anophelines were collected during the 2-year study. At Kombewa, 336 An. funestus were collected during November 2003 and 130 An. funestus and 54 An. gambiae during June–July 2004. At Iguhu, no mosquitoes were collected during November 2003. During June 2004, 259 An. funestus and 564 An. gambiae were collected.
Minimum–maximum temperatures and monthly rainfall in Kombewa during November 2003 and June–July 2004 were 19–30°C and 109 mm and 18–29°C and 40 mm, respectively. Similar data from Iguhu were 16–28°C and 138 mm during November 2003 and 14–26°C and 256 mm during June–July 2004.
Table 1 summarizes by location and date the portion of specimens from which we obtained complete blood meal profiles (i.e., amplification and visualization of alleles at all six loci). When digestion of meals was not taken into account, we obtained complete profiles from 29–32% of the engorged anophelines examined. If color of the triturated blood meal was considered (e.g., red denotes a recently imbibed and brown a digested meal), we obtained complete profiles for recently imbibed blood in 92% (36/39) of An. gambiae in the laboratory-based time series study, 61% (265/432) of An. funestus collected from houses, and 46% (236/499) of An. gambiae collected from houses (Table 2). The proportion of completely profiled meals decreased and failed profiles increased with increasing blood meal digestion, as indicated by color of the ground meal.
|
|
). Of the 55 people in the study, we detected blood from 32 (58%) in 421 engorged anophelines. Forty-two percent of people (N = 23) were not bitten by any mosquitoes whose blood meals we profiled. Conversely, 4% of people (N = 2) accounted for 20% (N = 83) of all meals and 16% of people (N = 9), who were bitten 
20 times accounted for 58% (N = 245) of all meals.
|
2 analysis did not reveal differences by the sex of the person bitten within or among years or anopheline species. To the best of our knowledge, there were no pregnant women sleeping in collection houses.
The distribution of bites was associated with human host age (Table 4). For An. funestus collected at Kombewa during November 2003, people 
20 years old were bitten less often and those 21–30 years old more often than expected (2 = 11.3, P = 0.045, df = 5). All blood meals from the 21- to 30-year-old group were from two individuals: a 25-year-old mother and a 25-year-old father who slept in different compounds. During June–July 2004, An. funestus in Kombewa bit 11–20 year olds less often and 21–50 year olds more often than expected (2 = 19.1, P = 0.0018, df = 5). During June 2004, An. funestus in Iguhu bit children 10 years old and adults 41–50 years old less often and people 11–20 years old more often than expected (2 = 13.16, P = 0.0105, df = 4). There were no differences by human host age for blood meals imbibed by An. gambiae from Kombewa during November 2003 (2 = 6.39, P = 0.27, df = 5), although the majority of meals were from people 11–20 and 31–40 years old. During June 2004, An. gambiae bit children 10 years old and adults 41–50 years old less often and 11–20 year olds more often than expected (2 = 36.2, P < 0.01, df = 4); the majority of meals were from people in the later age group.
|
). The following June, the mother and new infant son also slept in the house, and the entire family slept under a bed net. Only two profiled mosquitoes bit household residents; one each on the daughter and youngest son. In a different house during November 2003, the father, mother, and their infant daughter slept under a bed net. None of the mosquitoes profiled from that house had imbibed their blood. Their other four children (ages 7 years) did not sleep under a net and were bitten by a total of 20 An. funestus.
Some mosquitoes bit more than one person during the 36-to 48-hour detection period of our fingerprinting methodology (Figure 1). The rate at which An. funestus in Kombewa imbibed blood from two different people was 3% (3/109) during 2003 and 2% (1/41) during 2004. In Iguhu during 2004, 14% (8/58) of An. funestus contained blood from two people. During 2004, blood from two people was not detected in An. gambiae collected in Kombewa (0/15), but 11% (20/191) of An. gambiae collected in Iguhu bit two people. There was no evidence that any of the mosquitoes we profiled imbibed blood from three or more different people.
Across mosquito species, locations, and collection periods, most mosquitoes (80–98%) ingested blood from a person who slept in the house where the mosquito was collected (Table 5). The rate at which people who were fed on slept in a house in the compound, but not in the collection house, ranged from 0 to 14%. The rate at which the person bitten did not sleep in any house in the compound ranged from 0 to 7%.
|
|
DISCUSSION |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The converse of some people being bitten frequently was our observation that almost one half the people were not bitten by any of the mosquitoes we examined, including people who were in the age group that was bitten most often. An improved understanding of why the majority of people were not bitten at all, how they avoided bites, and whether this is a stable pattern would improve our understanding of malaria transmission dynamics. It seems reasonable to attribute differences in biting frequency by both species to the relative attractiveness of different people,8,21 even though the abundance of mosquitoes may have varied from one house or compound to another. Most people in the study population did not sleep under bed nets or use other means of interfering with mosquito contact, and within households, we detected significant differences among the people bitten.
Foremost among the affects on nonrandom distribution of bites, as has been reported previously, was human age. Unlike some previous investigators,11–13 we did not detect a difference by the sex of the person bitten. As far as we were aware, our study population did not include pregnant women. Pregnancy has been shown to affect the attractiveness of adults to An. gambiae.10,26 In our study, young adults (11–25 years) were the age group that was bitten most often, followed by adults (21–50 years), and children (< 10 years) were bitten least often. Considering that, at the area where we worked, malaria morbidity and mortality is most severe for the pediatric portion of the population,27 this result highlights the efficiency of parasite transmission by An. funestus and An. gambiae. The portion of the human population with the lowest relative risk of receiving infectious bites—children < 10 years old—is nevertheless repeatedly exposed to parasites and adversely affected by the infection. The two people who were bitten most often, 40 and 43 times, were 14 and 15 years -old, respectively. Of the nine people who were bitten at least 20 times, six were 14–25 years old, two were 38 and 50 years old, and one was 2 years old. Why a child in an age group that overall was bitten least often (< 10 years) was bitten so frequently, how often this kind of apparent anomaly occurs, and the epidemiologic implications to parasite transmission of elevated biting on toddlers merits additional investigation. Port and Boreham7 determined that the positive relationship between host age and proportion of An. gambiae blood feeding is likely attributable to total surface area and weight of the exposed person. A similar pattern of age-dependent feeding success based on DNA profiling blood meals was described for Aedes aegypti12,23 and Culex quinquefasciatus.11 Data reported by Soremekum and others13 supported this trend, but they did not provide mosquito species or person specific data. Lacroix and others28 reported that, in an olfactometer assay conducted near Mbita, Nyanza Province, Kenya, children infected with the sexual stage of Plasmodium falciparum, which is transmissible to mosquitoes, were about twice as attractive to An. gambiae as children who were uninfected or were infected with asexual parasites. We did not record measures of body mass for study participants nor did we determine if they were parasitemic. Future studies should examine the impact of age-dependent human exposure to mosquito bites on the basic reproductive rate of malaria and the extent to which variation in body mass and parasite infection accounts for differences in feeding frequency across different age and risk groups. For example, why were 35% (N = 9) of the people in the age cohort that was bitten most often (11–25 years old) not bitten by any of the profiled mosquitoes and why was a 2-year-old toddler bitten 20 times?
A notable exception to age-dependent exposure to mosquito bites concerned the impact of an untreated bed net in house K25B. Our data on this topic supports results from experimental hut studies29 and shows that this phenomenon also occurs in natural home settings. When the parents and their infant daughter slept under a net, bites seemed to be redirected from parents to their four children who did not sleep under a net. Further detailed examination of this phenomenon is merited, but our observation suggests that bed net use by some, but not all, household residents can shift exposure from preferred to less preferred human hosts. When the latter are children, in holoendemic areas of malaria transmission, the shift may increase the risk of morbidity and mortality. We would expect insecticide treatment of bed nets to reduce or eliminate this effect.15
Both mosquito species took blood meals from more than one person during a 36- to 48-hour period. Gilles17,18 reported that > 20% of An. funestus and nearly all An. gambiae require two blood meals to complete their first gonotrophic cycle. Our results for multiple feeding support the idea that imbibing multiple meals during a single egg laying cycle would be expected to increase the basic reproductive rate of malaria parasites.5,19 The frequency of biting multiple people in our study ranged from 2 to 14% for An. funestus and 0 to 11% for An. gambiae. These frequencies are best considered conservative, lower estimates for reasons explained by DeBenedictis and others.12 We did not determine whether parasite infection status of mosquitoes influenced blood feeding patterns. In an experimental hut study with male volunteers > 15 years of age, Koella and others30 reported that An. gambiae infected with sporozoites were more likely to bite several people in a single night than uninfected mosquitoes: 20 versus 10%, respectively.
The frequency at which we failed to amplify DNA from blood meals, which did not seem to be digested beyond our capacity to amplify their DNA, indicate that at least some of the engorged An. funestus and An. gambiae we collected fed on nonhuman hosts. In our laboratory-based time series experiment, we were able to profile all or some of the loci in all 39 An. gambiae blood meals that were red when triturated. Among field-collected mosquitoes, however, DNA failed to amplify in 26 and 47% of the red blood meals from An. funestus and An. gambiae, respectively. Other investigators previously reported that in Kenya ~90% or more of the engorged An. funestus and An. gambiae imbibed blood from humans.31–33 It would be surprising if one quarter to one half of the An. funestus and An. gambiae in our study had imbibed non-human blood. A more reasonable explanation is that ~10% of the failures were the result of mosquitoes that did not bite a person and the rest were caused by undefined factors associated with processing field-collected specimens.
Regarding residence status of the person bitten, our results—relatively few profiled An. funestus or An. gambiae contained blood from a person who did not sleep in the collection house—are similar to those reported by Soremekun and others.13 We determined that when blood from a nonresident of the collection house was detected, it most often came from a person who slept in a different house within the same compound. An exception to this pattern occurred when 20% of An. funestus in Kombewa during 2003 bit nonresidents of the collection house. We assume that mosquitoes containing blood from a non-resident of the collection house bit a person in their home and subsequently flew to the collection house. We did not confirm that people actually slept where they reported that they slept. It is possible that some people were bitten while visiting a home before residents went to sleep. The latter scenario seems unlikely, however, because Githeko and others34 reported that only 2–13% of the An. funestus and An. gambiae in the region of western Kenya where we worked bit people before 10:00 PM, and 90% of the villagers were in bed 1 hour before that. In most cases, we would not expect the proportion of mosquitoes that obtained their blood meal from people who did not slept in the collection house to affect adversely calculation of measures for malaria transmission risk, such as entomological inoculation rates. Before this is accepted as a general rule, however, the frequency of exceptions like the one in Kombewa during 2003 should be determined. Moreover, calculation of entomological inoculation rates could be affected by the combined proportions of mosquitoes that did not bite a collection house resident and bit a nonhuman host.
Trends in feeding behavior that we detected were consistent for both species in highland and lowland sites and across different sampling periods. We did not detect site-specific differences even though slightly cooler temperatures at the highland site (Iguhu) seem to have slowed blood meal digestion and increased the probability of successfully amplifying human host DNA and, thus, identifying the person bitten.
PCR-based human DNA identification techniques have significantly improved the discriminatory power for reconstruction of natural interactions between mosquitoes and their human hosts.10–13 Across the mosquito taxa studied to date—Anopheles, Aedes, and Culex—investigators are gaining an increasingly detailed understanding of patterns, such as the ones we describe herein, in the ecology of vector–host interactions and the details of pathogen transmission.
Received December 1, 2005. Accepted for publication April 7, 2006.
Acknowledgments: The authors thank the people of Kombewa and Iguhu for working with us and allowing us to collect mosquitoes from their homes. Harryson Atieli, Peter Wamae, Carolyne Okoth, Yaw Afrane, Bryson Ndenga, and Robison Oriango collected and processed mosquitoes in the field laboratory.
Financial support: This research was supported by National Institutes of Health Grant AI-50243.
* Address correspondence to Thomas W. Scott, Department of Entomology, University of California, Davis, CA 95616. E-mail: twscott{at}ucdavis.edu 
Authors’ addresses: Thomas W. Scott and Andrew Fleisher, Department of Entomology, University of California, Davis, CA 95616. Andre Githeko, Kenya Medical Research Institute, Kisumu, Kenya. Laura C. Harrington, Department of Entomology Cornell University, Ithaca, NY 14853. Guiyun Yan, Department of Biological Sciences, State University of New York, Buffalo, NY 14260.
Reprint requests: Thomas W. Scott, Department of Entomology, University of California, Davis, CA 95616. E-mail: twscott{at}ucdavis.edu.
|
REFERENCES |
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. M. Darbro, A. A. Dhondt, F. M. Vermeylen, and L. C. Harrington Mycoplasma gallisepticum Infection in House Finches (Carpodacus mexicanus) Affects Mosquito Blood Feeding Patterns Am J Trop Med Hyg, September 1, 2007; 77(3): 488 - 494. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
