Determination of Human Identity from Anopheles stephensi Mosquito Blood Meals Using Direct Amplification and Massively Parallel Sequencing

Abstract

DNA obtained from biological evidence can link individuals to a crime scene. DNA is typically obtained from body fluids deposited on various substrates such as fabric or common household objects. However, other unusual sources of human biological material can also be used to generate DNA profiles. Here we show that short tandem repeat (STR) DNA profiles can also be obtained from single source and mixtures of human DNA in the blood meals of Anopheles stephensi mosquitoes. Using direct amplification with the PowerPlex® Fusion 6C System, we have determined that full and partial profiles can be obtained by assessing degradation of DNA at various times post-feed up to 20–24 hours post-blood meal. Moreover, we can assign donor identity through both STR profiles, as well as through single nucleotide polymorphisms (SNPs) detected using massively parallel sequencing (MPS) with the Precision ID Identity Panel and Ion Chef™/Ion S5™ System up to 24–48 hours post-blood meal. Based on the results from a total of 490 mosquitoes fed on 11 different sources of human blood, we conclude that both STR and SNP technologies can be applied to mosquito blood meals as effective forensic approaches to determine the identity of specific individuals and establish the timing of their presence at a crime scene.

1. Introduction

Evidence containing DNA at a crime scene can be helpful in linking an individual to a crime. Common substrates from which DNA can be obtained at a crime scene include different types of fabric, objects used to drink or eat, and other common items [1]. Most of the time, DNA is obtained from body fluids deposited on articles of clothing or other substrates. Although not typically considered forensically-relevant substrates, mosquitoes can be a potential source for human DNA at a crime scene as female mosquitoes feed upon and digest human blood containing nucleated white blood cells as a protein source for the production of eggs [2]. Mosquitoes maintain a ubiquitous presence in the environment, especially in the summer or in geographical areas with hot climates, so it would not be unusual for these insects to be present at crime scenes [3]. In fact, mosquitoes have been previously observed and documented at crime scenes around the world [3, 4].

There are many factors to consider when assessing the practicality of using mosquitoes as a forensic substrate at a crime scene. First, mosquitoes (of the family Culicidae) include approximately 3,600 different species and are commonly found throughout much of the world [3]. While most mosquitoes prefer warm and tropical climates, their viable range expands in the summer months. Moreover, there are several behaviors of mosquitoes that are beneficial to their use in forensic applications. For instance, there is some variance in flight behaviors, and feeding times, while host preferences vary across mosquito species, which also enables a greater range and possibility for their forensic application [5]. Additionally, mosquitoes strongly prefer to consume warm blood, thus limiting the sources and timing of possible blood donors [6, 7]. While mosquitoes that have not fed on blood can travel significant distances, those that have taken a blood meal often remain near the location of feeding until they have digested enough of the blood to readily fly [3, 8]. This is advantageous from a forensic perspective as mosquitoes that have fed on blood are more likely to linger near a crime scene to digest after the feed. Additionally, mosquitoes have the ability to feed on multiple sources of human blood [9], which poses an opportunity for the forensic science community to detect and interpret profiles of more than one donor. Taken together, when a mosquito is found at a crime scene and a mixture profile is detected from the blood, investigators can pursue more than one individual who may be involved in the crime or were present near the crime scene.

The forensic community routinely uses a few approaches in order to determine the identity of individuals. One way in which the DNA from blood can be used for individualization is through analysis of short tandem repeats (STRs). STRs are found in noncoding regions of the genome and each STR is comprised of repeated units of DNA seen in tandem at a particular location on a chromosome [10]. The combination of different STRs observed at different locations can be used to generate a unique DNA profile for an individual [10]. Previous studies, involving extraction or isolation of the DNA prior to amplification, have examined whether or not DNA profiles can be obtained from blood in mosquito midguts [3, 5, 9, 11–15]. Literature indicates that these researchers were successful in determining the identity of the individual from which the blood meals were taken using blood meals from several different species of mosquito. One study even found that a full profile could be obtained up to 48 hours [5], while another determined the limit to be 24 hours [13]. However, DNA extraction and quantification steps are not always ideal for forensic applications as they can be time-consuming, labor-intensive and expensive. Moreover, these procedures routinely cause the consumption and loss of some or all of the sample. Many of these issues can be mitigated through the use of direct amplification approaches, and when sufficiently clean DNA is produced, profiles can be produced within a short period of time [16, 17]. While the practicality of using new commercially available swabs designed for direct amplification of crime scene samples has been investigated [18, 19], use of such swabs with mosquitoes as a substrate has not. More recently, individualization can also be determined through DNA sequencing [20], whereby massively parallel sequencing (MPS) can be used to identify single nucleotide polymorphisms (SNPs) in the sequence of donor DNA compared to a reference genome. This approach is beneficial over STRs because MPS targets a greater amount of markers, provides actual sequence information and requires less DNA to obtain a profile [21]. PCR amplification of SNPs is also more advantageous when analyzing potentially degraded sample types due to smaller amplicons relative to those of STR amplification [20]. Several studies have documented success using MPS to identify SNPs with trace amounts of DNA or low quality DNA similar to that commonly encountered in forensic casework [22, 23].

The purpose of this study was to assess the feasibility of using blood-fed mosquitoes as a potential source of human DNA in personal violent crime cases. We sought to demonstrate if complete STR profiles can be obtained from both single sources and mixtures of human blood in Anopheles mosquitoes using direct amplification and whether those profiles are concordant with a reference profile from donor blood. We also investigated how the amount of time between feeding and euthanizing the mosquitoes can affect the ability to obtain complete STR profiles, as well as the presence and extent of degradation of DNA in the mosquito midgut. Finally, we also sought to determine whether an MPS approach can also be used to identify both single and multiple individuals. Together, we provide a refined experimental workflow for the use of blood-fed mosquitoes as a forensic tool.

2. Materials and Methods

2.1. Preparation of Mosquitoes and Feeding of Donor Blood Samples

Anopheles stephensi mosquitoes were reared using standard laboratory conditions. Three to five days post-eclosure they were allowed to feed upon human blood warmed to 37°C in a glass feeding apparatus connected to a circulating water bath. The blood (BioIVT, Westbury, NY, USA) used in the study was obtained from anonymous donors and were assigned a unique identifier upon receipt (donors 1 – 11). Prior to feeding, an aliquot of the blood was reserved to serve as a reference sample. Mosquitoes were fed on blood from either a single donor or on a mixture of blood from two donors (50:50 mixture), and then were collected and euthanized at various time points.

Initial experiments were conducted with goals of determining the preferred method of euthanasia and the relative time window post-blood meal in which partial or full STR profiles could be obtained for the direct amplification portion of the study. For these experiments, mosquitoes were euthanized at 0, 24, 48 and 72 hours post-feed. Twenty-seven mosquitoes were euthanized through immersion in 70% v/v ethanol and 60 mosquitoes were euthanized through freezing at −20°C to complete both goals of the initial experiments. Five single source feeds (donors 1–5) and one feed with a mixture of two donors (mixture of donors 4 + 5) were conducted. The samples were then subjected to the sampling, direct amplification and capillary electrophoresis procedures described in sections 2.2 of this paper in order to observe profile completeness at each time post-feed. The time window was determined by considering the longest time post-feed that a full profile could be observed, the shortest time interval that a partial profile could be observed, and the interval(s) at which no alleles were observed.

For subsequent experiments, mosquitoes were euthanized at 0, 4, 8, 12, 16, 20, 24, and 48 hours post-feed solely by freezing at −20°C. Six single source feeds (donors 6–11) and three feeds of blood with a mixture of two donors (6 + 7, 8 + 9 and 10 + 11) were conducted. Unfed mosquitoes were collected following identical treatment and were used to establish a baseline negative control for MPS experiments. Mosquitoes were stored at −20°C until analyses were conducted.

All mosquitoes used in this study, both unfed and those containing blood, were processed in compliance with the Institutional Biosafety Committee (IBC) of The Pennsylvania State University under protocols #47937 and #48327.

2.2. Direct Amplification

2.2.1. Sampling of reference blood and blood from mosquito midgut

Reference blood and blood from the midguts of the mosquitoes were collected using microFLOQ® direct collection devices (Copan Italia, Brescia, Italy). For processing of reference blood, the tips of the swabs were quickly placed on liquid samples. For mosquitoes, the tips of each independent swab were used to puncture the abdomen of a single mosquito and the blood was collected on the tip of one microFLOQ® device per mosquito. The tips of the swabs were cut and transferred to PCR amplification tubes. In situations where the blood appeared to be dry prior to collecting the blood from the midgut, such as with mosquitoes frozen at longer post-feed times, 2 μL of PunchSolution™ reagent (Promega Corporation, Madison, WI, USA) was added to the tip of the swab to aid in sample collection. A total of 188 mosquitoes were used in the direct amplification portion of the study.

2.2.2. Direct amplification of DNA for STR analysis

Blood from mosquitoes and reference samples (donors 6, 7, 8, 9, 10 and 11 and mixtures 6 + 7, 8 + 9, and 10 + 11) were amplified in at least biological duplicate (two different mosquitoes from the same feed euthanized at the same number of hours post-feed). Time points that did not initially yield full profiles were conducted with up to six biological replicates (six different mosquitoes from the same feed euthanized at the same number of hours post-feed) using direct amplification with the PowerPlex® Fusion 6C amplification kit (Promega Corporation) according to the manufacturer’s instructions for direct amplification of DNA from storage card punches in a 12.5 μL reaction volume [24]. The PowerPlex® Fusion 6C kit uses a six-dye system to target 27 STR loci including Amelogenin and three Y-STR loci. Briefly, the tips of the microFLOQ® swabs containing blood were pretreated with 10 μL of the PunchSolution™ reagent and incubated at 70°C for 30 minutes. The manufacturer’s recommended protocol for amplification, performed in a Veriti Thermal Cycler (Applied Biosystems, Foster City, CA, USA), was followed for direct amplification of DNA from storage card punches in a 12.5 μL reaction volume (a total of 2.5 μL of 5X Master Mix, 5X Primer Pair Mix, and 5X Amp Solution per sample with 5 μL of amplification grade water) [24]. During the amplification procedure, the tips remained submerged in the reagent. For most samples a cycle number of 25 was used for amplification [24]; however, in cases where a full profile could not be obtained, the cycle number was increased from 25 to 27 or 28 using a new mosquito. To assess contamination and whether inhibition was occurring from the microFLOQ® swabs remaining in the tubes during amplification, positive and negative controls were performed. Swabs with 5 ng and 10 ng of 2800M control DNA served as a positive control while swabs containing no DNA served as a negative control. The positive and negative controls were amplified in parallel with samples with experimental blood from the mosquito midgut.

2.2.3. Capillary electrophoresis and data analysis

After amplification, Hi-Di™ formamide (Applied Biosystems) was added to each of the amplified samples, along with the internal lane standard for the PowerPlex® Fusion 6C kit with an appropriate allelic ladder. Samples were denatured for 3 minutes at 95°C and snap-cooled for 3 minutes prior to injection on a 3130xl Genetic Analyzer (Applied Biosystems) in which POP-4™ polymer (Applied Biosystems) and a 1:10 dilution of Buffer (10X) with EDTA (Applied Biosystems) were installed and used.

Data analysis and interpretation of profiles was completed using the GeneMarker® HID Software version 2.9.0 (SoftGenetics LLC, State College, PA, USA). The analytical threshold for peaks was set to 50 relative fluorescence units (RFU) for all profiles.

2.3. Assessment of DNA Degradation in Mosquito Midguts

Blood was collected by two methods. First, for donors 6, 7 and the mixture of donors 6 + 7, blood was obtained by puncturing the mosquito’s abdomen with a cotton swab (Puritan, Guilford, ME, USA) and absorbing blood off of a microscope slide, with a second swab moistened with 2 μL of G2 buffer (Qiagen, Germantown, MD, USA) used to absorb remaining blood on the slide. The tips of both of these swabs were cut into an extraction tube and were processed using the EZ1 DNA Investigator kit (Qiagen) and the manufacturer’s recommended protocol for blood and saliva stains [25]. The samples were then incubated on a thermomixer at 900 rpm at 56°C for 90 minutes. Upon completion of the incubation period, the swabs were transferred to sterile cut pipette tips to allow separation of the sample from the swab by centrifugation in a fixed-speed Labnet mini-centrifuge at room temperature. Alternatively, for donors 8, 9, 10 and 11 as well as the two mixtures of donors 8 + 9 and 10 + 11, blood-fed mosquitoes were placed into a tube containing 190 μL of G2 buffer and were pierced with a needle to release the blood, which was then processed according to the manufacturer’s recommended protocol for casework and reference samples [25] for the EZ1 DNA Investigator kit (Qiagen). These samples were incubated on a thermomixer at 900 rpm at 56°C for 120 minutes, and the liquid sample was separated from the mosquitoes and transferred to a fresh tube. In all instances described above, each sample contained the blood from only one mosquito and the DNA from three individual mosquitoes were extracted at each time point for each donor. Therefore, 72 mosquitoes were extracted using each method for a total of 144 mosquitoes used for the degradation assessment. Samples were extracted using the BioRobot EZ1 using the trace protocol (Qiagen) and DNA was eluted in a final volume of 50 μL of TE buffer [25]. The samples were then quantified using the Quantifiler™ Trio DNA Quantification kit (Applied Biosystems) and the 7500 Real Time PCR System (Applied Biosystems) following the manufacturer’s recommended protocols [26]. Only non-concentrated DNAs were used in the degradation study (donors 6, 7, 8, 9 and mixtures of donors 6 + 7 and 8 + 9) as these samples are most representative of the actual concentration of DNA in the mosquito midgut.

2.4. MPS Analysis

2.4.1. Concentration of samples prior to MPS analysis

In order to test the differences between concentrated and non-concentrated samples using the MPS system and test the power of this technology, a portion of the DNA samples in the study were concentrated prior to MPS analysis. The concentration of DNA samples was performed in an Eppendorf Vacfuge Plus Concentrator System (Eppendorf, Hauppauge, NY, USA). All concentration procedures were performed at room temperature for 25 minutes at the fixed speed associated with this instrument. One extract of each triplicate set from donors 8, 9, and the mixture of donors 8 + 9 was concentrated and quantified before and after concentration. All samples from donors 10, 11 and mixture of donors 10 + 11 were concentrated and quantified only after concentration. A total of 215 mosquitoes involved in the extraction process were then analysed using MPS. Of the 215 samples, 120 samples had been concentrated prior to sequencing.

2.4.2. Detection of SNPs using MPS

All extracts from the degradation assessment (single donors 6, 7, 8, 9, 10, and 11 and mixtures of donors 6 + 7, 8 + 9, and 10 + 11) were subjected to sequencing analysis, with the exception of two single donor samples from 0 and 48 hour timepoints that failed to yield DNA by quantification. Unfed mosquitoes (negative control) or 5μL of reference blood from each donor (used for comparison with the profiles generated from the mosquitoes) were subjected to the same extraction and quantification processes as the direct extraction with blood-fed mosquitoes. Samples were used to prepare libraries using manufacturer’s protocols [27] for the Precision ID Identity Panel (ThermoFisher Scientific, Waltham, MA, USA), which targets 124 SNPs, including 43 Ken Kidd SNPs, 34 Y-clade SNPs and 48 SNP for ID SNPlex SNPs (Applied Biosystems). The Precision ID Library Kit, Precision ID IonCode™ Barcode Adapters 1–96 Kit (ThermoFisher Scientific) and Agencourt AMPure XP Beads (Beckman Coulter, Indianapolis, IN, USA) were used for manual library preparation of samples. The number of cycles used to amplify the target areas was adjusted based upon the concentration of input DNA: >0.15 ng, 23 cycles, 0.01 to 0.15 ng, 27 cycles, <0.01 ng, 30 cycles (ThermoFisher Scientific, personal communication). Sixteen low-quality samples from donors 8, 9 and 8 + 9 were subjected to automated library preparation using the Ion Chef™ System (ThermoFisher Scientific) and the Ion AmpliSeq™ Kit (ThermoFisher Scientific) according to manufacturer’s recommended protocols [27] with 29 cycles of amplification (ThermoFisher Scientific, personal communication).

Sample libraries were quantified using the Ion Library TaqMan™ Quantification Kit [28] and then templated onto an Ion 530™ Chip using the Ion Chef™ System (ThermoFisher Scientific). Sequencing was then performed with 30 pM of each library using the Ion 530™ Kit and Ion S5™ System using manufacturer’s protocols (ThermoFisher Scientific) [27]. Data analysis was performed using the Torrent Server, Torrent Suite Software and the HID SNP Genotyper plugin version 5.2.2 using a coverage threshold set to 6 reads (ThermoFisher Scientific).

2.5. Statistical analysis

All statistical analyses performed for the direct amplification (profile completeness, average RFU per locus and average heterozygote balance per locus) and MPS (profile completeness, average quality score, average coverage, and average heterozygote balance per locus) portions of the study were completed using RStudio statistical software (version 1.1.463) using the glm and ggplot functions. The statistical analysis performed for the degradation study was completed using RStudio but instead the geom_boxplot function was used for visualization of the data. Spearman’s correlations and generalized linear models were used for all statistical analyses performed in this study with the exception of the degradation study which used Pearson’s correlation. All p-values reported are unadjusted. The analyses performed for the direct amplification and MPS portions of the study sought to find the relationship between a variety of factors and the number of hours post-feed while also adjusting for cycle number.

Several calculations were performed in order to analyse the data. The completeness of the STR profiles was assessed by performing the following calculation:

$$\text{ Profile completeness }(\text{\%})=\frac{\text{ Number of alleles observed }}{\text{ Number of alleles expected }}\times 100$$

For single source samples, two alleles were expected per locus with the exception of the Y-loci. Homozygous loci were considered to contain two identical alleles. Because the Fusion 6C kit contains a total of 27 loci (23 autosomal loci, Amelogenin and 3 Y-loci), single source samples were expected to contain 51 alleles if a complete profile was obtained. Mixture samples were treated differently from single source samples. Amelogenin was expected to have only two alleles, the X and the Y. This was also true about the alleles detected at Y-loci as every donor in the study was determined to be male and therefore two male contributors were expected in mixture samples. All other loci were expected to have four alleles, providing a total of 100 expected alleles for the entire profile.

The total RFUs (GeneMarker software) for each locus were added and the average of these values was calculated in order to obtain an overall average RFU per locus for the profile. The heterozygote balance was calculated according to the following formula for each locus:

$$\text{ Heterozygote balance }=\frac{\text{RFU of smaller RFU allele }}{\text{RFU of larger RFU allele }}$$

This result produces a number between 0 and 1 that indicates how well the amplification performed (1 = equal amplification), and was applied only to single donor samples. Homozygous loci were omitted from the calculation, as well as three instances where heterozygous loci in the event that both alleles dropped out of the profile (one at timepoints at 4, 12, and 48 hours post-feed). For the remaining profiles, the average heterozygote balance was taken for all usable loci over the entire profile.

For the degradation study, the degradation ratio for each sample was calculated according to the following formula:

$$\text{ Degradation ratio }=\frac{\text{ Small autosomal concentration }}{\text{ Large autosomal concentration }}$$

Those samples that were highly degraded (“HD”) with a large autosomal concentration of 0 ng/μL, including the majority of samples from the 48 hour time point, were not used in the calculation of the linear regression model.

Profile completeness and average heterozygote balance per locus were also assessed for the MPS data using the same calculations as were used for the direct amplification data. For profile completeness, a total of 124 SNPs were expected for each profile to be considered complete. A SNP was considered to be correct in single source samples if it matched the SNP at the same SNP site in the reference samples. In mixtures, a SNP was considered to be correct if it matched the SNP at the same SNP site of either donor, or if the SNP was a mixture of the SNPs from the two donors. For heterozygote balance, the number of reads for each base of only heterozygous, biallelic SNPs were used for the calculation. MPS data was also accessed for its average coverage and average quality score, or genotype quality, of all 124 SNP sites for each sample.

2.6. Data availability statement

Here we have provided the comparisons of reference STR and MPS data to that of the samples isolated from mosquito midguts. Raw sequencing files are not provided in order to observe and preserve the donors’ right to privacy and to avoid the possibility of them being de-identified at any point.

3. Results

3.1. Direct Amplification

Based on results from the direct amplification experiments, it was first determined that the presence of the microFLOQ® swab in the reaction tube did not affect amplification compared to controls (Figures S1 and S2) as the expected electropherograms were obtained for each control in the experiment. The result of the initial direct amplification experiments showed that full STR profiles could be obtained from mosquitoes euthanized immediately following a blood meal (t = 0 hours post-feed), with partial STR profiles observed at 24 and 48 hours post-feed and no alleles detected at 72 hours post-feed. This finding revealed that at some point between 0 and 24 hours, the ability to obtain a complete profile with this set of mosquitoes was lost in the form of allelic dropout. In addition, while both methods of euthanasia were capable of producing complete STR profiles, the freezing method was more robust, as submersion in ethanol made isolation of individual mosquitoes more challenging and led to the release of some blood from the midgut that could potentially combine with blood from the midgut of other individual mosquitoes. Based upon these preliminary results, we proceeded with a direct amplification approach using mosquitoes collected by freezing for up to 48 hours and investigated the 0–24 hour time period further.

With the information gathered from the initial experiment, we assessed how long after a blood meal was taken that complete STR profiles could be obtained by direct amplification using single (donors 6,7, 8, 9, 10, and 11) and mixed (donors 6 + 7, 8 + 9, and 10 + 11) samples. We observed that at least one complete profile was obtained at the time intervals 0, 4, 8, 12, 16, 20 and 24 hours post-feed, and that each profile was concordant between and within samples, as well as with the donor-specific reference profile. Complete STR profiles could be obtained from individual mosquitoes collected at time points between 0–4 hours post-blood meal, with more instances of complete profiles being obtained between 0–20 hours. We did not observe one complete profile at every time interval for every donor. Of all of the replicates of each time interval per donor, 25% of all time intervals from 0–20 hours did not yield any full profiles for a single donor (1 of 6 time intervals for donor 6, 4 of 6 for donor 10 and 3 of 6 for donor 11) or mixture of two donors (1 of 6 time intervals for donors 6 + 7, 4 of 6 for donors 8 + 9 and 5 of 6 for donors 10 + 11). For a profile obtained from samples taken immediately after the blood meal was completed (Figure 1A), all expected alleles that correspond to the profile generated from the donor reference blood were present, including the Y-STR loci. This indicates that the mosquito fed on blood from a male individual. A representative complete STR profile from a single mosquito collected at 24 hours (Figure 1B) also demonstrated a “ski-slope” effect indicative of partial sample degradation (where smaller alleles amplified preferentially and were higher in RFU compared to the larger alleles). However, despite this partial degradation, it was still possible to detect all alleles (including the male-specific Y-STR loci) in this complete profile and thus could be used to definitively identify the donor.

Figure 1:

STR profiles can be obtained for blood consumed by mosquitoes using direct amplification. Mosquitoes that fed upon a single human blood source were collected (A) immediately following a blood meal (25 cycles) or (B) 24 hours later (27 cycles). (C) Alternatively, mosquitoes that had fed upon a mixed human blood source (50/50 from two donors) were collected at 20 hours post-feed (27 cycles). Complete and partial STR profiles were generated from individual, blood-fed mosquitoes using the PowerPlex® Fusion 6C amplification kit. Representative STR profiles are shown from two to six biological replicates with 1 mosquito sampled per replicate.

For mixtures of two sources of donor blood, we observed that complete STR profiles can be obtained from blood in the abdomens of mosquitoes out to 20 hours post-blood meal (Figure 1C). All expected alleles from both donors were observed, including two independent Y-STR loci indicating that both donors were male. As before, we observed partial sample degradation due to the presence of a ski-slope effect, and yet these profiles are sufficient to unambiguously determine donor identities.

Statistical analysis performed on the direct amplification data provided greater insight into the relationship between the post-feed time and various profile performance factors. We observed a negative linear relationship (Figure 2A, p=1.92×10^−13, Spearman’s correlation) when comparing the time point at which the sample was collected versus the completeness of the STR profile. For samples at 0 hours post-feed, the majority of samples exhibited profiles greater than 75% complete. As the amount of time post-feed increases, allelic dropout is more likely to occur. As a result, a greater spread in profile completeness values is observed for higher time intervals. Moreover, the average RFU per locus (Figure 2B, p=3.78×10^−9, Spearman’s correlation) and average heterozygote balance per locus (Figure 2C, p=2.71 × 10^−4, Spearman’s correlation) also were significantly impacted by the amount of time the sample was present in the mosquito midgut post-blood meal with negative linear relationships observed. Although the average RFU per locus shows variability among samples of the same time post-feed from 0–24 hours, most samples at the 48 hour post-feed time did not exhibit profiles over an average of 2500 RFU. The average heterozygote balance was determined to be at least 0.60 for the majority of samples in the 0–24 post-feed times. Lower values were observed at higher post-feed times, likely due to the impact of degradation on the samples causing the smaller alleles to be preferentially amplified over larger alleles. These relationships ultimately indicate that the DNA in the blood does degrade over time and this degradation will have an effect on the alleles observed, possibly causing heterozygous loci to appear homozygous or as if there is mixture of two donors present at different amounts.

Figure 2:

Linear regression plots of (A) STR profile completeness, (B) average RFU/locus, or (C) average heterozygote balance/locus against time post-feed adjusted for cycle number for direct amplification (STR) data.

3.2. Assessment of Sample Degradation

For both small and large autosomal concentrations, we observed that up to 10-fold more DNA could be obtained from the direct extraction approach compared to the cotton swab approach (Figures 3A, 3B, 3C and 3D). This was the expected result because all of the DNA from the blood in the mosquito is placed in the extraction tube, whereas using a cotton swab could allow for loss of DNA on the microscope slide or on the swab after incubation. In addition, the increase in incubation time may also have had an impact on the amount of DNA yield from the mosquito midgut. The small autosomal concentrations obtained for cotton swab extracted samples ranged from 0–0.6654 ng/μL with the highest concentration observed at 0 hours post-feed (Figure 3A). The large autosomal concentrations obtained for cotton swab extracted samples ranged from 0–0.7854 ng/μL. The highest concentration was observed at 0 hours post-feed for the same sample that had the highest small autosomal concentration (Figure 3B). The average concentration of DNA shows a general decreasing trend as the time post-feed increases with the greatest amount of variation among 0 and 4 hour post-feed samples for both small and autosomal concentrations (Figures 3A and 3B). Small and large autosomal concentrations for the direct extraction approach ranged from 0–6.1463 and 0–7.4390 ng/μL respectively. The highest concentration observed in each corresponds to the same 0 hour sample. The average small autosomal concentration of DNA observed from 0–12 hours is relatively consistent and drops beginning at 16 hours post-feed (Figure 3C). In contrast, the average large autosomal concentration of DNA shows a decreasing trend with increasing post-feed time with the 0, 4, 8 and 24 hour post-feed times exhibit the most variation amongst replicates (Figure 3D). Although the direct extraction method was able to produce higher overall DNA concentrations, the 0 hour time interval varied enough that even samples with minute DNA concentrations were observed. These results indicate that some variability exists in the concentration of DNA in the blood in the midgut of mosquitoes even when euthanized at the same number of hours post-feed. The exception to this was the 48 hour post-feed time, where the samples quantified close to 0 ng/μL regardless of extraction method.

Figure 3:

DNA concentrations quantified from extracted blood in mosquito midgut for 0–48 hours post-feed provided as box plots for (A) small autosomal concentrations from samples extracted using cotton swab approach, (B) large autosomal concentrations from samples extracted using cotton swab approach, (C) small autosomal concentrations from samples extracted using direct extraction approach and (D) large autosomal concentrations from samples extracted using direct extraction approach. Error bars denote standard deviation for 3 biological replicates with 1 mosquito sampled per replicate per donor.

Moreover, by using this data to calculate the degradation ratio, the extent of degradation over time post-blood meal was determined (Figures 4A and 4B). While the cotton swab extraction shows several highly degraded samples at various post-feed times (Figure 4A), we suspect that this finding is related more to the efficiency of the extraction technique as opposed to representing the actual extent of degradation in the mosquito midgut. For the direct extraction technique, although variability is still observed among samples at the same post-feed times, a general positive trend emerges (Figure 4B) indicating that as the amount of time post-feed increases, the degradation of DNA from blood in the mosquito midgut increases. This holds as the linear regression analysis of the data for the two experiments combined (Donors 6, 7, and 6+7 and Donors 8, 9, and 8+9) indicates that a weakly linear relationship (R^2 = 0.0688, Pearson’s correlation) exists between the degradation ratio and the time post-blood meal (Figure 4C) despite some variability observed.

Figure 4:

Assessment of DNA degradation in mosquito midguts with respect to time post-feed using (A) DNA extracted from cotton swabs or (B) direct extraction methods. (C) The degradation ratio (calculated by dividing the small autosomal concentration by the large autosomal concentration) over the time course is provided as box plots for 0–24 hours post-feed for DNA extracted from both cotton swabs and the direct extraction method. HD denotes highly degraded samples where the large autosomal concentration was 0. Error bars denote standard deviation for 3 biological replicates with 1 mosquito sampled per replicate per donor.

3.3. MPS to Determine Donor Identity

The results of the MPS analysis showed that DNA from the blood in the midgut of mosquitoes could be sequenced using the Ion Chef™ / Ion S5™ system. One representative sequencing coverage chart is shown from a sample that was isolated using cotton swabs 16 hours post-blood meal (Figure 5A, ThermoFisher Scientific HID SNP Genotyper plugin). The sequencing of reference samples that accompanied this set of samples yielded several SNP sites in which a genotype was not determined, as well as inconsistency between replicates of reference samples in terms of genotype. As a result, comparisons could not be made between the SNPs in the reference profiles and the SNPs from the DNA in the blood from the mosquito midgut.

Figure 5:

Sequencing coverage charts from individual mosquitoes fed on (A) a mixture of blood from two individual donors collected at 16 hours using the cotton swab extraction, (B) a single blood source collected at 48 hours using the direct extraction with sample concentration, and (C) a mixture of blood from two individual donors collected at 20 hours using direct extraction without sample concentration.

In contrast, blood isolated by direct extraction approaches were sufficient in terms of completeness and consistency to allow for the comparison of the reference SNPs to the SNPs observed in the blood from the mosquito midgut (Figures 5B and 5C). In general, the samples were higher in coverage compared to the DNA obtained from the blood collected on the cotton swab. The samples collected from individual mosquitoes that were allowed to feed on a single blood donor yielded sequencing results in which all 124 SNPs were concordant with the 124 SNPs of the corresponding donor’s reference sample even out to 48 hours post-blood meal for concentrated samples and 24 hours for non-concentrated samples. Moreover, mosquitoes that fed on a mixture of two donors also yielded robust sequencing results with all 124 concordant SNPs detected out to 24 hours post-blood meal for concentrated samples and 20 hours for non-concentrated samples. As with the direct amplification approach, there was noticeable variance in the number of concordant SNPs observed in samples collected at the same time point. However, direct extraction coupled with MPS approaches yielded a robust and conclusive identification of donor identity.

As seen with STR analyses, there was also a negative linear relationship between the number of hours after which the mosquito was euthanized post-feed and the completeness of the SNP profile (Figure 6A, p=0.00720, Spearman’s correlation). The majority of samples 0–24 hours post-feed had profiles that were greater than 90% complete. At 48 hours post-feed, greater variability was observed in terms of profile completeness with many samples under 60% complete. Moreover, we also observed a negative linear relationship between the number of hours after which the mosquito was euthanized post-feed and the average quality score of the sequencing data (Figure 6B, p=0.0243, Spearman’s correlation). The quality score performance factor took on a similar trend as profile completeness with most samples having a quality score greater than 75 until the 48 hour time interval. No significant effect was observed of time upon the average coverage (Figure 6C, p=0.985, Spearman’s correlation) or average heterozygote balance per locus of the profile (Figure 6D, p=0.297, Spearman’s correlation). Unlike in STRs, heterozygote balance for this MPS data is not as indicative of degradation.

Figure 6:

Linear regression plots of (A) sequencing-based profile completeness, (B) average quality score, (C) average sequencing coverage versus time post-feed as adjusted by cycle number, and (D) the average heterozygote balance per locus of the profile is provided using MPS data.

Finally, the use of MPS allows even greater confidence through the alignment of the sequencing results to a reference human genome (hg19) via the Torrent Suite Software (ThermoFisher Scientific). For the samples that were extracted using the cotton swab method, we detected that a substantial percentage of reads (41%) align to the human reference genome, but that over half of the reads did not (Figure 7). A similar result was also observed in direct extraction samples. By extracting the unaligned reads from the resulting BAM file using SAMtools (Genome Research Limited) and using NCBI BLAST, we identified that these reads aligned with the Indian strain of the Anopheles stephensi mosquito used in this study with 100% identity. Since these reads were unaligned to the human genome, we did not observe interference of mosquito DNA in the human DNA profiles. Together, this indicates that both human and mosquito identity can be confidently obtained using this approach to further support forensic investigations.

Figure 7:

Alignment of sequencing reads to the hg19 human genome assembly from samples extracted using a cotton swab (donors 6, 7, and 6 + 7).

4. Discussion

Expanding the scope of the types of samples that are available to forensic scientists can help to ensure justice for both the innocent and the victims. As mosquitoes are commonly found throughout the world, strongly prefer to feed on warm blood, and often remain close to their source of a blood meal after feeding, here we have explored the possibility of using of blood-fed anopheline mosquitoes for forensic investigations. As mosquitoes can take a blood meal from multiple sources in a short period of time, such as from both a victim and a suspect, or from multiple suspects, it was additionally important to explore whether donor identities could be determined from a mixed blood sample taken from mosquitoes. In this study, we demonstrate that both STR and MPS approaches can be used to confidently identify individuals from a mosquito blood meal. Importantly, these methods can determine the identity of single individuals or, in the case of STR analysis, deconvolute multiple individuals bitten by a mosquito when appropriate reference samples are available.

Like the previous studies that explored extraction of human DNA from the mosquito midgut prior to STR amplification [3, 5, 9, 11–15], full STR profiles concordant with the corresponding reference were able to be obtained using direct amplification. Compared to these previous studies, our finding that a full STR profile could be obtained up to 24 hours agreed with the result found previously by Rabelo et al. [13] but disagreed with the study by Curic et al., where full profiles were obtained up to 48 hours [5]. Although direct amplification may not have been successful in obtaining full profiles at the 48 hour time point, it was possible to obtain some profiles that were above 90% complete. Ultimately, direct amplification is a simple, time-saving technique that is able to produce results comparable to that obtained using extraction methods, making direct amplification a viable technique for STR analysis. In addition, profiles were also obtained using MPS analysis even for DNA that was contained in the live mosquito midgut for 48 hours prior to processing. Just as in two separate studies by Young et al. [22, 23], it was determined that even small amounts of potentially degraded DNA can be sequenced using this technology.

Here we have examined parameters that greatly influence the robustness of the profiles generated from these samples and provide recommendations for best practices for the adoption of mosquito blood meal sampling by the forensic and entomology communities.

First, we considered a typical forensic collection and sample processing scenario, along with the availability of resources, constraints and limitations that often are imposed upon it. As many forensic samples are collected using cotton swabs, we demonstrated that microFLOQ® swabs are effective to both collect blood samples from mosquitoes and to allow amplification of human DNA. Moreover, we demonstrated that leaving the swab in the sample tube does not influence sample preparation, thus streamlining the workflow further. This finding agreed with that previously documented by Sherier et al. and Chong et al. for direct amplification of DNA from bloodstains [18, 19].

Second, we considered two approaches for the collection of mosquitoes at a crime scene that would be amenable to current operations and resources. In field settings, mosquitoes can be collected through a modified handheld or backpack-based vacuum aspirator such as the Prokopak [29], which can be outfitted to collect mosquitoes directly into an ethanol-containing sample tube, or into a chamber that could be securely removed and placed into a standard −20°C freezer to kill the mosquito and preserve the sample. We find that while both approaches are effective, freezing of the mosquitoes allows for easier handling of individual mosquitoes and prevents the premature rupture of mosquito midgut contents that could cross-contaminate other individual mosquitoes.

Finally, frozen mosquitoes can also be readily stored at −20°C for processing at a later time, although we have not yet explored the effect of long-term cold storage of blood-fed mosquitoes upon the DNA samples that can be obtained from them. While our conclusion was that DNA profiles could be obtained from mosquitoes that remained frozen for at least one month and then analysed, additional studies are being pursued to determine the effect of longer periods of storage.

For forensic applications, it is important to know how wide the collection window can be and still obtain information that is admissible in a legal proceeding. As female mosquitoes feed upon and digest blood as a protein source for the production of eggs, it was anticipated that sample quality of the DNA in the blood would deteriorate over time. The implications of these findings for the forensic community are manifold. As mosquitoes strongly prefer drinking blood that is warmed to near body temperature (37°C), this would limit blood feeding to live and recently deceased individuals. When coupled with other forensic methods for determining the time of death, this information can help define the sequence and timing of events at a scene-of-interest. Similarly, as degradation of the blood meal and the DNA in it continues over time, the definition of when complete, partial, or no profiles can be obtained may help to further define the timing of events. We examined parameters such as the degradation ratio, profile completeness, average RFU per locus, and average heterozygote balance per locus, but conclude that the variability across samples did not allow for a robust diagnostic measurement that is predictive of the time that had elapsed since the blood meal was taken. However, exploration of other parameters in future work may do so. We observed variability in the completeness of the profiles obtained from individual mosquitoes given identical access and time to take a blood meal, and visibly we observed substantial variation in the amount of blood that individual mosquitoes consumed. We encourage sampling of multiple, blood-fed mosquitoes found at the scene-of-interest as the presence of additional people may be revealed that were not identifiable in a single mosquito. The concentration of extracted DNA can extend the useful window of sequencing-based applications. In our study, we found that meaningful sequencing profiles could be obtained out to 48 hours instead of just 20–24 hours without DNA concentration.

We therefore recommend several best practices for the forensic and entomology communities. Direct amplification using a commercial amplification kit is a simple, easy method to obtain STR profiles from blood in the midgut of a mosquito. While STR profiles are robust and are currently the method of choice in forensic cases, the advent of sequencing-based technologies provides even greater confidence and information for forensic applications. An important aspect of deploying these methods for forensic applications is to ensure that these extraction and amplification approaches are easy, rigorous, and reproducible. The use of the direct extraction and sequencing of DNA from mosquitoes collected by freezing outperforms swab-based extraction and sequencing and is easier to do. Moreover, as complete profiles can be obtained from individual mosquitoes, surveying all blood-fed mosquitoes at a location-of-interest can provide a wider view of which individuals were present. Finally, while we find that both STR and MPS approaches provide robust and confident identification of individuals, sequencing-based approaches (MPS) provide extended profiling windows with DNA concentration, additional information and greater confidence in assigning identity to available reference samples.

5. Conclusion

In this study we have demonstrated that blood found in Anopheles mosquitoes is an excellent forensic source for the identification of single or multiple individuals. We find that complete STR DNA profiles that are concordant with donor reference profiles can be best obtained using direct amplification from human blood ingested by Anopheles mosquitoes. Moreover, we show that the amount of time that has lapsed between when the blood meal is taken by the mosquito and when the sample is collected influences whether complete or partial STR profiles can be obtained. Finally, we show that the MPS approach also is an effective method to individualize single or multiple blood donors from mosquito blood meals. We find that while the MPS approach using SNPs does outperform STR analysis for degraded DNA in obtaining a complete profile, STR analysis remains superior for mixture interpretation. We conclude that either direct amplification or MPS approaches with blood taken from mosquitoes present at a crime scene can open a new avenue for identifying the individuals involved. In addition, these approaches can be used to aid in determination of the time when a victim or perpetrator may have been present at the crime scene.

Supplementary Material

1

Figure S1: Electropherogram obtained from 1.0 μL (10 ng) of 2800M control DNA on a microFLOQ® direct collection device (positive control).

Figure S2: Electropherogram obtained from microFLOQ® direct collection device with no DNA (negative control)