IgG antibody response against Anopheles salivary gland proteins in asymptomatic infections in Narino – Colombia.

Background: The humoral immune response against Anopheles salivary glands proteins in the vertebrate host can reflect the intensity of exposure to Anopheles bites and the risk of Plasmodium infection. In Colombia, the identification of exposure biomarkers is necessary due to the several Anopheles species circulating. The purpose of this study was to evaluate risk of malaria infection by measuring antibody responses against salivary glands extracts from An. (Nys.) albimanus and An. (Nys.) darlingi and also against the gSG6-P1 peptide of An. gambiae in people residing in a malaria endemic area in the Colombian Pacific coast. Methods: We eluted dried blood spots samples to measure the IgG antibodies against salivary gland extracts of An. (Nys.) albimanus strains STECLA (STE) and Cartagena (CTG) and An. (Nys.) darlingi and the gSG6-P1 peptide by ELISA in uninfected people and microscopic and submicroscopic Plasmodium carriers from the Colombia Pacific Coast. A multiple linear mixed regression model, Spearman correlation, and Mann-Whitney U-test were used to analyze IgG data. Results: Significant differences in specific IgG levels were detected between infected and uninfected groups for salivary glands extracts from An. (Nys.) albimanus and for gSG6-P1, also IgG response to CTG and gSG6-P1 peptide were positively associated with the IgG response to P. falciparum in the mixed model. Conclusion: The CTG and STE An. (Nys.) albimanus salivary glands extracts are a potential source of new Anopheles salivary biomarkers to identify exposure to the main malaria vector and to calculate risk of disease in the Colombian Pacific coast. Also, the gSG6-P1 peptide has the potential to quantify human exposure to the subgenus Anopheles vectors in the same area. Our recent studies revealed important differences in salivary content in arthropods collected in the field when compared to the same species maintained in a colony (34). Also, a previous study suggests the possibility of two An. (Nys.) albimanus lineages circulating two geographically distant regions of Colombia. Thus, we aimed to determine if risk of infection can be affected by the salivary content of mosquitoes from the same species but from different origin. So, we used a recently colonized strain (CTG) and a long term established laboratory colony (STE) each isolated from a distinct geographical region (Colombia and El Salvador) to account for potential changes in IgG responses based on salivary content. As our results indicate, the SGE from the CTG strain showed significant association with the Pf-MSP1 and not with the SGE from STE suggesting potential differences. Determination and confirmation of these differences are subject of further studies aimed to characterize salivary gland content of the two An. (Nys.) albimanus lineages circulating in Colombia and comparing those to An. (Nys.) albimanus isolates from other countries. This is important since the use of salivary antigens as vaccines for malaria are undergoing and characterization of the main immunogenic salivary proteins of the main vectors circulating in endemic areas are important for the success of such vaccine.

To design a proper vector control method, it is necessary to accurately determine human-vector interaction and the proportion of those vectors that are infected. Vectorial capacity (VC) and EIR are quantitative entomological indicators used to determine epidemiology of vector-borne diseases such as malaria. The VC is used as the measure of a mosquito population's proficiency to transmit an infectious agent to a susceptible population (14), while EIRs are useful to establish a direct estimation of transmission risk (15,16). In the case of malaria, the EIR is the gold standard for measuring transmission intensity. EIRs are based on the number of mosquitoes captured and the proportion of mosquitoes infected with Plasmodium (17). However, estimation of EIR is expensive and may be insufficient in areas of low or seasonal transmission (18,19). Human Landing Collection (HLC) is currently the only mosquito catching method that can directly measure the biting rates of humanseeking mosquitoes. Unfortunately, it is only applicable to mosquitoes seeking human adults and results are difficult to extrapolate to children or to pregnant women that are the most vulnerable to malaria (20). Furthermore, during HLC, the human bait is exposed to the diseases transmitted by the landing mosquitoes posing ethical concerns on implementation of this technique (21). As an alternative, catching traps such as the CDC (Center for Disease Control) light trap and the bed net traps have been developed and the data collected is useful in estimating vector populations when the studies are properly controlled. However, these trapping methods often differ in the number of hostseeking mosquito population sampled (22). Still, in spite the high number of mosquitoes captured on these studies (up to 12,000 specimens) a few mosquitoes (up to 4 specimens) were found positive for Plasmodium parasites even in their high abundance months (12,13). So, the question remains on how much is people being exposed to mosquito bites and acquiring the parasite. Thus, it is important to design alternative methods able to reflect the vector-human contact and complement the data collected by mosquito trapping methods.
Malaria is acquired when Plasmodium spp. sporozoites are injected into human skin through the bite of a female Anopheles mosquito along with the mosquito salivary proteins (23). Previous studies have 5 shown that a significant number of mosquito salivary proteins are immunogenic and able to induce antibody responses, mainly IgG isotype. These antibodies can reflect the intensity of human exposure to mosquito bites and represent good indicators of the risk of infection with Plasmodium spp. (24)(25)(26)(27)(28).
Thus, the use of salivary gland and saliva antigens has been previously validated as an indirect proxy to determine mosquito bite exposure. Significant higher IgG antibody levels against An. (Nys.) albimanus and An. (Nys.) darlingi salivary proteins have been observed in people with active malaria infection in Central and South America when compared to uninfected people living in the same region (24,29). A similar pattern has been observed in areas where An. (Cel.) gambiae and An. (Cel.) stephensi are among the most important vectors. A significant number of these studies were performed evaluating IgG responses against the An. gambiae salivary protein gSG6, a highly conserved protein among Anopheles species from the Subgenus Cellia and Anopheles (30). The peptide, gSG6-P1, was designed from the original An. gambiae gSG6 sequence. IgG responses specific to this salivary peptide has been validated as a biomarker of human exposure not only in Africa but also in Asia and South America (25,28,31). Although there are no known species of the subgenus Cellia in South America, the responses observed against the gSG6-P1 peptide could be hypothesized to result from the presence of mosquitoes belonging to the subgenus Anopheles such as An. (An.) pseudopunctipennis and An. (An. ) punctimacula and An. (An.) calderoni (32).
Consequently, it is necessary to characterize a broader panel of biomarkers able to identify the risk of disease more closely in areas with a great diversity of Anopheles mosquitoes. Our research group plans to identify exposure markers that include not only the primary malaria vectors but also markers for the majority of the circulating species playing an important role in malaria transmission in Latin America, even when these vectors species are in a smaller proportion. Since the use of salivary gland extract as antigen to indirectly measure exposure to mosquito species circulating in a region has been validated by several groups the main objective of this work was to measure IgG antibodies in humans living in an area where low-density P. falciparum infections are frequent. Thus, we explored if human IgG responses to Anopheles salivary gland extracts (SGE) are associated with low-density infections by P. falciparum and risk of disease we aimed to evaluate whether gSG6-P1 peptide continues as a useful marker to detect exposure in areas where mosquitoes from the sub-genus Anopheles are important vectors of malaria in Colombia.

Samples selection
The samples used in this study were collected as part of a longitudinal study in which the purpose was to evaluate the dynamic of submicroscopic Plasmodium infections in Colombia. No entomological data was collected during the time of our study (33).
To compare the vector exposure between infected and uninfected individuals, all positive P. falciparum samples were selected (n=63) from the 958 people that were enrolled in the main study.
All of these infections were afebrile (axillary temperature <37.5°C), and 48 (76.2%) were submicroscopic (detected by Loop-mediated isothermal amplification -LAMP or nested polymerase chain reaction-nPCR but not by light microscopy-LM). Furthermore, 50 uninfected samples were randomly selected by age (±5 years) and sex from the total of non-infected individuals by using an Excel random list. (Nys.) darlingi laboratory strain originated from Iquitos, Peru (35), and was maintained in the NAMRU-6 insectary (Iquitos, Loreto, Peru). Salivary glands from 8 to 10 days old female mosquitoes were extracted by dissection and pooled into 1X PBS (24). Mosquitoes were blood feed at day 3 or 4 after emergence. A pool of 100 salivary gland pairs from each strain was then frozen and thawed three times to prepare the SGE. The concentration of the SGE was determined using a NanoDrop™ (Thermo Scientific, Wilmington, DE, USA) and 50uL aliquots were stored at -80 o C until use. The An. gambiae gSG6-P1 peptide was synthesized by Genscript (Piscataway, NJ, USA) and the P. falciparum Pf-MSP (Plasmodium falciparum Merozoite Surface Protein) peptide (Fitzgerald, USA) was used to evaluate exposure to malaria parasites.

Indirect ELISA (Enzyme Linked Immunosorbent Assay)
ELISA conditions were standardized as described elsewhere (24,25). Also, DBS samples were (Blocking buffer) for 1.5 hours at 37°C. The DBS eluted was used to prepare a 1:50 sample dilution in blocking buffer, this optimal dilution had been determined by preliminary experiments and 50 µL of diluted samples were added to each well (individual samples were tested in duplicate). Plates were incubated at 37°C for 1.5 hours, washed three times, then incubated 1h at 37°C with 50 µL/well of a 1/1,000 dilution of goat monoclonal anti-human IgG conjugated with horseradish peroxidase (AbCam, Cambridge, MA). After three final washes, colorimetric development was carried out using tetramethyl-benzidine (Abcam) as a substrate. In parallel, each assessed microplate contained in 8 duplicate: a positive control, a negative control, and a blank; wells containing no sample. The positive control was a pool of DBS of people with positive malaria diagnosis. The negative control was a sample of people from US (n=36) with no exposure to malaria parasites. The blank was composed by wells containing no sample. The reaction was stopped with 0.25 N sulfuric acid, and the optical density (OD) was measured at 450 nm.

Statistical analysis
All data from questionnaires and forms were entered into a Microsoft Access database, and statistical analyses were conducted in STATA 14 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP) and GraphPad Software V5. OD normalization and plate to plate variation was performed as described elsewhere (25). Briefly, antibody levels were expressed as the ΔOD value: ΔOD = ODx − ODb, where ODx represents the mean of individual OD in both antigen wells and ODb the mean of the blank wells. For each tested peptide, positive controls of each plate were averaged and divided by the average of the ODx of the positive control for each plate to obtain a normalization factor for each plate as previously described. Each plate normalization factor was multiplied by plate sample ΔOD to obtain normalized ΔOD that were used in statistical analyses.
Assay variation of samples (inter and intra assay) tested in the study was below 20% and we only included in the analysis serum samples with a coefficient of variation ≤20% duplicates between duplicate (36). The mean ΔOD of negative US controls plus 3 standard deviations (SD) was used to determine cut-off value for responsiveness to antigens. The ΔOD cut off value to determine exposure to malaria antigens as 0.263. We estimated the median of antibody level for each antigen in Odd ratios (OR) were calculated to evaluate risk of malaria. For this, the median was used to classify IgG antibody levels as high (ΔOD higher than the median) and low (ΔOD equal or lower than the median) and the samples were classified as cases (Asymptomatic and submicroscopic infections) and controls (uninfected). In addition, Spearman correlation coefficients were calculated to measure the strength of association between each Anopheles antigen with Pf-MSP IgG levels. Finally, a Mann-Whitney U-test was used to estimate differences between medians of each Anopheles antigen by the status of infection in the whole sample and by sites and a Kruskal-Wallis test to estimate differences between groups of infection. A multiple linear mixed regression model was constructed to determine the correlation between anti-Anopheles IgG levels (anti-gSG6-P1, CTG, STE, and An. (Nys.) darlingi) with anti Pf-MSP IgG levels. A random intercept at the village level was introduced in the model to correct the inter-village variations. The model was adjusted by Plasmodium infection, age and time of residence in a malarial endemic area; these factors showed significant p values in simple models.

Study sample demographics, sociocultural variables and antibody responses to mosquito antigens
We studied exposure to mosquito bites in the area of Tumaco in Nariño (Colombia) (Figure 1). We also tested whether the difference observed in antibody level between infected and uninfected will be influenced by the village where samples were collected. Figure 3 shows the median of anti-

Detection of IgG antibody levels by P. falciparum detection threshold (microscopic vs. sub-microscopic)
All of our Plasmodium infected patients were afebrile and considered as asymptomatic carriers.
However, we grouped them according to the diagnostic test results into microscopic (if parasites were detected by LM and PCR) or submicroscopic if parasites were only detected by PCR (Figure 4).
Accordingly, our results showed that IgG levels might change according to parasitemia. Specifically, we observed a tendency of increased antibody levels in samples where parasitaemia was detected by light microscopy compared to infections only detected by molecular tests and also in uninfected specimens. There were significant differences in the median IgG antibody levels against CTG (Kruskal-Wallis test p =0.0016) and gSGS-P1 (Kruskal-Wallis test p value =0.0067) between the three groups of infections. Although the tendency was also observed when using STE and An. (Nys.) darlingi as antigen, the differences were not significant.

Association between exposure to Anopheles antigens and antibodies against
Plasmodium pf-MSP1 protein We evaluated whether there was any correlation between the level of IgG antibodies against the Pf-MSP1 protein and exposure to mosquito bite reflected by the levels of IgG antibodies against the salivary antigens. We observed a positive association between Pf-MSP IgG levels with anti CTG (Spearman r = 0.2722, p=0.0035) and gSG6-P1 peptide (Spearman r= 0.3872; p <0.000) (Figure 5) but not for An. (Nys.) darlingi and STE SGE.

Antibody-based model to evaluate factors of variation in responses against
Anopheles and Plasmodium antigens. there is a decreasing of IgG immune response with increased age ( Table 2). A similar situation occurred with the time of residence in an endemic area for malaria; IgG responses to gSG6-P1 peptide was 3.4% lower in samples from people who had lived in a malarial area for more than five years (RE= -0.035; 95% CI -0.070 to -0.003). Finally, no significant variation of specific anti-Anopheles IgG was observed according to the status of infection ( Table 2).

Discussion
The intensity of malaria transmission has been traditionally evaluated using the EIR, which is defined by the number of infected bites received per human per unit of time ; nevertheless, this strategy has shown limitations in low endemic settings for malaria (26,37  Interestingly, the relationship between parasitemia and IgG antibodies against Anopheles antigens was significant when using the antigen from the CTG strain and not the STE, suggesting that the antigens contained on the SGE from the CTG may be more closely related to the one the study subjects are exposed in the field. However, we did not find an association between antibodies against Previous studies suggest that An. (An.) calderoni is a primary malaria vector in Narino (12). This may 13 explain our current findings showing a high IgG response against gSG6-P1 peptide in samples from infected compared to uninfected people. These findings agree with our previous study in Colombian volunteers, where we found that the concentration of gSG6-P1 antibodies was significantly correlated with malaria infection status and that people with clinical malaria presented significantly higher levels of IgG anti-gSG6-P1 antibodies than healthy controls (25). Although, Anopheles species from the subgenus Nyssorhynchus are the main vectors of malaria in Colombia, at least six species from the sub-genus Anopheles have been described as potential malaria vectors in the region (38,39). Three of these species (An. This study has several limitations. First, because this study was cross-sectional, association with the anti-Anopheles IgG levels should be interpreted with caution as they do not imply causality. Second, due to the lack of a symptomatic group, we could not analyze risk factors for this kind of infection, and we could not explore the differences in the anti-Anopheles IgG levels between uninfected, asymptomatic (both, submicroscopic and microscopic infections) and symptomatic groups. Also, the lack of concurrent entomological data in our study is a significant limitation. Since this study did not

Consent for publication
Not applicable

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its supplementary information files. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.