Measuring antibody avidity to Plasmodium falciparum merozoite antigens using a multiplex immunoassay approach

Background Antibodies (Ab) play a significant role in immunity to Plasmodium falciparum malaria. Usually, following repeated exposure to pathogens, affinity maturation and clonal selection take place, resulting in increased antibody avidity. However, some studies suggest affinity maturation may not occur to malaria antigens in endemic areas. Information on development of antibody avidity is confusing and conflicting, in part, because different techniques have been used to measure avidity. Today, bead-based multiplex immunoassays (MIA) are routinely used to simultaneously quantitate antibody levels to multiple antigens. This study evaluated the feasibility of developing an avidity MIA with 5 merozoite antigens (AMA1, EBA-175, MSP1-42, MSP2, MSP3) that uses a single chaotropic concentration. Methods The most common ELISA protocols that used the chaotropic reagents guanidine HCl (GdHCl), urea, and ammonium thiocyanate (NH4SCN) were adapted to a multiplex MIA format. Then, different concentrations of chaotropes and incubation times were compared and results were expressed as an Avidity Index (AI), i.e., percentage of antibody remaining bound in the presence of chaotrope. Experiments were conducted to (i) identify the assay with the widest range of AI (discriminatory power), (ii) determine the amount of chaotrope needed to release 50% of bound Ab using plasma from adults and infants, and (iii) evaluate assay repeatability. Results Overall, 4 M GdHCl and 8 M urea were weaker chaotropes than 3 M NH4SCN. For example, they failed to release significant amounts of Ab bound to MSP1-42 in adult plasma samples; whereas, a range of AI values was obtained with NH4SCN. Titration of NH4SCN revealed that 2 M NH4SCN gave the widest range of AI for the 5 antigens. Binding studies using plasma from 40 adults and 57 1-year old infants in Cameroon showed that 2.1 M ± 0.32 (mean ± SD) NH4SCN (adults) and 1.8 M ± 0.23 M (infants) released 50% of bound Ab from the merozoite antigens. Conclusions An avidity MIA is feasible for the 5 merozoite antigens that uses a single concentration (2 M) of NH4SCN. The assay provides a simple method to quickly obtain information about Ab quantity and quality in the acquisition of immunity to malaria in endemic populations.

antibody avidity provides insight into the extent of somatic mutation of immunoglobulin hypervariable regions and subsequent clonal selection. Considering the importance of Ab in immunity to malaria, including to merozoite antigens, surprisingly little is known about antibody avidity in naturally-infected individuals. Early studies showed that Ab avidity to an extract of P. falciparum-infected erythrocytes increased after a few infections in a low transmission region [2,3]. However, in high transmission areas, most studies have found little or no increase in antibody avidity with age to merozoite antigens, including MSP1, MSP2, MSP3 and EBA-175 [4][5][6][7][8][9], although an increase with age for AMA1 has been reported [4,5]. Thus, there is much to be learned about the role of Ab avidity in immunity to malaria.
A variety of protocols have been used to study antibody avidity. Although a few investigators have employed plasma magnetic resonance or bilayer interferometry [4,7,10], most studies have used the ELISA format with different concentrations of chaotropic agents [2,3,5,6,8,9,[11][12][13][14][15][16][17]. In these assays, results are often expressed as an Avidity Index (AI), defined as the amount (e.g., O.D.) of Ab remaining bound to an antigen in the presence of a chaotropic agent divided by amount of Ab bound in its absence multiplied by 100. Studies of avidity to merozoite antigens with plasma from naturally-infected humans have used a variety of chaotropes, including 0.5 M, 2 M, 4 M, or 5 M guanidine HCl (GdHCl) [5,6,9,11]; 8 M urea [2,[14][15][16][17]; or 1 M or 2.4 M thiocyanate (SCN) [3,8]. Additional studies of other malarial antigens have used 3 M NH 4 SCN for VAR2CSA [18,19] and 1 M NH 4 SCN for volunteers immunized with the RTS/S vaccine [12,13]. A few studies, mainly laboratory-based, employed titration curves of urea or NH 4 SCN to determine amount of chaotrope needed to release 50% of bound Ab, usually in animal models vaccinated with malaria antigens [20,21]. Prior avidity studies have usually considered only 1 to 2 antigens, have not explained why the chaotrope was selected, or have compared results among various chaotropes for the same or multiple antigens. The use of different methodologies makes it impossible to compare results between studies.
Today, bead-based multiplex immunoassays (MIA) are commonly used to measure Ab to combinations of malarial antigens [22][23][24][25][26]. Currently, it is unclear if an avidity MIA with multiple malaria antigens can be developed that uses only one concentration of a chaotropic agent, since prior avidity studies of viral and bacterial antigens have found that different concentrations of salt were required to release 50% of bound Ab from different antigens [1,27].
This study explored the feasibility of an avidity MIA for 5 P. falciparum merozoite proteins (AMA1, EBA-175, MSP1-42, MSP2 and MSP3). The goal was to determine if accurate information could be obtained using a single concentration of only a single chaotrope. First, the most commonly used ELISA protocols for GdHCl, urea, and thiocyanate were adapted to the MIA format. These results allowed comparison of the amount of Ab released by the different chaotropes. Next, titration curves for GdHCl, urea and NH 4 SCN were examined; as well as, the length of incubation with different salt concentrations. Initial results revealed that 2 M NH 4 SCN consistently gave the widest range of AI values. Subsequent experiments determined (i) how 2 M NH 4 SCN compared with the amount of NH 4 SCN required to release 50% of Ab bound to the antigens, (ii) if the assay was appropriate for measuring avidity in infants with newly-developing malarial immunity as well as adults with high levels of immunity, and (iii) the reproducibility of the avidity MIA. Overall, the combined results lead to the conclusion that 2 M NH 4 SCN provides meaningful results in an avidity MIA for the 5 P. falciparum merozoite recombinant proteins evaluated.
The antibody levels were determined by combining 50 µl of plasma (diluted 1:300 with PBS plus 1% BSA, pH 7.2) with 50 µl of the 5-plex bead mixture (total plasma dilution of 1:600) in filter microtitre plates. Following incubation at room temperature on a rotary shaker for 1 h, beads were washed twice with PBS + 0.05% Tween 20 and once with 1%BSA in PBS. Then, 100 µl of secondary antibody (R-phycoerythrin-conjugated, Affini Pure F(abʹ) 2 fragment, Goat anti-human IgG Fc fragment specific, Jackson Immunoresearch, West Grove, PA) diluted to 2 µg/ml in PBS-1% BSA was added to each well and incubated in the dark on a rotary shaker for 1 h. Beads were washed, re-suspended in 100 µl PBS-1% BSA, and 85 µl of the microsphere suspension was analyzed using a Liquichip M100 reader (Qiagen, Valencia, CA). The reader was programmed to read a minimum of 100 beads per spectral address, DD Gate 7500-15,000. Results were expressed as median fluorescence intensity (MFI). The linear part of the Ab binding curve (MFI) was determined for each antigen and found to extend from ~ 1000 to > 20,000 MFI with r coefficients ranging between 0.913 and 0.995 (Additional file 1: Fig. S1).

Basic avidity MIA
The avidity MIA assay was performed as described above, with the inclusion of an additional step. After incubation with plasma, the bead-Ab complexes were washed and then incubated for 30 min with 100 µl of PBS (no salt) or specified concentrations of different chaotropes (with salt). After washing, 100 µl of R-phycoerythrin-conjugated goat-anti-human IgG (Fcγ specific) was added to beads for 1 h incubation. Percentage of high avidity Ab was calculated by the formula (AI = [MFI beads with salt]/[MFI beads with no salt] × 100). Since AI evaluate strength of Ab binding and MFI indicate the amount of Ab, the two measures should be independent. Thus, in investigating the feasibility of an avidity MIA, it was important to confirm independence of MFI and AI. Results showed that similar AI were obtained when different dilutions of the same plasma sample was used (Additional file 2: Fig. S2).

Plasma samples
A collection of de-identified, archival, plasma samples from malaria endemic regions of Cameroon with high transmission were used [29,30]. Samples included: plasma from 5 selected adults with known Ab levels for the antigens; 40 adults residing in Ngali II (pregnant women, non-pregnant women, males) who had been repeatedly infected with malaria and had developed high malarial immunity, and 57 12-month-old infants living in Ngali II who had just begun developing immunity to malaria [30]. Two pools of plasma were used as positive controls (PC), with PC consisting of plasma samples pooled from adults living in Ngali II village (a high malaria transmission area), Cameroon and positive control #2 (PC#2) including plasma pooled from pregnant women living in the same village [29]. Negative controls (NC), used to establish a cut-off for positivity, included plasma samples from US individuals who had never been exposed to malaria.

Developing the avidity MIA
The initial experiment adapted 3 ELISA protocols for measuring Ab avidity to the MIA format. The protocol employed 4 M GdHCl for 10 min [5,11], 8 M urea for 5-10 min [2] and 3 M NH 4 SCN for 30 min [18,19] (Fig. 1). In the second set of experiments, 8 M urea, and 1 M, 2 M and 3 M NH 4 SCN were compared using the following conditions: 50 µl of diluted plasma was incubated with 50 µl Ag-coupled bead mixture for 60 min; washed three times; beads were re-suspended in 100 µl of the chaotropic agent for 30 min; beads were washed and incubated with 100 µl of PE-anti-human IgG for 60 min; washed and examined using a MicroChip 100 as described above (Fig. 2). In the third experiment, 2 M, 4 M and 8 M GdHCl, 4 M, 8 M and 12 M urea, and 1.5 M, 3 M and 4.5 M NH 4 SCN were evaluated using the above protocol (Additional file 3: Fig. S3). In the fourth experiment, the above protocol was followed except beads were incubated for either 15 or 30 min with the respective chaotropic agents to evaluate the effect of time of exposure to various concentrations of chaotrope (Fig. 3). To determine the amount of NH 4 SCN needed to remove 50% of bound Ab from the merozoite antigens, titrations curves of 0 M, 1.5 M, 3 M and 4.5 M NH 4 SCN were constructed (as described in "Results" section) and the Forecast program was used to estimate the 50% concentration (results in Table 1). Finally, to evaluate the repeatability of the assay over time, archival data from 13 avidity assays, conducted over a 30-day period using MSP1-42, MSP2 and MSP3, were evaluated (Fig. 4).

Statistical analysis
Antibody values were recorded as median fluorescence intensity (MFI). The avidity Index (AI) was calculated as the MFI of the sample treated with chaotrope divided by the MFI in the absence of chaotrope times 100. Results from replicate assays were expressed as means and standard deviations (SD). Median MFI and AI values for adults and infants were reported (not normally distributed). The estimated amount of salt needed to remove 50% of bound Ab was determined using the Forecast Program (Excel).

Comparison of the three different chaotropes
In the initial experiment, ELISA protocols for 4 M GdHCl, 8 M urea and 3 M NH 4 SCN were adapted to the MIA format (see "Methods" section for details). The avidity MIA included the malaria antigen 5-plex, the PC and 5 selected plasma samples with high Ab levels ( Fig. 1). The PC was pooled from multiple adults and represents Ab in the "general population;" whereas, the 5 individual samples provided information about variation among different adults. As expected, a pattern similar to that with PC was observed among the 5 individuals, but substantial variation was also found. The 5 adults had different proportions of high avidity Ab (i.e., AI) for each of the antigens. For example, after 3 NH 4 SCN treatment, person #1 had 92% high avidity Ab to MSP1-42, 87% to EBA-175, but only 5% to MSP3. Thus, it was possible for a person to have a high AI for one antigen and relatively fewer high avidity Ab to another antigen. Likewise, the AIs for MSP3 were generally lower than for the other antigens, except for individual #2 who had AIs over 80 for MSP3 (Fig. 1e). Thus overall, AI for some antigens were higher than for others. In summary, the PC reflected the effect of combining plasma samples with different proportion   Figs. 1 and 2 show that the majority of mean AI have overlapping SD; however, a few differences are seen, especially with 8 M urea for AMA1 and MSP3 that are due to inter-assay variation of high avidity Ab, i.e., it did not have the highest AIs, but rather the average of the combined samples. Additionally, significant variation in AIs was observed among individuals who had been exposed malaria all their lives. Because of the relatively strong effect of 3 M NH 4 SCN compared to GdHCl and 8 M urea, a second experiment was conducted using the same plasma samples treated with 8 M urea and a titration of NH 4 SCN (1 M, 2 M and 3 M) (Fig. 2). In general, the effectiveness of 1 M NH 4  To further explore the influence of chaotrope concentration, one-half and twice the molar concentrations of 4 M GdHCl, 8 M urea and 3 M NH 4 SCN were evaluated on MSP1, MSP2 and MSP3 as test antigens (Additional file 3: Fig. S3). NH 4 SCN was the only chaotrope that dislodged Ab bound to MSP1, but at 3 M and 4.5 M released almost all Ab bound to MSP2 and MSP3. Urea remained the least stringent chaotrope, failing to release Ab from MSP1 at any concentration, and provided a weak doseresponse with MSP2 and MSP3 at 8 M and 12 M (Additional file 3: Fig. S3). In contrast, 4 M and 8 M GdHCl appeared to release almost all Ab bound to MSP2 and MSP3, but only 10-20% of Ab bound to MSP1 (Additional file 3: Fig. S3). Overall, none of the concentrations of the 3 chaotropes improved the range of AI compared to 2 M NH 4 SCN (Fig. 2).

Timing of incubation with chaotropes
Previous studies used various lengths of time for incubating antigen-Ab complexes with chaotropes, with some protocols simply washing bound antigen-Ab complexes with the salt solution whereas others incubated for 10, 15 or 30 min. The most common lengths were 10-15 and 30 min. In this experiment, Ab-antigen-bead complexes were incubated for 15 and 30 min with either 4 M GdHCl, 8 M urea or 2 M NH 4 SCN (Fig. 3). When directly compared in this experiment, results showed that AIs were only slightly, but not significantly, lower after 30 compared to 15 min of incubation regardless of the chaotrope. In subsequent experiments, a 30-min incubation period was used for convenience when conducting large multiple assays.  (Table 1). Overall, 90-100% of adults had Ab to AMA1, EBA-175, MSP1 and MSP2, but only 68% had Ab to MSP3. Similar median Ab levels (MFI) were detected to AMA1 and EBA-175 (16,799 and 13,815 MFI); equivalent median Ab levels to MSP1 and MSP2 (9054 and 9177); and lower levels of Ab to MSP3 (2066 MFI). The molar concentration estimated to obtain AI of 50 ranged from 1.7 to 2.3 M (Table 1), with a mean of 2.1 M ± 0.32 (mean ± SD) for adults living in a malaria-endemic area. Thus, 2 M NH 4 SCN appeared to release about half of the Ab bound to the 5 antigens, regardless of Ab levels.
The above plasma samples were from adults who had been exposed to P. falciparum throughout their lives and, therefore, most likely had acquired a mature malarial humoral response over time. The question then became, would 2 M NH 4 SCN be appropriate for infants who were just beginning to acquire immunity? Therefore, 57 plasma sample were screened from infants who were 1 year of age (mean ± SD: 362 ± 12 days; range: 326-385 days) living in Ngali II, who had been followed since birth and had begun to make their own Ab after the decline of maternal antibodies. As expected, the majority of infants had Ab to AMA1 (72%), MSP1-42 (93%) and MSP2 (88%); with fewer infants having Ab to EBA-175 (49%) and MSP3 (37%). Ab levels were much lower in the infants to all of the antigens compared to adults (Table 1), but the amount of NH 4 SCN needed to release 50% of bound Ab to the antigens was slightly lower, ranging from 1.6 to 2.1 M (compared to 1.7 to 2.3 in adults; Table 1), with a mean of 1.8 M ± 0.23 M. Although lower in young children, assay conditions to release 50% of bound Ab did not differ drastically between infants and adults.

Reproducibility of the avidity MIA
In a separate study on Ab avidity, MSP1-42, MSP2 and MSP3 were included as antigens on 13 different plates run over a 30-day period. The protocol included incubating antigen-Ab for 60 min, then 30 min with 2 M NH 4 SCN, followed by 60 min with secondary Ab. The archival data provided information on day-to-day and plate-to-plate variation of the avidity MIA assay. Figure 4 shows the variation in MFI and AI (mean ± SD) for 13 replicate assays. Coefficient of Variation (CoV) for each of the AIs was smaller than the corresponding CoV for MFI, e.g., the MFI CoV for MSP1-42 with PC#2 was 32.2%; whereas, the AI CoV was 6.3% (Fig. 4a, b). A few MFI value were outside of the mean ± 1 SD, but the corresponding AI remained within, demonstrating the independence of the two measures. Direct linear relationships with higher MFI (r = 0.898) and AI (r = 0.936) values and smaller CoV were detected (Fig. 4g, h). Thus, repeating the avidity MIA using 2 M NH 4 SCN for 30 min resulted in similar avidity indexes.

Discussion
Considering the importance of Ab in immunity to malaria and the extensive search for correlates for protection, it is surprising that Ab avidity has received relatively little attention. The exclusion method used in this study is a simple approach for measuring Ab avidity, whereby Ab-antigen complexes are treated with a chaotrope and the amount of Ab release is determined. Although the exact mechanism is unknown, it is hypothesized that the gap between antigens and bound Ab with good complementarity is tight enough to exclude the denaturing agent, unlike the gap in low avidity Abs [1]. As such, hydrophobic and ionic bonds stabilizing the complexes remain unbroken [31][32][33]. Since merozoite antigens have multiple epitopes, each of which induces Ab with different binding-strengths (affinity), AI represent the overall percentage of Ab with sufficient complementarity to remain bound after exposure to a specific chaotrope. Accordingly, the term "high avidity antibodies" is defined by the concentration of chaotrope used. For example, in Fig. 1a, plasma from Individual #3 would be described as having 82% and 75% high avidity Ab to AMA1 if 4 M GdHCl and 8 M were used, but only 8% high avidity to AMA1 in assays using 3 M NH 4 SCN. Likewise, in Fig. 2a, Individual #3 would be reported as having 95%, 30% and 8% high avidity Ab to AMA1 if 1 M, 2 M, or 3 M NH 4 SCN were used. Since "High Avidity Ab" is defined by the concentration of chaotrope, it is difficult to compare results between studies and interpret results when different protocols are used. Today, bead-based MIA are commonly used to measure Ab levels to malarial antigens [22][23][24][25][26]. The MIA format has many advantages, including the requirement for small amounts of antigen, speed, assaying > 100 replicates (beads) instead of only a few wells, and being internally controlled since all antigens and reagents are present within the same well (e.g., if Ab are not detected for one antigen, but are to others, then absence of Ab is not due to a technical error). For these reasons, having an avidity MIA using the exclusion method would allow researchers to quickly obtain quantitative data about Ab avidity. Based on the initial results, however, it was not clear if an avidity MIA using the 5 antigens and a single chaotrope was feasible.
Follow-up experiments sought to further characterize and refine the bead-based MIA using different concentrations of chaotropes ( Fig. 2 and Additional file 3: Fig.  S3) and different incubation periods (Fig. 3). Overall, 2 M NH 4 SCN provided the widest range of AI for the 5 antigens. Since using a concentration of chaotrope that is in the center of the exclusion curve is desirable, 2 M NH 4 SCN was compared with the actual molar amount of NH 4 SCN needed to release 50% of bound Ab to each of the 5 antigens (Table 1). Using plasma from 40 adults living in a highly endemic malaria region, mean molar concentrations of NH 4 SCN to release 50% of bound Ab ranged from 1.7 to 2.3 M, with an average of 2.1 ± 0.32 M for the 5 antigens (Table 1). Thus, 2 M NH 4 SCN proved to be in the middle of the exclusion curves. Incubation with antigen-Ab complexes with 2 M NH 4 SCN for 15 or 30 min didn't alter AIs significantly (Fig. 3), but the longer incubation period proved to be practical when running multiple plates. Using the avidity MIA, over 100 samples can be screened against the 5 or more antigens in a single afternoon with good reproducibility (Fig. 4). Since the assay provides information on both the amount (MFI) and the proportion of high avidity Ab (AI), data on both Ab quantity and quality is obtained in a single experiment. Without further experimentation, it is unclear if the protocol can be used for other malarial antigens, but results from this study provide a starting point for development of future avidity assays for other antigens.
In repeat experiments, AI values were more consistent than MFI (Fig. 4). In other words, AIs are technically more error-proof than raw MFIs. Technical and instrumental differences, for example due to pipetting errors or mis-calibration, result in variation of MFI when the same sample is used in multiple assays. However, the AI values are less affected since they measure the proportion of high binding antibodies. The wide dynamic range of MIA (e.g., 500 MFI to 25,000 MFI in this study) helps explain the very strong association between AI and the coefficient of variation (r = 0.935) (Fig. 4). Thus, these data support the feasibility of a simple, useful, repeatable avidity MIA for merozoite antigens.
Initially, it was hoped that by comparing the most commonly-used protocols in the same experiment, one might gain insight to help make comparisons and interpret data from previous studies. A conscientious search of the literature revealed that the 4 studies using 4 M GdHCl focused on merozoite antigens, including AMA1 [5,6,9,11], MSP1 [5,6,9,11], MSP2 [9], MSP3 [5]. Whereas, the 5 studies that used 8 M urea measured avidity to a schizont extract [2], Plasmodium vivax MSP1 and P. vivax Duffy Binding Protein [14,17], and Ab from individuals vaccinated with AMA1 and Pf25 [15,16]. In contrast, the 6 studies using SCN evaluated other malarial antigens including: a schizont-extract and 1 M NH 4 SCN [3], EBA-175 using 2.4 M NaSCN [8], the RTS/S vaccine and 1 M NH 4 SCN [12,13], and VAR2CSA with 3 M NH 4 SCN [18,19]. Thus, it does not appear that any of these prior studies are similar enough to be directly compared with each other. Accordingly, comparisons can be made within the same study among different cohorts or treatment groups, but direct comparison between studies remains unfeasible. Variation in methodologies and antigens used in previous studies may explain why an inconsistent picture of changes in Ab avidity with age and the role of Ab avidity in protection from malaria exist.
Results from this study provide hints about maturation of the Ab response in individuals living in malaria endemic areas. First, individuals may have high AI to one antigen but low AI to another antigen. For example, in Fig. 2, using 2 M NH 4 SCN, Individual #1 AI of 90 to MSP1-42, 85 to EBA-175, 25 to AMA1, but only 8% to MSP3. This result indicates that affinity maturation does not occur at the same rate (or reach the same level), for all antigens. Second, at the population level, Ab avidity tends to be greater for some antigens than others. For example, in Fig. 1, AI to MSP1-42 were high (> 90 with 3 M NH 4 SCN), but low AI to MSP2 and MSP3 (AI < 15 for 3 M NH 4 SCN). Thus, some antigens appear to induce affinity maturation better than others. Third, the amount of Ab was lower in infants than adults for the 5 merozoite antigens, requiring the use of 1:100 and 1:1000 dilutions of infant and adult plasma, respectively, for MFI to fall on the linear part of the binding curve. Interestingly, the amount of NH 4 SCN needed to release 50% of bound Ab was quite similar for Ab-positive infants and adults (Table 1), although lower amounts of chaotrope were required for infants for AMA1 and EBA-175, slightly lower for MSP1-42, but similar for MSP2 and MSP3 (Table 1). Studies in high transmission areas have reported little or no increase in Ab avidity with age for EBA-175, MSP1, MSP2, and MSP3 [4][5][6][7][8][9]. Plasma used in this study was from infants and adults living in a rural village with perennial transmission who received an estimated 257 infectious mosquito bites/person/year [29,30]. Clearly, it would be interesting to investigate affinity maturation with age in this high transmission setting to determine if and when affinity maturation occurs. Overall, the use of 2 M NH 4 SCN in future studies evaluating maturation of immunity from infancy to adulthood seems appropriate; whereas, investigators studying acquisition of immunity in young children might consider using 1.5 M NH 4 SCN as the lower concentration will give a wider range of AI values.

Conclusions
A multiplex bead-based avidity immunoassay is feasible for the merozoite antigens AMA1, EBA-175, MSP1-42, MSP2 and MSP3 that employs 2 M of NH 4 SCN. Further studies are needed to determine if 2 M NH 4 SCN is the best concentration of chaotrope for other malarial antigens. The assay is simple, measures avidity near the middle of the dynamic range, and AI are similar in repeated experiments. The assay provides a simple method to quickly obtain information about Ab quantity and quality in the acquisition of immunity to malaria in endemic populations.