Discovery of Specific Antigens That Can Predict Microfilarial Intensity in Loa loa Infection

ABSTRACT

Antigen-based immunoassays are currently needed for point-of-care quantification of Loa loa microfilariae (mf). Coupling transcriptomic approaches with bioinformatic analysis, we have identified 11 specific putative proteins (coding mRNAs) with potential utility as biomarkers of patent (mf⁺) L. loa infection. We successfully developed antigen capture immunoassays to quantify 2 (LOAG_14221 and LOAG_15846) of these proteins in individual plasma/serum samples. Of the 2 quantifiable circulating biomarkers, LOAG_14221 showed the highest degree of specificity, particularly with a monoclonal antibody-based immunoassay. Moreover, the levels of LOAG_14221 in L. loa mf⁺ patients were positively correlated to the mf densities in the corresponding blood samples (r = 0.53 and P = 0.008 for polyclonal assay; r = 0.54 and P = 0.004 for monoclonal assay). Thus, LOAG_14221 is a very promising biomarker that will be exploited in a quantitative point-of-care immunoassay for determination of L. loa mf densities.

INTRODUCTION

Loiasis, caused by the filarial parasite Loa loa, affects ∼14 million people in Central and West Africa (1). Because the majority of infections are clinically asymptomatic and the symptomatic infections are rarely life-threatening, loiasis has been largely neglected (2). However, L. loa infection has gained importance over the past 20 years because of its negative impact on the elimination programs for onchocerciasis and lymphatic filariasis. Indeed, individuals with very high L. loa microfilariae (mf) levels can develop serious adverse events (SAEs), most notably irreversible neurological conditions (mainly encephalopathies) that can lead to coma or even death, following administration of ivermectin (IVM) during mass drug administration (MDA) campaigns (3, 4). As a result, IVM-based MDA campaigns have been interrupted or delayed in areas of Central Africa where L. loa is coendemic with either Wuchereria bancrofti or Onchocerca volvulus (5).

Diagnostic methods that can accurately identify individuals that are at high risk of developing SAEs during IVM-based treatment are thus needed to achieve lymphatic filariasis and onchocerciasis elimination goals by 2020/2025 as targeted by the WHO. With the recent findings of an association between very high L. loa mf loads and mortality, independent of the effect of IVM treatment (6), identifying those at risk becomes all the more important.

Traditional methods of L. loa mf identification and quantification are based on the microscopic examination of midday blood samples (7), a tedious and sometimes inaccurate process that is neither point of care (POC) nor high throughput (8). Real-time quantitative PCR (qPCR) and loop-mediated isothermal amplification (LAMP) methods are credible alternatives to microscopy since they are high throughput and combine a high degree of sensitivity and specificity with the ability to accurately quantify L. loa mf levels (9–11). However, they require a well-equipped laboratory (for qPCR), relatively expensive reagents, and time-intensive DNA/RNA extraction processes. More recently, the CellScope Loa (or LoaScope), a mobile phone-based video microscopy system, has been described to identify individuals at high risk of developing SAEs (12) and is being used for the rapid and very accurate counting of L. loa mf at the POC. However, such a device has not yet been commercialized. We have recently developed an antigen capture immunoassay that is also capable of quantitating microfilaria-derived antigen(s) (13); however, it has not yet been developed as a POC tool because of some constraints (time-consuming protein expression and expensive reagents and equipment) associated with the luciferase immunoprecipitation system (LIPS) technology used.

With the recent advances in genomics, a variety of pathogens, including the filarial parasites Brugia malayi (14), O. volvulus (15), and L. loa (16, 17), have been fully sequenced. It has, therefore, become possible to describe the transcriptomes and proteomes of different stages of these filarial helminth parasites (18). When coupled with newer bioinformatic tools, they lead to the relatively rapid identification of potential vaccine, drug, and biomarker candidates (18). The present study aimed to identify new biomarkers through transcriptomics and bioinformatics that can be the basis of an antigen capture immunoassay for the detection and quantification of L. loa mf at the POC.

RESULTS

Biomarker candidates for immunoassays.

A total of 12,277 mRNAs (of 15,444 open reading frames predicted) were identified in transcriptome sequencing (RNA-seq) analyses of mf derived from L. loa mf (16). Filtering the data set for putative proteins with no or little sequence homology with human, W. bancrofti, B. malayi, and O. volvulus proteins and with significant SecretomeP scores (>0.6) resulted in the identification of 11 L. loa mf-specific proteins. All of the Loa mf-specific proteins were annotated as “hypothetical proteins” with variable expression levels in mf (fragments per kilobase per million [FPKM] ranging from <1 to 3,877) (Table 1).

TABLE 1.

Details of specific selected proteins specifically identified in the transcriptome of Loa loa microfilariae

Proteina	No. of amino acids	SecretomeP score	FPKM levelb	Homology base E value
				Human	O. volvulus	B. malayi	W. bancrofti
LOAG_14221	66	0.79	0				3
LOAG_15846	56	0.63	9.7
LOAG_14347	119	0.77	1.2	2.00E-09			5.3
LOAG_11259	106	0.81	1.3				0.14
LOAG_03292	93	0.78	24.9				2.3
LOAG_03165	49	0.75	3,877
LOAG_03646	115	0.82	0.8
LOAG_00816	109	0.85	18.1				0.031
LOAG_02895	83	0.64	499.3	0.18
LOAG_03624	49	0.64	29.6	7.1
LOAG_11280	125	0.71	0			7.5

^aBolded proteins were expressed in recombinant form with at least 75% purity.

^bFKPM data were obtained from Desjardins et al. (16).

Development of polyclonal immunoassays.

Five of the 11 selected proteins highlighted in bold in Table 1 were expressed in recombinant form with at least 75% purity; the other six failed at either the expression or purification level. Highly reactive specific polyclonal IgGs were obtained for 4 of the 5 recombinant proteins (see Fig. S1 in the supplemental material), LOAG_14221, LOAG_11259, LOAG_15846, and LOAG_03292, while immunization with LOAG_14347 in rabbits was unsuccessful (Fig. S1E). Capture enzyme-linked immunosorbent assays (ELISAs) were then developed for each of the four proteins using monospecific polyclonal IgG on both sides of the assays. As proof of principle, results show that all 4 proteins were detected in these sandwich ELISAs when either phosphate-buffered saline (PBS) (Fig. 1A) or human AB serum (Fig. 1B) was spiked with the appropriate recombinant protein over a range of concentrations. The analytical sensitivities of these ELISAs (detection limit) were much better for LOAG_14221 (10 ng/ml) and LOAG_15846 (50 ng/ml) compared to those of LOAG_03292 (100 ng/ml) and LOAG_11259 (500 ng/ml). In addition, higher signals with a larger dynamic range were observed in LOAG_14221 and LOAG_15846 assays. Moreover, both LOAG_14221 and LOAG_15846 proteins could be detected in Loa mf-infected sera, with the detected signals significantly increased when sera were pretreated with glycine (see Fig. S2 in the supplemental material).

FIG 1.

Development of capture ELISAs for L. loa microfilaria proteins. Serial dilutions (spiked with 0.005 μg/ml to 5 μg/ml) of each of the recombinant proteins (LOAG_14221 [red], LOAG_15846 [blue], LOAG_03292 [purple], and LOAG_11259 [green]) in PBS (A) and in human AB serum (B) were tested with their corresponding polyclonal antisera. The seroreactivities of the polyclonal sera are plotted with the protein concentrations (μg/ml) on the x axes and net optical densities (ΔO.D.) on the y axes.

Sensitivity and specificity of polyclonal immunoassays.

Using pooled human AB serum samples spiked with increasing concentrations of the appropriate antigen, we generated standard curves (see Fig. 2A and B) to estimate the levels of circulating LOAG_14221 and LOAG_15846 in glycine-pretreated sera. As can be seen, 14 of the 25 tested L. loa mf-positive (mf⁺) samples had detectible levels (range, 20 to 642 ng/ml) of LOAG_14221, which was detected in only 1 of 8 L. loa mf-negative (mf⁻), 4 of 15 O. volvulus mf⁺, 5 of 15 W. bancrofti mf⁺, and 0 of 13 tested uninfected serum samples (Fig. 2C). LOAG_15846 quantities were obtained for all 25 tested L. loa mf⁺ (range, 11 to 490 ng/ml), 5 of 8 L. loa mf⁻, 12 of 15 O. volvulus mf⁺, 5 of 15 W. bancrofti mf⁺, and 0 of 13 tested uninfected serum samples (Fig. 2D). Using receiver operating characteristic (ROC) analyses to determine a cutoff to give 100% specificity (95% confidence interval [CI], 96.38% to 100%), the LOAG_14221 assay sensitivity was 36% (9/25; 95% CI, 17.97% to 57.48%) of L. loa mf⁺ individuals while LOAG_15846 picked up 24% (6/25; 95% CI, 9.36% to 45.13%) of them.

FIG 2.

Sensitivity of LOAG_14221 and LOAG_15846 polyclonal assays and correlation between serum protein amounts and corresponding mf blood levels. The net optical densities as a function of spiked recombinant protein in human AB serum for LOAG_14221 (A) and LOAG_15846 (B) were used as standard curves to determine the quantities of LOAG_14221 (C) and LOAG_15846 (D) in sera for L. loa mf⁺ (red), L. loa mf⁻ (black), W. bancrofti-infected (green), and O. volvulus-infected (blue), and uninfected (purple) individuals. Each dot represents an individual serum sample. Correlation between the quantities of detected LOAG_14221 (E) and LOAG_15846 (F) by ELISA and the corresponding microscopic L. loa mf blood counts are shown. Undetermined amounts of protein (amounts below the detection limit = 10 ng/ml) were considered to be 10 ng/ml.

Correlation between the mf counts and amounts of proteins estimated by polyclonal assays.

The relative amounts of LOAG_14221 antigen estimated in the polyclonal-based assay in glycine-pretreated serum samples of L. loa-infected individuals ranged from 10 (assay detection limit) to 643 ng/ml and significantly correlated to the corresponding counts of L. loa mf (range, <10 to 7,740 mf/ml of blood) in the peripheral blood (Fig. 2E; r = 0.53 and P = 0.008). In contrast, there was no significant correlation between the amounts of LOAG_15846 estimated in the polyclonal-based assay in the same L. loa mf⁺ glycine-pretreated sera (range, 11 to 490 ng/ml) and the blood counts of L. loa mf (Fig. 2F; r = −0.04 and P = 0.85). There was also no correlation between estimated protein concentrations in sera and the counts of mf in blood in either assay when sera were not pretreated with glycine (see Fig. S3 in the supplemental material), suggesting that both antigens are found complexed to antibody in the human circulation.

Development of monoclonal-based assays.

Because the LOAG_14221 polyclonal ELISA appeared to be predictive of L. loa mf blood counts (Fig. 2E) with only a small bit of nonspecific reactivity to non-Loa filaria-infected sera (Fig. 2C), we raised monoclonal antibodies to LOAG_14221 for better assay performance. Five monoclonal antibodies (IgG) were chosen based on their specificity and reactivity to the LOAG_14221 recombinant protein. One of these monoclonal antibodies, named 4B10C8, was shown to specifically and strongly bind the LOAG_14221 protein (data not shown).

Sensitivity and specificity of monoclonal assays.

Using 4B10C8 monoclonal capture IgG, LOAG_14221 was captured in serum/plasma samples of 20 L. loa mf⁺ (n = 28), 1 L. loa mf⁻ (n = 8), 4 O. volvulus mf⁺ (n = 26), and 3 W. bancrofti mf⁺ (n = 28) individuals and 1 uninfected (n = 107) individual (Fig. 3A). Like for the polyclonal-based assays, using ROC analyses to determine a cutoff giving 100% specificity, the 4B10C8 assay sensitivity was 67.9% (19/28; 95% CI, 47.6% to 84.1%) of L. loa mf⁺ individuals.

FIG 3.

Sensitivity of the LOAG_14221 monoclonal assay and correlation between serum LOAG_14221 amounts and mf levels. LOAG_14221 levels were assessed for serum from L. loa mf⁺ (red), L. loa mf⁻ (black), W. bancrofti-infected (green), O. volvulus-infected (blue), and uninfected (purple) individuals using capture monoclonal antibody 4B10C8 IgG (A). Spearman correlation between the LOAG_14221 quantities detected with 4B10C8 IgG (B) in individual serum samples and L. loa mf count is also represented. Undetermined amounts of protein (amounts below the detection limit = 10 ng/ml) were considered to be 10 ng/ml.

As shown with the polyclonal assay, LOAG_14221 concentrations (range, <10 to 1,105 ng/ml) in L. loa mf⁺ sera captured by monoclonal 4B10C8 IgG also significantly correlated to the corresponding counts of L. loa mf in the peripheral blood (Fig. 3B; r = 0.54 and P = 0.004).

DISCUSSION

A quantifiable biomarker using an antigen-based immunoassay for L. loa mf can not only greatly benefit MDA programs by identifying individuals at high risk of SAEs, but it also would be useful for the field or laboratory diagnosis of active L. loa infections and, therefore, for treatment decision making. Biomarkers for most infectious pathogens (viruses being the exception) are proteins, as methodological constraints prevent RNAs from being easily utilized as biomarkers (19). Nevertheless, we utilized gene expression data to identify 11 mRNAs coding for proteins unique to L. loa (expressed by the mf) for potential biomarker discovery. Then, we were able to show that 2 of the 11 identified potential proteins encoded by these mRNAs could be detected in Loa mf-infected sera using capture ELISAs. Importantly, one of these, LOAG_14221, showed reasonable diagnostic performance in polyclonal and monoclonal antibody-based capture ELISAs.

It is well recognized that antigen-based assays are the sine qua non for diagnosis of infectious diseases and for some filarial infections as well (20). Thus, we first successfully developed ELISAs that allowed us to capture two proteins, LOAG_15846 and LOAG_14221, in serum samples of L. loa mf⁺ patients using monospecific polyclonal antibodies. The detected ELISA signal in the two assays was significantly increased with pretreatment of sera with glycine aimed to dissociate serum immune complexes, a pretreatment previously shown to be important for the detection of specific viral antigens (21–23). This suggests that significant amounts of both proteins are often found complexed with antibodies in the human circulation.

Sensitivity and specificity are the two most important measures of diagnostic test performance. Thus, we assessed them in this study using well-characterized serum samples from L. loa, O. volvulus, W. bancrofti, and uninfected individuals. There was a certain degree of reactivity in serum samples from O. volvulus and W. bancrofti using the polyclonal-based assays, especially with the LOAG_15846 assay. The nonspecific reaction observed with the two assays is surprising given the in silico data suggesting that LOAG_14221 and LOAG_15846 are unique to L. loa (Table 1). Fixing the assay's specificity at 100%, the sensitivity of both assays, especially the LOAG_15846 assay, was low due certainly to the mentioned nonspecific reactivity.

The optimization of the LOAG_14221-based capture assay, which showed higher sensitivity with equal specificity, required the use of a monoclonal antibody for capture. This use of monoclonal antibody significantly increased the assay's specificity, an expected finding. The marked increase in sensitivity is likely due to the higher affinity and binding capacity of the monoclonal antibodies to the specific antigen (LOAG_14221) within a mixture of related molecules (24). It also can be explained by the fact that polyclonal capture antibodies bind to multiple epitopes and may be more likely to give false positives than monoclonal antibodies (25).

In contrast to antibody assays, antigen-based immunoassays should be able to distinguish between previous and active infection, particularly for those mf⁺ individuals important for sustaining transmission, and probably can be used as a surrogate for cure following treatment with microfilaricides. That we were able to detect and quantify LOAG_14221 and LOAG_15846 proteins in the majority of microfilaremic L. loa mf-infected individuals and only in very few of those that were amicrofilaremic (L. loa mf⁻) suggests that both of these proteins can be markers of an active L. loa infection. This also suggests that these proteins are either not expressed by L. loa adult worms or L3s or they are expressed at very low levels in these stages.

A positive relationship between the detected amounts of protein and the mf levels in blood was only observed with LOAG_14221 in both polyclonal- and monoclonal (4B10C8)-based assays. This suggests that LOAG_14221 in the blood can be used as a surrogate of L. loa mf infection intensity and that the LOAG_14221 assay may be useful in providing a POC tool to identify those at risk for post-IVM SAEs in a context of MDA campaigns (3). However, the numbers of mf-tested samples were low with ranges from 1 to 7,740 mf/ml, with high microfilaremia being associated with a high risk of developing post-IVM SAEs. The relatively low mf numbers in tested samples may explain why the correlation (r = 0.54) between levels of the LOAG_14221 protein and the corresponding levels of mf is not higher. Detection of circulating mf-derived antigen in serum has been used with limited success for both B. malayi (26, 27) and W. bancrofti (28), but none of these assays examined the relationship between amounts of detected antigen and the number of mf in the blood.

There are currently no commercial products available for measuring circulating antigens in L. loa infection. To fill this gap, we had previously identified another microfilaria-derived protein, LOAG_16297, which is measureable in the serum/plasma samples of L. loa microfilaremic individuals using a less conventional competitive LIPS assay (13). It is likely that monoclonal antibodies directed against LOAG_16297 in combination with LOAG_14221 will increase the sensitivity without loss of specificity of a potential circulating antigen assay, which can be easily adaptable to lateral flow assays.

In summary, we have mined the genome and transcriptome of L. loa mf and identified 11 putative biomarkers. Among them, LOAG_14221 can be detected in the plasma/serum samples of L. loa mf-infected individuals in a capture ELISA format. Levels of LOAG_14221 correlate well with the quantity of L. loa mf found in the peripheral blood of the same individuals. Taken together, our data highlight the utility of the LOAG_14221 mf protein as a biomarker that can be the basis of a lateral flow assay for a POC assessment tool both in areas of L. loa endemicity and in the laboratory diagnosis of active L. loa infection.

MATERIALS AND METHODS

Selection of proteins for immunoassays.

The L. loa microfilarial transcriptome (16) was analyzed for potential candidate biomarkers. The putative proteins were primarily down-selected based on (i) comparison of sequence homologies against human proteins and those of other filarial species (B. malayi, O. volvulus, and W. bancrofti) or other nematodes (e.g., Caenorhabditis elegans) for which the genome was available, (ii) SecretomeP (nonclassical protein secretion pathway) scores, and (iii) RNA expression (fragments per kilobase of transcript per million mapped reads, or FPKM levels). Putative proteins that showed no or little homology (not significant) to non-Loa sequences, significant SecretomeP score (>0.6), and/or high FPKM levels were selected as candidates for recombinant expression and further study.

Production and purification of recombinant proteins.

Recombinant proteins were produced at the Protein Expression Laboratory (Leidos Biomedical Research, Frederick, MD). Briefly, the full-length open reading frame of the cDNA expressing the protein of interest was N-terminal 6×His tagged and cloned for bacterial expression using Gateway recombination cloning technology (Life Technologies, Grand Island, NY). Obtained constructs were used to transfect Escherichia coli, which was grown overnight at 37°C while shaking was maintained at 481 rpm. After centrifugation (5,000 × g for 20 min), cell pellets were immediately frozen at −80°C until use. Microscale immobilized metal affinity chromatography (IMAC) (PhyNexus, San Jose, CA) was used to purify E. coli protein expression constructs. The amount and purity of protein were evaluated by SDS-PAGE, and protein concentration was determined by the Bradford method (Bio-Rad Laboratories, Hercules, CA) with bovine IgG as control.

Generation of rabbit polyclonal antibodies.

Rabbit polyclonal antisera were raised at Spring Valley Laboratories (Woodbine, MD) using a standard protocol. Total IgG from antisera was purified using protein A/G purification (Pierce, Rockford, IL). Some of the purified IgG was biotinylated using a Pierce antibody biotinylation kit according to the manufacturer's recommendations.

Generation of mouse monoclonal antibodies.

Mouse monoclonal IgGs were raised using the GenScript MonoExpress protocol (GenScript, Piscataway, NJ). Briefly, five mice were immunized with the purified recombinant protein and T-max adjuvant. Spleen cells were isolated and fused with myeloma cells, and selected hybridomas were screened by ELISA for reactivity with the recombinant protein. Selected positive parental clones were subcloned by limited dilution. After 2 to 3 rounds of subcloning, stable monoclonal clones were obtained and injected intraperitoneally in mice. Ascites were obtained after 7 to 10 days, and antibodies were purified by protein A/G affinity columns.

Plasma samples.

Well-characterized plasma/serum samples used to validate the utility of the potential biomarkers by capture ELISA were from L. loa-infected individuals (n = 33 [25 microfilaremic, or mf⁺, and 8 amicrofilaremic, or mf⁻]) from Central (mostly Cameroon, Gabon, Central African Republic) and West Africa (Benin). Samples used as controls included those from subjects with W. bancrofti infection (mf⁺; n = 15) from India and the Cook Islands, with O. volvulus infection (mf⁺; n = 15) from Ecuador, and healthy adult volunteers (n = 107) from North America (United States and Canada). These North American volunteers (normal controls) had no history of exposure to filariae or other helminths, had never traveled outside North America, and were collected through a healthy volunteer research protocol from the Division of Transfusion Medicine at the National Institutes of Health. The diagnosis of mf⁺ L. loa and W. bancrofti was made based on the presence of mf by microscopy in a blood smear (and/or in a filtered 1-ml blood sample) or by PCR (11, 29). The presence of O. volvulus mf was determined by skin snip examination by microscopy. For those amicrofilaremic individuals with L. loa (mf⁻), a definitive diagnosis was made based on the presence of an adult worm on biopsy specimen or by having classical symptoms (eye worm or Calabar swellings), an appropriate exposure history, and positive antifilarial serology as previously described (30). All samples were collected from subjects as part of registered protocols approved by the Institutional Review Boards of the National Institute of Allergy and Infectious Diseases, National Institutes of Health, for the filaria-infected patients (NCT00001345) and for healthy donors (NCT00090662). Written informed consent was obtained from all subjects.

Capture ELISA for detection of proteins in plasma.

Immulon 4 flat bottom 96-well microplates (Dynatech Laboratories, Alexandria, VA) were coated with 100 μl of 0.5 μg/ml of polyclonal or 1 μg/ml of monoclonal IgG in phosphate-buffered saline (PBS) and incubated overnight at 4°C. Microplates were blocked with 5% bovine serum albumin in PBS for 1 h at 37°C. To dissociate potential immune complexes, sera were pretreated with 0.15 M glycine (pH 2.0), heated at 70°C for 10 min, cooled to room temperature, and neutralized with 3.5 M Tris (pH 10.8) as previously described (21). Then, 100 μl of the glycine-treated sera was added to duplicate wells and incubated at 37°C for 1 h. The microplates were washed and then incubated with 100 μl of biotinylated recombinant-specific polyclonal IgG (1:5,000) at 37°C for 1 h and washed. Next, we added 100 μl (1:10,000) of streptavidin conjugated to alkaline phosphatase (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) at 37°C for 1 h. After again washing, 100 μl of 1 mg/ml of p-nitrophenyl phosphate (Sigma Chemicals, Saint-Louis, MO) in sodium carbonate buffer (pH 8.6) was added, and absorbance of the reaction was measured at 405 nm after 15 min using a microplate ELISA reader (Molecular Devices, Sunnyvale, CA). For each assay, a dilution series of known concentrations (from 10 to 0.005 μg/ml) of the protein of interest in 5% human AB serum served as standards.

Statistical analysis.

Statistical analyses, including specificity and sensitivity calculations (ROC curve analysis), correlations (Spearman's rank method), and difference in protein amounts between multiple groups (Kruskal-Wallis test) were performed using Prism 6.0 (GraphPad Software, San Diego, CA). All differences were considered significant at a P level of <0.05.