Human genital tract secretions contain monomeric IgG and IgA antibodies, which are thought to be primarily derived from serum, and polymeric IgA and IgM antibodies, which are locally produced in mucosal tissues and transported across the epithelium via a specific receptor-mediated transport system involving the polymeric immunoglobulin receptor(1). Genital tract infections can lead to specific local IgA and IgM antibody responses and also the appearance of pathogen-specific IgG antibodies in local secretions. Pathogen-specific IgG and IgA isotype antibodies were detected in cervicovaginal secretions from women following intravaginal administration of live attenuated polio vaccine(2), and infection by Neisseria gonorrhoeae(3), herpes simplex virus type 2 (HSV-2)(4), and Chlamydia trachomatis(5). In men, pathogen-specific antibodies have been detected in semen following genital tract infection by Chlamydia trachomatis(6), hepatitis B virus(7), and Neisseria gonorrhoeae(8). Genital tract antibodies are thought to play an important role in defense against genital tract infections. Both IgG and IgA antibodies effectively neutralize bacterial and viral pathogens(9). Studies in mice have demonstrated resistance to HSV-2 infection following active vaginal immunization with attenuated virus(10), or passive vaginal immunization with HSV-2-specific antibodies(11). Studies on primates have shown protection against vaginal SIV transmission following immunization with microencapsulated vaccine and appearance of anti-SIV antibodies in vaginal secretions(12).
HIV-specific antibodies have been detected in genital tract secretions from HIV-seropositive men and women(13,14) and may influence the sexual transmission of HIV-1. Studies in vitro have demonstrated that HIV-1-specific IgA and IgG subclass antibodies can neutralize HIV-1(15,16); however, other studies have shown that antibodies can also promote HIV-1 infection via Fc or complement receptor-mediated mechanisms(18). The effects of genital tract HIV-1-specific antibodies on the mucosal transmission of HIV-1 are unknown.
The purpose of this study was to develop a method for the quantification of HIV-1-specific immunoglobulin isotypes in genital tract secretions of HIV-1-seropositive men and women to enable a better understanding of their potential role in viral transmission. Published reports on antibodies in genital tract secretions have failed to provide a consensus on concentrations of immunoglobulin isotypes in semen and cervicovaginal secretions. Furthermore, data from studies of mucosal anti-HIV antibody responses are usually presented as positive/negative or as ELISA optical density (OD) units, making it difficult to compare data obtained from different experiments and laboratories. In this report, we describe a quantitative assay based on HIV-specific immunoglobulin isotype standards for the accurate assessment of HIV-1-specific IgG, IgA, and IgM antibodies in genital tract secretions. An analysis of clinical parameters affecting mucosal antibody profiles in HIV-1-infected men and women will be published in a subsequent report.
MATERIALS AND METHODS
Cervicovaginal lavage (CVL) and semen specimens were from frozen(−70°C) archives of samples collected from 1991 through 1994 from the Heterosexual AIDS Transmission Study (HATS) and from Boston Partner's Study. Confirmed seronegative women (n = 15) and other low-risk women (n = 65) and low-risk men (n= 10) were recruited from the Harvard Medical School community to serve as negative controls. Clinical information was available for seropositive men (n = 14) and women (n = 62) from the Boston and Providence sites. In addition, all samples had recorded information concerning polymorphonuclear leukocyte (PMN) concentration, as assessed by Endtz test as described previously(19).
CVL and Semen Collection Procedure
Women were instructed to abstain from intercourse for at least 48 hours prior to sample collection. CVL samples were obtained using a modified 20-ml syringe filled with sterile Hank's balanced salt solution(HBSS), as described previously(20). The syringe was inserted into the vagina while the participant was in a supine position, and HBSS was expelled and drawn into the syringe three times. After the final collection, contents were emptied into a 50-ml sterile test tube and transported on ice to the laboratory. Local samples were sent via same-day courier; out-of-town samples were sent via overnight mail. In the laboratory, samples were centrifuged for 10 minutes at 1000 xg. Supernatants were aliquoted and stored at −70°C. Before testing for antibodies, samples were thawed and heat inactivated for 30 minutes at 56°C. Compliance with the 48-hour abstinence requirement was confirmed by an ELISA that detects seminal plasma (SEMA test, Humagen. Charlottesville, VA, U.S.A.).
Men were also instructed to abstain from intercourse for at least 48 hours prior to collection. Semen was collected by masturbation into sterile urinalysis containers. Upon receipt (usually within 1 hour of collection), samples were diluted 1:1 (vol:vol) in sterile phosphate-buffered saline (PBS) and centrifuged at 400 xg for 10 minutes. Seminal plasma was stored at −70°C until testing.
Protein concentration in CVL samples was determined using the BCA Protein Assay kit (Pierce, Rockford, IL, U.S.A.).
Capture ELISA for Quantitation of Immunoglobulin Isotypes
Goat antibodies [F(ab)'2] against total human immunoglobulins and IgA, IgM. and IgG antibody isotypes were obtained from Accurate Chemical and Scientific Corporation, Westbury, NY, U.S.A. Antibodies were diluted to 5 µg/ml in 15 mM Na2CO3/35 mM NaHCO3 buffer, pH 9.5, and 50 µl of this solution was added to each well of ELISA plates (96-well Falcon plates, Becton Dickinson, Lincoln Park, NJ, U.S.A.) and incubated overnight at 4°C. After excess antibody was removed, a blocking buffer consisting of 3% bovine serum albumin(BSA) and 0.05% Tween 20 solution in PBS (0.01 M PBS, 0.138 M NaCl, 0.0027 M KCl, pH 7.4) was added to each well and plates were incubated for 2 hours at room temperature before washing. CVL specimens, semen, or serial fivefold dilutions of immunoglobulin standards(total immunoglobulin [Calbiochem, La Jolla, CA, U.S.A.]; chromatographically purified IgG, IgA [from colostrum], or IgM [Cappel, Durham, NC, U.S.A.]) were added in 50-µl volumes to individual wells in the ELISA plates and were incubated for 30 minutes at room temperature. Plates were then washed and incubated with horseradish peroxidase(HRP)-conjugated goat F(ab)'2 antibodies specific for human immunoglobulins, IgG, IgA, or IgM (Accurate for 30 minutes at room temperature. Optimal dilutions for the secondary HRP-conjugated antibodies were 1:20,000 for anti-Ig and anti-IgG antibodies and 1:15,000 for anti-IgA and anti-IgM antibodies. Peroxidase substrate (0.03% H2O2, O-phenylenediamine, Sigma), was added to the wells in 100-µl volumes and was incubated for 30 minutes at room temperature. The color reaction was terminated by addition of 1 N H2SO4 (100µl/well); ODs were read on an ELISA reader with a 492-nm filter.
Screening for HIV-1-Specific Antibodies
HIV-1-specific immunoglobulins in CVL samples and semen were detected by ELISA (Abbott Laboratories, North Chicago, IL, U.S.A.). The Abbott kit was originally designed to detect antibodies in sera diluted 1:400. At the start of our study, we roughly estimated that 20-ml CVL specimens represented a 1:20 dilution of the original secretion, and diluted the samples an additional 20-fold to approximate the optimal serum dilution for the assay. Ten µl of CVL or 10 µl of a 1:20 dilution of seminal plasma were added to 200 µl of diluent and incubated with HIV-1-coated beads for 1 hour at 37°C. The beads were then washed using the Qwik Wash system (Abbott Laboratories, Irving, TX, U.S.A.) and incubated for 2 hours at 37°C with an anti-human, heavy-and light-chain specific IgG Fab that is also reactive with IgM and IgA isotypes. The absorption threshold above which a sample was considered positive for HIV-1 antibodies (mean OD of negative control+0.1 × OD of positive control) was calculated by the ELISA reader computer from positive and negative serum controls supplied with the kit. Because seronegative CVL and semen samples had a higher "background" than sera in the ELISA, the negative cutoff level for each fluid was refined further by determining the mean OD of three CVL samples from HIV-1-seronegative women or three seminal plasmas from HIV-1-seronegative men and using these values +0.1 × OD of the serum positive control as the threshold for positivity for CVL and seminal plasma specimens, respectively.
Protein-G Treatment to Remove IgG
Protein-G was used to remove IgG from samples prior to IgA and IgM assessment, as described previously(21). CVL samples were incubated with protein-G-coated sepharose beads (Sigma), 10:1 (vol-:vol), with constant mixing for 3 hours at room temperature. Samples were then centrifuged for 3 minutes at 1000 xg to pellet the beads, and the supernatant was collected. To ascertain the efficiency of IgG removal, IgG concentrations were measured by IgG-specific capture ELISA (Abbott) before and after protein-G treatment. IgG-depleted CVL and seminal plasma supernatants were used to quantify total and HIV-1-specific IgM and IgA isotype antibodies as described below.
Preparation of HIV-1 Specific Immunoglobulin Isotype Standards
Three HIV-antibody-positive CVL samples selected for high titers of all three immunoglobulin isotypes were pooled, and a 1.8-ml sample was incubated on a rocker overnight at room temperature with HIV-1 antigen-coated beads (600 µl/bead) (Abbott). To elute HIV-specific antibody from the beads after washing, 200 µl of 0.1 M glycine-hydrochloric acid, pH 2.7, was added to each well and mixed for 5 minutes. The eluted anti-HIV antibody was then transferred to tubes containing 20 µl of 1 M Tris-HCl, pH 9, for pH neutralization. The concentrations of HIV-1-specific total immunoglobulins, as well as immunoglobulin isotypes IgG, IgA, and IgM, were determined by ELISA using commercially available immunoglobulin standards as described earlier. The HIV-1-specific immunoglobulins were then serially diluted for use as standards.
Quantitation of HIV-1-Specific IgG, IgA, and IgM in CVL and Seminal Plasma Samples
A modified version of the Abbott ELISA was used to quantify HIV-1-specific antibody isotypes. CVL or seminal plasma samples were diluted 1:20 and 1:200 (vol:vol), respectively, and incubated with HIV-1 antigen-coated beads overnight at room temperature. After washing, the beads were incubated with one of the following peroxidase-conjugated secondary antibodies: goat anti-human IgG F(ab)'2 diluted 1:1200, goat anti-human µ-chain-specific IgM diluted 1:600. (Organon Tecknica Durham, NC, U.S.A.), and goat anti-humanα-chain-specific IgA 1:500 dilution, (Southern Biotech, Birmingham, AL, U.S.A.). Serial dilutions of the HIV-1-specific antibody standards were run in each assay to establish a standard curve. The positive cutoff level for each antibody isotype was calculated as the mean OD of three CVL or seminal plasma samples from seronegative subjects +10% of the mean OD of the positive serum control. The OD readings were converted to micrograms per milliliter using predicted concentration values from the standard curves for each immunoglobulin before treatment with protein-G for detection of IgG and after treatment with protein-G for detection of IgM and IgA (discussed later), then antibody concentrations were obtained by multiplying these values by 20 for CVL samples or 200 for seminal plasma samples to compensate for the assay dilution factor. Antibody concentrations were also calculated as percent of total protein.
Statistics and Calculations
Twofold dilutions of HIV-1-specific antibody from the CVL pool were used to establish a standard curve for each HIV-specific antibody isotype. OD values were plotted on the x axis and the corresponding known immunoglobulin concentrations of the standards in micrograms per milliliter were plotted on the y axis. Both linear and polynomial regression analyses were performed on raw and logarithmically transformed OD readings for the most accurate immunoglobulin quantification based on predicted values. The Stat View II statistical software program (Abacus Concepts, Berkeley, CA, U.S.A.) was used for this purpose.
The HIV-1-specific immunoglobulin concentrations obtained were analyzed by descriptive statistics (Stat View II) to determine the range of positive values and the intra- and interassay variation. Spearman rank order analysis was applied to assess the correlation between protein and total immunoglobulin content in CVL and seminal plasma samples.
Pilot experiments in which various concentrations of isotype standards were added to CVL or semen samples determined that antibody concentrations were not affected by inhibitors present in these samples or by up to two freeze-thaw cycles (data not shown).
Five of the original 201 CVL samples were deleted from the study because of extremely low (<150 µg/ml) protein content (n = 3) or SEMA positivity (n = 2). Various statistical methods were used to calculate experimental immunoglobulin concentrations from the standard curves. For high immunoglobulin concentrations, the most accurate values were obtained from polynomial regression analysis of raw or logarithmically transformed data. For low immunoglobulin concentrations, polynomial regression analysis of logarithmically transformed data was the most accurate (Table 1). Concentrations of IgG, IgA, and IgM in CVL and semen specimens analyzed in this study are presented in Table 2, and agree with previous reports (8,9).
Since almost all of the HIV-specific antibody values corresponded to the lower range of the standard curve, polynomial regression analysis of logarithmically transformed OD values was used to determine the predicted antibody values. The reproducibility of the ELISA procedure was determined by testing CVL samples from three seropositive women for each of the three antibody isotypes in triplicate in three separate experiments. The intra-assay coefficients of variation ranged from 2% to 20%; the interassay coefficients of variation ranged from 5% to 33% (Table 3). In both cases, best reproducibility was obtained for HIV-1-specific immunoglobulins and IgG, which were present in higher concentrations than IgA and IgM.
The sensitivity of the assay was based on the lowest concentration of HIV-1-specific immunoglobulin standard that resulted in an OD above the ELISA cutoff level(mean OD of three seronegative CVL samples +0.1 × OD of positive serum control). The sensitivity was >0.01µg/ml for IgG, >0.04 µg/ml for IgA, and >0.08 µg/ml for IgM. According to the positivity threshold established in our study, all 80 HIV-1-seronegative or lowrisk women were negative for HIV-1-specific immunoglobulin isotypes (IgG, IgA, and IgM), whereas CVL samples from all 196 HIV-1-seropositive women had HIV-specific immunoglobulin values above the positive threshold. All CVL samples from seropositive women contained detectable amounts of HIV-1-specific IgG(0.01-46.64 µg/ml), whereas only 16 CVL samples contained detectable levels of HIV-1-specific IgA (1.38-3.88µg/ml) and only 8 CVL samples contained detectable levels of HIV-1-specific IgM (1.79-8.61 µg/ml)(Table 4). In the vast majority of patients (>95%), concentrations of HIV-1-specific IgG antibodies were considerably higher than those of IgA and IgM antibodies. However, in nine patients, concentrations of HIV-1-specific IgA, IgM, or both exceeded those of HIV-1-specific IgG. HIV-1 antibody isotypes are also presented as percentage of total amount of protein, percentage of total immunoglobulins, and percentage of HIV-1-specific immunoglobulins (Table 4). Good correlations were observed between concentration of HIV-1-specific IgG and percentage of total protein, percentage of total IgG, and percentage of total HIV-1-specific immunoglobulins(Spearman rank order correlation coefficients+0.95, +0.89, and +0.95, respectively, all at p <.0001).
Preliminary analysis of clinical factors associated with HIV-specific immunoglobulin levels in CVL samples indicated that levels of IgG were significantly elevated in CVL specimens from women who had acquired HIV heterosexually in comparison with the intravenous drug user risk group (p < .01, Mann Whitney U-test), and that the majority (11 of 13) of women with elevated HIV-specific IgA levels in CVL samples had evidence of genital tract inflammation (>104 PMNs/ml CVL; p < .02, Fisher's exact test). None of the other clinical factors(e.g., menstrual cycle, peripheral CD4 count, antiviral therapy) were significantly associated with total or HIV-specific antibody titers.
HIV-1-specific IgG antibodies were detected in 13 of 14 (93%) of seropositive men, whereas IgA and IgM HIV-1-specific antibodies were only detected in 10 of 14 men (71%). Median concentrations of HIV-1-specific IgA and IgM subclass antibodies were considerably lower than those of IgG subclass antibodies in most men, but a few men had IgA:IgG ratios >1 for HIV-1-specific antibodies (Table 5).
We have developed a method to quantify isotypes of HIV-1-specific antibodies, and have used this method to assess concentrations of HIV-1-specific IgG, IgA, and IgM antibodies in semen and CVL specimens from seropositive men and women. Our data agree with previously published reports that IgG is the dominant immunoglobulin in genital tract secretions(1,8,9), and confirms that HIV-1-specific IgG antibodies are usually present in substantially higher concentrations than are classical mucosal antibodies (IgA and IgM)(13,14). All seropositive women and most seropositive men in our study had HIV-specific antibodies of the IgG isotype in genital tract specimens, whereas fewer subjects (<10% women, 71% men) had detectable levels of HIV-1-specific IgA or IgM antibodies. Furthermore, our study for the first time provides absolute concentrations of HIV-1-specific antibodies of the IgG. IgA, and IgM isotypes in semen and CVL samples. In the majority of cases, the concentration of IgG anti-HIV antibodies exceeded the concentration of HIV-specific IgA or IgM subclass antibodies. A study is currently underway to identify factors associated with high IgA:IgG or IgM:IgG ratios in genital tract specimens from an expanded population of seropositive men and women, and will be the subject of a future report.
Since many of the CVL samples used in this study were collected in the field under poorly controlled conditions, we performed two tests to assure high quality of the samples. To eliminate samples with extremely low protein concentrations as a result of poor collection technique, total protein content was determined for each CVL sample. The median protein concentration in CVL samples was 426 µg/ml, and samples with <150 µg/ml were extreme outliers (<2.5% of the population). Therefore, samples with <150 µg/ml were excluded from the study. To verify compliance with the request for no intercourse during the 48-hour period prior to collection of samples, we performed the SEMA test to detect the presence of semen from recent sexual intercourse(19). Two positive samples were found, and these were excluded from the study, because semen can be a source of anti-HIV antibodies(13,14).
Quantification of HIV-1-specific antibodies was achieved by converting OD values obtained from the ELISA reader to antibody concentrations through use of HIV-specific antibody standards of known concentrations. The linear relationship between logarithmically transformed ODs and known values in micrograms per milliliter was established by polynomial regression analysis of predicted values. Regression analysis permitted a unit of change in a dependent variable for each unit of change in an independent variable. By taking the antilog of the dependent values in micrograms per milliliter, we calculated the amount of HIV-1-specific antibody isotype in each CVL sample. The 1:20 dilution factor for the ELISA was taken into account when calculating the final value in micrograms per milliliter. To convert the antibody concentration per CVL sample to the original concentration in the undiluted cervicovaginal secretion, one would have to know the dilution factor inherent in the lavage technique. Belec et al.(22) used a method based on adding a known concentration of lithium chloride to the lavage solution prior to collection to more precisely determine the dilution factor in CVL samples prospectively collected for antibody assessment. They determined that, on average, 300 µl of cervicovaginal secretions are collected when a 3-ml amount of lavage fluid is used, but there was a large range. This figure could be extrapolated to a 60× dilution factor in a 20-ml lavage. Protein concentration cannot be used to determine the dilution factor because it varies in cervicovaginal secretions(1). The median protein concentration in 20-ml CVL samples in this study (427 µg/ml) was approximately 140- to 180-fold less than that in serum(60-80 mg/ml), but the range was considerable (300×− 12× less than serum). Unfortunately, most routine clinical CVL specimens are collected without the addition of lithium or other substances to enable precise determination of the dilution factor. Antibody measurements on undiluted cervicovaginal secretions would preclude the need for dilution factor assessment and would most likely produce a higher detection rate of HIV-1-specific IgA and IgM class antibodies and improve the accuracy of the test. Now that a precise immunoglobulin quantification system is available, either undiluted or dilution factor-corrected CVL samples should be collected in prospective studies for assessment of immunoglobulin concentrations. In addition, reagents for the quantitative assessment of polymeric IgA and IgM antibodies that contain secretory component and are therefore mucosally derived should also be developed. Existing commercially available antibodies to secretory component were tested for use in our study, but they provided an unacceptably low signal-to-noise ratio.
In summary, we present the basis for accurate quantitation of HIV-1-specific antibody isotypes in genital tract secretions. This approach should prove useful in standardizing the measurement of mucosal HIV antibodies in different laboratories, leading to the further identification of clinical and physiologic factors that affect anti-HIV mucosal immunity. Preliminary analysis from our study suggests that genital tract inflammation is associated with elevated levels of HIV-specific IgA in cervicovaginal secretions. We did not observe an effect of menstrual cycle, peripheral CD4 count, or antiviral therapy on CVL antibody concentrations; however, the cross-sectional design of this study and the large individual variation in immunoglobulin concentrations could have masked potential effects. Previous reports have indicated substantial effects of the menstrual cycle and oral contraceptives on immunoglobulin concentrations in human cervical mucus(9). These and other clinical factors should be further studied in carefully designed longitudinal studies using standardized quantitative methodology. Ultimately, such studies will contribute to a further understanding of the role of mucosal immunity in HIV-1 transmission and antiviral defense.
Acknowledgments: This research was supported by grants AI35564 and HD33205 from the National Institutes of Health, Bethesda, MD, U.S.A. The authors thank Lynne Tucker, B.S., and Drs. Joseph Politch, Marshall Posner, and Raina Fichorova for valuable technical assistance and Bernadette Aidonidis for typing the manuscript.
1. Kutteh WH, Mestecky J. The concept of mucosal immunology. In: Bronson RA, Alexander NJ, Anderson DJ. Branch DW, Kutteh WH, eds. Reproductive immunology
. Boston: Blackwell Science, 1996:28-51.
2. Ogra PL, Ogra SS. Local antibody response to polio vaccine in the human female genital tract. J Immunol
3. O'Reilly RJ, Lee L. Welch BG. Secretory IgA antibody responses to Neisseria gonorrhoeae
in genital tract secretions of infected females. J Infect Dis
4. Ashley RL, Corey L, Dalessio J, et al. Protein-specific cervical antibody responses to primary genital herpes simplex virus type 2 infections. J Infect Dis
5. Richmond SJ, Milne DJ, Hilton AL, Caul EO. Antibodies to Chlamydia trachomatis
in cervicovaginal secretions: relation to serum antibodies and current chlamydial infection. Sex Transm Dis
6. Kojima H, Wang SP, Kuo CC, Grayston JT. Local antibody in semen for rapid diagnosis of chlamydia trachomatis epididymitis. J Urol
7. Ayoola EA, Ladipo OA, Adelola HA. Antibody to hepatitis B core antigen, e-antigen and its antibody in menstrual blood and semen. Int J Gynecol Obst
8. Fowler JE. Antibody response to bacterial infections of the male urogenital tract. In: Bronson RA, Alexander NJ, Anderson DJ, Branch DW, Kutteh WH, eds. Reproductive immunology
. Boston: Blackwell Science. 1996:513-31.
9. Parr MB, Parr EL. Immunoglobulins in the female genital tract. In: Bronson RA, Alexander NJ, Anderson DJ, Branch DW, Kutteh WH, eds. Reproductive immunology
. Boston: Blackwell Science, 1996:275-308.
10. McDermott MR, Smiley JR, Leslie P, Brais J, Rudzroga HE, Bienstock J. Immunity in the female genital tract after intravaginal vaccination of mice with an attenuated strain of herpes simplex virus type 2. J Virol
11. Whaley KJ, Zeitlin L, Barrat RA, Hoen TE, Cone RA. Passive immunization of the vagina
protects mice against vaginal
transmission of genital herpes infections. J Infect Dis
12. Marx PA, Compans RW, Grettie A, et al. Protection against vaginal
SIV transmission with microencapsulated vaccine. Science
13. Wolff H, Mayer K, Seage G, Politch J. Horsburgh CR, Anderson DJ. A comparison of HIV-1 antibody classes, titers, and specificities in paired semen and blood samples from HIV-1 seropositive men. J Acquir Immune Defic Syndr Hum Retrovirol
14. Belec L, Tevi-Benissan C, Lu XS, Prazuck T, Pillot J. Local synthesis of IgG antibodies to HIV within the female and male genital tracts during asymptomatic and pre-AIDS stages of HIV Infection. AIDS Res Hum Retroviruses
15. Bukawa H, Sekigawa K, Hamaiima K, et al. Neutralization of HIV-1 by secretory IgA induced by oral immunization with a new macromolecular multicomponent peptide vaccine candidate. Nature Med
16. Burnett PR, VanCott TC, Polonis VR. Redfield RR, Birx DL. Serum IgA-mediated neutralization of HIV type 1. J Immunol
17. Kozlowski PA. Black KP, Shen L, Jackson S. High prevalence of serum IgA HIV-1 infection-enhancing antibodies in HIV-infected persons. J Immunol
18. Robinson WE, Montefiori DC, Gillespie DH, Mitchell WM. Complement-mediated, antibody-dependent enhancement of HIV-1 infection in vitro is characterized by increased protein and RNA syntheses and infectious virus release. J Acquir Immune Defic Syndr
19. Politch JA, Wolff H, Hill JA, Anderson DJ. Comparison of methods to enumerate white blood cells in semen. Fert Steril
20. Hill JA, Anderson DJ. Human vaginal
leucocytes and the effects of vaginal
fluid on lymphocyte and macrophage defense functions. Am J Obstet Gynecol
21. Weiblen BJ, Schumacher RT, Hoff R. Detection of IgM and IgA HIV antibodies after removal of IgG with recombinant protein-G. J Immunol Methods
22. Belec L, Meillet D, Levy M, Georges A. Tevi-Benissan C, Pillot J. Dilution assessment of cervicovaginal secretions obtained by vaginal
washings for immunological assay. Clin Diag Lab Immunol