All 63 HIV-1 infected women (Table 1) were recruited from the 1917 Clinic at the University of Alabama at Birmingham (UAB). At entry of the study, the average of absolute CD4⁺ T-cell count and plasma viral load was 601/µl and 30,860 vRNA copies/ml, respectively. Thirteen of the subjects had a level below the limit of detection (<50 vRNA copies/ml). The majority of patients (93%) had been diagnosed as having HIV-1 for at least 1 year. Four of the subjects were identified as HIV-1 positive for 2-7 months. After obtaining written and informed consent approved by the local institutional review board and biosafety committee, serum/plasma, CVL, and RL samples were collected. Freshly obtained serum and/or acid citrate dextrose-processed plasma samples were aliquoted and frozen at -80°C. Serum/plasma samples were heat-inactivated at 56°C for 1 h before being used in assays. The difference in HIV-1 neutralization activity between serum and plasma was not observed by us and another group [35]. The CVL samples were obtained by flushing the cervix and vagina with 5 ml of sterile saline; the washes were collected from the posterior vaginal vault into tubes containing protease inhibitors (Sigma-Aldrich, St. Louis, MO) [36]. The RL samples were collected after the instillation of 50 ml of sterile saline into the rectum; the fluids were retained for up to 5 min, and then expelled into a collection device with protease inhibitors [36]. Both CVL and RL samples were then centrifuged, aliquoted, and stored at -80°C until assayed.

The concentration of total Ig in serum/plasma, CVL, and RL samples was determined by ELISA as described previously [17, 18, 22]. Briefly, duplicates of 2-fold serially diluted samples and a human Ig reference serum (pool of calibrated human sera, Liquichek^TM Immunology Control, BioRad, Hercules, CA) were added to 96-well microplates (Nalge Nunc International, Rochester, NY) coated with goat F(ab’)₂ anti-human IgA, IgG, or IgM (Jackson ImmunoResearch Laboratories, West Grove, PA). The captured Ig was detected after consecutive incubations with biotin-labeled goat F(ab’)₂ specific for human IgA, IgG, or IgM (Geneway Biotech. Inc., San Diego, CA), horseradish (HRP)-labeled ExtrAvidin (Sigma-Aldrich), and the peroxidase substrate (O-phenylenediamine-H₂O₂, Sigma-Aldrich). The color reaction was stopped with 1 M sulfuric acid. The absorbance was measured in an EL312 Bio-Kinetics microplate reader (Bio-Tek Instruments Inc., Winooski, VT) at 490 nm. Total Ig levels were calculated by interpolating the optical densities on calibration curves constructed using the Delta Soft 3 computer program (BioMettalics Inc., Princeton, NJ), as previously described [37].

The presence of HIV-1-specific antibodies of the IgA or IgG isotype was determined by WB adapted from a previously described method [17, 18, 22, 38, 39] using commercial HIV-1 strips (Cambridge Biotech, Worchester, MA). The strips contain 9 HIV-1 gene products derived from HIV-1_IIIB-infected H9 T cells. These proteins are identified as gp160, gp120, p66, p55, gp41, p51, p31, p24, and p17. Strips were incubated overnight at 4°C with 1:500 diluted serum/plasma samples or 1:20-1:50 diluted CVL and RL samples. High background staining was observed when the dilution of CVL and RL collected from either infected or uninfected individuals was lower (e.g., 1:2-1:10). This was attributed to nonspecific absorption of mucosal proteins and glycoproteins [18]. Strips were washed, then incubated with biotinylated F(ab’)₂ fragments of goat anti-human IgA or IgG (Geneway Biotech. Inc., San Diego, CA), followed by ExtrAvidin alkaline phosphatase conjugate (Sigma-Aldrich), and finally developed with alkaline phosphatase substrate: p-nitro blue tetrazolium chloride enhanced 5-bromo-4-chloro-3-indolyl phosphate (Bio-Rad). Scoring of band intensity (from 0 to 3+) was done as previously described [17, 18, 22, 39].

The TZM-bl cell line (NIH ARRRP catalog no. 8129) and the human embryonic kidney cell line 293T (ATCC CRL-11268) were both maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), L-glutamine (2 mM), penicillin (100 U/ml), and streptomycin (100 µg/ml). TZM-bl cell is a genetically engineered HeLa cell clone that expresses CD4, CXCR4, and CCR5, and contains Tat-responsive reporter genes for firefly luciferase (Luc) and Escherichia coli β-galactosidase under regulatory control of an HIV-1 long terminal repeat (LTR) [40, 41]. Cultures were incubated at 37°C in a humidified 5% CO₂-95% air environment. Cell monolayers were split at 1:10 ratio by treatment with 0.25% trypsin in 1 mM EDTA solution (Invitrogen) when cells reached at about 90 percent confluency.

HIV-1_NL4.3 and pseudovirus (HIV-1_SF162, HIV-1_YU2, HIV-1_TV1, and HIV-1_92BR025.09) stocks were generated by transfecting HIV-1_NL4.3 full-length plasmid DNA (16 µg) or rev/env expression plasmid (5 µg) and an env-deficient HIV-1 backbone vector (pSG3Δenv 10 µg) into exponentially dividing 293T cells using FuGENE 6 according to the manufacturer’s protocol (Roche Applied Science, Indianapolis, IN). Virus-containing culture supernatants were harvested 60 h post-infection, centrifuged at 3,000 rpm for 10 min, passed through 0.22 µm filter, and frozen at -80°C. The newly generated virus stocks were subsequently used in the infectivity assay.

TZM-bl cells were seeded at a concentration of 10⁴ cells/well in 96-well flat-bottom culture plates (BD Biosciences, San Jose, CA) 1 day prior to infection. Virus stock was serially diluted (three-fold dilutions) in 240 µl of complete medium with 20 µg/ml DEAE Dextran in separate Nunc 96-well MicroWell plates (Nalge Nunc International, Rochester, NY). 100 µl of serially diluted virus was added to each well (duplicated at each dilution) after the medium in the culture plates was discarded. After a 48 h incubation, the medium was removed and 100 µl of 1 x lysing buffer (Promega, Madison, WI) was added to each well. Plates underwent two freeze-thaw cycles to allow complete cell lysis. 20 µl of cell lysate was then transferred to 96-well white solid plates (Thermo Fisher Scientific) to measure luminescence in a Victor 2 luminometer (Perkin-Elmer Life Sciences, Shelton, CT) after 100 µl of Luciferase Assay System substrate (Promega) was automatically injected into designated wells. Dilutions of virus stocks that yielded relative light units (RLU) from 150,000 to 250,000 on TZM-bl cells were identified and employed in the neutralization assay.

Neutralization was measured as the reduction in Luc reporter gene expression after a single round of virus infection in TZM-bl cells [42, 43]. Briefly, freshly trypsinized cells were cultured overnight in 96-well flat-bottom culture plates at a density of 10⁴ cells per well. Serial dilutions of test samples were prepared in 120 µl of complete medium in separate Nunc 96-well plates. Virus stocks were diluted (1:1200 for HIV-1_NL4.3, 1:240 for pseudoviruses) in 120 µl of complete medium with 40 µg/ml of DEAE Dextran and added to each well of the test sample plate, yielding a total volume in each well of 240 µl. Two sets of control wells were also included (each with at least 6 wells). One set contained cells plus virus (virus infectivity control), and another set contained cells only (background control). The virus-sample mixtures were incubated at 37°C for 1 h. The medium from the cell plates was discarded, and then 100 µl of the virus-sample mixtures was transferred into the corresponding wells (duplicate for each dilution). After a 48 h incubation, the luciferase activity expressed as RLU was measured as previously described for the virus infectivity assay. The reduction of viral infectivity was calculated by comparing the average RLU in wells cultured with virus alone after the subtraction of the background RLU. The highest dilution of a sample that inhibited viral infection by 50% was considered the neutralization titer (IC₅₀). Indinavir (NIH ARRRP catalog no. 8145), a protease inhibitor, was added to a set of HIV-1_NL4.3 control wells at a final concentration of 1-2 μM, which is sufficient to inhibit nearly all secondary infections [44]. The difference in the mean RLU from the HIV-1_NL4.3 control and from HIV-1_NL4.3 with Indinavir were not statistically significant (p=0.10), indicating that HIV-1_NL4.3 underwent only a single round of replication in TZM-bl cells during the 48 h incubation.

IgA and/or IgG was removed by incubating serum/plasma, CVL, and RL samples with immobilized Jacalin (Pierce, Rockford, IL) or Protein G Sepharose (GE Healthcare Bio-sciences Corp., Piscataway, NJ), respectively. The amount of Jacalin or Protein G to be used was calculated according to manufacturer’s indicated binding capacity for IgA or IgG. However, an approximately 20% excess of Jacalin or Protein G was used for Ig depletion in order to ascertain the complete removal of the selected Ig. Levels of total Ig in samples before and after depletion were determined by ELISA. The percentage of Ig depletion was calculated as follows: 100-[(Ig concentration after depletion / Ig concentration before depletion) x 100].

IgA was the dominant total Ig isotype in RL, while IgG was dominant in CVL as well as in serum/plasma samples (Table 2). Although the pattern of IgG>IgA>IgM levels was observed in both CVL and serum/plasma samples, the levels of total IgG or IgA in CVL were about 400-fold lower than those of the corresponding Ig in serum/plasma samples. Total IgM and IgG levels were found to be the lowest of all Ig isotypes present in CVL and RL, respectively.

The presence of HIV-1-specific antibodies of the IgA and IgG isotypes in serum/plasma, CVL, and RL of the HIV-1-infected women were determined by WB (Table 3). HIV-1-specific antibodies of the IgG isotype to the three ENV antigens (gp160, gp120, and gp41) were detected in the serum/plasma of all 63 HIV-1-infected women. In addition, bands of strong intensity representing HIV-1 p66, p55, p51, p31, p24, and p17 antigens were also prominent. HIV-1-specific antibodies to gp160, gp120, and gp41 of the IgA isotype were detected in 94%, 62%, and 86% of serum/plasma samples, respectively. However, the majority of the samples had fewer positive bands with much lower intensity (Fig. 1). HIV-1 ENV-specific IgG antibodies were readily detectable in CVL samples, though with lower intensity than in corresponding serum/plasma. 100%, 82%, and 73% of CVL samples were positive for IgA antibodies to gp160, gp120, and gp41, respectively. In spite of the 9-fold lower level of total IgG than IgA in RL samples (Table 2), the majority of RL samples contained IgG antibodies to HIV-1 ENV antigens. When present, IgA antibodies were mainly specific for gp160. These data demonstrated that IgG-binding antibodies to HIV-1 antigens were the dominant Ig isotype presented in the serum/plasma, CVL, and RL samples of HIV-1-infected women.

Fig. (1). HIV-1-specific IgG and IgA antibodies in serum/plasma samples determined by WB. Note the less intense bands and lower number of positive bands recognized by HIV-1-specific IgA antibodies.

Ig in external secretions are not only produced locally but also derived from the circulation, and the relative contribution of the two origins to the Ig pools varies with Ig isotype and the type of secretion [21]. To determine the relationship of antibody responses to HIV-1 antigens in serum/plasma and the corresponding secretions, the patterns of HIV-1-specific antibodies of the IgG or IgA isotype were compared in parallel among 11 samples (Table 4). IgG antibodies to all 9 HIV-1 antigens detected in all serum/plasma were also found in 6 CVL and 4 RL samples. The other 5 CVL samples had IgG antibodies to 7 or 8 HIV-1 antigens. Of the remaining RL samples, 5 had IgG antibodies to 3-8 HIV-1 antigens and 2 had no detectable IgG antibodies to any HIV-1 antigens. IgA antibodies to 9 HIV-1 antigens were found in 4 serum/plasma samples, but none of the CVL or RL samples tested had IgA specific for all 9 HIV-1 proteins. The remaining 7 serum/plasma and all 11 CVL samples had IgA antibodies to 2-8 HIV-1 antigens. Of the 11 RL samples, 6 had IgA antibodies to 3-5 HIV-1 antigens, 4 had IgA antibodies to 1 HIV-1 antigen and 1 had no detectable HIV-1-specific IgA antibodies to any HIV-1 antigens. These data demonstrated that the similarity of external secretions to serum/plasma samples regarding the patterns of HIV-1-specific antibodies were greater for the IgG isotype than for IgA. Moreover, both IgG and IgA HIV-1-specific antibodies presented in CVL were closely related to those in serum/plasma, indicating that Ig from circulation might contribute to the Ig pool in CVL.

Serum/plasma samples from 63 HIV-1-infected women were initially screened against three HIV-1 clade B viruses representing R5- (HIV-1_SF162 and HIV-1_YU2) and X4-tropic (HIV-1_NL4.3) strains (Table 5). Three-fold dilutions of serum/plasma samples starting at either 1:20 or 1:30 were used (range of 1:20 to 1:43,740, and 1:30 to 1:21,870, respectively), and for a few samples the dilution ranged either 1:40 to 1:87,480, or 1:180 to 1:393,660. Of the 63 samples evaluated, 21 had ability to neutralize three viruses, 21 to two viruses (19 to HIV-1_SF162 and HIV-1_NL4.3, and 2 to HIV-1_SF162 and HIV-1_YU2), and 21 to only one (20 to HIV-1_SF162, and 1 to HIV-1_NL4.3). The median of IC₅₀ titers to HIV-1_SF162, HIV-1_NL4.3, and HIV-1_YU2 were 2,422, 265, and <30, respectively. Of the 42 samples exhibiting neutralizing antibody to at least two viruses, the 19 with relatively higher IC₅₀ titers were selected and used to determine the contribution of IgA or IgG antibodies to neutralization activity (Table 7). Of the 19, 11 individuals’ corresponding CVL and RL samples were tested at 1:10 with two-fold dilutions. As expected, HIV-1-specific neutralization activity in CVL and RL samples was present at much lower levels than that in serum/plasma. The median of IC₅₀ titers to HIV-1_SF162, HIV-1_NL4.3, and HIV-1_YU2 were 23, 29, 19 among CVL samples and 28, 23, 19 among RL samples, respectively. The CVL from WARO and the RL from HUTA had the highest neutralization activity, and thus were selected to further determine the effect of Ig removal on virus neutralization (Table 9).

In addition, 5-6 samples that neutralized all three clade B viruses were selected and used to determine their ability to neutralize cross clade HIV-1. Experiments with clade C viruses, HIV-1_TV1 and HIV-1_92BR025.09, were subsequently performed. In general, neutralization activity was detectable but at relatively low levels (Table 5).

To determine whether the inhibition of virus infectivity was indeed antibody-mediated, and to identify the Ig isotypes responsible for HIV-1 neutralization, depletion of IgG and/or IgA was performed on serum/plasma samples from 19 HIV-1-infected women, and CVL and RL samples from 3 of the 19 women. Neutralization capacity was then examined in parallel on samples before and after Ig depletion. As shown in Table 6, the removal of IgA and/or IgG from samples was effective and efficient. Complete removal of IgG (>99%) from serum/plasma and CVL samples was consistently achieved. In some cases, removal of IgG from RL was performed twice on the same sample to ensure that IgG was depleted. Depending on the distribution of IgA subclasses in different body fluids and individuals [45], total IgA depletion varied, since immobilized Jacalin binds to IgA1, but not to IgA2 according to the manufacturer’s instructions. Total IgA elimination from serum/plasma samples reached a maximum of 97%, while some IgA, probably IgA2, remained in IgA-depleted CVL and RL samples. Furthermore, the high variability in the proportion of depleted IgA in CVL was probably due to the fact that CVL samples were collected irrespective of the stage of the menstrual cycle. It is well known that the levels of IgA1 and IgA2 in cervical mucus are hormonally dependent and display remarkable variation before and after ovulation [46-49].

The reduction in IC₅₀ to at least one clade B virus was observed after depleting IgA from all 19 serum/plasma samples evaluated (Table 7). Nevertheless, up to 3-fold reduction of IC₅₀ to two clade B viruses was only seen in samples from HALE and WARO. Among 19 samples, DASH apparently had the strongest IgA responses to HIV-1 ENV antigens (Table 8) but IgA antibodies were only responsible for about half of the HIV-1_SF162 neutralization, and had no effect on HIV-1_NL4.3 neutralization. IgA antibodies in serum from HAAL, which also demonstrated stronger binding to HIV-1 ENV antigens, contributed predominantly to HIV-1_SF162 neutralization, but little to HIV-1_NL4.3, and not at all to HIV-1_YU2 neutralization. In contrast, IgA antibodies in HALE contributed to more than half of neutralization activity to all three calde B viruses, although the capacity of IgA binding antibodies to HIV-1 ENV proteins was one of the weakest. These data suggested that the ability of IgA antibodies to bind HIV-1 ENV antigens or to neutralize HIV-1 viruses was not correlated. Removal of IgG resulted in a significant reduction in IC₅₀ to at least two clade B viruses; in some samples (8/19) IC₅₀ was reduced to under the detectable levels (lower than the starting dilution). Greater than 3-fold reduction in IC₅₀ to HIV-1_SF162, HIV-1_NL4.3, or HIV-1_YU2 was observed in 18, 14, and 9 samples, respectively, after IgG was removed. Taken together, these data demonstrated that both IgA and IgG antibodies in serum/plasma samples were responsible for inhibiting HIV-1 infectivity, but the neutralizing antibodies were present mainly in the IgG isotype.

As shown in Table 7, the removal of both IgG and IgA from 15/19 plasma/serum samples abolished neutralization activity to at least one clade B HIV-1. However, the inhibition of the three clade B HIV-1 infectivity remained at some levels in serum/plasma samples from 4 women (FOBI, MIUR, JOME, JODE) after both IgA and IgG were removed. Although HIV-1-infected women recruited for this study did not suppose to be on antiretroviral therapy (ART), these four individuals were reported later by the staff in 1917 clinic at UAB as ART users at the time of samples collection. Based on the pattern of IC₅₀ to three clade B viruses determined in these women, two others were identified and confirmed as well. Therefore samples from 6 HIV-1-infected women were excluded from further analysis.

Contribution of Ig to neutralize clade C viruses was also examined with samples (HAAL, HUCA, and WARO) that were available in sufficient volumes for further Ig depletion. IC₅₀ were reduced below detectable level after IgG removal, demonstrating that neutralizing antibodies to HIV-1_TV1 and HIV-1_92BR025.₀₉ were mainly IgG.

The neutralization activity detected in RL samples did not reflect antibody-mediated virus neutralization since IC₅₀ titers to the three clade B viruses did not change in samples before and after Ig depletion (Table 9). However, in CVL samples the dominance of IgG over IgA on HIV-1 neutralization activity was observed. Removal of both IgA and IgG resulted in decreases of IC₅₀ close to or below detectable levels. These data indicated that neutralization activity in CVL mainly resulted from HIV-1-specific neutralizing antibodies of the IgG isotype and to a lesser degree of the IgA isotype, and the contribution of other humoral innate factors was minimal.