Prof. L.W. Poulter, Department of Immunology, Royal Free and University College Medical School, Royal Free Campus, Rowland Hill Street, London NW3 2PF, UK. E-mail:email@example.com
This study investigates the distribution of immunocompetent cells in the ectocervix, and cytokine and immunoglobulin (Ig) levels in cervicovaginal secretions to determine whether they are altered in asymptomatic human immunodeficiency virus (HIV) infection. Ectocervical biopsies from 10 HIV+ and 10 presumed HIV-ve women were studied by immunocytochemistry. Levels of Igs in cervicovaginal secretions were quantified by radial immunodiffusion (RID) and cytokine levels by ELISA. HIV+ women had significantly increased numbers of CD8+ lymphocytes resulting in reversal of the CD4:CD8 ratio. There was a significant increase in the proportion of activated CD8+ HLA-DR+ and CD4+ HLA-DR + lymphocytes, but not in CD8+ TIA-1+ cells. The epithelium of the cervix from HIV+ subjects showed a significant increase in both numbers of macrophages (CD68+) and proportions of activated macrophages (CD68+ HLA-DR+) compared to normal. The stroma contained increased proportions of inductive (D1+) and suppressive (D1+ D7+) macrophages but a decrease in effector phagocyte (D7+) proportions and Langerhans' cells. Significantly lower tumour necrosis factor (TNF)-α levels were observed in cervicovaginal secretions from HIV+ subjects. IgG levels were 4 times higher and IgM levels twice higher in cervicovaginal secretions from HIV+ women, compared to results from normal subjects. These results suggest a response within the CD8+ cells in HIV+ women, yet these cells may have a low cytolytic capacity. The raised proportions of HLA-DR+ and D1+ CD4+ macrophages could act as antigen-presenting cells (APC) for CD4+ CD45RO+ lymphocytes, and represent a local acquired response. However, the close juxtaposition of these cells offers the potential for them to act as a local reservoir of virus and promote its proliferation. The increase of IgG over sIgA in secretions of HIV+ subjects provides evidence suggesting a dysregulation of local humoral immunity.
The immune system within the female lower genital tract (LGT) is the initial defence to sexually transmitted diseases (STD). Despite the fact that STDs represent a major public health problem on a world scale, there is only limited knowledge of the immune mechanisms active in the female LGT. Unlike the mucosa of the gut [1,2] and the lungs [3,4] where extensive investigation has described much of the immune system in these areas, immune defence in the LGT remains ill defined.
It is known that immunocompetent cells occur throughout the female genital tract with T and B-lymphocytes, natural killer (NK) cells, macrophages, dendritic cells (DCs) and granulocytes present in the fallopian tubes, uterus, cervix, vagina and vulva [5–7]. The cervical transformation zone has been shown to contain the largest number of lymphocytes in the female LGT, such that its lymphocytes, Langerhans' cells, plasma cells and endocervical columnar cells (which may be important in the transport of secretory Igs into cervical mucous) have been proposed to form part of the mucosal associated lymphoid tissue (MALT) as cervical lymphoid tissue or CLT .
DCs (Langerhans' cells) appear in equal numbers in the ectocervix, cervical transformation zone and vulva but only rarely in the vaginal mucosa . Macrophages and granulocytes have also been identified in the cervix and vagina . A secretory immune system has been demonstrated within the endocervix, comprising plasma cells containing IgA and J-chains, and epithelial cells producing secretory component [6,8]. IgG- and IgM-producing cells are also present at these sites [6,9].
Dysregulation of the immune system of the body is seen in association with HIV infection, with a fall in the CD4+ lymphocyte count [10,11] and a reversal of the CD4 : CD8 ratio  now being well recognized systemic manifestations of infection. HIV infection is associated with chronic and recurrent viral (Herpes Simplex Virus (HSV) and Human Papilloma Virus (HPV)) [13,14] and fungal (candida) infections of the female LGT , suggesting an alteration in local immunity. Furthermore, cervical dysplasia is an HIV-related condition and carcinoma of the cervix is an AIDS-defining illness . Both conditions occur more commonly in immunocompromised patients, and the rapid rate of progression of cervical intraepithelial neoplasia and its increased recurrence rate in HIV-positive women  clearly implicate altered immunity as a factor in these conditions.
Alterations in mucosal immunity in the presence of HIV infection have been clearly documented in both the gut [18,19] and the lungs . In contrast, the female LGT has not been as extensively studied. Pilot studies in this laboratory have reported distinct changes to several components of the immune repertoire at this site in association with HIV infection [21,22]. However, to this date no comprehensive study has been undertaken which relates the cellular and humoral components of the immune defence system in areas of the LGT of HIV+ women. As this area has been proposed as a possible route for vaccination [23,24], and trials of prophylactic compounds administered via the vagina are emerging , a fuller understanding of the immune system in the LGT in HIV infection is required.
This study has utilized immunohistological techniques to study tissue sections of the ectocervix as well as quantitative methods of measuring soluble components of the immune system (cytokines and Igs) present in cervicovaginal secretions. This comprehensive approach has allowed a comparison to be made between LGT immunity in HIV positive and normal subjects.
Materials and methods
Subjects HIV positive women at varying stages of disease were recruited to the study. They were all patients attending the Ian Charleson Day Centre at the Royal Free Hospital, London and were presenting for their routine annual gynaecological check up. Informed consent was obtained from all patients and the local ethics committee had approved the project. High vaginal and endocervical swabs were taken to determine the presence of the following infections; candida, trichomonas, bacterial vaginosus, gonococccus, chlamydia, herpes simplex virus and cytomegalovirus. A cervical smear was performed to exclude cervical dysplasia and a bimanual pelvic examination was done. An abnormal result on any of these investigations excluded the patient from this study. Colposcopically directed ectocervical biopsies were taken from areas of clinically normal native squamous epithelium. Ultimately, biopsies from 10 HIV+ women with no known LGT pathology were selected for further study. Each woman's age, contraceptive practice and day of cycle were recorded along with their CD4 lymphocyte count, viral load and current treatment. The control group was drawn from 30 healthy premenopausal women who had volunteered for a phase I/II study looking at the effects of a potential vaginal microbicide on the cervical epithelium. They had been recruited via advertisements in the free London papers. The women were aged between 16 and 40 years (mean age: 33.2 years), were not pregnant or lactating, had normal menstrual cycles, a normal recent cervical smear result, and a low risk of HIV infection (as determined by interview). Exclusion criteria were known HIV infection, a current STD or genital lesion, a history of postcoital bleeding, abnormal biochemistry or coagulation results (as the microbicide is an anticoagulant), allergy to heparin or sulfated polysaccharides, the use (within 30 days of enrolment) of anticoagulants, medications with antiplatelet activity, or drugs which are cleared via the reticulo-endothelial system, and the use of intravaginal products other than the study gel. The subjects were asked to refrain from sexual intercourse during microbicide use and cervical and vaginal swabs were taken to check for infection (as above). Ectocervical biopsies were taken in the proliferative phase of the menstrual cycle, prior to application of the microbicide. All biopsies were taken in the proliferative phase of the menstrual cycle. Six low risk women were recruited to provide secretions for serial Ig analysis. The mean age of this group was 21.3 years, and they were all using barrier methods of contraception. Each student was asked to collect her cervicovaginal secretions at three time-points during the menstrual cycle: in the follicular phase 3 days following the end of menstruation, at midcycle, and in the secretory phase just prior to the next period. Sample collection was performed as described below and repeated over three consecutive menstrual cycles. Cervicovaginal secretions for the study of Igs and cytokines (see below) were obtained from a further 10 HIV+ women and compared with a control group of another 10 low-risk women. Again, these women were within the reproductive age group. Exclusion criteria were a current STD or genital lesion, a history of postcoital bleeding, the use of hormonal contraception, application of any intravaginal products and any contra-indication to the use of tampons such as a history of toxic shock syndrome. Subjects were asked to refrain from sexual intercourse for 72 h prior to sample collection.
Preparation of biopsy specimens Biopsy specimens were covered in OCT medium (BDH Chemicals Ltd, Poole, UK), mounted on cork discs and snap frozen in isopentane cooled in a bath of liquid nitrogen. These were stored in liquid nitrogen. Six micron cryostat sections were cut, air-dried, fixed in chloroform/acetone [1:1] for 10 min at room temperature, wrapped in cling-film, and stored at − 20 °C until use. Sections of human palatine tonsil were prepared as above and used as monoclonal antibody (MoAb) reagent controls. Sections from all samples were stained with 0.1% toluidine blue and haematoxylin and eosin (both from Sigma Diagnostics, St Louis, MO, USA) in order to reveal tissue morphology and integrity.
Immunohistology The characteristics of the MoAb and polyclonal antisera used in this study are documented in Table 1. The study employed indirect immunoperoxidase, immunofluorescence, alkaline phosphatase antialkaline phosphatase and biotin/streptavidin alkaline phosphatase methods.
Table 1. List of MoAbs used in this study
IgG1 and IgG2a
Class II MHC restricted T-cell subset
T cells and some B cells
IgM and IgG
Class I MHC restricted T-cell subset
NK cells, granulocytes, monocytes
Activated T cells, costimulatory signal
Activated T cells, plasma cells, monocytes
Primed T cells
NK cells, B cells, T-cell subsets
Macrophages, monocytes, Langerhans' cells
Cytolytic effector cells
B-mix: CD19, CD22, CD37
Pan B cells
R & D
Immunoperoxidase staining To identify T and B-lymphocytes, Langerhans' cells and tissue macrophages a standard indirect immunoperoxidase method was used . Sections were counter-stained with Harris's haematoxylin (Sigma Diagnostics) and mounted in DPX (BDH Laboratory Supplies, Poole, UK). Three control preparations were employed. Sections of human tonsil, in which the distribution and pattern of staining could be tested against tissue architecture, were used as positive controls in each experiment. Control incubations to detect background staining were performed on sections of each ectocervical sample, omitting the primary antibody. Thirdly, isotype specificity was confirmed on tonsil sections by comparison to staining with irrelevant MoAbs of the same isotype as the MoAbs used. Normal serum from appropriate species was used as a control to the heterologous antisera. Within the context of this experiment, a large panel of MoAbs, of both IgG and IgM classes, was used. They represent irrelevant isotype controls for each other, and no nonspecific staining was seen with any of these antibodies. The only evidence of endogenous peroxidase was associated with small numbers of neutrophils within the tissue. This posed no problem to our methods of quantification and there was therefore no need to use hydrogen peroxide to block endogenous peroxidase activity.
Immunoperoxidase analysis The presence and distribution of immunoperoxidase stained cells was determined using an image analysis system (Seescan, Cambridge, UK) at × 40 magnification. Areas of epithelium and stroma were outlined, and positive cells within each measured framed area were counted. Five fields each were counted from epithelium and stroma. In the stroma, positive cells were counted to a depth of 15 cells below the epithelial basement membrane. Cell numbers are expressed as cells/unit area; the unit area being 104 micron2.
Immunofluorescence staining Double immunofluorescence staining  was used to determine the relative proportions of lymphocyte and macrophage subsets. Lymphocyte subset phenotype was determined using the appropriate MoAbs combinations (see Results). The MoAb RFD1, RFD7 and CD68 were used in order to identify the macrophage subsets. The first layer was applied using combinations of one IgG and one IgM class MoAb at predetermined optimal dilution and the samples incubated in a moist chamber for 45 min at room temperature. After rinsing in phosphate-buffered saline (PBS), a combination of goat antimouse IgG conjugated to fluorescein isothiocyanate (FITC) and goat antimouse IgM conjugated to tetramethylrhodamine isothiocyanate (TRITC) (both from Southern Biotechnology, Birmingham, AL, USA), were applied at pretitrated optimal concentrations. The slides were then incubated in the dark for a further 40 min, rinsed in PBS, fixed in 4% paraformaldehyde and mounted in Citifluor (AF1; Citifluor Products, Canterbury, UK). Background staining was identified by comparison with negative control samples of cervical tissue from which the specific MoAbs had been omitted. Positive specificity controls were prepared using sections of human palatine tonsil.
Immunofluorescent counts Immunofluorescent counting was performed on a fluorescence microscope (Zeiss, Oberkochen, Germany) at × 40 magnification with selective filters for FITC and TRITC. Cells were positive either for FITC, TRITC or both. CD4+ and CD8+ cells were counted in the epithelium and stroma of either three high power fields (hpf) or up to a total of 100 cells. The ratio was recorded as:
In order to measure the CD8+ subsets (CD8/CD5, CD8/CD45RO, CD8/TIA1), cells staining positively for both CD8 and the specified subset were counted and expressed as a percentage of CD8+ cells. For example for CD8/CD5 staining:
In order to determine relative proportions of macrophage subsets, three high power fields or 100 cells were observed on each section. Using the FITC and TRITC barrier filters, individual cells were scored as: RFD1 + (red), RFD7 + (green), and RFD1 + RFD7 + (red and green). Relative proportions were then calculated with the formula:
Proportions of Langerhans' cells expressing RFD1 antigen were also quantified using a similar approach with a CD1, RFD1 MoAb combination.
Tissue cytokine analysis by the Biotin/Streptavidin method Tissue sections were analyzed for interferon(IFN)-γ, TNF-α, T-cell growth factor-β1 (TGF-β1), IL-1β, interleukin(IL)-4 and IL-10 by the biotin-streptavidin method . Freshly cut cryostat sections were air-dried for 2 h, ringed with polysiloxane and fixed in cooled methanol : acetone at − 20 °C for 10 min. After rinsing in PBS at room temperature, the primary anticytokine antibody appropriately diluted in PBS with 0.5% bovine serum albumin (BSA) (Albumin, bovine fraction, Sigma Chemicals) was applied and the sections incubated overnight in a moist covered chamber at 4 °C. The slides were then rinsed in fresh Tris-buffered saline (TBS) (BDH) at pH 7.6 and then incubated further for 1 h with a compatible biotinylated second layer diluted to 1 : 100 in PBS–BSA. After rinsing in fresh TBS, a streptavidin-alkaline phosphatase conjugated third layer (Vector Laboratories) diluted in PBS–BSA was applied to the sections, which were then incubated in a moist covered chamber for 1 h at room temperature. Sections were again rinsed in fresh TBS and the reaction was developed by a 15 min application of filtered substrate solution (5 mg Napthol ASBI Phosphate (Sigma Chemicals), 10 ml Tris HCL at pH 8.0 (BDH Laboratories), 200 µl dimethylformamide (Sigma Chemicals), 10 mg Fast Red (TR) and 10 drops of Levamisole (Sigma Chemicals) added last). Sections were then washed in tap water and counterstained with Mayer's haematoxylin before mounting in PBS glycerol (9 : 1) (PBS from Oxoid Ltd, Basingstoke, Hampshire, UK and glycerol from BDH Laboratories). The controls were performed on ectocervical sections as above using the streptavidin/biotin second and third layers alone. Isotype specificity was confirmed by comparison to staining with an irrelevant IgG1 MoAb on cervical sections or by the use of sheep or rabbit serum not containing the relevant antibodies. The distribution of cytokines in the sections was recorded following microscopic examination.
Cervicovaginal fluid In order to collect undiluted cervicovaginal secretions, women were asked to wear a tampon (Tampax R, Tambrands Ltd, Hants, UK) for 6 h or overnight. On removal it was placed in a sterile 50 ml polypropylene tube (Becton Dickinson, San Jose, CA, USA). This was then centrifuged at 450 × g for 30 min and the fluid extracted aliquoted and frozen at − 20 °C until further investigation. This was performed in the secretory phase of the menstrual cycle, just prior to the next expected period, as previous studies have shown that maximal volumes of secretions could be obtained at this time. Cervicovaginal secretions have been similarly sampled for quantification of Ig levels . In order to ensure that this method of collection of cervicovaginal fluids did not cause retention of Ig molecules on the tampon, thereby resulting in inaccuracy in quantification of Ig levels, the following was done. Known volumes of calibrators of known concentrations were applied to the tampons. These were centrifuged as described, and the volume of fluid extracted was measured. The Ig concentration of each test sample was then measured by the method appropriate to the Ig. There was a consistent loss of sample volume of between 30 and 50%, but the Ig concentrations were always absolutely accurately measured. Thus, this method of fluid collection can be used to accurately measure Ig concentrations in cervicovaginal fluid but not to measure fluid volumes.
Ig and cytokine analysis in cervicovaginal fluid Ig levels were determined using radial immunodiffusion kits for IgG, IgM and secretory IgA (The Binding Site, Birmingham, UK). Ring diameters were measured on completion of ring development, and the values obtained correlated with concentrations on the standard tables relating ring diameters to concentrations for IgG and IgA. A standard curve was plotted for IgM of ring diameter2 against concentration, and IgM levels read off this graph. Levels of IL-1β, TNF-α and TGF-β1 were quantified in cervicovaginal secretions by the ELISA method using standard kits from R & D Systems, Minneapolis, USA.
Statistical analysis Unless otherwise stated, the median values and range are given. The Mann–Whitney U-test was used to compare values from the HIV+ and HIV-ve samples. A probability value (P) of < 0.05 was taken as indicating statistical significance. All tests of significance were two-tailed.
Characteristics of study population
All the volunteers were within the reproductive age group. Of the HIV+ women, five were of black African origin, four were white Europeans and one was Oriental (Thai), whereas the low-risk group were mainly of white European extraction. All biopsies were taken in the proliferative phase of the menstrual cycle.
Two of the HIV+ cohort were receiving anti-retroviral treatment. They were on a combination of the nucleoside analogues lamivudine (3TC) and stavudine (d4T), and a protease inhibitor, either ritonavir or saquinavir. One woman had been diagnozed to have AIDS.
Of the HIV+ women who provided samples of cervicovaginal secretions, four were of black African origin, five were white European and one was Oriental (Thai), whereas of the low-risk group the one woman who was not white European was South Asian. Five HIV+ women were on anti-retroviral therapy and five were not on any treatment. Two women had been diagnozed to have AIDS. These results are summarized in Tables 2A and 2B.
Table 2A. Characteristics of study population - Cervical biopsy
CD4 count × 106/l median and range
Viral load copies/ml median and range
270 (< 400–403 000)
Table 2B. Characteristics of study population - cervical secretions
CD4 count × 106/l median and range
Viral load copies/ml median and range
2200 (< 400–403 000)
Comparisons of the age distribution, CD4+ T-cell counts and viral load levels of the two cohorts of HIV+ women studied (those who provided cervical biopsies and those who provided secretions), did not reveal any significant differences. Both cohorts of HIV+ women reflected the racial mix within the Ian Charleson Day Centre (ICDC) clinic population. Sixty % of the women attending the ICDC are of black African origin and for cultural reasons use sanitary pads rather than tampons. It was therefore not possible to recruit the same HIV+ women as had provided cervical biopsies. Only two samples of cervicovaginal secretions were obtained from the same women who had provided cervical tissue, the rest being provided by women of other racial groups.
Haematoxylin and eosin staining confirmed normal histology and appropriate orientation of all tissue sections. No histological abnormalities were detected in any tissue and no fundamental histological difference was seen between any of the specimens studied.
In all tissue sections studied, the variability between the selected areas of any one specimen was far less than the variability between specimens. For example, when five areas of epithelium from a single HIV+ specimen stained for CD8+ T cells were analyzed, the median number of cells/unit area was 7.1 and the range 6.1–7.7, compared to a median of 4.5 and range of 2.4–8.2 cells/unit area when 10 specimens were analyzed (Fig. 1). A similar pattern was observed when five areas of stroma from a single low risk specimen stained for CD8+ T cells were analyzed, the median number of cells/unit area was 2.6 and the range 2.4–2.8. This was in comparison to a median of 2.4 and range of 0.5–5.3 cells/unit area when 10 similarly stained specimens were analyzed.
Each group of samples was similarly analyzed, with a single sample from within a group being randomly selected for comparison with the rest of the group. A similar pattern of variability was observed regardless of whether immunoperoxidase or double immunofluorescence methods were used. All results given quote the median and range throughout the group, so reflecting the greatest variability seen.
T cells were present in both the epithelium and the stroma in all samples; no difference in total T-cell numbers was detected when the HIV+ and low-risk groups were compared (data not shown). A significantly reduced CD4 : CD8 subset ratio was observed in the HIV+ group, both in the cervical stroma (0.52 HIV+ v 2.11 low risk) and in the epithelium (0.49 HIV+ v 1.60 low risk), P < 0.01 in both cases (Fig. 2). This change was associated with an overall increase in numbers of CD8+ cells in the HIV+ samples (Fig. 3). An example of double immunofluorescence for lymphocyte subsets is shown in Fig. 4(A).
In the stroma, median numbers of CD8+ cells were 5.66 cells/unit area compared to 1.67 cells/unit area in low risks (P < 0.05). In the epithelium, the incidence of CD8+ cells was 4.47 cells/unit area in the HIV+ samples compared to 2.36 cells/unit area in the low risk groups (P < 0.05),(Fig. 3). The percentage of CD8+ CD5+ cells was significantly increased in the HIV+ subjects (Fig. 5) compared to the low risk groups. This applied to both the stroma (79.1 HIV+ v 53.5 low risk) and in the epithelium (70.0 HIV+ v 47.5 low risk).
In the stroma and epithelium of the low-risk group, over 90% of CD4+ cells also expressed CD45RO. Investigation of the CD8+ cells revealed that some 59–66% of this population were CD45RO+ , and a similar proportion also expressed TIA-1. This was true both for HIV+ and low-risk samples (data not shown). The proportions of both CD4+ and CD8+ T cells expressing HLA-DR were raised in the samples from HIV+ compared to low-risk subjects. This was true for T cells in both the epithelium and the stroma (see Table 3). In some samples, clusters of CD4+/CD45RO+ T cells associated with DCs or macrophages were seen in the upper layers of the stroma, just below the basement membrane of the ectocervical epithelium (Fig. 4B).
Table 3. Proportions of CD4 and CD8 cells expressing HLA-DR
HIV + Median (range)
Control Median (range)
P-value (Mann– Whitney)
When the stromal CD8+ subsets were tested for evidence of activation, however, no difference between the HIV+ and low-risk subjects was seen in terms of the proportions of these cells expressing either CD28 orCD38 (Fig. 6A+B).
B-cell and NK-cell distribution
No B-lymphocytes (CD19 or CD20 positive cells) or plasma cells (RFD6+) were observed in any samples. No NK (CD11B+) cells were observed in the epithelium of any sample, although these cells were present in small numbers in the stroma of both the HIV+ and the low-risk subjects (1.5 HIV+ v 1.2 low risk, P > 0.05).
Within the stroma of the cervix no difference in the overall number of macrophages (CD68+ cells) was seen between the HIV+ and the low-risk samples (Fig. 7A). When the subset analysis was performed, however, it was revealed that the HIV infection was associated with significant changes to the balance of macrophage subsets within the tissue. In low risk samples the vast majority of stromal macrophages were D7+ effector cells (median 90.5%), with only small proportions of inductive cells (D1 + 3.5%) and suppressor cells (D1 + D7 + 5.5%) (Fig. 7B). In the HIV+ samples, however, only 7% of the macrophages were D7+ effector cells, while 57% were D1 + inductive cells and 31% were D1 + D7 + suppressive cells (Fig. 7C).
Within the epithelium there were more CD68+ cells in the HIV+ samples (median HIV + 2.7 cells/unit area, low risk 1.02 cells/unit area, P < 0.05), Fig. 8(A). An analysis of the macrophage subsets within the epithelium revealed that virtually all cells within the epithelium were D1 + D7-inductive cells with either none or very few D1/D7+ expressing cells. This was true both of the HIV+ and low-risk subjects (Fig. 8B). This observation was reflected in the proportion of CD68+ cells expressing HLA-DR which was significantly higher than in low-risk samples where only 8% of the CD68+ cells within the epithelium expressed HLA-DR (Fig. 8C). Interestingly, both in the HIV+ and the low-risk subjects, a relatively high proportion of D1 + APC were found to express CD4 (HIV + stroma 91.5%, epithelium 100% and low-risk stroma 87.5%, epithelium 96.6%).
Immunoperoxidase staining identified fewer CD1a + Langerhans' cells (LC) in the epithelial layer of HIV+ compared to low-risk women (median HIV+ 0.45 cells/unit area and low risk 2.36 cells/unit area). Of these cells, 71% of CD1a + cells in HIV+ subjects also expressed D1, while 76% of CD1a + cells were D1 + in the low-risk group. Immunofluorescent staining, however, identified CD1a + cells in the stroma as well. Here low-risk subjects showed virtually no dual staining CD1a + D1 + Langerhans' cells in the stroma, while all CD1a + cells expressed D1 in samples from HIV+ women, albeit in very small numbers (Figs 9 and 4C). There was no difference between the numbers of these cells in the epithelial layers of the two groups (Fig. 9).
Tissue cytokine staining
The distribution of tissue-associated cytokines in ectocervical tissue taken from HIV+ subjects was similar to that seen in the low-risk group. The basal layer of the epithelium expressed TNF-α, but no staining was seen in association with the basement membrane. Positively staining macrophage-like cells could be identified both just beneath the epithelial basement membrane as well as scattered throughout the stroma. The endothelial cells of the stromal vessels also showed positivity for TNF-α(Fig. 4D).
IFN-γ was distributed throughout the epithelium and was strongly associated with the epithelial basement membrane. It was also present on endothelial cells and some randomly distributed lymphoid and macrophage-like cells in the stroma. IL-10 was expressed diffusely within the epithelium, on small numbers of lymphoid and macrophage-like cells within the stroma and in association with stromal perivascular cells. No positivity for IL-4 or TGF-β1 was found in tissue sections from either HIV+ or low-risk samples, and only endothelial cells showed any degree of expression of IL-1β.
The cytokines TNF-α, TGF-β1 and IL-1β were quantified in cervicovaginal secretions collected during the secretory phase of the cycle (Table 4). In contrast to tissue-associated cytokines, differences between the HIV+ and low-risk groups were seen with these parameters. In HIV+ women TNF-α levels were significantly reduced from a median of 109 pg/ml (range 25–900 pg/ml) to 30 pg/ml (range 15–555 pg/ml) in HIV+ subjects compared to low risks (Table 4). IL-1β occurred at a median level of at 285 pg/ml compared to 370 pg/ml in low-risk subjects, but this difference did not reach statistical significance, whilst levels of TGF-β1 were very similar between the two groups.
Table 4. Levels of cytokines in cervicovaginal secretions of HIV+ and low risk women
Low risk (Mann–Whitney)
370, n = 6
285, n = 10
1.3-fold increase in HIV+ women
196, n = 5
200, n = 6
30, n = 9
105, n = 10
3.5-fold decrease in HIV+ women
IgG, sIgA and IgM were quantified during the secretory phase of the menstrual cycle and expressed in mg/l. The IgG levels were increased 1.4-fold (from 383 to 555 mg/l) and IgM levels increased two-fold (from 1.4 to 2.8 mg/l) in HIV+ women compared to low risks. There was minimal change in sIgA levels (Table 5).
Table 5. Levels of Ig in cervicovaginal secretions of HIV+ and low-risk women
As all the subjects did not yield adequate secretions for every Ig and cytokine analysis to be performed, differing numbers of sample results appear in each group.
555, n = 8
383, n = 12
1.4-fold increase in HIV+ women
301, n = 10
298, n = 10
2.8, n = 8
1.4, n = 13
2-fold increase in HIV+ women
Samples of cervicovaginal secretions obtained from six low-risk subjects were tested for levels of IgG, secretory-IgA and IgM at three time points of the menstrual cycle. Within each cycle, levels of IgG and secretory IgA tended to be higher in the follicular phase (IgG 574 mg/l, IgA 373 mg/l) and secretory phases (IgG 459 mg/l, IgA 410 mg/l) with a fall in Ig levels occurring at mid-cycle (IgG 399 mg/l, IgA 258 mg/l), although these differences were not statistically significant (P > 0.05). Levels of IgM were very low, showed only subtle variations and did not appear to follow this pattern (follicular phase 2.1 mg/l, mid-cycle 1.4 mg/l, secretory phase 1.4 mg/l). Analysis of data generated from any one individual did not reveal a consistent pattern between cycles. It was not possible to obtain adequate secretions for full analysis at each time point, so data from all three cycles have been combined to calculate the median and ranges of IgG, IgA and IgM concentrations at each phase of the cycle, Fig. 10. Although no significant differences in Ig levels across the cycle were noted on group analysis, a trend was demonstrated, as described above.
This study describes the distribution of immunocompetent cells in the cervix, and quantifies levels of IgG, IgM and sIgA as well as the cytokines TNF-α, TGF-β, and IL-1β in the cervicovaginal fluid of low-risk and HIV+ women. The mucosa of the female LGT in HIV+ women exhibits fundamental changes in lymphocyte and macrophage populations and in Ig and cytokine production even in the absence of LGT symptoms. Similar mucosal derangements are seen in the gut [18,30] and lung [31,32] of HIV-infected patients. Unlike the gut and lung, however, which are major sites of opportunistic infection [33–35], the LGT in HIV+ women does not appear to be as frequently affected .
The expected reversal of the CD4 : CD8 ratio is observed in the female LGT  as at other mucosal sites [37,38]. T-cell numbers are well maintained, as despite a fall in CD4+ cell numbers, there is a concomitant rise in CD8+ lymphocytes. This is in keeping with systemic events . A pulmonary lymphocytic infiltration has been demonstrated to varying degrees in lung mucosa [32,40]. In the gut no clear pattern has emerged and intraepithelial lymphocyte counts have been reported as increased , decreased  or unchanged  in the presence of HIV infection. However, the elevated numbers of CD8+/CD5+ cells seen in the ectocervix of HIV+ women (80% compared to 50% in low risks) demonstrates the active recruitment of nonresident CD8+ lymphocytes rather than an expansion of the resident intraepithelial population.
Increased proportions of these CD8+ cells are activated, a phenomenon that has also been demonstrated at other mucosal surfaces [30,32]. Relatively high proportions of activated CD8+ lymphocytes are present in the ectocervix, even in the absence of HIV infection. However, as elsewhere, these cells do not appear to be capable of mounting an adequate immune response to the virus. The MoAb TIA-1 identifies cytotoxic granules in cells. Although the majority of CD8+ T cells express TIA-1 in low-risk women, the lack of a significant increase in the proportions of CD8+ TIA-1+ cells in HIV+ women provides circumstantial evidence that there is no increase in the cytotoxic capacity of these CD8+ cells.
The CD8+ lymphocytes are dependent on CD4+ cell help, such that a depletion in CD4+ lymphocytes may contribute to the subsequent loss of the CD8+ cytotoxic function observed . As in the lung [32,38] and peripheral blood , an increased proportion of activated CD4+ HLA-DR+ cells is seen in the female LGT despite declining CD4 T-cell numbers, suggesting that this mucosal immune system of HIV+ women exists in a more activated state than that in seronegative individuals.
The raised proportions of activated macrophages (CD68+ HLA-DR+) in both HIV+ and low-risk women suggest that local mechanisms of T-cell activation are in place. However, the presence of high proportions of D1 + CD4+ cells, indicates the presence of a high number of DCs known to be susceptible to HIV infection [44,45].
Although absolute numbers of Langerhans' cells (CD1a+) are decreased in HIV infection, large proportions of LC in the epithelium are CD1a + D1 + 0. Similar proportions of CD1aD1 + cells are seen in low-risk women. D1 positivity is associated with antigen presentation, and in quiescent tissue Langerhans' cells are D1 negative . The expression of RFD1 by Langerhans' cells therefore indicates the active involvement in antigen presentation , despite a reduction in total Langerhans' cell numbers.
Whilst the gut and lung show a preponderance of suppressive macrophages (D1 + D7 +) [3,47], the majority of macrophages in the cervical epithelium of low-risk women are of the inducer type (D1+ D7-), and in the stroma are mature phagocytes (D1-D7+). The increased proportions of D1 + D7 + macrophages, which are considered capable of suppressing T-cell responses , in association with a decrease in phagocytic (D7 +) macrophages provide further evidence for a situation whereby mucosal responses to HIV are compromised. Following HIV infection, an inverse relationship between falling peripheral blood CD4+ lymphocyte levels and elevated numbers of suppressive macrophages has been observed in the lung [Beckles, personal communication]. Although this study does not address this issue, a similar relationship seems possible in the female LGT mucosa.
Based on studies of Simian Immunodeficiency Virus (SIV), two patterns of infection have been postulated. Firstly that HIV-1 infects DC or macrophages [49,50], which then convey infection to lymphoid tissues where CD4+ T cells become infected. Secondly, that macrophages, DC and CD4+ T cells can become productively infected at the portal of entry, with the majority of infected cells (> 80%) being CD4+ T cells . DCs are susceptible to HIV through their expression of CD4 antigen [44,45,52].
In some samples, clusters of CD4+ CD45RO + T cells were seen associated with DCs or macrophages in the upper layers of the stroma, just below the basement membrane of the ectocervical epithelium. Activated CD4+ CD45RO+ T cells are targets for HIV and are highly susceptible to this infection [53–55]. If macrophages or DC carry or become infected with HIV, it is possible that they could act to transmit the virus to these highly activated CD4+ lymphocytes. Furthermore, as HIV-1 is thought to cross mucosal barriers by infecting macrophages, DC and CD4+ T cells at the portal of entry, it is possible that this arrangement might allow the maximal opportunity for the cellular transmission of infection.
Despite a lack of B cells and plasma cells in the ectocervix, IgG, IgM and sIgA have been identified in cervicovaginal secretions. The minimal increase in sIgA compared to IgG and IgM levels seen in HIV+ subjects implies an impairment of mucosal sIgA production. The IgG and IgM form part of the transudate that contributes to cervicovaginal secretions. This overproduction of IgG could also result from systemic or local opportunistic infections, which occur with greater frequency in HIV+ than in HIV-ve women . Interestingly, IgG appears to be the dominant Ig in cervicovaginal fluid, both in low risk and HIV+ women.
This observation supports the finding of others [57,58], and suggests the presence of a responsive immune system at this site. Increased levels of IgG have also been identified in blood  of HIV+ women compared to low-risk subjects. Similar patterns of Ig appear in the lung  and the female LGT , with increased IgG compared to static or reduced sIgA levels. Studies on intestinal mucosa have suggested that this unexpected predominance of IgG over IgA may result from abnormal Ig production in the mucosa rather than from serum leakage .
TNF-α is produced by a wide variety of cells, but its main producers are activated mononuclear phagocytes, T cells and B cells. It appears to be involved in the defence against intracellular pathogens, and its cytotoxic activity is directed specifically against virus-infected cells. Raised levels of TNF-α have been demonstrated in the peripheral blood of HIV+ patients. Belec et al. also showed increased levels of this pro-inflammatory cytokine in both blood and cervicovaginal washings of HIV+ women . The results of our study, however, showed a definite fall in TNF-α levels in the lower genital tract of the HIV+ group. No data was obtained from peripheral blood. Therefore, while the current results may indicate a unique local phenomenon, this remains a speculation.
The normal female LGT possesses a reactive immune system with a high proportion of primed, activated cells. This balance is altered even in an asymptomatic HIV infection, where although a seemingly greater degree of activation exists but this is not accompanied by an increase in cytolytic activity. Indeed, the juxtaposition of activated macrophages and CD4+ T cells in subepithelial clusters creates a situation with the potential to facilitate the sexual transmission of HIV via the mucosa of the female LGT.
This study was supported by a grant to L.W. Poulter from the Medical Research Council (number G9715680). The authors gratefully acknowledge the co-operation and advice given by Professor A.B. MacLean of the Academic Department of Obstetrics and Gynaecology, Royal Free and University College Medical School, London, UK.
This work was carried out at the Royal Free and University College Medical School, Royal Free Campus, London, UK.