Influence of the mucosal epithelium microenvironment on Langerhans cells: Implications for the development of squamous intraepithelial lesions of the cervix
Article first published online: 5 NOV 2001
Copyright © 2002 Wiley-Liss, Inc.
International Journal of Cancer
Volume 97, Issue 5, pages 654–659, 10 February 2002
How to Cite
Giannini, S. L., Hubert, P., Doyen, J., Boniver, J. and Delvenne, P. (2002), Influence of the mucosal epithelium microenvironment on Langerhans cells: Implications for the development of squamous intraepithelial lesions of the cervix. Int. J. Cancer, 97: 654–659. doi: 10.1002/ijc.10084
- Issue published online: 8 JAN 2002
- Article first published online: 5 NOV 2001
- Manuscript Accepted: 27 AUG 2001
- Manuscript Revised: 9 JUL 2001
- Manuscript Received: 21 FEB 2001
- Belgian Fund for Scientific Research, The Centre de Recherche Interuniversitaire en Vaccinologie, Walloon Regions
- GlaxoSmithKline Biologicals
- CentreAnticancéreux près l'Université de Liège
- EU. Grant Number: BIO4-CT98-0097
- Langerhans cells (LC);
- transformation zone (TZ);
- squamous intraepithelial lesions (SILs);
- human papillomavirus (HPV);
- local immunity
We have addressed the notion that the initiation and progression of human papillomavirus associated cancer of the uterine cervix are associated with alterations of Langerhans cells (LC) within the mucosal squamous epithelium. Since the transformation zone (TZ) of the cervix is the site where the majority of squamous intraepithelial lesions (SIL) are initiated, in contrast to the exocervix, we decided to investigate the influence of the local microenvironment within the TZ on the function and density of LC. We show that the TZ is associated with a significant reduction in the density of immature LC (CD1a/LAG) compared to the exocervix. In contrast, the development of SILs is attributed with a relative increased density of immature LC, compared to the TZ. Furthermore, we show that this variability in LC density is correlated with a differential expression of TNFα and MIP3α within the micro-environment of the TZ and SILs. Both TZ and SIL epithelium-derived LC, in the presence of allogeneic PBMC, induced lower levels of proliferation and IL2 production and higher levels of the immunosuppressive cytokine IL10 in comparison to the exocervix. Nevertheless, the epithelium-derived LC in SILs exhibits a reduction in their functional activity, relative to the TZ. Together our studies suggest that the immunosurveillance within the epithelium of the TZ may be intrinsically perturbed due to the altered expression of chemokines/cytokines and the concomitant diminished density of LC. Furthermore, following HPV infection and the development of SILs, the function of LC may be further incapacitated by viral associated mechanisms. © 2001 Wiley-Liss, Inc.
The chronic infection of keratinocytes of the uterine cervix by the human papillomavirus (HPV) is associated with the development of cervical cancer. Despite the evidence that HPV is strongly implicated as the causative agent in the etiology of cervical cancer and its precursors (SIL, squamous intraepithelial lesion), HPV infection alone is not sufficient for cancer development.1 The role of the intrinsic immunity in controlling HPV infection and the subsequent development of SILs is shown indirectly by the increased frequency of HPV-associated lesions in patients with depressed cell-mediated immunity.2, 3 For the most part, the development of SIL and/or cervical cancer is preferentially associated with a local type II (IL4/IL6) and/or immunosuppressive (IL 10) cytokine pattern,4–7 not a cell-mediated type I immune response,which is more appropriate for tumor immunity. Given that immune responses in mucosal sites are frequently dominated by a “default” type 2 response,8, 9 it is essential to consider mechanisms that contribute to this predisposition in order to design vaccine strategies to promote a localized cell-mediated immunity.
The squamous epithelium of the cervix is composed mainly of keratinocytes, the primary target of HPV and a type of immature dendritic cell (DC), the Langerhans cells (LC), which are important for the immunosurveillance of the squamous epithelium. Hence, the intimate contact at the site of HPV infection between the target cells (keratinocytes) and the antigen presenting cells (LC) is potentially important for the functional activation, differentiation and chemotaxis of LC. Importantly, keratinocytes are capable of producing a large array of cytokines (e.g., GM-CSF, IL1β, TNFα and IL10) and/or chemokines (e.g., MIP3α and RANTES) and β-defensins all of which can importantly influence the migration, activation and/or differentiation potential of LC/DC.10–14 The production profile of these cytokines/chemokines is most likely influenced by the complex differentiation state of the keratinocytes and thus has the potential to be altered following tumorigenesis.15
Importantly, a substantial majority (∼87%) of cervical SILs and cancers develop within a specific microenvironment of the cervix, the transformation zone (TZ), where the glandular endocervical epithelium is transformed progressively into a mature squamous epithelium, a process called metaplasia.16 This implies that exogenous or endogenous factors specific to the anatomical milieu of the TZ may be conducive to SIL and cancer development. The classical explanation for the increased susceptibility to HPV infection and associated SILs is based on the proposed “mechanical accessibility”of the virus to target basal epithelial cells. Nevertheless, we and others have suggested that other factors may contribute to the sensitivity of the TZ to SIL development.6, 17, 18 For example, we have observed that the density of LC is significantly reduced and the immunosuppressive cytokine IL10 is more frequently expressed in the microenvironment of TZ in comparison to the exocervix.5, 6, 17 To the contrary, other groups have reported that the density of LC in the TZ is equal to that of the exocervix.19, 20
As an approach to understanding the factors involved in the initiation and progression of SILs, we sought to use a variety of methods to assess the differential density and the antigen presenting function of LC derived from TZ and SILs. Importantly, for these studies, we used paired biopsy specimens in order to control for inter-individual differences in LC density and function. Moreover, we evaluated the expression of a chemokine (MIP3α) and cytokines (TNFα and IL10) that play a fundamental role in the control of LC density and function.
MATERIAL AND METHODS
Normal biopsy material was obtained from women undergoing a routine hysterectomy (HPV−) and biopsies from women with SIL (HPV+) were obtained before surgical procedures. Biopsies (2–5 mm2) were either immediately frozen in Tissue-Tek (Miles, IN) and stored at −80°C or placed in DMEM with Gentamicin/Fungizone (GIBCO/BRL, Merelbeke, Belgium) for transport. For the frozen specimens, the diagnosis was confirmed and assigned to 1 of 4 categories based on the histologic findings after hematoxylin and eosin staining: (i) normal exocervix from healthy women; (ii) transformation zone from healthy women; (iii) low-grade (LG) SIL including condyloma and CIN I (Cervical Intra-epithelial neoplasia); and (iv) high-grade (HG)SIL, including CIN II and III. This study protocol was approved by Ethics Committee of the University Hospital of Liège. All biopsies were screened for HPV DNA by PCR.
Crude DNA extracts were prepared from all biopsies for the detection of HPV DNA by digesting tissue sections or cells with proteinase K (1 mg/ml) overnight (Boehringer Mannheim, Mannheim, Germany). PCR of DNA was carried out using PCRMaster Mix (Boehringer Mannheim) with a standard aliquot of the DNA preparations. The β-actin and the L1 HPV genes were amplified by PCR for each sample using published oligonucleotide sequences.21 The PCR products were analysed on ethidium bromide stained agarose gels (1.8%).
Eight micrometer frozen biopsy sections of CD1a (DAKO, Glostrup, Denmark), MIP3α (PeproTech, London, UK) and LAG (gift from Dr. Yoneda, Japan) were fixed in acetone and incubated in 0.3% H2O2 to block endogenous enzymatic activity. The sections were sequentially incubated with the primary antibody or isotype control followed by a biotinylated rat anti-mouse for CD1a, goat anti-rabbit for MIP3α and mouse anti-rat antibody for LAG. The staining buffer (PBS/2% BSA) for MIP3α included 0.1% saponin.The antibody complex was visualized using Strep-HRP (Vector Stain, CA) and DAB (Sigma, Bornem, Belgium) and counterstained with methylene blue.
Biopsies (2–5 mm2) were incubated in Dispase II (2.4 U/ml, Boehringer) for 45 min. Using fine forceps, the epithelium was peeled off, washed extensively and then incubated (37°C, 5% CO2) with media (Ham F12/DMEM, 2 mM glutamine, 1 mM sodium pyruvate, 50 μg/ml Gentamicin, 0.75 μg/ml [GIBCO/BRL], 5% human serum AB, 0.4 mg/ml hydrocortisone [Sigma], 2 ng/ml EGF [Boehringer Mannheim] and GM-CSF 10 ng/ml (gift from Novartis, Switzerland) for 48 hr. Following filtration with blutex, the remaining tissue fragments were digested using a combination of 0.03% trypsin and 0.02% DNAse (Boehringer Mannheim). Yields range between 0.05 and 0.2 × 106 total cells per biopsy.
Mixed epithelial lymphocyte respone assay (MELR)
Epithelium-derived cells (0.5–1 × 105) (including ∼500–1000 LC) were used as APCs (irradiated 1,500 rads, Stabilivolt Siemens) with equal numbers of peripheral blood mononuclear cells (PBMC), as responder cells. The PBMC were isolated from buffy coats of healthy donors (Lymphoprep, Nycomed, Oslo, Norway). The PBMC and irradiated epithelial cells were incubated in RPMI 1640 (2 mM glutamine, 1 mM sodium pyruvate, 50 μg/ml Gentamicin, 0.75 μg/ml Fungizone [GIBCO/BRL] and 5% human serum AB) in 96-well round bottom plates at 37°C in 5% CO2. Proliferation was assayed after 7 days, following an 18 h incubation with 1 μCi of [3H] thymidine (Amersham, IL) and analysed using a liquid scintillation counter (Packard Top Count,CT).
ELISA IL2 and IL10
An ELISA assay for IL10 (BD PharMingen, CA), IL2 and TNFα (R&D Systems, MN) was used to detect cytokines in the supernatants collected at 7 days for MELR assay (IL2/IL10) and 48 hr for TNFα and frozen (−80°C) until assayed.
Statistical evaluation of the results was done using the Wilcoxon (paired data) and Mann-Whitney (unpaired data) non-parametric tests (InStat, Graph Pad Software, CA).
Langerhans cell density
We evaluated the density of immature LC expressing CD1a and LAG (a protein associated with Birbeck granules22) in paired biopsies derived from the exocervix and TZ or the exocervix and SILs of individual patients. Figure 1 shows the results of our analysis of the density of CD1a and LAG positive cells in the TZ and SILs in comparison to the paired exocervix. We observed a 60% decrease in the relative percentage of LC (CD1a p = 0.0020, LAG p = 0.0312) in the TZ. In contrast, the low grade (LG) and high grade (HG) SILs show a relative reduction of 30% and 37%, respectively. Nevertheless, in comparison to the TZ the relative density of LC increases in SILs, being more significant in HG SILs (CD1a p = 0.089, LAG p = 0.041), compared to LG SILs (p = 0.287). Following this observation, we decided to determine if the decreased density of CD1a/LAG expressing cells could be correlated with the loss of CD1a/LAG expression, which is usually associated with the maturation of LC in situ. However, among the different groups we did not observe any difference in the density of CD80 and CD86 LC (0–2/mm2), 2 molecules associated with mature LC (data not shown).
Differential expression TNF α and MIP3 α
Owing to their importance in LC trafficking we studied the expression of several chemokines/cytokines within the microenvironment of TZ and SIL. TNFα, a cytokine known to be important for the activation of LC and to provoke their emigration from the epithelium, was found to be expressed at similar levels in whole biopsy specimens as assayed by semi-quantitative RT-PCR (data not shown). However, we found that the epithelium of the TZ and SIL produced higher levels of TNFα (ELISA) than the exocervix, as illustrated in Figure 2. The increased production of TNFα was significant in the TZ (p = 0.0312) compared to the exocervix. In contrast, MIP3α, a chemokine important for attracting immature LC to the epithelium, was found to be differentially expressed (immunohistochemistry) within the epithelium of the biopsies (Table 1). The exocervix more frequently presented a uniform staining (Fig. 3). In contrast, SILs were more likely to be associated with an intermittent pattern of staining and the TZ was frequently negative for MIP3α (Fig. 3). Analysis of the same biopsies did not reveal any difference in the percentage of biopsies expressing (RT-PCR) the chemokines RANTES and MIP1α, 2 chemokines also implicated in the chemotaxis of immature LC (data not shown).
|Biopsy||+ (%)||+/− (%)||− (%)|
Altered alloantigen presentation by Langerhans cells
To assay for the function of LC we used a MELR proliferation assay (Mixed Epithelium Lymphocyte Response) using allogeneic PBMC and epithelium-derived cells (LC and keratinocytes) from the exocervix, TZ and SIL paired biopsies. Since it has been shown that keratinocytes are unable to induce proliferation in a primary MELR, we could assay specifically for the function of LC.23 As shown in Figure 4, TZ and SIL (LG/HG) epithelium derived cells induced a significantly reduced proliferative response (p = 0.0391, p = 0.0020, respectively) in comparison with the exocervical epithelium-derived cells. Furthermore, the SILs were associated with a progressive decrease in their capacity to induce proliferation compared to the TZ (p = 0.0754).
We next evaluated the production of the cytokines IL2, an important indicator of T cells activation and IL10, an immunosuppressive cytokine, produced during the MELR assay. We observed that the average production of IL10 and IL2 was, respectively, elevated (75% of cases) and decreased (94% of cases) in cultures derived from SILs relative to the paired exocervix, as shown in Figure 5. However, the differential production of IL10 and IL2 in the TZ and SIL relative to the Exo was not considered significant (TZ p = 0.999/0.2500, SILs p = 0.577/0.1563, respectively).
In this study, we have provided further evidence that the epithelium of the TZ microenvironment is potentially “immunodeficient” due to an impaired immunosurveillance by LC. We have demonstrated that the TZ is associated with a lower density of LC as assessed by in situ analysis (CD1a/LAG), which correlates with the reduced in vitro stimulation of T cells (proliferation and IL2 production), as compared with the exocervix. Furthermore, we have shown that the development of SILs, although associated with a relative increase in immature LC, is deficient in their presentation of alloantigens to T cells (proliferation) compared to the TZ. In addition, as previously observed in situ,6 we confirm that the micro environment of the TZ and SILs is more frequently associated with IL10 production. Moreover, a potential correlation was also observed between the differential frequency of expression of MIP3α and/or TNFα and the density of immature LC (CD1a/LAG) in the exocervix, TZ and SILs.
Several groups have reported compelling evidence implicating multiple factors that may contribute to the sensitivity of the TZ to cervical carcinogenesis, all of which may be inter-related, i.e., the increased accessibility of HPV to target basal cells, the lack of factors which down regulate the expression of the viral oncogenes E6 and E7 or the sensitivity to estrogen.18, 24 Moreover, the process of metaplasia that occurs within the TZ is frequently associated with inflammation (i.e., TNFα) and proliferation, which are considered as risk factors for carcinogenesis.25, 26 Nevertheless, we propose that the expression of the immunosuppressive cytokine IL10 and the relative reduced density of LC and production of MIP3α may be additional risk factors for the development of SILs in the TZ.
The importance of cytokines/chemokines associated with DC/LC trafficking in normal tissues and tumors has been reported in numerous studies.10, 27, 28 MIP3α is considered as an important chemokine for attracting immature DC/LC to the epithelium.13 For example, MIP3α has been recently reported to be overexpressed in breast epithelium, resulting in an infiltration of immature DC into the tumor.29 In contrast, TNFα, which is known to stimulate the emigration of LC from the epithelium by downregulating the expression of E-cadherin,30 has been shown to be inversely correlated with LC density in skin tumors.28 The observed increased expression of MIP3α, along with the diminished frequency of TNFα in SILs, may contribute to the increased density of LC relative to the TZ, but additional factors within the microenvironment of SILs (i.e., HPV …) could interfere with the immunosurveillance capacity of LC.
The production of the immunosuppressive cytokine IL10 in the TZ and SILs may influence the chemotaxis of LC by modulating the production or effect of TNFα and GM-CSF.31, 32 Moreover, by a variety of mechanisms, IL10 has been implicated in the loss of function of LC/DC33 and the induction of apoptosis.34 Not surprisingly, many studies have implicated IL10 in the inefficient induction of tumor immunity.35–37 Consequently, the increased expression of IL10 within the microenvironment of the TZ and SILs in situ and the ability of TZ and SIL derived epithelial cells to induce IL10 in PBMC in vitro implicate this cytokine as a factor likely to influence the establishment and development of SILs.6
Other factors and/or mechanisms besides IL10 may contribute to the reduced alloantigen presentation by SIL-derived LC, such as apoptosis, the aberrant activation/differentiation of LC in the presence of HPV or other factors including the overexpression of VEGF or IL6 observed in SILs.4, 38, 39 Alternatively, HLA-DR+ keratinocytes that are frequently detected in SILs may interfere with LC function and induce T cell tolerance.40, 41 Moreover, it has also been demonstrated that tumors undergoing apoptosis, as opposed to necrosis, are capable of rendering DC tolerogenic.42 It is possible that the progressive increase in apoptosis as SILs progress to cancer43 may contribute to the tolerization of the immune response and this phenomenon may contribute to our in vitro observations (MELR). Nevertheless, nothing has been reported concerning the relationship between LC and keratinocytes in HPV+SILs and apoptosis. Several studies have reported that viruses (HIV, measles virus, etc.) can perturb the functional capacity and maturation of DC/LC.44–46 The effect(s) of HPV protein(s) on the function/maturation of DC is unknown, although the infection of keratinocytes by HPV has been associated with the perturbation of cytokine/chemokine expression,47–49 which may indirectly influence DC/LC function.
Studies concerning the different innate immunological surveillance mechanisms within the distinct microenvironments of the cervix are essential to distinguish between the immune alterations that are associated with either the predisposition or the progression of HPV-associated cervical lesions. Furthermore, an understanding of the interactions among the target cells of HPV infection (keratinocytes), LC and locally produced cytokines/chemokines, which contribute to the regulation of anti-tumor immunity, will aid in the design and development of prophylactic and therapeutic vaccines against cervical (pre) neoplastic lesions.
Dr. P. Delvenne is a research associate of the Belgian National Fund for Scientific Research. The authors thank Drs. C.F. Calvo, G. Thyphronitis, G. Butticé and M. Moutschen for their helpful comments and suggestions on the manuscript. We thank Mrs. H. Piron for her excellent technical assistance.