Alteration of the immunological synapse in lung cancer: a microenvironmental approach

Abstract

This study was designed to investigate the immunological properties of stroma reaction T cells and tumoral cells by comparison with non-tumoral lung tissue and local lymph nodes in order to explore interactions between tumour cells and the immune system. Immunodetection of major histocompatibility complex (MHC) molecules, CD3/T cell receptor (TCR) complex and T cell subsets markers was carried out in situ on frozen sections, and the semi-quantitative expression of CD3, CD4 and CD8 was examined in flow cytometry on lymphocytes of nodal, tumoral and healthy lung tissue from 62 patients with non-small cell lung cancer. This study showed alterations on lymphocytes and tumour cells in lung cancer, consistent with an impairment of T cell activation. CD3, TCRαβ and accessory molecules expression is down-modulated on peri- or intra-tumoral lymphocytes. MHC class I and class II molecules are down-modulated significantly on tumour cells. Other differences were noted, such as the reversed CD4/CD8 ratio of tumour infiltrating cells, compared to healthy lung tissues, consistent with the development of cytotoxic anti-tumoral responses. This study reports on the presence of a strong in vivo immunomodulating effect of tumour cells in human non-small cell lung cancer, likely to impair proper formation of the immunological synapse.

Introduction

The surveillance exerted constantly by the immune system to eradicate cells with potential for tumoral transformation has long been demonstrated in animal models or through human observations of high cancer incidence in immunocompromised individuals. Several mechanisms have been shown to be responsible for this immunological control, including down-modulation of major histocompatibility complex (MHC) antigens on tumour cells identified by natural killer (NK) cells and expression of tumour antigens recognized by cytotoxic T cells or antibodies involved in complement-dependent or cellular cytotoxicity [1,2]. The development of such anti-tumoral immune responses implies the participation of all mechanisms of specific immunity, involving cell proliferation and the generation of both effector and memory cells [2]. Once cancer is clinically obvious, it may be because these mechanisms have been overwhelmed and rendered inefficient, or because the tumour cells found a way to escape a still-functional immune system [1,3,4].

In many tumour types [5–7], the malignant tissue is partially infiltrated and surrounded by a stroma reaction, a non-tumoral tissue, which feeds and supports the tumour. In lung cancer, it comprises neoangiogenesis, extracellular matrix components and inflammatory cells which are, in large part, T cells, with a few macrophages and NK cells [8]. Tumour infiltrating T cells (TIL) have been demonstrated to be specific of tumour antigens in some instances [9,10], yet little is known about their specificity in the stroma reaction of lung cancer [11].

Decreased expression of MHC antigens occurs frequently in tumoral cells [12], limiting antigen presentation and thus recognition by T cells. Tumoral cells can also induce T cell apoptosis via the Fas/FasL complex or down-modulate the expression of CD3/T cell receptor (TCR) expression on T cells [13].

Decreased numbers of CD3/TCR complexes on T cells, as well as MHC down-modulation, would impair the proper constitution of immunological synapses, necessary to engage TCR-mediated specific responses [14]. Although seldom studied in lung cancer, a decreased expression of the CD3/TCR complex has been shown in other types of tumours [15] and could possibly have prognostic value [16].

Here, we explored the immunological characteristics of the immunological synapse, of tumoral (MHC molecules) and stromal (CD3/TCR complex and accessory molecules) cells by comparison with non-tumoral lung tissue and local lymph nodes. We show that a decrease of both CD3/TCR expression on T cells and MHC molecules on tumoral cells indeed exist in lung cancer, which could concur with the inefficacy of tumour growth control.

Materials and methods

Patients

The population studied comprised 62 patients, all eligible for surgery as first-intention therapy. Thirty-three patients had been diagnosed with epidermoid carcinoma, 23 with adenocarcinoma and six with large-cell carcinoma. According to TNM (tumour, lymph nodes, metastasis) classification, there were 26 patients with nodal extension (TxN1, TxN2) and four patients with metastases. Tobacco consumption was reported in 35 patients with a mean value of 44 ± 16 packs/year (Table 1). The study was approved by the local internal review board (IRB).

Table 1.

TNM (tumour, lymph node, metastasis) classification. Staging of the 62 lung cancer patients tested in immunofluorescent frozen sections and in flow cytometry, according to the TNM classification.

Number of patients	62
Males	55
Age	62 ± 10
Females	7
Age	61 ± 7
Epidermoid carcinoma	33
Adenocarcinoma	23
Large cell carcinoma	6
Nodal extension	26
Metastasis	4
T1	6
T2	49
T3–4	7
Tobacco consumption	Determined in 35 cases 44 ± 16 packs/year

Samples

The lung tissue samples were obtained immediately after curative surgery. Three types of samples were provided by the pathologist after macroscopic examination of the surgical material, within a few hours after the completion of surgery: (i) tumour tissue (n = 62); (ii) apparently normal lung tissue (n = 62) cut out as far as possible from the tumour; and (iii) an intrapulmonary lymph node (n = 27). Part of each sample was reserved for elution of lymphocytes from fresh tissue and analysis in flow cytometry. The remnant was snap-frozen in liquid nitrogen, then stored at −80°C until used for immunohistological analysis. Blood samples from 11 of these patients could be studied in flow cytometry.

For a few samples (n = 37), a small piece of fresh tissue was cultured briefly overnight in 1 ml of Quantum 007 medium (PAA Laboratories GmbH, Pashing, Austria) in sterile 12-well Costar® culture plates (Corning Inc., NY, USA) at 37°C in a water-saturated 5% CO₂ atmosphere.

Immunohistology

Immunodetection was carried out on 5 µm-thick serial frozen-cut sections (cryostat Jung CM 1800; Leica, Nussloch, Germany) prepared at −30°C. For each tissue, the first section was stained with toluidine blue for histological examination in light microscopy. Immunofluorescence staining was performed on the following sections using monoclonal mouse anti-human antibodies (mAbs) to β2 microglobulin (clone TÜ99; BD Pharmingen, San Diego, CA, USA) for MHC class I and to human leucocyte antigen D-related (HLA-DR) (clone B8·12·2; Beckman Coulter, Fullerton, CA, USA), HLA-DQ (clone SPVL3; Beckman Coulter) and HLA-DP (clone B7/21; BD Pharmingen) for MHC class II antigens. T cells were examined with anti-CD3 (clone UCHT1), anti-CD4 (clone 12T4D11), anti-CD8 (clone 21THY2D3), anti-TCR α/β (clone BMA031), anti-TCR γ/δ (clone IMMU510), anti-CD45RO (clone CHL1) and anti-CD45RA (clone 2H4). All the latter clones were from Beckman Coulter.

Incubations were carried out in a moist chamber at room temperature for 1 h with 5 µl of mAb per section. A control of autofluorescence was performed systematically, omitting the mAb step. The sections were washed with phosphate-buffered saline (PBS) and incubated for a further 30 min in a moist chamber at room temperature with fluorescein isothiocyanate (FITC)-conjugated rabbit anti-mouse antibody (Dako, Glostrup, Denmark) diluted 1:400 in PBS. After washes in cold PBS, the slides were mounted in PBS/glycerol (3/7). All slides were examined with an epifluorescence microscope (Olympus BX60, Tokyo, Japan) equipped with a digital camera and documented for scoring using Olympus software (DP70 Manager). Two investigators scored fluorescence intensity on a 0–4 scale in a blind manner. Cell numbers were expressed per mm² using an optical grid for enumeration.

Flow cytometry

For each tissue sample, the part reserved for lymphocyte elution was minced with scissors and shredded gently using the MediMachine (Dako). Eluted cells were placed in Hank's medium. Cell eluates or blood samples diluted in Hank's medium, layered gently over a Ficoll–Hypaque gradient (Lymphoprep; Nycomed, Oslo, Norway), were centrifuged for 30 min at 400 g at room temperature. The lymphocyte layer was then collected, washed once in Hank's medium and resuspended at 10⁶ cells/ml. T cell subsets in purified lymphocyte suspensions were assessed using flow cytometry using the four-colour panel of CD45-FITC/CD4-RD1/CD8-ECD/CD3-PC5 (TetraChrome®; Beckman Coulter). Results were expressed as percentages of lymphocytes and mean fluorescence intensity (MFI).

Statistical analysis

The normality of data distribution was assessed using the Kolmogoroff–Smirnoff test. Immunophenotyping results had a normal distribution and were analysed using Student's t-test.

Immunohistological scores were not distributed normally and were therefore expressed and compared as medians, using the Wilcoxon matched-pairs test (P = 0·05).

Statistics used the GraphPadPrism™ software (San Diego, CA, USA).

Results

Comparative histological study of healthy lung tissues, tumoral tissues and lymph nodes discloses significant anomalies

In healthy lung tissues, in 61% of the cases lymphocytes were absent or very scarce. In the other 39%, they were mainly scattered cells infiltrated between pneumocytes (mean value 2000 ± 152/mm²). In some cases, in smoker patients, lymphocytes were also present as small foci in the near vicinity of bronchiolar or vascular structures (mean value 4800 ± 316/mm²).

In tumoral tissues lymphocytes were present in 88% of the cases, mainly as large foci located within the stroma reaction around and sometimes within the tumour, with a mean value of 15 200 cells ± 3847/mm² in 78% of cases, significantly higher than in healthy lung tissues (P < 0·0001). Scattered cells were also often present in the surrounding lung tissue in 84% of the samples, with a low mean value of 800 ± 78 cells/mm². The six tumours without lymphocytic infiltration were all differentiated, with very dense tumoral foci in three cases. Five of them were stage T2 and did not present nodal extension. There was no significant difference in the presence of lymphocytes between patients with differentiated or undifferentiated tumours (Fig. 1).

Fig. 1.

Histological and immunohistological observations. Healthy (1) and tumoral (2) lung tissue toluidine blue staining (initial magnification ×100) showing the scarce infiltrates of lymphocytes and loose structure of normal lung tissue alveolae (1). By comparison, the tumour invades most of the parenchyma, obliterating alveolae and heavy lymphocytic infiltration can be seen in the vicinity of the tumour. Healthy (3-5-7-9-11) and tumoral (4-6-8-10-12) lung tissue immunofluorescent staining (initial magnification ×100 and ×400): β2 microglobulin is expressed strongly on healthy pneumocytes (3) clearly absent on tumour cells and clearly down-regulated in the adjacent lung tissue (4); human leucocyte antigen D-related is expressed strongly in normal alveolar macrophages and normal pneumocytes (5), but absent on tumoral tissue, while adjacent lymphocytes in the massive infiltrate on the left express it strongly (6); CD3 is expressed brightly on the small lymphocytic infiltrates of healthy lung tissue (7) and clearly down-regulated on intratumoral lymphocytes (8); a similar down-regulation of CD4 expression appears when comparing normal (9) and tumoral tissue (10); sparse brightly stained CD8 cells can be seen in healthy lung tissue (arrow, 11) while intratumoral CD8 cells appear much dimmer (12). L: lymphocytes, BV: blood vessel, M: macrophages, HP: healthy pneumocytes, TT: tumoral tissue, TIL: tumour infiltrating lymphocytes.

Paired analysis of tumoral and healthy tissues confirmed a significantly higher lymphocytic infiltration in tumoral tissue in most (74%) of the cases, (P < 0·0001). There was the same amount of lymphocytes in both tissues in 8% of the cases, fewer lymphocytes in tumoral tissue in 8%, and no lymphocytic infiltration in either tissue in 10%.

Within the group of patients with reported tobacco consumption, tumour-infiltrating lymphocytes, present in all cases, were significantly more frequent than in healthy lung tissue (P = 0·006), yet without relationship to the degree of tobacco exposure.

The architecture of lymph nodes was often disorganized, with metastatic cells. Germinal centres were visible in 12 cases and in the remaining patients the lymph nodes were infiltrated by tumoral cells and follicles were no longer distinguishable.

Expression of the major partners of the immunological synapse is altered in lung cancer

Beta-2 microglobulin was, as expected for nucleated cells, expressed brightly on healthy pneumocytes and lymphocytes. It was significantly dimmer or absent on tumoral cells (P < 0·005) and lower on lymphocytes from tumoral samples compared to healthy tissues (P = 0·0002).

More surprisingly, the same observation of a strong labelling of healthy pneumocytes and lymphocytes was made for HLA-DR expression, with a significant decrease or disappearance on tumoral cells (P = 0·008) and on tumour-infiltrating lymphocytes (P = 0·0001).

A decrease of β2 microglobulin expression on tumour cells compared to healthy pneumocytes was seen in 50 of 51 cases, of HLA-DR in 45 cases and of HLA-DQ in 19 cases.

Both β2 microglobulin and HLA-DR were expressed brightly on nodal lymphocytes.

DQ expression was not significantly different between healthy and tumoral tissues, while DQ⁺ nodal lymphocytes were slightly brighter than in other sites. No labelling was seen for HLA-DP (Figs 1 and 2).

Fig. 2.

Expression levels of major histocompatibility complex antigens in lymph nodes, tumoral and healthy lung tissue. The expression of different markers was assessed by immunofluorescent staining on frozen sections of the three types of tissues. Data expressed as median of scores of fluorescence intensity on a 0–4 scale.

In immunohistology, almost half the samples of healthy lung tissue contained CD3⁺ T lymphocytes, while this figure reached 86% for tumoral tissue. The expression of CD3 was visibly markedly decreased for 40 of 51 cases in tumoral tissue compared to healthy lung tissue (Fig. 1). This was confirmed on eluted cells by comparing MFI values, as shown in Fig. 3, reporting on the significantly lowered MFI of tumour-infiltrating CD3⁺ cells (P < 0·0001). Moreover, after a short organotypic culture in non-tumoral medium, CD3-MFI, compared in paired samples, was significantly higher on cells from healthy lung tissues and lymph nodes (P = 0·047 and P = 0·01, respectively). In addition, the mean percentage of peripheral blood CD3⁺ cells from patients was lower compared to the normal laboratory values (64% ± 17 versus 75% ± 5, P = 0·006) and CD3-MFI of peripheral blood lymphocytes was significantly higher than that of cells eluted from healthy and tumoral tissues (P < 0·0001).

Fig. 3.

Expression levels of CD3 on T cells eluted from fresh tissues or organotypic cultures in non-tumoral medium and on peripheral blood T cells from patients and controls. The cell were eluted from lymph nodes, tumoral and healthy lung tissues. The organotypic cultures were performed overnight in 1 ml of Quantum 007 medium. The purified lymphocyte suspension was obtained from these tissues and peripheral blood using a Ficoll–Hypaque gradient. The expression of CD3 was assessed by flow cytometry on CD45⁺ gated cells. Data expressed as median of mean fluorescence intensity (MFI).

TCRαβ was detected in situ as faintly expressed in all lymph nodes and with a significantly decreased expression on tumoral tissue lymphocytes compared to lymph nodes (P = 0·007). TCRγδ⁺ cells were very rare or absent from the three types of tissue for all patients.

CD4⁺ cells were present in 37% of the cases in healthy lung tissue and in 76% of the cases in tumoral tissue, without any significant difference in fluorescence intensity (Fig. 1). In flow cytometry, CD4⁺ cells were observed in all samples, but represented 13% of eluted lymphocytes from tumours and 19% from healthy tissues. The proportion of CD4⁺ cells was increased significantly in lymph nodes after a short culture in non-tumoral medium (P = 0·03). The MFI of peripheral CD4 was significantly higher than that of cells eluted from healthy and tumoral tissues, although decreased compared to normal laboratory values (P = 0·04). In lymph nodes, CD4⁺ lymphocytes were the most abundant population with a mean CD4/CD8 ratio of 4 ± 1·6. This CD4/CD8 ratio was reversed significantly in cells from both healthy (0·88 ± 0·17) and tumoral (0·58 ± 0·07) lung tissue (Fig. 4).

Fig. 4.

Expression levels of CD4 on T cells eluted from fresh tissues or organotypic cultures in non-tumoral medium and on peripheral blood T cells from patients and controls. Same experimental settings as in Fig. 3. The expression of this marker was assessed by flow cytometry on gated CD3⁺ T cells. Data expressed as median of mean fluorescence intensity (MFI).

Indeed, CD8⁺ cells were present in 47% of the cases in healthy lung tissue and in 86% of the cases in tumoral tissue, with a significantly brighter expression in healthy lung tissue (P = 0·03). CD8⁺ cells represented two-thirds of lymphocyte infiltrates in stroma reactions. In 41 cases CD8⁺ cells could be seen within the tumour patches, in close contact with tumour cells (Fig. 1). The proportion of CD8⁺ cells was increased significantly within lymphocytes eluted from tumours after a short organotypic culture in non-tumoral medium (P = 0·04) (Fig. 5).

Fig. 5.

Expression levels of CD8 on T cells eluted from fresh tissues or organotypic cultures in non-tumoral medium and on peripheral blood T cells from patients and controls. Same experimental settings as in Fig. 3. The expression of this marker was assessed by flow cytometry on gated CD3⁺ T cells. Data expressed as median of mean fluorescence intensity (MFI).

CD3 and CD8 expression on T cells was brighter in tumour and normal lung parenchyma, respectively, for patients with nodal extension according to the TNM classification. Conversely, β2 microglobulin and HLA-DR-positive nodal cells were lower in patients with nodal extension.

CD45RA and CD45RO were expressed more brightly on tumour infiltrating cells than healthy lung tissue T cells. In tumours, when present, CD45RO was expressed significantly at the surface of more T cells than CD45RA (P = 0·01). The same was true, in tumoral tissue, for all tumour types and all stages of cancer evolution. In tumoral tissue, CD45RO cell numbers decreased correlatively with tobacco consumption.

Discussion

In this study, in situ interactions between tumour cells and the immune system were explored in patients with non-small cell lung cancer (NSCLC), focusing upon partners of the immunological synapse. Tumoral and healthy lung tissue as well as locoregional lymph nodes were examined, demonstrating a gradient of immunological alterations in these various compartments.

In tumoral samples, T lymphocytes were organized mainly in peritumoral infiltrates within the stroma reaction, although CD8⁺ T cells were present inside tumoral tissues, as has been reported in breast cancer [17]. The main observation of this study is, however, that both intratumoral lymphocytes and lung tumour cells present alterations consistent with an impairment of T cell activation. The decreased expression of MHC class I on tumoral cells is likely to limit their recognition by CD8⁺ T cells [18]. Such a decrease is consistent with what has been reported in other types of tumours for both classical or non-classical class I molecules [1,12]. The similarly decreased level of MHC class II expression observed here would, moreover, impair CD4⁺ T cell activation. Increasing evidence points out the important role of CD4⁺ T cells in the generation of an effective immune response comprising T helper 1 (Th1) and Th2 cells [19].

These alterations of MHC expression could result from a decrease of interferon (IFN)-γ production [20], consistent with the apparently inactivated state of tumour infiltrating lymphocytes observed here not to express HLA-DR. Moreover, the loss of HLA-DR expression in tumour cells seems to be associated with the epigenetic inactivation of the transcriptional activator for MHC class II genes: co-activator class II transactivator (CIITA) [21].

Indeed, parallel to the altered expression of antigen-presenting molecules on tumoral epithelial cells, we observed alterations on the counterpart T cell population in situ. Modulation of the CD3/TCR complex has seldom been examined in lung cancer. In this study, immunohistology and flow cytometry data concurred in demonstrating a significant decrease of CD3 expression within tumours, consistent with what has been reported in various other cancer types [13,14,22–24].

Among the mechanisms suspected for this modulation of the CD3/TCR complex are activation of intracellular peptidases leading to cleavage of CD3 ζ chain or secretion of soluble factors suppressing the ζ chain without interfering with the functions of tyrosine kinases [13]. In addition, tumoral development may induce chronic antigenic stimulation, leading possibly to a rapid decrease of CD3 ζ and ε chains expression [13]. Alternatively, the turnover of these molecules could be not fast enough to compensate for CD3 degradation by reactive oxygen intermediates [13]. However, a recent study from Esendagli et al.[25] reported contradictory results, finding no difference of CD3 percentage of expression between malignant and non-malignant tissue using a flow cytometry technique in patients with NSCLC. A part of this difference could be explained by the fact that these authors examined the percentage of cells expressing CD3 and not MFI. The percentage of positive cells is not sufficient to determine whether CD3 is expressed in numbers large enough to form an immunological synapse. However, this cannot explain totally the difference of results between both studies. Moreover, we showed that CD3 was expressed less in tumoral tissue than in healthy lung tissue by two different techniques, and these results are consistent with the literature for many other cancer types [25].

Our data also show a gradient of T cell incapacitation, as both TCRαβ and CD3 were expressed less on the lymphocytes adjacent or infiltrating tumoral tissue than on those present in healthy lung tissue or lymph nodes. In parallel, MHC class I and class II molecules, which are expressed constitutively on the surface of pneumocytes [26], were both repressed significantly on tumour cells and to a lesser extent on the lymphocytes present in the tumoral tissue. As the recognition of antigens presented by MHC molecules requires the formation of immunological synapses involving the mobilization of large numbers of TCR molecules [14], these data indicate that this mechanism is impaired severely in lung tumours. This has been confirmed by showing that even when TCRαβ are expressed at the surface of CD8⁺ T cells, they are often anergic [27].

CD8 expression was often more intense in tumoral tissue with a low CD4/CD8 ratio. This is consistent with the development of cytotoxic anti-tumoral responses by T cells capable of direct lysis of tumoral cells [28]. That such immune responses develop is suggested indirectly by the high CD4/CD8 ratio observed in local lymph nodes. However, some degree of immunosuppression may also be mediated at this level by infiltrating tumour cells as an increase of CD4 expression was observed in lymph nodes after a short culture in non-tumoral medium. An increase of CD8⁺ detectable cells was also observed in tumours after a short culture in non-tumoral medium, confirming the notion of an immunosuppressive tumoral microenvironment, possibly cytokine-mediated [29]. These results are thus consistent with local suppression, the tumoral microenvironment playing a role in the inhibition of molecules involved in the immunological synapse, as reported in other, but not lung, cancer types [29].

Also consistent with previous reports, CD45RO⁺ lymphocytes were more numerous and brighter in tumours compared to healthy lung tissue, suggesting the presence of memory cells derived from lymphocytes having recognized locally present and thus possibly tumoral antigens. However, the expression of CD45RA was more intense in tumoral tissue than in healthy lung tissue, suggesting either the recruitment of naive lymphocytes at the tumoral site or the re-expression of CD45RA at the surface of T cells having just achieved an immune response [30]. Nevertheless, CD45RO⁺ cells were more numerous than CD45RA⁺ cells.

Healthy lung tissue contained only sparse lymphocytes, and lymphocyte foci only in some smokers, suggesting some degree of activation. Tobacco consumption has indeed been reported to cause various chronic disorders of the respiratory tract such as eosinophilic pneumonitis [31] or obstructive pulmonary disease [32] associated with inflammation of the lung tissue. The presence of lymphocytes in healthy lung tissue could also suggest an immune surveillance of tumoral cells development.

The decreased expression of molecules involved in the immunological synapse demonstrated here indicates that the tumour generates an immunosuppressor microenvironment, at the same time repressing the antigen-recognition complex (TCR) and the CD3 complex of signal transduction on infiltrating T cells and both MHC class I and class II antigen presentation complexes on tumoral cells.