Article Text

PDF

Gastric cancer cell lines lack Fas ligand (FasL) expression but kill T cells via a FasL independent pathway
  1. I TINHOFER*,
  2. H WYKYPIEL*,
  3. I MARSCHITZ,
  4. T HENN,
  5. R GREIL
  1. Laboratory of Molecular Cytology
  2. Department of Internal Medicine
  3. University of Innsbruck
  4. Anichstrasse 35
  5. A-6020 Innsbruck, Austria
  1. Dr Richard Greil (email:Richard.Greil{at}uibk.ac.at)
  1. M W BENNETT,
  2. J O'CONNELL,
  3. D ROCHE,
  4. C BRADY,
  5. J KELLY,
  6. J K COLLINS,
  7. F SHANAHAN
  1. Department of Medicine
  2. Cork University Hospital
  3. National University of Ireland
  4. Cork, Ireland
  5. Department of Surgery
  6. Mercy Hospital
  7. National University of Ireland
  8. Cork, Ireland
  1. Professor Shanahan, Department of Medicine, Clinical Sciences Building, University Hospital, Cork, Ireland
  1. G C O'SULLIVAN
  1. Department of Medicine
  2. Cork University Hospital
  3. National University of Ireland
  4. Cork, Ireland
  5. Department of Surgery
  6. Mercy Hospital
  7. National University of Ireland
  8. Cork, Ireland
  1. Professor Shanahan, Department of Medicine, Clinical Sciences Building, University Hospital, Cork, Ireland

Statistics from Altmetric.com

Editor,—Bennett et al(Gut1999;44:156–162) reported that in each of 30 paraffin wax specimens of human gastric adenocarcinomas, FasL mRNA and protein co-localised to neoplastic epithelial cells. TUNEL staining revealed that a high number of tumour infiltrating lymphocytes (TIL) displayed apoptotic features. From these results and from their findings of FasL expression in human colon1 and oesophageal cancer,2 the authors propose that FasL might be a mediator of immune privilege in gastrointestinal cancers.

We studied intrinsic FasL expression in gastric cancer cell lines derived from primary (RF-1, SNU-1) or from metastatic sites (SNU-16, Kato-III , N-87, RF-48). We did not detect FasL mRNA or protein in any of the six cell lines analysed by RT-PCR and by flow cytometry (table1).3 4 We then performed the JAM assay to rule out the presence of a functional FasL expression below the detection limit of our assays.5 Although we found that gastric cancer cells were able to induce DNA fragmentation in the Fas sensitive T-acute lymphocytic leukaemia cell line CEM-C7H2 (fig 1A), blocking FasL on the effector cell site did not reduce the extent of cytotoxicity. This result was confirmed by replacing the target cell line by a subclone stably expressing the viral cowpox protein crmA, which inhibits activation of caspases 1 and 8 and thereby mediates resistance to Fas triggering (fig1B).6

Table 1

Expression of FasL and Fas in gastric cancer cell lines and their sensitivity toward Fas triggering by the CH11 monoclonal antibody

Figure 1

CEM-C7H2 T-acute lymphocytic leukaemia cells are killed by gastric carcinoma cell lines via a FasL independent pathway. (A) CEM-C7H2 target cells were incubated with 10 μCi/ml [3H]-thymidine for 16 hours and cocultivated with the gastric cancer cell lines at a target:effector ratio of 1:10. Cocultivation of cells was performed for 72 hours at 37°C. The reduction in radioactivity was used to calculate the percentages of gastric cell mediated target cell killing. The bars represent mean (SEM) specific killing (n=8). Statistical analysis of the blocking experiments showed the following: untreated v anti-FasL monoclonal antibody treated effectors (RF-1, p=0.5; RF-48, p=0.5; SNU-1, p=0.2); untreated v anti-tumour necrosis factor (TNF) α treated effectors (RF-1, p>0.07; RF-48, p>0.15; SNU-1, p>0.5). (B) CrmA expressing CEM-C7H2 (C7H2/crm) cells were used as target cells. Experimental conditions were as for (A). Statistical analysis did not reveal any significant reduction in mean specific killing of the crmA expressing C7H2 cells by the gastric cancer cell lines (RF-1, p>0.3; RF-48, p>0.8; SNU-1, p>0.5).

Owing to the discrepancy between our results (all cell lines were FasL negative) and those of Bennett et al (all 30 primary neoplasias were FasL positive), we wondered whether tissue derived factors such as tumour necrosis factor (TNF) α and interferon (IFN) γ might upregulate FasL in vivo, thus explaining the differences observed. In our setting, neither of the cytokines was able to modify FasL expression on gastric cancer cell lines (table 1). In addition, killing of T cell lines was not mediated via secretion of TNF-α as blocking the cytokine using a monoclonal antibody did not influence the result of the JAM assay (fig 1A). How can the differences between in situ and in vitro results be explained?

Bennett et al mention that CD45+ TIL express FasL mRNA, but they did not analyse Fas expression and sensitivity, features that together characterise activation induced cell death. Although on morphological examination of slides the authors excluded the possibility of lymphocytes being killed by infiltrating neutrophils potentially attracted by the expression of FasL on the tumour cells,7 it is possible that lymphocytes succumbed to apoptosis owing to either fratricide or suicide. This mechanism could well be under the (cytokine) control of the tumour as has been discussed for other diseases.8 Alternatively, lymphocytes could indeed be killed by the tumour cells but by a mechanism independent of the Fas system, a hypothesis supported by our data (fig1).

Bennett et al did not use the standard Lauren classification system.9 It has been shown that gastric carcinoma cells of the intestinal and diffuse type (according to Lauren9) differ in morphology, growth pattern and risk factors, and also in their expression of molecules involved in apoptosis such as Fas or p53.10 There is evidence that at least in some tumour models Fas and FasL expression are under transcriptional control of p53.11 Loss-of-function mutations or deletions of p53 have been reported to be involved in gastric carcinogenesis11 and the frequency of these events differs between intestinal and diffuse gastric cancers.12Also, a correlation between p53 mutation, Fas expression and gastric carcinoma cell differentiation has been demonstrated.10Further studies of the impact of differentiation and p53 functional status on FasL expression are therefore mandatory in gastric carcinoma cells.

Insensitivity towards Fas is usually an early step in tumour development, allowing tumour cells to resist the attack of the immune system and to avoid suicide when FasL expression is acquired.13 Furthermore, a sequence of Fas resistance and FasL expression has been demonstrated for hepatocellular carcinogenesis.14 Secondary loss of the FasL gene or of its expression during continuous culture of gastric adenocarcinoma cells is unlikely for the following reasons: (I) the cell lines were resistant to Fas and thus loss of FasL expression does not seem to be a prerequisite for their survival, and (ii) to our knowledge, no data are available from other cell (line) systems that tumour cell lines lose FasL expression during long term culture.

In conclusion, we think that Bennett et al's data suggest that CD45+ lymphocytes die in the immediate proximity of neoplastic cells. Although their data are compatible with Fas induced TIL cell death, our functional data from cell lines suggest that other tumour mediated mechanisms of killing immunocompetent cells might also exist in gastric cancer. Further work clarifying the sequence of Fas/FasL expression and function during the transformation and metastatic processes is needed.

References

Reply

Editor,—We concur with the view expressed by Tinhofer et al that Fas ligand (FasL) mediated “counterattack” against antitumour lymphocytes is not the sole mechanism of immune evasion in gastric, or indeed any other form of cancer. Tumours evolve multiple immune evasive strategies. However, there is ample in vitro and in vivo evidence that constitutive expression of FasL enables cancers to promote apoptosis of antitumour immune effector cells.1-1 For example, FasL has been significantly associated with apoptosis and loss of tumour infiltrating lymphocytes in human oesophageal cancer1-2 and depletion of antitumour natural killer cells in a mouse model.

In stomach cancer, apart from our finding of FasL expression at the mRNA and protein level in vivo in all 30 gastric adenocarcinomas examined,1-3 Rudi and colleagues1-4 also showed FasL mRNA in all three gastric carcinoma cell lines examined—including one cell line, KATO III, in which Tinhofer et al failed to detect FasL mRNA. This poses serious questions regarding the sensitivity of the FasL RT-PCR performed by Tinhoferet al. In fact, appropriate positive controls have not been shown to verify that their negative findings are not merely owing to the insensitivity of their assays for detecting FasL mRNA and protein in adherent cells.

Successful use of the JAM assay depends on using target cells that exhibit good sensitivity to FasL mediated apoptosis. Even different cultures of cell lines that are regarded as Fas sensitive, such as Jurkat E6 cells, can vary in their Fas sensitivity for reasons which are unclear, and Fas resistant subclones are easily generated. Tinhofer et al need to demonstrate that their cultures of CEM-C7H2 target cells were indeed susceptible to apoptosis via Fas in order to validate their negative JAM results. Authentic FasL mediated killing of Fas sensitive target cells is normally detectable after eight hours of co-culture with FasL expressing effector cells.1-5 Tinhoferet al performed a prolonged co-culture of 72 hours. It is more likely that the cell death detected in target cells at this late stage was from non-specific effects, such as exhaustion of nutrients or growth factors in the presence of proliferating effector cells, rather than a specific killing mechanism. Tinhoferet al should repeat their JAM assay for a shorter length of time with highly Fas sensitive target cells and include a proved FasL expressing effector cell line as a positive control.

Tinhofer et al's findings that gastric carcinoma cell lines are relatively resistant to Fas mediated apoptosis is consistent with findings for several other types of cancer cell. Fas resistance is a prerequisite for expression of FasL. Colon adenocarcinoma cell lines, for example, are also Fas resistant, enabling most colon adenocarcinoma cell lines to coexpress Fas and FasL without undergoing Fas mediated suicide.1-5 1-6 We agree with Tinhofer et al that the sequence of Fas/FasL expression and function during gastric carcinogenesis merits further investigation. Their suggestion that these molecules should also be investigated in metastases of gastric cancer is also pertinent as recent evidence suggests that FasL contributes to the invasion of Fas sensitive organs, such as the liver, by colonic adenocarcinoma cells.1-7 1-8

References

  1. 1-1.
  2. 1-2.
  3. 1-3.
  4. 1-4.
  5. 1-5.
  6. 1-6.
  7. 1-7.
  8. 1-8.
View Abstract

Footnotes

  • * These authors contributed equally to this work.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.