In Vivo Analysis by in Situ Hybridization

Home , In situ

Proc. Natl. Acad. Sci. USA Vol. 86, pp. 5064-5068, July 1989 Immunology Regulation of tumor necrosis factor gene expression in colorectal adenocarcinoma: In vivo analysis by in situ hybridization (immunohistochemistry) STEFAN BEISSERT*, MICHAEL BERGHOLZt, INGE WAASE*, GERD LEPSIENt, ALFRED SCHAUERt, KLAUS PFIZENMAIER*, AND MARTIN KR6NKE*§ *Kfinische Arbeitsgruppe, Max-Planck-Gesellschaft, Gosslerstrasse 10 d, 3400 Gdttingen, Federal Republic of Germany; and tZentrum Pathologie and tZentrum Chirurgie, Universitit G6ttingen, Robert-Koch-Strasse 40, 3400 Gottingen, Federal Republic of Germany Communicated by Lloyd J. Old, March 9, 1989 (receivedfor review December 12, 1988)

ABSTRACT Tumor necrosis factor (TNF) produced by TNF to induce collagenase synthesis (15) may lead to tissue macrophages is thought to contribute to the host defense destruction facilitating the invasive growth of tumors. Thus, against development of cancer. However, since tumor cells demonstration of TNF production in cancer may reflect themselves are able to produce TNF, it is conceivable that TNF either protective or pathogenic processes. Clearly, a first step may also play an adverse pathological role in carcinogenesis. to understanding the pathophysiological role of TNF in To better understand the functional ignificance of TNF in cancer would be the identification of the TNF-producing cell neoplastic disease, we have determined the cellular source of type at the actual site of disease. Thus we analyzed samples TNF activity produced in 10 patients with colorectal cancer. from 10 patients with colorectal carcinoma. We show that Northern blot analysis of RNAs extracted from fresh biopsy tumor-infiltrating macrophages and not tumor cells present in specimens revealed detectable TNF mRNA levels in all in- colorectal adenocarcinoma express TNF mRNA and pro- stances. By using in situ hybridization of frozen sections, duce TNF protein. scattered cells expressing TNF mRNA could be discerned. Based on morphological criteria, these TNF-positive cells most MATERIALS AND METHODS likely belong to the macrophage lineage. Macrophages in normal tissue surrounding the tumor did not express TNF RNA Blotting Analysis. Total cellular RNA was extracted mRNA, suggesting that macrophage activation occurs locally at from fresh biopsies by the method of Chirgwin et al. (16), 30 the site of neoplastic transformatn. Immunohistochemistry gg of RNA was electrophoresed on formaldehyde gels, using anti-TNF monoclonal antibodies revealed that less than transferred to nitrocellulose filters, and hybridized to a 1% TNF TNF-specific cDNA probe (an 800-base-pair EcoRI fragment oftumor-infiltrating macrophages synthesize protein. of A42-4) (17) or to an HLA-B7-specific cDNA (18) labeled Thus we present evidence that in colorectal cancer only a small with [32P]dCTP by random priming. After hybridization, proportion of tumor-infiltrating macrophages produces TNF, filters were washed and exposed to Kodak XAR films by indicating that the microenvronment of the tumor provides using an intensifying screen. adequate, yet suboptimal, conditions for macrophage activa- In Situ Hybridization. In situ hybridization was performed tion. essentially as described by Hogan et al. (19). Freshly isolated biopsies were frozen immediately without any fixation. Spec- Tumor necrosis factor (TNF) has been originally described as imen blocks were cut at 8 glm in a cryostat at -200C. Serial a serum activity in lipopolysaccharide-treated mice that is frozen sections were collected on "subbed" slides (20), dried capable of causing hemorrhagic necrosis of animal tumors for 10 min at 550C, and fixed in 4% (wt/vol) paraformalde- (1). However, since the availability of highly purified recom- hyde. To remove proteins possibly associated with mRNA, binant TNF, evidence has accumulated indicating that TNF slides were incubated in 2x SSC (lx SSC = 0.15 M NaCl/ has a wide range ofbiological activities, affecting the growth, 0.015 M sodium citrate, pH 7.0) for 30 min at 700C, rinsed in differentiation, or function of virtually every cell type inves- phosphate-buffered saline (PBS, 20 mM sodium phosphate/ tigated (2-5). Although the complexity of TNF activities 0.7% NaCl, pH 7.4), and digested with Pronase (0.2 mg/ml) makes it at present difficult to assign a specific role to TNF for 10 min at room temperature. Slides were then refixed, in vivo, its direct tumoricidal activity and its role in inflam- acetylated, dehydrated, and finally air-dried. mation are thought to contribute to the host's defense against RNA probes were labeled by using SP6 or T7 RNA tumors. This proposal has been challenged by findings indi- polymerases and 35S-labeled UTP as described (21). Tem- cating that many cell types, including tumor cells themselves, plate TNF- and HLA-B7-specific DNAs were linearized by are able to produce either TNF or lymphotoxin (also called digestion with enough restriction enzyme to generate 35S_ TNF-,B) (6-10). Lymphotoxin is a TNF-related cytokine that labeled run-on transcripts 380-1100 bases long. Hybridiza- binds to the same receptor (11) and exerts similar biological tion was performed at 500C in a humid chamber. Slides were activities (12). TNF or lymphotoxin production by tumor washed in 50o (vol/vol) formamide/2x SSC followed by cells suggests that these cytokines have a possible pathogenic digestion with RNase A (20 gg/ml) and three further washes role in tumorigenesis. Indeed, the secretion of lymphotoxin in 50o formamide/2x SSC. For autoradiography, air-dried by human myeloma cells has been linked to osteoclastic bone slides were coated with NTB2 emulsion (Eastman Kodak) destruction and hypercalcemia in patients with myeloma (8). and exposed at 40C for 10 days. Slides were developed and In addition, because of its potent angiogenic activity (13, 14), counter-stained with hematoxylin/eosin. TNF may stimulate the growth of blood vessels, thereby Immunohistochemistry. For immunostaining frozen sec- promoting tumor development. Furthermore, the ability of tions were incubated with monoclonal antibodies as indicated. Monoclonal anti-human TNF antibody 154/6 was a gift The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviation: TNF, tumor necrosis factor. in accordance with 18 U.S.C. §1734 solely to indicate this fact. §To whom reprint requests should be addressed. 5064 Downloaded by guest on September 28, 2021 Immunology: Beissert et al. Proc. Natl. Acad. Sci. USA 86 (1989) 5065 of G. Trinchieri (Philadelphia). Monoclonal antibody 63D3, recognizing a monocyte-specific antigen (22), was purchased from the American Type Culture Collection. Alkaline phosphatase-anti-alkaline phosphatase complexes were prepared as described by Cordell et al. (23). Sections were counterstained with hematoxylin/eosin.

RESULTS TNF mRNA Expression in Colorectal Cancer. Tumorbiopsy specimens were obtained from 10 patients with colorectal cancer during abdominal surgery. Total RNA was extracted and analyzed for TNF mRNA levels. As shown in Fig. 1, the abundance of TNF mRNA varied depending on the tumor specimen investigated. Longer exposure of the RNA blot revealed that, indeed, TNF mRNA could be detected in all instances (data not shown). The tumor ofpatient 7 apparently expressed the highest level of TNF mRNA. Control hybridization to a HLA-A,B,C-specific cDNA probe confirmed the integrity of the RNA samples loaded. The different HLA- A,B,C mRNA levels observed most likely reflect variable HLA-A,B,C gene expression by individual tumor tissues since control ethidium bromide staining indicated equivalent quantities of RNA transferred to nitrocellulose filters. In Vivo TNF mRNA Expression by Tumor-Infiltrating Mac- rophages. To determine the cellular source of TNF mRNA, in situ hybridization offrozen tumor sections was performed using a 35S-labeled antisense TNF RNA probe. As exempli- fied by patient 7, the majority of cells have 5-10 grains per cell (Fig. 2a) corresponding to the background staining revealed by hybridization to a 35S-labeled sense HLA-B7 RNA probe (Fig. 2c). Only a few scattered cells had grain counts that were well above background (40-200 grains), which was considered specific hybridization to TNF mRNA. By histo- morphological criteria (azurophilic cytoplasm, large lobular nucleus, and dense chromatin), these cells were considered to be of the macrophage/monocyte lineage. To demonstrate the sensitivity of the in situ hybridization, we used an 35S-labeled antisense HLA-B7 RNA probe. As shown in Fig. 2b, greater than 80%o of cells were positive, indicating that most if not all of the cells contained intact mRNA that was accessible for effective hybridization. Similar results were obtained when tumor specimens from nine other patients were analyzed by in situ hybridization (Table 1). Tumor-Inflltrating Macrophages Produce TNF Protein. Since TNF gene expression is largely regulated at a post- transcriptional level-that is, translation of TNF mRNA and FIG. 2. In vivo TNF mRNA expression in the adenocarcinoma secretion of TNF require additional stimuli (10, 24, 25)-it from patient 7. Frozen sections (8 ,gm thick) were hybridized to was important to determine whether TNF mRNA-positive 35S-labeled antisense TNF (a) or HLA-B7 (b) complementary RNA macrophages actually produce TNF protein. As shown in probes. Control hybridization was performed with a sense HLA-B7 Fig. 3, immunohistochemistry using an anti-TNF monoclonal RNA probe (c). After a 6-day exposure, slides were developed and counterstained with hematoxylin/eosin. (For a, x400; for b and c, 1 2 3 4 5 6 7 8 9 10 x250.) cDNA probe antibody strongly stained cells with macrophage-like morphology, indicating that tumor-infiltrating macrophages do produce TNF protein in vivo. 40 pTNF Lack of TNF-Producing Macrophages in Normal Tissue Surrounding the Tumor. Normal colorectal tissue of patient 7 adjacent to the site of tumor had strong mononuclear infiltration (Fig. 4a) comparable to that observed in the adenocarcinoma (Fig. 5a). However, neither TNF mRNA nor TNF protein could be detected by in situ hybridization (Fig. 4b) or by immunostaining (data not shown), respec- o* it Sp pHLA-B7 tively. This finding indicates that it is the microenvironment FIG. 1. TNF mRNA in tumor biopsy specimens of 10 patients of colorectal cancer that provides the stimuli for the regional with colorectal carcinoma. Total RNA (30 tg per lane) was size- activation of macrophages. separated in 1% agarose/formaldehyde gels, transferred to nitrocel- Activation Status of Tumor-Infiltrating Macrophages. To lulose filters, and hybridized to 32P-labeled TNF and HLA-B7- determine the level of macrophage activation in colorectal specific cDNA probes. cancer in further detail, we compared the extent of tumor Downloaded by guest on September 28, 2021 5066 Immunology: Beissert et al. Proc. Nati. Acad. Sci. USA 86 (1989) Table 1. In situ analysis of TNF production in colorectal cancer Site of Frequency of TNF mRNA- TNF- adeno- infiltrating positive producing Patient carcinoma macrophages,* % cells,t % cells,*% 1 Sigmoid colon 15-25 2 0.2 2 Sigmoid colon 18-35 3 0.3 3 Colon 20-30 1.5 0.2 4 Colon 15-25 2 0.3 5 Sigmoid colon 30-40 1.5 0.1 6 Sigmoid colon 25-35 1.5 0.1 7 Rectum 25-35 6 0.3 8 Rectum 25-35 2 0.1 9 Colon 20-30 1 0.1 10 Rectum 30-40 <1 0 *Cryostat sections of tumor biopsy specimens were immunostained with the monocytic-specific antibody 63D3. Frequencies of infiltrating macrophages were revealed by microscopic evaluation of several visual fields corresponding to more than 1000 nucleated cells. Data are the ranges ofmonocyte infiltration that varied within a given tumor specimen depending on the particular section investigated. tTNF mRNA-positive cells were visualized by in situ hybridization with an antisense TNF complementary RNA probe. Frequencies of TNF mRNA-positive cells were estimated by evaluating >2000 cells. tTNF protein-producing cells were detected by immunostaining with anti-TNF antibody. Frequencies ofpositive cells were estimated by evaluating >2000 cells.

infiltration by macrophages with the frequency of macrophages expressing TNF mRNA or producing TNF protein. As shown in Fig. Sa, approximately 30%o of the cells in the rectal adenocarcinoma of patient 7 had macrophage-like morphology and were stained with 63D3, a monocyte- specific monoclonal antibody (22). Bright- and dark-field FIG. 4. Lack of TNF mRNA-positive cells in normal tissue. microscopic evaluation of the in situ hybridization using Cryostat sections of normal rectal tissue from patient 7 in the 35S-labeled antisense TNF RNA probe showed that =6% of immediate vicinity ofthe tumorwere immunostained with monocyte- the cells were TNF mRNA-positive (Fig. 5 c and d). When specific monoclonal antibody 63D3 (a) or hybridized to a 35S-labeled immunostained by the anti-TNF monoclonal antibody, only anti-sense TNF RNA probe (b). Arrows indicate macrophages 0.3% of the mononuclear cells were positive (Fig. 5b and identified by morphological criteria. (x250.) Table 1). Thus the data indicate that approximately every DISCUSSION fifth tumor-infiltrating macrophage only expresses TNF mRNA and even fewer synthesize detectable levels of TNF TNF mRNA has been detected (26) in RNA samples ex- protein. Similar results were obtained from corresponding tracted from either peripheral blood lymphocytes or solid analyses of nine other patients with colorectal adenocarci- tumor tissue from cancer patients. Determination of either noma (Table 1). TNF activity in the serum or TNF mRNA levels in the RNA extracted from heterogenous tumor tissue, however, does not answer the crucial questions as to the cellular source and functional significance of TNF production in neoplastic disease. In the present study, we have attempted to characterize the conditions of TNF production in colorectal cancer. We show by in situ hybridization and immunohistochemistry that in all cases of colorectal carcinoma investigated, TNF was produced by infiltrating macrophages. The observation that the mononuclear infiltrate expresses TNF mRNA in 10 out of .- .- wd 10 colorectal carcinoma patients suggests that in this tumor TNF production may be a defense reaction executed by ., -~~~~~~~~~~~~~. activated macrophages. One of the principal findings of this report is that only a fraction of tumor-infiltrating macrophages expresses TNF mRNA, and even fewer are able to produce TNF protein. Although it appears to be difficult to precisely judge the sensitivity of the in situ hybridization and immunostaining procedures, analogous analyses ofrenal cell carcinoma spec- FIG. 3. In vivo TNF production in rectal adenocarcinoma. imens have shown that it is possible to demonstrate TNF Cryostat sections from the adenocarcinoma ofpatient 7 were immnu- mRNA as well as TNF protein in greater than 50%o of the nostained with monoclonal anti-TNF-antibody 154/6 using the alka- tumor-infiltrating macrophages (I.W. and M.K., unpublished line phosphatase-anti-alkaline phosphatase method. Parallel sections data). Thus, the frequencies of TNF-positive macrophages analyzed with isotype-matched control immunoglobulins were not observed might be only slightly underestimated. The quan- stained. (x600.) titative representation of tumor-infiltrating macrophages ex- Downloaded by guest on September 28, 2021 Immunology: Beissert et al. Proc. Natl. Acad. Sci. USA 86 (1989) 5067

FIG. 5. Activation status of tumor-infiltrating macrophages. Frozen sections from patient 7 were immunostained with either monocyte- specific monoclonal antibody 63D3 (a) or monoclonal anti-TNF antibody 154/6 (b). In situ hybridization was performed using antisense TNF (c) and HLA-B7 (e) RNA probes. (d) Dark-field image of c. (f) Dark-field image ofe. Arrows indicate TNF protein- (b) or TNF mRNA-positive (c) cells. (x1OO.) pressing TNF mRNA or TNF protein detected in this study neoplastic transformation, since macrophages present in the suggests that the control mechanisms of TNF gene expres- normal tissue in the immediate vicinity of the tumor neither sion defined in vitro are recapitulated in vivo. It has been express TNF mRNA nor secrete TNF protein. It is therefore shown that murine macrophages require a sequence of two concluded that the colorectal tumor itself provides the mi- distinct signals to become activated for cellular toxicity (27, croenvironment for macrophage activation. 28), which may be TNF-mediated (29, 30). The first signal Our findings have implications for entering patients with serves as a "priming" factor and the second signal triggers colorectal cancer into clinical trials with biological response the final activation. The priming signal may mediate tran- modifiers. Since, in this tumor, the majority of infiltrating scriptional activation ofthe TNF gene, and the second signal macrophages seems to be suboptimally activated, treatment may be required to allow translation of the protein and TNF with recombinant TNF may supplement the host's defense secretion by macrophages (24, 25, 31, 32). As illustrated in against tumor progression and eventually benefit the patient. this report, the number of detectable TNF-producing mac- Furthermore, treatment with macrophage-activating agents, rophages is quite small, even though the tumor is heavily such as the interferons, might enhance the endogenous infiltrated. This suggests that the population of tumor- antitumoral activities of the mononuclear infiltrate. infiltrating macrophages is only suboptimally activated. Ac- tivation of macrophages apparently occurs at the site of We thank G. Trinchieri (Wistar Institute, Philadelphia) for pro- Downloaded by guest on September 28, 2021 5068 Immunology: Beissert et al. Proc. Natl. Acad. Sci. USA 86 (1989) viding anti-TNF monoclonal antibody 154/6. We thank A. Hofsom- 16. Chirgwin, J. M., Przybyla, A. E., MacDonald, R. J. & Rutter, mer, B. Junemann, and M. Kollhof for technical assistance and G. W. J. (1979) Biochemistry 18, 5294-5299. Schmidt for preparing the manuscript. We also thank Drs. P. Gruss 17. Pennica, D., Nedwin, G. E., Hayflick, J. S., Seeburg, P. A., and G. Dresslerfor helpful discussions. This work was done in partial Derynck, R., Palladino, M. A., Kohr, W. J., Aggarwal, B. B. fulfillment of the M.D. thesis for S.B. and I.W. & Goeddel, D. V. (1984) Nature (London) 312, 724-729. 18. Sood, A. K., Pereira, D. & Weissmann, S. M. (1981) Proc. 1. Carswell, E. A., Old, L. J., Kassel, R. L., Green, S., Fiore, N. Nat!. Acad. Sci. USA 78, 616-620. & Williamson, B. (1975) Proc. Nat!. Acad. Sci. USA 72, 3666- 19. Hogan, B., Constantini, F. & Lacy, E. (1986) Manipulating the 3670. Mouse Embryo (Cold Spring Harbor Lab., Cold Spring Harbor, 2. Old, L. J. (1985) Science 230, 630-632. NY), pp. 228-235. 3. Beutler, B. & Cerami, A. (1986) Nature (London) 320, 584-588. 20. Gall, J. & Pardue, M. (1971) Methods Enzymol. 38, 470-480. 4. Nathan, C. F. (1987) J. Clin. Invest. 79, 319-325. 21. Goebels, N., Waase, I., Pfizenmaier, K. & Kronke, M. (1988) 5. Le, J. & Vilcek, J. (1987) Lab. Invest. 56, 234-248. J. Immunol. 141, 1231-1235. 6. Rubin, B. Y., Anderson, S. L., Sullivan, S. A., Williamson, 22. Ugolini, V., Nunez, G., Smith, R. G., Stasny, P. & Capra, B. D., Carswell, E. A. & Old, L. J. (1986) J. Exp. Med. 164, J. D. (1980) Proc. Nat!. Acad. Sci. USA 77, 6764-6768. 1350-1356. 23. Cordell, J. L., Falini, B., Erber, W. N., Ghosh, A. K., Abdu- 7. Cuturi, M. C., Murphy, M., Costa-Giomi, M. P., Weinmann, laziz, Z., MacDonald, S., Pulford, K. A. F., Stein, H. & R., Perussia, B. & Trinchieri, G. (1987) J. Exp. Med. 165,1581- Mason, D. Y. (1984) J. Histochem. Cytochem. 32, 219-229. 1594. 24. Beutler, B., Krochin, N., Milsark, I. W., Luedke, C. & Ce- 8. Garrett, R., Durie, B. G. M., Nedwin, G. E., Gillespie, A., rami, A. (1986) Science 232, 977-980. Bringman, T., Sabatini, M., Bertolini, D. R. & Mundy, G. R. 25. Kriegler, M., Perez, C., DeFay, K., Albert, I. & Lu, S. D. (1987) N. Engl. J. Med. 317, 526-532. (1988) Cell 53, 45-53. 9. Spriggs, D., Imamura, K., Rodriguez, C., Horiguchi, J. & 26. Balkwill, F., Burke, F., Talbot, D., Tavernier, J., Osborne, R., Kufe, D. S. (1987) Proc. Nat!. Acad. Sci. USA 84, 6563-6566. Naylor, S., Durbin, H. & Fiers, W. (1987) Lancet ii, 1229-1232. 10. Kronke, M., Hensel, G., Schhiter, C., Scheurich, P. & Pfizen- 27. Hibbs, J. B. R., Taintor, R., Chapman, H. A. & Weinberg, maier, K. (1988) Cancer Res. 48, 5417-5421. J. B. (1977) Science 197, 279-281. 11. Aggarwal, B. B., Eessalu, T. E. & Hass, P. E. (1985) Nature 28. Ruco, L. P. & Meltzer, M. S. (1978) J. Immunol. 121, 2035- (London) 318, 665-667. 2042. 12. Paul, N. L. & Ruddle, N. H. (1988) Annu. Rev. Immunol. 6, 29. Zacharchuk, C. M., Drysdale, B.-E., Mayer, M. M. & Shin, 407-438. H. S. (1983) Proc. Nat!. Acad. Sci. USA 80, 6341-6345. 13. Frkter-Schroder, M., Risau, W., Hallmann, R., Gautschi, P. & 30. Urban, J. L., Shepard, H. M., Rothstein, J. L., Sugarman, Bohlen, P. (1987) Proc. Nat!. Acad. Sci. USA 84, 5277-5281. B. J. & Schreiber, H. (1986) Proc. Natl. Acad. Sci. USA 83, 14. Leibovich, S. J., Polverini, P. J., Shepard, H. M., Wiseman, 5233-5237. D. M., Shiveley, V. & Nuseir, N. (1987) Nature (London) 329, 31. Hensel, G., MAnnel, D. N., Pfizenmaier, K. & Kronke, M. 630-632. (1987) Lymphokine Res. 6, 119-125. 15. Dayer, J. M., Beutler, B. & Cerami, A. (1985) J. Erp. Med. 162, 32. Kornbluth, R. S., Gregory, S. A. & Edgington, T. S. (1988) J. 2163-2168. Immunol. 141, 2006-2015. Downloaded by guest on September 28, 2021