Interleukin-1 beta and interleukin-6 release by peripheral blood monocytes in head and neck cancer.

In patients with advanced head and neck squamous cell carcinoma (HNSC), evidence of cell-mediated immunity and monocyte functional abnormalities has been reported. We studied the production of interleukin 1 beta (IL-1 beta) and interleukin 6 (IL-6) by peripheral blood monocytes from 22 patients with HNSC (12 larynx and ten oral cavity cancers) in comparison with monocyte cytokine production of age-matched healthy subjects. Pure monocytes were incubated with and without lipopolysaccharides (LPS) (10 micrograms ml-1) for 4 h at 37 degrees C and IL-1 beta and IL-6 concentrations were determined in supernatants by specific ELISA. There was no significant difference in IL-1 beta levels in monocyte supernatants from cancer in comparison to control subjects; conversely, a higher IL-6 production by unstimulated and LPS-activated cells from HNSC patients than from controls was found. No relationship was observed between cytokine production and cancer stage. The regression analysis evidenced a significant correlation between IL-1 beta and IL-6 monocyte-release in HNSC patients and in controls, so suggesting a possible autocrine control of IL-6 production by other cytokines.     

Cytokine regulation of tumor necrosis factor-alpha and -beta (lymphotoxin)-messenger RNA expression in human peripheral blood mononuclear cells.

The immune response at the molecular level is characterized by a carefully coordinated interplay of both cytokine production and receptor induction. The regulation of these molecules including the closely related tumor necrosis factors alpha (TNF) and beta (lymphotoxin, LT) is still incompletely understood. We have examined the effects of various cytokines on the expression of TNF and LT mRNA in human peripheral blood mononuclear cells (PBMC). Northern blot analysis with total cellular RNA from mixed populations of PBMC revealed that genes coding for TNF and LT were not spontaneously expressed. Treatment of PBMC with recombinant interleukin (IL)-2 resulted in a high level expression of TNF and LT mRNA. Whereas IL-1 beta was equally effective as IL-2 in inducing both TNF and LT mRNA, granulocyte-macrophage colony-stimulating factor selectively induced only TNF mRNA. Both TNF and LT mRNA were minimally induced by IL-1 alpha, IL-3, interferon (IFN)-alpha, or IFN-gamma. Similarly TNF alone had little effect on induction of TNF and LT mRNA. In conjunction with IL-2, cytokines such as IFN-alpha, IFN-gamma, or TNF did not interfere with IL-2 induction of TNF and LT mRNA. Interestingly, IL-4 in combination with IL-2 inhibited the IL-2-driven induction of TNF and LT mRNA. This inhibitory effect of IL-4 was also observed at the level of TNF and LT protein secretion. Furthermore, IL-4 was also inhibitory of IL-2-mediated induction of Tac mRNA in PBMC. These results extend the interrelationship of cytokine regulation of TNF and LT expression. In particular, they reveal the previously unrecognized function of IL-4 in antagonizing the IL-2 induction of TNF, LT, and Tac mRNA in PBMC.     

Regulation of TNF-alpha, IL-1 and IL-6 synthesis in differentiating human monoblastoid leukemic U937 cells

The human monoblastoid tumor cell line U937 was induced to differentiate along the monocyte/macrophage lineage by treatment with 5 x 10(-9) M 12-O-tetradecanoyl phorbol-13-acetate (TPA). Between 2 h and 4 h following TPA-treatment U937 cells started to release significant amounts of TNF-alpha which remained detectable until 8-10 days. A significant IL-1 beta release was measured 24 h-48 h post stimulation and increased levels of IL-1 beta persisted until 20-22 days of culture. In contrast no release of either IL-1 alpha or IL-6 could be detected with 5 x 10(-9) M TPA during the whole time course of the experiments. The sequential induction of TNF-alpha and IL-1 beta appeared to be independently regulated since TNF-alpha release was not required for the onset of IL-1 beta production. Northern-blot analysis confirmed the sequential induction and the long term expression of TNF-alpha and IL-1 beta mRNAs. Western-blot analysis predominantly showed a high molecular weight IL-1 beta protein of about 35 kD. Further investigations on the regulation of cytokine production and release by TPA-differentiated U937 cells revealed that TNF-alpha and IL-1 beta synthesis was not influenced by exogenously added rhTNF-alpha or PGE2, whereas rh gamma-IFN specifically enhanced the IL-1 beta production. Thus, the regulation and intracellular processing of cytokines generated by differentiating U937 cells shows some differences when compared to mature monocytes/macrophages which may be related to the tumorigenic origin of U937 cells or to an incomplete differentiation.     

Specific inhibition of interleukin 1 beta gene expression by an antisense oligonucleotide: obligatory role of interleukin 1 in the generation of lymphokine-activated killer cells.

The purpose of the studies reported here is to determine whether interleukin 1 (IL-1) plays an important role in the regulation of lymphokine-activated killer (LAK) cell induction. The addition of exogenous IL-1 to peripheral blood lymphocyte culture containing suboptimal concentrations of interleukin 2 (IL-2) resulted in induction of cytoplasmic pore-forming protein expression. Polymerase chain reaction results revealed that the mRNAs of both IL-1 alpha and IL-1 beta were induced within 6 h when cultured in IL-2 alone or in a combination of IL-2 and IL-1; however, tumor necrosis factor alpha and beta mRNAs were expressed earlier in peripheral blood lymphocytes stimulated with the combination of IL-1 and IL-2. Furthermore, we have examined the functional role of endogenous IL-1 in LAK activity. The lytic potential was significantly inhibited by an IL-1 receptor antagonist, which could block IL-1-mediated effects, or by specific neutralizing antibodies for IL-1, suggesting that the extracellular autocrine/paracrine pathway of IL-1 is involved in LAK activation. However, a synthetic IL-1 beta antisense oligonucleotide, which could specifically inhibit intracellular IL-1 beta protein expression as detected by Western blot, was more effective in reducing LAK killing, but it could not suppress the cytotoxicity generated by exogenous IL-1 plus IL-2. These findings clearly indicate the existence of an intracellular IL-1 autocrine circuit. Taken together, our results strongly indicate that IL-1 should be considered an obligatory factor in the regulation of IL-2-mediated lymphocyte functions.     

Cell-cycle regulator cyclin D1 mRNA and protein overexpression occurs in primary malignant gliomas

We compared the in vitro effects of various cytokines on the expression of monocyte chemoattractant protein-1 (MCP-1) in glioma cell lines and found that MCP-1 expression was highly induced by tumor necrosis factor-alpha (TNF alpha) and interleukin-1 beta. The intra-tumoral injection of TNF alpha in rat glioma model increased the in vivo expression of MCP-1 at 1 to 12 h after the injection and induced macrophage infiltration into tumor tissue. The injection of TNF alpha into post-operative tumor cavity of human malignant glioma also increased the concentration of MCP-1 in the cavity fluid at 24 to 38 h after injection. These data, together with the previous finding that the growth of transplanted MCP-1-transfected cells was significantly inhibited by infiltrated macrophages, suggest that injection of TNF alpha inhibits turner growth via the induction of MCP-1 expression.     

The difference in p53 antioncogene transcription in human monocytes and lymphocytes.

The p53 gene is associated with malignant transformation as well as 'antioncogene' activity. In this report expression of p53 in resting and activated human blood monocytes and lymphocytes was studied. It is shown that human monocytes freshly isolated by continuous Percoll-gradient centrifugation contain detectable levels of p53 mRNA. Stimulation of monocytes by the potent activation inducer Staphylococcus aureus Cowan I (SAC) for 3-5 h caused the disappearance of p53 mRNA. In contrast, induction of a high level of tumor necrosis factor alpha mRNA was detected. The addition of cycloheximide did not increase the p53 mRNA content in stimulated monocytes, and decreased the mRNA level in resting cells. p53 mRNA was absent in freshly isolated lymphocytes and in resting cells cultured for 20 h. Activation of lymphocytes by phytohemagglutinin caused accumulation of p53 mRNA. We suggest that p53 gene regulation and functions might be different in human monocytes and lymphocytes.     

Involvement of leukocyte (beta 2) integrins (CD18/CD11) in human monocyte tumoricidal activity.

Appropriately activated mononuclear phagocytes mediate contact-dependent tumoricidal activity. Adhesion structures involved in contact-dependent tumor cytotoxicity have not been defined. The present study was aimed at identifying the adhesion structures involved in the tumoricidal activity of activated (IFN-gamma + LPS) human monocytes. Tumor cells of different histological origin were used as targets in a 48-hr cytolysis assay. Anti-CD18 (integrin beta 2 chain) monoclonal antibodies (MAbs) substantially (50-80%) inhibited human monocyte cytotoxicity. When the role of different a-chains was studied, anti-alpha L (CD11a, LFA1), anti-alpha M (CD11b, Mac-1) and anti-alpha X (CD11c, p150,95) caused marginal inhibition, but the effect of the 3 combined was comparable to that of anti-CD18. Anti-CD18 MAb did not affect the release of various cytotoxic molecules (e.g. TNF) by activated human monocytes. Activated monocytes showed augmented binding to target cells and anti-CD18 MAb inhibited the binding of resting and activated monocytes to tumor target cells. While IFN-gamma alone augmented expression of leukocyte integrins and LPS had no effect, the 2 activation signals, combined for optimal stimulation of tumoricidal activity, resulted in no appreciable increase in these leukocyte adhesion molecules, as assessed by flow cytometry. Our results suggest that the augmented CD18-dependent binding of activated monocytes on tumor cells depends mainly upon changes in the adhesive properties of these molecules rather than upon increased numbers on the cell surface. Anti-ICAM-1 MAb significantly reduced monocyte cytotoxicity on tumor cells, which is consistent with a role of the CD11/CD18 adhesion pathway. These results implicate "activated" leukocyte (beta 2) integrins (CD11/CD18) as important adhesion molecules in the contact-dependent tumoricidal activity of human monocytes.     

Sensitivity of resistant human tumor cell lines to tumor necrosis factor and adriamycin used in combination: correlation between down-regulation of tumor necrosis factor-messenger RNA induction and overcoming resistance.

A number of human tumor cell lines of various histological origin were examined for their sensitivity and resistance to tumor necrosis factor-alpha (TNF) and Adriamycin (ADR). Six ovarian lines, and one each of a renal, lung, and B-cell line, were tested for putative mechanisms of resistance to these agents. Cytotoxicity resulting from TNF or ADR showed no overall correlation in these lines. The combination of TNF and ADR produced enhanced cytotoxicity against these tumor lines and furthermore resulted in overcoming the resistance of TNF or ADR alone or in combination. A proposed mechanism of TNF resistance in tumor cells is the endogenous production of TNF mRNA and protein. There was a positive correlation between resistance to TNF and the constitutive production of TNF mRNA and protein. The TNF-resistant lines that did not constitutively produce TNF mRNA and protein and the three TNF-sensitive tumor lines exhibited up-regulation of their TNF mRNA in the presence of TNF or phorbol myristate acetate/ionophore, but did not secrete any detectable protein. Due to the enhanced cytotoxicity seen with the combination of TNF and ADR, the effect of this combination on the level of TNF mRNA was examined. ADR alone reduced the constitutive level of TNF mRNA and in combination with TNF reduced the level of induction produced by TNF. This down-regulation of TNF mRNA by ADR may play a role in the enhanced cytotoxicity seen with the combination of these 2 agents.     

The development of human tumor-cell resistance to TNF-alpha does not confer resistance to cytokine-induced cellular cytotoxic mechanisms.

We have derived a TNF-alpha-resistant clone (RA-I) from the parental TNF-sensitive human breast-adenocarcinoma cell line (MCF-7). The acquisition of TNF-resistance was not associated with endogenous TNF production or with differential levels of TNF receptors since both MCF-7 and RA-I display comparable TNF-receptor expression. We have investigated the relationship between acquisition of resistance to TNF and susceptibility to lysis by cytokine-activated effectors. Experiments were performed using human peripheral-blood monocytes stimulated with IL-2, IFN or GM-CSF, and lymphokine-activated killer cells as effector cytotoxic cells. Our data indicate that both TNF-resistant (RA-I) and TNF-sensitive (MCF-7) cells were killed by IL-2-activated monocytes. Incubation of monocytes with IFN also resulted in the activation of their tumoricidal activity against MCF-7 and RA-I. When stimulated monocytes were pre-incubated in the presence of a TNF-specific neutralizing monoclonal antibody, prior to co-culture with target cells, no effect on their lytic capacity was observed. Thus, the monocyte killing does not appear to involve the membrane form of TNF. These observations suggest that, in our experimental system, IL-2 and IFN are able to induce non-TNF-mediated mechanisms of cytotoxicity by monocytes. Experiments performed using GM-CSF and LPS for monocyte stimulation indicate that, although both reagents were efficient in inducing the membrane form of TNF on monocytes, they did not enhance the cell-killing capacity towards MCF-7 and RA-I targets. Furthermore, using IL-2-stimulated LGL as effector cells, we show in this study that the TNF-resistant clone RA-I was as sensitive as MCF-7 to human LAK cells.     

Prolonged exposure to IL-1beta and IFNgamma induces necrosis of L929 tumor cells via a p38MAPK/NF-kappaB/NO-dependent mechanism

Interleukin-1beta (IL-1beta) is a cytokine that shares with tumor necrosis factor (TNF) the ability to initiate largely similar signaling pathways, leading to proinflammatory gene expression. In contrast to TNF, however, IL-1beta is not believed to induce tumor cell death. Here we demonstrate that prolonged treatment with IL-1beta, in combination with interferon-gamma (IFNgamma), is cytotoxic for L929 tumor cells. IL-1beta/IFNgamma-induced cytotoxicity requires only minimal amounts of IL-1beta and shows morphological features of necrosis. Although TNF induces a similar response, we could exclude a contribution of endogenous TNF production in the effect of IL-1beta/IFNgamma. Cell death in response to IL-1beta/IFNgamma is independent of caspases, but requires the IL-1beta/IFNgamma-induced production of inducible nitric oxide synthase (iNOS) and NO. Moreover, necrosis and iNOS/NO production could be prevented by treatment of the cells with a p38 mitogen activated protein kinase (p38MAPK) or IkappaB kinase beta inhibitor. Altogether, these findings demonstrate that prolonged exposure to IL-1beta plus IFNgamma induces L929 tumor cell necrosis, via a p38MAPK and nuclear factor-kappaB (NF-kappaB)-dependent signaling pathway, leading to the expression of iNOS and the production of toxic NO levels.     

Correlation between in vivo induction of cytokine gene expression by flavone acetic acid and strict dose dependency and therapeutic efficacy against murine renal cancer.

The investigational chemotherapeutic drug flavone acetic acid (FAA) acts as an immunomodulator by augmenting natural killer activity in both humans and rodents after in vivo administration. The accumulated data derived from a series of experiments also demonstrates that FAA synergizes with interleukin 2 (IL-2) for the treatment of murine renal cancer. The immunomodulatory and immunotherapeutic effects of FAA are strictly dose dependent with doses of FAA greater than 150 mg/kg effectively synergizing with IL-2, and doses less than 150 mg/kg exhibiting very little therapeutic effect. The antitumor and immunomodulatory effects of FAA are more pronounced in vivo than in vitro. Collectively, these results suggested that cytokines induced by FAA may contribute to these effects, and that the induction of such cytokines may also be very dose dependent. Studies were therefore initiated to investigate whether the in vivo administration of FAA would alter the expression of cytokine mRNA in leukocytes. Splenic leukocytes or liver nonparenchymal cells from untreated and FAA-treated mice were used as a source of RNA for Northern blot analysis. Interferon alpha and interferon gamma mRNA in the spleen was upregulated within 1.5 h after FAA administration, with peak induction occurring by about 2 h. An upregulation of tumor necrosis factor alpha mRNA was detected in the spleen by 0.5-1 h after treatment with peak induction occurring by 1-1.5 h. Induction of tumor necrosis factor alpha mRNA was also detected in hepatic nonparenchymal cells. No up-regulation of splenic mRNA for tumor necrosis factor beta, IL-1 alpha or beta, or IL-2 was detected after FAA administration. IFN and TNF activities were detectable in the serum by bioassay immediately following the appearance of mRNA in FAA mice. The observed up-regulation by FAA of cytokine mRNA and the corresponding serum protein was strictly dose dependent with substantial induction of both mRNA and proteins occurring only at FAA doses greater than or equal to 150 mg/kg, a dose range also shown to be the minimum required for immunomodulatory and immunotherapeutic effects. In summary, these results demonstrate that FAA acts as a potent inducer of at least three cytokines in vivo, and suggest that the immunomodulatory and immunotherapeutic effects of FAA may be partially mediated by these induced cytokines.     

In vitro and in vivo effect of adriamycin therapy on monocyte activation by liposome-encapsulated immunomodulators

The purpose of these studies was to determine the effect of Adriamycin (ADR) on the ability of liposome-encapsulated immunomodulators to activate human blood monocytes to the tumoricidal state. We undertook these experiments because we envisioned using encapsulated activators in addition to chemotherapy to destroy pulmonary micrometastases in patients with osteosarcoma (OS). Prior to the initiation of such therapy, it was important to determine whether chemotherapy interferes with monocyte function. First, human peripheral blood monocytes were isolated from normal donors and preincubated with ADR (0.5-500 ng/ml) for 1 h and then washed prior to the addition of free or liposome-encapsulated activators. After 18-24 h incubation, the activating agents were washed off and [125I]IdUrd-labeled A375 melanoma cells were added. Lysis of radiolabeled tumor cells was quantified 72 h later. Monocytes were also incubated with ADR for 24 h in the presence of free or liposome-encapsulated activators and their cytotoxicity quantified. ADR had no effect on the ability of either free or liposome-encapsulated agents to activate monocyte tumoricidal function. We also studied the in vivo effect of ADR therapy on monocyte function in nine patients with OS. At the time of diagnosis and 1 month after ADR therapy (75 mg/m2) patient monocytes could be activated to the tumoricidal state by liposome-encapsulated agents at levels equal to or greater than pretherapy levels. Monocytes isolated from four patients with OS 1 day after ADR therapy and then activated by liposome-encapsulated agents also demonstrated tumoricidal activity. These studies indicated that the monocytes isolated from osteosarcoma patients treated with ADR can be activated in vitro to kill tumor cells and that additional therapy with liposome-encapsulated immunomodulators may be combined with ADR in the treatment of metastatic pulmonary OS.     

Growth regulation of the AML-193 leukemic cell line: evidence for autocrine production of granulocyte-macrophage colony-stimulating factor (GM-CSF), and inhibition of GM-CSF-dependent cell proliferation by interleukin-1 (IL-1) and tumor necrosis factor (TNF alpha)

The human leukemic cell line AML-193 was tested for its proliferative response to endogenously produced autocrine factors and to a variety of cytokines and colony-stimulating factors. Cells grown in the absence of GM-CSF incorporated tritiated thymidine, and this was partially reversed by adding neutralizing anti-GM-CSF antibodies to the culture medium, suggesting that it was due, at least in part, to autocrine GM-CSF production. This was confirmed by immunopurification of a GM-CSF-like activity from cell supernatant of AML-193 cells grown in serum free medium in the absence of exogenous GM-CSF. When AML-193 cells were cultured with GM-CSF in combination with other cytokines, Interleukin-1 alpha and beta (IL-1 alpha and beta), Interleukin-3 (IL-3), Interleukin-6 (IL-6), granulocyte colony-stimulating factor (G-CSF) and tumor necrosis factor alpha (TNF alpha), none of them affected the concentration of GM-CSF required to induce 50% of maximum proliferation (D50). However, the maximum proliferation induced by GM-CSF alone was drastically decreased by IL-1 alpha, IL-1 beta and TNF alpha. Inhibition caused by exposure of the AML-193 to IL-1 for up to 24 hr was reversible, ruling out a direct cytotoxic effect.     

Comparison of recombinant tumor necrosis factor and the monocyte-derived cytotoxic factor involved in monocyte-mediated cytotoxicity

A direct comparison of recombinant tumor necrosis factor (rTNF) and the monocyte-derived cytotoxic factor (CF) which is involved in monocyte-mediated cytotoxicity revealed immunological, physiochemical, and biological similarities, indicating that TNF is an effector molecule in monocyte-mediated cytotoxicity. Neutralizing antiserum raised against rTNF completely inhibited the ability of CF-containing monocyte supernatants to induce cytolysis and cell death of sensitive target cells and, conversely, antiserum raised against purified CF completely inhibited the cytotoxic activity of rTNF. Both CF and rTNF have an apparent isoelectric point of 5.8-5.9 as determined by chromatofocusing, and a molecular weight of about 40,000 as determined by gel filtration. Moreover, when present in monocyte supernatants with a total protein concentration of about 1 mg/ml and 0.1% sodium dodecyl sulfate, both CF and rTNF migrated with a molecular weight of about 35,000 upon sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Pure rTNF, however, migrated with a molecular weight of 17,000, suggesting that the relative amount of sodium dodecyl sulfate to protein is critical for dissociating the apparent dimeric structure of TNF. CF and rTNF were also similar with respect to their ability to kill various types of target cells the sensitivity of which to TNF differ, and the dose-response curves of cytotoxicity obtained with CF-containing monocyte supernatants and rTNF were similar. As is the case with anti-CF serum, anti-rTNF serum inhibited drug-dependent cellular cytotoxicity and cytolysis mediated by both freshly isolated monocytes and in vitro cultured unactivated and lymphokine-lipopolysaccharide activated monocytes, indicating that TNF is an effector molecule in both drug-dependent cellular cytotoxicity and "classical" monocyte-mediated cytotoxicity.     

Tumor necrosis factor as effector molecule in monocyte mediated cytotoxicity

A newly developed assay system which uses actinomycin D (Act D) pretreated Wehi 164 target cells allows for the measurement of human monocyte cytotoxicity in a 7-h 51Cr release assay. Using the monocyte specific monoclonal antibody M42 in a direct rosetting procedure we confirm herein that among human peripheral blood mononuclear cells cytotoxicity is restricted to monocytes. When applying stringent conditions that exclude exogenous lipopolysaccharide (LPS) we could demonstrate that as little as 0.1 ng of LPS per ml triggers this cytotoxicity. Further, a factor can be detected in supernatants of mononuclear cells which is also cytotoxic against Act D treated Wehi 164 cells. This cytotoxic factor can be triggered by LPS within 4 h, but at as low a LPS concentration as 0.001 ng/ml. Since one of the LPS triggered monocyte products is tumor necrosis factor (TNF), we tested the effect of recombinant TNF cloned from the U937 cell line and we could show potent lytic activity against Act D pretreated but not, or only minimally, against untreated Wehi 164 target cells. Recombinant TNF rapidly lysed the target with significant specific release occurring as early as after 3 h in the assay. By contrast, recombinant interleukin 1 gave no lysis while lymphotoxin derived from the RPMI 1788 cell line was effective. An affinity purified antiserum directed against TNF neutralized the lytic activity of recombinant TNF and also the cytotoxic factor produced by LPS triggered mononuclear cells, while the antiserum was ineffective against lymphotoxin. Further, the antiserum when added to the assay of effector cells and Act D treated Wehi 164 cells also completely ablated cytotoxic activity. Size fractionation of cytotoxic factor and recombinant TNF by high pressure liquid chromatography led to a superimposable peak of cytotoxicity in the molecular weight range of 9,500-17,000. Further, immunoblotting with the anti-TNF antibody revealed the same Mr 15,500-16,500 band for the recombinant TNF and LPS triggered cytotoxic factor. Taken together, our data demonstrate that the cytotoxic activity of human monocytes against Act D treated Wehi 164 is mediated entirely by a LPS triggered molecule that is very similar or identical to the human tumor necrosis factor. The assay system thus provides a powerful tool to analyze the biology of TNF in humans.     

Investigation of cytokine gene expression in human colorectal cancer

RNA was extracted from 28 samples of colorectal cancer and 26 samples of adjacent normal bowel. Northern blotting analysis showed the presence of mRNA for tumor necrosis factor (TNF) in 15 of 28 cancer samples and 6 of 26 matched normal areas. In 10 patients, TNF mRNA was found only in the tumor; in 5, TNF mRNA was seen in tumor and normal areas; and in only 1 was TNF mRNA seen in the normal, but not the malignant, area. The expression of TNF mRNA was not related to the stage of disease, degree of lymphocyte infiltration, or necrosis in the tumor. Blots were reprobed for gamma-interferon, interleukin (IL) 1 alpha and beta, IL-6, and transforming growth factor beta 1 mRNA. One tumor sample was positive for IL-1 beta, and one normal sample expressed interferon gamma mRNA. All samples had transforming growth factor beta 1 mRNA, and there was no obvious difference between levels in tumor tissues or adjacent normal areas. In situ hybridization studies with a TNF riboprobe showed that TNF mRNA was only detectable in a small minority of mononuclear and predominantly stromal cells. Immunohistochemistry on sequential sections showed that CD4- and CD8-positive lymphocytes, and macrophages, were present in the stroma. An antibody to the macrophage C3b receptor identified a minority population whose distribution corresponded closely to the cells labeled with the TNF riboprobe.