TNF-α Blockade Down-Regulates the CD40/CD40L Pathway in the Mucosal Microcirculation: A Novel Anti-Inflammatory Mechanism of Infliximab in Crohn’s Disease¹

The CD40/CD40L (CD40L)³ system is crucially involved in immunity (1, 2), because cognate interactions between CD40 and CD40L generate intra- and intercellular signals that result in up-regulation of a variety of cell surface and soluble molecules that ultimately impact on humoral and cellular immunity as well as inflammation (2, 3). CD40 and CD40L are overexpressed in both forms of inflammatory bowel disease (IBD), Crohn’s disease (CD), and ulcerative colitis (4, 5, 6, 7, 8, 9). In active IBD tissue, CD40 is up-regulated in several cell types including microvascular endothelial cells, whereas CD40L is overexpressed by lamina propria T cells and platelets, and its soluble form (sCD40L) is increased in the circulation of IBD patients (7, 8, 10, 11, 12). These observations strongly indicate that the CD40/CD40L pathway plays a key pathogenic role in intestinal inflammation (13). The up-regulation of CD40 in the microcirculation of IBD-involved mucosa is of particular interest because the CD40 pathway is intimately involved in vascular inflammation. Indeed, stimulation of CD40-bearing endothelial cells triggers multiple inflammatory signals, resulting in leukocyte recruitment and amplification of tissue injury (14, 15). We have previously reported that intestinal mucosal microvascular CD40 is biologically functional, as demonstrated by in vitro experiments showing that CD40 engagement is able to potently activate human intestinal microvascular endothelial cells (HIMEC), bringing about the production of proinflammatory cytokines and chemokines, up-regulation of cell adhesion molecules, and chemoattraction and adhesion of T cells (7, 8). These and other observations make the CD40/CD40L pathway a rational target for therapeutic intervention because disrupting CD40-CD40L interactions, either with blocking Abs or antisense oligonucleotides, could represent a novel approach to turn off intestinal inflammation. Supporting this possibility are reports showing that shutting down the CD40/CD40L system is very effective in dampening inflammation in in vitro cellular systems (7, 8, 9) as well as in vivo in animal models of experimental colitis (16, 17, 18, 19).

CD40 belongs to the TNFR superfamily, whereas CD40L is a member of the TNF gene superfamily (3, 20, 21, 22, 23). During inflammation CD40 and TNF-α influence each other’s biological activity: CD40 ligation leads to production of TNF-α in various cell types (24, 25, 26), while TNF-α up-regulates expression of CD40 and promotes CD40 downstream signaling pathways (27, 28). Because of its potent proinflammatory activity, several highly specific neutralizing Abs have been developed against this molecule for therapeutic purposes (29). Among these, infliximab is a chimeric mAb highly effective in neutralizing the biological activity of TNF-α and is currently used for severe, steroid-refractory, or fistulizing CD (30). Although some of the potential mechanisms of action of infliximab have been explored, it is still unclear what effects this Ab has on the multiple pathways that orchestrate chronic intestinal inflammation (31, 32). In the present study, we investigated whether some of infliximab’s anti-inflammatory effects are mediated through the CD40/CD40L pathway, possibly by inhibiting CD40 in the intestinal microvasculature or CD40L in platelets and T cells, either in its membrane-bound or soluble (sCD40L) forms. Our results show that infliximab effectively down-regulates in vivo CD40 expression by intestinal microvessels in CD mucosa and blocks TNF-α-induced CD40 up-regulation by HIMEC in vitro. Additionally, infliximab significantly reduces levels of sCD40L in the circulation of CD patients as well as CD40L surface expression on T cells.

Patients with active CD undergoing infliximab treatment for the first time were studied. This group consisted of 18 subjects (7 males and 11 females), with a mean age of 31.8 years (19–50 years), and their clinical characteristics are summarized in Table I. A group consisting of 36 healthy subjects, matched for age and sex with the CD patients, was studied as control. The diagnosis of CD was confirmed by clinical, radiological, endoscopic, and histological criteria, and clinical activity was measured by the CD activity index (CDAI) (33). Blood samples and mucosal biopsies were collected in CD patients before and 2–4 wk after completing infliximab therapy. In addition, intestinal tissues were obtained from patients admitted for bowel resection due to malignant and nonmalignant conditions, including colon cancer, benign polyps, and diverticulosis. This study was approved by the Institutional Review Board of the Catholic University (Rome, Italy) and Hospital Clinic (Barcelona, Spain).

Immunohistochemical staining for CD40 and VCAM-1 was performed as previously reported (8). For identification of CD40-positive endothelial cells in intestinal tissues, paraffin-embedded sections of histologically normal and CD-involved colonic mucosa were obtained before and after infliximab therapy, cut at 3-μm thickness, deparaffinized, hydrated, blocked for endogenous peroxidase using 3% H₂O₂/H₂O, and subsequently subjected to microwave epitope enhancement using a Dako Target retrieval solution (DakoCytomation). Incubation with the primary CD40 polyclonal Ab (Santa Cruz Biotechnology) and VCAM-1 (R&D Systems) was conducted at 1/200 and 1/100 dilution, respectively, for 30 min at room temperature. Detection was achieved using a standard streptavidin-biotin system (Vector Laboratories), and Ag localization was visualized with 3′-3-diaminobenzidine (Vector Laboratories). Staining intensity was scored blindly from 1 (absent) to 4 (strong).

Surgical specimens used for isolation of HIMEC were all of colonic origin. Mucosal strips were obtained from surgically resected tissue and processed as previously described (7). Briefly, processing consisted of enzymatic digestion followed by gentle compression to extrude endothelial cell clumps, which adhered to fibronectin-coated plates and were subsequently cultured in MCDB131 medium supplemented with 20% FBS, antibiotics, heparin, and endothelial cell growth factor. Cultures of HIMEC were maintained at 37°C in 5% CO₂, fed twice a week, and split at confluence. HIMEC were used between passages 3 and 10.

Platelets were isolated using an established consensus protocol that prevents stasis and activation, as previously reported (7). Briefly, peripheral venous blood samples were collected without tourniquet using 10% sodium citrate as anticoagulant and transferred into polypropylene tubes. Platelet-rich plasma was obtained by centrifugation at 180 × g for 8 min at room temperature. Platelets were isolated by centrifugation (1200 × g for 2 min), washed, and resuspended in Tyrode-HEPES buffer (10 mmol/L HEPES, 12 mmol/L NaHCO₃, 137 mmol/L NaCl, 2.7 mmol/L KCl, and 5 mmol/L glucose, pH 7.4). The resulting platelet population was essentially free (<0.1%) of erythrocytes and peripheral mononuclear cells and >99% pure as assessed by flow cytometry for expression of the platelet-specific Ag CD42b. To ensure that the isolation procedure did not artificially stimulate the platelets, their activation state was assessed before and after isolation by measuring P-selectin expression levels in whole blood and purified platelets by flow cytometry. Peripheral blood T cells (PBT) were isolated by negative selection as previously described (10).

For CD40 and VCAM-1 expression by HIMEC, endothelial cells were plated on fibronectin-coated wells of a 24-well cluster plate at a density of 5 × 10⁴/ml/well. Confluent monolayers were cultured for 24 h in the presence of 100 U/ml (equivalent to 1 ng/ml) of TNF-α, 4 μg/ml sCD40L, or 100 U/ml IL-1β (R&D Systems) in the presence and absence of therapeutic levels (5 μg/ml) of infliximab (34, 35). In some experiments, increasing doses of TNF-α or infliximab were used, and HIMEC were stained at days 0, 1, 3, 5, and 7. In other experiments, HIMEC were cultured for 24 h with TNF-α, washed, and infliximab was added to the monolayers for an additional period of 24 h. At the end of the culture, HIMEC were washed five times in cold PBS, and a single-cell suspension was obtained using a detaching buffer (PBS, 20 mM HEPES (pH 7.4), 10 mM EDTA, and 0.5% BSA), followed by vigorous pipetting. After additional washing, HIMEC were incubated with FITC-conjugated anti-CD40 or FITC-conjugated anti-VCAM-1, or the appropriate isotype control Ab (BD Pharmingen), for 30 min at 4°C, fixed in 1% paraformaldehyde, and analyzed by flow cytometry as previously described (7).

The expression of CD40L on the platelet surface was assessed before and after activation with 0.5 U/ml thrombin (Sigma-Aldrich), in the presence and absence of infliximab, for 15 min at room temperature. Platelets were fixed in 1% paraformaldehyde and incubated with Abs against CD40L (FITC-conjugated-anti-CD40L; BD Biosciences), or CD42b (Santa Cruz Biotechnology), for 30 min at room temperature. PBT were analyzed before and after a 24-h stimulation with polystyrene beads coated with murine mAbs to human CD3 and CD28 (Dynal), in the presence or absence of infliximab, and stained for CD40L as described elsewhere (36). In some experiments, PBT were stained with propidium iodide and analyzed for DNA content (37).

In some experiments, anti-CD3/CD28-activated PBT were incubated with HIMEC in the presence or absence of infliximab or 10 μg/ml anti-CD40L blocking Ab (M90; Immunex) for 24 h at the ratio 1:20 as previously reported (8). Parallel cocultures were conducted in tissue culture wells and in a Transwell system (3-μm pore size). At the end of the coculture, HIMEC were extensively washed to remove all PBT and then stained with FITC-conjugated anti-VCAM-1 as described above.

Samples were analyzed by quantitative flow cytometry using a Coulter Epics XL flow cytometer (Beckman Coulter), and each analysis was performed on at least 10,000 events. Quantification of CD40 and CD40L expression was obtained using the Winlist software program (Verity Software House).

Whole peripheral blood was collected into EDTA-containing tubes, immediately centrifuged at 1000 × g at 4°C, and plasma was stored at −70°C (10). Platelets and PBT were isolated from normal control and CD subjects. Freshly isolated platelets were cultured in Tyrode-HEPES buffer alone or in the presence of thrombin, whereas PBT were cultured in serum-free RPMI 1640 alone or stimulated with anti-CD3/CD28. Parallel cultures were conducted in the presence of infliximab as described above for the CD40 flow cytometric analysis. After 24 h, supernatants were harvested by centrifugation at 800 × g for 5 min at 4°C and stored at −70°C. Plasma and supernatants were thawed once and analyzed for sCD40L content in duplicate using a commercially available ELISA kit with an assay reproducibility of >95% (R&D Systems).

Data were analyzed by GraphPad software) and expressed as mean ± SEM. The Student’s t test or the ANOVA followed by the appropriate post hoc test were used when appropriate. Statistical significance was set at p < 0.05.

We have previously shown that CD40 expression is increased in the mucosal microvessels of CD patients (8). To assess whether such increase could be inhibited by TNF-α blockade, mucosal biopsies were obtained before and after (2–4 wk) infliximab administration to a group of 18 patients with active CD. Indications for infliximab therapy included severe, steroid-resistant or -dependent, fistulizing, or chronically active disease in which standard immunosuppressive therapy had previously failed. Patients’ clinical characteristics are summarized in Table I. As shown in Fig. 1, intestinal mucosal microvessels display abundant surface CD40 expression in the inflamed mucosa (panels 1 and 4). Notably, in biopsies taken from the original biopsy sites of the same patients, CD40 expression was strikingly reduced after infliximab treatment, reaching a level similar to that seen in the normal mucosa of control subjects (Fig. 1, panels 2, 3, and 5). VCAM-1 expression was also detected in the intestinal microvasculature of CD patients (Fig. 1, panel 7), but disappeared after infliximab treatment (Fig. 1, panel 8). Semiquantitative scores revealed a significant reduction of both CD40 (p < 0.001) and VCAM-1 (p < 0.05) expression by intestinal microvessels of CD patients after comparison to conditions before infliximab therapy (data not shown). Because these results suggest that TNF-α blockade is responsible for CD40 down-regulation in vivo, we directly evaluated the modulatory effect of infliximab on HIMEC CD40 expression in vitro by flow cytometry. As shown in Fig. 2, the low spontaneous CD40 expression on HIMEC was strongly up-regulated by TNF-α treatment in a dose-dependent manner, reaching peak expression at 3 days and returning to baseline levels by 7 days. When the same experiments were conducted in the presence of infliximab, up-regulation of CD40 was inhibited in a dose-dependent way (Fig. 2). Comparable results following similar time and dose kinetics were observed with regard to VCAM-1 induction and blockade (Fig. 2 B). That such effect was specifically dependent on TNF-α blockade was demonstrated by the lack of effect of infliximab in preventing CD40 up-regulation induced by the proinflammatory cytokine IL-1β (data not shown). Finally, to investigate whether infliximab could not only prevent but also inhibit established CD40 expression, HIMEC were first stimulated with TNF-α for 24 h, extensively washed, and then treated with infliximab. Under these conditions, blocking TNF-α did not revert TNF-α-induced CD40 expression, as seen by flow cytometric analysis (data not shown).

During gut inflammation, several adhesion molecules are overexpressed on microvascular endothelial cells, VCAM-1 being one of the most prominent (38). Because TNF-α blockade inhibited HIMEC CD40 expression, we investigated whether this inhibitory action was also observed with regard to VCAM-1. At baseline, HIMEC expressed VCAM-1 at negligibly low levels (Fig. 3), but when cells were exposed to TNF-α, VCAM-1 expression increased remarkably (p < 0.01), and infliximab was able to return this expression to baseline levels (Fig. 3). sCD40L was also able to up-regulate VCAM-1, but the combination of sCD40L plus TNF-α increased VCAM-1 expression to levels higher than TNF-α alone (Fig. 3), and exposure to infliximab significantly (p < 0.05) decreased VCAM-1 expression to the levels induced by sCD40L alone. However, infliximab failed to modulate the VCAM-1 expression induced by sCD40L.

Having shown that infliximab has a clear inhibitory effect on the expression of CD40 by mucosal endothelial cells both in vivo and in vitro, we next investigated whether infliximab could also inhibit the expression and secretion of CD40L, the CD40 counterreceptor. Circulating sCD40L levels were measured in the plasma of CD patients before and after infliximab treatment, as well as in normal controls. As shown in Fig. 4, plasma levels of sCD40L were dramatically (p < 0.001) higher in CD patients (11,900 ± 1,730 pg/ml) than in normal controls (110 ± 32 pg/ml), but they were significantly (p < 0.001) reduced after infliximab therapy (5,770 ± 1,120 pg/ml). In all patients, both the CDAI and sCD40L levels decreased after infliximab therapy, but without a significant correlation (r = 0.28) between the two parameters.

We previously reported that, although both activated platelets and T cells can express and release CD40L, platelets are the almost exclusive source of circulating sCD40L in CD patients (10). Because infliximab treatment decreases circulating levels of sCD40L in these patients (Fig. 4), we investigated whether TNF-α blockade hindered the shedding of this molecule from T cells, platelets, or both. All platelets and T cells per unit of blood were isolated, and CD40L surface expression and secretion were assessed in resting cells and cells activated by thrombin or anti-CD3/CD28, respectively, in the presence and absence of infliximab. As shown in Fig. 5, platelet CD40L expression was cryptic. After thrombin stimulation, a high number of platelets expressed CD40L, and this increase was not prevented by infliximab. Few resting PBT expressed surface CD40L but, like platelets, their number increased substantially after anti-CD3/CD28 activation. In marked contrast to platelets, TNF-α blockade resulted in a major reduction of the number of CD40L-positive PBT. To test whether infliximab could also have an inhibitory effect on sCD40L release (39, 40), we measured the concentration of sCD40L in the supernatants of cultures containing autologous platelets and PBT from normal control subjects. As shown in Fig. 6,A, in the absence of stimulation, platelets were virtually the only source of sCD40L, but this situation was drastically reverted upon activation, when PBT became the dominant source of sCD40L. Secretion of sCD40L by platelets was unaffected by infliximab, whereas that of PBT was significantly (p < 0.05) reduced by TNF-α blockade. When PBT from normal control and CD subjects were compared, secretion of sCD40L was significantly greater in unstimulated as well as activated cultures of CD compared with normal control cells (Fig. 6B ). Infliximab treatment significantly (p < 0.01) reduced production of sCD40L by both normal control and CD PBT compared with untreated activated cells (Fig. 6 B).

The above results demonstrate that infliximab is very effective in inhibiting the expression of membrane-bound CD40L and release of sCD40L by PBT (Figs. 5 and 6). Because activated T cells express CD40L and they can enhance endothelial VCAM-1 through the CD40/CD40L pathway (8, 13), we finally investigated whether infliximab could affect this mechanism of endothelial cell activation. Adding infliximab to cocultures of HIMEC and anti-CD3/CD28-activated PBT significantly (p < 0.05) reduced VCAM-1 expression (Fig. 7,A). The dependence of VCAM-1 induction on contact with CD40-positive T cells was confirmed by the significant (p < 0.05) inhibition achieved with anti-CD40L blockade (Fig. 7,A) and the failure to induce VCAM-1 expression when activated PBT were separated from HIMEC in a Transwell system (data not shown). Infliximab has also been reported to induce apoptosis of activated T cells, an effect believed to be of central importance in reducing mucosal inflammation in CD (29, 34, 41, 42). We confirmed this observation by assessing apoptosis of PBT in unstimulated and anti-CD3/CD28-activated cells. Unstimulated PBT displayed low levels of spontaneous apoptosis that increased only modestly after anti-CD3/CD28 stimulation (Fig. 7,B). However, when infliximab was added to anti-CD3/CD28-stimulated PBT apoptosis increased considerably (Fig. 7 B), but had no effect on resting cells (data not shown).

The results of this study demonstrate that infliximab down-regulates the CD40/CD40L pathway in CD by acting on both components of this molecular pair. It is now established that this pathway, in addition to playing an essential role in cellular immunity as traditionally acknowledged, is a key component of the pathophysiology of multiple inflammatory disorders, including vascular inflammation (14, 43, 44), which is dependent on the integrated signaling pathways of CD40 and TNF-α. This integration has been confirmed in murine models, where inhibition of CD40 signaling decreases inflammatory infiltrates in established atherosclerotic lesions (45) and TNF-α blockade reduces progression of atherosclerosis (46). A recent but growing body of evidence indicates that vascular inflammation is also an important component of IBD pathogenesis (47). We have previously shown that the CD40/CD40L system is involved in several proinflammatory events in IBD, including chemokine and cytokine production, cell adhesion molecule up-regulation, chemokine mobilization on the endothelial surface, and T cell adhesion to and transmigration across endothelial monolayers (8, 10). In addition, a recent preliminary report has shown that mice in which the CD40 or CD40L genes have been deleted are protected from experimental colitis, and adhesion of leukocyte and platelets to the intestinal microvasculature is deeply impaired in colitic animals in the absence of the CD40/CD40L pathway (48).

In view of the close functional relationship between CD40 and TNF-α and the established therapeutic effects of infliximab through its anti-TNF-α activity, we investigated whether such beneficial effects might be mediated by inhibition of CD40/CD40L-dependent vascular inflammation in the mucosa of CD patients. Supporting evidence for this possibility was initially obtained in vivo, as we found that in patients who received infliximab treatment the expression of CD40 was essentially abolished in the mucosal microvasculature in parallel with a reduction of the inflammatory infiltrates. The same was observed with regard to VCAM-1 expression, which disappeared with clinical and histological improvement. These observations were corroborated by our in vitro studies demonstrating that infliximab incubation effectively and dose-dependently prevented CD40 and VCAM-1 up-regulation on HIMEC. An additional effect of infliximab on HIMEC was the inhibition of endothelial cell activation, as quantified by a significant decreased of VCAM-1 expression. This is important because VCAM-1 expression is physiologically absent in the normal gut (49) but dramatically increased in active IBD, where it promotes recruitment of leukocytes into the inflamed mucosa (38). The crucial contribution of this phenomenon to IBD has been confirmed by a recent report showing that use of CD40 antisense oligonucleotides effectively improves experimental colitis and reduces vascular inflammation, as measured by the reduction of VCAM-1 expression by the mucosal microvasculature in treated animals (19). Taken together, these in vivo and in vitro data suggest that TNF-α plays a crucial role in governing CD40 expression by mucosal endothelial cells in CD and that infliximab abrogates the proinflammatory effects of TNF-α on these vascular cells.

The above results offer compelling evidence that infliximab acts on the CD40 component of the CD40/CD40L system. We next investigated whether TNF-α blockade could also act on the ligand component of the system by measuring levels of circulating sCD40L in CD patients before and after infliximab therapy. An elevation of plasma sCD40L levels is known to be present in CD patients (10), as we previously reported. Clinical improvement, as assessed by a major drop in CDAI, was observed after infliximab therapy, and this response was accompanied by a significant reduction of plasma sCD40L concentrations. To investigate the mechanisms responsible for such effect, we analyzed the individual contribution of each one of the two cellular sources of sCD40L, platelets and T cells. Under resting conditions, platelets are practically the single source of sCD40L in the peripheral circulation because unstimulated T cells produce only minute amounts of the ligand. However, when T cells become activated as it occurs in several inflammatory conditions, this balance is completely reverted, because we discovered that T cells turned out to be a proportionally more abundant source of sCD40L than platelets. This is a novel observation, considering that platelets have been long considered the major source of plasma sCD40L (40, 50). This reversal in the predominant source of sCD40L is likely to be explained by the fact that platelets contain only preformed CD40L and lack synthetic capacity even after stimulation, whereas activated T cells can produce massive quantities of inflammatory molecules for prolonged periods of time (51). When infliximab was added to activated T cells, levels of sCD40L were significantly reduced, while platelet-derived sCD40L remained unchanged. Translating these in vitro findings to our in vivo observations in infliximab-treated CD patients, it is reasonable to speculate that the clinical reduction of sCD40L levels in CD patients is mainly due to the inhibitory action of infliximab on T cells since this Ab fails to alter the amounts of sCD40L shed by platelets.

Among the many functional consequences of inhibiting activated T cells is the decrease in their capacity to bind to and activate endothelial cells, especially under inflammatory conditions such as IBD (52). Therefore, we explored whether infliximab could reduce activation of HIMEC secondary to binding of CD40L-positive T cells. Using VCAM-1 expression as an index of vascular activation (38, 52), we observed that TNF-α blockade significantly reduced the capacity of T cells to trigger vascular inflammation. Although leukocyte activation of endothelial cells depends on multiple factors, two are particularly important to our investigation and may explain the observed findings. The first is the decrease in CD40L expression by activated T cells exposed to infliximab as we discussed above. The second is the recognized ability of infliximab to induce T cell apoptosis (29, 34, 41, 42), an effect we reproduced in our experimental system and that would result in a diminished availability of activated T cells for binding to the inflamed mucosal microvasculature.

Therefore, considering the pathogenic role of the CD40/CD40L system in intestinal inflammation, one could envision a dual beneficial effect of infliximab in CD: one mediated by blocking TNF-α-dependent activation of endothelial cells and another mediated by inhibiting mucosal microvascular inflammation dependent on activated CD40L-positive T cells. The therapeutic implications of ameliorating vascular inflammation with infliximab go beyond IBD and could be relevant to other conditions where endothelial inflammation also plays a pathogenic role, such as rheumatoid arthritis, multiple sclerosis, psoriasis, and atherosclerosis. Moreover, because increased levels of sCD40L are considered a strong risk factor for cardiovascular events (53, 54), and IBD patients are at an increased risk of thromboembolic complications (55), lowering sCD40L levels with infliximab in CD patients could also decrease their risk of sCD40L-related thrombosis (11, 56).

In conclusion, to the best of our knowledge, this the first report showing an in vivo and in vitro effect of infliximab on the CD40/CD40L pathway in CD. This points to a novel mechanism of action of infliximab in this condition, i.e., the amelioration of mucosal inflammation through the disruption of TNF-α-dependent CD40/CD40L-mediated cognate interactions between the intestinal microcirculation and circulating T cells. This mechanism should be added to the growing list of alternative modes of actions of anti-TNF-α therapies that may better explain their immunomodulatory action besides direct neutralization of TNF-α and T cell killing (57).

Dr. Danese acknowledges the technical support provided by “Fondazione Ricerca in Medicina” and Dr. C. Terzini.

The authors have no financial conflict of interest.