Decreases in T-Cell Tumor Necrosis Factor α Binding With Interferon Beta Treatment in Patients With Multiple Sclerosis FREE

MULTIPLE SCLEROSIS (MS) is characterized pathologically by areas of primary demyelination and perivascular inflammatory infiltrates of lymphocytes and macrophages. Immune-mediated factors, perhaps induced by viral infection, are likely to be implicated in the pathogenesis of MS.^{1 - 2} Cytokines are important central nervous system immunomodulators³ ; unbalanced immune responses in patients with MS may depend on a cytokine network derangement.

The cytokine tumor necrosis factor α (TNF-α) is a 17-kd protein, 157 amino acids long, produced primarily by macrophages but also by T and B lymphocytes, natural cytotoxic cells, astrocytes, and microglial cells, in response to a wide variety of stimuli.^{4 - 5} Besides its antitumoral activity, TNF-α is involved in inflammation, viral replication, and immune modulation.

Tumor necrosis factor α is linked to the inflammatory demyelinating process of MS and experimental autoimmune encephalomyelitis, an animal model for human MS. It has been identified immunohistochemically in the lesions of brains of patients with MS.⁶ It is able to induce the expression of major histocompatibility complex class II molecules on gliocytes⁷ ; antigen presentation by astrocytes positive for major histocompatibility complex may enhance the local immune response and, thus, facilitate continuous growth of lesions. Several authors have demonstrated a significant rise in TNF-α messenger RNA and production by blood mononuclear cells before onset of an acute exacerbation.^{8 - 9}

Tumor necrosis factor α mediates its biological effects through binding to a single class of specific cell surface high-affinity receptors of 55- to 60-kd and 75- to 80-kd molecular size, referred to as p60 (type I) and p80 (type II). Increased concentrations of the former are measured in the serum of patients with MS.^{10 - 11} The p60 form of the TNF-α receptor contains 455 amino acid residues; the p80 form, 461 amino acids.^{12 - 14}

Both receptor types bind TNF-α with equal affinity and transduce signals independent of each other. Receptor expression is regulated by a variety of agents, including cytokines (interferons, interleukin [IL] 1, IL-2, IL-4, IL-6, IL-8, and TNF-α), protein kinase activators, corticosteroids, and Ca⁺⁺ ionophores.¹² After TNF-α binds its receptor (either the p60 or the p80 form), the nuclear factor-κB–related transcriptional pathway is activated.

Since the early 1980s, interferons have been proposed for the treatment of MS, because of the evidence that an immunoregulatory derangement or a viral infection, or both, may have a pathogenetic role.¹⁵ Interferon gamma, however, has been proved to activate MS,¹⁶ and no effect was observed when interferon beta was given intravenously.¹⁷ On the contrary, intrathecally administered interferon beta has been reported to reduce the exacerbation rate in patients with relapsing-remitting MS,¹⁸ although with discordant results in other trials.¹⁹ Studies of parenterally administered human recombinant interferon beta in patients with relapsing-remitting MS showed fewer relapses^{20 - 21} and a significant reduction in the average number of gadolinium-enhancing lesions on magnetic resonance images in treated patients.^{21 - 22}

In a pilot study, peripheral-blood monocytes from patients with MS who received interferon beta demonstrated decreased mitogen-driven IL-2 receptor (IL-2R) expression.²³ Brod et al^{24 - 25} found, in patients with stable relapsing-remitting MS treated with interferon beta-1b, that after 3 months of treatment peripheral-blood mononuclear cells stimulated with concanavalin A secreted significantly more TNF-α, interferon (IFN) γ, IL-2, IL-6, and IL-10 and less IL-4 than concanavalin A–stimulated mononuclear cells before treatment.²⁴ However, at 9 months, concanavalin A–induced TNF-α secretion decreased significantly below baseline, whereas IFN-γ secretion increased above baseline; there were no changes in concanavalin A–induced IL-2 and IL-10 secretion during the 9-month period or in IL-6 secretion during the 6-month period.²⁵ Similar findings of no persistenteffects on inflammatory or counterregulatory anti-inflammatory cytokine synthesis also have been reported in other studies.²⁶

The majority of patients with MS treated with interferon beta-1b show a decreased number of IFN-γ–producing blood lymphocytes,^{27 - 28} but there are also discordant findings in the literature.²⁹

Costimulatory molecules help determine T-cell responses. CD80 (B7-1) and CD86 (B7-2), costimulatory proteins on antigen-presenting cells, bind to CD28 on T cells. During MS exacerbations, the number of circulating CD80⁺ B lymphocytes is significantly increased; therapy with interferon beta-1b markedly reduces the number of these cells and increases CD86⁺ monocyte number.³⁰

Therapy with interferon beta-1b is associated with a remarkable increase in soluble vascular cell adhesion molecule 1, and this effect correlates with the resolution of contrast-enhancing lesions on magnetic resonance images³¹ ; moreover, patients with MS treated with interferon beta-1b show a reduced expression of vascular cell adhesion molecule 1 ligand, the very late antigen 4, on their lymphocytes.³²

We studied T-cell TNF-α binding in untreated patients with MS.³³ We found that lymphocytes from patients bear significantly more receptors than those from healthy controls; among patients with MS, subjects in relapse showed higher T-cell TNF-α binding than those in stable phase. In the present work, we assayed TNF-α receptor binding on T lymphocytes from patients with relapsing-remitting MS treated with human recombinant interferon beta-1b, to study the effects of interferon beta-1b on T-cell activation in vivo.

Eighteen patients with clinically definite stable relapsing-remitting MS³⁴ were consecutively enrolled at the Section of Neurology, Department of Neurosciences, University of Pisa, Pisa, Italy. Relapsing-remitting MS was characterized by clearly defined acute attacks with full recovery or with sequelae and residual deficit on recovery.³⁵ Clinical status was defined according to Kurtzke's³⁶ criteria (Expanded Disability Status Scale). There were 5 men and 13 women, aged 24 to 46 years (mean±SD, 32.6±7.1 years). Ages at onset ranged from 16 to 41 years; disease duration, 1 to 11 years; and Kurtzke Expanded Disability Status Scale scores, 1 to 3.5 (Table 1).

T-cell TNF-α binding was measured in all patients before the initiation of alternate-day subcutaneous treatment with 8×10⁶U of interferon beta-1b³⁷ and after 3 and 6 months during therapy. Serum IL-2 and soluble IL-2R levels were used as markers for circulating activated T cells³⁸ ; for IL-2 and soluble IL-2R immunoenzymatic assays, commercial kits were purchased (Genzyme, Milan, Italy).

Patients were monitored for side effects during the 6-month period, including systemic symptoms, local injection-site reaction, and changes in peripheral total leukocyte count and lymphocyte count.

At the time of the first observation and blood sampling, none of the patients had received corticosteroids for at least 1 month, or immunosuppressive drugs for at least 6 months; during interferon beta-1b treatment, none of them received corticosteroids or immunosuppressive medications. No subject showed any symptoms or signs of concurrent infection either before or during interferon beta-1b therapy.

The control population consisted of 18 age- and sex-matched healthy subjects selected among blood donors or laboratory personnel, aged 18 to 51 years (mean±SD, 31.7±9.6 years).

T-cell TNF-α binding was also assayed in 10 patients with rheumatoid arthritis (mean±SD age, 53.7±15.8 years), and in newly diagnosed neurologic patients suffering from myasthenia gravis (n=8; mean±SD age, 41.8±16.8 years), idiopathic Parkinson disease (n=7; mean±SD age, 68.4±17.6 years), and dementia of the Alzheimer type (n=10; mean±SD age, 72.3±14.4 years), as disease controls.

Patients and controls gave their informed consent before their inclusion in the study.

All subjects underwent venipunture between 7 and 8 AM to avoid circadian variations of lymphocyte subsets.³⁹ Patient and control samples were processed together in the same way, providing sex- and age-matched controls for patients with MS.

After mixing with an equal volume of Ca⁺⁺- and Mg⁺⁺-free Hanks balanced saline solution (Sigma-Aldrich, Milan, Italy), blood was centrifuged with Ficoll-Paque (Amersham Pharmacia Biotech Italia, Milan) at 400g for 30 minutes at room temperature. Buffy coats of mononuclear cells, carefully transferred with sterile pipettes into centrifuge tubes, were spun at 800g for 10 minutes at 4°C to obtain cell pellets. After cell counting with a Neubauer hemocytometer, lymphocytes and monocytes in Dulbecco modified minimal Eagle medium (Sigma-Aldrich) with 10% fetal calf serum (Bio-Whittaker International PBI, Milan) were plated overnight in plastic Petri dishes at 4°C to separate peripheral-blood lymphocytes from macrophages that adhere to the floor of Petri dishes. To obtain pure T cells, lymphocytes were incubated in Dulbecco modified minimal Eagle medium with 10% fetal calf serum at 4°C for 2 hours in petri dishes coated with mouse anti–human IgG (Sigma-Aldrich).

Such a panning procedure was repeated thrice, so that we obtained 98% pure T-cell suspensions; T lymphocytes were identified morphologically and as CD3⁺ cells by flow cytometry.

Recombinant human TNF-α was purchased (Sigma-Aldrich; specific activity, 2×10⁷ U/mg of protein). Fifty micrograms of concentrated TNF-α was reacted for 12 hours on ice with 37 MBq of Bolton-Hunter reagent labeled with iodine 125 (¹²⁵I) (DuPont de Nemours, Milan) in 0.1 mL of 50-mmol/L sodium phosphate buffer (pH 7.4) with 5% sucrose.

The reaction was quenched with 0.2-mol/L glycine (Sigma-Aldrich) in 50-mmol/L sodium phosphate buffer (pH 7.4). Unconjugated iodine was separated from conjugated iodine over a fine column (Sephadex G-25, Amersham Pharmacia Biotech) equilibrated with 50-mmol/L sodium phosphate buffer (pH 7.2) containing 5% sucrose (Sigma-Aldrich), 1-mmol/L dithiothreitol (Sigma-Aldrich), and 1-mg/mL gelatin (Sigma-Aldrich). The ¹²⁵I–TNF-α eluted from the gel column as a monomer with a molecular weight of 17 kd. The peak fractions containing ¹²⁵I–TNF-α were pooled and stored at −70°C in 20-µL samples. The ¹²⁵I–TNF-α was biologically active, as shown by the cytotoxicity assay on mouse L-929 fibroblast cells,⁴⁰ where TNF-α and ¹²⁵I–TNF-α are equipotent (Figure 1). The initial specific activity of the ¹²⁵I–TNF-α of different preparations ranged between 0.37 and 0.74 MBq/µg, defined as bindable counts per minute per microgram of biologically active TNF-α.⁴⁰

Before use in binding assays, ¹²⁵I–TNF-α was diluted in 50-mmol/L sodium phosphate buffer (pH 7.2), 5% sucrose, 1-mmol/L dithiothreitol, and 1-mg/mL bovine serum albumin (Sigma-Aldrich); each dilution was centrifuged for 5 minutes at 15,000g at 4°C. In standard binding assays, 2×10⁶ T cells from each subject were incubated in duplicate at 4°C in 750 µL of Dulbecco modified minimal Eagle medium with Hepes buffer (10 mmol/L, pH 7.2) and 10% fetal calf serum for 2 hours with different amounts of ¹²⁵I–TNF-α (from 0.1 to 0.5 ng). In competitive binding experiments, increasing amounts of unlabeled TNF-α were added to standard binding assays.

Nonspecific binding was determined by adding in duplicate a 100-fold excess of unlabeled TNF-α; at this point, the addition of greater amounts of unlabeled TNF-α did not further reduce the binding of ¹²⁵I–TNF-α. At the end of incubation, samples were layered over 300 µL of a di-n-butylphthalate and dinonylphthalate mixture (Sigma-Aldrich) in microfuge tubes and spun for 4 minutes at 15,000g at 4°C.

Supernatants (containing free ¹²⁵I–TNF-α) were discarded, and cell pellet radioactivity was counted in a gamma counter (Beckman Italia, Milan) with 55% efficiency. A blank reaction, without cells, bound less than 50 counts per minute over the machine background: such a blank was subtracted from all bound counts per minute reported. Final results about binding variables were achieved through the Scatchard equation, using McPherson's⁴¹ LIGAND program, which determines the maximal receptor number (B_max) and the dissociation constant (K_d).

To identify the receptor type, in a subset of patients' and controls' samples (n=10 each), we preincubated T lymphocytes (2×10⁶ cells) with 2 anti–TNF-α receptor monoclonal antibodies⁴² for 1 hour at 4°C.

In group 1, we used an anti-p60 antibody (htr-9), at a concentration of 15 µg/mL; in group 2, an anti-p80 antibody (utr-1), at 15 µg/mL; in group 3, both antibodies; and in group 4, none (only preimmune serum). Afterward, we assayed TNF-α binding as above.

In a limited set of patients (n=10) and controls (n=10), peripheral blood (3 mL per subject) was collected in EDTA test tubes. White blood cell count and leukocyte differentiation were performed routinely by means of an automated cell counter (Coulter Electronics Ltd, Luton, England) and May-Grünwald Giemsa staining.

One hundred microliters of blood was incubated at room temperature in the dark for 30 minutes with 10 µL of fluorescein isothiocyanate, phycoerythrin, or Quantum Red conjugated monoclonal antibodies against the following antigens: CD4, CD8, CD25 (IL-2Rα), CD120a (p60TNF-α receptor), and CD120b (p80TNF-α receptor).

With triple labeling technique, CD4⁺p60⁺IL-2R⁺, CD4⁺p80⁺IL-2R⁺, CD8⁺p60⁺IL-2R⁺, and CD8⁺p80⁺IL-2R⁺ cell subsets were assayed. The erythrocytes were lysed with a solution containing formic acid (0.1%), and the cells were stabilized and fixed by successive passages in sodium carbonate (50 mmol/L), sodium chloride (250 mmol/L), sodium sulfate (220 mmol/L), and paraformaldehyde (330 mmol/L). Then the samples were counted through a flow cytometer (Epics XL-MCL; Coulter Electronics Ltd) equipped with single 15-mW argon ion laser (excitation wavelength of 488 nm in combination with 530-nm, 575-nm, and 670-nm bandpass filters for fluorescein isothiocyanate, phycoerythrin, and Quantum Red, respectively).

Fluorescein isothiocyanate–conjugated mouse IgG1, phycoerythrin-conjugated mouse IgG1, and conjugated monoclonal antibodies–conjugated mouse IgG1 were used to evaluate nonspecific fluorescence (always less than 0.1%). As autofluorescence control, 10 µL of diluent (0.01-mol/L phosphate buffered saline, pH 7.4, containing 1% bovine serum albumin and 0.1% sodium azide) instead of 10 µL of fluorochrome-conjugated monoclonal antibodies was added to whole blood in test tubes. To exclude monocytes, other leukocytes, and debris, gates were selected on lymphoid cells, as determined by forward and right-angle scatter.

Data were recorded on xyz plots, with antibody-linked fluorescence intensity (in log unit) reported on each axis, ie, anti-CD4, anti-p60 (or alternatively anti-p80), and anti–IL-2R antibody fluorescence values for CD4⁺CD25⁺CD120a⁺ (or CD4⁺CD25⁺CD120b⁺) cells, and anti-CD8, anti-p60 (or anti-p80), and anti–IL-2R antibody fluorescence values for CD8⁺CD25⁺CD120a⁺ (or CD8⁺CD25⁺CD120b⁺) cells. Results were expressed as percentages of each subset to total lymphocytes, and as arbitrary units for fluorescence intensity. All analyses were performed without knowledge of the clinical diagnosis.

Statistical evaluation was performed by analysis of variance, with the SigmaStat software package (SigmaPlot 5.0; Jandel Scientific GmbH, Erkrath, Germany). A value of P<.05 was accepted as significant.

Human T cells from patients with MS and healthy controls constitutively expressed high-affinity TNF-α receptors. The binding of ¹²⁵I–TNF-α was specific, since it was inhibited only by unlabeled TNF-α, whereas IFN-α, IFN-β, and IFN-γ were ineffective at inhibiting TNF-α binding.

Figure 2 shows a representative experiment of competitive binding of ¹²⁵I–TNF-α and unlabeled TNF-α to T cells. Highly enriched T-cell suspensions were incubated with 500 pg of ¹²⁵I–TNF-α and different amounts of unlabeled TNF-α: a 100-fold excess of unlabeled TNF-α was able to prevent greater than 80% of the ¹²⁵I–TNF-α binding on T cells (Figure 2, left). Scatchard analysis of the data produced a linear plot (Figure 2, right) suggestive of a single-binding-site model. Saturation binding experiments gave similar results.

No significant differences in K_d values were found among patients with MS, rheumatoid arthritis, myasthenia gravis, Parkinson disease, and dementia of the Alzheimer type, and healthy controls (mean±SE, 67.9±4.4, 70.2±4.9, 67.2±3.8, 66.7±5.6, 67.5±4.7, and 69.1±4.7 pmol/L, respectively), or among pretreatment and treatment groups. However, highly significant (P<.001) differences in B_max values were observed between patients with MS and controls: 837±33 vs 135±5 (mean±SE) receptors per cell (Figure 3). Significant differences in T-cell TNF-α binding also were observed between patients with MS and subjects with rheumatoid arthritis (756±34 receptors per cell; P=.04), and among patients with MS and subjects with myasthenia gravis, Parkinson disease, and dementia of the Alzheimer type (663±20, 633±28, and 701±22 receptors per cell, respectively; P<.001 for all).

Interferon beta-1b therapy caused a decrease in T-cell TNF-α receptor number; B_max values were significantly (P<.001) lower 3 months after initiation of treatment (452±29 receptors per cell). At 6 months, B_max values significantly decreased further (345±35 receptors per cell). In any case, however, TNF-α binding values in patients with MS remained significantly (P<.001) higher than those of controls.

Linear regression analysis performed in patients with MS (both pretreatment and treatment groups) resulted in no significant correlations between TNF-α receptor B_max values and serum IL-2 levels (or soluble IL-2R levels).

In a subset of patients and controls (n=10), we found that anti-p60 antibody reduced T-cell TNF-α receptor B_max values by 39% and 37% in patients and controls, respectively, whereas with anti-p80 antibody treatment, TNF-α receptor binding decreased by 56% in patients with MS and 58% in healthy controls (Table 2). Therefore, in both subject groups, p60 and p80 forms were approximately 40% and 60% of total T-lymphocyte TNF-α receptors, respectively. No significant differences were observed in the relative percentages, nor in the K_dvalues of both TNF-α receptor forms, between patients and controls (data not shown).

Three-color fluorescence analysis showed that the proportions of CD4⁺CD25⁺CD120a⁺, CD4⁺CD25⁺CD120b⁺, CD8⁺CD25⁺CD120a⁺, and CD8⁺CD25⁺CD120b⁺ cells did not significantly differ in patients and controls, ranging from 0.010 to 0.040. In both subject groups, similar proportions of cells expressed the p60 and the p80 forms of TNF-α receptors: 0.410±0.130 and 0.590±0.100, respectively, in patients and 0.440±0.170 and 0.610±0.190, respectively, in controls (Table 3). By overlaying histograms from patients and healthy controls and comparing them by means of nonparametric statistics, we found that in MS TNF-α binding is significantly (P<.001) up-regulated in both CD4⁺ and CD8⁺ cells. There were no significant differences in anti-p60 (or anti-p80) antibody fluorescence intensity in both patients and controls between CD4⁺CD120a⁺ and CD8⁺CD120a⁺ histograms, and between CD4⁺CD120b⁺ and CD8⁺CD120b⁺ histograms, respectively; this suggests similar TNF-α receptor expression in CD4⁺ and in CD8⁺ cells. As far as IL-2R is concerned, anti-CD25 antibody fluorescence intensity was significantly (P=.02) higher in patients with MS than in healthy controls.

In MS, disease-related immune changes are also found in peripheral-blood lymphocytes.^{33 ,43 - 53} Data from various laboratories suggest that the signals resulting from the binding of TNF-α to its receptor have a role in lymphocyte activation^{54 - 56} ; an enhanced number of high-affinity TNF-α receptors on activated T cells with respect to resting T lymphocytes has been reported in studies on TNF-α receptor expression in vitro modulation.^{57 - 58}

We found previously³³ that untreated patients with MS have significantly more T-cell TNF-α receptors than healthy controls do. In the present work, we confirmed such findings and demonstrated that both p60 and p80 TNF-α receptor forms are up-regulated in patients with MS, with the same ratio (2:3) as in controls.⁴²

In the present study, moreover, we found that systemic treatment with interferon beta-1b is able to cause a decrease in T-lymphocyte TNF-α binding, in proportion to the duration of therapy. In separate in vitro experiments with T lymphocytes from normal subjects, we observed a decreased expression of high-affinity TNF-α receptors (K_d, 71.2±4.2 pmol/L) on concanavalin A–stimulated T cells treated with interferon beta-1b (500 U/mL), as compared with untreated control T lymphocytes activated with concanavalin A (5 ng/mL): 674±26 vs 1228±39 receptors per cell. Similar results were found in T-cell subsets: interferon beta-1b in vitro down-regulated TNF-α binding on concanavalin A–activated CD4⁺ and CD8⁺ T cells from normal donors, causing a significant (P<.001) decrease in TNF-α receptor B_max from 1187±33 to 662±31 receptors per cell and from 1243±44 to 707±29 receptors per cell in CD4⁺ and CD8⁺, respectively. Such findings in vitro parallel our results in vivo: it is likely that interferon beta-1b counteracts T-cell activation.

Given the immunopathogenesis of MS (namely, the activation of peripheral T cells that, after having traversed the blood-brain barrier interacting with endothelial cells, encounter target self-antigens presented by perivascular macrophages and microglia), lymphocytes, monocytes and macrophages, blood-brain barrier endothelia, and microglia emerge as main targets for the action of interferon beta.⁵⁹

Studies on microglia and astrocytes have shown complex findings. No inhibition by interferon beta on HLA-DR up-regulation by IFN-γ was found in adult human microglia,⁶⁰ and the reported antagonism of interferon beta on the IFN-γ–mediated major histocompatibility complex class II molecule expression on cultured human astrocytes and an astrocytoma cell line^{61 - 62} has not been observed in studies on fetal astrocytes.⁶⁰ However, Satoh et al⁶³ demonstrated a counteracting effect of interferon beta on IFN-γ–mediated proliferation of human astrocytes in vitro. Moreover, interferon beta inhibits the ability of IFN-γ–primed rodent microglia and human macrophages to mount a respiratory burst^{59 ,64} ; by contrast, interferon beta up-regulates microglial and macrophage FcR expression approximately 3- to 4-fold,⁵⁹ an effect implying enhanced phagocytosis. Microglia produce nitric oxide in response to both IFN-β and IFN-γ alone,⁶⁵ and each synergizes with the other and with TNF-α.⁵⁹ Pretreatment of astrocytes with interferon beta impairs IFN-γ induction of nitric oxide synthase.⁶⁶ Interferon β increases TNF-α secretion by tissue macrophages and stimulated microglia, and no antagonism between interferon beta and IFN-γ has been reported.⁵⁹

Interferon β inhibits IFN-γ–mediated up-regulation of major histocompatibility complex class II molecules on endothelial cells,⁶⁷ therefore decreasing lymphocyte adhesion, since class II molecules might act as lymphocyte adhesion receptors on endothelium.⁶⁸ Interferon beta-1a selectively down-regulates endothelial basal levels of the adhesion molecule very late antigen 4 and inhibits its up-regulation by interferon gamma.⁶⁹

In patients with MS, very late antigen 4 (and very late antigen 5) expression is significantly higher on cerebrospinal fluid lymphocytes and reduced on peripheral T cells compared with normal controls, suggesting that this adhesion molecule is instrumental in selective lymphocyte recruitment in MS.^{32 ,70} Recently, discordant findings of no down-regulation by interferon beta-1a on the inducible expression of intercellular adhesion molecule 1, vascular adhesion molecule, or E-selectin have been reported.⁷¹

In vitro studies demonstrate interferon beta–mediated inhibition of T-cell receptor–dependent and independent proliferation of both CD4⁺ and CD8⁺ T cells.^{23 ,72 - 74} Interferon beta partially suppresses both antigen- and IL-2–driven proliferation of myelin basic protein–specific T-cell clones from patients with MS that are secreting helper T (T_H) 1 cytokines.⁷⁵ The mechanisms of interferon beta–mediated inhibition of T-cell proliferation are not clear so far. Down-regulation of IL-2R-α chain expression might be responsible, indirectly supported by elevated IL-2 levels in culture supernatants because of reduced consumption. It seems that at least a proportion of T cells fail to express high-affinity IL-2R in the presence of IFN-β.^{23 ,72 - 73 ,75} Moreover, several IFN-β inhibitory effects on T cells may be caused by indirect rather than direct effects. In this regard, the observation of IL-10 induction by interferon beta could be important,^{73 ,76} since IL-10, which is a mainly inhibitory cytokine suppressing the production of proinflammatory T_H1 cells and favoring a T_H2 (regulatory) environment, may directly inhibit T-cell proliferation and IL-2 production.⁷⁷

In addition, interferon beta impairs other aspects of T-cell activation, namely up-regulation of IL-2R, transferrin receptors, CD2 expression, and the production of IFN-γ and TNF-α.^{23 ,72 ,74 - 75 ,78 - 79} Migration of activated T cells through the blood-brain barrier also depends on the ability of lymphocytes to cleave the component of the basement membrane surrounding cerebral endothelium.

Interferon beta-1b markedly decreases in vitro the activity of T-cell matrix metalloproteinases 2 (gelatinase A) and 9 (gelatinase B), lymphocyte enzymes that degrade subendothelial basement membrane constituents, such as fibronectin.^{80 - 81} Thus, interferon beta might hinder the migration of T cells through intracerebral vessel walls, leading to disruption of the blood-brain barrier and recruitment of other inflammatory cells. In support of this notion, interferon beta was most efficient in reversing blood-brain barrier dysfunction in patients with MS, documented by gadolinium-enhanced magnetic resonance imaging.⁸²

On human peripheral-blood monocytes, interferon beta exhibits contrasting effects on individual class II antigens: interferon beta inhibits IFN-γ induction of HLA-DQ but not that of HLA-DR or HLA-DP⁶⁹ ; on the contrary, in some reports, interferon beta actually increases the expression of HLA-DR and HLA-DP.^{83 - 84} In vitro interferon beta induces IL-10 secretion by peripheral-blood monocytes from patients with MS⁸⁵ and significantly improves the nonspecific suppressor activity of mitogen-treated peripheral-blood monocytes.⁸⁶

Interferon γ and TNF-α, the main T-cell–derived proinflammatory cytokines, are present in active MS plaques and have been functionally implicated in the disease process^{6 - 7 ,87 - 88} ; interferon beta given to patients with MS acts on a background of endogenous IFN-γ and TNF-α production. In the active disease, antagonism between exogenously administered interferon beta and endogenous IFN-γ and TNF-α would impair the ability of microglia to mount a respiratory burst and also to up-regulate major histocompatibility complex class II expression, leading to impaired cytotoxic function, decreased recruitment of activated T cells into the central nervous system, and decreased propagation and maintenance of the T_H-cell arm of the immune response. In addition, the direct antiproliferative effect of IFN-β on T cells and on secretion of regulatory cytokines, such as IL-10, would help maintain the anti-inflammatory environment.

A decrease in the number of T-cell TNF-α receptors dependent on interferon beta-1b treatment might be caused, at least in part, by the antagonistic action of interferon beta-1b on TNF-α, and IFN-γ, which is able to up-regulate TNF-α receptors.⁸⁹

Accepted for publication April 24, 1998.

Reprints: Paolo Bongioanni, MD, PhD, Department of Neurosciences, University of Pisa, Via Roma 67, 56100 Pisa, Italy (e-mail: Bongioanni@sssup1.sssup.it).