JOURNAL OF INTERFERON AND CYTOKINE RESEARCH 22:491–501 (2002)© Mary Ann Liebert, Inc.

Mechanisms for Regulation of Cellular Responsiveness to Human IFN-b1a

STEFAN A. DUPONT, SUSAN GOELZ, JAYA GOYAL, and MARIE GREEN

ABSTRACT

Interferons (IFNs) are potent, pleiotropic cytokines, and therefore it is likely that the cell has mechanisms tomodulate IFN activity in response to excessive or prolonged IFN exposure. To investigate this question, Jur-kat T cells were exposed to IFN-b1a in vitro. The effect of dose and frequency of IFN treatment on receptorexpression, the signal transduction pathway, and biologic activity was examined. Results demonstrate that ateven modest doses of IFN (60 IU/ml), cell surface expression of the IFN receptor subunit, IFNAR-1, decreasedsignificantly, and the cells were unresponsive to further IFN treatment. More interestingly, after an initialtreatment with very low concentrations of IFN (,10 IU/ml), even when receptor levels remained normal andphosphorylation of signaling molecules occurred, cells were still refractory to further IFN treatment. Afterwithdrawal of IFN, full cellular responsiveness was a progressive but surprisingly slow process. Cells retreated2 days or 4 days after the initial IFN treatment were still refractory to even high doses (500 IU/ml) of IFN.Cells retreated 1 week after the initial IFN treatment were fully responsive. High levels of Stat1 and Stat2correlated with the block in transcriptional activation of IFN-dependent genes and may be a mechanism bywhich cells can downmodulate an IFN response. Similar results were obtained when fresh peripheral bloodmononuclear cells (PBMC) were treated with IFN and expression of the endogenous IFN-dependent gene,MxA, was examined. Cell surface levels of IFNAR-1 decreased and Stat1 levels increased after IFN-b treat-ment, and retreatment with IFN resulted in an attenuated induction of Mx protein expression. In the contextof using IFNs as therapeutic agents in the treatment of human disease, our data suggest that increasing theamount or frequency of IFN administration may not yield desired biologic effects. Thus, issues concerningthe dosage and the frequency of IFN-b administration deserve careful consideration.

491

INTRODUCTION

IINTERFERONS (IFNS) EXERT THEIR BIOLOGIC effects by bind-ing to a multicomponent cell surface receptor.(1,2) As a con-

sequence of this binding, signaling molecules (signal transduc-ers and activators of transcription [Stat]) are activated and newgenes are transcribed.(3,4) The IFNs comprise a group of pro-teins that exhibit antiviral, antiproliferative, and immunomod-ulatory activities.(5) Type I IFNs (IFN-a, IFN-b, and IFN-v)are produced by numerous types of cells, including variousleukocytes and fibroblasts.(6) In contrast, type II IFN, com-monly known as IFN-g, which is induced by immune and in-flammatory stimuli, is primarily synthesized by T lymphocytesand natural killer (NK) cells. IFNs elicit their effects by bind-ing to specific receptors on the cell surface membrane. Thebinding of IFN-g to its cognate receptor, IFNGR, and the sub-sequent signaling events have been the subject of intense

study.(7) Indeed, the combined discoveries from several labo-ratories working on different aspects of IFN-g signaling haveculminated in the formulation of a receptor signaling model thatis currently one of the most complete.(7) The fine mechanisticdetails of the type I IFN signaling pathway, however, are as yetpoorly understood.

The human type I IFNs are a family of related molecules, allof which bind to a common receptor complex.(8–11) It has beenshown that the functional type I IFN receptor, IFNAR, has amultichain structure composed of at least two distinct subunits,IFNAR-1 and IFNAR-2.(12) The IFNAR-2 chain is the bindingsubunit. The IFNAR-1 chain, which on its own binds most typeI IFNs weakly at best, increases the binding affinity of IFN tothe receptor complex by about 10-fold.(13) IFNAR-1 also con-tributes to the differential recognition of type I IFNs by the IFNAR-1/IFNAR-2 complex(13,14) and is critical for signaltransduction. All type I IFNs activate receptor-associated tyro-

Biogen, Inc., Cambridge, MA 02142.

sine kinases, Jak1 and Tyk2, which in turn activate a series oflatent, cytoplasmic transcriptional activator proteins known asStat.(15) IFNAR-1 associates with Tyk2, and IFNAR-2 associ-ates with Jak1, Stat1, and Stat2.(16) The IFN-mediated associ-ation of the two receptor subunits facilitates the cross-phos-phorylation and activation of Tyk2 and Jak1, which in turnphosphorylate Y466 of IFNAR-1. The phosphorylated receptorsubunit provides a docking site for the SH2 domain of Stat1,positioning the protein for phosphorylation on Y701. Therefore,the IFNAR-1 subunit of the receptor complex performs a piv-otal role in the activation of Stat1 and the overall process ofsignal transduction.

One important area of research in the IFN field concerns themodulation of the IFN response. Under normal circumstances,when the cell is exposed to very low levels of IFN for a shorttime, modulation occurs in several ways. For instance, the IFN-g signal transduction pathway, which is normally activated byphosphorylation, can be shut off by the induction and activa-tion of specific phosphatases.(17) Also, after IFN-g binds to itsreceptor, the receptor/ligand complex gets internalized and de-graded.(17) As it is critical with such potent, pleiotropic mole-cules that the body can maintain homeostasis in the event ofexcess or prolonged exposure to IFN (e.g., with chronic treat-ment of multiple sclerosis (MS) with IFN-b), it was of interestto understand the various regulatory mechanisms for type I IFNactivity.

Cellular responses to other cytokines also provide interest-ing models of functional downregulation using various strate-gies. Monocytes exhibit a reduction in interleukin-6 (IL-6) receptor mRNA levels on exposure to the ligand.(18)

Myelomonocytic cells tightly control the action of IL-1 by ex-pressing membrane-bound or soluble, nonfunctional receptorsthat act as molecular traps for IL-1.(19) Other cytokines, suchas IFN-g, use a multitude of such negative regulatory strate-gies. IFN-g can induce desensitization by downregulating theexpression of IFNGR-2 mRNA and protein.(20,21) In addition,IFN-g induces the expression of a family of proteins termedSOCS/JAB/SSI (suppressor of cytokine signaling/Jak-bindingprotein/Stat-induced Stat inhibitor-1), which bind to and inhibitactivated Jak and desensitize cells in either a homologous or aheterologous manner.(22–24) The object of this study was to de-termine if similar or novel regulatory mechanisms apply to thetype I IFN signaling pathway.

In this study, we examined the extent of IFN-induced geneexpression on extended exposure to recombinant IFN-b1a. Itwas shown that cells can become less responsive after repeatedexposure to IFN-b1a and that this decrease in responsivenesscorrelates both quantitatively and temporarily with the in-creased expression of Stat1.

MATERIALS AND METHODS

Cell culture and antibodies

Jurkat clone E6.1 cells (American Type Culture Collection,TIB-152) were grown in RPMI 1640 supplemented with 10%fetal bovine serum (FBS) (JRH Biotechnologies, Lenexa, KS),10 mM HEPES, 2 mM L-glutamine, and 1 mM sodium pyru-vate. Anti-IFNAR1 and anti-IFNAR-2 antibodies were raised

in mice as previously described.(25) All other antibodies werepurchased from Cell Signaling Technology (Beverly, MA). Therecombinant IFN-b1a is produced at Biogen Inc. (Cambridge,MA) and marketed under the commercial name Avonex™. Cy-cloheximide (Calbiochem, La Jolla, CA) was dissolved in ethylalcohol (Aldrich, Milwaukee, WI) and used at a final concen-tration of 5 mM.

Human peripheral blood mononuclear cells (PBMC) wereisolated from whole blood of healthy donors. Separation wasachieved by density gradient centrifugation using Ficoll-Paque(Amersham Pharmacia Biotech, Paris, France). Whole blood(25 ml) was loaded onto 15 ml Ficoll-Paque in Uni-Sep Maxi1separation columns (Accurate Chemical & Scientific Corp.,Westbury, NY). Tubes were spun at 2000 rpm for 30 min. Themononuclear layer was aspirated and washed with 50 ml com-plete RPMI medium containing 10% FBS. Cells were platedand kept at 37°C, 5%CO2.

FACS analysis

To evaluate cell surface expression levels of IFNAR-1 andIFNAR-2, Jurkat E6.1 cells were washed with phosphate-buffered saline (PBS) and resuspended in 100 ml cold FACSbuffer (RPMI 1640 with 1% FBS and 0.1% sodium azide). Foreach sample, 5 3 105 cells were incubated for 1 h on ice inFACS medium containing 10 mg/ml anti-IFNAR-1 (or anti-IFNAR-2) antibody. Cells were washed with FACS mediumand incubated on ice for 45 min in 100 ml R-PE-conjugatedF(ab9)2 goat antimouse IgG (Biosource International, Camar-illo, CA) (1:100 in FACS medium). Cells were washed, sus-pended in 350 ml FACS medium, and analyzed using a BectonDickinson FACSCalibur (Mountain View, CA).

RNA isolation and RT-PCR analysis

Total RNA was isolated from 5 3 106 Jurkat E6.1 cells us-ing Qiagen minicolumns (Chatsworth, CA) and adjusted to aconcentration of 0.5 mg/ml in water. Reverse transcription wasperformed using Promega’s reverse transcription system kit(Madison, WI). Total RNA (1.5 mg) was transcribed to cDNAusing oligo (dT) primers and AMV Reverse Transcriptase en-zyme (Promega). Stat1-specific primers were used for the PCRreaction. The oligonucleotide sequences of the Stat1 primer set(forward and reverse) were 59-AGG ATA ATT TTC AGGAAG AC-39 and 59-CCT GAT TAA ATC TCT GGG CG-39.PCR products were run on a 2% Metaphor agarose gel (Biowhit-taker, Inc., Rockland, ME).

Transient transfections and reporter gene assays

For reporter gene assays, Jurkat E6.1 cells were transfectedat 80% confluence in 6-well plates in triplicate with expres-sion plasmids encoding an IFN-dependent luciferase gene.pISRE-Luc (Path Detect ISRE cis-Reporting System) (Strat-agene, La Jolla, CA) or a control vector, pCIS-CK, in addi-tion to a plasmid encoding a constitutively active secreted al-kaline phosphatase (SEAP) gene pCMV SEAP (Tropix,Bedford, MA) were transfected in Jurkat E6.1 cells using li-posome-based transfection reagent Fugene-6 (Roche, Nutley,NJ). Total DNA transfected was kept constant at 10 mg perwell. Six hours after transfection, cells were supplemented

DUPONT ET AL.492

with fresh medium containing 500 IU/ml IFN-b1a and cul-tured for another 18 h. Cells were rinsed with PBS and re-suspended in lysis buffer (Promega), and luciferase activitywas measured using a commercially available Bright-Glo as-say kit (Promega). SEAP was measured using a chemilumi-nescent assay (Tropix Phospha-Light kit) and was used as ameasure of transfection efficiency.

Cell extracts and SDS-PAGE

Whole cell extracts were prepared by lysing cells in a 23

SDS sample buffer (Novex™) supplemented with an equal vol-ume of RIPA buffer (1% NP-40, 1% deoxycholate, 0.1% SDS,150 mM NaCl, and 50 mM Tris-HCl, pH 7.0). Samples were

boiled briefly, and supernatants were collected after centrifu-gation (14,000 rpm) for 10 min. Supernatants were then heatedat 90°C for 5 min and subjected to SDS-PAGE on a 10% poly-acrylamide gel.

Western blotting

Proteins were transferred following electrophoresis to aPVDF membrane (Novex) by electroblotting for 2 h at 12 V intransfer buffer from Novex. The blots were blocked in 5% drymilk in Tris-buffered saline (TBST) (10 mM Tris, 150 mMNaCl, pH 7.5, with 0.1% Tween-20) for 2 h at ambient tem-perature, with gentle agitation. Blots were incubated with a1:2000 dilution of the primary antibody (Cell Signaling Tech-

REGULATION OF CELL RESPONSE TO HuIFN-b1a 493

FIG. 1. Effect of IFN-b1a on cell surface IFNAR-1 and IFNAR-2 expression levels and IFN-dependent gene expression. (A)Effect of IFN-b1a on cell surface IFNAR-1 and IFNAR-2 expression levels. Jurkat E6.1 cells were treated with 0–1000 U/mlIFN-b1a for a period of 24 h prior to analysis. Cell surface levels of the IFN receptor subunits were measured using a flow cy-tometry-based assay. Either a mouse antihuman IFNAR-1 or IFNAR-2 antibody and then a goat antimouse F(ab)2 carrying a flu-orescent tag were used to label the receptor molecules exposed on the surface membrane. In each case, the receptor density fromuntreated cells was assigned a value of 100%. (B) Effect of IFN-b1a on cell surface expression of various T cell receptors. Jur-kat E6.1 cells were left untreated or treated with 500 U/ml IFN-b1a for a period of 24 h. Indicated receptors were stained withappropriate phycoerythrin-conjugated antibodies prior to the flow cytometric analysis. (C) Effect of IFN-b1a on expression of atransfected ISRE-luciferase reporter gene. Jurkat E6.1 cells (5 3 106) were transfected with an ISRE-luciferase reporter plasmid.Following a 6-h incubation, transfected cells were treated with the indicated amounts IFN-b1a. Eighteen hours later, cells werelysed, and enzymatic activity was measured using a Bright-Glo luciferase assay system.

nology) for 2–3 h, washed with TBST, and incubated for 1 hwith horseradish peroxidase (HRP)-conjugated secondary anti-body (antirabbit HRP-conjugated antibody) (Cell SignalingTechnology). The blots were washed again with TBST and de-veloped using the ECL Western blot detection system (Cell Sig-naling Technology).

RESULTS

The immortalized T lymphocyte cell line, Jurkat E6.1, wasused to study the effects of dose and frequency of IFN treat-ment on cellular responsiveness to further IFN exposure. Ini-tially, cells were treated with modest concentrations (60–1000IU/ml) of Avonex, a recombinant human IFN-b (IFN-b1a), fora period of 24 h, and cell surface receptor levels were analyzedusing flow cytometry. These doses include physiologically rel-evant concentrations, because when MS patients are treatedwith IFN-b1a, levels in peripheral blood range from 5 to 100IU/ml.(26–28) A reduction in the surface density of IFNAR-1 onexposure to increasing concentrations of IFN-b1a was observed(Fig. 1A). The decrease in cell surface IFNAR-1 was inhibitedby low temperatures (4°C) and the presence of azide, suggest-ing energy-dependent internalization as a major mechanism forthe loss of IFNAR-1 (data not shown). This effect appeared tobe specific to IFNAR-1, as only a minimal reduction in the sur-face density of IFNAR-2, the second subunit of the receptorheterodimer complex, was seen (Fig. 1A). However, our flowcytometry-based assay could not distinguish between the twomembrane-bound forms of IFNAR-2 (IFNAR-2b [short, un-known function], IFNAR-2c [long, signaling-competent]).

Therefore, it remains a possibility that the remaining IFNAR-2 was mainly the inactive short form. The levels of other sur-face markers, such CD3 and IFNGR were unaffected (Fig. 1B)by treatment with even relatively high doses of IFN-b1a. Asexpected, MHC I expression was upregulated by IFN-b1a treat-ment.(29)

The effect of IFN-b dose on transcription of IFN-induciblegenes was examined by monitoring the expression of a lucif-erase reporter gene driven by a promoter (ISRE) that is specificto type I IFNs. Cells were transfected with the ISRE-luciferase(pISRE-Luc)-containing plasmid, allowed to recover for 6–8 h,and then treated with different doses of IFN-b. Figure 1C showsthat the ISRE-luciferase plasmid was not transcribed in the ab-sence of IFN-b and was transcribed in a dose-dependent fash-ion in the presence of IFN-b.

To determine how the IFN-induced reduction in cell surfacelevels of IFNAR-1 affected the IFN signal transduction path-way, expression of the ISRE-luciferase reporter gene was as-sessed in cells that had been pretreated with the same concen-trations of IFN-b as in the experiment shown in Figure 1. Jurkatcells were incubated with 0, 62.5, 125, 250, 500, or 1000 IU/mlIFN-b for 24 h, washed, and transfected with the ISRE-lucif-erase plasmid. The cells were stimulated with a second dose ofIFN-b1a (500 IU/ml) 6–8 h after transfection. Eighteen hourslater, the cells were analyzed for luciferase activity. The IFN-dependent expression of luciferase was dramatically reducedwhen cells had been pretreated with IFN-b, even at the lowestdose of 62.5 IU/ml (Fig. 2A). This result was specific to theIFN-b pathway, as these IFN-b-pretreated cells were still com-petent for expression of a non-IFN-dependent reporter gene,CMV SEAP (secreted alkaline phosphatase) plasmid (Fig. 2A).

DUPONT ET AL.494

FIG. 2. IFN-dependent expression of ISRE-luciferase reporter gene in cells that have had previous exposure to IFN-b. (A) IFN-dependent gene expression in response to a second dose of IFN. Jurkat E6.1 cells were treated with 0–1000 IU/ml IFN-b1a for24 h. After this initial treatment period, 5 3 106 cells were transfected with either the ISRE-luciferase reporter plasmid or thetransfection control plasmid containing a secreted alkaline phosphatase gene (SEAP) downstream of the cytomegalovirus (CMV)promoter sequence. Transfected cells were then challenged with 500 IU/ml IFN-b1a. Eighteen hours later, cells were lysed, andluciferase enzymatic activity was measured using the Bright-Glo luciferase assay system. SEAP levels were measured from thesupernatants as per the manufacturer’s protocol. Naive cells remained untreated during the initial 24-h period. Untransfected cellsdid not contain the reporter plasmid and provided a value for the background level of fluorescence inherent to the assays. (B)Kinetics of IFN resistance. Jurkat cells were treated with 50 IU/ml IFN-b1a for 1, 3, 8, or 24 h. Cells from each treatment groupwere transfected with the ISRE-luciferase reporter plasmid for a period of 6 h and then challenged with 500 IU/ml fresh IFN-b1a. Luciferase activity was assessed 18 h later and is expressed as the percent of untreated cells.

Pretreatment of at least 8 h was required to effectively renderthe cells refractory to a secondary challenge of IFN (Fig. 2B).

The extent of inhibition of IFN-dependent transcription atthe lowest pretreatment dose of IFN-b (i.e., 62.5 IU/ml) wasparticularly interesting. As illustrated in Figures 1A and 2A,cells pretreated with 62.5 IU/ml IFN-b showed only a 50% re-duction in cell surface IFNAR-1, whereas transcriptional acti-vation of the ISRE-luciferase plasmid was reduced by 80%. Tomore carefully determine the relationship between IFNAR-1cell surface expression and downstream activation of the IFNsignal transduction pathway, a lower range of pretreatmentdoses was used. Figure 3A and B show that inhibition of IFN-dependent transcription does not correlate with cell surface ex-pression levels of IFNAR-1. In fact, at the lowest concentra-tion of IFN-b tested (7.8 IU/ml), IFNAR-1 levels were virtuallyunchanged (Fig. 3A), whereas IFN-dependent gene expression

was dramatically reduced (Fig. 3B). This uncoupling of IFNAR-1 levels and gene expression was further explored by examin-ing the phosphorylation state of several of the signaling mole-cules known to be involved in the IFN pathway. Figure 3Cshows that when cells are retreated with IFN-b, activation ofthe signaling molecules (i.e., phosphorylation) correlated wellwith cell surface levels of IFNAR-1 but not with transcriptionalactivation. When IFNAR-1 levels were .80%, full phosphor-ylation could occur, but gene expression was almost completelyinhibited. When IFNAR-1 levels drop below 50% (e.g., at 62.5IU/ml), phosphorylation of signaling molecules is reduced. Asexpected, cells that had not been treated with IFN showed nobasal phosphorylation of Stat1, Stat2, Stat3, or Tyk2, and whenIFN-treated cells were examined .24 h postexposure, no re-sidual phosphorylation was seen (data not shown). Thus, thesedata suggest that a mechanism(s) other than cell surface levels

REGULATION OF CELL RESPONSE TO HuIFN-b1a 495

FIG. 3. Effect of IFN-b1a treatment on cellular responsiveness. Jurkat E6.1 cells were treated with the indicated concentra-tions of IFN-b1a for a period of 24 h. (A) Effect of IFN-b1a on expression levels of cell surface IFNAR-1. Cell surface levelsof the receptor subunit IFNAR-1 were measured after 24 h of treatment with the indicated concentrations of IFN-b using theflow cytometry-based assay. (B) IFN-inducible luciferase reporter assay. After the 24-h initial treatment period with 0–62.5 IU/mlIFN-b, the Jurkat E6.1 cells were washed and transfected with an ISRE-luciferase reporter plasmid for a period of 6 h and thenrechallenged with 500 IU/ml fresh IFN-b1a. Luciferase activity was assessed 18 h later. (C) IFN-induced phosphorylation of re-ceptor-associated molecules. After the 24-h initial treatment period with 0–62.5 IU/ml IFN-b, the Jurkat E6.1 cells were washedand transfected with an ISRE-luciferase reporter plasmid for a period of 6 h. The cells were then rechallenged with 500 IU/mlfresh IFN-b1a for an additional 15 min and lysed in sample buffer. The total levels and phosphorylation status of Stat1, Stat2,Stat3, and Tyk2 were assessed by Western blot analysis.

of the IFN receptor and tyrosine phosphorylation of signalingmolecules can regulate a cell’s ability to respond to IFN treat-ment.

Interestingly, of the parameters that were investigated, it wasa striking increase in total Stat1 and, to a lesser extent, Stat2levels that correlated best with the inhibition of IFN-dependenttranscription (Fig. 3C). At pretreatment concentrations of 7.8IU/ml, the level of total Stat1 protein was significantly in-creased. Stat2 levels also increased, although not as dramati-cally as Stat1 levels. The kinetics of this increase in total Stat1levels is shown in Figure 4A. By 8 h post-IFN-b treatment (50IU/ml), the levels were significantly elevated. Figure 1C showsthat the kinetics of the inhibition of IFN-induced gene expres-sion is very similar to that seen with Stat1 levels. When cellshad been pretreated with 50 IU/ml IFN-b for at least 8 h, theIFN-dependent induction of luciferase was reduced by .50%.In order to compare mRNA levels with the protein levels, thekinetics of Stat1 mRNA expression is shown in Figure 4B. Thepeak levels of Stat1 mRNA do not seem to correlate with Stat1protein levels or with Mx mRNA levels (a known, tightly reg-ulated IFN-stimulated gene [ISG]). As a control, the kinetics ofStat1 phosphorylation is shown in Figure 4C. After IFN treat-ment with high doses of IFN-b (500 IU/ml), PhosphoStat1 lev-els increase rapidly (15 min) and are virtually back to baselineafter 4 h.

To determine how long IFN-treated cells remain refractoryto further IFN treatment, cells that had been exposed to vari-ous relatively low (7.8–125 U/ml) doses of IFN were retreatedat either 48, 96, or 168 h after the initial treatment, and subse-quent expression of the ISRE-luciferase plasmid was assessed.The cells were retreated with the higher dose of IFN-b (500IU/ml) in order to observe the maximal effect of the retreat-ment. Figure 5 shows that 48 h after pretreatment (even withlow doses of IFN), the cells were only partially responsive toretreatment with IFN-b. For example, after a physiologic doseof ,30 IU/ml, the cell’s response to a second dose of IFN wasreduced by 60% compared with cells that had not received apretreatment. Ninety-six hours after the initial IFN treatment,cells were still only partially responsive (50% responsive to asecond dose after an initial treatment of ,30 IU/ml). By 168 h(1 week) after the initial treatment, the cells had regained fullresponsiveness to treatment with IFN-b. Stat1 levels correlatedwell with cellular responsiveness; while Stat1 levels remainedelevated, IFN-dependent transcription remained low, and whenStat1 levels returned to pretreatment amounts, the cells werefully responsive to further IFN treatment.

An experiment to determine the effect of repeated IFN treat-ment is shown in Figure 6. Figure 6A is a schematic of the ex-perimental protocol. Figure 6B shows that when cells are treatedthree times per week with 500 IU/ml IFN-b, Stat1 levels re-main high. Figure 6C demonstrates that IFNAR-1 levels remainrepressed as a result of frequent treatment with IFN-b. Thesefindings suggest that frequent dosing of cells with IFN-b mightinhibit, rather than increase, the cellular response.

The experiments described above utilized an immortalizedT cell line and an exogenously added IFN-inducible reportergene and were performed in vitro. As an initial step to deter-mine if these findings can be extrapolated to in vivo situations,the expression of an endogenous IFN-inducible protein (MxA)from fresh human peripheral blood mononuclear cells (PBMC)

was examined. Within this blood fraction are numerous celltypes that respond to IFN-b and that are thought to mediatemany of the therapeutic effects of IFN-b in MS. Thus, PBMCwere isolated and pretreated with IFN-b (100 IU/ml) or vehi-cle control. After 48 h, half of each sample was challenged withanother dose of IFN, and the expression of MxA was assessedby Western blot (Fig. 7A). The results obtained with these pri-mary human cells were consistent with the Jurkat T cell linefindings in that PBMC appear to regulate the type I IFN sig-naling pathway in a manner similar to that of the Jurkat cells.In the absence of pretreatment, MxA is highly induced after ex-

DUPONT ET AL.496

FIG. 4. Effect of IFN-b1a treatment on cellular Stat1 con-tent. (A) Time course of Stat1 upregulation. Jurkat cells weretreated with 50 IU/ml IFN-b1a for either 1, 3, 8, or 24 h andthen analyzed for total Stat1 content by lysing the cells and per-forming Western analysis. (B) Stat1 and Mx mRNA were re-verse transcribed and amplified using specific oligonucleotideprimers to determine Stat1 and Mx mRNA levels after 0, 1, 2,4, 6, 8, and 24 h of IFN-b treatment. Actin was used as a con-trol to confirm equal loading. (C) Kinetics of Stat1 phosphor-ylation. Jurkat cells were treated with 500 IU IFN-b for 15, 60,or 240 min. The cells were then lysed in sample buffer, sub-jected to SDS-PAGE, and analyzed by Western blot with aStat1-specific or Phospho-Stat1-specific antibody.

posure to 100 IU/ml IFN (Fig. 7A, lane 2). However, eventhough levels of MxA were still elevated 48 h after the initialtreatment (Fig. 7A, lane 3), retreatment did not induce a re-sponse even equivalent to that seen in naive cells (Fig. 7A, com-pare lane 4 with lane 2). Further indications that the Jurkat sys-tem is predictive for in vivo models come from the observationthat with treatment of PBMC with high doses of IFN (500IU/ml), IFNAR-1 is downregulated and total Stat1 levels areupregulated to an extent similar to that observed with the Ju-rkat cells pretreated with 500 IU/ml IFN-b (Fig. 7B, C).

DISCUSSION

Cells frequently undergo a process of desensitization in re-sponse to almost any type of stimulus. In this study, we dem-onstrate that T cells become desensitized as a result of persis-tent IFN-b1a stimulation.

Adaption to extracellular signals can occur in various ways.In some cases, it results from a decrease in the number of spe-cific cell surface receptors, which can take hours to complete.A more rapid adaptation response can result from a molecularchange in the receptor or in any downstream component of thesignaling pathway. Although the observation that at the levelof transcription, cells become desensitized to further IFN-a

treatment has been reported,(30) the mechanistic details of howcells adapt to excessive, prolonged, or repeated exposure toIFN-b are not well understood. The experiments reported herebegin to address this question.

Jurkat cells and a reporter gene were used as a model sys-tem. Jurkat cells were selected because they are derived fromT cells, which are among the targets of IFN immunodulatoryaction in vivo. When examining the cell’s response to repeatedIFN treatment, a reporter gene was used in order to obtain asnapshot of the cell’s responsiveness without the confoundingbackground levels of previously induced endogenous geneproducts. By tracing the pathway of IFN-b-mediated signalingin mammalian cells, several points where negative regulationcould occur (e.g., cell surface receptor levels, activation of sig-nal transduction molecules, and IFN-dependent transcription)were examined. Treating Jurkat cells with increasing concen-trations of IFN-b resulted in a dose-dependent loss of the IFNreceptor subunit, IFNAR-1, from the cell surface membraneeven at concentrations that are routinely achieved in therapeu-tic settings (Fig. 1). Initially, it was assumed that a decrease inboth the activation (i.e., phosphorylation) of signaling mole-cules and the transcription of an IFN-inducible reporter genewould correlate with this receptor downregulation. However,when lower concentrations of IFN-b were tested, a discrepancybetween receptor levels and transcription was observed. When

REGULATION OF CELL RESPONSE TO HuIFN-b1a 497

FIG. 5. Long-term effects of IFN-b1a treatment on cellular responsiveness and correlation between cellular Stat1 content andresponsiveness to IFN-b1a treatment. Jurkat E6.1 cells were treated with various concentrations of IFN-b1a (0–125 IU/ml). TheISRE-luciferase gene was introduced into cells at (A) 48 h, (B) 96 h, and (C) 168 h after treatment. The cells were retreated 6h later with 500 IU/ml fresh IFN-b1a and, after 18 h, assayed for luciferase activity. The rest of the cells were analyzed for to-tal Stat1 content by lysing the cells at 48, 96, or 168 h and performing Western blot analysis.

cells were pretreated with extremely low amounts of IFN-b,normal cell surface receptor levels were maintained. Whenthese cells were retreated with IFN-b, despite the activation ofsignal transduction molecules (as determined by tyrosine phos-phorylation), IFN-dependent transcription of the IFN-depen-dent reporter gene was strikingly reduced (Fig. 3).

Although there are several possible mechanisms that couldmediate this desensitization, some of the best-studied regula-tory mediators (such as suppressor of cytokine signaling [SOC]or phosphatases) are unlikely candidates, as tyrosine phospho-rylation of the Stat still occurred in the unresponsive cells. Al-though the exact mechanism of the desensitization has yet tobe determined, an intriguing correlation between levels of to-tal Stat1 and Stat2 and cellular responsiveness was observed.Cells treated with low doses of IFN-b have a marked and sus-tained increase in the cellular content of Stat1, a key compo-nent of the IFN signaling pathway. This increase in Stat1 lev-els correlated with a decrease in expression of the IFN-inducible

reporter gene both temporally and quantitatively (compare Figs.1C and 4A). This correlation did not hold true with respect toStat1 mRNA, which remains at a relatively steady-state afterIFN treatment. This was not the case with Mx, a classic type IIFN ISG. Stat2 levels were also elevated in the desensitizedcells but did not correlate with desensitization as well as Stat1did. The role of Stat1 in IFN-a/b or IFN-g signaling has beenestablished unequivocally through the generation and charac-terization of mice with targeted disruptions of the Stat1 gene.(31)

Cells from Stat1-null mice are incapable of manifesting bio-logic responses to IFN-a/b or IFN-g. The involvement of Stat1in positive, negative, and constitutive regulation of gene ex-pression has been well studied. For example, in a study by Chenet al., there is a tight correlation between high levels of Stat1and insensitivity to IFN-g.(32) These authors suggest that in-creases in Stat1 levels serve as a mechanism for attenuating theeffects of persistent exposure to IFN-g on cellular processes,perhaps by direct impairment of IFN-g signal transduction or

DUPONT ET AL.498

FIG. 6. Effect of treatment frequency on Stat1 recovery. Naive Jurkat E6.1 cells were adjusted to a density of 5 3 105 cells/mland incubated for 24 h with 500 IU/ml IFN-b1a (37°C, 5% CO2). Cells were then washed and resuspended in fresh medium.Cells in the 13 treatment group received only the initial IFN-b treatment. Cells in the 33 treatment group received two addi-tional doses of 500 IU/ml at 48 h and 96 h. (A) A schematic of the experiment. (B) Western analysis of Stat1 levels. Cells werelysed, subjected to SDS-PAGE, and blotted with a specific Stat1 antibody at 48, 96, and 168 h posttreatment. (C) Cell surfaceexpression levels of IFNAR-1 following once a week or three times a week treatment. Cell surface levels of the IFNAR-1 re-ceptor subunit were measured at each time point by flow cytometric analysis.

by cross-talk with other cytokine signaling pathways.(32) In theChen et al. study (as in our own) the cells did not regain theirresponsiveness to IFN-g until the cytokine had been absent forapproximately 6 days.(32)

Our investigation of Stat1 levels in cells exposed to IFN-b re-vealed a similar specific, drug-induced increase in protein levelsas early as 4 h after treatment. It is possible that the modulationof Stat1 levels may serve as a general mechanism for diminish-ing the effects of IFNs on gene expression. The observed increasesin Stat2 levels may also have an effect on alterations of gene ex-pression. The mechanism by which the excess of Stat proteinscould exert negative effects on IFN responsiveness is unclear.However, two reasonable possibilities include the excess Stat re-maining in the cytosol and competing with phosphorylated Statin their binding to other proteins or by translocating to the nu-cleus and binding DNA as a transcriptional repressor.(33,34)

The phenomenon of desensitization in the context of re-peated, frequent dosing was also addressed. Figure 5 shows that

Jurkat cells are not fully responsive to a second dose of IFNeven 4 days after the initial treatment. After a full week, thecells are fully responsive once again. Even more relevant forsome patients is the experiment shown in Figure 6, where cellswere treated to mimic a three times a week dosing schedule. Inthis case, Stat1 levels remained high, and IFNAR-1 levels re-mained relatively suppressed.

In order to begin to examine if the results obtained in the Jurkat cell-reporter gene system will extend to more physio-logic situations, primary PBMC were used. In these cells (un-like the Jurkat cells), the turnover of MxA protein was rela-tively rapid, thereby allowing an assessment of the effects ofrepeated dosing on the transcriptional activation of an IFN-inducible gene. Although both protein (Fig. 7A) and mRNAlevels (data not shown) were examined, from a therapeutic per-spective, it is the actual gene products that are most relevant.The results of the preliminary experiments with the PBMC (Fig.7) are consistent with the results obtained in the Jurkat cells.

REGULATION OF CELL RESPONSE TO HuIFN-b1a 499

FIG. 7. Effects of IFN-b1a on normal human PBMC. (A) Effect of IFN-b1a on MxA induction in normal human PBMC.PBMC were isolated by Ficoll-Paque density gradient centrifugation from whole blood. Cells were adjusted to a density of 5 3105/ml and treated with 0 or 100 U/ml IFN-b1a for 24 h at 37°C and 5% CO2. After 24 h of treatment, cells were washed andleft in fresh medium for 24 h to recover. Pretreated or untreated cells were challenged with 100 U/ml IFN-b1a for an additional24 h prior to lysis. The level of MxA was measured by Western blot analysis. (B) PBMC were treated with 500 IU/ml IFN-b.Cell surface levels of the receptor subunit IFNAR-1 were measured using a flow cytometric assay. (C) PBMC were pretreatedwith either 0 or 500 IU/ml IFN-b. After 24 h, the cells were washed and then challenged with either 0 or 500 IU/ml IFN-b. Thecells were lysed with sample buffer 15 min after challenge, and the levels and phosphorylation status of Stat1 were measured byWestern blot analysis.

Cell surface IFNAR-1 levels can decrease after IFN treatment,Stat1 levels increase, and pretreatment of the cells with IFNcauses a desensitization of IFN-dependent gene expression.

The phenomenon of cellular desensitization as a consequenceof high or repeated exposure to cytokines is well documented.The observations reported here with IFN-b raise issues con-cerning the impact of aggressive IFN-b treatment regimens inclinical settings. IFNs were discovered on the basis of their anti-viral activities. A variety of cells produce IFNs in response toinfections by various viruses, bacteria, and mycoplasma.(35,36)

In addition to these effects, however, IFNs exhibit pleiotropicbiologic activities, such as antitumor and immunomodulatoryeffects. Consequently, the use of IFNs in treatment of humandisease has become prevalent. For instance, IFN-b is currentlyused for the treatment of MS.(37) MS is a T cell-mediated dis-ease of the central nervous system in which activated macro-phages effectively demyelinate the neuronal tissue. The mostcompelling hypothesis concerning the immunologic basis of thetherapeutic effect of IFN-b involves stimulating a Th2 shiftwhereby secretions of protective and anti-inflammatory cyto-kines (e.g., IL-10) help suppress the effector cells.(38,39) There-fore, the ability of cytokine-secreting cells to respond to IFN-b is of paramount importance in mediating its therapeuticeffects. In this context, the data in this report suggest that in-creasing the amount or frequency of IFN administration maynot yield the desired biologic effects.(40) Further experimentsto explore the consequences of frequent IFN administration inpatients are underway.

ACKNOWLEDGMENTS

We thank M. Zafari and A. Larner for the IFNAR-1 and Stat2antibodies. We also thank Joe Rosa, Wendy Jones, and WendyGabel for critical reading of the manuscript.

REFERENCES

1. COLAMONICI, O.R., PFEFFER, L.M., D’ALESSANDRO, F.D.,PLATANIAS, L.C., GREGORY, S.A., ROSOLEN, A., NORDAN,R., CRUCIANI, R.A., and DIAZ, M.O. (1992). Multichain struc-ture of the IFN-alpha receptor on hematopoietic cells. J. Immunol.148, 2126–2132.

2. HU, R., GAN, Y., LIU, J., MILLER, D., and ZOON, K.C. (1993).Evidence for multiple binding sites for several components of hu-man lymphoblastoid interferon-alpha. J. Biol. Chem. 268, 12591–12595.

3. DARNELL, J.E., Jr., KERR, I.M., and STARK, G.R. (1994). Jak-Stat pathways and transcriptional activation in response to IFNsand other extracellular signaling proteins. Science 264, 1415–1421.

4. DARNELL, J.E., Jr. (1997). Stats and gene regulation. Science 277,1630–1635.

5. TYRING, S.K. (1995). Interferons: biochemistry and mechanismsof action. Am. J. Obstet. Gynecol. 172, 1350–1353.

6. DIAZ, M.O., POMYKALA, H.M., BOHLANDER, S.K., MAL-TEPE, E., MALIK, K., BROWNSTEIN, B., and OLOPADE, O.I.(1994). Structure of the human type-I interferon gene cluster de-termined from a YAC clone contig. Genomics 22, 540–552.

7. STARK, G.R., KERR, I.M., WILLIAMS, B.R.G., SILVERMAN,R.H., and SCHREIBER, R.D. (1998). How cells respond to inter-ferons. Annu. Rev. Biochem. 67, 227–264.

8. MERLIN, G., FALCOFF, E., and AGUET, M. (1985). 125I-labeledhuman interferons alpha, beta and gamma: comparative receptor-binding data. J. Gen. Virol. 66, 1149–1152.

9. BRANCA, A.A., and BAGLIONI, C. (1981). Evidence that typesI and II interferons have different receptors. Nature 294, 768–770.

10. FLORES, I., MARIANO, T.M., and PESTKA, S. (1991). Humaninterferon omega (omega) binds to the alpha/beta receptor. J. Biol.Chem. 266, 19875–19877.

11. PESTKA, S., LANGER, J.A., ZOON, K.C., and SAMUEL, C.E.(1987). Interferons and their actions. Annu. Rev. Biochem. 56,727–777.

12. NOVICK, D., COHEN, B., and RUBINSTEIN, M. (1994). The hu-man interferon alpha/beta receptor: characterization and molecularcloning. Cell 77, 391–400.

13. COHEN, B., NOVICK, D., BARAK, S., and RUBINSTEIN, M.(1995). Ligand-induced association of the type I interferon recep-tor components. Mol. Cell. Biol. 15, 4208–4214.

14. CUTRONE, E.C., and LANGER, J.A. (1997). Contributions ofcloned type I interferon receptor subunits to differential ligandbinding. FEBS Lett. 404, 197–202.

15. KOTENKO, S.V., IZOTOVA, L.S., MIROCHNITCHENKO,O.V., LEE, C., PESTKA, S. (1999). The intracellular domain ofinterferon-alpha receptor 2c (IFN-alphaR2c) chain is responsiblefor Stat activation. Proc. Natl. Acad. Sci. USA 96, 5007–5012.

16. KOTENKO, S.V., and PESTKA, S. (2000). Jak-Stat signal trans-duction pathway through the eyes of cytokine class II receptor com-plexes. Oncogene 19, 2557–2565.

17. CELADA, A., and SCHREIBER, R.D. (1987). Internalization anddegradation of receptor-bound interferon-gamma by murine mac-rophages. Demonstration of receptor recycling. J. Immunol. 139,147–153.

18. BAUER, J., LENGYEL, G., BAUER, T.M., ACS, G., and GEROK,W. (1989). Regulation of interleukin-6 receptor expression in hu-man monocytes and hepatocytes. FEBS Lett. 249, 27–30.

19. COLOTTA, F., RE, F., MUZIO, M., BERTINI, R., POLEN-TARUTTI, N., SIRONI, M., FIRI, J.G., DOWER, S.K., SIMS, J.E.,and MANTOVANI, A. (1993). Interleukin-1 type II receptor: a de-coy target for IL-1 that is regulated by IL-4. Science 261, 472–475.

20. BACH, E.A., SZABO, S.J., DIGHE, A.S., ASHKENAZI, A.,AGUET, M., MURPHY, K.M., and SCHREIBER, R.D. (1995).Ligand-induced autoregulation of IFN-gamma receptor beta chainexpression in T helper cell subsets. Science 270, 1215–1218.

21. GREENLUND, A.C., FARRAR, M.A., VIVIANO, B.L., andSCHREIBER, R.D. (1994). Ligand-induced IFN gamma receptortyrosine phosphorylation couples the receptor to its signal trans-duction system (p91). EMBO J. 13, 1591–1600.

22. STARR, R., WILLSON, T.A., VINEY, E.M., MURRAY, L.J.L.,TAYNER, J.R., JENKINS, B.J., GONDA, T.J., ALEXANDER,W.S., METCALF, D., NICOLA, N.A., and HILTON, D.J. (1997).A family of cytokine-inducible inhibitors of signaling. Nature 387,917–921.

23. NAKA, T., NARAZAKI, M., HIRATA, M., MATSUMOTO, T.,MINAMOTO, S., AONO, A., NISHIMOTO, N., KAJITA, T.,TAGA, T., YOSHIZAKI, K., AKIRA, S., and KISHIMOTO, T.(1997). Structure and function of a new Stat-induced Stat inhibi-tor. Nature 387, 924–929.

24. ENDO, T.A., MASUHARA, M., YOKOUCHI, M., SUZUKI, T.,SAKAMOTO, H., MITSUJ, K., MATSUMOTO, A., TAN-IMURA, S., OHTSUBO, M., MISAWA, H., MIYAZAKI, T.,LEONOR, N., TANIGUCHI, T., FUJITA, T., KANAKURA, Y.,KOMIYA, S., and YOSHIMURA, A. (1997). A new protein con-taining an SH2 domain that inhibits Jak kinases. Nature 387,921–924.

25. GOLDMAN, L.A., ZAFARI, M., CUTRONE, E.C., DANG, A.,BRICKELMEIER, M., RUNKEL, L., BENJAMIN, C.D., LING,L.E., and LANGER, J.E. (1999). Characterization of anti-human

DUPONT ET AL.500

IFNAR-1 monoclonal antibodies: epitope localization and func-tional analysis. J. Interferon Cytokine Res. 19, 15–26.

26. ALAM, J., McALLISTER, A., SCARAMUCCI, J., JONES, W.,and ROGGE, M. (1997). Pharmacokinetics and pharmacodynam-ics of interferon beta-1a (IFN-b1a) in healthy volunteers after in-travenous, subcutaneous or intramuscular administration. Clin.Drug Invest. 14, 35–43.

27. MUNAFO, S., TRINCHARD-LUGAN, I.I., NGUYEN, T.X., andBURAGLIO, M. (1998). Comparative pharmacokinetics and phar-macodynamics of recombinant human interferon beta-1a after in-tramuscular and subcutaneous administration. Eur. J. Neurol. 5,187–193.

28. KHAN, O.A., and DHIB-JALBUT, S.S. (1998). Serum interferonbeta-1a (Avonex) levels following intramuscular injection in re-lapsing-remitting MS patients. Neurology 51, 738–742.

29. SELIGER, B., DUNN, T., SCHWENZER, A., CASPER, J., HUBER,C., and SCHMOLL, H.J. (1997). Analysis of the MHC class I anti-gen presentation machinery in human embryonal carcinomas: evi-dence for deficiencies in TAP, LMP and MHC class I expression andtheir upregulation by IFN-gamma. Scand. J. Immunol. 46, 625–632.

30. LARNER, A.C., JONAK, G., CHENG, Y.-S.E., KORANT, B.,KNIGHT, E., and DARNELL, J.E., Jr. (1984). Transcriptional in-duction of two genes in human cells by b interferon. Proc. Natl.Acad. Sci. USA 81, 6733–6737.

31. MERAZ, M.A., WHITE, M.J., SHEEHAN, K.C.F., BACH, E.A.,TODIG, S.J., DIGHE, A.S., KAPLAN, D.H., TILEY, J.K.,GREENLUND, A.C., CAMPBELL, D., CARVER-MOORE, K.,DUBOIS, R.N., CLARK, R., AGUET, M., and SCHREIBER, R.D.(1996). Targeted disruption of the Stat1 gene in mice reveals un-expected physiologic specificity in the Jak-Stat signaling pathway.Cell 84, 431–442.

32. CHEN, G., HOHMEIER, H.E., and NEWGARD, C.B. (2001). Ex-pression of the transcription factor Stat-1 in insulinoma cells pro-tects against cytotoxic effects of multiple cytokines. J. Biol. Chem.276, 766–772.

33. WAITE, K.J., FLOYD, Z.E., ARBOUR-REILY, P., and STE-PHENS, J.M. (2001). Interferon-g-induced regulation of peroxi-some proliferator-activated receptor g and Stats in adipocytes. J.Biol. Chem. 276, 7062–7068.

34. RAMANA, C.V., CHATTERJEE-KISHORE, M., NYUYEN, H.,and STARK, G.R. (2000). Complex roles of Stat1 in regulatinggene expression. Oncogene 19, 2619–2627.

35. GRESSER, I. (1997). Wherefore interferon? J. Leukocyte Biol. 61,567–574.

36. MORRIS, A., and ZVETKOVA, I. (1997). Cytokine research: theinterferon paradigm. J. Clin. Pathol. 50, 635–639.

37. ALAM, J.J. (1995). Interferon-beta treatment of human disease.Curr. Opin. Biotechnol. 6, 688–691.

38. PETEREIT, H.F., BAMBORSCHKE, S., ESSE, A.D., and HEISS,W.D. (1997). Interferon gamma producing blood lymphocytes aredecreased by interferon beta therapy in patients with multiple scle-rosis. Mult. Scler. 3, 180–183.

39. RUDICK, R.A., RANSOHOFF, T.M., LEE, J.-C., PEPPLER, R.,YU, M., MATHISEN, P.M., and TUOHY, V.K. (1998). In vivo ef-fects of interferon beta-1a on immunosuppressive cytokines in mul-tiple sclerosis [published erratum appears in Neurology 1998; 51,332]. Neurology 50, 1294–1300.

40. ALAM, J. GOELZ, S., RIOUX, P., SCARAMUCCI, J., JONES,W., McALLISTER, A., CAMPION, M., and ROGGE, M. (1997).Comparative pharmacokinetics and pharmacodynamics of two re-combinant human interferon beta-1a (IFN beta-1 a) products ad-ministered intramuscularly in healthy male and female volunteers.Pharm. Res. 14, 546–549.

Address reprint requests to:Dr. Jaya Goyal

Biogen Inc. 14 Cambridge Center

Cambridge, MA 02142

Tel: (617) 679-2921 Fax: (617) 679-3523

E-mail: susan_goelz@biogen.com

Received 15 October 2001/Accepted 14 January 2002

REGULATION OF CELL RESPONSE TO HuIFN-b1a 501