Constitutive and Interleukin-7- and Interleukin-15-StimulatedDNA Binding of STAT and Novel Factors in Cutaneous TCell Lymphoma Cells

Jian-Zhong Qin,1 Jivko Kamarashev, Chun-Lei Zhang,2 Reinhard Dummer, GuÈnter Burg, and Udo DoÈbbelingDepartment of Dermatology, University Hospital of ZuÈrich, ZuÈrich, Switzerland

On testing cutaneous T cell lymphoma cell lines andskin lesions, we found that the transcription factorsSTAT2, STAT3, STAT5, and STAT6 (STAT, signaltransducer and activator of transcription) were pre-sent in the nuclei of these cells and that the bindingto their speci®c DNA binding sites was stimulated byinterleukin-7 and interleukin-15. DNA binding stud-ies also revealed the presence of three additionalDNA factors in cutaneous T cell lymphoma cellsthat bound to the same sequences and could also bestimulated by interleukin-7 and interleukin-15. One

of these novel factors was also present in the adult Tcell leukemia cell line Jurkat and malignant T cellsfrom the blood of SeÂzary syndrome patients, but notin normal peripheral blood lymphocytes. It maytherefore be a marker of T cell leukemia. It seems tointerfere with the binding of STAT1 to the sis indu-cible element, suggesting that the DNA bindingactivity of STAT1 in cutaneous T cell lymphomacells is disturbed. Key words: gene regulation/oncogenes/signal transduction/transcription factors. J Invest Dermatol117:583±589, 2001

Cutaneous T cell lymphoma (CTCL) is a group oflymphoproliferative disorders of the skin (Burg etal, 1997). The most frequent forms of CTCL aremycosis fungoides (MF) and its leukemic counter-part the SeÂzary syndrome (SS). MF usually

progresses very slowly (over a period of 5±20 y) but leadsmostly to death by systemic spread of CTCL tumors orimmune disbalance and superinfections. In addition to general-ized erythroderma, patients with SS also have malignant T cellsin the blood. Their life expectancy is generally shorter (3 y)than that of patients who suffer from MF. Except in late stagesof disease lymphocyte accumulation in CTCL remains restrictedto the skin. This fact implies that CTCL cells may remaindependent on growth factors preferentially produced in the skin.Candidates for such factors are interleukin-7 (IL-7), which isproduced by keratinocytes among other cell types (Dalloul et al,1992; Matsue et al, 1993), and IL-15, which is expressed innormal skin (Mohamadzadeh et al, 1995; Blauvelt et al, 1996).Both IL-7 and IL-15 have been shown to increase the survivalof freshly isolated CTCL cells in vitro (DoÈbbeling et al, 1998).IL-15 is also produced by CTCL cells in the skin of patients

(DoÈbbeling et al, 1998), and increasing endogenous IL-15production during tumor progression (Asadullah et al 2000) mayhelp CTCL cells to become independent of IL-15 produced inthe cutaneous environment.

IL-15 has been identi®ed as a T cell growth stimulating cytokineproduced by many cell types and tissues (Grabstein et al, 1994). Thereceptors of IL-2, IL-7, and IL-15 are structurally related: the IL-15receptor contains the b and g chain of the IL-2 receptor and arecently identi®ed speci®c a chain (Giri et al, 1994; 1995; Grabsteinet al, 1994). IL-15 could replace IL-2 in several systems (Giri et al,1994; Grabstein et al, 1994; Armitage et al, 1995), but mostlyproved to be less effective than IL-2. The IL-7 high af®nityreceptor also contains the IL-2 receptor g chain (see Nowak, 1993,for review).

The IL-2 receptor b and g chains interact with the tyrosinekinases Janus kinase 1 (Jak1) and Jak3 (Miyazaki et al, 1994; Russellet al, 1994), which in turn activate the transcription factors STAT3(signal transducer and activator of transcription 3) and STAT5 (Giriet al, 1994; Nielsen et al, 1994; Fujii et al, 1995; Wakao et al, 1995).The STAT family factors bind to DNA elements that mediate theeffects of hormones and growth factors such as prolactin, plateletderived growth factor, erythropoietin, granulocyte macrophagecolony stimulating factor, and interferons. As the IL-7 and IL-15receptors contain one or two subunits that are also part of the IL-2receptor, one may therefore expect that IL-7 and IL-15 may causethe same effects as IL-2 (Miyazaki et al, 1995). Recent ®ndings doindeed point in this direction (Johnston et al, 1995; Schindler andDarnell, 1995). It may also be possible, however, that the recentlyidenti®ed IL-15 receptor a subunit activates additional STATfactors.

To study which STAT factors are activated by IL-7 and IL-15we used the IL-7/IL-15 dependent CTCL cell line SeAx (Kaltoft etal, 1987) as a test system. These cells are able to survive 4±6 d in theabsence of both cytokines. To determine the effects of IL-7 and IL-15, the cells were kept for 3 d in the absence of both interleukins

Manuscript received February 22, 2001; revised April 10, 2001; acceptedfor publication April 17, 2001.

Reprint requests to: Dr. Udo DoÈbbeling, Department of Dermatology,University Hospital of ZuÈrich, CH-8091 ZuÈrich, Switzerland. Email:doebbeli@derm.unizh.ch

Abbreviations: EMSA, electrophoretic mobility shift assay; Jak, Januskinase; MF, mycosis fungoides; SIE, sis inducible element; SS, SeÂzarysyndrome; STAT, signal transducer and activator of transcription.

1Present address: Skin Disease Research Laboratories, OncologyInstitute, Loyola University Medical Center, Maywood, IL, U.S.A.

2Present address: The University of Texas, MD Anderson CancerCenter, Department of Dermatology, Houston, TX, U.S.A.

0022-202X/01/$15.00 ´ Copyright # 2001 by The Society for Investigative Dermatology, Inc.

583

and after this time the cells were stimulated with IL-7 or IL-15 orleft unstimulated. This system allows us to distinguish the effects of

IL-2, IL-7, and IL-15 and to detect irregularities of interleukinsignal transduction in CTCL cells.

MATERIALS AND METHODS

Cell cultures The cell line HUT 78 (SS) was obtained from theEuropean Collection of Animal Cell Cultures. The cell lines MyLa 2059(MF) and SeAx (SS) were kind gifts of Dr. Keld Kaltoft, University ofAarhus, Denmark. HUT 78, MyLa 2059, and SeÂzary cells from patientsat our department were grown in HEPES-buffered RPMI 1640 mediumwith 2 mM glutamine, supplemented with 10% fetal bovine serum(FBS), 0.25 mg per ml amphotericin B, 100 U penicillin G, 100 Ustreptomycin, and 1 mM pyruvate. SeAx cells were grown under thesame conditions with the exception that 10% human serum instead of10% FBS was used. The concentrations of the cytokines were as follows:IL-2, 50 ng per ml (100 U); IL-7, 5 ng per ml (10 U); IL-15, 10 ng perml (10 U).

Blood and skin samples Skin biopsies and blood samples were takenprimarily for diagnostic purposes with informed consent of the patients.The specimens used were surplus material available after all the routinediagnostic procedures. The assignment to a certain stage was doneaccording to the recommendations of the EORTC CutaneousLymphoma Project Group. SeÂzary cells from patients' blood wereisolated by Ficoll gradient centrifugation and sorted by two-color¯uorescence-activated cell sorter using antibodies against CD 4 and thevb region of their T cell receptor (Dummer et al, 1996).

Electrophoretic mobility shift assay (EMSA) and ``supershift''experiments EMSA were performed according to Barberis et al(1987). Double stranded g-32P ATP labeled oligonucleotides (30,000cpm) containing the STAT factor binding motifs of the sis inducibleelement (SIE) and the IL-4 responsive element (IL-4RE) were incubatedwith 3 mg of nuclear extracts of the investigated cells in binding bufferconsisting of 10 mM HEPES pH 7.9, 60 mM KCl, 4% Ficoll, 1 mMdithiothreitol, and 1 mM ethylenediamine tetraacetic acid (EDTA)pH 8.0. Two micrograms of poly[d(I±C)] (poly deoxy-inosinic±deoxy-citidylic acid) were used as competitor for unspeci®c DNA bindingactivities. The total volume of the reaction was 30 ml. In the ``supershift''experiments, 2±4 mg of the corresponding antibody (Santa CruzBiotechnology, TransCruz gel supershift reagent) were added. Nuclearextracts were prepared according to Gerber et al (1992). The reactionmixture was incubated for 30 min at 4°C and then loaded on a 4%native polyacrylamide gel (0.25 3 Tris borate EDTA buffer). Theelectrophoresis was run at 4°C for 8±10 h in 0.25 3 TBE buffer at 10 Vper cm. The oligonucleotide sequences for the SIE and IL-4RE were 5¢GTGCATTTCCCGTAAATCTTGTCTACA 3¢ and 5¢ GAGCCTGA-TTTCCCCGAAATGATGAGC 3¢, respectively. The oligonucleotideswere synthesized by Microsynth (Balgach, Switzerland).

Western blots For Western blotting 15±30 mg protein of nuclearextracts or 30±60 mg protein of cytoplasmatic extracts were loaded on a7.5%±9% sodium dodecyl sulfate polyacrylamide gel and separated bypolyacrylamide gel electrophoresis. The percentage of polyacrylamideused depended on the molecular weight of the protein to beinvestigated. The proteins were transferred to a nitrocellulose ®lter usinga Mini Trans Blot Cell (Bio-Rad) following the instructions of thesupplier. Unspeci®c antibody binding sites were blocked by incubationof the ®lter overnight at 4°C in Tris-buffered saline (TBS), pH 8.0, 0.3%Tween 20, and 2% milk powder. The ®lter was incubated with thecorresponding ®rst antibody (Santa Cruz Biotechnology, 1:1000 dilution)for 4 h at room temperature in TBS, 0.3% Tween 20, and 1% milkpowder. The incubation with the second antibody (antirabbit, SantaCruz Biotechnology, 1:1000 dilution) was done in TBS, 0.3% Tween 20for 4 h at room temperature. The signal was detected by incubation withBM purple AP substrate (Roche Biochemicals) following the instructionsof the supplier.

RESULTS

IL-7 and IL-15 stimulate the binding of proteins to theSIE On studying the SIE oligonucleotide for STAT factors innuclear extracts from SeAx cells by EMSA we detected two DNA±protein complexes CS1 and CS2 (CS, complex with SIE), whichwere stimulated by IL-2, IL-7, and IL-15 (Fig 1A). Thestimulation occurred within 30 min, indicating that alreadypresent transcription factors were directly activated and that noprotein synthesis was necessary. To determine whether these

Figure 1. Binding of STAT to the SiE. (A) The in¯uence of IL-2(100 U), IL-7 (10 U), and IL-15 (10 U) on DNA binding activities tothe SIE oligomer in SeAx cells. Lane 1, negative control with peripheralblood lymphocytes; lanes 2, 3, positive controls with HUT 78 and MyLa2059 cells; lane 4, no interleukin; lane 5, IL-2 treated SeAx cells; lane 6,IL-7; lane 7, IL-15. (B) Identi®cation of the SIE binding proteins inSeAx cells by antibodies against the transcription factors STAT1±STAT6.Lane 1, negative control with peripheral blood lymphocytes; lane 2,untreated SeAx cells; lanes 3±9, IL-15 treated SeAx cells. The STATproteins were identi®ed by speci®c antibodies against the individualSTAT1±STAT6 proteins (lanes 4±9). The speci®city of the addedantibody is given above the corresponding lane. Only the binding ofSTAT3 could be detected (lane 5, CS2). (C) Determination of theidentity of the SIE binding activities in HUT 78 cell line by STATantibodies. Lane 1, negative control with peripheral blood lymphocytes;lanes 2±9, HUT 78 cells. The speci®city of the added antibody is givenabove the corresponding lane. Lane 9, competition test with 100 timesexcess of unlabeled SIE oligonucleotide (100 3 SIE). (D) SIE bindingactivities in MyLa cells. The DNA binding proteins were determined byantibodies against STAT1±STAT6. Lane 1, negative control withperipheral blood lymphocytes; lanes 2±8, MyLa cells. The speci®city ofthe added antibody is given above the corresponding lane. CSM, MyLa-speci®c SIE complex; STAT2/5-Ab, STAT2 and STAT5 antibodycomplexes.

584 QIN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

complexes contain STAT proteins, we added speci®c antibodiesdirected against STAT1±STAT6 to the DNA binding reaction.The recognition of a DNA binding protein either causes thedisruption of the protein±DNA complex, when the antibody isdirected at the DNA binding domain of the tested protein, orcreates a supershift complex, consisting of the DNAoligonucleotide, DNA binding protein, and the antibody. Thesesupershift complexes migrate even more slowly than the DNA±

protein complexes. Testing the complexes in SeAx cells withantibodies against STAT1±STAT 6, we found that CS2 wasdisrupted by an antibody against STAT3, indicating that thiscomplex contained STAT3 (Fig 1B). CS1 was not signi®cantlyaffected by any STAT antibody. In some experiments we found aslight reduction of CS1 by STAT1 antibodies, but the result wasmarkedly weaker than the supershift that we obtained with the SIEand interferon-g stimulated peripheral blood lymphocytes fromnormal donors (data not shown). The nuclear extracts from theHUT 78 and MyLa cell lines also contained CS1 and CS2. CS1 isprobably speci®c for leukemia T cell lines, as it was found in theacute T cell leukemia cell line Jurkat, but not in hepatoma(HepG2), breast cancer (MCF-7), or melanoma cells (J.-Z.Q. andU.D., unpublished data). We therefore called the CS1 formingfactor SBPT (SIE binding protein in T cell lymphoma/leukemia).

Also in HUT 78 and MyLa 2059 cells CS2 was disrupted by aSTAT3 antibody, and CS1 behaved as in SeAx cells (Fig 1C). InMyLa 2059 cells an additional complex (CSM) was found, whichcould be disrupted by antibodies against STAT2 and STAT5(Fig 1D). In later experiments we found that CSM could also bevisualized when higher amounts of nuclear extracts of HUT 78 andSeAx were loaded (not shown). The signal of CS2 in MyLa cellswas somewhat reduced by antibodies against STAT4 and STAT5.As we found no DNA binding of STAT4 on a STAT4-speci®coligo (see below), we interpret this signal reduction as a cross-reaction between STAT3 and STAT4 antibodies. The sameexplanation may also be true for the reduction of the CS2 signalby the STAT5 antibody, as STAT5 is present in the CSM complex.The presence of low amounts of STAT5 in CS2 cannot be formallyexcluded, however. In MyLa 2059 and HUT 78 the stimulatingeffect of IL-2, IL-7, and IL-15 was much weaker, which wasprobably due to the high basal levels of the SIE binding activities inthese cell lines (data not shown). The complexes below CS2 werenot always reproducible and may be due to sequence-unspeci®cDNA binding proteins.

IL-7 and IL-15 stimulate the binding of STAT2, STAT5,and STAT6 to the IL-4 responsive element of the Fcgreceptor I gene (IL-4RE) As not all STAT factors bind to thesame sequence we tested another sequence, which is known tobind STAT factors: the IL-4RE (Hou et al, 1994). Using the IL-4RE and nuclear extracts from SeAx cells we found two maincomplexes, C-IL-4RE 1 and C-IL-4RE 2. IL-7 and IL-15increased the DNA binding of both complexes. In normalperipheral blood lymphocytes (Fig 2B, lane 1) IL-15 inducedanother complex, which perhaps is also present in the CTCL celllines in low amounts. Supershift experiments with antibodiesagainst STAT1±STAT6 (Fig 2B) showed that C-IL-4RE 2 wastotally supershifted by an antibody against STAT5 and partiallysupershifted by antibodies against STAT2 and STAT6. From thisresult one can conclude that C-IL-4RE 2 consists mostly of STAT5and a lower amount of STAT2 and STAT6. The presence ofseveral supershift bands suggests that C-IL-4RE 2 consists ofdifferent homodimers and heterodimers of these molecules. C-IL-4RE 1 showed no supershift with any STAT antibody and only aweak reduction of the signal was observed when the STAT1antibody was used. In contrast to the other factors the C-IL-4RE 1forming factor could also bind to mutated IL-4RE, which lackedthe characteristic thymidine triplet (J.-Z.Q. and U.D., unpublisheddata). In HUT 78 and MyLa 2059 cells C-IL-4RE 2 was alsopresent and reacted with antibodies against STAT2, STAT5, andSTAT6 (Fig 2C, D). Using the so-called IL-4 gamma interferonactivation sequence by which STAT 4 has been identi®ed, weobserved no DNA binding of STAT4, and only a weak interactionof STAT2, STAT5, and STAT6 with this oligo could be found(data not shown).

Western blot analysis of STAT factors To see whether theSTAT factors that we found in the supershift experiments wereintact and of the expected size we performed Western blots usingnuclear extracts of the three CTCL cell lines. All the STAT factors

Figure 2. Binding of STAT to the IL-4 responsive element (IL-4RE) in CTCL cells. (A) IL-7 and IL-15 stimulate the binding ofproteins to the IL-4RE of the Fcg RI gene in SeAx cells. Lane 1, normaluntreated peripheral blood lymphocytes; lane 2, unstimulated SeAx cells;lanes 3±5, IL-2 (100 U), IL-7 (10 U), IL-15 (10 U) stimulated SeAxcells. (B) Identi®cation of the proteins binding to the IL-4RE byantibodies against STAT1±STAT6 in IL-15 stimulated SeAx cells. Lane1, negative control with IL-15 (10 U) treated peripheral bloodlymphocytes; lanes 2±8, SeAx cells. The speci®city of the added antibodyis given above the corresponding lane. STAT-Ab, STAT±DNA antibodycomplexes. (C) Determination of Fcg RI IL-4RE binding activities inHUT 78 cells. Lane 1, negative control with untreated peripheral bloodlymphocytes; lanes 2±8, SeAx cells. The speci®city of the added antibodyis given above the corresponding lane. (D) Determination of Fcg RI IL-4RE binding activities in MyLa 2059 cells. Lane 1, negative control withuntreated peripheral blood lymphocytes; lanes 2±8, SeAx cells. Thespeci®city of the added antibody is given above the corresponding lane.Only C-IL-4RE 2, consisting of STAT2, STAT5, and STAT6, could bedetected in HUT 78 and MyLa 2059.

VOL. 117, NO. 3 SEPTEMBER 2001 STAT FACTORS IN CTCL 585

seen in the supershift experiments (STAT2, STAT3, STAT5,STAT6) were detected by Western blotting (Fig 3) in all three celllines. STAT1 and STAT4, which showed no DNA binding inEMSA, were also tested and surprisingly STAT1 was detected in allthree nuclear extracts (Fig 3A). When we used the gammainterferon activating sequence in supershift experiments, however,a STAT1-speci®c antibody revealed DNA binding of STAT1 tothese sequences (U.D., J.-Z.Q., and C.-L.Z., unpublished data).

The only size difference of STAT proteins in the CTCL celllines was seen for STAT2. In HUT 78 and MyLa 2059 cells aSTAT2 form of 105 kDa was prevalent, whereas in SeAx cells a120 kDa form dominated (Fig 3B). In all the three cell lineswe found the wild-type STAT3 protein. Some preparations ofthe K-15 and C-20 antibodies detected proteins of 120±130 kDa (Fig 3C) and 60 kDa (not shown); however, incontrast to the STAT3 wt signal these bands were not alwaysreproducible. No consistent signal for a larger mutant STAT3protein in MyLa 2059 cells that has been previously reported byNielsen et al (1994) was detected, although we used the sametwo antibodies.

No signal of the expected size of 88 kDa was detected for STAT4(Fig 3D). The signals seen in this blot may derive from mutant orSTAT4-related proteins from which only expressed sequence tags(EST) or partial cDNAs have been cloned. Both STAT5 proteins,STAT5a and STAT5b, were present in all three cell lines (Fig 3E).

In general, the concentrations of STAT factors were highest inMyLa 2059 cells and lowest in SeAx cells when equal amounts ofprotein were loaded.

In vivo expression and localization of STATs in CTCL skinlesions When we tested the skin lesions of CTCL patients for thepresence of STAT1, STAT2, STAT3, STAT5, and STAT6, wefound cytoplasmic staining in the malignant T cells of all six MFand six SS patients tested. The expression and distribution ofSTAT1, STAT2, STAT3, and STAT6 were nearly identical andcorrespond to the four samples given in Fig 4(A)±(D) for STAT3.The staining was predominant in the cytoplasm, but nuclearstaining was also detected in skin lesions of all tested CTCLpatients. In one half of the patients we observed signi®cant nuclearstaining in many malignant cells (Fig 4C, D), whereas in the otherhalf only a few cells showed nuclear staining (Fig 4A, B). Therewas no strict correlation between nuclear localization and diseaseprogression.

All samples tested by a STAT5 antibody showed strong STAT5staining in the nucleus (Fig 4E±H). Strong nuclear STAT5 stainingwas already evident in early stages (Fig 4B, top left) and did notincrease signi®cantly in later stages.

SBPT is present in malignant T cells in the blood of SSpatients As no antibodies against SBPT are available and nofunctional nuclear extracts from skin biopsy material can beprepared, we investigated nuclear extracts of malignant T cellsfreshly isolated from the blood of patients with SS for the presenceof SBPT. EMSA showed that a complex migrating at the sameposition as CS1 in the CTCL cell lines was present in all sevenpatients tested, whereas CS2, which may represent STAT3, waspresent in only one patient (Fig 5A). In one patient a very strongCS1 signal was found ± possibly two different complexes migratingclosely together composed this signal. As in SeAx cells IL-7 and IL-15 could stimulate the CS1 complex in malignant T cells of patients(Fig 5B). This complex also did not react signi®cantly with STATantibodies. Due to this behavior and the common migrationbehavior on the gel we suppose that CS1 is the same in all the testedsamples and represents SBPT. As SBPT is absent in the peripheralblood lymphocytes of healthy donors we suppose that it isindicative for CTCL cells in the blood.

When we performed corresponding experiments with the IL-4RE we found no binding of STATs to this DNA sequence(J.-Z.Q. and U.D., unpublished data).

DISCUSSION

Our results show that the STAT factors 1, 2, 3, 5, and 6 areconstitutively active in CTCL cell lines and that they can bestimulated by IL-7 and IL-15 in SeAx cells. This is also true forthe presumed novel factors that bind to the SIE and the IL-4RE. The ®ve STATs are also present in the nuclei ofmalignant T cells in CTCL skin lesions, where STAT5, as inthe cell lines, seems to be the most prominent factor. Thebinding of STAT1 to the SIE may be inhibited by the novelSIE binding factor SBPT.

With respect to constitutive STAT activities, CTCL cellsresemble leukemic cells from acute myeloid leukemia, Burkitt'slymphoma, and adult T cell leukemia patients (Gouilleux-Gruart etal, 1996; Weber±Nordt et al, 1996; Takemoto et al, 1997). Ingeneral the unstimulated STAT factor binding is higher in the IL-7/IL-15 independent cell lines HUT 78 and MyLa 2059 than in theIL-7/IL-15 dependent SeAx cells. IL-7 and IL-15 increase theDNA binding of STAT and the novel STAT motifs binding factorsin SeAx cells and thus compensate for the lower constitutiveactivity of these factors in this cell line. As SeAx cells survive in the

Figure 3. Expression of STAT1±STAT6 in CTCL cells. (A)STAT1 expression in HUT 78 (lane 1), MyLa 2059 (lane 2), and SeAx(lane 3) cells. (B) STAT2 expression in HUT 78 (lane 1), MyLa 2059(lane 2), and SeAx (lane 3) cells. (C) Expression of STAT3 isoforms inHUT 78 (lanes 1, 4), MyLa 2059 (lanes 2, 5), and SeAx (lanes 3, 6) cellstested with antibody K-15 (lanes 1±3) and C-20 (lanes 4±6). (D) STAT4,(E) STAT5, and (F) STAT6 expression in HUT 78 (lane 1), MyLa 2059(lane 2), and SeAx (lane 3) cells. The 97.4 kDa band of the protein sizemarker is given on the left.

586 QIN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

absence of IL-7 or IL-15 only for 4±6 d, a certain threshold level ofSTAT factor activities may be necessary for the survival of thesecells. Due to their high constitutive DNA binding activities, IL-7and IL-15 increase the DNA binding of STATs in HUT 78 andMyLa 2059 cells only weakly (data not shown).

Despite the inactivity of STATs, SBPT is also constitutivelypresent in the nuclei of unstimulated malignant T cells of SSpatients. As the DNA binding activity of this protein is not detectedin the blood of normal donors, it may play a role for thecancerogenesis of CTCL. SBPT may be typical for leukemia Tcell lines, as it was found in the acute T cell leukemia cell lineJurkat, but not in peripheral blood lymphocytes of heathy donors.It was also absent in hepatoma (HepG2), cervix carcinoma (HeLa),breast cancer (MCF-7), and melanoma cells (J.-Z.Q. and U.D.,unpublished data).

IL-7 and IL-15 stimulated in SeAx cells a broader range of STATand novel STAT motifs binding factors than has been reported forother cell lines (Schindler and Darnell, 1995; Leonard and O'Shea,1998). Our results indicate that the speci®city of IL-7 and IL-15signaling to STAT factors has been lost in CTCL cell lines: evenSTAT2, a factor that is selectively induced by interferon-a, wasstimulated by these two interleukins. This loss of speci®city and theconstitutive binding activity of STAT factors cannot be explainedby a constitutive stimulation through the IL-7 and IL-15 regulatedtyrosine kinases Jak1 and Jak3 (Haque et al, 1997) alone, as theyshould only stimulate STAT3 and STAT5.

Other tyrosine kinases that have been reported to stimulateSTAT DNA binding, such as c-Src (Yu et al, 1995; Olayioye et al,1999) or Bmx (Saharinen et al, 1997; Wen et al, 1999), may also beinvolved in the constitutive activation of STATs in CTCL cells.

Figure 4. Expression and localization ofSTAT proteins in CTCL cells in skin lesions.(a±d) STAT3 antibody staining experiments withcryostat sections from skin lesions of two MFpatients (a, c) and two SS patients (b, d). (a) and(b) represent early stages, (c) and (d) late stages. (e±h) Corresponding stainings with a STAT5antibody: (e) early MF stage, (f ) early SS stage, (g)late MF stage, (h) late SS stage. Themagni®cations are 100x (a±c, f ±h) and 62.5x (d, e).

VOL. 117, NO. 3 SEPTEMBER 2001 STAT FACTORS IN CTCL 587

The constitutive activities of STATs may be due to thefunctional loss of STAT inhibitors. Recently a family of STATinhibitors SOCS1±SOCS3 (suppressor of cytokine signaling),which is involved in a feedback regulation of STAT activation,has been identi®ed (Endo et al, 1997; Naka et al, 1997; Starr et al,1997). A failure of these regulators could cause the perpetuation ofa cytokine signal. Indeed, it was found that CTCL cell lines lackfunctioning SOCS3 proteins (Brender et al, 2001). The highconcentrations of some STATs, especially STAT5, in the cytoplasmof CTCL cells in skin lesions suggest that over-expressed STATscould also titrate STAT inhibitors and that therefore a fraction ofthese proteins can migrate into the nucleus.

The Western blots detected the two known forms of STAT2(p105 and p120). For STAT3 our ®ndings with MyLa 2059 cellsdiffer from those of Nielsen et al (1997), who detected a slowlymigrating form of STAT3 but no STAT2 and STAT5 binding.Their ®ndings may be due to the long culture time of this cell line.STAT3 is constitutively present in all three cell lines and in themalignant T cells of one patient. The presence of STAT3 in themalignant T cells of at least one patient suggests that the activationof STAT3 is a late but normally occurring event during theprogression of CTCL. It may be that SeÂzary cells get the ability toproliferate in very late stages, as we found expression of theproliferation-promoting c-myc and junD genes in SeÂzary cells ofpatients with very high numbers of leukemic T cells (Qin et al,1999).

No STAT4 protein of the expected size and DNA binding couldbe detected in nuclear extracts of CTCL cells. The absence of thisprotein or its inability to become activated may explain why IL-12induces no STAT4-mediated gene expression in CTCL cells(Showe et al, 1999). We detected no DNA binding of STAT

proteins in the leukemic T cells of SS patients. Zhang et al (1996)reported phosphorylated STAT5 proteins in 70% of their SSpatients; however, they did not analyze their DNA bindingactivities. Thus it may be possible that STAT5 needs additionalserine phosphorylation for effective binding to the IL-4RE.

CTCL cells are a good source of new IL-7 and IL-15 stimulatedtranscription factors as, besides SBPT, we have already discoveredtwo novel E-box binding proteins in these cells (Qin et al, 1999).The stimulation of the DNA binding of these factors is speci®c, as itoccurs within less than 30 min and is therefore not due to de novoprotein synthesis; it can occur during a general recovery frominterleukin starvation of the cells after the addition of IL-7 or IL-15.The speci®city of the effects is also underscored by the fact that thetranscription factors c-Myc, Max, and OCT-1 (Qin et al, 1999) arenot stimulated by these two interleukins. IL-7, IL-15, and theSTATs may play an important role for the survival of CTCL cells,as we observed that they increased the binding of STAT2, STAT5,STAT6, and c-Myb to their binding sites in the promoter of thebcl-2 gene and consequently increased the amount of Bcl-2 proteinin CTCL cells (Qin et al, 2001). CTCL cells may gain IL-7 and IL-15 independence when signal transduction molecules like tyrosinekinases mutate to constitutively active enzymes.

The authors would like to thank Dr. Keld Kaltoft for the generous gift of the MyLa

2059 and SeAx cell lines and M. Johnson and M. BaÈr for the preparation of the

photographs. This work was supported by the Swiss Cancer League (grant KFS

275±1-1996 to G.B.), the Swiss National Science Foundation (grant 3100-

43244.95/1) to G.B., the United Bank of Switzerland (UBS), and the Kanton

of ZuÈrich.

REFERENCES

Armitage RJ, Macduff BM, Eisenman J, Paxton R, Grabstein KH: IL-15 hasstimulatory activity for the induction of B cell proliferation and differentiation.J Immunol 154:483±490, 1995

Asadullah K, Haeussler-Quade A, Gellrich S, et al: IL-15 and IL-16 overexpression incutaneous T-cell lymphomas: stage-dependent increase in mycosis fungoidesprogression. Exp Dermatol 9:248±251, 2000

Barberis A, Superti-Furga G, Busslinger M: Mutually exclusive interaction of theCCAAT-binding factor and of a displacement protein with overlappingsequences of a histone gene promoter. Cell 50:347±359, 1987

Blauvelt A, Asada H, Klaus-Kovtun V, Altman DJ, Lucey DR, Katz SI: Interleukin-15 mRNA is expressed by human keratinocytes, Langerhans cells, and bloodderived dendritic cells and is downregulated by ultraviolet B radiation. J InvestDermatol 106:1047±1052, 1996

Brender C, Nielsen M, Kaltoft K, et al: STAT3-mediated constitutive expression ofSOCS-3 in cutaneous T cell lymphoma. Blood 97:1056±1062, 2001

Burg G, Kempf W, HaÈffner AC, et al: Cutaneous Lymphomas. Curr Probl Dermatol9:139±204, 1997

Dalloul A, Laroche L, Bagot M, et al: Interleukin-7 is a growth factor for SeÂzarylymphoma cells. J Clin Invest 90:1054±1060, 1992

DoÈbbeling U, Dummer R, Laine E, Potoczna N, Qin J-Z, Burg G: IL-15 is anautocrine/paracrine viability factor for cutaneous T cell lymphoma cells. Blood92:252±258, 1998

Dummer R, Heald PW, Nestle FO, et al: SeÂzary syndrome T-cell clones display T-helper 2 cytokines and express the accessory factor-1 (interferon-g receptor b-chain). Blood 88:1383±1389, 1996

Endo TA, Masuhara M, Yokouchi M, et al: A new protein containing an SH2domain that inhibits JAK kinases. Nature 387:921±924, 1997

Fujii H, Nakagawa Y, Schindler U, Kawahara A, Mori H, Taniguchi T: Activation ofSTAT5 by interleukin 2 requires a carboxy-terminal region of the interleukin 2receptor beta chain but is not essential for the proliferative signal transmission.Proc Natl Acad Sci USA 92:5482±5486, 1995

Gerber H-P, Georgiev O, Harshman K, Schaffner W: In vitro transcriptioncomplementation assay with mini extracts of transiently transfected COS-1cells. Nucl Acids Res 20:5855±5856, 1992

Giri JG, Ahdieh M, Eisenman J, et al: Utilization of the beta and gamma chains of theIL-2 receptor by the novel cytokine IL-15. EMBO J 13:2822±2830, 1994

Giri JG, Kumaki S, Ahdieh M, et al: Identi®cation and cloning of a novel IL-15binding protein that is structurally related to the alpha chain of the IL-2receptor. EMBO J 14:3654±3663, 1995

Gouilleux-Gruart V, Gouilleux F, Dasaint C, et al: STAT-related transcription factorsare constitutively activated in peripheral blood cells from acute leukemiapatients. Blood 87:1692±1697, 1996

Grabstein KH, Eisenman J, Shanebeck K, et al: Cloning of a T cell growth factor thatinteracts with the beta chain of the interleukin-2 receptor. Science 264:965±968,1994

Figure 5. SIE-oligo binding activities in the malignant T cells ofSS patients. (A) Lane 1, positive control with HUT 78 cells; lanes 2±8,patients 1±7. The CS1 complex found with nuclear extracts of patientcells is probably the same as in HUT 78 cells (lane 1). CS2 is found inonly one patient. (B) The effect of IL-7 and IL-15 on the SIE bindingactivities of malignant T cells of two SS patients (lanes 3±8). Lane 1,negative control with normal peripheral blood lymphocytes; lane 2,positive control with MyLa cells. Lanes 3±6, patient 2: lane 3, nointerleukin; lane 4, IL-7 (10 U); lane 5, IL-15 (10 U); lane 6, IL-7 plusIL-15 (5 + 5 U). Lanes 7±8, patient 4: lane 7, no interleukin; lane 8,IL-7 plus IL-15 (5 + 5 U).

588 QIN ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Haque SJ, Wu Q, Kammer W, et al: Receptor-associated constitutive proteintyrosine phosphatase activity controls the kinase function of JAK1. Proc NatlAcad Sci USA 94:8563±8568, 1997

Hou J, Schindler U, Henzel WJ, Ho CH, Brasseur M, McKnight SL: An interleukin-4-induced transcription factor. IL-4 Stat Sci 265:1701±1706, 1994

Johnston JA, Bacon CM, Finbloom DS, et al: Tyrosine phosphorylation andactivation of STAT5, STAT3, and Janus kinases by interleukins 2 and 15. ProcNatl Acad Sci USA 92:8705±8709, 1995

Kaltoft K, Bisballe S, Rasmussen HF, Thesstrup-Pedersen K, Thomsen K, Sterry W:A continuous T-cell line from a patient with SeÂzary syndrome. Arch DermatolRes 279:293±298, 1987

Leonard WJ, O'Shea JJ: Jaks and STATs: biological implications. Annu Rev Immunol16:293±322, 1998

Matsue H, Bergstresser PR, Takashima A: Keratinocyte-derived IL-7 serves as agrowth factor for dendritic epidermal T cells in mice. J Immunol 151:6012±6019, 1993

Miyazaki T, Kawahara A, Fuji H, et al: Functional activation of Jak1 and Jak3 byselective association with IL-2 receptor subunits. Science 266:1045±1047, 1994

Miyazaki T, Liu Z-J, Kawahara A, et al: Three distinct IL-2 signaling pathwaysmediated by bcl-2, c-myc, and lck cooperate in hemapoietic cell proliferation.Cell 81:223±231, 1995

Mohamadzadeh M, Takashima A, Dougherty I, Knop J, Bergstresser PR, Cruz PD:Ultraviolet B radiation up-regulates the expression of IL-15 in human skin. JImmunol 155:4492±4496, 1995

Naka T, Narazaki M, Hirata M, et al: Structure and function of a new STAT-inducedSTAT inhibitor. Nature 387:924±928, 1997

Nielsen M, Svejgard A, Skov S, Odum N: Interleukin-2 induces tyrosinephosphorylation and nuclear translocation of STAT3 in human Tlymphocytes. Eur J Immunol 24:3082±3086, 1994

Nielsen M, Kaltoft K, Nordahl M, et al: Constitutive activation of a slowly migratingisoform of STAT3 in mycosis fungoides: tyrpphostin AG490 inhibits STAT3activation and growth of mycosis fungoides tumor cell lines. Proc Natl Acad SciUSA 94:6764±6769, 1997

Nowak R: `Bubble boy' paradox resolved. Science 262:1818±1818, 1993Olayioye MA, Beuvink I, Horsch K, Daly JM, Hynes NE: ErbB receptor induced

activation of STAT transcription factors by Src tyrosine kinases. J Biol Chem274:17209±17218, 1999

Qin J-Z, Dummer R, Burg G, DoÈbbeling U: Constitutive and IL-7/ IL-15stimulated DNA-binding of Myc, Jun, and novel Myc-like proteins incutaneous T cell lymphoma (CTCL) cells. Blood 93:260±267, 1999

Qin J-Z, Zhang C-L, Kamavashev J, Dummer R, Burg G, DoÈbbeling U: IL-7 andIL-15 regulate the expression of the bcl-2 and c-myb genes in cutaneous T celllymphoma (CTCL) cells. Blood 98: in press, 2001

Russell SM, Johnson JA, Noguchi M, et al: Interaction of IL-2R beta and gamma cchains with Jak1 and Jak3: implications for XSCID XCID. Science 266:1042±1045, 1994

Saharinen P, Ekman N, Sarvas K, Parker P, Alitalo K, Silvennoinen O: The Bmxtyrosine kinase induces activation of the STAT signaling pathway, which isspeci®cally inhibited by protein kinase Cdelta. Blood 90:4341±4353, 1997

Schindler C, Darnell JE: Transcriptional responses to polypeptide ligands: the JAK-STAT pathway. Annu Rev Biochem 64:621±651, 1995

Showe LC, Fox FE, Williams D, Au K, Niu Z, Rook AH: Depressed IL-12mediated signal transduction in T cells from patients with SeÂzary syndrome isassociated with the absence of IL-12 receptor beta 2 mRNA and highlyreduced levels of STAT4. J Immunol 163:4073±4079, 1999

Starr R, Wilson TA, Viney EM, et al: A family of cytokine-inducible inhibitors ofsignalling. Nature 387:917±921, 1997

Takemoto S, Mulloy JC, Cereseto A, et al: Proliferation of adult T cell leukemia/lymphoma cells is associated with the constitutive activation of JAK/STATproteins. Proc Natl Acad Sci USA 94:13897±13902, 1997

Wakao H, Harada N, Kitamura T, Mui AL-F, Miyajima A: Interleukin 2 anderythropoietin activate STAT5/MGF via distinct pathways. EMBO J 14:2527±2535, 1995

Weber-Nordt RM, Egen C, Wehinger J, Ludwig W, Gouilleux-Gruart V,Mertelsmann R, Fink J: Constitutive action of STAT proteins in primarylymphoid and myeloid leukemia cells and in Epstein±Barr virus (EBV) -relatedlymphoma cell lines. Blood 88:809±816, 1996

Wen X, Lin HH, Shih HM, Kung HJ, Ann DK: Kinase activation of the non-receptor tyrosine kinase Etk/Bmx alone is suf®cient to transactivate STAT-mediated gene expression in salivary and lung epithelial cells. J Biol Chem274:38204±38210, 1999

Yu CL, Meyer DJ, Campbell GS, Larner AC, Carter-Su C, Schwartz J, Jove R:Enhanced DNA-binding activity of a Stat3-related protein in cells transformedby the Src oncoprotein. Science 269:81±83, 1995

Zhang Q, Nowak I, Vonderheid EC, et al: Activation of Jak/STAT proteinsinvolved in signal transduction pathway mediated by receptor forinterleukin 2 in malignant T lymphocytes derived from cutaneousanaplastic large T-cell lymphoma and SeÂzary syndrome. Proc Natl AcadSci USA 93:9148±9153, 1996

VOL. 117, NO. 3 SEPTEMBER 2001 STAT FACTORS IN CTCL 589