The precursor form of IL-1␣ is an intracrine proinflammatory activator of transcription

Ariel Werman*†, Rachel Werman-Venkert*, Rosalyn White*, Jae-Kwon Lee†, Batsheva Werman†, Yakov Krelin*, Elena Voronov*, Charles A. Dinarello†, and Ron N. Apte*‡

*Department of Microbiology and Immunology, Faculty of Health Sciences and Cancer Research Center, Ben-Gurion University of the Negev, Beer-Sheva 84105, Israel; and †Department of Medicine, University of Colorado Health Sciences Center, Denver, CO 80262

Contributed by Charles A. Dinarello, December 31, 2003 Although most are studied for biological effects after additional external stimulus. For example, mice deficient in engagement of their specific surface membrane receptors, IL-1␣ are resistant to experimental colitis (unpublished obser- increasing evidence suggests that some function in the nucleus. In vations), and IL-1␣-deficient but not IL-1␤-deficient mice fail to the present study, the precursor form of IL-1␣ was overexpressed prime antigen-specific T cells (9). in various cells and assessed for activity in the presence of satu- The possibility that, upon nuclear localization sequence- rating concentrations of IL-1 antagonist to prevent recep- dependent nuclear translocation, pIL-1␣ and propiece IL-1␣ tor signaling. Initially diffusely present in the cytoplasm of resting (ppIL-1␣) might exhibit functionality has been the focus of cells, IL-1␣ translocated to the to nucleus after activation by several studies. Indeed, overexpression of the propiece results in endotoxin, a Toll-like receptor . The IL-1␣ precursor, but not transformation to a malignant phenotype (10). The propiece also the C-terminal mature form, activated the transcriptional machin- has been shown to induce in some tumor cells by ery in the GAL4 system by 90-fold; a 50-fold increase was observed affecting alternate splicing of RNA (11). Antisense directed using only the IL-1␣ propiece, suggesting that transcriptional against IL-1␣ mRNA reverses the natural senescence of human activation was localized to the N terminus where the nuclear umbilical vascular endothelial cells (12). Such findings support localization sequence resides. Under conditions of IL-1 receptor the hypothesis of a nuclear site of action for IL-1␣. blockade, intracellular overexpression of the precursor and pro- The intracellular accumulation of cytokines like IL-1␣ lacking piece forms of IL-1␣ were sufficient to activate NF-␬B and AP-1. a signal likely represent an evolutionary advantage, Stable transfectants overproducing precursor IL-1␣ released the because readily secreted in aquatic life forms would not cytokines IL-8 and IL-6 but also exhibited a significantly lower be able to function during regenerative processes. An example is threshold of activation to subpicomolar concentrations of tumor the ability of the Asterias forbesi starfish to regenerate a severed necrosis factor ␣ or IFN-␥. Thus, intracellular functions of IL-1␣ arm. The macrophage-like coelomocytes of these animals con- might play an unforeseen role in the genesis of inflammation. tain a preformed IL-1-like molecule (13, 14) that is delivered to During disease-driven events, the cytosolic precursor moves to the the site of injury as an intracellular . If growth nucleus, where it augments transcription of proinflammatory factors in starfish were secreted proteins, diffusion into the genes. Because this mechanism of action is not affected by extra- surrounding seawater would markedly reduce the concentration cellular inhibitors, reducing intracellular functions of IL-1␣ might needed at the site for repair. That IL-1␣ and other intracellular prove beneficial in some inflammatory conditions. cytokines persist after the evolution of extracellular receptors in terrestrial life suggests a duality of function, one function being he IL-1 family has 10 known members, and the precursor solely in intracellular processes and the second as a mechanism Tform of each member lacks a clear signal peptide, suggesting to trigger the extracellular receptor and to initiate signal trans- persistence of an early evolutionary role as intracellular proteins duction. The two functions are not necessarily related. In the (1). Although one form of the IL-1 receptor (IL-1R) antagonist case of high-mobility group box 1 (HMGB1), one function is to (IL-1Ra) possesses a signal peptide and is secreted by means of facilitate gene transcription by stabilizing nucleosomes and the Golgi, the IL-1Ra gene has alternative splice͞insertion sites; allowing bending of DNA, and the second function is to trigger thus, in nearly all cells, variants of the antagonist extracellular receptors as typical proinflammatory cytokine pro- remain intracellular (reviewed in ref. 2). Although recombinant duced late in endotoxemia (15). forms of mature IL-1␣ (mIL-1␣), mIL-1␤, and mIL-18 are active In the present report, we studied a nuclear function of ␣ agonists, the precursor forms of IL-1␤ and IL-18 (pIL-1␤ and intracellular IL-1 as an activator of transcriptional machinery with its effects on the progression of inflammation. Expression pIL-18) are inactive. These precursors use the intracellular ␣ cysteine protease caspase-1 to cleave the precursor to an active vectors encoding the three forms of human IL-1 were con- ␣ structed: the full-length pIL-1␣ (amino acids 1–271), ppIL-1␣ mature form and exit the cell. The precursor form of IL-1 ␣ (pIL-1␣), in contrast, is fully active when it engages the IL-1 (amino acids 1–112), and mIL-1 (amino acids 113–271). We surface receptor. Unlike IL-1␤, IL-1␣ is rarely found in the also blocked all by means of the surface extracellular compartment but rather is primarily associated with receptors by using saturating concentrations of IL-1Ra. It ap- pears that one of the functions of intracellular IL-1␣ is to lower the cell as either an intracellular or a membrane form (3, ␬ 4). A membrane-associated calcium-dependent protease, cal- the threshold of NF- B- and AP-1-dependent gene expression to pain, can process pIL-1␣ to a mature form (5). However, many subpicomolar concentrations of inflammatory stimuli. Two ob- vious implications of these results are priming (memory) and cells are devoid of this processing ability, leaving the intracellular ␣ precursor form in abundance. pIL-1␣ possesses a KVLKKRR chronicity. Cells previously primed to express IL-1 will hyper- nuclear localization site in its N-terminal propiece (6). In fact, pIL-1␣ translocates to the nucleus in a variety of cells, and this Abbreviations: TNF, tumor necrosis factor; CFP, cyan fluorescent protein; YFP, yellow appears to be necessary for specific downstream events (7). fluorescent protein; IL-1R, IL-1 receptor; IL-1Ra, IL-1R antagonist; pIL, precursor IL; ppIL, pIL-1␣ is found in keratinocytes and epithelial cells of healthy propiece IL; mIL, mature IL; HMGB1, high-mobility group box 1; DBD, DNA binding domain; subjects, and mice deficient in IL-1␣ do not exhibit an apparent UAS, upstream activator sequence; MIP-2, macrophage inflammatory protein 2. phenotype (8) unless challenged by disease processes, suggesting ‡To whom correspondence should be addressed. E-mail: [email protected]. that a putative function for the intracellular pIL-1␣ requires an © 2004 by The National Academy of Sciences of the USA

2434–2439 ͉ PNAS ͉ February 24, 2004 ͉ vol. 101 ͉ no. 8 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0308705101 Downloaded by guest on September 26, 2021 respond to minute concentrations of various inflammatory chal- lenges as compared with previously unprimed cells. In contrast, overexpression of intracellular IL-1␣ often seen in chronic diseases might play a pivotal role in the pathogenesis and maintenance of chronicity in an IL-1R-independent manner. Materials and Methods Cloning of Human IL-1␣, IL-18, and IL-1␤ cDNA. IL-1␣ and IL-18 were cloned by PCR amplification using a human keratinocyte cDNA library as a template. The primers contained a random pen- tamer, an endonuclease restriction site, and the first and last 24 bp of the coding region of the gene for the 5Ј and 3Ј primers, respectively. pIL-1␤ was amplified by using cDNA derived from human peripheral blood mononuclear cells. Amplified PCR Fig. 1. pIL-1␣ is translocated to the nucleus upon inflammatory stimuli. fragments were cloned into pCR3.1 or pcDNA6͞His-Myc (In- NIH͞3T3 cells (20,000 per well) were plated on a cover slip placed in six-well vitrogen). Positive clones were verified by direct sequencing. plates. Twenty-four hours later, cells were transfected with both the vectors encoding the fusion proteins pIL-1␣͞EYFP-N1 and pIL-1␣͞ECFP-C1 (a). Twenty- Fluorescence Constructs. pECFP-C1 and pEYFP-N1 were pur- four hours after transfection, an aliquot of cells were stimulated with lipo- chased from Clontech. IL-1␣–enhanced cyan fluorescent pro- polysaccharide (10 ␮g͞ml). Twenty-four hours later, cover slips were observed tein (ECFP) constructs were generated by inserting a PCR- by fluorescent microscopy. (a) Fusion proteins used (calpain cleavage site marked by arrow). (b and c) Unstimulated cells observed through yellow and generated fragment into a BglII and EcoRI sites in pECFP-C1 ␣ cyan filters, respectively. (d–f) Stimulated cells observed through light micro- keeping an ORF from ECFP through pIL-1 coding region. The scopy and yellow and cyan filters, respectively. pEYFP construct was produced by ligating a PCR-generated fragment of pIL-1␣ synthesized by using a 3Ј primer lacking the TAG stop codon. This fragment was ligated to the BamHI͞ of 10% pRL Null-Renilla (Promega), 20% reporter vector (when EcoRI sites of the enhanced yellow fluorescent protein (EYFP) used), and 70% expression vector. Stable transfectants were vector, creating an ORF from pIL-1␣ through EYFP. performed similarly, except that 3 days posttransfection the cells were trypsinized and plated in 10-cm-diameter plates in the Fluorescence Microscopy. Cells were observed on cover slips on a presence of 2 ␮g͞ml Blasticidin (Invitrogen) and maintained in DMR fluorescent microscope (Leica, Wetzlar, Germany). CFP the selection medium for at least 21 days. Cloning, when and YFP were observed through XF114 and XF104 filters indicated, was performed by limiting dilution using 25 cells per (Omega Optical, Brattleboro, VT), respectively. plate in a 96-well plate.

GAL4 Constructs. PCR-generated constructs using primers con- Luciferase Assays. Luciferase and Renilla were measured by using taining a BamHI site following a random pentamer (CGCGG) Ј the dual-luciferase assay kit (Promega) according to the manu- to ameliorate restriction digestion in the 5 and a PstI site in the facturer’s instructions. 3Ј were inserted into the respective sites in pG5-LUC (Promega). The primers were designed to amplify the coding regions of the Cell Culture and Cytokine Measurements. Murine endothelial MS1, entire IL-1␣ precursor (pIL-1͞GAL4), the propiece (ppIL-1͞ ͞ ␣ ͞ murine fibroblast NIH 3T3, murine macrophage Raw264.7 cells, GAL4), and the mature form of IL-1 (mIL-1 GAL4), each rhesus COS-7 kidney fibroblast-like cells, and murine embryo with an ORF with the GAL4 DNA binding domain (DBD) fibroblast cells were used in the experiments described. The cells coding region (amino acids 1–147). Sequences were verified by were maintained in DMEM (Cellgro, Waukesha, WI) supple- direct sequencing at the University of Colorado Health Sciences mented with L-glutamine, 10% FCS, 100 units͞ml penicillin, and Center. 0.25 ␮g͞ml streptomycin. Recombinant human IL-1Ra was Expression Vectors for Human IL-1␣, IL-1␤, and IL-18. pIL-1␣, ppIL- supplied by Amgen Biologicals. Cytokine concentrations in 1␣, and mIL-1␣ were constructed by PCR amplification by using supernatants or cell lysates were measured by liquid-phase 5Ј primers containing a random pentamer, an EcoRI site, and a electrochemiluminescence assays by using an ORIGEN Ana- KOZAK sequence upstream of the 5Ј 24 bp of the coding region lyzer (IGEN, Gaithersburg, MD) (16). for each form of IL-1␣, respectively. 3Ј primers contained a Mouse Embryo Fibroblasts. Mouse embryo fibroblasts were pre-

random pentamer, a PstI site, and an added TAG stop codon for IMMUNOLOGY the ppIL-1 construct. The primers were ligated into pcDNA6- pared as previously described (17). Cells from passages 4–10 His-myc-B (Invitrogen) previously linearized by using the same were used in this study. restriction endonucleases without keeping an open frame with the His-myc tag. The sequence for each plasmid was confirmed Statistical Analysis. Cytokine and luciferase were measured in Ϯ by restriction endonuclease digestion and verified by direct triplicate, and the mean SEM of at least three experiments was sequencing. pIL-18 and pIL-1␤ were constructed similarly by used in the statistical analysis. inserting a PCR-generated fragment into pcDNA6-His-Myc Results using primers containing a BamHI and EcoRI in the 5Ј and 3Ј, respectively. Nuclear Translocation of IL-1␣ upon Stimulation. Using immunoflu- orescence and transfections with GFP IL-1␣ fusion proteins in Transfections. Transient transfections were performed by using a variety of cells, we observed that pIL-1␣ is evenly distributed Lipofectamine 2000 (Invitrogen). MS1, COS-7, NIH͞3T3, and within the cytoplasm in resting cells but translocates to the RAW264.7 cells were split the day before transfection at num- nucleus after activation with inflammatory stimuli such as bers sufficient to reach 90% confluence the next day in either 96- lipopolysaccharide, IL-1, or tumor necrosis factor (TNF). Two or 24-well plates. Transfection was performed by using 0.2 ␮gof vectors encoding pIL-1␣ tagged to YFP and CFP on the N and DNA and 0.5 ␮l of lipid per well for 96-well plates and 0.8 ␮gof C terminus, respectively, were constructed (Fig. 1a). Transfec- DNA and 2 ␮l of lipid per well for 24-well plates. DNA consisted tion of both these constructs into NIH͞3T3 cells shows an even

Werman et al. PNAS ͉ February 24, 2004 ͉ vol. 101 ͉ no. 8 ͉ 2435 Downloaded by guest on September 26, 2021 Fig. 3. IL-1␣ activates transcription from a heterologous promoter. MS1 cells were transfected with empty plasmid (mock) or plasmids expressing the GAL4 DBD, either alone or upstream of pIL-1␣, mIL-1␣, or ppIL-1␣, represented by pIL-GAL4, mIL-GAL4, and ppIL-GAL4, respectively. Cells were also cotrans- fected with pG5-LUC containing five copies of the GAL4 enhancer site up- stream of the luciferase gene and with pNULL-Renilla to control for transfec- tion efficiency. Data represent the mean Ϯ SEM of three separate experiments, each performed in triplicate.

calcium ionophore A23187 activates calcium sensitive calpain Fig. 2. The UAS͞GAL4 system. (Top)The two vectors used in this experiment: and results in the release of active IL-1␣ (5, 19). As shown in Fig. IL-1 fused to GAL4-DBD (Top Left) and the GAL4 responsive reporter (Top 4, in the presence of saturating concentrations of IL-1Ra, Right). The yeast GAL4 contains a DBD in the N terminus calcium ionophore-induced macrophage inflammatory protein 2 that lacks significant transcriptional activity. Without a transcription activator ␣ in close proximity to the DBD, the latter is unable to recruit coactivators and (MIP-2) expression was similar for pIL-1 -overexpressing cells stimulate transcription of the reporter upon binding the enhancer site (Mid- and controls, whereas vast amounts of the chemokine were dle). Fusion of an activation domain, in this case forms of IL-1␣,tothe present in the supernatants of cells stimulated without IL-1Ra. GAL4-DBD results in an active transcription factor, capable of recruitment of This indicates that 10 ␮g͞ml recombinant IL-1Ra is sufficient to coactivators and transcription of the luciferase reporter driven by UAS͞GAL4 block downstream effects of IL-1␣ released from cells in our (Bottom). system. To test the possibility that intracellular IL-1␣ activates NF-␬B, MS1 cells were transfected with constructs encoding the various distribution of both fusion proteins in the cytoplasm of resting forms of IL-1␣ together with a NF-␬B reporter vector. In the cells (Fig. 1 b and c) but nuclear localization in lipopolysaccha- presence of saturating concentrations of IL-1Ra, NF-␬B was ride-treated cells (Fig. 1 c–e). In this system, if IL-1␣ were ␣ ␣ ␣ ␣ activated by pIL-1 and ppIL-1 (Fig. 5 Upper). A construct of processed by calpain, ppIL-1 and mIL-1 would be visualized pIL-18 that is structurally related to IL-1␤ but, like IL-1␣,is as yellow and cyan, respectively. Both the yellow and cyan ͞ ␣ ubiquitously expressed and found mainly in the cytosol (20), fluorescent IL-1 fusion proteins are evenly distributed in the exhibited minimal effects on NF-␬B activity. To test whether nucleus, suggesting, in addition, that the translocation is IL-1␣ is necessary for NF-␬B activation, mouse embryo fibro- achieved without processing. blast cells derived from either IL-1␣-deficient mice or wild-type ␣ mice were studied. Cells from IL-1-deficient mice exhibited a pIL-1 Activates Transcription from a Heterologous Promoter. Be- reduced ability to activate NF-␬B compared with cells from cause under in vitro inflammatory conditions pIL-1␣ translo- ␣ wild-type mice (Fig. 5 Lower). cates to the nucleus, it is plausible that pIL-1 exerts its effects AP-1 was activated in a pattern similar to that seen with through a nuclear mechanism. We used the upstream activator NF-␬B, but again intracellular IL-18 had minimal effect (Fig. 6). sequence (UAS) of GAL4͞GAL4 (UAS͞GAL4) system to ␣ ␣ To determine whether the presence of intracellular IL-1 forms assess whether pIL-1 participates in transcription machinery. might sensitize MS1 cells to further stimulation, pooled stable Fig. 2 depicts the construction of the test system in which the ␣ transfected MS1 cells transfected with the various IL-1 forms, various forms of IL-1 are inserted into plasmids containing the vector, or pIL-1␤ expression vectors, were stimulated with IFN-␥ DBD of GAL4 creating GAL4-DBD-IL-1 chimeras. As shown in Fig. 3, pIL-1␣ and ppIL-1␣ were able to activate transcription of the heterologous promoter by Ͼ90- and Ͼ50-fold, respec- tively, whereas mIL-1␣ was without activity.

Functional Role for pIL-1␣. If pIL-1␣ acts through transcriptional activation and is located in the nucleus upon inflammatory stimulus, part of the role of IL-1␣ in inflammation or in host defense might involve intracellular mechanisms without engag- ing surface membrane IL-1R. It has been shown that both extracellular pIL-1␣ and mIL-1␣ as well as membrane IL-1␣ are biologically active by means of the engagement of surface IL-1R (3, 4, 18). To demonstrate a functional role of intracellular IL-1␣ ␣ in the absence of triggering the IL-1 surface receptor (because Fig. 4. Recombinant IL-1Ra abrogates activity of endogenous IL-1 released ␣ to supernatants. Raw264.7 cells were stably transfected with either pcDNA6͞ of leakage of IL-1 from dying cells), cells expressing forms of His-myc (vector) or pIL-1␣ expression vector. Cells were plated in the presence IL-1␣ intracellularly were cultured in the presence of saturating ␮ ͞ ␮ ͞ of 10 g ml IL-1Ra as indicated. On the next day, cells were washed with PBS concentrations of the IL-1Ra at 10 g ml. The mouse macro- and replenished with medium containing nothing (control), A23187, IL-1Ra, phage cell line Raw264.7 was stably transfected with either or both A23187 and IL-1Ra. Twenty-four hours later, medium was removed pIL-1␣ or empty vector. Treatment of these cells with the and concentrations of MIP-2 were measured by ECL.

2436 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0308705101 Werman et al. Downloaded by guest on September 26, 2021 Fig. 5. Expression of intracellular IL-1␣ activates NF-␬B in a surface IL-1R- Fig. 7. Expression of intracellular IL-1␣ augments MS1 cells responses to independent manner. MS1 cells were transfected with the empty vector IFN-␥ and TNF-␣. Stable, pooled MS1 cells transfected with either vector or (vector) or with expression vectors encoding pIL-1␣, mIL-1␣, ppIL-1␣, or pIL-18. expression vectors for pIL-1␣ or pIL-1␤ were plated in the presence of satu- In each case, cells were cotransfected with a vector containing the NF-␬B rating concentrations of IL-1Ra (10 ␮g͞ml). Twenty-four hours later, cells were enhancer upstream of the luciferase gene and pNULL-Renilla. Transfection washed and treated with IL-1Ra together with either TNF-␣ (Upper) or IFN-␥ and culture were performed in the presence of saturating concentrations of (Lower). After 24 h, MIP-2 was measured. Data represent the mean Ϯ SEM of IL-1Ra (10 ␮g͞ml) for IL-1 constructs or IL-18BP (1 ␮g͞ml) for the pIL-18 three separate experiments. construct. MS-1 WT, cells transfected with the pNF␬B-LUC and pNULL-Renilla only (Upper). Mouse embryo fibroblasts from BALB͞c and IL-1␣-deficient (IL-1␣Ϫ͞Ϫ) mice were transfected with pNF␬B-LUC and pNULL-Renilla in the was sufficient to induce synthesis of mouse IL-1␣ in an intracrine presence of IL-1Ra as indicated. On the next day, medium was removed, cells manner. As can be seen in Fig. 9, this is indeed the case. were washed, and fresh medium with or without IL-1Ra was added. Cells were activated with 10 ng͞ml TNF-␣. Twenty-four hours later, cells were lysed and Discussion Ϯ luciferase was measured (Lower). Data represent the mean SEM of three Our findings indicate that heterologous extracellular stimuli, separate experiments performed in triplicate. including cytokines and Toll-like receptor ligands, lead to the

or TNF-␣ and then secondary cytokines were measured. As shown in Fig. 7, expression of pIL-1␣ led to sensitization of MS1 cells to stimulation by subpicomolar concentrations of IFN-␥ (Upper) or TNF-␣ (Lower) by releasing increased levels of MIP-2, an NF-␬B-regulated chemokine. We further tested the NF-␬B-regulated cytokines, IL-6 in NIH͞3T3 stable transfec- tants (Fig. 8 Upper) and IL-8 in COS-7 transiently transfected cells (Fig. 8 Lower). Even under saturating concentrations of IL-1Ra in the cultures, cells expressing pIL-1␣ produced these NF-␬B-regulated cytokines without additional stimuli. IL-1 is known to up-regulate its own synthesis in an autocrine, receptor-mediated positive-feedback loop (21). Because we have shown above that ppIL-1␣ activates NF-␬B and NF-␬B- regulated genes, we assessed whether transfection of ppIL-1␣ IMMUNOLOGY

Fig. 8. Stable expression of intracellular pIL-1␣ provides a complete signal for the synthesis of cytokines independent of the IL-1 surface receptor in different cell lines. NIH͞3T3 cells were stably transfected with the various IL-1␣ constructs or the empty vectors as described in Fig. 2. Clones expressing similar amounts of cytokines were replated in the presence of saturating concentra- tions of IL-1Ra. After 24 h, the cells were washed and treated again with IL-1Ra for 24 h. On the subsequent day, supernatants were removed and measured Fig. 6. Expression of intracellular IL-1␣ activates AP-1 in a surface IL-1R- by ELISA for murine IL-6 (Upper). COS-7 cells were transiently transfected in independent manner. Experiments were performed similarly to those in Fig. the presence of IL-1Ra with the various IL-1 constructs. After 24 h, the cells 5 Upper, except using a reporter vector containing the AP-1 enhancer up- were washed and replenished with fresh medium containing IL-1Ra. Super- stream of the luciferase gene. Data represent the mean Ϯ SEM of three natants were removed on the subsequent day and measured by ECL for human separate experiments performed in triplicate. IL-8, which crossreacts with rhesus IL-8 (Lower).

Werman et al. PNAS ͉ February 24, 2004 ͉ vol. 101 ͉ no. 8 ͉ 2437 Downloaded by guest on September 26, 2021 mIL-1␣, bind and activate IL-1R (5). Leakage of pIL-1␣ into the supernatants from transfected MS1, NIH͞3T3, or COS-7 cells was below the detection level in our assays. Moreover, saturating concentrations of IL-1Ra were used throughout to further ensure that the intracellular events were independent of activa- tion of the extracellular receptor. The concentrations of IL-1Ra were sufficient to block exogenous as well endogenously pro- duced pIL-1␣ in overexpressing macrophages activated by the calcium ionophore A23187 (5). We conclude that these studies demonstrate the role of intracellular IL-1␣ in the absence of receptor triggering. ␣ Fig. 9. Expression of ppIL-1 is a sufficient signal for up-regulation of In our reporter experiments, intracellular overproduction of endogenous IL-1␣. MS1 cells stably transfected with either the empty vector or ␣ ␣ ␬ ␣ pIL-1 or ppIL-1 was sufficient to activate NF- B and AP-1 in an expression vector for human ppIL-1 were plated in the presence of IL-1Ra. ␣ ␬ The next day, cells were washed and medium containing IL-1Ra was added. nonstimulated MS1 cells. That IL-1 is part of the NF- B and Twenty-four hours later, cells were lysed and murine IL-1␣ was measured AP-1 transcription complex remains to be demonstrated by by ECL. coimmunoprecipitation or two-hybrid experiments. However, IL-1␣ might act upstream, leading to the activation of NF-␬B and AP-1. The fact that intracellular IL-1␣ activates these transcrip- translocation of intracellular pIL-1␣ to the nucleus. There, tion factors by an IL-1R-independent mechanism is a previously pIL-1␣, in an N-terminal propiece-dependent manner, results in uncharacterized role for cytokines in general and this cytokine, activating transcriptional machinery together with synthesis of in particular, in inflammation. Interestingly, pIL-18, which, like proinflammatory cytokines; each process is independent of IL-1␣, is constitutively present in cells from healthy subjects (20) membrane IL-1R activation. To date, our knowledge of IL-1␣ and remains in the cytosol of many epithelial cells (28), did biology is based mostly on studies using the 17-kDa recombinant not activate NF-␬B and AP-1, using the same experimental mature form, neutralizing antibodies, or soluble receptors in procedure. which the activity of IL-1␣ follows membrane IL-1R ligation (1). Only pIL-1␣ and ppIL-1␣ stable transfectants of MS1 cells However, in vivo, IL-1␣ is not readily released from cells even responded to picomolar concentrations of TNF-␣ and IFN-␥ by upon activation, and, unlike IL-1␤, IL-1␣ is not detected in the inducing MIP-2, a known NF-␬B-regulated gene, in the absence serum of various infectious and inflammatory diseases, with the of surface membrane IL-1R engagement. We have observed the exception of severe pathological states in which cell death likely intracellular activation of proinflammatory cytokines by the results in release of the cytokine (22). In tissues from healthy nuclear localization sequence-containing pIL-1␣ in a variety of subjects, pIL-1␣ is stored intracellularly in a variety of cells cells, including the two examples in this report of IL-6 in (mostly epithelial cells), where it is not readily released in the NIH͞3T3 cells and IL-8 in COS-7 cells. absence of cell membrane breakdown. As an intracellular mol- Using the yeast two-hybrid system, other investigators have ecule, IL-1␣ has been also shown to be inserted in the membrane reported that pIL-1␣ binds to a nuclear protein called necdin, and act in a juxtacrine manner (3, 4, 23). However, there is which possesses growth suppressor activity (29). In mammalian growing evidence for intracellular functions of IL-1␣, including cells pIL-1␣ can associate with necdin to reverse its negative promotion of senesence (12), cell growth and differentiation (7) effect on cell growth and production of procollagen (29). Also and regulation of gene expression (24). using the two-hybrid method, ppIL-1␣ has been shown to When NIH͞3T3 cells were transfected with pIL-1␣ fused to associate with HAX-1 (30). HAX-1 contains amino acid motifs fluorescent proteins on either the N terminus or the C terminus, for binding to several intracellular proteins including polycys- localization of IL-1␣ was found to depend on cell activation by tin-2 (a protein linked to polycystic kidney disease), cortactin, inflammatory stimuli. In resting cells, pIL-1␣ was found in the and Epstein–Barr virus nuclear antigen leader protein (30). We ␣ cytoplasm, whereas, upon activation, pIL-1␣ translocated to the also reported HAX-1 association with the pIL-1 in the yeast § nucleus. To determine whether IL-1␣ affected transcriptional two-hybrid system. However, the significance of HAX-1 or machinery, the UAS͞GAL4 system was used (25). The chimeras necdin to the transcriptional activation of proinflammatory ␣ ␣ cytokines by the ppIL-1␣ remains uncertain. Similar to the of the GAL4 DBD fused to either pIL-1 or ppIL-1 led to ␣ activation of transcription from a GAL4-responsive promoter, present studies, others have reported that ppIL-1 activates 90- and 50-fold, respectively. These results must be interpreted transcription in the GAL4 system and is increased further by carefully. Ruden et al. (26) have shown that Ϸ1% of sonicated proteins with histone acetyltransferase activity (31). Escherichia coli DNA sequences are able to activate transcription There are several examples of proteins and that act from a LexA- and a GAL4-responsive promoter in a similar both through a specific extracellular receptor and also in intra- assay of UAS͞GAL4 in yeast. In contrast, the level of activity of cellular compartments; the term ‘‘intracrine’’ was suggested for pIL-1␣͞GAL4 (90-fold) seen in our experiments was similar to these proteins. These include insulin, fibroblast growth factors A that of GAL4 fused to herpesvirus VP16 (data not shown), a and B, platelet-derived growth factor, nerve growth factor, epidermal growth factor, growth , prolactin, para- construct considered to be of near maximal potency in this ␥ system (27). Furthermore, only the chimeras encompassing thyroid hormone-related protein, angiogenin, IFN- , and others (reviewed in ref. 32). pIL-1␣ and ppIL-1␣, each of which incorporates a nuclear Of relevance to the present study is the example of a well localization sequence, were active in this system. mIL-1␣, which known transcription factor that was rediscovered as a cytokine. binds to the extracellular IL-1R, was inactive when inserted in HMGB1 is a ubiquitously expressed, highly conserved protein, the GAL4 vector. This difference indicates a distinction between initially characterized as a DNA binding protein, augmenting the functions of mIL-1␣ compared with intracellular pIL-1␣, gene transcription and stabilizing nucleosome formation (15, suggesting evolutionary diversion. 33). However, in searching for late mediators in macrophages To establish the intracellular role of IL-1␣, it was essential to show that the activity of the intracellular IL-1␣ forms was present in the absence of engagement of its membrane receptor, §Werman, A., White, R. M., Dobkin, M., Voronov, E., Werman-Venkert, R., Dinarello, C. A. because both pIL-1␣ and the calpain cleavage product of pIL-1␣, & Apte, R. N. (2001) Scand. J. Immunol. 54, Suppl. 1, 63 (abstr.).

2438 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0308705101 Werman et al. Downloaded by guest on September 26, 2021 exposed to inflammatory stimuli for prolonged periods Ͼ8h, interactions that subsequently evolved ligand receptor binding large amounts of HMGB1 were found in the supernatants (15). properties. Despite evolutionary development, the cytokines HMGB1 is a late and essential mediator of endotoxemia in mice, retained part of their original intracellular roles. In this report we because blockade of HMGB1 by neutralizing antibodies pro- demonstrate that pIL-1␣, found in the intracellular compart- tected mice, whereas injection of recombinant HMGB1 dimin- ment of many cells, has a proinflammatory role during tran- ished survival from lethal endotoxemia. HMGB1 was shown to scription in an IL-1R-independent manner. Modalities of treat- activate inflammatory pathways when released from necrotically ment based on reduction of nuclear activity of pIL-1␣ might lysed cells but not apoptotic cells (34). In many ways, HMGB1 prove beneficial in diseases hallmarked by elevated intracellular is the chronological ‘‘other side’’ of IL-1␣, which was initially pIL-1␣. studied for its activity in triggering its extracellular receptor. In vivo, both molecules appear to act intracellularly by transcrip- We thank Soo-Hyun Kim, Tania Azam, Leonid Reznikov, Xiaoping tional modulation. However, they can be released under certain Song, and Philip Bufler for help and discussions during these studies. conditions and exert receptor-specific events. This work was supported by National Institutes of Health Grants AI-15614 and HL-68743 (to C.A.D.), the Colorado Cancer Center These ‘‘dual-function’’ cytokines likely originated in unicellu- (C.A.D.), the United States–Israel Binational Foundation (R.N.A. and lar organisms as exclusively intracellular mediators, responding C.A.D.), the Israel Science Foundation of the Israel Academy of Sciences to stress by regulating genes. It is possible that these molecules (R.N.A.), and the Israel Ministry of Health Chief Scientist Office developed from DNA binders through later protein–protein (R.N.A.).

1. Dinarello, C. A. (1996) Blood 87, 2095–2147. 17. Huleihel, M., Douvdevani, A., Segal, S. & Apte, R. N. (1990) Eur. J. Immunol. 2. Arend, W. P., Malyak, M., Guthridge, C. J. & Gabay, C. (1998) Annu. Rev. 20, 731–738. Immunol. 16, 27–55. 18. Brody, D. T. & Durum, S. K. (1989) J. Immunol. 143, 1183–1187. 3. Kurt-Jones, E. A., Beller, D. I., Mizel, S. B. & Unanue, E. R. (1985) Proc. Natl. 19. Miller, A. C., Schattenberg, D. G., Malkinson, A. M. & Ross, D. (1994) Toxicol. Acad. Sci. USA 82, 1204–1208. Lett. 74, 177–184. 4. Kaplanski, G., Farnarier, C., Kaplanski, S., Porat, R., Shapiro, L., Bongrand, 20. Puren, A. J., Fantuzzi, G. & Dinarello, C. A. (1999) Proc. Natl. Acad. Sci. USA P. & Dinarello, C. A. (1994) Blood 84, 4242–4248. 96, 2256–2261. 5. Kobayashi, Y., Yamamoto, K., Saido, T., Kawasaki, H., Oppenheim, J. J. & 21. Dinarello, C. A., Ikejima, T., Warner, S. J., Orencole, S. F., Lonnemann, G., Matsushima, K. (1990) Proc. Natl. Acad. Sci. USA 87, 5548–5552. Cannon, J. G. & Libby, P. (1987) J. Immunol. 139, 1902–1910. 6. Wessendorf, J. H. M., Garfinkel, S., Zhan, X., Brown, S. & Maciag, T. (1993) 22. Watanabe, N. & Kobayashi, Y. (1994) Cytokine 6, 597–601. J. Biol. Chem. 268, 22100–22104. 23. Resch, K., Martin, M., Lovett, D. H., Kyas, U. & Gemsa, D. (1986) Immuno- 7. Maier, J. A. M., Statuto, M. & Ragnotti, G. (1994) Mol. Cell. Biol. 14, 1845–1851. biology 172, 336–345. 8. Horai, R., Asano, M., Sudo, K., Kanuka, H., Suzuki, M., Nishihara, M., 24. Kawaguchi, Y., Hara, M. & Wright, T. M. (1999) J. Clin. Invest. 103, 1253–1260. Takahashi, M. & Iwakura, Y. (1998) J. Exp. Med. 187, 1463–1475. 25. Sadowski, I. & Ptashne, M. (1989) Nucleic Acids Res. 17, 7539. 9. Nakae, S., Naruse-Nakajima, C., Sudo, K., Horai, R., Asano, M. & Iwakura, Y. 26. Ruden, D. M., Ma, J., Li, Y., Wood, K. & Ptashne, M. (1991) Nature 350, 250–252. (2001) Int. Immunol. 13, 1471–1478. 27. Sadowski, I., Ma, J., Triezenberg, S. & Ptashne, M. (1988) Nature 335, 563–564. 10. Stevenson, F. T., Turck, J., Locksley, R. M. & Lovett, D. H. (1997) Proc. Natl. 28. Pizarro, T. T., Michie, M. H., Bentz, M., Woraratanadharm, J., Smith, M. F., Acad. Sci. USA 94, 508–513. Jr., Foley, E., Moskaluk, C. A., Bickston, S. J. & Cominelli, F. (1999) 11. Pollock, A. S., Turck, J. & Lovett, D. H. (2003) FASEB J. 17, 203–213. J. Immunol. 162, 6829–6835. 12. Maier, J. A. M., Voulalas, P., Roeder, D. & Maciag, T. (1990) Science 249, 29. Hu, B., Wang, S., Zhang, Y., Feghali, C. A., Dingman, J. R. & Wright, T. M. 1570–1574. (2003) Proc. Natl. Acad. Sci. USA 100, 10008–10013. 13. Beck, G. & Habicht, G. S. (1991) Mol. Immunol. 28, 577–584. 30. Yin, H., Morioka, H., Towle, C. A., Vidal, M., Watanabe, T. & Weissbach, L. 14. Willenborg, D. O. & Prendergast, R. A. (1974) J. Exp. Med. 139, 820–833. (2001) Cytokine 15, 122–137. 15. Wang, H., Bloom, O., Zhang, M., Vishnubhakat, J. M., Ombrellino, M., Che, 31. Buryskova, M., Pospisek, M., Grothey, A., Simmet, T. & Burysek, L. (2003) J., Frazier, A., Yang, H., Ivanova, S., Borovikova, L., et al. (1999) Science 285, J. Biol. Chem., 10.1074͞jbc.M306342200v1. 248–251. 32. Re, R. N. (2000) Hypertension 35, 1189–1190. 16. Puren, A. J., Fantuzzi, G., Gu, Y., Su, M. S.-S. & Dinarello, C. A. (1998) J. Clin. 33. Yang, H., Wang, H. & Tracey, K. J. (2001) Shock 15, 247–253. Invest. 101, 711–724. 34. Scaffidi, P., Misteli, T. & Bianchi, M. E. (2002) Nature 418, 191–195. IMMUNOLOGY

Werman et al. PNAS ͉ February 24, 2004 ͉ vol. 101 ͉ no. 8 ͉ 2439 Downloaded by guest on September 26, 2021