Autocrine Interleukin-1β Production in Leukemia: Evidence for the Involvement of Mutated RAS6¹

There is an increasing body of evidence that indicates that IL-1β³ can act as an autocrine and/or paracrine growth factor for many of the major forms of leukemia (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14). Under normal physiological conditions, cells of the hematopoietic system produce IL-1β only when a stimulus is present (5). Leukemic cells, however, can produce this cytokine without an exogenous source of activation (1, 3, 4, 7, 8, 9, 10, 12, 14). Indeed, constitutive expression of IL-1β has been detected in chronic myelogenous leukemia (8, 14, 15), JCML (1, 12, 16), acute myelogenous leukemia (4, 7, 11, 17), and ALL (3, 18). Once released, IL-1β can operate in an autocrine manner and induce the proliferation of leukemic blasts (1, 3, 4, 8, 11). Due to the prominent role that IL-1β appears to play in myeloid leukemias, antagonist strategies using IL-1 receptor antagonist, soluble IL-1 receptors, and IL-1-converting enzyme inhibitors have been suggested for therapeutic purposes (6, 10).

Over the past decade, the signaling pathways and promoter elements regulating inducible Il-1β gene expression have been extensively studied (19, 20, 21, 22, 23). However, the molecular mechanisms by which constitutive expression occurs in leukemic cells remain unexplained. Interestingly, there are several lines of evidence that suggest a possible association between Ras-related pathways and the induction of the IL-1β gene. First, it has been demonstrated that stable transfection of an activated RAS oncogene into fibroblasts and mesothelioma cells results in IL-1β expression (24). Also, macrophages expressing v-Raf acquire growth factor-independent properties as well as the ability to constitutively produce IL-1β (25, 26). The latter observation is significant because Raf is a downstream mediator of Ras signal transduction (27). Furthermore, it has been shown that growth factor-independent leukemic cells have elevated levels of phosphorylated Raf and increased Raf kinase activity (28).

An additional link, albeit circumstantial, between Ras and IL-1β is provided by the observation that RAS mutations are frequently found in leukemias in which aberrant expression of IL-1β has been demonstrated. Indeed, mutations of RAS have been detected in over 30% of patients with chronic myelomonocytic leukemia and in 20%–30% of patients with acute myelogenous leukemia, JCML, and ALL (17, 29, 30, 31). In addition to mutation, Ras is activated in many types of leukemia by alternative mechanisms (29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41). For instance, in chronic myelogenous leukemia, in which RAS mutations are rare (29, 32, 33) and IL-1β deregulation is common (14, 15), it has been suggested that activation of Ras-mediated pathways occurs via the interaction between Grb2 and the Bcr-Abl oncoprotein (34). In JCML, many of the patients who do not have Ras mutations have a loss of neurofibromin (NF1) function (35, 36). Because NF1 contains a GTPase-activating protein domain that is thought to down-regulate Ras (37), loss of NF1 may account for the increased levels of Ras-GTP found in JCML bone marrows (38). Taken together, activation of Ras, either directly through mutation or indirectly, may be a pathogenic factor in many of the hematopoietic malignancies in which IL-1β is expressed. However, despite the wealth of data available on the regulation of IL-1β in normal cells (19, 20, 21, 22, 23, 42, 43, 44, 45, 46, 47) and on the role of Ras in signaling pathways (48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60), a direct link between activation of Ras and IL-1β deregulation in neoplasia has not been explored.

To determine whether activation of RAS by mutation mediates IL-1β production in leukemia, we analyzed two leukemic cell lines. B1 and W1 cells were derived from a patient in first relapse with ALL (3) and a 10-month-old patient diagnosed with JCML (1), respectively. Both cell lines produce high levels of IL-1β, use IL-1β as an autocrine growth factor (1, 3), and have either a K-RAS (B1 cells) or N-RAS (W1 cells) mutation. We demonstrate that interference with Ras pathways by FTase inhibitors (either monoterpenes or peptidomimetics) or antisense targeted to Ras suppresses the production of IL-1β in these cells. Furthermore, decreased IL-1β levels are associated with a decrease in binding at the NF-IL6/CREB site (a major activation site in the IL-1β promoter, Refs. 19, 20, 23, 42). Our results suggest that in these leukemic cell lines, mutated RAS mediates IL-1β production via signaling pathways that lie upstream of the Il-1β promoter.

B1 is a human ALL cell line (3), and W1 is a human JCML cell line (1). THP-1 is a human monocytic leukemia cell line (61), HL60 is a human promyelocytic leukemia cell line, and CCD969SK is a normal human fibroblast cell line (all from American Type Culture Collection, Rockville, MD). ALL-1 is a human Philadelphia chromosome-positive ALL cell line (kindly donated by G. Rovera, Wistar Institute, Philadelphia, PA). B1 and W1 cells were maintained in α-MEM (Life Technologies, Inc., Grand Island, NY); THP-1, ALL-1, and HL60 cells were maintained in RPMI 1640 (BioWhittaker, Walkerville, MD); and CCD969SK cells were maintained in Eagle’s MEM with Earle’s salt containing MEM nonessential amino acids and sodium pyruvate (Life Technologies, Inc.). Media for all cell lines contained 10% FCS (Sigma, St. Louis, MO) and a 1× antibiotic:antimycotic solution (penicillin, streptomycin, and amphotericin B; Life Technologies, Inc.).

The IL-1β probe is a 0.6-kb cDNA (clone designation, YEpsec1-hIlβ; Ref. 62); the β-actin probe is a 1.1-kb cDNA (clone designation, HHCI89; Ref. 63; The American Type Culture Collection); and the C/EBP (NF-IL6) consensus and mutant oligonucleotide probes contain either the TTGCGCAA consensus binding motif or an 8-bp substitution (Santa Cruz Biotechnology, Santa Cruz, CA). The Oct-1 oligonucleotide probe contains the ATGCAAAT consensus sequence, a regulatory element important for the transcription of a number of housekeeping genes (Santa Cruz Biotechnology).

The anti-K-Ras antibody is a mouse monoclonal antibody (Santa Cruz Biotechnology); the anti-G protein α-subunit internal (40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53) antibody is a rabbit polyclonal antibody (Calbiochem, San Diego, CA). Isotypic antibody (anti-laminin) was obtained from DAKO Corp. (Carpinteria, CA), and normal rabbit serum was purchased from Sigma. Anti-CREB antibodies were kindly donated by Dr. Marc Montminey (Salk Institute, San Diego, CA), and the rabbit polyclonal anti-C/EBPβ (C19; NF-IL6) antibodies were obtained from Santa Cruz Biotechnology.

The monoterpenes limonene (Aldrich Chemical Co., Milwaukee, WI), perillic acid (Arcos), and perillyl alcohol (Aldrich Chemical Co.) and the peptidomimetic B581 (LC Laboratories, Woburn, MA) were used. B581 was dissolved in 4% DMSO, α-MEM, and 10% FCS to a final stock concentration of 5 mm; the monoterpenes were dissolved in α-MEM with 20% FCS to a final stock concentration of 5 mm.

Isolation of mRNA from cell lines was performed using the RiboSep mRNA Isolation Kit (Becton Dickinson, Bedford, MA). The mRNA was quantitated and run on a 5% formaldehyde gel containing 1% agarose. Transfer of RNA to nitrocellulose (Nitropure; Micron Separations Inc., Westboro, MA) was accomplished using the PosiBlot Pressure Blotter System (Stratagene, La Jolla, CA). Probe labeling was performed using the Multiprime DNA Labeling System (Amersham, Arlington Heights, IL) with radioactive [α-³²P]CTP (DuPont New England Nuclear, Boston, MA). Probes were purified from free nucleotides using a G50 column (5′3′ Company, Boulder, CO), and hybridization was carried out as described previously (64). After washing, blots were rehybridized with a β-actin probe to assess equal loading of lanes. Cycloheximide and dexamethasone were purchased from Sigma.

Isolation of genomic DNA and Southern blotting were performed as described previously (65, 66). Twelve μg of DNA were digested overnight at 37°C with 20 units of BstXI, BclI, HindIII, and SspI (Boehringer Mannheim, Indianapolis, IN). Digested DNA was run on a 0.7% agarose gel. The gel was denatured, and DNA was transferred to nylon (Nitran Plus; Schleicher & Schuell, Keene, NH) using the PosiBlot Pressure Blotter System. Probe labeling and hybridization were performed as described for Northern blotting. Washed membranes were exposed to Kodak X-OMAT AR or Kodak X-OMAT RP film (Eastman Kodak, Rochester, NY) at −70°C for 16 h using an intensifying screen.

Isolation of nuclear extracts was performed as described by Newell et al. (67). Extracts were quantitated using the Bio-Rad Protein Assay Kit (Bio-Rad Laboratories, Hercules, CA). Single-stranded oligonucleotide probes for gel shift analysis were made by Life Technologies, Inc. The NF-IL6, NFB1, NF-IL6/CREB, and NFκB probe sequences corresponded to those that have been published previously (21, 22, 23). Probes were designed with 5′ overhangs, and labeling was performed using Klenow enzyme (Boehringer Mannheim) and [α-³²P]CTP (10 uCi/μl; DuPont New England Nuclear). Purification from free nucleotides was accomplished using a G50 column. Nuclear extracts were analyzed by gel shift assay as described previously (67). Phorbol 12-myristate 13-acetate and LPS (LPS-Escherichia coli) were purchased from Sigma.

³²P-end-labeled primers were used to amplify codons 12, 13, and 61 of the H-, K-, and N-RAS genes using PCR. PCR products were denatured and used to hybridize membrane-fixed synthetic probes (20 nucleotides in length) specific for mutations in RAS codons 12, 13, and 61.

Cells (1 × 10⁷) were washed two times in ice-cold PBS and then incubated with 50 μl of NP40 lysis buffer [20 mm HEPES (pH 7.4), 250 mm NaCl, and 2 μg/ml leupeptin; Boehringer Mannheim], 2 μg/ml aprotinin (Sigma), and 1 mm phenylmethylsulfonyl fluoride (Boehringer Mannheim) for 30 min at 4°C, with vortexing every 10 min. Samples were centrifuged, the supernatants were removed, and aliquots were used to measure protein levels using the Bio-Rad DC protein Assay Kit (Bio-Rad Laboratories). An equal volume of SDS sample buffer [100 mm Tris (pH 6.8), 4% SDS, 2.5 m β-mercaptoethanol, and 10% glycerol] was added to the remaining supernatants, and the samples were boiled for 5 min and stored at −70°C until use. Lysates were run on 15% SDS-polyacrylamide gels and transferred to nitrocellulose membranes (Nitropure; Micron Separations, Inc.) using the Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad Laboratories). Membranes were blocked overnight with 10% nonfat milk in PBS with 0.5% Tween 20 and then exposed to primary antibody (anti-K-Ras or anti-G protein) at 0.5 μg/ml for 1 h at room temperature. Secondary antibodies containing the horseradish peroxidase detection system for chemiluminescence were used as recommended by the manufacturer (Amersham Life Science, Buckinghamshire, United Kingdom).

Membranes exposed to anti-K-Ras and anti-G protein were stripped and then exposed to isotypic antibody (anti-laminin) or normal rabbit serum, respectively. Recombinant K-Ras (Oncogene Science, Uniondale, NY) and recombinant G protein from bovine brain (Calbiochem) were used as positive controls.

A retrovirus system has been described that specifically inhibits the production of K-ras mRNA and not that of H-ras or N-ras (68). The antisense RNA contains the second and third exons of the K-ras gene, along with adjoining intron sequences. This 2-kb antisense molecule has been molecularly cloned into the E1 region of an adenovirus (Ad5) vector in both orientations (sense and antisense; Ref. 56). In both vectors, the RNA is driven by a cytomegalovirus promoter and has a SV40 polyadenylation sequence at its 3′ end. This vector has been shown to inhibit the proliferation of K-RAS-transformed lung carcinoma cells in vitro (56). For our experiments, B1 cells were plated the day before infection at 4 × 10⁵ cells/ml. On the day of infection, 1 × 10⁷ cells/sample were washed twice in PBS and then resuspended in 1 ml of PBS for infection. Cells were infected with a multiplicity of infection of 1000 viral particles/cell for 1 h at 37°C. After infection, 9 ml of α-MEM/10% FCS were added to each sample, and the cells were plated at 1 × 10⁶ cells/ml overnight. The next morning, cells were washed in fresh media and replated at a density of 5 × 10⁵ cells/ml. Cells were counted daily, and viability was assessed by trypan blue exclusion. Cell lysates were harvested on day 3 after replating for Western blotting of Ras and to measure IL-1β levels, cell viability, and total protein.

To determine levels of viral transduction and expression, B1 cells were infected with an adenovirus expressing the β-galactosidase gene. Infections were carried out the same way as described for the antisense experiments using 1000 virus particles /cell. After incubation, B1 cells were washed once in PBS and then stained with C₁₂ fluorescein digalactoside as recommended by the manufacturer (Molecular Probes, Eugene, OR). Cells were analyzed by FACScan (Becton Dickinson) with a FITC filter set (488 excitation, 530 emission) using the Lysis II software.

Leukemic cell lysates were harvested by three freeze/thaw cycles in PBS diluted 1:3. Lysates were loaded on an IL-1β-specific ELISA plate (R&D Systems, Minneapolis, MN) by equal protein concentrations or equal cell numbers, as indicated in the figure legends. The lower limit of sensitivity for this assay is 0.3 pg/ml.

Northern blotting and hybridization with a 0.6-kb IL-1β cDNA probe revealed that the B1 and W1 cells expressed a 1.6-kb IL-1β mRNA transcript (data not shown). The level of message was 5–10-fold greater in W1 cells as compared to B1 cells. Consistent with previously published results (5), no IL-1β message was seen in normal leukocytes or in purified populations of unstimulated lymphocytes or myeloid cells (data not shown).

These results show that B1 and W1 leukemic cells constitutively express IL-1β, whereas normal hematopoietic cells do not.

B1 and W1 cell genomic DNA was digested with BstXI, BclI, HindIII, and SspI. Southern blotting and hybridization with the 0.6-kb IL-1β cDNA probe revealed germ-line DNA bands of normal intensity as compared to control CCD969SK cells (normal human fibroblasts; data not shown). [After digestion with the above restriction enzymes, this IL-1β probe detects fragments spanning from 3 kb upstream of the first coding exon to 1 kb downstream of the last exon (43)]. The promoter binding elements that have been described for the IL-1β gene lie within the 3-kb-upstream sequences (20, 21, 22, 23).

These results suggest that constitutive expression of IL-1β in B1 and W1 cells is not due to amplification or rearrangement within the parts of the IL-1β gene studied above.

To analyze binding of the transcription factor sites in the IL-1β promoter (20, 21, 22, 23), gel shift analyses were performed. The sites studied included the NF-IL6/CREB site (−2764), the upstream NF-IL6 site (−2866 to −2896), and the NFB1 site (−2846 to −2866) as well as the downstream NFκB site (−300). These promoter elements are bound in cell lines such as THP-1 or U937 after exposure to exogenous inducers of IL-1β such as LPS, tumor necrosis factor, or phorbol 12-myristate 13-acetate (20, 21, 22, 23, 42). The NF-IL6/CREB site is generally considered to be the most crucial because mutation of this site decreases responsiveness to multiple stimuli (20).

Fig. 1,A is a gel shift assay using the region I probe that contains the NF-IL6/CREB binding site (23). The unstimulated leukemic cell lines B1 and W1 have a banding pattern identical to that of LPS-stimulated THP-1 cells (Fig. 1,A, Lanes 1, 4, and 7), and the use of a cold region I competitor suggests that the three major bands formed are specific (Fig. 1,A, Lanes 3, 6, and 9). The two slower-moving bands migrate to the same location as the slow and fast complexes described previously in THP-1 cells (Ref. 23; Fig. 1,A, Lanes 1, 4, and 7). Consistent with the finding that these complexes are comprised of NF-IL6/CREB and NF-IL6/NF-X (19, 23), respectively, a supershift of the slow complex was seen in THP-1, B1, and W1 cells using anti-CREB antibodies (Fig. 1,A, Lanes 2, 5, and 8), and cold NF-IL6 site competitors reduced the binding of both complexes (Fig. 1,B, Lane 3). A NF-IL6 mutant probe, however, had no effect on the binding of these complexes (Fig. 1,B, Lane 4). In addition, a NF-IL6 antibody produced a supershifted band (Fig. 1,B, Lane 2). A faster-migrating band (the fastest band not previously described) was also detected in all cell lines (Fig. 1,A, Lanes 1, 4, and 7). Because this band is present in unstimulated THP-1 and HL60 cells (data not shown) and disappears in the presence of anti-CREB antibodies (Fig. 1 A, Lanes 2, 5, and 8), it may be comprised of inactive CREB homodimers. Analysis of the NFκB, NF-IL6, and NFB1 elements in the IL-1β promoter also demonstrated constitutive binding in B1 and W1 cells (data not shown).

The combined gel shift data therefore indicated that multiple transcriptional activation sites were constitutively bound in the B1 and W1 cell lines. These observations suggest that an upstream factor is responsible for IL-1β induction.

In the presence of cycloheximide, LPS stimulation of IL-1β mRNA is superinduced (19, 44, 45, 46), suggesting the presence of protein synthesis-independent transcription factor(s) and a protein synthesis-dependent suppressor. Similarly, in our experiments, IL-1β transcripts in cycloheximide-treated B1 and W1 cells were 2-fold those of untreated cells (Fig. 2 A).

Therefore, it is conceivable that IL-1β-regulating factors similar to those seen after LPS induction are operative in our cell lines. Although the LPS pathway for IL-1β induction is incompletely defined, it has been demonstrated in macrophages that LPS can stimulate the activity of Raf (26) and MAP kinase (47), molecules that are known to be the downstream effectors of Ras (27).

Dexamethasone is known to inhibit IL-1β mRNA production at multiple levels. One mechanism of inhibition is thought to involve NF-IL6/CREB transcription factors because after LPS stimulation, dexamethasone can decrease the activity of a region I-containing CAT construct in THP-1 cells (19). In our cell lines, a 4-fold and 10-fold decrease in IL-1β mRNA (equalized for loading with a β-actin probe) was seen after dexamethasone exposure of B1 and W1 cells, respectively (Fig. 2,B). Furthermore, a dose-dependent decrease in the binding of region I complexes was observed in both W1 (Fig. 2 C) and B1 cells (data not shown).

These data suggest that decreased binding at the NF-IL6/CREB site (a site generally considered crucial for inducible IL-1β expression; Refs. 19, 20, 23, and 42) after dexamethasone exposure may have been at least partially responsible for the decrease in IL-1β mRNA in our leukemic cells and support an important role for this site in the constitutive production of IL-1β in B1 and W1 cells. Of interest in this regard are the observations of Nakajima et al. (48) and Xing et al. (49), who have demonstrated that NF-IL6 and CREB are phosphorylated and activated by a RAS-induced MAP kinase pathway.

Using the reverse dot blot procedure, B1 cells were found to possess a K-RAS mutation at codon 12 due to a GGT→GAT (a glycine to aspartic acid amino acid) change. W1 cells had a mutation in codon 12 of the N-RAS gene due to a GGT→CGT (glycine to arginine) conversion (data not shown).

Two structurally distinct types of FTase inhibitors (monoterpenes and a peptidomimetic) were used to inhibit Ras. These agents suppress the enzyme FTase that is required for the association of Ras with the plasma membrane, its site of function (50, 51, 52, 53). These compounds have also been shown to decrease Ras protein levels without a change in the levels of other G proteins (54). The monoterpenes used, in order of increasing potency, included limonene, perillic acid, and perillyl alcohol (54, 55).

A dose-dependent decrease in IL-1β levels (as measured by ELISA) was seen after exposure of the B1 and W1 cells to the monoterpenes perillic acid and perillyl alcohol, but little change was seen after exposure to the weak parent compound limonene (Figs. 3 and 4). Viability (as assessed by trypan blue) was not significantly decreased after exposure to the highest concentrations of perillyl alcohol (1 mm; viability = 85% in B1 cells and 89% in W1 cells) or perillic acid (3 mm; viability = 83% in B1 cells and 91% in W1 cells). Nor was there a significant decrease in total protein levels as assessed by the Bradford assay (Fig. 3). [Furthermore, washing and replating the cells previously exposed to 1 mm perillyl alcohol or limonene in fresh media revealed similar cell numbers at 48 and 72 h (data not shown), indicating that the cells exposed for 3 days to the highest concentration of perillyl alcohol had not lost their growth potential.] Western blotting using an anti-Ras antibody revealed a decrease in Ras levels in the perillyl alcohol-treated cells but not in the limonene-treated cells (Fig. 5, top panel). There was no effect on the closely related heterotrimeric G protein α-subunit at any of the concentrations tested (Fig. 5, bottom panel).

To confirm the above results, B1 cells were exposed to the peptidomimetic B581, a FTase inhibitor whose mechanism of action differs from the monoterpenes (51). After 2 days of exposure to this molecule, a dose-dependent decrease in IL-1β levels was seen, with a maximum reduction of 50% at a concentration of 500 μm (Fig. 6). Cell viability was unaffected by these concentrations of B581 (data not shown).

These data show that two structurally distinct FTase inhibitors selectively decrease IL-1β levels in W1 and B1 cells, thus indicating that interference with Ras activation inhibits constitutive IL-1β expression in these leukemic cells.

B1 and W1 cells constitutively produce IL-1β and use IL-1β as an autocrine growth factor (1, 3). To determine the effect of the FTase inhibitors on cell growth, cells were treated with 3 mm limonene, 1.5 mm perillic acid, and 3 mm perillic acid for 3 days. In B1 cells, the less active parent compound limonene produced only a 20% decrease in cell growth compared to untreated controls, as assessed by the number of trypan blue-excluding cells (Fig. 7). However, perillic acid decreased cell growth by 40% and 96% at concentrations of 1.5 and 3 mm, respectively. Interestingly, growth arrest was partially reversed at 1.5 and 3 mm perillic acid after treatment with 10 ng/ml rIL-1β. Higher concentrations of rIL-1β were tested, but no significant increase in proliferation was observed (data not shown). (In cells that were not treated with perillic acid, rIL-1β was able to accelerate growth at the plated density; however, once a maximum density was achieved at approximately 48 h, the cells remained static. This allowed the untreated control cells to achieve similar population numbers on day 3). Growth of the B1 cells was also inhibited in a dose-dependent manner after treatment with 0.3, 0.6, and 1 mm perillyl alcohol, whereas 1 mm limonene had no effect on cellular proliferation (data not shown). In these cells, 1 mm perillyl alcohol caused a complete growth arrest. However, there was little effect on cell viability.

In W1 cells, exposure to 3 mm limonene for 3 days decreased cell growth by 15%, whereas 1.5 and 3 mm perillic acid decreased cell growth by 50% and 90%, respectively (Fig. 7). Growth inhibition was partially reversed by IL-1β. Perillyl alcohol also caused a dose-dependent decrease in W1 cell proliferation at concentrations of 0.3, 0.6, and 1 mm (data not shown). (Similar to the B1 cells, untreated W1 cells responded to recombinant IL-1β, but only at low densities).

These results suggest that growth inhibition by FTase inhibitors is partially reversible by adding IL-1β back into the system. Therefore, interference with Ras activation appears to result in decreased proliferation of these leukemic cells via both an IL-1β-dependent and IL-1β-independent pathway.

To determine whether suppression of Ras by FTase inhibitors was associated with an effect on IL-1β transcriptional elements, we assessed the effects of the monoterpenes on binding at the NF-IL6/CREB site using the region I probe and gel shift assays. As demonstrated in Fig. 8, A and B, a decrease in binding to the region I probe was observed with perillyl alcohol but not with limonene, with the effects of various concentrations of perillyl alcohol paralleling the effects on IL-1β levels (Fig. 4). In addition, binding to the housekeeping promoter element Oct-1 was affected to a much lesser extent by perillyl alcohol when compared to region I (data not shown), implying that the decrease in transcription factor binding was not generalized.

It therefore appears that the monoterpenes decrease IL-1β levels in B1 and W1 cells at least partially by inhibiting binding to the crucial cAMP-responsive element in the IL-1β promoter.

To more specifically link Ras to IL-1β production, we used an adenovirus-based antisense approach to inhibit Ras expression. The adenoviral vector used has been shown to inhibit K-Ras expression and the proliferation of K-RAS-transformed lung carcinoma cells in vitro (56). To deduce viral infectivity, B1 cells were infected with a recombinant adenovirus expressing the E. coli LacZ gene (β-galactosidase) in place of the antisense K-RAS cassette. Fluorescence-activated cell sorting of virally infected cells revealed a shift of fluorescence (Fig. 9,B) when compared to uninfected cells (Fig. 9,A), consistent with nearly 100% infectivity of the B1 cell population. However, cells expressing high levels of β-galactosidase only comprised approximately 56% of the population analyzed (Fig. 9,B). B1 cells (which have a K-RAS mutation) were next infected using sense and antisense K-RAS cassettes. IL-1β levels do not change appreciably in sense-infected cells but drop by about 50% in antisense-infected cells (Fig. 9,C). Western blotting also showed a similar reduction in Ras protein levels in the antisense-infected cells (Fig. 9, D and E). Viability and total protein levels were unaffected by viral infection (data not shown; Fig. 9,C). In terms of cell growth, the sense-containing virus had little effect on B1 cell proliferation (Fig. 10). However, the antisense K-RAS virus decreased growth by approximately 60% on day 3 compared to the sense control. To determine whether the effect was reversible, B1 cells were treated with an exogenous source of rIL-1β. Similar to the monoterpene studies, IL-1β partially reversed the adenovirus-induced growth inhibition (Fig. 10, AS K-RAS + IL-1β).

These results indicate that interference with mutated Ras function via exposure of the leukemic cells to antisense also reduces IL-1β levels and suppresses cell proliferation at least in part by decreasing autocrine growth factor production. These observations further strengthen the conclusions derived from the use of the FTase inhibitors, i.e., that Ras activation via mutation results in constitutive IL-1β production which then drives the autocrine growth of the leukemic cells.

Our current observations suggest that mutated RAS can induce the expression of IL-1β, which can then serve as an autocrine growth factor for leukemic cells. Whereas an explicit connection between IL-1β and RAS has not been previously established in leukemia, there were several lines of experimental evidence that prompted our investigation of such a role for mutated RAS. For instance, transfection of activated RAS into a mesothelioma cell line stimulates the transcription of the IL-1β gene (24). In addition, macrophage cell lines infected with a v-RAF-containing retrovirus (Raf being an immediate downstream effector of Ras signaling) are growth factor independent and produce IL-1β constitutively (26). The above reports on the mesothelioma and macrophage experimental models suggested the possibility that signal transduction pathways that lie downstream of Ras are capable of activating transcription factors that bind the IL-1β promoter.

A wealth of studies describe constitutive expression of IL-1β in diverse leukemias (1, 3, 4, 7, 8, 9, 10, 12, 14). Furthermore, IL-1β can serve as an autocrine and/or paracrine growth factor for leukemic blasts (1, 2, 3, 4, 8, 11). To determine whether activated Ras acts as an inducer of IL-1β expression in human hematopoietic malignancies, we analyzed two leukemic cell lines that constitutively produce high levels of IL-1β, use IL-1β as an autocrine growth factor (1, 3), and have RAS mutations. Our preliminary data indicated that the aberrant expression of IL-1β in these two cell lines (B1 and W1) was most likely due to an upstream molecular defect. This data included Southern blotting, which revealed no rearrangement or amplification in the IL-1β gene or its promoter, and gel shift assays (Fig. 1,A; data not shown), which demonstrated constitutive binding of multiple critical promoter elements (NF-IL-6/CREB, NFκB, NFB1, and NF-IL-6) in unstimulated B1 and W1 cells. It is believed (19) that maximal expression of IL-1β is induced through synergistic activity of the NFB1, NF-IL6, and NFκB binding elements, which cooperate with the NF-IL6/CREB site to overcome the effects of a negative regulatory element located in the IL-1β promoter. The NF-IL6/CREB heterodimer may be the most crucial transcription factor for high level expression of IL-1β (19, 20, 23, 42) because in THP-1 cells, mutation of this binding site in an IL-1β promoter construct leads to a 75% reduction in expression after multiple stimuli (20). In our cell lines, the finding that a decrease in IL-1β mRNA after exposure to dexamethasone (Fig. 2,B) was accompanied by a decrease in NF-IL6/CREB binding (Fig. 2 C) provided indirect support for the relevance of this site. Importantly, both NF-IL6 (48) and CREB (49) have been shown to be activated by Ras-induced pathways. Furthermore, activation of other IL-1β-inducing transcription factors may also be regulated by pathways that overlap or are downstream of Ras. For instance, the Stat1-related protein NFB1 (19) can be activated by MAP kinase in a fashion mimicking that described for Stat1 (57). Also, NFκB is known to lie downstream of Raf-mediated pathways (58, 59) and can also be activated in the presence of a RAS mutation.

To more directly assess the possibility that activated RAS is responsible for the anomalous expression of IL-1β, B1 and W1 cells were exposed to molecules designed to inhibit Ras function and/or Ras expression. Two structurally different classes of FTase inhibitors were used to inhibit Ras: (a) monoterpenes; and (b) a peptidomimetic. These compounds inhibit the enzyme FTase, which is required for the addition of a 15-carbon farnesyl group to the Ras COOH terminus (50, 51, 55). This modification allows Ras to associate with the plasma membrane, its site of function (52, 53). The monoterpenes resemble the farnesyl group and therefore serve as competitive inhibitors of FTase (60, 69, 70). These molecules can also decrease Ras protein levels without affecting the levels of other G proteins in leukemic cell lines (54). The monoterpenes used included limonene, perillic acid, and perillyl alcohol, with limonene previously shown to have weak anti-Ras activity, perillic acid previously shown to be of intermediate potency, and perillyl alcohol previously shown to be the most potent (54, 55). Treatment of B1 and W1 cells with increasingly potent or increasing concentrations of monoterpenes caused a dose-dependent decrease in IL-1β protein levels (Figs. 3 and 4). The potency-related effects were consistent with those published by Crowell et al. (55). At the concentrations tested, the monoterpenes appeared to have low levels of nonspecific toxicity because cell viability and total protein levels (Fig. 3) were not significantly changed. In addition, cells treated with the highest concentrations of perillyl alcohol (1 mm) retained their growth potential after washing and replating in fresh media. Importantly, binding at the crucial NF-IL6/CREB site was inhibited in both the B1 and W1 cells by perillyl alcohol (Fig. 8). The effects on binding closely mimicked the effects on IL-1β expression (Fig. 4), suggesting that the monoterpenes reduced IL-1β levels by interfering with pathways that resulted in transcription of the IL-1β gene. Similar to the findings of Hohl and Lewis (54), perillyl alcohol was able to reduce Ras protein levels, whereas limonene had no effect (Fig. 5). It is possible that farnesylated Ras is more stable than the nonfarnesylated form or, alternatively, that that the inhibitory affects of the monoterpenes on Ras may be independent of the enzyme FTase (54). In addition to the effects on IL-1β, the monoterpenes perillic acid and perillyl alcohol were effective inhibitors of leukemic cell proliferation (Fig. 7). Growth inhibition induced by perillic acid was partially reversed with an exogenous source of recombinant IL-1β, further strengthening the link between IL-1β-related autocrine proliferation and activated Ras.

To further rule out the possibility of nonspecific drug effects, B1 cells were treated with B581, a peptidomimetic that is structurally distinct from the monoterpenes. B581 is a tetrapeptide analogue that mimics the Ras COOH-terminal CAAX motif (51), therefore serving as a competitive inhibitor of FTase. The results were similar to those seen with the monoterpenes, with B581 reducing IL-1β protein levels (Fig. 6) without significantly affecting cell viability. To more specifically inhibit Ras function, B1 cells (which harbor a K-RAS mutation) were treated with an antisense adenovirus that has previously been shown to inhibit K-Ras expression (56). In the presence of antisense K-RAS, the expression of IL-1β was decreased by 50% in the B1 cells (Fig. 9,C) with a corresponding decrease in Ras protein levels (Fig. 9, D and E). Viability and total protein were unaffected by the adenoviral infections. In terms of cell growth, the antisense adenovirus inhibited B1 cell proliferation as compared to the sense control (Fig. 10). As with the monoterpenes, growth inhibition was partially reversed when antisense-treated cells were exposed to an exogenous source of IL-1β (Fig. 10). These data confirmed that interference with mutated RAS expression reduced IL-1β levels and autocrine growth in our leukemic cells.

It appears that in both experimental models and in human hematopoietic neoplasms in vitro, Ras-mediated pathways can lead to the expression of IL-1β. It therefore seems plausible that under some circumstances, activated RAS can induce the binding of multiple transcription factors that ultimately lead to the activation of IL-1β transcription. The fact that Ras pathways are frequently activated in several types of hematological malignancies either directly by mutation or indirectly by alternate genetic defects is consistent with a central role for Ras in the induction of IL-1β in these diseases. In hematopoietic cells, this aberrant expression can help propagate the transformed phenotype by autocrine growth stimulation.

Finally, our study implies that the proliferation of our leukemic cell lines can occur through IL-1β-dependent and IL-1β-independent mechanisms, both of which rely, at least in part, on Ras. Previously published reports as well as data from our laboratory have demonstrated that the proliferation of B1 and W1 cells can be suppressed with neutralizing strategies targeted to IL-1β (1, 3). However, the suppressive effects were only partial (in the range of 50%). In contrast, interference with Ras function and/or expression by the monoterpenes lead to a virtually complete inhibition of growth. Growth arrest was only partially reversed by an exogenous source of IL-1β. It is therefore possible that activated RAS promotes the proliferation of leukemic blasts by stimulating the expression of cytokines such as IL-1β, which can then serve as autocrine growth factors, and simultaneously activating components of the cell cycle machinery or by inducing other pathways that function independently of cytokine production.

In conclusion, our findings imply that activation of RAS by mutation is responsible, at least in some cases, for autocrine IL-1β production in leukemias. Of interest in this regard is that RAS mutations are also ubiquitous in solid tumors, as is constitutive IL-1β expression (5, 71). Further work should reveal whether these two events are related in nonhematopoietic tumors as well.

We thank Dr. P. Auron, W. Waterman, and Dr. J. Gray for scientific input regarding IL-1β expression and possible avenues of deregulation; Dr. R. Hohl for input regarding the use and mechanisms of action of the monoterpenes; Sandy Wiehle for help in virus preparation; and technicians S. Ku and M. Leonard for technical support.