Suppression of Human Ribosomal Protein L23A Expression During Cell Growth Inhibition by Interferon-B

Suppression of human ribosomal protein L23A expression during cell growth inhibition by interferon-b

Hongping Jiang1, Jiao Jiao Lin2, Jing Tao1 and Paul B Fisher1,2

Departments of 1Pathology and 2Urology, Columbia-Presbyterian Comprehensive Cancer Center, College of Physicians and Surgeons of Columbia University, 630 West 168th Street, New York, NY 10032, USA

Interferons inhibit cell growth in normal and tumor- Recent studies are providing a precise de®nition of derived cells. The molecular basis of interferons anti- the biochemical and molecular events underlying IFN proliferative activity remains to be de®ned. Using action (Pestka et al., 1987; Darnell et al., 1994). An subtraction hybridization, a human melanoma dierentia- IFN response begins with the binding of interferon to tion associated gene, mda-20, has been identi®ed that is speci®c cell surface receptors resulting in the activation down-regulated by treatment with interferon. Sequence of speci®c enzymes, JAK kinases, that phosphorylate a analysis indicates that mda-20 is human ribosomal protein family of proteins, STAT proteins, that translocate to L23a (rp L23a). The mRNA levels of rp L23a and growth the nucleus and bind to speci®c DNA response are diminished in a variety of human tumor cell lines elements (Hunter, 1993; Darnell et al., 1994). This following treatment with human ®broblast interferon, cascade of biochemical changes ultimately results in the interferon-b (IFN-b). Expression of rp L23a is also modi®ed expression of distinct sets of genes in IFN- reduced in human melanoma cells treated with human treated cells (Darnell et al., 1994). Many IFN-induced leukocyte (IFN-a) and immune (IFN-g) interferons, but genes have been identi®ed and characterized, and their not by growth inhibition resulting from serum starvation. roles in regulating cellular changes are under active These ®ndings suggest that growth suppression alone is investigation (for review see Pestka et al., 1987). Two not sucient to reduce rp L23a expression. Instead, IFN-inducible pathways, the 2'-5' oligoadenylate reduced rp L23a mRNA results from biochemical changes synthesis (2-5 A) and the ds-RNA-dependent protein mediated by interferons. Ectopic expression of an kinase (PKR) systems appear to be involved in IFN- antisense rp L23a sequence in human HeLa cervical induced growth inhibition (Romeo, 1989; Silverman, carcinoma cells results in a reduction in colony formation 1994). The 2-5 A system is an interferon regulated indicating a direct antiproliferative eect by inhibiting RNA degradation pathway (Hassel et al., 1993). IFN rp L23a expression. The mechanism underlying inhibition treatment induces synthesis of the enzyme 2-5 A in rp L23a expression in IFN-b-treated cells may involve polymerase that catalyzes the production of 2',5'- antisense rp L23a RNA. These results suggest that linked oligoadenylates known as 2-5 A (Sen and rp L23a may be one of the target molecules involved in Lengyel, 1992). A 2-5 A-dependent endoribonuclease mediating growth inhibition by interferon. is activated that cleaves mRNAs and rRNAs resulting in growth inhibition (Zhou et al., 1993). The dsRNA- Keywords: interferon-b; melanoma dierentiation asso- activated protein kinase (p68) is induced by IFNs ciated gene-20 (mda-20); ribosomal protein L23a; cell resulting in autophosphorylation in human cells growth inhibition; antisense transcripts (Szyszka et al., 1989). A target substrate of the p68 kinase is the a-subunit of eIF2, which when phosphorylated results in a general inhibition of protein synthesis and growth suppression. Down- Introduction regulation of expression of speci®c genes by IFNs may also negatively regulate growth (Pestka et al., Interferons (IFNs) consist of a family of proteins that 1987). This possibility is suggested by several lines of induce numerous changes in target cells, including evidence: (1) IFNs down-regulate expression of growth induction of an antiviral state, growth suppression, factor receptors thereby antagonizing the growth modulation of dierentiation and alterations in promoting eects of speci®c growth factors (Zoon et histocompatibility and tumor associated antigen al., 1986); (2) IFNs inhibit the activity of ornithine expression (Fisher and Grant, 1985; Greiner et al., decarboxylase which is a rate-limiting enzyme in the 1985; Pestka et al., 1987). As antiproliferative agents biosynthesis of polyamines (Sreevalsan et al., 1980; of both normal and tumor cells, exogenous IFNs Sekar et al., 1983); and (3) IFNs inhibit genes involved prolong all phases of the cell cycle and inhibit the in cell growth (Harel-Bellam et al., 1988; Chatterjee transition of cells from G0/G1 to S phase (Romeo et and Savarese, 1992). These results suggest that multiple al., 1989). Autocrine interferons are involved in mechanisms may underlie the ability of IFNs to inhibit feedback control of cell proliferation, dierentiation cell growth. and gene expression in cultured cells (Birnbaum et al., IFNs display variable success in treating patients 1990; Abdollahi et al., 1991; Belhumeur et al., 1993; with metastatic melanomas (Legha, 1989; Gutterman, Jiang et al., 1993a). 1994). All three types of IFNs, leukocyte (IFN-a), ®broblast (IFN-b) and gamma (IFN-g), induce a dose- Correspondence: PB Fisher dependent antiproliferative eect on melanoma cell Received 8 July 1996; revised 30 September 1996; accepted 1 October growth in vitro with the greatest inhibition occurring 1996 with IFN-b (Garbe and Krasagakis, 1993). IFN-b also Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 474 facilitates melanoma cell dierentiation induced by a protein kinase C activating (TPA-like) compound, a mezerein (Fisher et al., 1985). The combination of 24 hr 48 hr 72 hr 96 hr β β IFN-b and mezerein results in induction of terminal β β CON IFN- CON IFN- cell dierentiation accompanied by an irreversible loss CON IFN- CON IFN- of proliferative capacity in human melanoma cells (Jiang et al., 1993a, 1995a). mda–20 Subtraction hybridization between cDNAs constructed from actively proliferating HO-1 human melanoma cells and dierentiation inducer (IFN- GAPDH b+MEZ) treated HO-1 cells identi®ed novel melanoma dierentiation associated (mda) genes whose expression is induced or enhanced in dierentiated b versus undierentiated human melanoma cells (Jiang α β γ

and Fisher, 1993). Two of the mda genes, mda-6 and CON IFN– IFN– IFN– mda-7, can inhibit cell growth when transfected and overexpressed in human melanoma and other cell types mda–20 (Jiang et al., 1995b,c, 1996). Using the same subtraction hybridization protocol, mda genes display- ing altered expression following IFN treatment have GAPDH been identi®ed and cloned (Jiang and Fisher, 1993; and unpublished data). A component of ribosomal Figure 1 Northern blot analysis of melanoma dierentiation- proteins, human ribosomal protein L23a (rp L23a), associated gene-20, mda-20, expression in HO-1 human melanoma identi®ed as mda20, is one of the genes down-regulated cells grown in the presence or absence of IFN. Total RNA from in human melanoma and other cell types treated with control and IFN-b-treated (1000 units/ml) HO-1 cells was isolated, electrophoresed (10 mg/lane), and transferred to nylon IFN-b. We have cloned the rp L23a cDNA and membranes. Filters were hybridized with random primed 32P- analysed the regulation of expression of this gene in dCTP-labeled mda-20 cDNA fragments. Membranes were the context of IFN-b-induced growth inhibition. Our stripped and rehybridized with random primed 32P-dCTP-labeled results indicate that a modi®cation of human rp L23a GAPDH cDNA fragments. Changes in mda-20 expression were expression accompanies and may contribute to cell quantitated by comparing levels of mda-20 to GAPDH by densitometer analyses. (a) HO-1 human melanoma cells treated growth inhibition induced by IFN-b. The regulation of with IFN-b for 24, 48, 72 and 96 h. (b) HO-1 cells treated with rp L23a is exerted at a post-transcriptional level. We 1000 units/ml of IFN-a, IFN-b or IFN-g for 24 h demonstrate that antisense rp L23 sequences (rp L23a (AS)) are transcribed de novo and steady state levels of rp L23a (AS) transcripts decrease with IFN-b treat- most eective interferon in reducing mda-20 expression, ment. Moreover, transfection of rp L23a (AS) into i.e. 79% reduction with IFN-g, 65% reduction with IFN- HeLa cells results in a reduction in colony formation. b and 45% reduction with IFN-a. These observations support the hypothesis that IFN-b- Double-stranded sequencing of the mda-20 cDNA induced growth suppression may in part result from clone (209 bps) indicates extensive homology (89%) to the down-regulation of rp L23a expression, mediated rat rp L23a (Suzuki and Wool, 1993) and almost by antisense rp L23a transcripts. identical homology (599%) to partial sequences of several human cDNA clones in Genbank accession numbers H56983, H80209, L13799 and U02032. The mda-20 cDNA insert was used as a probe to screen an Results HO-1 cDNA library (Jiang and Fisher, 1993; Jiang et al., 1995b,c). Several cDNA clones that hybridized to Identi®cation of human rp L23a, mda-20, as an mda-20 were isolated and found to contain sequences interferon-suppressible gene in human melanoma cells identical to the mda-20 clone. From these cDNA clones, Using a subtraction hybridization library screening a complete cDNA sequence was obtained, which protocol (Jiang and Fisher, 1993), we isolated cDNAs contains an open reading frame (ORF) encoding a dierentially expressed in HO-1 human melanoma cells protein of 156 amino acids that is identical to rat grown in the presence or absence of IFN-b. Several rp L23a (Suzuki and Wool, 1993). Therefore, the gene known and novel, not previously reported in current originally isolated as mda-20 is a full-length human DNA data bases, interferon-inducible genes were rp L23a cDNA. The sequence of mda-20 is shown in identi®ed (Jiang and Fisher, 1993 and unpublished Figure 2 and consists of 550 nucleotides, apart from a data). In addition, a cDNA clone (mda-20) was found poly(A) tract, with the open reading frame starting at to be interferon suppressible (Figure 1). A reduction in position 22 and terminating at position 492. The open mda-20 expression is apparent in HO-1 cells after 24 h reading frame is ¯anked by 5'-and3'-untranslated growth in IFN-b and this inhibition persists in 96 h sequences of 21 and 54 nt, respectively. There is a growth-suppressed IFN-b-treated cells (Figure 1a; Jiang hexamer, AATATA, 24 nt upstream of the start of the et al., 1993a). In the experiment shown in Figure 1a, mda- poly(A) tract, which is an infrequent variant from the 20 expression is reduced by 41% after 24 h, 58% by 48 h, canonical polyadenylation signal AATAAA (Sheets et 76% by 72 h and 89% by 96 h. Suppression of mda-20 al., 1990). To determine the initiation site of the expression is also observed in HO-1 cells treated with transcripts, the RACE technique was used (Jiang et IFN-a or IFN-g (Figure 1b). When treated with a similar al., 1995b,c). Nested primers complementary to the amount of antiviral units (1000 units/ml), IFN-g is the rp L23a mRNA were used for the reverse transcription Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 475

a GBM–18 HeLa SW613 T47D HTB–113 β β β β β CON IFN– CON IFN– CON IFN– CON IFN– CON IFN–

rp L23a

GAPDH

b Figure 2 The cDNA and protein sequence of human mda-20 is identical to human ribosomal protein L23a. The top lane indicates the DNA sequence of the human rp L23a (mda-20) cDNA. The bottom lane indicates the predicted amino acid sequence of rp L23a (mda-20). The probable polyadenylation signal is underlined. The numbering begins at the 5' end of the rp L23a mRNA, determined as described in the text. Accession number U443701 (Genbank) and PCR reaction. The ampli®ed 5'-end fragments were cloned in plasmids and sequenced. The sequencing results indicate that there are two types of 5' ends: one is shown as in Figure 2, and the other has seven thymidines in place of nucleotides 1 to 15. This second type of 5' end was also found in mda-20 cDNA clones isolated from the HO-1 cDNA library. The ability of this cDNA to encode a protein was veri®ed by in vitro translation (data not shown). These two types of 5' ends Figure 3 Eect of IFN-b on cell growth and rp L23a expression in various human tumor cell lines. (a) Northern blot analysis of may result from transcription of rp L23a alleles. Both rp L23a expression. The cells were treated with IFN-b cDNAs contain pyrimidine rich sequences at their 5' (1000 units/ml) for 24 h. (b) Cell growth inhibition by IFN-b. ends. Similar pyrimidine rich sequences at the 5' ends The cells were treated with IFN-b (1000 units/ml) for 96 h and are also present in rat rp L23a and most eukaryotic triplicate samples were counted. The results are the average of ribosomal protein mRNAs and presumably play a role triplicate plates that varied by 410%. Cell lines: GBM-18, human glioblastoma multiforme; HeLa, human cervical carcino- in the regulation of their translation (Suzuki and Wool, ma; SW613, human colon carcinoma; T47D, human breast 1993; Levy et al., 1991). Southern blot analyses of the carcinoma; HTB-113, human endometrial carcinoma restriction digests of human genomic DNA demonstrate several hybridization bands indicative of multiple copies of the rp L23a genes (data not shown), although a treatment. Reduction of rp L23a mRNA may result majority of them may be pseudogenes as is the case with directly from treatment with IFN-b or this gene other mammalian rp genes (Wagner and Perry, 1985; expression change may be a consequence of changes Chan and Wool, 1995). in growth status. To distinguish between these possibilities, several experiments were performed. HeLa, GBM-18, and HO-1 cells were cultured in the Down-regulation of human rp L23a is associated with absence of serum (DMEM-0) which results in growth IFN-b-induced growth inhibition in a variety of cell types inhibition, as indicated by a reduction in cell number The rp L23a mRNA is detected on Northern blots in a and an inhibition in DNA synthesis (Jiang et al., variety of cells, including HeLa, GBM-18, SW613, 1995c; data not shown). In serum deprived cells, no T47D and HTB-113 (Figure 3a). A variable reduction suppression in rp L23a mRNA is apparent (Figure 4a). in rp L23a mRNA levels is observed in all cell lines Indeed, if one compares the ratios of GAPDH to mda- treated with IFN-b. In the experiment shown in Figure 20 expression by densitometry analysis, DMEM-0 3a, rp L23a mRNA levels are reduced by 20% in actually increases rp L23a expression by an average GBM-18, 35% in HeLa, 50% in SW613, 50% in T47D of *65% above control serum containing cultures. and 44% in HTB-113. Similar reductions+15% have These results indicate that the changes observed in also been observed in replicate studies with the same rp L23a mRNA levels induced by IFN-b can be cell types (unpublished data). Eects of IFN-b on cell dissociated from changes in cell proliferation. Addi- growth were determined after incubating cells in IFN-b tional experiments were conducted to con®rm this for 4 days. A substantial reduction in cell number (24 hypothesis. Total cellular RNAs were isolated from to 85%) is observed in all of the cell lines treated with HeLa cells treated with dierent doses of IFN-b and IFN-b (Figure 3b). A direct correlation between the these were tested for rp L23a gene expression. An IFN- amount of rp L23a mRNA and the degree of growth b dose-dependent reduction in rp L23a mRNA levels suppression induced by IFN-b is not apparent in the was observed after 24 h treatment, without signi®cant dierent cell types. inhibition in cell growth, especially at low IFN-b doses, An important question is whether suppression of i.e. mda-20 expression is reduced by 53 and 62% after rp L23a mRNA is a direct or indirect eect of IFN-b treatment with 0.1 or 1 units/ml of IFN-b, respectively Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 476 a GBM 18 HeLa HO–1 CON DMEM–0 CON DMEM–0 CON DMEM–0

rp L23a

GAPDH

b Figure 5 Eect of rp L23a (AS) expression on colony formation IFN–β in HeLa cells. HeLa cells (46105 cells/plate) were transfected (Units/ml) with the pREP4 (vector) or pREPrpL23a (AS) (antisense rp L23a construct) respectively. Selection with hygromycin began after 48 h and continued for approximately 3 weeks. The plates were 0 0.1 1 10 100 1000 ®xed and stained with methylene blue rp L23a pREPrpL23a (AS) resulted in 716+62. This represents *43% reduction in colony formation, similar reduc- GAPDH tions in colony formation have been observed in several independent transfection assays. A representa- tive transfection assay is shown in Figure 5. A reasonable conclusion is that suppression of rp L23a c expression may contribute to cell growth inhibition induced by IFN-b. CON 15 min 30 min 60 min 2 hr 4 hr 8 hr 24 hr

rp L23a Antisense rp L23a transcripts may mediate down- regulation of rp L23a expression in IFN-b treated cells To determine whether down-regulation of rp L23a GAPDH mRNA occurs by a transcriptional or post-transcriptional mechanism, RNA degradation assays and Figure 4 Northern blot analysis of rp L23a expression in cells nuclear run-on assays were performed. In RNA starved of serum or treated with dierent doses of IFN-b and for various time periods. (a) The cells were cultured in medium degradation assays, cellular RNA synthesis was without serum (DMEM-0) for 96 h and total RNA was isolated blocked with actinomycin D and RNA degradation for the analysis. (b) HeLa cells were treated with dierent doses of rates were assessed by Northern blot analysis (Jiang et IFN-b for 24 h and total RNA was isolated and analysed by al., 1993b). This assay indicated no change in the Northern blotting (c) HeLa cells were treated with IFN-b stability of rp L23a mRNA in the absence of (1000 units/ml) for dierent time periods transcription after IFN-b treatment (data not shown). Nuclear run-on assays were then conducted to measure the transcription rate of the rp L23a gene before and (Figure 4b and data not shown). In addition, analysis after IFN-b treatment. Nylon membranes bearing sense of mRNAs isolated at early time points indicates a and antisense rp L23a RNAs transcribed in vitro were decrease of *20 and *43% in rp L23a mRNA levels used to detect the nascent antisense and sense 1 and 8 h after IFN-b treatment, respectively (Figure transcripts from isolated nuclei of HeLa cells 4c). These ®ndings support the hypothesis that untreated or treated for 1 and 24 h with IFN-b. suppression of rp L23a gene expression precedes cell Transcription of rp L23a (AS) sequences was apparent growth inhibition, and results from a direct eect of in the untreated HeLa cells and HO-1 human IFN-b treatment rather than from cell growth melanoma cells (Figure 6a and data not shown). The inhibition. rate of transcription from the rp L23a sense orientation An equally important question is whether IFN-b was unaected, but the transcription rate from the inhibition of cell growth can occur as a result of rp L23a antisense orientation was increased by IFN-b directly changing rp L23a expression. To address this after 1 h treatment (Figure 6a). In contrast, no question, DNA transfection assays were performed dierence was apparent in the transcription rate of with an expression vector containing antisense rp L23a rp L23a (AS) after 24 h treatment with IFN-b.Asa sequences and the control vector. HeLa cells were control, transcription of the GAPDH gene was transfected with the pREP4 vector construct lacking determined and found to be unaected by IFN-b rp L23a sequences (control vector) or an antisense treatment (Figure 6a). pREP4 construct containing rp L23a in an antisense To con®rm the existence of the rp L23a (AS) orientation (pREPrpL23a (AS)) plasmids and selected transcripts in cells, Northern blot analysis and RT ± in hygromycin-containing medium. After 3 week PCR assays were used. The Northern blot analysis selection, colonies of 450 cells were determined. indicated high background hybridization, and the Transfection with pREP4 vector resulted in rp L23a (AS) transcripts were barely detected (data 1257+121 colonies, whereas transfection with not shown). There are, at least, two possible Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 477 Discussion

a β

CON IFN– Although the mechanism of communication between IFN receptors and the nucleus has been extensively GAPDH investigated, the IFN-targeted genes responsible for IFN-induced cell growth inhibition remain to be rp L23a (S) identi®ed and their precise roles in modulating rp L23a (AS) cellular proliferation remain to be elucidated. In this investigation, we report the identi®cation of human rp L23a, mda-20, as an IFN-suppressible gene. Expression of rp L23a is down-regulated by IFNs in variety of tumor cell lines including human melanoma b cells from which this gene was isolated. Down- regulation of rp L23a in IFN-treated cells, but not 0 15 min 30 min 60 min 2hr 4hr 8hr 24hr in serum-starved cells, suggests that rp L23a expres- rp L23a (AS) sion can be dissociated from growth arrest although its expression directly correlates with IFN treatment. However, an involvement of rp L23a as a component in cell growth inhibition induced by IFNs is suggested GAPDH by the fact that ectopic expression of antisense rp L23a can suppress cell proliferation, probably Figure 6 Eect of IFN-b on rp L23a (AS) expression. (a) through a reduction of this component of the cellular Transcription rates of sense and antisense rp L23a in HeLa cells protein synthesis machinery. The present study also treated with IFN-b. Nuclei were isolated from cells untreated or identi®es rp L23a (AS) transcripts in cells. These treated with IFN-b (1000 units/ml) for 1 h. RNA labeled during nuclear run-on transcription was hybridized to the single-stranded antisense transcripts may mediate degradation of L23a RNA transcribed in vitro from both directions. GAPDH rp L23a mRNAs, thereby regulating levels of the was used as a control. (b) Correlation between rp L23a (AS) rpL23a protein and directly contribute to IFN-induced RNA levels and IFN-b treatment. Total RNA was isolated from growth suppression. HeLa cells treated with IFN-b for the time periods indicated. Expression of ribosomal proteins is subject to both Reverse transcription was performed with a primer complementary to rp L23a (AS) RNA and then PCR was carried out to transcriptional and post-transcriptional (translational) amplify the rp L23a (AS) cDNAs. GAPDH was used as a control regulation (for reviews see Mager, 1988). The promoters of the rp genes in most higher eukaryotes display striking similarities in architecture that are explanations for this observation: (a) immediately typical features of housekeeping genes, i.e. the absence after synthesis of the antisense transcripts, they of a canonical TATA box and the presence of a hybridize to the sense RNA and are digested by pyrimidine-rich transcriptional start region (Mager, RNases; and/or (b) the antisense transcripts them- 1988). Accordingly, accumulation of rp mRNAs will selves have high turnover rates in cell nuclei. A more not ¯uctuate with the cell cycle and they will not be sensitive assay than Northern blotting, RT ± PCR, was aected by cell growth status (Jacobs et al., 1985; used to determine if rp L23a (AS) transcripts are Meyuhas et al., 1987; Mager, 1988). The level of present in cells. An oligo primer complementary to the rp L23a mRNA remains constant in growth arrested rp L23a (AS) sequence was used in reverse transcrip- versus actively growing HO-1, HeLa and GBM-18 tion to synthesize an rp L23a (AS) cDNA. These cells and expression is not regulated transcriptionally cDNAs were then ampli®ed by PCR. As shown in by IFNs. These observations are in agreement with Figure 6b, similar to the nuclear-run-on results, the the ®ndings of a number of investigators studying rp antisense transcripts were present in HeLa cells. genes (Jacobs et al., 1985; Meyuhas et al., 1987; However, their abundance was reduced as was Mager, 1988). However, transcriptional control of rp rp L23a mRNA in cells treated with IFN-b.A genes has been observed during mouse myoblast reduction in L23a (AS) mRNA of 75% was apparent dierentiation, suggesting the presence of regulatory by 15 min and after 8 and 24 h there was a 595% elements in the promoter regions of the rp genes in decrease in rp L23a (AS) mRNA in IFN-b-treated muscle cells that may be aected by dierentiation HeLa cells. The possibility that the reduction was status (Agrawal and Bowman, 1987). Moreover, NF- caused by diminution of the transcription rate was a1 (E2F) and NF-b1 recognition sequences have been excluded by nuclear run-on assays (data not shown). found in diverse mammalian housekeeping genes As a result of the possible rp L23a (AS)-mediated including ribosomal proteins (Overman et al., 1993). degradation, the steady-state RNA levels of rp L23a These results support the hypothesis that multiple decrease by 1 h after IFN-b treatment (Figure 4c). To mechanisms may underlie the regulation of ribosomal con®rm the ability of rp L23a (AS) to hybridize proteins during cell growth arrest and dierentiation (based on sequence homology) with rp L23a mRNA, in dierent cell types. the antisense PCR products were cloned and The constant rate of rp L23a expression during the sequenced. The sequence of the rp L23a (AS) PCR cell cycle, the constant level of rp L23a mRNA when clones is complementary to its rp L23a (S) counterpart cells are treated with actinomycin D, and the presence (data not shown). However, the sizes of the full length of antisense rp L23a transcripts de novo in tumor cells rp L23a (AS) transcripts and whether the transcripts suggest an alternative mechanism underlying post- contain rp L23a-unrelated sequences beyond the PCR- transcriptional regulation of rp L23a in IFN-treated cloned region remain to be determined. cells. Regulation at the level of pre-mRNA maturation Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 478 through endogenous antisense transcripts has been ship between antisense rp L23a expression and the observed in several genes, including p53 in mouse cells 2-5 A system in regulating cell growth. Additional (Khochbin and Lawrence, 1989), c-myc and N-myc in IFN-inducible or suppressible genes are also likely human cells (Krystal et al., 1989; Celano et al., 1992), contributors to growth inhibition following cytokine and rp L1 in Xenopus laevis (Caarelli et al., 1987). treatment. Antisense RNAs can pair to speci®c complementary DNA and RNA targets to alter transcription or translation. It has also been suggested that a block in Materials and methods the splicing of nuclear mRNA precursors by antisense transcripts may lead to destabilization of the pre- Cell lines and culture conditions mRNA or interfere with pre-mRNA maturation (Caarelli et al., 1987; Khochbin and Lawrence, The HO-1 human melanoma cell line was established from a metastatic inguinal lymph node lesion from a 49 year-old 1989). The enhanced transcriptional rate of rp L23a female (Giovanella et al., 1976; Fisher et al., 1985). HeLa (AS) 1 h post-treatment with IFN-b and the is a human cervical carcinoma cell line (Kao and Nevins, signi®cantly reduced amount of the antisense 1983). Additional human tumor cell lines include: GBM-18 rp L23a transcripts in IFN-b-treated cells suggest a (glioblastoma multiforme), SW613 (colon carcinoma), novel mechanism for IFN-b modulation of rp L23a T47D (breast carcinoma) and HTB-113 (endometrial levels in IFN-treated cells. The double-stranded RNAs adenocarcinoma). Cells were cultured in Dulbecco's formed by interaction of the sense and antisense L23a modi®ed Eagle's medium (DMEM) supplemented with transcripts may be degraded by an endoribonuclease 10% fetal bovine serum (FBS) and penicillin/streptomycin (100 U/100 mg/ml)at378Cina5%CO 95% air- that is activated in IFN-treated cells. This hypothesis 2 is supported by several observations. Continuous humidi®ed incubator. To determine the eect of IFN on various parameters, cells were cultured in the presence or transcription of speci®c genes, such as rp L23a (AS), absence of IFN for various times and the cultures were may be mandatory for degrading rp L23a mRNAs, harvested for cell counting, RNA isolation or nuclei that are normally very stable following inhibition of isolation. For serum starvation studies, the medium was RNA transcription. Growth inhibition by the forced changed to 0% FBS when cells were subcon¯uent, cultures expression of rp L23a (AS) in HeLa cells suggests that were incubated in this medium for 96 h and cells were rp L23a is down-regulated during growth suppression harvested for RNA isolation. in these cells. The rp L23a (AS) sequence is complementary to rp L23a ensuring the formation of Cell growth assay ds RNAs between rp L23a (S) and rp L23a (AS) 4 transcripts, that can then be cleaved by RNases Cells were seeded at 2.5610 /35 mm plate and 24 h later the medium was changed with and without IFN-b present in cells (Schein et al., 1990; Chiang et al., (1000 units/ml). The cells were cultured for 96 h and then 1991; Schein and Haugg, 1995). Of interest is the triplicate samples for each condition were counted using a observation that the catalytic activity of a ribonu- ZM Coulter counter (Hialeah, FL). The data was analysed clease (bovine seminal ribonuclease) toward double- for statistical signi®cance using the Student's t-test. stranded RNA is enhanced following IFN treatment. To directly prove our hypothesis, it will be necessary Gene cloning and sequence analysis to rule out the possibility that the rp L23a (AS) transcripts are degraded without forming double- The rp L23a cDNA, mda20, was identi®ed and cloned by stranded RNA structures and to identify the subtraction hybridization as described previously (Jiang interferon-activated RNAses that degrade double- and Fisher, 1993). The full-length cDNA was isolated by screening an HO-1 cDNA library and con®rmed by the stranded RNA in human cells. Considering the RACE (rapid ampli®cation of cDNA ends) technique importance of 2-5 A dependent RNase (RNase L) in (Frohman et al., 1988; Jiang et al., 1995b,c). The DNA controlling cell proliferation (Silverman, 1994), it will sequence was determined using a sequenase kit (Sequenase be of interest to determine whether RNase L functions ver 2.0; USB). The entire sequence was determined for in degrading ds RNA. both strands. Nucleotide and amino acid sequence analyses The amino acid sequences of human and rat rp L23a were performed using the GCG sequence analysis software are identical (Suzuki and Wool, 1993). Rat L23a is package (Pearson and Lipman, 1988; Genetics Computer related to Saccharomyces cerevisiae L25 and to Group, 1991). Escherichia coli L23 that bind to a speci®c site in domain III of 26S and of 23S rRNAs, respectively RNA analysis and quantitation (Vester and Garrett, 1984; El-Baradi et al., 1985; Total cellular RNA from cell cultures was prepared using Suzuki and Wool, 1993). Based on the similarity in the guanidinium/phenol procedure (Chomczynski and amino acid sequences of L23a, L23 and L25, it was Sacchi, 1987). For Northern blot analysis, 10 mgof inferred that L23a likely binds to the homologous site RNA/sample were denatured with glyoxal/DMSO, electro- in rat 28S rRNA and participates in initiation of the phoresed on 1% agarose gels, transferred to nylon assembly of the large ribosomal subunits (Suzuki and membranes, and hybridized with radioactively labeled Wool, 1993). The importance of human rp L23a in the probes prepared from the cloned rp L23a cDNA and assembly of the protein synthesis machinery is GAPDH cDNA fragments as described previously (Jiang et demonstrated by the observation that exogenous al., 1993). To detect the rp L23a antisense transcripts, 1 mg expression of rp L23a (AS) inhibits cell growth. The of DNase-treated RNA was reverse transcribed with primers (5'-GGCCAAGAAGGCAGTGTTGA-3') speci®c present study indicates that the ability of IFNs to for the rp L23a (AS) transcripts, and PCR was performed inhibit cell growth can be partially attributed to the as described using the primers (5'-GGCCAAGAAGG- suppression of rp L23a expression. Further experi- CAGTGTTGA-3' and 5'-GTAGCATTAGTCCTCGGT- ments are necessary to determine the relation- CA-3' (Jiang et al., 1995b,c). As controls, total RNAs Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 479 were reverse transcribed with oligo(dT) primers and diameter plates at 46105/plate and transfected with 10 mg GAPDH cDNA fragments were ampli®ed with GAPDH- of the DNA plasmids (pREP4 vector and pREPrpL23a speci®c primers (5'-TCTTACTCCTTGGAGGCCATG-3' (AS)) using the lipofectin method (Gibco BRL) (Jiang et and 5'-CGTCTTCACCACCATGGAGAA-3') (Jiang et al., 1995c, 1996). Forty-eight hours later, the cells were al., 1995b,c). RNA blots were quantitated by densito- selected with hygromycin (100 mg/ml; Sigma) for approxi- metric analysis using a Molecular Dynamics densitometer mately 3 weeks. The plates were ®xed with formaldehyde (Sunnyvale, CA) (Jiang et al., 1993b). solution and stained with methylene blue solution. Colonies containing 550 cells were counted. Nuclear run-on assays Isolation of nuclei, nuclear run-on reactions, and extrac- tion of radioactively labeled nuclear RNAs were performed as described (Jiang et al., 1993b). Sense and antisense Abbreviations GAPDH and rp L23a transcripts were prepared by in vitro IFN: interferon; rp: ribosomal protein; rp L23a: ribosomal transcription and blotted onto nylon membranes which protein L23a; rp L23a antisense: rp L23a (AS); mda: were then hybridized to the radioactively labeled nuclear melanoma dierentiation associated. RNAs. After hybridization, the membranes were washed and exposed for autoradiography.

Acknowledgements DNA transfection assays This work was supported in part by National Cancer rp L23a expression vectors were constructed by cloning Institute Grant CA35675, the Samuel Waxman Cancer the rp L23a cDNA into the pREP4 vector (Invitrogen) Foundation and the Chernow Research Endowment. PBF downstream of a Rous sarcoma virus (RSV) promoter in is a Chernow Research Scientist in the Departments of an antisense (pREPrpL23a (AS)) orientation using Pathology and Urology. JT was a visiting student from previously described techniques (Jiang et al., 1995c). the Bronx High School of Science, Bronx, New York HeLa cells were seeded as triplicate groups in 100 mm- 10468.

References

Abdollahi A, Lord KA, Homan-Liebermann B and Gutterman JU. (1994). Proc. Natl. Acad. Sci. USA, 91, Liebermann D. (1991). Cell Growth Dier., 2, 401 ± 407. 1198 ± 1205. Agrawal MG and Bowman LH. (1987). J. Biol. Chem., 262, Harel-Bellan A, Brini AT and Farrar WL. (1988). J. 4868 ± 4875. Immunol., 141, 1012 ± 1017. Belhumeur P, Lanoix J, Blais Y, Forget A, Steyaert A and Hassel BA, Zhou A, Sotomayer C, Maran A and Silverman Skup D. (1993). Mol. Cell. Biol., 13, 2846 ± 2857. RH. (1993). EMBO J., 12, 3297 ± 3304. Birnbaum M, Trink A, Shainberg A and Saizberg S. (1990). Hunter T. (1993). Nature, 366, 114 ± 116. Dierentiation, 45, 138 ± 145. Jacobs FA, Bird RC and Sells BH. (1985). Eur. J. Biochem., Caarelli E, Fragapane P, Gehring C and Bozzoni I. (1987) 150, 255 ± 263. EMBO J., 6, 3493 ± 3498. Jiang H and Fisher PB. (1993). Mol. Cell. Dier., 1, 285 ± Celano P, Berchtold CM, Kizer DL, Weeraratna A, Nelkin 299. BD, Baylin SB and Casero RA Jr. (1992). J. Biol. Chem., Jiang H, Su Z-Z, Boyd J and Fisher PB. (1993a). Mol. Cell. 267, 15092 ± 15096. Dier., 1, 41 ± 66. Chan Y-L and Wool IG. (1995). Biochim. Biophys. Acta, Jiang H, Waxman S and Fisher PB. (1993b). Mol. Cell. 1260, 113 ± 115. Dier., 1, 197 ± 214. Chatterjee D and Savarese TM. (1992). Cancer Chemother. Jiang H, Lin J, Su Z-Z, Herlyn M, Kerbel RS, Weissman BE, Pharmacol., 30, 12 ± 20. Welch DR and Fisher PB. (1995b). Oncogene, 10, 1855 ± Chiang M-Y, Chan H, Zounes MA, Freier SM, Lima WF 1864. and Bennett CF. (1991). J. Biol. Chem., 266, 18162 ± Jiang H, Lin J, Young S-M, Goldstein NI, Waxman S, 18171. Davila V, Chellappan SP and Fisher PB. (1995a). Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, Oncogene, 11, 1179 ± 1189. 156 ± 159. Jiang H, Lin JJ, Su Z-Z, Goldstein NI and Fisher PB. Darnell JE Jr, Kerr IM and Stark GR. (1994). Science, 264, (1995c). Ongogene, 11, 2477 ± 2486. 1415 ± 1421. Jiang H, Su Z-Z, Lin JJ, Goldstein NI, Young CSH and El-Baradi TTAL, Raue HA, de Regt VCHF, Verbree EC and Fisher PB. (1996). Proc. Natl. Acad. Sci. USA, 93, 9160 ± Planta RJ. (1985). EMBO J., 4, 2101 ± 2107. 9165. Fisher PB and Grant S. (1985). Pharmacol. Ther., 27, 143 ± Kao H-T and Nevins JR. (1983). Mol. Cell. Biol., 3, 2058 ± 166. 2065. Fisher PB, Prignoli DR, Hermo H Jr, Weinstein IB and Khochbin S and Lawrence J-J. (1989). EMBO J., 8, 4107 ± Pestka S. (1985). J. Interferon Res., 5, 11 ± 22. 4114. Frohman MA, Dush MK and Martin GR. (1988). Proc. Krystal G, Birrer M, Way J, Nau M, Sausville E, Thompson Natl. Acad. Sci. USA, 85, 8998 ± 9002. C, Minna J and Battey J. (1988). Mol. Cell. Biol., 8, 3373 ± Garbe C and Krasagakis K. (1993). J. Invest. Dermatol., 100, 3381. (Suppl 2), 239s ± 244s. Legha SS. (1989). Seminars in Oncology, 1, Suppl 1, 34 ± 44. Genetics Computer Group. (1991). Program Manual. Levy S, Avni D, Hariharan N, Perry RP and Meyuhas O. Giovanella BC, Stehlin JS, Santamaria C, Yim SO, Morgan (1991). Proc. Natl. Acad. Sci. USA, 88, 3319 ± 3323. AC, Williams LJ, Leibovitz A, Fialkow PY and Mumford Mager WH. (1988). Biochim. Biophys. Acta, 949, 1±15. DM. (1976). J. Natl. Cancer Inst., 56, 1131 ± 1142. Meyuhas O, Thompson EA Jr and Perry RP. (1987). Mol. Greiner JW, Schlom J, Pestka S, Langer JA, Giacomini P, Cell. Biol., 7, 2691 ± 2699. Kusama M, Ferrone S and Fisher PB. (1985). Pharmac. Overman PF, Rhoads DD, Tasheva ES, Pyle MM and Roufa Ther., 31, 209 ± 236. DJ. (1993). Somatic Cell Mol. Genet., 19, 347 ± 362. Ribosomal protein L23a; (mda-20): a target of IFN-b growth suppression HJianget al 480 Pearson WR and Lipman D. (1988). Proc. Natl. Acad. Sci. Sreevalsan T, Rozengurt E, Taylor-Papadimitriou J and USA, 85, 2444 ± 2448. Burchell J. (1980). J. Cell. Physiol., 104, 1±9. Pestka S, Langer JA, Zoon KC and Samuel CE. (1987). Suzuki K and Wool IG. (1993). J. Biol. Chem., 268, 2755 ± Annu. Rev. Biochem., 56, 727 ± 777. 2761. Romeo G, Fiorucci G and Rossi GB. (1989). Trends Genet., Szyszka R, Kudlicki W, Kramer G, Hardesty B, Galabru J 5, 19 ± 24. and Hovanessian A. (1989). J. Biol. Chem., 264, 3827 ± Sen GC and Lengyel P. (1992). J. Biol. Chem., 267, 5017 ± 3831. 5020. Vester B and Garrett RA. (1984). J. Mol. Biol., 179, 431 ± Schein CH, Haugg M and Brenner SA. (1990). FEBS Lett., 452. 270, 229 ± 232. Wagner M and Perry RP. (1985). Mol. Cell. Biol., 5, 3560 ± Schein CH and Haugg M. (1995). Biochem. J., 307, 123 ± 3576. 127. Zhou A, Hassel BA and Silverman RH. (1993). Cell, 72, Sekar V, Atmar VJ, Joshi AR, Krim M and Kuehn GD. 753 ± 765. (1983). Biochem. Biophys. Res. Commun., 114, 950 ± 954. ZoonKC,KarasakiY,zurNeddenDL,HuRand Sheets MD, Ogg SC and Wickens MP. (1990). Nucleic Acids Arneiter H. (1986). Proc. Natl. Acad. Sci. USA, 83, Res., 18, 5799 ± 5805. 8226 ± 8230. Silverman RH. (1994). J. Interferon Res., 14, 101 ± 104.