Changes in gene expression during the growth arrest of ... - Nature

Oncogene (1998) 16, 2935 ± 2943  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc

Changes in gene expression during the growth arrest of HepG2 hepatoma cells induced by reducing agents or TGFb1 Andrea Cabibbo1, G Giacomo Consalez1, Milena Sardella1, Roberto Sitia1 and Anna Rubartelli2 1

Department of Biological and Technological Research (DIBIT), San Raaele Scienti®c Institute (HSR), Via Olgettina 58, 20132 Milano; 2Instituto Nazionale per la Ricerca sul Cancro L.go R.Benzi 10, 16132 Genova, Italy

The growth of hepatoma cells can be inhibited by treatment with TGFb1 or with exogenous reducing agents. To gain information on the molecular mechanisms underlying growth arrest, we visualized and compared gene expression pro®les of proliferating versus non proliferating HepG2 cells by computer-assisted gene ®shing, an improved technique of RNA ®ngerprinting that allows the selective ampli®cation of coding regions within transcripts. While many transcripts are selectively regulated by either treatment, a set of bands appear to be coordinately regulated by 2ME and TGFb1, suggesting their possible involvement in the mechanisms of growth arrest. Display tags corresponding to 18 dierentially expressed genes were cloned and, in most cases, identi®ed as known genes or, more frequently, as their homospeci®c/cross-speci®c homologues. A novel member of the kinesin superfamily was identi®ed amongst the genes induced by both 2ME and TGFb1. This gene, KIF3C, is upregulated in several cell lines undergoing growth arrest. Taken together, our ®ndings show that computer-assisted gene ®shing is a powerful tool for the identi®cation and cloning of genes involved in the control of cell proliferation and indicate that extracellular reducing agents can regulate cell growth through modulation of gene expression. Keywords: redox regulation; redox signalling; proliferation; RNA ®ngerprinting; placental type alkaline phosphatase; Cytokeratin K17

Introduction The extra- and intra-cellular redox potentials have been shown to play key roles in the regulation of cell proliferation and of other important cellular functions (Nakamura et al., 1997). Recent studies point to reactive oxygen species (ROS) as key signal transducers of the mitogenic signal in Ras transformed cells (Irani et al., 1997). Oxidants can also trigger activation of receptor (Knebel et al., 1996) and non-receptor tyrosine kinases (Bauskin et al., 1991), and are important regulators of the NF-kB transcription factor (Sen and Packer, 1996). A reversed perspective on the same paradigm can be achieved by focusing the attention on the biological activity of antioxidants and reducing agents. Thiol reagents block the heat shock response in human and rodent cells in culture (Huang

Correspondence: A Cabibbo Received 5 July 1997; revised 29 December 1997; accepted 6 January 1998

et al., 1994). Molecules such as 2-mercaptoethanol (2ME) or N-acetyl-cysteine inhibit the growth of the HepG2 human hepatoma cell line (Rubartelli et al., 1995) while stimulating the proliferation of normal and neoplastic B-lymphocytes (Iwata et al., 1994; Rubartelli et al., 1995). Many key regulatory molecules are sensitive to thiols and may act as redox sensors. Amongst extracellular proteins, Fibroblast Growth Factor-1 is released by the cells as an inactive, disul®de-linked dimer which must be reduced in order to acquire biological activities (Jackson et al., 1995). Owing to the presence of a critical redox sensitive site on its luminal domain (Sullivan et al., 1994), the Nmethyl-D-aspartate (NMDA) receptor may be present in synapses in two forms: a reduced, potentiated state and an oxidized, less active one. Treatment with reducing agents activates this calcium channel, while oxidation decreases its activity (Aizenman et al., 1989). Thiols may also act on intracellular proteins: for instance reducing agents increase the DNA binding activity of the thyroid transcription factor-1 (Arnone et al., 1995), while preventing the activation of NF-kB (Schreck et al., 1992). Fos and Jun DNA binding in vitro is regulated by reduction/oxidation of a single conserved cysteine residue (Abate et al., 1990). Moreover, zinc-®nger DNA binding proteins, in particular those belonging to the Sp-1 family, contain SH groups whose redox state controls the ability of these proteins to activate the transcription of target genes (Wu et al., 1996). The above examples constitute potential targets that could mediate the activity of exogenous thiols on cell proliferation. Indeed, many of these redox-sensitive molecules can modulate gene expression, either directly, as in the case of transcription factors, or indirectly, acting on various signal transduction pathways. We thus decided to monitor changes in gene expression in HepG2 cells possibly arising during the proliferative block induced by 2ME treatment. To this end, a computer-assisted RNA ®ngerprinting approach was used to visualize and compare the gene expression pro®les of actively proliferating and 2ME-growth arrested HepG2 cells. This technique utilizes, for the polymerase chain reaction (PCR), computer-selected single dodecamer oligonucleotides that are biased towards the ampli®cation of protein-coding regions of mRNAs (Consalez et al., unpublished). We present evidence that the growth arrest induced in the HepG2 hepatoma by reducing agents is paralelled by the upor down-regulation of a number of genes, some of which were partially cloned and identi®ed. Changes in gene expression have also been monitored in HepG2 cells in which growth arrest had been induced by exposure to TGFb1. A small set of genes seems to be regulated in a similar way following the two

Transcripts expressed in growth arrested HepG2 A Cabibbo et al

2936

treatments, suggesting their possible correlation with growth arrest.

Results Both 2ME and TGFb1 inhibit HepG2 cell proliferation

Figure 1 Inhibition of HepG2 cell proliferation by 2ME and TGFb1. HepG2 cells were cultured for 24 h in DMEM containing 1% Nutridoma HS, alone (NI) or supplemented with 1.4 mM 2ME or 5 ng/ml TGFb1 and counted. Means of counts from three dierent cultures+standard deviations are shown

In order to induce growth inhibition, HepG2 cells were cultured with or without 5 ng/ml TGFb1 (Lin et al., 1992) or 1.4 mM 2ME (Rubartelli et al., 1995). As expected, both treatments caused a proliferative arrest in HepG2 already evident after 24 h (Figure 1). The cell cycle analysis of HepG2 cells under the dierent conditions was performed by ¯ow cytometry after 24, 48 or 72 h of culture. As shown in Figure 2, the percentage of cells in G1 phase increased in both 2ME and TGFb1 treated cultures, while a concomitant reduction in the proportion of cells in S phase was observed. Under these conditions, no evident peaks corresponding to apoptotic cells (Darzynkiewicz et al., 1992) were detected even after 72 h of culture. Taken together these data suggest that 2ME and TGFb1 inhibit HepG2 cell proliferation by inducing a G1 block.

Figure 2 Cell cycle analysis of HepG2 cells upon treatment by 2ME and TGFb1. HepG2 cells were cultured in the presence or absence of 1.4 mM 2ME or 5 ng/ml of TGFb1 for 24, 48 or 72 h as indicated. Cells were then ®xed, stained with propidium iodide and the nuclear DNA content was analysed by ¯ow cytometry on a Epix Elite instrument as described in materials and methods. One representative experiment out of four is shown. The peak on the left corresponds to cells in G1, the one on the right to cells in G2 and the intermediate area to cells in S phase. Apoptotic cells would yield a peak to the left of the one corresponding to G1 cells (Darzynkiewicz et al., 1992)


Comparison of gene expression patterns of proliferating versus non proliferating HepG2 cells In order to visualize the gene expression patterns of HepG2 cells, before or after induction of growth arrest, we employed computer-assisted gene ®shing, an innovative technique of RNA ®ngerprinting. In this technique, the PCR reactions are performed with a set of single dodecamers which were selected by computer

simulations for being ecient PCR primers and for targeting preferentially the coding regions of mRNAs (Consalez et al., unpublished). To control RT or PCR artifacts, total RNA was always prepared from duplicate cultures, from either treated or untreated cells. One RT reaction corresponding to each RNA preparation was used as a template for many PCR ampli®cations, each performed with a dierent dodecamer. In a ®rst set of experiments, the

Figure 3 Changes in gene expression pro®les in growth-arrested HepG2. To monitor changes in gene expression pro®les in untreated, proliferating HepG2 cells or in cells in which growth arrest had been induced by either 2ME or TGFb1 treatment, total RNA was extracted from two control HepG2 cultures (the two left lanes of each group of six, indicated as -), two 2ME- (center lanes, 2ME) and two TGFb1(right lanes, TGF) treated cultures and the corresponding reverse transcription rections were used for PCR ampli®cations with dierent single oligonucleotides (24, 83, 121, 125, 130d, 143d, 188d) as described in the methods. Bands corresponding to prominent dierentially expressed transcripts are highlighted by arrows. M: MspI digest of pBR322 end-labelled with 32P-dCTP. At lower exposures of the autoradiogram, the inducibility of band 143D by TGFb1 is evident

2937


2938

transcriptional pro®les of untreated, proliferating HepG2 cells were compared to the pro®les of cells treated for 24 h with 2ME. In these experiments, 12 single and 13 degenerate primers were individually used for the PCR ampli®ctions, yielding a total of 2245 display products (data not shown). Out of these, 56 appear to be regulated by the 2ME treatment, 35 in a positive and 21 in a negative fashion. In total, it appears that about 2.5% of the transcripts are modulated by treatment with 2ME. A second set of experiments was performed, in which, in addition to 2ME, TGFb1 was also used as a growth arrest inducer for HepG2 cells. In this `triple display' experiment only the ten dodecamers that proved to be most ecient in the ®rst experiments were used, yielding a total of 877 displayed bands. A display with seven of these oligos is shown in Figure 3. With respect to controls, 45 bands appeared to be regulated by TGFb1, of which 30 in a positive and 15 in a negative fashion. A priori, by comparing treated with untreated cells, eight dierent qualitative regulatory categories can be envisaged: genes induced or repressed by 2ME only, by TGFb1 only, by both, induced by 2ME and repressed by TGFb1 and vice versa. As shown in Table 1, at least one transcript was identi®ed for each of these categories, except for the one comprising genes repressed by 2ME and induced by TGFb1.

lished), one random clone was sequenced also for bands 143B and 188A, for which none of the clones scored signi®cantly positive in the dot assay (not shown). Homologies of these sequences with previously cloned genes were searched through the BLAST algorithm (Altschul et al., 1990) on available databases, accessed through the National Center for Biotechnology Information (NCBI) BLAST network service. As shown in Table 2, the majority of the cloned sequences turn out to be either identical to known human genes, or to share signi®cant homologies to the coding sequences of genes from dierent species present in the databases. Among the clones isolated from this screen, two encode proteins related to the cytoskeleton, i.e. cytokeratin K17, induced by TGFb1, and a novel kinesin-related protein, KIF3C (Sardella et al., 1998), induced by both treatments; three clones encode proteins potentially involved in intracellular signalling: the tyrosine phosphatase 1C (down-regulated by 2ME), a protein similar to a G-protein coupled receptor and a protein similar to a G-protein b subunit, induced by TGFb1 and 2ME respectively.

Cloning and identi®cation of dierentially expressed genes Eighteen dierent PCR products derived from bands belonging to the dierent regulatory categories outlined in Table 1 were cloned. In order to avoid the cloning of false positive bands, which is one of the main problems of the classical display strategies, a prescreening of many individual clones for each band was performed by the `spot test' (Figure 4; see also Consalez et al., 1996). From each band, two clones that scored positive for induction in the dot test were sequenced. The clones contained inserts ranging from 150 ± 900 bp, which were fully sequenced. The two selected clones always turned out to be either identical or to contain the same insert in opposite orientations. To increase the information on ampli®cation of open reading frames by our primer set (Consalez et al., unpub-

Table 1 Dierentially expressed bands in HepG2 cells treated with 2ME or TGFb1 2ME + 7 = = + + 7 7 a

Treatmenta

TGFb1

DD-bandsb

= = + 7 + 7 + 7

19 6 27 7 3 1 0 7

Regulation pattern as observed in the RNA ®ngerprinting. + indicates an upregulation, 7 indicates a downregulation and = indicates no regulation of the band by 2ME or TGFb1. b This column indicates the number of bands belonging to each regulatory class out of a total of 877 bands inspected in the triple display experiments

Figure 4 Spot test analyses are essential to isolate truly dierentially expressed transcripts. Five or six inserts of individual transformants obtained from the cloning of four dierentially expressed bands (83-1; 121A; 143C; 143D) were spotted on duplicate or triplicate ®lters, which were then hybridized with the display reactions from which the dierent bands derive, as described in the Materials and methods


2939

Regulation of gene expression during HepG2 growth arrest For some of the cloned genes, Northern blot experiments were performed in order to con®rm the expression patterns displayed in the ®ngerprinting experiments. In all cases tested so far, the results of Northern blot and display/spot tests were superimposable. Three examples (121-A, cytokeratin 17 and placental-type alkaline phosphatase) of the correspondence between the patterns observed in ®ngerprinting and in Northern blots are outlined by the comparison between Figures 3 and 5. The sequence of 121-A reveals a signi®cant homology to a C. elegans open reading frame (Table 2) of unknown function. In the display, this gene seems to be strongly upregulated by 2ME and downregulated by TGFb1 (see arrow on Figure 3). This pattern is con®rmed by Northern blotting: in untreated HepG2 cells, a band of 2.1 kb is detected that is strongly enhanced by 2ME (Figure 5, lane 4) and downregulated by TGFb1 (lane 2). Hybridization of the same blot with the insert 143dD, corresponding to cytokeratin K17, yielded a barely detectable band of 1.9 kb in untreated and 2ME treated samples. A strong induction was evident in cells treated with TGFb1 con®rming the display patterns (Figure 3, see arrow). In the case of the placental type alkaline phosphatase (PLAP; 121E) we were unable to detect any signal in control or 2ME treated cells, either by display, which can be considered as a bona ®de semiquantitative RT ± PCR technique, or by Northern blotting. Conversely, both the display (Figure 3, see arrow) and the Northern blot experiments (Figure 5, lane 2) reveal a strong band in the TGFb1 treated samples. In Northern blots, the PLAP transcript migrates at the expected size of 3 kb.

Figure 5 Regulation of gene expression by 2ME and TGFb1 in HepG2 cells. Total RNA (10 mg) from untreated (7, lane 1) or TGFb1 treated cells (+, lane 2), or from untreated (7, lane 3) or 2ME treated cells (+, lane 4) was fractionated on a denaturing agarose-formaldehyde gel, transferred to a nylon support and hybridized with dierent probes: 121-A, cytokeratin K17, placental type alkaline phosphatase (PLAP) and Kif3c (band 83-1b2) all derived from bands cloned from the ®ngerprinting experiments (see Table 2). p50 (Z47747), p65 (Z22948) and I-kB (G699497) are human probes for subunits of the NF-kB transcription factor

Table 2 Cloning and identi®cation of dierentially expressed bands Banda

Sizeb

Regulationc (Spot test) 2ME TGFb1

24A 83 ± 1b2 83 ± 1w1 83 ± 2 83 ± 3 121A 121B 121C 121D 121E

550 458 458 270 266 875 795 456 371 187

+ + + + + + 7 + = =

ND + ND ND ND 7 = = + +

125d2 143dA 143dB 143dC 143dD 143dE 143dF 188A

244 850 700 650 420 172 700 358

+ = = = = + = =

= + = + + = 7 =

a

Identityd

Tyrosine phosphatase 1C (U15528) Alkaline phosphatase placental type (M15694)

Cytokeratin K17 (X62571) Thyroid hormone receptor (M24898)

Similarityd No match KiF3C (AF018164) No match No match Mouse gene E25 (L38971) C. elegans gene cm01g4 (U42437) C. elegans gene cm01g4 (U42437) No match Mouse transforming protein Int-3 (P31695) Leucocyte antigen CD97 precursor (P48960) inter-alpha trypsin inhibitor (D89287) G-protein coupled receptor (Z54306, gene B0457.1) G-protein beta subunit (U12232 and Z23105) Proline oxidase precursor (L07330)

Scored (blast) 1.6e7181 9.7e713 2.9e76 2.0e737 9.4e75 1.3e754 1.1e74 2.2e78 6.7e74 5.7e710 7.4e723 1.4e710 3.5e77 1.4e765

The name of each band includes the number of the oligo from which the band was derived. Some of the bands are highlighted in Figure 2. b Size of the inserts in base pairs. c Regulation pattern as revealed by the spot test. + indicates an upregulation, 7 indicates a downregulation and = indicates no regulation of the band by 2ME or TGFb1. In two cases (143dB and 188A) we were unable to identify positives among the analysed clones. ND means not determined because these bands derive from the initial experiments in which untreated cells were compared only with 2ME treated cells. d Identities normally score very high (e7204).


2940

Since the activation of the NF-kB transcription factor has been shown to be inhibited by 2ME (Schreck et al., 1992), we evaluated the regulation of the p50, p65 and IkB NF-kB subunits in our experimental system, at the level of RNA abundance. While there was a slight increase in p65 upon TGFb1 treatment, 2ME had no eects on the levels on any of the three NF-kB mRNAs tested (Figure 5). In order to investigate the kinetics of induction of dierentially expressed genes following incubation with 2ME, a time course experiment was performed and the expression of KIF3C and 121A was analysed by Northern blot. Figure 6 shows that 121A was already upregulated after 1 h from 2ME stimulation and its expression increased after 6 h. In contrast, a signi®cant increase of KIF3C was detected only after 6 h of treatment.

Kif3c expression is increased in dierent cell lines undergoing proliferative arrest The complete sequence of the transcript corresponding to the 83-1b2 band, which is induced by both 2ME and TGFb1 (Table 1), revealed that its product is a novel member of the kinesin superfamily, which we named KIF3C (Sardella et al., 1998). Northern blot analysis con®rm that the corresponding mRNA band is upregulated by both antiproliferative stimuli (Figure 5). In order to investigate whether the upregulation of KIF3C is restricted to HepG2 cells or is a more general feature of growth arrest, we analysed its expression in dierent cell lines following dierent antiproliferative treatments. Figure 7 shows that upregulation of KIF3C is also observed in the hemopoetic cell line HL60 upon treatment with DMSO or retinoic acid, which induce growth arrest and dierentiation toward the granulocyte lineage, but not by PMA, which promotes macrophage dierentiation of these cells (Collins, 1987). Moreover, growth inhibition and induction of neuronal dierentiation by NGF in the rat pheocromocytoma PC12 cells (Greene et al., 1976) and by retinoic acid in the SH-SY5Y neuroblastoma cell line (Sidell, 1982), result in increased expression of KiF3C transcripts (Figure 7). Discussion

Figure 6 Time course of expression of 121A and Kif3c upon 2ME treatment. Ten mg of total RNA (normalized by ethidium bromide staining) extracted from HepG2 cells treated with 1.4 mM 2ME for 0, 1 or 6 h as indicated, were electrophoresed on a denaturing agarose-formaldehyde gel, transferred to a nylon support and hybridized with a probe speci®c for 121A. The blot was then melted and re-hybridized with a Kif3c probe

Changes in gene expression are at the basis of many crucial physiological and pathological processes, such as dierentiation or neoplastic transformation. A number of genes potentially associated with tumor progression have been recently isolated by DD (Liang et al., 1995; Salesiotis et al., 1995; Francia et al., 1996; Wang et al., 1996).

Figure 7 Upregulation of Kif3c during growth arrest in dierent cell lines. Total RNA was extracted from PC12 phaeochromocytoma cells untreated or treated with Nerve Growth Factor (NGF), from SH-SY5Y neuroblastoma cells untreated or treated with retinoic acid (RA) and from HL60 cells treated with dimethylsulfoxide (DMSO), phorbol-myristate-acetate (PMA) or RA (see methods for details) and probed with Kif3c or gapdh in Northern blot experiments


In this study we have addressed a dierent problem, that is the changes of gene expression in tumor cells following treatments which inhibit cell proliferation (Schneider et al., 1988). To identify new genes associated with growth arrest, we have employed `computer-assisted gene ®shing' a modi®cation of the DD technique (Consalez et al., unpublished). This technique has several advantages over the classical DD, in which a major limitation is the frequent ampli®cation of untranslated regions, deriving from the use of poly-T primers together with random primers. Our RNA ®ngerprinting technique overcomes this problem by utilizing single dodecamers from a set of reagents selected by computer simulation for their eciency as PCR primers and for targeting preferentially open reading frames (Consalez et al., unpublished). This modi®cation results in a substantial improvement in the ampli®cation of translated portions of transcripts: indeed, as shown in Table 2, out of 18 cloned sequences, 14 were found to be identical or homologous to open reading frames present in the databases. Thus, the sequencing of a few hundred bases of the reampli®ed bands often leads to the rapid identi®cation of the corresponding gene. This eliminates the need of screening cDNA libraries and sequencing longer clones, that can later result in the identi®cation of already reported sequences. An essential feature of our RNA ®ngerprinting is the extreme reproducibility of the banding pattern, resulting from the use of single oligonucleotides (compare the duplicate lanes in Figure 3). Furthermore, the introduction of a S (A and T) or W (G and C) degeneration in the last position of some oligonucleotides, which generates a primer pair in which the two primers have the same expected melting temperature, leads to a consistent increase in the number of bands obtained in a single experiment. Last but not least, unlike in the classical approach, in which duplicate RT are performed on the same RNA sample, two individual RNA preparations from duplicate cultures were always used in our experiments: this allows one to rule out the cloning of bands arising from RT or PCR artifacts or from dierential DNA contaminations of the RNA samples (A.C., unpublished observations). A major problem of display experiments is that the cloning of dierentially displayed bands often results in the co-cloning of contaminating bands (Consalez et al., 1996). Figure 4 illustrates the importance of performing a pre-screening of several independent clones for each band, in order to isolate the truly dierentially expressed bands: the ratio of positive to negative clones is highly variable among the dierent bands, ranging from no positives (not shown) to all positives, e.g. band 143D, probably depending on how many bands were present in the region of the gel from which the band of interest was excised. We employed computer-assisted gene ®shing to identify genes dierentially expressed in actively proliferating and in growth arrested HepG2 cells after 24 h of culture with 2ME or TGFb1. This time point was chosen as it is slightly shorter than the doubling time of HepG2 cells (28 h, data not shown) and as after 24 h of treatment the antiproliferative eects of 2ME or TGFb1 are well evident with no sign of cell toxicity (Figure 2, and Rubartelli et al., 1995). We have

previously shown that HepG2 cells undergo growth arrest when cultured in the presence of exogenous reducing agents, such as 2ME or N-acetyl-cysteine (Rubartelli et al., 1995). A similar inhibition of proliferation is achieved by treating HepG2 cells with the inhibitory factor TGFb1. Both treatments block cells in the G1 phase of the cell cycle (see Figure 2): interestingly, although TGFb1 was shown to induce apoptosis in Hep3B hepatoma cells (Lin et al., 1992), we failed to detect DNA fragmentation in non-dividing HepG2 cells treated with either TGFb1 or 2ME. It is not easy to determine precisely the levels of redundancy in the display gels, that is transcripts yielding more than one band. In the case of bands 121A and 121C, the ampli®cation products appear to derive from two splicing variants of the same transcript (not shown). Assuming that a cell transcribes an average of 1.46104 genes at a given time (Zhang et al., 1997), the banding patterns that we compared in actively proliferating and growth arrested HepG2 cells should cover approximately 1/6 of their total transcript pool. About 2.5% and 5% of the bands were found to be dierentially expressed by treatment with 2ME and TGFb1, respectively. From these data we can extrapolate that a few hundred genes are modulated in HepG2 upon induction of growth arrest. These ®gures are higher than the estimated number of dierentially expressed genes (100) calculated from DD experiments in human breast cancer versus normal mammary epithelial cells (Liang et al., 1992). This may be due to a higher sensitivity of our protocol and to the higher number of bands that have been analysed in our experiments. Out of the 67 bands dierentially detected in 2ME or TGFb1 treated cells, most are either repressed or induced by the single treatment. Some of these bands likely correspond to gene products involved in the transduction of the anti-proliferative signals delivered by 2ME or TGFb1, respectively. Amongst the ®ve TGFb1-induced genes identi®ed, two are known genes: the cytokeratin K17 and the placental type alkaline phosphatase. The latter is a tumor marker (Narayanan, 1983); the ®nding that the expression of this gene can be modulated by TGFb1 may shed light on its regulation in tumor cells. The other three are new genes. Two of them were found to share similarities to receptor proteins: a 7-span transmembrane protein and a G-protein receptor respectively. Such receptor proteins are involved in the ®rst steps of signalling cascades, including those which regulate cell proliferation (Hamm and Gilchrist, 1996). Thus, these new genes may encode proteins playing some roles in the TGFb1-dependent signal transduction that leads to growth arrest (MassagueÂ et al., 1997). As many molecules involved in the control of gene expression, including a wide array of transcription factors, are sensitive to the redox conditions of the environment, the observation that up to 2% of the transcripts are regulated by 2ME is not surprising. Noteworthily, treatment with either 2ME or TGFb1 modulates a small set of common genes: this suggests that common signalling pathway(s), possibly correlated with the onset and/or the maintainance of growth arrest, can be activated or inhibited by the two antiproliferative treatments. One of the transcripts induced by both 2ME and TGFb1 has been identi®ed as a novel member of the kinesin superfamily. In view of its similarities with KiF3B, a murine kinesin

2941


2942

involved in vesicular transport along microtubules, this gene has been named KIF3C (Sardella et al., 1998). Although its precise role remains to be investigated, it is interesting to note that KIF3C transcripts are upregulated by various antiproliferative treatments in cells of dierent origins, suggesting a possible functional correlation of KIFC with growth arrest.

Materials and methods Cell cultures and cell cycle analysis HepG2 cells, derived from a well dierentiated human hepatocellular carcinoma, and the human myeloid leukemia cell line HL-60 (kind gift of Dr B Parodi, Genova, Italy) were cultured in DMEM or RPMI medium (Seromed, Milano, Italy) respectively, supplemented with 100 mg/ml each of penicillin and streptomycin (Seromed), 2.5 mM L-glutamine (Seromed) and 10% FCS (Gibco ± BRL, Life technologies, Milano, Italy). To monitor cell growth, HepG2 cells were plated at 0.16106 in 24-well plates in DMEM ± FCS. After overnight incubation, cells were washed and medium was replaced with 1 ml of DMEM ± 2.5 mM L-glutamine ± 1% Nutridoma HS (Boehringer Mannheim Italia, Monza, Italy) alone or supplemented with 1.4 mM 2ME or 5 ng/ml of TGFb1 as above. Cell number was assayed after 24 h of culture in the dierent conditions by counting in a Burker chamber as described (Rubartelli et al., 1995). Counts of triplicate cultures were performed. The number of viable cells was determined by Trypan blue (Sigma-Aldrich, Milano, Italy) dye exclusion. For cell cycle analysis, HepG2 cells were cultured as above in the presence or absence of 2ME or TGFb1 for dierent periods of time. At the end of incubation, cells were washed with phosphate buered saline (PBS), ®xed in 85% cold ethanol/15% PBS and kept at 7208C. Cells were stained in propidium iodide (50 mg/ml, Sigma) solution containing 0.1% RNase (Sigma) and the nuclear DNA content was analysed by ¯ow cytometry on a Epix Elite instrument (Coulter, Miami, Florida) as described (Darzynkiewicz et al., 1992). For RNA extraction, HepG2 cells were plated at subcon¯uence in 75 cm2 ¯asks in DMEM ± FCS. After overnight incubation, cells were washed and medium was replaced with DMEM-2.5 mM L-glutamine-1% Nutridoma HS alone or supplemented with 1.4 mM 2ME or 5 ng/ml of recombinant TGFb1 (generous gift of Dr Vincenzo Sorrentino, DIBIT, Milano). HL-60 cells were plated at 0.56106 ml in RPMI-1 mM L-glutamine-1% Nutridoma HS with or without 100 mM retinoic acid (RA, kind gift of F Tosetti, Genova) or 1.25% DMSO (Sigma) or 40 ng/ml of phorbol 12-myristate 13-acetate (PMA) as described (Collins, 1987). RNA was extracted after 24 h of culture. PC12 cells and SH-SY5Y cells were grown and dierentiated to a neuronal phenotype as described previously (Vignali et al., 1996). Computer-assisted gene ®shing RNA was extracted by the cesium chloride method (Sambrook et al., 1989) from duplicate cell cultures and treated with RNase-free DNase (RQ1 Promega, Madison, WI, USA). Reverse transcription (RT) reactions were carried out in a ®nal volume of 20 ml using a (dT)16 primer on 1 mg total RNA with M-MLV reverse transcriptase (Gibco ± BRL) according to the manufacturer instructions, in the presence of 4 units of RNasin (Promega). Radioactive PCR reactions were performed

from 1 ml of each RT reaction in 50 ml ®nal volume with individual 12-mers used at a ®nal concentration of 4 mM. The following oligonucleotides were used: 24: 5'-GGAGAAGCTGCC-3'; 121: 5'-AGCAGGGTCTGG-3'; 143deg: 5'-CAAGACG-GCCTG/C-3'; 188deg: 5'-CAGGTCCTGCAC/G-3'; 125: 5'-GCTGGTGGTGCT-3'; 130deg: 5'GGTGCTCAGCAG/C-3'; 83: 5'-GCGAAGGAGGAG-3'; 124deg: 5'-AG-CTTCGCCAGG/C-3'; 135deg: 5'-CGTGCTGCAGCG/C-3'; 95deg: 5'-TCGATGCCGCTC/G-3'; 188deg: 5'-CAG-GTCCTGCAG/C-3'; 24: 5'-GGAGAAGCTGCC-3'; 38: 5'-CGTGGATCCAGG-3'; 133deg: 5'CAGTCCTGGCC-A/T-3'; 138deg: 5'-CCTGTCCGTCCT/ A-3'; 128: 5'-TGG-TCGTCCACG-3'; 192: 5'-AGCCGGAGGA-TG-3'; 179deg: 5'-TGTTGCGGTGGC/G-3'; 212deg: 5'-CTC-TCCGATGCC/G-3'; 133deg: 5'CAGTCCTGGCCA/T-3'; 129deg: 5'-GCCAGCATGCTG/ C-3'; 27deg: 5'-GTGG-AGAGCTGC/G-3'; 123: 5'GCTTCCAGCAGC-3'; 122: 5'-CATGGCTGCCAG-3'; 4: 5'-GAGCAGACCCTC-3'. In some cases the oligonculeotides had a partial degeneration at the 3' position introduced with the aim of increasing the number of PCR products per primer (degenerate primers, Consalez et al., submitted). PCR conditions were as follows: 3 min at 948C, 2 min at 808C at which Taq polymerase was added (hot start), followed by 35 cycles of 40 s at 948C, 1 min at 508C, 1 min at 728C with a 1 s extension for each cycle, with a ®nal elongation step of 5 min at 728C. 0.2 ml [a32P]dCTP were added to each reaction, which was performed in a ®nal concentration of 1.5 mM MgCl2 and 0.4 mM dNTPs. Ampli®cation products were separated on a 5% denaturing polyacrylamide gel and visualized by autoradiography. Blunt-end cloning of dierentially displayed bands and `spot tests' Bands were excised from the gel and eluted by boiling in water for 10 min. Acrylamide debris were removed by centrifugation and the DNA was ethanol precipitated in the presence of glycogen at a ®nal concentration of 0.05 mg/ml. Bands were reampli®ed using the fingerprinting primers and cloned into a modi®ed pBluescript II SK + (Stratagene GmbH, Heidelberg, Germany) as described by Consalez et al. (1996). In order to reduce the background of non recombinant clones, after digestion and dephosphorylation, the bluescript vector was ligated on itself overnight at 168C in standard reaction conditions. After this ligation step, the linear form of the plasmid was puri®ed from an agarose gel and used for the ligation in the presence of the inserts. The ligation of the vector alone allows the circularization or concatenation of the residual phosphorylated molecules and permits to separate them from the linear unphosphorylated plasmid. In our hands, this procedure produces a vector which yields virtually no background of `empty' clones. Clones corresponding to dierentially displayed bands were selected from the background of unrelated products as described by Consalez et al. (1996): brie¯y, inserts of many (3 ± 8) independent clones for each band were ampli®ed from the pBluescript vector with T3 and T7 primers. One volume of 206SSC was added to these inserts, which were then denatured and spotted in duplicate or triplicate on two or three nylon ®lters pre-wetted in 106SSC. The ®lters were then crosslinked and hybridized with the original display rections re-labelled to high speci®c activity with a random priming kit (Decaprime II, Celbio, Milano, Italy). Hybridization and washes were performed in standard stringent conditions at 658C. Standard molecular biology techniques were performed as described (Sambrook et al., 1989).


Acknowledgements We are indebted with Francesca Navone and Giovanna Vignali (University of Milano, Italy) for invaluable help in performing the Northern experiments with PC12 and SHSY5Y cells. We thank Sara Costigliolo (IST, Genova, Italy) for technical assistance, Barbara Parodi (IST-CBA, Genova) for the kind gift of HepG2 and HL-60 cells, Vincenzo Sorrentino (DIBIT-HSR, Milano) for generously

providing us with TGFb1, Francesca Tosetti for retinoic acid and our colleagues at the Cytometry Unit (IST-CBA, Genova) for cell cycle analyses. This work was supported in part by grants from AIRC, CNR (target projects ACRO, progetto ®nalizzato biotecnologie and 5% biotecnologie) and Ministero SanitaÁ (target project: AIDS). AC is a recipient of a fellowship from AIRC.

References Abate C, Patel L, Rauscher FJ III and Curran T. (1990). Science, 249, 1157 ± 1162. Aizenman E, Lipton SA and Loring RH. (1989). Neuron, 2, 1257 ± 1263. Altschul SF, Gish W, Miller W, Myers EW and Lipman DJ. (1990). J. Mol. Biol., 215, 403 ± 410. Arnone MI, Zannini M and Di Lauro R. (1995). J. Biol. Chem., 270, 12048 ± 12055. Bauskin AR, Alkalay I and Ben-Neriah Y. (1991). Cell, 66, 685 ± 696. Collins S. (1987). Blood, 70, 1233 ± 1244. Consalez GG, Corradi A, Ciarmatori S, Bossolasco M, Malgaretti N and Stayton CL. (1996). Trends in Genetics, 12, 455 ± 456. Darzynkiewicz Z, Bruno S, Del Bino G, Gorczyca W, Hotz MA, Lassota P and Traganos F. (1992). Cytometry, 13, 795 ± 808. Francia G, Mitchell SD, Moss SE, Hanby AM, Marshall JF and Hart IR. (1996). Cancer Res., 56, 3855 ± 3858. Greene LA and Tischler AS. (1976). Proc. Natl. Acad. Sci. USA, 73, 2424 ± 2428. Hamm HE and Gilchrist A. (1996). Curr. Opin. Cell Biol., 8, 189 ± 196. Huang LE, Zhang H, Bae SW and Liu AYC. (1994). J. Biol. Chem., 269, 30718 ± 30725. Irani K, Xia Y, Zweier JL, Sollott SJ, Der CJ, Fearon ER, Sundaresan M, Finkel T and Goldschmidt-Clermont PJ. (1997). Science, 275, 1649 ± 1652. Iwata S, Hori T, Sato N, Ueda-Taniguchi Y, Yamabe T, Nakamura H, Masutani H and Yodoi J. (1994). J. Immunol., 152, 5633 ± 5642. Liang P and Pardee AB. (1992). Science, 257, 967 ± 970. Liang P, Averboukh L, Keyomarsi K, Sager R and Pardee AB. (1992). Cancer Res., 52, 6966 ± 6968. Liang P, Bauer D, Averboukh L, Warthoe P, Rohrwild M, Muller H, Strauss M and Pardee AB. (1995). Methods Enzymol., 254, 304 ± 321. Lin Jk and Chou CK. (1992). Cancer Res., 52, 385 ± 388.

Jackson A, Tarantini F, Gamble S, Friedman S and Maciag T. (1995). J. Biol. Chem., 270, 33 ± 36. Knebel A, Rahmsdorf HJ, Ullrich A and Herrlich P. (1996). EMBO J., 15, 5314 ± 5325. MassagueÂ J, Hata A and Liu F. (1997). Trends Cell Biol., 7, 187 ± 192. Nakamura H, Nakamura K and Yodoi J. (1997). Annu. Rev. Immunol., 15, 351 ± 369. Narayanan S. (1983). Annal. Clin. Lab. Sci., 13, 133 ± 136. Rubartelli A, Bonifaci N and Sitia R. (1995). Cancer Res., 55, 675 ± 680. Salesiotis AN, Wang CK, Wang CD, Burger A, Li H and Seth A. (1995). Cancer Lett., 91, 47 ± 54. Sambrook J, Fritsch EF and Maniatis T. (1989). Molecular Cloning: a Laboratory Manual. 2nd edition. Cold Spring Harbor Laboratory Press. Sardella M, Navone F, Rocchi M, Rubartelli A, Viggiano L, Vignali G, Consalez GG, Sitia R and Cabibbo A. (1998). Genomics, 47, 405 ± 408. Schreck R, Meier B, Mannel DN, Droge W and Baeuerle PA. (1992). J. Exp. Med., 175, 1181 ± 1194. Schneider C, King RM and Philipson L. (1988). Cell, 54, 787 ± 793. Sen CK and Packer L. (1996). FASEB J., 10, 709 ± 720. Sidell N. (1982). J. Natl. Cancer Inst., 68, 589 ± 593. Sullivan JM, Traynelis SF, Chen HS, Escobar W, Heinemann SF and Lipton SA. (1994). Neuron, 13, 929 ± 936. Vignali G, Niclas J, Sprocati MT, Vale RD, Sirtori C and Navone F. (1996). Eur. J. Neurosci., 8, 536 ± 544. Wang FL, Wang Y, Wong WK, Liu Y, Addivinola FJ, Liang P, Chen LB, Kanto PW and Pardee AB. (1996). Cancer Res., 56, 3634 ± 3637. Wu X, Bishopric NH, Disher DJ, Murphy BJ and Webster KA. (1996). Mol. Cell. Biol., 16, 1035 ± 1046. Zhang L, Zhou W, Velculescu VE, Kern SE, Hruban RH, Hamilton SR, Vogelstein B and Kinzler KW. (1997). Science, 276, 1268 ± 1272.

2943