Effect of the 3′-untranslated region on the expression levels and rnRNA stability of α1(I) collagen gene

~,

:

BB

15

ELSEVIER

Biochimica et Biophysica Acta 1260 (1995) 294-300

Biochi~ic~a et BiophysicaA~ta

Effect of the 3'-untranslated region on the expression levels and mRNA stability of a l(I) collagen gene Arto Miiiitt~i *, Erika Ekholm, Risto P.K. Penttinen Department of Medical Biochemistry, University of Turku, FIN-20520 Turku, Finland Received 1 June 1994; accepted 29 September 1994

Abstract

Changes in the synthesis of type I collagen, a major extracellular matrix component in skin and bones, are associated with both normal growth or repair processes and with several pathological conditions such as lung fibrosis and liver cirrhosis. The expression of the c~1(1) collagen gene is regulated by transcriptional and post-transcriptional mechanisms. Regulation at both these levels are usually utilised when extensive changes occur in collagen synthesis• We constructed plasmids carrying the whole or partially deleted 3'-UTR sequences of the c~1(1) collagen gene, fused to two hGH exons and to the promoter of the c~1(I) collagen gene. A control plasmid contained the 3'-UTR of the hGH gene. In transient transfections into Rat-1 fibroblasts, no significant differences between plasmids were found, which suggests that although 3'-end of the gene has been shown in previous studies to contain DNaseI hypersensitive sites and to bind sequence-specific nuclear proteins it does not seem to function as a transcriptional regulator. This was further supported by the finding that TGF-/3 treatment induced a 2.5-fold expression of hGH mRNA from plasmids containing collagen promoter and either hGH or a l(I) collagen 3'-UTR. In stable transfections, mRNAs using the first polyadenylation site were not as stable as those transcibed from the endogenous a 1(I) collagen gene. We suggest that the 3'-UTR alone may not be sufficient to determine the stability of the shorter a 1(I) collagen mRNA species. Keywords: a 1(I) Collagen; Gene expression; mRNA stability

I. Introduction

Type I collagen, the major constituent of the extracellular matrix and the most abundant human protein. During embryonic development and growth its expression is controlled in a tissue and time-dependent manner [1-3]. In adults, collagen synthesis is repressed except when activated during repair of tissue trauma [4-6] and in fibrotic conditions such as scleroderma and cirrhosis [7]. The regulatory regions in the promoter, in the first intron, and in the more upstream sequences have been recently studied [1,8,9], but less attention has been paid on the role of the rest of the gene. For example, DNase I hypersensitive sites

Abbreviations: CAT, chloramphenicol acetyltransferase; COLIA1, human pro ~1(I) collagen gene; CTF, chicken tendon fibroblasts; OADPH, glyceraldehyde-3-phosphate dehydrogenase; hGH, human growth hormone; SSC, 0.15 M sodium chloride, 0.015 M sodium citrate. * Corresponding author. Present address: Centre for Biotechnology, University of Turku, Turku, Finland. Fax: +358 21 6338000. 0167-4781/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved

SSDI 0 1 6 7 - 4 7 8 1 ( 9 4 ) 0 0 2 0 7 - X

have been found in the 3'-UTR and flanking regions [10,11] but their functions are not known. Furthermore, recent evidence indicataes that sequences mediating downregulation by Ras reside in the coding region or far downstream flanking region of the gene [12]. Messenger RNA stability is probably a common factor in the expression of collagen genes. Particularly, when a rapid change occur in the expression, a concerted alteration is detected both in the transcription rate and in the m R N A stability [13-15]. The low rate of type I collagen synthesis in malignant transformation is due to both decreased tran~ scription and stability of the m R N A [12,16]. The a l ( I ) collagen m R N A stability is also regulated by cell adhesion. Re-attachment of suspension cultured fibroblasts increases both transcriptional activity and m R N A stability of collagen [17], whereas the opposite is seen in fibroblasts grown in collagen gels [18]. Several m R N A s have sequences controlling stability in their 3'-UTR [19]. The rapid turnover of lymphokine and c-los m R N A s is, in part, mediated by AU-rich sequences of the 3'-UTR [20,21], and stem-loop secondary structures

295

A. Miiiittii et al. / Biochimica et Biophysica Acta 1260 (1995) 294-300

are involved in the regulation of transferrin receptor and transferrin synthesis [22]. In a previous report we presented the sequence of the a 1(I) collagen 3'-UTR with two highly conserved regions flanking the two alternate polyadenylation sites pAl and pA2 [23]. Our goal was to find out whether the presence of several consensus sequences for transcription factors at the 3'-UTR [23] confers an effect on the expression of the gene. Another goal was to study the effect of the 3'-UTR on the stability of a l(I) collagen mRNAs. For these purposes we constructed plasmids containing sequences of the a 1(I) collagen 3'-UTR and deleted defined fragments from the full length 3'-end. The collagen 3'-UTR carrying plasmids were used in stable and transient transfection studies and the corresponding reporter hGH RNA was quantified by Northern or solution hybridisations.

A Plasmid

Structure

E-hGH-2.2

~

mRNA size I

0.9 + (A)n 2.0 + (A)n

E-hGH-0.7

~

I -NN,//~-

E-hGH-AMS

0.9 + (A)n

~

I

I

0.9 + (A)n 1.55 + (A)n

E-hGH-ASA

~

| "NN¢/~

0.9 + (A)n

E-hGH 0.7 + (A)n

2. Materials and m e t h o d s

2.1. Plasmid constructions For the reporter gene, a 620 bp HindllI-PvulI fragment of the human growth hormone (hGH) gene containing the fourth and most of the fifth exon and excluding the polyadenylation site was inserted into the HindlII-SmaI digested plasmid plC19 (plC plasmids bear an extended polylinker compared to pUC plasmids from which they have been developed). A 2.3 kb HindlII-HindllI fragment of a subclone R26 of the cosmid clone CG 103 [10] containing the promoter, the first exon and most of the first intron of human c~1(I) collagen gene was ligated into the HindlII site immediately 5' to hGH sequences. A 3.0 kb EcoRI-EcoRI fragment of the resultant plasmid consisted of the COL1A1 promoter/enhancer-hGH elements and was then inserted into the EcoRI site of the collagen 3'-end plasmids described below. The following strategy was used to obtain the COL1A1 3'-end plasmids. A 5.7 kb EcoRI-EcoRI fragment of the clone CG 103 contains the 3'-sequences of COL1A1 gene starting from the EcoRI site of the last exon. A 0.3 kb TaqI/HindlII fragment in the 5.7 kb EcoRl subclone starting 22 bp upstream from the stop codon includes the first polyadenylation site. The following 1.9 kb HindlII/HindlII fragment contains the second polyadenylation site and about 900 bases of 3' flanking sequences. These 0.3 kb TaqI/HindlII and 1.9 HindlII/HindlII fragments were ligated into pUC19 to produce plasmid pUC 2.2. In the subcloning procedure, intermediate pIC plasmids were used to avoid excess polylinker sequences between the 3'-end and the reporter gene in the final minigene plasmids. Plasmid pUC 0.7 was constructed by digesting the pUC 2.2 plasmid with MluI, filling the ends with the Klenow enzyme, and cutting a 0.7 kb fragment with TaqI. This fragment was inserted first in plC 19 R and therefrom to the EcoRI site of pUC 19. The plasmids

B I a b c d

-

I 0.15 hGH riboprobe 0.68 HindlII-EcoRI 2.3 EcoRI-EcoRI 1.9 HindlI/-HindIIl

Fig. 1. Structure of the COL1AI-hGH minigenes used in transfections. (A) The structure, size, predicted transcript and predicted size of the processed transcript are schematically presented (not drawn in scale). The plasmids were constructed using standard methods as described in the text. Open rectangles represent human growth hormone gene sequences starting from a Hindlll site in the 3rd intron; the hatched rectangle represents a 2.3 kb HindlII fragment (nucleotides - 8 0 4 to + 1442) encompassing o~1(I) collagen promoter and the first intron; bold black line COL1A1 3'-UTR and flanking sequences from TaqI site 22 nucleotides before the stop codon; thin lines deletions and plasmid sequences and; black vertical bars polyadenylation sites. The predicted mRNA transcribed from the plasmid is presented with a thin line under each construction. The introns and deletions are indicated with V-shaped interruptions. The abbreviations of the shown restriction sites are as follows: A = Accl, H = HindlII, M = MluI, N = Nsil, P = PstI, Sau = Sau3A1, Sm = SmaI, St = Stul, T = Taql. (B) The DNA and RNA-probes used for the determination of the copy number and the integrity of the plasmids in the stably transfected polyclonal cells and the structure and abundance of the respective mRNAs. The open box in probe (a) indicates the T7 RNA-polymerase promoter used for in vitro RNA synthesis.

pUC 0.7 and pUC 2.2 were then linearized with EcoRI and fused with the 3.0 kb EcoRI promoter-enhancer-reporter cassette described above to constitute minigene plasmids pEhGH-0.7 (6.5 kb) and pEhGH-2.2 (8.1 kb) (Fig. 1). In some transfection experiments into CTF cells plasmid pEhGH-2.4 was used instead of pEhGH-2.2. This plasmid, which was constructed before any other minigene plasmids, contains additional 25 bp of pBR 322 sequences and 177 bp of the COL1A1 5'-flanking region between the collagen promoter reporter gene and COL1A1 3'-UTR. Deletions AMS (338 bp) and ASA (205 bp) were produced by digesting the pUC 2.2 plasmid with MluI and StuI, or SmaI and AccI, followed by blunt ending and

296

A. Miiiittiiet al. / Biochimica et BiophysicaActa 1260 (1995) 294-300

ligation. The control plasmid pEhGH contains in addition to the 0.68 kb reporter gene cassette the intact 3'-UTR, polyadenylation signal and about 200 bp of the 3'-flanking DNA from the hGH gene. The 2.3 kb COL1A1 promoterenhancer cassette was ligated to a HindlII site immediately 5' of the hGH sequences. The plasmid pEH' is identical to the pEhGH except that it contains the whole first intron of the COL1A1 gene until a SmaI site 160 bp downstream from the HindlII site where the E-fragment ends. In transient transfections, both pEhGH and pEH' give identical results. The SV-CAT-SH plasmid was constructed by inserting a Sau3A1-HindlII (blunt ended) fragment downstream from the latter polyadenylation site of the a 1(I) collagen gene into BamHI-SalI (blunt ended) opened pCATpromoter (Promega) plasmid. The fragment was, thus, in correct oritentation in a enhancer polylinker downstream of the CAT gene. 2.2. Stably transfected polyclonal cell populations Rat-1 fibroblasts were grown in DMEM supplemented with 10% calf serum on 100 mm dishes (Nunc, Denmark). Subconfluent cultures were co-transfected by the calciumphosphate co-precipitation method as described by Bornstein and McKay [24] with 10 /zg of the described COLhGH minigene plasmids and 1 /xg of the plasmid pBGS carrying a SV-40 promoter driven neomycin phosphotransferase gene (a gift from Dr. Bruce Gragner, University of Montana, USA). Neomycin resistant colonies were selected in 5 0 0 / x g / m l G418 (Sigma) and after 2 to 4 weeks the colonies (over 50 in each transfection) were pooled, expanded and tested for the presence of COL-hGH minigene DNA and RNA. In all analyses described in this paper, pooled cells between the second and third passage were used. RNA degradation rates were measured by application of 60 ~ M DRB (5,6-dichlorobenzimidazoleriboside, Sigma) to cells and isolation of total RNA in indicated time points. 2.3. Transient transfections Rat1 cells, NIH-3T3 cells or chicken tendon primary fibroblasts (CTF) grown overnight on 100 mm diameter dishes were transfected for 16-24 h as in the production of the stable cell populations. Transfection efficiency was estimated by co-transfection with a plasmid containing /3-galactosidase gene driven by RSV promoter (1-3 /zg). The cells were washed with PBS, fed with 10 ml of fresh medium, chased for 24 h and split for the determination of /3-galactosidase activity and isolation of total RNA [24,25]. For the measurement of the RNA transcribed from the transfected plasmids, a pGEM-plasmid carrying a 0.15 kb BgllI-PvulI fragment of the fifth exon of human growth hormone gene was linearized and antisense riboprobes were synthesized using the TransProbe T kit (Pharmacia,

Sweden/USA). The protection assays for the RNA samples from transiently transfected cells were carried out as described [24]. The CAT activities were measured by a phase partitioning method as desribed by Sherwood and Bornstein [26] from cell lysates obtained by rapid freezing and thawing of the cells. 2.4. Southern blots Genomic DNA from the stably transfected cell populations and control DNA from human peripheral leucocytes (10 /xg) was digested with HindlII, size fractionated in an 1% agarose gel, transferred to Gene Screen Plus nylon membranes and hybridised at +65°C according to the manufacturer's recommendations with a 0.68 kb HindllI/PvulI fragment encompassing the hGH reporter gene (probe b in the Fig. 1B); a 2.3 kb HindlII promoterenhancer fragment of the COL1A1 gene (probe c, Fig. 1B); and an 1.9 kb HindlII fragment from the 3'-UTR and flanking genomic sequences of the COL1A1 gene (probe d, Fig. 1B). The filters were washed at +65°C with 2 × SSC, 1% SDS for 30 min and in some cases further with 0.2 × SSC, 0.1% SDS at +65°C. The probes were detected by autoradiography on Kodak XAR films at - 7 0 ° C using intensifying screen and the signals were quantitated by laser densitometry (Ultroscan, LKB, Sweden). 2.5. Northern blots Total RNA from polyclonal cell populations was isolated essentially as described [25] and samples corresponding 15 /zg of total RNA were fractionated in 1% agarose/formaldehyde gels, transferred on Gene Screen Plus membranes and sequentially hybridised with the 0.68 kb hGH probe; pMCOL1, a cDNA for murine pro a(I) collagen [27]; and with a rat GAPDH probe [28]. The filters were processed like the Southern blots, except that 50% deionised formamide was included in the prehybridisation and hybridisation buffers and the temperature was 42°C. For the rehybridisations, the previous probes were removed from the filters by boiling in 0.1% SSC, 0.1% SDS for 15 min.

3. Results and discussion

3.1. Construction of plasmids to study a l (I) collagen gene expression To investigate, whether the 3'-UTR sequences are involved in the regulation of the a l(I) collagen mRNA levels, we constructed a series of expression plasmids bearing a l(I) collagen promoter and different fragments of the 3'-UTR of the human a 1(I) collagen gene (Fig. 1A).

A. Miiiittii et al. / Biochimica et Biophysica Acta 1260 (1995) 294-300

We used the collagen promoter for the following reasons. Firstly, it is a strong constitutive promoter in cultured fibroblasts [8,29]; secondly, it has successfully been used in experiments with the same reporter gene [3,12,24]; and finally, we assumed that the concomitant use of the promoter with the 3'-UTR of the same gene would better mimic the physiologic al regulation than the use of other promoters. The activity of the a 1(I) collagen promoter and first intron sequences have, furthermore, recently been evaluated in stably transfected cell lines by several groups [9,30]. Pavlin et al. [9] found no difference in the promoter activity in stably transfected Rat-1 cells, when either a promoter construct including 3 kb 5'-genomic DNA or 0.8 kb promoter DNA used in this work. The first intron of the minigenes is a hybrid intron containing sequences from both a l ( I ) collagen and hGH gene. It contains an AP-1 like sequence, which has been shown to enhance collagen transcription by 4-5-fold [8]. In the control pEhGH plasmid, the 3'-UTR, polyadenylation signal, and 3'-flanking DNA are those of the hGH gene, whereas respective elements of the a l ( I ) collagen gene were used in all other constructs. Plasmid pEhGH-2.2 bears a full length 3'-UTR coding for both c~1(I) collagen mRNAs whereas plasmid pEhGH-0.7 contains only the first polyadenylation site and the rest of the first highly conserved region. Plasmid pEhGH-ASA contains a deletion that removed the latter polyadenylation site and conserved region [23]. Plasmid pEhGH-AMS has a 338 bp deletion in the middle of the 1.1 kb separating the polyadenylation sites. The AMS deletion removes also one of the a DNaseI hypersensitive sites of the u 1(I) collagen locus [10].

3.2. Transient transfections in fibroblasts We conducted first transient transfections in Rat 1, 3T3, and CTF fibroblasts with the plasmids to find out, whether the 3'-UTR sequences contribute to the steady state mRNA levels of the a l(I) collagen gene. Previously, several consensus sequences of known transcription factors were detected in the region between the alternate polyadenylation sites of a l ( I ) collagen gene [23]. Results of three independent transfections into Rat 1 cells are presented in Fig. 2A. No significant differences were found between the plasmids. The amounts of hGH mRNA were somewhat higher when plasmids pEhGH or pEhGH-ASA were used, but within the limits of normal experimental variation. Similar results were obtained with transfections into 3T3 cells (not shown). In some CTF transfections the collagen 3'-UTR carrying plasmids yielded less hGH RNA than the pEhGH or pEH' plasmids but the difference varied and was not detected in all experiments. Constant results were obtained when the effect of TGF-fl was investigated on the RNA levels in the CTF transfections. TGF-/3 increased the reporter gene RNA levels approx. 2.5-times (Fig. 2B) with both control (pEH') plasmid and collagen 3'-UTR (pEhGH-2.4) plasmids. As similar increase in mRNA lev-

A

297

2.5

~1.5 a~

0.5

pEhGH-2.2 pEhGH-0.7 pEhGH*aSA pEhGH-~MS

B

pEhGH

3

J gg

2.5 2

1.5. 1 0.5, 0. control

TGF.13

Fig. 2. Reporter gene RNA levels in transient transfections in Rat-1 cells. (A) The RNA levels were determined using RNase protection with antisense RNA probe directed to the 5th exon of the hGH gene. The values obtained with densitometric quantitation of hybridisation signals were corrected with fl-galactosidase activity expressed by a co-transfected RSV-/3-galactosidase plasmid. Results from three independent transfections including all the plasmids with two to three replicates are summarised. The average expression from the plasmid pEhGH-2.2 (the one bearing the full length 3'-UTR) is designated to 1.00 and all other values are reported in relation to that. Error bars indicate S.E. (B) The CTF cells were chased after the transfection in the presence or absence of 100 pM TGF-/3 in depleted medium [15]. Data from five independent measurements for both plasmids are presented. The control values were adjusted to 1.00. The S.E. for the amount of induction was 0.3 for the pEH' and 0.5 for pEhGH-2.4. The plasmids pEhGH-2.2 and pEhGH-2.4 (see Materials and methods) act equally in the transfection analysis.

els after TGF-fl stimulation were obtained in transfections with or without the COLIA1 3'-UTR we suggest that in these experiments the promoter/first intron sequences have mediated the effect. This result is consistent with the notion that in subconfluent cells, TGF-/3 acts mainly at the transcriptional level [15]. Interestingly, our constructs lack the upstream TGF-/3 responsive element [31], indicating that several promoter elements might contribute to the transcriptional enhancement of TGF-fl. The maximal induction in the endogenous a 1(I) collagen mRNA levels by TGF-/3 has been shown to be over 20-fold in confluent cells made quiescent by culturing in growth factor depleted serum [15], whereas in transient transfections using subconfluent cells the enhancement is only up to 5-fold [31]. Thus, it remains possible that there are additional elements elsewhere in the gene or other mechanisms that are responsible for the maximal induction in confluent cells. Plasmid SV-CAT-SH was used to evaluate the effect of a region encompassing a DNaseI hypersensitive site in the

298

A. Miiiittii et al. / Biochimica et Biophysica Acta 1260 (1995) 294-300

flanking DNA downstreeam from both polyadenylation sites [10]. No differences in the chloramphenicol acetyltransferase activity was found in transfections compared to a c o n t r o l p l a s m i d c o n t a i n i n g no e n h a n c e r s (pCATpromoter). When mean of the control values corrected with transfection efficiency was adjusted to 1.00 (six measurements, standard error 0.24) the value for the plasmid bearing the Sau3A1-HindlII fragment was 0.81 + 0.14. Results of the CAT-assays are, thus, consistent with those of the transient transfections with collagen-hGH minigenes. We concluded from these experiments that the 3' UTR sequences used in our plasmid constructs have only a minor effect on the transcriptionally induced steady state expression levels in proliferating Rat 1 fibroblasts. This is in accordance with the finding that t~l(I) collagen promoter sequences alone are sufficient for high-level tissue specific expression in transgenic mice [3,9]. These sequences might be responsible for additional control of the expression of a 1(I) collagen gene in cell types other than embryonic fibroblasts in which the expression of COL1AI gene is constantly at a high level. Furthermore, growth factors and cytokines modulating c~l(I) collagen mRNA levels may have different effects if the collagen 3'-UTRis present in the studied plasmids. Earlier, we have found a clear decrease in the hGH reporter gene expression when the transfected cells are treated with dexamethasone [32].

3.3. Stable transfections and measurements of the mRNA half-life Further investigation on the role of the 3'-UTR sequences was performed by construction of stably transfected polyclonal cell lines. We chose to make stable transfections to accurately mimic the processing and transport of the endogenous mRNA. Furthermore, other investigators had suggested that changes in the mRNA stability are more accurately detected in stable than in transient transfections [33]. The plasmids were co-transfected into Rat 1 cells together with a plasmid conferring neomycin resistance to cells and G418-resistant colonies were pooled to form polyclonal cell populations as described [3,9,30]. The colonies were pooled to overcome the effect of random integration of plasmid DNA in single clones. The pools were split for cryopreservation and analysis of DNA

and RNA. The average copy number (Table 1) and integrity of the plasmids in the polyclonal populations was evaluated by Southern blots using the probes described (Fig. 1B). The copy number varied from low to intermediate (Table 1). Southern analysis showed also that the majority of the transgenes in the cell populations contained an intact promoter and 3' sequences (not shown). It might be that the transcriptional efficiency is not linearly correlated with the copy number of integrated plasmids. Titration of a low abundance trans-acting factor by several copies of the target sequence may become rate-limiting and interfere with transcription. This is supported by the finding of Rowe and co-workers, who could not show a correlation between transgene expression and the copy number of the plasmid in stably transfected cell populations [9]. Thus, we did not compare the steady state mRNA levels between the different populations but concentrated in the investigation of the half-life of the cytoplasmic mRNAs. The steady state levels for the endogenous rat pro a l ( I ) collagen mRNA were determined by Northern hybridisation to exclude the possibility that the transfection and selection procedures would have affected the endogenous a l(I) collagen mRNA levels in the cells. No statistical changes were seen between the different populations (Table 1). To measure mRNA half-lives transcription was inhibited by DRB and mRNA levels for transfected minigenes were determined by Northern hybridisations. The hGH reporter gene mRNA was quantitated by laser densitometry and corrected with the values from GAPDH reference probe or 28S ribosomal RNA and compared to the expression of the endogenous rat pro a 1(I) collagen gene. Although drugs like actinomycin D and DRB that block transcription have in fact been shown to stabilize some mRNAs [19], they can be safely used to measure c~l(I) collagen mRNA stability, because approximately similar half-lifes are obtained by pulse-label techniques [14] and transcriptional blockage [15]. The stability of endogenous a l ( I ) collagen mRNA was analysed from the same RNA samples as the stability of transgene mRNA. There was a clear difference in the stabilities of the mRNAs transcribed from plasmid pEhGH-0.7 and from the endogenous rat a l(I) collagen gene (Fig. 3). The half-life of the 4.8 kb endogenous mRNA species was about 10 h and that of the

Table 1 The average copy number of the transfected plasmids and the steady state levels of the endogenous ct 1(I) collagen mRNA Cell population

mRNA levels of endogenous oil(I) collagen

Copy number of the transgene

E-hGH E-hGH-0.7 E-hGH-ASA E-hGH-AMS

1.12 1.20 0.81 1.03

4 2 10 8

+ 0.58 _+ 0.35 + 0.19 + 0.34

Pro a l ( I ) collagen steady state mRNA levels were determined by Northern blotting and densitometric quantitation of the hybridisation signals (mean and S.E. of 5 to 8 measurements presented). The average oil(1) mRNA level in Rat-1 cells is indicated as 1.00. The copy number was approximated by Southern blotting using genomic probes from hGH exons 4 and 5, oil(1) collagen promoter, and otl(I) collagen 3'-UTR.

A. Miiiittii et al. / Biochimica et Biophysica Acta 1260 (1995) 294-300

A

pEhGH

Hours withl0 DRB

3

6

pEhGH-0.7 9110 3 6 9 I

ii

hGH

~ :~ ~

1 2 3 4 5 6 7 8

B 100 RNA left

iu %

Y

75

T

50

25

1 0

0'

3'

;

9' time/hours

Fig. 3. Analysis of the mRNA half-life. (A) mRNA transcribed from the plasmid pEhGH-0.7 is less stable than the endogenous od(I) collagen mRNA or pEhGH mRNA from control transfection. The mRNA synthesis was inhibited by 60 /xM DRB, and total RNA was isolated at indicated time points. The Northern blot filters were sequentially hybridised with probes for hGH and otl(I) collagen chain. Lanes 1-4, RNA from pEhGH transfection; lanes 5-8, pEhGH-0.7 transfection. (B) Quantification of mRNA half-life. Differences in the decay rates of the 4.7 and 5.8 kb a l ( I ) collagen and transgene mRNA with the shorter otl(I) 3'-UTR polyadenylated at the first site were estimated by densitometry of autoradiograms. Each time point represents the mean+S.E, of three measurements. Explanation of the symbols [] = 4.7 kb endogenous od(1) collagen mRNA, O = 5.8 kb endogenous al(I) collagen mRNA, • = pEhGH-0.7 mRNA.

5.7 kb mRNA species about 7 h (Fig. 3A and B). The mRNA from the plasmid pEhGH-0.7 was less stable than the 4.8 kb a 1(I) collagen mRNA, although it has the same 3'-UTR. In repeated experiments pEhGH-0.7 mRNA almost disappeared within 6 h (Fig. 3). This effect is not dependent on the size of the mRNA molecule because the shorter pEhGH mRNA of the control plasmid carrying the hGH 3'-UTR was more stable (Fig. 3A, lanes 1-4) than mRNA from the plasmid pEhGH-0.7. The half-life of pEhGH-ASA mRNA was similar to pEhGH-0.7 mRNA (not shown), although due to a higher copy number, the

299

steady state pEhGH-ASA mRNA levels were higher than those of pEhGH-0.7 mRNA. It is striking that the mRNA utilising the first polyadenylation site was unstable independently of the transfected plasmid. Thus, it is unlikely that this reesult was due to an artefact in the transfection or selection procedures. Some caution has to be reserved due to different species origin of the cell line and transfected chimaeric gene. It might be possible that rat cells, e.g., do not have appropriate proteins to stabilize human mRNA. However, the Y-UTR in the pEhGH-0.7 mRNA is highly conserved between man, mouse and rat [23,34,35] and rat human and mouse ce1(I) collagen mRNAs have similar half-lifes. We have also studied the a l(I) collagen mRNA binding proteins in the human skin fibroblasts, murine 3T3 cells, and Ratl-fibroblasts [32]. All these cells contain the same cytoplasmic a l(I) collagen mRNA-binding protein. We favour the possibility that the 3'-UTR of the shorter a 1(I) collagen mRNA alone is not sufficient for a maximal stability. In that case a 1(I) collagen mRNA would contain several regions that govern the mRNA half-life. Other stability determinants might be located in the coding region of the mRNA. Regions that control the rapid decay of the mRNA have been found both in the coding region and in the 3'-UTR of c-myc and c-fos mRNAs [21,36]. The stably transfected cel(I) collagen minigenes used by Oisen et al. [30] contained also sequences from the coding region. In their experiments, a strong expression of minigene mRNA apparently utilising the first polyadenylation site was found but surprisingly no mRNA corresponding to the use of the second site [30]. Although no evidence for the intact integration of the 3'-UTR sequences was provided and the cell line was different, it is possible that the parts of the coding region (mostly corresponding to propeptides) involved in minigenes contributed to the prevalence of the shorter mRNA species. Our preliminary results with a stably transfected pEhGH-2.2 cell population indicate that the longer mRNA polyadenylated at the second site would better mimic the stability of the endogenous gene, but the results need to be confirmed with further experiments. Recent results by Bornstein and co-workers further support the possibility that a 1(I) collagen mRNA stability determining region(s) are located outside of the Y-UTR. In ras-transformed Rat-1 fibroblasts, the half-life of a l ( I ) mRNA is decreased [12]. However, no decrease was found in the expression of transfected plasmids carrying the whole 2.2 kb long 3'-end in these cells [12], which suggests that parts of the gene other than 3'-UTR are sensitive to destabilizing signals in these transformed cells. We have also mapped RNA binding proteins that recognise the shorter 0.3 kb Y-UTR of o~1(I) collagen mRNA. A 67 kDa protein, named oz1-RBF~,7, in human and murine fibroblast cytoplasmic extracts bound specifically this region [32]. The amount of the binding activity was decreased in extracts from dexamethasone treated cells, which correlates the decreased stability of a l(I) collagen mRNA

300

A. Miiiittii et al. / Biochimica et Biophysica Acta 1260 (1995) 294-300

after corticosteroid application to fibroblasts [32]. Interestingly, also the mRNA levels from transiently transfected plasmid pEhGH-2.2 decreased by dexamethasone treatment [32]. Clearly, this particular protein and its target sequence are not the only complexes involved in the determination of mRNA stability albeit they are putatively involved in the regulation by glucocorticoid hormones. Although sequences located shortly before the first polyadenylation site might be, in part, responsible for the destabilizing effect caused by glucocorticoids [32], also other parts of the mRNA might be of importance and should be studied in future experiments.

Acknowledgements The expert technical assistance of Mrs. Marita Potila and Mrs Tuula Oivanen is greatly appresciated. We are grateful to Dr. Paul Bornstein for reading and discussing the manuscript and for providing the plasmid pEH'. This study was financially supported by Turku University Foundation, The Research and Science Foundation of Farmos Ltd. (to A.M.), and by Southwestern Funds of the Finnish cultural Foundation (to A.M.).

References [1] Bornstein, P. and Sage, H. (1989) Prog. Nucleic Acids Res. 37, 67-106. [2] Vuorio, E. and De Crombrugghe, B. (1990) Annu. Rev. Biochem. 59, 837-872. [3] Slack, J.L., Liska, D.J. and Bornstein, P. (1991) Mol. Cell. Biol. 11, 2066-2074. [4] Haukipuro, K., Melkko, J., Risteli, L., Kairaluoma, M.I. and Risteli, J. (1991) Ann. Surg. 213, 75-80. [5] Scharffetter, K., Kulozik, M., Stolz, W., Lankat-Bungereit, B., Hatamochi, A., S6hnchen, R. and Krieg, T. (1989) J. Invest. Dermatol. 93, 405-412. [6] Sandberg, M.M., Aro, H.T. and Vuorio, E.I. (1993) Clin. Orthop. 289, 292-312. [7] Kivirikko, K.I. (1993) Ann. Med. 25, 113-126. [8] Liska, D.J., Slack, J.L. and Bornstein, P. (1990) Cell Regul. 1, 487-498.

[9] Pavlin, D., Lichtler, A.C., Bedalov, A., Kream, B.E., Harrison, J.R., Thomas, H.F., Gronowicz, G.A., Clark, S.H., Woody, C.O. and Rowe, D.W. (1992) J. Cell Biol. 116, 227-236. [10] Barsh, G.S., Roush, C.L. and Gelinas, R.E. (1984) J. Biol. Chem. 259, 14906-14913. [11] Brenner, D.A., Westwick, J. and Breindl, M. (1993) Am. J. Physiol. 264, G589-G595. [12] Slack, J.L., Parker, M.I., Robinson, V.R. and Bornstein, P. (1992) Mol. Cell. Biol. 12, 4714-4723. [13] K~ih~iri, V.-M., Chen, Y.Q., Su, M.W., Ramirez, F. and Uitto, J. (1990) J. Clin. Invest. 86, 1489-1495. [14] H~im~il~iinen, L., Oikarinen, J. and Kivirikko, K.I. (1985) J. Biol. Chem., 720-725. [15] Penttinen, R.P., Kobayashi, S. and Bornstein, P. (1988) Proc. Natl. Acad. Sci. USA 85, 1105-1108. [16] Schalk, E.M., Gosiewska, A., Prather, W. and Peterkofsky, B. (1992) Biochem. Biophys. Res. Commun. 188, 780-785. [17] Dhawan, J., Lichtler, A.C., Rowe, D.W. and Farmer, S.R. (1991) J. Biol. Chem. 266, 8470-8475. [18] Eckes, B., Mauch, C., Hiippe, G. and Krieg, T. (1993) FEBS Lett. 318, 129-133. [19] Hentze, M.W. (1991) Biochim. Biophys. Acta 1090, 281-292. [20] Shaw, G. and Kamen, R. (1986) Cell 46, 659-667. [21] Shyu, A-B., Greenberg, M.E. and Belasco, J.G. (1989) Genes Dev. 3, 60-72. [22] Klausner, R.D., Rouault, T.A. and Harford, J.B. (1993) Cell 72, 19-28. [23] Mii~itt~i,A., Bornstein, P. and Penttinen, R.P.K. (1991) FEBS Lett. 279, 9-13. [24] Bornstein, P. and McKay, J. (1988) J. Biol. Chem. 263, 1603-1606. [25] Chomczynski, P. and Sacchi, N. (1987) Anal. Biochem. 162, 156159. [26] Sherwood, A.L. and Bornstein, P. (1990) Biochem. J. 265, 895-897. [27] Mets~iranta, M., Toman, D., De Crombrugghe, B. and Vuorio, E. (1991) Biochim. Biophys. Acta 1089, 241-243. [28] Fort, P., Marty, L., Piechaczyk, M., El Sabrouty, S., Dani, C., Jeanteur, P. and Blanchard, J.M. (1985) Nucleic Acids Res. 13, 1431-1442. [29] Simkevich, C.P., Thompson, J.P., Poppleton, H. and Raghow, R. (1992) Biochem. J. 286, 179-185. [30] Olsen, A.S., Geddis, A.E. and Prockop, D.J. (1991) J. Biol. Chem. 266, 1117-1121. [31] Ritzenthaler, J.D., Goldstein, R.H., Fine, A. and Smith, B.D. (1993) J. Biol. Chem. 268, 13625-13631. [32] M~i~itt~i,A. and Penttinen R. (1993) Biochem. J. 295, 691-698. [33] Wilson, T. and Treisman, R. (1988) Nature 336, 396-399. [34] Mooslehner, K. and Harbers, K. (1988) Nucleic Acids Res. 16, 773. [35] Herget, T., Burba, M., Schmoll, M., Zimmermann, K. and Starzinski-Powitz, A. (1989) Mol. Cell. Biol. 9, 2828-2836. [36] Wisdom, R. and Lee, W. (1991) Genes Dev. 5, 232-243.