THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 258. No. 10, Issue of May 25, pp. 6043-6050, Printed in U.S.A. 1983 Expression of the Human Insulin Gene and cDNA in a Heterologous Mammalian System* (Received for publication, October 18, 1982) Orgad LaubS and William J. Rutter From the Departmentof Biochemistry and Biophysics, University of California, Sun Francisco, California 94143 The human insulin gene or the corresponding cDNA has been inserted into the early region of a simian virus 40 vector in which all SV40 splice junctions were de- leted while the early promoterand polyadenylation regions remained intact. The expression of insulin-cod- ing sequences was tested in permissive monkey COS cells. The insulin cDNA was transcribed from the early promoter to produce a stable polyadenylated RNA which was translated, and immunoreactive human proinsulin accumulated in the medium. Thus RNA splicing is not obligatory for insulin expression in this system. The genomic insulin transcript was also initiated from the SV40 promoter and terminated at the SV40 polyadenylation site. S1 endonuclease mapping re- vealed that the transcript is processed via two alter- native splicing pathways within the insulin gene. About one-third of the total transcripts are processed nor- mally with removal of the two insulin-specific introns. This transcript is apparently translated normally since immunoreactive proinsulin accumulates in the me- dium. About two-thirds of the transcripts are processed via an alternative splicing pathway involving a new splice acceptor site located within the coding region of the insulin gene. This results in a codon frameshift such that translation would produce a novel chimeric pep- tide containing the insulin NH2-terminal B chain, but a different COOH terminus containing human and SV40 sequences. A peptide of the predicted size is detected in the COS cell extract. Most eukaryotic genes are mosaic structures in which the coding regions (exons) represented in the mRNA are inter- rupted by intervening sequences (introns) which are subse- quently removed from the primary transcript by RNA splicing (Sharp, 1982). In mostcellulargenes thus far studied, the primary transcript gives rise to a single mature mRNA species. There are however notable exceptions in which alternative splicing is employed: in the early gene of SV40, two donor sites are spliced to one common acceptor site (Berk and Sharp, 1978). In the adenovirus 2 late genes, one donor site is spliced to several acceptor sites (Chow et al., 1977). Alternative RNA splicing pathways also contribute to the diversity of the Grants GM 28520 and AM 21344 and by Grant 1-745from the March * This research was supported by National Institutes of Health of Dimes. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Cystic Fibrosis Foundation. i. Recipient of the Weizmann Fellowship and a fellowship from the expression of immunoglobulin genes (Marcu, 1982),calcitonin (Rosenfeld et al., 1982; Amara et al., 1982), and to the tissue- specific expression of salivary and liver amylase (Young et al., 1981). The nucleotide sequence of the humaninsulin gene and its flanking regions has been determined (Bell et al., 1979, 1980). Acomparison of the human insulingenewith the insulin cDNA and with other insulin genes indicates that this gene has two intervening sequences. Although the DNA sequence provides crucial structural information, it does notdecisively locate the boundary region for the RNA splicing junction nor the regionsregulatinggeneexpression. To elucidatethese features biological systems must be employed. We have used SV40 as a vectorto express the human insulin gene in permissive monkey cells. This system is particularly useful for these and other studies because during infection the virus reaches high titer within the cells and consequently high levels of transcription are achieved. For these experiments our construction employed the SV40 early promoter but the SV40 splice sites were eliminated. A fulllength cDNA was inserted to test for the effect of introns on the expression of insulin- coding sequences. Introns are required for effective expression of the mouse pmHJ globin gene in another SV40 vector-host system (Hamer and Leder, 1979).’ We show that cultured transformed monkey cells infected with this SV40-insulin recombinant express high levels of insulin-coding mRNA without splicing and secrete immunoreactive proinsulin into the medium. The transcripts from the genomic human insulin DNA were processed by two splicing pathways. In the first, the precursor RNA is processed normally; in the second, a new splice acceptor site is recruited from within the insulin- coding sequences. MATERIALS AND METHODS Enzymes and Radioisotopes-Restriction enzymes were pur- chased from Bethesda Research Laboratories or New England Bio- labs. Escherichia coli polymerase I, T4 polynucleotide kinase, and T4 DNA ligase were from New England Biolabs. SI nuclease was obtained from Miles Laboratories. All radioisotopes were from Amer- sham. Cell Transfection and Virus Strain-The construction of SV40- insulin recombinants isdescribed in the results sections. Transformed African green monkey (COS-7) cells (Gluzman, 1981)were maintained in Dulbecco’s modified Eagle’s medium containing penicillin, strep- tomycin, and 10% fetal calf serum. Twenty-four hours after seeding (1.5 X lo6 cells/lO-cm plate), the cultures were transfected by the CaPO, procedure (Graham and Van Der Eb, 1973; Parker and Stark, 1979). 5-10 pg of circularized SV40-insulin recombinant DNA were used for each plate and virus stocks were prepared from cell lysates 14-21 days after transfection. The titer of the recombinant viruses was estimated by comparison with wild type virus stock of known titer, assayed under the same conditions. The comparisons were performed by comparing cytopathic effects of the two stocks and alternatively by comparing amounts of free viral DNA (Hirt, 1967) 48-h post-infection. A. Buchman and P. Berg, personal communication. 6043 by guest on March 21, 2019 http://www.jbc.org/ Downloaded from
Transcript
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 258. No. 10, Issue of May 25, pp. 6043-6050, Printed in U.S.A.
1983
Expression of the Human Insulin Gene and cDNA in a Heterologous Mammalian System*
(Received for publication, October 18, 1982)
Orgad LaubS and William J. Rutter From the Department of Biochemistry and Biophysics, University of California, Sun Francisco, California 94143
The human insulin gene or the corresponding cDNA has been inserted into the early region of a simian virus 40 vector in which all SV40 splice junctions were de- leted while the early promoter and polyadenylation regions remained intact. The expression of insulin-cod- ing sequences was tested in permissive monkey COS cells.
The insulin cDNA was transcribed from the early promoter to produce a stable polyadenylated RNA which was translated, and immunoreactive human proinsulin accumulated in the medium. Thus RNA splicing is not obligatory for insulin expression in this system.
The genomic insulin transcript was also initiated from the SV40 promoter and terminated at the SV40 polyadenylation site. S1 endonuclease mapping re- vealed that the transcript is processed via two alter- native splicing pathways within the insulin gene. About one-third of the total transcripts are processed nor- mally with removal of the two insulin-specific introns. This transcript is apparently translated normally since immunoreactive proinsulin accumulates in the me- dium.
About two-thirds of the transcripts are processed via an alternative splicing pathway involving a new splice acceptor site located within the coding region of the insulin gene. This results in a codon frameshift such that translation would produce a novel chimeric pep- tide containing the insulin NH2-terminal B chain, but a different COOH terminus containing human and SV40 sequences. A peptide of the predicted size is detected in the COS cell extract.
Most eukaryotic genes are mosaic structures in which the coding regions (exons) represented in the mRNA are inter- rupted by intervening sequences (introns) which are subse- quently removed from the primary transcript by RNA splicing (Sharp, 1982). In most cellular genes thus far studied, the primary transcript gives rise to a single mature mRNA species. There are however notable exceptions in which alternative splicing is employed: in the early gene of SV40, two donor sites are spliced to one common acceptor site (Berk and Sharp, 1978). In the adenovirus 2 late genes, one donor site is spliced to several acceptor sites (Chow et al., 1977). Alternative RNA splicing pathways also contribute to the diversity of the
Grants GM 28520 and AM 21344 and by Grant 1-745 from the March * This research was supported by National Institutes of Health
of Dimes. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cystic Fibrosis Foundation. i. Recipient of the Weizmann Fellowship and a fellowship from the
expression of immunoglobulin genes (Marcu, 1982), calcitonin (Rosenfeld et al., 1982; Amara et al., 1982), and to the tissue- specific expression of salivary and liver amylase (Young et al., 1981).
The nucleotide sequence of the human insulin gene and its flanking regions has been determined (Bell et al., 1979, 1980). A comparison of the human insulin gene with the insulin cDNA and with other insulin genes indicates that this gene has two intervening sequences. Although the DNA sequence provides crucial structural information, it does not decisively locate the boundary region for the RNA splicing junction nor the regions regulating gene expression. To elucidate these features biological systems must be employed.
We have used SV40 as a vector to express the human insulin gene in permissive monkey cells. This system is particularly useful for these and other studies because during infection the virus reaches high titer within the cells and consequently high levels of transcription are achieved. For these experiments our construction employed the SV40 early promoter but the SV40 splice sites were eliminated. A full length cDNA was inserted to test for the effect of introns on the expression of insulin- coding sequences. Introns are required for effective expression of the mouse pmHJ globin gene in another SV40 vector-host system (Hamer and Leder, 1979).’ We show that cultured transformed monkey cells infected with this SV40-insulin recombinant express high levels of insulin-coding mRNA without splicing and secrete immunoreactive proinsulin into the medium. The transcripts from the genomic human insulin DNA were processed by two splicing pathways. In the first, the precursor RNA is processed normally; in the second, a new splice acceptor site is recruited from within the insulin- coding sequences.
MATERIALS AND METHODS
Enzymes and Radioisotopes-Restriction enzymes were pur- chased from Bethesda Research Laboratories or New England Bio- labs. Escherichia coli polymerase I, T4 polynucleotide kinase, and T4 DNA ligase were from New England Biolabs. S I nuclease was obtained from Miles Laboratories. All radioisotopes were from Amer- sham.
Cell Transfection and Virus Strain-The construction of SV40- insulin recombinants is described in the results sections. Transformed African green monkey (COS-7) cells (Gluzman, 1981) were maintained in Dulbecco’s modified Eagle’s medium containing penicillin, strep- tomycin, and 10% fetal calf serum. Twenty-four hours after seeding (1.5 X lo6 cells/lO-cm plate), the cultures were transfected by the CaPO, procedure (Graham and Van Der Eb, 1973; Parker and Stark, 1979). 5-10 p g of circularized SV40-insulin recombinant DNA were used for each plate and virus stocks were prepared from cell lysates 14-21 days after transfection. The titer of the recombinant viruses was estimated by comparison with wild type virus stock of known titer, assayed under the same conditions. The comparisons were performed by comparing cytopathic effects of the two stocks and alternatively by comparing amounts of free viral DNA (Hirt, 1967) 48-h post-infection.
FIG. 1. SV40-Human insulin re- combinant designed to express the human insulin sequences. The insulin sequences are indicated by the dark solid line, and the coding sequences are indicated by three blocks. The details for these constructions are summarized in the text.
(GM48) pSV40
LSVinsCZ LSVinrZ LSVinrlP LSVinrLPZ LSVins
DNA Preparations"SV40 strain 777 DNA was used for the con- struction of the SV40 vectors. The HincII/BamHI human insulin DNA fragment was purified from pIns96 which is a subclone of pHi3OO (Bell et al., 1980). All SV40 and human DNA fragments were purified by agarose gel electrophoresis. DNA fragments were eluted by shaking the gel in 0.2 M NaCl, 1 mM EDTA, 10 mM Tris, pH 7.5. Eluted DNA was filtered through GF/C filters and concentrated with butanol-1. Fragments were ligated as described by Maniatis et al. (1978) and SV40-insulin recombinants were cloned in pBR322 and amplified in E. coli (Clewell and Helinski, 1969). SV4O-insulin recom- binant DNA was extracted from infected COS cells as described (Hirt, 1967; Randloff et al., 1967). Preparation of uniformIy labeled viral DNA has been previously described (Zasloff et al., 1982). End-labeled DNA probes were prepared by using T4 DNA polymerase for 3' end labeling (O'Farrell et al., 1980) and T4 polynucleotide kinase for 5' end labeling (Maxam and Gilbert, 1980).
Analysis of RNA-Polyadenylated and nonpolyadenylated cyto- plasmic mRNAs from COS cells infected at a multiplicity of 10-100 plaque-forming units/cell with the LSV*-insulin recombinants were isolated as described previously (Laub and Aloni, 1975). Mapping of RNAs by the SI method of Berk and Sharp (1977) was done with labeled insulin probes prepared from LSV,,. DNA. After hybridization at 50-52 "C for 4 h, the DNA:RNA hybrids were digested with lo00 units of SI nuclease, denatured with formamide, and analyzed by electrophoresis on a 5% polyacrylamide, 8 M urea gel (Maxam and Gilbert, 1980). For Northern analysis polyadenylated RNA was elec- trophoresed on a methyl mercury gel (Alwine et al., 1977) and blotted onto nitrocellulose paper. The resulting blot was hybridized to a nick- translated insulin probe (Maniatis et aE., 1978), washed with 0.1 X SSC at 50 "C, and autoradiographed.
Protein Analysis-COS cells were infected with the LSVIns2 or LSVi,.C2 virus stocks at 10-100 plaque-forming units/cell. Cells were maintained for 24 h in cysteine-depleted medium followed by 6-h labeling with 10 pCi/ml of ~-rS]cysteine. Cell lysate or culture medium was immunoprecipitated with guinea pig anti-bovine insulin serum and analyzed on a 12.5% acrylamide/sodium dodecyl sulfate
The abbreviations used are: LSVi,, late simian virus 40-insulin recombinant; ESVi,, early simian virus 40-insulin recombinant; LSV,,C2, late simian virus 40-insulin cDNA recombinant; LSVk2, late simian virus 40-insulin genomic recombinant; Pipes, 1,4-pipera- zinediethanesulfonic acid; pLSV, late simian virus 40 vector.
gel (Laemmli, 1970). Quantitative radioimmunoassays were per- formed as described (Rall et al., 1973).
RESULTS
Construction of SV40-Human Insulin DNA Recombi- nants-Fig. 1 summarizes the procedure for constructing the SV40-insulin recombinants. The LSV vector was generated from an SV40 genome inserted in the B a m H I site of pBR322 (pSV40) and amplified in E. coli GM48 cells. pSV40 was linearized by partial digestion with Hind11 restriction endo- nuclease (SV40 nucleotide 5107) followed by S1 treatment to produce blunt ends. The linear pSV40 DNA was digested with BcZl restriction endonuclease (SV40 nucleotide 2706) and the 7.2-kilobase pair pLSV vector was purified by preparative agarose gel electrophoresis. The resulting pLSV vector con- tains the SV40 origin of replication and the coding information for the SV40 late genes, Most of the coding region for the SV40 early genes (nucleotides 5107 to 2706) was deleted from the pLSV vector. The vector retains the early SV40 promoters and 105 noncoding nucleotides downstream from the 5' end as well as 136 nucleotides prior to the polyadenylation site at the 3' end of the early gene region. In pLSVi,,C2 a 545-bp BamHI/ EcoRI blunt-ended insulin cDNA fragment was inserted in this site. The human insulin gene including 69 nucleotides in the 5' flanking and 119 nucleotides in the 3' flanking region was present in the 1603-bp HincII/BamHI fragment that was used in all the insulin genomic recombinants. In LSVi,,2 the 1603-bp insulin fragment was inserted 105 bp downstream from the early SV40 major cap site. In LSVin,LP the late SV40 leader sequences (SV40 nucleotides 234-437) are placed be- tween the SV40 early promoter and the human insulin gene. In LSV,,LP2, the late SV40 promoter and leader (SV40 nucleotides 140-437) were inserted between the early SV40 promoter and insulin gene. The constructions of the late SV40 replacement recombinant, ESVi,,,, and the nonexpressing
Human Insulin Gene and cDNA in Heterologous Mammalian System 6045
a c
LsVisCP M RNA
- b LSVinsCP RNA
Nu. INSULIN
CY.
"
Nu. SV40
(RNase) Nu.
FIG. 2. Transcription of the LSVrmC2 recombinant. COS cells were infected with 10-100 plaque-forming units/cell of the LSVi,C2 virus stock and incubated for 48 h at 37 "C. Nuclear (Nu.) and cytoplasmic (Cy.) RNA was prepared as described previously ( h u b and Aloni, 1975). a, polyadenylated cytoplasmic RNA was denatured in 10 m~ methyl mercury, electrophoresed through a 5 KIM methyl mercury, 1.5% agarose gel, and transferred to nitrocellulose. The resulting blots were preincubated in 50% formamide, 4 X SSC, 3 X Denhardt's, and hybridized at 42 "C to an insulin 32P-labeled DNA probe. The blot was washed in 0.1 X SSC at 50 "C and autoradiographed. LaneM contains DNA size markers labeled with [y-32P]ATP and polynucleotide kinase (Maxam and Gilbert, 1980). b, 5-fold dilutions of the nuclear and cytoplasmic RNAs and controls of RNase-treated samples were spotted on nitrocellulose and hybridized to nick-translated SV40 and insulin r3*P]DNA probes.
LSVi, hybrid have been described elsewhere:' All insulin DNA fragments had a blunt 5' end and a BamHI 3' end. These fragments were ligated between the HindIII/Sl blunt end and the Bell end within the pLSV vector.
Ampicillin-resistant colonies were screened by hybridiza- tion to 32P-labeled insulin and SV40 probes, and the positive colonies were analyzed by restriction endonuclease mapping. The resulting plasmids contain a BamHI insert which includes the SV40 origin of replication, a functional set of late SV40 genes and human insulin sequences inserted in the sense direction relative to the early SV40 promoter and polyaden- ylation sites. These BamHI fragments containing the LSV- insulin hybrids were self-ligated to form circular DNA and subsequently infected into permissive monkey COS cells (Gluzman, 1981).
Transcription of Insulin Sequences from the SV40-Insulin cDNA Recombinant (LSV&2)-The intronless LSVi,,C2 re- combinant produces high levels of cytoplasmic polyadenylated RNA containing insulin coding sequences. A Northern blot analysis of this RNA (Fig. 2a) revealed one insulin-specific band which corresponds in size (-900 bases) to an insulin &NA which is initiated at the SV40 early promoter and is polyadenylated at the early SV40 termination signal. This RNA contains the 748 nucleotides of nonspliced insulin coding RNA and about 150 poly(A) residues.
Because of the previous reports suggesting a role of intron removal in mRNA stability (Hamer and Leder, 1979) and transport from the nucleus to the cytoplasm (Lai and Khoury, 1979), we have assayed the steady state levels of insulin-coding sequences in the nucleus and the cytoplasm. The results (Fig. 2b) show that the partitioning of nonspliced insulin mRNA between the nucleus and the cytoplasm is indistinguishable from the partition of normally spliced late SV40 mRNA. Thus, in this system, there is no apparent barrier to the transport of intronless mRNA from the nucleus to the cytoplasm.
3Laub, O., Rall, L., Bell, G. I., and Rutter, W. J. (1983) J. Biol. C h m . 258,603743042.
Transcription of Insulin-coding Sequences from SV40-In- sulin Gene Recombinants-The RNA present in the cyto- plasm of COS cells infected with each of the LSV-insulin recombinants was analyzed by the method of Berk and Sharp (1978). A uniformly labeled insulin-specific probe was pre- pared from COS cells infected with the LSVi, virus stock and labeled with (:"P)orthophosphate for 12-16 h. Viral super- coiled ["'PIDNA was purified (Hirt, 1967; Randloff et al., 1967) and the insulin probe was isolated as a 1.7-kilobase pair HincII fragment. This labeled probe contains the 1603 nu- cleotides of the human insulin insert and an additional 103 bases derived from the 3' end of the early SV40 gene. The 'v2P probe was denatured and hybridized to poly(A') and poly(A-) RNA isolated from infected COS cells under conditions favor- ing RNA-DNA duplexes (Casey and Davidson, 1977). The resulting hybrids were digested with S1 nuclease and analyzed on a 5% acrylamide, 8 M urea sequencing gel (Maxam and Gilbert, 1980). As shown in Fig. 3, lane A, poly(A) RNA from the ESVi.. recombinant produced six bands; the band 206- and 218-nucleotides long correspond to the two insulin-coding exons and co-migrate with the S1-protected fragments ob- tained with human insulinoma RNA treated in a similar manner.3 LSVi., is not transcribed (Fig. 3, lane B). This is due to the strong inhibitory effect of an SV40 sequence (map units 0.76 to 0.86) interposed between the promoter and down- stream sequences. If this inhibitory fragment is removed (LSVh2) or replaced (LSVhLP and LSVhLP2), the insulin sequences are transcribed and expressed at high levels.
LSVi.,2, LSVi.,LP, and LSVi..LP2 (Fig. 3, lanes C, D, and E, respectively) are expressed efficiently, and produce 5- to 10-fold more RNA than the late SV40 replacement recombi- nant. The second exon of the human insulin gene, 206 nucleo- tides in length, is present in the poly(A') fraction of all LSV- insulin constructions; however, exon 1 (42 bp) and exon 3 (218 bp) are not detected in any of the LSV recombinants. This result indicates that most RNA transcripts initiate and ter- minate at SV40 early signals. The band, 99 f 3 nucleotides in
Human Insulin Gene and cDNA in Heterologous Mammalian System
3%
206
99
4Y bp YO6 bp "
9V bp "
SVIO Polyodenylotion +, . 1101. Ironr<rlpll
SV40 Polyodenylotion c( - 1€011* lronl'rlpll
- 340 bp
I 434 t p
F'IG. 3. S1 analysis of RNA from SV40-insulin-infected cells. Purified non-poly(A) (-)- and poly(A)-containing (+) cytoplasmic RNAs from infected COS cells were used. RNAs were hybridized to an insulin/HincII DNA fragment purified from uniformly labeled LSV,, ["'PIDNA. Hybridization mixture in 80% formamide, 0.4 M NaCI, 0.01 M Pipes, pH 6.4 was denatured 2 min at 80 "C, annealed 3 h at 52 "C, digested 1 h with IO00 units of S1 nuclease, and subjected to a 5% acrylamide, 8 M wea sequencing gel. Tracks MI and MZZ contain size markers; Lane A, ESVi, RNA; B, LSVi, RNA; C, L S V , 4 RNA D, LSVi,LP RNA E, LSVi..LP2 RNA; 0, control of probe without RNA. The diagram represents the predicted S1-protected fragments. The band 315 nucleotides in length was unpredicted.
length, corresponds to RNA that is initiated at the SV40 cap site and spliced at the donor site of insulin exon 1. The band, 443 f 8 nucleotides long, corresponds to the RNA extending from the splice acceptor site of insulin exon 3 to the 3' end of the insulin probe and therefore results from transcripts poly- adenylated at the early SV40 site. The largest protected band, 483 f 8 nucleotides in length, was also observed with the ESVi, , recombinant (Fig. 3, lane A ) . This fragment is derived from a partially spliced RNA in which only the second intron is removed, thus this particular RNA class contains exon 1, exon 2, and the first insulin intron. The RNA, 315 f 6
nucleotides in length, was not predicted; the next set of experiments was aimed at the mapping and characterization of this transcript.
Mapping the Novel Transcript-Hue111 restriction endo- nuclease is one of the few restriction enzymes which cuts single-stranded DNA at its specific recognition site (CCGG). The DNA 315 nucleotides in length derived from the S1 endonuclease protection experiment was eluted from the se- quencing gel and digested with an excess of HaeIII. The cleavage products were analyzed on a 5% acrylamide, 8 M urea sequencing gel. As shown in Fig. 4, two new bands, 192 f 3 and 123 f 3 bases long, were detected. Only two locations within the insulin gene insert could produce a 315-nucleotide
315
192
123
603
463
194 188
152
118
102
72 68
I I I I I I C I I 4 4 4 4 I 14 w "-4
c """"" 4 I- - - - - - - - - - -, 0 I
1 500 1000 1500 1700
Base Pairs FIG. 4. HaeIII mapping of the 315 transcript. The 315-base
[32P]DNA derived from the SI nuclease protection experiment (Fig. 3) was eluted from the sequencing gel by passive shaking of the gel as described under "Materials and Methods." The eluted ["'PIDNA was digested with 3 units of HaeIII, and the cleavage products were analyzed on a 5% acrylamide, 8 M urea sequencing gel. Lane MI and MZZ are end-labeled ["*P]DNA size markers. The diagram represents the HaeIII restriction map of the insulin DNA. The dotted lines represent the predicted location of the 315-nucleotide transcript.
Human Insulin Gene and cDNA in Heterologous Mammalian System 6047
fragment which, when cleaved with HaeIII, would yield 123- and 192-bp fragments (Fig. 4). Because the HaeIII products appear as doublets, we reasoned that the fragment must be located at the 3' end of the insulin insert. In this region there are two HaeIII sites separated by 6 bases and located on opposite DNA strands, thus the single-stranded DNA can form a loop that contains a double-stranded HaeIII site, which is presumed to be the substrate for this enzyme (Bron and Murray, 1975).
Mapping the 3' E n d of the New Exon-A probe for map- ping the new exon was constructed from LSVi,, DNA digested with PuuII labeled with "'P using T4 DNA polymerase (O'Farrell et al., 1980) and subsequently cleaved with HincII.
MI A B C MI
9
5 - 0 .
c 5 - * -
c
8. c
18
72
Pvu n +I """-4
4 4 5 b p *- - -- 4
I I Barn HI
98 b p *- 2 1 5 b p *- 380 bp *
FIG. 5. Mapping the 3' end of the new exon. LSVi, DNA was cut with PuuII restriction endonuclease and labeled with T4 DNA polymerase (O'Farrell et aZ., 1980) using [y-32P]dCTP. The labeled DNA fragments were cut with HincII, and a 445-bp PuuII/HincII- labeled fragment was purified by gel electrophoresis. This DNA probe was annealed with L S V d RNA and analyzed by S1 nuclease as described for Fig. 3. Lane MI, end-labeled ["PIDNA size markers; A, SI mapping of RNA from a late SV40-replacement recombinant (ESVi,)", B, RNA from COS cells infected with the LSVi,2 recom- binant; C, untreated labeled probe. The diagram represents the predicted SI-protected fragment, terminating at the insulin poly(A) site (98 bp) or early SV40 poly(A) site (380 bp).
The 445-bp PuuII/HincII-labeled fragment was purified by gel electrophoresis. This DNA probe contains 98 bp derived from the third insulin exon, 117 noncoding nucleotides from the 3' end of the insulin insert, and 227 bp derived from the 3' end of the SV40 early genes. This [''"PIDNA probe was hy- bridized to cytoplasmic polyadenylated RNA extracted from COS cells infected with the ESVi,, or the LSVi.,2 virus stocks. As shown in Fig. 5, lane A, the ESVi., transcripts protected two DNA bands; the 98 f 2-base fragment corresponds to insulin transcripts polyadenylated at the insulin poly(A) site and the 215 f 4-base fragment extends beyond the end of the insulin insert and reflects polyadenylation at the late SV40 poly(A) site. Hybridizing the probe with LSV,,,2 RNA (Fig. 5,
21!
353 078 872
603
3lO
281 271
234
194
118
72
Srna I Born HI
4 1 I I 563 bpt *
340 bpb * 215 bp-*
FIG. 6. Mapping the 5' end of the new exon. The 1603-bp HincII/BamHI insulin-coding DNA fragment was end-labeled with [y-"P]ATP and polynucleotide kinase. The labeled fragment was digested with restriction endonuclease SmaI and the resulting 563-bp labeled fragment was purified by gel electrophoresis. This DNA probe was annealed with LSVi-2 RNA and analyzed by SI nuclease as described for Fig. 3. Lane MI, DNA size markers; A, untreated probe; B, no RNA; C, S1 mapping of RNA extracted from COS cells infected with LSV,&. The diagram represents the S1-protected fragments.
6048 Human Insulin Gene and cDNA in Heterologous Mammalian System
lane B ) results in protecting only one DNA fragment 380 f 6 nucleotides in length which corresponds to polyadenylation at the early SV40 site. The unique 380-base band detected in this experiment suggests that the new exon has an overlapping 3' end with the correct insulin transcript. Thus, the normal and the anomalous RNAs terminate at the early SV40 site, but differ in their 5' acceptor site.
Mapping the 5' End of the New Exon-A probe for map- ping the 5' end of the new exon was prepared from the 1603- bp HincII/BamHI insert by labeling with [yJ2P]ATP and polynucleotide kinase. The labeled fragment was digested with the restriction endonuclease SmaI and the resulting 563- bp labeled fragment was purified by gel electrophoresis. This DNA probe extends from the 3' end of the insulin DNA insert into the second intron within the insulin gene; thus, it includes exon 3. The ['''PIDNA probe was denatured and annealed to poly(A) RNA extracted from COS cells infected for 48 h with LSVins2. As shown in Fig. 6, lane C, two S1-resistant bands were detected in the analytical sequencing gel. The 340-base band corresponds to the expected insulin transcript spliced at the acceptor site of insulin exon 3. The 222 k 4-base band corresponds to the new insulin splice acceptor site which is located within the coding region of exon 3.
The protected fragments shown in Fig. 6 were sized by gel electrophoresis alongside a sequencing ladder of chemical degradation products (Maxam and Gilbert, 1980) generated from the same BamHI/SmaI probe. As summarized in Fig. 7, the position of the two protected fragments within the se-
C C A C G C T C T C T G G G T G t C C A C A G G T G C C G t . . . GCTCCGAGAGACCCACGGGTGTCCACGGTTGCGGCC.. .
c-( C
5'
.. b m
!$ I T
G
G C
t%'
RG. 7. S u m m a r y of the new splice acceptor site in the hu- man insulin gene. a, summary of the mapping of the two alternative splice sites on the sequence of the human insulin gene (Bell et al., 1980). b, the S1 analysis described for Fig. 6 was sized by gel electro- phoresis alongside a sequencing ladder of chemical degradation prod- ucts (Maxam and Gilbert, 1980) generated from the end-labeled SmaI/BamHI probe. c, a schematic representation of the two alter- native splice acceptor sites for the third exon coded by the human insulin gene.
M A R -E- .-g- P... s..- PA E- - - m
mhaln
FIG. 8. Synthesis of human-related peptides in infected COS cells. Monolayers of lo6 COS cells were infected with the L S V d or LSVhC2 virus stocks. Control COS cells were infected with an LSV vector carrying the hepatitis B virus surface antigen gene. Cells were grown 12 h in a cysteine-depleted medium followed by 6-h labeling with [?3]cysteine (10 pCi/ml). Secreted proteins and the cellular lysate (0.5% Nonidet P-.iO/deoxycholate) were analyzed by immuno- precipitation with guinea pig anti-bovine insulin serum (Rall et al., 1973) followed by 12.5% acrylamide, sodium dodecyl sulfate gel elec- trophoresis (Laemmli, 1970). Lane M, size markers: 69, 46, 30, 18.4, 12.3 Da; A and a, cellular and media from control COS cells infected with LSV-Herpes simplex virus 1. B and b, cellular media from COS cell infected with LSV-2 virus stock. C and c, cellular and media from COS cells infected with the LSVi,C2 virus stock.
quence is accurately mapped within the insulin gene. Synthesis of Human Insulin-related Peptides in COS
Cells-To test for synthesis of insulin-related peptides, COS cells were infected with LSVin,C2 (insulin cDNA recombinant) and LSVi2 (insulin genomic recombinant) virus stocks. Cells were grown in cysteine-depleted medium and labeled for 12 h with ["Slcysteine. Insulin-related material was immunopre- cipitated from both culture media and cellular extracts ( R a l l et al., 1973) and analyzed on a 12.5% polyacrylamide-sodium dodecyl sulfate gel (Laemmli, 1970). As shown in Fig. 8, both LSVi,C2- and LSVi,Z-infected COS cells synthesize and se- crete human proinsulin. Thus, insulin cDNA is expressed as efficiently as genomic DNA in this SV40 vector.
The two splicing pathways within the human insulin gene should on translation produce two distinct peptides which share a common insulin NHa-terminal region but are divergent in the COOH-terminal region. The aberrant splicing within the insulin gene expressed in the LSV vector system generates a chimeric translatable insulin-SV40 exon. A =15,000-Dal chimeric peptide was detected in the cellular fraction of COS cells infected with the genomic recombinant, LSVi,2 (see protein x in Fig. 8). This peptide was not detected in COS cells infected with the LSVin,C2 virus stock or with control COS cells infected with an LSV recombinant carrying the hepatitis surface antigen gene.
DISCUSSION
In this paper we describe the use of SV40 early gene replacement vectors (LSV) for studying transcription and processing of a cloned human insulin gene expressed in SV40- transformed monkey kidney cells (COS). The LSV-derived expression system has several useful features. First, the host monkey COS cell line produces SV40 large T antigen and therefore is permissive to early SV40 replacement recombi- nants, thus the LSV-insulin recombinants can be propagated in this host without a helper virus. Second, in this SV40 vector-host system, the early SV40 promoter produces high levels of RNA. Third, all SV40 splice sites were deleted from the LSV expression vector, thus it is possible to study the RNA splicing signals within a gene cloned in this vector.
Human Insulin Gene and cDNA in Heterologous Mammalian System 6049
Finally, the SV40-COS system faithfully reproduces in vivo transcription processing (Mellon et al., 1981).
When the human insulin cDNA was placed under the control of the SV40 early promoter, proinsulin accumulated in the medium at approximately the same rate as that result- ing from a similar construction employing the intact gene containing its two introns (LSV,,2). Gruss et al. (1981) have obtained similar results with a late SV40 replacement recom- binant carrying an intronless rat insulin DNA fragment. Anal- ysis of the mRNA in LSVin,C2-infected COS cells demon- strated normal levels of stable polyadenylated RNA of the expected size. Further, a dot blot analysis of insulin-coding mRNA indicated no significant difference in the nuclear/ cytoplasmic ratio of this RNA as compared to the ratio found in a similar analysis with the intact insulin gene or of normally spliced SV40 late mRNAs. Thus, in this system, introns and/ or splicing are not an obligatory requirement for the produc- tion of stable translatable mRNA. This contrasts to the results obtained with /3-globin (Hamer and Leder, 1979). Apparently no general rule can be made concerning the role of splicing in the recruitment of primary transcripts for translation in the cytoplasm.
Nearly all of the constructed LSV-insulin gene recombi- nants produced high levels of mRNA which was translated to give a secreted peptide having the properties of proinsulin. However, one LSV recombinant (LSVi,,), which replicated efficiently and produced normal levels of SV40 late gene products, did not transcribe the human insulin insert. This is due to the effect of a strong cis-acting inhibitory SV40 se- quence (SV40 nucleotides 457 to 982) on the transcription of downstream insulin sequences. If this inhibitory fragment is removed (LSVins2) or replaced (LSVi,LP and LSVi,LP2), the insulin gene is expressed at high levels (-lo6 molecules/in- fected cell). We assume that this DNA fragment contains an efficient transcriptional termination signal. Ordinarily this sequence is in the nontranscribed strand of the sV40 late gene region, and is therefore nonoperational.
In a late SV40 replacement recombinant (ESVi,,), both SV40 and insulin signals were used.3 However, the late pro- moter and polyadenylation sites are about 10-fold more effi- cient than the insulin signals. In the LSV-insulin recombinants studied here, the SV40 early promoter appears 5-10-fold more efficient than the late SV40 promoter. T antigen is known to autoregulate the early SV40 promoter. We assume that the amounts of T antigen is COS cells is lower than produced during a lytic infection. This results in decreased early pro- moter autoregulation. We presume that no transcription was detected from the insulin promoter because of the relative strength of the sV40 early promoter. In addition only SV40 early termination signals were used. It is significant that the SV40 promoters are not necessarily strong with respect to intracellular normal promoters. The early SV40 promoter under optimal circumstances may produce in this system 1- 4% of the total cellular mRNA, but this RNA is derived from 105-106 SV40 copies/cell. Thus, the rate of transcription/pro- moter is small. Since not all SV40 molecules are transcription- ally active, a precise estimate of relative promoter efficiency cannot be made. In contrast, a single human insulin gene produces higher levels of insulin mRNA in pancreatic B cells. The high levels (-10%) of insulin mRNA in B cells can be inferred from high levels of insulin sequences in a cDNA library prepared from mRNA from hamster islets4 or angler- fish islets.5 This very large discrepancy between expression in the SV40 and B cells suggests a cell- or chromosome-specific
G. Bell, personal communication. ’ P. Hobart, personal communication.
mechanism for enhancing expression of the insulin gene. The heterologous monkey COS cell system seems to be fully competent to process insulin mRNA and secrete human proin- sulin. It can be assumed therefore that the processing systems for mRNA and the secretory mechanisms are not gene- or cell-specific.
Intriguingly, the strength of the termination sites in these systems appear to correspond to the strength of the promoter (early SV40 > late SV40 > insulin). It remains to be demon- strated whether the strength of the initiation and termination signals are intrinsically coupled via some interaction or are affected by some other structural feature (e.g. the stronger promoters are always upstream and the stronger terminators are downstream).
RNA mapping by S1 nuclease analysis demonstrated that two alternative splicing pathways are used in the formation of stable insulin-specific mRNA derived from the human insulin gene, One processing pathway uses the authentic insulin-splic- ing sites to produce the mRNA coding for human proinsulin, the other pathway uses an alternative splice junction to pro- duce a novel mRNA. Since these vectors were constructed independently, the alternative splicing pathway is not a result of a cloning rearrangement (as also confirmed by DNA se- quencing, see Fig. 7). Moreover, all these LSV-insulin recom- binants express insulin mRNA with different 5’ untranslated sequences and promoters; thus the promoter region appears to have no effect on the splicing pattern of this gene.
The novel cryptic splice site was accurately mapped within the coding region of the third exon of insulin. This splice site is within a sequence TACCAGC which is located in front of a pyrimidine-rich DNA stretch and resembles a perfect “con- sensus’’ sequence (Breathnach et al., 1978 Lewin, 1982). The sequence TACCAGC is repeated 19 nucleotides upstream from the cryptic splice site, and another similar sequence, ACCAGC, is found at the site of polyadenylation. Neither of these sequences are employed in splicing to a significant degree in our experiments; this suggests that the splicing process is not determined solely by sequences at the junction but by some other characteristic, for example some structural features of the entire molecule.
All SV40-insulin recombinants express the alternative splice sites to some degree, suggesting that this additional processing site is an intrinsic characteristic of the human insulin gene. However, the constructions employed in the early promoter region (LSVi,,2, LSVi,LP, and LSVi,,LP2) produce domi- nantly the aberrant transcript, while those in the late gene promoter (ESVi,) produce largely the normal insulin mRNA p r ~ d u c t . ~ This suggests that the recruitment of the cryptic splice site appears to be partially influenced by extragenic sequences. A t present we can only conjecture whether these new splice sites are normally employed. There are precedents for the use of alternative splicing sites in the regulation of expression of a single gene in two different cell types (e.g. liver and salivary amylase) (Young et al., 1981) or in the production of two separate proteins from the same genetic sequence (Berk and Sharp, 1978). Rosenfeld et al. (1982) and Amara et al. (1982) have recently presented evidence suggesting that two structurally different mRNAs encoding for different calci- tonin-related protein products arise as a consequence of alter- native RNA-splicing events from the same gene. Our data on insulin RNA processing may provide another illustration of the concept that a single gene can encode more than one protein product. The alternative splicing pathways produce two mRNA species that would on translation generate two distinct insulin-related proteins. The two peptides share the NHz-terminal B chain of insulin but differ in their COOH termini. The normal product proinsulin is readily detected in
6050 Human Insulin Gene and cDNA in Heterologous Mammalian System
the cell medium. We have immunoprecipitated the novel protein in the cell extract of COS cells infected with LSVi,,2. Interestingly it is not secreted ;.n significant quantities despite having the preproinsulin signal peptide. This presumably is due to an alteration at the COOH end of the protein as shown also for p-lactamase secretion (Koshland and Botstein, 1980). Whether an analogous molecule is produced normally in islet B cells or in other cells is unknown. If such molecules were generated from the human insulin gene sequences, it would differ from the molecule detected here because it would not have the sequences derived from SV40. The human DNA sequences in the new exon contain a long open reading frame which could code for 106 amino acids. However, the sequence AATAAA which is part of the recognition site for polyaden- ylation (Proudfoot and Brownlee, 1974; Fitzgerald and Shenk, 1981) is not present in the immediate 3’ end of the alternative human exon. Thus, if this novel splice is used productively in vivo, it is likely to involve an additional splicing step as well. It is difficult to predict the size and structure of this molecule in human cells because of the variety of possible splice junc- tions present in the DNA.
Whether or not this alternate splicing pathway is utilized in uiuo, our results demonstrate that the nucleotide environment external to the gene can influence the selection of splicing alternatives. This suggests an evolutionary route for the origin of new genes from a given genetic sequence.
Acknowledgments-We would like to thank Drs. D. Standring, Y. Shaul, and J. Ou for their careful reading of the manuscript and Leslie Spector for preparing the manuscript.
REFERENCES
Alwine, J. C., Kemp, D. J. & Stark, G. R. (1977) Proc. Natl. Acad.
Amara, S. G., Jonas, V., Rosenfeld, M. G., Ong, E. S. & Evans, R. M.
Bell, G. I., Swain, W. F., Pictet, R., Cordell, B., Goodman, H. M. &
Bell, G. I., Pictet, R. L., Rutter, W. J., Cordell, B., Tischer, E. &
Sei. U. S. A. 74,5350-5354
(1982) Nature (Lond.) 298,240-244
Rutter, W. J. (1979) Nature (Lond.) 282,525-527
Goodman, H. M. (1980) Nature (Lond.) 284, 26-32
Berk, A. J. & Sharp, P. A. (1977) Cell 12, 721-732 Berk, A. J. & Sharp, P. A. (1978) Proc. Natl. Acad. Sci. U. S. A. 75,
1274-1278 Breathnach, R., Benoist, C., OHare, K., Gannon, F. & Chambon, P.
(1978) Proc. Natl. Acad. Sci. U. S. A. 75,4853-4857 Bron, S. & Murray, K. (1975) Mol. Gen. Genet. 143,25-33 Casey, J. & Davidson, N. (1977) Nucleic Acids Res. 4, 1539-1552 Chow, L. T., Gelinas, R. E., Broker, T. R. & Roberts, R. J. (1977) Cell
Clewell, D. B. & Helinski, D. R. (1969) Proc. Natl. Acad. Sei. U. S. A.
Fitzgerald, M. & Shenk, T. (1981) Cell 24, 251-260 Gluzman, Y. (1981) Cell 23, 175-182 Graham, F. L. & Van Der Eb, A. Y. (1973) Virology 52,456-467 Gruss, P., Efstratiadis, A., Karathanasis, S., Konig, M. & Khoury, G.
Hamer, D. H. & Leder, P. (1979) Cell 18, 1299-1302 Hirt, B. (1967) J. Mol. Biol. 26, 365-369 Koshland, D. & Botstein, D. (1980) Cell 20, 749-760 Laemmli, U. K. (1980) Nature (Lond.) 227,680-685 Lai, C. & Khoury, G. (1979) Proc. Natl. Acad. Sci. U. S. A. 76, 71-75 Laub, 0. & Aloni, Y. (1975) J. Virol. 16, 1171-1183 Lewin, B. (1980) Cell 22,324-326 Maniatis, T., Hardison, R. C., Lacy, E., Laur, J., O’Connell, C., Quon,
Marcu, K. B. (1982) Cell 29,719-721 Maxam, A. M. & Gilbert, W. (1980) Methods Enzymol. 65,499-560 Mellon, P., Parker, V., Gluzman, Y. & Maniatis, T. (1981) Cell 27,
O’Farrell, P. H., Kutter, E. & Nakanishi, M. (1980) Mol. Gen. Genet.
Parker, B. A. & Stark, G. R. (1979) J. Virol. 31, 360-369 Proudfood, N. J. & Brownlee, G. G. (1974) Nature (Lond.) 252,359-
Rall, L. B., Pictet, R. L., Williams, R. H. & Rutter, W. J. (1973) Proc.
Radloff, R., Baur, W. & Vinograd, J. (1967) Proc. Natl. Acad. Sei. U.
Rosenfeld, M. G., Lin, C. R., Amara, S. G., Stolarsky, L., Roos, B. A., Ong, E. S. & Evans, R. M. (1982) Proc. Natl. Acad. Sei. U. S. A. 79, 1717-1721
12, 1-8
62, 1159-1166
(1981) Proc. Natl. Acad. Sei. U. S. A. 78,6091-6095
D., Sim, G. K. & Efstratiadis, A. (1978) Cell 15, 687-701
279-288
179,421-435
362
Natl. Acad. Sei. U. S. A. 70,3478-3482
S. A. 57,1514-1521
Sharp, P. A. (1981) Cell 23,643-646 Young, R. A., Hagenbiichle, 0. & Schibler, U. (1981) CeEl23,451-458 Zasloff, M., Santos, T. & Hamer, D. H. (1982) Nature (Lond.) 295,
