Biochimica el Biophysica Acre, 1130 (1992) 259-266 259 © 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4781/92/$05.{11)

BBAEXP 92360

The use of chimeric gene constructs to express a bacterial endoglucanase in mammalian cells

Judith Hall a,,, Barry H. Hirst ", Geoffrey P. Hazlewood ~ and Harry J. Gilbert ~ Departments of "Bioloh, ical & Nutritional Sciences and i, Physiological Sciences, Unit'ersit.v of Ne.'casth" upon Tyne.

Newcastle upon Tyne (UK) and " Agricultural and Food Research Council bastitute Of Anbnal Ph)'shdo~q." and Geneth's Research, Babraham. Cambridge (UK)

(Received 3 September 1991 )

Key words: Endogllicanase; lnlron; Protein sccrelion: Gone exprc.,~sion

The synthesis and secretion of a truncated CIostridium thermocelhtm endoglucanase (EGE')encoded by the celE' gcnc was investigated in Chinese hamster owiry (CHO) cells. Fusion genes consisting of the human growth hormone (hGIt) gcnc and celE', transcribed from the SV40 early enhancer/promoter, were constructed and stably transfectcd into CHO cells. A gene consisting of celE' inserted into the first exon of the hGH gene resulted in the synthesis of truncated proteins ( _< 22 kDa) lacking endoglucanase activity. Cloning celE' into the second exon of the hGH gene, resulted in the synthesis and secretion of a 50 kDa protein with endoglucanase activity. A 50 kDa protein was also synthesised by cells transfected with celE' cloned into the fifth exon of the hGH gene. However, despite a 5-fold increase in enzyme activity compared to the exon 2 transfeetcd cell line less than 40% of the protein was secreted. Constructs devoid of introns, in which celE' was fused to the SV40 early promoter and to the rabbit /3-globin polyadcnylation sequence resulted in a 2-18-fold increase in endoglucanasc activity compared to the constructs containing introns. In addition more than 75% of the synthesised protein was sccreted. Analyses of EGE' encoded mRNA from the transfected cell lines suggests that the presence of introns results in the aberrant splicing of message by the use of cryptic splice sites in the celE' gene. These results demonstrate that introns are not required for the efficient expression of a bacterial endoglucanase in mammalian cells, rather introns appear to reduce expression of the encoded protein.

Introduction

The majority of mammalian genes contain exons which code for mRNA and are interrupted by non-cod- ing sequences of variable length known as introns. Introns provide the means by which exons can be excised, without interruption of the coding sequence, fused to other exons and thus facilitate the rapid evolution of novel proteins [1]. In some instances, intervening sequences are the sites where regulatory elements reside. For example, sequence elements that direct specific DNA rearrangements [2] or that regu- late transcription [3] have been demonstrated to occur in introns.

Early observations [4] suggested that splicing, in which introns are excised from the primary RNA tran- script, was necessary for mRNA accumulation in the cytoplasm. The subsequent use of eDNA vectors has not, however, always resulted in the loss of mRNA

Correspondence and present address: J. Hall, Department of Oral Biology, University of Newcastle upon Tyne, Dental School, Newcas- tle upon Tyne NE2 4~W, UK.

production and protein expression, at least in cell culture. For instance, the cDNAs coding for dihydrofo- late reductase [5] and apolipoprotein D [6] are effi- ciently expressed in mammalian cells. Furthermore, the genes for bean phaseolin [7] and thymidine kinase [8], produce equal amounts of mRNA whether introns are present or not. Additionally, eukaryotic genes such as the histones [9] and heat shock proteins [10] function despite an absence of introns. While these genes pro- vide examples where splicing appears unnecessary for their expression, Buchman and Berg [11] determined that the addition of heterologous splicing signals into the expression vector containing rabbit fl-globin eDNA were effective in promoting /3-globin formation. The authors suggest that the splicing mechanism may play an important role in transcription initiation and elon- gation, nuclear stability and/or cleavage of the RNA transcript. In contrast, Brinster et ai. [12] reported that while introns increased transcription rates of human growth hormone (hGH) gene in transgenic mice there was no significant difference in hGH expression from the eDNA or the complete gene in cultured mam- malian cells.

260

From the foregoing discussion it is apparant that the importance of introns for the efficient expression of cukaryotic genes in mammalian cells is still unresolved. It appears that expression is dependent upon factors such as the gene itself, the expression system em- ployed, the type of transfection and possibly the recipi- ent cells. In recent years our research has focussed on the expression of bacterial endoglucanases in a variety of organisms [13,14]. As prokaryote genes do not con- tain endogenous introns, the role of intervening se- quences or the splicing process per se in mammalian gone expression, are important questions when consid- ering the synthesis of bacterial proteins in animal cells, The aim of this study was to determine the effect of introducing heterologous introns, derived from a highly expressed mammalian gene, on the expression of a bacterial protein in mammalian cells. The bacterial sequence utilised the celL: gone from Clostridium ther- mocellum, which encodes endoglucanase E (EGE), an enzyme consisting of 780 residues with a Mr of 84016 [15]. Introns were supplied by the hGH gene which when fused to the appropriate enhancers is highly expressed in a variety of cell-types [12,16]. A truncated

form of celE (ceiE') was inserted into either the first, second or fifth exon of the hGH gone, respectively, and the effect of intron position on the expression of the bacterial gene, at the level of mRNA and protein, assessed in Chinese hamster ovary (CHO) cells, The introduction of introns into celE' reduced the expres- sion of EGE' in CHO cells. This was probably at- tributed to aberrant splicing of the hnRNA.

Materials and Methods

Plasmid constntction Truncated celE' [15], encoding residues 4-367 of

mature EGE (EGE'), was excised from pHGB2 and mutagenised to introduce a translational stop codon and delete an internal EcoRI site [14]. Modified celE' was cloned into HindIll/Ssti cleaved pMTL22p [17] to generate pMSTI. The modified gene was introduced into the second or fifth exon of the hGH gene, to give pX2hGHcelE' and pXShGHcelE', respectively (Fig. la and b). In the exon 2 chimeric construct celE' was fused to hGH so that the resultant fusion protein contained a leader sequence encoded by hGH. The

ISp Bg l Sp TAAKSs

$V40 p~moter/ on~3ncor

Sp TAAKSs

phGH

pX2FC

Endoglucanase Activity (nmoles MUCB)

Cell Lysate MecSa % Secrelmn

06 40 87

pX6FC 2 7 I 7 38

N ~ TAAKSs

"OATCGATCaG fl ) MATG,~R TSt- LLAFOLLCLPWLQEGSAFPTIPLS

Eukart0te signal Impt~

~,GATCCATOG (2) MGKKIVSLVCVLVMLVSILGSFSWAASPVKGFQVS

Pmka~ote 8~nall~otKle

pXIEFC 0 0 0

pXIPFC 0 0 0

El H SO TAA

pXIE&FC 0 5 52 92

N H~o T/~kKSs R B9 89 Sa X pXIPAFC 0 0 0

"" k~\\\"~\':,,\\\\'~l pE.G~c 42 6.s nal~oat i~.uIob,n I~ly A

s~uenee pPBGFC 89 308 76

Fig, I, Effect of introns on chimeric gene expression. For construct (a), a 1 kb celE' Hindlll.Sstl fragment from pMSTI was blunt-ended by "1"4 DNA polymerase treatment and cloned into the Prull site of exon 2 of the hGH gene. For (b) a I kb BamHI.Bglil celE' fragment excised from pMSTi was cloned into the Bglil site of exon 5 of the hGH gene, In both constructs, the orientation of the prokaryote gene was determined by restricting with SphL For construct (c), the 1,2 kb celE' fragment encoding either the eukaryote or prokaryote signal peptide and endoglucanase gene was cloned into the gamH! site of exon 1 of the hGH gene. The primary structures of the two signal peptides and the nucleotide sequence between the ATG start codon and the 3' nueleotide of the SV40 promoter are shown, For construct (d)o a 0.46 kb fragment between the 3' end of the ce/E' gene (Kpni site) and the P~'ull site of exon 2 of the hGH gene was excised. The Kpnl site was blunt-ended by DNA polymerase and fused to the Pout| site of exon 2. Exons are indicated by hatched boxes (//>), the celE' gene is indicated by an open box ra, signal peptide sequences by a filled boxes • and rabbit/3-globin denoted by hatched boxes (\\\) Restriction sites are designated as follows: B, BamHl; Pv, Pcull; P, Pstl; Bg, BglIi; Sp, Sphl; R, EcoRi; K, KpnI; Ss, Sstl; N, Ncol; H, H/ndIII; Sa, Sail; X, Xhol. Endoglucanase activity for each cell line was determined in cell iysates and media as described in Materials and Methods, and expressed as nmol MUCB. Relative amounts secreted

into the media are expressed as a % of the total activity.

point of fusion of the two genes was chosen so that the cleavage site remained intact, and the N-terminal seven amino acids of the mature protein were from hGH. In the exon 5 chimeric construct, secretion was again predicted to be mediated by a signal sequence donated by the hGH gene, but the resultant protein comprised a fusion between residues 137 and 4 of the mature forms of hGH and EGE', respectively.

Four more constructs were engineered in which signal peptide coding sequences were fused to the 5' end of celE'. Fusion in phase of BamHl/Hindlll restriction fragments encoding the secretion signals described by Hall et al. [14], to the 5' end of celE', created genes encoding EGE' containing an N-termi- nal eukaryotic (celE'E) and prokaryote (celE'P) signal peptide. Both celE'E and celE'P were inserted into the first exon of the hGH gene 26 bp downstream from the transcriptional start and upstream of the hGH translation initiation codon. The resultant constructs designated pXIEhGHcelE' and pXIPhGHcelE' en- code EGE' fused to either a eukaITotic (hGH) signal peptide or a prokaryote (EGE) signal peptide (Fig lc). Two additional constructs, pXIEAhGHcelE' and pXIPAhGHceIE', were devised in which l0 bp of the first exon, the first intron and 134 bp of the second exon of hGH, were removed from pXIEhGHcelE' and pXIPhGHcelE' (Fig. ld). The two remaining con- structs in which the hybrid genes lack introns are described in Hall et al. [14] (Fig. le).

Plasmids X2hGHcelE', X5hGHcelE', XI(E/P) hGHcelE' and XI(E/P)AhGHcelE' were restricted with BamHI/EcoR! and the respective 2.85 kb (2.39 kb for XI(E/P)AhGHceIE') fragments, encoding the celE'/hGH chimeric genes, cloned into pSV2DHFR digested with BamH! and EcoRl. Finally a 330 bp HindIIl DNA fragment, encoding the early SV40 en- hancer/ promoter was blunt ended with sequenase and cloned into the BamHI site (filled in with sequenase), which was immediately upstream of the ATG transla- tional start codon of the respective structural genes. The junctions of the chimeric gene sequences were checked by DNA sequencing. The final constructs are termed pX2FC, pX5FC, pXIEFC, pXIPFC, pXIEAFC, pXIPAFC, pEBGFC and pPBGFC.

Cell culture, DNA tran~fection and cell line selection CHO cells deficient in dihydrofolate reductase

(CHO-DHFR-) were maintained in medium supple- mented with 4 / z g / m l hypoxanthine, 4 # g / m l thymi- dine and 10% dialysed foetal calf serum with gentam- icin 25 /~g/ml. pX2FC, pX5FC, pXI(E/P)FC, pXI(E/P)AFC were linearised with EcoRl and pre- cipitated with ethanol. CHO-DHFR- cells were trans- letted with each plasmid (5 p,g/106 cells) by electropo- ration. Cells transfected with pSV2DHFR alone acted as controls. Details of the electroporation conditions,

261

cell line selection and gene amplification are described in Hall et al. [14].

DNA and RNA analysis High molecular weight DNA was isolated from each

cell line (10 7 cells) as described by Sambrook et al. [18] and 10 p,g digested to completion with Hindlll and Sstl. The DNA was separated by agarose gel elec- trophoresis and blotted onto Hybond N (Amersham) nylon membranes. The membranes were probed with celE', excised from pMSTI, on a BamHl/Bglll frag- ment.

RNA was isolated by the RNAzol blue method [19]. Total RNA (20 tzg) was fractionated on denaturing Mops/formaldehyde agarose gels, transferred to Hy- bond N membranes and probed with the 1 kb ceiE' fragment. DNA probes were labelled using random primers and the hybridisations were performed as de- scribed by Sambrook et al. [20]. Alterations in gene copy number due to amplification with methotrexatc were determined from the Southern hybridisation autoradiograms using a gel documentation system (UVP).

Synthesis of radiolabeiled endoglucanase E' Synthesis of EGE' was monitored by labelling cells

(105) in methionine free medium containing 200 #Ci [35S]methionine (specific activity 1000 Ci/mmol). Cells were incubated for 120 min at 37°C and 5% CO,. The radioactive medium was removed and replaced with complete medium containing an excess of unlabelled methionine (2 mM). Following 60 min incubation at 37°C and 5% CO 2, the radioactive medium was re- moved and cell lysate prepared using a single-deter- gent lysis buffer [18]. lmmunoprecipitation of EGE' in media, chase media and cell lysates was performed as described by Hall et al. [14].

Detection of endoglucanase acticity Following a 24 h incubation medium (5 ml) was

harvested from confluent monolayers of transfected cells (107), clarified by centrifugation and stored at -80°C. The cell monolayer was washed twice with phosphate-buffered saline and the cells collected by trypsinisation. Following iysis and centrifugation the cell lysates (0.5 ml) were stored frozen at -80°C.

Endoglucanase activity was determined by incubat- ing media and cell lysate samples with 0.125 #tool of endoglucanase substrate, 4-methylumbelliferyl-/3-o-cel- iobiopyranoside (MUC) for 60 min at 60°C. After stop- ping the reaction by the addition of 50 mM glycine pH 10.2 the fluorescent activity of the product 4-methyl- umbelliferone (MUCB) was determined at 365 nm excitation and 460 nm emission in a Perkin-Elmer LS5 spectrofluorimeter. Nanomoles of product per reaction was calculated from a standard curve of MUCB.

262

Results

Transfection of CHO cells Transfection of the mammalian cells with the plas-

raids pX2FC, pX5FC, pXIEFC, pXIPFC, pEBGFC, pPBGFC, pSV2DHFR gave rise to seven stably trans- fected ~ l l lines. Cells that had incorporated the for- eign DNA were selected for DHFR expression by growth in the absence of the nucleosides hypoxanthine and thymidine. The transformants of each particular transfection were pooled and selected for growth in sequentially increasing concentrations of methotrexate. Results from Southern blot analysis, indicated that upon selection for methotrexate resistance (0.5 ,~M) the celE' gene was amplified by approx. 20 copies per cell.

Northern analysis of the transfected cell lines Northern blot analysis of the transfected cell lines

containing pX2FC, pX5FC, pXIEFC, pXIPFC, pE- BGFC and pPBGFC using celE' as a probe are shown in Fig. 2. All cell lines synthesised message which hyuridised to celE' except those containing pXIEFC and pSV2DHFR, although on longer exposure of the filter a faint high molecular weight band ( ~ 3 kb) was observed with pXIEFC (data not shown). Two tran- scripts, of 1.5 and 0.9 kb, respectively, were detected

from the cell line containing pXIPFC. A similar pat- tern of two transcripts, 2.0 kb and 1.7 kb, respectively, was observed for transfected cells containing pX2FC. For each of these latter two cell lines a single tran- script of 2.0 kb was expected. Positioning the celE' gene in exon 5 of hGH (pX5FC) resulted in only one species of celE'oderived mRNA encoding message of 1.4 kb. This result, when compared to the predicted transcript size of 1.8 kb, again suggested that the hnRNA was incorrectly processed. Analysis of the cells transfected with either pEBGFC or pPBGFC con- structs, which contain celE' devoid of introns, revealed only one transcript of 1.0 kb for each cell line.

To investigate whether aberrant splicing was associ- ated with the first intron of hGH, two further con- structs, pXIEAFC and pXIPAFC, were stably trans. fected into CHO cells. In these constructs the splicing signals associated with the first intron of hGH were removed (Fig. ld). Two celE' mRNA transcripts were produced by each transfected cell line (Fig. 2), indicat- iffg that introns other than the first were associated with aberrant splicing of message.

EGE synthesis in the transfected cell lines To determine if the mRNA derived from the

hGH/celE' constructs could direct the synthesis and secretion of active protein, the concentration of func-

16s ==-

18s~-

18s

t

1 2 3 4 5 6 7 8

Fig, 2, Northern blot analysis of total RNA from the endoglucanase transfected cell lines. RNA was probed with 1 kb BamHl/Bgill fragment isolated from pMSTI and encoding the endoglucanase gene. Lanes: 1, pEBGFC; 2, pPBGFC; 3, pXIPFC; 4, pXIEFC; 5, pX5FC; 6, pX2FC; 7,

pXIPAFC; 8, pXIEAFC. Arrows indicate 18S (HeLa) and 16S (Escherichia coil) rRNA bands.

tional E G E in cell lysates and media f rom all the t ransfected cell lines was analysed (Fig. 1). Despi te the presence of celE' m R N A , E G E activity ~ a s not de-

263

tected in e i ther cell lysates or media from cell lines containing p X I E F C , p X I P F C and pXIPAFC. Equiva- lent activity was detected in cells containing pX2FC

a

50 46

s0

50 . j 46

~F ' ' ~ ~

1 2

50 ~ !

3 4 5

50 tN , . , a l=q t ~ o / . ~

1 2 3 4 5 6 7 8 9 10 11

Fig. 3. Synthesis of radiolabeiled EGE'. Autoradiograms of fluorographed $D$-PAGE gels of EGE' immunoprecipitated from cell lysates, media and methionine chase of (a) pX2FC, (b)pXIEAFC and pX5FC, and (c) pXIPFC transfected cells. (a) pX2FC transfected cells: lane l, cell lysate; lane 3, chase medium; lane 5, medium, lanes 2 and 4, controls (proteins immunoprccipitated with EGE antiserum from cell lysate (2) and medium (4) of CHO cells containing pSV2DHFR only). (b) pXIEAFC and pXfFC transfected cells: lane 1, pXIEAFC cell lysate; lane 2, pXSFC cell lysate; lane 3, pXIEAFC cell lysate (after chase); lane 4, pXSFC cell lysate (after chase); lane 5, pSV2DHFR control cell lysat¢: lane 6. pXIEAFC chase medium; lane 7, pXSFC chase medium; lane 8, pSV2DHFR control chase medium; lane 9, pXIEAFC medium; lane 10, pXSFC medium; lane 11, pSV2DHFR control medium. (c) pXIPFC transfected cells: lane 1, cell lysate; lane 2, cell lysate (after chase); lane 3 pSV2DHFR control cell lysate; lane 4, medium; lane 5, chase medium; lane 6, pSV2DHFR control medium. The Mr values x 10 3 of

immunoprecipitated proteins are shown.

264

22 21 19

a

22 21 19

1 2 3

Fig. 3 (continued).

~ l I P q U I I D

4 5 6

and pXIE/IFC, respectively, with more than 85% of protein detected in the media• A 5-fold increase in activity was observed in the cell lysates from pX5FC transfected cells, compared to other intron containing genes, although less than 40% of the protein was secreted. The cell lines containing the intronless con- structs expressed 2 to 3-fold more activity than the construct with upstream introns (pXSFC), but 8 to 184old more activity than the constructs with down- stream introns. Moreover, more than 75% of the syn- thesized protein was secreted.

Previous studies using ['~S]methioninc labelling have shown that three forms of anti-EGE immunoprecip- itable protein (50,46, 37 kDa) could be identified in the cell lysates from pEBGFC and pPBGFC transfected cells [14]. The 50 kDa form comigrated with the extra- cellular form of EGE. Cells transfected with pX2FC, pXIEAFC and pXSFC constructs, revealed a similar pattern with 50 and 46 kDa proteins identified in the cell lysates (Fig. 3a, b). When radiolabeled proteins were chased with an excess of unlabeled methionine there was a rapid disappearance ( < 1 h) of the radiola- belled 50 kDa protein from the cell lysates of the pX2FC and pXIE,IFC transfected cells and its appear- ance in the media (Fig. 3b). No immunoprecipitable protein was detected in the medium or chase medium from the pXSFC transfected cells (Fig. 3b).

Labelling studies of the cell line transfected with pXIP,4FC revealed no anti-EGE immunoprecipitable protein in either the cell lysate or medium (data not shown). However, three low molecular weight proteins (19, 21 and 22 kDa) were precipitated from the cell lysate and medium of the pXIPFC transfected cell line (Fig. 3c). After chasing with unlabeled methionine only

a weak 19 kDa protein band was observed in the cell lysate, while all three bands were observed in the chase medium (Fig. 3c), suggesting that the smaller proteins are not intermediates in the synthesis of the 22 kDa protein.

Discussion

The expression of a gene encoding a bacterial en- doglucanase was studied in CHO cells. Inserting celE' into either the first or second exons of the hGH gene resulted in reduced endoglucanase activity. Analysis of the transcription products suggested that problems may have arisen due to the organisation and subsequent transcription of the chimeric genes, it is possible that the size of the 5' exon containing celE' favoured the use of alternative splice sites. Hawkins [20] has ob- served that 5' coding regions and internal exons in eukaryotes are rarely greater than 800 nt and 300 nt, respectively. In our exon I and exon 2 constructs the extended exons ( > 1 kb) may have promoted the recog- nition and activation of 5' cryptic splice sites within the celE' gene. The presence of more than one celE' encoded message suggests aberrant splicing, possibly arising from the use of cryptic splice sites. Inspection of the celE' sequence reveals three cryptic 5' splice sites, two of which show 7/9 bp homology with the consensus sequence (CA) /AG/CT(AG) /AGT. A sim- ilar finding where the introduction of an intron into the normally unspliced yeast YP2 gene influenced the splicing efficiency of the resultant transcripts was re- ported by Klintz and Gallwitz [21]. Moreover, the au- thors concluded that the further the distance between the mRNA Cap site and the intervening sequence the more inefficient the splicing of the resultant tran- scripts. More recently, Huang and Gordon [22], while observing an increase in cytoplasmic CAT RNA from CV! cells transfected with the CAT gene, containing the small t intron at its 3' end, also reported a 6-8-fold reduction in enzyme activity corr,,~ ~red to the intronless construct. Again the result was attributed to the activa- tion of cryptic splice sites within the CAT gene. in the present studies, however, it cannot be excluded that the presence of more than one transcript was associ- ated with the incorrect initiation of transcription.

The absence of celE' mRNA from the cell line transfected with the pXIEFC construct suggests that the prokaryote gene was either not transcribed or that the RNA derived from the construct was very unstable. The presence of a faint high molecular weight celE' mRNA species (~ 3 kb) following long exposures of the filters indicates that the hnRNA derived from this construct is inefficiently processed in the nucleus. Pro- cessing may have been inhibited by the secondary or tertiary structure of the RNA affecting the complexing of proteins associated with either the spliceosome, or

in the protection of the RNA from nucleases. The coupling of the splicing mechanism to a polyadenyla- t ion/transport process [23] may then explain the ab- sence of celE' cytoplasmic message. While removal of the first intron from the. hGH gene to generate pXIEAFC resulted in the synthesis and secretion of active protein the splicing mechanism still appeared inefficient. Indeed large amounts of message appeared incapable of producing functional protein probably due to aberrant splicing involving introns 3' to the celE' gene.

Absence of active EGE' from the cell lines trans- retted with pXIPFC and pXIPAFC suggests that the observed transcripts are either incorrectly processed, resulting in the premature termination of translation products, or incorrectly initiated. Cell labelling studies lend support to the former as low molecular weight proteins ( _< 22 kDa) were immunoprecipitated by anti- EGE from the cell lysates and media of cells trans- fected with pXIPFC, consistent with the synthesis of proteins with a functional signal peptide, but with premature termination of protein synthesis. Expression of the XIEAFC containing cell line but not the compa- rable XIPAFC cell line may be attributable to two cryptic 5' splice sites in the prokaryote signal peptide. In siting celE' in the fifth exon of the hGH gene splicing problems were apparently overcome as ob- served by a 5-fold increase in the cell synthesis of EGE'. However, removal of the eukaryote introns and fusing celE' 5' to a SV40 early promoter and 3' to rabbit /3-globin stop, poly(A) sequence resulted in a significantly greater EGE' activity (2-18-fold), Thus, the presence of introns in constructs containing nor- mally unspliced bacterial genes does not appc,~r to enhance expression in mammalian cells.

The endoglucanases synthesized and secreted by the cells containing the intronless constructs have been described previously [14]. Interestingly, the cell lines transfected with pX2FC, pX5FC or pXIEAFC all syn- thesize a similar pattern of proteins (Fig. 3a, b). This suggests that the proteins from these cell lines are also post-translationally modified, probably by N and O-lin- ked glycosylation. The presence of a 50 kDa protein with MUC activity in the lysate of the cells transfected with pX5FC indicates that the resultant h G H / E G E ' fusion protein is catalytically active. However, the de- creased MUC activity observed in the bathing medium suggests that a large percentage of the enzyme is trapped in the secretory apparatus of the cells. The inability to detect aSS-labelled proteins in the same m~d!um probably reflects the decreased sensitivity of the radiolabelling/immunoprecipitation system.

It is interesting that the intronless celE' genes were efficien.tly transcribed to a single mRNA species; there was no evidence of aberrant hnRNA splicing. This indicates that only some of the intron excision se-

265

quences are present in ceiE'. Analysis of the bacterial gene sequence revealed 5', but no 3' splice sites. From an evolutionary view point there is no rationale for eukaryote splice sites to be selected against in bacterial structural genes. Thus, many bacterial genes would be predicted to also contain cryptic splice sites, which when brought into close proximity with eukaryote in- trons increases the likelihood of incorrect processing of hnRNA.

Conclusions

The role of introns in maximising gene expression remains controversial. In this paper we show that the presence of introns strongly influences the expression of a bacterial gene in CHO cells. While the presence of introns increased the amount of cytoplasmic mRNA, it did not result in increased levels of functional protein. Aberrant splicing between cryptic splice sites in celE' and introns located downstream of the gene appeared to res~lt in truncated proteins lacking EGE activity. These results demonstrate that introns are not re- quired for the efficient expression of a bacterial en- doglucanase in CHO cells and in f~ct appear to be detrimental to expression.

Acknowledgements

This work was supported by AFRC Grant TAP39, and a Royal Society Equipment Grant for electropora- tion apparatus. Jane Daniels is thanked for excellent technical assistance and Dr. Kathleen Sot~le for advice on the MUC assay.

References

I Breitbarl, R.E.. Andreadis, A. and NadaI-Ginard. B. 119871Annu. Rev. Biochem. 56, 467-495.

2 Tonegawa, S. (19831 Nature 3112, 575-581, 3 Fr;mklin, G.C., Donovan, M., Adam. G.I.R.. Holmgren, L.,

Pfeifer-Ohlsson, S. and Ohlsson R. (ITS91) EMBO J. 10, 1365- 1373.

4 Gruss, P. and Khoury G. 11981))Nature 286, 634-637. 5 Gasser, C.S., Simonscn. C.C., Schilling, J.W. and Schimkc, R.T.

(1982) Proc. Natl. Acad. Sei. USA 79, 6522-6526. 6 Drayna, D., Fielding, C., McLean, J., Baer, B,, Castro G., Chum

E., Comstock, L., Henzel. W., Kohr, W., Rhec. L.. Wion. K. and Lawn, W. (19861 J. Biol. Chem. 261. 16535-16539.

7 Chee, P.P., Klassy, R.C. and Slightom. J.L. [1986) Gentr 41, 47-57.

8 Gross, M.K., Kainz, M.S. and Merrill, G.F. (1987l Mol. Cell. Biol. 7, 4576-4581.

9 Kedes, L.It. (1979) Annu. Rev. Biochem. 48. 837-8711. I[1 Wu, B., Hunt. C. and Morimoto. R, (1'~85) Mol. Cell. Biol. 5.

33[)-341. 11 Buchman, A.R. and Berg. P. (19881 Mol. Cell, Biol. 8. 4395-44115. 12 Brinster. R.L., Allen, J.M., Behringer. R.R., Gelinas, R.E. and

Palmiter, R.D. (19881 Proe. Natl. Acad. Sci. USA 85. 836-840, 13 Bates, E.E.M., Gilbert. H.J., Hazlewood, G.P., Huckle, J., Laurie.

J.I. and Mann, S.P. (19891 Appl. Env. Microbiol. 55. 2~}95-2097.

266

14 Hall, J., Hazlewood, G.P., Surani, M.A., Hirst, B.H. and Gilbert, HJ. (1990)J. Biol. Chem. 265, 19996-19999.

15 Hall, J., Hazlewood, G.P., Barker, P,J. and Gilbert, HJ. (1988) Gene 69, 29-38.

16 Friedman, J.S., Corer, C.L., Anderson, C.L., Kushner, J.A., Gray, P.P., Chapman, G.E., Stuart, M.C., Lazarus, L., Shine, J. and Kushner, P.J. (1989) Bio/Technology 7, 359-362.

17 Chambers, S.P., Prior, S.E., Bartsow, D.A. and Minton, N,P. (1988) Gene 68, 139-149.

18 Sambrook, J., Fritsch, E.F. and Manialis, T. (1989} Molecular

Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor.

19 Chomczynski, P. (1989) Cinna/Biotecx Bull., No. 3. 20 Hawkins, J.D. (1988) Nucleic Acids Res. 16, 9893-9905. 21 Klinz. F.-J. and Gallwitz, D. (1985) Nucleic Acids Res. 13, 3791-

3805. 22 Huang, M.T.F. and Gorman, C.M. 0990) Mol. Cell. Biol. 10,

1805-1810. 23 Huang, M.T.F. and Gorman, C.M. (1990) Nucleic Acids Res. 18,

937-947.