Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987

YEAST CELLS MAY USE AUC OR AAG AS INITIATION CODON FOR PROTEIN SYNTHESIS by OLE OLSEN

Department of Physiology, Carlsberg Laboratory, Gamle Cadsbergvej 10, DK-2500 Copenhagen Valby

Keywords: Recombinant DNA, initiation, secretion

A yeast expression plasmid without an ATG codon for initiation of mouse a-amylase protein synthesis directs the synthesis and secretion of active enzyme indistinguishable from both native mouse a-amylase and amylase synthesized from plasmids with normal AT(3 initiation codons. The initiation of amylase synthesis directed by this plasmid is at either an AUC or an AAG codon. In either case the sequence of the hydrophobic core and peptidase cleaving region of the signal peptide are normal, and the protein translation remains in frame with the structural of the mouse a-amylase.

1. INTRODUCTION mal context for initiation was 6NNAUGGA A where Initiation of protein synthesis in is a purine in position -3 (three nucleotides up- catalysed by a relatively large number of specific stream of the AUG initiator codon) is most eukaryotic initation factors (eIFs) bringing highly conserved. about the formation of the 80S initiation com- Until recently it was generally believed that plex composed ofmRNA, an 80S and eukaryotic initiate exclusively at met-tRNAi (l 3). Eukaryotic met-tRNAi differs AUG codons although it had been shown by from its prokaryotic counterpart by being asso- RAJBHANDARYand GHOSH (24) that yeast met- ciated with the 40S pre-initation complex before tRNAi may function with either AUG or GUG binding to mRNA (13). This suggests that the as initiation codon in a cell-free translation cell has a requirement for met-tRNA~ during an system from E. coli. At present, a few examples early step of the initiation process leading to of non-AUG codons for initiation of protein assembly of the 80S complex. The ability of the synthesis are known from eukaryotes. In yeast 40S complex to recognize an AUG initiation cells UUG- and AUA-codons may be used in codon in proper context is probably dependent initiation of protein synthesis (34) but it is on codon-anticodon interaction. Among the emphasized that the start codon must appear in important features (13), the cap structure a proper context. BECERRAet at. (2) have shown (mTG5'pppN) at the 5" end ofeukaryotic mRNA that the methionyl-tRNA dependent initiation is essential for efficient initiation which usually ofadeno-associated virus (AAV) B protein starts occurs at the first AUG triplet downstream from from an ACG codon, and ANDERSON et al. (1) the cap structure. From a survey of published have shown that an ACG codon (replacing the DNA sequences it was concluded that the opti- AUG used in vivo to initiate the synthesis of

Abbreviations: ADHI (PR) = alcohol dehydrogenase I (promoter); bp = base pair(s).

Springer-Verlag 0105-1938/87/0052]0083/$01.60 O. OLSEN:Non-AUG initiation of translation

bacteriophage T7 gene 0,3 protein) may func- yeast cells as follows. Total DNA was prepared tion as initiation codon in a wheat germ cell-free as described by STRUHLet al. (28). After incuba- protein synthesizing system. tion with RNase A, the DNA samples were In the present study a yeast plasmid directing extracted twice with an equal volume of phenol the synthesis and secretion of low amounts of and once with chloroform, precipitated from mouse salivary a-amylase has been sequenced in 70% ethanol, washed, and resuspended in 10 the region flanking the promoter and the cDNA mM-Tris-HCl, 1 mM-EDTA (pH 7.5). This solu- coding for the secreted protein. It was found that tion was used to transform competent E. coli the a-amylase coding region was without an JM83 cells. ATG codon for initiation of translation. This E. coli plasmid DNA was prepared by the example of a heterologous gene in yeast cells method of BmNBOIM and DOLV (3) which for being expressed without an AUG initiation large scale preparations was followed by caesium codon shows that yeast cells may use either AAG chloride equilibrium centrifugation. Purifica- or AUC codons for translation initiation. The tion of specific DNA fragments from low melt- secretion level of a-amylase directed from this ing temperature agarose gels was carried out as plasmid is low compared to plasmids with an described by MANIATlSet al. (16). ATG start codon. However, the pre-a-amylase synthesized from mRNA without an AUG codon is apparently processed and secreted in 2.4. Transformations the same manner as a-amylase initiated from an Transformation ofE. coli with plasmid DNA AUG codon. was performed as described by COHEN et al. (8). Yeast cells were transformed according to the method OfITo et al. (12) using lithium acetate to 2. MATERIALS AND METHODS induce competence. Yeast transformants were 2.1. Strains and media selected on SC medium without and E. coil K-12 strain JM101 (17) was used to cells secreting a-amylase were detected on simi- propagate M 13mp phage vectors as described by lar medium supplied with 1% starch (29). MESSING (18). Otherwise the E. coli K-12 strain JM83 (17) was used as host in bacterial cloning experiments as described by VIEIRAand MESS- 2.5. DNA sequence determination ING (30). The Saccharomyces cerevisiae strain The nucleotide sequence of the a-amylase DBY746 (29) was grown in either YPD or SC cDNA gene (9) and its 5' flanking regions was media (22). determined by the dideoxynucleotide chain ter- mination method of SANGER et al. (25, 26) except that the elongation reaction temperature 2.2. Enzymes and chemicals was 55 ~ Radioactive nucleotides were purchased from New England Nuclear. All enzymes as well as deoxy- ~nd dideoxynucleotides were from 2.6. Activity staining of secreted a-amylase Boehringer Mannheim. Yeast cells grown in liquid medium were spun down. The supernatant was concentrated and dialysed in an Amicon ultrafiltration unit. Non- 2.3. Plasmids and DNA purification denaturing electrophoretic separation of the The yeast expression plasmid, pMA56, was concentrated supernatant polypeptides was car- kindly provided by Dr. B.D. HALL,University of fled out in 13% polyacrylamide gels using a Washington, Seattle, USA. pMS 15 was one of Tris- buffer system, pH 8,3 (14). several plasmids constructed by inserting a Amylase activity was visualized by incubating mouse a-amylase cDNA gene behind the ADH! the gels for 2 hours at 37 ~ in a 50 mM-phos- gene promoter of plasmid pMA56 (29). phate buffer (pH 6,9) containing 4% starch and Recombinant plasmids were recovered from 0,08% NaCI (4, 27). After rinsing in water the

84 Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 O. OLSEN: Non-AUG initiation of translation

E A E E directing different levels of a-amylase secretion from yeast (Figure 1). The plasmids were con- structed by inserting Bal31 exonuclease treated a-amylase eDNA in the yeast B, C E pMA56 using EcoRI linkers (29). Among these plasmids one, pMS 15, gave rise to low level of amylase secretion from yeast. It was chosen for further analysis and the nucle- otide sequence flanking the ADHI gene pro- moter and the a-amylase cDNA was deter- mined. In Figure 2 the nucleotide sequence is shown in comparison with the sequence of the // analagous region from plasmid pMSI2 which directs high level of amylase secretion from yeast cells. While plasmid pMS 12 contains a complete a-amylase coding region the nucleotide se- quence found in pMS15 seems to represent an aberrant cloning event where the a-amylase Figure 1. Physical map of a-amylase expression plas- mids. Mouse salivary a-amylase eDNA is inserted eDNA has been inserted in the vector via an adjacent to the yeast ADHI gene promoter. The plas- EcoRI* site present in position +5 of the coding mid contains a yeast TRP1 gene for selection and the region, i.e. within the second codon. origin of replication from the yeast 2Ix circular plasmid. The sequence specificity of EcoRI* activity is The plasmid also contains most of the E. coli plasmid not quite clear, but the activity may be induced pBR322 including both the ampicillin resistance gene as a result of increased glycerol concentration in and the origin of replication. Cleavage sites for restric- the reaction mixture (6). POLISKY et al. (23) tion enzymes Apa| (A), BamHI (B), ClaI (C) and EcoRI observed EcoRI* activity under conditions of (E) are shown. elevated pH and low ionic strength. They sug- gest that EcoRI* cleaves between the first and second base of the sequence NAATTN thereby generating the same type of 5' protruding ends gels were stained for 5-15 minutes in a 1% I2-KI as normal EcoRI digestion. During the con- solution. Bands with amylase activity appeared struction of the pMS plasmid series a 30 fold as clear regions against a dark background. molar excess of EcoRI linker was used; subse- quently the linkers were digested with EcoRI using standard conditions. Yet it seems that a 3. RESULTS AND DISCUSSION few EcoRI* sites were cleaved. The generated This report presents results of the analysis of restriction fragments were cloned in the EcoRI an aberrant plasmid among a series of plasmids site of the vector pMA56.

,,, CCAAG~ATACAATCAAG6AATTCC~6~AAGAATACTG~AA~A~CATA~CAAAAT~AAATTCTTCCTGCT(~:TTTCC~' . , pMS12 , , , ~GTTC~TAT~TTAGTTCCTTA'~CCCTTCTTAT~AC~TT~TCGTATC~TTTTACTTTAA~AA66ACGAC~AAA6~'~- , ,

,,, CC~AGCATACA~TC~AG G~,ATTCTTC CTGCTGCTTTCCC, , , pMS15 ,,, GGTTCGTATGTTAGTTCCTTA~GAAG~iACGACGAAAGG6.,,

Figure 2. Comparison of the nucleotide sequence flanking the promoter and the a-amylase eDNA of plasmids pMS 12 and pMSl 5. Vertical an-rows indicate where the a-amylase eDNA is inserted into the EcoRI site of the yeast expression plasmid pMA56. The common sequence between the a-amylase sequence of pMSI 2 and pMSI 5 is underlined. The start codon ATG is marked in pMSl 2, while the putative start codons are marked in pMSl 5.

Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 85 O. OLSEN: Non-AUG initiation of translation

RV ~ I RI ~ I RV RI solution. The gel system used for this analysis is very sensitive. It has been used to separate ~oo~ mouse pancreatic a-amylase isoenzymes (10) which subsequently have been shown to differ q D only by one neutral amino acid (5). The amylase 9 banding patterns from pMS 12 (a plasmid con- Figure 3. Sequencing strategy. Restriction map of the struction with normal ATG initiator codon) and 3' region ofthe ADHlgene promoter (thin line) and the pMS 15 are indistinguishable from that of mouse 5' region of the a-amylase cDNA (heavy line). Shown salivary a-amylase. The occurrence of multiple are the restriction sites for AluI (I), EcoRl (RI), EcoRV bands of a-amylase activity may be due to (RV) and HaellI (III). Horizontal arrows indicate glycosylated and unglycosylated forms of the regions where nucleotide sequence determination was a-amylase enzyme. These results indicate that obtained by dideoxynucleotide sequencing. plasmid pMS 15 directs transcription of an mRNA encoding a-amylase with a signal pep- tide which is cleaved offin the same position as However, this aberrant cloning event might in native mouse a-amylase. have happened during subcloning of the 5' Two ATG triplets are located in the ADHI EcoRI fragment of the cDNA in preparation for gene promoter 189 and 251 bp upstream from sequencing, inferring that the yeast cells might the cDNA gene. Both these ATGs are in frame contain an intact 5' EcoRI fragment. Therefore, with a TAA and cannot be used for non-EcoRI restriction fragments flanking the a-amylase synthesis. In the normal a-amylase promoter and the cDNA were isolated from cDNA the second ATG codon is located 288 bp pMS 15, subcloned in M 13 and the nucleotide downstream from the initiation codon. sequences determined (Figure 3). The results From a survey of available sequence data from this analysis showed that the sequence KOZAK (13) found that functional initiator previously derived was not an artifact intro- duced during subcloning. Further evidence for plasmid pMS15 direct- ing synthesis of a-amylase in spite of missing the ATG initiation codon was obtained by isolating the plasmid from yeast cells. The nucleotide sequence of the DNA fragment of interest was determined and it was found to be identical with the originally obtained sequence. A functional in vivo control test was also performed. The BamHI-ClaI fragment includ- ing the ADHI gene promoter and most of the a-amylase cDNA (Figure l) in pMSl5 was replaced by the analagous fragment from pMS 12. The new recombinant plasmid pMS 12- 15 directs an a-amylase secretion level compara- ble to pMS 12 (Figure 4). A plasmid constructed by the reciprocal exchange of fragments, pMS 15-12, directs an a-amylase secretion level comparable to pMS 15 (Figure 4). Figure 4. Secretion of a-amylase from yeast. Yeast cells The a-amylase from concentrated yeast cul- transformed with plasmid pMSI2 (1), pMSI2-15 (2), ture supernatants was analysed in non-denatur- pMS15 (3) and pMSI5-12 (4) were incubated over ing polyacrylamide gels (Figure 5). Zones of night on plates containing 0.196 starch. The plates were amylase activity were revealed by iodine stain- stained with iodine vapour. A halo shows that the cells ing after incubating the gel in a 4% starch secrete a-amylase to the medium.

86 Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 O. OLSEN:Non-AUG initiation of translation

AGNNAUGG A pMS12 AAAAUGA pMS15- L ACCAAGC pMS15- I ACAAUCA pMS15-S ATCAAGG

Figure 6. The most frequently used sequence context around AUG triplets in eukaryotic mRNAs (top line) is compared to the sequence around the initiator codon from pMS12 and the three possible initiation codons from pMS 15. Initiator codons are underlined.

peptide synthesis does not occur ifa codon-anti- codon recognition mismatches by two or three nucleotides (19). A polypeptide initiated from one of the three proposed initiation codons shown in Figure 2 and Figure 6 can clearly be precursor of a secreted protein (32, 33) and indeed each of the amino acid sequences shown in Figure 7 has the essential characteristics of a signal peptide: A variable N-terminal sequence followed by a core of hydrophobic amino acids preceeding the site Figure 5. Activity staining of secretion products. Elec- at which the signal sequence region is cleaved off trophoresis was carried out on a polyacrylamide gel from the mature protein. The only difference which was stained for a-amylase activity. Migration among the putative signal peptides is the N-ter- was from the top of the gel. Lane l: native mouse minal part, but this region is highly variable in salivary a-amylase. Lane 2: a-amylase (initiation secreted proteins (32, 33). All have the hydro- codon ATG) secreted from yeast cells transformed with phobic core as well as the region around the site plasmid pMS 12. Lane 3: a-amylase secreted from yeast for signal peptidase cleavage in common with cells transformed with plasmid pMS 15. normal pre-a-amylase. An unfavourable aspect is the residue between the N and the core region of each putative signal peptide but that has been found in other eukaryotic codons in eukaryotic cells occur in a restricted signal peptides as well (32, 33). None of the three sequence context where cNNAUGcA A is most aligned signal peptides has a net negative charge, frequently observed. Applying this observation which, in E. coli, is known to be detrimental to on the potential initiation codons in frame with translation (11, 31). the coding sequence of a-amylase (Figure 2 and Misreading of genetic information is believed Figure 6) it may be seen that all potential to take place at a low frequency either due to initiation sites have an A in position -4. The incorrect aminoacyl-tRNA selection or due to sequence pMS 15-I with ATC as potential initia- improper codon-anticodon recognition. Several tion codon resembles the normal sequence in suppressor tRNA are known (7, 15, 20, pMS 12 and the consensus most. It is not be- 21), and one cannot exclude that is lieved that other initiation codons than those the N terminal amino acid in the signal peptide shown in Figure 2 and Figure 6 are used because synthesized from pMS 15. Thus, the main diffe-

Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 87 O. OLSEN" Non-AUG initiation of translation

Sp12 Met Lys Phe Phe Leu Leu ......

pMS12 ...... TGC CAA CAG CAT AGC AAA ATG AAA TTC TTC CTG CTG ......

pMS15 ...... TAT ACC AAG CAT ACA ATC AAG GAA TTC TTC CTG CTG ......

sp15-L MetLys His Thr Ile Lys Glu Phe Phe Leu Leu ...... Sp15-1 Met Lys Glu Phe Phe Leu Leu ...... Ile Met Glu Phe Phe Leu Leu ...... Sp15-S Lys

Figure 7. Comparison of signal peptides possibly synthesized from pMSI5 with the authentic a-amylase signal peptide. The nucleotide sequences around the initiation codon of pMS 12 and the putative initiation codons of pMS 15 are shown. The authentic signal peptide, sp 12, is aligned with the putative signal peptides spl 5-L, spl 5-I and sp 15-S. Suppressor mutations may result in a methionyl residue at the N-termini of the pMS 15 signal peptides instead of either a or an residue.

rences between the signal peptides that may be ACKNOWLEDGEMENTS synthesized from pMS15 and the authentic KARL KRISTIANTHOMSEN is acknowledged for signal peptide, spl 2, is the length of the amino scientific and practical guidance and encourage- terminal region and the glutamic acid residue ment throughout this work. KARINA BOSERUP is already mentioned. thanked for excellent technical assistance, and From the alignment of the signal peptides in NINA RASMUSSENfor preparing the figures. JORN Figure 7 it is not possible to predict which TONNES NIELSEN, University of Aarhus, is initiation codon is used by the translation ap- thanked for providing mouse salivary a-amy- paratus, but translation of mRNAs (transcribed lase. DITER VON WETTSTEIN is thanked for sup- from a transcription plasmid) in a yeast ho- port and encouragement and for critical reading mologous translation system followed by N-ter- of the manuscript. This work was supported by minal amino acid analysis of produced ~t-amy- a grant No. 1984-133/001-84.127 from Tekno- lase protein should make such a determination logistyrelsen of The Ministry of Industry to possible. DITER VON WETTSTEIN and M.C. KIELLAND- BECERRA et al. (2) speculate that use of non- BRANDT. AUG as initiation codon for translation of the AAV-B protein could regulate the relative syn- REFERENCES thesis of proteins B and C in vivo if these proteins 1. ANDERSON,C. W. &E. BUZASH-POLLERT:Can ACG are translated from the same mRNA. Assuming serve as initiation codon for protein synthesis in the initiation signal for AAV-B protein (initia- eucaryotic cells? Mol. Cell. Biol. 5, 3621-3624 tion codon ACG) is weak compared to the signal (1985) for AAV-C protein (initiation codon AUG), 2. BECERRA, S, P., J. A. ROSE, M. HARDY, B. M. then it would be possible for the cell to obtain a BAROUDY& C. W. ANDERSON: Direct mapping of fixed ratio of B protein/C protein. adeno-associated virus capsid proteins B and C: A Analagous control systems may operate in possible ACG initiation codon. Proc. Natl. Acad. Sci. USA 82, 7919-7923 (1985) yeast cells. Perhaps the sequence context around 3. BIRNBOIM,H. C. &J. DOLY:A rapid alkaline extrac- one - or more - of the non-AUG initiation tion procedure for screening recombinant plasmid codons reported on here is homologous to se- DNA. Nucleic Acids Res. 7, 1513-1523 (1979) quences in the yeast genome where translation 4. BLOOR,J. H. &M. H. MElSLER:Additional evidence initiation events are regulated in a way compara- for close linkage of Amy-1 and Amy-2 in the ble to synthesis of AAV-B and AAV-C proteins. mouse. J. Hered. 71,449-451 (1980)

88 Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 O. OMEN: Non-AUG initiation of translation

5. BODARY, S., G. GROSSI, O. HAGENBUCHLE & P. K. codon-anticodon recognition in a simplified cell- WELLAUER:Members of the Amy-2 alpha-amylase free system. Biochemistry 25, 6391-6397 (1986) gene family of mouse strain CE/J contain duplicat- 20. NORMANLY,J. -M. MASSON, L. G. KLEINA, J. ABEL- ed 5' termini. J. Mol. Biol. 182, 1-10 (1985) SON & J. H. MILLER: Construction of two ES- 6. CHIRIIOIAN,J. G. & J. GEORGE: In: Gene Amplifica- cherichia coli amber suppressor genes: tRNAcuPhe A tion and Analysis I, p. 74. Ed. J. G. Chirikjian, and tRNAcucys A. Proc. Natl. Acad. Sci. USA 83, Elsevier North Holland, Inc., New York (1984) 6548-6552 (1986) 7. CIGAN,M. &T. E DONAHUE:Translation initiation: 21. PEARSON, D., I. WILLIS, H. HOTTINGER,J. BELL, A. mutational analysis oftRNAiMet. Cold Spring Har- KUMAR, U. LEUPOLD & D. SOLE: Mutations pre- bor Lab., Abstr. Pap. Meet. Mol. Biol. Yeast, p. 40 venting expression of sup_3 tRNA s~ nonsense sup- (1985) pressors of Schizosaccharomyces pombe. Mol. 8. COHEN,S. N., A. C. CHANG& L. HSU: Nonchromo- Cell. Biol. 5, 808-815 (1985) some antibiotic resistance in bacteria. Genetic 22. PETERSEN, J. G. L., M. C. KIELLAND-BRANDT, S. transformation of by R-factor HOLMBERG & T. NILSSON-TILLGREN: Mutational DNA. Proc. Natl. Acad. Sci. USA 69, 2110-2114 analysis of isoleucine- biosynthesis in Sac- (1972) charomyces cerivisiae. Mapping of ilv2 and ilv5. 9. HAGENBI3CHLE.O., R. BOVERY& R. A. YOUNG:Tis- Carlsberg Res. Commun. 48, 21-34 (1983) sue-specific expression of mouse a-amylase genes: 23. POLISKY, B., P. GREENE, D. E. GAR~qN, B. J. MC- nucleotide sequence of isoenzyme mRNAs from CARTHY, H. M. GOODMAN & n. W. BOYER: pancreas and salivary gland. Cell 21, 179-187 Specificity of substrate recognition by the EcoRI (1980) restriction endonuclease. Proc. Natl. Acad. Sci. I0. HJORT, J. P., A. J. LUSIS & J. T. NIELSEN: Multiple USA 72, 3310-3314 (1975) structural genes for mouse amylase. Biochem. 24. RAJBHANDARY, U. L. & n. P. GHOSH: Studies on Genet. 18, 281-301 (1980) polynucleotides. J. Biol. Chem. 244, 1104-1113 11. INOUYE, S., X. SOBERON, T. FRANCESCHINI, K. (1969) NAKAMURA, K. ITAKURA& M. INOUYE: Role of 25. RASMUSSEN,S. IC, H. E. HOPP & A. BRANDT:Nucle- positive charge on the amino-terminal region of otide sequences of cDNA clones for B1 hordein the signal peptide in protein secretion across the polypeptides. Carlsberg Res. Commun. 48, 187- membrane. Proc. Natl. Acad. Sci. USA 79, 3438- 199 (1983) 3441 (1982) 26. SANGER, F., S. NICKLEN & A. R. COULSON: DNA 12. Ixo. H., Y. FUKUDA, K. MURATA& A. KIMURA: sequencing with chain-terminating inhibitors. Transformation of intact yeast cells treated with Proc. Natl. Acad. Sci. USA 74, 5463-5467 (1977) alkali cations. J. Bacteriol. 153, 163-168 (1983) 27. SICK,K.&J.T.NIELSEN: Genetics of amylase isoen- 13. KOZAK, M.: Comparison of initiation of protein zymes in the mouse. Hereditas 5 l, 291-296 (1964) synthesis in procaryotes, eucaryotes, and or- 28. STRUHL, I(~, D. T. STINCHCOMB, S. SCHERER& R. W. ganelles. Microbiol. Rev. 47, 1-45 (1983) DAVIS: High-frequency transformation of yeast: 14. LAEMMLI.U. K.: Cleavage of structural proteins Autonomous replication of hybrid DNA during the assembly of the head of bacteriophage molecules. Proc. Natl. Acad. Sci. USA 76, 1035- T4. Nature 227, 680-685 (1970) 1039 (1979) 15. LIEBMAN,S. W. & J. A. ALL-ROBYN:A non-Mende- 29. THOMSEN,K. K.: Mouse a-amylase synthesized by lian factor, [eta+], causes lethality of yeast omnipo- Saccharomyces cerevisiae is released into the cul- tent-suppressor strain. Curt. Genet. 8, 567-583 ture medium. Carlsberg Res. Commun. 48, 545- (1984) 555 (1983) 16. MANIAT1S. T., E. F. FRITSCH & J. SAMBROOK: In: 30. VIEIRA, J. & J. MESSING: The pUC plasmids, an Molecular Cloning: A laboratory manual. Cold M 13mp7-derived system for insertion mutagene- Spring Harbor Laboratory, Cold Spring Harbor, sis and sequencing with synthetic universal New York, pp 1-540 (1982) primers. Gene 19, 259-268 (1982) 17. MESSING,J.: A multi-purpose cloning system based 31. VLASUK,G. P, S. INOUYE, H. ITO, K. ITAKURA& M. on the single-stranded DNA bacteriophage M 13. INOUYE:Effects of the removal of basic amino acid Recombinant DNA Technical Bulletin, NIH pub- residues from the signal peptide on secretion of lications No. 79-99, 2, No. 2, pp. 43-48 (1979) lipoprotein in Eschericia coli. J. Biol. Chem. 258, 18. MESSlNG,J.: New M13 vectors for cloning. Meth- 7141-7148 (1983) ods Enzymol. 101, 20-79 (1983) 32. YON HEIJNE, G: Patterns of amino acids near 19. NEGRE, D., A. J. COZZONE & Y. CENATIENO: Accura- signal-sequence cleavage sites. Eur. J. Biochem. cy of natural messenger translation: analysis of 133, 17-21 (1983)

Carlsberg Res. Commun. Vol. 52, p. 83-90, 1987 89 O. OLSEN: Non-AUG initiation of translation

33. VON HEIJNE, G.: How signal sequences maintain cleavage specificity. J. Mol. Biol. 173, 243-251 (1984) 34. ZITOMER,R. S., D. A. WALTHALL,B. C. RYMOND& C. P. HOLLENBERG:Saccharomyces cerevisiae recog- nize non-AUG initiation codons. Mol, Cell. Biol. 4, 1194-1197 (1984)

Accepted by E LUND

90 Cadsberg Res. Commun. Vol. 52, p. 83-90, 1987