The EMBO Journal Vol.5 no.l2 pp.32l9—3225, 1986
`
`A synthetic operon containing 14 bovine pancreatic trypsin
`inhibitor genes is expressed in E. coli
`
`Mylan V. Genentech
`
`IPR2016-00710
`
`Genentech Exhibit 2071
`
`Brigitte von Wilcken-Bergmann, Daniela Tils, Jiirgen
`Sartorius, Ernst August Auerswald‘, Werner Schroder‘
`and Benno Miiller-Hill
`
`Institut fiir Genetik der Universitat zu Koln, Weyertal 121, 50()() Koln 41,
`and ‘Bayer AG, Institut fiir Biochernie, Postfach 10 17 01, 560() Wuppertal
`l, FRG
`
`Communicated by B.Miiller-Hill
`
`A synthetic gene encoding the protein sequence of mature
`bovine pancreatic trypsin inhibitor (BPTI) has been cloned
`into a novel E. coli expression vector. After in vitro gene
`amplification by successive DNA duplications, more than
`600 000 mostly inactive inhibitor molecules may be recovered
`from a single cell. After purification the inhibitory activity
`can be reconstituted almost completely. The specificity of
`BPTI for trypsin is abolished by a single amino acid exchange
`from lysine to isoleucine at position 15. The altered protein
`is shown to be an efficient inhibitor of human leukocyte
`elastase.
`
`Key words: E. coli expression vectors/in vitro gene amplifica-
`tion/overproduction of BPTI/protein engineering/change of
`specificity of BPTI.
`
`Introduction
`
`Bovine pancreatic trypsin inhibitor (BPTI, or aprotinin) is a small,
`rather basic protein of 58 amino acids, which is purified from
`several bovine organs i.e.
`lung, pancreas or parotid glands
`(Kunitz, 1947; Fritz and Wunderer, 1983). BPTI inhibits tryp-
`sin most efficiently but it also acts as an inhibitor of chymotryp-
`sin, plasmin and kallikrein® (Fritz and Wunderer, 1983). Its
`structure (Deisenhofer and Steigemann,
`1975),
`fimction
`(Laskowski and Kate, 1980) and folding pathway (Creighton,
`1978) have been studied intensively.
`The recent analysis of cloned bovine DNA confirms that the
`mature inhibitor is processed from a larger precursor polypep-
`tide by proteolytic cleavage of both termini (Laskowski and Kate,
`1980; Anderson and Kingston, 1983) and that the bovine BPTI
`gene may be a member of a family of closely related proteins
`that have acquired different inhibitory specificities by few amino
`acid exchanges (Kingston and Anderson, 1986). Substitution of
`one amino acid may alter the substrate specificity of a protein.
`This has been shown for the lactose permease by Markgraf et
`al. (1985), and data concerning the serpins are reviewed by Carrel
`(1984). BPTI could be converted semi—synthetically into efficient
`human leukocyte elastase inhibitors by substituting valine,
`leucine, or methionine for the lysine residue at position 15
`(Wenzel and Tschesche, 1981; Wenzel et al., 1985). It was
`shown further that monosubstitution of lysine-15 by isoleucine,
`oz-amino butyric acid, norvaline or norleucine yields inhibitors
`of leukocyte elastase (Tschesche et al., 1985). Similarly,
`chymotrypsin is inhibited more efficiently than trypsin when tryp-
`tophan or phenylalanine are incorporated at position 15 of the
`inhibitor (Jering and Tschesche, 1976; Wenzel and Tschesche,
`
`1984). Proteases and their specific inhibitors play an important
`role in several biological processes such as food digestion, blood
`coagulation and fibrinolysis, and some aspects of the immune
`response (Holzer and Tschesche, 1979). Diseases such as
`pulmonary emphysema or rheumatoid arthritis have been related
`to an imbalance between a proteinase and its specific natural in-
`hibitor (Mittmann, 1972; Barrett, 1978; Menniger and Mohr,
`1981). The clinical applications of BPTI and its properties as a
`therapeutic agent (Trasylol®) for the treatment of various diseases
`such as hyperfibrinolytic haemorrhage, acute pancreatitis,
`myocardial
`infarction, or traumatic haemorrhagic shock are
`reviewed in detail by Fritz and Wunderer (1983). Because of
`its comparatively small size, its stability and its frequent use as
`a model globular polypeptide (Wagner and Wuthrich, 1982;
`Karplus and McCamrnon, 1981) BPTI seems to be particularly
`well suited as a model system to study the possibilities of pro-
`tein design as has been suggested by Marks and Anderson (1984).
`Thus we have set up a system where a synthetic BPTI gene
`may be manipulated in vitro for fundamental scientific, as weH
`as for practical purposes, and expressed in E. coli to yield the
`altered protein.
`
`Results
`
`In order to facilitate the cloning of the synthetic bovine pancreatic
`trypsin inhibitor (BP'I'I) gene we constructed new plasmid vec-
`tors by combining slightly modified DNA fragments from
`pBR322 (Sutcliffe,
`1978), pKO60 (Besse et al., 1986),
`bacteriophage fdll (Beck et al., 1978) and synthetic promoter
`sequences (Figure 1A). The upstream sequences and the -35
`region of the synthetic promoter correspond to the bacteriophage
`T5 P25 promoter (von Gabain and Bujard, 1979). The Pribnow
`box and the downstream sequences are close to the consensus
`sequences (von Hippel et al., 1984). These sequences (see Figure
`3) proved to be a very efficient promoter as judged by the amounts
`of B-galactosidase synthesized under the control of this promoter
`(see Figure 4, lane 1).
`The BPTI gene was constructed from 14 synthetic complemen-
`tary overlapping oligonucleotides to yield a double-stranded DNA
`segment of an overall length of 204 bp, which was cloned be-
`tween the }07aI and HindIII sites of piWiT9 (Figure 1A). The
`synthetic sequences include a ribosomal binding site at an ap-
`propriate distance upstream of the initiation codon ATG. The
`coding sequence for mature BPTI was linked to the lacZ gene
`in phase by an intervening amber codon that is partially translated
`into a tyrosine in the recipient strain Su3. Thus colonies har-
`bouring recombinant plasmids expressed a BPTI-B-galactosidase
`fusion protein and could be identified by their ability to hydrolyse
`X—Gal on indicator plates.
`The DNA sequence was determined completely from both
`strands (Maxam and Gilbert, 1983); it proved that the synthetic
`oligonucleotides had been joined and cloned correctly. The
`amount of fusion protein was quantified by determining
`fl-galactosidase activity in crude cell extracts (Miller, 1972). It
`
`© IRL Press Limited. Oxford, England
`
`3219
`
`Mylan v. Genentech
`IPR2016-00710
`Genentech Exhibit 2071
`
`

`
`teolytic activities, and we used the unique Xbal site situated
`upstream of the ribosomal binding site, and the unique HindIH
`site downstream of the amber stop codon, to duplicate the BPTI
`gene. For this purpose plasmid DNA was digested with BamHI
`and HindI[I and the smaller fragment, containing the 3' end of
`the tetracycline resistance genes, the origin of replication, the
`promoter and the BPTI gene was purified. A second sample of
`the same plasmid DNA was digested with BamHI and Xbal to
`Ecom
`
`ma 1
`
`Hind lll
`
`B.von Wilcken-Bergmarm et al.
`
`was ~50% of the B-galactosidase activity observed in the
`presence of a fully induced episomal wild-type lac operon, but
`this extract did not specifically inhibit trypsin. The reason for
`this failure was a rapid breakdown of the small peptide as
`demonstrated by pulse chase experiments. Figure 2 shows the
`results obtained with the [on mutants SG935 and SG936.
`
`An attempt was then made to raise the rate of synthesis to a
`level above what was needed to saturate the intracellular pro-
`
`A
`
`Clal
`
`ECORI
`
`
`
`7
`
`
`
`piWi T10 wL1
`~ 6.6 kb
`
`3
`
`a
`
`b
`
`Fig. 1. (A) Physical map of the cloning vector piWiT9. The 332-bp SauA fragment from bacteriophage fdll which carries a transcription termination signal
`(ter -1 ) is shown black. It was linked to the 2.070—bp EcoRI—PvuH fragment from pBR322 which confers resistance against tetracycline (Tck). The pBR322
`derived origin of replication (ori ->) was excised from pK060 with Ahalll and M101, and the lacZ fragment extending from codon 6 to a few bp beyond the
`three temiination codons was also taken from pK060. Synthetic promoter sequences » were inserted between the XhoI and the HindIH site. The lacZ gene
`is virtually not expressed at all from piWi'I‘9 because it lacks an initiation codon for the start of translation. Some restriction sites are indicated. The unique
`Jdml site and the unique HindHI site which were used for the cloning and for the amplification of the synthetic inhibitor gene are emphasized by bold letters.
`Parallel to the cloning of the synthetic gene a small fragment providing a ribosomal binding site and a start codon for the [M2 gene was inserted between the
`Hull and HindIII sites in order to allow a test of the quality of the synthetic promoter. This IacZ+ derivative was named piWiT10 and used as a control
`plasmid (see Figure 3). (B) Physical map of the expression plasmid piWiTl0wLl. The 14 inhibitor genes numbered 1 to 14 are shown black. For further
`details see Figure 2. All the other symbols are the same as in A. (C) 0.9% agarose gel showing piWiTl0wLl DNA digested completely with HindIII and
`partially with BssHII for different times (a) 15 min, (b) 35 min, (c) 75 min. Bssl-III cuts between every two inhibitor genes (see Figure 2).
`
`3220
`
`

`
`— 6.5
`
`-12.5
`
`— 21.0
`
`Fig. 2. Autoradiograph of a l7.5% SDS—polyacrylarnide gel. SG935 (lanes
`1 and 3) and SG936 (lanes 2 and 4) cells harbouring plasmids with two
`tandemly repeated BPTI genes were pulse labeled in viva according to the
`protocol of McCarthy er al. (1985) using [358]-cysteine. The chase was
`1 min (lanes 1 and 2) and 1 h (lanes 3 and 4). The samples of
`unfractionated cell lysates correspond to -107 cells. lane 5: mol. wt
`standards.
`
`Expression of synthetic operon in E. coli
`
`yield a fragment which contained the BPTI gene again, the lacZ
`gene, the transcription termination signal, and the 5’ end of the
`tetracycline resistance genes. Both fragments were ligated in the
`presence of the synthetic adapter molecule
`5’ AGCTAATGAGCGCGC 3’
`3’
`TTACTCGCGCGGATC 5’
`
`and transformed into E. coli BR17, a RecA— strain. Since the
`adapter molecules carry protruding ends complementary to the
`HindIII— and }02aI—generated single strands which are readily
`ligated but do not regenerate either site, the resulting new plasmid
`again has only one unique )02aI site upstream of the first BPTI
`gene, and one unique Hindlll site downstream of the second BPTI
`gene. The series of reactions described above were repeated with
`this new plasmid DNA to yield another plasmid with four tandem-
`ly repeated BP'TI genes, each preceded by its own ribosomal bind-
`ing site. Prior to the next duplication the 3—kb EcoRI fragment
`containing the bulk of the lacZ gene was excised with EcoRI and
`deleted from the plasmid in order to reduce its overall size. After
`the third doubling of BPTI genes we happened to pick up a
`plasmid with seven BPTI genes instead of eight. Plasmids with
`odd numbers of BPTI genes were usually found among the pro-
`ducts of the ligation reactions at varying frequencies. We attribute
`their rise to partial denaturation with subsequent false hybridisa-
`tion during the melting of the seaplaque agarose (Maniatis et al.,
`1982). We never observed any deviation from a given number
`of genes when these plasmids were propagated further uninduc—
`ed in a RecA— strain.
`
`.
`
`.
`
`.
`
`8
`E
`. . . ..pBR322 DNAN
`. .GCGAGTCAGTGAGCGAGGAAGCGGAAGAGCG&CAATACGCAAAACCGCCTCTCCCCGCGCC CGAGATAAAAAATTTACAGGTACAATTCTTGA
`
`codons 322-341 of lacI
`
`-35
`
`promoter
`
`TATTTTTTAAAT-
`
`"'GTCCATGTTAAGTTCA
`
`. .
`
`.
`
`. . . ..CGCTCAGTCACTCGCTCCTTCGCCTTC GCGGGTTATGCGTTTTGGCGGAGAGGGGCGCGGGAGC
`HaeII
`XhoI
`
`lac operator
`mRNA
`-10
`*
`l\/\/\>
`TATAA 'TTATCATCTAGCAAATXGEGAQQGGATAAQAATTTGCACACAGCTAGATAATAAAAATTTAAQQQA TCTT ATG ACA AGC TTT CTA GAT TAAAQQAQTTATCTT
`A ATTTCCTCAATAGAA
`ATATT‘ AATAGTAGATCGTTTAACACTCGCCTATTGTTAAACGTGTGTCGATCTATTATTTTTAAATTCCCT
`AGAA TAC TGT TCG
`GAT
`EcoRV
`HindIII XbaI
`
`RBS
`
`RBS
`
`10
`5
`1
`25
`20
`15
`met arg pro asp phe cys leu glu pro pro tyr thr gly pro cys lys ala arg ile ile arg tyr phe tyr asN ala lys ala gly leu
`TGC AAA GCT CGT ATC ATC CGT TAC TTC TAC AAT GCA AAG GCA G
`CTG
`ATG AGA CCA GAT TTC TGC
`C GAG CCG CCG TAC ACT GGG C
`TAC TCT GGT CTA AAG ACG GAG C
`GGC GGC ATG ATG
`GAG
`C GGG ACG TTT CGA GCA TAG TAG GCA ATG AAG ATG TTA CGT TTC CGT C
`StuI
`XhoI
`ApaI
`
`40
`35
`30
`55
`S0
`45
`cys g1N thr phe val tyr gly gly cys arg ala lys arg asN asN phe lys ser ala glu asp cys met arg thr cys gly gly ala
`CGC
`CTT CTG A
`TAC GCA TGA ACG CCA CCA CGA ATC
`ACA GTC TGG AAG CAT
`G CCG CCG ACG TCT CGA TTC GCA TTG TTG RAG TTT AGG
`TGT CAG ACC TTC G
`TAC GGC GGC TGC
`A GCT AAG CGT AAC AAC TTC AAA TCCf§§3 GAA GAC TGC ATG CGT ACT TGC GGT GGT GCT TAG
`AccI
`PstI
`SstII
`SphI
`
`GC-‘GCTAATGAO
`CGTTCG-ITACTCGCGO
`BssHII
`
`3
`2
`1
`RES
`met arg pro asp phe . . .
`0 TAGATTAAAQQQQTTATCTT ATG AGA CCA GAT TTA ...
`0 AATTTCCTCAATAGAA TAG TCT GGT CTA AAT ...
`
`. ..
`
`.
`
`
`58
`S7
`56
`55
`... cys gly gly ala
`.. TGC GGT GGT GCT TAG
`. ACG CCA CCA CGA ATC
`
`GC GCTTTCCGGG
`CGTTCG
`GGCCCCTT
`
`TTCCA TG
`GT AC
`
`13 HindIII
`
`EcoRI PvuII
`
`2
`AGCGCCGGTCGCTACCATTACCAGTTGGTCTGGTGTCAAAAATAATAATAACCGGCAAGGGGATEGATCCCGCAAAAGCGGCCTTTGACTCCCT.....
`TCGCGGCCAGCGATGGTAATGGTCAACCAGACCACAGTTTTTATTATTATTGGCCGTTCCCCTAGGTAGGGCGTTTTCGCCGGAAACTGAGGGA.....
`codons 1007-1023 of lacz
`(Cla'I)
`§
`§ fdll DNA”...
`
`Fig. 3. DNA sequence of the right half of plasmid piWiT10wL1. Restriction enzyme cutting sites are indicated by arrows. The Clal site in the bottom line is
`bracketed because the enzyme will not cut here unless the DNA is prepared from a dam_ strain. The — l0 and -35 promoter elements are shown in boxes.
`The presumed start of transcription is indicated by a wavy arrow. The lac operator sequences are shown by a dotted line between the DNA strands and by an
`asterisk above the central G/C pair. The ribosomal binding sites (RBS) are underlined. The protein sequence of the inhibitor is shown above its DNA. The
`adapter oligonucleotides used for joining the sticky HindIH and fial ends during the amplification of the genes are boxed. The 3183 bp from the Xhol site
`upstream of the promoter and the HindIH site downstream of the fourteenth inhibitor gene are entirely of synthetic origin. The numbering of the flanking
`pBR322 and fdll DNA corresponds to the numbering of Sutcliffe (1978) and Beck et al. (1978).
`
`322 l
`
`

`
`RPDFCLZPPYTGPCK —>
`
`
`
`PDPCLBPPYTGPCK
`
`bacterial arr:
`
`
`
`cjrclll IP11
`of bovine orlqlne
`
`
`
`‘III:
`
`[min]
`
`Fig. 5. Comparison of tryptic peptides of bacterial BPTI and commercial
`BPTI. The superimposed HPLC elution profiles are identical except for the
`N—terminal peptide which carries an additional methionine in case of the
`bacterial BPTI. For comments on the small extra peak in the bacterial BPTI
`peptide profile between 48 and 49 min see Discussion.
`
`upstream of the first BPTI gene in addition to the initial pro-
`moter sequences.
`In the presence of these sequences )8-
`galactosidase expression dropped to about 20-25 % of the yields
`obtained in their absence,
`in contrast to the BPTI synthesis
`(Figure 4).
`This alteration raised the BP'TI synthesis from levels insuffi—
`cient to be seen on protein gels, to more than 2% of the total
`cellular protein (Figure 4, lane 2). With 32 such genes (Figure 4,
`lane 3) the amount of protein is almost doubled, but with seven
`or eight genes no band can be seen neither in complete cell ex-
`tracts nor in the insoluble fraction (data not shown).
`The BPTI synthesized in E. coli was found to be predominantly
`insoluble when the cells were lysed with lysozyme (Figure 4).
`Inactive BP'I‘I was obtained almost pure after washing cell debris
`twice with 2 M urea (Figure 4). It was then dissolved in 6 M
`guanidinium hydrochloride and renatured according to the pro-
`cedure of Creighton (1985, see also Materials and methods).
`Bacterial BPTI was obtained in pure form after perchloric acid
`precipitation and affinity chromatography on trypsin sepharose.
`The amino acid composition was determined (data not shown)
`and agreed well with the predicted values and the protein sequence
`of the first twenty residues was verified. A comparison of pep-
`tide maps of bovine and bacterial BPTI is shown in Figure 5
`(Mayes, 1985).
`Inhibition of trypsin was measured (Nakajima et al., 1979) and
`found to be almost as efficient as with bovine BPTI (Table I).
`The yields were 1 pg soluble BPTI per 109 cells, with 5 pg
`renatured BPTI from the insoluble material. This corresponds
`to at least 2% of the total cellular protein or 600 00() molecules
`per cell.
`Making use of the unique Apal and Stul sites within the DNA
`sequence of the synthetic BPTI gene we exchanged the lysine
`codon at position 15 (Figure 3) for an isoleucine codon, amplified
`the modified gene by five successive duplications as described
`above, and analysed the gene product. The I1e,5BPTI produced
`from piWiT10wi7 migrates slightly faster on SDS—polyacryl—
`amide gels (Figure 4). A similar behaviour has been reported
`for semi-synthetically produced Valls BPTI (Wenzel et al., 1985).
`The purified Ile,5BP'I‘I no longer inhibits trypsin, but it is an ef-
`ficient inhibitor of human leukocyte elastase (Table 1).
`
`Discussion
`
`From the known amino acid sequence (Kassel et al., 1965) of
`bovine pancreatic trypsin inhibitor (BPTI) we derived a cor-
`
`B.von Wilcken-Bergmann et al.
`
`KD
`92.0—
`66.2—
`
`45.0 A
`
`31.0—
`
`21.5-
`
`14.4 —
`
`Fig. 4. 15% reducing SDS-polyacrylarnide gel stained with Coomassie Blue.
`M 1
`: mol. wt standards; lanes 1-3: complete cell lysates corresponding to
`2 X 10‘ cells each. The host is E. coli BRI7 canying (lane 1) a li-
`galactosidase overproducing plasmid (piWiTl0, see legend to Figure IA),
`(lane 2) the BPTI plasmid piWiTl0wL1, and (lane 3) the altered l]e,5BPTI
`producing plasmid piWiTl0wi7). Lanes 4 and 5 show the soluble protein
`fractions of the lysates of lanes 2 and 3 respectively. Lanes 6 and 7: pellets
`from 10° (6) and 2 X 109 (7) cells of E. coli BRI7 harbouring plasmid
`piWiTl0wL1, The pellets have been washed with 0.5% SDS and 2 M urea.
`Lanes 8 and 9 same as 6 and 7 but with plasmid piWiTl0wi7. Lane 10:
`2.5 pg commercial BPTI. The inhibitor band is indicated by an arrow.
`
`Table I. Comparison of trypsin and elastase inhibition
`
`Inhibitor
`
`Commercial
`BPTI
`
`Bacterial
`BPTI
`
`I]e,5BPTI
`
`50% Inhibition
`Trypsm
`
`250 ng
`
`300 ng
`
`-3
`
`Elastase
`
`-3
`
`-3
`
`200 ng
`
`2 pg trypsin and 0.5 pg elastase were used for the inhibition assays.
`“There was no measurable inhibition in the presence of 10 pg of the
`respective inhibitors.
`
`When 14 BPTI genes had been assembled in this manner
`(Figure 1, B and C) cell extracts were examined on protein gels
`after silver staining (Biirk et al. , 1983). There was a faint band
`co—migrating with commercial BP'TI, which seemed to be slightly
`more intensive in the extract prepared from the cells harbouring
`the BPTI producing plasmid than in the control extract which
`did not contain any BPTI. We then altered the sequences between
`the promoter and the first structural gene of this artifical operon,
`because we had observed that the length and composition of the
`5’ transcribed sequences could have a dramatic effect upon the
`expression of B-galactosidase or the BP'TI—ga1actosidase fusion
`protein. The various mRNA leader sequences that have been
`studied in connection with different genes will be published and
`discussed elsewhere. The sequence which gives the highest yield
`with the BPTI genes is presented in Figure 3. It contains the
`natural lac operator sequences and a short open reading frame
`
`3222
`
`

`
`responding DNA sequence (Figure 3) using, preferentially,
`codons that occur frequently in highly expressed genes of E. coli
`(Grosjean and Fiers, 1982). However, we did not hesitate to
`depart from this rule whenever we could introduce a unique
`restriction site. Thus we used the AGA codon for Arg3, in order
`to generate the Pstl site (Figure 3). We also used the AGA codon
`for Arg, in order to enhance ribosome binding (Scherer et al.,
`1980) since Robinson et al. (1984) have reported that transla-
`tion is not impaired by a cluster of four such unfavourable codons
`so long as the protein production does not exceed 14% of the
`total cellular protein.
`Seven unique restriction sites distributed throughout the gene,
`in addition to the flanking Xbal and HindIII sites, allow the altera-
`tion of the primary structure of the gene by exchanging small
`restriction fragments for freshly prepared oligonucleotides. Thus
`nearly all amino acids may be exchanged, and the influence of
`codon usage as well as the function of sequences in the 5' non-
`coding region can be investigated. We have introduced an
`isoleucine at position 15 and obtained an altered inhibitor which
`has lost its affinity for trypsin, but efficiently and specifically
`inhibits human leukocyte elastase (Table I), an enzyme which
`has been implicated to cause severe damage to lung tissue when
`released in excess (Cochrane et al., 1983; Gadek et al., 1979).
`It remains to be seen whether this engineered elastase inhibitor
`will be of therapeutic use for the treatment of pulmonary
`emphysema.
`The most severe impediment that usually obstructs the pro-
`duction of small foreign proteins in E. coli is the fact that these
`are liable to be degraded quickly by the host proteases. Such
`foreign peptides have been protected successfully against pro-
`teolytic attack by fusing them to E. coli proteins (Maniatis et al. ,
`1982; Kempe et al., 1985) or to each other (Shen, 1984; Kempe
`et al. , 1985). However, such a strategy is only applicable as long
`as there are means to cleave the fused peptides and gain the
`desired product intact. In our case, pulse—chase experiments
`(Figure 2) had indicated that BP'I'I synthesized at a high rate from
`two copies of the gene was equally rapidly degraded. We
`therefore employed a different strategy which should be applicable
`to a wide range of small proteins, especially those which cannot
`be cleaved intact from a fusion because they contain all the amino
`acid residues in their sequence which can be used for specific
`cleavage at the fusion points. Leeiet al. (1984) have reported
`previously on the cloning of up to four tandem repeats of an in-
`terferon gene and pointed out the economy of such a procedure.
`We have amplified the synthetic BPTI gene in order to accelerate
`the synthesis until the production of BPTI was so much faster
`than the proteolytic breakdown that the product could accumulate
`and finally reach an intracellular concentration, where we hoped
`it would precipitate and thus be inaccessible to the host proteases.
`Similarly temperature-sensitive runaway replicons have been used
`to enrich cell extracts for unstable proteins (Brent and Ptashne,
`1981). A further enhancement could be envisaged by combin-
`ing both strategies. From our yields we have calculated a final
`cellular concentration of 0.5 mM for the BPTI peptide which
`is a 50-fold excess over the concentration of B-galactosidase
`monomers in a fully induced wild-type E. coli. These values
`make it highly probable that the insoluble fraction of the inhibitor
`recovered from the sediment indeed reflects the degree of in-
`tracellular precipitation of the overproduced peptide.
`The use of two different restriction sites and a synthetic adapter,
`which after ligation destroys both sites, greatly facilitates the ac-
`cumulation of more than four genes. In our case the ribosomal
`
`Expression of synthetic operon in E. coli
`
`binding site was an integral part of the gene segment. Altema-
`tively it might be positioned on the adapter molecule. Our adapter
`carries additional termination codons in phase (TAATGA, see
`Figure 3) because the single amber codon proved to be slightly
`leaky in terminating transcription. The additional peak in the tryp-
`tic map of the bacterial BPTI at 48-49 min (Figure 5) is most
`probably due to trace amounts of C-terminal peptide lengthened
`by three residues.
`Plasmid piWiT10wL1, as well as its analogue piWiT10wi7
`which carries 32 Ile,5BPTI genes, is completely stable as long
`as it is propagated in any RecA‘ lacIQ strain in the absence of
`the inducer IPTG. Plasmid piWiT10wL1 which carries 14 BP'I'I
`genes is also reasonably stable when grown in the presence of
`IPTG in E. coli BR 17, the strain that gives the highest protein
`yields; after > 60 generations less than 5 % of the plasmid DNA
`exhibited a reduction in size that would correspond to a loss of
`one or two BP'I‘I genes. Under the same conditions plasmid
`piWiT10wi7 with 32 Ile15BPTI genes was considerably less
`stable, the protein yield dropped significantly after 20 genera-
`tions in fully induced state. We do not yet know whether the
`selective pressure against the l1e,5BPTI synthesis originates from
`overstraining the synthesis capacity of the cells or from a toxic
`quality of the Ile15BPTI peptide, which is absent from nomial
`BPTI.
`
`During the last years data has been gathered which will con-
`tribute to an understanding of mRNA translational efficiency
`(Scherer et al., 1980; Hui et al., 1984; Schoner et al., 1984;
`Stanssens et al. , 1985). We can add another item to this collec-
`tion. We have constructed a strong promoter which caused an
`extremely high level of B-galactosidase synthesis (Figure 4,
`lane 1), but it did not suffice to express the BPTI gene to such
`an extent that the product could be detected on protein gels. The
`additional sequences introduced into the 5’ untranslated region
`of the mRNA which significantly enhanced the BPTI expression
`had an inverse effect when combined with the lacZ gene. These
`observations indicate that the translational efficiency of a mRNA
`does not so much depend on the sequences upstream of the in-
`itiation codon but on the interaction of the mRNA leader and
`
`the coding sequences of the gene.
`
`Materials and methods
`Bacterial strains
`
`E. coli Su3 is lac-pro mefargam SupF mi" and was kindly provided by J.Miller.
`E. coli SG 935 and SG 936 are F_, ladam), trp(am), pho(am), SupC(t5), rpsL,
`mal(am), htpR(am), tsx::Tn10, lon(am) (Buell et al., 1985). E. coli BBl.8 which
`is mal’ and XR and was obtained from P.Starlinger. BB1.8 was chosen because
`of its high efficiency of protein synthesis in comparison to about a dozen other
`E. coli strains (Weidemann, 1985). A RecA' allele was introduced by standard
`procedures (Miller, 1972) into this strain which was designated BR17.
`Plasmids
`
`pBR322 (Sutcliffe, 1978), pKO60 (Besse et al., 1986) and pUR278 (Riither and
`M|'iller—Hill, 1983) have been described. Bacteriophage fdll DNA (Beck et al.,
`1978) was a gift of B. Gronenbom.
`Enzymes and chemicals
`The restriction endonucleases, T4 DNA ligase, polynucleotide kinase, and DNA
`polymerase 1,
`large fragment, were purchased from Bethesda Research
`Laboratories (Neu-Isenburg, FRG), Boehringer (Mannheim, FRG) or New
`England Biolabs (Bad Schwalbach, FRG) and employed as recommended by the
`manufacturers.
`[3’P]-nucleotides were obtained from Amersham Buchler
`(Braunschweig, FRG); 5-bromo-4-chloro-3-indolyl-B-D-galactoside (X—Gal) and
`isopropyl-3-D—thiogalactoside (IPTG) from Bachem Fine Chemicals (Torrance,
`USA). The chemicals for DNA synthesis were purchased fi'om Applied Biosystems
`(Pfungstadt, FRG) and the chemicals for DNA sequence analysis (Maxam and
`Gilbert, 1983) have been listed previously (Biichel et al., 1980).
`
`3223
`
`

`
`B.von Wilcken-Bergrnann et al.
`
`Methods
`
`The methods for the construction of plasmids are standard techniques (Maniatis
`et al., 1982). Synthetic oligonucleotides were prepared automatically on an Ap-
`plied Biosystems 380A DNA synthesizer. Full-length molecules were purified
`from preparative denaturing polyacrylarnide gels after ethidium bromide stain-
`ing. When only two complernentary oligonucleotides were to be cloned they were
`used without adding 5’ phosphates. For the simultaneous cloning of up to 14
`oligonucleotides, the 5' ends were phosphorylated and "P was incorporated into
`a tenth of each oligonucleotide sample prior to a second polyacrylarnide gel elec-
`trophoresis, and the labelled bands were then cut from the gels. Removing the
`majority of unphosphorylated oligonucleotides by a second gel purification prior
`to cloning greatly enhances the yields when more than two oligonucleotides are
`to be cloned simultaneously.
`Transformation of E. coli was carried out according to the method of Hanahan
`(1983). E. coli proteins were analysed essentially as described by Marston et al.
`(1984) with the following modifications worked out by Weidemann (1985): cells
`from 3rnl ovemight cultures were harvested by a short centrifugation and
`resuspended in 122 ;:.l 40 mM Tris—HCl, 5 mM EDTA, pH 8 containing 0.3
`mg/ml lysozyme. Afier 30 min at room temperature 18 pl 8% sodium deoxycholate
`were added, after further 10 min the suspension was mixed with 150 pl 10 mM
`MgCl2, 20 mM Tris-HCl, pH 8 and 10 p.g DNaseI/ml, which was allowed to
`react for about 20 min. The sample was then either mixed with 600 pl reducing
`sample buffer (Laemmli, 1970), containing 8M urea and incubated at 37°C for
`1 h or fractioned first by 10 min centrifugation in an Eppendorf centrifuge. The
`pelleted cell debris was then washed with 40 mM Tris-HCl, 5 mM EDTA, pH 8
`containing 0.5% SDS or 2 M urea before it was dissolved in sample buffer and
`incubated at 37°C as above. Aliquots of the samples were then analysed on
`polyacrylamide gels as described (Laemrnli, 1970). Pulsechase experiments were
`performed as described by McCarthy et al. (1985).
`Active inhibitor was purified from the soluble fraction of the lysate by per-
`chloric acid precipitation followed by affinity chromatography on trypsin sepharose
`(Fritz et al., 1970). For the preparation of renatured BPTI by solid state folding
`(Creighton, 1985) cells from 5 1 overnight culture were harvested, disrupted in
`a French Press, and centrifuged for 20 min at 20 00() r.p.m. The pellet was dissolv-
`ed in 10 ml buffer A [8 M urea,
`1 mM bis-(hydroxyethyl)-disulfide,
`1 mM
`2-mercaptoethanol, 50 mM Tris-HCl, 1 mM EDTA, pH = 8.2] and reduced
`for 1 h at 50°C under nitrogen. The solution was then applied to an Econo col-
`umn (25X10 mm, Biorad, Miinchen, FRG) filled with ~10 ml CM Sepharose
`Fast Flow (Phamiacia, Freiburg, FRG) which was equilibrated with buffer A.
`The column was then developed by a linear gradient of 100 ml buffer A and
`100 ml buffer B, which is identical to buffer A except that it contains no urea.
`The column was washed with buffer C (50mM Tris—HCl,
`1 mM EDTA,
`pH = 8.2) until the baseline was stable. Native BPTI was then eluted from the
`column with a second linear gradient formed from 100 ml buffer C and 100 ml
`buffer C which contained 0.6M NaCl.
`Purified inhibitor was used for amino acid analysis (Benson and Hare, 1975)
`carried out on a Biotronic LC 5000 Analyser from Biotronic (Maintal, FRG).
`Protein sequence analysis was performed on a gas phase sequenator from Ap-
`plied Biosystems (Pfungstadt, FRG) according to Hewick et al. (1981). The
`phenylthiohydantoin derivatives were analysed on a cyano—HPLC column from
`DuPont (Wilmington, USA) as described by Beyreuther et al. (1983). Peptide
`mapping has been described (Mayes, 1985).
`Trypsin inhibition was assayed (Erlanger et al., 1961) using N-benzoyl—DL-
`arginine-p-nitsoanilide (BANA) from Merck (Dannstadt, FRG). Human leukocyte
`elastase from Elastin Products Company (Pacific, USA) and methoxysuccinyl-
`L-alanyl-L-alanyl-L-prolyl-L-valine-p-nitroanilide from Bachem (Budendorf,
`Switzerland) were used to assay leukocyte elastase inhibition (Nakajima er al. ,
`1979).
`
`Acknowledgements
`We thank E.Truscheit for encouragement, E.Schnabe1, G.Reinhardt and D.H6rlein
`for help with the protease inhibition assays and K.Otto for expert technical
`assistance. This work was supported by Bundesministerium fiir Forschung und
`Technologie through BCT 0365/2.
`
`References
`
`(1983) Proc. Natl. Acad. Sci. USA, 80,
`
`Anderson,S. and Kingston,l.B.
`6838-6842.
`Barret,A.J. (1978) Agents Actions, 8, 11-18.
`Beck,E., Sommer,R., Auerswald,E.A., Kurz,Ch., Zink,B., 0sterburg,G. and
`Schaller,H. (1978) Nucleic Acids Res. 5, 4495 -4503.
`Benson,I.R. and Hare,P.E. (1975) Proc. Natl. Acad. Sci USA 72, 619-622.
`Besse,M., von Wi1cken—Bergmarm,B. and Miiller-Hil1,B. (1986) EMBO J., in
`press.
`
`3224
`
`(1975) Acta Crystallogr., Sect. B, 31,
`
`Beyreuther,1(., Biese1er,B., Bovens,.l., Di1drop,R., Neifer,K., Stiiber,K., Zais,S.,
`Ehring,R. and Zabel,P. (1983) In Modern Methods in Protein Oiemistry. Walter
`de Gruyter & Co., Berlin—New York, pp. 303-325.
`Brent,R. and Ptashne,P. (1981) Proc. Natl. Acad. Sci. USA, 78, 4204-4208.
`Biichel,D.E., Gronenbom,B. and Miiller-Hill,B. (1980) Nature, 283, 541-545.
`Buell,G., Schu1z,M.-F., Selzer,G., Chollet,A., Movva,N.R., Semon,D.,
`Escanez,S. and Kawashima,E. (1985) Nucleic Acids Res., 13, 1923-1938.
`Biirk,R.R., Eschenbruch,M., Leuthard,P. and Steck,G. (1983) Methods Enzymol.
`91, 247-254.
`Carrell,R. (1984) Nature, 312, 14.
`Cochrane,C.G., Spragg,R.G., Revak,S.D., Cohen,A.B. and McGuire,W.W.
`(1983) Am. Rev. Respir. Dis., 127, Suppl. 2, 25-27.
`Creighton,T.E. (1978) Prog. Biophys. Mol. Biol., 33, 231-297.
`Creighton,T.E. (1985) In