Appl Microbiol Biotechnol (1992) 36:640-649

Applied Microbiology Biotechnology © Springer-Verlag 1992

High expression vectors for the production of recombinant single-chain urinary plasminogen activator from Escherichia coli Regina Brigelius-Floh~, Gerd Steffens, Wolfgang Strassburger, and Leopold Floh6* Griinenthal GmbH, Centre of Research, Zieglerstrasse 6, W-5100 Aachen, Federal Republic of Germany Received 8 August 1991/Accepted 10 October 1991

Summary. An expression cassette containing a synonymous gene for human single-chain urokinase-type plasmin0gen activator (Rscu-PA) 5'-flanked by a trp promoter and the Shine-Dalgarno sequence of the xyl A operon of Bacillus subtilis and terminated by the terminators trp A and TnlO was constructed and inserted into a pBR322 derivative to yield pBF160. When compared to pUK54 trp 207-1 containing the natural scuPA gene without the Shine-Dalgamo sequence and terminator, the expression efficiency of pBF160 in Escherichia coli strains was improved by one order of magnitude. Replacement of the trp by the tac promoter (pBF171) did not affect expression. Inserting the ShineDalgarno sequence and TnlO terminator into pUK54 trp 207-1 (pWH1320) slightly increased the expression level, whereas elimination of the Shine-Dalgarno sequence and the terminators from pBF160 with almost complete conservation of the synonymous structural gene (pBF191) significantly reduced the expression. Variation of the distance between the Shine-Dalgarno sequence and the start codon between 8 and 10 bp (pBF163) proved irrelevant. In conclusion, poor expression of mammalian genes in E. coli may result from both improperly designed regulatory elements and structural features of the coding region and therefore de-novo synthesis of the gene may be required to obtain satisfactory expression.

Introduction Plasminogen activators are of great importance for the treatment of acute occlusive blood vessel diseases. One of the plasminogen activators was isolated from urine by Williams (1951) and named urokinase. The different forms of urokinase all proved to be derived from the * Present address: Gesellschaft fiJr Biotechnologische Forschung (GBF), Mascheroder Weg 1, W-3300 Braunschweig, Federal Republic of Germany Offprint requests to: L. Floh6

same precursor molecule pro-urokinase (Floh6 et al. 1985). This zymogen, the single-chain urokinase-type plasminogen activator (Scu-PA), was first described by Bernik and Oiler (1973). It is now kown to consist of three domains comprising a total of 411 amino acids (Gianzler et al. 1982; Steffens et al. 1982; Holmes et al. 1985), and, in its naturally occurring form, to be glycosylated at Asn 302 (Steffens et al. 1982) and Thr 18 (Kentzer et al. 1990). Single-chain urokinases have a higher fibrin specificity than the two-chain urokinases, and are therefore considered superior for therapeutic use (Floh6 1985). If single-chain urinary plasminogen activator (ScuPA) is produced by recombinant techniques with Escherichia coli, an inactive insoluble material is generated, from which an unglycosylated form of Scu-PA called Rscu-PA or saruplase (INN) can be obtained. Rscu-PA proved to be clinically more efficient and better tolerated than the conventional fibrinolytic agent streptokinase (PRIMI Trial Study Group 1989). With regard to the economic impact on the health care system, high-level expression systems for the production of such recombinant proteins may be of crucial importance. The use of bacteria for the expression of mammalian genes, however, is often hampered by poor yields for various reasons: 1. The heterologous proteins may be toxic to the host. 2. The regulatory elements of the expression system may not properly meet the biological requirements of the host. 3. Non-random usage of synonymous codons has been reported for both pro- and eukaryotes (Maruyama et al. 1986; Sharp et al. 1988) and optimizing the codon usage of the heterologous genes for a specific host may lead to increased expression levels (Williams et al. 1988; Makoff et al. 1989). Since also the amount of Rscu-PA obtained with the construct of Holmes et al. (1985), i.e. pUK54 trp 207-1, and various modifications thereof (Hibino et al. 1988; Surek et al. 1991) were far from satisfactory, we tried to design high expression vectors (pBF160; pBF171) according to the following requirements:

641

1. They should contain a combination of regulatory elements shown previously to adequately function in E.

M1

o21 (7¢) Xba I

Eco RI

O21A {74)

coil 2. S y n o n y m o u s c o d o n s s h o u l d b e o p t i m i z e d for u s a g e i n E. coil 3. S e c o n d a r y structure i n p r e s u m a b l y critical r e g i o n s o f the m R N A s h o u l d b e m i n i m i z e d .

M2

Materials and methods

!i3

Nde I

01 (57}

02 (60)

OIA (51)

02A (79)

05 (sg) Pstl

General methods. All recombinant work was done according to standard procedures (Maniatis et al. 1982; Ausubel et al. 1987). The DNA fragments used for the construction of the plasmids described herein were separated in an agarose gel, removed therefrom by electroelution and purified on DE-52. Restriction enzymes and other enzymes were from Pharmacia (Uppsala, Sweden), New England Biolabs (Schwalbach, FRG) or BRL (Basel, Switzerland). All chemicals were of the highest grade commercially available.

Construction ofplasmids. Plasmid pBF160 was constructed by first removing the nic/bom region as NdeI fill-in x PvulI fragment from pBR322. Self-ligation of the remaining vector resulted in plasmid pBF157 (see Fig. 2). From pBF157 an essential part of the tetracycline resistance (Tc R) gene was removed as an NruI x EcoRV fragment resulting in plasmid pBF158. Then a multi-cloning site (MCS) was inserted into EcoRIxHindlII of pBF158. This multi-cloning site consisted of 223 bp with the restriction sites indicated in Fig. 2. The nucleotide sequence between the restriction sites was arbitrarily chosen with the exception of the sequence between XbaI and NdeI, which represents the ribosome binding site in a distance of 8 bp from the start codon ATG, which is part of the NdeI site (see Fig. 2 and for sequence Fig. 4). This sequence of the multi-cloning site is already part of the final construct and therefore not removed during the following cloning steps. The resulting plasmid is pBF158-01. The tet A/orf L terminator from TnlO was taken as the ClaI x HindlII fragment from pRT61 (Jorgensen and Reznikoff 1979) and inserted between ClaI x HindlII of pBF158-01, resulting in pBF158-01 T.

•

• Pst I

O4A (70)

07 (62) Sacl

O6A {56)

O7A (83)

0 8 (72)

0 9 (79)

O10 (75)

Sac I

Eag[

OSA (51)

lil,'i

O88 (70)

011 (61)

I~@

$1

O14A (60)

O12A (83)

01411(30) • •

O17 {73)

Spe I

• O17A (57}

~11

015 (51) •

O15A (45)

•

O10A (62)

O13 (74) • O13A (52}

• (72)

0141 {411 Kpn I

0 9 A (51)

012 (80)

Eag I

Olla

Synthesis of the structural gene encoding Scu-PA. Before starting the synthesis possible coding regions were analysed for the formation of secondary structures between promoter and structural gene sequences and between sequences of the structural gene using the programme RNAFOLD (Zuker and Stiegler 1981) implemented in GENMON (GBF, Braunschweig, FRG). Analyses were repeated with the programme MFOLD (Jaeger eral. 1990), which uses updated energy rules. The gene thus designed was synthesized as follows. Singlestranded oligonucleotides of 40-80 bp length were synthesized in a 1-1xmol scale with a DNA synthesizer (model Biosearch 8600, New Brunswick Scientific, Edison, N J, USA) by the solid-phase method described by Adams et al. (1983). Incoming nucleotides were fl-cyanoethyl-diisopropylamino-phosphoramidites(Applied Biosystems, Weiterstadt, FRG) with dimethyloxytrityl-blocked 5'OH groups (Beaucage and Caruthers 1981). After being synthesized the oligonucleotides were cleaved from the support by a 1-h treatment with concentrated ammonium hydroxide and further purified by reversed-phase HPLC. The expression cassette for Rscu-PA is built up of eight fragments (M1-M8), each consisting of two to nine single-stranded oligonucleotides (see Fig. 1). The 5' ends of the single-stranded oligonucleotides not representing the external 5' ends of the pro-urokinase gene fragments were phosphorylated by means of polynucleotide kinase. Then the respective single-stranded oligonucleotides were annealed and then ligated. The resulting double-stranded fragments were ligated into the corresponding vector, which was prepared from the respective precursor plasmid (see Fig. 2).

o6 (68)

•

04 (54)

•

O3A (40)

= 0 5 A (50)

M4

03 (75)

• 0161 (37) • 01611 {46} •

O16A(45)

Kpn I

Spe I

O16B (63)

0 1 8 {43)

Barn HI O18A {59}

O19 {76)

Barn HI"

Cla I O19A {74}

Fig. 1. Strategy for the synthesis of the single-chain urokinasetype plasminogen activator (scu-PA) gene. The oligonucleotide building blocks M1-M8, which are defined by their restriction sites, consist of the respective overlapping single-stranded oligonucleotides 01-021 and 01A-021A with the numbers of bases in parentheses. Phosphorylated oligonudeotides are indicated by a black point: Assembly of the oligonucleotides was achieved by annealing and ligation of the phosphorylated ends. Then M1 to M8 were ligated into the respective vectors; see Fig. 2 and Materials and methods for further details The synthetic gene for human pro-urokinase with the codon usage optimized for E. coli was inserted stepwise into the multicloning site. Fragment M8 was cloned between BamHI and ClaI, consisting of the T-end of the pro-urokinase gene from BamHI to the stop codon TAA, immediately followed by an NheI site. The resulting plasmid is pBF158-02T with the sequence from NheI to ClaI: GCTA GCCCGCCTAATGAGCG G G C T Y I T I T ~ A TCGA T NheI Clal representing the trp A terminator (Christie et al. 1981). Then the sites from BamHI to NdeI were filled in with the respective fragments of the synthetic pro-urokinase gene (see Fig. 2). The insertion of the synthetic trp promoter between EcoRI and XbaI was the last step, thereby creating the final construct pBF160. The full sequence of the expression cassette from EcoRI to ClaI coding for the amino acid sequence of the human prourokinase in the eDNA and of the synthetic gene together with the relevant restriction sites is shown in Fig. 4. For further details see (Brigelius-Floh6 et al. 1991). Plasmid pBF171 is constructed by insertion of the expression cassette for human pro-urokinase as an EcoRI x HindIII fragment

642

~°°~'~'°~° P

s

t

I

,

P s t I ~

~

i ~

NdeI

H~uI

Ndel, fill in, x Pvull Isolation of the larger fragment Ligation

~oal P~uII

EcoRI H i n a l I I

x

EcoRV Nrul Isolation of the larger fragment L|gation

EcoRI x Hindlll isolation of the larger fragment Ligatlon with the synthetic multi cloning site Ps~I EcoRI

Ndel

Sacl

Knnl

BamHI

'' I,~,~all~t ' hg' l~p~' Ic', 26 lg 24 21 20 26 20 22 24 21

){~aI

I

bp-

X~aZ

~;:~

i

I

"'°"'"

Sa¢I

~"2~

Ea~l ~pnI Spel

BaMHI .

Ps~I Ps%I

"~~ !n&lI

the

Clal x H~ndlll

isolation of larger fragment Ligation with the Clal x Hindlllfragment from pRT 61 (Terminator} b

successive insertion of M8 to M1 into the respective restriction sites of the cloning site '~

multi

~aI N~oI E~oRI~II~N~eI / ~ ' Sacl P ~ t l ~ E a s l /~ ..... .~ \)~.o, I

~

.~o

I~ "°' a~HI hel Clal dIII

Fig. 2. Synthesis of pBF160. By removing the nic/bom region and an essential part of the tetracycline resistance gene a high safety vector pBF158 was obtained. The strategy adopted for the synthesis of the scu-PA gene included the incorporation of a multi-cloning site into pBF158, the transcription terminator from TnlO (obtained from pRT61) and subsequent replacement of the multi-

•

cloning site by insertion of the building blocks of the gene from M8 to M1 (see Fig. 1) into the restriction sites indicated: Synthetic sequences of the scu-PA gene are shown in black. For further details see Materials and methods. Ap R, ampieillin resistance gene; puk, gene for Scu-PA; T, TnlO terminator; Tc~, tetracycline resistance gene; P, trp promoter

643 EcoRI

E~oRIX~al

IS

..~"7oo~

Insertion of a ShineDalg~rno sequence upstream of the sta~ codon ATG, flanked by Xba I and Nde I, and the Tn 10 terminator into

/'~L"~c°l

P.t

r ~r~.~ ~Ec°RI

Cla I XHind 111'

"

pHX54t~2~?_i l~coRi

v

"/._-"SaMHt //~;:V

\

~.-. ~,.

...///

~

~MH ~ 2 ~

.

U

.,-.~,,

i~;5"~

.~'"

~a~l

~aI IHd,e I 1~ HcoI

EeoRl

~-~L

Isolation of the smaller Ncol x BamHl fragment

,_ P-,t~S ~

Ligatlon

..

~ H ~ o l

P~tlS

~'

V "L~ " ~

~ \~,.. .~

~W~,~ ~?~"

~ ~ )

"

Inser,tion of the

#~ooRl

EcoRI x H~ndtllfragment into pER 322 resulting

.,.//~:~..~

~

? -o.

:~:~

/

~ ~xux~ E°°Nl

_

lP~*I

~-.

~

EeoRt I Xl~aI

I II

~-

F

~

..... ~_.~o o~I

\ __.

Isolatfon of the larger BamHI x Ncol. fragment ~btained by partial dlgest|on ~{ the Ncol llnearized plasmld with B a m H l

/

X~I

I

I

/ MpBF161

~HI

/

~ I [

I

Replacement of the trp promoter by the tac promoter

~

~III

for detailed descdptlon see fig,2 and Bri~efius-

~ ~oh~ et al,(I~9~)

I

]~ooRI

~oo.t,~'-~.'.'~.

I ,~:~ooi p

,,i..//~

//.,.

..~

\.,~ ,..

\W,ot

7 \ ~ o..

~ltn&I I I

~

t

i

~

~

I ~z, txl

~aI

Fig. 3. Summary of the plasmids constructed for this study and their relationship. All restriction sites relevant for this study are indicated. Synthetic sequences of the scu-PA gene are indicated by

the black parts in the arrows, which indicate genes and their transcription direction. Abbreviations as in Fig. 2; tac, tac promoter; MCS, multi-cloning site

obtained form pBFI60 into the EcoRI and HindlII sites from pBR322. From the resulting pBF161 the trp promoter is removed by digestion with EcoRI x XbaI and the tac promoter inserted as the E e o R I × X b a I fragment obtained from plasmid ptae-SDT (DSM 5018).

Plasmid pBF191. For the construction of pBFI91 (see Fig. 3) the NcoI x BamHI fragment from pBF160 representing an essential part of the pro-urokinase gene was eluted from an agarose gel after digestion with the respective enzymes. The vector was obtained by partial digestion of pUK54207-1 with BamHI x NcoI

644 and eluting the larger fragment containing the intact BamHI site within the TcR gene from an agarose gel. Both fragments were ligated and positive clones selected by testing for specific expression as described below. Construction of recombinant microorganisms. The hosts E. coli JM 103 and E. eoli K12 ATCC 31446 were obtained from the American Type Culture Collection (Rockville, MD, USA) and transformed with the plasmids described by means of the methods described by Maniatis et al. (1982) or Hanahan (1983). Fermentation. Cells were grown in 1 1 fermentation units (BraunDiessel, Melsungen, FRG) in a mineral salt medium containing glucose, yeast extract, thiamine and ampicillin (Ap) as described in (Brigelius-Floh6 et al. 1991). Cells with trp-promoter-containing plasmids were induced by addition of 62 mg indoleacrylic acid/1 medium and cells containing plasmid with tac promoters were induced by addition of isopropylthiogalactoside (IPTG) (0.5 mM final concentration). Enzymatic activity of Rscu-PA. Before induction of the scu-PA gene expression and 1-6 h thereafter, an amount of cells equivalent to 1 ml suspension with an optical density of 1.0 at 578 nm was collected by centrifugation. Cells were then broken with lysozyme, the cell contents solubilized with 4-5 M guanidinium hydrochloride and the proteins refolded (Winkler and Blaber 1986). After treatment with plasmin, which converts Rscu-PA to twochain urokinase (Rtcu-PA), the activity was measured with the synthetic peptide substrate $2444 (pyroglu-gly-arg-p-nitroanilide; Kabi Diagnostica, Stockholm, Sweden; for details see suppliers instructions and Brigelius-Floh6 et al. 1991). The activity thus obtained is expressed in Ploug units per millilitre of cell suspension having an optical density of 1.0 (PU/ml OD).

Results

Table 1. Codons used for highly expressed proteins by Escherichia cob Amino acid

Phe Leu lie Met Val Ser Pro Thr Ala Tyr His Gin Ash Lys Asp Glu Cys Trp Arg Gly

Total number

12 31 19 7 19 30 23 26 16 18 17 20 19 27 18 20 24 8 22 35 411 -

Optimal codon

Number of optimal codons in the human pro-urokinase gene

TTC CTG ATC ATG GTT TCT=-TCC CCG ACT----ACC GCT TAC CAC CAG AAC AAA GAC GAA TGC TGG CGT GGT

-

-

Natural

Synthetic

5 11 10 7 2 15 2 18 5 13 13 13 13 9 12 8 12 8 2 2 172 (~41.8%)

11 20 18 7 19 28 21 26 15 17 16 17 17 26 17 18 21 8 22 32 376 (~91.5%)

-

"Typical" codon according to Sharp et al. (1988) and incidence of such codons in the eDNA encoding single-chain urokinase-type plasminogen activator (Scu-PA) and the synonymous synthetic structural gene

The synthetic synonymous gene encoding Rscu-PA The e D N A sequence encoding h u m a n Scu-PA contains a large variety of codons not frequently used in E. coli. As can be seen f r o m Table 1, most o f the codons for valine, proline, lysine, arginine, and glycine of the natural scu-PA gene are not those preferred by E. coli (Sharp et al. 1988). On average only 41.8% of the e D N A codons can be considered typical codons in E. coli. Although it is b y no means certain that codons rarely used in a certain species can nevertheless be read efficiently, we decided to a d a p t the structural gene for Scu-PA as far as possible to what we believed to be a s y n o n y m o u s gene similar to typical E. coli genes. Due to the large discrepancy between the natural scu-PA gene and the p r e s u m a b l y better-suited one for E. coli (see Table 1), the intended adaption could not possibly be done step by step by means o f site-directed mutagenesis. Therefore a structural gene was designed considering typical c o d o n usage in E. coli but also guaranteeing technical feasibility by inserting singular restriction sites. We further tried to exclude strong secondary structure f o r m a t i o n between regulatory elements and the beginning o f the coding region and within the coding region (see Fig, 4). A possible secondary structure formation with a A G of - 1 1 . 5 k c a l / m o l was calculated for positions 55 through 157, i,e. start of the m R N A and the consecutive 90 bp, in p U K 5 4 trp 207-1 by means o f

the R N A F O L D p r o g r a m m e (Zuker and Stiegler 1981). According to the same p r o g r a m m e this relatively strong secondary structure was reduced to a A G o f 3.8 k c a l ! mol in pBF160 in order to improve translation efficiency when the Shine-Dalgarno sequence was inserted. A recalculation according to the m o r e recent M F O L D p r o g r a m m e described by Jaeger et al. (1990), however, yielded essentially identical A G values for this m R N A region (9.1-9.5 k c a l / m o l ) in p U K 5 4 trp 207-1, pWH1320, pBF160 and pBF191. D e p e n d i n g on the m o d e o f calculation, the starting region o f the synthetic s y n o n y m o u s gene could thus be considered similar to, or different from the natural one in terms o f m R N A secondary structure. The rscu-PA gene thus designed, which contains 91.5% the codons frequently used in E. coli (Table 1 and Fig. 4), was synthesized in toto as described in Materials and methods, and used for plasmid constructions.

Description o f the plasmids used P l a s m i d p U K 5 4 trp 207-1. This contains the gene for hum a n pro-urokinase, obtained by screening a e D N A library f r o m m R N A of h u m a n pharyngeal carcinoma cells, u n d e r the control of the trp p r o m o t e r in a pBR322 derivative. The construction was described in detail by

645 1 1 35 35

T ATT CTG AAA TGA GCT GTT GAC AAT TAA TCA TCG ~ ATT ~TG AAA TGA GCT GTT GAC AAT TAA TCA TCG AAC TAG TTA ACT AAC TAG TTA ACT

EcoRI

AGT ACG CAA GTT CAC GTA AAA AGG AGT ACG CAA GTT CAC GTA AAA AGG

1 n - P U K 71 s-pIv~ 71

Met Set Asn GIu GTA TC~ AGA ........... r ATT ATG AGC AAT GAA G T A T C T ~ G A T ~ A G G A G G A A A T C A T ~T__G A G C A A T G A A

5 107 107

L e u H i s G l n V a l P r o S e r A s n Cys A s p C y s L e u A s h CTT CAT CAA GTT CCA TCG AAC TGT GAC TGT CTA AAT CTT CAT CAA GTT CCA TCG AAC TGT GAC TGT CTA A~T

17 143 143

Gly Gly Thr Cys Val Ser ASh Lys Tyr Phe Set ASh GGA GGA ACA TGT GTG TCC AAC AAG TAC TTC TCC AAC GGC GGA ACC TGC GTT TCT AAC AAA TAT TTC ~CT AAC

29 179 179

Ile His Trp Cys ASh Cys Pro Lys Lys Phe Gly Gly ATT CAC TGG TGC AAC TGC CCA AAG AAA TTC HGA GGG ATC CAC TGG TGT AAC TGC CCG AAA AAA TTC GGT GGT

41 215 215

G l n H i s Cys G l u I l e A s p L y s S e r Lys T h r C y s T y r CAG CAC TGT GAA ATA HAT AAG TCA AAA ACC TGC TAT CAG CAC TGC GAA ATC GAC AAA TCT AAA ACC TGC TAC

53 251 251

GIu Gly ASh Gly His Phe Tyr Arg Gly Lys Ala Set GAG GGG AAT GGT CAC TTT TAC CGA GGAAAG GCC AGC GAA GGT AAC GGT CAC TTC TAC CGT GGT AAG GCT TCT

65 287 287

Thr Asp Thr Met Gly Arg Pro Cys Leu Pro Trp Ash ACT GAC A~C ATG ~GC CGG CCC TGC CTG CCC TGG AAC ACC GAC ACC ATG GGT CGT CCG TGC CTG CCG TGG AAC

Ncol Ncol

77 323 323

Set Ala Thr Vsl Leu Gin Gin Thr Tyr His Ala His TCT GCC ACT GTC CTT CAG CAA ACG TAC CAT GCC CAC TCT GCT ACC GTT ~G CAG CAG ACC TAC CAC GCT CAC

Pstl

89 359 359

Arg Set Asp Ala Leu Gin Leu Gly Leu Gly Lys His AGA TCT HAT GCT CTT CAG CTG GGC CTG GGG AAA CAT CGT TCT GAT GCA TTG CAG CTG GGT CTG GGT AAACAC

i01 395 395

Asn Tyr Cys Arg ASh Pro Asp ASh Arg Arg Arg Pro AAT TAC TGC AGG AAC CCA GAC AAC CGG AGG CGA CCC AAC TAC TGC CGT AAC CCG GAC AAC CGT CGT CGT CCG

113 431 431

T r p Cys T y r V a l G l n V a l G l y L e u Lys P r o L e u V a l TGG TGC TAT GTG CAG GTG GGC CTA AAG CCG CTT GTC TGG TGC TAC GTT CAG GTT GGT CTG AAA CCG CTA GTT

125 467 467

Gln GIu Cys Met Val His Asp Cys Ala Asp Gly Lys CAA GAG TGC ATG GTG CAT GAC TGC GCA GAT GGA AAA C~G GAA TGC ATG GTT CAC GAC TGC GCT GAC GGT AAA

137 503 503

Lys Pro Set Set Pro Pro GIU GIu Leu Lys Phe Gin AAG CCC TCC TCT CCT CCA GAA GAA TTA AAA TTT CAG AAA CCG TCT TCT CCG CCG GAA GAG CTC AAA TTC CAG

149 539 539

Cys Gly Gln Lys Thr Leu Arg Pro Pro Phe Lys lie T G T G G C C A A A A G A C T C T G A G G C C C C C C T T T 2kAG A T T TGC GGT CAA AA~ ACC CTA CGT~CCG CGT TTT AAA ATC

161 575 575

Ile Gly Gly Glu Phe Thr Thr lie Glu Ash Gin Pro ATT GGG GGA GAA TTC ACC ACC ATC GAG AAC CAG CCC ATC GGT GGT GAG TTC ACC ACC ATC GAA AAC CAG CCG

173 611 611

Trp TGG TGG

185 647 647

Set Val Thr Tyr Val Cys Gly Gly Ser Leu ~le Set TCT GTC ACC TAC GTG TGT GGA GGC AGC CTC ATC AGC TCT GTT ACC TAC GTT TGC GGT GGT TCT CTG ATC TCT

197 683 683

P r o Cys T r p V a l I l e S e t A l a T h r H i s Cys P h e l l e CCT TGC TGG GTG ATC AGC GCC ACA CAC TGC TTC ATT CCG TGC TGG GTT ATC TCT GCT ACC CAC TGC TTC ATC

2O9 719 719

A s p T y r P r o L y s L y s G l u A s p T y r Ile V a l T y r L e u HAT TAC CCA AAG AAG GAG GAC TAC ATC GTC TAC CTG GAC TAC CCG AAA AAA GAA GAC TAC ATC GTT TAC CTC

221 755 755

Gly Arg Ser Arg Leu ASh Set Asn Thr Gln Gly GIu GGT CGC TCA AGG CTT AAC TCC AAC ACG CAA GGG GA~ GGC CGT TCT CGT TTA AAC TCT AAC ACC CAG GGT GAA

233 791 791

Met Lys Phe GIu Val GIu ASh Leu Ile Leu His Lys ATG AAG TTT GAG GTG GAA AAC CTC ATC CTA CAC AAG ATG AAA TTC GAA GTT GAA AAC CTG ATC CTG CAC AAA

245 827 827

Asp Tyr Ser Ala Asp Thr Leu Ala His His ASh Asp GAC TAC AGC GCT GAC ACG CTT GCT CAC CAC AAC GAC GAC TAC TCT GCT GAC ACC CTG GCT CAC CAC A~C GAC

257 863 863

I l e A l a L e u L e u Lys I l e A r g S e t L y s G I u H l y A r g ATT GCC TTG CTG AAG ATC CGT TCC AAG GAG GGC AGG A T C G C T C T G C T A A A A A T C C G T T C T Alia G A A G G T C G T

269 899 899

Cys Ala ~in Pro Set Arg Thr Ile Gln Thr Met Cys TGT GCG CAG CCA TCC CGG ACT ATA CAG ACC ATG TGC TGC GCT CAG CCG TCT CGT ACC ATC CAG ACC ATC TGC

281 935 935

Leu CTG CTG

293 971 971

Ser Cys Glu Ile Thr Gly Phe Gly Lys GIu ASh Ser AGC TGT GAG ATC ACT GGC TTT GGA AAA GA G AAT TCT TCT TGC GAA ATC ACC GGT TTC GGT ~ GAA AAC TCT

Xbal XbaI,

NdeI

305 1007 1007

T h r A s p T y r L e u T y r P r o G l u G l n L e u Lys M e t T h r ACC GAC TAT CTC TAT CCG GAG CAG CTG AAA ATG ACT ACC GAC TAC CTG TAC CCG GAA CAG CTG AAA ATG ACC

317 1843 1043

Val Val Lys Leu Ile Set His Arg Glu Cys Gln Gln GTT GTG AAG CTG ATT TCC CAC CGG GAG TGT CAG CAG GTT GTT AAA CTG ATC TCT CAC CGT GAA TGC CAG CAG

329 1079 1079

Pro His Tyr Tyr CCC CAC TAC TAC CCG CAC TAC TAC

341 ii15 Ii15

Leu Cys Ala Ala Asp Pro Gln Trp Lys Thr Asp set CTA TGT GCT GCT GAC CCC CAA TGG AAA ACA GAT TCC CTG TGC GCT GCT GAC CCG CAG TGG AAA ACC GAC TCT

353 llS1 1151

Cys G l n G l y A s p S e t G l y G l y P r o L e u V a l C y s S e t TGC CAG GGA GAC TCA GGG GGA CCC CTC GTC TGT TCC TGC CAA GGT GAC TCT GGT GGT CCA CTA $~T TGC TCT

365 i187 1187

L e u G l n G l y A r g M e t T h r L e u T h r G l y Ile V a l S e t CTC CAA GGC CGC ATG ACT TTG ACT GGA ATT GTG AGC CTC CAG GGT CGT ATG ACC CTG ACC GGT ATT GTT TCT

377 1223 1223

Trp Gly Arg Gly Cys Ala Leu Lys Asp Lys Pro Gly TGG GGC CGT GGA TGT GCC CTG AAG GAC AAG CCA GGC CCG GGT TGG GGT CGT GGT TGC GCT CTG AAA GACAAA

389 1259 1259

V a l T y r T h r A r g V a l S e r H i s P h e L e u P r o T r p Ile GTC TAC ACG AGA GTC TCA CAC TTC TTA CCC TGG ATC HamHI GTT TAC ACC CGT GTT TCT CAC TTC CTG CCG TGG ATC Ba~qI

401 1295 1295

Arg Set His Thr Lys GIu GIu ASh Gly Leu Ala Leu ~GC AGT CAC ACC AAG GAA GAG AAT GGC CTG GCC CTC ~GT TCT CAC ACC AAA GAA GAA AAC GGT CTG GCT CTG

1331 1331

TGA TAA GCT AGC

1367

CG~ T

Gly Set Glu Val Thr Thr Lys Met GGC TCT GAA GTC ACC ACC AAA ATG GGT TCT GAA GTT ACC ACC AAA ATG

Spel

C C G C C T A A T G A G C G G G C T T T T T T T TA__T ClaI

SacI

Phe Ala Ala Ile Tyr Arg Arg His Arg Gly Gly TTT GCG GCC ATC TAC AGG AGG CAC CGG GGH GGC TTC GCT GCT ATC TAC CGT CGT CAC CGT GGT GGT

P r o S e r M e t T y r A S h A s p P r o G l n Phe G l y T h r CCC TCG ATG TAT AAC GAT CCC CAG TTT GGC ACA CCG TCT ATG TAC AAC GAC CCG CAG TTC ~GT ACC

Eagl

KpnI

H o l m e s et al. (1985). The plasmid with the restriction sites relevant for this study is s h o w n in Fig. 3.

Plasmid p WH1320. In this plasmid is inserted the ribosome binding site from the Xyl A gene from B. subtilis at a distance o f 8 bp from the start codon, flanked by two restriction sites (XbaI and NdeI). The tetA/orfL terminator from TnlO (Schollmeier et al. 1985) is placed behind the scu-PA gene into the restriction sites ClaI and HindlII, thereby inactivating the Tc rt gene pro-

Fig. 4. Sequence of the human cDNA (n-puk), the synthetic gene (s-puk), the amino acid sequence derived therefrom, and the trp promoter (bases 1-74). The Shine-Dalgarno sequence missing in pUK54 trp 207-1 is boxed in the synthetic gene. The restriction sites relevant for the construction of the synthetic gene are shown by underlining the recognition sequence with the enzyme indicated on the right side

moter (Kolot et al. 1989). The resulting plasmid pWH1320 does not, therefore, transfer tetracycline resistance to bacterial strains. The detailed construction is described by Surek et al. (1991).

Plasmid pBF160. This is derived from pBR322 as described in Materials and methods. It contains the synthetic structural gene encoding Scu-PA under the control o f the trp promoter. The Shine-Dalgarno sequence is derived from the xyl A gene from B. subtilis as in

646 Table 2. Expression plasmids referred to, brief characteristics and human scu-PA (rscu-PA) expression levels given as Ploug units per millilitre of cell suspension with an optical density of 1.0 in E. coli JM 103, which were obtained 6h after induction (2h with pMUT4L)

Plasmids

Characteristics

pUK54 trp 207-1 pMUT4L

Ap R, TcR, n-puk Modified sequence of the first 24 bp of the natural puk gene resulting in a weaker secondary structure tac, rrnB T1, T2 terminator Ap ~, trp-xyl A RBS, n-puk, Tnl0-terminator Ap rt, trp s-puk expr. in pBR322 TcR-deletion, nit/bornAp R, trp s-puk expr. in pBR322 Ap R, trp s-puk expr. in pBR322, Tc~t-deletion Ap ~, trp s-puk expr. in pBR322, D-10 Ap R, tac s-puk expr. in pBR322 Ap R, tac s-puk expr. in pBR322, TcR-deletion Ap rt, TcR, n-puk = s-puk hybrid

pWH1320 pBF160 pBF161 pBF162 pBF163 pBF171 pBF172 pBF191

Yield 150- 200 140a

200- 250

1200-1800 1300-1800b 1300b 1300b 1400 1100b 700

RBS: ribosome binding site (Shine-Dalgarno sequence); n-puk, scu-PA gene isolated from a human eDNA bank; s-puk, synthetic scu-PA gene; s-puk expr., expression system for scu-PA consisting of the synthetic puk gene flanked by trp or tac promoter, the Shine-Dalgarno sequence from the xylA operon of Bacillus subtilis, the trp A terminator and the Tn 10 terminator; D-10, distance of the RBS prolonged from 8 to 10 bp by filling in the NdeI site; Ap R, ampicillin resistance; TcR, tetracycline resistance a Calculated from Hibino et al. (1988) b Reported by Brigelius-Floh6 et al. (1991)

pWH1320 and is localized at a reasonable distance of 8 bp from the starting codon. Transcription termination is guaranteed by two terminators (trpA plus Tnl0). An essential part o f the Tc R gene is removed. Mobilisation o f the plasmid is minimized by elimination of the n i c / born region o f the pBR322 vector. According to Winnacker (1985) and Covarrubias et al. (1981) this plasmid can therefore be rated as a high-safety vector and is considered particularly suited for large-scale production.

Only the first 200 5'-terminal and the last 37 3'-terminal base pairs of the natural scu-PA gene are retained and those contain only a few codons rarely used in E. coli. The relationship of the plasmids described here is summarized in Fig. 3. Further plasmids referred to in this paper are listed in Table 2 (see Discussion).

Expression experiments With all constructions investigated in E. coli, the expression o f the scu-PA gene results in insoluble inactive material which, however, can be refolded (Winkler and Blaber 1986) to yield native Rscu-PA. By means o f the refolding procedure and analytical tests used here (Brigelius-Floh6 et al. 1991) the plasminogen activator activities measured can be taken as a reliable, measure of the amount o f material expressed at given times, if a standard variation o f 10% is taken into account. Interassay variance is slightly higher. For this reason, all data given in one figure are derived from experiments run under strictly identical conditions at the same time thus allowing best comparability. Time-dependent yield of Rscu-PA from E. coli K12 JM 103 strains transformed with plasmids having the scu-PA gene under the control o f the trp promoter is shown in Fig. 5. Generally a low level of expression is seen before induction o f expression by indoleacrylic acid (time 0). After induction, the yield of Rscu-PA rises sharply and usually tends to cease 6 h later. It was

lOOO

O

500

.m

< I

lOO

Plasmid pBF171. This differs from pBF160 essentially by having the scu-PA gene controlled by the tac promoter (instead o f the trp promoter). Plasmid pBF171, in contrast to pBF160, still contains the nic/bom region, which, however, is functionally irrelevant in respect to expression level (Brigelius-Floh6 et al. 1991). Plasmid pBF191. This was designed particularly in order to investigate the importance of the codon usage in the constructs presented here. In essence it is the plasmid p U K 5 4 trp 207-1, with most of its natural scu-PA gene replaced by the optimized synthetic scu-PA gene.

o 0

1

2

3

4

5

6

hours after induction

Fig. 5. Yields of Rscu-PA obtained with pUK54 trp 207-1 ( I ) , pWH1320 ([]), pBF191 (O), and pBF160 (0) in Escherichia coli K12 JM 103. Cells were grown as described in Materials and methods and induced by addition of indole acrylic acid. Samples were taken before induction (time 0) and 1-6 h thereafter, cells broken and Rscu-PA refolded and activated as described in Materials and methods. The expression is measured as two-chain-PA (Rtcu-PA) activity in Ploug units (PU) per millilitre of cell suspension having an optical density of= 1.0

647 pBF160 and pBF171. Any relevant influence of the host strain was not detectable. 1500

Discussion JM 103

~ II

~ , ~

ATCC 31446

1000

.~

°

,~ I~.

~O

~

~

~

o.

~

~

~

;

~ ~ ~

~ ~

b ~ i

~ ~ ~ ~

hours after induction

F~g. 6. Yields of Rscu-PA obtained with pUK54 t~ 207-1 (~), pBF160 (O) and pBF171 (~) in E. coli K12 JM 103 and E. coli K12 ATCC 31~6 given as two-chain PA activity as described in the legend to Fig. 5. Cells containing pBF160 (t~ promote0 were induced with indoleac~lic and, cells containing pBF171 (tac promote0 were induced with isopropylthiogalactoside. For N~her detNls see Materials and methods

confirmed that E. coli transformed with pUK54 trp 2071 is an extremely poor producer strain. With plasmid pWH1320, which contains the very same structural gene, but is modified by addition of a Shine-Dalgarno sequence and a transcription terminator downstream of the structural gene, the expression is consistently better but still low. With pBF160 containing the synthetic gene, the yield is about ten times the activity obtained from pUK54 trp 207-1. The plasmid pBF191, containing a structural gene largely composed of the optimized synthetic gene and some residues from the natural cDNA (200 bp at the 5'-end and 37 bp at the 3'-end) provides an intermediate expression level. In general, the choice of the host (E. coli K12) did not substantially affect the productivity (data not shown). Figure 6 shows a comparison of expression levels obtained with the plasmids pBF160 and pBF171 having the synthetic gene under the control of the trp or the tac promoter, respectively, each in two different E. coli strains. To support also ATCC 31446 with the lac I gene, which enables a regulated control of the tac-derived transcription form pBF171 and which is already present in JM 103, the plasmid pACYC lac i q was also transferred to ATCC 31446, when the expression should be induced from pBF171. Apart from some variability around the sixth hour after induction no difference in efficiency was seen between the plasmids

The attempts to construct vectors suitable for the production of urokinase-type plasminogen activators started in the early 1980s and have now reached an optimum with plasmids such as pBF160 and pBF171. A yield of dose to 20% of total E. coli protein, as calculated from scanned electropherograms of the total E. coli protein (Brigelius-Floh6 et al. 1991), may indeed be considered satisfactory and surprisingly high in the case of a protein of 46.344 kDa containing 12 disulphide bridges. The practical goal being met, the question remains as to which of the various gene modifications finally lead to reasonable expression. This theoretical problem certainly can not be solved by comparing the few plasmids described here but might become slightly more transparent, if previous attempts are compared with the present results. For this purpose we have compiled some pertinent data in Table 2. In none of the experiments presented here did the choice of the host have any substantial influence on productivity. This finding seemingly contrasts with the conclusion of Surek et al. (1991), who described considerable host dependency of Rscu-PA yield. We can in fact confirm the importance of host choice stressed by Surek et al. (1991). The apparent discrepancy is due to the fact that in the present study only closely related hosts (all E. coli K12) were used, which were previously selected for product stability, because we intended to investigate plasmid function with minimum disturbance from interfering proteolytic events. In agreement with Surek et al. (1991) we do, however, believe that pronounced productivity differences in recombinant E. coli are due rather to different proteolytic activity than to substantial differences in host-plasmid interaction. Undoubtedly, a bacterial-type ribosome binding site in a reasonable distance of 6-12 bp to the start codon should improve translation, and Rscu-PA yield improvements are also seen upon proper design of the ShineDalgarno sequence. Plasmid pWH1320 differs from the poorest expression vector, pUK54 trp 207-1, primarily in this respect, and pBF191, having the 5'- and 3' terminal regulatory elements of the pUK54 trp 207-1 flanking a partially optimized structural gene, is substantially less efficient than pBF160, which is more related to pWH1320 in its regulatory elements. Unfortunately, the examples quoted are not decisive with respect to the importance of the Shine-Dalgarno sequence, since the superiority of pWH1320 over pUK54 trp 207-1 could also result from the TnlO terminator in the former plasmid, and the gap in productivity between pBF191 and pBF160 might result from a variety of structural differences between these plasmids. Also, the messenger RNA derived from pUK54 trp 207-1 and pWH1320 could also differ in the ability to form secondary structures between the 5' regulatory unit and the first base

648 pairs of the structural gene (see below). However, what we can learn from the results obtained with pBF191 is that lack o f a state-of-the-art ribosome binding site is well compatible with a reasonable, although not satisfactory expression in E. coli. Further, the distance between the Shine-Dalgarno sequence and the start codon appears to tolerate some variability. Surek et al. (1991) reported only marginal variations in efficiency when this distance was varied between 6 pb and 10 bp and we could not detect any difference in expression when the distance of 8 bp in pBF160 was enlarged to 10bp in pBF163 (BrigeliusFloh6 et al. 1991). To what extent the terminators downstream of the structural genes determine the expression rate cannot be answered from the present investigation. Lack of any terminator downstream of the structural gene inevitably leads to futile transcription and is therefore generally considered to impair transcription efficiency. Therefore, at least one terminator was used in all but one of our constructions. Plasmid pBF191, lacking any terminator directly following the structural gene but being reasonably efficient, again shows us that such a terminator cannot be the most crucial factor for determining expression. Based on the comparison of the trp and tac promoter constructs, also the choice of the promoter cannot be made responsible for the dramatic differences in the expression o f the scu-PA gene described. In general, no difference between trp and tac promoter systems were observed when the remaining gene structure was identical or similar (e.g. pBF160-pBF171 or pBF162pBF172, see Table 2). According to the R N A F O L D programme (Zuker and Stiegler 1981) a weak secondary structure formation ( A G = 3.8 kcal/mol) in the remaining construct was anticipated. Recalculation of the secondary structure according to the new algorithm o f Jaeger et al. (1990), however, revealed the possibility of an other m R N A secondary structure formation with a AG of - 9 . 1 k c a l / m o l in pUK54 207-1 and pBF191 and o f - 9 . 5 k c a l / m o l in pWH1320, pBF160, pBF161 and pBF162. Since therefore the free energy values of possible m R N A secondary structures in the Y-terminal gene region may be practically identical for the plasmids investigated, they are not likely to be responsible for the drastically different expression rates. Furthermore, the possible secondary structures are precisely identical in p U K 5 4 trp 207-1 and pBF191 but their efficiency differs markedly nevertheless. Left with no alternative we have to conclude that the structure of the coding region provides the most relevant increment to the height of expression. Whether this can be attributed to codon usage alone or also to secondary structures cannot be worked out without unreasonable effort and thus we are limited to guessing. One o f the crucial steps o f vector improvement might have been the systematic elimination o f the codon A G G frequently encoding Arg in the human c D N A and practically never used in E. coli (Sharp et al. 1988; Bonekamp and Jensen 1988). This assumption would still be compatible with the reasonable efficiency of

pBF191, which still contains some unoptimized codons (including two Arg codons) in its 5'- and 3'-ends o f the coding region, but not a single A G G codon. Alternatively, the slightly lower expression observed with pBF191 with respect to pBF160 may result from the lacking Shine-Dalgarno sequence, a less favourable m R N A secondary structure or a lacking terminator sequence. In conclusion, major uncertainties regarding the relative importance of regulatory elements, secondary structure p h e n o m e n a and codon usage still have to be resolved before foreign gene expression in E. coli may be reliably predicted. Furthermore, it appears unlikely that generally applicable rules can be developed, since more than one effect may determine the expression rate in individual cases. In the case of scu-PA gene expression retrospective analysis favours the assumption that p o o r codon usage of the constructions based on the human c D N A and lacking or unfavourable Shine-Dalgarno sequences were the major obstacles for satisfactory expression experiments in the past. Acknowledoements. The skilful technical assistance of Mrs. Alice

Dressen, Mrs. Ulrike Leinz-Hanke and Mr. Heinz-G0nter DOteberg is greatfully acknowledged. This study was supported by the Bundesministerium ftir Forschung und Technologic (BMFT) of the Federal Republic of Germany. The plasmids pWH1320 and pACYC lac, iq, which is a lac iq-containing pACYC 184, were kind gifts from W. Hillen, Erlangen, FRG.

References

Adams SP, Kavka KS, Wykes El, Holder SB, Gallupi GR (1983) " Hindered dialkylamino nudeoside phosphite reagents in the synthesis of 2 DNA 51-mers. J Am Chem Soc 105:661-663 Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JJ, Smith JA, Struhl K (1987) Current protocols in molecular biology, vols 1 and 2. Massachusetts General Hospital, Harvard Medical School Beaucage SL, Caruthers MH (1981) Deoxynucleoside phosphoamidites - a new class of key intermediates for deoxypolynudeotide synthesis. Tetrahedron Lett 22:1859-1862 Bernik MB, Oiler EP (1973) Increased plasminogen activator (urokinase) in tissue culture after fibrin deposition. J Clin Invest 52: 823-834 Bonekamp F, Jensen KF (1988) The AGG codon is translated slowly in E. eoli even at very low expression levels. Nucleic Acids Res 16:3013-3024 Brigelius-Floh6 R, Floh6 L, Hiilen W, Steffens G J, Strassburger W, Wilhelm M (1991) Piasmide, deren Herstellung und deren Verwendung bei tier Gewinnung eines Plasminogenaktivators. Offenlegungsschrift DE 4020438 A1 Christie GE, Farnham PJ, Platt T (1981) Synthetic sites for transcription termination and a functional comparison with tryptophan operon termination sites in vitro. Proc Natl Acad Sci USA 78:4180-4184 Covarrubias L, Cervantes L, Covarrubias A, SobOron X, Vichido I, Blanco A, Kupersztoch-Portnoy YM, Bolivar F (1981) Construction and characterization of new cloning vehicles. V. Mobilization and coding properties of pBR322 and several deletion derivatives including pBR327 and pBR328. Gene 13:2535 Floh+ L (1985) Single-chain urokinase-type plasminogen activators: new hopes for dot-specific lysis. Eur Heart J 6:905-908 Floh6 L, Steffens GJ, Giinzler WA, Otting F, Heynecker H, Holmes W, Rey M, Shephard HM, Seeburg P, Hayflick H, Ve-

649 har G (1985) Insight into biosynthesis of human urokinase forms. In: Davidson JF, Donati MB, Coccheri S (eds) Progress in Fibrinolysis. Churchill Livingstone, Edinburgh London Melbourne and New York, pp 213-216 Gtinzler WA, Steffens GJ, Otting F, Kim S-M, Frankus E, Floh6 L (1982) The primary structure of high molecular mass urokinase from human urine. The complete amino acids sequence of the A chain. Hoppe Seyler's Z Physiol Chem 363:11551165 Hanahan D (1983) Studies on transformation of Escherichia coli with plasmids. J Mol Biol 166:557-580 Hibino Y, Miyake T, Kobayashi Y, Ohmori M, Miki T, Matsumoto R, Numao N, Kondo K (1988) Enhanced expression of human pro-urokinase eDNA in Escherichia coli. Agric Biol Chem 52:329-336 Holmes WE, Pennica D~ Blaber M, Rey MW, G~nzler WA, Steffens GJ, Heynecker HL (1985) Cloning and expression of the gene for pro-urokinase in Escherichia coli. Biotechnology 3:923-929 Jaeger JA, Turner DH, Zuker M (1990) Predicting optimal and suboptimal secondary structure for RNA. Methods Enzymol 183:281-306 Jorgensen RA, Reznikoff WS (1979) Organization of structural and regulatory genes that mediate tetracycline resistance in transposon Tnl0. J Bacteriol 138:705-714 Kentzer EJ, Buko A, Menon G, Sarin VK (1990) Carbohydrate composition and presence of a fucose-protein linkage in recombinant human pro-urokinase. Biochem Biophys Res Commun 171:401-406 Koiot MN, Kashlev MV, Gragerov AI, Khmel JA (1989) Stability of the pBR322 plasmid as affected by the promoter region of the tetracycline-resistance gene. Gene 75:335-339 Makoff AJ, Oxer MO, Romanos MA, Fairweather NF, Ballantine S (1989) Expression of tetanus toxin fragment C in E. coli: high level expression by removing rare codons. Nucleic Acids Res 17:10191-10202 Maniatis T, Fritsch EF, Sambrook J (1982) Molecular Cloning. A laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, N. Y.

Maruyama T, Gojobori T, Aota SJ, Ikemura T (1986) Codon usage tabulated from the GenBank sequence data. Nucleic Acids Res 14:r151-r197 PRIMI Trial Study Group (1989) Randomized double-blind trial of recombinant pro-urokinase against streptokinase in acute myocardial infarction. Lancet, April 22, 863-868 Schollmeier K, Gartner D, Hillen W (1985) A bidirectionally active signal for termination of transcription is located between tet A and orf L on transposon TnlO. Nucleic Acids Res 13:4227-4237 Sharp PM, Cowe E, Higgins DG, Shields DC, Wolfe KH, Wright F (1988) Codon usage patterns in Escherichia coli, Bacillus subtilis, Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila melanogaster and Homo sapiens; a review of the considerable within-species diversity. Nucleic Acids Res 16:8207-8211 Steffens GJ, Giinzler WA, Otting F, Frankus E, Floh6 L (1982) The complete amino acid sequence of low molecular mass urokinase from human urine. Hoppe Seyler's Z Physiol Chem 363 : 1043-1058 Surek B, Wilhelm M, Hillen W (1991) Optimizing the promoter and ribosome binding sequence for expression of human single chain urokinase-like plasminogen activator in Escherichia coli and stabilization of the product by avoiding heat shock response. Appl Microbiol Biotechnol 34:488-494 Williams DP, Regier D, Akiyoshi D, Genbauffe F, Murphy JR (1988) Design, synthesis and expression of a human intedeukin-2 gene incorporating the codon usage bias found in highly expressed Escherichia coli genes. Nucleic Acids Res 16" 1045310467 Williams JRB (1951) The fibrinolytic activity of urine. Br J Exp Pathol 32:530 Winkler ME, Blaber M (1986) Purification and characterization of recombinant single-chain urokinase produced in Escherichia coli. Biochemistry 25:4041-4045 Winnacker EL (1985) Gene und Klone. VCH, Weinheim, p 298 Zuker M, Stiegler P (1981) Optimal computer folding of large RNA sequences using thermodynamics and auxiiliary information. Nucleic Acids Res 9:133-148