FEMS Microbiology Revie~,s 88 (1992) 73-92 © 1992 Federation of Eurot~ean Microbiological Societies 0168-6445/92/$05.00 Published by Elsevier

73

FEMSRE 00218

Gene expression in Lactococcus lactis

M a a r t e n van de G u c h t e m, Jan Kok and G e r a r d V e n e m a

Department of Genetics, Centre of Biological Sciences, Unh'ersity of Groningen. Haren, The Netherlands

Received 24 April 1991 Revision received 28 June 1991

Accepted 2 July 1991

Key words: Lactococcus lactis; Gene expression signal; Heterologous gene expression; Translational coupling; Protein secretion

1. SUMMARY

Lactic acid bacteria are of major economic importance, as they occupy a key position in the manufacture of fermented foods. A considerable body of research is currently being devoted to the development of lactic acid bacterial Strains with improved characteristics, that may be used to make fermentations pass of more efficiently, or to make new applications possible. Therefore, and because the lactococci are designated ' G R A S ' organisms ('generally recognized as safe') which may be used for safe production of foreign pro- teins, detailed knowledge of homologous and het- erologous gene expression in these organisms is desired. An overview is given of our current knowledge concerning gene expression in Lacto- coccus loctis. A general picture of gene expres- sion signals in L. lactis emerges that shows con- siderable similarity to those observed in Es-

Correspondence to: G. Venema, Department of Genetics, Centre of Biological Sciences, University of Groningen, Kerk- laan 30, 9751 NN Haren, The Netherlands. I Present address: National Food Biotechnology Centre, Uni-

versity College, Cork, Ireland.

cherichia coli and Bacillus subtilis. This feature allowed the expression of a number of L. lactis- derived genes in the latter bacterial species. Sev- eral studies have indicated, however, that in spite of the similarities, the expression signals from E. coli, B. subtilis and L. lactis are not equally efficient in these three organisms.

2. INTRODUCTION

Lactic acid bacteria, including members of the genera LactobaciUus, Lactococcus, Leuconostoc, Pediococcus, and Streptococcus, have long been known for their use in the manufacture of fer- mented foods. These include a broad range of products derived from a variety of raw materials such as vegetables, cereals, meat and milk [1]. The fermentations not only serve the preserva- tion of the food [2], they also add to the develop- ment of flavour and texture of the products [3]. Moreover, some cultured milk products are dis- cussed to have certain health and additional nu- tritional benefits [4].

These features explain the major economic importance of the lactic acid bacteria for the food

74

industry, and the dairy industry in particular. Therefore, considerable efforts are made to un- derstand and improve the important characteris- tics of lactocoecal starter strains. Increasing our knowledge of genetics and physiology of these organisms will almost certainly lead to the devel- opment of starter strains with improved proper- ties, whereas the acquisition of more specific knowledge of the ru~es governing (heterologous) gene expression may be used to add desirable new traits. In addition, since the lactococci are designated 'GRAS' organisms ('generally recog- nized as safe') [5], they may be used in new applications for the production of commercially important heterologous proteins in a safe way. Depending on the nature and application of newly synthesized proteins, several expression strategies are conceivable. One option is to (over)produce the new protein intracellularly, after which the cells can be easily pelleted in order to harvest the protein. The second option is to make the lactic acid bacteria secrete the new protein into the culture medium. For special applications, e.g, in cheese making, a combined strategy may be pre- ferred, in which the protein is produced intracel- lularly, and subsequently released by a delivery system based on autolysis or induced lysis of the host [11_

3. GENE EXPRESSION IN PROKARYOTES

As in eukaryotes, gene expression in prokary- otes is governed by the processes of transcription and translation. In the cascade of events that results in the production of a protein, a promoter sequence in the DNA mediates the interaction between the RNA-polymerase and the DNA to initiate transcription. In Escherichia coli two hexameric sequences in the promoter area are well-conserved: the so-called - 3 5 region (TI'GACA), and the - 10 region (TATAAT) [6]. Usually, these hexamers are separated by a se- quence of 17 4-1 nucleotides, while the starting point of transcription is usually situated 7 + 1 nueleotides downstream of the - 1 0 hexamer. The same features are conserved in promoters active in vegetative cells of the Gram-positive bacterium Bacillus subtilis [7].

In prokaryotes, transcription is immediately followed by the translation of the nascent RNA messenger [8,9]. Translation is initiated at the ribosome binding site, which comprises the fol- lowing conserved elements: the Shine Dalgarno (SD) sequence that is complementary to the 3'- end of the 16S-rRNA (E. coli 16S-rRNA has the 3'-end sequence: 3' AUUCCUCCA-5' [10]), and the translational start codon. These two elements are usually separated by a window of approxi- mately 10 bases (from the central AGGA string of the SD sequence to the start codon). In the window, and also immediately upstream of the SD sequence, A and U are preferred nucleotides [11]. In general, Gram-positive translation initia- tion sites have SD sequences that show a higher complementarity to the 16S-rRNA sequence than Gram-negative SD sequences [12].

4. GENE EXPRESSION IN LACTOCOCCUS LACTIS

The first successful transformation of Lacto- coccus lactis ssp. lactis protoplasts with plasmid DNA was reported in 1982 [13]. Since that time, several modified transformation protocols have been described [14-18], including the now gener- ally preferred method of electrotransformation [19-22]. Plasmid-free host strains became avail- able such as the widely used L. lactis ssp. lactis strains MG1363 [23] and IL1403 [24], allowing the rapid analysis of transformation and gene cloning experiments.

General purpose gene cloning vectors were developed, either based on cryptic L. lactis ssp. cremoris plasmids [25,26], or on non-lactococcal plasmids of Gram-positive origin [14,27]. The for- mer class of vectors offered the advantage of being able to replicate in several hosts, including E. coli and B. subtilis, thereby facilitating DNA manipulation and selection procedures. From these vectors, special purpose vectors were de- rived to study specific problems. Among these are promoter-probe vectors [28-32], terminator-probe vectors [32], vectors for the isolation of secretion signals [33,34], expression vectors [35], secretion vectors [36], and vectors for integration into the chromosome [37].

Owing to these developments, the species L. lactis became accessible to sophisticated genetic research techniques, and the cloning and analysis of a gradually increasing number of homologous genes and expression signals became possible. The accumulating information on lactococcal genes and their expression signals is currently being used to establish the expression of heterol- ogous genes of interest in L. lactis. In this paper, data available on homologous and heterologous gene expression in L. lactis will be reviewed.

4.1. Homologous gene expression

4.1.1. Cloning and characterization o f lactococcal expression signals using promoter and terminator probe vectors

A straightforward approach to obtain more insight in the organization of gene expression signals employs promoter probe vectors. The frst vector of this kind for use in L. lactis was de- scribed by Van der Vossen et al. [32,38]. This vector, pGKV210, contained the promoterless chloramphenicol acetyltransferase (cat-86) gene from B. pumilus, preceded by a multiple cloning Site. In this multiple cloning site, chromosomal DNA fragments obtained from L . lactis ssp. cre- moris We2 were inserted. The ligation mixtures were used to transform L. lactis ssp. lactis MG1363, either directly or via precloning in B. sttbtilis. Subsequently, the chloramphenicol acetyltransferase activity in the transformants was determined. Some of the chromosomal DNA in- serts that acted as a promoter for the expression of the cat-86 gene were further characterized by n/eans of DNA sequencing, and determination of the starting points of transcription. The main conclusion that could be drawn from this work was that the lactococcal promoter sequences closely resembled the consensus E. coli and B. subtilis 0 .43 promoter sequences. Because of the broad host range nature of the plasmids used in this study, it was possible to make a comparison of chloramphenicol acetyltransferase activity lev- els in L. lactis and in R subtilis. The activity levels observed in R subtilis were, on the aver- age, about 10 to 20 times higher than those in L. lactis. However, since the overall expression level

75

was mea.~ ared in this system, in addition to differ- ences in promoter activity also differences in plasmid copy numbei, mRNA stability, transla- tion (initiation) efficiency, or product stability may be involved. Some of these considerations also apply to the comparison of different promoters within one strain of L. lactis using this system, as is evident from the results obtained by Koivula et al. [30], who isolated DNA fragments with pro- moter activity from the chromosome of L. lactis ssp. lactis, in an approach similar to that used by Van der Vossen et al. Several promoters were characterized by DNA sequencing, mapping of the transcriptional starting points, measurement of CAT activities, and measurement of the rela= tive amounts of cat gene-specific mRNAs. The results showed a discrepancy between the relative promoter strenghts in L. lactis as determined by the measurement of CAT activities, and those determined by means of mRNA measurements, indicating that CAT activity can only be used to obtain a rough initial estimate of promoter strength. FurthermOre, CAT activities obtained in L. lactis were substantially (about 500-fold on the average) lower than in B. subtilis, while the mRNA levels in both species were roughly the same, ::ndicating host-specific differences in trans- lation (initiation) efficiency, or product stability.

Several other promoter probe vectors compa- rable to pGKV210 have been described [28,29], that either employed the promoterless cat-86 gene of B. pum.:!is, or the cat-194 gene of Staphy- lococcus aureus. Recently, Simons et al. [31] de- scribed the vector pNZ336, that utilizes the ho- mologous phospho-fl-galactosidase (lacG) gene of L. lactis ssp. lactis.

Once a suitable promoter has been cloned in vectors of the type described above, they can be conveniently used as terminator probe vectors, by inserting DNA fragments in between the pro- moter and the reporter gene. Van der Vossen [39] used this approach to show that the pre- sumed transcription terminator downstream of the L. lactis ssp. cremoris We2 protease gene [40] was functional in L. lactis ssp. lactis. Again, data obtained from this kind of experiments should be interpreted with care, because of the indirect mode of measurement of transcript formation:

76

Table i

Lactococcus lactis-derived transcription initiation sites

Promoter Sequence (a) (b) (I)

(2)

(3) (4) (5) (6)

(7) (8) (9) (lO) (11)

(12)

(13)

(14)

(14a) (15) (16) (17)

(18) (19)

(20) (21)

G A T C

TTTTTTCTTGACAGAAGAAGGCGAAAAATGGTATTATATTTAGGTACTGTT 18 8

CCTAAGACTGATGACAAAAAGAGCAAATTTTGATAAAATAGTATTAGAATT 17 6 GTGAGCTTGGACTAGAAAAAAACTTCACAAAATGCTATACTAGGTAGGTAA 22/17 4

AGCTAAACTCTTGTTTTACTTGATTTTATGTTAAAATAATTAATGAGTGTA 15/18 8/5 ACATTAAATTCTTGACAGGGAGAGATAGGTTTGATAGAATATAATAGTTGT 17 5

AAAGAAAGACTTGCATTTGTTGTTGAAAAATGCTAAAATACATAAGTCCGA 17 6

TGACTACGAATCAGGCGGTTGTAGGTTCGAATCCTACCGCTTGCATAAATA AGTT6ACCTTGAAAAAAAACTGAAAATCTGTTATCATAAATAATGGACATT 17 8

TACTATTGTATCAAAATTCTGTCAAATTTGATAAAATAAGGTACGAGTAAG 20 8 TGTCATCAA6CGACCTTTGAG6GGGCATTATGTTATAATTAAGCT_ATGAAG I I 6

ATTTTCGTTGAATTTGTTCTT CAATAGTATATAATATAATAGTATATAATA 21/15 5

ACC CTACGCTTGATGTAGTTAAGATTATATTATATAATATTATATACTATT ]7 7

TAAACG6CTCTGATTAAATTCTGAAGTTTGTTAGATACAATGATTTCGTTC 20 4

TAACATTTGTTAACGAGTTTTATTTTTATATAATCTATAATAGATTTATAA 23/20

ACATTTGTTAACGAGTTTTATTTTTATATAATCTATAATAGATTTATAAAA 23/20 6

TCATAAAGAAATATTAAGGTGGGGTAGGAATA6TATAATAT~'TTTATTCAA 22/14 6 TAATTTTTTGTTTTTTTTTATTTGTTTTTTTAAAAAATA6ATAACACCGTT 19 7

TCGAATTTTTCTTCCATATTTTCAAAGAATCCGTTACTATCTAACGATC 18 5

TATGAAAAATG6ACTGATGTAACTCC6TTGACT6TAATGGATGCA_GAAGAT 17 7

TTCATTATTTTTATAATCCTCACTAGTTATACATATAGTATTTGG_GTTTTT 18 7 TATGGAAAAATACA6ACAAGCAAACTAAGGAGGGTATATTGAATG_ACCGAC 16 5

AAAAT6TAAGATTG6A6TTACTAAAACAGTAACTTACTCCAACTGG_AGGTA 17 5

1323222527134472445365553142451443001123426"455 778598857573955X6646476779887796869 FYY966X773Z Y4689868X699665699?Y7647877898974X9657888866X2 16440242123 51552114032512312001341203113201420

T TG T A T T AAA A TA TTGACA . . . . . . . 17 . . . . . . . . TATAAT-- -7- - -Pu

Putative -35 and - 10 hexanucleotides are doubly underlined. Starting points of transcription are singly underlined. (a), spacing between putative -35 and -10 hexanucleotides; (b), spacing between putative -10 hexanucleotlde and the starting point of transcription; (1-5), P21, P23, P32, P44, P59 [38]; (6-10), PI, PI0, P21, ss45, ss80 [30]; (11, 12), pWV05 prtP, prtM [39,40,97l; (13), nisin precursor [98]; (14, 14a), p9BA-60RFAI = IcnMa, ORFB1 =IcnA ([42], M. van Belkum personal communication); (15), usp45 [43]; (16), lacR [99]; (17-21), Pal, Pf2, Pa3, Pg2, Pfl [41l. Promoters (17-21) are lactococcal phage promoters, The proposed -35 and - 10 hexanucleotides in entry (7) are situated upstream of the sequence shown [30]. GATC, number of G, A, T and C residues at each position totalled over entries (1-7), (9-12) and (14-21). Entries (8) and (13) were left out because of the inaccuracy in the determination of the transcription starting point. X = 10, Y= 11, Z = 13, F = 14. Figures printed in bold face indicate that the residue indicated occurred at least 10 times, or at I~ast 5 times more than the corresponding complementary residue (i.e. G versus C, and A versus T). *, seemingly preferred nucleotides in L lactis (as derived from this table); **, consensus E. coil and B. subtilis 0 .43 promoter [6,7]; Pu, Purine residue.

insert ion of possible te rmina tor sequences, or o the r sequences, in f ront of the coding sequence

may in te r fe re w i t h t rans la t ion in i t i a t ion , r a the r

t han te rm ina te t ranscr ip t ion .

4.1.2. Transcription initiation signals o f Lactococ- cus lactis-derived genes

Now that several lactococcal genes have been cloned and sequenced, their expression signals

can be compared with those obtained using the promoter-probe strategy described in the preced- ing paragraph. In Table 1 a compilation of lacto- coccal promoter sequences is shown. Entries 1 to 10 were isolated using promoter probe vectors [30,38], whereas entries 11 to 16 were isolated together with specific L. lactis genes. Entries 17 to 21 are iactococcal bacteriophage promoters that were isolated using a promoter probe vector [41]. Only promoters of which the functionality was confirmed by mapping of the starting points of transcription are included in the Table. Be- cause all known lactococcal promoters also func- tion in B. subtilis and E. coli, an obvious way to analyse these sequences is to look for the pres- ence of the well conserved sequences constituting a consensus promoter in these organisms [6,7]. Sequences resembling these so-called - 3 5 (TTGACA) and - 1 0 (TATAAT) boxes can in- deed be recognized in the L. lactis promoters and are doubly underlined in Table 1. It is often not possible, however, to unambiguously identify one particular hexanucleotide as the only possible - 3 5 or - 10 analog in the iactococcal sequences. Therefore, the sequences in Table 1 were aligned with respect to the starting points of transcrip- tion, and the numbers of G, A, T, and C residues at each position were totalled (Table 1, bottom). This procedure indicated a strong prevalence of A residues at the starting point of transcription (underlined in Table 1), and of T residues at the - 1 position, while A residues are present throughout the transcription initiation sites stud- ied, very few A residues were observed at the - 1 position and at the position immediately down- stream of the starting point of transcription. A - 35 "VI'G sequence can be recognized, as well as a region which is rich ia A residues, and there- fore resembles a consensus - 1 0 sequence. The distance separating these -35 and - 1 0 se- quences seems to comprise a few nucleotides more than the 17 to 18 usually found in B. subtilis and E. coli. It is important, however, to realize that only a small set of promoters has been anal- ysed. The whole promoter area contains very few C or G residues. Inspection of Table 1 reveals the presence of a TGN sequence directly preceding the - 1 0 hexanucleotide in sequences 1 to 6 and

77

8 to 10, but not in sequences 11 to 21. Several of the former promoters had been selected for their activity in B. subtilis, however, in which this TG sequence is strongly conserved [7,38]. Therefore, an inadvertent selection of promoters containing this sequence may have occurred.

Promoters 14 and 14a contain exactly the same sequence. Also the sequences upstream of the promoters, which contain an open reading frame the function of which is yet unknown, are identi- cal for at least 1 kbp (M. van Belkum, personal communication). The sequences downstream of the promoters are identical up to the 7th codon of the coding regions [42]. In spite of this high degree of similarity, a 2-bp difference in the starting points of transcription was observed [42].

It is generally believed that high levels of gene expression result from promoters that closely re- semble the consensus sequence. This does not seem to be an absolute requirement, however, as is apparent from sequence nr. 15 in Table 1. No clear - 35 sequence is present in this promoter belonging to the gene usp45, which was selected because of the high level production (secretion) of its gene product [43].

4.1.3. Translation initiation signals o f Lactococcus lactis-derived genes

Table 2 gives a compilation of L. lactis-derived translation initiation sequences. In all these se- quences, the translational start codon is AUG, although it is known that UUG and GUG act as start codons in L. lactis in the expression of the B. pumilis cat-86 gene [38], and of the B. subtili~ nprE gene [44], respectively. As far as available, the 30 nucleotides preceding the translational start codon, and the 27 nucleotides following this codon are shown in Table 2. In the region pre- ceding the start codon, the putative SD se- quences are underlined. These sequences are complementary, to the 3'-end of L. lactis 16S- rRNA (3'-UCUUUCCUCCA-5' [45]). The free energies of complementarity of the putative SD sequences and the 16S-rRNA (AG), and the sizes of the windows are also given in Table 2. The latter range from 4 to 15 bases, the most common value being 9. These sizes correspond to those usually observed in E. coli and B. subtilis [12,46].

78

Table 2

Lactococcus lactis-derived translation initiation sites

Gene Sequence AG

(!)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(!i)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(22)

UAIJGUUAAAAUAGCUUCUAGGAGUAUAGCCAUGAGUUUAAAUUUAAGAGAUCUUGAAUAU --ll .6 . . . . . 8 - - M e t S e r L e u A s n L e u A r g A s p L e u G t u T y r

AAUUAAAAAAGAUAGGAAAAUUUCAUGACUUACGCAGAUCAAGUUUUUAAACAA --9.4

. . . . 7 - - M e t T h r T y r A l a A s p G L n V a l P h e L y s G L n UAUUACGGAGGAUUUAAAAUGCGCUUUAACCAUUUUUCAAUUGUUGAC -14.4

. . . . . . 9 - -Me tA rgPheAsnH isPheSerZLeVa lAsp CAAAUGCUUUUUUUGAAA6GACUUACACUUAUGACUAAAACACUUCCUAAAGAUUUUAUU -14.0

. . . . . . 9 - - M e t T h r L y s T h n L e u P r o L y s A s p P h e I l e AAAUAGAUUUUUAGAACA6GGAGUAGGUAAAUGAUAACUUUGCAACACCAAGAUUGGGAA -9 .4

. . . . . 8 - - M e t l L e T h n L e u G L n H i s G L n A s p T r p G L u

GCUACUUGUUUUUGAUAAGGUAAUUAUAUCAUGGCUAUUAAAAAUACUAAAGCUAGAAAU -8 .4

. . . . . . 9 - - M e t A L a I L e L y s A s n T h r L y s A L a A r g A s n UUGAUUGGAGUUUUUUAAAUGGUGAUUUCAGAAUCGAAAAAAAGAGUU --9.4

. . . . . . 9 - - R e t V a L I l e S e r G L u S e r L y s L y s A r g V a L

UUAUAAAAAUAAGGAGAUUAUUAUGAAAAAUCAAUUAAAUUUCGAAAUCCUA -12.8 . . . . 7 - - H e t L y s A s n G L n L e u A s n P h e G t u l L e L e u

UAUAAAAAUUGAAAGGAUUCAGGUACUAAAAUGAAAAAAGAUGAAGCAAAUACAUUUAAA -14.0

- - - - - 3 - - = = = 6 = = M e t L y s L y s A s p G l u A l a A s n T h r P h e L y s / - 7 . 4 CCA6UACACUAAAEAAAGGCUUACAAAUUAAUGAAAAAUGAUAAUUUUUUAAUAAAUAGA -14.0

. . . . . . l O - - M e t L y s A s n A s p A s n P h e L e u I l e A s n A r g

CGGGUUGCACCAUUGAGGAUUAGUUAAGAUAUGAAAAAAAAACAAAUAGAAUUUGAAAAC --9.4

- 1 4 = = = = = = = 1 1 = = R e t L y s L y s L y s 6 L n l L e 6 L u P h e G l u A s n / - 9 . 4 CUAAUAAAAAAGAACUGAGGUUUAGAGUUAAUGAAAAAAAAAGUUGAOACAGAAAAACAA --9.4

. . . . . . . . 1 2 - - R e t L y s L y s L y s V a L A s p T h r G L u L y s G L n

ACCGUGGGAACAAAUGUACACUAUCGGUUGGGGUCAUUAUGGA -7 .2 - 4 - - M e t T y r T h e I l e G L y T r p G L y H i s T y r G L y

AUGAAUAAAAAU6ACAGCGAGGAUAUAUCAAUGAACUAUUUUAAAGGUAAACAAUUUCAA -9 .4

- l O = = = = 7 = = M e t A s n T y r P h e L y s G l y L y s G l n P h e G l n / - 9 . 4 AAGGUUUCUCGCUCAG6UUUCUAUGAAUACEUGCAUCGU¢GUCCUUCAAAACAACAAGU6 -7 .5

. . . . . . . . . . . . 1 5 - - R e t H i s A n g A n g P r o S e r L y s G L n G L n V a L

AUGAAACUUUUGGAAAGUGGAGGAUAUUGGAUGCAAAGGAAAAAGAAAGGGCUAUCGAUC -14.4

. . . . . . 9 - - R e t G L n A r g L y s L y s L y s G L y L e u S e r I L e ACUGUAAGCAUUUCAGAGGAGACCGAAUCGAUGAAGAAAAAAAUGCGCCUUAAAGUAUUA --11.6

. . . . . . l O - - R e t L y s L y s L y s R e t A r g L e u L y s V a l L e u

AAUGGGAGGAAAAAUUAAAAAAGAACAGUUAUGAAAAAAAAGAUUAUCUCAGCUAUUUUA - 4 . 6

- - S - - R e t L y s L y s L y s l L e I t e S e r A t a I l e L e u UACAAAAUAAAUUAUAAQGAGGCACUCAAAAUGAGUACAAAAGAUUUUAACUUGGAUUUG -17.8

. . . . . . l O - - M e t S e r T h r L y s A s p P h e A s n L e u A s p L e u

GGUAAAAAAAUAUUC6GAGGAAUUUUGAtAUGGCAAUCGUUUCAGCAGAAAAAUUCGUA -14.4

. . . . . . . 1 1 - - H e t A L a I L e V a L S e r A L a 6 L u L y s P h e V a L

CGUGAUGUGUGAGGGAAAG6A6UC6CUUUUAUf6¢CAAAAGUGGACUUUAUACA66C6UA -16.2

. . . . . . 9 - - M e t A L a L y s S e r G L y L e u T y r T h r 6 1 y V a L

UAUAAGACAGAUAUAAAUGGAGAUAGAAUUAUGAUGAAUCACCCGCAUUCA~GUCAUAGG --9,4

. . . . . . ~ - - ~ e t R e t A s n H i s P r o H i s S e r S e r H i s I L e

Table 3

lqucleotide frequencies at positions + 4 - + 30

79

l o s G A U C

Min Obs Ma:i Min Obs Max Min Obs Max Min Obs Max

4 4 4 4 12 14 15 1 I 3 2 3 3 5 1 3 3 11 11 11 3 3 3 5 5 7 6 1 3 18 0 9 18 0 6 13 0 4 13

7 0 0 0 16 17 18 3 4 4 0 I 3 8 2 2 2 12 12 12 5 5 5 3 3 3 9 0 1 13 0 9 16 0 10 15 0 2 15

10 4 4 4 11 12 14 1 3 4 2 3 4 11 1 2 3 13 13 13 4 4 4 2 3 4 12 0 2 15 0 10 16 0 7 14 0 3 14

13 7 7 7 6 6 7 0 3 4 5 6 8 14 2 2 3 11 11 I t 6 6 6 2 3 3 15 1 3 16 0 9 16 0 10 15 0 0 15

16 4 4 4 5 6 9 4 6 7 4 6 7 17 4 4 6 6 6 6 6 6 6 4 6 6 18 1 2 13 0 7 14 0 10 19 0 3 19

19 6 6 6 8 8 11 2 6 7 1 2 3 20 2 2 5 12 12 12 4 4 4 1 4 4 21 0 1 17 0 13 17 0 6 14 0 2 14

22 6 6 6 7 8 8 2 3 6 3 5 6 23 0 I 1 10 10 10 7 7 7 4 4 5 24 0 ! 15 0 11 17 0 10 15 0 0 15

25 6 6 6 5 7 8 6 7 7 2 2 4 26 4 4 5 9 9 9 8 8 8 0 1 1 27 1 2 12 0 8 13 0 9 16 0 3 16

28 7 7 7 6 7 7 1 4 5 3 4 8 29 2 2 2 9 9 9 11 II 11 0 0 0 30 0 2 15 0 13 18 0 4 17 0 3 17

Nucleotide frequencies in the 5'-parts of the 22 coding sequences presented in Table 2. The positions + 1 - + 3 . Pos, position hr.; Obs, observed frequency: Min and Max, minimal and changing the amino acid sequence, respectively.

AUG translational start codon covers maximal frequencies possible without

Notes to Table 2. The functionality of the translation initiation sites has been inferred from expression studies (entries (1-13, 16-19, 22)), lacZ translational fusions (entries (1, 20. 21)), or homology to known DNA or protein sequences (entries (14. 15)). AG, free energy (kcal/mol) of complementarity between the SD sequence (doubly underlined) and the 3'-end of L. lactis 16S-rRNA (3' UCUUUCCUCCA-5 ' [45]) [100]. Figures preceding the deduced amino acid sequences indicate the 6ze of the window between the SD sequence and the translational start codon. (1), mleR [101]; (2), thyA [102]; (3), X.PDAP [103,104]; (4), pbg (lacG) [29,105,106]; (5), pTR2030 hsp [107]; (6), pWV01 repA [108], nearly identical to pSH71 repA [29]; (7), pWV01 ORFC [108], and pSH71 ORFC [29]; (8-12), p91M-60RFA1 = IcnMa, ORFA2 = IcnMb, O R F A 3 = IciM, ORFB2 = Ici,4, ORFC2 = IciB ([42]; M. van Beikum personal communication); (13), d, vML3 lysin [109]; (14). ISS1 putative transposase [110] and IS946 putative transposase [111], nearly identical to: ISS1N and ISS1W putative transposases [112]; (15), IS904 putative transposase [113,114]; (16, 17), pWV05 prtP, prtM [40,97]; (18), usp45 [43]; (19), nisin precursor [98,113,115]; (20,21), ORF32, ORF44 [38]; (22), citP 1116]. Sequences downstream of the (putative) starting points of transcription are given for entries (2, 3, 7, 8, 13) and (20). Entries (9, 10, 11) and (12) show sequences that are part of an operon. The translational stop codons of the upstream open reading frames are printed in bold face.

80

.a

° ° i ° . . . . . . . .

I

Lys

A

AG

3

5 8

12

11

7 1

1 0

1 0

1 0

1 2

8 8

66

18

5 1

5 13

17

7 30

A

AA

22

17

48

22

44

18

10

2

4 11

3

9 5

12

8 16

17

68

21

26

5

19

5 41

2 70

Met

A

UG

6

9 15

8

10

4 3

5 2

4 2

3 3

3 6

4 11

21

7

6 3

5 22

16

2 10

0

Ph

e U

UU

17

12

31

13

33

7

1 1

1 14

3

1 2

3 2

6 7

31

5 2

1 6

13

212

68

UU

C

3 4

11

12

10

0 0

1 1

3 1

2 1

0 2

3 1

22

4 3

0 5

10

99

32

Pro

C

CG

2

0 3

3 !

0 0

0 0

0 0

0 0

0 1

0 0

18

0 1

0 0

2 31

15

C

CA

3

7 10

10

3

I 0

1 1

1 0

2 1

1 2

2 1

37

2 3

1 1

8 98

47

C

CU

0

4 11

7

4 4

0 0

1 5

0 0

0 0

2 0

5 11

0

5 0

2 2

63

30

CC

C

2 0

0 0

3 0

0 0

0 0

0 0

0 0

0 1

2 3

2 0

l 1

0 15

7

Ser

A

GU

8

1 11

2

17

6 0

0 0

2 0

3 2

0 2

3 2

29

ll

17

3 4

6 12

9 24

A

GC

2

1 2

1 3

3 0

0 0

0 0

0 1

0 2

1 3

23

11

13

1 3

2 72

13

U

CG

5

0 1

0 0

0 1

1 0

0 1

1 0

0 0

2 1

17

1 0

2 1

I 35

6

UC

A

5 3

7 8

14

5 3

1 3

3 2

1 0

1 6

1 7

26

4 32

2

1 11

14

6 27

U

CU

5

4 23

4

15

2 2

1 2

2 0

2 1

5 l

3 7

27

2 14

1

3 6

132

24

UC

C

1 0

2 0

4 0

0 0

2 1

0 1

0 2

0 1

0 14

1

0 0

2 1

32

6

Th

r A

CG

1

2 9

3 3

2 0

0 1

0 0

0 0

0 1

4 2

63

5 0

0 0

8 10

4 19

A

CA

1

6 7

7 15

5

2 0

1 5

3 2

6 2

6 3

4 57

6

18

5 9

9 17

9 32

A

CU

2

4 16

9

16

2 0

1 2

1 2

1 2

1 l

6 4

57

1l

20

l 7

l 16

7 30

A

CC

4

1 8

i 2

0 0

0 0

0 1

1 0

0 1

2 6

61

8 0

0 5

I 10

2 18

Trp

U

GG

0

6 16

10

3

3 0

1 2

l l

l 2

2 6

5 2

6 2

4 0

0 I

74

100

Ty

r U

AU

9

12

21

23

35

12

! 1

2 5

2 2

4 2

8 15

11

44

7

8 0

3 7

234

70

UA

C

2 3

9 14

8

l 0

0 0

3 2

0 0

1 3

3 5

25

6 4

0 3

7 99

30

Val

G

UG

0

1 6

4 2

1 1

0 0

0 0

0 0

0 1

2 3

28

4 2

0 2

5 62

14

G

UA

3

3 6

7 l0

q

0 1

3 5

0 0

1 2

4 3

3 l I

2

8 2

4 4

86

19

GU

U

6 7

19

14

13

7 2

2 2

5 3

0 3

6 10

2

3 65

9

16

1 2

25

222

50

GU

C

I 5

I l

3 4

l 0

0 1

0 0

0 0

1 l

4 3

29

2 0

0 1

7 74

17

End

U

GA

0

0 0

0 0

1 0

0 0

0 1

0 0

0 0

0 0

0 0

0 0

0 0

2 9

UA

G

0 0

0 0

0 0

0 0

0 1

0 0

0 0

0 0

0 1

0 i

0 1

0 4

17

UA

A

1 1

1 I

1 0

1 1

1 0

0 1

1 1

I 1

1 0

1 0

1 0

1 17

74

Th

e fi

gure

s in

dica

te t

he

codo

n fr

eque

ncie

s us

ed i

n th

e ge

nes

indi

cate

d at

the

top

of

the

Tab

le.

AA

, am

ino

acid

; C

od,

codo

n; a

, ru

ler

[101

]; b

, th

yA [

102]

; e,

X-P

DA

I"

[103

];

d,

pbg

(lac

G)

[29,

105,

106]

; e,

pT

R~:

II30

hsp

[1

07];

f-g

, pw

V01

re

pA,

OR

FC

[1

08];

h-n

, p

9B

4-6

0R

FA

I=Ic

nM

a,

OR

FA

2=Ic

nMb,

O

RF

A3=

IciM

, O

RF

BI

=Ic

nA

, O

RF

B2

= I

ciA

, OR

FC

I =

lcn

B,

OR

FC

2 =

Ici

B ([

42];

M.

van

Bel

kum

per

sona

l co

mm

unic

atio

n); O

, dbv

ML

3 ly

sin

[109

]; p

, IS

SI

puta

tive

tra

nspo

sase

[1

10];

q,

IS90

4 pu

tati

ve t

rans

posa

se [

113]

; r-

s, p

WV

05 p

rtP

, prt

M [

40,9

7]; t

, us

p45

[43]

; u,

nis

in p

ree.

urso

r [9

8,11

3,11

5];

v, l

acR

[99]

; w

, ci

tP [

116]

. Tot

, ro

w to

tals

; %

,

codo

n us

age

per

amin

o a

cid

in %

. T

he

(G+

C)

cont

ent

calc

ulat

ed f

rom

thi

s T

able

is

38.6

%.

8 2

Most ZIG values range from - 8 to - 14 kcal/mol, with a mean value of -11 kcal/mol. In this respect the SD sequences of L. lactis resemble those of E. coli, where z ig values ranging from - 9 to - 1 2 kcal/mol are common, with a mean of -11 to - 1 2 kcal/mol [12]. In B. subulis, on the other hand, ZIG values ranging from - 14 to

- 19 kcal/mol are common, with a mean value of - 16 to - 17 kcal/mol [12]. In the area preceding the SD sequence often long stretches of A or U residues are present.

The SD sequence, the start codon, and the window between the two, are generally consid- ered to be the main determinants of translation initiation efficiency. Additional sequences, lo- cated outside of this region, may have an effect on the efficiency of translation initiation, how- ever. The nucleotide sequence of the mRNA per se may influence the interaction between the mRNA and the ribosome [8,12,47,48]. Alterna- tively, secondary structures may be formed in the mRNA that influence the efficiency of translation initiation [47,49].

In order to examine whether certain nu- cleotides are preferred at certain positions within the coding sequence, the frequencies of occur- rence of the nucleotides present in the coding sequences from Table 2 were listed in Table 3. Together with the observed frequency of a given nucleotide at a given position, the minimal and maximal frequencies of that nucleotide at that position are given. The maximal frequency was obtained by introducing that nucleotide at that position in all coding sequences where it was possible t o do this without affecting the amino acid sequence of the gene product, i.e. by using synonymous codons. Similarly, the minimal fre- quency was obtained by avoiding the use of the nucleotide of interest at a given position as much as possible. When the AUG start codon is set to cover positions + 1 to + 3, a strong prevalence of A residues can be seen at positions +4, +5, +7, +8, +10, +11, +20, +21 and +30. A similar high preference of A residues in this region was observed in B. subtilis [12]. From Table 3 it is dear, however, that the high incidence of A residues at the majority of these positions is di- rectly linked to the amino acids encoded by the

codons covering these positions, i.e. the observed nucleotide frequencies more or less equal the minimal frequencies. Of the positions mentioned, only at positions + 21 and + 30 there is no strong need for an A in the coding sequence (the mini- mal frequency is 0), and yet this residue prevails. Concerning the positions +4, +5, +7, +8, +10, + 11 and + 20, the question arises as to whether a certain nucleotide is preferred in the mRNA, or whether the preference applies to a certain (set of) amino acid(s) in the gene product. Table 2 shows that there is no clear preference of one particular amino acid in the deduced protein sequence at either of the positions mentioned. Analysis of an extended number of translation initiation sequences is required to reveal whether the seeming nudeotide preferences are real pref- erences. Although at some positions G, A, U and C residues are present in more or less equal numbers (e.g. positions + 13, + 16, + 17), at many other positions there seems to be a bias towards A and U residues. This may reflect the generally high (A + T) content of L. lactis (see also Section 4.1.4 on codon usage).

4.1.4. Codon usage in Lactococcus lactis Little is known about codon usage in L. lactis.

Kok et al. [40] made an analysis of the codon usage in the L. lactis ssp. cremoris protease gene, and Porter et al. [50] analysed the L. lactis ssp. lactis pbg gene. Table 4 includes these data and those of a number of other lactococcal genes. The Table shows that the codon usage can vary markedly between genes. Although in many cases this variation may be a consequence of chance because of the small numbers of the various codons involved per gene, some differences are striking, as, tbr example, the noticeable underre- presentation of the CAG codon for glutamine in usp45. Another example concerns isoleucine which is encoded predominantly by the AUU codon in most genes. In addition to this codon, a large number of AUA codons is present in the hsp gene, whereas in the. prtP gene a large num- ber of AUC codons is found. Furthermore, in the prtP gene the AAG and AAA codons for lysine are equally represented, whereas the overall fig- ures (column: %) show a marked preference for

Table 5

Codon usage in L. lactis. B. subtils and E. coli

83

AA cod LI Bs Ec AA cod LI Bs Ec AA cod LI Bs Ec AA cod LI Bs Ec

Ala GCG 13 22 31 Gin CAG 24 48 76 Leu UUG 24 14 9 Ser AGU 24 11 6 GCA 29 29 22 CAA 76 52 24 UUA 30 24 7 AC_~ 13 26 26 GCU 40 27 26 Glu GAG 19 30 27 CUG 8 21 6g UCG 6 11 11 GCC 18 22 21 GAA 8! 70 73 CUA 11 6 2 UCA 27 20 7

Arg AGG 5 11 1 Gly GGG 11 13 8 CUU 20 25 7 UCLJ 24 23 23 AGA 24 22 1 GGA 26 32 5 CUC 8 10 7 UCC 6 9 27 CGG 4 14 3 GGU 43 24 48 Lys AAG 30 27 24 ",.'hr ACG 19 24 18 CGA 19 12 3 GGC 21 31 39 AAA 70 73 76 ACA 32 45 7 CGU 36 22 56 His CAU 71 67 54 Phe UUU 68 57 37 ACU 30 15 25 CGC 11 19 36 CAC 29 33 46 UUC 32 43 63 ACC 18 16 50

Asn AAU 70 56 26 lie AUA 18 13 3 Pro CCG 15 40 65 Tyr UAU 70 65 40 AAC 30 44 74 AUU 58 50 36 CCA 47 21 16 UAC 30 35 60

Asp GAU 68 61 46 AUC 23 37 61 CCU 30 30 12 Val GUG 14 24 27 GAC 32 39 54 End UGA 9 21 17 CCC 7 9 7 GUA 19 21 22

Cys UGU 74 68 43 UAG 17 8 8 Met AUG ** ** ** GUU 50 31 36 UGC 26 32 57 UAA 74 71 75 Trp UGG ** ** ** GUC 17 24 15

The figures represent the codon usage in L. lactis (LI) (Table 4), B. subtilis (Bs) [117] and E. coli (Ec) [117], expressed in percentage per amino acid. AA, amino acid; cod, codon; * *, 100%.

the A A A codon. In the same gene, all four codons for threonine are equally represented, whereas the overall figures seem to favour the ACA and ACU codons. In usp45 these two codons are the only ones present for threonine. At this time we can only guess about the meaning of these differ- ences. Also in other bacterial species, and higher organisms, a considerable within-species diversity in codon usage has been observed [51]. Especially in E. colt this diversity has been associated with gene expression, with a strong bias towards a particular subset of codons in highly expressed genes, and a more even codon usage in moder- ately expressed genes [51]. In contrast, a markedly unbiased codon usage has been reported for B. subtilis [52].

In Table 5, the overall codon usage in L. lactis as given in Table 4 is compared with that of E. coli and B. subtilis. In general, L. lactis shows a preference for codons with an A or a U at the wobble position, which is reflected in the high (A + T) content of the organism (approx. 62% [53], Table 4). L. lactis, E. coli and B. s~btilis resemble each other in the preference for ,some codons, e.g. GAA for glutamic acid, AAA for lysine, and U A A as the translational stop codon.

For some amino acids L. lactis and E. coil show opposite codon preferences, while R subtilis takes an intermediate position, as for instance with respect to the asparagine and glutamine codons. For some amino acids, the codon choice of L. lactis resembles that of R subtilis, whereas for others it resembles that of E. coli. However, neither L. lactis nor R subtilis show the clear preference of E. coil for one or two codons in those cases where 6 codons are available, i.e. for arginine and leucine.

In the preceding paragraphs an overview has been given of some features that are, or are likely to be, important in gene expression. It was shown that also in L. lactis considerable variation exists in promoter sequences, translation initiation se- quences, and codon usage. Unfortunately, little is known about the optimal characteristics of each of these features, and their possible interactions, necessary to achieve high level gene expression. Whereas it is common belief that high level ex- pression is obtained using promoters and ribo- some binding sites with high similarity to the consensus sequences, it is intriguing that the usp45 gene [43], selected for its high level expres- sion, lacks both.

84

4.2. Heterologous gene expression

Relatively few studies have dealt with the ex- pression of heterologous genes in L. lactis. In the fc, llowing paragraphs an overview of heterologous gene expression in L. lact/s is presented.

4.2.1. Antibiotic resistance markers The first class of heterologous genes expressed

in L. lactis is composed of genes specifying an- tibiotic resistances. Plasmids from non-lactococ- ca' Gram-positive bacteria, or derivatives thereof, carrying these antibiotic resistance markers were used in conjugation experiments, or as gene cloning vectors [14,27,54]. Alternatively, the genes xpeci~ing antibiotic resistance were used to label cryptic lactococcal plasmids in order to use these as selectable gene cloning vectors [25,26]. The fact that iaetococeal expression signals closely resemble those generally present in Gram-posi- tives, allowed the expression of these heterolo- gnus genes to sufficient levels without the need for further adjustments. This noticeable feature subsequently permitted the rapid development of the genetic analysis of L. lactis. Among the an- tibiotic resistance-conferring genes most com- monly used are the ehioramphenicol transferase (cat) gene of the S. aureus plasmid pC194 [55], and the MLS resistance genes of the S. aureus plasmi.'l pE194 [56] and the Enterococcus faecalis plasmid pAM/31 [57]. In addition, the B. pumilis cat-86 gene [30,38], and the E. coli TEM-/3- lactamase gene [33,34] have been used to isolate and study lactococcal expression and secretion signals, respectively.

4.Z2. Other heterologous genes Heterologous genes other than antibiotic resis-

tance markers have been expressed in L. lactis by transcriptional or translational fusion to L. /act/s-derived expression signals. Among these is the E. coli /acZ gene, encoding /3-galactosidase. De Vos and Simons [58] fused this gene to the ninth codon of the L. lact/s ssp. cremor/s SKl l prtP gene, and showed that the hybrid gene could provide the lactose-deficient L. lactis ssp. /act/s strain MG1363 with the ability to grow on lac- tose. The lacZ gene has, also in L. lac#s, proven

to be a useful tool in the study of translation [59,60]. Recently, Haandrikman et al. [61] used the lacZ gene in combination with the Cyamops/s tetragonoloba (Guar) eDNA gene encoding a- galactosidase, to study divergent expression sig- nals. The a-gal gene was fused to the signal sequence of the B. subtilis a-amylase gene [62], and both galaetosidases were provided with L. ~act~s-derived transcription and translation initia- tion signals. The B. stearothermophilus and B. licheniform/s a-amylase genes, deprived of their signal sequences, have been used to isolate and study lactococeal secretion signals [33,36].

Other heterologous gene products produced by L. lactis have potential applications in dairy industries, such as the bovine prochymosin (Sec- tion 6; [63,64]), the B. subtilis neutral protease (Section 6; [44]), and several lysozymes ([35]; Van de Guehte et al., unpublished results).

Lysozymes (EC 3.2.1.17), defined as 1,4-/3-N. acetylmuramidases cleaving the glycosidic bond between the C~ of N-acetylmuramie acid and the Ca of N-acetylghcosamine in the bacterial pepti- doglycan [67,68], may find an application because of their antimicrobial activity against several bac- teria involved in food spoilage and food-borne disease [69-71]. L. lactis was transformed to pro- duce these enzymes using a iactococcal expres- sion vector that contained gene expression signals and a short Y-truncated open reading frame of lactococcal origin [35]. Copy DNA encoding the mature hen egg white lysozyme was fused in-frame to the lactococeal open re2ding frame, which resulted in the production of a hen egg white lysozyme fusion protein in L. lactis [35]. Also, the unfused mature hen egg white lysozyme coding eDNA was expressed in L. lactis, albeit with reduced efficiency [65]. The differences in the expression levels of the fused and unfused gene are probably (partly) due to differences in the efficiency of translation initiation [59,60].

An enzyme closely resembling the hen egg white lysozyme, is encoded by the E. coli bacte- riophage T4 e gene. In order to express this gene in L. lact/s, it was provi~led with L. ~act~s-specific expression signals that were positioned immedi- ately upstream of the coding sequence [65]. Alter- natively, the gene was expressed by means of

translational coupling to a short 3'-truncated L. lactis derived open reading frame [65].

The E. coli bacteriophage A lysozyme, the A R gene product, was produced in L. lactis after the lysozyme coding sequence and translation initia- tion signals were transcriptionally fused to a 3'- truncated iactococcal open reading frame [66]. In this configuration, efficient expression of the AR gene was shown to be dependent on the transla- tion of the preceding lactococcal open reading frame.

5. IMPROVEMENT OF GENE EXPRESSION IN LACTOCOCCUS LACTIS

In the preceding paragraphs, it was shown that several heterologous genes could be expressed in L. lactis employing L. !actis-specific expression signals. Depending on the nature of the heterolo- gous gene product and its application, however, improved gene expression may be desired. Simi- larly, enhancement of the expression of homolo- gous genes may be advantageous for certain ap- plications. Several strategies can be envisaged to accomplish enhanced gene expression. One way is to increase the gene dosage, for example by incorporating the gene in a high copy number plasmid, or by multiple insertions of the gene into the chromosome. Several high copy number plas- mids for use in L. lactis have been described, for example pCK21 [72], plL253 [27], pNZ12 [25], and derivatives thereof. De Vos et al. [63] re- ported an increased ~production of the L. lactis ssp. cremoris SK11 protease after cloning of the genes involved in its production, prtP and prtM, on a plasmid with a copy number higher than that of the plasmid that originally carried these genes. Recently, Leenhouts et al. [73] realized the inte- gration of the L. lactis ssp. cremoris Wg2 prtP and prtM genes into the chromosome of L. lactis in a Campbell-like way. The protease activity displayed by the transformants was shown to be dependent on the number of copies of the genes present, although no linear relationship was ob- served. /

In the following we will focus on methods of increasing gene expression without affecting the

85

gene dosage. One reason for this is, that in food- grade applications it may be preferable to insert a heterologous gene at a specific site in the chro- mosome of the host. In a fixed copy number situation, gene expression may be enhanced by improvement of the efficiency of transcription initiation. Since, as described in paragraph 4.1.2., it is as yet not clear what exactly constitutes an optimal transcription initiation region, the most obvious strategy to enhance transcription is to replace the promoter present upstream of a gene of interest with another, more efficient, pro- moter. This strategy was applied by Van der Vossen ([39]; unpublished results), who isolated a number of promoter sequences from the chromo- some of L. lactis ssp. cremoris Wg2, and used these to improve the transcription of the L. lactis ssp. cremoris prtP and prtM genes in L. lactis ssp. lactis, resulting in elevated mRNA levels. A concomitant increase in protease activity was ob- served.

A second important determinant of gene ex- pression is the efficiency of translation initiation. The characteristics of a translation initiation re- gion that are considered to be of prime impor- tance in determining translation initiation effi- ciency are: the extent of complementarity be- tween the SD sequence and the Y-end of 16S- rRNA, the start codon, and the window between these two sequences [12,46]. Furthermore, natu- ral ribosome binding sites tend to have a low potential for secondary structure formation [49,74], which is in part a consequence of the generally low (G + C) content of these sites. The nucleotide sequence in the region surrounding the start codon might also be important in estab- lishing additional contacts between the ribosome and the mRNA [8,12,47,75]. Changes introduced in a ribosome binding site to increase the comple- mentarity to the 16S-rRNA, and to convert the window to a size that is frequently observed, can in some cases increase gene expression [59]. How- ever, this method, or that of precisely fusing a coding sequence to an existing ribosome binding site, may not be generally preferable, because the exchange of coding sequences and the introduc- tion of changes in ribosome binding sites can easily affect mRNA secondary structure forma-

86

tion in the translation initiation region, and thus affect gene expression [60]. In fact, results of many changes introduced in ribosome binding sites with the aim to improve the interaction between the ribosome and the mRNA, may to a considerable extent reflect changes in mRNA secondary structure [46]. Therefore, a configura- tion may be preferred in which translation of a gene of interest is coupled to that of a preceding efficiently expressed open reading frame [59],

The principle of translational coupling was first recognized by Oppenheim and Yanofsky [76], and has since been described for many naturally oc- curring and artificial systems [77-83]. It ensures the efficient translation of a gene by making it dependent on the translation of a gene immedi- ately preceding it. Translation of the upstream gene may serve to resolve secondary structures that would otherwise occlude the ribosome bind- ing site of the downstream gene, thereby render- ing this ribosome binding site accessible to free ribosomes [76]. Alternatively, ribosomes that have translated the first gene may proceed to translate the second gene after a translational restart [76]. Especially in the latter situation, translation of the downstream gene would profit from an effi- cient initiation of translation of the upstream gene. Therefore, if an efficient translational cou- pling could be realized, it would be possible to make use of efficient expression signals of an upstream gene, without disturbing the delicate interplay that may exist between transcription and translation signals and the (5'-end of the) coding sequence. In reference [59] a model sys- tem is described in which the juxtaposition of the stop and start codons of two tandemly arranged open reading frames, the lactococcal ORF32 and the E. coil lacZ, was varied, in order to optimize the translational coupling between the two open reading frames in L. lactis. This study showed the result of translational coupling, which in this case most likely occurred via a translational restart mechanism, to be critically dependent on the configuration in which the two open reading frames were juxtaposed. Only configurations in which the stop codon of the upstream ORF32 and the start codon of the downstream lacZ were contiguous or partially overlapping, resulted in

translational coupling. The best result was ob- tained with an AUGA configuration, in which UGA represents the stop codon of the upstream ORF32, and AUG the start codon of iacZ. It was shown that translational coupling could be used to enhance the expression of lacZ as compared to the expression of this gene in the absence of ORF32.

Other processes that may affect gene expres- sion, but are difficult to optimize, if at all, are the rates of transcription and translation. The latter variable may be correlated to codon usage, al- though opinions are divided on this issue [8,51,84-86]. In E. coli, a strong correlation is observed between high levels cf gene expression, the use of a particular subset of codons, and the high abundance of specific tRNAs [51,85], sug- gesting a causal connection between these fac- tors. Bonekamp et al. [84], however, have pre- sented evidence suggesting that translation rates of individual codons do not reflect the concentra- tions of the corresponding tRNAs or the frequen- cies with which the codons are used in E. coli. Furthermore, several authors state that initiation, and not elongation, is usually the rate-controlling step in translation [8,87]. For L. lactis, there is insufficient information to answer the question whether a particular codon usage might raise translation efficiency (paragraph 4.1.4.).

No information is available on the possibility to increase the abundance of specific mRNAs by improving the rate of transcription. As in transla- tion, elongation in transcription usually seems not to be rate-limiting [8]. mRNA stability, however, may be improved by the addition of a 3'-exten- sion that is capable of secondary structure forma- tion, such as a rho-independent transcription ter- minator, thus protecting the transcript from degradation by exonuclease activity [8~89].

In conclusion, strategies to improve gene ex- pression heavily depend on improvement Gf initi- ation of transcription and translation. Although transcription initiation signals, translation initia- tion signals and coding sequences are often thought of as distinct boxes that can easily be exchanged to improve gene expression, there are strong indications that this is an oversimplifica- tion of reality. Therefore, it may be preferable to

use an undivided set of expression signals, i.e. the promoter, the ribosome binding site, and (the 5'-part of) the coding sequence of a highly ex- pressed gene, to express a heterologous gene, provided that translation of the latter can be efficiently coupled to translation of the former. The feasibility of this approach has been demon- strated in references [59] and [65], describing the expression of the E. coli lacZ, and the E. coli bacteriophage T4 e gene in L. lactis, respectively.

6. SECRETION OF HETEROLOGOUS PRO- TEINS

Various approaches have been used in order to make L. lactis secrete heteroiogous gene prod- ucts. Van Asseldonk et al. [43] screened several L. lactis strains for secreted proteins by SDS- polyacrylamide gel electrophoresis of culture su- pernatants, and showed that all strains produced extracellular proteins, one of which, banding at 50-60 kDa, was particularly abundant. The chro- mosomally located gene usp45, encoding this pro- tein in L. lactis ssp. lactis MG1363, was isolated. DNA and protein sequence analysc~ revealed that the gene most probably contained a signal se- quence of 27 codons. This signal sequence was used to direct the secretion of the B. stearother- mophilus a-amylase [36]. Similarly, the signal se- quence of the L. lactis ssp. cremoris SKl l cell envelope-associated protease gene, which most likely has a length of 33 eodons, could be used to effect secretion of the bovine prochymosin [63], a protein of major importance in the dairy industry. Both signal peptides conform to the structural characteristics of signal peptides sensitive to sig- nal peptidase 1 cleavage, as described by Von Heijne and Abrahmsen [90].

In the strategy described above, secreted ho- mologous gene products were identified, after which the corresponding genes were cloned and analysed. The putative signal sequences were subsequently used to direct the secretion of het- erologous gene products. Another approach was used by P6rez-Martinez et al. [33] and Sibakov et al. [34]. In both studies the 5'-truncated E. coli TEM-g-lactamase gene was used as a reporter

87

gene in a signal sequence probe vector. In addi- tion, P6rez-Martinez et al. [33] used the 5'-trun- cated B. licheniformis a-amylase gene as the re- porter gene. Fragments of the L. lactis ssp. lactis chromosome were cloned upstream of these genes, after which gene expression and secretion of the gene products was studied. In both studies secretion of the heterologous gene products was observed with signal sequences resembling those described by Von Heijne and Abrahmsen [90]. Furthermore, P~rez-Martinez reported species- specific differences in the amounts of protein secreted. The chromosomal DNA inserts had not only been selected for the presence of a signal sequence, however, but also for the presence of gene expression signals. Therefore, it is not clear whether the differences observed in protein pro- duction in E. coil, B. subtilis, and L. lactis were due to species-specific requirements with respect to the signal sequences, or with respect to the gene expression signals. An effect of species- specific signal sequence requirements on secre- tion efficiency is quite conceivable, as pointed out by Von Heijne and Abrahmsen [90], who showed that signal sequences from Gram-positive bacte- ria tend to be longer than those used in E. coll.

The functionality of a heterologous signal se- quence in L. lactis has been demonstrated by the secretion of the B. subtilis neutral protease using the nprE gene specific signal sequence [44]. The secretion of this enzyme by L. lactis may provide a solution to the problems involved in the appli- cation of the purified enzyme to accelerate the process of cheese ripening, which are the often uneven distribution and low level of retention in the cheese curd [91-95]. The B. subtilus nprE gene, encoding the neutral protease, was pro- vided with L. lactis derived transcription signals, and introduced into L. lactis [44]. The original Npr signal peptide, which most probably contains 27 amino acids [96], served to direct the secretion of the enzyme into the culture medium. Like the above-mentioned signal sequences derived from L. lactis, the nprE signal sequence resembles the canonical signal sequences described by Von Hei- jne and Abrahmsen [90].

The signal sequence of the B. subtilis a-amylase gene has been used successfully to secrete the

88.

Cyamopsis tetragonoloba (Guar) a-galactosidase (Section 4.2.2; [61]), providing another example of the use of heterologous signal sequences in L. lactis.

7. CONCLUDING REMARKS

The general picture of gene expression signals in L. lactis emerging from the data now available very much resembles those observed in Es- cherichia colt and Bacillus subtilis. This feature allowed the expression of a number of L. lactis- derived genes in the latter bacterial species. Sev- eral studies have indicated, however, that in spite of the similarities, the expression signals from E. colt, B. subtilis and L. lactis are not equally efficient in these three organisms. As it remains obscure which features of the gene expression signals determine their efficiency in L. lactis, the most straightforward approach to heterologous gene expression in this bacterium makes use of L. lactis-specific gene expression signals. Since, in addition to the generally conserved expression signals (promoter and ribosome binding site), mRNA secondary structure and the nucleotide sequence of the (5 '-part of the) coding region per se may also be important, coupling the transla- tion of a heterologous gene to that of an effi- ciently expressed (3'-truncated) homologous gene may be the most universally applicable method to accomplish efficient expression of the heterolo- gous gene.

REFERENCES

ill McKay, L.L. and Baldwin, K.A. (1990) Applications for biotechnology: present and future improvements in lac- tic acid bacteria. FEMS Microbiol. Rev. 87, 3-14.

[2] Lindgren, S.E. and Dobrogosz, WJ. (1990) Antagonistic activities of lactic acid bacteria in food and feed fermen- tations. FEMS Microbiol. Rev. 87, 149-164,

[3] Olson, N.F. (1990) The impact of lactic acid bacteria on cheese flavor. FEMS Microbiol. Rev. 87, 131-148.

[4~ Gilliland. S.E. (1990) Health and nutritional benefits from lactic acid bacteria. FEMS Microbiol. Rev. 87, 175-188.

[5] Teuber, M. (1987) Genetisch ver~inderte Mikroorganis- men in Lebensmitteln: M6glichkeiten und Grenzen. Kieler Milehwirts. Forschungsb. 39, 265-274.

[6l Harley, C.B. and Reynolds, R.P. (1987) Analysis of E. coli promoter sequences. Nucleic Acids Res. 15, 2343- 2361.

[7] Moran. C.P., Lane, N.. LeGrice, S.F.J.. Lee, G., Stephens, M., Sonenshein, A.L., Pero, J. and Losick, R. (1982) Nucleolide sequences thai signal the initiation of transcription and translation in Bacillus subtilis. Mol. Gen. Genet. 186, 339-346.

[8] Jacques, N. and Dreyfus, M. (1990) Translation initia- tion in Escherichia colt: old and new questions. MoL Microbiol. 4, 1063-1067.

[9] Miller, O.L., Hamkalo, B.A. and Thomas, C.A. (1970) Visualization of bacterial genes in action. Science 169, 392-395.

[10] Shine. J. and Dalgarno, L. (1974) The 3'-terminal se- quence of Esdterichia colt 16S ribosomal RNA: comple- mentarity to nonsense triplets and ribosome binding sites. Proc. Natl. Acad. Sci. USA 71, 1342-1346.

[11] Kozak, M. (1983) Comparison of initiation of protein synthesis in procaryotes, eucaryotes, and organdies. Mi- crobiol. Rev. 47, 1-45,

[12] Hager, P.W. and Rabinowitz, J.C. (1985) Translational specificity in Bacillus subtilis. In: The Molecular Biology of the Bacilli (Dubnau, D.A., Ed.), Vol IL, pp. 1-31, Academic Press, Orlando, FL.

[13] Kondo, J.K. and McKay, L.L. (1982) Transformation of Streptococcus lactis protoplasts by plasmid DNA. Appl. Environ. Microbiol. 43, 1213-1215.

[14] Kondo, J.K. and McKay, L.L. (1984) Plasmid transfor- mation of Streptococcus lactis protoplasts: optimization and use in molecular cloning. Appl. Environ. Microbiol. 48, 252-259.

[15] Simon, D., Rouault, A. and Chopin, M.-(~. (1986) High- efficiency transformation of Streptococcus lactis proto- plasts by plasmid DNA. Appl. Environ. Microbiol. 52, 394-395.

[16] Van der Vossen, .I.M.B.M., Kok, J., Van der Lelie, D. and Venema, G. (1988) Liposome-enhanced trar.~for- marion of Streptococcus lactis and plasmid transfer by intergenerlc protoplast fusion of Streptococcus lactis and Bacillus subtilis. FEMS Microbiol. Lett. 49, 323-329.

[17] Von Wright, A., Taimisto, A.-M. and Sivela, S. (1985) Effect of Ca z+ ions on plasmid transformation of Strep- tococcus lactis protoplasis. Appl. Environ. Microbiol. 50, 110o-1102.

[18] Woskow, S.A. and Kondo, J.K (1987) Effect of proteo- lyric enzymes on transfection and transformation of Streptococcus lactis protoplasts. Appl. Environ. Micro- biol. 53, 2583-2587.

[19] Harlander, S.K. (1987) Transformation of Streptococcus lactis by electroporation. In: Streptococcal Genetics (Ferretti, J.J. and Curtiss, R. 111, Eds.), pp. 229-233. American Society for Microbiology, Washington, D.C.

[20] Holo, H. and Nes, l.F. (1989) High-frequency transfor- mation, by ¢lectroporation, of Lactococcus lactis ssp. cremoris grown with glycine in osmotically stabilized media. Appl. Environ. Microbiol. 5.~, 3119-3123.

[21] Powel, I.B., Achen, M.G., Hillier, A.J. and Davidson, B.E. (1988) A simple and rapid method for genetic transformation of lactic streptococci by electroporation. Appl. Environ. Mierobiol. 54, 655-660.

[22] Van der Lelie, D., Van der Vossen, J.M.B.M. and Venema, G. (1988) Effect of plasmid incompatibility on DNA transfer to Streptococcus cremoris. Appl. Environ. Microbiol. 54, 865-871.

[23] Gasson, M.J. (1983) Plasmid complements of Strepto- coccus lactis NCDO 712 and other lactic streptococci after protoplast-induced curing. J. Bacteriol. 154, 1-9.

[24] Chopin, A., Chopin, M.C., Moillo-Batt, A. and Lan- gella, P. (1984) Two plasmid-determined restriction sys- tems in Streptococcus lactis. Plasmid 11. 260-263.

[25] De Vos, W.M. (1986) Gene cloning in lactic strepto- cocci. Neth. Milk Dairy .I. 40, 141-154.

[26] Kok, .I., Van der Vossen, J.M.B.M. and Venema, G. (1984) Construction of plasmid cloning vectors for lactic streptococci which also replicate in Bacillus subtilus and Escherichia coll. Appl. Environ. Microbiol. 48, 726-731.

[27] Simon, D. and Chopin, A. (1988) Construction of a vector plasmid family and its use for molecular cloning in Streptococcus lactis. Biochimie 70, 559-566.

[28] Achen, M.G., Davidson, B.E. and Hillier, A.J. (1986) Construction of plasmid vectors for the detection of streptococcal promoters. Gene 45, 45-49.

[29] De Vos, W.M. (1987) Gene cloning and expression in lactic streptococci. FEMS Microbiol. Rev. 46, 281-295.

[30] Koivula, T., Sibakov, M. and Palva, I. (1991) Isolation and characterization of Lactococcus lactis ssp. lactis promoters. Appl. Environ. Micrublol. 57. 333-340.

[31] Simons, G., Buys, H., Hogers, R., Koenhen, E. and De Vos, W.M. (1990) Construction of a promoter-probe vector for lactic acid bacteria using the lacG gene of Lactococcus lactis. Dev. Ind. Microbiol. 31, 31-40.

[32] Van der Vossen, J.M.B.M., Kok, J. and Venema, G. (1985) Construction of cloning-, promoter screening- and terminator screening shuttle ,ectors for Bacillus subtilis and Streptococcus lactis. Appl. Environ. Micro- biol. 50, 540-542.

[33] P~rez Marffnez, G., Smith, H., Bron, S., Kok, .I. and Venema, G. (1990) Protein export functions isolated from Lactococcus lactis DNA. FEMS Microbiol. Rev. 87, P23.

[34] Sibakov, M., Koivula, T., Yon Wright, A. and Palva, l. (1991) Secretion of "i"EM-~8-1actamase with signal se- quences isolated from the chromosome of Lactococcus lactis ssp. lactis. Appl. Environ. Microbiol. 57, 341-348.

[35] Van de Guchte, M., Van der Vossen, J.M.B.M., Kok, J. and Venema, G. (1989) Construction of a lactococcal expression vector: expression of hen egg white lysozyme in Lactococcus lactis ssp. lactis. Appl. Environ. Micro- biol. 55, 224-228.

[36] Simons, G., Van Asseldonk, M., Rutten, G., Braks, A., Nijhuis, M., Homes, M. and De Vos, W.M. (1990) Analysis of secretion signals of lactococci. FEMS Micro- biol. Rev. 87, P24.

89

[37] Leenhouts, K.J., Kok, J. and Venema, G. (!.¢89) Camp. bell-like integration of heterulogous plasmid DNA into the chromosome of Lactococcus lactis ssp. lactis. Ap[,I. Environ. Microbiol. 55, 394-400.

[38] Van der Vossen, J.M.B.M., Van der Lelie, D. and Venema, G. (1987) Isolation and characterization of Streptococcus cremoris Wg2-specific promoters. Appl. Environ. Microbiol. 53. 2452-2457.

[39] Van der Vossen, J.M.B.M. (1988) Development and use of host-vector systems for the characterization of lacto- coccal expression signals. PhD thesis, University of Groningen, Oroningen.

[40] Kok, J., Leenhonts, KJ., Haandrikman A.J., Ledeboer A.M. and Veneraa, G. (1988) Nucleotide sequence of the cell wall proteinase gene of Streptococcus cremoris Wg2. Appl. Environ. Microbiol. 54, 231-238.

[41] Lakshmidevi, G., Davidson, B.E. and Hillier, A.J. (1990) Molecular characterization of the Lactococcus lactis ssp. ~remoris temperate bacteriophage BKS-T and identifi- cation of a phage gene implicated in the regulation of promoter activity. Appl. Environ. Microbiol. 56, 934- 942.

[42] Van Belkum, M.J., Hayema, B.J., Jeeninga, R.E., Kok, J. and Venema, G. (1991) Organization and nucleotide sequence of two lactococcal bacteriocin operuns. Appl. Environ Microbiol. 57, 492--498.

[43] Van Assekdonk, M., Rutten, G., Oteman, M., Siezen, R.J., De Vos, W.M. and Simons, G. (1990) Cloning of usp45, a gene encoding a secreted protein from Lact~- coccus lactis ssp. lactis MG1363. Gene 95, 155-160.

[44] Van de Guchte. M., Kodde, J., Van der Vossen, J.M.B.M., Kok, J. and Venema, G. (1990) Heterologous gene expression in Lactococcus lactis ssp. lactis: synthe- sis, secretion, and processing of the Bacillus subtilis neutral protease. Appl. Environ. Microbiol. 56, 2606- ~.t~.l 1.

[45] Ludwig, W., Seewaldt, E., Klipper-Balz, R., Schleifer, K.H., Magrum, L., Woese, C.R., Fox, G.E. and Staeke- brandt, E. (1985) The phyiogenetic position of Strepto- coccus and Enterococeus. J. Gen. Microbiol. 131, 543- 551.

[46] Gold, L. (1988) Posttranscriptional regulatory mecha- nisms in Escherichia coll. Ann. Rev. Biochem. 57, 199- 233.

[47] Dreyfus, M. (1988) What constitutes the signal for the initlation of proteiv, synthesis on Escherichia coil mRNAs? J. Mol. Biol. 204. 79-94.

[48] Tomich, C.S.C., OIson, E.R., OIsen, M.K., Kaytes, P.S., Rockenbach, S.K. and Hatzenbuhlcr, M.T. (1989) Effect of nucleotide sequences directly downstream from the AUG on the expre:.~sion of bovine somatotrupin in E. coll. Nucleic Acids [~:es. 17, 3179-3197.

[49] De Smit, M. and Van Duin, l. (1990) Control of prokaryotic translational initiation by mRNA secondary structure. In: Progress in Nucleic Acid Research and Molecular Biology (Cohn, W.E. and Moldave, K., Eds.), Vol. 3g, pp. 1-35. Academic Press, San Diego, CA.

90

[50] Porter, E.V. and Chassy, B.M. (|988) Nucleotide se- quence of the/]-D-phosphogalactoside galaetohydrolase gene of Lactobacillus cusei: comparison to analogous pbg genes of other Gram-positive organisms. Gene 62, 203-2"16.

[51] Sharp, P.M., Cowe, E., Higgins, D.G., Shields, D.C., Wolfe, K.H. and Wright, F. (1988) Codnn usage pat- terns in Escherichia coli, Bacillus sabtilis Saccharomyces cerevisiae, Schizosaccharomyces pombe, Drosophila melanogaster and Homo sapiens; a review of the consid- erable within-species diversity. Nucleic Acids Res. 16, 8207-8211.

[52] Ogasawa, N. (1985) Markedly unbiased codon usage in Bacillus subtilis. Gene 40, 145-150.

[53] Holt, J.G. (1986) Bergey's Manual of Systematic Bacte- riology, Vol. 2. Williams & Wilkins, Baltimore, MD.

[54] Gasson, M.J. and Davies, F.L. (1980) Conjugal transfer of the drug resistance plasmid pAM/3 in the lactic streptococci. FEMS Microbiol. Lett. 7, 51-53.

[55] Horinouchi, S. and Weisblum, B. (1982) Nucleotide sequence and functional map of pC194, a plasmid that specifies inducible chloramphenicol resistance. J. Bacte- riol. 150. 815-825.

[56] Horinouchi, S. and Weisblum, B. (1982) Nacleotide sequence and functional map of pE194, a plasmid that specifies inducible resistance to macrolide, lincosamide, and streptogramin type B antibiotics. J. Bacteriol. 150, 804-814.

[57] Brehm, J., Salmond, G. and Minion, N. (1987) Sequence of the adenine methylase gene of the Streptococcus faecalis plasmid pAMfll. Nucleic Acids Res. 15, 3177.

[58] De Vos, W.M. and Simons, G. (1988) Molecular cloning of lactose genes in dairy lactic streptococci: the vh,-s- pho-fl-galactosidase genes and their expression prod- ucts. Biochimie 70, 461-473.

[59] Van de Guchte, M., Kok, J. and Venema, G. (1991). Distance-dependent translational coupling and interfer- ence in Lactococcus lactis. Mol. Gen. Genet. 227, 65-71.

[60] Van de Guchte, M., Van der Lende, T., Kok, J. and Venema, G. (1991) A possible contribution of mRNA secondary structure to translation initiation efficiency in Lactococcus lactis. FEMS Microbiol. Lett. 81, 201-208.

[61] Haandrikman, A.J. (1990) Development and use' of a broad-host-range vector for the simultaneous analysis of divergent promoters. PhD Thesis, University of Gronin- gen, Groningen.

[62] Overbeeke, N., Termorshuizen, G.H.M., Giuseppin, LF., Underwood, D.R. and Verrips, C.T. (1990) Secre- tion of the a-galactosidase from Cyamopsis te- tragonoloba (Guar) by Bacillus subtilis. Appl. Environ. Microbiol. 56. 1429-1434.

[63] De Vos, W.M., Vos, P., Simons, G and David, S. (1989) Gene organization and expression in mesophilic lactic acid bacteria. J. Dairy Sci. 72, 3398-3405.

[64] Klessen, C., Schmidt, K.H., Ferretti, J.J. and Malke, H. (1988) Tripartite streptokinase gene fusion vectors for

gram-positive and gram-negative procaryotes. Mol. Gen. Genet. 212, 295-300.

[65] Reference omitted. [66] Reference omitted. [67] Joll~s, P. and Joll~s, J. (1984) What's new in lysozyme

research? Mol. Cell. Biochem. 63, 165-189. [68] Schleifer, K.H. and Kandler, O. (1972) Peptidoglycan

types of bacterial cell walls and their taxonomic implica- tions. Bacteriol. Rev. 36, 407-477.

[69] Crawford, R.J.M. (1987) The use of lysozyme in the prevention of late blowing in cheese, Bulletin of the International Dairy Federation no. 216. International Dairy Federation, Brussels.

[70] Hughey, V.L. and Johnson, E.A. (1987) Antimicrobial activity of lysozyme against bacteria involved in food spoilage and food-borne disease. Appl. Environ. Micro- biol. 53, 2165-2170.

[7t] Hughey, V.L., Wilger, P.A. and Johnson, E.A. (1989) Antibacterial activity of hen egg white lysozyme against Lysteria monocytogenes Scott A in foods. Appl. Environ. Microbiol. 55, 631-638.

[72] Gasson, M.J. and Anderson, P.H. (1985) High copy number plasmid vectors for use in lactic streptococci. FEMS Microbiol. Lett. 30, 193-196.

[73] Leenhouts, KJ., Gietema, J., Kok. J. and Venema, G. (1991) Chromosomal stabilization of the proteinase genes in Lactococcus lactis. Appl. Environ. Microbiol. 57, 2568-2575.

[74] Ganoza, M.C., Kofoid, E.C., Marlire, P. and Louis, B.G. (1987) Potential secondary structure at translation-ini- tiation sites. Nucleic Acids Res. 15, 345-360.

[75] Sprengart, M.L., Fatscher, H.P. and Fuchs, E. (1990) The initiation of translation in E. coli: fipparent base pairing between the 16SrRNA "and downstream se- quences of the mRNA. Nucleic Acids Res. 18, 1719- 1723.

[76] Oppenheim, D.S. and Yanofski, C. (1980) Translational coupling during expression of the tryptophan operon of Eseherichia coli. Genetics 95, 785-795.

[77] Das, A. and Yanofsky, C. (1984) A ribosome binding site sequence is necessary for efficient expression of the distal gene of a translationally-coupled gene pair. Nu- cleic Acids Res. 12, 4757-4768.

[78] Gatenby, A.A., Rothstein, S.J. and Nomura, M. (1989) Translational coupling of the maize chloroplast atpB and atpE genes. Proc. Natl. Acad. Sci. USA 86, 4066- 4070.

[79] Harms, E., Higgins, E., Cheu, J.W. and Umbarger, H.E. (1988) Translational coupling between the ilvD and ilvA genes of Escherichia coli. J. Bacteriol. 170, 4798-4807.

[80] Lindahl, L., Archer, R.H., McCormick, J.R., Freedman, L.P. and Zengel, J.M. (1989) Translational coupling of the two proximal genes in the SI0 ribosomal protein operon of Escherichia coli. J. Bacteriol. 171, 2639-2645.

[81] Little, S., Hyde, S., Campbell, C.J., Lilley, R.J. and Robinson, M.K. (1989) T~anslational coupling in the

91

threonine operon of Escherichia coil K-12. J. Bacteriol. 171, 3518-3522.

[82] Schoner, B.E., Hsiung, H.M., Belagaje, R.M., Mayne, N.G. and Schoner, R.G. (1984) Role of mRNA transla- tional efficiency in bovine growth hormone expression in Escherichia coil Proc. Natl. Acad. Sci. USA 81, 5403-5407.

[83] Schiimperli, D., McKenney, K., Sobieski, D.A. and Rosenberg. M. (1982) Translational coupling at an in- tercistronic boundary of the Escherichia coil galactose operon. Cell 30, 865-871.

[84] Bonekamp, F., Dalbge, H., Christensen, T. and Jensen, K.F. (1989) Translation rates of individual codons are not correlated with tRNA abundances or with frequen- cies of utilization in Escherichia coll. J. Bacteriol. 171. 5812-5816.

[85] Grosjean, H. and Fiers, W. (1982) Preferential codon usage in prokaryotic genes: the optimal codon-anti- codon interaction energy and the selective codon usage in efficiently expressed genes. Gene 18, 199-209.

[86] Ohno, S. (1988) Codon preference is but an illusion created by the construction principle of coding se- quences. Proc. Natl. Acad. Sci. USA 85, 4378-4382.

[87] Gualerzi, C.O. and Pon, C.L. (1990) Initiation of mRNA translation in prokaryotes. Biochemistry 29, 5881-5889.

[88] Brawerman, G. (1987) Determinants of messenger RNA stability. Cell 48, 5-6.

[89] Kubo, M., Higo, Y. and Imanaka, T. (1990) Biological threshold values of procaryotic gene expression which is controlled by the DNA inverted repeat sequence and the mRNA secondary structure. J. Ferment. Bioeng. 69, 305-307.

[90] Von Heijne, G. and Abrahmsen, L. (1989) Species- specific variation in signal peptide design; implications for protein secretion in foreign hosts. FEBS Left. 244, 439-446.

[91] Ard6, Y. and Pettersson, H.E. (1988) Accelerated cheese ripening with heat treated cells of Lactobacillus helueti- cus and a commercial proteolytic enzyme. J. Dairy Res. 55, 239-245.

[92] Law, B.A. (1984) The accelerated ripening of cheese. In: Advances in the Microbiology and Biochemistry of Cheese and Fermented Milk (Davies, F.L. and Law, B.A., Eds.), pp. 209-228. Elsevier, London.

[93] Law, B.A. and King, J.S. (1985) Use of liposomes for proteinase addition to Cheddar cheese. J. Dairy Res. 52, 183-188.

[94] Law, B.A. and Wigmore, A. (1982) Accelerated cheese ripening with food grade proteinases. J. Dairy Res. 49, 137-146.

[95] Law, B.A. and Wigmore, A.S. (1983) Accelerated ripen- ing of Cheddar cheese with a commercial proteinase and intracellular enzymes from starter streptococci. J. Dairy Res. 50, 519-525.

[96] Yang, M.Y., Ferrari, E. and Henner, DJ. (1984) Cloning of the neutral protease gene of Bacillus subtilis and the

use of the cloned gene to create an in vitro-derived deletion mutation. J. BacterioL 160. 15-21.

[97[ Haandriknlan, A.J., Kok, J., Laan, H., Soomitro, S., Ledeboer, A.M.o Konin~, W.N. and Venema, G. (1989) Identification of a gene required for maturation of an extracellular lactococcal scrine proteinase. J. BacterioL 171, 2789-2794.

[98] Buchman, G.W., Banerjce, S. and Hansen, J.N. (1988) Structure, expression, and evolution of a gene encoding the precursor of nisin, a small protein antibiotic. J. Biol. Chem. 263, 16260-16266.

[99] van Rooijen, RJ. and De Vos, W.M. (1990) Molecular cloning, transcriptional analysis, and nucl©otide se- quence of lacR, a gene encoding the repressor of the lactose phosphotransferasesystem of Lactococcus lactis. J. Biol. Chem. 265. 18499-18503.

[100] Tinoco, 1., Borer, P.N., Dengler, B., Levine, M.D., Uh- lenbeck, O.C., Crothers, D.M. and Gralla, J. (1973) Improved estimation of secondary structure in ribonu- cleic acids. Nature New Biol. 246. 40-41.

[101] Renault, P., Gaillardin, C. and Heslot, H. (1989) Prod- uct of the Lactococcus lact~s gene required for malolac- tic fermentation is homologous to a family of positive regulators. J. Bacteriol. 171, 3108-3114.

[102] Ross, P., O'Gara, F. and Condon, S. (1990) Cloning and characterization of the thymidylate synthase gene from Lactococcus lactis ssp. lactis. Appl. Environ. Microbiol. 56, 2156-2163.

[103] Mayo, B., Kok, J., Venema, K., Bockelmann, W., Teu- ber, M.. Reinke, H. and Venema, G. (1991) Molecular cloning and sequence analysis of the x-prolyl dipeptidyl aminopeptidase gene from Lactococcus lactis ssp. cre- moris. Appl. Environ. Microbiol. 57, 38-44.

[104] Nardi, M., Chopin, M., Chopin, A., Cals, M. and Gripon, J. (1991) Cloning and DNA sequence analysis of an x-prolyl dipcptidyl aminopeptidase gene from Lactococ- cus lactis ssp. lactis NCDO763. Appl. Environ. Micro- biol. 57, 45-50.

[105] Boizet, B., Villeval, D., Slos, P., Novel, M., Novel, G. and Mercenier, A. (1988) Isolation and structural analy- sis of the phospho-/3-galactosidase gene from Strepto- coccus laetis Z268. Gene 62, 249-261.

[106] De Vos, W.M. and Gasson, M.J. (1989) Structure and expression of the Lactococcus lactis gene for phospho- /3-galactosidase (laeG) in Eschericha coil and L. lactis. J. Gen. Microbiol. 135, 1833-1846.

[107] Hill, C., Miller, L.A. and Klaenhammer, T.R. (1990) Nucleotide sequence and distribution of the pTR2030 resistance determinant (/up) which aborts bacterio- phage infection in lactococci. Appl. Environ. MicrobioL 56, 2255-2258.

[108] Leenhouts, KJ., Tolner, B., Bron, S., Kok, J., Venema, G. and Seegers, J.F.M.L (i991) Nucleotide sequence and characterization of the broad-bost-range lactococcal plasmid pWV01. Plasmid 26, 55-66.

[109] Shearman, C., Underwood, H., Jury, K. and Gasson, M.

92

(1989) Cloning and DNA sequence analysis of a Lacto. coccus bacteriophage lysin gene. Mol. Gen. Genet. 218, 214-221.

[110] Polzin, K.M. and Shimizu-Kadota: M. (1987) Identifica- tion of a new insertion element, similar to gram-nega- tive IS26, on the lactose plasmid of Streptococcus lactis ML3. J. Bacteriol. 169, 5481-5488.

[111] Romero, D.A. and Klaenhammer. T.R. (1990) Charac- terization of insertion sequence IS946, an iso-ISSl ele- ment, isolated from the conjugative lactococcal plasmid p 1R2030. J. Bacteriol. 172, 4151-4160.

[112] Haandrikman, A.J., Van I.xeuwen, C., Kok, J., Vos, P., De Vos, W. and Venema, G. (1990) Insertion elements on lactococeal proteinase plasra~.~. Appl. Environ. Mi- crobiol. 56, 1890-1896.

[113] Dodd, H.M., Horn, N. and Gasson, M.J. (1990) Analysis of the genetic determinant for production of the peptide antibiotic nisin. J. Gen. Microbiol. 1:36, 555-566.

[114] Rauch, P.J,G., Bccrthuyzen, M.M. and De Vos, W.M. (1990) Nucleotide sequence of I8904 from Lactococcus lactis ssp. lactis strain NIZO R5. Nucleic Acids Res. 18, 4253-4254.

[115] Kaletta, C. and Entian, K. (1989) Nisin, a peptide an- tibiotic: cloning and sequencing of the his,4 gen¢ and posttranslational processing of its peptide product. J. Bacteriol. 171, 1597-1601.

[116] David, S., Van der Rest, M.E., Driessen, A.J.M., Si- mons, (3. and De Vos, W.M. (1990) Nucteotide se- quence and expression in Escherichia coli of the Lacto- coccus lactis citrate permease gene. J. Bacteriol. 172, 5789-5794.

[117] Piggot, P.J. and Hoch, J.A. (1985) Revised genetic link- age map of Bacillus subtilis. MicrobioL Rev. 49, 158-179.