THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry end Molecular Biology, Inc.

Vol. 263, No. 31, Issue of November 5, pp. 16260-16266, 1988 Printed in U.S.A.

Structure, Expression, and Evolution of a Gene Encoding the Precursor of Nisin, a Small Protein Antibiotic*

(Received for publication, June 22, 1988)

George W. Buchman, Sharmila Banerjee, and J. Norman Hansen From the Department of Chemistry and Biochemistry, University of Maryland, College Park, Maryland 20742

We have cloned and sequenced a gene (spaN) from Streptococcus lactis ATCC 11454 which encodes the peptide precursor of the small protein antibiotic nisin. The encoded precursor is 57 amino acids long, with a 23-residue leader region and a 34-residue structural region. The structural region contains serines, threo- nines, and cysteines at exactly the positions required to give mature nisin by a series of post-translational modifications involving dehydration of serines and threonines to dehydro forms, and cross-linking with cysteine residues. S1 mapping revealed a 267-nucleo- tide transcript of the nisin gene that is expressed dur- ing vegetative growth and stationary phase. It has a half-life of 7-10 min. The absence of an identifiable promoter or rho-independent terminator and the de- tection of two different 5‘-ends of the transcript sug- gested it is a processing product from a larger RNA. This may represent a polycistronic mRNA which may also encode proteins involved in processing the nisin precursor peptide. Open reading frames were found in regions flanking the nisin gene. The one downstream had a ribosome binding site and appeared to be tran- scribed by read-through from the nisin gene. The one upstream had significant homology to a putative trans- posase from the Escherichia coli IS2 insertion element. Comparison of gene sequence homologies between nisin and the other lanthionine antibiotics, subtilin and epidermin, indicated that they all evolved from a com- mon ancestor. Corresponding leader peptide sequences showed mediocre amino acid homology, but nearly per- fect hydropathic homologies, suggesting a common function. It is proposed that this function includes rec- ognition signals or other information required for post- translational processing.

Understanding the relationship between sequence and properties of proteins is crucial for the rational design of new proteins with novel structures and functions. Study of this relationship in small proteins can simplify the interpretation of experimental results. We have been studying subtilin, produced by Bacillus subtilis ATCC 6633, and nisin, produced by Streptococcus h t i s ATCC 11454. They are small protein antibiotics, with 32 and 34 amino acids, respectively. We recently cloned and sequenced the subtilin gene to confirm that it is gene-encoded. It has been uncertain, because subtilin

*This project was supported by National Institutes of Health Grant R01-AI24454 (to J. N. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accord- ance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 504057.

contains a plethora of unusual amino acids, including dehy- droalanine and lanthionine. The gene sequence confirmed an old hypothetical scheme of post-translational modification in which precursor serines and threonines undergo dehydration to double-bond dehydro forms which then react with neigh- boring cysteines to give lanthionine, which contains a thioether group (1). This scheme is depicted in Fig. 1. Shortly after we reported this sequence (2), Schnell et al. (3) reported the gene for epidermin, which is also a lanthionine-containing peptide antibiotic. In this paper, we report the sequence of the nisin gene and its expression, and characterize the tran- script.

Subtilin, nisin, and epidermin are representative of a class of antibiotic that is characterized by the presence of lanthion- ine, and have been termed lantibiotics (3). It appears to be a class with many members, because several others have been reported (reviewed in Ref. 3). The structures of subtilin, nisin, and epidermin (Fig. 2) suggest a common ancestor, and the homologies among their genes that we present here confirm this.

The nisin gene is important for several reasons. It was the first lanthionine antibiotic to be discovered (4), the first to have its structure determined (5) , and has been the most studied (reviewed in Ref. 6). Its total chemical synthesis was recently achieved (7). Nisin is of significant contemporary use, being important in dairy fermentations and as a food preservative (6). The availability of the nisin gene allows us to better discern conserved and diverged features at the nu- cleic acid level, and allows us to make more confident conclu- sions about evolutionary relationships among these antibiot- ics. The results we present increase our expectation that the lanthionine antibiotics can serve as a model for studying structure-function relationships in small proteins. The fact that they contain unusual amino acids has an added benefit of providing an opportunity to study ways of side-stepping the limitations imposed by the genetic code to allow the introduction of unusual amino acids into gene-encoded pro- teins. This would provide new scope for the functional prop- erties that genetic engineers could incorporate into their designs.

MATERIALS AND METHODS

Bacterial Strains, Cloning Vectors, and Culture Conditwns-Nisin- producing S. &tis ATCC 11454 was obtained from the American Type Culture Collection (Rockville, MD). Strains were stored at -20 “C in ATCC Medium 17 (100 g of skim milk powder, 100 g of tomato juice, 5 g of yeast extract to pH 7.0) containing 25% glycerol. Working stocks were maintained on 1.2% LB agar plates (10 g of Bacto-tryptone, 5 g of Bacto-yeast extract, 10 g of NaCl/liter). M17 culture medium (S), consisting of 5 g of Bacto-peptone (Difco), 5 g of Bacto-soytone (Difco), 2.5 g of yeast extract (Difco), 5 g of beef extract (Difco), 0.5 g of ascorbic acid, 5 g of lactose (or glucose), 19 g of @-disodium glycerophosphate (Eastman), and 0.12 g of anhydrous MgSOJiter, was used to culture S. lactis for nisin production, ge-

16260

Small Protein Antibiotic Nisin Gene 16261 I , Precursor peptide contains serines, threonines, and cysteines

peptide

IL' C-CHPH

2 . Dehydration of serines and threonines

FIG. 1. Scheme for formation of unusual amino acids. A hypothetical series of reactions for formation of the dehydro residues and the thioether cross-linkages in nisin is shown. Serine and threo- nine are dehydrated to dehydroalanine (DHA) and dehydrobutyrine (DHB), respectively. Nucleophilic addition of the sulfhydryl of cys- teine to the dehydro residues proceeds with stereo-inversion. Lan- thionine and methyl lanthionine result from the formation of thioether cross-linkages. Scheme is based on that proposed by Ingram (1).

15 -Lw

MSRJ ,+O %,

B z h

EPDWMN 15

Ala-LWGly-ya' \Arw

10 2-Am-Al0-E 22 Lya -Ala-Ala-le-)\HZ 's/ II M

s 21

1

1

FIG. 2. Primary structures of nisin, subtilin, and epidermin. Nisin and subtilin structures were determined by Gross and co- workers (5), and epidermin by Allgaier et al. (23). Unusual amino acids are as defined in the legend to Fig. 1. ABA, aminobutryic acid.

nomic library construction, and total RNA isolation. The organism was grown at 32 "C without aeration using a 2% inoculum into an appropriate volume of M17 medium.

Bacillus cereus T spores used in the assay for nisin production were prepared and stored as previously described (9). Antibiotic activity assays were performed as previously described using fractions of the S. &tis culture supernatant (10, 11).

DNA Isolation Procedure-S. &tis ATCC 11454 was incubated in 500 ml of M17 medium for 30 h at 32 "C without aeration. Cells were

buffered saline (8 g of NaCl, 1.4 g of Na2HP04, 1.2 ml of 1 N HCl/ collected by centrifugation, and washed in 25 ml of phosphate-

liter). The cells were resuspended in 15 ml of 50 mM Tris-HC1 (pH 7.6) and subsequently digested with 33 pg/ml mutanolysin (Sigma) for 15 min at 37 "C with gentle agitation (12). Then 5 ml of STEP solution (13) (0.5% sodium dodecyl sulfate, 50 mM Tris-HC1 in 0.4 M

EDTA, and 1 mg/ml proteinase K) was added and incubation per- formed at 37 "C for 30 min with occasional mixing. The mixture was extracted with 1 volume of CHC13, 1 volume of 5050 phenokCHC4, and finally with 1 volume of CHCl3. One-tenth volume 3 M sodium acetate and 2 volumes of ethanol were added; the DNA was spooled and resuspended in 20 ml of 50 mM Tris-HC1 and 4 mM EDTA containing 50 pg/ml pancreatic RNase (Sigma). The solution was dialyzed against a buffer of 50 mM Tris-HC1 and 4 mM EDTA for 16 h at 4 "C with one buffer change. The DNA was ethanol-precipitated two times in the presence of 2.5 M ammonium acetate and finally dissolved in 2 ml of 10 mM Tris-HC1 (pH 7.6).

Probe Construction, Radiolabeling, and Hybridization Procedures- Several different probes were used to search for the nisin gene in S. lactis ATCC 11454 DNA. Hybridization conditions were optimized as previously described (2). Two oligomeric probes were prepared by chemical synthesis using a Biosearch model 8700 DNA synthesizer. One was a 20-mer mixed probe designed against a region of low codon degeneracy within the putative nisin precursor sequence. The second was a single-sequence 103-mer oligonucleotide probe designed using the strategy of Lathe (14). A natural DNA probe was also employed, which was a 1.l-kb' restriction fragment containing the subtilin gene that had previously been cloned from B. subtilis ATCC 6633 (2).

Library Construction ana' Isolation of the Nisin Gene-A total genomic library of S. lactis ATCC 11454 DNA in XJ1 was constructed and screened as previously described (2). Positive clones were mapped by restriction analysis and subcloned into pUC9 and pTZ19U plasmid vectors for further analysis, and into M13mpl8 and M13mp18 for sequencing. Sequence determination was performed by the dideoxy termination method using modified T7 polymerase and the protocol in a Sequenase kit obtained from the United States Biochemical Company.

RNA Isolation and Northern Blot Analysis-Total RNA isolation was performed according to the method of Ulmanen et al. (15). RNA fractionation was performed on a denaturing acrylamide gel, electro- blotted onto Zeta-Probe (Bio-Rad) nylon membrane, and hybridized as described previously (2).

Protein Analysis-Proteins were analyzed by electrophoresis on the polyacrylamide gel system of Swank and Munkres (16), and silver- stained using Bio-Rad reagents. Nisin activity was determined by the method of Morris et al. (17).

SI Mapping and Messenger Half-life Measurements-S1 mapping was performed by the method of Davis et al. (18), using the strategy previously described for subtilin transcripts (2). The half-life of the nisin mRNA was measured using the rifampcin procedure of Bechofer and Dubnau (19). The level of rifampicin used was 0.20 mg/ml.

RESULTS

Characterization of Probes Directed toward the Nisin Gene- The synthetic 103-mer, synthetic mixed 2O-mer, and the natural subtilin gene were hybridized to genomic blots of restricted genomic DNA from S. lactis ATCC 11454. Both the short mixed probe and the 103-mer hybridized to a 9.5-kb genomic EcoRI restriction fragment. The 103-mer gave a stronger signal and the hybrid melted at a higher temperature than the mixed probe. The 103-mer gave the best signals when hybridized at 56 "C in 6 X SSC and washed at 51 "C in 2 x SSC. The best signals for the mixed probe were obtained with a 51 "C hybridization and a 48 "C wash. We originally expected that the subtilin gene would be a good probe for the nisin gene, but it was not; and indicated that the nisin and subtilin genes are highly diverged (cf. Fig. 7).

Identification, Restriction Analysis, and Sequencing of a Clone Which Contains the Nisin Gene-A total genomic li- brary of s. &tis 11454 constructed in bacteriophage XJ1 was screened with the 20-mer mixed oligonucleotide probe and the 103-mer probe using the optimized hybridization and wash conditions determined above. Clones that hybridized to both probes were selected for further study. Only two such clones were obtained from a screen of more than 10,000 recombi- nants, and both contained a truncated EcoRI fragment of 5

'The abbreviations used are: kb, kilobase; ORF, open reading frame.

kb. We interpreted this to mean that the complete 9.5-kb EcoRI genomic fragment was difficult to obtain, and that the 5-kb fragment was the result of having been truncated by Mbof during the original restriction enzyme sizing of the genomic DNA. The problem DNA had presumably been re- moved from this truncated fragment. The possibility exists that the problem DNA is somehow related to the nisin gene locus. One of these clones was subjected to further restriction analysis (Fig. 3). This clone contained a 9.5-kb PstI fragment,

AGGATAGTATTTTCTTAGTTCACkCATGGhTACTlfCCZIATAACTTA~~AGCC 692 3' end of nisin mRNA-->II<---inverted---------

I<--downstream ORF---- 6 1 2

CTCAACCGTTTTTAGTAACAAATACAATTTTATCTCCAAACGATAAACCGAGTTTTACTC 624 636 648 660

l s G l n P r o P h e L e u V a l A r g A a n T h r I l e L e u S e r P r o 4 ~ p L ~ s A r ~ S e r P ~ ~ T h r C

AlTATACTCAAGTCATTGACACTCTAAGTAAAAATAA~~TTTTTTTCCAACAGTTACTAC 672 684 696 708 720

luT~rThrGlnValIieCltThrVslSerLysA~nLysV~lPheLe~Gl~ClnLe~LeuL

TAGCTAATCCTAAACTCTATGATCTTATCCACAAATATAATGCTCGT---~~nt.-~--~ 732 741 756 768 780

c u A l a A s n P r o L ~ s L e u T ~ r A ~ p ~ a l H E T C l n L y 8 T ~ ~ A ~ ~ A l ~ C I ~ - - - c ~ ~ t . - - - - ~

k

lh 5 10 15

FIG. 3. Sequence of the nisin gene and its flanking regions. The sequence obtained for the gene of the precursor of the small protein antibiotic nisin (spaN), is shown translated. The leader region is overscored with asterisk, and the structural region is numbered as in Fig. 2. Amino acids which undergo m ~ i ~ c a t i o n according to the scheme in Fig. 1 are in botd ktters and correspond to the locations of the unusual amino acids in mature nisin (Fig. 2) . The 5'- and 3'- termini of the nisin transcript were determined by S1 mapping (Fig. 5). The inverted repeat downstream from the nisin sequence is shown folded in Fig. 5. The ribosome binding sites (r.b.s.1 were identified by their proximity to the methionine codons, together with their consen- sus to known B. subti~is Shine-Dalgarno sequences (24). The down- stream ORF is discussed in the text. An ORF which lies about 750 base pairs upstream from the nisin gene is not included in this figure, but is shown translated in Fig. 7. The 103-mer probe sequence is shown above the gene sequence with homologies in dashes. The region corresponding to the 20-mer mixed probe is shown bridging across amino acids 17-23 in the structural gene sequence. The restriction map is of the S. lactis genomic fragment cloned into XJ1. It shows major restriction sites and the location of the nisin gene sequence. The EcoRI site a t position 0 kb on the restriction map is an artificial site created by the cloning process. The 5-kh EcoRI fragment is a truncated form of a 9.5-kb genomic fragment, and the site of tm- cation is upstream from the nisin structural gene.

20

a 4-kb Hind111 fragment, and a 1.7-kb Sau3A fragment, and fragments corresponding to these were also present in the genome, indicating that the truncated fragment had not rear- ranged. The Sau3A fragment was subcloned and partially sequenced. The sequence shown in Fig. 3 contained an open reading frame (ORF) in which all of the unusual amino acids in mature nisin had appropriate serines, threonines, and cysteines in exactly the positions required to give mature nisin by the scheme in Fig. 1.

Expression of the Nisin Gene and Characterization of the ~ ra~c r ip t -The ultimate result of expression of the nisin gene is the appearance of the antibiotic activity associated with the mature nisin peptide. A growth curve of S. &tis ATCC 11454 is shown in Fig. 4. Samples were removed at different stages of growth, and the cells were pelleted. The supernatant fluids were assayed for nisin activity, and elec- trophoresed on sodium dodecyl sulfate-polyacrylamide gels, The gels were stained for protein to reveal any nisin present. Total RNA was isolated from the pelleted cells, which was electrophoresed and electroblotted for Northern analysis in which a labeled 20-mer oligonucleotide which was homologous to the nisin structural gene was used as a probe. These results are shown in Fig. 4. The growth curve shows that stationary phase is reached at about 6 h. A peptide band with a sue corresponding to nisin could be seen at 5 h, whereupon antibiotic activity could be detected as well. Because of the requirement for processing of the precursor, the nisin gene would have to be transcribed before this. The Northern analy- sis reveals a hybridizing band with an apparent size of about 265 nucleotides that is present at all the times tested. The nisin gene is therefore expressed during vegetative growth as well as stationary phase. In addition, there is another band just below the wells, possibly indicating the presence of a larger species, but it was not possible to estimate its size.

S1 Mapping of the Nisin mRNA-The 5'-end of the tran- script was mapped as the SI-resistant hybrid formed with a radioactive cDNA fragment extending from within the nisin gene to the flanking Sau3A site, about 1000 nucleotides up- stream from the nisin gene. The RNA used was isolated from mid-log cells, and so should contain the transcripts revealed in Fig. 4. The results in Fig. 5 show that the nisin transcript is "ragged" at its 5'-end, terminating at two slightly different locations separated by about 3-5 nucleotides. The locations of these ends are shown in the DNA sequence in Fig. 3. Inspection of the sequence upstream from this terminus does not reveal an obvious promoter, although this cannot be ruled out without appropriate experimental tests. If the transcript is a processing product, one would not expect to find a promoter. Several larger bands are present among the S1 products. The size of the largest band corresponds to the distance from the nisin gene to the end of the Sau3A insert, or about 1000 nucleotides. These bands may reflect the pres- ence of a polycistronic mRNA, but additional experiments will be required to confirm whether any is a precursor to the nisin transcript.

The 3'-end of the t ransc~pt was determined by hybridizing the transcript to a radioactive cDNA of the entire Sau3A insert, treating with SI, and measuringthe lengths of resistant hybrids using a sequencing ladder as a size standard. The major transcript species has a length of about 267 nucleotides (Fig. 5). Measuring down this distance from the 5'-end deter- mined above places the 3'-end of the transcript in the region of an inverted repeat (Fig. 5; cf. Fig. 3). This inverted repeat does not have the stretch of Ts that is characteristic of a rho- independent terminator (24), so the 3"terminus is either the result of termination by a rho-dependent process, or process-

Small Protein Antibiotic Nisin Gene

A 16263

B FIG. 4. Northern analysis of RNA

isolated from different growth stages of S. factis ATCC 11454. Panel A shows a growth curve in medium M17, monitored by a Klett-Summerson colorimeter using Klett units. Panel B shows total protein in the supernatant fluids from samples removed at different times as revealed by silver staining. The lane labeled std consisted of a mixture of proteins of known molecular weight. A band corresponding to the molecular weight of nisin was present at all times from 5 to 24 h. Nisin antibiotic activity was observed whenever this band was intense enough to be seen, and was pre- sent in the 5-24-h samples. Panel C shows the total RNA populations as re- vealed by ethidium bromide staining. Panel D shows a Northern blot of the gel in Panel C. The blot was hydridized to a 5’-end-labeled 20-mer probe that was homologous to the nisin structural gene.

2 6 10 Hous

C

LJ

3 4 5 6 8 1 2 f n ‘I, s - 23s - 5 s

stained for RNA

ing of a larger RNA transcript. We are interested in the possibility that such a transcript may encode proteins in- volved in processing of the nisin precursor. The S1 mapping results also show other minor transcript lengths, including several that are apparently shorter degradation products of the nisin transcript.

The Nisin Transcript Is a Moderately Stable RNA Species- We observed earlier that the subtilin transcript synthesized by B. subtilis ATCC 6633 has a half-life of about 45 min (Z), which is remarkably long for a prokaryote. We therefore determined whether the nisin transcript is similarly stable. Northern analysis of RNA isolated from rifampicin-inhibited cells revealed that the half-life of the nisin transcript is about 7-10 min (Fig. 6). Although this is somewhat longer than the 2-4-min half-life that is characteristic of prokaryotes (19), it is dramatically less than that of the subtilin mRNA.

Open Reading Frames Flanking the Nisin Gene-An ORF was found immediately below the inverted repeat of the nisin gene, and is shown translated in Fig. 3. A typical consensus ribosome binding site is present at a location that is appro- priate to initiate at a methionine codon. The protein data bases (BIONET) were searched for homologies to this se- quence, but none was found. Considering the fact that the inverted repeat near this ORF may be a processing site rather than a terminator, it appears likely that this ORF is tran- scribed by read-through from the nisin gene sequence, and is a candidate to be a component of a polycistronic mRNA. An ORF was also discovered upstream from the nisin structural gene, and is shown in Fig. 7. When the protein data base was searched for homologies to this ORF, a good match was found to the putative transposase from the Escherichia coli IS2 insertion element (20). The significance of this is uncertain, but it suggests the possibility that the nisin gene is, or was at

d * ? e 6 5 4 3 z

- nisin

stained for protein

D

”

autoradiographed

4 - 265

some time, associated with an insertion element or a transpo- son. This in turn suggests a mechanism for transmission of this antibiotic among the various bacterial strains in which it is found.

Sequence Homologies between the Genes and the Peptide Precursors of Nisin, Subtilin, and Epidermin-Nisin, subtilin, and epidermin are produced by S. lactis ATCC 11454, B. subtilis ATCC 6633, and Staphlococcus epidermis Tu3298, respectively. These are all Gram-positive eubacteria that have evolved to fit very different ecological niches, but the similar- ities between the structures of these antibodies (Fig. 2) and their unusual processing requirements suggest a common ancestor. Fig. 7 compares the homologies among the genes and the encoded peptides of their respective leader peptide regions, as well as among the mature structural regions. The homologies of sequence and organization support the idea of a common ancestor, but the differences in both amino acid and nucleic acid sequences indicate that they have been evolving separately for a long time. Indeed, inspection of the silent codon positions suggests that they have become com- pletely randomized (calculations not presented). This diver- gence accounts for our observation that the subtilin gene was a poor hybridization probe for the nisin gene, and indicates that the remaining homologies are a consequence of conser- vation of structure by the selective pressure of conservation of function.

Hydropathic Homologies of the Precursor Sequences-Since random processing of cytosolic proteins would probably be lethal, the processing system must be specific for the antibi- otic precursor peptide. This could be the result of substrate recognition, or the processing system could be compartment- alized, and only those substrates that are directed to the processing compartment would undergo modification. Be-

16264 Small Protein Antibiotic Nisin Gene

A T T

5’ 3‘ - full length 1 4

0

.-

minutes after rifampicin addition

o v A 9 8 0 N o m 9 I S a

A A

T3

. ? m

T G T G G

G-C A-T T-A T-A T-A A-T T-A

T G G-C A-T A-T T-A A -T T-A T-A

T ritl A T C C T-A A-T AAGAGAGG

C I I I I I A T A G G

T A C

FIG. 5. S1 mappig of 6‘- and 3’-ends of the nisin mRNA. Total RNA was isolated from S. lactis ATCC 11454 cells growing exponentially in M17 medium. The 5 ’ - and 3’-ends of transcripts that contain sequences encoded in the nisin gene were mapped using the strategy described in the text. The 5’ lane shows the S1 protection pattern from which the location of the 5’-end of the transcript was determined. There are two major cutting sites that are in close proximity to the nisin gene sequence, and the sequence around these sites is shown. They are also shown on the nisin gene sequence in Fig. 3. There are also larger fragments. The position marked as full- length corresponds to a protected fragment that would extend to the end of the cDNA, or a size of about 1000 nucleotides. The 3’ lane shows the S1 protection pattern from which the location of the 3’- end was determined. The location of the region containing the cutting site which would produce the major 267-nucleotide fragment is shown on the sequence of the inverted repeat and in Fig. 3.

- 267

FIG. 6. Determination of the half-life of nisin mRNA. The stability of nisin mRNA was determined after addition of rifampicin to inhibit RNA synthesis and fractionation on a formaldehyde- agarose gel (18) followed by Northern analysis of remaining nisin mRNA using a 5”end-labeled 20-mer probe of the nisin structural gene. Panel A shows an experiment with time points ranging from 0 to 120 min after addition of rifampicin. Panel B shows a similar experiment with time points taken a t shorter intervals.

cause of sequence variability around the processed amino acids, we think the latter is more likely. We noted previously that the hydropathic profile of the subtilin leader peptide was unusual for an exported protein, and that it could play a role in processing (2). Fig. 8 shows the hydropathic profiles of all three antibiotics. Except for the N-terminal hexapeptide that occurs only in epidermin, the profiles of the leader regions are virtually identical. In view of the comparatively large number of actual sequence differences, it is clear that hydro- pathic character underlies the amino acid functions that are actually being selected for. The hydropathic character within the structural region is much less conserved. We speculate that it is this conserved hydropathic character of the leader region that destines the precursor to be processed.

DISCUSSION

Typical antibiotics are organic molecules synthesized by biosynthetic pathways, and require laborious organic chem- istry to create structural variants. Gene-encoded peptide an- tibiotics are unique among antibiotics in permitting the use of genetic engineering to introduce controlled structural mod- ifications. Using mutagenesis of gene-encoded peptides, it is practical to make thousands of variants which can be sorted by screening techniques. With this report of the sequence of a gene encoding the antibiotic nisin, together with previous reports of genes for subtilin (2) and epidermin (3), the fact that the lanthionine-containing antibiotics are gene-encoded and ribosomally synthesized is firmly established. It is of added interest and importance that the precursors of these antibiotics undergo post-translational modifications to intro- duce unusual amino acids. In addition to its intrinsic interest as a biological phenomenon, this processing system creates an opportunity to introduce unusual amino acids such as dehydroalanine and lanthionine into any peptide or protein,

Small Protein Antibiotic Nisin Gene 16265 Amino Acid Homologies

MSTK DFNLDLVSVSKKDSGASPR ITSISLCTPGCKTGALMGCNMKTATCHCSIHVSK Nisina (Leader Region) (Structural Region)

MEAVKEKNDLFNLDVKVNAKESN----E-- -A-KFI---"AKTGSFNSYCC --KFD--D--v-K---Q--KIT-Q WK-E ------_ V _ _ _ _ QT-FLQ-L--N-K- -- Subtilinb

Epidermin'

Nucleic Acid Homologies

spaN:a Leader Region--> ATGAGT ACAAAAGATTTTAACTTGGATTTGGTATCTGTT *S:b -E:

---TCAAAGTTCG-T-----CG-T------G-T--GAAA--C ATGGAAGCAGTAAAAGAA-AA-A-GATCTTTTTA--C--G-TG-TA-AG-TAATG-AAAA

FIG. 7. Homologies of nisin amino acid and nucleic acid sequences to others. Top shows amino acid homolo- gies between nisin, subtilin, and epider- min. A single-residue gap in the nisin leader and a 2-residue gap in the subtilin structural region are inserted to improve homology. Middle shows homologies of the nisin (SPUN), subtilin (spas), and epidermin (spaE) genes, using the same gaps as above. The spaE gene is the same as the epiA gene (3). The percent ho- mology between amino acid sequences (AA) and nucleic acid sequences (NA) is calculated between the nisin leader (NL), nisin structural region (NS) , and the corresponding regions for subtilin (SL, SS) and epidermin (EL, ES). Bot- tom shows the amino acid homology in an ORF located about 750 base pairs upstream from the nisin precursor gene, with a putative transposase from E. coli IS2 insertion element (20). The homol- ogy was found by a search of the Gene- bank protein data base. The significance is discussed in the text.

FIG. 8. Comparison of the hydro- pathic profiles of subtilin, nisin, and epidermin. Hydropoathic indexes were calculated using the algorithm of Kyte and Doolittle (25), and are plotted as unaveraged values of individual amino acid residues. A hydropathic index of 4.5 (horizontal dotted lines) is neutral, with relatively hydrophobic residues having higher values. Amino acid alignments are the same as in Fig. 7. Unconnected points in the nisin leader and the subtilin structural region reflect the gaps which were inserted to improve homology (see legend to Fig. 7).

spaN : TCGAAGAAAGATTCAGGTGCATCACCACGC <--Leader Region = S : - -T--AC----C---AAAATCA-T--G-AA Structural Region--> spaE : GAATCT--C--------A--TGA----A-A

spaN : ATTACAAGTATTTCGCTATGTACACCCGGTTGTAAAACAGGAGCTCTGATGGGTTGTAACATGAAA = S : TGG-A----GAA--A--T--------A--A---GT---T--T--AT--CAAAC---CTT-C-TC-- spaE : ---G-T----AA-TTA-------T--T--A---GC--A-AC--G-AGTT-TAACA--T-TTGTTGT

s p a N : ACAGCAACTTGTCATTGTAGTATTCACGTAAGC =S : ---CT--C----A-C-GC-AA-TC TCT

Per Cent Homology Among Regions NL

NA A A N A A A NS

NA A A NA A A SL ss

SL 50 54 - - ss - - 51 59 EL 42 25 - - ES - -

34 20 - - " 41 2 3

"

- - "

"

4a 3 2

10 20 30 40 50 60 Homology of Upstream ORF to IS2 Transposase

I I I I I I CGKEHPQPGLIVHTDQGSQYTSSRYQSTLRQVGAQSSMSRKGNPYDNAMMESFYKSLKRELINDAHF

LWRRKRPRNVIVHTDRGGQYCSADYQAQLKRHNLRGSMSAKGCCYDNACVESFFHSLKVECIHGEHF I I I I I I I I I I I I I l l I I I I I I I l l I l l I I I I

70 80 90 100 I I I I I I I l l I I I I I I I / I I

ETRAEATQEIFKYIETYYNTKRMHSGLDYKSPKDFEKYNSZ (Upstream ORF)'

ISREIMRATVFNYIECDYNRWRRHSWCGGLSPEQFENKNLA (Portion of E. colid

athis work bref 2 'ref 3 dref 2 0 IS2 Transposase)

10 LEADER REGION I STRUCTURAL REGION

10 x ) 90 40

RESIDUE NUMBER

16266 Small Protein Antibiotic Nisin Gene

so long as appropriate processing signals and information are incorporated into the precursor sequences. To achieve this, it will be necessary to understand how the signals and infor- mation are encoded in the precursor, and how the processing steps are actually carried out.

One possibility is that all the processing signals are in the leader region of the precursor. The hydropathic homologies in the leader regions of the nisin, subtilin, and epidermin precursors provide circumstantial evidence that this is the case. Accordingly, one might predict that a fusion precursor constructed from the subtilin leader and the nisin structural region would be recognized and processed by B. subtilis to normal active nisin. Even if reality is not so simple, it seems likely that the processing system in B. subtilis can tolerate sequence changes in the subtilin structural region without destroying processing information, thus permitting many sub- tilin variants to be made.

The known existence of still other lanthionine-containing antibiotics suggests that many variants have evolved natu- rally. This seems consistent with the small protein nature of these antibiotics, which provides them access to the same mechanisms of mutation and selection as other proteins. This evolutionary adaptability should permit the antibiotic to con- form readily to the specific needs of its host, which is in competition with other species for a particular ecological niche. The structural variations among nisin, subtilin, and epidermin show these adaptations have been extensive. To illustrate the completeness of this adaptation, it has been shown that the most nisin-sensitive species is Streptococcus cremoris, which occupies essentially the same ecological niche as nisin-producing S. la& (21). It has also been shown that subtilin has a different spectrum of action than nisin (22). If the other lanthionine antibiotics are similarly adapted to fill the needs of their hosts, it is a logical conclusion that their protein nature has permitted them to evolve to the same level of exquisite perfection of specificity for their target that is observed with other evolved proteins such as enzymes and antibodies. Extension of this logic suggests that the number of antibiotic forms that could be created is vast, and the corresponding number of targets they could recognize is of the same order as the number of antigens that can be recog- nized by the repertoire of antibodies made by the mammalian immune system. If so, these antibiotics could be adapted to many uses.

REFERENCES 1. Ingram, L. (1970) Biochim. Biophys. Acta 224,263-265 2. Banejee, S., and Hansen, J. N. (1988) J. Biol. Chem. 263,9508-

3. Schnell, N., Entian, K. D., Schneider, U., Gotz, F., Zahner, H.,

4. Mattick, A. T. R., and Hirsch, A. (1944) Nature 154, 551 5. Gross, E. (1977) in Protein cross-linking (Friedman, M., ed) pp.

131-153, Plenum Press, New York 6. Hurst, A. (1981) in Advances in Applied Microbiology (Perlman,

D., and Laskin, A. I., eds) Vol. 27, pp. 85-123. Academic Press, New York

7. Fukase, K., Kitazawa, M., Sano, A., Shimbo, K., Fujita, H., Horimoto, S., Wakamiya, T., and Shiba, T. (1988) Tetrahedron Lett. 29,795

8. Terzaghi, B. E., and Sandine, W. E. (1975) Appl. Microbiol. 29,

9. Vary, J. C., and Halvorson, H. 0. (1965) J. Bacteriol. 8 9 , 1340- 1347

10. Morris, S. L., and Hansen, J. N. (1981a) J. Bacteriol. 148 , 465- 471

11. Buchman, G. W., and Hansen, J. N. (1987) Appl. Enuiron. Mi- crobiol. 5 3 , 79-82

12. Siegel, J. L., Hurst, S. F., Liberman, E. S., Coleman, S. E., and Bleiweis, A. S. (1981) Infect. Zmmun. 31,808-815

13. Silhavy, T. J., Berman, M. L., and Enquist, L. W. (1984) Exper- iments With Gene Fwions, pp. 137-139, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

9514

Kellner, R., and Jung, G. (1988) Nature 333,276-278

807-813

14. Lathe, R. (1985) J. Mol. Biol. 183, 1-12 15. Ulmanen, L., Lundstrom, K., Lehtovaara, P., Sarvas, M., Ruo-

16. Swank, R. T., and Munkres, K. D. (1971) Anal. Biochem. 39,

17. Morris, S. L., Walsh, R. C., and Hansen, J. N. (1984) J. Biol. Chem. 269,13590-13594

18. Davis, L. G., Dibner, M. D., and Battey, J. F. (1986) Basic Methods in Molecular Biology, Elsevier Science Publishers, New York

19. Bechhofer, D. H., and Dubnau, D. (1987) Proc. Natl. Acad. Sci.

20. Timmerman, K. P., and Tu, C. P. D. (1985) Nucleic Acids Res.

21. Hurst, A., and Collins-Thompson, D. (1979) Adv. Microb. Ecol.

22. Jay, J. M. (1983) in Food Microbiology (Rose, A-H., ed) Vol. 8,

23. Allgaier, H., Jung, G., Werner, R. G., Schneider, U., and Zahner,

24. Band, L., and Henner, D. J. (1984) DNA (NY) 3,17-21 25. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Biol. 167 , 105-132

honen, M., and Palva, I. (1985) J. Bacterwl. 162,176-182

462-477

U. S. A. 84,498-502

13,2127-2139

3 , 79-133

pp. 117-143, Academic Press, London

H. (1986) Eur. J. Biochem. 160,9-22