INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Restriction and modification (R-M) systems are widespread among prokaryotes, where they have a role in counteracting bacteriophage proliferation in a manner akin to a primitive immune system (reviewed in reference 20). Their presence in lactococci in conjunction with other natural phage defense mechanisms (i.e., adsorption blocking, injection blocking, and abortive infection [Abi] mechanisms) is of particular economic importance to the dairy fermentation industry, since phage infection of commercial lactococcal starter cultures persists as one of the primary sources of failed or suboptimal fermentations. The persistent challenge posed by phage infection has required the continued search for strains with superior phage-resistant traits to meet the demands of the dairy fermentation industry.

To date, at least eight complete lactococcal R-M systems have been sequenced. Two of these have been classified as type I systems (46, 47), and five have been classified as type II systems; the latter include LlaDCHI (32), LlaBI (34), LlaCI (26), LlaDII (27), and ScrFI (7, 54, 55). The other R-M system described, LlaI, has characteristics reminiscent of both type IIS and type I R-M systems (17, 35). In general, type II systems are composed of two structural genes, an endonuclease and a methyltransferase. However, both the LlaDCHI and ScrFI R-M systems encode two methyltransferases. Comparison of LlaDCHI with the well-characterized DpnII system indicated that in addition to the conventional double-stranded DNA methyltransferase, a single-stranded DNA methyltransferase was encoded, a property which potentially facilitates the uptake of intact DNA via conjugation in Lactococcus lactis (32). Two 5-methylcytosine methyltransferases were identified in the ScrFI R-M system; while the exact role of each has not been established, the second methyltransferase may have evolved to counteract the degeneracy of the ScrFI endonuclease (55).

A prerequisite to fully developing and enhancing phage resistance systems from a biotechnological standpoint is gaining an understanding of how these defense mechanisms are expressed and regulated in vivo. A number of factors have been shown to have an influence on the phenotypic expression of a number of native lactococcal phage resistance determinants. These factors include heat, whereby elevated temperatures similar to those used during the Cheddar cheese cooking process (40°C) render such mechanisms as the LlaI R-M system and the AbiA abortive infection mechanism (originally designated Hsp due to its heat-sensitive phage resistance) less effective in counteracting phage. In addition, insertion sequence (IS) elements have been demonstrated to play a role in the expression of some Abi determinants. A promoter sequence within an iso-ISS1 element located upstream of the abiB gene from L. lactis subsp. lactis IL416 was required for transcription of this gene (6), whereas insertion of IS981 into a fragment originating from the native lactococcal plasmid, pKR223, accounted for the loss of the Abi phenotype (37). Furthermore, IS-mediated rearrangements have contributed to altered resistance phenotypes in the conjugal plasmid pTR2030 (43).

The native lactococcal phage resistance plasmid pKR223, first identified in L. lactis subsp. lactis biovar diacetylactis KR2, was shown to mediate two distinct mechanisms of resistance to bacteriophage infection: an R-M system active against the small isometric-headed phage sk1, and an Abi mechanism which retards the proliferation of the prolate-headed phage c2 (25, 30). Other phages, such as stl1 and eb1, also exhibited a reduced plaque size on hosts harboring pKR223, an observation predicted to be linked to the presence of the Abi mechanism. By using a streptococcal cloning vector, both resistance mechanisms of pKR223 were subcloned on a 19-kb HpaII fragment, generating a construct designated pGBK17 (25). A combination of deletion derivatives and a natural mutant arising from the fortuitous insertion of a novel lactococcal insertion sequence permitted the general location of both defense mechanisms to be established (30, 37). In this study, we determined the nucleotide sequence and established the precise location of the R-M determinants which contained an IS982 element. We subcloned this region and investigated the effect of the IS element on expression of the R-M phenotype.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria, plasmids, and phage. The strains used in this study included L. lactis subsp. lactis biovar diacetylactis KR2, an industrial starter strain containing seven plasmids including pKR223, which encodes the LlaKR2I R-M system and an Abi mechanism (25); L. lactis subsp. cremoris LM0230, a plasmid-free derivative of L. lactis subsp. cremoris C2 (12) (formerly called subsp. lactis); and L. lactis GBK17, an LM0230 transformant with a 19-kb HpaII fragment of pKR223 cloned into the HpaII site of pGB301 in a construct designated pGBK17 (25). The 5.7-kb Escherichia coli/Lactococcus shuttle vector pCI372 (16) was used to construct the following plasmids: pDOT29, containing a 1.4-kb PCR fragment encoding llaKR2IM and generated by using the primers 5' GCACTAGTTATAACATATGAAATAG 3' and 5' GCGTCGACGATTGATGATAAGGCTG 3', incorporating the restriction sites SpeI and SalI (underlined), respectively, and cloned into the XbaI and SalI sites of pCI372; pDOT31, containing a 4.1-kb PCR fragment incorporating the entire LlaKR2I R-M system including IS982 and generated by using the primers 5' GCGGATCCAATCAGTGCCTCTGTTAC 3' and 5' GCGTCGACGATTGATGATAAGGCTG 3', which incorporated the restriction sites BamHI and SalI, respectively, and was cloned into the BamHI and SalI sites of pCI372; pDOT46, containing two separate PCR fragments harboring llaKR2IR (amplified by using the primers 5' CGGGATCCTTTTAGTATCAACCTGTC 3' and 5' CGACTAGTGTAAATAGTTCAGTAGG 3', incorporating BamHI and SpeI sites, respectively) and llaKR2IM (amplified by using the primers used to construct pDOT29) fused together at an SpeI site; and pDOT51 through pDOT70, 20 8.7-kb PCR-generated constructs consisting of pCI372 and the LlaKR2I R-M system without IS982, amplified by using primers 5' GTAAATAGTTCAGTAGGTATGAGTATGG 3' and 5' TATCATTATAACAGATGAAATAGTTGATCG 3' and religated following treatment with polynucleotide kinase. E. coli XL1Blue MRF' (Stratagene, La Jolla, Calif.) served as the host for the construction of pUC19 chimeras containing fragments of pGBK17 for sequencing. The small isometric-headed bacteriophage sk1 was used for monitoring the R-M phenotype of the LlaKR2I determinants in L. lactis.

Media, enzymes, and culture conditions. Lactococcal strains and their phage were routinely grown without shaking at 30°C in M17 medium (51) supplemented with 0.5% glucose (M17G). For the enumeration of phage, the base medium and the soft agar (0.75%) overlay were supplemented with 5 mM CaCl₂. Where necessary, the antibiotics chloramphenicol (5 µg/ml) and erythromycin (5 µg/ml) were added to media. E. coli was grown in Luria-Bertani (LB) medium (45) at 37°C with aeration. E. coli cells containing pUC19 chimeras were maintained by the addition of ampicillin (100 µg/ml), and recombinants were selected for on LB agar supplemented with 5-bromo-4-chloro-3-indolyl--D-galactosidase (X-Gal; 30 µg/ml) and isopropyl--D-thiogalactopyranoside (IPTG; 30 µg/ml).

Molecular cloning procedures. Restriction enzymes, T4 DNA ligase, polynucleotide kinase, and the Klenow fragment from DNA polymerase I were purchased from either New England Biolabs (Beverly, Mass.) or Promega Corporation (Madison, Wis.) and used according to the manufacturers' instructions. Plasmid DNA for sequencing was isolated from E. coli by using a QIAGEN plasmid mini kit (QIAGEN Inc., Chatsworth, Calif.). A similar kit was used for the isolation of lactococcal DNA except that the following modifications were incorporated into the procedure: 10 ml of an overnight culture of L. lactis was pelleted; the cells were resuspended in 300 µl of buffer P1 (50 mM Tris-HCl, 10 mM EDTA [pH 8.0]) supplemented with RNase A (100 µg/ml) and lysozyme (30 mg/ml) and then incubated at 37°C for 30 min. Thereafter, the procedure outlined for the isolation of E. coli DNA was conducted. Large amounts of lactococcal plasmid DNA were isolated by the method of Anderson and McKay (4) and purified by CsCl-ethidium bromide density gradient centrifugation. Electrotransformation of DNA was performed with a Bio-Rad Gene Pulser apparatus (Bio-Rad Corp., Richmond, Calif.) according to manufacturer's instructions for E. coli or by the method described by Holo and Nes (18) for L. lactis.

PCR. All PCRs were performed with a Robocycler Gradient 40 temperature cycler (Stratagene). For routine PCR applications, Taq DNA polymerase (Promega) was used. In the construction of plasmids pDOT29, pDOT31, and pDOT46, the cloned high-fidelity-proofreading Pfu DNA polymerase from Pyrococcus furiosus (Stratagene) was used to amplify the inserts prior to their ligation to pCI372. The Expand high-fidelity PCR system comprising a blend of Taq DNA polymerase and Pwo polymerase (Boehringer Mannheim Corp., Indianapolis, Ind.) was used to amplify an 8.7-kb fragment whose ends were religated to generate constructs pDOT51 through pDOT70.

DNA sequencing and analysis. A 4.6-kb segment of pGBK17, extending from an EcoRV site to an XbaI site, was subcloned in a variety of fragments in pUC19, and the nucleotide sequences of both DNA strands were determined. Sequencing reactions were performed with an ABI Prism Dye Terminator cycle sequencing kit, using AmpliTaq DNA polymerase FS, and the products were separated in an ABI 377 automatic sequencer (Applied Biosystems, Foster City, Calif.). In addition to universal pUC forward and reverse primers, custom-designed specific primers were used in sequencing reactions. DNA sequences were compiled and analyzed with DNASTAR* software (DNASTAR, Madison, Wis.) and compared to database sequences by using the BLAST suite of programs (2).

Phage assays. The R-M status of lactococcal hosts was monitored by plaque assays using the small isometric-headed sk1. To titer the phage, 1 ml of the relevant phage dilution was added to 4 ml of prewarmed M17G medium (46°C), which already contained 0.1 ml of an overnight culture of the appropriate host and 5 mM CaCl₂; contents were mixed and poured onto M17G medium supplemented with 5 mM CaCl₂. The efficiency of plaquing of the phage was defined as (phage titer on the host of interest)/(phage titer on a nonrestricting host). Phage DNA modification was established by purification of phage from single-plaque isolates and repropagation on the same host culture.

Nucleotide sequence accession number. The nucleotide sequence presented here has been deposited in the GenBank database and has been assigned the accession no. AF051563.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Sequence organization of the LlaKR2I R-M system. The native lactococcal plasmid pKR223 has previously been shown to encode an R-M system whose genetic determinants were subcloned on a 19-kb HpaII fragment in pGBK17 (25). The approximate location of the genetic loci encoding restriction and modification activities of pKR223 were previously mapped by deletion analysis to a region of ~5 kb (30). In this study, multiple fragments from this location were subcloned in E. coli by using pUC19, and the nucleotide sequence of 4,612 bp was determined (Fig. 1). A number of open reading frames (ORFs) were identified, and the deduced protein sequences were compared to data bank sequences, permitting identification of the endonuclease and methyltransferase genes. This analysis confirmed that the R-M system had not previously been characterized; accordingly, it was designated LlaKR2I, based on the conventional nomenclature for R-M systems (50). The LlaKR2I R-M system comprised two divergently transcribed genes, an endonuclease gene (llaKR2IR) and a methyltransferase gene (llaKR2IM); there was a complete copy of the lactococcal IS element IS982 in the intergenic region, with the putative transposase gene oriented in the same transcriptional direction as llaKR2IM (Fig. 2). This organization is unique for all R-M systems described to date. The position of the IS element was also verified in the parental industrial strain L. lactis subsp. lactis biovar diacetylactis KR2 by using PCR (data not shown), indicating that insertion of this element did not occur following passage through the laboratory strain, L. lactis subsp. cremoris LM0230.

View larger version (85K): [in this window] [in a new window]   View larger version (46K): [in this window] [in a new window]   FIG. 1.   Nucleotide sequence of the LlaKR2I R-M system harboring a copy of the IS element IS982. The deduced protein sequences of the major ORFs observed are presented beneath the nucleotide sequence, along with the gene designation and an arrow indicating the direction of transcription. Potential ribosome binding sites (RBS) are underlined. Several potential translational start codons are available for LlaKR2I; however, the only one preceded by an appropriately spaced ribosome binding site is indicated. Inverted repeat sequences at the left and right termini of IS982 are indicated by arrows labeled IR-L and IR-R, respectively. Direct repeat sequences flanking IS982 are indicated by thick lines labeled DR. Sequences homologous to 10 and 35 promoter consensus sequences are labeled accordingly and are accompanied by the letter P and an arrow indicating the direction of transcription. A potential stem-loop transcriptional terminator structure located after llaKR2IM is indicated by facing arrows.

View larger version (11K): [in this window] [in a new window]   FIG. 2.   Restriction map of the LlaKR2I R-M locus. The positions and frequencies of relevant restriction sites are shown. ORFs are represented by arrows. A predicted stem-loop secondary structure located after llaKR2IM which potentially acts as a transcriptional terminator is indicated. The G value of this structure was calculated by the method of Tinoco et al. (53).

View larger version (53K): [in this window] [in a new window]   FIG. 3.   (A) Alignment of the deduced protein sequences of the endonucleases Sau3AI (48) and LlaKR2I. Shaded residues indicate amino acids identical in the two sequences. (B) Alignment of segments of Sau3AI and LlaKR2I which exhibit sequence similarity with the GATC-specific MutH proteins of E. coli (15) and H. influenzae (13). Identical residues are shaded, and residues identical in three of the four sequences are boxed.

View larger version (42K): [in this window] [in a new window]   FIG. 4.   Alignment of deduced protein sequences of M · Sau3AI and M · LlaKR2I. Shaded residues indicate amino acids identical between the two sequences; dashed lines represent breaks introduced to optimize alignment. Ten motifs (I to X) characteristic of the 5-methylcytosine methyltransferase family of proteins (39) are indicated by lines above the sequence A TL dipeptide present in both proteins which has been used to anchor alignments of numerous TRDs in other 5-methylcytosine methyltransferases is marked by two dots above the residues toward the C-terminal end of the variable region. The catalytic active site of these proteins is located within motif IV and is marked by an asterisk.

Subcloning of the LlaKR2I R-M system. Phage sk1 was restricted and modified when passed through L. lactis subsp. lactis biovar diacetylactis KR2 (25, 30). To confirm that the loci identified by sequence analysis were the only regions required for expression of the R-M phenotype, a 4,057-bp PCR fragment encompassing llaKR2IR, IS982, and llaKR2IM was generated with high-fidelity-proofreading Pfu DNA polymerase, using primers which included restriction sites to facilitate cloning into pCI372. Following ligation of the PCR fragment into pCI372, attempts to efficiently transform E. coli XLIBlue MRF' with this DNA proved unsuccessful; we now know that this difficulty was due to the toxicity of the methyltransferase M · LlaKR2I (56). This finding is similar to that observed for the Sau3AI R-M system, where efforts to clone the entire R-M system in a variety of E. coli strains failed and the toxicity was tentatively linked to the expression of sau3AIM (48). Since the methyltransferase was detrimental to E. coli, all subsequent transformations with DNA incorporating llaKR2IM were carried out in L. lactis LM0230. A number of L. lactis LM0230 transformants harboring pCI372 with an insert containing llaKR2IR, IS982, and llaKR2IM were obtained (plasmid designated pDOT31), and infection with sk1 indicated that all encoded an active endonuclease. The degree of restriction observed was similar to (or slightly better than) that observed with hosts carrying pGBK17 (Table 1). As expected, sk1 propagated on hosts harboring pDOT31 were not restricted by pGBK17, demonstrating M · LlaKR2I activity. These data confirmed that the 4.1-kb fragment of pDOT31 encoded a functional R-M system.

Deletion of IS982 from the LlaKR2I R-M system. The organization of the LlaKR2I R-M system suggests that originally, expression of llaKR2IR or llaKR2IM may have been directed from divergent promoters within the intergenic region, a region disrupted by the insertion of IS982. As it is well established that IS elements may disrupt or contribute to gene expression, we investigated how the LlaKR2I R-M system functions in the absence of IS982. From the alignments of all of the boundaries of IS982 elements sequenced to date, it is clear that the IS element present in the LlaKR2I R-M system has 16-bp indirect repeats at its ends which are flanked by 7-bp direct repeats, regions believed to have been duplicated during the transposition event (Fig. 5). To generate a putative precursor of the LlaKR2I R-M system prior to the insertion of IS982, the entire IS element and one of the direct repeats (5' TATCATT 3') flanking this element should be eliminated.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Alignment of the left and right termini of IS982 elements fully sequenced to date and their flanking DNA sequences. The termini and flanking sequences of the partially sequenced element from Wg2 are also presented (60). The IS982 element identified in the LlaKR2I R-M system is denoted by pKR223(5'), whereas a second copy located 7.2 kb downstream is denoted by pKR223(3'). A 16-bp inverted repeat sequence present in all elements, which may extend up to 18 bp in some instances, is highlighted by an arrow above the sequence. When present, direct repeats flanking the inverted repeat sequences which are believed to have been duplicated during the transposition event are boxed. The internal highly homologous portion of these IS elements is represented by dashed lines.

View larger version (25K): [in this window] [in a new window]   FIG. 6.   Diagrammatic representation of the LlaKR2I R-M system and the effect of removing IS982. Inverted repeat sequences at the left and right termini of IS982 (arrows labeled IR-L and IR-R, respectively) are flanked by directly repeated sequences (DR) depicted as hashed boxes. Putative promoters are represented by P with arrows indicating the direction of transcription. Sequences equivalent to 10 and 35 consensus sequences are labeled accordingly and represented by a bar above the sequence. A Sau3AI site overlapping the putative 35 llaKR2IM promoter and which may have a role in regulation of R-M expression is highlighted below the sequence. Short facing arrows represent a putative stem-loop transcriptional terminator structure located immediately after llaKR2IM. PCR fragments cloned in pCI372 during this study demonstrating the effect of the presence or absence of IS982 on the R-M phenotype are depicted below the sequence. The inclusion of an SpeI restriction site in primers A and B during the construction of pDOT46 contributed to an R phenotype, presumably due to alterations to an extended 10 promoter sequence overlapping this site. pDOT70* is a representative of 20 transformants harboring a plasmid with IS982 specifically deleted by using primers C and D as PCR primers and pDOT31 as the template. RBS, ribosome binding site.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The organization of the LlaKR2I R-M system is unique for type II R-M systems, having an IS element inserted between divergently transcribed endonuclease and methyltransferase genes. These elements are prevalent in lactococci and have been associated with a variety of important cellular processes, including phage resistance, proteinase activity, nisin production, and ability to utilize lactose, sucrose, and citrate (8, 14, 44). The proximity of these mobile, transposable elements to many of these important traits presumably reflects the evolutionary role of IS elements in helping cells adapt to their environment. To date, five classes of lactococcal IS elements have been characterized: ISS1 (38), IS904 (9), IS981 (37), IS905 (10), and IS982 (60). IS982 was first identified between the oligopeptide permease gene cluster and origin of replication in the lactose-utilizing plasmid pSK11L from L. lactis subsp. cremoris SK11 (60). In addition to the two IS982 elements identified in this study, the sequences of three further complete highly homologous IS982 elements have been reported for the citrate utilization plasmid pCIT264 (8), the exopolysaccharide encoding plasmid pNZ4000 (57), and the chromosome of the phage resistance strain L. lactis subsp. lactis biovar diacetylactis S94 (40). All six fully characterized IS982 elements and a partially sequenced element from L. lactis subsp. lactis WG2 (60) contained identical perfect inverted repeats at the ends. Many of these inverted repeats were flanked by direct repeats. However, the lengths of the observed direct repeats vary from between 2 and 8 bp (Fig. 5), indicating that unlike for many other lactococcal IS elements, the length of the duplicated region is not conserved for this IS family.

Why IS982 is located in the LlaKR2I R-M system of pKR223 is unclear, as the system can operate independent of this element and originally may even have been more efficient at restricting phage proliferation when this element was absent. It is conceivable that the transposition of IS982 between llaKR2IR and llaKR2IM may have been an evolutionary response to stress imposed on the cell by aberrant restriction or modification activity. It is plausible that if methylation activity was diminished in the presence of an active functional endonuclease, autodigestion of the hosts genomic DNA may have ensued, provoking the insertion of IS982 and permitting either an enhancement of modification activity and/or a reduction in endonuclease activity. Therefore, the insertion of IS982 may have been an induced temporal control mechanism.

The restriction and modification components of LlaKR2I shared 32.9 and 54.6% identity with the equivalent proteins of the Sau3AI R-M system (48). While all DNA methyltransferases share some common sequence motifs (28), no recurring primary sequence motifs exist for the type II endonucleases despite similar catalytic functions and cofactor requirements, preventing any systematic grouping of these heterogeneous proteins (reviewed in references 1, 42, and 59). To date only some isoschizomeric type II endonucleases exhibit homology. Therefore, while the degree of homology observed among the Sau3AI and LlaKR2I endonucleases was not quite as high as that observed for their methyltransferases, the similarities nevertheless are quite significant and presumably highlight the likely common sequence specificities of these proteins. These similarities point to a common evolutionary origin for the Sau3AI and LlaKR2I R-M systems. Neither of the endonucleases of these systems shared any significant sequence similarity with DpnI, DpnII, MboI, or LlaDCHI, all of which are GATC endonucleases but differ markedly with respect to the ability to be affected by DNA methylation. DpnII, MboI, and LlaDCHI all resemble each other (32) and are prevented from digesting their host DNA by a cognate N6-methyladenine methyltransferase, whereas the methylation-dependent endonuclease DpnI cleaves GATC sites only when the adenine has been modified to N6-methyladenine (24). Therefore, primary sequence comparisons of these GATC-specific systems indicated that one of the factors which appears to have had a pronounced influence on how the endonuclease evolved was the type of accompanying methyltransferase, or the absence of one in the case of DpnI. While many type II endonucleases lack discernible similarity at the primary sequence level, recent X-ray crystallography data indicate that despite the absence of primary sequence similarity, these proteins may share similar tertiary structures especially throughout the catalytic active-site residues (1, 5, 36). Therefore, originally all six endonucleases may have had a common ancient progenitor but then divergently evolved into enzymes which are compatible with their accompanying methylation requirements. The MutH proteins of E. coli and H. influenzae appear to be more closely related to Sau3AI and LlaKR2I than other type II GATC-specific endonucleases. In E. coli, MutH has a role in postreplicative mismatch repair, where it creates a nick 5' of the guanine in a GATC site in a newly synthesized DNA strand harboring a mismatched base (58) and requires the GATC specific N6-methyladenine methyltransferase, Dam, to discriminate the nascent DNA strand harboring the mismatch (reviewed in reference 33). The homology observed among these sequence-specific endonucleases is in agreement with a previous proposal that the MutH mismatch repair proteins evolved from an ancient R-M system where the mutH gene originally accompanied a GATC-specific methyltransferase gene but evolved into a mismatch correction process (29). The similarities between LlaKR2I and the MutH homologs maybe further evidence of this common ancestry.

Extensive studies of a range of 5-methylcytosine methyltransferases, particularly the Bacillus phage-encoded multispecific methyltransferases, have demonstrated that TRDs reside toward the C-terminal end of the variable region and generally contain the dipeptide T(L,I,V) (reviewed in reference 33). Furthermore, X-ray crystallography data for the 5-methylcytosine methyltransferases M · HhaI and M · HaeIII complexed with their specific substrates revealed that along with the catalytic active site, residues toward the C-terminal end of the variable region form the majority of base-specific contacts with the target sequence (22, 41). The homology observed between M · Sau3AI and M · LlaKR2I extended through the C-terminal end of the variable region and included a conserved TL dipeptide (Fig. 4). These data suggest that similar to TRDs of the other 5-methylcytosine methyltransferases, the TRDs of M · Sau3AI and M · LlaKR2I reside toward the C-terminal end of the variable region and that these two proteins methylate similar or identical sequences.

Resistance to phage infection is an industrially important trait in lactococcal starter cultures. In this respect R-M systems are potentially of immense benefit to the dairy fermentation industry, as they are more versatile than other bacteriophage resistance mechanisms which are often specific for certain phage groups or species. Theoretically R-M systems require only one target site in the phage to be effective, with a greater degree of restriction observed with an increasing number of sites (32). However, the relative ease at which fully methylated phage develop, the ability of phage to eliminate certain restriction sites from the genome, and the presence of phage antirestriction mechanisms requires R-M systems to be used in conjunction with other compatible, complementary resistance mechanisms. A number of strategies combining multiple R-M systems either in a single strain or in a strain rotation strategy have been devised for dairy starter cultures and shown to be effective in counteracting phage proliferation (11, 19, 49). Localization of the determinants of the LlaKR2I R-M system and characterization of its sequence specificity provide an additional tool for devising rational strategies to counteract phage infection and proliferation. In a complex milk habitat, the association of R-M systems and other phage resistance mechanisms with IS elements may facilitate their dissemination throughout other lactococcal strains via transposition or homologous recombination, permitting phage-insensitive variants to evolve. However, in a defined starter rotation strategy, the evolution of variants may not be desired and may even be detrimental if certain critical traits are disrupted. Furthermore, the close association of an IS element with an R-M system may allow phage to acquire the modification component more readily due to the increased mobility of these elements. Indeed, it is interesting to speculate that the acquisition of a functional domain of the LlaI methyltransferase (M · LlaI) by a phage may have been mediated by an IS element (IS946) located upstream of llaIM, when the LlaI R-M system was introduced into a commercial dairy fermentation environment (17). Therefore, removal of IS982 from the LlaKR2I R-M system should provide it with an extra element of stability and may enhance the chances of the integrity of the resistance mechanism being maintained during the fermentation process.