Cloning and Characterization of the Lactococcal Plasmid-Encoded Type II Restriction/Modification System, LlaDII

Abstract

The LlaDII restriction/modification (R/M) system was found on the naturally occurring 8.9-kb plasmid pHW393 in Lactococcus lactis subsp. cremoris W39. A 2.4-kb PstI-EcoRI fragment inserted into the Escherichia coli-L. lactis shuttle vector pCI3340 conferred to L. lactis LM2301 and L. lactis SMQ86 resistance against representatives of the three most common lactococcal phage species: 936, P335, and c2. The LlaDII endonuclease was partially purified and found to recognize and cleave the sequence 5′-GC↓NGC-3′, where the arrow indicates the cleavage site. It is thus an isoschizomer of the commercially available restriction endonuclease Fnu4HI. Sequencing of the 2.4-kb PstI-EcoRI fragment revealed two open reading frames arranged tandemly and separated by a 105-bp intergenic region. The endonuclease gene of 543 bp preceded the methylase gene of 954 bp. The deduced amino acid sequence of the LlaDII R/M system showed high homology to that of its only sequenced isoschizomer, Bsp6I from Bacillus sp. strain RFL6, with 41% identity between the endonucleases and 60% identity between the methylases. The genetic organizations of the LlaDII and Bsp6I R/M systems are identical. Both methylases have two recognition sites (5′-GCGGC-3′ and 5′-GCCGC-3′) forming a putative stem-loop structure spanning part of the presumed −35 sequence and part of the intervening region between the −35 and −10 sequences. Alignment of the LlaDII and Bsp6I methylases with other m⁵C methylases showed that the protein primary structures possessed the same organization.

Lactococcus strains are widely used as starter cultures for the manufacture of dairy products. During the fermentation processes, the lactococci are often challenged by a variety of bacteriophages. The most common phages belong to the 936, P335, and c2 species and are responsible for most milk fermentation failures (6, 42). Although this detrimental effect has been recognized for many years, the traditional problem-solving methods have only recently been supplemented with genetic and molecular techniques (8, 12, 17, 27, 53, 56). The approaches are based mainly on natural phage defense mechanisms, which are classified into three groups on the basis of their mode of action: blocking of phage adsorption, restriction/modification (R/M) systems, and abortive infection. In addition, a fourth bacteriophage resistance mechanism, phage DNA penetration blocking, has recently been reported (16). Several mechanisms have been found in individual strains. Generally the determinants are plasmid encoded, although one example of a chromosomal location has been described (21).

Due to the diversity of lactococcal phages, phage defense mechanisms with broad activity are required. R/M systems have the potential to fulfill this requirement. Despite the efficiency of R/M systems, any surviving phages will become methylated, with dramatic consequence to fermentations, as the phage DNA is no longer recognized as hostile and can replicate normally within the cell. Phages may develop counterdefense mechanisms, which provide immunity against bacterial resistance mechanisms (37). Examples of phage-encoded methylases found in Lactococcus and Bacillus species and in Escherichia coli have been reported (23, 25, 60). Furthermore, phages undergo selection against recognition sites of host restriction enzymes through evolution (30, 55). These inconveniences may be eliminated or at least reduced by ensuring a sufficiently high bacteriophage resistance level. This can be obtained by stacking R/M systems or combining them with other bacteriophage resistance mechanisms.

Several R/M systems have been identified in Lactococcus species. Despite this, only a few systems have been characterized with respect to their recognition or nucleotide sequence (21). The first biochemical evidence of a type II restriction endonuclease in lactic acid bacteria was for the chromosomally encoded ScrFI from Lactococcus lactis subsp. cremoris UC503. This endonuclease specifically recognizes 5′-CC↓NGG-3′, where N is A, C, G, or T and the arrow indicates the cleavage site (14). The complete nucleotide sequence of the ScrFI R/M system has now been published (10, 61, 62). The first lactococcal R/M system sequenced was the LlaI system on the conjugal plasmid pTR2030 isolated from L. lactis subsp. lactis ME2. The methylase of the pTR2030 system was similar to the type IIs methylase M · FokI (23). Downstream of the methylase, four open reading frames were identified, of which three were involved in LlaI endonuclease activity (49). Furthermore, pTR2030 also encoded an abortive-infection mechanism (22). Another LlaI endonuclease (renamed Lla497I) has been reported. This endonuclease, from L. lactis subsp. lactis NCDO 497, belongs to type II and recognizes the sequence 5′-CCWGG-3′, where W is A or T (41). From L. lactis subsp. cremoris DCH-4 the plasmid-encoded type II R/M system LlaDCHI, which recognizes 5′-↓GATC-3′ was cloned and sequenced. LlaDCHI strongly resembles the DpnII R/M system in organization of the operon as well as in sequence similarity (44).

Our laboratory has previously reported results for two plasmid-encoded type II R/M systems, LlaAI and LlaBI, recognizing 5′-↓GATC-3′ and 5′-C↓TRYAG-3′, respectively, where R is A or G and Y is C or T (47). The LlaAI system contains three open reading frames and shows homology to the DpnII system (48). The nucleotide sequence of LlaBI has recently been published (46).

Here we report on another type II R/M system, isolated from the Danish mixed cheddar starter TK5. The 8.9-kb plasmid pHW393 from L. lactis subsp. cremoris W39 expresses LlaDII activity and confers resistance against phages of the 936, P335, and c2 species. LlaDII cleaves the sequence 5′-GC↓NGC-3′ and is thus an isoschizomer of Bsp6I, BsoFI, and Fnu4HI. The genes of LlaDII were cloned and sequenced. Analysis of the nucleotide sequence predicted that the endonuclease gene precedes the methylase gene. The methylase displayed typical amino acid sequence motifs indicative of an m⁵C methylase (35, 36, 50).

MATERIALS AND METHODS

Bacterial strains and media.

The strains and plasmids used in this study are listed in Tables 1 and 2. L. lactis was grown at 30°C in M17 medium (Oxoid) supplemented with 0.5% glucose (GM17). E. coli was grown at 37°C in Luria broth (52). When appropriate, antibiotics were added as follows: for L. lactis, 6 μg of chloramphenicol per ml, and for E. coli, 100 μg of ampicillin per ml, 10 μg of tetracycline per ml, or 20 μg of chloramphenicol per ml. Blue-white screening with pBluescript SKII+ was performed as described by Sambrook et al. (52). pVS2-cured derivatives were obtained with 0.1 μg of novobiocin per ml in GM17 broth.

TABLE 1.

Bacterial strains and bacteriophages

Bacterial strain or phage	Relevant characteristicsa	Source or reference
L. lactis
W39	Industrial strain, multiple plasmids	28
LM2301	Plasmid free, host for 936 and c2 phages	64
SMQ86	Derivative of the industrial strain L. lactis UL8, host for P335 phages	43
E. coli
XL1-Blue MRF′	Transformation host, Tcr	Stratagene, La Jolla, Calif.
Sure	Transformation host, Tcr	Stratagene
Phages
p2	Small, isometric-headed 936 species	20
jj50	Small, isometric-headed 936 species	29
sk1	Small, isometric-headed 936 species	7
ul36	Small, isometric-headed P335 species	43
Q30	Small, isometric-headed P335 species	43
Q33	Small, isometric-headed P335 species	43
c2	Prolate-headed c2 species	20
λb2	Carries a deletion of the b region that damages the att site and therefore prevents lysogenization	13

^aTc^r, tetracycline resistance. 

TABLE 2.

Plasmids

Plasmid	Relevant characteristicsa	Source or reference
pHW393	Resident plasmid of W39, R+/M+, 8.9 kb	This study
pVS2	Shuttle vector, Cmr Emr, 5.0 kb	63
pSA3	Shuttle vector, Cmr Tcr Emr, 10.2 kb	9
pCI3340	Shuttle vector, Cmr, 5.7 kb	18
pCAD1	2.4-kb PstI-EcoRI fragment from pHW393 cloned into pCI3340; Cmr, R+/M+, 8.1 kb	This study
pEE1	1.4-kb EcoRV-EcoRI fragment from pHW393 cloned into pCI3340; Cmr, R+/M+, 7.1 kb	This study
pBluescript SKII+	Cloning vector for sequencing, Amr, 3.0 kb	Stratagene
pPX1	0.9-kb PstI-XhoI fragment from pHW393 cloned into pBluescript SKII+; Amr, 3.9 kb	This study
pXE1	1.5-kb XhoI-EcoRI fragment from pCAD1 cloned into pBluescript SKII+; Amr, 4.5 kb	This study

^aAm^r, ampicillin resistance; Cm^r, chloramphenicol resistance; Em^r, erythromycin resistance; Tc^r, tetracycline resistance; R⁺/M⁺, active restriction and modification enzymes. 

Bacteriophage propagation and assays.

The bacteriophages used in this study are listed in Table 1. Lactococcal phages were propagated as described by Terzaghi and Sandine (58), and λb2 was propagated as described previously (52). Plaque assays were conducted by the method of Jarvis (26), and the efficiency of plaquing (EOP) was calculated as the ratio of plaques formed on the resistant host to those formed on the sensitive host. Screening for phage resistance was done by cross streaking the bacteria against suitable phage dilutions on GM17 agar containing 5 mM CaCl₂.

Transformation.

E. coli was transformed by the standard CaCl₂ procedure (52). Electrotransformation of L. lactis was performed as described by Holo and Nes (24).

Restriction enzyme purification.

One liter of a fresh overnight culture of L. lactis LM2301 containing pCAD1 was harvested at 16,000 × g, washed in 200 ml of wash buffer (50 mM Tris-HCl [pH 7.6], 10 mM MgCl₂), and resuspended in 12 ml of ice-cold lysis buffer (50 mM Tris-HCl [pH 7.6], 10 mM MgCl₂, 25 mM NaCl, 7 mM β-mercaptoethanol). Cells were disrupted with a French press (Aminco, Silver Spring, Md.) at 1,500 lb/in². After centrifugation to remove cell debris and ribosomes, the supernatant (crude extract) was purified by one-step fast protein liquid chromatography anion-exchange chromatography on a Mono Q column in buffer A (50 mM Tris-HCl [pH 7.6], 10 mM MgCl₂, 5 mM β-mercaptoethanol) with a KCl salt gradient. The fractions were assayed for endonuclease activity with nonmethylated phage lambda DNA (Pharmacia Biotech, Uppsala, Sweden) as the substrate. The digestions were performed at 37°C for 1 h in 33 mM Tris-acetate (pH 7.9)–10 mM Mg-acetate–66 mM K-acetate–0.5 mM dithiothreitol. DNA samples were analyzed in 0.7% agarose gels in Tris-acetate-EDTA as described previously (52).

DNA isolation and manipulation.

Large quantities of lactococcal plasmid DNA were extracted by the method of Anderson and McKay (1) and further purified by CsCl-ethidium bromide gradients (52). Small-scale preparations were performed by a method modified from that of Andresen et al. (3). Large quantities of plasmid DNA from E. coli were isolated with a Qiagen (Chatsworth, Calif.) Maxi kit, while minipreparations were performed by the TELT method described by Ausubel et al. (5). Restriction endonucleases (Boehringer, Mannheim, Germany, or New England Biolabs, Beverly, Mass.), Klenow enzyme and calf intestine alkaline phosphatase (Boehringer), mung bean nuclease (New England Biolabs), and T4 ligase (U.S. Biochemicals, Cleveland, Ohio) were used according to the manufacturer’s instructions.

Sequencing.

The 2.4-kb PstI-EcoRI fragment from pHW393 comprising the LlaDII genes was sequenced. The 0.9-kb PstI-XhoI fragment from pHW393 and the 1.5-kb XhoI-EcoRI fragment from pCAD1 were subcloned into pBluescript SKII+, resulting in pPX1 and pXE1, respectively. Nested deletions were made in both directions with a nested deletion kit (Pharmacia Biotech). However, it was not possible to obtain deletions from the EcoRI end of the XhoI-EcoRI fragment. Instead, the sequence was obtained by use of custom-made primers (Pharmacia Biotech). Single-stranded DNA for sequencing was purified by using paramagnetic beads with covalently coupled streptavidin (Dynabeads M280; Dynal AS, Oslo, Norway) for the capture of biotin-labeled DNA fragments. Incorporation of biotin was performed by using a biotinylated primer in a PCR. The sequencing reactions were performed by the standard dideoxy sequencing procedure with an Auto Read sequencing kit and fluorescein-labeled universal and reverse primers (Pharmacia Biotech). The nucleotide sequence was determined on an Automated Laser Fluorescent (A.L.F.) DNA sequencer (Pharmacia Biotech). The DNA sequence was analyzed with the Genetics Computer Group (Madison, Wis.) sequence analysis software, version 8. Comparisons of DNA and amino acid sequences were performed by using gap analysis with default parameters.

RNA extraction and primer extension.

Total RNA was extracted from L. lactis LM2301 and L. lactis LM2301(pCAD1) by the RNA isolation procedure described by Arnau et al. (4). The oligonucleotide used for primer extension was complementary to the sequence shown in Fig. 4 and had the sequence 5′-GGTTCACTAATTGCATCAGGCATATTCATTCCTCG-3′ (positions 892 to 858). Furthermore, it was Cy5 labeled on the 5′ end, enabling primer extension analysis with the aid of an ALFexpress DNA sequencer as described by Myöhänen and Wahlfors (45). With minor modifications, the primer extension was performed as described previously (45). Reverse transcriptase and RNasin were purchased from Gibco BRL (Roskilde, Denmark).

FIG. 4.

Nucleotide sequence of the 2,355-bp PstI-EcoRI fragment from plasmid pHW393 and the deduced amino acid sequences of the LlaDII endonuclease and methylase. Relevant restriction sites are shown. Start codons and presumed promoter signals are underlined. The transcription start of llaDIIR and inverted repeats are indicated by boldface and arrows. Identical or conserved amino acid changes found with the alignment are shown in boldface, and the motifs are underlined. Motif numbers are indicated. The Pro-Cys catalytic site is situated in motif IV.

Strain deposition.

L. lactis subsp. cremoris LM2301 with pCAD1 containing the LlaDII R/M system has been deposited at the Belgian Coordinated Collections of Microorganisms, Laboratorium voor Microbiologie-Bacteriënverzameling, Universiteit Gent, Ghent, Belgium, under accession no. LMG P-16901.

Nucleotide sequence accession number.

DNA sequence information is available in the EMBL database through accession no. Y12707.

RESULTS

Identification of a plasmid encoding the LlaDII R/M system.

L. lactis subsp. cremoris W39 was isolated from the mixed cheddar starter culture TK5 (28). This starter has been exceptionally resistant to bacteriophages and was used daily for 12 years for industrial cheese production without any phage-associated problems (63a). L. lactis subsp. cremoris W39 was one of the very phage-resistant strains and had previously been shown to express the type II endonuclease activity LlaDI (48). In order to isolate this R/M system, which was expected to be located on a plasmid, total plasmid DNA from L. lactis subsp. cremoris W39 was isolated and cotransformed with the marker plasmid pVS2 into the phage-sensitive, plasmid-free strain L. lactis LM2301 as described previously (29). Chloramphenicol-resistant colonies were screened for phage resistance, and a few phage-resistant transformants were obtained and analyzed. One of them contained pVS2 and a 8.9-kb plasmid, pHW393. The transformant was cured of pVS2 by using novobiocin. Phages propagated on the resulting strain circumvented restriction, indicating that plasmid pHW393 encodes an R/M system. Crude lysate from L. lactis LM2301(pHW393) was found to mediate type II endonuclease activity on phage lambda DNA (data not shown).

Cloning of the LlaDII R/M system.

A restriction map of pHW393 is presented in Fig. 1. Different restriction fragments were shotgun cloned into the E. coli-L. lactis shuttle vector pCI3340 and electroporated into L. lactis LM2301. Transformants were screened for phage resistance. In this way a plasmid with a 2.4-kb PstI-EcoRI insert was found and designated pCAD1. In previous experiments phage lambda DNA digested by partially purified LlaDI lysate from L. lactis subsp. W39 gave six bands (48). In contrast, phage lambda DNA digestions performed with partially purified lysate from L. lactis LM2301(pCAD1) expressed type II endonuclease activity that was found to digest phage lambda DNA into multiple fragments of about 1 kb or smaller (data not shown). The new endonuclease activity was named LlaDII. Phages propagated on L. lactis LM2301(pCAD1) were capable of overcoming the resistance, indicating that pCAD1 encodes a classical R/M system. Phages were not restricted by L. lactis LM2301 carrying a 1.4-kb EcoRV-EcoRI fragment cloned into pCI3340, resulting in pEE1. Since phages propagated on L. lactis LM2301(pEE1) could circumvent the LlaDII activity, plasmid pEE1 probably codes for the LlaDII methylase.

FIG. 1.

Restriction map of plasmid pHW393.

Effectiveness of the LlaDII R/M system against lactococcal phages.

The LlaDII R/M system on plasmid pCAD1 was assessed for its ability to function as a defense mechanism against seven phages belonging to the phage species 936, P335, and c2. The EOP was determined (Table 3). The 936-type phages, represented by the small, isometric-headed phages p2, sk1, and jj50, showed EOPs in the range of 10⁻² to 10⁻⁴. Phage c2, the only prolate-headed phage tested, had an EOP of 10⁻². In order to determine the effect of LlaDII against the small isometric-headed phages belonging to the P335 phage species, pCAD1 was transformed into L. lactis SMQ86. The P335 phages ul36, Q30, and Q33, propagated on L. lactis SMQ86, had EOPs in the range of 10⁻⁵ to 10⁻⁶.

TABLE 3.

Levels of resistance of the LlaDII R/M system against lactococcal phages^a

Phage species	Phage	EOPb
936	p2	2.1 × 10−4
	jj50	2.8 × 10−4
	sk1	1.2 × 10−2
P335	ul36	8.0 × 10−6
	Q30	4.5 × 10−6
	Q33	4.9 × 10−6
c2	c2	2.4 × 10−2

^aThe EOPs of the 936 and c2 phages were tested on L. lactis LM2301(pCAD1). The EOPs of these phages are 1 on L. lactis LM2301. The EOPs of the P335 phages were tested on L. lactis SMQ86(pCAD1). The EOPs of these phages are 1 on L. lactis SMQ86. 

^bCalculated as the ratio of plaques on the resistant host to plaques on the sensitive host. The results are averages of at least three independent determinations. 

Determination of the recognition sequence and cleavage site.

Digestions of nonmethylated phage lambda DNA with partially purified LlaDII resulted in DNA fragments smaller than 1 kb. These fragments were blunt ended with Klenow enzyme or mung bean nuclease and subsequently ligated into pBluescript SKII+ digested with SmaI (47). Plasmid DNA from white colonies was sequenced by using universal and reverse primers. The recognition sequence and cleavage site were determined to be 5′-GC↓NGC-3′, where N is A, C, G, or T and the arrow indicates the cleavage site. This sequence occurs 380 times in phage lambda DNA. The restriction patterns of the vectors pSA3, pCI3340, and pBluescript SKII+ digested with LlaDII and with the commercial isoschizomer Fnu4HI (38) were identical (Fig. 2). Attempts to cut pCAD1 with Fnu4HI were not successful, indicating that LlaDII modification of pCAD1 protected the plasmid from restriction by Fnu4HI (Fig. 3).

FIG. 2.

Digestions of plasmids pSA3, pCI3340, and pBluescript SKII+. Lane 1, Supercoiled ladder (Gibco BRL); lane 2, 1-kb ladder (Gibco BRL); lane 3, uncut pSA3; lane 4, pSA3 cut with Fnu4HI; lane 5, pSA3 cut with LlaDII; lane 6, uncut pCI3340; lane 7, pCI3340 cut with Fnu4HI; lane 8, pCI3340 cut with LlaDII; lane 9, uncut pBluescript SKII+; lane 10, pBluescript SKII+ cut with Fnu4HI; lane 11, pBluescript SKII+ cut with LlaDII; lane 12, 1-kb ladder (Gibco BRL); lane 13, Supercoiled ladder (Gibco BRL). The digestions were performed for 7.5 h at 37°C in the case of Fnu4HI and at 30°C in the case of LlaDII.

FIG. 3.

Digestions of EcoRI-linearized plasmids pCAD1 and pXE1. Both plasmids were initially digested with EcoRI. Lane 1, 1-kb ladder (Gibco BRL); lane 2, pCAD1; lane 3, pCAD1 cut with Fnu4HI; lane 4, pCAD1 cut with LlaDII; lane 5, pXE1; lane 6, pXE1 cut with Fnu4HI; lane 7, pXE1 cut with LlaDII; lane 8, 1-kb ladder (Gibco BRL). The digestions were performed for 3 h at 37°C in the cases of EcoRI and Fnu4HI and at 30°C in the case of LlaDII.

In vivo expression of the LlaDII genes.

The isoschizomeric R/M system Bsp6I from Bacillus sp. strain RFL6 showed phage restriction against phage λvir in E. coli RR1 (40). We therefore evaluated the functional expression of the LlaDII genes in E. coli Sure.

Chromosomal DNAs from both L. lactis LM2301 and E. coli Sure containing pCAD1 were protected from restriction by LlaDII and Fnu4HI but not from that by EcoRI (data not shown). The XhoI-EcoRI fragment from pCAD1 was cloned into pBluescript SKII+, resulting in pXE1. This plasmid was not digested by LlaDII or Fnu4HI after 3 h of incubation (Fig. 3), indicating that plasmid pXE1 carries the LlaDII methylase gene and that llaDIIM is expressed in E. coli.

Plasmid pCAD1 in E. coli Sure had an EOP of 1 against phage λb2. Additionally, it was not possible to detect an active LlaDII endonuclease in the crude lysate from E. coli Sure(pCAD1). Both of these results indicated that R · LlaDII is not active in E. coli Sure. Plasmid pCAD1 purified from E. coli Sure encoded an active LlaDII R/M system when transformed back to L. lactis LM2301, demonstrating that the inability of the LlaDII endonuclease to restrict phage λb2 was not due to a mutation.

Gene organization and DNA sequence.

The PstI-EcoRI insert of pCAD1 was subcloned and sequenced on both strands, and the complete nucleotide sequence, comprising 2,355 bp, is presented in Fig. 4. The average G+C content of the insert is 31.9%. The DNA sequence contained two major open reading frames (ORF1 and ORF2) of 543 and 954 bp with coding potentials of 180 and 317 amino acids, respectively. ORF1 and ORF2 are arranged tandemly and separated by a 105-bp intergenic region. ORF1 showed 54% identity to the endonuclease gene of the Bsp6I R/M system, while ORF2 showed 65% identity to the corresponding methylase gene. The Bsp6I R/M system is from Bacillus sp. strain RFL6 (40, 57).

The DNA sequence upstream of ORF1 contained a putative ribosome binding site (AAGGTGA) 4 bp from the presumed start codon TTG. Putative promoter signals were an extended −10 sequence (T-TG-TAAAAT) and a −35 sequence (TTTAGA) separated by 17 bp. Since TTG is a rare start codon, primer extension was performed. This indicated that transcription starts with an A (or, alternatively, a G) 28 bp upstream of the putative translation start, TTG (Fig. 4 and 5). The results of the primer extension do not exclude the proposed start codon. In addition, a comparison with R · LlaDII and R · Bsp6I (40) showed that the endonucleases have the same putative translation start positions and that they are of equivalent length.

FIG. 5.

Mapping of the 5′ end of the LlaDII endonuclease transcript. The oligonucleotide primer for primer extension is complementary to nucleotides 892 to 858. The retention time of the primer extension peak is about 169 min, indicating that the transcription starts with an A or, alternatively, a G.

ORF2 has the typical start codon ATG. A possible ribosome binding site (AGGAGAA) was located 5 bp upstream of ATG. A putative −10 sequence (TATACT) and −35 sequence (TTGCGG) were separated by 17 bp. Interestingly, the intervening sequence and part of the putative −35 sequence contain two tandem LlaDII recognition sites (GCGGC and GCCGC) separated by 2 bp and forming a putative stem-loop structure with 5 bp in the stem and 2 bp in the loop. Another putative stem-loop structure consisting of 5 bp in the stem and 2 bp in the loop was found upstream but with no associated recognition sites. These features are indicated in Fig. 4.

Analysis of the amino acid sequence.

The predicted amino acid sequence for ORF1 was 41% identical to that of R · Bsp6I, while the predicted amino acid sequence for ORF2 showed 60% identity to that of M · Bsp6I. These comparisons correlated with the methylase activity of plasmid pEE1 and with the resistance of plasmid pXE1 to digestion with LlaDII and Fnu4HI and indicated that ORF2 encodes the methylase M · LlaDII and that ORF1 encodes the endonuclease R · LlaDII.

A catalytic sequence motif characteristic of endonucleases, PDX₃₃EXK, was observed in R · LlaDII. The corresponding sequence for R · Bsp6I is PEX₃₂EXK. Whether they have any effect is unknown, but the motifs are longer than those observed previously (PDX_10-30[D/E]XK) (2).

On the basis of a BLAST search, 10 bacterial methylases with highest similarity to M · LlaDII were chosen for multiple-sequence alignment. These enzymes, including M · LlaDII and M · Bsp6I, have a common architecture comprising 10 motifs occurring in an invariant order as previously described (35, 36, 50). The 10 motifs are shown in Fig. 4.

DISCUSSION

A 8.9-kb plasmid, pHW393, encoding the type II R/M system LlaDII was isolated from L. lactis subsp. cremoris W39. The endonuclease was partially purified and was found to recognize and cleave the DNA sequence 5′-GC↓NGC-3′, where N is A, C, G, or T and the arrow indicates the cleavage site. Thus, LlaDII is an isoschizomer of the R/M systems Bsp6I (40), Fnu4HI (38), and BsoFI (11). The LlaDII R/M system cloned into the E. coli-L. lactis shuttle vector pCI3340 in L. lactis LM2301 and L. lactis SMQ86 confers resistance to species of the small, isometric-headed phages 936 and P335 and to the prolate-headed phage c2, as shown by their reduced EOPs (Table 3). LlaDII is an efficient R/M system and probably is partially responsible for the strong phage resistance shown by L. lactis subsp. cremoris W39. Apart from LlaDII, L. lactis subsp. cremoris W39 contains another type II R/M system, designated LlaDI (48), and presumably other bacteriophage defense mechanisms.

In general, the small isometric-headed phages of the P335 and 936 species were more affected by the LlaDII restriction than the prolate-headed phage c2. This is presumably due to their larger genomes and correspondingly numerous recognition sites. As reported previously, the EOP decreases logarithmically as the number of sites in the viral DNA molecule increases (44, 66). The genome sequence of phage c2 contains six LlaDII sites (39), giving a restriction of 2 log units on L. lactis LM2301(pCAD1). This is in the same range as the restriction of phage sk1, which contains five LlaDII sites (accession no. AF011378). The variation of restriction of the 936 phages may reflect differences in the number of LlaDII recognition sites in the phage genomes. Digestion of phage jj50 and p2 DNAs with LlaDII and Fnu4HI indicated that they contain more restriction sites than phage sk1 (data not shown), which could account for the more severe restriction of the two former phages. Comparison of the restriction of the 936 phage species to the restriction of the P335 phage species indicates that LlaDII restriction against the newly emerged P335 phages is more efficient than that against the more common 936 phages. This phenomenon has been described earlier and is explained by an unusually high number of type II endonuclease sites in the P335 phages compared to the more common lytic phages of the 936 species (43).

The LlaDII region harbors two tandemly arranged genes, llaDIIR and llaDIIM, encoding a restriction endonuclease and a methylase, respectively. llaDIIM on the plasmids pEE1 and pXE1 was expressed in L. lactis LM2301 and E. coli XL1-Blue MRF′, respectively, suggesting that llaDIIM could be transcribed as a monocistronic mRNA. However, further experiments will be needed to identify transcriptional units. The endonuclease of LlaDII was apparently not functionally expressed from its natural promoter in E. coli XL1-Blue MRF′, as it was not possible to detect endonuclease activity in the E. coli XL1-Blue MRF′(pCAD1) crude lysate. This was supported by the observation that phage λb2 was not restricted following infection of cells harboring pCAD1. When transformed back to L. lactis LM2301, pCAD1 expressed R/M activity, demonstrating that mutations were not responsible for the abolished phage restriction in E. coli XL1-Blue MRF′. Conversely, bsp6IR was cloned and expressed in E. coli RR1. However, this was possible only after premethylation of the recipient strain DNA to prevent the possible suicide effect of introducing the complete R/M system (40). Differences in the promoter signals of llaDIIR and bsp6IR may cause this difference.

The DNA sequence of the LlaDII R/M system is highly related to that of Bsp6I, which is its only sequenced isoschizomer (40, 51). In both R/M systems the endonuclease gene precedes the methylase gene. The intervening regions of 105 and 99 bp, respectively, contain two recognition sites (5′-GCGGC-3′ and 5′-GCCGC-3′) forming a putative stem-loop structure with 5 bp in the stem and 2 bp in the loop. The stems are identical, while the loops vary. Both stem-loops are situated in exactly the same position as part of the presumed −35 sequence and part of the intervening region between the −35 and −10 sequences. We speculate that the recognition sites forming the stem-loop structure may be part of a regulation system controlling the expression of the methylase gene, for example, by methylation status. It would therefore be interesting to establish whether the operons of other isoschizomers show the same pattern as in the LlaDII and Bsp6I R/M systems. Based on sequence information for several R/M systems in the REBASE and GenBank databases, this phenomenon is exceptional, even though repeats or recognition sites preceding the endonuclease and methylase genes occur (51). Examples of this are (i) an inverted repeat in front of the SsoII methylase and endonuclease genes (31) and (ii) one in front of the SinI methylase and endonuclease genes (but neither is associated with recognition sites) (32), (iii) two recognition sites of the PaeR7I R/M system situated about 30 bp upstream of the start codon of the methylase gene (but neither forms an inverted repeat) (59), and (iv) two FokI recognition sites situated immediately upstream of the start codon of the FokI methylase but without forming an inverted repeat or part of it. In this context the translation start of the FokI endonuclease is the most interesting example, since two recognition sites are situated in front of the start codon, and both of them are contained in an inverted repeat (34). Removal of this stem-loop structure was essential for overproduction of the FokI endonuclease, indicating the presence of a regulation mechanism (33). Finally, the second stem-loop in the LlaDII operon was not conserved in Bsp6I.

The G+C content of the LlaDII R/M system is 31.9%, which is lower than the usual lactococcal G+C content, which ranges from 34.8 to 35.6% as determined from melting temperature with the type strain L. lactis subsp. cremoris NCDO 607 (54). This phenomenon has been recognized in other bacteriophage resistance mechanisms found in lactococci (10, 15, 46, 61). Despite the low G+C content in lactococcal DNA, recognition sequences from lactococcal R/M systems published so far have demonstrated a preference for G+C-rich restriction enzyme recognition sequences (ScrFI [14], LlaI [41], and LlaDII) or for equal contents of G+C and A+T (LlaDCHI [44], LlaAI [47], and LlaBI [47]).

Although all of the above-mentioned isoschizomeric endonucleases recognize and cut the unmethylated sequence 5′-GCNGC-3′, the methylation sensitivities might differ. It has been reported that the Fnu4HI endonuclease is unable to cut the methylated sequences 5′-G^mCGGC-3′ and 5′-GCGG^mCG-3′, while only the double-methylated sequence 5′-G^mCGG^mCGG-3′ escapes cleavage by the BsoFI endonuclease (11). Earlier experiments showed that BsoFI was not able to digest LlaDII-methylated plasmids, implying similar methylation patterns for these two isoschizomeric R/M systems (data not shown).

The protein primary structure of the LlaDII methylase has the same overall organization as those of the 11 selected m⁵C methylases. The proteins contain 10 conserved core sequences among variable regions. These motifs are indicated in Fig. 4. The core sequences apparently comprise components of the methylation reaction that are common to all of the enzymes (35, 36, 50). This suggests that M · LlaDII is an m⁵C methylase, modifying the fifth carbon of cytosine to yield 5-methylcytosine, as found with its isoschizomers. The key catalytic residue in the methylation reaction is cysteine in the Pro-Cys motif found in motif IV. The Pro-Cys motif, conserved for all m⁵C methylases, can also be found in M · LlaDII. In contrast, this motif is not found in the N⁴C or the N⁶A methylases (67). Among the variable regions, one differs from the others due to its remarkable length. This is the proposed variable region responsible for target recognition, also known as the target recognition domain (TRD). The size varies and is dependent on the target of recognition and the degree of multispecificity (65). The TRD of M · LlaDII is one amino acid longer than that of M · Bsp6I. The overall identity of the two methylases is 60%, and their TRDs show 57% identity. Conversely, TRDs of the other methylases vary considerably (data not shown), as expected with nonisoschizomers.

The results presented here show that the LlaDII R/M system on its native plasmid, as well as cloned in a vector, functions as a classical R/M system and restricts phages. Preliminary experiments (not presented here) indicate that LlaDII together with other bacteriophage resistance mechanisms shows an additive resistance effect with regard to EOP determination and the Heap-Lawrence test (19).