|Home | About | Journals | Submit | Contact Us | Français|
A nomenclature is described for restriction endonucleases, DNA methyltransferases, homing endonucleases and related genes and gene products. It provides explicit categories for the many different Type II enzymes now identified and provides a system for naming the putative genes found by sequence analysis of microbial genomes.
There are three main groups of restriction endonucleases (REases) called Types I, II and III (1,2). Since 1973, REases and DNA methyltransferases (MTases) have been named based on an original suggestion by Smith and Nathans (3). They proposed that the enzyme names should begin with a three-letter acronym in which the first letter was the first letter of the genus from which the enzyme was isolated and the next two letters were the first two letters of the species name. Extra letters or numbers could be added to indicate individual strains or serotypes. Thus, the enzyme HindII was one of four enzymes isolated from Haemophilus influenzae serotype d. The first three letters of the name were italicized. Later, a formal proposition for naming the genes encoding REases and MTases was adopted (4). When there were only a handful of enzymes known, these schemes were very useful, but as more enzymes have been found, often from different genera and species with names whose three-letter acronyms would be identical, considerable laxity in naming conventions has appeared. In addition, we now know that each major type of enzyme can contain sub-types. This especially applies to the Type II enzymes, of which more than 3500 have been characterized (5). In this paper we revisit the naming conventions and outline an updated scheme that incorporates current knowledge about the complexities of these enzymes. We describe a set of naming conventions for REases and their associated MTases. Since the homing endonucleases (6) have been named in an analogous fashion, we propose that similar guidelines be applied to that group of enzymes. Finally, it is important to realize that the aim of this document is to provide a nomenclature for these enzymes, not to provide a rigorous classification.
First, we introduce a number of general changes, standard abbreviations and definitions that are recommended for use.
The original subdivision of Types I, II and III will be maintained and a new Type IV added to accommodate a class of methyl-dependent restriction enzymes. The previously proposed candidates for new types, such as Eco57I and GsuI, will be incorporated as subtypes of existing Type II enzymes.
The key characteristics of the Type I R-M systems are that these enzymes are multisubunit proteins that function as a single protein complex and usually contain two R subunits, two M subunits and one S subunit (10). The symbol for Type I systems is hsd, thus the genes are hsdR, hsdM and hsdS, and their protein products are HsdR, HsdM and HsdS, respectively. The protein products can be abbreviated by omitting Hsd. The S subunit is the specificity subunit that determines which DNA sequence is recognized. The R subunit is essential for cleavage (restriction) and the M subunit catalyzes the methylation reaction: in all known cases the methylated base formed is m6A. When Type I enzymes act on unmethylated substrates, they function mainly as REases (they may also methylate unmodified sites with a low probability) and have an absolute requirement for ATP during cleavage. They cleave the DNA at variable positions away from their recognition sequence. The location of the cleavage sites is determined by either the collision and stalling of two such complexes during translocation along a DNA chain, or the stalling of a single enzyme on a single-site circular substrate following DNA translocation. The biochemical nature of the termini produced upon cleavage is unknown and the enzymes do not turn over in the cleavage reaction. In contrast, when these complexes encounter a hemimethylated substrate, in which one strand of the recognition sequence is methylated, as would occur immediately after DNA replication of a fully methylated substrate, then the complex functions as a DNA MTase, using S-adenosylmethionine (AdoMet) as the donor of the methyl group. A complex of two M subunits and one S subunit is fully functional as an MTase. Probably the best known Type I enzyme is EcoKI (11). The REase is referred to as R.EcoKI or EcoKI, but it is important to remember that it is also an MTase. The MTase complex of two HsdM and one HsdS is referred to as M.EcoKI. When referring to phenotypes the preferred convention is rKI+ mKI+ etc.
Four sub-categories of Type I enzymes (A, B, C and D) are in common use (12). These are based on genetic complementation and their use will be continued. If experimental evidence defines new subtypes, then additional letters may be used as suffixes to describe them. A number of artificially created hybrid enzymes have been described (13), which often include those with new specificities. These should be named as deemed appropriate, but without a Roman numeral at the end.
The Type II REases recognize specific DNA sequences and cleave at constant positions at or close to that sequence to produce 5′-phosphates and 3′-hydroxyls. Usually they require Mg2+ ions as a cofactor, although some have more exotic requirements (see below). They may act as monomers, dimers or even tetramers and usually act independently of their companion MTase. The MTases usually act as monomers and transfer a methyl group from the donor S-adenosyl-l-methionine directly to double-stranded DNA and form m4C, m5C or m6A. Because of the interest in these Type II REases for recombinant DNA technology, more than 3500 have been characterized (5). Given the assay that is used to find them, which detects any activity yielding a consistent DNA fragmentation pattern, it is no surprise that they come in a large variety of ‘flavors’. Early on it was recognized that while then-normal Type II enzymes recognized palindromic sequences and cleaved symmetrically within them, the Type IIS enzymes cut outside their normally asymmetric sequences and differed in other interesting ways (14). We now know of additional enzymes that cleave on both sides of their recognition sequence (e.g. BcgI), are activated by AdoMet (e.g. Eco57I), interact with two copies of their recognition sequence (e.g. EcoRII) or have unusual subunit structures (e.g. BbvCI).
These additional kinds of enzymes will be considered subdivisions of Type II. It should be recognized that for the purposes of nomenclature some enzymes would fall into more than one subdivision. Specifically, some of the criteria are based on the sequence cleaved and others on the structure of the enzymes themselves, so not all subdivisions are mutually exclusive, e.g. BcgI is both Type IIB and IIH. Type IIS enzymes, originally designated as enzymes with cleavage sites shifted away from their recognition sequence (4), will be retained, but a new Type IIA will be defined that includes all Type II REases that recognize asymmetric sequences. A new Type IIP will be used to designate the enzymes that recognize symmetric sequences (palindromes).
The overriding criterion for inclusion as a Type II enzyme would be that it yields a defined fragmentation pattern and cleaves either within or close to its recognition sequence at a fixed site or with known and limited variability. In general, the Type II REases and their associated MTases are separate, independent enzymes, but in several classes (e.g. IIB, IIG and IIH) the R and M genes are fused into a single composite gene. The nomenclature for the subtypes of the Type II enzymes currently known is shown below. It should be noted that these designations are not intended to be exclusive, but rather to permit enzymes with common characteristics to be referred to as a group. Conservation of structural domains with associated enzymatic activities is observed between different classes of Type II enzymes and also between other types of R-M enzymes.
The Type II subdivisions are summarized in Table Table11 and described in more detail below.
This would be used as a generic description for all enzymes that recognize symmetric sequences, often termed palindromes, and cleave at fixed symmetrical locations either within the sequence or immediately adjacent to it. The recognition sequences and cleavage sites of these enzymes should be represented as in the following example: EcoRI: G↓AATTC. In full double-stranded form this corresponds to:
5′ G↓A A T T C
3′ C T T A A↑G
Note that enzymes such as SinI (recognition sequence: GGWCC), BglI (recognition sequence: GCCNNNN↓NGGC) and HindII (recognition sequence: GTYRAC) belong to Type IIP because the recognition mechanism still involves a symmetric homodimer.
This would be used as a generic designation for any Type II enzymes that recognize asymmetric sequences irrespective of whether they cleave away from the sequence or within the sequence. Typically these systems have one REase gene and two MTase genes, one to modify each strand of the asymmetric recognition sequence. However, occasionally two R genes are found as with Bpu10I (15), or both M genes are fused as with M.FokI (16). When more than one R or M gene is present the genes and their protein products should be named with either an Arabic 1 or 2 in the prefix of the name. Thus, the two MTases of the SapI system would be named M1.SapI and M2.SapI if the proteins are being referred to, or sapIM1 and sapIM2 for the genes. However, the two subunits of the Bpu10I REase would be designated R.Bpu10IA and R.Bpu10IB and their genes bpu10IAR and bpu10IBR. The recognition sequences and cleavage sites of the Type IIS REases should be represented as in the following example:
HphI: GGTGA(8/7) where the first numeral in the parentheses indicates the position of cleavage on the strand written and the second numeral indicates the cleavage position on the complementary strand. In full double-stranded form this corresponds to:
Note that when recognition sequences are assigned, the convention is to write the single-stranded sequence such that cleavage lies downstream of the sequence. If cleavage takes place within the sequence, then the single-strand designation is always written so that the sequence of the strand is first alphabetically.
This would be used for enzymes that cleave on both sides of the recognition sequence. At present there are many well defined members of this class (AloI, BplI, Bst44I, BaeI, BcgI, BsaXI, Bsp24I, CjeI, CjePI, HaeIV, Hin4I and PpiI). In this case the recognition sequence and cleavage sites should be represented as exemplified for BcgI:
BcgI—recognition sequence: (10/12)CGANNNNNNTGC(12/10)
Here the (10/12) preceding the recognition sequence indicates that cleavage occurs 10 bases in front of the sequence on the strand written and 12 bases before the sequence on the complementary strand. The (12/10) following the recognition sequence indicates cleavage 12 bases after the recognition sequence on the strand written and 10 bases after the sequence on the complementary strand. In double-stranded form this would be written:
This would be used as a generic term for all enzymes that have a hybrid structure containing both cleavage and modification domains within a single polypeptide. Examples include all of the Type IIB, IIG and some Type IIH enzymes.
This would be used for enzymes that interact with two copies of the recognition sequence, one being the actual target of cleavage, the other being the allosteric effector. The best studied examples are EcoRII (17) and NaeI (18). FokI, MboII and Sau3AI were found to exhibit similar properties. Other enzymes such as Acc36I, AtuBI, BsgI, BpmI, Cfr9I, Eco57I, HpaII, Ksp632I, NarI, SacII and SauBMKI are likely to be members of this group because they are reported to be stimulated by oligonucleotide duplexes containing the specific recognition site.
This would be used for enzymes that interact with, and cleave coordinately, two copies of their recognition sequence. Examples include BspMI, Cfr10I, NgoMIV, SfiI and SgrAI.
This would be used for enzymes that have both R and M domains fused to form single polypeptides and that may be stimulated or inhibited by AdoMet, but otherwise resemble Type II enzymes. These include Bce83I, BseMII, BseRI, BsgI, BspLU11III, Eco57I, GsuI, MmeI and Tth111II. The recognition sequences may or may not be asymmetric. Thus, both Type IIA and Type IIP enzymes may be of Type IIG.
This would be used for enzymes that contain genetic features resembling Type I enzymes, but biochemically behave as Type II enzymes. At present three examples have been characterized: AhdI and PshAI, both of which comprise a three gene system akin to that of a typical Type I enzyme (G.G.Wilson, unpublished results), and BcgI, which is a two gene system. Several hypothetical systems have gene organizations that resemble that of BcgI.
This would be used for DpnI and similar enzymes that recognize a specific methylated sequence in DNA and cleave at a fixed site. Note that the methyl-dependent enzymes such as McrA, McrBC are not considered members of this subclass, because they do not have well defined recognition sequences and cleavage sites. They are included within the Type IV enzymes.
This would be used for Type IIA enzymes that cleave at least one strand of the DNA duplex outside of the recognition sequence (i.e. cleavage is shifted relative to the recognition sequence). Note that for some enzymes, such as BsmI (recognition sequence: GAATGC), cleavage of the strand written takes place outside of the recognition sequence, whereas cleavage of the complementary strand takes place within the recognition sequence. This is still considered a Type IIS enzyme. However, in most cases both strands are cleaved away from the recognition sequence, which therefore remains intact. These were the earliest sub-classes of the Type II restriction enzymes to be recognized (14).
This would be used for enzymes that are composed of heterodimeric subunits. This subtype includes enzymes like BbvCI, Bpu10I and BslI.
Two types of nicking enzymes are known. One type includes those that behave functionally like REases, but cleave only one strand of the DNA substrate. These enzymes should be named with the prefix N and their recognition sequences should be written such that the strand displayed is the strand nicked. Thus, N.BstSEI has the recognition sequence: GAGTCNNNN↓ which is abbreviated to GAGTC(4). Similarly, the mutants of AlwI and MlyI that have interrupted the dimerization function, and which have become nicking enzymes, are named N.AlwI (19) and N.MlyI (20). For enzymes such as Bpu10I, where the wild-type REase has two subunits, each of which nicks a different strand, the mutant nicking enzymes made by inactivating one or the other subunit should be named Nt.Bpu10I for the enzyme that nicks the top strand of the normal recognition sequence and Nb.Bpu10I for the enzyme that nicks the bottom strand.
In full double-stranded format Nt.Bpu10I would recognize
5′ C C↓T N A G C
3′ G G A N T C G
while Nb.Bpu10I would recognize
5′ C C T N A G C
3′ G G A N T↑C G
Alternatively this may be written
5′ G C↓T N A G G
3′ C G A N T C C
A single-stranded representation of their recognition sites would be Nt.Bpu10I (recognition sequence: CC↓TNAGC) and Nb.Bpu10I (recognition sequence: GC↓TNAGG or CCTNA↑GC). Note that the use of ↑ always denotes cleavage of the lower strand.
A second type of nicking enzyme is found exclusively in association with m5C-MTases, where it serves to nick the G/T mismatches that can result from deamination of m5C within the recognition sequence of the MTase. The best studied of these is the Vsr protein that accompanies the Dcm MTase of E.coli K-12, M.EcoKDcm. Vsr recognizes the specific G/T mismatch that occurs if there is deamination of the methylated cytosine residue within the context of the CCWGG recognition sequence of M.EcoKDcm (21). These kinds of mismatch nicking enzymes are named with the prefix V and should be given the acronym of the MTase gene with which they are associated. Thus, Vsr, the product of the V gene that overlaps with the gene for M.EcoKDcm, is systematically named V.EcoKDcm. However, the trivial name Vsr, which was originally designated for this protein, is an acceptable synonym. For other V genes and their products the systematic names are preferred. Thus, V.HpaII is the preferred name for the mismatch nicking endonuclease that accompanies M.HpaII.
Some R-M systems are found to have an additional gene that encodes a protein involved in the control of expression of the R gene. The best studied examples are the PvuII and BamHI systems, where the products of the C genes, C.PvuII (22) and C.BamHI (23), serve as transcriptional activators; this prevents the expression of the R genes following transfer of the systems into naive hosts, until such time as C protein has accumulated and methylation is sufficient to provide protection against what would otherwise be the deleterious action of the REase.
These systems are composed of two genes (mod and res) encoding protein subunits that function either in DNA recognition and modification (Mod) or restriction (Res) (10,24,25). Both subunits are required for restriction, which also has an absolute requirement for ATP hydrolysis. For DNA cleavage, the enzyme must interact with two copies of a non-palindromic recognition sequence and the sites must be in an inverse orientation in the substrate DNA molecule. Cleavage is preceded by ATP-dependent DNA translocation as with the Type I REases. The enzymes cleave at a specific distance away from one of the two copies of their recognition sequence. The Mod subunit can function independently of the Res subunit to methylate DNA: in all known cases the methylated base formed is m6A and full modification is actually hemimethylation. This is not deleterious because of the requirement for two unmodified sites in inverse repeat orientation for cleavage. DNA replication puts all of the unmodified sites in the same orientation. The best-known examples of Type III enzymes are EcoP1I and EcoP15I. Putative Type III R-M systems are easily recognized because of their similarity at the sequence level. When naming the genes for these enzymes the mod gene of EcoP1I would be systematically named ecoP1Imod, but the abbreviation mod is acceptable when it does not result in confusion.
These systems are composed of one or two genes encod ing proteins that cleave only modified DNA, including methylated, hydroxymethylated and glucosyl-hydroxymethylated bases. Their recognition sequences have usually not been well defined except for EcoKMcrBC, which recognizes two dinucleotides of the general form RmC (a purine followed by a methylated cytosine—either m4C or m5C) and which are separated by anywhere from 40 to 3000 bases. Cleavage takes place ~30 bp away from one of the sites. The best studied example at both the genetic and biochemical level is EcoKMcrBC of E.coli (26,27), but on the basis of sequence similarity it is likely that there are many such systems in other bacteria and archaea. As with the genes of the Type I and Type III systems, the abbreviations McrBC for the enzyme and mcrBC for the gene are acceptable.
Hypothetical REases and DNA MTases can often be found by similarity searching in DNA sequences or their presence may be inferred when specific sequences in plasmid or bacterial DNAs are found to be methylated. It is convenient and useful to be able to refer to such hypothetical enzymes by name. The following convention for naming these enzymes is proposed. They should be named as though they were normal R-M systems, but should carry the suffix ‘P’ to indicate their putative nature. Once biochemical or unequivocal genetic activity, such as phage restriction, is demonstrated the suffix ‘P’ and any open reading frame (ORF) designations can be dropped allowing the main element of the name to be retained. Furthermore a Roman numeral should be included to indicate whether it is the first, second, third, etc. enzyme to be found in that organism. Note that the P extension should remain with the gene until such time as a gene product has been demonstrated to be functional.
This ‘P’ convention is illustrated with genes from H.influenzae serotype d. Two Type II REases, HindII and HindIII, and their associated MTases had been characterized biochemically (28–31). One Type I system had been demonstrated genetically (32) and the MTase, presumably associated with this system, had been partially characterized biochemically (30,31). In the genome there are two putative Type I systems, although only one has a complete set of intact genes (33). The intact system therefore carries the designation HindI. In addition to these three systems, there was also known to be a Dam-like MTase, now called M.HindDam. However, also in the genome are putative m5C-MTase and REase genes (genes HI1040 and HI1041) that show high similarity to the known R-M system, HgiDI (34). The MTase encoded by HI1041 leads to a functional protein with specificity identical to that of M.HgiDI (R.D.Morgan, J.Patti and R.J.Roberts, unpublished results). It is therefore named M.HindV. However, the adjacent gene for the putative endonuclease is inactive and so it is named HindVP. One other R-M system can also be seen in the genome, this time encoding a Type III system. Neither the R nor the M gene have yet been demonstrated to be active and so these are named HindORF1056P and M.HindORF1056P. If they are shown to be active they would be renamed HindVI and M.HindVI. The convention here is to name the system after the ORF encoding the MTase gene. This is to ensure that the two genes are given names that indicate they are part of the same R-M system.
Homing endonucleases have been classified into four families according to conserved sequence motifs. These are the LAGLIDADG, GIY-YIG, H-N-H and His-Cys box families (35). Nomenclature of the homing endonucleases is patterned after that of REases, with a three-letter genus-species designation, followed by a Roman numeral (6). Whereas intron endonucleases are characterized by the prefix I- (for intron), the intein endonucleases are characterized by the prefix PI- (for protein insert), and where the endonuclease is not intron- or intein-encoded, the prefix is F- (for freestanding). The systematic nomenclature does not preclude maintaining historic names. Counter to the original conventions proposed (6), the above nomenclature will extend to putative homing endonucleases without demonstrated catalytic activity. As with hypothetical REases, the suffix P will be used to denote the putative nature of the assignment, and the P will be dropped once nuclease activity has been confirmed. Hybrid homing endonucleases will be preceded by the prefix H-, followed by the authors’ designation, e.g., an I-DmoI/I-CreI chimera could be H-DreI, or an I-TevI/I-BmoI hybrid could be H-TevBmo. Those homing endonucleases that have been characterized biochemically will continue to be listed within REBASE (5).
The authors of this proposal have all agreed to follow these recommendations and it is hoped that other authors and journals will also adhere to these conventions. If further changes become appropriate, then REBASE (5) should be consulted for the latest modifications and practices.