Article

ISSN 0006-2979, Biochemistry (Moscow), 2025, Vol. 90, No. 4, pp. 502-512 © Pleiades Publishing, Ltd., 2025.

Published in Russian in Biokhimiya, 2025, Vol. 90, No. 4, pp. 559-570.

502

Restriction–Modification Systems

Specific toward GGATC, GATGC, and GATGG.

Part1. Evolution and Ecology

Sergey Spirin

1,2,3,a

*, Ivan Rusinov

, Olga Makarikova

Andrei Alexeevski

1,3

, and Anna Karyagina

1,5,6

Belozersky Institute of Physico-Chemical Biology, Lomonosov Moscow State University,

119234 Moscow, Russia

Higher School of Economics National Research University, 109028 Moscow, Russia

NRC “Kurchatov Institute” - SRISA, 117218 Moscow, Russia

Moscow Institute of Physics and Technology, 117303 Moscow, Russia

Gamaleya National Research Center for Epidemiology and Microbiology,

Ministry of Healthcare of the Russian Federation, 123098 Moscow, Russia

All-Russia Research Institute of Agricultural Biotechnology, 127550 Moscow, Russia

e-mail: sas@belozersky.msu.ru

Received January 21, 2025

Revised March 19, 2025

Accepted March 25, 2025

Abstract— The article presents the results of studies on the evolution of proteins from restriction–modifica-

tion systems consisting of restriction endonucleases with the REase_AlwI family domain and either two DNA

methyltransferases, each with the MethyltransfD12 family domain, or a single DNA methyltransferase with

two domains of this family. It was found that all such systems recognized one of the three DNA sequences,

namely GGATC, GATGC or GATGG. Based on the sequence similarity, restriction endonucleases of these sys-

tems could be attributed to three clades that unambiguously corresponded to the RM system specificity. The

DNA methyltransferase domains of these systems were classified into two groups based on sequence simi-

larity, with the two domains of each system belonging to different groups. Within each group, the domains

were attributed to three clades according to their specificity. An evidence of multiple interspecific horizontal

transfer of entire restriction-modification systems has been found, as well as the transfer of individual genes

between the systems (including the transfer of one of DNA methyltransferases accompanied by changes in its

specificity). Evolutionary relationships of DNA methyltransferases from the studied systems with other DNA

methyltransferases, including orphan DNA methyltransferases, have been revealed.

DOI: 10.1134/S0006297925600115

Keywords: restriction–modification system, molecular evolution, DNA methyltransferase, restriction endonucle-

ase, horizontal gene transfer

* To whom correspondence should be addressed.

INTRODUCTION

Restriction–modification (RM) systems are de-

fence systems of prokaryotes that protect these or-

ganisms against introduction of foreign DNA, in par-

ticular DNA of bacteriophages [1]. RM systems are

traditionally classified into several types [2], of which

Type II is the most studied. Each Type II system con-

tains genes coding for proteins with two enzymat-

ic activities: a restriction endonuclease (REase) that

recognizes and hydrolyzes a specific DNA sequence,

and at least one DNA methyltransferase (MTase) that

methylates host DNA within the target sequence, thus

preventing its recognition by the REase. MTases meth-

ylate DNA either by cytosine bases to form C

-meth-

ylcytosine (5mC) or N

-methylcytosine (4mC), or by

RM SYSTEMS: EVOLUTION AND ECOLOGY 503

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

adenine bases to form N

-methyladenine (6mA). Most

TypeII RM systems recognize palindromic sequences,

although some of them (subtype IIA) have asymmetric

recognition sites. Typically, subtype IIA systems con-

tain not one, but two MTases. In the case when a sub-

type II RM system has only one MTase, this enzyme

contains two catalytic centres and is a fusion of two

MTase proteins. Two MTases are necessary to provide

methylation of both DNA strands in the asymmetric

site, which excludes the appearance of unmodified

sites after replication [3, 4].

REases and MTases constituting RM systems can

belong to different protein families (according to se-

quence homology). Earlier [5], we classified RM sys-

tems containing one REase and two MTases based on

the catalytic domain families identified in them by the

Pfam database tools [6] and investigated in detail the

evolution of RM systems consisting of one REase with

the NOV_C family domain and two 5mC MTases, each

with the DNA_methylase family domain. We showed

that these systems might have descended from a sin-

gle ancestral system of the same composition. Hori-

zontal gene transfer of entire systems has played a

major role in their evolution. We also observed the

evidence of relatively rare gene exchange events be-

tween the systems.

Here, we investigated the evolution of RM sys-

tems consisting of a REase with an RE_AlwI family

domain and either two 6mA MTases, each with a

MethyltransfD12 family domain, or a single fusion

MTase with two MethyltransfD12 family domains. All

systems with this domain composition, for which their

specificity has been determined, belonged to subtype

IIA and recognized one of the three DNA sequences:

GGATC/GATCC, GATGC/GCATC or GATGG/CCATC. And

vice versa, almost all RM systems with such confirmed

specificities had this domain composition, with very

few exceptions.

MATERIALS AND METHODS

The composition of the studied systems and the

sequences of MTases and REases were extracted from

REBASE [7], v.303 of 28.02.2023. The evolutionary do-

mains in protein sequences were identified accord-

ing to the Pfam database v.35 [6]. Protein sequenc-

es were aligned with Muscle [8]. Phylogenetic trees

were inferred with FastME [9], rooted at the midpoint,

and visualised with MEGA [10] or iTOL service [11].

Clustering of protein sequences was performed with

CD-HIT [12].

The list of MTases with confirmed methylation

sites was compiled from MTases satisfying the follow-

ing two conditions: (i)their names in REBASE did not

end with the letter ‘P’ (in REBASE, ‘P’ at the end of

protein name means ‘putative’); (ii) the page for the

MTase on the REBASE website stated that the nucle-

otide sequence methylated by the enzyme has been

confirmed by the host genome sequencing using the

PacBio technology.

MTases homologous to MTases from the stud-

ied RM systems were found among the sequences

of MTases with confirmed methylation sites using

BLASTP; the E-value threshold was 0.001. The search

for REases homologous to the studied REases was per-

formed in the same way among REases belonging to

RM systems containing MTases with confirmed meth-

ylation sites.

The contrast, i.e., the ratio of the observed num-

ber of sites in the genome to the expected one, was

calculated using the formula proposed by Burge etal.

[13] (see also [14]).

RESULTS

In this work, we investigated RM systems con-

taining REases of the Pfam family called ‘AlwI re-

striction endonuclease’ (Pfam ID RE_AlwI, Pfam AC

PF09491) and either two MTases of the family ‘D12

class N

adenine-specific DNA methyltransferase’

(MethyltransfD12, PF02086) or one MTase with two

MethyltransfD12 family domains. A total of 493 such

systems with two MTases and 227 such systems with

one MTase were found in REBASE v.303 (see Online

Resource1 for the list of identified systems). All these

systems belonged to Type II. The genes for most of

them were located on chromosomes; only 36 out of

these 720 RM systems were encoded in plasmids. For

some of the identified systems, REBASE indicated one

of the three recognition sites: GGATC (or GATCC on

the other chain), GATGC (GCATC), or GATGG (CCATC).

Out of 97 RM systems for which these sites were con-

firmed with PacBio, 91 had the domain composition

listed above (a REase with the RE_AlwI family domain

and two MethyltransfD12 family domains in one or

two MTases). Among the six exceptions (see Online

Resource  2), three RM systems consisted of one REase

with the RE_AlwI domain and one MTase with the

MethyltransfD12 domain (according to REBASE), but

next to the genes coding for these proteins there was

a gene coding for an MTase with the MethyltransfD12

domain, so we assumed that the composition of these

RM systems in REBASE was defined incorrectly. In the

remaining three cases, some of the three typical do-

mains was not detected in proteins of RM systems.

Phylogeny of REases and MTases. The phylo-

genetic tree inferred from the full-length REase se-

quences contained three clades, and REases with the

same annotated specificity always belonged to the

same clade. The phylogenetic tree presented in Fig.  1a

SPIRIN et al.504

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

Fig.  1. Phylogenetic trees for REs and MTases from three RM systems with two MTases and three systems with a single

(fusion) MTase. The recognition sequence (according to REBASE) is shown after the system name. a)  Phylogenetic tree of

REases. Enzymes from RM systems with the fusion MTase are marked with asterisks. b) Phylogenetic tree based on the

MTase domain sequences. Domains from individual MTases are named according to the corresponding MTases from REBASE;

‘N’ and ‘C’ designate N- and C-terminal domains, respectively, of the fusion MTases; A and B denote two MTase groups.

for six REases illustrates a general trend, so we were

able to predict with a high reliability the specificity

of all systems studied in our work.

Below, we refer to RM systems specific toward

GGATC/GATCC, GATGC/GCATC, and GATGG/CCATC as

‘red’, ‘blue’, and ‘green’, respectively (Fig.1). Only the

GGATC, GATGC, and GATGG variants were used to de-

note the recognition sites; the choice between the two

complementary variants was determined by the spec-

ificity of the two MTases predicted by us [see Part 2.

Functionality and Structure, Biochemistry (Moscow),

vol. 90, issue 4]. It should be noted that REases from

RM systems with the two-domain (fusion) MTase

(marked with asterisks in Fig. 1a) or two individual

MTases were not distinguished on the tree.

To study the evolution of MTases, we aligned sep-

arately N- and C-terminal domains from the fusion

MTases to the MTase domains from RM systems with

two MTases. The phylogenetic trees inferred from the

resulting alignments showed the two MTases from the

same RM system have evolved independently. Thus,

the tree of the MTase domains branched into two

clades, and for each of the studied systems, the two

MTase domains of same system always belonged to

different clades. Figure 1b illustrates this observation

for the MTases of the same six RM systems. Hencefor-

ward, we will refer to MTases of the upper and lower

clades (Fig.1b) as group A and B MTases, respectively.

Both A and B groups of MTases in Fig.1 were sub-

divided into three clades based on the enzyme spec-

ificity. The phylogenetic tree for all MTases with the

confirmed specificity mostly showed the same pattern,

with one exception in groupA (see “Gene recombina-

tion between RM systems” below).

Mutual arrangement of genes in RM systems.

Table 1 contains data on the mutual arrangement of

genes for all studied RM systems with the available

relevant information in REBASE. In all the cases, the

genes for A and B group MTases were located on the

same DNA strand, although their order could be differ-

ent. The REase gene was usually located on the same

strand as the MTase genes, but with some exceptions.

Table 1 demonstrates that the ‘red’ and ‘green’

RM systems possessed a typical mutual genes

RM SYSTEMS: EVOLUTION AND ECOLOGY 505

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

Table 1. Mutual arrangement of genes for RM systems recognizing GGATC, GATGC, and GATGG sequences

Gene order*

GATGG GATGC GGATC GATGG GATGC GGATC

Number of systems

†

Number of REase clusters

ABR 338/0 50/147 0/0 205/0 30/42 0/0

ABr 0/0 0/7 0/0 0/0 0/5 0/0

BAR 0/0 0/0 7/68 0/0 0/0 6/45

BAr 0/0 0/0 0/1 0/0 0/0 0/1

BRA 2/– 7/– 0/– 2/– 3/– 0/–

BrA 0/– 1/– 0/– 0/– 1/– 0/–

RAB 1/1 13/3 0/0 1/1 6/2 0/0

rAB 0/0 74/1 0/0 0/0 54/1 0/0

Total 341/1 145/158 7/69 208/1 94/50 6/46

*  A and B denote MTases or MTase domains (in two-domain MTases) from groups A and B, respectively; R and r denote

REase genes with the same or opposite orientation, respectively, as the MTase genes.

†

  The numbers before and after the slash sign correspond to the numbers of systems or clusters with separate or fusion

MTases, respectively; dash after the slash indicates that location of the REase gene between the MTase genes is impossible

for the systems with fusion MTases. The most typical variants are highlighted in bold.

  Number of clusters with 98% identity per 98% length of the shorter sequence for REases from RM systems with the cor-

responding gene arrangement.

arrangement characteristic for the system with each

particular specificity. The typical sequence for the

‘green” systems was (group  A  MTase)  →  (group  B

MTase)  →  (REase), with only four exceptions. Only one

‘green’ system had a fusion MTase. The typical gene

sequence for ‘red’ systems was (group  B MTase)  →

(group  A MTase)  →  (REase), with only one exception.

Most ‘red’ systems had fusion MTases. In contrast, no

dominant location of the REase gene was found for

the ‘blue’ systems, although genes for groupA MTases

often preceded genes for group  B MTases (as in ‘green’

systems) and there were about as many ‘blue’ systems

with the fusion MTases as with two individual en-

zymes. Analysis of REase phylogeny in the ‘blue’ RM

systems (Fig. 2) showed that during the evolution of

this group, fusion and separation of two MTases genes,

as well as rearrangement of MTase and REase genes,

have occurred repeatedly. The most parsimonious, and

thus most probable, scenario is as follows. The order

of the ancestral genes in the ‘blue’ systems was ABR,

but during the evolution, the transfer of the REase

gene has occurred several times with the formation of

the RAB or rAB arrangements. The BRA arrangement

was observed only in ‘blue’ systems that contained

MTase  A which was closer to the ‘green’ MTase  A (see

“Gene recombination between RM systems” below).

Distribution of RM systems across bacterial

taxa. The studied RM systems were found to be rep-

resented in 11 phyla and 21 classes of bacteria, al-

though their distribution across these taxa was very

uneven (see Online Resource  1). Closely related RM

systems were found in bacteria from different phyla,

while poorly related systems were present in closely

related bacteria (see phylogenetic trees for REases in

Online Resource 3). For example, among 70% REase

clusters with the GATGC specificity, there was a cluster

that included RM systems from 14 bacterial species

from six different phyla. For one of these species (the

livestock pathogen Mannheimia haemolytica), 62  RM

systems from different strains were represented in

REBASE, while the remaining 13 species possessed one

RM system each. For seven of these 13  RM systems,

REase sequences demonstrated less than 10% differ-

ence between each other and with sequences from

M.  haemolytica. In the same cluster, two RM systems,

whose proteins showed over 98% sequence similarity,

have been found in unrelated human oral microflora

bacteria, Streptococcus oralis (Bacillota) and Fusobac-

terium nucleatum (Fusobacteriota).

The distribution of RM systems across the strains

within species with a large number of different strains

with fully sequenced genomes (see Online Resource  4)

demonstrated that for about a half of the species for

which genomes of five or more strains had been com-

pletely sequenced (i.e., assembled up to whole chro-

mosomes), the RM system with one of the studied

SPIRIN et al.506

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

Fig.  2. Phylogenetic tree of REases from RM systems with the GATGC specificity (with the gene order). Cluster representa-

tives were selected based on 70% sequence identity. Letter F at the beginning of the name indicates fusion MTase; gene

order is shown with three letters (similar to the Table1). The numbers on the branches indicate the bootstrap support.

RM SYSTEMS: EVOLUTION AND ECOLOGY 507

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

specificities was found in the genome of only one

strain. Among genomes of 274 Streptococcus pneumo-

niae strains, RM systems with the GCATC specificity

were found in 15 genomes (5.5%). For only ten species,

such systems were found in half or more of the strains,

and in all these species except one (M.  haemolytica),

RM systems from different strains were highly simi-

lar to each other (>85% identity of complete protein

sequence). As for M.haemolytica, this was true for all

but one strain, which contained the Mha10208II system

relatively unrelated to the others (25 and 59% identity

of the REase and the fusion MTase, respectively, with

the corresponding enzymes from other strains).

Avoidance of RM system recognition sites in

genomes. A prolonged presence of an active RM sys-

tem often leads to the avoidance of its recognition site

in the host genome [15,  16]. Three bacterial species

(Bacteroides caccae, Campylobacter upsaliensis, Kin-

gella kingae) containing RM systems specific toward

GATGG and one species (Helicobacter cinaedi) with

the GATGC-specific system demonstrated a significant

underrepresentation of the corresponding recognition

sites in their genomes, i.e., the ratio of the observed

number of sites to the expected one was less than 0.8.

The underrepresentation of these sites was found in

the genomes of all strains of these species, although

in the case of K. kingae and H.  cinaedi, the genes for

the RM system were found in less than a half of the

strains. In particular, genes for RM systems with the

GATGC specificity were identified in three out of 12

genomes of H.  cinaedi strains, while other two ge-

nomes contained genes for RM systems with the GAT-

GG specificity. No avoidance of GATGG was observed

in any H.  cinaedi genome. An example of the opposite

situation is Campylobacter concisus, in which two (out

of 16) strains carried genes for the GATGC-specific RM

systems, while the genes for the GATGG-specific sys-

tems were absent in all the strains; at the same time

all 16 genomes demonstrated a strong underrepresen-

tation of GATGG, but not significant deviations of the

GATGC frequency from the expected one. The aver-

age contrasts across all genomes of each species (i.e.,

the ratios of observed word counts to the expected

ones) for the words GATGC, GATGG, and GGATC are

available in the Supplementary materials (Online Re-

source 4; columns G, H, and I).

Gene recombination between RM systems

(horizontal transfers of RM system components).

We found evidence of horizontal gene transfer be-

tween the RM systems with different specificities.

In particular, we identified several related RM systems

in which REases and MTases  B were similar to the

corresponding enzymes in the RM systems with the

GATGC specificity, but their MTases  A were more sim-

ilar to MTases  A of the RM systems with the GATGG

specificity. Among others, such RM systems included

Cup11541IV, Hfe11613I, and Hmu12714II, whose speci-

ficity toward GATGC has been confirmed with PacBio.

It is likely that the ancestor of these RM systems had

once acquired an ‘alien’ group A MTase, which has

changed its specificity from GATGG to GATGC. This

change in the specificity could occur either simultane-

ously with a mutation in the MTase gene (which led to

changes in the recognition DNA sequence) or through

a temporary weakening of the MTase specificity, for

example, to GATGS. The molecular basis of changes

in the enzyme specificity is discussed in Part2.Func-

tionality and Structure, Biochemistry (Moscow),

vol. 90, issue 4.

We also found evidence of repeated exchange of

components within each of the three groups of RM

systems with different specificity. For example, for the

three ‘blue’ RM systems (Teq529ORF1000P, Mha10208II,

and Mva1312II), the REs from the first two systems

were similar to each other (the score of the local

alignment constructed by the water program of the

EMBOSS package with default parameters was 1502)

but significantly less similar to the REase from the

third system (scores of 464.5 and 501, respectively).

MTases from the first and third systems were close to

each other (score, 1850), whereas MTase M.Mha10208II

was more distant from them (scores, 1512 and 1574).

This favours the suggestion that the ancestor of one

of these systems was formed from REase and MTase

genes that had originated from different RM systems

with the same specificity (GATGC). Several more sim-

ilar examples can be given for RM systems with each

of the three specificities. However, a detailed analysis

of the frequency of such events was obstructed by the

poor quality of the MTase phylogeny reconstruction:

the bootstrap support for many branches of the MTase

phylogenetic tree was below 50%.

Evolutionary relationships of RM systems spe-

cific toward GGATC, GATGC, and GATGG and RM

systems with homologous MTases. It appeared in-

teresting to study the evolution of systems specific

toward GGATC, GATGC, and GATGG in the context of

evolution of RM systems with other specificities and

homologous MTases and REs. The phylogenetic trees

of MTases with the recognition sites confirmed with

PacBio and close to the studied group A and B MTases

(see Materials and Methods) are shown in Fig.  3.

A large number of MTases specific toward GATC

were related to group B MTases, but did not form

a monophyletic group within them. Some of these

MTases formed sister with ‘red’ group  B MTases

(with the GGATC specificity), while others, includ-

ing the well-studied M.EcoKDam, occupied a more

basal position. Special attention should be paid to

M.EcoT4Dam. The specificity of M.EcoT4Dam toward

GATC has not been confirmed by the PacBio sequenc-

ing, but we added it to our set because this enzyme

SPIRIN et al.508

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

Fig.  3. Phylogenetic trees of group A (a) and group B (b) MTases of MethyltransfD12 family from the RM systems spe-

cific toward GGATC, GATGC, and GATGG and related enzymes. Branches with the bootstrap support less than 40% were

removed. Monophyletic groups of MTases with the same specificity were combined and indicated with diamonds; the

recognition site is given for each of these groups followed by parentheses with variants of RM system composition, where

M is for MTase, 2M is for two MTases, R is for REase, and MR for fusion bifunctional protein containing MTase and

REase domains. For RM systems with REases, the identifiers of catalytic domain families found in the REases are given

in the second set of parentheses; Unk means that no domains were identified by Pfam profiles in the REase sequence.

Thenumber of systems for which the recognition site has been experimentally confirmed by the PacBio technology is given

after the system composition. Group B MTase includes M.EcoT4Dam MTase, whose specificity toward GATC has not been

confirmed by PacBio.

RM SYSTEMS: EVOLUTION AND ECOLOGY 509

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

Fig.  4. Phylogenetic tree of REases close to REases specific toward GGATC, GATGC, and GATGG. The clades consisting of

REases of the same specificity are indicated by diamonds; the recognition sequence is followed by the number of REases

of this clade belonging to RM systems with MTases, whose specificity was confirmed by PacBio.

has been extensively studied (in particular, a number

of structures of its complexes with DNA were solved).

M.EcoT4Dam appeared to be closer to the ‘green’ and

‘blue’ group B MTases, whereas all MTases with the

GATC specificity confirmed by PacBio were closer to

the ‘red’ ones. A significant fraction of MTases with

the GATC specificity that were close to M.EcoKDam

[clade ‘GATC (M, M+R(DpnII)) 609’ in Fig.  3b] belonged

to the resident Dam-MTases according to the termi-

nology of [17], i.e., represented orphan (not included

in the RM systems) MTases of Gammaproteobacteria

with the vertical inheritance, as their phylogeny coin-

cided with the phylogeny of the bacteria themselves.

The phylogenetic tree of REases close to REases

from the studied RM systems is shown in Fig.4. Beside

the studied REases, it included only REases with the

GASTC specificity. The REases specific towards all four

sites were approximately equidistant from each other.

The REase PbaD1IIIP, a member of the M.PbaD1III sys-

tem with one MTase, whose specificity against GATGC

was confirmed by PacBio-confirmed but which con-

tained only one MethyltransfD12 family domain, was

an exception that did not cluster with other REases

of the same specificity.

DISCUSSION

REases and MTases with the same set of do-

mains can have different specificity. The phyloge-

netic trees for REases and group  B MTases (Fig.  3b)

exhibited clearly distinguishable clades corresponding

to the three specificities (GGATC, GATGC, and GATGG).

The phylogenetic tree for group  A MTases had a small

fraction of ‘blue’ MTases that formed a clade within

‘green’ MTases, but the remaining group  A MTases

were also well separated by the specificity (Fig.  3a).

In the REase phylogenetic tree, the ‘red’, ‘blue’, and

‘green’ clades were almost equidistant from each oth-

er, while MTases with the GATGC specificity (‘blue’)

were significantly closer to MTases with the GATGG

specificity (‘green’) than to enzymes with the GGATC

specificity (‘red’) (Figs. 1 and 3). Based on these data,

as well as joint trees including MTases with other

specificities (Fig.  3), we can conclude that the RM

systems containing REases with the RE_AlwI family

domain and MTases with two MethyltransfD12 family

domains have appeared at least twice in the evolution.

MTases of RM systems specific toward GATGG and

GATGC were found to be closely related to MTases

of the FokI-like systems with the GGATG specificity.

The phage MTase M.EcoT4Dam and related MTases of

other phages with the GATC specificity have probably

originated from an ancestor common with group  B

MTases of such systems. Meanwhile, other MTases

with the GATC specificity, in particular, Dam MTases

from various E.  coli strains (including M.EcoKDam),

were closer to group  B MTases with the GGATC speci-

ficity. It is likely that the specificity toward GATC has

been acquired by MTases more than once.

Gene order and fusion/separation of genes. The

systems with two MTases and one (fusion) MTase were

approximately equally distributed among RM systems

with the GATGC specificity (‘blue’). The overall tree for

REases from these and other systems (Fig.  2) shows

that the fusion or separation of the MTase genes

have occurred at least 10 times. These RM systems

were also characterised by a different location of the

REase gene relative to the two MTase genes (Table  1

and Fig.  2). Beside the main ABR variant (see Table  1),

six more rare variants of the mutual arrangement of

the three genes were observed. It should be noted that

SPIRIN et al.510

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

the rAB variant, which was represented in 75 systems

and 55 clusters based on 98% similarity of REase se-

quences, was found almost exclusively in Mycobacte-

rium bovis strains. Systems with the rAB arrangement

formed a single clade on the tree (Fig. 2). In this re-

gard, the rAB order should be considered less com-

mon than the main ABR order.

The situation was different for RM systems with

the other two specificities. Thus, only one RM sys-

tem with a fusion MTase, Cpe832ORF840, was found

among the systems specific toward GATGG; all other

RM systems contained two single-domain MTases. This

suggests that the fusion MTases with this specificity

are less functional (or not functional at all) for some

reason, so the fusion event has not been fixed in evo-

lution. It is possible that some structural features of

recognition of the GATGG site prevent proper binding

of fusion MTases to DNA. In contrast, most RM systems

with the GGATC specificity contained fusion MTases,

except only seven systems that had separate MTases

(see Table1). Also, a non-standard REase gene location

was less common in ‘red’ and ‘green’ systems than

in ‘blue’ systems. Therefore, the gene mobility within

an RM system is more common in the systems with

the GATGC specificity. This is consistent with the fact

that some RM systems with this specificity (‘blue’) con-

tained group A MTases apparently ‘borrowed’ from

RM systems with a different specificity (‘green’).

The evolutionary advantage of fusion of two

MTases is not obvious. In a standard situation, the

substrate of these enzymes is a half-methylated DNA

formed after replication, so that only one of the two

MTases acts at each site. It is likely that a fusion

MTase in a ‘red’ or ‘blue’ system is equally efficient

compared to separate enzymes and has been fixed or

lost in the neutral evolution.

Horizontal transfer and lifetime of RM systems

in bacteria. Comparison of the REase phylogenetic

tree with the taxonomy of host bacteria showed that

some groups of these RM systems have been trans-

ferred from genome to genome quite often, includ-

ing transfer between phylogenetically distant bacte-

ria, and have been lost rather quickly. For example,

the Kki66ORF4915P, Kki10529IP, Hpa4058ORF9600P,

and Aur25976ORF26P systems (GATGG specificity)

were very close to each other (>85% identity between

REases); however, they were present in bacteria of two

different classes: the first two systems were found in

two strains of K. kingae from the class Betaproteo-

bacteria, Hpa4058ORF9600P – in Haemophilus para-

influenzae from the class Gammaproteobacteria, and

Aur25976ORF26P – in Actinobacillus ureae, also from

the class Gammaproteobacteria, while no other simi-

larly related RM systems are represented in REBASE,

and the RefSeq protein database contains only one

other REase with the same level of similarity, which

is from Pasteurella multocida (Gammaproteobacteria).

All four listed bacteria are human pathogens. Of the

remaining RM systems, the three closest ones (over

40% REase sequence identity) belonged to three dif-

ferent bacterial phyla. This suggests that such systems

are very easily transferable, including between unre-

lated bacteria, but the duration of their existence in

the host is relatively short.

Another group (specific toward GATGC) included

16 RM systems from bacteria belonging to 14 differ-

ent families and eight different classes; at the same

time, 62 identical systems from different M.haemolyt-

ica strains belonged to the same group. The sequence

identity of REs from different systems in this group

was at least 75%. Analysis of fully assembled genomes

of M.  haemolytica strains showed that these systems

were present in about half of the strains. Apparently,

in M.  haemolytica, such RM systems have become es-

sential for some reason, whereas in other hosts, RM

systems of this group are easily acquired and rapidly

lost. It is interesting to note that all organisms host-

ing RM systems of this group, except M.haemolytica,

belonged to the human microflora, pathogenic or nor-

mal, which may be related to the frequency of RM

system exchange between them.

No other such closely related groups of RM sys-

tems from equally diverse bacteria have been found,

but the fact that RM systems are generally present

ina very small fraction of strains of each species also

indirectly indicates relatively frequent acquisition and

loss of RM systems.

Besides M. haemolytica, several other bacterial

species contained studied RM systems in a significant

proportion of strains. It is likely that in these species,

such RM systems also fulfil essential functions, for ex-

ample, a role in maintaining subspecies identity [18].

The lifetime of an RM system can be indicated by

the avoidance of its recognition site in the genome [15,

16]. For example, the avoidance of the GATGC site and

the absence of avoidance of the GATGG site in the ge-

nomes of all H.cinaedi strains (see Online Resource4)

may indicate that the systems specific toward GATGC

had been acquired by this species a long time ago but

have been then lost by most strains, whereas RM sys-

tems with the GATGG specificity have been acquired

relatively recently. A strong underrepresentation of

GATGG in the genomes of C. concisus strains may in-

dicate a long-term presence of RM systems with this

specificity in this bacterial species, although no such

RM systems have been detected in the sequenced ge-

nomes of this species.

Evolution of RM system specificity. The phy-

logenetic tree of MTases specific toward GGATC,

GATGG, and GATGC and closely related MTases (Fig.  3)

suggests a rather complex pattern of evolution of

the system specificity. It is likely that RM systems

RM SYSTEMS: EVOLUTION AND ECOLOGY 511

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

withthe GGATC specificity have been assembled from

two MTases and a REase independently of systems

with the GATGG and GATGC specificities. We can as-

sume a parallel evolution of the two MTases and REas-

es in most lineages of the ‘blue’ and ‘green’ systems.

Inother words, all these systems had a common ances-

tor, and no exchange of components have occurred in

their evolution, with one exception. Systems with the

GGATG specificity and the FokI_cleav_dom domains in

REases have probably acquired their MTases from an

ancestor common with the present-day RM systems

specific toward GATGG and GATGC.

The assumption made in [19] about a possible

origin of two MTases of the M.FokI family (GGATG

specificity) by duplication of the gene of the ances-

tral MTase with the SSATSS recognition sequence does

not fit well with the fact that the two MTases of such

systems are less related to each other than each of

them is to MTases with other specificities (Fig.3). Most

likely, MTases of this family have originated from

MTases with other specificities and in the course of

evolution, have undergone a change in the recogni-

tion sequence, which must have occurred in parallel

in two MTases that methylate different DNA strands.

Such change was possible if the genome contained an

RM system with a broader specificity or through the

weakening of the system own specificity. The same is

true for MTases specific toward GATGG (‘green’) and

GATGC (‘blue’), which have a relatively recent com-

mon ancestor with M.FokI-like MTases. As for REases,

they might have been acquired independently by RM

systems with different specificities, since FokI REase

contains a catalytic domain of a different family, and

REases of the systems specific toward GATGG and

GATGC, although homologous, are significantly less re-

lated than MTases of the same systems: the sequence

similarity between them is about the same level as

with the REases recognising GGATC (Figs. 1a and 4).

The evolution of MTases of all these systems is

closely related to the evolution of Dam MTases. It is

possible that all groupB MTases have originated from

orphan MTases that recognized GATC. In this case, the

phage MTase M.EcoT4Dam and related phage MTases

that also recognize the GATC site, appear to have

evolved from an ancestor of group  B MTases with

the GATGC specificity and have undergone a rever-

sal of specificity. This scenario is supported by the

differences in the mechanisms by which M.EcoKDam

and M.EcoT4Dam MTases recognize the same GATC

site [20].

CONCLUSION

All REs of RM systems specific toward GGATC,

GATGC, and GATGG are homologous to each other.

The same is true for MTases of these systems, al-

though some of RM systems include one fusion MTase

with two catalytic domains, while others contain two

MTases. The N- and C-terminal parts of the fusion

MTases were compared with each other and with

the sequences of separate MTases. Based on the se-

quence similarity, MTases were classified into two

rather distant groups. Two separate MTases of the

same system and two parts of fusion MTases always

belong to two different groups. Analysis of phyloge-

netic trees inferred from REases and from each of the

MTase groups showed that the fusion and separation

of MTases have occurred repeatedly in the evolution.

We found the evidence of rearrangements of the

gene structure of RM systems, as well as horizontal

transfer of entire RM systems and individual genes to

other RM systems. In particular, a group of RM sys-

tems contained MTases that have probably originated

from an MTase with a different recognition specificity.

Most of the studied RM systems are represented

in a very small fraction of strains of the correspond-

ing bacterial species, indicating a high probability

of their loss by their hosts. At the same time, a few

groups of these RM systems might have become es-

sential to their hosts because they are represented in

the majority of strains.

Abbreviations. MTase, DNA methyltransferase;

REase, restriction endonuclease; RM system, restric-

tion–modification system.

Supplementary information. The online version

contains supplementary material available at https://

doi.org/10.1134/S0006297925600115.

Acknowledgements. The authors would like to

thank Dr. A. V. Grishin for his help in preparing the

text.

Contributions. S. Spirin, A. Alexeevski, and

A. Karyagina developed the concept and supervised

the study; S. Spirin, I. Rusinov, O. Makarikova, and

A. Karyagina prepared the data, developed program

support, and analyzed the results; S. Spirin and

A. Karyagina wrote and edited the manuscript.

Funding. This work was supported by the Rus-

sian Science Foundation (project no.21-14-00135).

Ethics approval and consent to participate.

This work does not contain any studies involving hu-

man or animal subjects.

Conflict of interest. The authors of this work de-

clare that they have no conflicts of interest.

REFERENCES

1. Williams, R.J. (2003) Restriction endonucleases: clas-

sification, properties, and applications, Mol. Biotech-

nol., 23, 225-244, https://doi.org/10.1385/mb:23:3:225.

SPIRIN et al.512

BIOCHEMISTRY (Moscow) Vol. 90 No. 4 2025

2. Roberts, R. J. (2003) A nomenclature for restriction

enzymes, DNA methyltransferases, homing endonu-

cleases and their genes, Nucleic Acids Res., 31, 1805-

1812, https://doi.org/10.1093/nar/gkg274.

3. Madhusoodanan, U.K., and Rao, D.N. (2010) Diversity

of DNA methyltransferases that recognize asymmetric

target sequences, Crit. Rev. Biochem. Mol. Biol., 45,

125-145, https://doi.org/10.3109/10409231003628007.

4. Vasu,K., and Nagaraja,V. (2013) Diverse functions of

restriction-modification systems in addition to cel-

lular defense, Microbiol. Mol. Biol. Rev., 77, 53-72,

https://doi.org/10.1128/mmbr.00044-12.

5. Fokina, A.S., Karyagina, A.S., Rusinov, I.S., Moshensky,

D. M., Spirin, S. A., and Alexeevski, A. V. (2023) Evo-

lution of restriction–modification systems consisting

of one restriction endonuclease and two DNA meth-

yltransferases, Biochemistry (Moscow), 88, 253-261,

https://doi.org/10.1134/S0006297923020086.

6. Mistry, J., Chuguransky, S., Williams, L., Qureshi, M.,

Salazar, G. A., Sonnhammer, E. L. L., Tosatto, S. C. E.,

Paladin, L., Raj, S., Richardson, L. J., Finn, R. D., and

Bateman, A. (2020) Pfam: the protein families da-

tabase in 2021, Nucleic Acids Res., 49, D412-D419,

https://doi.org/10.1093/nar/gkaa913.

7. Roberts, R. J., Vincze, T., Posfai, J., and Macelis, D.

(2014) REBASE – a database for DNA restriction and

modification: enzymes, genes and genomes, Nucle-

ic Acids Res., 43, D298-D299, https://doi.org/10.1093/

nar/gku1046.

8. Edgar, R.C. (2004) MUSCLE: multiple sequence align-

ment with high accuracy and high throughput, Nu-

cleic Acids Res., 32, 1792-1797, https://doi.org/10.1093/

nar/gkh340.

9. Lefort,V., Desper,R., and Gascuel,O. (2015) FastME 2.0:

A comprehensive, accurate, and fast distance-based

phylogeny inference program, Mol. Biol. Evol., 32,

2798-2800, https://doi.org/10.1093/molbev/msv150.

10. Kumar, S., Stecher, G., and Tamura,K. (2016) MEGA7:

Molecular Evolutionary Genetics Analysis version7.0

for bigger datasets, Mol. Biol. Evol., 33, 1870-1874,

https://doi.org/10.1093/molbev/msw054.

11. Letunic, I., and Bork, P. (2021) Interactive tree of life

(iTOL) v5: an online tool for phylogenetic tree display

and annotation, Nucleic Acids Res., 49, W293-W296,

https://doi.org/10.1093/nar/gkab301.

12. Li, W., and Godzik, A. (2006) Cd-hit: a fast program

for clustering and comparing large sets of pro-

tein or nucleotide sequences, Bioinformatics, 22,

1658-1659, https://doi.org/10.1093/bioinformatics/

btl158.

13. Burge,C., Campbell, A.M., and Karlin,S. (1992) Over-

and under-representation of short oligonucleotides in

DNA sequences, Proc. Natl. Acad. Sci. USA, 89, 1358-

1362, https://doi.org/10.1073/pnas.89.4.1358.

14. Rusinov, I.S., Ershova, A. S., Karyagina, A. S., Spirin,

S.A., and Alexeevski, A.V. (2018) Comparison of meth-

ods of detection of exceptional sequences in prokary-

otic genomes, Biochemistry (Moscow), 83, 129-139,

https://doi.org/10.1134/S0006297918020050.

15. Karlin, S., Burge, C., and Campbell, A. M. (1992)

Statistical analyses of counts and distributions of

restriction sites in DNA sequences, Nucleic Ac-

ids Res., 20, 1363-1370, https://doi.org/10.1093/nar/

20.6.1363.

16. Rusinov, I., Ershova, A., Karyagina, A., Spirin, S., and

Alexeevski, A. (2015) Lifespan of restriction-modifi-

cation systems critically affects avoidance of their

recognition sites in host genomes, BMC Genomics,

16, 1084, https://doi.org/10.1186/s12864-015-2288-4.

17. Brézellec, P., Hoebeke, M., Hiet, M. S., Pasek, S.,

and Ferat, J. L. (2006) DomainSieve: a protein do-

main-based screen that led to the identification of

dam-associated genes with potential link to DNA

maintenance, Bioinformatics, 22, 1935-1941, https://

doi.org/10.1093/bioinformatics/btl336.

18. Murray, N. E. (2002) 2001 Fred Griffith review lec-

ture. Immigration control of DNA in bacteria: self

versus non-self, Microbiology, 148, 3-20, https://

doi.org/10.1099/00221287-148-1-3.

19. Friedrich, T., Fatemi, M., Gowhar, H., Leismann, O.,

and Jeltsch, A. (2000) Specificity of DNA binding

and methylation by the M.FokI DNA methyltransfer-

ase, Biochim. Biophys. Acta, 1480, 145-159, https://

doi.org/10.1016/s0167-4838(00)00065-0.

20. Horton, J. R., Liebert, K., Bekes, M., Jeltsch, A., and

Cheng, X. (2006) Structure and substrate recognition

of the Escherichia coli DNA adenine methyltransfer-

ase, J.Mol. Biol., 358, 559-570, https://doi.org/10.1016/

j.jmb.2006.02.028.

Publisher’s Note. Pleiades Publishing remains

neutral with regard to jurisdictional claims in published

maps and institutional affiliations. AI tools may have

been used in the translation or editing of this article.