Article

ISSN 0006-2979, Biochemistry (Moscow), 2026, Vol. 91, No. 7, pp. 1174-1186 © The Author(s), 2026. This article is an open access publication.

1174

Substrate Specificity of T4 RNA Ligase 1

in Intramolecular Ligation of Single-Stranded DNA

Assol R. Sakhabutdinova

, Ayzilya A. Shakirova

and Ravil R. Garafutdinov

1,a

Institute of Biochemistry and Genetics, Ufa Federal Research Center, Russian Academy of Sciences,

450054 Ufa, Bashkortostan, Russia

e-mail: garafutdinovr@gmail.com

Received November 10, 2025

Revised June 18, 2026

Accepted June 19, 2026

Abstract— Ligation of nucleic acids by DNA and RNA ligases is a routine method in molecular biology.

Enzymatic ligation is increasingly used in the development of novel methods for detecting specific nucle-

otide sequences, as well as in non-biological DNA application. In this study, we investigated how the type

of 5′- and 3′-terminal nucleotides affects the efficiency of single-stranded DNA cyclization. Model oligode-

oxyribonucleotides (ODNs) 23, 39, and 45 nucleotides in length, representing all possible combinations of

the donor and acceptor terminal nucleotides, were used as substrates. The substrate specificity of T4  RNA

ligase  1 was assessed by analyzing the primary structure of amplicons generated through rolling circle

amplification (RCA) of ligation products. For longer ODNs (39 and 45  nt), T4  RNA ligase  1 most efficiently

cyclized substrates containing 5′-CC or 5′-TT donor dinucleotides and 3′-AA acceptor dinucleotide. Among

the 23-nt ODNs, efficient cyclization was observed predominantly for a single substrate containing 5′-TT

and 3′-AA terminal dinucleotides, likely reflecting a greater flexibility of AT-rich DNA. Substrates contain-

ing 5′-dGG and 3′-dGG dinucleotides showed the lowest propensity for cyclization. These findings provide

guidance for selecting substrate sequences that maximize the yield of circular DNA products.

DOI: 10.1134/S0006297925603983

Keywords: RNA ligases, DNA cyclization, rolling circle amplification, nanopore sequencing, sequencing reads

* To whom correspondence should be addressed.

INTRODUCTION

T4 RNA ligase  1 (T4Rnl1) is an enzyme most

commonly used in molecular biology for joining sin-

gle-stranded nucleic acids (NAs) through the forma-

tion of a 3′,5′-phosphodiester bond. A notable feature

of T4Rnl1 is its ability to catalyze both inter- and

intramolecular template-independent ligation of RNA

and single-stranded DNA (ssDNA), as well as the li-

gation of double-stranded DNA  [1-3]. Intramolecular

ligation enables the cyclization of NAs without the

need for a complementary splint or supporting tem-

plate and generates circular molecules that can subse-

quently serve as substrates for rolling circle amplifi-

cation (RCA), one of the most widely used isothermal

NA amplification techniques  [4]. RCA produces long

concatemeric products consisting of tandem repeats

of the circular template sequence, with lengths that

are integer multiples of the original template. RCA

has been extensively applied for the detection and

quantification of specific NAs, proteins, small organic

molecules, and even inorganic ions [5, 6]. In recent

years, RCA has become a key component in sever-

al high-throughput sequencing approaches designed

for “difficult” NA targets, primarily various classes of

small RNAs [7,  8]. Its application is especially valu-

able in the analysis of degraded and ancient DNA

samples  [9]. Beyond biological applications, T4Rnl1 is

increasingly employed in emerging DNA-based tech-

nologies, such as DNA nanotechnology, DNA comput-

ing, and DNA data storage, among others [10, 11].

Recently, several studies have revealed that the

efficiency of DNA strand joining by DNA ligases

depends on the type of the donor (5′-terminal) and

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1175

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

acceptor (3′-terminal) nucleotides. These findings

were enabled by the development of high-through-

put analytical approaches, including DNA microarray

technology coupled with mass spectrometry detec-

tion  [12], capillary electrophoresis  [13], and massive-

ly parallel sequencing  [14-16]. Among these, sequenc-

ing-based methods are particularly powerful because

they allow simultaneous analysis of ligation outcomes

for a large number of substrates within a single sam-

ple. For example, PacBio sequencing combined with

RCA for enriching the target NAs was used to identi-

fy donor–acceptor nucleotide combinations that max-

imize DNA ligation efficiency by T4 DNA ligase  [16].

The substrate specificity of T4Rnl1 remains less char-

acterized. Early studies using mono-, di-, and triribo-

nucleotide substrates produced contradictory results

regarding its specificity toward linear RNA substrates

[17-19]. More recently, attempts have been made to

investigate T4Rnl1 substrate preferences by sequenc-

ing-based analysis of miRNA libraries  [20,  21]. How-

ever, subsequent studies demonstrated that the ob-

served variations in the target abundance (commonly

referred to as ligation bias) were primarily caused by

secondary structures formed between the NA targets

and adapter sequences, rather than by the types of

the terminal nucleotides  [22]. To date, no systemat-

ic data are available on the substrate specificity of

T4Rnl1 toward single-stranded DNA, particularly in

the context of NA cyclization.

It is worth noting that specialized DNA ligases,

namely CircLigase  I and CircLigase  II, have been

developed for the synthesis of small circular DNA

molecules and are commercially available. However,

their activity appears to be sequence-dependent, as

the efficiency of circular product formation depends

on the substrate nucleotide context

. Recently, an

alternative ssDNA ligase has been reported for this

application, although it is not yet commercially avail-

able  [23]. Consequently, despite a relatively low effi-

ciency of ssDNA cyclization catalyzed by T4Rnl1, this

enzyme remains the most available and convenient

tool for the preparation of circular DNA structures.

Several approaches have been proposed to improve

the yield of circular ligation products [2,  9,  24]. Fur-

ther optimization of circular DNA synthesis, partic-

ularly in applications involving multiple substrates,

where unbiased sequence representation is essential,

requires a better understanding of how 5′- and 3′-ter-

minal nucleotides influence ligation efficiency. There-

fore, this study aimed to investigate the substrate

specificity of T4Rnl1 in the intramolecular ligation

of ssDNA.

MATERIALS AND METHODS

Reagents. T4Rnl1, exonuclease  I (Exo  I), polyeth-

ylene glycol 4000 (PEG4000), bovine serum albumin

(BSA) from ThermoFisher Scientific (USA), adenosine

triphosphate (ATP) from Sigma (Germany), Bst 2.0

WarmStart DNA polymerase from New England Bi-

olabs (USA), T4 polynucleotide kinase, dNTPs from

SibEnzyme (Russia), and dsGreen intercalating dye

from Lumiprobe (Russia) were used in the study. All

solutions and samples were prepared using ultrapure

water (>18  MΩ) (Millipore, France).

Oligodeoxyribonucleotide (ODN) substrates

and primers were artificial nucleotide sequences

(Table  1) designed and analyzed using OligoAnalyzer

to minimize the possibility of secondary structure

formation. All ODNs were synthesized by Lumiprobe

(Russia).

DNA cyclization. All ODN substrates were pre-

liminarily phosphorylated with T4 polynucleotide

kinase according to a standard protocol. Enzymatic

ligation was performed for individual ODNs (A0A, …,

T0T, A8A, …, T8T, A11A, …, T11T) and their complete

mixtures mN0N, mN8N, and mN11N corresponding

to substrate sets N0N, N8N, and N11N, respectively

(Table  1; 16 substrates in each mixture) Ligation re-

actions were carried out in the presence of PEG4000

and BSA in the reaction mixture (20  μL) containing

10  μM ODN, T4Rnl1 (1, 2, 4, 8, or 10  U depending

on the experiment), 1×  T4Rnl1 ligase buffer (50  mM

Tris-HCl, pH  7.8 at 25°C, 10  mM MgCl

, 10  mM dith-

iothreitol, 1  mM ATP), 15%  PEG4000, and 1  μg  BSA.

All ligation reactions were prepared in triplicates

and incubated at room temperature for 1, 2, 4, 6, or

24  h depending on the experiment, after which the

enzyme was inactivated at 65°C for 20  min. Following

ligation, 4.0  U of Exo  I, 1  μL of Exo  I buffer, and 4  μL

of water were added to 5  μL of the ligation mixture

and incubated for 1.5 h at 37°C. Exo  I was then in-

activated for 15  min at 85°C. The resulting reaction

mixtures from three replicates were pooled, diluted

10-fold with water, and used for RCA.

Rolling circle amplification. RCAwas performed

in a two-primer format (ramification). RCA samples

were prepared in a UVC/TM-AR PCR workstation (Bio-

san, Latvia). Prior to use, the working area, pipettes,

and plastic consumables ware irradiated with UV light

for 20  min. Amplification reactions were carried out

in a CFX96 Opus Real-Time PCR System (Bio-Rad Lab-

oratories, USA). Each reaction (20  μL) was prepared

in triplicate and contained 1  μL of the tenfold-dilut-

ed, exonuclease-treated ligation product, 4  pmol of

https://www.biosearchtech.com/molecular-biology-enzymes/dna-and-rna-modifying-enzymes/dna-and-rna-ligases/c/

dna-rna-ligases

https://www.idtdna.com/pages/tools/oligoanalyzer

SAKHABUTDINOVA et al.1176

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Table 1. ODN substrates and primers used in the study

No. ODN set Code Sequence, 5′→3′ Length, nt

N0N

A0A AACTGTAGTCTGCTATGTGCTAA

2 A0C AACTGTAGTCTGCTATGTGCTCC

3 A0G AACTGTAGTCTGCTATGTGCTGG

4 A0T AACTGTAGTCTGCTATGTGCTTT

5 C0A CCCTGTAGTCTGCTATGTGCTAA

6 C0C CCCTGTAGTCTGCTATGTGCTCC

7 C0G CCCTGTAGTCTGCTATGTGCTGG

8 C0T CCCTGTAGTCTGCTATGTGCTTT

9 G0A GGCTGTAGTCTGCTATGTGCTAA

10 G0C GGCTGTAGTCTGCTATGTGCTCC

11 G0G GGCTGTAGTCTGCTATGTGCTGG

12 G0T GGCTGTAGTCTGCTATGTGCTTT

13 T0A TTCTGTAGTCTGCTATGTGCTAA

14 T0C TTCTGTAGTCTGCTATGTGCTCC

15 T0G TTCTGTAGTCTGCTATGTGCTGG

16 T0T TTCTGTAGTCTGCTATGTGCTTT

N8N

A8A AA(n)

CTGTAGTCTGCTATGTGCT(n)

18 A8C AA(n)

CTGTAGTCTGCTATGTGCT(n)

19 A8G AA(n)

CTGTAGTCTGCTATGTGCT(n)

20 A8T AA(n)

CTGTAGTCTGCTATGTGCT(n)

21 C8A CC(n)

CTGTAGTCTGCTATGTGCT(n)

22 C8C CC(n)

CTGTAGTCTGCTATGTGCT(n)

23 C8G CC(n)

CTGTAGTCTGCTATGTGCT(n)

24 C8T CC(n)

CTGTAGTCTGCTATGTGCT(n)

25 G8A GG(n)

CTGTAGTCTGCTATGTGCT(n)

26 G8C GG(n)

CTGTAGTCTGCTATGTGCT(n)

27 G8G GG(n)

CTGTAGTCTGCTATGTGCT(n)

28 G8T GG(n)

CTGTAGTCTGCTATGTGCT(n)

29 T8A TT(n)

CTGTAGTCTGCTATGTGCT(n)

30 T8C TT(n)

CTGTAGTCTGCTATGTGCT(n)

31 T8G TT(n)

CTGTAGTCTGCTATGTGCT(n)

T8T TT(n)

CTGTAGTCTGCTATGTGCT(n)

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1177

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Table 1 (cont.)

No. ODN set Code Sequence, 5′→3′ Length, nt

N11N

A11A AA(n)

CTGTAGTCTGCTATGTGCT(n)

34 A11C AA(n)

CTGTAGTCTGCTATGTGCT(n)

35 A11G AA(n)

CTGTAGTCTGCTATGTGCT(n)

36 A11T AA(n)

CTGTAGTCTGCTATGTGCT(n)

37 C11A CC(n)

CTGTAGTCTGCTATGTGCT(n)

38 C11C CC(n)

CTGTAGTCTGCTATGTGCT(n)

39 C11G CC(n)

CTGTAGTCTGCTATGTGCT(n)

40 C11T CC(n)

CTGTAGTCTGCTATGTGCT(n)

41 G11A GG(n)

CTGTAGTCTGCTATGTGCT(n)

42 G11C GG(n)

CTGTAGTCTGCTATGTGCT(n)

43 G11G GG(n)

CTGTAGTCTGCTATGTGCT(n)

44 G11T GG(n)

CTGTAGTCTGCTATGTGCT(n)

45 T11A TT(n)

CTGTAGTCTGCTATGTGCT(n)

46 T11C TT(n)

CTGTAGTCTGCTATGTGCT(n)

47 T11G TT(n)

CTGTAGTCTGCTATGTGCT(n)

48 T11T TT(n)

CTGTAGTCTGCTATGTGCT(n)

primers

PrF CTGTAGTCTGCTATGTGCT 19

50 PrR AGCACATAGCAGACTACAG 19

Note. Variable nucleotides are highlighted in bold; the letter “n” denotes degenerate nucleotides.

PrF primer, 10  pmol of PrR primer, 0.25 mM of each

dNTP, 1×  ThermoPol buffer, 3.0  U of Bst  2.0 Warm-

Start DNA polymerase, and 0.5× dsGreen intercalating

dye. Amplification was performed under quasi‐iso-

thermal conditions using mild thermocycling in the

following regime: 300 cycles of 68°C for 2  s and 63°C

for 20  s.

Nanopore sequencing of amplicons was per-

formed by Cloning Facility (Russia). To minimize the

influence of variability introduced during ligation

and amplification, amplification products from the

three experimental replicates were pooled prior to

sequencing. The resulting RCA products were puri-

fied using a Mini DNA/RNA Isolation Kit for Reac-

tion Mixtures (Biolabmix, Russia). Each purified DNA

preparation was then divided into three technical

replicates to evaluate sequencing accuracy and to re-

duce the impact of random errors. The samples were

lyophilized and subjected to nanopore sequencing.

Sequencing data were processed using an in-house

algorithm comprising the following steps:

(i) From the complete set of reads obtained for each

sample, only those containing the inter-primer

region with a characteristic central tetranucle-

otide insertion (5′-NNNN-3′) were selected.

(ii) For reads derived from ODNs of the N8N and

N11N sets, sequences containing the inter-primer

region of the expected length (20 and 26  nt, re-

spectively), as well as variants shortened by one

or two nucleotides, were retained for analysis.

(iii) The total number of 5′-NNNN-3′ insertions and

the counts of individual insertion types were

determined. The frequencies (%) of all 5′- and

3′-terminal dinucleotides within the tetranucle-

otide insertions were then calculated.

RESULTS AND DISCUSSION

Until recently, it had been believed that NA ligas-

es join NA strands with an equal efficiency regardless

of the type of 5′- and 3′-terminal nucleotides (donor

SAKHABUTDINOVA et al.1178

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

and acceptor sites, respectively). However, accumulat-

ing evidence has challenged this assumption. Similar

to other NA-processing enzymes, such as DNA poly-

merases [25], NA ligases appear to exhibit substrate

specificity, i.e., the rate of enzymatic conversion de-

pends on the nucleotide sequence context. The data

on the ligase substrate specificity are particularly

important when working with low-copy-number NA

molecules, such as degraded or ancient DNA, as well

as short nucleic acids, including miRNAs. Under cer-

tain reaction conditions, miRNAs can undergo intra-

molecular ligation and form circular molecules [2,

24]. Although the yield of circular products is typi-

cally low, it is sufficient for successful RCA or roll-

ing circle transcription. These approaches enable re-

al-time analysis and enrichment of NA samples for

downstream sequencing, as demonstrated previously

[15, 16]. Because RCA generates long concatemers

composed of tandem repeats of the original circular

template, this strategy provides sufficient sequencing

coverage even for low-abundant NAs.

To investigate the substrate specificity of T4Rnl1,

three sets of model ODNs containing a common 19-nt

core sequence were designed (Table 1). The first set

(N0N) consisted of 23-nt ODNs, the second set (N8N)

of 39-nt ODNs, and the third set (N11N) of 45-nt

ODNs. In all cases, ODNs contained variable terminal

homodinucleotides (5′-NN and 3′-NN), and each set

included all 16 possible combinations of these termi-

nal dinucleotides. ODNs from the N8N and N11N sets

additionally contained two regions of eight or elev-

en degenerate nucleotides, respectively, that flanked

the conserved core sequence (N0N ODNs lacked such

regions). The N8N and N11N sets were designed as

models of “universal” substrates, in which the in-

fluence of nucleotides located three to ten positions

from either end (i.e., broader sequence context) was

averaged out through sequence degeneracy. ODNs of

the N0N set, which contained no degenerate nucleo-

tides, served as reference molecules for comparison.

The use of terminal homodinucleotides allowed for

unambiguous identification of ligation junctions in

sequencing data, as ligation generated characteristic

5′-NNNN-3′ tetranucleotide signatures that marked the

site of DNA circularization.

All ODNs were phosphorylated using T4 polynu-

cleotide kinase, and then ligated in the presence of

PEG4000 and BSA to increase the yield of circular

DNA [9]. Ligation was performed for both individual

ODNs (A0A, …, T0T, A8A, …, T8T, A11A, …, T11T; 48

samples in total) and their mixtures (mN0N, mN8N,

and mN11N, 16 individual ODNs each). Following li-

gation, residual linear ODNs and linear ligation prod-

ucts were digested with Exo I, using the incubation

time three times longer than that recommended in

the standard protocol. The resulting circular DNA

preparations were subjected to RCA to generate suffi-

cient quantities of amplicons for sequencing analysis

(RCA enrichment) (Fig. 1).

Amplification was performed in a two-primer

variant (PrF and PrR), referred to as ramification [26].

The primers were fully complementary to each other,

with PrF corresponding to the core sequence common

to all ODNs. The PrR primer initiated the synthesis

of a DNA strand complementary to the ligation prod-

ucts; therefore, it was added in slight excess to pro-

mote efficient amplicon accumulation. Ramification

was carried out using Bst 2.0 WarmStart DNA poly-

merase, which combines strong strand-displacement

activity with robust polymerase performance, high

processivity, and sufficient fidelity [27]. Real-time RCA

was conducted under conditions previously shown to

suppress non-specific isothermal amplification (mul-

timerization), namely at 63°C in the presence of 0.5×

intercalating dye [28, 29]. Under these conditions, no

amplification curves were generated for negative con-

trol samples, whereas successful intramolecular liga-

tion was indicated by the appearance of amplification

signals in experimental samples. Ramification of liga-

tion products derived from individual ODNs result-

ed in the accumulation of amplification products in

most cases. Relatively low threshold time (Tt) values

indicated efficient cyclization of ODNs under selected

ligation conditions, which included the presence of

PEG4000 and BSA (Table 2).

It is generally accepted that lower Tt values

correspond to higher copy numbers of the original

DNA template in a sample. However, this relation-

ship holds only when amplifying the same target se-

quence under identical experimental conditions. In

the present study, DNA templates with substantially

different nucleotide composition were amplified, for

which such direct correlation may not be valid due to

sequence-dependent differences in the amplification

efficiency, including variation in nucleotide incorpo-

ration rates during DNA chain polymerization (con-

text-dependent amplification) [30, 31]. Analysis of Tt

values indicated that, in most cases, accumulation of

amplification products started earlier for the ligation

products derived from individual ODNs of the N8N

and N11N sets that contained CC dinucleotides at the

5′ and/or 3′ ends. Within the N0N set, the lowest Tt

values were observed for products resulting from li-

gation of 5′-TT substrates. ODNs containing 3′-CC or

3′-GG in combination with 5′-AA, 5′-CC, or 5′-GG did

not undergo detectable ramification, suggesting the

absence of circular DNA formation in these samples.

This pronounced trend for a more efficient amplifica-

tion of dC- and dT-rich substrates cannot be attribut-

ed simply to the nucleotide preference of DNA poly-

merase, as purine nucleotides, particularly dATP, are

generally incorporated at higher rates during DNA

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1179

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Fig. 1. Scheme of the experiment. N, variable 5′- and 3′-terminal nucleotides; n, degenerate nucleotides.

synthesis  [28]. Ramification was also performed for

mixed samples (mN0N, mN8N, and mN11N), which

exhibited intermediate Tt values. Overall, the results

of amplification experiments should be considered

preliminary, as they do not allow for an unambigu-

ous conclusion on the substrate specificity of T4Rnl1.

RCA enrichment of ligation products enabled the

generation of DNA samples suitable for high-through-

put sequencing. Amplification enrichment of low-copy

NAs, most commonly by polymerase chain reaction,

has become a widely used approach in such stud-

ies [32]. In our work, nanopore sequencing was ap-

plied for the first time to address a similar challenge.

This approach is particularly well suited for deter-

mining the primary structure of amplicons because

it does not require DNA fragmentation and thus pre-

serves long, informative reads. Although nanopore

sequencing is characterized by comparatively lower

accuracy [33], its use is especially justified for ana-

lyzing the concatemeric structure of RCA products

composed of tandem repeats, where sequence peri-

odicity facilitates unambiguous interpretation of the

obtained data.

Sequencing was performed on amplicons de-

rived from the ligation products of individual ODNs,

as well as mN0N, mN8N, and mN11N mixtures. For

each amplicon, several thousand reads with an av-

erage length of ~400  bp were obtained. All reads

contained 5′-terminal sequences non-homologous to

the original ODNs, which have likely been formed

by non-specific DNA synthesis or co-amplification of

background DNA  [34]. Some reads also contained nu-

cleotide substitutions and deletions, presumably due

to sequencing errors or polymerase-induced errors.

From the total dataset, informative reads were se-

lected that contained tandem repeats in the expect-

ed head-to-tail orientation. These reads also includ-

ed either a full-length inter-primer region (4  nt for

N0N, 20  nt for N8N, and 26  nt for N11N) or versions

shortened by 1-2  nt (except N0N), together with the

characteristic central 5′-NNNN-3′ tetranucleotide in-

sertion (Fig.  2).

In general, for each ligation product, the num-

ber of informative reads did not exceed 35% of the

total read count. Processing of informative reads en-

abled determination of the frequency of each type of

5′-NNNN-3′ insertion, as well as the contribution of

terminal dinucleotides to their formation. For reads

containing both direct and reverse sequences, the

distribution of corresponding dinucleotides was ap-

proximately identical, so further data are present-

ed only for direct sequences. For ligation products

SAKHABUTDINOVA et al.1180

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Table 2. Mean threshold time (Tt, min) in the amplification of ligation products of individual ODNs categorized

according to5′- and3′-terminal dinucleotides and their mixtures (data were obtained using 2  U of T4Rnl1 and

a 24-h ligation duration)

5′-Dinucleotides

3′-Dinucleotides

AA CC GG TT

Individual ODNs of the N0N set

AA 12.3 ± 0.9 nd nd 13.0 ± 0.7

CC 7.4 ± 0.3 nd nd 11.9 ± 0.2

GG 16.1 ± 1.0 nd nd 10.3 ± 0.5

TT 4.5 ± 0.3* 10.2 ± 0.6 9.8 ± 0.2 9.8 ± 0.3

Mixed sample mN0N 9.4 ± 0.9

Individual ODNs of the N8N set

AA 7.1 ± 0.3 3.2 ± 0.4 4.0 ± 0.7 7.7 ± 0.3

CC 2.9 ± 0.5 1.6 ± 0.0 5.1 ± 0.4 2.9 ± 0.2

GG 6.2 ± 0.1 2.9 ± 0.2 10.1 ± 0.3 9.9 ± 0.4

TT 4.4 ± 0.3 2.1 ± 0.1 9.4 ± 0.2 6.8 ± 0.8

Mixed sample mN8N 5.8 ± 0.4

Individual ODNs of the N11N set

AA 6.8 ± 0.4 5.9 ± 1,1 8.1 ± 1.0 3.7 ± 0.1

CC 5.7 ± 0.2 1.4 ± 0.2 7.8 ± 0.3 2.4 ± 0.2

GG 10.2 ± 0.5 4.6 ± 0.4 9.8 ± 0.9 7.3 ± 0.3

TT 6.1 ± 0.1 3.5 ± 0.3 8.5 ± 0.8 4.2 ± 0.5

Mixed sample mN11N 7.5 ± 0.7

Note. Cells with the minimum Tt value in the corresponding rows and columns are highlighted in gray; nd,samples showing

no increase in amplification curves; * minimum Tt values.

derived from individual ODNs, each set of informa-

tive reads was expectedly represented by sequences

containing a single type of tetranucleotide insertion

(data not shown). In contrast, samples obtained from

the mN0N, mN8N, and mN11N mixtures contained

reads with diverse 5′-NNNN-3′ insertions. For the li-

gation products derived from mN0N mixture, nucle-

otide sequences with the corresponding tetranucle-

otide insertions were identified (Fig.  2, read  1). For

the mN8N and mN11N mixtures, in addition to the

expected products (Fig.  2, reads  2 and  3), reads with

shortened inter-primer regions were also detected,

including both sequences containing the 5′-NNNN-3′

insertion (Fig.  2, read  4) and those lacking it (data not

shown). The presence of these truncated reads can

likely be attributed to a minor fraction of incorrect-

ly synthesized oligonucleotides in the starting ODN

pools, which nevertheless supported efficient forma-

tion of circular DNA molecules of expected size and

subsequent amplification of the corresponding prod-

ucts. Notably, each informative read represented an

amplification product of a single circular DNA tem-

plate, as indicated by the presence of a single repeat-

ing motif throughout each sequence.

Among the informative reads obtained from

the N0N mixture, the vast majority contained the

5′-AATT-3′ insertion (83.2  ±  3.1%), while the second

most frequent was 5′-AACC-3′ (12.5  ±  1.4%). In the

case of N8N mixture, the most abundant insertions

were 5′-AATT-3′ (18.2  ±  2.4%), 5′-AACC-3′ (14.0  ±  1.5%),

5′-CCTT-3′ (13.1  ±  1.2%), and 5′-TTAA-3′ (10.7  ±  1.7%).

For the N11N mixture, the most frequent insertions

were 5′-TTCC-3′ (12.1  ±  1.2%), 5′-AAGG-3′ (10.7  ±  1.4%),

5′-AACC-3′ (10.6  ±  0.6%), and 5′-GGTT-3′ (10.2  ±  1.1%)

(Fig.  3a). Overall, shorter ODN substrates (N0N

group), characterized by the terminal dinucleotides

5′-AA, 5′-GG, 3′-GG, and 3′-TT, rarely formed circular

DNA, with cyclization occurring predominantly with

a single substrate – oligonucleotide T0A – in the N0N

mixture (Fig.  3b). This is likely explained by a greater

conformational flexibility of the T0A backbone com-

pared with other ODNs of the same length [35-37].

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1181

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Fig. 2. Examples of primary structures of amplicons obtained by RCA enrichment of ligation products derived from mN0N

(read 1), mN8N (read 2), and mN11N (reads 3 and 4) mixtures. Characteristic dinucleotides and tetranucleotide insertions

(5′-NNNN-3′) are highlighted. Target sequences are flanked at the 5′ end by non-homologous sequences (shown in red and

truncated for clarity). Motifs confirming RCA from a single DNA template are enclosed in frames; PrF primer sequence is

underlined; boundaries of tandem repeats are indicated by dotted lines.

With increasing ODN length, the contribution to

the circular DNA formation of substrates containing

5′-GG and 3′-GG increases, although these dinucleo-

tides remain the least preferred. Thus, ODN length

has a pronounced effect on the DNA cyclization ef-

ficiency during ligation, which may be explained by

the presence of conformational states optimal for

specific enzymatic transformations  [38,  39]. This in-

terpretation is supported by the tendency toward a

more uniform representation of all dinucleotides in

ligation products as ODN length increases, likely re-

flecting greater conformational freedom of DNA mol-

ecules.

To further validate the observed substrate speci-

ficity of T4Rnl1, additional kinetic experiments were

performed in which either the amount of ligase (1, 2,

4, 8, and 10  U per reaction; 24-h reaction duration) or

the ligation time (1, 2, 4, 6, and 24  h; 2  U ligase per

SAKHABUTDINOVA et al.1182

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Fig. 3. Frequency of characteristic tetranucleotide 5′-NNNN-3′ insertions (a) and dinucleotides (b) in reads (data obtained

using 2  U of T4Rnl1 for the ligation duration of 24 h).

reaction) were varied. These experiments were based

on the assumption that a lower enzyme-to-substrate

ratio would favor ligation of more preferred sub-

strates. The experimental protocol, including ligation,

RCA enrichment, and sequencing (see above) was

applied to the mN8N mixture. Analysis of dinucleo-

tide frequencies in sequencing reads from the corre-

sponding samples enabled assessment the effects of

the enzyme concentration and ligation time on the

T4Rnl1 substrate specificity (Fig.  4). With increasing

enzyme amount, the frequencies of 5′-AA, 5′-CC, and

3′-AA decreased, the frequencies of 5′-TT and 3′-CC

increased, while the frequencies of 5′-GG, 3′-GG, and

3′-TT remained largely unchanged (Fig.  4a). The most

pronounced changes were found for 5′-TT (~16%)

and 3′-AA (~15%). A similar trend was observed in

the time-course experiments: increasing the reaction

time led to a marked decrease in the frequency of

5′-CC and increase in 5′-TT (Fig.  4b). Incontrast to the

enzyme titration experiment, the levels of 5′-AA and

3′-CC remained essentially the same. Overall, these

experiments demonstrated that under conditional en-

zyme deficiency and during the early reaction phase

(i.e., at the lowest enzyme-to-substrate ratio), T4Rnl1

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1183

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Fig. 4. Frequency of dinucleotides (%) in sequencing reads as a function of (a)T4Rnl1 amount (24-h ligation) and (b)ligation

duration (2  U T4Rnl1) for the mN8N set.

preferentially circularizes ssDNA substrates contain-

ing 5′-AA, 5′-CC, and 3′-AA. As these substrates become

depleted, ligation progressively shifts toward ODNs

containing 5′-TT and 3′-CC. ODNs with 5′-GG and 3′-GG

consistently represent the least favored substrates.

Summarizing the obtained data, it can be con-

cluded that in mixtures of ODN substrates, T4Rnl1

preferentially cyclizes ssDNA molecules containing

CC and TT dinucleotides at the donor end and AA at

the acceptor end. This preference may be explained

by stronger interactions between more hydrophobic

pyrimidine bases and amino acid residues in the en-

zyme’s active site  [40]. However, this hypothesis re-

quires further validation through detailed studies, in-

cluding molecular dynamics simulations. In addition,

steric factors such as DNA flexibility and conforma-

tional mobility likely play an important role in ssDNA

cyclization and warrant separate investigation. Final-

ly, it should be noted that enzymatic ssDNA cycliza-

tion competes with intermolecular ligation, proceeds

with lower efficiency, and is therefore influenced by

the dominance of competing reaction pathways.

CONCLUSION

Detailed understanding of the properties of DNA

and RNA ligases is essential for developing novel

molecular biology methods, molecular diagnostics

technologies, and non-biological application of DNA.

SAKHABUTDINOVA et al.1184

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

Here, we investigated the substrate specificity of

T4Rnl1 in the intramolecular ligation (cyclization) of

single-stranded DNA using ODNs of varying lengths

(23, 39, and 45  nt) with different 5′- and 3′-termi-

nal homodinucleotides. Ligation reactions were per-

formed using ODN mixtures representing all possi-

ble combinations of donor and acceptor nucleotides,

followed by the enrichment of ligation products by

RCA and nanopore sequencing of amplicons. This ap-

proach enabled the identification of sequence-depen-

dent preferences of T4Rnl1 toward DNA substrates.

For conformationally more flexible ODNs, the enzyme

preferentially cyclized substrates containing CC and

TT dinucleotides at the donor end and AA at the ac-

ceptor end, likely reflecting their higher affinity for

the enzyme active site. Substrates containing 5′-GG

and 3′-GG were least favored. For the shortest (23-nt)

ODNs, cyclization predominantly occurred with a

single substrate variant containing 5′-TT and 3′-AA

dinucleotides, presumably due to a higher ligation

efficiency for more flexible AT-rich DNA sequences.

The results of the study can provide guidance for

selecting nucleotide sequences to maximize circular

DNA formation efficiency.

Abbreviations

BSA bovine serum albumin

Exo I exonuclease I

NA nucleic acid

ODN oligodeoxyribonucleotide

PEG4000 polyethylene glycol 4000

RCA rolling circle amplification

T4Rnl1 T4 RNA ligase 1

Tt threshold time

Acknowledgments

Equipment used in this work was provided by the

Agidel and Biomika Centers for Collective Use.

Contributions

A.R.S. developed the study concept, supervised the

study, conducted experiments, analyzed and discussed

results, and wrote the text of the article; A.A.Sh. con-

ducted experiments, analyzed and discussed results;

R.R.G. analyzed and discussed results and edited the

manuscript.

Funding

This work was supported by the Russian Science

Foundation (project no. 25-24-00356).

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 declare that they have no

conflicts of interest.

Open access

This article is licensed under a Creative Commons

Attribution 4.0 International License, which permits

use, sharing, adaptation, distribution, and reproduc-

tion in any medium or format, as long as you give

appropriate credit to the original author(s) and the

source, provide a link to the Creative Commons li-

cense, and indicate if changes were made. The im-

ages or other third party material in this article are

included in the article’s Creative Commons license,

unless indicated otherwise in a credit line to the

material. If material is not included in the article’s

Creative Commons license and your intended use

is not permitted by statutory regulation or exceeds

the permitted use, you will need to obtain permis-

sion directly from the copyright holder. To view a

copy of this license, visit http://creativecommons.org/

licenses/by/4.0/.

REFERENCES

1. Uhlenbeck, O. C., and Gimport, R. I. (1982) T4 RNA Ligase/The enzymes, XV, Academic Press Inc. New York,

https://doi.org/10.1016/0968-0004(83)90258-X.

2. Tessier, D. C., Brousseau, R., and Vernet, T. (1986) Ligation of single-stranded oligodeoxyribonucleotides by T4

RNA ligase, Anal. Biochem., 158, 171-178, https://doi.org/10.1016/0003-2697(86)90606-8.

3. Bullard, D. R., and Bowater, R. P. (2006) Direct comparison of nick-joining activity of the nucleic acid ligases

from bacteriophage T4, Biochem. J., 398, 135-144, https://doi.org/10.1042/BJ20060313.

4. Bodulev, O. L., and Sakharov, I. Y. (2020) Isothermal nucleic acid amplification techniques and their use in

bioanalysis, Biochemistry (Moscow), 85, 147-166, https://doi.org/10.1134/S0006297920020030.

5. Garafutdinov, R. R., Sakhabutdinova, A. R., Gilvanov, A. R., and Chemeris, A. V. (2021) Rolling circle amplifica-

tion as a universal method for the analysis of a wide range of biological targets, Russ. J. Bioorg. Chem., 47,

1172-1189, https://doi.org/10.1134/S1068162021060078.

6. Xu, L., Duan, J., Chen, J., Ding, S., and Cheng, W. (2021) Recent advances in rolling circle amplification-based

biosensing strategies – a review, Anal. Chim. Acta, 1148, 238187, https://doi.org/10.1016/j.aca.2020.12.062.

7. Del Piano, A., Kecman, T., Schmid, M., Barbieri, R., Brocchieri, L., Tornaletti, S., Firrito, C., Minati, L., et al. (2022)

Phospho-RNA sequencing with circAID-p-seq, Nucleic Acids Res., 50, e23, https://doi.org/10.1093/nar/gkab1158.

SUBSTRATE SPECIFICITY OF T4 RNA LIGASE 1 IN DNA CYCLIZATION 1185

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

8. Maguire, S., and Guan, S. (2022) Rolling circle reverse transcription enables high fidelity nanopore sequencing

of small RNA, PLoS One, 17, e0275471, https://doi.org/10.1371/journal.pone.0275471.

9. Sakhabutdinova, A. R., Maksimova, M. A., and Garafutdinov, R. R. (2017) Synthesis of circular DNA templates with

T4 RNA ligase for rolling circle amplification, Mol. Biol., 51, 639-646, https://doi.org/10.1134/S0026893317040161.

10. Chidchob, P., and Sleiman, H. F. (2018) Recent advances in DNA nanotechnology, Curr. Opin. Chem. Biol., 46,

63-70, https://doi.org/10.1016/j.cbpa.2018.04.012.

11. Garafutdinov, R. R., Chemeris, D. A., Sakhabutdinova, A. R., Kiryanova, O. Yu., Mikhaylenko, C. I., and

Chemeris, A. V. (2022) Encoding of non-biological information for its long-term storage in DNA, Biosystems,

215-216, 104664, https://doi.org/10.1016/j.biosystems.2022.104664.

12. Kim, J., and Mrksich, M. (2010) Profiling the selectivity of DNA ligases in an array format with mass spectrom-

etry, Nucleic Acids Res., 38, e2, https://doi.org/10.1093/nar/gkp827.

13. Lohman, G. J., Bauer, R. J., Nichols, N. M., Mazzola, L., Bybee, J., Rivizzigno, D., Cantin, E., and Evans, T. C., Jr.

(2016) A high-throughput assay for the comprehensive profiling of DNA ligase fidelity, Nucleic Acids Res., 44,

e14, https://doi.org/10.1093/nar/gkv898.

14. Becker, H. F., Héliou, A., Djaout, K., Lestini, R., Regnier, M., and Myllykallio, H. (2017) High-throughput sequenc-

ing reveals circular substrates for an archaeal RNA ligase, RNA Biol., 14, 1075-1085, https://doi.org/10.1080/

15476286.2017.1302640.

15. Potapov, V., Ong, J. L., Langhorst, B. W., Bilotti, K., Cahoon, D., Canton, B., Knight, T. F., Evans, T. C. J., and

Lohman, G. J. S. (2018) A single-molecule sequencing assay for the comprehensive profiling of T4 DNA ligase

fidelity and bias during DNA end-joining, Nucleic Acids Res., 46, e79, https://doi.org/10.1093/nar/gky303.

16. Duckworth, A. T., Bilotti, K., Potapov, V., and Lohman, G. J. S. (2023) Profiling DNA ligase substrate specificity with

a pacific biosciences single-molecule real-time sequencing assay, Curr. Protoc., 3, e690, https://doi.org/10.1002/

cpz1.690.

17. Romaniuk, E., McLaughlin, L. W., Neilson, T., and Romaniuk, P. J. (1982) The effect of acceptor oligoribonucleotide

sequence on the T4 RNA ligase reaction, Eur. J. Biochem., 125, 639-643, https://doi.org/10.1111/j.1432-1033.1982.

tb06730.x.

18. Maistrenko, V.F., Pustoshilova, N.M., Klyagina, V.P., Sedelnikova, E.A., Smolyaninova, O.A., and Zhenodarova,S.M.

(1984) Substrate specificity of T4 RNA ligase: Effect of the nucleotide sequence of the minimal phosphate ac-

ceptor on the efficiency of intermolecular ligation [in Russian], Bioorg. Khim., 10, 498-505.

19. Zhenodarova, S. M., Klyagina, V. P., Maistrenko, V. F., Pustoshilova, N. M., Smolyaninova, O.A., and Soboleva,I.A.

(1989) Substrate specificity of T4 RNA ligase. Effect of nucleotide composition of substrates and length of the

phosphate donor on the efficiency of intermolecular ligation [in Russian], Bioorg. Khim., 15, 478-483.

20. Hafner, M., Renwick, N., Brown, M., Mihailović, A., Holoch, D., Lin, C., Pena, J. T., Nusbaum, J. D., Morozov, P.,

Ludwig, J., Ojo, T., Luo, S., Schroth, G., and Tuschl, T. (2011) RNA-ligase-dependent biases in miRNA represen-

tation in deep-sequenced small RNA cDNA libraries, RNA, 17, 1697-1712, https://doi.org/10.1261/rna.2799511.

21. Jayaprakash, A. D., Jabado, O., Brown, B. D., and Sachidanandam, R. (2011) Identification and remediation of

biases in the activity of RNA ligases in small-RNA deep sequencing, Nucleic Acids Res., 39, e141, https://doi.org/

10.1093/nar/gkr693.

22. Zhuang, F., Fuchs, R. T., Sun, Z., Zheng, Y., and Robb, G. B. (2012) Structural bias in T4 RNA ligase-mediated

3′-adapter ligation, Nucleic Acids Res., 40, e54, https://doi.org/10.1093/nar/gkr1263.

23. Ma, Z., Chen, H., Yang, Y., Gao, S., Yang, J., Cui, S., Zhou, S., Jiang, B., Zou, B., Sun, M., and Wang, L. (2024)

Characterization of an ssDNA ligase and its application in aptamer circularization, Anal. Biochem., 685, 115409,

https://doi.org/10.1016/j.ab.2023.115409.

24. Harrison, B., and Zimmerman, S. B. (1984) Polymer-stimulated ligation: enhanced ligation of oligo- and poly-

nucleotides by T4 RNA ligase in polymer solutions, Nucleic Acids Res., 12, 8235-8251, https://doi.org/10.1093/

nar/12.21.8235.

25. Garafutdinov, R.R., Kupova, O.Y., and Sakhabutdinova, A. R. (2024) Influence of nucleotide context on non-specif-

ic amplification of DNA with Bst exo

–

DNA polymerase, Biochemistry (Moscow), 89, 53-64, https://doi.org/10.1134/

S0006297924010036.

26. Zhang, D. Y., Brandwein, M., Hsuih, T. C., and Li, H. (1998) Amplification of target-specific, ligation-dependent

circular probe, Gene, 211, 277-285, https://doi.org/10.1016/s0378-1119(98)00113-9.

27. Oscorbin, I., and Filipenko, M. (2023) Bst polymerase – a humble relative of Taq polymerase, Comput. Struct.

Biotechnol. J., 21, 4519-4535, https://doi.org/10.1016/j.csbj.2023.09.008.

28. Garafutdinov, R. R., Gilvanov, A. R., and Sakhabutdinova, A. R. (2020) The influence of reaction conditions on

DNA multimerization during isothermal amplification with Bst DNA polymerase, Appl. Biochem. Biotechnol., 190,

758-771, https://doi.org/10.1007/s12010-019-03127-6.

SAKHABUTDINOVA et al.1186

BIOCHEMISTRY (MOSCOW) Vol. 91 No. 7 2026

29. Sakhabutdinova, A. R., and Garafutdinov, R. R. (2025) Mechanism of DNA multimerization caused by strand-dis-

placement DNA polymerases, Anal. Biochem., 703, 115876, https://doi.org/10.1016/j.ab.2025.115876.

30. Doronin, S. V., Nevinsky, G. A., Malygina, T. O., Podust, V. N., Khomov, V. V., and Lavrik, O. I. (1989) The effi-

ciency of interaction of deoxyribonucleoside-5′-mono-, di- and triphosphates with the active center of E.  coli

DNA polymerase I Klenow fragment, FEBS Lett., 259, 83-85, https://doi.org/10.1016/0014-5793(89)81500-5.

31. Kolocheva, T. I., Nevinsky, G. A., Volchkova, V. A., Levina, A. S., Khomov, V. V., and Lavrik, O. I. (1989) DNA

polymerase I (Klenow fragment): role of the structure and length of a template in enzyme recognition, FEBS

Lett., 248, 97-100, https://doi.org/10.1016/0014-5793(89)80439-9.

32. Hook, P. W., and Timp, W. (2023) Beyond assembly: the increasing flexibility of single-molecule sequencing

technology, Nat. Rev. Genet., 24, 627-641, https://doi.org/10.1038/s41576-023-00600-1.

33. Senol Cali, D., Kim, J. S., Ghose, S., Alkan, C., and Mutlu, O. (2019) Nanopore sequencing technology and tools

for genome assembly: computational analysis of the current state, bottlenecks and future directions, Brief Bio-

inform., 20, 1542-1559, https://doi.org/10.1093/bib/bby017.

34. Zyrina, N. V., and Antipova, V. N. (2021) Nonspecific synthesis in the reactions of isothermal nucleic acid am-

plification, Biochemistry (Moscow), 86, 887-897, https://doi.org/10.1134/S0006297921070099.

35. Johnson, S., Chen, Y. J., and Phillips, R. (2013) Poly(dA:dT)-rich DNAs are highly flexible in the context of DNA

looping, PLoS One, 8, e75799, https://doi.org/10.1371/journal.pone.0075799.

36. Geggier, S., and Vologodskii, A. (2010) Sequence dependence of DNA bending rigidity, Proc. Natl. Acad. Sci. USA,

107, 15421-15426, https://doi.org/10.1073/pnas.1004809107.

37. Ortiz, V., and de Pablo, J. J. (2011) Molecular origins of DNA flexibility: sequence effects on conformational and

mechanical properties, Phys. Rev. Lett., 106, 238107, https://doi.org/10.1103/PhysRevLett.106.238107.

38. Zhang, Y., and Crothers, D. M. (2003) High-throughput approach for detection of DNA bending and flexibility

based on cyclization, Proc. Natl. Acad. Sci. USA, 100, 3161-3166, https://doi.org/10.1073/pnas.0530189100.

39. Joffroy, B., Uca, Y. O., Prešern, D., Doye, J. P. K., and Schmidt, T. L. (2018) Rolling circle amplification shows a

sinusoidal template length-dependent amplification bias, Nucleic Acids Res., 46, 538-545, https://doi.org/10.1093/

nar/gkx1238.

40. El Omari, K., Ren, J., Bird, L. E., Bona, M. K., Klarmann, G., LeGrice, S. F., and Stammers, D. K. (2006) Molecu-

lar architecture and ligand recognition determinants for T4 RNA ligase, J. Biol. Chem., 281, 1573-1579, https://

doi.org/10.1074/jbc.M509658200.

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.