structural and kinetic aspects of chemical reactions in dna

© 7997 Oxford University Press Nucleic Acids Research, Vol. 19, No. 11 3073

Structural and kinetic aspects of chemical reactions inDNA duplexes. Information on DNA local structureobtained from chemical ligation data

Nina G.Dolinnaya*, Alia V.Tsytovich, Vladimir N.Sergeev, Tatyana S.Oretskaya andZoe A.ShabarovaThe Belozersky Laboratory of Molecular Biology and Bioorganic Chemistry and ChemistryDepartment, Moscow State University, 119899, Moscow, USSR

Received January 18, 1991; Revised and Accepted May 3, 1991

ABSTRACT

Chemical ligation of oligonucleotides in double-stranded helices has been considered in its structural-kinetic aspect. A study was made of (i) two series ofDNA duplexes with various arrangements of reactinggroups in the ligation junction induced by mispairingor by alteration of furanose structure (the replacementof dT unit with rU, all, IU, xU, dxT ones) and of (ii) eightsynthetic water-soluble carbodiimides with differentsubstituents at N1 and N3 atoms. We assumed thatsome information on the local structure of modifiedsites in the duplex can be obtained from kineticparameters of oligonucleotide coupling reaction. Theratio of kinetic constants k3/(k2 + k3) for productive andnonproductive decomposition of the activatedphosphomonoester derivative apparently reflects thereaction site structure: for a given duplex thisparameter is virtually independent of the condensingagent composition. Based on the analysis of thechemical ligation kinetics a suggestion has been madeabout the conformation of some modified units in thedouble helix.

INTRODUCTION

The interest in chemical reactions in spatially organized nucleotidesystems, stirred up by the works on RNA self-cleavage and RNAself-splicing, has markedly increased in recent years. This classof reactions includes chemical ligation, i.e. covalent joining ofoligonucleotide blocks on a complementary template under theaction of chemical agents [1]. Studies on the chemical ligationpatterns allow valuable information to be obtained on the reactiveability and local structure of double-helical nucleic acids. Suchapproach are also quite promising in respect of designing thenonenzymic means of assembling extended DNAs including thosewith preset modifications.

Obviously the efficiency of chemical ligation is associated withmutual arrangement of the reacting groups in the double helix.

We have earlier shown that a local structural alterations at thenick, caused e.g. by a mismatched base pair or a modified sugarresidue, greatly affects the reaction rate [2].

The present work analyzes the interrelationship between thekinetic parameters of chemical ligation and the structure of thereaction site. To this end, we have introduced a parameter whichis a relation of the rate constants for individual reaction steps.This parameter is shown to be a structural invariant of a givenduplex, i.e. to be practically independent of the composition ofchemical agents. The latter were a set of specially synthesizedwater-soluble carbodiimides differing in the nature of substituentsat Nl and N3:

C2H5NCNCH2CH2CH2NH(CH3)2C1 CDI 1

CH3NCNCH2CH2CH2NH(CH3)2C1 CDI 2

C2H5NCNCH2CH2NH(CH3)2C1 CDI 3CH3NCNCH2CH2NH(CH3)2C1 CDI 4C2H5NCNCH2CH2CH2N(CH3)3I CDI 5CH3NCNCH2CH2CH2N(CH3)3I CDI 6C2H5NCNCH2CH2N(CH3)3I CDI 7CH3NCNCH2CH2N(CH3)3I CDI 8

The substrates were DNA duplexes la (synthesis of aphosphodiester bond) and Ia'(synthesis of a pyrophosphate bond):

5' A- A-C-C-T- A-C-C-Tl pG-G-T-G-G-T3' T-G-G-A-T-G-G-A- C-C-A-C-C-A

duplex la

5' A-A-C-C-T-A-C-C-Tp' pG-G-T-G-G-T3' T-G-G-A-T-G-G-A - C-C-A-C-C-A

duplex la'

Then the kinetic aspects of chemical ligation were studied induplexes with a variable reaction site, using such modificationsas would alter the relative positions of the groups while leaving

* To whom correspondence should be addressed at Department of Molecular Biology, Lewis Thomas Lab., Princeton University, Princeton, NJ 08544-1014,USA (until November 1991)

Downloaded from https://academic.oup.com/nar/article-abstract/19/11/3073/2387773by gueston 09 April 2018

3074 Nucleic Acids Research, Vol. 19, No. 11

Scheme t

Duplex ] 5' A AC C T AC CtiffilG T G GT

3 ' d ( T G G A T G G A C C A C C A )

a b cT G

.oj-^pj0-

V f..oJoHp^O"

J X.'sfpj*

•°FPI°

5 ' d ( A A C C T A C

3' d (T GGA TGlG ACClACCA)

y V'G G| T G G T)

C T pGG C TpS G OT pGGSA CC

I

Duplex J 5' A c s G AIM c l c A G G A G T G A C '3' d ( G C C T A G G T C C T C A C )

bT C

.o|OMp|°-d

T C

A i°-.0\| HOsl

T C

e

u c

g

u c

..0$ Psl

T C

oQ p iI

u cHO] 1

Up U.Os| "HOsl

their chemical nature unchanged. Two families of DNA duplexesof different primary structure were constructed (Scheme 1).

Modifications at the nick (indicated by arrow on the Scheme,the structure expanded in the squares) included:

1. Altered position of the phosphate group: 3'- or 5'-terminal(duplexes Ha and lib).

2. Replacement of the 3'-terminal thymidine with ribouridine(duplexes Ic, Eg) or a nucleotide with reversed configuration atC-2' and/or C-3' atoms of furanose, such a modified units beingeither the phosphate acceptor (duplexes Ib, Id—f, lie and He)or donor (duplexes lid and Ilf).

3. Insertion of an 'extra' purine or pyrimidine residue at thedonor or the acceptor end (duplexes Ig—k).

4. Replacement of the dG • dC pair one base step away fromthe ligation junction with a dG-dG mismatch (duplex H).

Analysis of the kinetic parameters of chemical ligation alloweda suggestion about the local conformation of the region adjacentto the nick.

MATERIALS AND METHODSGeneralThe following reagents were used: l-ethyl-3-(3'-dimethylamino-propyl)-carbodiimide, hydrochloride, CDI 1 (Merck), the restof the water-soluble carbodiimides (CDI 2 -8 ) were synthesizedas described previously [3], y-32V ATP (Amersham), T4poly nucleotide kinase, T4 DNA ligase (USSR), PNase A(Reanal), PNase T2 (Sankyo).

Buffer solution: A (for CDI-induced ligation)—50 mM MES[4], pH 6.0, 20 mM MgCl2; B (for PNase A)-0.1 M Tris-HCl,pH 7.2, 2 mM EDTA; C (for RNase T2)—0.04 M ammoniumacetate, pH 4.5, 2 mM EDTA.

Preparation of synthetic oligomersOligodeoxynucleotides employed to form DNA duplexes I andII were synthesized by the phosphoramidite method on a Cyclon

(Biosearch) DNA synthesizer. Nucleosides with reversedconfiguration at C2' and (or) C3' atoms were obtained fromcorresponding natural nucleosides via the anhydroderivatives [5].The introduction of protecting groups was effected by the standardprocedure. To phosphitylate MeOTrdxT and MeOTrr°U thereaction time was increased to 5 and 24 hours, respectively.Modified hexanucleotides with the 3'-terminal phosphate groupwere obtained from heptanucleotides d(ACGGA)MrU, where Mis dxT or aU, by periodate oxidation of the 2',3'-cis-hydroxylsystem followed by /^-elimination [6]; oligonucleotidesd(AAAAp) and d(AACCTACCTp) were prepared in the sameway. Synthesized oligomers were finally purified by HPLC ona Lichrocorb C-18 column (4.6x250 mm, Merck) in an Altexchromatograph.

Optical measurementsNucleotide concentration was determined on a Spectronic 2000spectrophotometer (Bouch and Lomb). The molar extinctioncoefficients, e260, of anomalous nucleotide residues were takenequal to ejeo °f natural analogs. The melting curves of duplexesI and II were measured on a Cary 219 spectrophotometer (Varian)in thermostatted quartz cells of 1 mm path length.

Hydration of water-soluble carbodiimidesInitial carbodiimide concentration was 0.2 M. Hydration wasstudied in buffer A or in buffer A containing 5 • 10~4 M (permonomer) dp A, d(AAAA) or d(AAAAp). Carbodiimideconversion was assayed spectrophotometrically using aquantitative color reaction for the -N=C=N- group [3,7].

Chemical and enzymatic ligation of oligonucleotidesNucleotide concentration of duplexes I and II was about 10~3

M (per monomer); 5'-32P labelled oligonucleotides were addedin a 1.5-fold deficiency as compared to the unlabelled ones. Theconcentration of CDI 1-8 and incubation temperature werevaried in different experiments (see footnotes to Tables 1 and3). The changes in the concentrations of compounds werefollowed by withdrawing samples; carbodiimide conversion wasassayed spectrophotometrically as described above, oligo-nucleotide conversion by electrophoresis in denaturing 20%polyacrylamide gel. The nucleotide fraction was precipitated with2% LiClO4 in acetone beforehand.

Enzymatic ligation in duplexes Ic, Ilg under the action of T4DNA ligase was carried out as in [8].

Analysis of ligation product32P labelled oligonucleotide was treated with 0.3 N NaOH at37°C for 3 hours. Then the reaction mixture was neutralized by40% aqueous solution of acetic acid, oligonucleotide fraction wasprecipitated and analyzed by 20% polyacrylamide gelelectrophoresis.

Hydrolysis of the ligation product with RNases A or T2 wascarried out in buffer B or C, respectively, for 1.5 hours at 37°C.

Statistical treatment of the data was carried out by the leastsquares method.

RESULTS AND DISCUSSION

Kinetics of chemical ligation

The general kinetic scheme of chemical ligation induced witha water-soluble carbodiimide as applied to homooligomercondensation in a triple-helical complex was proposed by Popov


et al. [9]. The heterogeneous single-nicked DNA duplexesconsidered here offer a much better model allowing quantitativeinterpretation of the experimental data. Our theoreticalconsideration is based on Popov's kinetic scheme:

S + X —

X •IV

^ S + YII

IIIP + Y

Nucleic Acids Research, Vol. 19, No. 11 3075

Scheme 2

HO

where S is the nicked duplex,P is the condensation product (ligated duplex),X is the activating agent,Y is the hydration product of the activating agent,SX is the activated adduct.The scheme was constructed with the following assumptions:1. In the reaction conditions the duplex is regarded as a perfect

helix.2. Reactions II (hydration of the activated adduct) and IV

(hydration of the activating agent) are regarded as pseudofirst-order reactions.

3. Modification of bases in the oligonucleotides forming theduplex is disregarded.

4. The internucleotide bond has no catalytic influence onhydration of the activating agent.

This scheme is described by a system of five equations whichcannot be completely solved in analytical form without additionalassumptions. Let us consider a possible case when steps I andIV i.e. reactions with nucleophiles, are rate-limiting. Besides,let us assume that k,[S] < < IC4. Transformation of the kineticequations taking into account these assumptions yielded thefollowing equation [9]:

In[S]o = k,[X]0

[S] k4 k? + k.(1 - (1)

The rate of accumulation of the ligation product is determinedboth by the relation between k,[S] and ^(showing what portionof the reagent is used to activate the phosphate group in the nick,as related to the amount undergoing hydration) and by the relationbetween constants k2 and k3 (showing the portion of theactivated adduct SX decomposing by the productive pathway).

Let us consider the possibility of determining the rate constantsfor individual chemical ligation steps. The constant IC4 isobtained in a separate experiment. The rate constant for formationof the activated adduct, k,, can be determined with the use ofmodel systems, e.g. a mononucleotide or a phosphorylatedoligonucleotide, whereby the general reaction scheme issimplified:

+ X —i- SX -2-S

X "V

where S is the mono- or oligonucleotide with phosphomonoestergroup,

X is the activating agent,Y is the hydration product of the activating agent,SX is the activated adduct.

Solution of the equations corresponding to this scheme,assuming a limiting step of adduct formation, leads to thefollowing expression:

k)[S]0 + k} = - • In——?

Thus, knowing the slope a of the straight line in the plot ln[X]vs t, the rate constant for activated adduct formation can beobtained from the equation:

k, = k* -

[S]owhere k* = tga

Within the assumptions made (SX formation is the limiting step)the absolute values of the decomposition rate constants of theactivated adduct cannot be determined. However, with the useof the transformed equation (1) the slope of the straight line inthe plot

- 1)

yields the apparent rate constant of the coupling reaction,k3-ki/(k2 + k3). Knowing k1; it is easy to determine the ratiok3/(k2+k3) reflecting the portion of the activated adductdecomposing to give the chemical ligation product.

Let us analyze the factors affecting this ratio. In the courseof breakdown of SX (Scheme 2) the activated phosphate canparticipate in two competing reactions-with the hydroxyl of theneighbouring nucleotide unit and with a water molecule.Obviously, the k2 and k3 values are proportional to the partialpositive charge on the phosphorus atom and nucleophilicity ofthe reacting groups. When different condensing agents are usedthe absolute values of the constants would change, but if thereactivity of the attacking groups is the same and there are nosubstantial distortions in the structure of the ligation site, thenthe k3/(k2+k3) ratio for a given complex will remain constant.

The rightfulness of the assumptions made in composing thechemical ligation kinetics is confirmed by the following data.

1. Analysis of the melting curves of duplexes la and la' showsthat at the temperature of 10°C the amount of unpaired moleculescan be neglected. The Tm of duplexes la and la' are 35.0 and34.0°C, respectively, and the water-soluble carbodiimide lowersthe Tm of nucleotide complexes only by a few degrees.

2. The kinetic curves of hydration of CDI 1 - 8 are linearizedwell in the plot ln[X] vs t [3], which confirms the pseudofirstorder of the reaction.

3. During chemical ligation under conditions when the doublehelix is stable, modification of the nucleotide material by CDIis insignificant [2].



Table 1. Kinetic characteristics of water-soluble carbodiimides*

Numberof CDI

1

2:

5

6

7

8

1C4-102,h"1

0.19±±0.03

0.12±±0.02

4.8±±0.3

8.6±±1.0

1.3±±0.1

2.0±±0.3

Buffer

k|.M-' -h- '

3.6±±1.0

4.2±±0.9-

-

6.1 ±±3.1—

A, 0°Ck,-k3**

k2+k3

0.64±±0.02

0.66±±0.12

1.33±±0.14

1.8±±0.5

0.37 ±±0.04

0.68 ±±0.06

k3**

k2+k3

0.17±±0.03

0.15±±0.03—

-

0.06 ±±0.04-

k.,102,h"1

0.55 ±±0.05

0.41 ±±0.0613.1 ±

±0.717.3 ±

±1.13.6±

±0.34.9 ±

±0.4

Buffer

M - ' h " 1

8.4 ±±1.7

9.5±±2.3-

-

14.2 ±±6.2-

A, 10°Ck,-k3***

k2 + k3

0.038 ±±0.007

0.044 ±±0.004-

-

0.054±±0.008_

k3***

k2 + k3

0.0046 ±±0.0007

0.0046±±0.0007-

-

0.0038±±0.0015-

* The k! and 1^ values taken from 13/.** Pyrophoshate bond synthesis in duplex la'; carbodiimide concentration 0.1 M; the ratios of constants are presented per onephosphomonoester group in the reaction site.*** Phosphodiester bond synthesis in duplex la; nucleotide concentration (per monomer) 10"3 M, carbodiimide concentration 0.2 M.

4. In contrast to reactions in organic solvents wherecarbodiimide is also consumed in the interaction with theinternucleotide phosphate [10], in the aqueous solution we couldobserve no appreciable influence of phosphodiester groups oncarbodiimide hydration. Thus, in the presence of 5 • 10~4 M (permonomer) of d(AAAA) the CDI 1 hydration curve practicallycoincides with the control one obtained in buffer A. The ratesof interaction of CDI 1 with the 3'- and the 5'-terminal phosphategroups are approximately the same. This is evidenced by closek] values obtained in buffer A containing dpA or d(AAAAp):3.2 and 3.3 h- 'M" 1 , respectively.

Template-directed ligation of oligomers with different water-soluble carbodiimidesWhen designing the set of CDI 1 to 8, the rationale was asfollows. CDI 1 is known to undergo tautomeric cyclization inaqueous solutions [11]. The structural alterations introduced byus allow elucidation of the roles of both tautomeric forms incoupling reactions. The propensity of CDI 1—8 to ring-chaintautomerism and the hydration constants in the absence and inthe presence of a phosphate donor—mononucleotide—arepresented elsewhere [3]. It turned out that carbodiimidescontaining two methylene links instead of three and a terminaltertiary amino group are prone to form a stable five-memberedring hindering their interaction with nucleophiles. Carbodiimideswith a quaternary ammonium group can exist in an aqueous buffersolution exclusively in the linear form which is the reactive one.

For compounds that are short-lived in buffer A, such as CDI5, 6 and 8, k, cannot be determined exactly, since thecomponent corresponding to carbodiimide interaction with thephosphomonoester group is hard to discern on the backgroundof its own rapid hydration. The k) and K» values for somecarbodiimides that are most promising as ligation agents arepresented in Table 1.

The activity of carbodiimides was first assessed by their abilityto produce a pyrophosphate bond in duplex la'. This duplex waschosen as a test system owing to the high reaction rate, whichspeeds up and facilitates 'selection' of carbodiimides by theiractivity. As can be seen in Fig. 1, CDI 3 and 4 forming a stablefive-membered ring proved ineffective in the chemical ligation

3 2 4 5 6 7

XC-

BPB-

Figure 1. Electrophoretic analysis (20% denaturing PAG) of reaction mixturescontaining duplex la', after 24 h of incubation in buffer A at O°C with variouscarbodiimides (initial CDI concentration is 0.1 M). The 5'-32P label wasincorporated in hexanucleotide d(pGGTGGT). The lane number corresponds tothat of carbodiimide. The chain length of oligonucleotides is indicated to the right.The arrows indicate the positions of markers: bromophenol blue (BPB) and xylenecyanol (XC).

Figure 2. Accumulation of the duplex la' ligation product under the action ofvarious carbodiimides: (1) CDI 1, (2) CDI 5, (3) CDI 7, (4) CDI 2, (5) CDI6, (6) CDI 8; and hydration of the corresponding carbodiimides at 0°C in bufferA: (1') CDI 1, (2') CDI 5, (3') CDI 7, (4') CDI 2, (5') CDI 6, (6') CDI 8.For reaction conditions and assay techniques see Materials and Methods andfootnotes to Table 1.

reactions. With the other carbodiimides kinetic curves wereobtained for accumulation of the ligation product with apyrophosphate bond (Fig. 2). It should be noted that in thestandard reaction time, 6 h, only 20-40% of even the most activecondensing agents undergoes hydration (Fig. 2).


1

3 5 days 5 days

Figure 3. (a) Hydration of various carbodiimides at 10°C in buffer A: (1) CDI1, (2) CDI 2, (3) CDI 6, and (4) CDI 7. (b) Accumulation of the duplex la ligationproduct with various carbodiimides: (1) CDI 1, (2) CDI 2, (3) CDI 6, and (4)CDI 7. For reaction conditions and assay techniques see Materials and Methodsand footnotes to Table 1.

Figure 4. Linear anamorphoses of the accumulation curves for the duplex laligation product with (1) CDI 1, (2) CDI 2, and (3) CDI 7.

The same group of carbodiimides were used as condensingagents to synthesize a phosphodiester bond in duplex la (Fig. 3).As expected, the efficiency of intemucleotide bond formation withlinear CDI 5 - 8 was low because of their rapid hydration underreacting conditions. The most efficient phosphodiester bondformation was achieved with the less reactive but 'longer living'CDI 1 and 2.

The kinetic data on the accumulation of the ligation productin duplexes la and la' under the action of CDI 1, 2 and 5 - 8allowed the apparent reaction constants ki-k3 / (k2+k3) to bedetermined (Table 1). Evaluation of k3/(k2 + k3) was onlypossible for CDI 1, 2 and 7 for which the kj is reliably known(Table 1). In the case of duplex la ' (pyrophosphate bondformation), taking into account that there are two phosphategroups in the reaction site, the k3 /(k2+k3) value was divided bytwo.

Analysis of the data obtained allows the following conclusionsto be made.

1. There is an interrelationship between the composition ofcarbodiimides and the kinetic parameters of their reactions, thehydration rate constants and the apparent constants of chemicalligation.

2. The k3/(k2+k3) values for a given duplex are close fordifferent carbodiimides (taking into account the significantinaccuracy in determining the k, for CDI 7): ca. 0.004 withduplex I (phosphodiester bond formation) (Fig. 4, Table 1) andca. 0.13 for duplex n (pyrophosphate bond formation) (Table 1).


1 2 3 4XC —

BPB 6

Figure 5. Electrophoretic analysis (20% denaturing PAG) of reaction mixturescontaining duplexes (1) II, (2) la, (3) Ig, and (4) Di, after four days of incubationin buffer A at 10°C, initial CDI 1 concentration 0.3 M. The 5'-32P label wasincorporated in hexanucleotide d(pGGTGGT). Mixtures 2 and 4 also contain a5'-32P labelled 14-membered template oligonucleotide. Designations as in Fig. 1.

1 a17

BPB —

Figure 6. Electrophoretic analysis (20% denaturing PAG) of reaction mixturescontaining duplexes (1) Ilf, (2) lib and (3) lid, after three days of incubationin buffer A at 10°C, initial CDI1 concentration 0.2 M. The 5'-32P label wasincorporated in d(pACGGA)Mp. Designations as in Fig. 1.

Consequently, this parameter can serve as an invariant of theduplex. Obviously this is only true when the active phosphategroup derivative of the oligonucleotide does not distort thestructure of the reactive site.

3. Only a minor portion of the water-soluble carbodiimide isspent to activate the phosphomonoester groups: ca. 7% of thecarbodiimides capable of cyclization (CDI 1 and 2) and ca. 1.3%of noncyclizing ones (CDI 7). In the course of pyrophosphatebond formation ca. 13% of the activated adduct decomposes viathe productive pathway, whereas with phosphodiester bondformation this portion decreases to 0.4%. The results obtainedprovide an assessment of the relative nucleophilicity of thehydroxyl and the phosphomonoester groups attacking theactivated phosphate (there is a 30-fold difference, which doesnot contradict the published data [9]).

Interconnection between kinetic parameters and DNA localconformation at the ligation junctionLigation in duplexes I and II was carried out using CDI 1 asthe condensing agent. Among the set tested, it was thiscarbodiimide that proved the most suitable for the formation ofa phosphodiester bond.

A preliminary study was made of thermal stability of theduplexes I and II. It turned out that none of the modificationsused except the G • G mismatch (which lowered the Tm of the



1 2 3 A 5 6 7 6 9 10 11 12

BPB-

106

Figure 7. Radioautographs of 20% PAG electrophoresis of oligonucleotides:d(AACCTACC)rU*pd(GGTGGT) obtained by chemical ligation (1,2), controlr(*pAAAUUU) (3,4), d(AACCTACC)rU*pd(GGTGGT) obtained by enzymaticligation (5,6), d(AACCTACC)xU*pd(GGTGGT) (7,8), d(AACCTA-CC)aU*pd(GGTGGT) (9,10) and d(AACCTACC)lU*pd(GGTGGT) (11,12)obtained by chemical ligation, before (odd lanes) and after (even lanes) the treatmentwith PNase A. Designations as in Fig. 1. For reaction conditions see Materialsand Methods and footnotes to Table 3.

Table 2. Enzymatic and chemical hydrolysis of ligation products in duplexes Ic-f.For reaction conditions see Materials and Methods

No

1.

2.3.4.5.

Oligonucleotide structure

d(AACCTACC)rUd(GGTGGT)d(AACCTACC)lUd(GGTGGT)d(AACCTACC)aUd(GGTGGT)d(AACCTACC)xUd(GGTGGT)d(AAAUUU)*

RNase

30000

100

Hydrolysis extent, %A RNase T2

40---100

0.3 N NaOH

100100

00

100

* Hexaribonucleotide was used as a control.

double helix by 10°C) had an appreciable effect on duplexstability (the Tm of duplexes I was about 33 °C, duplexes IIabout 47°C).

Under optimal conditions (buffer A, temperature 10°C,0.2—0.3 M CDI) coupling reaction proceeded in all duplexestested but with different efficiency. Figures 5 and 6 exemplifythe results of electrophoretic analysis of some reaction mixtures.

Before analyzing the kinetic patterns, we shall present dataconfirming the structure of ligation products. The primarystructure of 15 — 17-membered oligomers containing noanomalous nucleotide units was confirmed by Maxam-Gilbertprocedure. The nature of the bond formed with the participationof rU as well as aU, 1U, or xU was proved by alkaline andenzymic hydrolysis (with RNases A and T2) of thecorresponding oligomers. The conditions of hydrolysis werechosen so as to provide complete hydrolysis of a controloligoribonucleotide r(AAAUUU) to tetranucleotide r(AAAU)with RNase A or to mononucleotides with RNase T2. Theresults of RNase A cleavage are shown in Fig. 7, and the restof the data are summarized in Table 2. As can be seen, theproduct of enzymatic ligation in duplex Ic, which contains aphosphodiester bond between rU and dG, is completelyhydrolyzed by the ribonucleases to the initial nona- andhexanucleotides (see, for instance, Fig. 7, lanes 5 and 6), whereasthe chemical ligation product is hydrolyzed with the sameenzymes by 30-40% (Table 2); i.e., the portion of 3'-5' linkedoligomers amounts to about 40%. It has earlier been shown usingthe same techniques [2] that the chemical ligation product ofduplex Ilg is also a mixture of isomers, but in this case the ribo-and deoxyribonucleotide residues preferentially (in 87%) formthe natural 3'-5' phosphodiester bond.

Oligonucleotides 2—4 (Table 2) are not substrates for RNaseA, which points to the presence of unnatural bonds. These

CM

u 0 70 210 350

Figure 8. Linear anamorphoses of the accumulation curves for the ligation product(P) with CDI 1 (X) in duplexes (S): (1) la, 2 (II), (3) Ig, (4) Di, (5) Ic, (6) Ikand (7) Ib. For reaction conditions see footnotes to Table 3.

Figure 9. Linear anamorphoses of the accumulation curves for the ligation product(P) in duplexes (S): (1) lib, (2) Ha, (3) Ilg at O°C (a) and (4) Hf, (5) He, (6)lid, (7) He at 10°C (b). For reaction conditions see footnotes to Table 3.

oligomers and d(AACCTACCxTGGTGGT) proved to be moreslowly cleaved by snake venom phosphodiesterase with maximalretardation observed for the dxT-containing oligomer (data notshown).

Let us now analyze the kinetic parameters of coupling reactionin duplexes I and II. The linear anamorphoses of the kineticcurves are presented in Figs. 8 and 9. The k3/(k2+k3) ratioswere obtained from their slopes (Table 3); the values of kj andIC4 are given in Table 1. It can be seen that the ratio of theconstants varies greatly for different duplexes. It would be riskyto attempt to derive conformational conclusions from these valuesalone. However, by analyzing the whole body of data for theseries of duplexes and experimental facts from earlierpublications, interesting information can be obtained on theinfluence of various modifications on the local structure of thedouble helix.

So, the k3/(k2+k3) values for duplexes la and D are practicallythe same (Table 3). This indicates that the structural perturbationcaused by the dG • dG pair does not affect the orientation of thereacting groups of the neighbouring unit and is therefore onlylocal. This conclusion does not contradict the data on the influenceof dGdG pairs on the physicochemical parameters ofoligonucleotide duplexes [12]. On the other hand 3IP NMR



Table 3. Kinetic parameters of CDI 1-unduced ligation in duplexes I and II*

Duplexnumber

Ligation site Coupling yieldin 6 days, %

Scheme 3S N

• 103

Duplex Iabcdefghk1

Duplex IIabcdefg

-dT'pdG--dxfpdG--rU'pdG--aU'pdG--xU'pdG--lU'pdG--d(TA)'pdG--d(TT)'pdG--dfpd(TG)--d(GT)'pdG-

-dT'pdC-dTp'dC--dxT'pdC--dxTp'dC--aU'pdC--aUp'dC--rU'pdC**

69283655548453767

80921218237524

4.75 ±0.281.27±0.161.75 ±0.54**

2.35 ±0.802.15±0.481.76 ±0.404.12±0.41

18.5±0.925.8±8.00.72 ±0.221.02 ±0.381.39±0.488.2±2.0

2.93±0.45*:

* Conditions: buffer A, nucleotide concentration (per monomer) 10 3 M,carbodiimide concentration 0.3 M for duplexes la—1, 0,2 M for duplexes Ila-g.Chemical ligation in duplexes Ha, b,g was carried out at O°C, for the rest at10°C.** Here k3 means the sum constant of both 3'-5' and 2'-5' phosphodiester bondformation.

The ratio of kinetic constants calculated on the basis of data in [2].

studies indicate that the distortion unduced by dG • dG mismatchis still felt several base pairs away [13].

An 'extra' nucleotide residue decreases the efficiency and therate of the chemical reaction; the k3/(k2+k3) value for duplexesIg-k falls by half as compared with the nonmodified duplex.According to the published data, an 'extra' adenine base [14]is integrated into the double helix, whereas thy mine at lowtemperature most probably remains outside the helix [15]. It canbe supposed that a single-strand break at the place of the defectincreases the extent of freedom of the unpaired unit. If the 'extra'unit is the phosphate donor (duplex Ik), then there is an additionalfactor diminishing the efficiency of coupling reaction; theactivated derivative proves less protected by the duplex from theattack by water.

An important and nontrivial observation is that the ligationparameters differ for the nonmodified duplexes la and Ha. Thismay be explained by the different nature of joined units: dT anddG in duplex la, dT and dC in duplex Ha (such a sequencedependence of the chemical ligation efficiency has been notedbefore [16]) as well as by the different secondary structure ofduplexes I and II. Elements of the A form in the structure ofduplexes I have been suggested on the basis of analysis of theCD spectra and the literature data [17 and its references]. Canthe lower efficiency of chemical ligation in duplex la as comparedwith duplex Ha be due to A-like conformation of the former?It has been calculated that the accessibility of the phosphateoxygen atoms to the solvent (and to the condensing agent) in theA form is almost twice lower than in the B form [18]. This mustsubstantially affect the constant k| (in our approximation ki istaken as constant for all contact combinations studied). Thesecondary structure of the helix must also tell on the ratio ofconstants k3/(k2 + k3). In duplexes Ic and Ilg containing rUinstead of dT at the phosphate acceptor end the coupling yield

Deoxyribose Ribose

declines appreciably as compared to the duplexes la and Ilacomposed entirely of deoxyribonucleotide units (Table 3). Onecan suggest that this is due to a drastically altered conformationof the 3'- terminal unit. Typical for B-DNA is the C2'- endoconformation of the furanose rings with pseudo-axial orientationof the 3'-hydroxyl group (Scheme 3). At the same time, in ahybrid duplex comprising covalently bound RNA and DNAblocks the ribonucleotide units, because of their structuralconservatism, retain the C3'-endo conformation typical of theA form [19], with the 3'-hydroxyl group occupying the pseudo-equatorial orientation which sterically hinders its interaction withthe phosphate group of the adjacent molecule. If this suggestionis true then chemical ligation in A-form duplexes would be muchless efficient than in B-like helices. Indeed, low yields of ligationproducts were observed in our experiments with synthetic RNAfragments (next publication).

The structural unequality of duplexes I and II is reflected inthe fact that the relative decline in coupling efficiency (ascompared with nonmodified counterparts) is different for duplexesIc and Ilg (Table 3). Also different are the percentages of their3'-5'- and 2'-5'-linked isomers: 87% of the natural bond in theligated site of duplex Ilg and 30—40% in duplex Ic. Obviouslyin these duplexes the activated phosphate is in dissimilar positionwith respect to the 3'- and 2'- hydroxyls.

Now we shall consider duplexes Ib, Id-f , I Ic-f containingnucleotide residues with reversed configuration at the furanoseC2' and/or C3' atoms. As can be seen in Table 3, the presenceof an anomalous nucleoside residue (dxT, aU, 1U or xU)markedly attenuates the ligation efficiency. The most interestingdata were obtained with arabinosyl-containing systems. Thek3/(k2+k3) value for duplex He with an aU unit in the acceptorend is one order of magnitude lower than that for the nonmodifiedduplex Ila (Table 3). At the same time, when the arabinosederivative is the phosphate donor (duplex Hf containing a -paUpfragment), the k3/(k2+k3) value is comparable to that of thenonmodified duplex lib (Table 3). Apparently the conformationof an anomalous unit adjacent to the nick differs depending onwhether it contains a 3'-phosphate or not. It is known that, unlikenatural monomers, the arabinonucleotides show a clear preferencefor one or the other conformation of the furanose ring dependingon the position of the phosphate group: (C3'-endo) N-type forpaA, and (C2'-endo) S-type for aAp [20]. This dependence isalso maintained at the oligomeric level. Thus a dinucleosidephosphate aApaA is a mixed S-N conformer (the strong biastoward S-type conformation of the aAp-residue and preferencefor N-type in -paA residue), a study of a trimer aApaApaAallowed it to be described as an S-S-N conformer [20]. Basedon the results of chemical ligation and on the above data onecan postulate with good probability that the -paU residue induplexes Id and He has N-type conformation with pseudo-equatorial position of the 3'-hydroxyl group (Scheme 4). Whenthe arabinose derivative contain the phosphate group at 0 3 '(duplex Hf) the conformation of the anomalous furanose changesfrom N- to S-type, ensuring the pseudo-axial position of thereacting group which is optimal for interaction.



Scheme 4N S

-o-p=o

Arabinose °"

From these suggestions it follows that in a DNA duplex thearabinonucleoside residues retain their structural autonomy, i.e.persist in the conformation typical of them at the monomeric oroligomeric levels. Since our findings and the NMR spectroscopicdata on nicked and non-nicked DNA duplexes [21] argue infavour of at least partially fixed conformation of the nucleotideresidues at the nick, one can suppose that in the covalentlycontinuous DNA the arabinonucleotide residues would also bein S-type conformation. This is in line with the data [22] on EcoRIcleavage of DNA duplexes containing aA in the cleavage site.

Different kinetic patterns are observed for dxT-containingduplexes IIc,d. The coupling yield is not appreciably affectedby the position of the phosphate group in the ligation site, andin both cases it is significantly lower than with the nonmodifiedduplex la (Table 3). This may point either to retention of thesame conformation of the 3'-terminal anomalous unit that isunfavorable for ligation, or to steric hindrances due to inversionof the 3'-hydroxyl group. Since there are no data on theconformation of dxT-derivatives, to check our suggestions weundertook an analysis of the molecular models of thesecompounds. This analysis demonstrates that for dxT and its3'-phosphate the S-type conformation is the spatially unobstructedone (Scheme 5), whereas in the N-type the substituent at C3'situated above the sugar ring is in sterically unfavorable proximityto the hydroxyl at C5' and to the heterocyclic base. For -dxTand -dxTp residues in the nick an S-type conformation determinesthe pseudo-equatorial position of 03' atom, which, as alreadymentioned, tells adversely on the efficiency of oligonucleotideligation.

The chemical ligation technique used in the present work toelucidate the local structure of DNA is in a sense similar to theenzymic one [22], but unlike restriction endonucleases whichcleave DNA in a strictly definite place, a condensing agent joinsany nucleotide units at the single-strand break. Variousconformational alterations at the nick, while affecting to a %'aryingextent the reaction efficiency, nevertheless do not block itcompletely. This approach, supplementing the conventionalphysicochemical methods and theoretical conformational analysis,allows a more reliable assessment of the conformation ofnucleotide residues adjacent to the nick and even 'within' an intactdouble helix.

CONCLUSION

Studying the relationship between the kinetic parameters ofchemical ligation and the local structure of nucleotide duplexesis of key importance for further development of the method bothin theoretical and applied aspects. First, this expands ourknowledge of the peculiarities of reactions in spatially organizedcomplexes stabilized by weak multipoint interactions, of the extentof restraint on the reacting groups therein, and helps to improvethe kinetic description of such enzyme-like reactions etc. Second,such an approach allows evaluation of the prospects of introducing

Scheme 5S . S

~ S'"O-P-

OM

Deoxy - threo - pentof uranose

a certain modification in the DNA sugar-phosphate backbone bychemical ligation, provides an assessment of the selectivity ofreactions involving polyfunctional units (e.g., ribose derivatives),and opens the way to chemical probing of the fine structure ofnucleotide duplexes.

REFERENCES

1. Shabarova, Z.A. (1988) Biochimie 70, 1323-1334.2. Dolinnaya, N.G., Sokolova, N.I., Grayznova, O.I. and Shabarova, Z.A.

(1988) Nucleic Acids Res. 16, 3721-3738.3. Gertzuk, M.N., Sergeev, V.N., Tsytovich, A.V., Dolinnaya, N.G. and

Kuchar, V.P. (1990) Bioorgan Khim. (Russ.) 16, 1268-1276.4. Abbreviations: MES, 2-morpholinoethane sulfonate; dxT, l-(|3-D-2-deoxy-

r/ireo-pentofuranosyl)thymine; aU, l-(/3-D-arabinofuranosyl)uracil; 1U, 1-(J3-D-xylofuranosyl)uracil; 1U, l-(/3-D-lixofuranosyl)uracil; *p, 32P-labelledphosphate group; °U, 2O,2'-anhydrouridine.

5. Tsytovich, A.V., Oretskaya, T.S., Dolinnaya, N.G., Sokolova, N.I. andShabarova, Z.A. (1987) Khimija Prirodn. Soedineny (Russ.) 3, 421-425.

6. Krynetskaya, N.F., Zayakina, G.V., Oretskaya, T.S., Volkov, E.M. andShabarova, Z.A. (1986) Nucleosides and Nucleotides 5, 33-43.

7. Jacobson,B.S. and Fairman, K.R. (1980) Anal. Biochem. 106, 114-117.8. Tsytovich, A.V., Dolinnaya, N.G. and Shabarova, Z.A. (1988) Molecul.

Biologija 22, 690-699.9. Melamed, N.V., Popov, S.G. and Shamovsky, G.G. (1974) Izvestija SO

AN SSSR 12, 90-98.10. Ohtsuka, E., Moon, M.W. and Khorana, H.G. (1965) J.Am.Chem.Soc. 87,

2956-2970.11. Tenforde, T., Fawwaz, R.A. and Freeman, N.K. (1972) J. Org. Chem. 37,

3372-3374.12. Werntges, H., Steger, G., Riesner, D. and Fritz, H.-G. (1986) Nucleic Acids

Res. 14, 3773-3790.13. Roongta, V.A., Jones, C.R. and Gorenstein, D.G. (1990) Biochemistry 29,

5245-5258.14. Roy, S., Sklenar, B., Appella, E. and Cohen, J.S. (1987) Biopolymers 26,

2041-2052.15. Kalnik,M.W., Norman, D.G., Li, B.F., Swann, P.F. and Patel, D.J. (1990)

J. Biol.Chem. 265, 636-646.16. Shabarova, Z.A., Veiko, V.P., Dolinnaya, N.G., Drutsa, V.L., Metelev,

V.G., Oretskaya, T.S. and Purmal, A.A. (1987) Bioorgan. Khim. (Russ.)13, 628-642.

17. Kuznetsova, S.A., Kubareva, E.A., Oretskaya, T.S., Dolinnaya, N.G.,Krynetskaya, N.F., Gromova, E.S., Shabarova, Z.A. and Cech, D. (1987)Biopolymers i Kletki (Russ.) 3, 283-289.

18. Saenger, W. (1984) Principles of Nucleic Acid Structure. Springer-Verlag,New York.

19. Mellema, J.-R., Haasnoot, C.A.G., van der MarelBoeckel, C.A.A., van Boom, J.H. and Altona, C.Res. 11, 5717-5738.

20. Doornbos, J., Barascut, J.-L., Lazrek, H., Imbach, J.-L., van Westrenen,J., Visser, G.M., van Boom, J.H. and Altona, C. (1983) Nucleic AcidsRes. 11, 4583-4600.

21. Pieters, J.M.L., Mans, R.M.W., van den Elst, H., van der Marel, G.A.,van Boom, J.H. and Altona, C. (1989) Nucleic Acids Res. 17, 4551 -4565.

22. Ohtsuka, E., Ishino, Y., Ibaraki, K. and Dcehara, M. (1984) Eur.J.Biochem.139, 447-450.

G.A., Wille, G., van(1983) Nucleic Acids


structural and kinetic aspects of chemical reactions in dna

Documents