Improved synthesis of trinucleotide phosphoramidites and generation of randomized oligonucleotide libraries

Article abstract of DOI:10.1080/15257770701426260

A new method to produce a set of 20 high quality trinucleotide phosphoramidites on a 5-10 g scale each was developed. The procedure starts with condensation reactions of P-components with N-acyl nucleosides, bearing the 3′-hydroxyl function protected with 2-azidomethylbenzoyl, to give fully protected dinucleoside phosphates 13. Upon cleavage of dimethoxytrityl group from 13, dinucleoside phosphates 16 are initially transformed into trinucleoside diphosphates 19 and then the 2-azidomethylbenzoyl is selectively removed under neutral conditions to generate trinucleoside diphosphates 5 in excellent yield. Subsequent 3′-phosphitylation affords target trinucleotide phosphoramidites 7. When mutagenic oligonucleotides are synthesized employing mixtures of building blocks 7 as well as following the new synthetic protocol, representative oligonucleotide libraries are generated in good yields. Copyright Taylor & Francis Group, LLC.

Full text of DOI:10.1080/15257770701426260

This article was downloaded by: [BYU Brigham Young University]
On: 15 January 2015, At: 00:53
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Nucleosides, Nucleotides and Nucleic
Acids
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/lncn20
Improved Synthesis of Trinucleotide
Phosphoramidites and Generation of
Randomized Oligonucleotide Libraries
Andrey Yagodkin a , Alex Azhayev a , Jarkko Roivainen b , Maxim
Antopolsky b , Alexei Kayushin c , Maria Korosteleva c , Anatoly
Miroshnikov c , John Randolph d & Hugh Mackie d
a Department of Pharmaceutical Chemistry , University of Kuopio,
Kuopio, Finland and Metkinen Chemistry , Kuopio, Finland
b Department of Pharmaceutical Chemistry , University of Kuopio ,
Kuopio, Finland
c Shemiakin and OvchinnikovInstitute of Bioorganic Chemistry ,
Russian Academy of Sciences , Moscow , Russia
d Glen Research , Sterling, Virginia, USA
Published online: 19 Jun 2007.
To cite this article: Andrey Yagodkin , Alex Azhayev , Jarkko Roivainen , Maxim Antopolsky , Alexei
Kayushin , Maria Korosteleva , Anatoly Miroshnikov , John Randolph & Hugh Mackie (2007) Improved
Synthesis of Trinucleotide Phosphoramidites and Generation of Randomized Oligonucleotide Libraries,
Nucleosides, Nucleotides and Nucleic Acids, 26:5, 473-497, DOI: 10.1080/15257770701426260
To link to this article: http://dx.doi.org/10.1080/15257770701426260
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions

Nucleosides, Nucleotides, and Nucleic Acids, 26:473–497, 2007
Copyright ꢀC Taylor & Francis Group, LLC
ISSN: 1525-7770 print / 1532-2335 online
DOI: 10.1080/15257770701426260
IMPROVED SYNTHESIS OF TRINUCLEOTIDE PHOSPHORAMIDITES
AND GENERATION OF RANDOMIZED OLIGONUCLEOTIDE
LIBRARIES
Andrey Yagodkin and Alex Azhayev
Department of Pharmaceutical Chemistry,
2
University of Kuopio, Kuopio, Finland and Metkinen Chemistry, Kuopio, Finland
Jarkko Roivainen and Maxim Antopolsky
Department of Pharmaceutical
2
Chemistry, University of Kuopio, Kuopio, Finland
Alexei Kayushin, Maria Korosteleva, and Anatoly Miroshnikov
Shemiakin
2
and OvchinnikovInstitute of Bioorganic Chemistry, Russian Academy of Sciences,
Moscow, Russia
John Randolph and Hugh Mackie
Glen Research, Sterling, Virginia, USA
2
A new method to produce a set of 20 high quality trinucleotide phosphoramidites on a 5–10
2
g scale each was developed. The procedure starts with condensation reactions of P-components
with N-acyl nucleosides, bearing the 3 ꢁ-hydroxyl function protected with 2-azidomethylbenzoyl, to
give fully protected dinucleoside phosphates 13. Upon cleavage of dimethoxytrityl group from 13,
dinucleoside phosphates 16 are initially transformed into trinucleoside diphosphates 19 and then
the 2-azidomethylbenzoyl is selectively removed under neutral conditions to generate trinucleoside
diphosphates 5 in excellent yield. Subsequent 3 ꢁ-phosphitylation affords target trinucleotide phos-
phoramidites 7. When mutagenic oligonucleotides are synthesized employing mixtures of building
blocks 7 as well as following the new synthetic protocol, representative oligonucleotide libraries are
generated in good yields.
Keywords Trinucleotide phosphoramidites; 2-azidomethylbenzoyl protection group;
randomized oligonucleotide libraries
INTRODUCTION
Protein mutagenesis can be used to fine tune a variety of properties,
such as improved stability to high temperatures, denaturants, or nonaque-
ous solvents; higher affinity binding to a target molecule; increased rates
Received 5 October 2006; accepted 30 March 2007.
Financial support by the Technology Development Centre of Finland (TEKES) is gratefully acknowledged.
Address correspondence to Alex Azhayev, Department of Pharmaceutical Chemistry, University of
Kuopio, FIN-70211 Kuopio, Finland. E-mail: Alex.Azhayev@uku.fi
473

474
A. Yagodkin et al.
of enzymatic reactions; or changes of specificities. However, generating
and finding these improved proteins can be a difficult task. One of the
most popular methods is to make pools of degenerate oligonucleotides,
which can be incorporated into the genes as cassettes or by PCR by
using the degenerate oligo as a primer.[1] Degenerate oligonucleotides are
synthesized as a mixture of A/C/G/T phosphoramidites (N) at the site of
the codons to be mutated. Problems arise, though, from using an equimolar
solution of each base. First there is a coding bias. Of the 64 possible codon
combinations of A, C, G and T, eighteen code for leucine, arginine, or
serine, but only two for tryptophan or methionine. As a result, only 3%
of the mutagenic oligonucleotides will contain methionine or tryptophan,
and over 28% will contain either leucine, arginine, or serine. In addition,
the three nonsense codons will lead to chain termination in 4.7% of the
sequences. There are ways to improve this situation. For instance, using two
degenerate mixes of bases, N and G/C, on the DNA synthesizer to insert
NNG/C into the sequence will halve the number of the most degenerate
codons, but still code for all 20 amino acids. However, still 59% of the
clones will code for just eight amino acids and 3% will have a stop codon
inserted. The generation of redundant sequences and stop codons makes
searching a clonal library inefficient. However, it is possible to improve
the efficiency of this process by using a mixture of trinucleotide (trimer)
phosphoramidites.[2–5] By synthesizing a set of trimers that cover all 20
amino acids, the mutation of a gene can be carried out at the codon
level rather than at individual bases. Therefore, unlike other methods of
mutagenesis, trimer phosphoramidites lead to no codon bias, no frame-shift
mutations, and no production of stop codons, making them one of the
most efficient tools to explore sequence space in protein regions that are
important for function[6]—even in nonsaturating conditions.[7,8]
RESULTS AND DISCUSSION
Several reports[9–11] describing the synthesis of trimer phospho-
ramidites have been published. To our mind, the basic approach de-
scribed by Kayushin et al.[2–4] still seems straightforward and timely.
The general procedure, reported in detail in the initial publication[4]
utilizes an approach employing condensation of the P-component
(triethylammonium salt of 5ꢁ-O-dimethoxytrityl-N -acyl-2ꢁ-deoxynucleoside-
3ꢁ-O-(2-chlorophenyl)phosphate or 5ꢁ-O-dimethoxytritylthymidine-3ꢁ-O-(2-
chlorophenyl)phosphate) with the N -acyl nucleoside containing an unpro-
tected 3ꢁ-hydroxyl [12,13] in the presence of 1-(2-mesitylenesulfonyl)-3-nitro-1
H-1,2,4-triazole (MSNT) in pyridine in the first step (Scheme 1). The
mildest way for internucleotide link formation was selected in this case[7]
in order to exploit the difference in reactivity of primary and secondary
hydroxyl functions of N -acyl nucleosides.

Improved Synthesis of Trinucleotide Phosphoramidites
475
SCHEME 1 Synthesis of dinucleotide according to[7]
.
Initially our goal was to synthesize all 20 Trimer phosphoramidites
following the above referenced method (structures being based on codon
usage in E. coli). We started with preparation of 12 intermediate dimers
and in the process of this work we have noticed that indeed[4] in all 12
condensation reactions the main product was always the desired dinu-
cleotide 3a. The isomeric dimer 3b (both 3a and 3b are regioisomeric pairs)
along with the trimer 4 were always present in the reaction mixtures as
well (Scheme 1). The amount of unwanted 3b and 4 varied, depending on
the target dimer in question, being up to 7–8% in certain cases (data not
shown). The silica gel chromatographic separation of trimer 4 appeared to
be an easy task. However, the separation of a few percent of unwanted dimer
3b was never a straightforward procedure. In some cases, the complete
separation of 3b (if possible at all) required a long silica gel column and
large volumes of chromatographic solvents. In any case, an analytical confir-
mation (RP HPLC) of the purity of 3a as well as the complete absence of 3b
in the target dimer, was absolutely required before the next condensation
step. In cases when the dimer 3b was not removed completely from 3a
(fast flash chromatography), subsequent condensation generated trimer 5
contaminated with three isomeric trimers, 6a, 6b, and 6c (Scheme 2). It
appeared very difficult, if possible at all to remove these unwanted trimers

476
A. Yagodkin et al.
SCHEME 2 Synthesis of trinucleotides according to[7]
.
from the target product 5. One has to keep in mind that all of these trimers
were in fact mixtures of six diastereomers. Obviously, this fact seriously
hampered the isolation of pure trimers 5. Subsequent conversion of 5
into trimer phosphoramidite 7 was a straightforward and simple procedure
(Scheme 3). However, preparative reverse phase (RP) chromatography does
not allow separation of target trimer phosphoramidite 7 from undesired
3ꢁ-phosphoramidites of 6a, 6b and 6c, in the case initial trimer 5, used
SCHEME 3 Synthesis of trinucleotides 7, employing intermediate protection of 3ꢁ-terminal hydroxy-
groups. Typical fragment ions of 7, labeled as reported earlier,[5] are also shown.

Improved Synthesis of Trinucleotide Phosphoramidites
477
for conversion into trimer phosphoramidite 7, was not pure enough. It is
noteworthy that any trimer 3ꢁ-phosphoramidite comprises a mixture of 8
isomers, irresolvable by preparative low pressure RP chromatography.
When a mutagenic oligonucleotide was synthesized using trimers 7
contaminated with 3ꢁ-phosphoramidites of 6a, 6b, and 6c, and subsequently
employed for construction of libraries by introducing it into a plasmid used
for E. coli transformation, a number of resulting clones often incorporated
DNA sequences with frame shifts and unexpected mutations.[14]
Evidently, the repetitive production of a set of 20 trimer phospho-
ramidites has to rely on a procedure which guarantees the absence of
unwanted dimer 3b and subsequently trimers 6a, 6b, and 6c on the
way to the final target trimer phosphoramidites 7. In addition, the new
approach has to be a straightforward synthesis, employing simple and fast
flash chromatography purification of intermediate compounds. Hence, the
method of production of trimer phosphoramidites has to be considerably
revised and improved to meet these requirements.
SYNTHESIS AND CHARACTERIZATION OF TRIMER
PHOSPHORAMIDITES
To our mind, a new strategy to produce a set of 20 high quality
trimer phosphoramidites has to start from the condensation reactions of
P-components with N -acyl nucleosides with properly protected 3ꢁ-hydroxyl
group. This will ensure the absence of unwanted isomeric dimers and
trimers present in the intermediate and final products and, more impor-
tantly, will considerably simplify the necessary purification steps. Due to
the nature of blocking functions in P-components (DMTr, 2-chlorophenyl
group on phosphorous and N -acyl group), this protection group must be
neither acid-labile nor strongly base-labile. In another words, the protection
of 3ꢁ-terminal nucleoside 3ꢁ-hydroxy group has to be easily removable either
under very weak basic conditions or under neutral specific conditions.
In the present investigation, we tested the suitability of three
such candidates—2,4-dichlorophenoxyacetyl-, 4-azidobutanoyl- and 2-
azidomethylbenzoyl protection groups (Scheme 4). To our knowledge,
the weakly base-labile 2,4-dichlorophenoxyacetyl group has not been
used in nucleoside chemistry earlier. This function may be easily in-
troduced into 5ꢁ-O-dimethoxytrityl-N -acyl-nucleoside by reaction with 2,4-
dichlorophenoxyacetylimidazole prepared in situ to give the corresponding
5ꢁ-O-dimethoxytrityl-3ꢁ-O-(2,4-dichlorophenoxyacetyl)-N -acyl-nucleosides in
a good yield. Subsequent removal of 5ꢁ-O-dimethoxytrityl functions under
conditions specific for each N -acylnucleoside or thymidine (see Exper-
imental section) gave 3ꢁ-O-(2,4-dichlorophenoxyacetyl)-N -acyl-nucleosides
8b–c and 3ꢁ-O-(2,4-dichlorophenoxyacetyl)-thymidine 8a in 80–90% overall

478
A. Yagodkin et al.
SCHEME 4 Synthesis of Dinucleotides 14–16, bearing protection group at 3ꢁ- terminal hydroxyl.
yield. Subsequent synthesis of completely protected trinucleotides was based
on classical phosphotriester method in solution (Scheme 3) with minor
technical modifications (see Experimental section). The overall yields of
3ꢁ-O-dichlorophenoxyacetyl-trinucleotides 17, when calculated based on
initial 3ꢁ-O-(2,4-dichlorophenoxyacetyl)-N -acylnucleosides 8a–c were rea-
sonably good. No unwanted isomeric trimers were found in reaction
mixtures and therefore all chromatographic separations of intermediate
3ꢁ-O-dichlorophenoxyacetyl-dinucleotides and 3ꢁ-O-dichlorophenoxyacetyl-
trinucleotides appeared to be straightforward.
As we have found, the 2,4-dichlorophenoxyacetyl- protecting group may
be removed under very mind conditions (10% triethylamine in pyridine-
methanol, 2:1 for 3 hours at room temperature (rt)) from the 3ꢁ-hydroxyl
of trinucleotides 17–19 (Scheme 3) to give trimers 5 in a reasonable yield of
65–85%. Under these conditions, most of the 2-clorophenyl groups as well
as the N -acyl groups (Scheme 3) remained intact. However, minor amounts
of initial 2,4-dichlorophenoxyacetylated products 17–19 (10–20%) were still
present in reaction mixtures. It is noteworthy that even such a mild basic
treatment of trinucleotides (especially incorporating 2ꢁ-deoxyguanosine
nucleoside units) was always accompanied by considerable darkening of
the reaction mixture. Although the longer (5 hours) treatment with

Improved Synthesis of Trinucleotide Phosphoramidites
479
triethylamine-pyridine-methanol mixture reduced the amount of initial
2,4-dichlorophenoxyacetylated trimers 17–19 to about 2–3%, the loss of
2-clorophenyl groups was becoming already noticeable and, in general,
reaction mixtures were turning very dark. The additional drawback of using
the strategy described above, is the formation of stable emulsions upon
cleavage of 3ꢁ-terminal 2,4-dichlorophenoxyacetyl protection group and
subsequent workup (see Experimental section). This workup—successive
washings of dichloromethane or chloroform solution of trinucleotide 5
containing a free 3ꢁ-terminal hydroxyl (Scheme 3) with sodium hydrogen
carbonate and then triethylammonium hydrogen carbonate normally took
several hours to accomplish due to very slow separation of the organic and
aqueous layers.
Thus, although the use of 2,4-dichlorophenoxyacetyl protecting group
basically helps to improve the quality of final trinucleotides by eliminating
the problem of unwanted isomeric trimers in the target trimer phospho-
ramidites, it brings forward certain disadvantages as well. Darkening of
reaction mixtures and hence colored target trimer phosphoramidites, the
need to waste time on slow separation of layers during the workup and,
more importantly, somewhat reduced yields of target compounds impelled
us to probe other protecting groups for the 3ꢁ-terminal hydroxyl function.
In order to avoid the drawbacks of the method described above,
we investigated the possibility to use 4-azidobutyryl group[15] for the
protection of the 3ꢁ-terminal hydroxyls of trinucleotides (Scheme 4). 4-
Azidobutyryl esters may be cleaved under neutral conditions via reduc-
tion (e.g., with triphenylphosphine in the presence of water) of the
azido function to give esters of 4-aminobutyric acid. The formation of
aminoacyl ester is then spontaneously followed by intramolecular cy-
clization, to generate the free hydroxyls (Scheme 5). The 4-azidobutyryl
protection group may be easily introduced into 5ꢁ-O-dimethoxytrityl-N -
acyl-nucleosides by reaction with 4-chlorobutyryl chloride to give corre-
sponding 5ꢁ-O-dimethoxytrityl-3ꢁ-O-(4-chlorobutyryl)-N -acyl-nucleosides in
good yield. Subsequent azidation reaction with sodium azide in DMF
followed by removal of 5ꢁ-O-dimethoxytrityl functions (see Experimen-
tal section) gave 3ꢁ-O-(4-azidobutyryl)-N -acyl-nucleosides 9b–c and 3ꢁ-O-
(4-azidobutyryl)-thymidine 9a in 70–80% overall yield. The subsequent
chemistry leading to 3ꢁ-O-(4-azidobutyryl)-trinucleotides was basically the
same as described above (see Experimental section). The overall yields of
3ꢁ-O-(4-azidobutyryl) trinucleotides 18 when calculated on initial 3ꢁ-O-(4-
azidobutyryl)-N -acylnucleosides 9a–c were in a range of 80–90%. As in the
case with 2,4-dichlorophenoxyacetyl temporary protection, no unwanted
isomeric trimers were found in these reaction mixtures. All chromato-
graphic separations of intermediate 3ꢁ-O-(4-azidobutyryl)-dinucleotides 15
and 3ꢁ-O-(4-azidobutyryl)-trinucleotides 18 proved to be fast and simple.
The 4-azidobutyryl protection group was removed from the 3ꢁ-hydroxyl

480
A. Yagodkin et al.
SCHEME 5 Cleavage of 3ꢁ-terminal protection groups, reported herein.
of trinucleotides (Scheme 3) upon reduction, followed by O → N acyl
migration (Scheme 5 and Experimental section) in good yield. Under
these conditions, no loss of 2-clorophenyl groups as well as N-acyl groups
(Scheme 3) was detected and no initial 4-azidobutyrylated trimers were
found in reaction mixtures. It is noteworthy that no darkening of reaction
mixtures was observed. However, a small amount (5–10%) of a very polar
DMT-containing material was present in all cases. The yields of trimers 5
when synthesized using 4-azidobutyryl temporary protection were 70–80%.
Most of the resulting trimer phosphoramidites 7 generated by this method
proved to be of high purity and were white to off-white in color.
Thus, the use of the 4-azidobutyryl protection group in the synthesis
of trimer phosphoramidites appears to be a considerable improvement,
when compared to the original procedure.[4] The potential to remove the
blocking group from the 3ꢁ-hydroxyl function of trimers 5 under neutral
conditions also has certain minor advantages, when compared to the
method utilizing 2,4-dichlorophenoxyacetyl- protection (no coloration of
reaction mixtures; no colored target trimer phosphoramidites). Neverthe-
less, the formation of 5–10% of side products in the process of 4-azidobutyryl
cleavage makes these advantages, although significant, not absolutely sat-
isfactory. In order to further improve the method to produce a set of

Improved Synthesis of Trinucleotide Phosphoramidites
481
20 trimer phosphoramidites, we have investigated the possibility of using
one more 3ꢁ-terminal protection group—2-azidomethylbenzoyl.[16] The 2-
azidomethylbenzoylation of the 3ꢁ-hydroxy functions of 5ꢁ-O-dimethoxytrityl-
N -acylnucleosides and 5ꢁ-O-dimethoxytritylthymidine was a simple proce-
dure. The removal of 5ꢁ-O-dimethoxytrityl functions (see Experimental sec-
tion) gave 3ꢁ-O-(2-azidomethylbenzoyl)-N -acyl nucleosides 10b–d and 3ꢁ-O-
(2-azidomethylbenzoyl)-thymidine 10a in good yields (65–77%). It is note-
worthy that adenine, cytidine, and thymidine 3ꢁ-O-(2-azidomethylbenzoyl)-
nucleosides were purified by simple crystallization. Preparation of dinu-
cleotides 16 and trinucleotides 19 was very similar to that of 15 and 18 and
brings about high yields of the desired products. The 2-azidomethylbenzoyl
group is basically similar to 4-azidobutyryl protection. It may be also
removed from 3ꢁ-terminal hydroxyls of trinucleotides 19 (Scheme 3) under
neutral conditions via reduction of the azido function (triphenylphos-
phine in the presence of water) to give esters of 2-aminomethylbenzoic
acid, followed by spontaneous intermolecular O → N acyl migration to
generate the free hydroxyls (Scheme 5). However, in this case, trans-
formation of trinucleotides 19 into trimers 5 (Scheme 3) proceeded in
nearly quantitative yield. The yields of trimers 5 when synthesized using 2-
azidomethylbenzoyl temporary protection were 85–95%. Most of the target
trimer phosphoramidites 7, generated by this method in excellent yield,
proved to be white in color and of high purity as shown below in Figures 1A
and 1B.
Noteworthy that in all cases when reactions were performed with 2ꢁ-
deoxyguanosine derivatives and MSNT, a somewhat higher content of side-
products in reaction mixtures was observed. However, when the reaction
conditions were strictly followed as described in the Experimental section,
these unwanted modifications were minor and they did not hamper the
isolation of desired products.
Quality control of trimer phosphoramidites appears to be very im-
portant and challenging. Trimer phosphoramidites have chiral centers at
all three phosphorus positions. There are 23 = 8 diastereomers in each
phosphoramidite, which are at least partially separated on RP HPLC. The
best separation of diastereomers is nearly achieved when the RP HPLC is
performed under basic conditions (pH 8.4 or higher). Figure 1 A shows the
RP HPLC traces of diastereomers of trimer phosphoramidite 7 (TAC), ob-
tained at pH 8.7. When the trimer phosphoramidite 7 is converted into the
corresponding trimer H-phosphonate by activation with tetrazole, followed
by subsequent hydrolysis, all of the diastereomeric trimer phosphoramidites
are converted into a faster eluting diastereomeric H-phosphonates (Figure
1B). This simple experiment normally proves the absence of contaminants
not incorporating a phosphoramidite function. In addition to the HPLC
data, the 31P-NMR spectrum of trimer 7 always has to show the presence of

482
A. Yagodkin et al.
FIGURE 1 A) RP HPLC trace of trimer phosphoramidite 7 (TAC); B) RP HPLC trace of trimer
phosphoramidite 7 (TAC) after treatment with ethylthiotetrazole and water (trimer H-phosphonate
diester TAC).
phosphotriester and phosphoramidite functions in a ratio of 2:1 (Figure 2
is given as an illustrative example).
There is also the concern that the sequence of the trimers has to be
verified (e. g., CTG, has to be differentiated from GTC). Electrospray mass
spectrometric analysis[5] (ESI MS) of the trimers provides data confirming
molecular mass (Figure 3A) and sequence (Figure 3B). In the latter

Improved Synthesis of Trinucleotide Phosphoramidites
483
FIGURE 2 31P-NMR spectrum of trimer phosphoramidite 7 (TAC).
case, collision induced negative ion MS/MS analysis of the deprotonated
molecule gives two types of complementary fragment ions, which clearly
indicate the base sequence[5] (Figure 3b and Scheme 3).
In our opinion, all three methods (RP HPLC, 31P-NMR, and ESI MS)
have to be routinely employed for characterization of all 20 trimer phos-
phoramidites 7, employed for generation of randomized oligonucleotides.
Thus, we have worked out an appropriate procedure to synthesize a
set of 20 trimer phosphoramidites 7 on a production scale. The com-
pounds generated following our method, utilizing protection of 3ꢁ-terminal
hydroxyl function with 2-azidomethylbenzoyl group, are normally of a
very high purity and are not contaminated with undesired isomeric 3ꢁ-
phosphoramidites of 6a, 6b, and 6c.
SYNTHESIS OF RANDOMIZED OLIGONUCLEOTIDE LIBRARIES
The set of 20 trimer phosphoramidites 7 was used in randomized
oligonucleotide synthesis. As pointed out in the original communication on
trimer preparation and oligosynthesis[4] reaction factor (RF) values have
to be assigned to each trimer 7. The reaction factor is critical since the
trimers likely will be mixed and they have differing reactivity in the coupling
reaction. For example, the RF for AAC is 1.0 and for TAC is 1.6. Therefore,
1.6 equivalents of TAC are needed for every 1.0 equivalent of AAC for
equal coupling. Since the publication by Kayushin et al.[4] the values of RFs

484
A. Yagodkin et al.
FIGURE 3 A) Negative ion zoom scan ESI spectrum of trimer phosphoramidite 7 (CTG) showing
the isotopic pattern of the molecule; B) collision induced negative ion MS/MS spectrum of trimer
phosphoramidite 7 (CTG), showing typical fragment ions.[15]
were re-evaluated and determined more precisely.[14] Several changes in RF
values were introduced most probably because of much higher purity of
trimers 7 reported in this study in comparison to trimer phosphoramidites
tested in original communication.[4] The trimers, their coding amino acids
and their revised reaction factors (RF) are listed in Table 1.
In the original communication[4] a “slightly modified standard cycles
and procedures from the synthesizer producer” were used. These modifica-
tions were—double or triple couplings and 2-fold extended reaction time.
In addition to these modifications, the trimers were dissolved in a mixture
of acetonitrile—dichloromethane (1:3, v/v) to give 0.15 N concentration. It
is noteworthy that since most of trimers 7 display a poor solubility in pure
acetonitrile, the use of acetonitrile—dichloromethane mixture to dissolve
trimers or trimer mixtures appears to be an obligatory and unavoidable
measure. When running the oligosynthesis with the 19 trimer mixture it was
observed that coupling yields ranged from 71% (the first trimer coupling)
to almost 98%.[4]
In the present work we attempted to repeat the synthesis of random-
ized oligonucleotides, following the method described above with a one

Improved Synthesis of Trinucleotide Phosphoramidites
485
TABLE 1 Trimer phosphoreamidites 7, their coding amino acids, and their revised
reaction factors (RF)
Trimer∗
MW
RF
Trimer∗
MW
RF
AAA (Lys)
AAC (Asn)
ACT (Thr)
ATC (Ile)
ATG (Met)
CAG (Gln)
CAT (His)
CCG (Pro)
CGT (Arg)
CTG (Leu)
1911.5
1887.5
1774.5
1774.5
1780.5
1869.5
1774.5
1845.5
1756.5
1756.5
1.1
1.0
1.3
1.2
1.3
2.0
1.9
1.8
1.1
1.2
GAA (Glu)
GAC (Asp)
GCT (Ala)
GGT (Gly)
GTT (Val)
TAC (Tyr)
TCT (Ser)
TGC (Cys)
TGG (Thr)
TTC (Phe)
1893.5
1869.5
1756.5
1762.5
1667.5
1774.5
1661.4
1756.5
1762.5
1661.4
1.9
1.3
1.5
1.1
1.9
1.6
1.3
1.5
2.4
1.3
∗Trimer phosphoramidites 7 are abbreviated with three-letter code for simplicity.
FIGURE 4 RP HPLC traces of randomized oligonucleotide RO-3. Target RO-3 elutes at about 18
minutes.

486
A. Yagodkin et al.
exception—we employed 0.1 N concentration of the 20 trimer mixture
rather than using 0.15 N concentration. Obviously our attempt to reduce
trimer consumption and hence the cost of randomized oligonucleotides
looked reasonable. However, the concentration reduction resulted in yields
of target randomized oligonucleotides ranging in total from 2% to 5%.
Under these circumstances, we decided to adjust the oligosynthesis proce-
dure to achieve better overall yields but still employing 0.1 N concentration
of trimers. When reviewing the synthetic procedure published earlier[4]
it seems obvious that, whereas trimers were initially dissolved in a mix-
ture of acetonitrile—dichloromethane, their solutions become diluted with
acetonitrile when passing through the synthesizer tubings, mixing with
activator and finally on the synthesis column, pre-washed with acetonitrile.
Obviously, dilution with acetonitrile seemed to be undesirable since we
observed the formation of emulsions with some trimers upon dilution of
their acetonitrile—dichloromethane (1:3, v/v) solution with acetonitrile. To
avoid this we decided to dissolve activator (5-ethylthio-1H-ethylthiotetrazole,
1 g) in acetonitrile-dichloromethane mixture (1:3 v/v, 31 mL), introduce
a washing step with acetonitrile-dichloromethane mixture (1:3 v/v) before
the trimer mixture addition step and employ a 10 min trimer mixture
addition step (in three increments) into the randomized oligonucleotide
synthesis protocol.
Hereafter, we report the comparison of results of synthesis of three
randomized oligonucleotides, obtained by the original method[4] and the
procedure described above. These randomized oligonucleotides RO-1 and
RO-2 were libraries of 204 48mers and RO-3 was a library of 2010 60mers:
5ꢁ-TTGTGTCACGACGTTGTGTCACGA(X)4AGCTCAGATCAC-3ꢁ RO-
1,
5ꢁ-TTGTGTCACGAC(X)2GTTGTGTCACGA(X)2AGCTCAGATCAC-3ꢁ
RO-2,
5ꢁ-TTGTGTCACGAC(X)5GTGTCA(X)5AGCTCAGATCAC-3ꢁ RO-3,
where X = 20 trimer phosphoramidite mixture.
In all of our test syntheses (0.1 µmolar scale), we programmed the
synthesizer to perform “trityl-on” for more convenient randomized oligonu-
cleotide isolation from the reaction mixture. Since we performed all our
syntheses on the Universal solid support,[17] our cleavage /deprotection
procedure differed slightly from the original one.[4] We employed a 45
minutes treatment of oligomer-bound solid support with 3.5 N ammonia in
methanol before 3 times diluting the methanolic solution with 32% aqueous
ammonia and leaving the resulting mixture for 36 hours at rt.
In the case of the original procedure, yields of target RO-1—RO-3
upon deprotection were as low as 3–4% for RO-1 and RO-2 and less than
1.5% for RO-3. However, our new modified method allowed preparation
of the same randomized oligomers in a yield of about 30%. In both

Improved Synthesis of Trinucleotide Phosphoramidites
487
FIGURE 5 Photograph of 10% PAGE. Lane 1: standard 40 mer; Lanes 2, 3, 4: randomized oligonu-
cleotides RO-1, RO-2, and RO-3 correspondingly, synthesized according to original protocol[7] and
purified as in Figure 6; Lanes 5, 6, 7: randomized oligonucleotides RO-1, RO-2, and RO-3 correspond-
ingly, synthesized according to modified protocol, reported herein and purified as in Figure 6; Lane 8:
standard 57 mer.
cases, yields of compounds were assessed using RP HPLC by integrating
total “trityl-on” peaks at 260 nm. Figure 4 shows the analytical RP HPLC
trace of crude RO-3 as an illustrative example. The chromatographic run
was performed by using a linear gradient of acetonitrile concentration
(0–20%) in 0.1 M triethylammonium acetate (pH 7.5) over 12 minutes,
then sharply raising the acetonitrile concentration from 20 to 64% over 1
minutes, and finally performing an isocratic elution with 64% acetonitrile
in 0.1 M triethylammonium acetate (pH 7.5) over additional 10 minutes.
According to integration of peaks from this trace, the yield of randomized
RO-3 (the last peak, eluting at 18 minutes) was 37%. All three oligomers,
synthesized employing original[4] and new procedures were purified on
a preparative scale using conditions, described above, de-tritylated and
finally analyzed on the 10% PAGE (Figure 5). One can clearly see that,
whereas all three randomized oligonucleotides were synthesized very well
employing our new synthetic procedure, only reasonably pure RO-1 and
RO-2 were obtained following original protocol.[4] The PAGE also has
supported the length of target oligomers and the absence of shorter failure

488
A. Yagodkin et al.
sequences, separated by RP HPLC. This evidence supports the claim that
the new synthetic protocol is superior to original one.[4] The new procedure
generates reasonable amounts of randomized oligonucleotides, more than
sufficient for subsequent biological experiments.
CONCLUSION
We have developed
a
new procedure to produce valuable
reagents—high quality trinucleotide phosphoramidites, on a 5–10 g scale
each. The trimers are now commercially available (Metkinen Chemistry,
Finland Kuopio) individually, as a mixture of all 20 trimers or any other
specific mixture.
We conclude that when mutagenic oligonucleotides are synthesized
using these trimers 7 following the modified synthetic protocol,
oligonucleotide libraries resulting from cloning will be most likely
free from oligomers containing DNA sequences with frame shifts and
unexpected mutations.
EXPERIMENTAL
General
5ꢁ-O-Dimethoxytrityl-N6-benzoyl-2ꢁ-deoxyadenosine, 5ꢁ-O-dimethoxytri-
tylN4-benzoyl-2ꢁ-deoxycytidine, 5ꢁ-O-dimethoxytrityl-N2-isobutyryl-2ꢁ-deoxy-
guanosine, and 5ꢁ-O-dimethoxytrityl-thymidine were purchased from
Rasayan Inc. (Encinitas, CA, USA). N,Nꢁ-Carbonyldiimidazole was
obtained from Aldrich (St. Louis, MO, USA) and recrystallized
from dry toluene. Dichlorophenoxyacetic acid was obtained from
Lancaster (Eastgate, Lancashire, UK), 4-chlorobutyryl chloride,
1-(2-mesitylenesulfonyl)-3-nitro-1 H-1,2,4-triazole (MSNT) and
7
N
ammonia in methanol were obtained from Aldrich and used without
further purification. 2-Azidomethylbenzoyl chloride was synthesized
as described earlier.[16] Benzenesulfonic acid and 4-toluenesulfonic
acid monohydrate were purchased from Fluka. Acetonitrile, pyridine,
and 1-methylimidazole (Fluka Buchs, Switzerland) were distilled over
calcium hydride. 2-chlorophenyldichlorophosphate was from Sigma
(St. Louis, MO, USA). Triethylammonium salts of 5ꢁ-O-dimethoxytrityl-
N -acyl-2ꢁ-deoxynucleoside-3ꢁ-O-(2-chlorophenyl)phosphates
and
5ꢁ-O-
dimethoxytritylthymidine-3ꢁ-O-(2-chlorophenyl)phosphate (P-components)
were prepared as reported earlier[4] or as described below. column
flash chromatography was run using B-681 and B-688 piston pumps
and borosilicate glass columns (Bu¨chi). Silica gel 60 (40–63 µm) or
Lichroprep RP-18 (40–63 µm), both from Merck (Darmstadt, Germany),
were used for column flash chromatography. A gradient of methanol in
dichloromethane (from 0 to 5–7%, flow rate 50 ml/minute) was used in a

Improved Synthesis of Trinucleotide Phosphoramidites
489
former case and gradient of acetonitrile in water (from 50% to 100%, 25
ml/minute) was used in a latter case for purification of compounds. RP
HPLC of Trimer phosphoramidites 7 and trimer phosphoramidites 7 after
treatment with ethylthiotetrazole and water was performed employing a
Gilson gradient chromatograph on a xBridge C18 column (4.6 × 150 mm,
5 µm, Waters (Milford, MA, USA)), using a gradient elution from 60 to
80% acetonitrile in 0.1 M triethylammonium acetate buffer, pH 8.7, over
20 minutes, followed by isocratic elution with 80% acetonitrile in 0.1 M
triethylammonium acetate buffer, pH 8.7 over 10 minutes at a flow rate
1 ml/minute. RP HPLC of randomized oligonucleotides was performed
employing a Merck-Hitachi (Darmstadt, Germany) gradient chromatograph
on a xTerra C18 column (4.6 × 150 mm, 5 µm, Waters), using a gradient
elution from 0 to 20% acetonitrile in 0.1 M triethylammonium acetate
buffer, pH 7.5 over 12 minutes, followed by gradient elution from 20 to 80%
acetonitrile in the same buffer over 1 min., and finally isocratic elution with
80% acetonitrile in the same buffer over 10 minutes; flow rate 1 ml/minute.
ESI MS data were obtained with Finnigan (San Jose, CA, USA) MAT
LCQ ion trap mass spectrometer, using negative ion mode. The structures
of trimer phosphoramidites were verified as described earlier.[5] NMR
spectra were taken with Avance 500 spectrometer (Bruker, Silberstetten,
Germany) in DMSO-d6 (1H-NMR) or CDCl3 (31P-NMR). Oligonucleotide
syntheses were performed with an 8-column ASM-800 synthesizer (Biosseti
Novosibirsk, Russia) on 0.1 µmolar scale, employing a Universal solid
support USII-AMPS-M-40 (Metkinen Chemistry, Kuopio, Finland). Cleav-
age/deprotection of oligonucleotides was performed as recommended by
the manufacturer.
Synthesis of 3ꢁ-O-2,4-dichlorophenoxyacetyl)-N-acyl-2ꢁ-
deoxynucleosides and 3ꢁ-O-(2,4-dichlorophenoxyacetyl)thymidine
The synthesis of 3ꢁ-O-2,4-dichlorophenoxyacetyl)-N -acyl-2ꢁ-deoxynucleo-
sides and 3ꢁ-O-(2,4-dichlorophenoxyacetyl)thymidine may be divided
in two steps. Step A, the preparation of a 5ꢁ-O-dimethoxytrityl-3ꢁ-O-
dichlorophenoxyacetyl-N -acyl-2ꢁ-deoxynucleosides, is common for all four
2ꢁ-deoxynucleosides. Step B, the removal of 5ꢁ-O- dimethoxytrityl residue
and further workup is base-specific and will be described for each nucleoside
separately hereafter.
3ꢁ-O-(2,4-Dichlorophenoxyacetyl)thymidine (8a)
Step A. The solution of carbonyldiimidazole (8.910 g, 55 mmol) in
100 ml of anhydrous acetonitrile was added to the solution of 2,4-
dichlorophenoxyacetic acid (17.155 g, 55 mmol) in 200 ml of anhydrous
acetonitrile under intensive stirring within 10 minutes. In nearly 5 min-
utes, the precipitate started to form. In 45 minutes the precipitate was

490
A. Yagodkin et al.
filtered off and the resulted solution was added to a suspension of 5ꢁ-
O-dimethoxytritylthymidine (15 g, 27.5 mmol) in 200 ml of anhydrous
acetonitrile. The suspension was concentrated to a gum by evaporation;
during evaporation the precipitate started to dissolve and finally the clear
solution has been obtained. In 2 hours 2 ml of water was added, in 20
minutes the solution was evaporated to dryness, the residue was dissolved in
nearly 200 ml of chloroform and washed with 200 ml of 0.1 M triethylammo-
niumbicarbonate (TEAB). The aqueous layer was extracted with chloroform
(3 × 200 ml), the combined organic layers were washed with 0.1 M TEAB
(3 × 200 ml), dried over sodium sulfate and evaporated to dryness.
Step B. The residue obtained after completing the Step A was dissolved
in 100 ml of 1,2-dichloroethane. The solution of (10.1 ml, 137.5 mmol)
anhydrous trifluoroacetic acid in 100 ml of 1,2-dichloroethane was added
under intensive stirring. In 1 minute 140 ml of saturated aqueous solution
of sodium hydrogen carbonate was added, the aqueous layer was washed
with chloroform (3 × 150 ml), the combined organic layers were washed
with 0.1 M TEAB (3 × 150 ml), dried over sodium sulfate and evaporated
to a gum. The residue was evaporated with chloroform (3 × 100 ml) and
the desired product was isolated by column chromatography on silica gel
(column 4 × 34 cm, linear gradient of methanol in chloroform, 0–6%, 2 l,
flow rate 1 l/hour). Yield 10.89 g (89%); 1 H NMR (DMSO-d6), δTMS, ppm;
J , Hz: 11.33 (s, 1H, NH of Thy), 7.73 (d, 1H, J = 1.1, H-6 of Thy), 7.59
(d, 0.5 H, J = 2.6, H-3 of 2,4-dichlorophenoxyacetyl), 7.34 (dd, 1H, J =
2.6, J = 8.9, H-5 of 2,4-dichlorophenoxyacetyl), 7.18 (d, 1H, J = 9.0, H-6
of 2,4-dichlorophenoxyacetyl), 6.19 (dd, 1H, J = 6.0, J = 8.6, H-1ꢁ), 5.34
(m, 1H, H-3ꢁ), 5.21 (t, 1H, J = 5.2, 5ꢁ-OH), 5.01 (s, 2H, CH2-OAryl), 4.02
(m, 1H, H-4ꢁ), 3.63 (m, 2H, H-5ꢁa,b), 2.29 (m, 2H, H-2ꢁa,b), 1.78 (d, 5-CH3
of Thy); MS (ESI-MS, CH3CN): Calcd. for C18H18Cl2N2O7 [M-H]−: 445.25.
Found: 443.1.
3ꢁ-O-(2,4-Dichlorophenoxyacetyl-N2-isobutyryl-2ꢁ- deoxyguanosine (8b)
Step A was performed with (20 g, 31.25 mmol) of 5ꢁ-O-dimethoxytrityl-
N2-isobutyryl-2ꢁ-deoxyguanosine. Step B was performed as described for
3ꢁ-O-(2,4-Dichlorophenoxyacetyl)thymidine, except for duration of acidic
1
treatment (30 seconds instead of 1 minute). Yield 15.38 g (91%); H NMR
(DMSO-d6), δTMS, ppm; J , Hz: 12.08 (s, 1H, NH of Gua), 11.66 (s, 1H,
2-NH of Gua), 8.27 (s, 1H, H-2 of Gua), 7.61 (d, 0.5 H, J = 2.6, H-3 of
2,4-dichlorophenoxyacetyl), 7.36 (dd, 2H, J = 2.6, J = 8.9, H-5 of Cyt
and H-5 of 2,4-dichlorophenoxyacetyl), 7.18 (d, 1H, J = 9.0, H-6 of 2,4-
dichlorophenoxyacetyl), 6.21 (dd, 1H, J = 5.6, J = 9.0, H-1ꢁ), 5.45 (m, 1H,
H-3ꢁ), 5.13 (t, 1H, J = 5.5, 5ꢁ-OH), 5.02 (d, 2H, J = 3.9, CH2-OAryl), 4.10 (m,
1H, H-4ꢁ), 3.60 (m, 2H, H-5ꢁa,b), 2.88 (m, 1H, H-2ꢁa), 2.77 (m, 1H, J = 6.9,
CH of ibu), 2.54 (m, 1H, H-2ꢁb), 1.13 (d, 6 H, J = 6.8, 2CH3 of ibu); MS (ESI-
MS, CH3CN): Calcd. for C22H23Cl2N5O7 [M-H]−: 538.10. Found: 538.2.

Improved Synthesis of Trinucleotide Phosphoramidites
491
3ꢁ-O-(2,4-Dichlorophenoxyacetyl-N4-benzoyl-2 ꢁ-deoxycytidine (8c)
Step A was performed with (15.85 g, 25 mmol) of 5ꢁ-O-dimethoxytrityl-
N4-benzoyl-2ꢁ-deoxycytidine. Step B was performed as described for 3ꢁ-O-
(2,4-Dichlorophenoxyacetyl)thymidine. After addition of sodium hydrogen
carbonate the precipitate was formed. The precipitate was filtered off,
the aqueous layer of the filtrate was washed with chloroform (5 × 100
ml), combined organic layers were washed with 0.1 M TEAB (3 × 150
ml), dried over sodium sulfate and evaporated to dryness. The residue
was combined with the precipitate obtained above and re-crystallized from
1
isopropanol. Yield 10.55 g (79%); m.p 198–201◦C; H NMR (DMSO-d6),
δTMS, ppm; J , Hz: 11.33 (s, 1H, 4-NH of Cyt), 8.37 (d, 1H, J = 7.6, H-6
of Cyt), 8.01 (m, 2H, Bz), 7.62 (m, 1H, Bz), 7.59 (d, 0.6 H, J = 2.6,
H-3 of 2,4-dichlorophenoxyacetyl), 7.51 (m, 2H, Bz), 7.36 (m, 2H, H-5
of Cyt and H-5 of 2,4-dichlorophenoxyacetyl), 7.19 (d, 1H, J = 9.0, H-6
of 2,4-dichlorophenoxyacetyl), 6.19 (dd, 1H, J = 5.9, J = 8.0, H-1ꢁ), 5.37
(m, 1H, H-3ꢁ), 5.28 (br. s, 1H, 5ꢁ-OH), 5.02 (s, 2H, CH2-OAryl), 4.18 (m,
1H, H-4ꢁ), 3.67 (m, 2H, H-5ꢁa,b), 2.54 (m, 1H, H-2ꢁb), 2.30 (m, 1H, H-2ꢁa);
MS (ESI-MS, CH3CN): Calcd. for C24H21Cl2N3O7 [M-H]−: 532.08. Found:
532.1.
3ꢁ-O-(4-Azidobutyryl)thymidine (9a) and
3ꢁ-O-(4-Azidobutyryl)-N-acyl-2ꢁ-deoxynucleosides (9b–9d).
General Procedure
A solution of 4-chlorobutyryl chloride (2.1 g, 14.9 mmol) in
dichloromethane (10 ml) was added dropwise with stirring to a
cold (0◦C) solution of 1-methylimidazole (1.5 g, 18.3 mmol) in
dichloromethane (10 ml). A solution of dried 5ꢁ-O-(4,4ꢁ-dimethoxytrityl)-N -
acyl-2ꢁ-deoxynucleoside or 5ꢁ-O-(4,4ꢁ-dimethoxytrityl)thymidine (10 mmol)
in dichloromethane (50 ml) was then added dropwise with stirring and the
resulting solution was left at (0◦C) overnight. The solution was washed with
saturated sodium hydrogen carbonate (2 × 50 ml), 5% oxalic acid until
pH < 5, water. organic layer was dried over sodium sulfate, evaporated
to dryness and the residue was dissolved in N,Nꢁ-dimethylformamide (100
ml). Sodium azide 1.3 g (20 mmol) was added to the solution and the
reaction mixture was stirred at 60◦C for 12 hours. The reaction mixture
was then evaporated to dryness, the residue was dissolved in toluene (150
ml). The toluene solution was washed with water (3 × 100 ml). The organic
layer was separated, dried over sodium sulfate and evaporated to dryness.
A 1% solution of 4-toluenesulphonic acid monohydrate in a mixture of
dichloromethane–methanol (7:3, 150 ml) was added to the oily residue
with vigorous stirring. After 10 min at rt, the acidic reaction mixture
was neutralized by the addition of saturated sodium hydrogen carbonate

492
A. Yagodkin et al.
(50 ml). The organic layer was separated, additionally washed with saturated
sodium hydrogen carbonate, dried over sodium sulfate and evaporated to
dryness. The residue was either chromatographed on the silica gel in a
gradient of methanol in dichloromethane (from 0 to 5%) to give 2.9 g
of 9a and 3.4 g of 9d as foams in 79 and 72% overall yield correspond-
ingly, or crystallized from acetone to give 2.9 g of 9c in 66% yield.; m.p
170–172◦C.
1
9a: H NMR (DMSO-d6), δTMS, ppm; J , Hz: 11.31 (s, 1H, NH of Thy),
7.73 (d, 1H, J = 1.2, H-6 of Thy), 6.18 (dd, 1H, J = 5.9, J = 8.7, H-1ꢁ), 5.24
(m, 1H, H-3ꢁ), 5.20 (t, 1H, J = 5.2, 5ꢁ-OH), 3.98 (m, 1H, H-4ꢁ), 3.63 (m,
2H, H-5ꢁa,b), 3.38 (t, 2H, J = 6.8, -CH2-N3), 2.44 (t, 2H, J = 7.3, -CH2-CO-),
2.26 (m, 2H, H-2ꢁa,b), 1.80 (m, 2H, -CH2-CH2-CH2-), 1.78 (d, 5-CH3 of Thy);
MS (ESI-MS, CH3CN): Calcd. for C14H19N5O6 [M-H]−: 352.13. Found:
352.1.
1
9c: H NMR (DMSO-d6), δTMS, ppm; J , Hz: 11.26 (s, 1H, 4-NH of Cyt),
8.39 (d, 1H, J = 7.4, H-6 of Cyt), 8.01 (m, 2H, Bz), 7.63 (m, 1H, Bz), 7.51
(m, 2H, Bz), 7.38 (d, 1H, H-5 of Cyt), 6.17 (dd, 1H, J = 6.0, J = 7.8, H-1ꢁ),
5.26 (m, 1H, H-3ꢁ), 5.22 (t, 1H, J = 5.2, 5ꢁ-OH), 4.14 (m, 1H, H-4ꢁ), 3.67
(m, 2H, H-5ꢁa,b), 3.39 (t, 2H, J = 6.8, -CH2-N3), 2.50 (m, 1H, H-2ꢁb), 2.45 (t,
2H, J = 7.3, -CH2-CO-), 2.26 (m, 1H, H-2ꢁa), 1.81 (m, 2H, -CH2-CH2-CH2-);
MS (ESI-MS, CH3CN): Calcd. for C20H22N6O6 [M+H]+: 442.16. Found:
441.2, 442.1. 9d: 1H NMR (DMSO-d6), δTMS, ppm; J , Hz: 11.21 (s, 1H, 6-NH
of Ade), 8.76 (d, 0.6 H, H-8 of Ade), 8.71 (s, 1H, H-2 of Ade), 8.05 (m, 2H,
Bz), 7.65 (m, 1H, Bz), 7.55 (m, 2H, Bz), 6.51 (dd, 1H, J = 6.1, J = 8.4, H-1ꢁ),
5.44 (m, 1H, H-3ꢁ), 5.20 (t, 1H, J = 5.7, 5ꢁ-OH), 4.14 (m, 1H, H-4ꢁ), 3.65 (m,
2H, H-5ꢁa,b), 3.41 (t, 2H, J = 6.8, -CH2-N3), 3.06 (m, 1H, H-2ꢁa), 2.58 (m,
1H, H-2ꢁb), 2.48 (t, 2H, J = 7.3, -CH2-CO-), 1.84 (m, 2H, -CH2-CH2-CH2-);
MS (ESI-MS, CH3CN): Calcd. for C21H22N8O5 [M-H]−: 465.17. Found:
465.3.
3ꢁ-O-(2-Azidomethylbenzoyl)-N-acyl-nucleosides (10b–d) and
3ꢁ-O-(2-Azidomethylbenzoyl)thymidine (10a). General Procedure
A solution of 2-azidomethylbenzoyl chloride (2.9 g, 15 mmol) in
dichloromethane (10 ml) was added dropwise with stirring to a cold (0◦C)
solution of 1-methylimidazole (1.5 g, 18.3 mmol) in dichloromethane (10
ml). A solution of dried 5ꢁ-O-(4,4ꢁ-dimethoxytrityl)-N -acylnucleoside or 5ꢁ-
O-(4,4ꢁ-dimethoxytrityl)thymidine (10 mmol) in dichloromethane (50 ml)
was then added dropwise with stirring and the resulting solution was left at
(0◦C) overnight. The solution was washed with saturated sodium hydrogen
carbonate (2 × 50 ml), 5% oxalic acid until pH < 3, water. organic layer
was dried over sodium sulfate and evaporated to dryness. A 1% solution of
4-toluenesulphonic acid in a mixture of dichloromethane–methanol (7:3,
150 ml) was added to the oily residue with vigorous stirring. After 10 minute

Improved Synthesis of Trinucleotide Phosphoramidites
493
at rt, the acidic reaction mixture was neutralized by the addition of saturated
sodium hydrogen carbonate (50 ml). The organic layer was separated,
additionally washed with saturated sodium hydrogen carbonate, dried over
sodium sulfate and evaporated to dryness. Crystallization of the residue from
1
dichloromethane gave 3.1 g (77% yield) of pure 10a, m.p 204–207◦C; H
NMR (DMSO-d6), δTMS, ppm; J , Hz: 11.33 (s, 1H, NH of Thy), 8.00 (dd, 1H,
J = 1.2, J = 7.8, H-6 of 2-azidomethylbenzoyl), 7.79 (d, 1H, J = 1.2, H-6 of
Thy), 7.68 (dt, 1H, J = 1.4, J = 7.5, H-5 of 2-azidomethylbenzoyl), 7.58(dd,
1H, J < 1.0, J = 7.0, H-3 of 2-azidomethylbenzoyl), 7.54 (dt, 1H, J = 1.1, J
= 7.5, H-5 of 2-azidomethylbenzoyl), 6.29 (dd, 1H, J = 6.5, J = 8.3, H-1ꢁ),
5.49 (m, 1H, H-3ꢁ), 5.26 (t, 1H, J = 5.2, 5ꢁ-OH), 4.80 (s, 2H, -CH2-N3), 4.20
(m, 1H, H-4ꢁ), 3.72 (m, 2H, H-5ꢁa,b), 2.43 (m, 2H, H-2ꢁa,b), 1.80 (d, 5-CH3
of Thy); MS (ESI-MS, CH3CN): Calcd. for C18H19N5O6 [M-H]−: 400.13.
Found: 400.1.
Crystallization of the residue from methanol gave 3.5 g (72% yield)
1
of pure 10c, m.p 144–147◦C; H NMR (DMSO-d6), δTMS, ppm; J , Hz:
11.28 (s, 1H, 4-NH of Cyt), 8.39 (d, 1H, J = 7.4, H-6 of Cyt), 8.00 (m,
3 H, H-6 of 2-azidomethylbenzoyl and Bz), 7.69 (t, 1H, J = 7.5, H-5 of
2-azidomethylbenzoyl), 7.63 (t, 1H, J = 7.4, Bz), 7.59 (d, 1H, J = 7.6, H-3 of
2-azidomethylbenzoyl), 7.55 (t, 1H, J = 7.6, H-4 of 2-azidomethylbenzoyl),
7.52 (t, 2H, J = 7.7, Bz), 7.41 (d, 1H, H-5 of Cyt), 6.28 (dd, 1H, J = 6.2,
J = 7.5, H-1ꢁ), 5.51 (m, 1H, H-3ꢁ), 5.29 (t, 1H, J = 5.2, 5ꢁ-OH), 4.81 (s, 2H,
-CH2-N3), 4.35 (m, 1H, H-4ꢁ), 3.75 (m, 2H, H-5ꢁa,b), 2.70 (m, 1H, H-2ꢁb),
2.39 (m, 1H, H-2ꢁa); MS (ESI-MS, CH3CN): Calcd. for C24H22N6O6 [M-H]−:
489.16. Found: 489.2. crystallization of the residue from ethanol gave 3.4 g
(67% yield) of pure 10d, m.p 110–113◦C; 1H NMR (DMSO-d6), δTMS, ppm;
J , Hz: 11.22 (s, 1H, 6-NH of Ade), 8.78 (d, 0.6 H, H-8 of Ade), 8.76 (s, 1H,
H-2 of Ade), 8.05 (m, 3H, H-6 of 2-azidomethylbenzoyl and Bz), 7.70 (dt,
1H, J = 1.4, J = 7.6, H-5 of 2-azidomethylbenzoyl), 7.65 (m, 1H, Bz), 7.61
(dd, 1H, J <1.0, J = 7.6, H-3 of 2-azidomethylbenzoyl), 7.57 (m, 3H, H-4 of
2-azidomethylbenzoyl and Bz), 6.63 (dd, 1H, J = 6.0, J = 8.4, H-1ꢁ), 5.69 (m,
1H, H-3ꢁ), 5.28 (t, 1H, J = 5.7, 5ꢁ-OH), 4.85 (s, 2H, -CH2-N3), 4.34 (m, 1H,
H-4ꢁ), 3.74 (m, 2H, H-5ꢁa,b), 3.20 (m, 1H, H-2ꢁa), 2.78 (m, 1H, H-2ꢁb); MS
(ESI-MS, CH3CN): Calcd. for C25H22N8O5 [M-H]−: 513.17. Found: 513.2.
Purification of the residue on a silica gel, using a gradient of methanol
in methylene chloride from 0 to 5% gave 3.22 g (65% yield) of pure 10b as
1
foam; H NMR (DMSO-d6), δTMS, ppm; J , Hz: 12.08 (s, 1H, 1-NH of Gua),
11.62 (s, 1H, 2-NH of Gua), 8.32 (s, 1H, H-2 of Gua), 8.05 (dd, 1H, J = 1.0,
J = 7.7, H-6 of 2-azidomethylbenzoyl), 7.70 (dt, 1H, J = 1.2, J = 7.5, H-5 of
2-azidomethylbenzoyl), 7.60 (d, 1H, J = 7.6, H-3 of 2-azidomethylbenzoyl),
7.56 (dt, 1H, J <1.0, J = 7.6, H-4 of 2-azidomethylbenzoyl), 6.60 (dd, 1H,
J = 5.7, J = 8.9, H-1ꢁ), 5.59 (m, 1H, H-3ꢁ), 5.19 (t, 1H, J = 5.5, 5ꢁ-OH), 4.82
(s, 2H, -CH2-N3), 4.27 (m, 1H, H-4ꢁ), 3.74 (t, 2H, J = 4.8, H-5ꢁa,b), 2.99 (m,

494
A. Yagodkin et al.
1H, H-2ꢁa), 2.75 (m, 1H, J = 6.9, CH of ibu), 2.69 (m, 1H, H-2ꢁb), 1.11 (d,
6 H, J = 6.8, 2CH3 of ibu); MS (ESI-MS, CH3CN): Calcd. for C22H24N8O6
[M-H]−: 495.18. Found: 495.2.
Triethylammonium Salts of 5ꢁ-O-dimethoxytrityl-N-acyl-2ꢁ-
deoxynucleoside-3ꢁ-O-(2-chlorophenyl)phosphates and
5ꢁ-O-dimethoxytritylthymidine-3ꢁ-O-(2-chlorophenyl)phosphate
(P-Components). General Procedure
A solution of 2-chlorophenyldichlorophospate (8 ml, 49 mmol) in
dichloromethane (40 ml) was added to a cold (0◦C) mixture of triazole
(8 g, 116 mmol), pyridine (20 ml, 250 mmol) and dichloromethane (80
ml). A solution of 5ꢁ-O-dimethoxytrityl-N -acyl-2ꢁ-deoxynucleoside or 5ꢁ-O-
dimethoxytritylthymidine (30 mmol) in dichloromethane (100 ml) was
added dropwise after 30 minutes stirring and the resulting mixture was
stirred for additional 2 hours at 0◦C. A 1 M triethylammonium hydrogen
carbonate (80 ml) was added to the reaction mixture. The organic layer
was separated, washed additionally with 1M triethylammonium hydrogen
carbonate (3 × 70 ml) and dried over sodium sulfate. The solution was evap-
orated, co-evaporated with toluene (2 × 80 ml) to give pure P-components
as white foams in nearly quantitative yield. These foams were used for
preparation of di- and trinucleotides without further purification.
5ꢁ-O-Dimethoxytrityl-3ꢁ-O-(acyl) trinucleotides (17–19).
General Procedure
A mixture of 3ꢁ-acyl-2ꢁ-deoxynucleoside 8a–d, 9a–d, or 10a–d (15 mmol)
and P-component (20 mmol) was dissolved in dry pyridine (150 ml).
MSNT (12.4 g, 42 mmol) was added in portions with stirring. The reaction
mixture was left for 1 hour at rt and quenched with water. The mixture
was evaporated to dryness, co-evaporated with toluene (2 × 100 ml) and
dissolved in dichloromethane (150 ml). The organic solution was washed
with aqueous 5% oxalic acid until pH<3, dried over sodium sulfate and
evaporated to dryness. A 1% solution of 4-toluenesulphonic acid in a
mixture of dichloromethane—methanol (7:3, 150 ml) was added to the oily
residue with vigorous stirring. After 10 minutes at rt, the acidic reaction
mixture was neutralized by the addition of saturated sodium hydrogen
carbonate (50 ml). The organic layer was separated, additionally washed
with saturated sodium hydrogen carbonate (50 ml), dried over sodium
sulfate and evaporated to dryness. purification of the residue on a silica gel,
using a gradient of methanol in dichloromethane from 0 to 7 gave pure
14–16 as foams in 70–80% overall yield.
A mixture of 3ꢁ-O-acyl dinucleotides 14–16 (15 mmol) and P-component
(20 mmol) was dissolved in dry pyridine (150 ml). MSNT (12.4 g, 42 mmol)
was added in portions with stirring. The reaction mixture was left for 1

Improved Synthesis of Trinucleotide Phosphoramidites
495
hour at rt and quenched with 1 M triethylammonium hydrogen carbonate
buffer, pH 7.8 (TEAB). The reaction mixture was concentrated in vacuum,
the residue was co-evaporated with toluene (2 × 100 ml), dissolved in
dichloromethane (150 ml). The dichloromethane solution was washed with
saturated hydrogen sodium carbonate (2 × 100 ml), dried over sodium
sulfate and evaporated to dryness. Purification of the residue on a silica gel,
using a gradient of methanol in dichloromethane from 0 to 7 gave pure
17–19 as foams in 80–90% yield. These foams were used for preparation of
trinucleotides without further purification and characterization.
5ꢁ-O-Dimethoxytrityltrinucleotides (5)
Removal of 3ꢁ-O-(2,4-Dichlorophenoxy)acetyl Group. General Procedure
Fully protected trinucleotide 17 (5 mmol) was dissolved in a mixture
of 100 ml pyridine, 50 ml MeOH, and 17 ml triethylamine. In 3 hours the
reaction mixture was evaporated to a gum, the residue was co-evaporated
with toluene (2 × 100 ml) and with chloroform (2 × 100 ml). The residue
was dissolved in 200 ml chloroform, washed with saturated aqueous sodium
hydrogen carbonate (3 × 100 ml), 0.1 M TEAB (2 × 100 ml), dried over
sodium sulfate and concentrated. The traces of pyridine were removed by
co-evaporation with toluene (3 × 50 ml), the residual gum was dissolved in
chloroform and concentrated to a minimal volume. The desired products 5
were isolated by column chromatography on silica gel. Yields 65–85%. These
trinucleotides were used for preparation of trinucleotides phosphoramidites
7 without further purification and characterization.
Removal of 4-Azidobutyryl Group. General Procedure
Fully protected trinucleotide 18 (10 g) was dissolved in a solution of
triphenylphosphine (2.5 g, 9.5 mmol) in a mixture of dioxane (200 ml)
and water (20 ml) and heated at 50–55◦C for 7 hours. The solution was
evaporated to dryness. Purification of the residue on a silica gel, using a
gradient of methanol in dichloromethane from 0 to 7% gave pure trimers
5 as foams in 80–90% yield. These trinucleotides were used for preparation
of trinucleotides phosphoramidites 7 without characterization.
Removal of 2-Azidomethylbenzoyl Group. General Procedure
Fully protected trinucleotide 19 (10 g) was dissolved in a solution of
triphenylphosphine (2.5 g, 9.5 mmol) in a mixture of dioxane (200 ml)
and water (20 ml) and left overnight at rt. The solution was evaporated
to dryness. purification of the residue on a silica gel, using a gradient
of methanol in dichloromethane (0–7%) gave pure trimers 5 as foams
in 85–95% yield. These trinucleotides were used for preparation of trinu-
cleotides phosphoramidites 7 without characterization.

496
A. Yagodkin et al.
Trinucleotide 3ꢁ-Phosphoroamidites 7. General Procedure
Trinucleotide 5 (5 mmol) was dissolved in a mixture of dry
dichloromethane (22 ml), dry acetonitrile (7 ml) and 2-cyanoethyl-
N,N,Nꢁ,Nꢁ-tetraisopropylphosphorodiamidite (4.75 ml, 4.5 g, 15 mmol). A
0.25M solution of 5-ethylthio-1H-tetrazole (4 ml, 1 mmol) was added and
the mixture was stirred at rt overnight. The reaction was quenched by
addition of saturated sodium hydrogen carbonate (20 ml). The organic
layer was separated, dried over sodium sulfate, and evaporated to dryness.
The residue was vigorously shaken with n-hexane (20 ml), the hexane
extract was decanted and discarded. The residue was dried in vacuum,
dissolved in acetonitrile (30 ml). Water (30 ml) was added to the acetonitrile
solution and the resulting emulsion was applied onto the Lichroprep RP-18
column (4 × 19 cm), equilibrated with 50% aqueous acetonitrile. Gradient
elution of acetonitrile in water (50–100%) afforded pure trimer phospho-
ramidites 7 as foams in 87–93% yield. All 20 Trimer phosphoramidites were
characterized as described above. Their structure was confirmed as reported
earlier.[5]
REFERENCES
1. Zon, G.; Gallo, K.; Samson, C.; Shao, K.; Michael, F.; Byrd, R. Analytical studies of ‘mixed sequence’
oligodeoxyribonucleotides synthesized by competitive coupling of either methyl- or beta-cyanoethyl-
N,N-diisopropylamino phosphoramidite reagents, including 2ꢁ-deoxyinosine. Nucleic Acids Res. 1985,
13, 8181–8196.
2. Kayushin, A.; Korosteleva, M.; Miroshnikov, A. Large-scale solid-phase preparation of 3ꢁ-unprotected
trinucleotide phosphotriesters—Precursors for synthesis of trinucleotide phosphoramidites. Nucleo-
sides Nucleotides Nucleic Acids 2000, 19, 1967–1976.
3. Kayushin, A.; Korosteleva, M. A new approach to the synthesis of trinucleotide phosphoramidites—
Synthons for the generation of oligonucleotide/peptide libraries. Nucleosides Nucleotides 1999, 18,
1531–1533.
4. Kayushin, A.; Korosteleva, M.; Miroshnikov, A.W.; Zubov, D.; Piel, N. A convenient approach
to the synthesis of trinucleotide phosphoramidites—Synthons for the generation of oligonu-
cleotide/peptide libraries. Nucleic Acids Res. 1996, 24, 3748–3755.
5. Mauriala, T.; Auriola, S.; Azhayev, A.; Kayushin, A.; Korosteleva, M.; Miroshnikov, A. HPLC
electrospray mass spectrometric characterization of trimeric building blocks for oligonucleotide
synthesis. J. Pharm. Biomed. Anal. 2004, 34, 199–206.
6. Neylon, C. Chemical and biochemical strategies for the randomization of protein encoding DNA
sequences: library construction methods for directed evolution. Nucleic Acids Res. 2004, 32, 1448–
1459.
7. Sondek, J.; Shortle, D. A general strategy for random insertion and substitution mutagenesis:
substoichiometric coupling of trinucleotide phosphoramidites. Proc. Natl. Acad. Sci. USA 1992, 89,
3581–3585.
8. Ya´n˜ez, J.; Jorge Argu¨ello, M.; Joel Osuna, J.; Sobero´n, X.; Gayta´n, P. Combinatorial codon-based
amino acid substitutions. Nucleic Acids Res. 2004, 32, e158.
9. Virnekas, B.; Liming Ge, L.; Pluckthun, A.; Schneider, C.; Wellnhofer, G.; E. Moroney, S. Trinu-
cleotide phosphoramidites: ideal reagents for the synthesis of mixed oligonucleotides for random
mutagenesis. Nucleic Acids Res. 1994, 22, 5600–5607.
10. Lyttle, M.; Napolitano, E.; Calio, B.; Kauvar, M.L. Mutagenesis Using Trinucleotide Beta-Cyanoethyl
Phosphoramidites. BioTechniques 1995, 19, 274–281.

Improved Synthesis of Trinucleotide Phosphoramidites
497
11. Ono, A.; Matsuda, A.; Zhao, J.; Santi, D. The synthesis of blocked triplet-phosphoramidites and their
use in mutagenesis. Nucleic Acids Res. 1995, 23, 4677–4682.
12. Catlin, J.; Cramer, F. Deoxyoligonucleotide synthesis via the triester method. J. Org. Chem. 1973, 38,
245–250.
13. Agarwal, K.; Riftina, F. Chemical synthesis of a self-complementary octanucleotide, dG-G-T-T-A-A-C-
C by a modified triester method. Nucleic Acids Res. 1978, 5, 2809–2824.
14. Sidhu, S. Personal communication. Department of Protein Engineering, Genentech Inc., South San
Francisco, CA, USA.
15. Kusumoto, S.; Sakai, K.; Shiba, T. 4-Azidobutyryl Group for Temporary Protection of Hydroxyl
Functions. Bull. Chem. Soc. Jpn. 1986, 59, 1296–1298.
16. Wada, T.; Ohkubo, A.; Mochizuki, A.; Sekine, M. 2-(Azidomethyl)benzoyl as a new protecting group
in nucleosides. Tetrahedron Lett. 2001, 42, 1069–1072.
17. Azhayev, A.; Antopolsky, M. Amide-group assisted 3ꢁ-deposphorylation of oligonucleotides synthe-
sized on universal A-supports. Tetrahedron 2001, 57, 4977–4986.