The efficiency of light-directed synthesis of DNA arrays on glass substrates

Article abstract of DOI:10.1021/ja964427a

New methods based on photolithography and surface fluorescence were used to determine photodeprotection rates and stepwise yields for light-directed oligonucleotide synthesis using photolabile 5'-(((α-methyl-2- nitropiperonyl)-oxy)carbonyl)(MeNPOC)-2'-deoxynucleoside phosphoramidites on planar glass substrates. Under near-UV illumination (primarily 365 nm) from a mercury light source, the rate of photoremoval of the MeNPOC protecting group was found to be independent of both the nucleotide and length of the growing oligomer (t( 1/4 ) = 12 s at 27.5 mW/cm²). A moderate dependence on solvent polarity was observed, with photolysis proceeding most rapidly in the presence of nonpolar solvents or in the absence of solvent (e.g., t( 1/4 ) = 10 - 13 s at 27.5 mW/cm²). In solution, the photolysis rate was linearly dependent on light intensity over the range 5-50 mW/cm². Average stepwise yields for the synthesis of dodecamer oligonucleotides were in the range of 92-94%, using monomers based on N⁶-(phenoxyacetyl)-2'-deoxyadenosine, N²- isobutyryl-2'-deoxyguanosine, N⁴-isobutyryl-2'-deoxycytidine, and thymidine. By comparison, an efficiency of 98%/step was obtained using a conventional 5'-dimethoxytrityl monomer with acid deprotection on the same support. The lower yields associated with the photochemical process appears to be due to incomplete recovery of free 5'-hydroxyl groups after photolysis on the support, although high yields of 5'-OH nucleosides (≤96%) are consistently observed when 5'-MeNPOC monomers are photolyzed in solution.

Full text of DOI:10.1021/ja964427a

VOLUME 119, NUMBER 22
J UNE 4, 1997
© Copyright 1997 by the
American Chemical Society
The Efficiency of Light-Directed Synthesis of DNA Arrays on
Glass Substrates
Glenn H. McGall,* Anthony D. Barone, Martin Diggelmann,† Stephen P. A. Fodor,
Erik Gentalen, and Nam Ngo
Contribution from Affymetrix, Inc., 3380 Central Expressway, Santa Clara, California 95051
ReceiVed December 27, 1996X
Abstract: New methods based on photolithography and surface fluorescence were used to determine photodeprotection
rates and stepwise yields for light-directed oligonucleotide synthesis using photolabile 5′-(((R-methyl-2-nitropiperonyl)-
oxy)carbonyl)(MeNPOC)-2′-deoxynucleoside phosphoramidites on planar glass substrates. Under near-UV illumina-
tion (primarily 365 nm) from a mercury light source, the rate of photoremoval of the MeNPOC protecting group was
found to be independent of both the nucleotide and length of the growing oligomer (t1/2 ) 12 s at 27.5 mW/cm2).
A moderate dependence on solvent polarity was observed, with photolysis proceeding most rapidly in the presence
of nonpolar solvents or in the absence of solvent (e.g., t1/2 ) 10-13 s at 27.5 mW/cm2). In solution, the photolysis
rate was linearly dependent on light intensity over the range 5-50 mW/cm2. Average stepwise yields for the synthesis
of dodecamer oligonucleotides were in the range of 92-94%, using monomers based on N6-(phenoxyacetyl)-2′-
deoxyadenosine, N2-isobutyryl-2′-deoxyguanosine, N4-isobutyryl-2′-deoxycytidine, and thymidine. By comparison,
an efficiency of 98%/step was obtained using a conventional 5′-dimethoxytrityl monomer with acid deprotection on
the same support. The lower yields associated with the photochemical process appears to be due to incomplete
recovery of free 5′-hydroxyl groups after photolysis on the support, although high yields of 5′-OH nucleosides (g96%)
are consistently observed when 5′-MeNPOC monomers are photolyzed in solution.
Introduction
building blocks in combination with polymeric semiconductor
photoresist films as the photoimageable component.3
The development of chemistry and processes for DNA array
Arrays of immobilized oligonucleotide probes (DNA “chips”)
have emerged as powerful new tools for parallel hybridization-
based analysis of DNA and RNA sequence.1 Light-directed
oligonucleotide synthesis provides an efficient and versatile
method for microfabricating probe arrays with densities as high
as 106 unique sequences/cm2.2 In this approach, 5′-terminal
protecting groups are selectively removed from growing oli-
gonucleotide chains in predefined regions of a glass support by
controlled exposure to light through photolithographic masks.
This technique has been implemented using oligonucleotide
building blocks with photolabile 5′-protecting groups2 and, more
recently, using conventional 5′-(4,4′-dimethoxytrityl) (“DMT”)
(1) (a) Lipshutz, R. J.; Morris, D.; Chee, M.; Hubbell, E.; Kozal, M. J.;
Shah, N.; Shen, N.; Yang, R.; Fodor, S. P. A. Biotechniques 1995, 19, 442-
447. (b) Cronin, M. T.; Fucini, R. V.; Kim, S. M.; Masino, R. S.; Wespi,
R. M.; Miyada, C. G. Hum. Mutat. 1996, 7, 244-255. (c) Sapolsky, R. J.;
Lipshutz, R. J. Genomics 1996, 33, 445-456. (d) Kozal, M. J.; Shah, N.;
Shen, N.; Yang, R.; Fucini, R.; Merigan, T. C.; Richman, D. D.; Morris,
M. S.; Hubbell, E.; Chee, M. S.; Gingeras, T. R. Nat. Med. 1996, 2, 753-
759. (e) Yershov, G.; et al. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4913-
4918. (f) Chee, M. S.; Huang, X.; Yang, R.; Hubbell, E.; Berno, A.; Stern,
D.; Winkler, J.; Lockhart, D. J.; Morris, M. S.; Fodor, S. P. A. Science
1996, 274, 610-614. (g) Lockhart, D. J.; Dong, H.; Byrne, M. C.; Follettie,
M. T.; Gallo, M. V.; Chee, M. S.; Mittmann, M.; Wang, C.; Kobayashi,
M.; Horton, H.; Brown, E. L. Nat. Biotechnol. 1996, 14, 1675-1680. (h)
Hacia, J. G.; Brody, L. C.; Chee, M. S.; Fodor, S. P. A.; Collins, F. S. Nat.
Genet. 1996, 14, 441-47. (i) Shoemaker, D. D.; Lashkari, D. A.; Morris,
D.; Mittmann, M.; Davis, R. W. Nat. Genet. 1996, 14, 450-456. (j)
Southern, E. M. Trends Genet. 1996, 12, 110-115.
† Present address: Ciba-Geigy Ltd., CH-4002 Basel, Switzerland.
X Abstract published in AdVance ACS Abstracts, May 15, 1997.
S0002-7863(96)04427-7 CCC: $14.00 © 1997 American Chemical Society

5082 J. Am. Chem. Soc., Vol. 119, No. 22, 1997
McGall et al.
fabrication requires access to routine procedures for measuring
on-chip yields and efficiencies of the chemical and photochemi-
cal steps involved. Methods commonly employed in standard
oligonucleotide synthesis, such as spectrophotometric monitoring
of released trityl cation4 and HPLC or CE analysis of products
after cleavage from the support,5 are not readily adaptable to
DNA synthesis on planar, silanized glass supports. This is due
to the fact that the surface coverage of the hydroxyalkylsilane
synthesis sites (∼10-30 pmol/cm2)6 is too low for accurate
determination of oligomers or protecting groups by absorbance
spectrophotometry, either bound to the support or after subse-
quent release.
Scheme 1
In this paper, we present a quantitative study of light-directed
oligonucleotide synthesis on planar supports using procedures
based on photolithography and surface fluorescence. These
methods involve quantitative “labeling” of the unprotected 5′-
hydroxyl groups associated with the bound nucleotides or
oligonucleotides using a fluorescent phosphoramidite derivative
and measuring surface fluorescence by confocal microscopy.
Photochemical protection rates and stepwise yields for oligo-
nucleotide synthesis can both be monitored directly on the
support. Although the present study focuses on photolitho-
graphic synthesis using 5′-(R-methyl-2-nitropiperonyl)oxycar-
bonyl(“MeNPOC”)-2′-deoxynucleoside phosphoramidites,2b the
techniques that are described here can be applied to the analysis
of any reagent or process using in situ synthesis to fabricate
arrays of biopolymers such as DNA, peptides, and analogs
thereof.
Results and Discussion
The 5′-MeNPOC-protected phosphoramidite building blocks
7a-k were prepared according to the method outlined in
Scheme 1.2b In order to develop reagents and processes for
light-directed DNA array synthesis, one must be able to measure
and monitor two basic parameters. One parameter is the rate
of photolytic release of the 5′-protecting group, related to the
reaction quantum efficiency, which is necessary to determine
the minimum amount of light required to completely deprotect
terminal residues on the support-bound oligomers during
synthesis. The second parameter is the stepwise yield of
monomer addition over the number of steps required to
synthesize oligomers of the desired length.
Surface Fluorescence Analysis. In this work, the aformen-
tioned parameters were obtained by measuring the relative
surface coverage of unprotected 5′-terminal hydroxyl groups,
in one case as a function of the time and intensity of light
exposure and, in the other case, as a function of the number of
complete cycles of combined monomer addition and photolysis.
Photolithographic masking was used to confine light exposure
and monomer coupling to specified regions of the substrate
which could then be addressed separately for comparison.
Surface hydroxyl groups were quantitatively reacted with a
fluorescein phosphoramidite derivative (8) using the standard
sequence of tetrazole-catalyzed coupling followed by capping
(acetic anhydride/1-methylimidazole/2,6-lutidine/THF) and oxi-
dation (I2/pyridine-H2O).7 After removing the acyl protecting
groups from the bound fluorescein, relative densities of hydroxyl
groups in different regions of the support could then be
determined from surface fluorescence intensities.
(2) (a) Fodor, S. P. A.; Read, J. L.; Pirrung, M. C.; Stryer, L. T.; Lu, A.;
Solas, D. Science 1991, 251, 767-773. (b) Pease, A. C.; Solas, D.; Sullivan,
E. J.; Cronin, M. T.; Holmes, C. P.; Fodor, S. P. A. Proc. Natl. Acad. Sci.
U.S.A. 1994, 91, 5022-5026.
(3) (a) McGall, G.; Labadie, J.; Brock, P.; Wallraff, G.; Nguyen, T.;
Hinsberg, W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 13555-13560. (b)
Wallraff, G.; Labadie, J.; Brock, P.; Nguyen, T.; Huynh, T.; Hinsberg, W.;
McGall, G. Chemtech 1997, February, 22-32.
For the purpose of this study, it was not necessary to achieve
an absolute measure of the amount of bound fluorescein in any
given region of the substrate, although the photon-counting
capability of the fluorescence microscope would, in principle,
enable one to do so. Instead, differences in surface fluorescence
were used to obtain relatiVe values for surface density, providing
a simple, internally consistent method for measuring chemical
and photochemical efficiencies.
(4) Atkinson, T.; Smith, M. In Oligonucleotide Synthesis. A practical
Approach; Gait, M. J., Ed.; IRL Press: Oxford, 1984; pp 35-81.
(5) (a) Warren, W. J.; Vella, G. In Protocols for Oligonucleotides and
Conjugates; Agrawal, S., Ed.; Humana Press: Totowa, New Jersey, 1994;
pp 233-264. (b) Andrus, A. Ibid; pp 277-300.
(7) Beaucage, S. L. In Protocols for Oligonucleotides and Analogs;
Agrawal, S., Ed.; Humana Press: Totowa, New Jersey, 1993; pp 33-61.
(6) McGall, G. H.; Barone, A. D. Unpublished results.

Light-Directed Synthesis of DNA Arrays on Glass
J. Am. Chem. Soc., Vol. 119, No. 22, 1997 5083
Scheme 2
Figure 1. Observed surface fluorescence intensity of hydroxyalkyl-
silanated glass after coupling mixtures 8 and 5′-DMT-T-CEP in varying
proportions. Total phosphoramidite concentration was 50 mM in all
cases. Fluorescence data was obtained from images acquired using a
scanning confocal fluorescence microscope with photon-counting
electronics.
One potential source of interference with this kind of analysis
is fluorescence quenching due to energy transfer interactions
between adjacent fluorophores on the surface. The initial
density of surface functional groups on the silanated glass
substrates that were used in this work have been estimated to
be in the range of 10-30 pmol/cm2.6 Assuming that the initial
silanation of the support gives a uniform distribution of
molecules, the mean distance between sites will be e∼40 Å.
With such close proximity between fluorophores on the surface,
significant quenching interactions would be expected,8 and
quantitative analysis would be complicated by the lack of direct
proportionality between surface fluorescence intensity and
fluorophore density. Quite strong fluorescence quenching was,
in fact, observed when substrates were labeled with stoichio-
metric amounts of the fluorescent phosphoramidite 8. However,
by diluting the fluorescent reagent with an equally reactive, but
nonfluorescent compound, quenching effects could be mini-
mized. Figure 1 shows the dependence of surface fluorescence
as a function of the mole fraction of 8, when used with 5′-
DMT-thymidine-3′-O-(â-cyanoethyl)-N,N-diisopropylphosphor-
amidite (DMT-T-CEP) as the diluent reagent. Fluorescence
intensity was seen to increase linearly with concentration of 8
up to a mole fraction of 0.2. The observed fluorescence reached
a maximum and then dropped as the amount of fluorescein was
increased further, consistent with increased quenching at higher
surface fluorophore densities. These data established that
quenching effects would be insignificant, provided that the
diluent amidite was present in at a least a 5-fold excess over
the fluorescent amidite. In order to ensure that a reliably linear
relationship was maintained between surface fluorescence and
fluorophore density, a 10-fold excess of diluent amidite was
used in all of the experiments described below.
Figure 2. Surface fluorescence image of a substrate bearing a
5′-MeNPOC-protected nucleoside, exposed through verticle stripes in
a photolithographic mask to near-UV light for increasing times, covering
a range of 3-450 s, as outlined in Scheme 2. The image was acquired
after “staining” the substrate with a 1:10 mixture of fluorescein- and
5′-DMT-thymidine phosphoramidites.
point for subsequent light-directed DNA synthesis. A photo-
lithographic mask is positioned over the back of the substrate
to allow a selected portion of the substrate, in this case an open
verticle aperture 0.8 × 12.8 mm, to be photolyzed by exposure
to near-UV light from a constant-intensity mercury light source
for a predetermined duration. The mask is then translated
horizontally 0.8 mm, placing the aperture of the mask over an
adjacent region for a subsequent, longer exposure. This process
is continued, increasing the exposure time each time, to generate
an array of stripes across the chip with a gradient of increasing
exposure dose, and therefore increasing extent of deprotection.
The pattern of surface deprotection is then “stained” by coupling
the fluorescein phosphoramidite mixture.
In a final step, the isobutyryl protecting groups are removed
from the bound fluorescein by immersing the substrate in a base
solution (1,2-diaminoethane in ethanol). The resulting surface
fluorescence image, acquired with a scanning confocal micro-
scope, is shown in Figure 2, and the fluorescence intensity
change vs exposure time, extracted from the image, is plotted
Surface Photolysis Rates. Scheme 2 illustrates a typical
strategy for determining the rates of photolytic deprotection for
monomers coupled to the support. In all experiments, hydroxy-
alkylated glass substrates were first derivatized with a MeNPOC-
protected hexaethyleneglycol (HEG)-based linker phosphora-
midite, MeNPOC-[OCH2CH2]6O-CEP (9), to provide a starting
(8) Forster, T. Naturwissenschaften 1946, 6, 166-175. (b) Fung, B. K.
K.; Stryer, L. Biochemistry 1978, 17, 5241-5248.

5084 J. Am. Chem. Soc., Vol. 119, No. 22, 1997
McGall et al.
Table 2. Dependence of Photolysis Half-Life on Solventa
entry
solvent
(none)
toluene
dioxane
DCM
DMF
acetonitrile
methanol
t1/2 (s)
1
2
3
4
5
6
7
8
10
11
16
19
20
30
a Data is for 5′-MeNPOC-dA(pac) in dioxane under near-UV output
from Hg source (365 nm, 35 mW/cm2).
mophore (λmax ) 345 nm, ꢀ ) 5 × 103 M-1 cm-1) becomes
negligible at wavelengths above 400 nm. This was confirmed
by the observation of a 10-fold reduction in photolysis rates
when filters were added to remove the 365 nm emission peak.
No significant effect was observed upon eliminating the 400
nm or 436 nm emission lines. Photocleavage of the MeNPOC
group is also induced by irradiation at lower wavelengths (data
not shown). However, wavelengths below 340 nm were avoided
due to the potential for photochemically-induced modification
of the DNA.9
Figure 3. Plot of fluorescence intensity vs exposure time, obtained
from the image shown in Figure 2. Intensity values were taken as the
average pixel intensity in counts per second, in each region of the
substrate exposed for a given time. The data have been fitted to a first-
A brief examination of solvent effects (Table 2) revealed that
photocleavage rates proceeded most rapidly under dry condi-
tions, or when the substrate was maintained under a nonpolar
solvent such as toluene or dioxane. A reasonably good
correlation (not shown) between ln(kobsd) and the Dimroth ET
solvent polarity scale10 is suggested by the data. However, given
the present understanding of the photochemistry of 2-nitrobenzyl
compounds,11 the mechanistic significance of the solvent effect
is unclear. It is also worth noting that oligomer length does
not significantly affect the photocleavage rate of the terminal
nucleotide (Table 1, entries 7-11). For the automated multistep
synthesis of arrays, this is a matter of some convenience as it
allows a single photolysis time to be used for every cycle of
monomer addition throughout the synthesis. Another practical
consideration for synthesis of DNA probe arrays is the use of
increasingly higher light intensities to reduce the photolysis time
and therefore the overall duration of each synthesis cycle.
Saturation of the excited state at high photon flux would present
a potential obstacle or limitation to this approach. However,
over the range of 5-50 mW/cm2 at 365 nm (Figure 4),
photocleavage half-lives obeyed an inverse-linear dependence
on light intensity indicating the absence of any saturation of
the excited state at these powers. At 50 mW/cm2, less than 1
min is required for complete (i.e., 10 half-lives) photolysis of
the MeNPOC group. In principle, even higher intensity light,
such as that provided by a nitrogen laser, could be used to
achieve even shorter exposure times.
Stepwise Synthesis Efficiency. Scheme 3 illustrates how a
similar methodology was used for the determination of stepwise
synthesis efficiencies. In this experiment, a mask with a
rectangular aperture was positioned over the substrate and
exposed for sufficient time to allow complete photolysis (10
half-lives) of the linker. A MeNPOC-nucleoside phosphora-
midite was coupled to the exposed region of the support, and
after capping and oxidation, the mask was offset horizontally
by 1/n × W, where W is the width of the open reticle and n is
the oligomer length to be tested. The photolysis was repeated,
and a second building block was added. This process was
order exponential (solid line). Inset shows linear plot of ln(∆I ) I∞
It) vs time.
-
Table 1. Photolysis Half-Lives for Base-Protected 5′-MeNPOC
Nucleotides on Glass Substratesa
entry
monomer
HEG
T
linker
HEG
HEG
HEG
HEG
HEG
HEG
HEG
HEG
HEG
(dA)4(HEG)
(dA)8(HEG)
(dA)12(HEG)
(dT)12(HEG)
t1/2 (s)
1
2
3
3
5
4
4
4
6
7
8
9
10
11
9
12
11
12
12
13
11
12
12
12
10
12
12
13
T(ibu)
dC(ibu)
dG(ibu)
dG(pac)
dG(DMT)
dG(pac/dpc)
dI
dA(pac)
dA(pac)
dA(pac)
dA(pac)
dA(pac)
a Photolysis carried out in the presence of dioxane under near-UV
output from Hg source (365 nm/27.5mW/cm2).
in Figure 3. The change in surface fluorescence with exposure
time followed a first-order exponential increase from which the
photolysis rate constant, or half-life (t1/2), was obtained by simple
curve-fitting analysis.
To measure photocleavage rates for the other MeNPOC
building blocks, portions of a MeNPOC-linker modified support
were completely photolyzed (10 half-lives) to allow addition
of the desired monomer, and the chip was then subjected to the
same sequence of exposures, staining, imaging, and analysis
described above. The photolysis data are summarized in Table
1. Rates for photolysis of MeNPOC monomers bound to the
glass substrate were found to be very similar for all of the
building blocks studied and in reasonably close agreement with
values measured previously for several 5′-MeNPOC-deoxy-
nucleosides in dioxane solution with monochromatic 365 nm
light.2b In this work, exposures were not carried out with
monochromatic light, but with collimated light from a com-
mercially available high-pressure mercury arc lamp. The lamp
was fitted with dichroic reflectors to selectively pass light in
the near-UV region of the spectrum, which consists primarily
of the Hg emission lines centered at 365, 405, and 436 nm.
The 365 nm emission is almost exclusively responsible for
photochemistry, as the absorbance of the MeNPOC chro-
(9) Cadet, J.; Vigny, P. In Bioorganic Photochemistry; Morrison, H.,
Ed.; Wiley: New York, 1990; Vol. 1, pp 1-272.
(10) Dimroth, K.; Reichardt, C.; Siepman, T.; Bohlmann, F. Ann. Chem.
1984, 661, 1.
(11) For a recent review, see: Corrie, J. E. T.; Trentham, D. R. In
Bioorganic Photochemistry; Morrison, H., Ed.; Wiley: New York, 1993;
Vol. 2, pp 243-305.

Light-Directed Synthesis of DNA Arrays on Glass
J. Am. Chem. Soc., Vol. 119, No. 22, 1997 5085
Figure 4. Dependence of the photolysis half-life on light intensity at
365 nm for compound 10 in dioxane at 200 mm.
Figure 5. Surface fluorescence image of an array of poly(dC)
oligomers ranging in length from 1-12, synthesized as outlined in
Scheme 3. The image was acquired after completely photolysing the
upper half of the array and “staining” with 8 as described in the text.
Scheme 3
% yield (step n) ) 100 × (In/In-1
)
(1)
where In ) fluorescence intensity for oligomer of length n.
Representative data for the synthesis of (dC)12 and T12 using
5′-MeNPOC building blocks are plotted in Figure 6. This figure
serves to illustrate a common feature observed with MeNPOC
amidites, namely, an apparent increase in stepwise yields over
the first ∼6 steps of oligomer synthesis to maximum values in
the range 92-98%. Also shown in Figure 6, for comparison,
is the stepwise efficiency observed for the synthesis of T12 using
conventional 5′-DMT-protected phosphoramidite on the same
substrates. The latter was determined in an experiment using
a masking sequence similar to the one employed in the
photolysis kinetics experiment (Vide supra), wherein an open
verticle aperture was translated horizontally across the substrate,
completely deprotecting regions of “fresh” linker-modified
substrate each time. In this case, after each exposure, a 5′-
DMT-protected nucleoside phophoramidite was coupled to the
support and then detritylated by washing the support with acid
(3% trichloroacetic acid in dichloromethane). The resulting
(dN)0-12 array of oligomers was stained and analyzed according
to the protocol described above. As expected, synthesis with
the DMT-monomers proceeded uniformly with very high
efficiency (∼98 ( 1%/step).
repeated for a total of “n” cycles of photolysis and coupling to
generate a duplicate set of stripes on the support comprising a
complete set of oligomers of length 1 to n. Upon completion,
half of the array was subjected to a final wide-field photolysis,
as shown, to release photolabile groups from the 5′-termini of
the finished oligomers, including a region of the previously
unphotolyzed linker for comparison (“n ) 0” control). Fluo-
rescein-CEP was then added to label the free 5′-hydroxyl groups
for quantitation as described above. The unphotolyzed half of
the array was left to provide an internal control for background
fluorescence due to nonspecific binding of the fluorescein in
regions of the substrate where synthesis had occurred.
Table 3 summarizes the stepwise efficiency data for synthesis
of a series of homopolymers with MeNPOC-protected mono-
mers. Interestingly, the chemical protecting groups on the
nucleobase can, in some cases, influence the observed efficien-
cies. For deoxyguanosine, substantially higher yields were
observed when the exocyclic amine was protected with isobu-
tyryl (iBu) compared to phenoxyacetyl (pac) or DMT. The
addition of protecting groups (iBu, benzoyl) to the amide N3-
position of thymidine had little discernable effect on the cycle
yields, while supplemental protection of the O6 (amide) of
deoxyguanosine with N,N-diphenylcarbamoyl (dpc) reduced
yields.
Figure 5 shows the fluorescence image of a (dC)0-12-0 array
synthesized with 5′-MeNPOC phosphoramidites. Fluorescence
decreases somewhat toward the center of the pattern as the yield
of the full-length oligomer decreases with increasing length.
After correcting for background fluorescence, the relative yield
for each step in the synthesis was calculated from the ratio of
the fluorescence intensities in adjacent stripes (eq 1).
The accuracy of the values determined by the methods
described above relies on the assumption that the fluorescence
quantum efficiency of the terminal fluorescent label is inde-
pendent of oligomer length. Two tests confirm that this is a

5086 J. Am. Chem. Soc., Vol. 119, No. 22, 1997
McGall et al.
complete (10t1/2) photolysis is affected, see above). Attempts
to enhance yields by carrying out the photolysis step in the
presence of mild acids, bases, oxidants, reductants, and other
reagents during the photolysis step failed to identify conditions
under which efficiencies were substantially improved over those
obtained when exposures are performed under neat solvent or
dry conditions. With respect to the latter approach, it should
be noted that the range of possible conditions which one might
contemplate using is somewhat restricted by the need to maintain
compatibility with the surface oligonucleotide and linkage
chemistry during the photolysis step.
Another possible explanation for the observed efficiencies
was that the MeNPOC protecting group might be prone to either
modification or removal during the activation/coupling, capping,
or oxidation steps, such that competent, photolyzable synthesis
sites were being capped or otherwise lost. However, tests which
involved extended or repeated exposure to each of these reagents
prior to the photolysis step, did not lead to any reduction in the
observed cycle yields, establishing the overall stability of the
MeNPOC group toward the reagents and conditions used in
oligonucleotide synthesis.
Figure 6. Observed stepwise synthesis efficiencies for (dC)12 (O) and
T12 (4), using 5′-MeNPOC building blocks 7h,i. Values were calculated
Solution Phase Photochemistry. Subquantitative yields for
the photochemical cleavage may be due to an intrinsic partition-
ing of products from the excited state of the nitrobenzyl
chromophore, such that, in addition to the desired products
resulting from heterolytic cleavage of the benzylic C-O bond,
one or more stable side products may formed which remain
attached to the 5′-oxygen, effectively blocking it from subse-
quent reaction. This hypothesis led us to study the photolysis
of several model compounds in solution, where reactions could
be performed on a scale that would allow a direct analysis of
the products, to determine whether any unusual species were
formed and to what extent.
from surface fluorescence as described in the text. Corresponding data
for T12 using conventional 5′-DMT building blocks (9) is shown for
comparison.
Table 3. Mean Stepwise Yields for the Synthesis of (dN)12
Homopolymers Using 5′-MeNPOC-Nucleoside Phosphoramidites
mean stepwise yield
(steps 6-12/1-12)
entry
nucleoside
dA(N6-pac)
a
b
c
d
e
f
g
h
i
96/94
93/91
88/85
84/81
88/87
86/83
90/89
98/94
96/92
91/88
96/93
dG(N2-ibu)
dG(N2-pac)
dG(N2-DMT)
dG(N2-ibu/O6-dpc)
dG(N2-pac/O6-dpc)
dI
Previously, base-protected 5′-MeNPOC (3′-OH) nucleosides
were photolyzed in dioxane solution and analyzed by HPLC to
determine photolysis rates, and only clean conversion to the
free 5′-OH nucleoside was reported.2b This result was confirmed
in the present study (data not shown). In addition, we also
prepared the model nucleotide 5′-MeNPOC-N2-pac-2′-deox-
yguanosine-3′-(2-cyanoethyl)methylphosphate (10, Scheme 4)
which enabled us to study a closer structural analog to the
terminal 5′-MeNPOC nucleotides involved in oligonucleotide
synthesis. The pac-protected dG nucleotide was chosen in this
case, since the surface-fluorescence analysis indicated the lowest
stepwise yields with the corresponding phosphoramidite 7c (av
85%/step, Table 3) and, hence, the greatest possibility of
observing side products.
dC(N4-ibu)
T
j
k
T(N3-ibu)
T(N3-bz)
a Photodeprotection carried out in the presence of dioxane under near-
UV output from Hg source (365 nm/27.5 mW/cm2).
reasonably good approximation. One is the stepwise yields
determined using a DMT monomer with acid deprotection,
which were uniformly high at every step, in accord with the
known yields for these reagents on particulate supports. The
second test was performed by “spiking” a MeNPOC phosphora-
midite with known amounts of a chain-terminating phosphora-
midite, such as 5′-O-methylthymidine-CEP, to compete for and
eliminate active synthesis sites when the amidite is coupled to
the substrate. This resulted in the expected decrease in the
observed stepwise yield, proportional to the amount of termina-
tor added (a 5% decrease yield was seen with a 95:5 mixture
of MeNPOC/methyl-T-CEP.
The somewhat lower stepwise yields obtained with MeNPOC
compounds, as compared to the standard DMT chemistry,
appears to be related to the net yield of free 5′-hydroxyl groups
after photolysis of the MeNPOC group. Inefficiency of the
tetrazole-catalyzed coupling step itself is not responsible for the
observed yields, since in all cases, yields were unaffected by
increasing either the coupling time, reagent (i.e., tetrazole
activator, amidite) concentration, or the total number of repeti-
tions of “fresh” amidite addition for each coupling step.
Stepwise yields were also unaffected by prolonged exposure
times (up to 100 half-lives) or by variation of the solvent, light
intensity, or wavelengths used (only the total time required for
As shown in Figure 7, HPLC analysis indicated that the
photolysis of 10 in dioxane under near-UV illumination resulted
in quite clean conversion to the free 5′-OH nucleoside 12 and
nitrosoketone 11, which were identified by comparison with
the authentic materials. The HPLC also revealed the presence
of several minor, unidentified byproducts, but these accounted
for e5% of the total peak integral (excluding 11), so this would
not explain the 85-88% stepwise yield (Table 3) observed with
7c on the glass support. Moreover, these byproducts appeared
to be related to the subsequent photochemical breakdown of
the primary photoproduct 11, since the same peaks were
observed in the HPLC when samples of authentic 11 were
photolyzed independently under identical conditions.
The solution studies described thus far were performed at
low concentrations (∼200 mM) to avoid quenching the photo-
cleavage rate by internal filtering. These concentrations would
not specifically replicate the average close spacing (∼40 Å) of
MeNPOC groups linked to oligonucleotide chains on the surface

Light-Directed Synthesis of DNA Arrays on Glass
J. Am. Chem. Soc., Vol. 119, No. 22, 1997 5087
Scheme 4
of the glass substrate. For this reason, the photolysis of 1,3-
bis-MeNPOC-propanediol (13, Scheme 5) was examined in
order to model the close spacing of chromophores on a support
and to allow for the possibility that the nitrobenzyl excited states
or reactive intermediates might undergo intermolecular interac-
tions with neighboring molecules that would account for the
observed photolysis efficiencies. However, as illustrated in
Figure 8, compound 13 photolyzed rapidly and cleanly, generat-
ing 1,3-propanediol and nitrosoketone 11 almost exclusively
(>96% yield). Further studies, in which 10 was photolyzed at
high concentrations in solution (60 mM), failed to reveal any
significant byproducts other than the nitrosoketone and its
photochemical breakdown products.
Figure 7. Analysis of the photolysis of compound 10 in dioxane by
reverse-phase HPLC ([10] ) 200 µm) (a) before photolysis and (b)
after 300 s of exposure to near-UV light from a mercury arc source
(10 mW/cm2 at 365 nm). The products (11 and 12) were identified by
comigration with authentic compounds.
Scheme 5
Summary and Conclusions
With this work we have established a number of routine
techniques, based on surface fluorescence, that can be generally
applied to the analysis of light-directed oligonucleotide synthesis
on planar glass supports. These methods were used to determine
photodeprotection rates and stepwise efficiencies for oligo-
nucleotide synthesis using a series of 5′-MeNPOC-nucleoside
phosphoramidite building blocks. Using building blocks pre-
pared from N6-pac-dA, N2-iBu-dG, N4-iBu-dC, and T, DNA
probe arrays can be synthesized with average stepwise yields
in the range of 92-94%. These yields are somewhat lower
than those achievable with conventional 5′-DMT phosphora-
midites and chemical deprotection (98%/step), apparently due
to incomplete recovery of free 5′-hydroxyl groups after pho-
tolysis. A working hypothesis is that photolysis on the support
is accompanied by the occurence of residual byproducts which
remain attached to the 5′-hydroxyl termini and prevent further
chain extension. Model experiments, however, indicate that
photorelease of the MeNPOC protecting group proceeds with
high yield (g96%) in solution. In order to gain a better
understanding of the mechanisms and factors influencing
synthesis efficiency on glass substrates, further efforts are being
undertaken to detect and characterize photolysis products
directly on the support. Other photolabile protecting groups12,13
are also being studied in order to identify potentially more
efficient alternatives to the MeNPOC group.
(12) For recent reviews, see: ref 11. Pillai, V. N. R. In Organic
Photochemistry; Padwa, A., Ed.; Marcel Dekker: New York, 1987; Vol.
9, pp 225-323.

5088 J. Am. Chem. Soc., Vol. 119, No. 22, 1997
McGall et al.
important objective, however, as greater homogeneity of probe
length will provide greater flexibility in the design and applica-
tion of these arrays.
Experimental Section
General Methods and Materials. HPLC analyses were performed
on Beckman System Gold using absorbance detection at 260 nm and
on a Beckman reverse-phase column (5 µm C18-silica, 4 mm I.D. ×
250 mm L) eluted with a linear gradient of acetonitrile in 0.1 M aqueous
triethylammonium acetate (TEAA)/pH 7.2 at a flow rate of 1 mL/min.
UV-vis spectra were obtained on a Varian Cary 3E spectrophotometer.
1H and 31P nuclear magnetic resonance (NMR) spectra were recorded
on a Varian Gemini-400 spectrometer. Mass spectra were recorded
on the following instruments: electrospray ionization (ESI-MS), Hewlett
Packard HP59987A; electron impact ionization (EI-MS), Hewlett
Packard HP5989B; positive ion fast atom bombardment (FAB-MS),
MicroMass ZABZ-EQ. Elemental analyses were performed by Quan-
titative Technologies, Inc., Whitehouse, NJ.
All reagents and anhydrous solvents were obtained from Aldrich,
except for the following: 20% w/v phosgene in toluene, from Fluka;
thymidine, 2′-deoxyinosine, N2-(phenoxyacetyl)-2′-deoxyguanosine, N2-
isobutyryl-2′-deoxyguanosine, N4-isobutyryl-2′-deoxycytidine, N6-(phe-
noxyacetyl)-2′-deoxyadenosine, and diisopropylammonium tetrazolide,
from Chem-Impex (Wood Dale, IL); 2-(cyanoethyl)-N,N,N′,N′-tetra-
isopropylphosphorodiamidite, from ChemGenes (Waltham, MA); fluo-
rescein phosphoramidite (Fluoreprime) and 5′-DMT-2′-deoxynucleoside
3′-phosphoramidites, from Pharmacia Biotech; DNA synthesis reagents
and anhydrous acetonitrile from Glen Research (Sterling, VA); silica
gel (60 Å pore size, 230-400 mesh for flash chromatography), E.
Merck. N3-Benzoylthymidine was prepared as previously described.14
N2-(4,4′-Dimethoxytrityl)-2′-deoxyguanosine (DMT-dG)15 and O6-(N,N-
diphenylcarbamoyl)-N2-(phenoxyacetyl)-2′-deoxyguanosine (dpc/pac-
dG)16 were prepared using modifications of reported methods which
are available as Supporting Information. O6-(N,N-Diphenylcarbamoyl)-
N2-isobutyryl-2′-deoxyguanosine (dpc/iBu-dG) was prepared in the
same way as dpc/pac-dG and converted directly to the 5′-MeNPOC
derivative without isolation.
Substrate Preparation. Glass microscope slides (2 in × 3 in ×
0.027 in, from Erie Scientific) were cleaned by soaking successively
in Nanostrip (Cyantek, Fremont, CA) for 15 min, 10% aqueous NaOH/
70 °C for 3 min, and then 1% aqueous HCl (1 min), rinsing thoroughly
with deionized water between each step, and then spin drying for 5
min under a stream of nitrogen at 35 °C. The slides were then silanated
for 15 min in a gently agitating 1% solution of N-(3-(triethoxysilyl)-
propyl)-4-hydroxybutyramide (PCR, Inc., Gainesville, FL) in 95:5
ethanol/water, rinsed thoroughly with 2-propanol, then deionized water,
and finally spin-dried for 5 min at 90-110 °C.
Photochemical Oligonucleotide Synthesis. Phosphoramidites were
used at a concentration of 50 mM in dry acetonitrile, and coupling
reactions were performed on an Affymetrix Array Synthesizer, which
consisted of a custom-built automated exposure tool and a 1 in diameter
flowcell linked to a modified Applied Biosystems model 392 DNA
synthesizer. Reagent delivery from the synthesizer was controlled using
OligoNet software, with minor adjustments being made to the standard
programmed coupling protocol to accommodate the particular volume
and mixing requirements of the flowcell. In cycles adapted for
photochemical synthesis, the detritylation step was replaced with a pause
to allow the exposure tool to perform the automated mask positioning/
alignment and timed light exposure. Exposures were made through a
chrome-on-quartz mask in contact with the back of the substrate while
the substrate remained clamped to the flowcell. Light was projected
horizontally from a 500 W collimated light source (model 87330, Oriel
Instruments, Stratford, CT) equipped with an Ushio model ush508sa
super high pressure mercury lamp and dichroic reflectors to provide
Figure 8. Analysis of the photolysis of compound 13 (Scheme 5) in
dioxane by reverse-phase HPLC ([13] ) 200 µm) (a) before photolysis,
(b) after 30 s of exposure to near-UV light from a mercury arc source
(10 mW/cm2 at 365 nm), and (c) after 300 s of exposure. Intermediate
14 and products 11 and 12 were identified by comigration with the
authentic compounds.
A detailed discussion of the impact of synthesis yields on
the hybridization properties of photolithographically generated
oligonucleotide probes is outside the scope of this report.
However, as this question is relevant to the use of probe arrays
for sequencing applications, a few general comments should
be made. The observed photochemical synthesis yields imply
that full-length probes will be accompanied on the support by
significant quantities of incomplete or “truncated” sequences.
Nonetheless, probe arrays prepared in this fashion exhibit
excellent performance in a variety of applications in which
sequence information can be “read” from hybridization data with
single-base resolution.1b-i The thermodynamics of hybridization
are such that conditions (e.g., temperature, ionic strength) can
generally be manipulated to eliminate the hybridization of
oligonucleotides which are substantially shorter than the full-
length probe. Truncated sequences which are shorter by only
a few nucleotides may still hybridize, but these comprise a
relatively small fraction (∼10%) of the total amount of
hybridizable probes on the surface, and consequently contribute
little to the observed hybridization signal. Further improvement
in the chemical efficiency of array synthesis remains an
(14) Sekine, M.; Fujii, M.; Nagai, H.; Hata, T. L. J. Chem. Soc., Chem.
Commun. 1987, 1119-1120.
(15) Hata, T. L.; Gokita, N.; Sakairi, N.; Yamaguchi, K.; Sekine, M.;
Ishido, Y. Bull. Soc. Chem. Jpn. 1982, 55, 2949-2955.
(16) Kamimura, T.; Tsuchiya, M.; Urakami, K.-I.; Koura, K.; Sekine,
M.; Shinozaki, K.; Miura, K.-I.; Hata, T. L. J. Am. Chem. Soc. 1984, 106,
4552-4457.
(13) (a) Pirrung, M. C.; Shuey, S. W. J. Org. Chem. 1994, 59, 3890-
3897. (b) Pirrung, M. C.; Bradley, J.-C. J. Org. Chem. 1995, 60, 1116-
1117. (c) Pirrung, M. C.; Bradley, J.-C. J. Org. Chem. 1995, 60, 6270-
6276.

Light-Directed Synthesis of DNA Arrays on Glass
J. Am. Chem. Soc., Vol. 119, No. 22, 1997 5089
output in the near-UV spectral range (g340 nm). Prior to the exposure
step, the flowcell was usually filled with a solvent (see Table 2),
delivered automatically by the DNA synthesizer from bottle no. 10.
Typically, the final step of a photolysis or synthesis efficiency
experiment (see Results and Discussion) is fluorescence “staining”,
wherein fluorescein phosphoramidite 8 (5 mM, in a solution containing
50 mM DMT-T-CEP in acetonitrile) is coupled to the free hydroxyl
groups on the substrate using the standard coupling protocol. Substrates
were then removed from the flowcell and deprotected in a solution of
1,2-diaminoethane in ethanol (50%/vol) for 2 h at ambient temperature,
rinsed with deionized water, and then dried under a stream of nitrogen.
The pattern and intensity of surface fluorescence was imaged with
a specially constructed scanning laser confocal fluorescence microscope,
which employed excitation with a 488 nm argon ion laser beam focused
to a 2 µm spot size at the substrate surface. Emitted light was collected
through confocal optics with a 530((15) nm bandpass filter and
detected with a photomultiplier equipped with photon-counting elec-
tronics. Output intensity values (photon counts/second) are proportional
to the amount of surface-bound fluorescein, so that relative yields of
free hydroxyl groups within different regions of the substrate could be
determined by direct comparison of the observed surface fluorescence
intensities. All intensity values were corrected for nonspecific back-
ground fluorescence, taken as the surface fluorescence within the
nonilluminated regions of the substrate.
Solution Photolysis Studies. Compounds to be photolyzed in
solution were dissolved in dioxane at a concentration of 200 µm.
Solutions were fully exposed, in a 1 mm path length quartz cuvette, to
near-UV light from a 500 W collimated mercury light source (see
above). Aliquots were removed at intervals for direct analysis by
reversed phase HPLC (20-100% CH3CN over 25 min, A260 detection).
The primary photolysis products were identified by co-injecting
authentic standards. Photolysis was followed by monitoring the
disappearance of starting material, and half-times were calculated from
ln[peak area] vs time plots (t1/2 ) 0.693/slope), using linear regression
analysis.
Methyl 3,4-(Methylenedioxy)-6-nitrophenyl Ketone (2). A solu-
tion of 410 g (2.5 mol) 3,4-(methylenedioxy)acetophenone (1, Aldrich)
in 1600 mL of glacial acetic acid was added dropwise over 45 min to
3400 mL of cold 70% HNO3. The solution was stirred continuously
and maintained at 3-5 °C throughout the addition and for another 60
min afterward. The mixture was allowed to stir at ambient temperature
for another 2 h and then slowly poured into 10 L of crushed ice. The
resulting yellow solid was collected by filtration, washed with water,
dried under vacuum, and recrystallized from THF/hexane to afford 345g
(66%) of 2. 1H NMR (CDCl3, 400 MHz): δ 7.55 (s, 1H), 6.75 (s,
1H), 6.18 (s, 2H), 2.49 (s, 3H). ESI-MS: 210 (M + H+), 232 (M +
Na+). Anal. Calcd for C9H7NO5: C, 51.68; H, 3.37; N, 6.70.
Found: C, 51.56; H, 3.21; N, 6.40. The product was contaminated
with a small amount (3-5% by HPLC) of an inert byproduct, 1,2-
(methylenedioxy)-4-nitrobenzene. 1H NMR (DMSO-d6, 400 MHz): δ
7.93 (d, J ) 10 Hz, 1H), 7.80 (s, 1H), 7.17 (d, J ) 10 Hz, 1H), 6.30
(s, 2H).
(R,S)-1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethanol (3). Sodium
borohydride (55 g, 1.45 mol) was added over 60 min to a cold, stirring
suspension of 3,4-(methylenedioxy)-6-nitroacetophenone (690 g, 3.3
mol) in 3200 mL of methanol, maintaining a temperature of 5-15 °C
with external cooling. Stirring was continued at ambient temperature
until the reaction was complete (2-3 h), at which time the mixture
was combined with 1600 mL of saturated aqueous ammonium chloride.
The resulting suspension was extracted three times with CH2Cl2, and
the combined extracts were washed with brine, dried with MgSO4, and
evaporated to give 680 g (98%) of product after drying under vacuum.
1H NMR (CDCl3, 200 MHz): δ 7.46 (s, 1H), 7.27 (s, 1H), 6.11 (s,
2H), 5.46 (quartet, J ) 6.5 Hz, 1H), 1.54 (d, J ) 6.5 Hz, 3H). ESI-
MS: 234 (M + Na+). Anal. Calcd for C9H9NO5: C, 51.19; H, 4.30;
N, 6.63. Found: C, 51.24; H, 4.25; N, 6.59.
removed under low vacuum with an aqueous NaOH trap, before the
remaining solvents were removed on a rotary evaporator. The residue
was triturated with hexane to obtain a solid brown cake, which was
then recrystallized from THF/hexane to afford 205 g of product as a
light-brown powder which was at least 95% pure according to 1H NMR.
1H NMR (CDCl3, 200 MHz): δ 7.52 (s, 1H), 7.05 (s, 1H), 6.47 (quartet,
J ) 6.5 Hz, 1H), 6.16 (s, 2H), 1.78 (d, J ) 6.5 Hz, 3H). Anal. Calcd
for C10H8NO6Cl: C, 43.90; H, 2.95; N, 5.12; Cl, 12.96. Found: C,
44.01; H, 2.98; N, 5.02; Cl, 12.96. The material was stored dessicated
at -20 °C.
Synthesis of 5′-MeNPOC-2′-deoxynucleosides (6a-k). All but one
(6j) of the 5′-O-MeNPOC-2′-deoxynucleosides were prepared from the
base-protected nucleosides using the following general procedure: The
base-protected nucleoside (90 mmol) was dried by coevaporating three
times with 250 mL of anhydrous pyridine, dissolved or suspended in
300 mL of anhydrous pyridine under argon, and then cooled to -15
°C with an ice/methanol bath. A solution of 27.5 g (100 mmol)
MeNPOC-Cl in 100 mL of dry CH2Cl2 was then added dropwise with
stirring. After 30 min, the ice bath was removed, and the solution
allowed to stir overnight at room temperature. After evaporating the
solvents, the crude material was taken up in EtOAc and extracted with
water and brine. The organic phase was dried over Na2SO4 and
evaporated to obtain a yellow foam. The crude products were generally
purified by flash chromatography on silica gel (0-6% MeOH gradient
in CH2Cl2 or 1:1 CH2Cl2/EtOAc), except 6g, which was recrystallized
from DCM. Products (mixture of diastereomers) were obtained in 65-
85% yield, with purity g96% as determined by reverse-phase HPLC
(15-100% CH3CN over 15 min).
5′-O-[(R,S)-(1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethoxy)carbo-
nyl]-N3-isobutyryl-2′-thymidine (6j). Chlorotrimethylsilane (2.48 g,
2.9 mL, 23 mmol) was added to a solution of MeNPOC-T (6i, 5.5g,
11.5 mmol) in 50 mL of dry DCM containing 2.5 g (3.27 mL, 25 mmol)
of triethylamine (TEA). Stirring was continued at room temperature
until HPLC analysis indicated complete silylation (∼4 h) and then 2
mL each of triethylamine and pyridine were added, followed by 1.5 g
(1.5 mL, 14.3 mmol) of isobutyryl chloride. HPLC analysis indicated
complete acylation after stirring overnight, at which time the mixture
was evaporated and redissolved in 100 mL of THF. The resulting
solution was filtered to remove TEA‚HCl, then 50 mL of water and
25 mL of HOAc were added, and stirring was continued for another
30 min to desilylate the nucleoside. EtOAc and saturated aqueous
NaHCO3 were then added (100 mL of each), and the organic phase
was separated and washed with brine (100 mL), dried with Na2SO4,
and evaporated to dryness. The crude product was purified by flash
chromatography (DCM/EtAc (9:1 to 7:3 step gradient) to obtain 3.5g
of 6j (55%).
Synthesis of 5′-MeNPOC-2′-deoxynucleoside 2-Cyanoethyl 3′-
N,N-Diisopropylphosphoramidites (7a-k). All 5′-MeNPOC-deoxy-
nucleosides were phosphitylated using the procedure of Barone et al.17
and purified by flash chromatography (silica gel, DCM/EtOAc contain-
ing 1% triethylamine). Yields were in the range of 60-90% with purity
g96%, as determined by reverse-phase HPLC (40-100% CH3CN over
15 min).
18-O-[(R,S)-(1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethoxy)car-
bonyl]-3,6,9,12,15,18-hexaoxaoctadec-1-yl O′-2-Cyanoethyl-N,N-Di-
isopropylphosphoramidite (MeNPOC-HEG-CEP, 9). Hexaethyl-
eneglycol (Aldrich, 200 g, 710 mmol) was coevaporated twice with
400 mL of dry pyridine and then dissolved in 500 mL of dry pyridine
and DCM (4:6). MeNPOC-Cl (4) (64.5 g, 236 mmol) in 500 mL of
CH2Cl2 was then added dropwise with stirring over 2 h. After another
hour of stirring, the mixture was transferred to a separatory funnel and
washed three times with 150 mL of water and once with 150 mL of
saturated brine. The organic phase was dried (NaSO4) and evaporated
to give ∼125 g of crude material as an oil, which, according to HPLC
analysis, consists of an 8:2 mixture of mono- and bisacylated products.
The crude material was purified by flash chromatography (0-5%
MeOH in DCM/EtOAc) to give 75 g of 18-[O-(R,S)-(1-(3,4-(methyl-
enedioxy)-6-nitrophenyl)ethoxy)carbonyl]-3,6,9,12,15,18-hexaoxaocta-
decan-1-ol (g98% purity by RP-HPLC). 1H NMR (CDCl3, 400
(R,S)-1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethyl chloroformate
(Menpoc-Cl) (4). A solution of phosgene in toluene (1250 mL of 20%
w/v, 2.4 mol) was added to a stirring suspension of (R,S)-1-(3,4-
(methylenedioxy)-6-nitrophenyl)ethanol (211 g, 1.0 mol) in 500 mL
of dry THF. Stirring was continued at ambient temperature under argon
until the reaction was complete (36-48 h). Excess phosgene was
(17) Barone, A. D.; Tang, A.-Y.; Caruthers, M. H. Nucleic Acids Res.
1984, 12, 4051-4061.

5090 J. Am. Chem. Soc., Vol. 119, No. 22, 1997
McGall et al.
MHz): δ 7.50 (s, 1H, HAr-MeNPOC), 7.09 (s, 1H, HAr-MeNPOC), 6.27
triester was detritylated by dissolving it in 2.5 mL of DCM/MeOH
(6:1) containing 6% trichloroacetic acid. After 5 min, 10 mL of EtOAc
was added, and the solution was washed with saturated aqueous
NaHCO3 and brine (5 mL each). The organic phase was dried (Na2SO4)
and evaporated, and the crude product purified by flash chromatography
(DCM/MeOH, 9:1) to obtain 12 mg (79%) of 12. 1H NMR (CDCl3,
400 MHz): δ 9.72 (br s, 1H, HN2), 8.02 (2s, 1H, HC8), 7.32 (t, 2H,
HAr-pac), 7.05 (t, 1H, HAr-pac), 6.99 (d, 2H, HAr-pac), 6.39 (t, 1H, HC1′),
5.31 (br s, 1H, HC3′), 4.62 (s, 2H, Hpac), 4.38 (br s, 1H, HC4′), 4.34-
(quartet, J ) 6.5 Hz, 1H, HArCH-MeNPOC), 6.12 (br s, 2H, HOCH O-MeNPOC),
2
4.80 (br s, 1H, HOH1), 4.31-4.19 (m, 2H, HC17), 3.73-3.59 (m, 22H,
H
C1-16), 1.64 (d, J ) 8Hz, 3H, HCH -MeNPOC). ESI-MS: 520 (M +
3
H+), 537 (M + H2O+), 542 (M + Na+).
MeNPOC-HEG-OH (80 g, 154 mmol) was coevaporated three times
with 300 mL of dry toluene, and combined with 2-cyanoethyl N,N,N′,N′-
tetraisopropylphosphorodiamidite (55.7 g, 185 mmol) and diisopropyl-
ammonium tetetrazolide (6.4 g, 77 mmol) in 600 mL of dry DCM under
argon at room temperature. The solution was stirred overnight and
then transferred to a separatory funnel and washed twice with saturated
aqueous NaHCO3 and once with saturated brine (400 mL each). After
drying with Na2SO4, the organic phase was evaporated and the crude
product was purified by flash chromatography (0-50% EtOAc in DCM/
0.5% triethylamine) to give 87 g (78.5%) of 9 (purity 98% pure by
HPLC; 100% by 31P NMR). 1H NMR (CDCl3, 400 MHz): δ 7.50 (s,
1H, HAr-MeNPOC), 7.09 (s, 1H, HAr-MeNPOC), 6.27 (quartet, J ) 6.5 Hz,
4.27 (m, 2H, HCE), 3.98-3.93 (m, 1H, HC5′a), 3.84 (d, 3H, HCH O-P),
3
3.84-3.76 (m, 1H, HC5′b), 3.03-2.94 (br m, 1H, HC2′a), 2.81 (br m,
1H, HCE), 2.72-2.64 (m, 1H, HC2′b). 31P NMR (CDCl3, 162 MHz): δ
+0.93. Anal. Calcd for C22H25N6O9P: C, 48.18; H, 4.59; N, 15.32;
P, 5.65. Found: C, 48.65; H, 4.58; N, 14.73; P, 5.40. FAB-MS(HR):
m/z calcd for C22H26N6O9P (M + H+) 549.14989, obsd 549.14916.
1,3-bis-[(R,S)-((1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethoxy)-
carbonyl)oxy]propane (13). A solution of 2.5 g (9.15 mmol) in 3
mL of dry DCM was added slowly with stirring to a cold solution of
1,3-propanediol in 5 mL of dry pyridine under argon. After stirring
overnight at room temperature, 50 mL of EtOAc was added, the solution
was filtered to remove pyr‚HCl, and the solvents were evaporated. The
crude products were purified by flash chromatography (hexane/EtOAc,
3:2) to obtain 1.2 g of 13 (mixture of diastereomers). UV-vis
(dioxane): 245 nm (λmax, 2.5 × 104), 294 nm (sh, 6.5 × 103), 344 nm
(λmax, 1.03 × 104). 1H NMR (DMSO-d6, 400 MHz): δ 7.49 (2s, 2H),
7.06 (s, 2H), 6.27-6.21 (m, 2H), 6.12 (s, 4H), 4.22-4.11 (m, 4H),
1.99 (quintet, 2H), 1.63 (d, 6H). EI-MS: 573 (M+•). Anal. Calcd
for C13H15NO8: C, 49.84; H, 4.83; N, 4.87. Found: C, 49.55; H, 4.82;
N, 4.55.
1H, HArCH-MeNPOC), 6.12 (br s, 2H, HOCH O-MeNPOC), 4.30-4.18 (m, 2H,
2
H
C17), 3.90-3.80 (m, 3H, HC1, CE-a), 3.76-3.58 (m, 23H, HC2-16,CH-iPr,
CE-a), 2.65 (t, J ) 8 Hz, HCE-b), 1.65 (d, J ) 8 Hz, 3H, HCH -MeNPOC),
3
1.21-1.16 (m, 12H, HCH -iPr). 31P NMR (CDCl3, 162 MHz): δ +148.7.
3
ESI-MS: 720 (M + H+), 742 (M + Na+), 821 (M + Et3NH+). Anal.
Calcd for C31H51N3O14P: C, 51.66; H, 7.13; N, 5.83; P, 4.30. Found:
C, 51.70; H, 6.82; N, 5.73; P, 4.04.
5′-O-(R,S)-(1-(3,4-(Methylenedioxy)-6-nitrophenyl)]ethoxy)carbo-
nyl]-N2-(phenoxyacetyl)-2′-deoxyguanosine 3′-O-(2-Cyanoethyl)-
methyl Phosphate (10). Anhydrous methanol (1.73 mL, 39 mmol)
was added to a solution of MeNPOC-(pac)dG-CEP (7c, 3.26 g, 3.9
mmol) in 10 mL of dry CH3CN under argon, and a solution of tetrazole
(0.45 M in CH3CN, 10 mL, 4.5 mmol) was added slowly with stirring.
After 15 min, 50 mL of EtOAc was added, and the solution was washed
with saturated aqueous NaHCO3 and brine (50 mL each). The organic
phase was dried (Na2SO4) and evaporated, and the crude product
purified by flash chromatography (DCM/MeOH, 97:3) to obtain 1.7 g
(57%) of 5′-MeNPOC-N2-(phenoxyacetyl)-2′-deoxyguanosine 3′-O-(2-
cyanoethyl)methyl phosphite (mixture of diastereomers). 1H NMR
(CDCl3, 400 MHz): δ 9.50 (br s, 1H, HN2), 7.86, 7.80 (2s, 1H, HC8),
7.40, 7.36 (2s, 1H, HAr-MeNPOC), 7.34-7.28 (m, 2H, HAr-PAC), 7.05 (t,
1H, HAr-PAC), 7.00-6.94 (m, 2H, HAr-PAC), 6.99, 6.86 (2s, 1H,
3-[(R,S)-((1-(3,4-(Methylenedioxy)-6-nitrophenyl)ethoxy)carbo-
nyl)oxy]propan-1-ol (14) was prepared by the method described for
9. UV-vis (dioxane): 244 nm (λmax, 1.23 × 104), 294 nm (sh, 3.4 ×
103), 345 nm (λmax, 5.2 × 103). 1H NMR (DMSO-d6, 400 MHz): δ
7.50 (s, 1H), 7.07 (s, 2H), 6.26 (quartet, 1H), 6.12 (s, 2H), 4.31-4.19
(octet, 2H), 3.70 (br m, 2H), 1.89 (quintet, 2H), 1.64 (d, 3H). ESI-
MS: 314 (M + H+), 336 (M + Na+). Anal. Calcd for C23H22N2O14:
C, 50.19; H, 4.03; N, 5.09. Found: C, 50.18; H, 3.95; N, 4.99.
Methyl 3,4-(Methylenedioxy)-6-nitrosophenyl Ketone (11).
A
solution of 1.2 g R-methyl-3,4-(methylenedioxy)-6-nitrobenzyl alcohol
(3) in 40 mL of acetonitrile was circulated through a 75 × 75 mm (1
mm path length) quartz flowcell under continuous exposure to near-
UV irradiation (10 mW/cm2 at 365 nm) from a 500 W collimated
mercury arc light source (model 87330, Oriel Instruments, Stratford,
CT) for 4 h. The partially photolyzed mixture was evaporated to
dryness and purified by flash chromatography (DCM). The product
eluted as a green band (TLC, Rf ) 0.35/DCM), which was reduced in
Vacuo to induce crystallization of the product as green-yellow crystals
(250 mg), purity 97% by HPLC (5-80% CH3CN over 25 min). UV-
vis (dioxane): 263 nm (λmax, 1.4 × 104), 328 nm (sh, 4.3 × 103), 382
nm (λmax, 9.5 × 103). 1H NMR (DMSO-d6, 400 MHz): δ 7.49 (s,
1H), 6.35 (s, 2H), 6.82 (s, 1H), 2.79 (s, 3H). 13C NMR (DMSO-d6,
100 MHz): δ 210.8, 161.0, 154.7, 150.4, 143.7, 106.9, 104.2, 89.0,
33.2. EI-MS: 193 (M+•). Anal. Calcd for C9H7NO4: C, 55.96; H,
3.65; N, 7.25. Found: C, 55.73; H, 3.75; N, 7.07.
H
Ar-MeNPOC), 6.32-6.26 (m, 1H, HC1′), 6.24-5.98 (m, 3H, HMeNPOC),
5.10-4.95 (br d, 1H, HC3′), 4.72-4.62 (m, 2H, Hpac), 4.43-4.32 (m,
3H, HC5′,4′), 4.08-4.02 (m, 2H, HCE), 3.60 (d, 3H, HCH O), 2.93-2.82
3
(m, 1H, H2′a), 2.70 (t, 2H, HCE), 2.65-2.47 (m, 1H, HC2′b), 1.62-1.56
(m, 3H, HCH -MeNPOC). 31P NMR (CDCl3, 162 MHz): δ +140.6.
3
The nucleoside 3′-phosphite triester (1.5g, 1.95 mmol) was combined
with 5 mL of 10% aqueous H2O2 in 5 mL of THF. After 5 min, 25
mL of EtOAc was added, and the solution was washed with saturated
aqueous NaHCO3 and brine (25 mL each). The organic phase was
dried (Na2SO4) and evaporated, and the crude product purified by flash
chromatography (DCM/MeOH, 97:3) to obtain 1.2 g (77%) of 10
(mixture of diastereomers). 1H NMR (CDCl3, 400 MHz): δ 9.77, 9.51
(2d, 1H, HN2), 7.85, 7.79 (2s, 1H, HC8), 7.40, 7.36 (2s, 1H, HAr-MeNPOC),
7.35-7.29 (m, 2H, HAr-pac), 7.06 (t, 1H, HAr-pac), 7.02-6.96 (m, 2H,
H
Ar-pac), 6.98, 6.85 (2s, 1H, HAr-MeNPOC), 6.32-6.26 (m, 1H, HC1′),
6.24-5.98 (m, 3H, HMeNPOC), 5.33-5.16 (br d, 1H, HC3′), 4.74-4.58
(m, 2H, Hpac), 4.46-4.26 (mm, 5H, HC5′,4′, HCE), 3.86 (d, 3H, HCH O-P),
3
3.07-2.93 (m, 1H, HC2′a), 2.82 (t, 2H, HCE), 2.80-2.62 (m, 1H, HC2′b),
1.62-1.56 (m, 3H, HMe-MeNPOC). 31P NMR (CDCl3, 162 MHz): δ
+0.15. Anal. Calcd for C32H33N7O15P: C, 48.92; H, 4.11; N, 12.48;
P, 3.94. Found: C, 48.54; H, 4.00; N, 12.06; P, 3.89. FAB-MS(HR):
m/z calcd for C32H33N7O15P (M + H+) 786.17723, obsd 786.17741.
N2-(Phenoxyacetyl)-2′-deoxyguanosine 3′-O-(2-Cyanoethyl)methyl
Phosphate (12). The same method described above for the preparation
of 10 was used to prepare 5′-DMT-N2-pac-2′-deoxyguanosine 3′-O-
(2-cyanoethyl)methyl phosphate from the commercially available
phosphoramidite. 1H NMR (CDCl3, 400 MHz): δ 9.50 (s, 1H, HN2),
7.85 (2s, 1H, H8), 7.42-6.78 (m, 18H, HAr-DMT,pac), 6.30 (t, 1H, H1′),
5.34-5.22 (br m, 1H, H3′), 4.69 (s, 2H, Hpac), 4.29-4.25 (m, 1H, H4′),
4.25-4.15 (m, 2H, HCE), 3.82 (d, 3H, HMeO-P), 3.77 (s, 6H, HMeO-DMT),
3.42-3.32 (m, 2H, H5′), 2.88-2.65 (m, 4H, H2′, HCE). 31P NMR
(CDCl3, 162 MHz): δ +0.90.
Acknowledgment. The authors thank the NIH for financial
support. We also thank the following individuals at Affymetrix
for their generous assistance: Jim Winkler, David Stern, JoAnn
Iwasa, Martin Goldberg, Peter Feikowsky, Rita Wespi, and
Richard Rava.
Supporting Information Available: Synthetic procedures
and analytical data (NMR, MS, elemental) are available for N2-
(4,4′-dimethoxytrityl)-2′-deoxyguanosine, O6-(N,N-diphenylcar-
bamoyl)-N2-(phenoxyacetyl)-2′-deoxyguanosine, all 5′-MeNPOC-
2′-deoxynucleosides (6a-k), and phosphoramidites (7a-k) (12
pages). See any current masthead page for ordering and Internet
access instructions.
A portion (180 mg, 220 µmol) of the DMT nucleoside phosphate
JA964427A