ABSTRACT
We have investigated the use of spacer molecules to reduce steric interference
of the support on the hybridisation behaviour of immobilised oligonucleotides.
These spacers are built up from a variety of monomeric units, using
phosphoramidite chemistry, by condensation onto an amine-functionalised polypropylene support. The optimal spacer length was
determined to be at least 40 atoms in length, giving up to 150-fold increase in the yield of hybridisation. The effects of different
charged groups in the spacer were also examined, and it was shown that both
positively and negatively charged groups in the spacer diminish the yield of
hybridisation. Steric hindrance in hybridisation can also be a problem if the
oligonucleotides attached to the support are too close to each other. Surface
coverage was varied using a combination of cleavable and stable linkers, giving
the highest hybridisation yields for surfaces containing
~
50% of the maximum concentration of oligonucleotides.
The combinatorial approach to synthesising and screening very large numbers of
ligands attached to a solid support has many applications in the life sciences
including new drugs development, enzyme/substrate reactions (
1
,
2
) and nucleic acid sequence analysis (
3
-
6
). Techniques using solid support-bound oligonucleotides as probes are finding a wide range of applications
(
7
). For all these applications, the hybridisation of the oligonucleotides to
target nucleic acids should be as free as possible from interference from the
solid support.
In this study we used combinatorial techniques for simultaneous analysis of
different factors influencing oligonucleotide duplex formation. We have
previously used such methods to study the effects of base composition, length
and structure on the hybridisation properties of oligonucleotides (
8
-
11
). A variety of ways of synthesis of combinatorial oligonucleotide arrays have
been described (
4
-
6
). We used a physical masking method developed in our laboratory (
13
) to direct the synthesis of spacers to specific regions.
Though spacers have been used to mitigate the effects of the solid support (
5
,
7
,
14
), there has been no systematic study of the various factors affecting
hybridisation behaviour. Properties of the spacer which are likely to be
important include length, charge, hydrophobicity and solvation. We made novel
monomers which can be used to build spacers with different chemical
composition. Combinatorial methods were used to make arrays in which a probe
oligonucleotide was attached to the surface by different spacer molecules. The
arrays were hybridised with labelled complementary targets of high and low
molecular weight. In the following we present the results of our study of the
optimal composition and length of spacer in between oligonucleotides and a
polypropylene support and of the concentration of oligonucleotides on the
support.
We found that the spacers have a large effect on hybridisation yield, the most
important property of the spacer being its length.
Array synthesis was carried out as previously described (
13
) adapting phosphoramidite chemistry (
15
). Oligonucleotides used for hybridisation were made in an Applied Biosystems
392 DNA/RNA synthesiser using the same standard chemistry. 3'-DMTr-5'-phosphoramidites of base-protected nucleosides and aminated
polypropylene were a gift from Beckman Instruments. Oligonucleotides and tRNA
Phe
(Sigma) were labelled according to standard methods using radioisotopes
purchased from Amersham International. Mass spectra were recorded on Voyager
TM
-Elite Biospectrometry
TM
Research Station, PerSeptive Biosystems.
1
H-NMR spectra were recorded on a Varian Gemini 200 MHz spectrometer.
PhosphorImages were obtained using Molecular Dynamics PhosphorImager Model
400A. All chemicals were purchased from Aldrich Chemical Company. Silica gel
for column chromatography and solvents were purchased from BDH/Merck.
O
1
-DMTr-O
n
-cyanoethoxydiisopropylaminophosphoramidites of diols (type
1-3
spacers) and the cleavable reagent, (2-cyanoethoxy)-2-(2'-
O
-DMTr-oxyethylsulfonyl)ethoxy-
N
,
N
-diisopropylaminophosphine (
4
), were synthesised as described (
16
-
18
).
Serinol (2.05 g, 22.5 mmol) was dried by evaporation with 10 ml dry pyridine,
dissolved in 120 ml dry pyridine and ice-cooled. FmocCl (5.7 g, 22.5 mmol, dissolved in 5 ml dry dioxane) was
slowly added to this solution during 1 h under stirring. The solution was then
stirred at room temperature for 4 h, ice-cooled and DMTrCl (7 g, 20.1 mmol) was slowly added during 30 min together
with 3 ml
N
,
N
'-diisopropylethylamine. The reaction mixture was left overnight,
then evaporated to an oil, taken up to 100 ml non-acidic chloroform, washed with water (3 * 250 ml) and brine (250 ml), evaporated to an oil, dissolved in 7
ml toluene and purified by flash chromatography on a 6 * 12 cm column using mixtures: methylene chloride-hexane (3:1, 2:1, 1:1, 300 ml of each, all containing 0.5% Et
3
N) as eluents. Two main fractions were collected, evaporated to foam and dried
in vacuo
. The first fraction (
R
f
0.56, CH
2
Cl
2
, 0.5% Et
3
N) appeared to be a ditrytilated derivative of
N
-Fmoc-serinol. Yield 5.0 g (25%).
1
H-NMR: (DMSO-D6):
3.32 (m, 1H), 3.69 (s, 12H, OCH
3
), 4.23 (m, 4H),
5.7 (m, 1H), 6.21 (m, 2H), 6.7-7.6 (m, 30H, arom.), 7.7 (d, 2H), 7.89 (d, 2H).
The second fraction (
R
f
0.22) was determined to be monotritylated
N
-Fmoc-serinol (
5
). Yield: 6.0 g (46%).
1
H-NMR (DMSO-D6): 3.41 (m, 1H), 3.69 (s, 6H, OCH3), 4.2 (m, 2H), 4.63 (m, 2H),
5.71 (m, 1H), 6.26 (m, 2H), 6.7-7.5 (m, 17H, arom.), 7.7 (d, 2H), 7.89 (d, 2H). Both fractions, when
treated with piperidine, showed decrease of
R
f
(cleavage of Fmoc-group) and positive test with fluorescamine (NH2- group). The monotritylated
N
-Fmoc-serinol (2 g, 3.6 mmol) was dried for 24 h
in vacuo
over P
2
O
5
and transferred under argon into 35 ml dry acetonitrile, and 250 mg (3.6 mmol)
dry tetrazole and 1.12 ml (3.6 mmol) bis-
N
,
N
-diisopropylamino-2-cyanoethoxyphosphite were added and the reaction was stirred
under argon at room temperature. After 4 h the reaction mixture was diluted
with 100 ml ethyl acetate, washed with saturated sodium bicarbonate solution (2
* 200 ml) and brine (200 ml), dried over anhydrous Na
2
SO
4
, evaporated, dissolved in 5 ml toluene and rapidly purified by flash-chromatography (dichloromethane:hexane 3:1, then dichloromethane, all with
0.5% Et
3
N). Product-containing fractions were evaporated, dissolved in 5 ml toluene and
precipitated into 500 ml ice-cold pentane. The solvent was decanted, and the residue was dried in vacuo
over P
2
O
5
to give 2.4 g (82%) of a yellowish powder.
R
f
0.83 (CH
2
Cl
2
, 0.5% Et
3
N), Calculated for C
48
H
54
N
3
O
7
P: 815.044. Found: (mass-spectrum, m/z): 836.92 (MI + Na
+
).
N
-Fmoc-glycine pentafluorophenyl ether (6 g, 12.9 mmol) was added to a
stirred solution of serinol (1.17 g, 12.8 mmol) in 100 ml dried pyridine and
left overnight at room temperature. The solution was then ice-cooled, and DMTrCl (4.2 g, 12.5 mmol) was slowly added during 30 min. The
reaction mixture was left overnight, then evaporated to an oil, taken up to 100
ml non-acidic chloroform, washed with water (3 * 250 ml) and brine (250 ml), evaporated to an oil, dissolved in 7
ml toluene and purified by flash chromatography on a 6 * 12 cm column using mixtures: methylene chloride-hexane (3:1, 2:1, 1:1, 300 ml of each, all containing 0.5% Et
3
N) as eluents. Two main fractions were collected, evaporated to foam and dried
in vacuo
. The first fraction (
R
f
0.54, CHCl
3
, 0.5% Et
3
N) appeared to be a ditrytilated derivative of
N
-Fmoc-glycyl-serinol. Yield: 2.2g (46%).
1
H-NMR (CDCl
3
): 3.19 (m, 2H), 3.52 (m, 2H), 3.73 (s, 12H), 3.80 (m, 1H), 4.08 (m, 2H), 4.12
(m, 1H), 4.30 (m, 2H), 5.25 (m, 1H), 5.94 (m, 1H), 6.7-7.9 (m, 34H). Calculated for C
62
H
58
N
2
O
9
: 975.149. Found: (mass-spectrum, m/z): 997.494 (MI + Na
+
), 1012.75 (MI + K
+
).
The second fraction (
R
f
0.08) was determined to be monotritylated
N
-Fmoc-glycyl-serinol (
7
). Yield: 2.08 g (25%).
1
H-NMR (CDCl
3
): 3.34 (m, 2H), 3.78 (s, 6H), 3.81 (m, 1H), 3.87 (m, 2H), 4.12 (m, 2H), 4.2 (m,
1H), 4.39 (d, 2H), 5.41 (m, 1H), 6.45 (m, 1H) 6.7-7.9 (m, 21H). Calculated for C
41
H
40
N
2
O
7
: 672.776. Found: (mass-spectrum, m/z): 696.708 (MI + Na
+
), 713.518 (MI + K
+
).
Both fractions, when treated with piperidine, showed decrease of
R
f
(cleavage of Fmoc-group) and positive test with fluorescamine (NH
2
- group). The monotritylated
N
-Fmoc-glycyl-serinol (2 g, 3 mmol) was phosphitylated and dried as
described above to give 2.03 g (78%) of a yellowish powder.
R
f
: 0.49 (CHCl
3
, 0.5% Et
3
N). Calculated for C
50
H
57
N
4
O
8
P: 872.995. Found: (mass-spectrum, m/z): 896.197 (MI + Na
+
).
[alpha],[alpha]-bis-(4-methoxyphenyl)[ring-U-
14
C]benzyl alcohol (Amersham, 777 MBq/mmol, 21 mCi/mmol, 150 mg) was mixed with
4.8 g DMTrCl and dissolved in 40 ml toluene. Acetyl chloride (5 ml) was added
and the mixture was boiled with stirring in a 100 ml round-bottom flask fitted with vertical condenser with drying tube on the oil
bath for 3 h. After distilling off about half the solvent, the rest was cooled
to room temperature and poured into 600 ml cold (-20oC) hexane to precipitate the DMTr*Cl which was collected by
filtration and dried
in vacuo
to give 4.4 g (87%) of slightly pink solid. This
14
C-labelled DMTrCl was used to synthesise the
14
C-labelled T phosphoramidite as described (
19
).
The following oligonucleotides were used for hybridisation to arrays: 1, 5'-GGT.GCG.AAT.TCT; 2, 5'-GAC.CTC.CAG.ATT. In all cases, `reversed' means 5'-immobilised oligonucleotide. Unless
otherwise specified, oligonucleotides are covalently attached to the solid
support through their 3'-phosphate.
Phosphoramidite reagents for oligonucleotide synthesis are bulky enough to
inhibit coupling to reactive groups on the support. The aminated polypropylene
support contains ~0.3 nmol/cm
2
of NH
2
groups (
20
), which, assuming the surface is flat, gives a spacing between adjacent amino
groups of ~8 Å. This distance is smaller than the size of tritylated nucleoside
phosphoramidites (20-25 Å) or the diameter of the DNA helix (~18-20 Å); clearly these groups cannot be accommodated
on the surface at the full density of the amino groups and the density of
coupled groups will be determined by the dimensions of the monomers used to
synthesise oligonucleotides. We have carried out a detailed physical study of
the surface before and after coupling the spacer, and of the reaction of the
spacer with phosphoramidites (
21
). An area of a sheet of aminated polypropylene was derivatised by reaction with
a mixture of all four phosphoramidites (A, C, G and T) under standard
conditions, but with a slightly longer period (2 min) of condensation step.
Without removing the DMTr protecting group, a second, radiolabelled
phosphoramidite (
14
C-DMTrT phosphoramidite) was applied to an area which included that which
had been coupled previously, and an adjacent area of untreated aminated
polypropylene. This reaction was carried out over an extended period of 1 h.
The
14
C-DMTr group was not removed. The polypropylene was first exposed to a
storage Phosphor screen and then treated with ammonia under conditions normally
employed in oligonucleotide synthesis. Radioactivity was measured using a
PhosphorImager. The coupling of the radioactive monomer was inhibited by only
15-25% by the previous coupling of non-radioactive monomers. This surprising result suggests that coupling
may be a much slower process than has been previously assumed. It is possible
that the first couplings are rapid, but that subsequent couplings are inhibited
by steric crowding of the surface. Comparing the strips of this polypropylene
support before and after ammonia treatment, it was also noted that the area
which contained only
14
C-DMTrT lost ~20% of the
14
C during ammonolysis, indicating that the bond formed after phosphitamide
condensation onto the surface of the aminated polypropylene support is not
completely stable or that there is something other than alkylamine groups on
the support which can act as functional groups for phosphoramidite coupling to
give a labile linkage.
The more an immobilised molecule is spatially removed from the solid support the
closer it is to the solution state and the more likely it is to react freely
with dissolved molecules. This is especially important in the case of
hydrophobic supports such as polypropylene (
20
), which are still rather solvent repellant despite amination. To assess the
influence of hydrophilic spacers of different length on the hybridisation
properties of immobilised oligonucleotides, we used appropriately protected
phosphitamide derivatives of aliphatic diols (obtained by dimethoxytritylation
of one hydroxy group of diols of different length and subsequent
phosphorylation of the other). To examine the effect of the length of the
uncharged part of the spacer attached to two phosphate groups on the properties
of the whole spacer, synthons of different length were synthesised. We did not
use longer (OCH
2
CH
2
)
n
O- units, because it has been shown (
23
) that the yield of phosphoramidite condensation with long (
n
> 5) PEG phosphoramidite synthons does not exceed 60%. This would lead to an
uncontrolled decrease of oligonucleotide yield with increasing length of
spacer. Therefore we derivatised only propanediol, di- and triethyleneglycols (
1-3
). These synthons allowed us to produce three different types of spacers. We
used previously described techniques (
13
) to synthesise arrays of spacers and oligonucleotides directly on the aminated
polypropylene support (Scheme 2).
Oligonucleotides were synthesised in both 3' to 5' and 5' to 3' directions using appropriate monomers
10
and
11
(Scheme 3). Synthon
11
allows one to synthesise oligonucleotides with free 3'-ends, which is important for some array applications where
enzymatic elongation is employed. Arrays synthesised only for hybridisation
experiments can be built up using conventional synthons
9
, which are slightly more effective in the oligonucleotide synthesis due to the
higher reactivity of the primary hydroxyl group. Sequences used for
hybridisation experiments were complementary to selected regions of tRNA
Phe
; for use as a hybridisation target, this tRNA was labelled at the 3'-end by ligation of
32
P-5',3'-ribocytidine diphosphate according to (
24
). In all cases hybridisation of arrays to labelled oligonucleotides gave
greater discrimination than hybridisation to tRNA, but the relative intensities
were similar, so for the greater part of the experiments we hybridised arrays
to oligonucleotides.
Unless otherwise specified, hybridisations were carried out at 36oC to achieve the maximum differentiation of spacer influence by minimizing
non-specific interactions. For both 3'- and 5'- immobilised oligonucleotides, a steady increase
of hybridisation yield was observed with increasing length of spacer up to 8-10 units of glycol synthons
1-3
(Fig.
2
, hybridisation to tRNA and oligonucleotides) in a preliminary experiment.
There is growing interest in the use of solid state devices for tests which
involve molecular interactions. Attachment of one of the ligands to a solid
phase introduces problems which are not encountered when the interactions take
place in a homogeneous solution as it constrains the ways in which the ligands
can interact with each other. The bound ligand is not so free to diffuse as it
would be in solution; this must reduce the reaction rate. The dissolved ligand
may be prevented from making a close approach to the bound ligand by a number
of steric factors: clearly, it cannot make an approach from the direction of
the solid support unless there is a spacer between it and the bound ligand; it
may also be hindered in its approach from the solution phase if the ligands on
the solid support are too close together. Close approach will also be affected
by the nature of the surface to which the bound ligands are attached; its
charge, hydrophobicity and degree of solvation are all likely to influence
significantly the environment in which the interaction of the ligands takes
place.
We devised new spacer chemistries and procedures for making arrays of spacers
which allowed us to study these factors systematically, and applied the method
to the behaviour of oligonucleotides in molecular hybridisation. We found that
the surface charge density had a small effect on duplex yield when varied
between 6 and 11 atoms per charge. We found that, as expected, the surface
density of the oligonucleotides was determined by the bulk of the units used
for oligonucleotide synthesis; the amino groups on the surface were only ~20% occupied by oligonucleotides. However, this packing is close enough to
inhibit duplex formation with target oligonucleotides by a factor of ~50%. Increasing the length of the spacer between the oligonucleotides and
the solid support had by far the largest effect on duplex yield, with a maximum
of 150-fold increase. Surprisingly, increasing the length of the spacer beyond 10-12 units diminished duplex yield until at ~25-30 units it equalled that found with no spacer.
These results suggest that the main effects of tethering a ligand to the solid
support are the prevention of diffusion of the bound ligand, and steric
constraints on the approach of the dissolved ligand. The combinatorial approach
using arrays of spacer molecules described in this paper is one which could be
adapted to other systems using solid state devices to study the interaction of
ligands or to the efficiency of chemical coupling to solid supports.
This work was supported by Beckman Instruments and the Medical Research Council.
*To whom correspondence should be addressed. Tel: +44 1865 275263; Fax: +44 1865
275283; Email: misha@bioch.ox.ac.uk
REFERENCES
Return