Comparison of Coupling and Deprotection
Protocols for RNA Synthesis
ABRF 2004, Integrating Technologies in Proteomics &
Genomics, Portland, Oregon, Feb. 28 – Mar. 2, 2004.
Richard T. Pon and Shuyuan Yu, UCDNA Services, Dept.
Biochemistry & Molecular Biology, University of Calgary.
e-mail: [email protected], tel.: 403-220-4225
1. Abstract
We have investigated different protocols in order to implement a
robust, efficient, and economical RNA synthesis service suitable
for a core facility environment. We used capillary gel electrophoresis (CGE) as a quantitative tool to systematically compare
results from different protocols under controlled conditions.
2. Introduction
RNA differs from DNA synthesis by:
1. 2’-protecting groups slow or reduce coupling reactions.
2. 2’-protecting groups may slow cleavage from the supports.
3. 2’-protecting groups reduce aqueous solubility.
4. Hydrolysis of 2’-protecting groups causes chain cleavage.
5. Incomplete 2’-O-deprotection reduces overall product yield
6. Secondary structures hinder deprotection and separations.
7. Single-stranded RNA is sensitive to enzymatic degradation.
We have been investigating how RNA chemical synthesis can be
optimized, especially with regards to the first five points above.
3. Methods
DMTO
B
• RNA phosphoramidites with 2’-O-tO
CH
butyldimethylsilyl protecting groups
O
O Si
(tBDMSi).
P
CH
CEO
N(iPr)
• 20-mer oligonucleotides of rU19T, rA19T, and
rC19T synthesized on Q-linker supports.
• Average coupling efficiencies (AY) calculated from amount
of FL product determined by CGE, not trityl analysis.
• Chain cleavage during base deprotection determined by
simulated base deprotection of rU19T sequences.
• Rates of cleavage from supports determined by quantitative
colorimetric determination of residual DMT groups.
• CGE analysis on Agilent system with PVA coated capillaries
(see Agilent bulletin #5988-4303EN for protocol).
3
2
3
tBu
4. Optimization of base deprotection
Base deprotection of RNA is a critical step because it is well
known that accidental loss of 2’-O-protecting groups will cause
chain cleavage and significantly reduce yield (1).
NH4OH, RT, 48 h
71.2% FL
10
15
3:1 NH4OH/EtOH
RT, 48 h
75.6% FL
20
NH4OH, 55o, 16 h
9% FL
10
15
10
15
20
3:1 NH4OH/EtOH
55o, 16 h
44.5% FL
20
10
15
20
Figure 1 shows the affect of various NH4OH deprotection
protocols on rU19T. Heating in NH4OH (55o, 16 h) causes severe
degradation, even in the presence of EtOH. Room temperature
deprotection with 3:1 NH4OH/EtOH (48 h) is the preferred method
of reducing chain degradation. However, the long deprotection
time may be unsatisfactory and so “fast” deprotection protocols
with methylamine were also evaluated.
4. Optimization of base deprotection continued
1:1 MeNH2/EtOH :
DMSO, 65o, 0.5 h
78.7% FL
10
15
20
Figure 2 shows a “fast” base deprotection protocol using
methylamine solutions for rU19T. Other base deprotection
protocols are summarized in Table 2 (below).
Reagent
NH3/MeOH
NH4OH:EtOH 3:1
MeNH2/EtOH:DMSO 1:1
MeNH2:NH4OH 1:1
NH3/EtOH
40% MeNH2/H2O
NH4OH
33% MeNH2/EtOH
Condition
%FL
Condition
% FL
48h @ R.T.
48h @ R.T.
79%
75%
16h @ 55
16h @ 55o
30m @ 65o
77%
45%
79%
1.5h @ R.T.
48h @ R.T.
1.5h @ R.T.
48h @ R.T.
1.5h @ R.T.
77%
76%
73%
66%
62%
16h @ 55o
10m @ 65o
16h @ 55o
15m @ 65o
74%
73%
9%
65%
o
Fast base deprotection with methylamine reagents is
advantageous because it reduces exposure of the RNA to
elevated temperatures and aqueous bases, conditions which both
contribute to chain degradation. However, N-acetyl protection on
cytidine is required.
5. Comparison of phosphoramidite activators
Dicyanoimidazole (DCI), tetrazole (TET), ethylthiotetrazole (ETT)
and benzylthiotetrazole (BMT) were used to synthesize rU19T
under otherwise identical synthesis and deprotection conditions.
DCI
(0.5 M)
20
FL = 68.1%
AY = 98.0%
Tetrazole
(0.45 M)
20
FL = 72.3%
AY = 98.3%
BMT
(0.25 M)
20
FL = 75.6%
AY = 98.5%
ETT
(0.25 M)
20
FL = 81.0%
AY = 98.9%
Figure 3 shows that the amount of full-length product was:
DCI < TET < BMT < ETT
Subsequent syntheses have confirmed that the yield of product is
only marginally higher when using ETT over BMT. Although the
difference in yield is too small to be significant, ETT is the
preferred activator because of its much lower cost.
Yields of full-length product were also compared using different
concentrations of activator. We found that the same concentration
of activator normally used for DNA synthesis works fine for RNA
synthesis. Contrary to some promotions, increasing the
concentration of activator for RNA synthesis does not increase
coupling performance (just the cost).
5. Comparison of Phosphoramidite Activators continued
ETT (0.25 M)
20
FL = 75.2%
AY = 98.5%
N+ = 2.1%
BMT (0.25 M)
20
FL = 72.8%
AY = 98.3%
N+ = n.d.
rAGUACAUUUGCCUAGAGCGC
(20-mer)
ETT (0.25 M)
25
FL = 62.4%
AY = 98.4%
N+ = 10.6%
BMT (0.25 M)
25
FL = 60.9%
AY = 98.3%
N+ = 10.4%
rAGUACAUUUGCCUAGAGCGCAGUACAUUUG
(30-mer)
Deprotection: 1. MeNH2/DMSO/EtOH, 65o, 0.5h; 2. NMP/TEA/TEA-3HF, 3h, 65o
Figure 4 further compares the ETT and BMT activators. Two
mixed based 20- and 30-mer RNA sequences were prepared. In
each case ETT produced more full-length product, but the
difference was very small. More significantly, the crude 30-mer
showed a large amount (~10%) of slower migrating impurities.
These are believed to be molecules which were not completely
desilylated. The residual tBDMSi protecting groups create multiple
impurities which appear as N+mer like peaks.
Efficient 2’-deprotection rather than coupling efficiency appears to
be the most limiting step in obtaining high yields of full-length
product when using 2’-O-tBDMSi protected phosphoramidites.
6. Removal of the 2’-O-protecting group
We have found that the 2’- deprotection step is the most critical
step in obtaining high quality RNA. Our CGE analyses have
shown that incomplete desilylation leads to the appearance of
what look like “N+mers” in the final product mixture.
• Amount of N+mers is usually small (1-5%).
• Partially silylated “N+” impurities may be confused with “N+”
impurities caused by double couplings.
• Extended desilylation decreases “N+” impurities with
unwanted 2’-silyl protecting groups.
• However, desilylation may be sequence dependent.
0.5 h, 65o
20
FL = 45.3%
N+mers = 42.3%
1.5 h , 65o
20
FL = 80.7%
N+mers = 4.1%
3 h , 65o
20
FL = 82.5%
N+mers = 2.5%
24 h , 65o
20
FL = 83.1%
N+mers = 1.4%
Figure 5 shows how 2’- partially silylated impurities (“N+” mers) of
rU19T decrease over time. NMP/TEA/TEA-3HF reagent (2).
2’-Desilylation proceeded cleanly for both rU19T and rA19T
oligonucleotides with only small amounts (< 5%) of N+mer like
impurities. However, for best results, we recommend extending
the desilylation time from 1.5 h to 3 h.
6. Removal of the 2’-O-protecting group continued
NMP/TEA/TEA3HF, 65o, 1.5 h
25
FL = 50.1%
N+mers = 27.3%
NMP/TEA/TEA
-3HF, 65o, 24 h
25
FL = 69.5%
N+mers = 7.7%
TEA-3HF
RT, 24 h
20
FL = 56.7%
N+mers = 19.4%
1M TBAF/ THF
RT, 24 h
20
FL = 65.2%
N+mers = 11.0%
Figure 6 shows a previously undocumented problem with
desilylation of an rC19T test sequence. Partially silylated impurities
remained, even after extended treatment with TEA-3HF or TBAF.
This problem is limited to long runs of rC (≥ 16 bases) since
rC4rU15T, rC8U11T, and rC12U7T were cleanly deprotected, but
rC16U3T was not.
We believe the problem with oligo-rC desilylation is due to some
solution structure, which hinders the 2’-deprotection. This is supported by literature precedents (3) which show that rC is more
slowly deprotected. Also, desilylation with TEA-3HF in other solvents, i.e. MeNH2 (4) or DMSO give clean deprotection. It is also
possible to get a good result with NMP/TEA/TEA-3HF (3h, 65o) by
working up the crude material, and then redissolving it and subjecting it to a second treatment with NMP/TEA/TEA-3HF (3h, 65o).
7. Rate of cleavage from the support.
The time required for succinyl linker cleavage is not well
documented when methylamine reagents are used (5). Therefore,
we verified the rate of cleavage as shown in Figure 6.
Figure 6. Rate of succinyl linker cleavage using
methylamine reagents for RNA nucleosides
100%
90%
% Cleavage
80%
70%
60%
50%
40%
30%
20%
10%
0%
0
5
A
B
10
C
15
D
20
E
25
30
Time (min)
Time required for 90% cleavage from a succinic acid linker:
A
B
C
D
E
40% aqueous methylamine
1:1 NH4OH + 40% aqueous MeNH2
33% ethanolic methylamine
1:1 NH4OH + 33% ethanolic MeNH2
1:1 DMSO + 33% ethanolic MeNH2
5 min
13 min
20 min
25 min
53 min
Generally, cleavage with ethanolic methylamine was noticeably
slower than aqueous methylamine and extended treatment (~ 1h)
was required. Thus, RNA requires longer cleavage than DNA.
Alternatively, the more labile Q-Linker arm (6) can be used to
reduce the cleavage time to only 2-5 min.
8. Recommendations for RNA synthesis
Difficulty
slower and less
efficient coupling
slower cleavage
from supports
•
•
•
•
•
Chain cleavage
during base
deprotection
reduced
solubility
2’-deprotection
•
•
•
•
Solutions
avoid tetrazole
use ETT or BMT
1 h+ cleavage with succinic
acid supports
5 min cleavage with QLinker supports
cleave at same time as
base deprotection
avoid prolonged heating
avoid heating in aqueous
NH4OH
avoid water with MeNH2
use DMSO
• use longer desilylation (3
not 1.5 h)
• try different reagents for
problem sequences
Preferred protocol
• 0.25 M ETT
solutions
• use supports
containing the more
rapidly cleaved QLinker arm
• 3:1 NH4OH/EtOH
for 48 h at RT
• 30 min @ 65o fast
deprotection
• 1:1 DMSO + 33%
ethanolic MeNH2
• NMP/TEA/TEA3HF (3h @ 65o)
• 4:1 DMSO/TEA3HF (3 h @ 65o)
9. Conclusions
• Good coupling efficiencies (98-99%) can be obtained, even
with t-butyldimethylsilyl protecting groups
• Coupling efficiency is not the limiting factor in RNA
synthesis. Efficient deprotection is the critical process.
• Chain cleavage during base deprotection is easily minimized
by selection of optimum conditions.
• 2’-O-deprotection of tBDMSi groups works well for most
RNA sequences. However, better methods and
documentation of efficiency are required for long RNA’s.
10. References
1.
2.
3.
4.
5.
6.
T. Wu, K.K. Ogilvie and R.T. Pon, 1989, Nucl. Acids Res. 17,
3501.
F. Wincott, A. DiRenzo, C. Schaffer, S. Grimm, D. Tracz, C.
Workman, D. Gonzalez, S. Scaringe and N. Usman, 1995,
Nucl. Acids Res. 23, 2677.
a. R. I. Hogrefe, A. P. McCaffrey, L.U. Borozdina, E.S.
McCampbell and M.M. Vaghefi, 1993, Nucl. Acids Res. 21,
4739. b. R. Vinayak, A. Andrus and A. Hampel, 1995,
Biomed. Pept. Prot. Nucl. Acids 1, 227.
L. Bellon in Current Protocols in Nucleic Acid Chemistry, Unit
3.6, Eds. S.L. Beaucage, D.E. Bergstrom, G.D. Glick and
R.A. Jones, John Wiley & Sons, 2000.
a. M.P. Reddy, N.B. Hanna and F. Farooqui, 1994, Tet. Lett.
35, 4311. b. M.P. Reddy, F. Farooqui and N.B. Hanna, 1995,
Tet. Lett. 36, 8929. c. M.P. Reddy, N.B. Hanna and F.
Farooqui, 1997, Nucleos. Nucleot. 16, 1589-1598.
R.T. Pon and S. Yu, 1997, Nucl. Acids Res. 25, 3629.
Acknowledgements
We would like to thank Heidi Huettler, Maria Loskot, and Bernd
Kalisch for technical assistance. We also thank Yogesh Sanghvi
(formerly Isis Pharmaceuticals, now Rasayan Inc.) and John
Ackroyd (Transgenomic) for donation of reagents.