ABSTRACT
We investigated the effects of internal primer-template mismatches on the efficiency of reverse transcription and PCR
amplification. As models, RNA transcripts representative of different HIV-1 group M subtypes were evaluated with a previously described
gag
primer pair system. We observed that the presence of two to four mismatches in
the primer-template duplexes did not have a significant effect on RT-PCR. However, the presence of five and six mismatches with the 28
and 30 base primers reduced PCR product yield by
~22- and 100-fold respectively, relative to the homologous template. The amount
of reduction was reproducible from experiment to experiment and was independent
of the initial copy number input. Under the conditions used, viral RNA
measurements of the more divergent HIV-1 subtypes (A and E) would be underestimated, while isolates of subtypes
B, C, D and F-H are expected to be efficiently amplified and accurately measured. The
reduced amplification efficiency for targets similar to HIV subtypes A and E
can be improved 4- to 10-fold by lowering the annealing temperature and implementing a
reverse transcription step that gradually increases in temperature. The
additional substitution of either 5-methylcytosine for cytosine throughout or the substitution of inosine at
positions of variable bases resulted in a <4-fold difference in product yield between the homologous and most
divergent templates.
Nucleic acid-based assays such as the polymerase chain reaction (PCR), the ligase chain
reaction (LCR), nucleic acid sequence-based amplification (NASBA) and branched chain DNA (bDNA) rely on the
efficient hybridization of oligonucleotides to the targeted sequence.
Mismatches between the oligonucleotides and the targeted nucleic acid can
affect duplex stability and may compromise the ability of a system to amplify
and/or detect the targeted sequence. The effects of mismatches depend on numerous factors, including the length of the oligonucleotide, the nature and position
of the mismatches, the temperature of hybridization, the presence of co-solvents and the concentrations of oligonucleotides as well as monovalent and divalent cations (
1
).
The sequence heterogeneity of human immunodeficiency virus type 1 (HIV-1) challenges efficient detection with nucleic acid-based assays. HIV-1 is divided into groups M and O (
2
-
5
). There are presently eight known subtypes within Group M, designated A-H. Subtype A is predominant in Central Africa, B in North America,
Europe, South America and Thailand, C in South Africa and India, D in Central
Africa, E in Thailand, India and Central Africa and subtypes F-H have been reported in Central Africa. Group O, not yet divided into
subtypes, is considered to be more divergent than Group M and to date has been
found in a few infected individuals from Cameroon, Gabon and France (
4
,
5
). In the US, only six non-B subtypes have been reported so far; one subtype D isolate from a Zairian
student (
6
) and three subtype E, one subtype D and one subtype A from five US servicemen (
7
).
Using HIV-1 as a model system, we previously reported on the effects of single 3'-terminal mismatches on PCR product yield (
8
). We found that primers that terminated in a T allowed significant amplification even when mismatched with C, G or T. In this study we
systematically examined the effects of multiple internal primer-template mismatches on RT-PCR using templates that represent various HIV-1 subtypes. The amount of product generated with each
template was determined with a quantitative RT-PCR assay (
9
).
HIV-1 primers SK462, d(AGTTGGAGGACATCAAGCAGCCATGCAAAT), and SK431,
d(TGCTATGTCAGTTCCCCTTGGTTCTCT), and probe SK102, d(GAGACCATCAATGAG- GAAGCTGCAGAATGGGAT), were used in this study. The primers were specifically
designed to be longer than usual in order to better accommodate mismatches.
These primers amplify a 142 bp HIV-1
gag
region which is highly conserved among the subtype B isolates. However, in non-subtype B isolates, as many as six mutations have been observed in the
upstream primer binding region of subtype A isolates and five in the downstream
primer binding region of subtype F. To systematically evaluate the effect of
mismatches on RT-PCR, a series of templates were constructed that harbored mutations
representative of different subtypes (Fig.
1
). The templates (designated M1-M8) were engineered to contain mismatches to either the SK462 (M1-M4) or SK431 (M5-M8) primer binding regions; the internal sequences, including the probe binding region were identical in all
constructs. A template that is completely homologous to both SK462 and SK431 (M0) served as
the control. The templates were further engineered to have common sequences
flanking the 5'- and 3'-termini. The sequences at the 5'- and 3'-termini encode the T7 promoter
transcription signal and a poly(A) tail respectively. The common sequences served two purposes. First, RNA transcripts were
generated directly from the amplified products, which eliminated the need to
subclone the products into a transcription vector. Second, primers to the
common sequences were used to amplify and normalize the templates for
subsequent analyses. Different derivatives of SK462 and SK431 were also
examined. These derivatives included different base substitutions that either
reduced the number of mismatches (as in SK145) or contained different modified
bases, such as inosine, 5-methylcytosine, propynyl dU or propynyl dC (
15
), to increase primer-template stability.
The mutant and control templates were generated by amplification of an HIV-1-containing recombinant plasmid, pSYC1857 (
10
), with mutagenic primers. Two sets of primers were used to construct the
templates (see Fig.
2
). To generate templates with various mutations in the SK462 region, common
primer D was coupled with mutagenic C primers for the first round of
amplification. The second round of amplification was performed with common
primers A and B. To generate templates with mutations in the SK431 region,
mutagenic D primers were coupled with common primer C for the first round of amplification and common primers A and B for the second round of amplification. Primer A
overlaps with primer C by 18 bases; primer B overlaps primer D by 17 bases.
Mutations in the SK462 and SK431 regions were selected based on known mutations
found in different HIV-1 subtypes (
4
). A control template that is completely homologous to both primers was similarly constructed.
RNA transcripts were prepared directly from the amplified products using a MEGAscripttm transcription kit (Ambion, Austin, TX). Residual DNA was removed from the RNA transcripts by treating with
DNase. Proteins were removed by extraction with phenol:chloroform and the RNA
was precipitated with ethanol. After resuspension, the transcripts were
purified over an Oligotex-dTtm column (Qiagen) as recommended by the manufacturer.
The RNA transcripts were normalized to ensure that the same copy number of each
test template was used in the evaluation of primer-template mismatches. To normalize the input, the templates were amplified with primers that were completely homologous to all eight
templates. Following amplification, the products were quantified on microwell
plates coated with SK102 probe as previously described (
9
). Either 10
4
or 10
5
copies of each template were subsequently used in the mismatch study.
To test the effect of mismatches on RT-PCR, the standard cycling conditions recommended for this primer pair were first used: 2 min at 50oC (for uracil N-glycosylase cleavage of any potential carryover of dUMP-containing PCR product from previous reactions) (
11
); 60oC for 30 min (for reverse transcription); four cycles consisting of a
denaturation step (95oC) for 10 s, annealing (50 or 55oC) for 10 s and extension (72oC) for 10 s; followed by 24 cycles of 90, 60 and 72oC for 10 s each. Amplified products were serially diluted
and quantified on microwell plates coated with bovine serum albumin-conjugated probe SK102 as previously described (
9
). Modifications to the standard cycling conditions were subsequently tested.
These included lowering the annealing temperature to 50oC and introducing a gradual `ramp-up' RT step (50-55oC for 5 min, 55-60oC for 15 min, followed by 60oC for 10 min).
To ensure that the synthetic templates harbored the expected mutations, each
template was purified by gel electrophoresis, cloned into a plasmid vector and
the sequence was verified using a Taq DyeDeoxytm Terminator Cycle Sequencing Kit (Applied Biosystems).
Under the standard RT-PCR conditions, the presence of up to four mismatches in either primer had
little or no effect on PCR product yield (Table
1
). The presence of five mutations in SK462 reduced PCR product yield by 22-fold relative to the homologous template; six mutations by 80- to 100-fold. In SK431, the largest difference (13-fold reduction) was observed with the template that had
five mutations. Mismatch tolerance was not affected by the initial copy number
input; the differences observed were identical regardless of whether 10
3
or 10
6
copies of the template were amplified (data not shown). Furthermore, the differences were consistent from run to run.
Table 1
Since the presence of a large number of mutations in either primer affects
amplification, two different approaches were taken to alleviate the reduced
efficiency. First, the annealing temperature during PCR was reduced from 55 to
50oC to decrease the hybridization stringency. Second, to enhance
hybridization, the temperature of the reverse transcription reaction was slowly
raised from 50 to 60oC, as described in Materials and Methods. Lowering the annealing
temperature to 50oC for the entire 28 cycles allowed for greater tolerance of the mismatches, particularly in those templates that contained mutations in the SK462 binding region.
However, the lower annealing temperature by itself was insufficient, as a 9-fold (versus 13-fold) reduction in yield was observed with the five mutations to
SK431 (data not shown). When both a `ramp-up' reverse transcription and a 50oC annealing temperature were applied to all templates, the effect of mutations was substantially reduced in all of the templates (Table
1
). The largest difference was a 10-fold reduction in PCR product with the templates that contained six
mutations in the SK462 region, while a 4-fold reduction was seen with the template that contained five mutations in
the SK431 binding region.
Given the adverse effects of multiple mismatches, we further investigated whether modified bases would allow greater mismatch tolerance. Base analogs such as 5-methylcytosine (5-MeC) have been shown to influence not only the binding equilibrium
but also increase the dissociation temperature (
T
d
) of oligonucleotides by changing the base stacking pattern that permits
intimate contact between the halogen atom and the adjacent base (
12
,
13
). Modified versions of SK462 and SK431 were synthesized containing 5-MeC in place of cytosine (Fig.
3
). By using a 50oC annealing temperature, mismatch tolerance was significantly improved.
Less than a 5-fold reduction in PCR product yield was observed in templates with the
most mismatches (data not shown). With the exception of one position in SK462,
the 5-MeC substitutions were not at positions of mismatches.
To improve amplification of mismatched templates, numerous changes can be
implemented. In this study we examined the effect of different annealing
temperatures and various modified bases in the primers on mismatch tolerance. A
comparison of the results from 50 and 55oC annealing indicates that mismatches can be better accommodated when a
lower annealing temperature is used. Modifications in the reverse transcription step also improved mismatch tolerance in the RT primer. The gradual temperature increase
presumably facilitates the priming and extension of mismatched templates. The
substitution of modified bases such as 5-meC further stabilized binding of the primer to the template. We have also
examined the effect of mismatches on probe hybridization and found that five
mismatches (presently found only in some group O isolates) had no effect on
detection under the hybridization conditions used for the microwell plate
system (data not shown).
Quantitative assays have been extensively used in monitoring the effect of anti-retroviral drugs and for use as a prognostic indicator (
21
-
24
). The clinical efficacy of these drugs was established based upon the ability
of the drugs to significantly reduce viral load as determined by quantitative
RNA assays. Given the heterogeneity of the HIV genome, we embarked on this study to determine the extent to which primer-template mismatches can affect amplification efficiency. The templates used in this study
were intentionally constructed to represent worst case scenarios. For example,
mutant templates of SK462 represent the most divergent of HIV isolates, the A
subtypes. The downstream reverse transcribing primer binding region, SK431, is more conserved, with a
maximum of five mismatches to a subtype F isolate.
We infer from the results of this study that isolates of subtype B, which are
predominant in the US and Europe, will be efficiently amplified and detected
given that they harbor at most three mismatches with SK462 and two mismatches
with SK431. Similarly, isolates of subtypes C and D are expected to be
efficiently amplified. On the other hand, the viral RNA titers of some subtype
A isolates will be underestimated due to the presence of as many as six
mismatches to the SK462 region. Subtype E isolates, although distinct from
subtype A over the
env
region, are indistinguishable from subtype A in the
gag
gene (
3
). Consequently, amplification of some subtype E isolates will also be
underestimated. The number of available sequences for subtypes F-H are limited, but based on the sequences available to date, are expected
to be amplified efficiently with SK462-431. It is important to note that although the absolute copy number
determinations for subtypes A and E may be compromised, the reduction in amplification efficiency was consistent from experiment to experiment, suggesting that the assay can still be used to monitor viral load
in an individual over time. These studies indicate that replacing SK462 with
SK145 would significantly improve quantification of the more divergent isolates. Furthermore, the incorporation of modified bases into the primers would further enhance
mismatch tolerance.
This study extends earlier work on the effects of primer-template mismatches on PCR amplification. We have demonstrated that mismatch
tolerance can be improved by reducing the number of primer-template mismatches, reducing the annealing temperature, using a slow temperature ramp during reverse transcription and using
modified bases to minimize destabilization of mismatches. The results of this
experimental study should serve as guidelines in the design of primers for other systems. Our future plans include extending these studies to clinical specimens of known HIV-1 subtypes.
We thank Tom White and Tom Myers for critical review of the manuscript, Sheng-Yung Chang for technical advice, Fred Reichert and Agnes Cavalli for sequencing the cloned templates, Corey Levenson,
Tomas Martinez, Laura Jung and Olga Budker for providing oligonucleotides and
Annie Yoon for manuscript preparation.
*To whom correspondence should be addressed. Tel: +1 510 814 2891; Fax: +1 510
522 1285; Email: cindy.christopherson@roche.com
Template
No. of mutations
Difference relative to homologous template
SK462
SK431
Standard conditions
Modified conditions
M0
0
0
M1
6
0
108
10.3
M2
6
0
79
9.9
M3
5
0
22
4.3
M4
4
0
2
1.4
M5
0
3
1.5
1.7
M6
0
5
13
4.3
M7
0
2
0.9
0.6
M8
0
1
1.9
1.6
REFERENCES
Return