Professional Documents
Culture Documents
Artigo Importante
Artigo Importante
Abstract
The polymerase chain reaction (PCR) is an extremely sensitive method owing to the repetitive multipli-
cation of template molecules. This property is a drawback for quantitative measurements because small
differences in the multiplication factor lead to large differences in the amount of product. Two methods can
be used to solve the problem of quantification: kinetic methods based on the determination or comparison of
the amplification factor; and coamplification methods, which compare the amount of product to that of a
simultaneously amplified standard template. An overview of the theoretical background of both methods is
presented. For selection of a suitable method, both theoretical and practical considerations are important.
Kinetic methods are the most convenient if PCR can be performed without opening the tubes, as in some
apparatus using fluorescence detection. Coamplification methods can be done without expensive equip-
ment but requires the parallel running of several PCR tubes. When the number of initial template molecules
is close to one, as in the limiting dilution technique, statistical considerations become important.
Index Entries: PCR; quantitative; PCR, competitive; kinetics theory.
n
1. PCR Amplification Pn = P 0 + ∑ P i-1 × Ei (3)
The polymerase chain reaction (PCR) is a re- i=1
petitive amplification process by which in each in which P0 denotes the starting amount of tem-
step (designated i), the copy number of product plate.
(P) already accumulated during the previous step A mathematically equivalent expression for
(Pi - 1) is multiplied by a factor that depends on the Eq. 3 is the iterated product:
efficiency Ei of DNA synthesis during that step. n
Ei is a measure of the relative increment of prod- Pn = P 0 × ∏ (1 + E i). (4)
uct in one step, defined as i=1
Ei = (Pi - Pi-1)/Pi-1 (1) Because the values of Ei are not known a priori,
Because the copy number can at most double in it follows from these equations that the absolute
one step, Ei has a value between 0 and 1. The abso- amount of P0 cannot be determined from one
lute value of the increase of the copy number in single measurement of Pn. However, the problem
one cycle is of quantitative PCR can be easily solved (in prin-
ciple) in different ways. The procedures that have
Pi - Pi-1 = Pi - 1 × Ei (2) been applied can be divided into two main catego-
The amount of product accumulated after n cycles ries: kinetic methods and coamplification meth-
is obtained by summation of the one-step incre- ods. Another useful distinction that can be made is
ments: between absolute quantification i.e., (the determi-
*Author to whom all correspondence and reprint requests should be addressed: Laboratorium voor Fysiologie,
KULeuven, Campus Gasthuisberg O/N, B3000 Leuven, Belgium.
Molecular Biotechnology 2000 Humana Press Inc. All rights of any nature whatsoever reserved. 1073–6085/2000/15:2/115–122/$12.00
may accumulate up to concentrations that inhibit The phenomenon is probably a result of the
amplification, whereas the efficiency of amplifi- higher background staining in agarose gels.
cation of the target sequence is hardly diminished. 3. A slope deviating from –1 may be caused by
One of the advantages of competitive PCR, at the unequal amplification efficiencies of tar-
least in theory, is its insensitivity to the effect of get and standard. The shift of the slope is ac-
saturation of the PCR. However, an interference companied by a shift of the point of
of saturation with the quantification cannot be equivalence, resulting in erroneous quantifi-
fully excluded for some templates, as will be ex- cation. It can been shown by computer simu-
plained in the Subheading 3.2. lation (14) that a deviation of the slope occurs
only if the difference between the amplifica-
3.2. Standard Curves of Competitive PCR tion efficiencies of target and standard varies
Having a Slope Different from –1 among PCR tubes of the dilution series on
which the standard curve is based. (A differ-
In the original method of constructing a log– ence between ET and ES that is identical in all
log standard curve to evaluate competitive PCR, tubes in all cycles results in a parallel shift of
the predicted property that the slope should equal the graph, thus maintaining the slope = 1
–1 has not been mentioned (10). As a conse- property but resulting in a shift of the point of
quence many papers show standard curves that do equivalence.) A possible cause is still specu-
not conform to theory. Although this fact does not lative. However, it is reasonable to suppose
necessarily imply that the quantifications based that in some cases, two similar sequences that
on these curves are grossly wrong (see explana- amplify with the same efficiency during the
tions below), it is obvious that such errors should exponential phase may start to amplify with
be avoided in the future. different efficiencies during the later linear
Because of the frequent occurrence of standard stages of the PCR. Small differences in the
curves with deviating slopes it is important to con- relevant properties of the templates may not
sider possible causes. There are at least three show up in conditions when DNA polymerase
types of explanations: and all substrates are in abundance, and the
concentration of the reaction products is still
1. If the PCR is run into saturation, a systematic below inhibiting levels. These differences may
bias against the more abundant PCR products become important, however, if the binding of
may occur if their sequences differ signifi- substrates or of polymerase, the rate of tem-
cantly (12). The consequence of this phenom- plate annealing, or the rate of strand dissocia-
enon is that the ratio of the products (Tn/Sn) is tion become rate-limiting. Because the
smaller than the ratio of the initial copy num- different samples constituting the standard
ber (T0/S0) when Tn is greater than Sn (or larger curve contain different copy numbers of tem-
when Tn is less than Sn). As a result, the slope plate, each tube will spend a different number
of the standard curve will be smaller than 1 in of cycles in the nonexponential phase of the
absolute value. The position of the point of PCR and will be differentially affected by the
equivalence is not shifted, so that the quanti- difference between ET and ES.
fication based on the position of this point
is correct. 3.3. Stochastic Effects in the Quantifica-
2. Systematic errors may arise in some methods tion of Small Numbers of Molecules
of quantification of the PCR products. It has It should be noted that the equations given in
been observed that ethidium bromide-stained the first part of this article are valid only if the
bands yield a tilted standard curve when ana- magnitude of influence of statistical variations on
lyzed on agarose gels but not on polyacryla- the outcome of PCR can be neglected. Statistical
mide gels (13). Also in this situation, the point considerations become important when the num-
of equivalence remains at the same position. ber of template molecules is small and when the
efficiency is significantly smaller than 1, because lar copy numbers of standard and target (15). Sto-
the amount of PCR product that is produced in chastic effects also may be important when PCR is
one cycle depends on molecular fluctuations. For used in combination with the limiting dilution tech-
example, starting from a single copy of a DNA nique, as this method requires that many samples
sequence amplified with an efficiency of 0.8, the contain one or a few template molecules (17,18). It
probability that one copy remains after the first follows that the method is reliable only if the effi-
cycle is 20%. In theory, the final copy number ciency equals or is very close to one.
after n cycles may be any number between 1 and 3.4. Other Confusions
1.8n. As a consequence, the analytical equations About Quantitative PCR
given above do not apply when the initial copy Besides concerns over the slope of competitive
number is low. A more rigorous description of the PCR’s standard curve, the reader should be warned
PCR process in these conditions should be based of some other illegitimate—but nevertheless pub-
on the theory of branching processes; thorough lished—simplifications in order to avoid a chain-
mathematical descriptions of PCR reactions in reaction multiplication of errors in the literature.
these conditions have been published (15,16). The It has been stated that reliable quantification is
simulation of PCR trajectories and the calcula- possible with competitive PCR, even when the ef-
tion of the expected outcome can be implemented ficiencies of the target (ET) and the standard (ES)
rapidly in suitable computer programs because are different, assuming that the ratio ET/ES is a
the distribution of Pi is the binomial distribution constant value (19). This statement should be
with parameters Pn-1 and E (16)*. As expected made more precise in that it applies only to rela-
the confidence interval of the estimated initial tive quantification and not to absolute quantifica-
copy number of the target is larger for a low ini- tions, as can be seen from Eq. 14, which itself is
tial copy number. The uncertainty also increases based on Eqs. 8 and 13:
with decreasing amplification efficiency. For
instance, when the initial copy number is 100, the log (Tn/Sn) = log (T0/S0) + n× log ([1 + ET]/[1 + ES]). (14)
relative uncertainty (ratio of uncertainty over true
value) is 0.1% for E = 0.9, and 25% for E = 0.5 Absolute quantification is possible only if the
(when the number of cycles is >20). The relative term at the right of the plus sign n × log ([1 + ET]/[1
uncertainty computed with one single initial copy + ES]) is equal to zero, i.e., if ET=ES. The deviation
is 2.55 for E = 0.5, whereas it is 0.99 for E = 0.9. of the quantification from the real value increases
Because the uncertainty increases with decreas- with difference in amplification efficiency and with
ing initial copy number, it follows that the accu- number of cycles. In theory, relative quantification,
racy of coamplification PCR of a very low copy i.e., comparing T0 in different samples, is still pos-
number of target will be higher when using a sible as long as n and ET/ES remain constant. How-
larger copy number of standard than by using simi- ever, it seems that reliable relative quantification
The following is a simulation in the Mathcad program (version 6.0) of 500 PCR runs of 20 cycles, starting from one copy of
template that is amplified with an efficiency of 0.6 P denotes the copy number and m denotes the amplification factor (1+p), where p
is the probability of duplication, which for large copy numbers corresponds to the amplification efficiency. Rnd(1) generates a random
number between 0 and 1.
PROGRAM:
p=0.6 n=20 trials=500 j=1..trials P0j=1
i=1..n uijrnd(1) Pij=Pi–j+qbinom(uij,Pi-l,j,p) mij=Pij/Pi-l,j
RESULTS
Mean number of copies at 20 cycles Pn (standard deviation): 1.22 × 104 (5.984 × 103).
Mean amplification factor m (standard deviation):1.60 (0.007).
As described by Peccoud and Jacob (16), the mean value of m is an estimation of the real amplification factor that converges to the
real value as i tends to infinity. For a limited number of cycles, an estimation of P0 can be calculated for each run from the mean value
of mi according to Eq. 7. The value of P0 in this particular simulation was 1.01± 0.497 (mean ± standard deviation).
in these circumstances is possible only in theory, (coamplification methods). The practical impli-
as it would require too many controls to be feasible cations of this requirement should not be under-
in practice. estimated. When using a kinetic method,
It has been stated that the ratio T0/S0 is propor- frequent opening of the PCR tubes for sampling
tional to the ratio of the slope of the line relating during the run is inconvenient. Therefore, using
Tn to the number of cycles n, divided by the slope continuous florescent detection (as in the Perkin
of a similar graph for Sn (if both slopes are deter- Elmer ABI PRISM™ 7700) is an excellent solu-
mined during the linear phase of the PCR, i.e., tion, although the equipment is expensive.
close to saturation; (20). There is neither a theo- Coamplification can be done with less invest-
retical nor a practical reason why this should the ment, but each quantification requires several
case. On the contrary, one would expect the PCR tubes, each of which must be further pro-
inverse because a sample containing more start- cessed for separate quantification of the target
ing material would run closer to saturation, and and the standard. When doing competitive PCR,
consequently, show a less steep increase in the perpetuation of the method needs especially
amount of product as a function of n. stringent precautions to avoid contamination of
A simple PCR method for relative samples and tools with the standard sequence, as
quantitation has been proposed as an alternative the latter is often handled from concentrated
to other methods, such as competitive PCR (21). stock solutions. In all cases, appropriate controls
The authors describe a method consisting of mak- should be done for each new sequence to be
ing a series of progressive dilutions by mixing quantified. Only then is it possible to obtain the
the two samples to be compared in different ra- ideal combination of accurate quantification and
tios. According to the authors, the alignment of the extreme sensitivity offered by PCR.
the quantities of amplified product in each tube References
along a line would demonstrate that the ampli- 1. Schnell, S. and Mendoza, C. (1997) Enzymological
fication efficiency in each tube was equal, al- considerations for a theoretical description of the
lowing direct comparison between the two quantitative competitive polymerase chain reaction
samples. It is clear that this method does not (QC-PCR). J. Theor. Biol. 184, 433–440.
eliminate the trap of running the PCR close to or 2. Schnell, S. and Mendoza, C. (1997) Theoretical de-
scription of the polymerase chain reaction. J. Theor.
into saturation, thereby compressing the differ- Biol. 188, 313–318.
ence between the amount of product in the two 3. Wiesner, R. J. (1992) Direct quantification of
samples. A close to linear regression line may be picomolar concentrations of mRNAs by mathemati-
obtained, even in conditions of near-saturation, cal analysis of a reverse transcription/exponential
polymerase chain reaction assay. Nucleic Acids Res.
particularly if one allows for errors in the quan-
20, 5863–5864.
tification of the products. Furthermore, this pa- 4. Ferre, F. (1992) Quantitative or semi-quantitative
per contains several mistakes in the calculations, PCR: reality versus myth. PCR Methods Appl. 2, 1–
and it also shows graphs relating the amount of 9.
PCR product to the number of cycles according 5. Cross, N. C. (1995) Quantitative PCR techniques
and applications. Br. J. Haematol. 89, 693–697.
to which the amplification factor would be much 6. Reischl, U. and Kochanowski, B. (1995) Quantita-
larger than 2, which is theoretically impossible. tive PCR. A survey of the present technology. Mol.
Biotechnol. 3, 55–71.
Conclusions 7. Raeymaekers, L. (1995) A commentary on the prac-
The description of the PCR process shows that tical applications of competitive PCR. Genome Res,
the problem of making PCR quantitative in a reli- 5, 91–94.
8. Kwok, S., Kellogg, D. E., McKinney, N., Spasic, D.,
able way is easily solved in theory. However, each
Goda, L., Levenson, C., et. al., (1990) Effects of
quantification requires several measurements, ei- primer-template mismatches on the polymerase
ther on several samples from one PCR tube (ki- chain reaction: human immunodeficiency virus type
netic methods) or on several PCR tubes 1 model studies. Nucleic Acids Res. 18, 999–1005.
9. Wang, A. M., Doyle, M. V., and Mark, D. F. (1989) 16. Peccoud, J. and Jacob, C. (1996) Theoretical uncer-
Quantitation of mRNA by the polymerase chain re- tainty of measurements using quantitative polymerase
action. Proc. Natl. Acad. Sci. USA. 86, 9717–9721. chain reaction. Biophys. J. 71, 101–108.
10. Gilliland, G., Perrin, S., Blanchard, K., and Bunn, H. F. 17. Villarreal, X. C., Grant, B. W., and Long, G. L.
(1990) Analysis of cytokine mRNA and DNA: detec- (1991) Demonstration of osteonectin mRNA in mega-
tion and quantitation by competitive polymerase chain karyocytes: the use of the polymerase chain reaction.
reaction. Proc. Natl. Acad. Sci. USA 87, 2725–2729. Blood. 78, 1216–1222.
11. Siebert, P. D. and Larrick, J. W. (1992) Competitive PCR. 18. Sykes, P. J., Neoh, S. H., Brisco, M. J., Hughes, E.,
Nature. 359, 557,558. Condon, J., and Morley, A. A. (1992) Quantitation of
12. Mathieu Daude, F., Welsh, J., Vogt, T., and McClelland, targets for PCR by use of limiting dilution.
M. (1996) DNA rehybridization during PCR: the “Cot Biotechniques. 13, 444–449.
effect” and its consequences. Nucleic Acids Res. 24, 19. Zachar, V., Thomas, R. A., and Goustin, A. S. (1993)
2080–2086. Absolute quantification of target DNA: a simple com-
13. Bouaboula, M., Legoux, P., Pessegue, B., Delpech, B., petitive PCR for efficient analysis of multiple
Dumont, X., Piechaczyk, M., et. al., (1992) Standardiza- samples. Nucleic Acids Res. 21, 2017–2018.
tion of mRNA titration using a polymerase chain reaction 20. Santagati, S., Bettini, E., Asdente, M., Muramatsu,
method involving co-amplification with a multispecific M., and Maggi, A. (1993) Theoretical considerations
internal control. J. Biol. Chem. 267, 21,830–21,838. for the application of competitive polymerase chain
14. Raeymaekers, L. (1993) Quantitative PCR: theoretical reaction to the quantitation of a low abundance
considerations with practical implications. Anal. mRNA: estrogen receptor. Biochem. Pharmacol. 46,
Biochem. 214, 582–585. 1797–1803.
15. Nedelman, J., Heagerty, P., and Lawrence, C. (1992) 21. Nicoletti, A. and Sassy-Prigent, C. (1996) An alter-
Quantitative PCR with internal controls. Comput. Appl. native quantitative polymerase chain reaction
Biosci. 8, 65–70. method. Anal. Biochem. 236, 229–241.