JOURNAL OF

IMMUWOlOGICAL
METHODS
Journal of Immunological Methods 187 ( 1995) 201-2 11

Mathematical considerations of competitive polymerase chain

reaction
Ashley R. Connolly a, Leslie G. Cleland , Bruce W. Kirkham a-*
Rheumatology Unit, The Queen Elizabeth Hospital, Adelaide, South Australia, Australia
b Royal Adelaide Hospital, Adelaide, South Australia, Australia

Received 17 May 1995; accepted 5 July 1995

Abstract

Reverse transcriptase polymerase chain reaction (PCR) is used frequently to monitor gene expression. It is generally
regarded as a qualitative technique, although refinements have been made to improve quantification. The object of this study
was to develop competitive PCRs to allow reliable quantification of the rat T cell cytokines interferon-y (IFN-y),
interleukin-2 (IL-2) and interleukin-4 (IL-4). Truncated constructs of cDNA for these cytokines were prepared using
appropriate pairs of standard and specially constructed primers designed to allow subsequent co-amplification of the purified
competitor construct and the target cDNA. A high resolution capillary electrophoresis (CE) system was used for PCR
product detection. The performance of the system was compared with a mathematical model that describes and predicts the
exponential nature of the PCR reaction. Co-amplification of the competitor and target were achieved. A high level of
resolution and accuracy was achieved using CE to detect and quantify the PCR products. The rates of generation of the
respective products conformed closely but not exactly to the predictions of the mathematical model. The competitive PCRs
estimated initial numbers of target cDNA within l.l-5.0-fold relative to the amount of starting material as assessed by
conventional spectrophotometric absorbance prior to dilution and amplification. A convenient and flexible competitive PCR
strategy has been developed with accurate resolution of products and reliable quantification. Assay variability was far less
than biological variability likely to be encountered in experiments investigating immunological responses in rats or other
animals.

Keywords: Quantitative polymerase chain reaction; Rat cytokine; Capillary electrophoresis

1. Introduction

The reverse transcription polymerase chain reaction (PCR) is a very sensitive technique that employs
oligonucleotide primers to amplify a specific cDNA target from within a heterogeneous cDNA sample.

Abbreviations: PCR, polymerase chain reaction; CE. capillary electrophoresis; IFN-y. interferon-y; IL-2, interleukin-2; IL-4, inter-
leukin-4
Corresponding author. At: The Rheumatology Department, St George Hospital, Kogarah, New South Wales 2217, Australia. Tel.:
(02)3502604; Fax: (02J5881156.

0022-1759/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved

SSDl 0022..1759(95)00185-9
202 A.R. Connolly et al. /Journal of Immunological Methods 187 (I 995) 201-211

Although PCR has been widely used as a qualitative technique, the exponential accumulation of products has
the potential to confound quantification. Several methods have been devised to make PCRs more quantifiable.
For example, the target cDNA can be quantified relative to a constitutively expressed gene (Noonan and
Roninson, 1991) or to a known amount of the target cDNA that is amplified in a separate reaction
(Ballagi-Pordany et al., 1991; Melby et al., 1993). These methods may not compensate for the exponential
nature of the amplification process, where tube to tube variations can result in differential accumulation of the
products.
Quantification using a competitive reaction involves the generation of an internal standard curve by
co-amplifying serial dilutions of a mutated cDNA competitor with replicate amounts of the target cDNA.
Competitive PCR has the advantage that it obviates variability in the rate of amplification between different
tubes. Numerous competitive PCR procedures have been published with a limited number addressing the
theoretical requirements of the amplification reaction (Santagati et al., 1993). A recently described mathematical
model of competitive PCR provides a logical theoretical framework to analyse this key issue more formally
(Raeymaekers, 1993). W e present here results of our competitive PCR systems, which we use to quantify the T
cell rat cytokines interleukin-2 (IL-2), interferon-y (IFN-7) and interleukin-4 (IL-4), and compare their
performance with that predicted by the mathematical model of competitive PCR. The quantitative PCR includes
the use of a capillary electrophoresis (CE) detection system that allows more accurate and reproducible
quantification of the PCR products than that achieved using standard gel electrophoresis and densitometry. We
have assessed the accuracy of our competitive PCRs by comparison with spectrophotometric quantification of
purified target cDNA fragments prior to dilution and subsequent PCR amplification.

2. Materials and methods

2.1. Cell culture

Spleen cells were obtained from Dark Agouti rats that were killed according to institutional guidelines. A
single cell suspension was prepared and erythrocytes were lysed with ammonium chloride (140 mM NH&l, 17
mM Tris, pH = 7.6). Cells were cultured in RPM1 (Gibco) supplemented with 2 mM L-glutamine, 50 mM
2-mercaptoethanol and 10% (v/v) heat inactivated fetal calf serum. 2 X lo6 cells/ml were incubated with 5
pg/ml of concanavalin A in 24 well plates. After 24 h cells were harvested and washed with phosphate
buffered saline (PBS). Cell pellets were lysed in 500 ~1 of solution D (4.2 M guanidine thiocyanate, 26.4 mM
sodium citrate, pH = 7, 0.5% N lauroyl sarcosine and 5 mM 2-mercaptoethanol).

2.2, RNA extraction

The method of Chomczynski and Sacchi (1987) was used to isolate total RNA from 4 X lo6 stimulated cells.
Briefly, 500 ~1 of water saturated phenol, 100 ~1 of chloroform: isoamyl alcohol (49:1, v/v) and 50 ~1 of
sodium acetate (2 M, pH = 4) were added to the lysates. Mixtures were vortexed for 30 s then stored on ice for
a further 15 min prior to centrifugation for 30 min at 4C. The upper aqueous phase was retained and mixed
with 500 ~1 of propan-2-01 before being cooled to - 70C. for 30 min. The RNA was pelleted at 6500 X g for
30 min at 4C. The supernatant was decanted and the pellet vortexed with 200 ~1 of solution D and 200 ~1 of
propan-2-01. Following incubation at - 70C for 30 min the RNA was pelleted as described above then washed
twice with 1 ml of 75% ethanol (-20C). The supernatant was decanted and the RNA pellet was left to dry for
1 h at room temperature. RNA was redissolved in 52.5 ~1 of diethyl pyrocarbonate treated water (DEPC water).
2.5 ~1 of RNA was diluted to 50 ~1 with DEPC water then quantified by measuring its optical density at 260
nm on a Beckman DU 650 spectrophotometer.
 A.R. Connolly et al./ Journal of Immunological Methods 187 fl995) 201-211 203

2.3. First strand cDNA synthesis

RNA was incubated for 10 min at 65C prior to use. 500 ng of total RNA was diluted to 10.1 ~1 with DEPC
water. To this mixture, 4 ~1 of reverse transcription buffer (500 mM KCl, 200 mM Tris and 20 mM MgCl,), 4
~1 of dNTPs (10 mM each dCTP, dATP, dGTP and dTTP), 1 ~1 random hexamers (500 pg/ml) and 2.8 U of
RNAsin (Promega) were added. The components were incubated together with 64 U of Moloney murine
leukemia virus (M-MLV) reverse transcriptase (Gibco BRL) at 37C for 1 h to generate cDNA which was then
stored at -20C.

2.4. Polymerase chain reaction

5 ~1 of cDNA sample was added to 5 ~1 PCR buffer (15 mM MgCl,, 500 mM KCl, 100 mM Tris), 34.75
~1 Mini-Q water, 1 FM of each oligonucleotide of the respective primer pairs (IFN-y sense primer = 5-
ATGAGTGCTACACGCCGCGTCTTGG-3, IFN-y antisense primer = 5-
GAGTTCATTGACAGCTITGTGCTGG-3, IL-2 sense primer = S-CATGTACAGCATGCAGCTCGCATCC-
3, IL-2 antisense primer = 5-CCACCACAGTTGCTGGCTCATCATC-3, IL-4 sense primer = 5-
TGATGGGTCTCAGCCCCCACCTTGC-3, IL-4 antisense primer = 5-
CTTTCAGTGTTGTGAGCGTGGACTC-3) and 1.25 U of Taq polymerase (Perkin Elmer). Samples were
overlaid with mineral oil then denatured at 94C for 7 min prior to cycling. Cycling was carried out in a Perkin
Elmer Cetus DNA thermal cycler using the following conditions: 94C for 1 min, 60C for 1 min and 72C for 2
min. A final extension at 72C for 7 min was carried out after 35 cycles unless otherwise stated. The IFN-y
primers amplified a 405 bp cDNA fragment and the IL-2 and IL-4 primers produced PCR fragments of 410 bp
and 378 bp respectively.

2.5. Preparation of the target templates

The product of the reaction using the appropriate sense and antisense primer was electrophoresed through an
8% polyacrylamide gel at 150 V for 4 h. Gels were stained for 10 min in 1.3 FM ethidium bromide then the
cDNA was visualised using W radiation. The correct molecular sized bands were excised from the gel, finely
dissected then soaked in 250 ~1 of TE buffer (10 mM Tris, 1 mM EDTA) for 2 days at 37C. The aqueous layer
was retained and the cDNA extracted with phenol and chloroform (Sambrook et al., 1989) then purified by
proteinase K digestion (Merck). cDNA was precipitated by adding 15 ~1 of sodium acetate (3 M, pH = 4) and
300 ~1 of ethanol before cooling the mixture to -70C for 30 min. The cDNA was pelleted at 6500 X g then
washed with 1 ml of 70% ethanol. The supernatant was decanted and the cDNA pellet was left to dry for 1 h at
room temperature. The quantity of cDNA was calculated by measuring its absorbance at 260 nm. This purified
PCR product, which represents a cDNA fragment of the appropriate cytokine, was stored for use in subsequent
experiments and is referred to below as the target cDNA.

2.6, Preparation of the competitor templates

Competitor cDNA templates were made from cDNA according to the method of Celi et al. (1993). Briefly, a
nested antisense primer was selected for each cytokine in addition to the two primers outlined above. The nested
antisense primers for IFN-y, IL-2 and IL-4 were as follows IFN-7 primer = 5-
CAGGTGCGATTCGATGACAC-3, IL-2 primer = 5-ATCCAACACACGCTGCAGAG-3, IL-4 primer = 5-
CCTCAGTICACCGAGAACCC-3. The primers were designed to amplify cDNA fragments of the following
sizes: 322 bp (IFN-y), 299 bp (IL-2), and 324 bp (IL-4). The antisense primers for each cytokine are
complementary to sequences approximately 100 base pairs apart. The nested antisense 20mer primer for each
cytokine was 5 coupled to the 3 end of the framing antisense primer. PCR using this extended primer and the
conventional sense primer produced a product that was harvested and purified as outlined above then stored in
204 A.R. Connolly et al. /Journal of Immunological Methods 187 (I 995) 201-211

aliquots for use in subsequent experiments. These purified PCR products, which are approximately 80 base pairs
shorter than their respective target cDNA fragments are referred to below as the competitors. The target and
corresponding competitor cDNA fragments can be co-amplified using the appropriate framing oligonucleotide
primers.

2.1. cDNA storage and dilution

Aliquots of the purified target and competitor cDNA fragments were stored at -70C then incubated for 10
min at 65C prior to dilution and amplification by PCR. Storage under these conditions was shown to maintain
stable, reproducibly quantifiable amounts of cDNA as assessed by absorbance at 260 nm.

2.8. Competitive PCR

The competitive PCR was executed as outlined above except the 5 ~1 of cDNA was comprised of 2.5 ~1 of
the competitor and 2.5 ~1 of the target cDNA. The volume of Milli-Q water was reduced to 32.25 ~1 to
accommodate the addition of 2.5 ~1 of dNTPs (2 mM of each dCTP, dATP, dGTP and dITP).

2.9. Detection limits of PCR assays

To determine the detection limit of the PCR assays each target and competitor cDNA fragment was serially
diluted then amplified using the appropriate framing oligonucleotide primers. IFN--y and IL-2 PCRs were
conducted over 35 cycles. The IL-4 PCR was carried out over 40 cycles and required a 3 min preincubation at
94C before the addition of the primers in order to achieve the desired level of sensitivity.

2.10. Gel electrophoresis of PCR products

Following PCR, 25 /.~l of each product was added to 6 ~1 of loading buffer (40% w/v sucrose in water,
0.25% w/v bromophenol blue, 0.25% w/v xylene cyanole FF). 25 ~1 of this mixture was electrophoresed
through a 2% agarose gel at 150 V for 90 min. The gel was stained for 1 h in 1.3 PM ethidium bromide then
photographed using UV illumination. pUC 19 digested with HpaII was used as the molecular weight marker.

2. II. Capillary electrophoresis of PCR products

Prior to capillary electrophoresis (CE), the PCR products were concentrated and desalted as follows: 20 ~1 of
the PCR product was mixed with 2 ~1 of sodium acetate (3 M, pH = 5.5) and 60 ~1 of ethanol. After vortexing,
the samples were cooled to - 70C for 40 min then precipitated at 6500 X g for 15 min at 4C. The supernatant
was removed and the pellet was washed with 180 ~1 of 70% ethanol (- 20C). The cDNA was left to dry for 1
h at room temperature then redissolved in 10 ~1 of Mini-Q water for analysis using a Waters Quanta 4000
capillary electrophoresis system. cDNA was detected using an on column 254 nm UV detector. For DNA
separation a 57 cm DB17 (t.~ Sil) silica column (i.d. = 100 pm> was equilibrated with 0.5% hydroxymethyl-
propyl cellulose in TBE buffer (89 mM Tris, 89 mM boric acid and 2 mM EDTA). The PCR product was
introduced into the capillary by electrokinetic injection at 5 kV for 1 min and separation was achieved at 25C
using a negative power supply operating at 10 kV. Data was processed by on line computational analysis using a
NEC Power Mate 1 advanced personal computer equipped with Maxima 820 chromatography software.
Quantification of the PCR products was undertaken using CE unless otherwise indicated.

2.12. Mathematical modelling of PCR

The exponential accumulation of PCR products (before plateau is reached) is well documented and the
reaction has been mathematically described by Raeymaekers (1993) as follows: the initial amount of target
 A.R. Connolly et aL/Journal of ImmunologicalMethods I87 (1995) 201-21 I 205

sequence (T,,) amplifies with an efficiency (E,) over (j) cycles to yield a final amount of PCR product (T,)
according to the equation:

?=7-,(l +E,)

Similarly for a competitor (C)

C,=C,,(l +E,)

From these two equations it can be shown that

log(q/C,) = logT,-logC,, + j log[(l+E,)/(l+E,)]

= C + x m (1)
Y

Eq. 1 predicts a straight line of the form y = c + xm when fixed amounts of the target (T,,) and competitor (C,)
are co-amplified for different cycle numbers (j) to give rise to the products T, and Cj. Log[(l + E,/(l + E,)]
represents the gradient of the line cm>, while log T, - log C, constitutes the y intercept (c). When the
amplification efficiency of the target (E,) is identical to that of the competitor (E,), log[(l + E,)/(l + E,)]
equals zero and Eq. 1 is simplified to Eq. 2.

log(q/cj) = log To - log C,

(2)
Y = c -1 x

Eq. 2 predicts a straight line of the form y = c -x when different amounts of the competitor (C,) are
co-amplified with a constant amount of the target (T,) for a fixed number of cycles. When the ratio of PCR
products, log(q/C,), is plotted as a function of the competitor copy number (C,) spiked into each tube prior to
amplification, a line with a theoretical gradient of - 1 is expected. Under ideal conditions the number of target
copies (T,) can be quantified by interpolating the x axis value when the final number of the target and
competitor copies are equal (therefore log (T,/Cj) = 0 and log T,, = log Co).

3. Results

3. I. Capillary electrophoresis analysis of PCR products

To assess the reproducibility and accuracy of CE, the migration time and area beneath two adjacent peaks
from a pUC 19 molecular weight ladder were monitored. A 25.8 s deviation in migration time was observed
over a 30 min acquisition time while a 1.4% deviation in the ratio of the peak areas was obtained (n = 6). The
detection limit of CE was reproducible at 0.32 ng/pl of cDNA with a signal to noise ratio of 10. The resolving
power of CE is demonstrated in Fig. 1 which depicts the constituent DNA fragments of the pUC 19 ladder.
Ten different aliquots containing known amounts of the purified IFN-y target and competitor were made,
then each was separated and quantified using CE. The ratio of the measured peak areas and the relative weight
of each fragment varied by less than 10% (data not shown). Fig. 2 shows a representative electropherogram of
CE sieving as applied to PCR amplified cDNA.

3.2. Amplification efJiciencies of IFN- y and its competitor

To assess the relative amplification efficiencies of the IFN-7 target and competitor cDNA fragments, similar
amounts of each were co-amplified for 23, 25, 27, 30, 33 and 35 cycles (Fig. 3). Following amplification for the
respective number of cycles, the target : competitor ratio was calculated using capillary electrophoresis. A plot
206 A.R. Connolly et al. /Journal of Immunological Methods 187 (1995) 201-2 I I

Fig. I Electropherogram showing resolution of pUC 19 DNA digested with Hpall having fragment lengths of 501 bp (34 mm), 489 bp, 404
bp, 331 bp, 242 bp, 190 bp, 147 bp, I1 I bp and 110 bp. A 57 cm (from injection to UV window), DB-17 capillary having an internal
diameter of 100 pm was filled with 0.5% hydroxymethylpropyl cellulose. DNA was introduced into the capillary using electrokinetic
injection at 5 kV for 1 min and separation was achieved between 25C and 26C using an applied voltage of 260 Vcm-. A 254 nm UV
detector measured DNA migration.

0.270 r B

0.265 -
01

8
5 A i .,
t I
:: 0.260 - . \
, .._. _,_.
s

0.255 1 I
25 26 30 32 35

Time (minutes)

Fig. 2. A representative electropherogram showing separation of IL-2 target (1.5X IO6 copies) and IL-2 competitor (2.5 X 10 copies) PCR
products after co-amplification for 35 cycles. Peak A represents the IL-2 competitor (299 bp) and peak B represents the IL-2 target (410 bp).
PCR samples were desalted and concentrated (as outlined in the text) before CE analysis. A 57 cm (from injection to UV window), DB-I 7
capillary having an internal diameter of 100 @rn was filled with 0.5% hydroxymethylpropyl cellulose. DNA was introduced into the
capillary using electrokinetic injection at 5 KV for 1 minute and separation was achieved between 25C and 26C using an applied voltage
of 260 Vcm- A 254 nm UV detector measured DNA migration.

1FN-y Amplification EMency

20 25 30 35 40

Cycle Number(j)

Fig. 3. IFN-y amplification kinetics. Replicate samples containing 96.3 fg of IFN-y target and 86.5 fg of IFN-y competitor were
co-amplified for 23, 25, 27, 30, 33 and 35 cycles. The PCR products were quantified using capillary electrophoresis then plotted as a
function of the cycle number according to Eq. 1. The line can be described by the equation y = - 0.012 x + 0.323r2 = 0.920.
 A.R. Connollyet al./ Journal of ImmunologicalMethods 187 (1995) 201-211 201

Table 1
IFN-y, IL-2 and IL-4 targets co-amplified with their respective competitors
Cytokine Experimental gradient r* E, = x%E,

IFN-y -0.012 0.920 94-95%

- 0.011 0.832 95%

IL-2 - 0.003 0.464 98-99%

+ 0.007 0.875 103-104%

IL-4 +0.019 0.970 109%

+0.011 0.658 105%

Amplification reactions were performed on separate occasions for the number of cycles specified in the text. The PCR products were
quantified using capillary electrophoresis. The gradient of each line was obtained by plotting log(q /Cj) versus cycle number(j) according
to Eq. I. Gradients of lines for E, = x%E, were obtained by plotting log(T,/C,) versus cycle number (j) using notional values of E,
relative to E, in FQ. I, r* represents the correlation coefficient.

of this ratio as a function of the cycle number provided a straight line with a gradient approximating zero within
the range of cycles used. The gradients obtained in separate experiments utilising this strategy are shown in
Table 1. The negative gradient of each line indicates that the target amplified slightly slower than the
competitor.
Under ideal conditions the ratio of target to competitor should remain constant throughout the PCR reaction.
Notional vaIues of E, relative to EC (i.e. E, = x%E,) were substituted into Eq. 1 and log(q/C,) was calcuIated
at different cycle numbers (j>. The gradients generated by our experimental data (n = 2) indicate that the
amplification efficiency of the target (E,) is between 94% and 95% of the competitor efficiency (E,) (Table 1).

3.3. Relative amplification efJiciencies of IL-2 and its competitor

The relative amplification efficiency of the IL-2 target and competitor cDNA fragments were examined by
co-amplifying fixed amounts of each for 2.5, 27, 30, 33 and 35 cycles (n = 2). A plot of the target: competitor
ratio as a function of the cycle number yielded a straight line with a gradient close to zero within the range of
cycles used (Table 1). The positive gradient obtained from the first experiment indicates that the target has a
slightly faster amplification rate than the competitor whereas the negative gradient obtained from the second
experiment indicates that the target has a slightly slower amplification rate than the competitor.
The gradients generated by our experimental data indicate that the amplification efficiency of the target (E,)
is between 98% and 104% of the competitor efficiency (EC) (Table 1).

3.4. Relative amplification ejticiencies of IL-4 and its competitor

The relative amplification efficiency of the IL-4 target and competitor cDNA fragments was assessed by
co-amplifying fixed amounts of each for 30, 32, 34, 36, 38 and 40 cycles (n = 2). A plot of the target : competi-
tor ratio as a function of the cycle number yielded a straight line with a gradient approximating zero within the
range of cycles used (Table I). The positive gradient obtained from each experiment indicates that the target has
a faster amplification rate than the competitor.
The gradients generated by our experimental data were similar to those obtained with a notional amplification
efficiency of the target (E,) was between 105% and 109% of the competitor efficiency (E,).
208 A.R. ConnoNy et al./ Journal of Immunological Methods 187 (1995) 201-21 I

IFNr Competitive PCR

I

log [Competitor Copy Number (Co)]

Fig. 4. IF?&y competitive PCR. Replicate samples containing 52.9 I fg (I .3 X IO5 copies) of purified IFN-y target were each spiked with a
different dilution of purified IFN-y competitor representing respectively 3.57 X lo6 copies, 1.79~ lo6 copies, 3.57 X IO5 copies, 1.79X IO5
copies and 3.57 X lo4 copies. All samples were co-amplified for 3.5 cycles. Quantification of the PCR products was undertaken using CE.
The line can be described by the equation y = 6.714- 1.176 log x, r2 = 0.997.

3.5. Assessment of accuracy of competitive PCR

To assess the accuracy of the IFN-y competitive PCR at quantifying cDNA, serially diluted samples of the
competitor were co-amplified with a constant known amount of the IFN-7 target (n = 4). Prior to amplification,
the IFN-y target and competitor cDNA fragments were each spectrophotometrically quantified which provided
an estimate of the number of copies present. A single dilution of the IFN-y target was co-amplified with various
dilutions of the IFN-y competitor for 35 cycles. In accordance with Eq. 2 a log plot of the target: competitor
product ratio as a function of competitor copy number spiked into each tube prior to amplification produced a
straight line with a gradient approximating - 1 (Fig. 4). The starting number of IFN-y target copies established
by competitive PCR is compared to the initial number of copies estimated spectrophotometrically in Table 2.

Table 2
Comparison between spectrophotometric and competitive PCR quantification of cDNA
Cytokine Gradient r2 Target copy Target copy Fold
number estimated number estimated difference
by PCR spectrophotometrically
IFN-y - 1.176 0.997 5.119x 10s 1.309x IO5 3.9
- 1.180 0.992 4.131 x 105 1.644x 10s 2.5
-1.067 0.996 1.526x 10 2.326 x IO5 0.7
- 1.250 0.999 4.804 X lo5 2.326 X 10 2.1

IL-2 - 1.324 0.972 8.316x IO4 7.486 x lo4 1.1

- 1.136 0.995 9.859X IO5 7.486 X IO5 1.3
- 1.253 0.986 1.841x 105 8.751 X IO4 2.1
- 1.218 0.998 1.371 x 105 a.751 x 104 1.6

IL-4 - 0.849 0.967 8.568 x IO5 4.457 x 106 0.2

- 0.929 0.973 1.646X IO6 4.457 x 1o6 0.4
- 0.719 0.976 1.531 x 107 4.457 x 1o7 0.3

The gradient of the line for each competitive PCR was calculated when the molar ratio of PCR products, log(q / C,), was plotted as a
function of the competitor copy number Co spiked into each tube prior to amplification. The target copy number estimated by PCR was
calculated by determining the x axis value at log(q / Cj> = 0. The target copy number was estimated spectrophotometrically by measuring
the optical density of the target cDNA at 260 nm prior to dilution and PCR amplification. The ratio of the copy number estimated by PCR to
that predicted spectrophotometrically represents the fold difference. r* represents the correlation coefficient.
 A.R. Connolly et al. /Journal of Immunological Methods I87 (1995) 201-21 I 209

The accuracy of the competitive PCRs for IL-2 and IL-4 was assessed by co-amplifying serially diluted
samples of the respective competitor with a constant amount of the IL-2 (n = 4) or IL-4 (n = 3) target. The
target and competitor cDNA fragments were quantified spectrophotometrically prior to dilution and PCR
amplification for 35 cycles. A log plot of the target: competitor ratios as a function of competitor copy number
resulted in a straight line in each case. Each line had a gradient that approximated - 1 (Table 2). The
competitive PCR consistently overestimated the starting amount of IL-2 target and underestimated the starting
amount of IL-4 target obtained spectrophotometrically.

4. Discussion

In this paper we have analysed the target and competitor amplification efficiencies for each PCR system
using a well characterised mathematical model of PCR. We then tested the accuracy of our competitive PCR
systems and assessed the mathematical requirements predicted by the theoretical model of competitive PCR.
Differences in the rate of amplification of the target and competitor fragments will limit the quantitative nature
of the competitive PCR (Gilliland et al., 1990). The decreased length of the competitor relative to the target
cDNA fragments led us to investigate potential differences in their amplification rates. The amplification
efficiency of the target relative to the competitor cDNA fragment was calculated by comparing our experimental
data to Eq. 1 which describes competitive PCR mathematically. This equation describes a straight line of the
form y = c + mx when the final molar ratio of PCR products log(TJCj) at different cycles j is plotted. The
gradient of the line m represents log[(l + E,)/(l + ,?,)I which reflects the difference in the amplification
efficiencies of the target (E,) and competitor (E,). When both amplify with equal efficiencies, the ratio
[(1 + E,)/(l + EC>]becomes one, and a straight line with a gradient of zero is expected. As E, and EC diverge
there is preferential accumulation of one PCR product. The larger the difference between E, and EC, the further
the gradient will deviate from zero and the less accurate will be the competitive PCR.
Co-amplifying the target with its respective competitor cDNA fragment for different cycle numbers provided
in all cases data which when fitted into Eq. 1 produced a straight line having a gradient which reflected the
relative amplification efficiencies (Table 1). The amplification efficiency of the target relative to its competitor
was 94-95% for IFN-y, 98-104% for IL-2 and 105-109% for IL-4. These results demonstrate that a minor
difference in the length of the competitor relative to the target cDNA fragment was not associated with a
systematic difference in the respective amplification rates which supports previous findings (Wang et al., 1989).
Taq polymerase extends at a rate of 2000-4000 bp/min at optimal temperature suggesting that a 2 min
extension step during thermal cycling should be adequate for complete extension of both the competitor
(approximately 300 bp) and target (approximately 400 bp) cDNA fragments, as indeed appears to be the case.
Under ideal conditions, when the amplification efficiency of the target equals that of the competitor, Eq. 1
can be simplified to Eq. 2. The theoretical ideal for Eq. 2 predicts that a straight line with a gradient of - 1 will
be obtained when different concentrations of the competitor are co-amplified with a constant amount of the
target. To test the quantitative nature of our competitive PCRs using Eq. 2, known amounts of the target cDNAs
were co-amplified with different concentrations of the competitors. The gradient obtained from each competitive
experiment approximated the theoretical optimum value of - 1. Quantification of the target can be accurately
achieved by selecting a range of competitor copy numbers C, such that the target to be assayed T,, falls within
the range of C, values. Interpretation of the x axis value at a point when the final number of target copies
equals that of the competitor (log I;/Cj = 0 therefore log 7s = log C,) provides a more accurate interpolated
estimation of T, than that obtained by extrapolating to the y intercept (c) since the latter estimate will be more
influenced by errors in the slope of the line.
The relative amplification efficiency of the target and respective competitor fragment may have influenced
the accuracy of quantification. The IFN-y target amplified 5-6% slower than its respective competitor and
accordingly, the competitive PCR overestimated (except in one reaction) the number of target copies predicted
210 A.R. Connolly et al. / Journal of Immunological Methods 187 (1995) 201-21 I

spectrophotometrically. The IL-2 target amplified with an efficiency very similar to its competitor and
consequently, the number of target copies predicted by the competitive PCR accurately portrayed that estimated
spectrophotometrically. The IL-4 target amplified substantially faster than the competitor, and as a result the
PCR consistently underestimated the amount of target. Strategies to correct for variations in the amplification
efficiencies did not increase the accuracy of the competitive PCRs.
The apparent accuracy of the competitive PCRs will also be influenced by the accuracy of spectrophotomet-
ric quantification of the target and competitor cDNA fragments prior to dilution and PCR amplification. This
spectrophotometric detection and subsequent dilution will also be subject to random and possibly systematic
errors. Our results demonstrate that our competitive PCR systems were sensitive enough to consistently amplify
less than 200 copies of each cytokine mRNA transcript and sufficiently accurate to measure between l.l- and
5.0-fold changes in cDNA levels. This compares favourably to a 120-fold increase in IFN-y gene expression
that we have observed during concanavalin A stimulation of splenocytes. Non-competitive PCR systems report
detection differences in the order of lo-fold changes between cytokines based on lo-fold serial dilutions of the
starting material (Dallman et al., 1991). The accuracy of our quantitative PCRs, which is comparable to that of
other competitive PCR systems (Pannetier et al., 1993; Piatak et al., 1993), represents a substantial increase in
accuracy.
The quantitative nature of competitive PCR has been demonstrated by exploiting the sensitivity, reproducibil-
ity and highly quantitative nature of CE. Despite the difference in the length of the target and competitor
fragments both had similar mobilities when electrokinetically introduced into the capillary column and the
relative weight of the PCR products was accurately represented by the area beneath the respective peaks in the
electropherogram. The relative weight of the PCR products was converted into relative copy number by dividing
the area of each peak by the respective molecular weight of the PCR product. CE analysis of PCR products
increases the accuracy of quantification compared with slab gel electrophoresis which uses ethidium bromide as
a revealing agent, and in which the intensity of staining is dependent on the size of the PCR products.
Allowance for differences in the extinction coefficients of the PCR products could increase the accuracy of CE
with UV detection, although this refinement would be of marginal value in the present context. Thus, on line
UV detection along with its high resolution, reproducibility and full automation makes CE superior to
conventional slab gel electrophoresis and densitometry for analysing the PCR products.
We have calculated small differences in the amplification efficiencies of the targets and competitors and, in
accordance with a mathematical model, we have demonstrated that competitive PCR can be used to accurately
quantify small changes in cDNA levels relative to those observed in biological systems. Further sources of error
of potentially equal or greater importance in practice include variability in extraction of mRNA and the reverse
transcription step. In conclusion, the accuracy of competitive PCR is sufficient for reliable quantification of
cDNA obtained from biological samples as demonstrated by this favourable comparison of the performance of
competitive PCR to a mathematical model of optimal PCR performance.

References

Ballagi-Pordany, A., Ballagi-Pordany, A. and Funa, K. (1991) Quantitative determination of mRNA phenotypes by the polyrnerase chain
reaction. Anal. B&hem. 196, 89.
Celi, F.S., Zenilman, M.E. and Shuldiner, A.R. (1993) A rapid and versatile method to synthesize internal standards for competitive PCR.
Nucleic Acids Res. 21, 1047.
Chomczynski, P. and Sac&i, N. (1987) Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform
extraction. Anal. Biochem. 162, 156.
Dallman, M.J., Larsen, C.P. and Morris, P.J. (1991) Cytokine gene transcription in vascularised organ grafts: Analysis using semiquantita-
tive polymerase chain reaction. J. Exp. Med. 174, 493.
Gilliland, G., Perrin, S., Blanchard, K. and Bunn, H.F. (1990) Analysis of cytokine mRNA and DNA: Detection and quantitation by
competitive polymerase chain reaction. Proc. Natl. Acad. Sci. USA 87, 2725.
 A.R. Connolly et al. /Journal of Immunological Methods 187 (1995) 201-21 I 211

Melby, P.C., Damell, B.J. and Tryon, V.V. (1993) Quantitative measurement of human cytokine gene expression by polymerase chain
reaction. J. Immunol. Methods 159, 235.
Noonan, K.E. and Roninson, I.B. (1991) Molecular and Cellular Biology of Multidrug Resistance in Tumor Cells. Plenum, New York, p.
319.
Pannetier, C., Delassus, S., Darche, S., Saucier, C. and Kourilsky, P. (1993) Quantitative titration of nucleic acids by enzymatic
amplification reactions run to saturation. Nucleic Acids Res. 21, 577.
Piatak, Jr., M., Luk, K.C., Williams, B. and Lifson, J.D. (1993) Quantitative competitive polymerase chain reaction for accurate quantitation
of HIV DNA and RNA species. Biotechniques 14, 70.
Raeymaekers, L. (1993) Quantitative PCR: Theoretical considerations with practical implications. Anal. Biochem. 214, 582.
Sambrook, J., Fritsch, E.F. and Maniatis. T. (1989) Molecular Cloning. A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory,
Cold Spring Harbor, NY, p. E3.
Santagati, S., Bettini, E., Asdente, M., Muramatsu, M. and Maggi, A. (1993) Theoretical considerations for the application of competitive
polymerase chain reaction to the quantitation of a low abundance mRNA: Estrogen Receptor. Biochem. Pharmacol. 46, 1797.
Wang, A.M., Doyle, M.V. and Mark, D.F. (1989) Quantitation of mRNA by the polymerase chain reaction. Proc. Natl. Acad. Sci. USA 86,
9717.