CLIN. CHEM.
39/6, 965-971
(1993)
Determination of One Thousandth of an Attomole (1 Zeptomole) of Alkaline
Phosphatase: Application in an Immunoassay of Proinsulin
David B. Cook and Colin H. Self’
Enzyme amplification has proved to be a highly sensitive
quantification technique for immunoassays. We have
shown that by using a fluorescent end-point, even more
sensitiveenzyme amplification assays can be generated
than hitherto reported. We describe some general properties of this system and demonstrate its application in an
assay for human proinsulin in plasma. The detection
system can be used to measure less than one thousandth
of an attomole (1 zeptomole) of alkaline phosphatase,
equivalent to about 350 molecules of alkaline phosphatase per well of a microtiter plate. We have used this
system to construct a proinsulin assay with a sensitivityof
0.017 pmol/L.
IndexIng Terms: enzyme amplification y
say
fluommetric as-
The major advances
in immunoassay
have been conspecificity,
sensitivity,
speed, and convePressures
for these advances have come from
both clinical and research needs. Within these areas the
search for new diagnostic markers of disease has been
simulated
by increases in sensitivity,
which have also
brought the possibility
of earlier therapeutic intervention than was possible previously. The high sensitivity
of modern methods enables tests to be conducted with
smaller sample volumes; such tests are less invasive,
because capillary blood (1), saliva (2), and urine (3) can
be used as samples. The increases in speed have brought
the advantages
of convenience and have reduced the
chance of perturbation of the test or analyte change
occurring during otherwise long incubations.
Among the most sensitive methods described
is the
cerned
mence.
with
technique of enzyme amplification (4), which is capable
of detecting as little as 0.01 attomole (amol) of alkaline
phosphatase
(ALP) (5) in an assay sensitive to 0.0013
mllJ/L thyroid-stimulating
hormone.
The technique of enzyme amplification depends on an
enzyme label giving rise to a catalytic intermediate that
amplifies
the detectable change. The well-established
form of this (4) depends on ALP dephosphorylating a
phosphorylated form of a thcotinainide
adenine dinucleotide, for example, NADP to NAD. The dephosphorylated cofactor then enters a highly NAD-speciflc
redox
cycle, where it is reduced by NAD-speciflc
alcohol
dehydrogenase, after which the oxidized form is regenerated by diaphorase
with the concomitant
reduction of
Department of Clinical Biochemistry, The Medical School, Urnof Newcastle upon Tyne, Newcastle upon Tyne, NE2 41111,
varsity
UK
‘Author for correspondence.
Received April 2, 1992; accepted
February
5, 1993.
p-iodonitrotetrazolium
violet (INT-violet)
reagent
to an
intensely
purple formazan
dye. The oxidized form is
thus continuously
cycled with the formation
of detectable product with every turn of the cycle. The cycle
works analogously
if it is initiated
from the opposite
side, i.e., NADH generated
from NADPH.
For each
molecule of NADH generated,
-600 molecules
of formazan dye may be produced by a redox amplification
system in -10 mm (5). Colorimetric
read-out
of the
system is convenient,
and the dynamic
range of the
assays may be increased,
when required, by means of
kinetic
recording techniques.
Although these techniques are exquisitely sensitive,
fluorometric techniques
are often found to be inherently
more sensitive than the corresponding
spectrophotometnc procedures.
We have, therefore,
investigated
the
degree of sensitivity
that might be achievable
by incorporating a fluorometric end-point in a standard
enzymeamplified
system as shown in Figure 1. In doing this, we
anticipated
an additional
advantage:
a greater dynamic
range. Unlike colorimetric
systems, in which quantifying transmission
changes
becomes difficult when the
absorbance
becomes relatively high, the practical
limitations on fluorescence
usually involve exhaustion
of
substrate
or fluorescence
quenching.
In the present study we used the dye resazurin, which
has been used in the dairy industry since the 1930s (6).
Although this compound might not be ideal for ultrasensitive fluorescent assays, it could clearly be substituted
for the INT-violet in enzyme-amplified
procedures.
Resazurin is a blue compound that, on reduction,
produces
resorufin, a compound with visible (bright pink) fluorescence. It was previously
applied to the analysis of dehydrogenase enzymes by Guilbault
and Kramer (7) in
conjunction
with the NAD’7NADH
system.
The considerable
interest
in highly sensitive assays
for proinsulin,
coupled with our experience
with this
substance, led us to use prounsulin as the model analyte
in this investigation.
MaterIals and Methods
The dye resazurin was obtained from Aldrich Chemicals Ltd. (Gilhingham,
Dorset,
UK). Some batches
of
commercially
available
NADP
contain
sufficient
NAD
to have an adverse effect on background;
consequently, we used “NAD-free”
grade from Boehringer
Mannheim
(Lewes, Sussex
UK), which has proved
suitable for very sensitive detection of ALP. NAD
was
also obtained from Boehringer
Mannheim.
ALP of high specific activity (1660 glycine
units/mg
protein, reagent no. P-7923), diaphorase,
and alcohol
dehydrogenase
were purchased
from Sigma Ltd. (Poole,
CLINICAL CHEMISTRY, Vol. 39, No. 6, 1993 965
NBDP4 subs trot e)
olkeilne phosphotose-lObelled
conjugate
Resorufln
NnD
Diaphorese
Resozurln
(thenoI
Amplification
SOil
NADH
J
Step
Rcetoldeh!Jde
Fig. 1. The principleof enzymeamplification
Dorset, UK). We found it necessary
to dialyze diaphorase and alcohol dehydrogenase
extensively
against 20
mxnol/L sodium phosphate,
pH 7.2, before use to remove
contaminating
materials
such as NAD.
(In subsequent
work we have also used ALP from Boehringer
Mannhelm, described as “enzyme label for enzyme immunoassay,” with a specific activity
of about 3000 kU/g when
nitrophenyl
phosphate is used as substrate.)
it is important
to note that reagents
from
some
sources are not of sufficient
quality
to achieve the
extreme
levels of sensitivity
reported
in the present
work. it is also important
that scrupulous
purity, including
that of the water used to make up buffer
solutions, be adhered to for the ultrasensitive
detection
of ALP. We used water from a Milhi-Q#{176}
water purification system with fresh cartridges
(Millipore
UK Ltd.,
Watford, UK).
Fluorescence
of NAD
solutions was measured
in
transparent
microtiter
plates (Maxisorp
Nunc; Gibco
BRL, Gaithersburg,
MD). Preliminary
experiments
had
demonstrated
that there was no substantial
difference
between transparent
plates and the black microtiter
plates or strips specifically
manufactured
for fluorometry.
The assay mixture was made by putting in each well
100 /LL of NAD
solution (1-10 mol/L) in 50 mmol/L
diethanolainine
buffer, pH 9.5, followed by 100 L of
amplifier solution, which consisted of 600 U of alcohol
dehydrogenase
and 15 U of diaphorase in 12 mL of 0.1
mol/L sodium phosphate
buffer, pH 7.4, containing
700
L of ethanol and appropriate
concentrations
of rosazurin (final concentration, 0.058-1.46 mmol/L).
The fluorescence
was measured
at intervals
on a
Fluoroskan
II microplate
fluorometer (Labsystems,
Basingatoke, UK) with filter set 3, giving an excitation
of
544 rim and measuring
fluorescence
at 590 rim, or with
filter set 4, giving an excitation of 548 nm and emission
of 612 nm, depending
on the sensitivity
required. The
spectral peaks for resoruf in were determined
as 570 rim
(excitation)
and 590 nm (emission)
on a Perkin-Elmer
(Beaconsfield,
UK) spectrophotofluorometer
(Figure
2);
these values differed slightly from the spectra described
by Guilbault
and Kramer (8) (560 nm for excitation and
580 rim for emission).
However, it was not possible to
verify the calibration
of our instrument.
966
CUNICAL CHEMISTRY, Vol. 39, No. 6, 1993
Amplified Enzyme Quantification
Substrate
(200 L)-200
mol/L
NADP
in 50
mmol/L diethanolamine
buffer, pH 9.5, containing
1
mmolfL MgC12 and 1 mol/L
ZnC12-was
added to the
wells and incubated
at room temperature
for 20 mm. At
the end of this period, 100 L of the enzyme amplifier
solution was added and the developing fluorescence
measured
at intervals
of 1-60 miii. The procedure
was
used with resazurin
at final concentrations
in the amplifier solution of 0.29 and 0.875 mmoIJL.
Comparisonof FluorescenceReagent with INT-Violet
To compare the two reagents
directly,
we made a
solution of amplification
enzymes up to 20 mL in phosphate buffer containing
1.16 mL of ethanol, and divided
this solution into four equal portions. To one portion we
added
2 mmol/L INT-violet
(Sigma); to the others we
added 0.29, 0.058, or 0.029 mmol/L resazurin.
We determined NAD
over the range 0.2-20 nmol/L with the
ifiter sets decnbed
for fluorescence
and at 490 rim for
absorbance,
using a Dynatech
700 microplate
reader,
and read the results after 30 min of amplification.
Sensitivity
(detection
limit)
was determined
as the
concentration
of NAD
equivalent
to a signal 2.5 SDs
greater than the blank. Determinations
were performed
on six separate
occasions and the mean ± SD of the
sensitivity observed was calculated for the three concentrations of resazurin and the INT-violet reagent.
5
V
.2
V
480
500
520
540
560
Wavelength
Fig. 2.
cence
580
600
620
640
(m)
Excitation(#{149})
and emission (0) spectra for resorufin fluores-
Estimationof Maximum Sensitivityfor ALP
The ultimate
sensitivity
of the amplified enzyme
detection technique
depends on the concentrations
of
cycling enzymes used. By determining
the optimal ratio
of diaphorase to alcohol dehydrogenase
and increasing
the concentration
of enzymes in this optimal ratio, it is
possible to speed the rate of fluorescence development
and increase the intensity of fluorescence
at low concentrations of NAD
so that lower concentrations
of NAD
may be detected (9). Improvement
in the sensitivity
of
detection of ALP can thus be achieved notwithstanding
the inevitable
increase in background
due to NADP
in
this situation.
Determination
of NAD
was performed
in the presence of 1800 U of alcohol dehydrogenase
and
90 U of diaphorase
in 12 mL of phosphate buffer containing 700 jL of ethanol and 0.29 mmol/L resazurin.
At these concentrations
of amplification
enzymes
a
sensitivity
better
than
1 nmol/L
of NAD
could be
achieved after 30 miii of amplification.
For determination
of ALP, dilutions of ALP of high
specific activity were made in 50 mmol/L diethanolamine buffer. For the purpose of calculating
its concentration, the preparation
was taken to be 100% pure
enzyme with a molecular mass of 140 kDa. To 50 jL of
400 moI/L
NADP
in diethanolamine buffer containing 2 mol/L
zinc chloride and 2 mmol/L magnesium
chloride we added 100 &L of ALP at appropriate concentrations in diethanolamine buffer. The microtiter strips
were incubated
for 5 h at room temperature, after which
100 L of amplifier solution at optimal cycling enzyme
concentrations was added. Fluorescence was recorded 10
nun after the addition of the amplifier solution.
This experiment was performed with a Milhipore
Cytofluor 2300 microplate fluorometer in which the
excitation and reading of fluorescence takes place below
the plate rather than above it, as with the Fluoroskan
(this design precludes the use of opaque microtiter
plates available for fluorometry). Sensitivity setting no.
1 (least sensitive) was used to achieve the necessary
compromise
between a low background and keeping the
response within the limit recordable by the instrument.
In this instrument the wavelengths available from the
standard
filters were 530 nm for excitation and 590 rim
for fluorescence. The 530-nm excitation wavelength is
slightly farther from the optimum of 570 nm than is the
544 rim available on the Fluoroskan instrument.
Effect
of Serum on Detectionof NAD
The application of fluorometric techniques to studies
with serum is often hindered
by the contribution
of
serum constituents
to background
fluorescence.
This
potential
problem was investigated
by using optimal
ratios of enzymes for amplification
and constructing
standard
curves for the detection of NAD
in triplicate
microtiter wells in which 200 &L of human serum had
been incubated
overnight at 4#{176}C,
and comparing
them
with control wells, which had not been so treated. Before
measurement
of NAD
the wells were washed four
with 390 cL of 1 mJ.IL Tween 20 solution, exactly
as they would have been treated in an immunoassay.
times
Effect of Dilution on Final Detection of Fluorescence
Because
the initial
blue color of the resa.zurin
quenches the pink fluorescence
of the resorufin
produced, we investigated the effects of diluting the samples in the wells after the reaction
had been completed
at appropriate concentrations
of NAD
and amplifier.
NAD
solution (100 L) was incubated
with amplifier
solution for 90 miii and the fluorescence
recorded.
Immediately thereafter, we added 100 L of distilled water
and recorded the fluorescence
again.
ImmunometricAssay of Human Proinsulin
Maxisorp
microtiter plates of
coated with the IgG fraction of a
guinea-pig
antiserum raised against human prounsulin
(a gift from Eli Lffly Research Center, Indianapolis,
IN).
Dose-response
curves for human prounsulin were constructed in these strips by incubating 200 cL of promsulin standards
(Eli Lilly) in a 60 g/L solution of human
serum albumin in 0.04 mol/L phosphate buffer, pH 7.4,
for 6 h at room temperature.
The solutions were then
removed by shaking from the plate, and the wells were
washed three times with a 1 milL solution of Tween 20
in 9 g/L NaCl solution. Finally, 200 pL of a solution of
mouse monoclonal
antibody to human C-peptide (PEP
001, a gift from Novo Industri A/S, Bagsvrd,
Denmark) labeled with ALP was added in the phosphate
buffer and incubation was continued at room temperature for a further 18 h.
The excess conjugated
monoclonal antibody solution
was then removed by shaking
and the strips were
washed with the 1 mLIL Tween 20 solution. Activity of
the enzyme bound to the wells was then measured
by
the enzyme-amplifying
procedure.
strips
up to 96 wells
for constructing
were
Results
The development of fluorescence in NAD
solutions
with 0.29 and 1.45 mmol/L resazurin is shown in Figure
3. These experiments were performed with the nonoptima! wavelength ifiter set 4 on the Fluoroskan
instrument, thus reducing the amount of fluorescence detected
and enabling us to monitor the exhaustion
of the resazurin solution at high concentrations
of NAD.
The
fluorescence
developed
very quickly and, with 1.46
mmol/L
resazurun,
reached
a peak at 10 miii with 10
mol/L
NAD.
No further fluorescence
was produced
with greater
concentrations
of NAD,
or with longer
development times, because all of the resazurun had
been exhausted. At this concentration of resazurin, -50
nmol/L
NAD
was detected at 10 min; the resorufin
fluorescence produced at lower concentrations of NAD
was quenched by the blue color of the starting resazurin.
This was overcome by using lower concentrations of
resazurun (Figure
3B), when 20 nmolIL NAD
was
detected after 10 min of fluorescence development with
a starting
concentration of 0.29 mmol/L resazurin.
However, with this low concentration
of resa.zurin, an upper
CLINICAL CHEMISTRY, Vol. 39, No. 6, 1993 967
A
2000
B
2000
C
1000
1000
tO
20
30
40
50
60
70
0
tIme (mm)
Fig. 3. Rate of development of fluorescence with enzyme amplificationsystem
(A) 1.48 mmol/L resazurinand NAD at 0(x), 0.625 (is), 1.25 (0), 5(0), and 10 Mmol/L (A); (8)0.29
nmol/L (0). 625 nmol/L (0), 1.25 ,moVL (A), and 5 imol/L (#{149})
10
20
30
40
50
60
time (mm)
mmol/L resazunn and NAD at 0(x), 40 nmol/L (is),
156
in the proinsulin
immunometric
assay system to give
both sensitivity
and a wide operating range.
Standard
curves
for the proinsulin
immunoassay
for
the enzyme-amplified
fluorescence system are shown in
Figure 5A (0.875 mmol/L resazurin)
and Figure 5B (0.29
mmol/L resazurin).
Fluorescence
was developed for different times. For increased
sensitivity
these measurements were performed with filter set 3 (excitation
544
nm, fluorescence
590 nm), the wavelengths
provided in
the standard
filter set of the Fluoroskan
instrument
most suitable for resorufin fluorescence.
With 0.875 mmol/L
resazurin a large response could
be seen within 5 mm of the addition of the amplifying
reagent,
with detection of <0.1 pmol/L proinsulin.
As
incubation continued, the higher concentrations
of proinsulin (>4.05 pmol/L) exhibited
fluorescence
in excess
of the response of the Fluoroskan
instrument
(beyond
the limit of fluorometry).
After 25 min of incubation,
although the background
was increased,
0.05 pmol/L
15000
proinsulin
could be detected and the instrument response was exceeded by 4.05 pmol/L proinsulin. By 45
mm, although the background
was increased further,
ioooo
the response to 0.05 pmol/L proinsulin
was easily visible
to the unaided eye.
The limitation that quenching by resazurin
places on
sensitivity
was
shown
by
the
use
of
0.29
mmol/L
dye
*
(Figure
SB), when, after 10 miii of amplification, the
5000
fluorescence
at low concentrations of proinsulin was
greater than with the dye at the higher concentration.
Proinsulin was detected under these conditions at a
concentration
of 0.0 17 pmol/L
as a significantly in0
I’,.
creased response
(>2.5 SD) over the background
flue010
100
1000
10000
rescence, which corresponded
to 3.4 amol of proinsulin
(NADI nM
in the 200-L
serum
sample.
The sensitivity
was
FIg. 4. Fluorescence at various NAD concentrations
roughly the same at 30 miii of incubation, when the
0.875 mmol/l. reeazunn (E andA) and 0.058 mmol/L resazunn (0 and #{149});
absolute fluorescence was increased
but at the expense
opensymbols,fluorescence at 20 mm, and solid symbols, fluorescence at 2 h.
The points marked with an asterisk note the lowest NAD concentration at
of increased
background.
The upper limit imposed by
which fluorescence was easily visible to the unaided eye. Background at 2 h
the use of this limited startmg
concentration
of resawas 348 fluorescence units wIth 0.875 mmol/L resazurin and 1752 fluoreszurun was about 10 pmol/L proinsuhn.
Also, as a result
cence units wIth 0.058 mmol/L resazurin
limit
of NAD
determination
was reached at -1000
nmolJL as a result of the resazurin
being exhausted.
By leaving
the fluorescence
to develop for longer
periods (up to 1 or 2 h) and by using a lower concentration of resazurun
to reduce quenching
further,
lower
concentrations
of NAD can be detected, even by eye, at
some expense
of increasing
background
fluorescence.
This is shown by the response curves for NAD
in
Figure 4, in which resazurin was used at 0.05 and 0.875
mmol/L and the fluorescence
was measured
at the optimal available
wavelengths.
Concentrations
of resazurin
<0.058 mmol/L are of
little advantage because the dye is rapidly exhausted at
concentrations of NAD that would be too low to be of
use in an assay system with a wide dynamic range, but
greater sensitivity
can thereby be achieved.
To compromise
between these two limitations,
we
used resazurin concentrations of 0.29 and 0.875 mmol/L
968 CUNICAL CHEMISTRY, Vol. 39, No. 6, 1993
Beyond limit of fluoranetry
30000-
Table 1. Fluorescence Unlts after Enzyme
Amplification of SolutIons of NAD
nmol/L
0
20000
6
DilutedL
6 447
37.6
Increase on
0.1
1
10
100
3037
4085
34.5
200
20000
5723
7606
32.9
7059
8976
27.2
100
mmol/L resazurun
(0.48 nmo]/L NAD)
and 0.29 mmol/L
(0.96 nmol/L NAD)
was still substantially
greater
than with the INT-violet, in spite of the greater
quenching of resorufun fluorescence
at these concentrations.
The determination
of the sensitivity
of detection of
ALP at optimal enzyme concentrations
is shown in
Figure 6. Optimum enzyme concentrations
were determined by varying the ratios of diaphorase
to ALP in the
presence of NAD
(9).
Sextuplet
determinations
of fluorescence at each ALP
determination
were made and 10 determinations
were
made at zero concentration.
Sensitivity was assessed as
±2.5 SD of the fluorescence of the blank determinations,
which represented
an enzyme concentration
of 6 x
10 18 mol/L (6 amolfL) ALP, equivalent
to 6 x 1022
mol of ALP in a 100-L
sample.
In experiments
where serum was incubated
in wells
prior to detection of NAD,
no significant difference was
observed compared with control wells. Thus, the addition of serum resulted in no subsequent contribution to
the fluorescence
produced under the conditions
used,
which included washing.
resazurin
20000
C
I
0-
pmol/l.prolnsulin
FIg. 5. Calibration curves for human proinsulin assay
(A) Different times of fluorescence developmentwith enzymeamplification
and 0.875 mmoi/Lresazurin:5 mm(0); 15 mm (A); 25 mm (0); 45 mm (#{149}).
(B)
Different times of fluorescence developmentwith enzyme amplification and
0.29 mmol/L. resazuiln:5 mEn(0); 10 mm (A); 20 mEn (0); 30 mEn(C). Points
shown are mean of triplicate estimations, except the zero standard, which was
estimated as 10 replicates for statistical calculation
concentration
of dye used, the response
of the Fluoroskan
fluorometer was not exceeded as it had been with 0.875 mmol/L resazurin.
The effect of diluting the sample to reduce quenching
is shown in Table 1. The increase in fluorescence varied
from 37.6% at 12.5 nmol/L NAD
to 32.9% at 200
nmolfL NAD.
Thus, a significant increase in sensitivof the
50
1475
2034
37.9
100
3Q.
e
25
711
979
37.7
r\s
0
(b)
325
dilution,
%
a Recorded on the Fiuoroskan fluororneter.
“Diluted samples are the same wells remeasured after the addition of
4. of distilled water.
10000-
0.
12
Undiluted
lower
capability
ity can be gained by the simple expedient of diluting the
sample at the end of the amplification reaction to reduce
the quenching effect of the blue resazurin. However, an
increase was also noted in fluorescence recorded on the
Fluoroskan
instrument with a solution of 20 /Lmol/L
NAD
in which all the resazurun
had been used up.
Thus, increasing
the volume in the well by the dilution
appears to improve the performance of the Fluoroskan
instrument
significantly. Apparently, therefore, two independent benefits may accrue from such dilution.
Direct comparisons of the sensitivity
of detection of
NAD
with resazurin at three concentrations
and 2
mmoIJL INT-violet, all in the same enzyme solution, are
shown in Table 2.
The mean sensitivity
for 0.029 mmoIJL resazurin
(0.4
nmolJL NAD)
was approximately
nine times greater
than with INT-violet.
The mean sensitivity at 0.058
Discussion
The use of resazurin to produce resorufin fluorescence
in an enzyme amplification
immunoassay
has been
shown to be capable
of achieving
an extremely
high
sensitivity,
in excess of that previously
reported
by
means of INT-violet.
If the concentration
of aniplification enzymes is optimized (5), the INT-violet system is
itself capable of exquisite sensitivity
such that 0.01
amol of ALP can be detected. However,
an important
feature of the fluorometric system described here is that
increased sensitivity
was achieved with standard enzyme concentrations
(4, 10). Furthermore,
it was possi-
Table 2. SensitIvity of Detection of NAD (nmol/L) with
Resazurin and NT-Violet
Resazurin conc, mmol/L
0.029
0.058
Meansensitivity
SD
0.4
0.15
Range
0.2-0.65 0.3-0.85
0.48
0.23
0.29
0.96
0.44
0.3-1.4
INT-violat,
2 mmol/L
3.675
1.95
1.6-6.4
n = 6 each.
CLINICAL CHEMISTRY, Vol. 39, No. 6, 1993
969
1200-
enzyme amplification
system actually exceeded the response of the Fluoroskan fluorometer at concentrations
of proinsulin as low as 4 pmol/L. With respect to the
application
of the system to human samples,
we have
found that the rigorous washing
procedure
(used routinely with the coated microtiter
plate immunoassay
format) removes any significant contribution
to fluores-
C
600U-
U.
0
Sensltivlty
400
(+ 2.5 S.D.)
6x
-
0
10
100
1000
mols alkaline phosphatase (x 1o22)
Fig. 6. Detection of ALP at optimal ratio and concentration of
amplification enzymes
Mean ± 2.5 SD of replicates. Sensitivity, calculated as mean ± 2.5 SD of blank
deteryTlinatlOn, represented by 0.6 zrnol (6 x 10
moO of ALP
ble to achieve this sensitivity
and a particularly wide
dynamic range without recourse to kinetic analysis
of
the enzyme reaction. However, by analog’ with spectrophotometric systems (5), kinetic reading and analysis
should improve the system even further.
We have shown resazurin
to be highly effective in the
enzyme amplification system in spite of the fact that, to
some extent, it quenches the fluorescence
of the resorufin product.
This is important
at high resazurin
concentrations
and low NAD
concentrations.
We overcame
this problem by using the resazurin
at relatively
low
concentrations and thus accepting a reduced upper limit
of the assay for very high sensitivity. This precluded
achieving
the theoretical dynamic range for a fluorescence assay. Resazurin
is, therefore, not an ideal dye for
the technique;
a nonquenching colorless starting material would not be subject to this limitation.
Although at
the lower concentrations used, resazuri.n
was quickly
exhausted at moderate
concentrations of NAD
or of
analyte in an immunoassay,
a range of 500 times the
concentration was demonstrated with 0.29
and 2000 times with 0.875 rnmol/L
resazurin. However, at the higher concentration, the
fluorescenceresponse at 0.05 pmol/L proinsulin,
though
detectable, was quenched
and therefore lower. Thus,
precision of the estimation at such a low concentration
would be impaired.
The sensitivities
of previously reported immunometnc assays for proinsulin have been on the order of 1-5
pmol/L with ‘I used for detection, and a limit of 1.2
pmol/L was reported for an ELISA technique
(11).
Using the reagents described here, we have achieved a
sensitivity of 0.5 pmol/L for an immunoradiometnic
assay and 0.1 pmolIL with a spectrophotometric enzyme
amplification system (12). With fluorescence detection,
this has been reduced to 0.0 17 pmol/L proinsulin
at
nonoptimal concentrations of amplification enzymes. In
view of the sensitivity
of previously reported assays, it is
notable that under certain conditions the fluorescent
proinsulin
mmol/L
resazurin
970 CUNICAL CHEMISTRY, Vol. 39, No. 6, 1993
cence from components
of the serum matrix of the assay,
thus removing
a problem frequently encountered
with
fluorometric analysis.
Thus, this technique
should be
capable of high-precision
analysis at hitherto undetectable concentrations
of this analyte in serum samples as
well as in dilute buffer solutions.
Filters necessazy to isolate the optimum wavelengths
required for measurement of resorufin fluorescence are
not available as standard
on the Fluoroskan
or the
Millipore instruments. A filter of 590-nm
emission
wavelength
is fitted as standard. This corresponds exactly to the maximum emission wavelength
determined
for resorufin on a spectrophotofluorometer.
The nearest
available
excitation
wavelength
filters were 544 nm
(Fluoroskan) and 530 rim (Millipore), slightly removed
from the 570-nm maximum
determined. However, the
excitation spectrum is sufficiently wide to accommodate
this difference, although some 25% greater excitation is
obtained
at the optimum wavelength. Therefore, one
may expect a further
increase in sensitivity to be
achievable with an excitation filter at the correct wave-
length of 570 nm.
Notwithstanding
these obstacles, an ultimate sensitivity for the detection of 6 x 10_22 mol of ALP was
observed
at optimal cycling enzyme concentrations,
some 18 times greater
than the sensitivity previously
reported with the use of INT-violet: 1.1 x 10#{176}
mol (5).
The calculated sensitivity
of the present work, about
350 molecules of ALP per well, assumes that the protein
content of the high-specific-activity ALP reagent represented 100% enzyme. Any lower enzyme content implies
that an even lower detection limit was reached.
Although it is possible to detect even smaller numbers
of enzyme molecules [Rotman
(13) was able to detect
single molecules of 3-galactosidase
by observing fluorescent droplets through a microscope after 15 h of incubation with substrate], and highly sensitive
-galactosidase-based
immunoassays
have been reported
(14),
such extreme sensitivity has not been translated
into
immunoassay
determinations. The detection limit for
-galactosidase
in a standard system with practical
substrate incubation times applicable to immunoassay
was reported by Imagawa et al. (14) to be 0.02 amol.
The theoretical advantage
of a large dynamic range
has not been fully demonstrated in this present work
because
of our having to limit the starting concentration
of the dye and because the assay exceeded the limit of
response of the fluorometer, but this is under further
investigation.
The extremely high sensitivity reported here should
allow the investigation of very small samples, possibly
obtained
less invasively than required
for methods in
which amplification is not used, and should allow detec-
tion of analytes
present at very low concentrations.
Potential
applications include early detection of neoplastic development and infectious states. In experimental settings, the technology should prove useful in mon-
itoring the surgical removal of tumors and endocrine
tissues and, as with thyroid-stimulating hormone, for
example
(15, 16), such determinations may lead to
methods of important clinical relevance. Applications of
this technology
to receptor and gene probe analysis
now under investigation.
are
We thank J. Fuller of Novo Nordisk Diagnostics (now Dako
Diagnostics
Ltd.) for performing
the ALP conjugation to the
anti-C-peptide
monodonal antibody. We also thank Marie-Claude
Fawcett for her most capable and valuable assistance.
References
1. Axelsson 0, EnghmdJ. A radioimmunologic
method applied for
analyses of human placental lactogen on minimni
amounts of
capillary whole blood.Am J Obstet Gynaecol 1982;144:9&4-5.
2. Walker RI, Fahmy DR, Read GF. Mrenal status assessed by
direct radioimmunoassay
of cortisol in whole saliva or parotid
saliva. Clin Chem 1978;24:1460-3.
3. Connell JA, Parry JV, Mortimer PP, Duncan RJS, McClean
KA, Johnson AM, et al. Preliminary report: accurate assays for
anti-HIV in urine. Lancet 1990;355:1366-9.
4. Self CII. Enzyme amplification-a
general method applied to
provide an immunoassisted
assay for placental sflrsi1ine phosphatase. J Immunol Methods 1985;76:389-93.
5. Johannsson A, Ellis DH, Bates DL, Plumb AM, Stanley CJ.
Enzyme amplification for immunoassays.
Detection limit of one
hundredth of an attomole. J Immunol Methods 1986;87:7-11.
6. Ramsdell GA, Johnson WT, Evans FR. Investigation
of resazurin as an indicator of the sanitary condition of milk. J Dairy Sd
1935;18:705-15.
7. Guilbault GC, Kramer DN. Fluorometric procedures
suring the activity
21.
8. Guilbault
of dehydrogenases.
for meaAnal Chein 1965;37:1219-
GC, Kramer DN. Resorufin butyrate
acetate as fluorogenic
1965;37:120-123.
substrates
for cholinesterase.
and indoxyl
Anal Chem
9. Johannsson A, Bates DL. Amplification by second enzymes. In:
Kemeny DM, Challacombe
SJ, eds. ELISA and other solid phase
immunoassays:
theoretical and practical aspects. New York: Wiley, 1988:85-106.
10. Johannsson A, Stanley CJ, Self CH. A fast highly sensitive
colorimetric enzyme immunoassay
system demonstrating benefits
of enzyme amplification in clinical chemistry. Clin Chim Acta
1985;148:119-24.
11. Hartling SH, Dunesen B, Kappelgrd A-hI. Faber 0K, Binder
C. ELISA for human prounsulin. Cliii Chin Acta 1986; 156;289-98.
12. Dhahir F, Cook DB, Self CH. An amplified enzyme-linked
immunoassay
for human proinsulin. Clin Chem 1992;38:227-32.
13. Rotman B. Measurement of activity of single molecules of
-D-ga]actosidase.
Proc Natl Acad Sci USA 1961;47:1981-91.
14. Imagawa hI, Hashida 5, Ohta T, Ishikawa E. Evaluation
of
-D-galactosidase
from Escherichia
coli and horseradish peroxi.
dase as labels by sandwich enzyme inununoassay technique. Ann
Cliii Biochem 1984;21:310-17.
15. Woodhead JS, Weeks I. Circulating thyrotrophun as an index
of thyroid function.
Ann Cliii Biochem
1985;22:455-9.
immunoasaays:
a
of thyrotropin
evaluated. Cliii Chem
16. Clark PMS, Price CP. Enzyme-amplified
new ultrasensitive
1986;32:88-92.
assay
CUNICAL CHEMISTRY, Vol. 39, No. 6, 1993
971