Download Highly specific radioimmunoassay for human insulin

Clinical Chemistry 44:7
1504 –1513 (1998)
Endocrinology and
Metabolism
Highly specific radioimmunoassay for human
insulin based on immune exclusion of all insulin
precursors
Michelle Deberg,1 Paule Houssa,1 Bruce H. Frank,2 Françoise Sodoyez-Goffaux,1* and
Jean-Claude Sodoyez1†
We describe a rapid and simple insulin RIA in which
proinsulin and conversion intermediates do not interfere. Three monoclonal antibodies (S1, S2, and S53)
were selected for their specificity (directed, respectively,
against the B10 region, the junction between A chain
and C-peptide, and the junction between B chain and
C-peptide), their affinity constant (;1010 L/mol), and
their interactive properties in mixture. S2 and S53 were
able to bind simultaneously to the same proinsulin
molecule, whereas neither could bind simultaneously
with S1. Preincubation of serum samples with an excess
of S2 resulted in capture of proinsulin and conversion
intermediates modified at the junction between B chain
and C-peptide into immune complexes that no longer
reacted with S1. Similarly, preincubation with S53 prevented proinsulin and conversion intermediates modified at the junction between A chain and C-peptide from
reacting with S1. Preincubation with an excess of both
S2 and S53 left insulin as the sole reactant with S1. Thus,
separation of insulin precursors from insulin by mutually exclusive antibodies is feasible, and on the basis of
this new principle, a highly specific RIA for insulin was
designed. The detection limit was 11 pmol/L, and the
inter- and intraassay coefficients of variation were 11%
and 5%, respectively. The potential of the assay for use
in clinical studies was verified by application to serum
samples from control subjects and patients with diabetes or insulinoma.
1
University of Liège, Division of Nuclear Pediatrics, Sart Tilman, 4000
Liège, Belgium.
2
Lilly Research Laboratories, 307 East Mccarty Street, Indianapolis, IN
46285.
*Author for correspondence. Fax 32-4-366-82-55.
†
Deceased.
Parts of these data were presented in poster form at the 29th meeting of
European Association for the Study of Diabetes, Istanbul, Turkey, September
1993.
Received September 23, 1997; revision accepted March 9, 1998.
Proinsulin is converted to insulin and C-peptide in beta
cell granules by two distinct site-specific endopeptidases
(PC1/3 and PC2) and carboxypeptidase H yielding, respectively, split 32,33; split 65,66 proinsulins; and des
31,32; des 64,65 proinsulins as conversion intermediates
(1– 4). The proinsulin conversion route via des 31,32
proinsulin appears to be the predominant pathway. Indeed, PC1 cleaves intact proinsulin at Arg32, Glu33 more
readily than PC2 at Arg65, Gly66; and PC2 has a stronger
affinity for des 31,32 proinsulin than for intact proinsulin
(5). Moreover, substantial amounts of des 31,32 proinsulin
and negligible amounts of des 64,65 proinsulin are found
both in the circulation (6, 7) and in human islets (8). In the
beta cell and in the circulation, it is generally accepted that
the major conversion intermediates found are des 31,32
and des 64,65 proinsulins and not split proinsulins (9),
because of the rapid cleavage of the newly exposed
C-terminal amino acids by carboxypeptidase H (3).
In many insulin assays, intact proinsulin, des 31,32
proinsulin, and des 64,65 proinsulin cross-react with
insulin. In healthy subjects, the contribution of intact
proinsulin and conversion intermediates is relatively low,
representing at most 10 –20% of immunoreactive insulin.
However, in patients suffering from type I (10) or type II
diabetes (11–13), familial hyperproinsulinemia (14, 15), or
insulinoma (16), the insulin concentration can be overestimated by the relatively high concentrations of proinsulin
and conversion intermediates.
Because insulin has no unique sequence of amino acids
of its own, RIAs or ELISAs based on the use of polyclonal
or even monoclonal antibodies (MoAbs)3 are never entirely specific. Specificity can be achieved only if a preliminary step of separation of insulin from proinsulin and
conversion intermediates followed by a nonspecific assay
3
Nonstandard abbreviations: MoAb, monoclonal antibody; RIT, radioimmune titration; RT, room temperature; and IC, immune complex.
1504
Clinical Chemistry 44, No. 7, 1998
is included or if MoAbs were raised against epitopes
specific for insulin by their conformation. We have developed a third approach based on a new principle of
exclusion between MoAbs and, in this study, present a
RIA specific for insulin.
Materials and Methods
This study was in accordance with the Ethical Committee
of the University of Liège.
peptides
Human, bovine, and porcine insulins and insulin modified in B10 (histidine replaced by aspartic acid) were
generously provided by Novo Research Institute (Copenhagen, Denmark). Human intact proinsulin, des 31,32
proinsulin, and des 64,65 proinsulin were from the Eli
Lilly Research Laboratories (Indianapolis, IN).
MoAbs
Preparation. Hybridomas were produced from spleen lymphocytes of mice immunized with rDNA proinsulin as
described in Sodoyez et al. (17). Briefly, BALB/c mice
were injected intraperitoneally with 20 mg of proinsulin
emulsified in complete Freund’s adjuvant, followed by
additional injections of the same amount of proinsulin in
incomplete Freund’s adjuvant at 2-week intervals. One or
two months after the third injection, the mice received a
booster of 40 mg proinsulin without adjuvant, and 4 days
later, their spleens were aseptically dissected out. Splenocytes were fused with SP2/O myeloma cells. Fused cells
were transferred to 96-well plates containing naive mouse
peritoneal macrophages as feeders in hypoxanthine-aminopterine-thymidine medium. Supernatants of wells containing growing hybridomas were screened for the presence of anti-proinsulin and anti-insulin antibodies by
radiobinding assays, and the hybridomas producing an
antibody with a high percentage of binding with proinsulin and insulin or proinsulin alone were cloned by
limiting dilutions.
The selected antibodies were produced in cell culture
devoid of fetal calf serum [culture medium: Hybridoma
High Protein (Gibco), 5 3 1025 mol/L 2-mercaptoethanol,
2 3 1023 mol/L glutamine, 100 000 units/L penicillin, 100
mg/L streptomycin, 1024 mol/L hypoxanthine, and 1.6 3
1025 mol/L thymidine] and purified on Hi-Trap™ Protein
G columns (Pharmacia) and stored at 220 °C.
Characterization. Epitope specificity was tested by radioimmune titration (RIT) or by competitive inhibition. Affinity constants of the selected monoclonal antibodies
were measured using the Scatchard method (18). Isotypes,
subclasses, and light chains were determined with the
Mouse-Hybridoma Subtyping kit (cat. no. 1183117) from
Boehringer Mannheim: an enzyme immunoassay for the
determination of immunoglobulin class, subclass, and
light chain type of mouse monoclonal antibodies.
1505
radiobinding assay
A constant amount of 125I-Tyr A14 human proinsulin or
125
I-Tyr A14 human insulin (25 000 cpm in 100 mL) was
mixed with 100 mL of culture supernatants and 100 mL of
serum pretreated with dextran-coated charcoal (peptidefree serum). After 2 h incubation at 37 °C, 1 mL of 150 g/L
polyethylene glycol was added to each tube. After 15 min
at room temperature (RT), the tubes were centrifuged,
supernatants were discarded, and the radioactivity of the
precipitates was measured using a multiwell gamma
counter. The percentage of antibody-bound radioligand
was calculated as the percentage of the total radioactivity
found in the precipitates. Blank values obtained by substituting peptide-free serum for monoclonal antibody
solution were subtracted from samples values.
rit
RIT was performed in duplicate by mixing a constant
amount of 125I-Tyr A14 human insulin, 125I-Tyr A14 B10
(Asp) insulin, 125I-Tyr A14 human proinsulin, 125I-Tyr A14
des 31,32 proinsulin, or 125I-Tyr A14 des 64,65 proinsulin
(25 000 cpm in 100 mL) with 100 mL of culture supernatants doubling dilutions, 100 mL of peptide-free serum,
and 100 mL of phosphate buffer containing 1 g/L bovine
serum albumin. After 90 min at RT, 1 mL of 180 g/L
polyethylene glycol was added to each tube. After 15 min
at RT, the tubes were centrifuged, the supernatants were
discarded, the precipitates were washed with 1 mL of 125
g/L polyethylene glycol, and their radioactivity was
measured using a multiwell gamma counter as described
for the radiobinding assay.
competitive inhibition
Epitope specificity was also investigated with 125I-labeled
proinsulin as tracer and proinsulin, insulin, des 31,32
proinsulin, des 64,65 proinsulin, and C-peptide as competitors. One hundred microliters of radioligand (25 000
cpm in 100 mL), 100 mL of buffer with or without the
different competitors at increasing concentration (from
10212 to 1025 mol/L), 100 mL of antibody diluted to bind
50% of the radioligand, and 100 mL of peptide-free serum
were incubated 90 min at RT. Bound and free hormone
were separated as described for RIT.
immune exclusion and cooperative binding
Immune complex (IC) size. Each MoAb S1, S2, and S53 was
incubated separately with 125I-labeled proinsulin (mol/
mol) overnight at 4 °C. In addition, equimolar mixtures of
MoAbs S1 and S2, S1 and S53, and S2 and S53 were
incubated similarly with 125I-labeled proinsulin (2 mol/
mol). ICs were submitted to gel filtration on a Superose 6
column connected to a fast protein liquid chromatography system (Pharmacia). The column was equilibrated
and run with 120 mmol/L NaH2PO4, 1 mmol/L EDTA,
and 3 mmol/L NaN3, pH 7.4. Bovine thyroglobulin (Mr
669 000), mouse IgG (Mr 160 000), bovine serum albumin
1506
Deberg et al.: RIA for insulin
(Mr 60 000), and 125I-proinsulin were used as molecular
weight markers to calibrate the column. The samples were
eluted at 0.4 mL/min, and fractions of 200 mL were
collected. The absorbance at 280 nm was read by an online
UV detector, and the radioactivity in each fraction was
measured in a well-type counter.
Cooperativity. MoAbs S2 and S53 were diluted to bind 50%
of the radioligand and then mixed in different ratios. A
constant amount of 125I-proinsulin (25 000 cpm in 100 mL)
and 100 mL of peptide-free serum were added to 100 mL of
mixed antibodies. After 90 min incubation at RT, the ICs
were precipitated as described above, and the radioactivity of precipitates was counted. Results were plotted as
the percentage of bound radioligand vs the ratio of the
two antibodies.
ria procedure specific for insulin
An equilibrium RIA was developed to quantify insulin
without interference of proinsulin or des 31,32 and des
64,65 proinsulins.
Plasma samples (100 mL) were preincubated at 37 °C
with an excess of either S2 (2.5 pmol in 50 mL) or S53 (1
pmol in 50 mL) or a mixture of S2 and S53, 100 mL of
peptide-free serum, and 125I-labeled insulin (25 000 cpm
in 100 mL). After 30 min, 100 mL of S1 (diluted to bind 50%
of the radioligand) was added, and the tubes were incubated 90 min at RT. Bound hormone was precipitated as
described for RIT.
Calibration solutions of insulin calibrated with the 1st
International Reference Preparation 66/304 and plasma
samples, when needed, were diluted in peptide-free
serum.
To validate the assay, insulin, intact proinsulin, des
31,32 proinsulin, and des 64,65 proinsulin were also
diluted in human peptide-free serum and calibrated on an
insulin calibration curve, using the MoAb directed against
an epitope common to all molecules (S1).
Preliminary studies showed that lipids (Intralipid 20%,
Pharmacia) and bilirubin (concentrations up to 5 g/L and
500 mmol/L, respectively) did not interfere with insulin
measurement in serum. We did not observe an adverse
effect of hemolysis. However, in view of its known
deleterious effect on insulin (19), it should be avoided.
patients
Serum samples were collected from 20 fasting healthy
subjects [14 women and 6 men; mean age, 30 years (range,
19 – 65 years); mean body mass index, 21.9 kg/m2 (range,
17.5–24.9 kg/m2); mean fasting plasma glucose, 4.2
mmol/L (range, 3.3–5.1 mmol/L)], from 6 patients with
non-insulin-dependent diabetes mellitus [5 women and 1
man; mean age, 66 years (range, 40 – 82 years); mean body
mass index, 29 kg/m2 (range, 22.8 –36.5 kg/m2); mean
fasting plasma glucose, 6.3 mmol/L (range, 4.1– 8.4
mmol/L)] after fasting and 120 min after a test meal, and
from 9 patients suffering from recurrent episodes of
hypoglycemia and with surgically proven insulinomas.
The patients with non-insulin-dependent diabetes mellitus had not been treated with insulin but were on oral
hypoglycemic treatment at the time of the study. Insulin
was measured with the specific assay described above.
Total insulin was measured with a RIA using the MoAb
S1. In patients with insulinomas, the concentrations of des
31,32 and des 64,65 proinsulins were estimated by subtracting the insulin value from the values obtained when
sera were preincubated with S53 or with S2, respectively,
and the concentration of proinsulin was estimated by
subtracting insulin and each conversion intermediate
from total insulin.
statistical analysis
The data presented in Fig. 5 were compared by ANOVA.
Multiple comparisons using Scheffé’s simultaneous confidence intervals were made to compare experimental
conditions two by two. The limit of statistical significance
was set at P ,0.05.
The clinical data were analyzed using the nonparametric test of Wilcoxon for paired or unpaired data.
Results
MoAbs
The characteristics of MoAbs S1, S2, and S53 are outlined
in Table 1.
Specificity. The specificity of the different MoAbs (S1, S2,
and S53) was investigated by RIT or by competition
experiments.
Table 1. Properties of the three developed MoAbs.
Clone name
IgG isotype
Specificity
S1
IgG1, k
Hi, Hpi
S2
S53
IgG1, k
IgG1, k
Hpi
Hpi
a
b
RI, radioimmune; Hi, human insulin; and Hpi, human proinsulin.
Coefficient of correlation for linearity of Scatchard plot.
RI titera
Affinity as
determined by
Scatchard analysis,
L/mol
r 2b
3200 (Hpi)
6400 (Hi)
2000
1600
7.9 3 10
1.0 3 1011
1.75 3 1010
1.47 3 1010
0.92
0.93
0.86
0.96
10
Clinical Chemistry 44, No. 7, 1998
S1 bound 125I-human insulin; it also bound 125I-intact
human proinsulin (60% of insulin binding), 125I-des 31,32
proinsulin (66%), and 125I-des 64,65 proinsulin (86%). It
did not bind 125I-human insulin modified in B10, thereby
suggesting specificity for the B chain portion including
B10 (Fig. 1A).
The competition curves obtained with 125I-labeled pro-
Fig. 1. Radioimmune titration of S1 and competitive inhibition of S2
and S53.
(A) Titration of S1 with insulin (F), proinsulin (f), des 31,32 proinsulin (Œ), des
64,65 proinsulin (), and insulin modified in B10 (E). Competitive inhibition of
binding of 125I-proinsulin to S2 (B) and S53 (C) with insulin (F), proinsulin (f),
C-peptide (‚), des 31,32 proinsulin (Œ), and des 64,65 proinsulin (). B/Bo, the
ratio of radioactivity bound in the presence of unlabeled antigen to the radioactivity bound in the absence of unlabeled antigen.
1507
insulin as tracer and insulin, proinsulin, C-peptide, des
31,32 proinsulin, and des 64,65 proinsulin as competitors
indicated that tracer binding to S2 (Fig. 1B) was inhibited
equally by proinsulin and des 31,32 proinsulin but not by
human insulin, C-peptide, and des 64,65 proinsulin, suggesting that S2 recognizes only the junction between A
chain and C-peptide. 125I-proinsulin binding to S53 (Fig.
1508
Deberg et al.: RIA for insulin
1C) was inhibited equally by proinsulin and des 64,65
proinsulin, suggesting that the epitope of S53 is the
junction between B chain and C-peptide.
Affinity constant. The affinity constant of the three MoAbs
for proinsulin was determined by Scatchard analysis.
Individual values of affinity and the coefficient of correlation for linearity of the Scatchard plots are shown in
Table 1. The three MoAbs were characterized by very
high affinity constants, and all Scatchard plots were
linear.
MoAb heavy and light chain composition. MoAb heavy and
light chain composition is shown in Table 1. S1, S2, and
S53 all belonged to the IgG1 isotype, and all had the same
type of light chain (k).
cooperativity and immune exclusion
Description of IC size. ICs formed with 125I-proinsulin and
a mixture of S2 and S53 yielded an asymmetrical peak of
radioactivity preceding the elution volume of S2 alone
with 125I-proinsulin and S53 alone with 125I-proinsulin
Fig. 2. Elution profiles of ICs made with
125
(Fig. 2A). These results indicate that, in the presence of
proinsulin, S2 and S53 are able to form ICs of a size
greater than one IgG. In the absence of the antigen, the
mixture of S2 and S53 eluted in the volume of one IgG, as
demonstrated by subsequent incubation of each fast protein liquid chromatography fraction with 125I-proinsulin
(data not shown). Moreover, the combination of these two
MoAbs in variable proportions did not increase the
amount of proinsulin bound (Fig. 3), indicating that these
two monoclonals have an additive effect and no positive
cooperativity.
ICs formed by 125I-proinsulin and a mixture of S1 and
S2 eluted in the same volume as that of S1 alone and
labeled proinsulin or S2 alone and labeled proinsulin,
thus in the volume of one IgG (Fig. 2B). The same elution
profiles were obtained if S2 was replaced by S53 (data
similar but not shown). This indicates that, in the presence
of proinsulin, S1 and S2 or S53 do not form immune
complexes of a size greater than one IgG. This inability of
S1 to form complexes with S2 and S53 in the presence of
antigen was further demonstrated by the following experiment: an insulin calibration curve performed with S1 was
I-proinsulin.
(A) With S2 alone (f), S53 alone (F), and a mixture of S2 and S53 (M); (B) with S1 alone (Œ), S2 alone (f), and a mixture of S1 and S2 (‚). Calibration was performed
using thyroglobulin (1), IgG (2), bovine serum albumin (3), and 125I-proinsulin (4).
Clinical Chemistry 44, No. 7, 1998
1509
exclusion (Fig. 5, column 2). If each mixture was preincubated with S2 (Fig. 5, column 3), the values obtained
corresponded either to insulin or to the sum of insulin and
des 64,65 proinsulin. Indeed, S2 bound to proinsulin and
des 31.32 proinsulin and excluded them. If the same
mixtures were preincubated with S53 (Fig. 5, column 4),
S53 bound to proinsulin and des 64,65 proinsulin and
excluded them, and only insulin or insulin with des 31,32
proinsulin were available for S1. The measurements of
each mixture after preincubation with S21S53 (Fig. 5,
column 5) were between 141 and 187 pmol/L, corresponding to insulin only.
Fig. 3. Cooperativity curve of MoAbs S2 and S53 mixed in different
ratios.
displaced to the left when a constant amount of proinsulin
was added to each insulin calibrator but resumed its
original position if the mixed insulin and proinsulin
calibrators were preincubated with S2 (Fig. 4A) or S53
(data similar but not shown). Similar displacements were
obtained by addition of des 31,32 proinsulin preincubated
or not with S2 (Fig. 4B) or by addition of des 64,65
proinsulin preincubated or not with S53 (Fig. 4C). Thus,
MoAbs S1 and S2 or S53 are mutually exclusive.
ria specific for insulin
The above described characteristics of MoAbs S1, S2, and
S53 allowed us to design a new insulin RIA specific for
insulin. Indeed, immune exclusion of all insulin precursors by two MoAbs directed against the junction between
A chain and C-peptide and the junction between B chain
and C-peptide, respectively, leaves insulin as the sole
reactant with the MoAb directed against the B chain.
Effects of known amount of proinsulin and conversion intermediates on insulin measurement. Different concentrations of
insulin, proinsulin, des 31,32, and des 64,65 proinsulins
were measured with S1 (Fig. 5). A known concentration of
each proinsulin-like molecule was added to a known
concentration of insulin (Fig. 5, column 1) and the concentration of each mixture was measured before immune
Assay characteristics. The calibration curve of this human
insulin assay is shown in Fig. 6. The assay characteristics
were as follows:
1. Limit of detection/limit of quantification.
The detection limit of this assay, calculated by the
mean 1 3 SD of 10 zero signals, was 11 pmol/L.
The working range of this assay was established by
calculating the CV of each calibrator in 10 independent
calibration curves. CVs were ,20% between 10 and 600
pmol/L and ,10% between 10 and 300 pmol/L.
2. CVs.
The interassay CV determined from the mean of duplicate measurements of three serum samples in 10 independent assays was 8.5%, 11.8%, and 4.2%, respectively,
at 71, 169, and 655 pmol/L insulin. The intraassay CV
estimated by analysis of 10 duplicates of three serum
samples in one assay was 5.4%, 5%, and 4.5%, respectively, at 73, 173, and 655 pmol/L insulin.
As can be expected in a RIA, the interassay CV of
samples is lower than the interassay CV of calibrators
because of the inclusion of a calibration curve in each run.
3. Analytical recovery.
Known amounts of purified insulin (ranging from 20 to
1000 pmol/L) were added in five assays to 10 sera of
known insulin contents (ranging from 28 to 245 pmol/L).
The recovery of human insulin was between 105.1% and
111.2%.
4. Dilution test.
Insulin recovery after serial dilutions (2 to 8) of 10
human serum samples (ranging from 88 to 283 pmol/L)
varied between 97.5% and 106% (mean, 101.8%), showing
a good linearity between the degree of dilution and the
measured concentrations.
clinical applications
The data presented correspond to small groups of subjects. They have essentially a value of example and will
require validation by larger clinical studies.
In fasting healthy subjects, insulin concentrations (median, 25 pmol/L; range, ,11– 62 pmol/L) were slightly
lower (2a #0.05) than total insulin concentrations (median, 36 pmol/L; range, ,18 –108 pmol/L).
In patients with non-insulin-dependent diabetes mellitus,
the median fasting concentrations of insulin and of total
1510
Deberg et al.: RIA for insulin
Fig. 4. Human insulin calibration curves obtained with MoAb S1 (F).
The same human insulin calibrators to which was added a constant amount (500 pmol/L) of: (A) proinsulin preincubated with S2 (f) or not (M); (B) des 31,32 proinsulin
preincubated with S2 (Œ) or not (‚); or (C) des 64,65 proinsulin preincubated with S53 (l) or not (L). B/Bo, the ratio of radioactivity bound in the presence of unlabeled
antigen to the radioactivity bound in the absence of unlabeled antigen.
insulin were higher than in healthy subjects (2a #0.01), at
100 pmol/L (range, 43–139 pmol/L) and 111 pmol/L (range,
50 –141 pmol/L), respectively; 120 min after a meal, the
concentrations of insulin and total insulin were 180 pmol/L
(range, 103–267 pmol/L) and 218 pmol/L (range, 127–371
pmol/L), respectively.
In nine patients with insulinoma, insulin values were
relatively low (range, 30 –248 pmol/L) compared with
total insulin (range, 130 –709 pmol/L). The insulin/total
insulin ratio and the contribution of each proinsulin-like
molecule were highly variable. In six patients, insulin
values were lower than proinsulin-like molecules values.
Proinsulin and des 31,32 proinsulin were usually predominant, and des 64,65 proinsulin was usually undetectable.
In one patient, the insulin value represented 67% (248
pmol/L) of the total immunoreactive insulin (367 pmol/
L), and the two conversion intermediates were present in
almost equal amounts (estimated concentrations of des
31,32 proinsulin, 48 pmol/L; estimated concentration of
des 64,65 proinsulin, 42 pmol/L). The concentrations of
des 31,32 proinsulin, des 64,65 proinsulin, and proinsulin
have not been corrected according to their affinity for S1
(Fig. 1A); the concentration of each insulin precursor is
therefore underestimated in these uncorrected data.
Discussion
After the original insulin RIA described by Yalow and
Berson (20), polyclonal antibodies that recognized all
insulin-like molecules and that had a relatively weak
sensitivity were used for many years. With the development of MoAbs (21), more specific and sensitive assays
were developed. An IRMA (7) and an immunoenzymometric assay (22) for insulin, designed with the same two
MoAbs, had a sensitivity for insulin of 2.3 pmol/L and 0.8
pmol/L, respectively, but both cross-reacted with intact
(5.3%) split 65,66 (62%) and split 32,33 (5%) proinsulins.
Andersen et al. (23) described a sensitive ELISA for
insulin (5 pmol/L) that cross-reacted with split 65,66
(30%) and des 64,65 (63%) proinsulins. The more recently
described assays for insulin were an immunoenzymometric assay (24) and an IRMA (25), both sensitive (6 pmol/L
and 1.2 pmol/L, respectively) and rapid but that crossreacted with des 64,65 proinsulin [77% (26) and 100%
(Sanofi Pasteur, personal communication) respectively].
Here we describe a RIA specific of insulin, based on a
new principle of separation of insulin precursors from
insulin by mutually exclusive antibodies: it requires three
antibodies selected for their specificity, affinity, and their
interactive properties when mixed. Fast protein liquid
Clinical Chemistry 44, No. 7, 1998
1511
Fig. 6. Calibration curve of human insulin (10 –1240 pmol/L), using
MoAb S1 after preincubation with a mixture of MoAbs S2 and S53.
Results are expressed as the mean 6 SD of 10 interassay calibration curves.
B/Bo, the ratio of radioactivity bound in the presence of unlabeled antigen to the
radioactivity bound in the absence of unlabeled antigen.
Fig. 5. Validation of immune exclusion method by addition of known
concentrations of proinsulin-like molecules to insulin.
A known concentration of each proinsulin-like molecule was added to a known
concentration of insulin (column 1), and the concentration of each mixture was
measured after immune exclusion or not (columns 2–5). Results are expressed
as the mean 6 SD of 10 interassay experiments. Hpi, human proinsulin; Hi,
human insulin; Des 31,32, des 31,32 proinsulin; and Des 64,65, des 64,65
proinsulin. * These two conditions were significantly different from each other
(observed difference, 45 pmol/L; Scheffé’s critical difference, 36 pmol/L) but
were not different from the expected concentration of insulin alone.
chromatography profiles show that S2 and S53 form with
proinsulin ICs of a size greater than one IgG. These ICs
are the result of simultaneous binding of S2 and S53 to the
same proinsulin molecule and are not the result of linkage
between S2 and S53 by their crystallizable fraction, as
demonstrated by the elution profile of mixtures of the two
MoAbs in the absence of antigen. No positive cooperativity was observed between these two MoAbs. This could
be related to the fact that individually they already have
a very high affinity for proinsulin. ICs formed by proinsulin and a mixture of S1 and S2 (or S53) elute in the
volume of one IgG, indicating an inability to bind to the
same proinsulin molecule in spite of a molar ratio between antibodies and antigen selected to favor formation
of large complexes (27). This immune exclusion might be
due to a modification of the conformation of proinsulinlike molecules when they are already bound by one
antibody or to a steric hindrance by portions of IgGs
distinct from the binding site. This exclusion between
MoAbs is responsible for the absolute specificity of this
assay and, in addition, has the advantage of suppressing
the step of separation of insulin from proinsulin-like
molecules, providing a specific, fast, and easy-to-perform
RIA. The sensitivity of this competitive assay (11 pmol/L)
is less than the sensitivity achieved by recent immunoenzymometric assays or IRMA assays but, nevertheless, is
adequate for most clinical studies.
In fasting healthy subjects, the concentrations of
plasma insulin measured with polyclonal antibodies are
relatively high, 96 pmol/L (28). Our values are much
lower, 25 pmol/L, and close to those measured with other
assays using MoAbs: 20 pmol/L (7) and 33 pmol/L (24).
The very small difference between total insulin and insulin in fasting healthy subjects is not surprising in view of
the very low concentration of proinsulin: 2.7 pmol/L
(Houssa, data to be published). Thus, insulin precursors
do not seem to account for the large differences in
1512
Deberg et al.: RIA for insulin
immunoreactive insulin values obtained with polyclonal
vs monoclonal antibodies. The numerous epitopes looked
at by polyclonal antibodies could allow recognition of
structurally related molecules, degraded insulin, or its
breakdown products.
In our small series of type II diabetic patients, basal
insulin is higher than in control subjects, in good agreement with Reaven et al. (13), and the ratio of insulin/total
insulin is ;90% in fasting conditions and 80% 120 min
after a test meal.
At variance in our patients with insulinoma, the insulin/total insulin ratio was usually lower and highly
variable; indeed only a few patients had predominantly
insulin-secreting tumors. This was also the observation of
Hale et al. (29). Monti et al. (24) do not mention individual
values of insulin or precursors, but their mean values for
insulin (119 pmol/L) and total insulin (349 pmol/L) were
very close to ours (116 and 298 pmol/L, respectively).
This heterogeneity of insulinomas was recognized years
ago, and classifications have been proposed on the basis
of the inverse relationship between the degree of beta cells
differentiation and the ratio of insulin precursors/insulin
in tumors (30) and serum (31).
If an increased proportion of insulin precursors/insulin is suggestive of impaired storage of insulin in tumoral
cells (30), the variable proportion of conversion intermediates that we have observed suggests, in addition, anomalies of proinsulin conversion mechanisms, des 31,32
proinsulin not always being the predominant pathway. In
this respect, a markedly reduced expression of PC3 has
been reported in a human insulinoma (32) and in the
insulin-producing cell line INS derived from a rat insulinoma (33). In the latter, no des 31,32 proinsulin was
detectable, whereas striking accumulation of des 64,65
proinsulin took place during proinsulin conversion.
In view of the heterogeneity of the insulin-like components released by these tumors, a nonspecific assay of
insulin cross-reacting with all precursors probably remains an appropriate screening tool for these patients
during a fasting test. The increasing availability and use of
highly specific assays will require measurement of each
component in patients suspected of insulinoma but will
concomitantly improve our knowledge of abnormal proinsulin conversion mechanisms.
This work was supported by grants from the Fonds de la
Recherche Scientifique Médicale (Brussels, Belgium). We
thank the “Service de Transfusion Sanguine de Liège” for
their hospitality. We are grateful to A. Albert for precious
advice regarding statistical analysis and to M.C. Requier
for technical assistance.
References
1. Steiner DF, Kemmler W, Tager HS, Peterson JD. Proteolytic
processing in the biosynthesis of insulin and other proteins. Fed
Proc 1974;33:2105–15.
2. Docherty K, Hutton JC. Carboxypeptidase activity in the insulin
secretory granule. FEBS Lett 1983;162:137– 41.
3. Davidson HW, Hutton JC. The insulin-secretory-granule carboxypeptidase H. Purification and demonstration of involvement in
proinsulin processing. Biochem J 1987;245:575– 82.
4. Davidson HW, Rhodes CJ, Hutton JC. Intraorganellar calcium and
pH control proinsulin cleavage in the pancreatic b cell via two
distinct site specific endopeptidases. Nature 1988;333:93– 6.
5. Rhodes CJ, Lincoln B, Shoelson SE. Preferential cleavage of
des-31,32-proinsulin over intact proinsulin by the insulin secretory
granule type II endopeptidase. Implications of a favored route for
prohormone processing. J Biol Chem 1992;267:22719 –25.
6. Temple RC, Carrington CA, Luzio SD, Owens DR, Schneider AE,
Sobey WJ, Hales CN. Insulin deficiency in non-insulin-dependent
diabetes. Lancet 1989;I:293–5.
7. Sobey WJ, Beer SF, Carrington CA, Clark PMS, Frank BH, Gray IP,
et al. Sensitive and specific two-site immunoradiometric assays
for human insulin, proinsulin, 65-66 split and 32-33 split proinsulins. Biochem J 1989;260:535– 41.
8. Sizonenko S, Irminger JC, Buhler L, Deng S, Morel P, Halban PA.
Kinetics of proinsulin conversion in human islets. Diabetes 1993;
42:933– 6.
9. Halban PA. Proinsulin conversion intermediates: a possible source
of confusion [Letter]. Diabetologia 1991;34:202–3.
10. Ludvigsson J, Heding L. Abnormal proinsulin/C-peptide ratio in
juvenile diabetes. Acta Diabetol Lat 1982;19:351– 8.
11. Mako ME, Starr JI, Rubenstein AH. Circulating proinsulin in
patients with maturity onset diabetes. Am J Med 1977;63:865–9.
12. Yoshioka N, Kuzuya T, Matsuda A, Taniguchi M, Iwamoto U.
Serum proinsulin levels at fasting and after oral glucose load in
patients with type 2 (non-insulin-dependent) diabetes mellitus.
Diabetologia 1988;31:355– 60.
13. Reaven GM, Chen YDI, Hollenbeck CB, Sheu WHH, Ostrega D,
Polonsky KS. Plasma insulin, C-peptide and proinsulin concentrations in obese and non obese individuals with varying degrees of
glucose tolerance. J Clin Endocrinol Metab 1993;76:44 – 8.
14. Gabbay KH, DeLuca K, Fisher JN, Mako ME, Rubenstein AH.
Familial hyperproinsulinemia. An autosomal dominant defect.
N Engl J Med 1976;294:911–5.
15. Barbetti F, Raben N, Kadowaki T, Cama A, Accili D, Gabbay KH, et
al. Two unrelated patients with familial hyperproinsulinemia due to
a mutation substituting histidine for arginine at position 65 in the
proinsulin molecule: identification of the mutation by direct sequencing of genomic deoxyribonucleic acid amplified by polymerase chain reaction. J Clin Endocrinol Metab 1990;71:164 –9.
16. Cohen RM, Given BD, Lucinio-Paixao J, Provow SA, Rue PA, Frank BH,
et al. Proinsulin radioimmunoassay in the evaluation of insulinomas
and familial hyperproinsulinemia. Metabolism 1986;35:1137– 46.
17. Sodoyez JC, Koch M, Lemaire I, Sodoyez-Goffaux F, Rapaille A,
François-Gerard C, Sondag D. Influence of affinity of antibodies
upon their detection by liquid phase radiobinding assay and solid
phase enzyme linked immunosorbent assay. Demonstration using
monoclonal antibodies raised against rDNA human proinsulin.
Diabetologia 1991;34:463– 8.
18. Scatchard G. The attraction of proteins for small molecules and
ions. Ann N Y Acad Sci 1949;51:660 –72.
19. Brodal BP. The influence of haemolysis on the radioimmunoassay
of insulin. Scand J Clin Lab Invest 1971;28:287–90.
20. Yalow RS, Berson SA. Assay of plasma insulin in human subjects
by immunological methods. Nature 1959;184:1648 –9.
21. Köhler G, Milstein C. Continuous cultures of fused cells secreting
antibody of predefined specificity. Nature 1975;256:495–7.
22. Alpha B, Cox L, Crowther N, Clark PMS, Hales CN. Sensitive amplified
immunoenzymometric assay (IEMA) for human insulin and intact
proinsulin. Eur J Clin Chem Clin Biochem 1992;30:27–32.
Clinical Chemistry 44, No. 7, 1998
23. Andersen L, Dinesen B, Jorgensen PN, Poulsen F, Roder ME.
Enzyme immunoassay for intact human insulin in serum or
plasma. Clin Chem 1993;39:578 – 82.
24. Monti LD, Sandoli EP, Phan VC, Piatti PM, Costa S, Secchi A,
Pozza G. A sensitive and reliable method for assaying true human
insulin without interaction with human proinsulin-like molecules.
Acta Diabetol 1995;32:57– 63.
25. Siohan E, Allauzen S, Mani JC, Demuynck A, Rousset M, Laprade
M, et al. A highly sensitive and specific immunoradiometric assay
for human insulin using monoclonal antibodies [Abstract]. Diabetologia 1996;39(Suppl 1):A86.
26. Chevenne D, Letailleur A, Trivin F, Porquet D. Dosage des proinsulines, insuline et peptide-C plasmatiques [Review]. Ann Biol Clin
1995;53:3–9.
27. Petersen K-G, Storch M-J, Rother K, Licht T, Kerp L. Insulin and
anti-insulin antibody interaction. Evidence for the formation of 7S
and 10S structures. Diabetes 1985;34:799 – 802.
28. Hales CN, Randle PJ. Immunoassay of insulin with insulin-antibody precipitate. Biochem J 1963;88:137– 46.
1513
29. Hale PJ, Djurup R, Baddeley RM, Nattrass M. Insulin and proinsulin concentrations in patients with insulinoma before and after
surgical treatment. Diabetes Nutr Metab 1991;4:113– 6.
30. Creutzfeldt W, Arnold R, Creutzfeldt C, Deuticke U, Frerichs H,
Track NS. Biochemical and morphological investigations of 30
human insulinomas. Correlation between the tumor content of
insulin and proinsulin-like components and the histological and
ultrastructural appearance. Diabetologia 1973;9:217–31.
31. Berger M, Bordi C, Cüppers H-J, Berchtold P, Gries A, Münterfering
H, et al. Functional and morphologic characterization of human
insulinomas. Diabetes 1983;32:921–31.
32. Smeekens SP, Montag AG, Thomas G, Albige-Rizo C, Carroll R,
Bening M, et al. Proinsulin processing by the subtilisin-related
proprotein convertases furin, PC2, and PC3. Proc Natl Acad Sci
U S A 1992;89:8822– 6.
33. Neerman-Arbez M, Sizonenko SV, Halban PA. Slow cleavage at the
proinsulin B-chain/connecting peptide junction associated with
low levels of endoprotease PC1/3 in transformed b cells. J Biol
Chem 1993;268:16098 –100.