| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Department of Clinical Chemistry and Central Laboratory, Philipps University, Marburg, Germany.
a Address for correspondence. Abteilung für Klinische Chemie und Zentrallaboratorium, Klinikum der Philipps-Universität, Baldingerstraße, 35033 Marburg, Germany. Fax (+49) 6421 28 5594; e-mail kropf{at}mailer.uni-marburg.de and www.med.uni-marburg.de/labmed.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
2-macroglobulin at
physiological concentrations suppressed most of the TGF-ß1 signal.
Plasma fibronectin showed no effect, even at concentrations exceeding
its physiological range. Equilibrium concentrations computed from a
model system confirmed the experimental results. Dilutional
nonlinearity could be markedly reduced by an appropriately designed
activation procedure that minimized the effects of reassociation
between TGF-ß1 and its binding partners during restoration of a
neutral pH. Plasma should be used for measuring TGF-ß1 in blood.
Because serum TGF-ß1 is highly significantly correlated with the
platelet count, probably most of the TGF-ß1 is released by platelet
degranulation. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
-granules of platelets (2). Cells secrete TGF-ß in a
biologically inactive, latent form bound to an amino-terminal
propeptide called latency-associated peptide (LAP). Transient
acidification is commonly used to activate this so-called small, latent
TGF-ß (LTGF-ß), i.e., to release TGF-ß from its noncovalent
association with LAP. Numerous other proteins, including proteoglycans,
type IV collagen, fibronectin, and other components of the
extracellular matrix have been reported to noncovalently bind TGF-ß
(3)(4). In blood, TGF-ß supposedly occurs
associated mainly with 2-macroglobulin (2M)
(5)(6). The determination of TGF-ß in blood
has been advocated for diagnosis of various diseases, e.g., cancer
(7)(8), immunological disorders
(9), and hematological (10)(11)(12) or
fibrotic (13) diseases. Bioassays for assessment of functional TGF-ß (14) and immunoassays for determination of TGF-ß concentrations (15)(16)(17)(18)(19) have been described. However, review of the literature reveals different ranges given for the concentration of TGF-ß in blood, and several different recommendations for preanalytical sample handling can be found. Given the abundance of TGF-ß in platelets, some authors argue that serum measurements are not useful and that plasma concentrations have to be corrected by simultaneous measurement of markers of platelet degranulation (20). In their opinion, direct measurement of TGF-ß is not possible in plasma or serum by use of growth-inhibition assays, radioreceptor assays, or immunological assays because the dilution curves are not parallel to the calibration curves for these assays. An extraction procedure has been proposed to eliminate interfering substances. On the other hand, others report that serum and platelet-poor plasma concentrations are similar and that direct measurements of serum specimens thus "provide a reliable estimate of active and total TGF-ß in plasma" (16). The striking differences between the findings reported by different groups can at least in part be explained by different assay methodologies used. For example, the assays developed by Grainger et al. (16), which are based on receptor binding of TGF-ß and do not use an activation procedure, measure only a fraction of TGF-ß. These assays are not affected by platelet-derived TGF-ß, which is explained by the insensitivity of the assays to detect the large latent TGF-ß complex, which contains the large TGF-ß-binding protein.
Although several commercial assay kits have been used recently to investigate the diagnostic efficacy of TGF-ß blood concentrations, none of these assays has been evaluated by standardized method comparison methodology.
Here we describe a newly developed sandwich-type enzyme immunoassay for measuring TGF-ß1. Additionally, we compare this "in-house" assay with 4 commercially available immunological assays for determining TGF-ß1 in serum or plasma samples and define some important preanalytical conditions. The in-house assay is used as an (arbitrary) reference for the 4 commercial assays as well as for a detailed investigation of some fundamental aspects of measuring TGF-ß in blood. In particular, to explain the nonlinearity encountered when measuring serial sample dilutions, we compare experimental data with the results of a simulation model. By using a modified acid-activation scheme, our newly developed assay can measure TGF-ß1 in blood samples without any significant influence from sample dilution.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Commercial immunoassay kits for the comparison determinations of
TGF-ß1 were purchased: Predicta (Genzyme), Quantikine (R&D Systems,
Minneapolis, MN), Amersham (Amersham International), and Promega
(Promega). All assays were performed strictly as recommended by the
manufacturers. A comprehensive description of the respective assay
procedures has been compiled in Table 1
.
|
Recombinant human TGF-ß1 (rhTGF-ß1; cat. no. 47281, Serva) was
stored in 0.05 mmol/L HCl. Chicken anti-TGF-ß1 antibody (cat. no.
AB-101-NA), recombinant human TGF-ß3 (cat. no. AB-244-NA),
recombinant human LAP (rhLAP; cat. no. 246-LP), and recombinant human
LTGF-ß1 (rhLTGF-ß1; cat. no. 299-LT-005) were purchased from R&D
Systems. Recombinant human TGF-ß2 (cat. no. AB-112-NA) was from DRG
Instruments (Marburg, Germany). The mouse monoclonal anti-TGF-ß1,2,3
antibody (cat. no. 1835–01) was obtained from Genzyme. Human plasma
fibronectin, human 2M, biotinylated anti-mouse IgG antiserum
(Fab-specific; cat. no. B-7151), streptavidin peroxidase, and RIA-grade
bovine serum albumin (BSA) were purchased from Sigma. Peroxidase
substrate solution "BM blue POD" was obtained from Boehringer
Mannheim.
Low-binding polypropylene tubes and pipette tips were used throughout. Assay plates (MaxiSorp Immuno modules) were obtained from Nunc. A Milenia kinetic analyzer microplate photometer (Diagnostic Products Corp.) was used for absorbance measurements.
in-house assay for tgf-ß1
Solutions and calibrators.
To coat the assay plates, the
stationary antibody was dissolved in 0.05 mol/L NaHCO3, pH
9.1. The washing solution contained 9 g/L NaCl and 0.5 g/L Tween 20.
Assay buffer was prepared by adding 5 g of BSA, 0.5 g of
Tween 20, and 0.05 mol of Tris to 1 L of H2O and adjusting
the pH to 7.7.
For calibration we used recombinant human TGF-ß1 at 667, 222, 74, 25, and 8.2 ng/L dissolved in assay buffer.
Acid activation of samples.
Procedure 1: This procedure
was designed according to the procedures typically used by the
commercial assays and mainly was used for comparison purposes. After
50-fold dilution in assay buffer, 200- µL plasma or serum samples
were acidified with 20 µL of 1 mol/L HCl for 15 min at room
temperature and then neutralized with 1 volume of 1 mol/L NaOH. The
neutralized samples were added to the microwells. This procedure
produces a time lag of 10 to 15 min between neutralization of an
individual sample and the start of its exposure to the solid-phase
antibody. We confirmed that pH values of 2–3 (acidified) and 7–8
(neutralized) were obtained.
Procedure 2: Plasma or serum samples were diluted 50-fold in assay buffer and subsequently acidified by addition of 20 µL of 1 mol/L HCl to 200 µL of diluted sample. If desired, further dilution was performed in acidified assay buffer (assay buffer: 1 mol/L HCl, 10:1 by vol). After thorough mixing and 10 min of incubation at room temperature 11 volumes of the sample were neutralized with 1 volume of 1 mol/L NaOH. Care was taken that each sample was neutralized only 10–15 s before addition to the coated microwell plate. We confirmed that by this procedure pH values of 2–3 (acidified) and 7–8 (neutralized) were obtained.
Assay procedure.
The volume of calibrators, samples, and
antibodies was each 100 µL/well. Between incubation steps, the
microwells were washed 3 times with washing solution by a mechanized
washer. Assay plates were coated with chicken anti-TGF-ß1 diluted in
coating buffer (2 mg/L) at 4 °C overnight with shaking and were then
blocked for 2 h at room temperature or overnight at 4 °C with
the assay buffer, 250 µL/well.
Calibrators or activated test samples were incubated in the microwells in duplicate for 2 h at room temperature or overnight at 4 °C with shaking. The secondary antibody was monoclonal panspecific mouse anti-TGF-ß1,2,3 (5 mg/L, 2 h at room temperature with shaking). Subsequently, biotinylated anti-mouse IgG antibody (0.5 mg/L in assay buffer) was added and incubated for 45 min at room temperature with shaking. Incubation for 30 min at room temperature with streptavidin–peroxidase (50 µg/L) followed. Finally, BM blue POD substrate was added to each well. Color development was stopped after 4–5 min by addition of 50 µL of 1 mol/L H2SO4, and the absorbance was measured at 450 nm.
statistical and mathematical methods
The Astute statistical software package (DDU Software; University
of Leeds, Leeds, UK) supplement for the Microsoft Excel spreadsheet
package and the Statgraphics statistical package (STSC, Rockville, MD)
were used for statistical calculations. Detection limits were
calculated as
3SDBlank[S0]/(AS0- ABlank), where SDBlank is
the SD of the assay blank, [S0] is the
concentration of the lowest-concentration calibrator, and
AS0 and ABlank
are the absorbances of the lowest calibrator and the blank,
respectively. Comparisons between different assay kits were evaluated
according to Passing and Bablok (21). Spearman rank
correlation was used throughout. The calibration curves of the assays
were fitted through use of the four-parameter logistic function
provided by the Milenia analyzer.
calculation of equilibrium concentrations
The simple iterative method of Perrin and Sayce
(22)(23) is used for the computation of
equilibrium concentrations. From the Law of Mass Action the equilibrium
concentration cj of
complex j in a system of M reactants
Ai in N chemical reactions with stability
constants Kj is given by:
 |
 |
 |
 |
|
Results and Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
. Variation of the sample incubation period between 2 h and
overnight had no significant effect on the sensitivity (detection
limit) of the assay; for practical reasons, however, the latter was
generally used. The TGF-ß1 assay did not show any cross-reactivity
with isoforms TGF-ß2 and TGF-ß3 at 100 µg/L. Only the activated
form of TGF-ß was detected by the assay: LTGF-ß1 at concentrations
up to 400 µg/L gave signals below the detection limit, whereas the
expected concentrations were obtained after activation of the
LTGF-ß1. Some additional analytical characteristics of the assay are
shown in Table 1
, in comparison with those of the commercial assays.
|
The in-house assay was evaluated additionally for measurement of TGF-ß1 in cell culture supernatants. Because in these fluids the protein content varies considerably compared with that of serum or plasma, we investigated the influence of the protein concentration on nonspecific binding of TGF-ß to the assay plates (24). A BSA concentration of at least 1 g/L is sufficient to prevent nonspecific adsorption of rhTGF-ß1 (recovery >90%), whereas in protein-free DMEM >50% of added TGF-ß is lost. Further increasing the BSA concentration up to 20 g/L did not improve the results.
method comparison
Regression analysis.
For assay comparison, aliquots of
frozen serum (n = 16) and plasma samples (n = 5 to 15) were
thawed and measured with 2 or 3 assay kits in parallel. The
recommendations of the respective manufacturers were strictly followed.
For the comparison method, we generally used our newly developed
in-house assay. Therefore, the results of a regression analysis
according to the robust method of Passing and Bablok are given pairwise
in relation to the in-house assay (Table 1
). Aliquots of a serum pool
(~45 µg/L TGF-ß1) were used for assessment of the within-series
and interassay CVs.
The Promega immunoassay generally measured much higher concentrations of TGF-ß1 in serum or plasma samples than did the other assays. Additionally, these concentrations did not show any correlation to those obtained by the other kits. The differences could not, however, be attributed to a different calibration because our TGF-ß1 preparation gave nearly identical results by both the Promega and the in-house assay. We did not further investigate the reason for these differences, but the most probable explanation might be that, when assaying blood samples, the Promega assay is strongly influenced by interfering substances not identical with TGF-ß1.
Imprecision.
Except for the Promega assay, the
intermethod comparability of the other assays shown in Table 1
was
satisfactory. Although the correlation coefficients were generally
acceptable (0.71–0.95), the linear regression slopes nevertheless
ranged between 1.83 for the Predicta assay and 0.89 for the Amersham
assay. These discrepancies can be explained at least in part by the use
of different calibrators; different activation procedures and different
susceptibility of the assays to matrix effects also will lead to
variations. The analytical imprecision, i.e., within-run and
within-series CVs, of the assays investigated were acceptable for
immunoassays. Because the Amersham assay was performed only twice, we
estimated its within-series CV by computing the median value of
the CV obtained from 16 serum samples measured in duplicate.
Detection limit and TGF-ß1 determination in nonactivated
samples.
The detection limits of all assays are sufficient to
allow determination of TGF-ß1 in serum or plasma. However, for
determinations in other biological samples, e.g., cell culture
supernatants, a lower detection limit is advantageous. The in-house
assay and the Quantikine assay had the lowest detection limits, 1.9 and
3.9 ng/L, respectively. That of both the Predicta and the Amersham
assay was ~10 times higher. In nonactivated plasma (sodium citrate or
EDTA as anticoagulant) or serum samples, all assays reported TGF-ß1
concentrations well <100 ng/L or did not detect any.
These results differ from those of Grainger et al. (16),
who described two assay variants specific for active and (active +
latent) TGF-ß1 but with the samples needing no acidification. They
reported fractions of active/(active + latent) TGF-ß1 between <0.1
and 1 in serum or plasma samples. At present we have no explanation for
these discrepancies but it should be noted that the concentration range
found by Grainger et al. for (active + latent) TGF-ß1 in serum
(<0.1–35 µg/L) is considerably less than the ranges found by any
other assay tested (see Table 2
). These differences are most probably the result of using
different methodologies because the assays described by Grainger et al.
operate (a) without acid activation and (b) do
not detect the large latent complex released from platelets, whereas
the assays we tested generally measure total TGF-ß1 after activation.
|
Cross-reactivity with isoforms TGF-ß2 and TGF-ß3.
The cross-reactivity for the TGF-ß isoforms 1 and 2 was tested by
measuring recombinant TGF-ß2 and TGF-ß3 at concentrations of 100
µg/L in the diluent provided by the respective manufacturers. In all
assays compared, the cross-reactivity was negligible for all practical
purposes, being below or near the detection limit.
Effects of sample dilution.
Deviations from linearity
were noticed when human plasma or serum samples were diluted. This
problem has been described many times in the literature and has been
attributed to "matrix effects," the nature of which, to our
knowledge, was never analyzed in detail. To circumvent these effects
from producing "bizarre dilution curves," some authors advocate an
obligatory acid–ethanol extraction of biological samples
(20).
To further investigate this phenomenon, we prepared serial dilutions of identical serum samples, which we measured with the four commercial TGF-ß1 assays and the in-house assay. According to the manufacturers' recommendations, the samples were prediluted 30- to 60-fold (except for the Promega assay, which required a 120-fold starting dilution) to obtain a starting concentration at the upper end of the measuring range of the respective assay. Then, the samples were further diluted in twofold serial dilutions. For direct comparison, we normalized the results by dividing the result of every dilution step by the result obtained for the starting concentration. This approach showed a steady increase of the normalized (calculated) concentrations, reflecting an increasing signal relative to the respective dilution.
As shown in Fig. 2
, all assays displayed nonlinearity, with the Promega assay
performing worst, giving variations of the apparent TGF-ß1
concentrations of >2.5-fold, depending on the extent of dilution. Only
the Quantikine and in-house assays (the latter with activation
procedure 2) gave consistent results that were largely independent of
the extent of dilution.
|
The most obvious difference between the Quantikine and in-house assays
and the remaining assays is the activation procedures used. Therefore,
we tested whether an activation procedure similar to those of the other
assays led to dilutional nonlinearity. Indeed, using activation
procedure 1 (see Materials and Methods) with both the
Quantikine assay and the in-house assay gave nonlinear dilution curves
similar to those of the other assays (Fig. 2
).
Nonspecific matrix effects caused by (e.g.) different protein concentrations of sample and diluent are an unlikely cause for the behavior described. In this case one would expect that, in highly diluted samples, differences between sample and calibrator matrices would have been leveled out and lead to linear responses. However, even dilutions of several hundredfold are not sufficient to abolish the nonlinearity. Therefore, other factors must be responsible.
The dilution effects shown above will be even more adverse under routine conditions, given the substantial dilution of the sample required by all assays. Methods with strong dilution bias cannot, therefore, be used reliably to measure TGF-ß1 in blood.
Effects of LAP, 2M, and fibronectin on TGF-ß1
measurement.
To test the hypothesis that the dilutional
nonlinearity is caused by reassociation of TGF-ß1 with specific
binding proteins that compete with the solid-phase antibody of the
assays after the activation procedure, we assessed the use of LAP,
2M, and fibronectin as model substances. For the experiments
described below we used an earlier variant of the in-house assay, which
used a different streptavidin–peroxidase concentration that extended
the calibration curve to an upper limit of 2 µg/L (80 pmol/L). All
other properties of this assay were identical to the in-house assay
described above.
Various concentrations of rhLAP , 2M, or fibronectin were mixed with
rhTGF-ß1 (40 pmol/L) in assay buffer to yield different proportions
of the concentrations of the respective molecules. For 2M, the
highest concentration used corresponded to the upper limit of the
reference range for adults (3 g/L) (25). After incubation
at room temperature for 1 h, the TGF-ß1 concentration was
measured by the in-house assay with overnight sample incubation. As
Fig. 3
shows, under the conditions described, rhLAP and 2M but not
fibronectin are able to nearly completely mask TGF-ß1. The
suppression caused by rhLAP and 2M was reversible; i.e., after
transient acidification, the expected amount of TGF-ß1 was largely
recovered (data not shown).
|
To further investigate the nonlinearity of the dilution curves in a
different model system, we diluted rhLTGF-ß1 (40 nmol/L) in assay
buffer and subjected it to acid activation according to procedure 1.
After neutralization and twofold serial dilution, the samples were
measured with the in-house assay. rhTGF-ß1 served as the control,
being treated and measured in parallel. As shown in Fig. 4
, this simple system gives results comparable with those
obtained for the complex serum/plasma samples shown in Fig. 2
.
|
model computations
Numerous proteins are known to form strong noncovalent
associations with TGF-ß1. For the sake of simplicity, our model
system contained only 2M and LAP as major binding partners for
TGF-ß. The following reactions were considered:
TGF-ß1 + LAP 
LAP–TGF-ß1
(K1)
TGF-ß1 + 2M 2M–TGF-ß1 (K2)
We assume that the total concentrations of LAP
(Mr = 75 000) and TGF-ß
(Mr = 25 000) are identical and equal to the
mean concentration of latent TGF-ß found in citrated plasma, i.e., 4
µg/L. The total concentration of 2M (Mr =
725 000) is derived from its mean serum concentration of 2 g/L. We
ideally assume that no other binding partners for 2M exist
besides TGF-ß1. The dissociation constants K1
and K2 are values published previously as
10-9 and 10-7 mol/L, respectively
(5)(26).
For simulation of the immunoassay situation, we introduce a reaction between TGF-ß1 and an antibody Ab, which for simplicity we assume are equimolar:
TGF-ß1 + Ab Ab–TGF-ß1
(K3)
The dissociation constants K3 of the Ab binding to TGF-ß are varied between 10-9 (relatively low affinity) and 10-11 mol/L (high affinity). We further assume that the signal finally obtained by the immunoassay procedure is proportional to the concentration of the species Ab–TGF-ß, i.e., the amount of TGF-ß bound by the solid-phase antibody. The total concentration of the antibody is estimated from the situation found for the coated microwell. We approximate the volume concentration by the amount of antibody bound to the plastic surface. If the saturating surface concentration is taken as 100 ng/cm2, the upper limit for the antibody concentration in conventional microwells corresponds to ~1 mg/L (27).
For the above system with an assumed antibody affinity
(K3) of 10-9 mol/L, we obtained the
following equilibrium concentrations: [Ab–TGF-ß] = 29 pmol/L,
[2M–TGF-ß] = 126 pmol/L, and [LAP–TGF-ß] = 0.38 pmol/L. In
this system the most prominent fraction of TGF-ß is bound to 2M;
very minor amounts are bound to LAP, and 4.7 pmol (2.9%) is present as
free TGF-ß.
Next, we simulated the effect of sample dilution by serially diminishing the total concentrations of all components except for the concentration of the antibody. The latter concentration remains constant because the antibody is immobilized at the surface of the microwells. The concentrations of Ab–TGF-ß obtained at various dilutions were compared with those of an identically diluted simple calibrator solution that contained Ab and TGF-ß only. For better comparability, we normalized the dilution curves by dividing all values by the concentration of the most concentrated solution.
The resulting curves for different antibody affinities (Fig. 5
) are strikingly similar to the dilution curves obtained
experimentally by the assays we evaluated. According to the simulation,
the different susceptibilities of the assays to sample dilution are
strongly dependent on the affinity of the respective solid-phase
anti-TGF-ß antibody. This conforms with the influence of dilution
being only marginal for the Quantikine assay (Fig. 2
), because this
assay uses the high-affinity type II TGF-ß receptor (28)
(KD = 5–50 pmol/L) as the
solid-phase capture agent. However, an additional effect of the special
activation procedure of the Quantikine assay, which uses acidified 10
mol/L urea, might also play a role. Perhaps reassociation between
TGF-ß and its binding partners is suppressed by the urea
concentration remaining after neutralization (~0.3 mol/L) or maybe
the binding proteins are degraded in the presence of 10 mol/L urea.
|
In human blood samples 2M presumably is most important in binding to
TGF-ß, but other proteins with similar or higher affinities, such as
proteoglycans (3)(4), might also be effective.
2M also binds numerous other proteins, e.g., cytokines, and might
therefore not be fully available for binding TGF-ß. However, because
in the system described above only a very small fraction (<0.00005) of
total 2M was saturated by binding TGF-ß, the competition between
TGF-ß and other proteins for 2M is probably not important. We
stress, however, that the simulation results have been derived by
assuming a reversible equilibrated system. Because kinetic effects are
completely ignored by this simulation model, we also performed some
experiments on the kinetics of reassociation.
kinetics of reassociation
The aim of the following experiments was to determine the
influence of the time lag between neutralization of samples and start
of exposure to the solid-phase antibody. The in-house assay was used
for these experiments. Serum samples were diluted 1:50, 1:100, 1:200,
1:400, and 1:800 in assay buffer. The samples were activated by mixing
2000 µL of sample with 200 µL of 1 mol/L HCl for 15 min at room
temperature and were neutralized by addition of 200 µL of 1 mol/L
NaOH. The first aliquot was transferred to the microwells immediately
after neutralization; subsequent aliquots were transferred after 5, 10,
30, and 120 min. As a control, one aliquot was left acidified for 120
min and neutralized only afterwards. To assess the possibility of
losses from adsorption to the walls of the sample tubes, we transferred
this aliquot to a fresh tube before neutralization. Every row of 12
wells underwent the usual washing procedure immediately before addition
of the samples, to prevent the microwells from drying out when standing
for a prolonged time after washing. The sample incubation was continued
overnight for 18 h and the assay was performed as usual.
The apparent concentration of TGF-ß1 steadily declined with
incubation time after neutralization (Fig. 6
). This effect was statistically significant (one-way ANOVA,
multiple range tests) for the concentrations measured 30 and 120 min
after neutralization and also confirms the assumption of a partial
reassociation of TGF-ß1 and its binding partners as a cause for the
dilutional nonlinearity. Although the concentration of the control was
lower than the concentration of the sample measured immediately after
neutralization (0 min), this difference was not statistically
significant. A slight degradation of TGF-ß1 during the 120 min of
acidification might be responsible for the lower concentration of the
control. In a similar experiment we investigated the reassociation
kinetics of the activation procedure used by the Quantikine assay,
i.e., activation with 2.5 mol/L acetic acid/10 mol/L urea for 10 min
followed by neutralization with 2.7 mol/L NaOH in 1 mol/L HEPES. The
time-dependent decrease in the apparent concentration of TGF-ß1 found
with this procedure was less pronounced than in the HCl/NaOH procedure.
The acetic acid/urea activation is not completely irreversible,
however; 120 min after neutralization the fraction of TGF-ß1 found
was 0.8 compared with the 0.68 found with the HCl/NaOH procedure (Fig. 6
). Activation with 2.5 mol/L acetic acid/10 mol/L urea might therefore
be an acceptable alternative when our recommended neutralization
procedure is not feasible.
|
comparison between serum and plasma samples
TGF-ß1 was measured in serum or plasma samples from patients
submitted for routine analysis in our laboratory. TGF-ß1 was
activated and the samples were measured as described. The median
concentrations obtained with the four commercial assays and the
in-house assay, compiled in Table 2
, show that all assays measure
clearly higher concentrations in serum than in plasma. Although the
serum and plasma samples compared were not obtained from identical
patients, the ratio of the median serum/plasma concentrations was
nonetheless similar for all assays compared. Because -granula of
platelets contain TGF-ß1 in high concentrations (29),
the fact that serum concentrations are relatively greater than in
plasma is most likely caused by platelet degranulation during the
clotting process (20). This explanation is supported
further by the strong positive correlation between platelet count and
the serum concentration of TGF-ß1; correlations between platelet
count and TGF-ß1 concentrations in citrated plasma were generally not
significant (Table 2
). Only the Promega assay markedly differed from
the other assays, yielding a negative correlation between platelet
count and TGF-ß1 serum concentration as well as a clearly lower
TGF-ß1 serum/plasma ratio.
concluding remarks
There has been much controversy concerning the immunological
determination of TGF-ß in biological samples. The main problems
discussed were (a) the nonparallelism between calibration
curves and sample dilution curves, (b) the suitability of
different kinds of samples (serum vs plasma anticoagulated with citrate
or EDTA), and (c) the influence of preanalytical factors.
We were able to explain the nonlinearity of sample dilution curves, which is often attributed to unspecific matrix effects, by assuming a simple competitive binding model. Theoretical considerations indicated that the dilution linearity mainly depends on the affinity between solid-phase reactants and TGF-ß. This comfortably explains the differences in dilution linearity in the assays compared. This model is based on the assumption of at least a partial reassociation between TGF-ß and its binding partners after neutralization. However, as the system is obviously far from having reached its equilibrium state immediately after neutralization, reassociation kinetics must be considered in addition. Both assumptions combined provide the rationale for an appropriately redesigned activation/neutralization procedure that avoids dilutional nonlinearity. Therefore, direct measurement of total TGF-ß in blood becomes possible without resorting to acid–ethanol extraction procedures, which activate latent TGF-ß1. Use of the latter procedures renders measurement of active TGF-ß in blood generally not possible. These time- and labor-consuming extraction procedures depend on the reliability of the assessment of the overall recovery of TGF-ß. Because in vivo TGF-ß might occur associated to various binding partners in variable amounts, it seems doubtful whether the simple use of 125I-labeled TGF-ß for assessment of recovery is sufficient. Furthermore, the different kinetics of naturally occurring (i.e., latent) TGF-ß and free (labeled) TGF-ß will influence the extraction efficacy. Measurement of native (biologically active) TGF-ß can be performed by omitting the activation step and preferably by using an assay that is based on receptor binding. However, the amount of TGF-ß that is available for binding to the receptor will depend in a nonlinear way on the composition of the system of binding proteins, receptor, and TGF-ß, i.e., on sample dilution. For a realistic approximation of biologically active TGF-ß, undiluted samples are therefore preferred.
The assays tested should not be used to measure TGF-ß concentration in serum samples. The highly significant correlation to platelet counts strongly indicates that a major part of the TGF-ß detected in serum is released from platelets during the clotting process. Therefore, investigations focused on determining the diagnostic value of circulating TGF-ß in serum should be viewed critically. This topic has been addressed extensively by Wakefield et al. (20), who recommend measurement of TGF-ß in plasma and subsequent correction for platelet degranulation via measurement of platelet factor 4.
2M and possibly several other binding proteins might play a specific
physiological role as scavengers for a whole range of various cytokines
in addition to TGF-ß. Therefore, we assume that the effects presented
here should be considered as having general importance for immunoassay
design.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2M, 2-macroglobulin. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  H. Tohyama, K. Yasuda, H. Uchida, and J. Nishihira The responses of extrinsic fibroblasts infiltrating the devitalised patellar tendon to IL-1{beta} are different from those of normal tendon fibroblasts J Bone Joint Surg Br, September 1, 2007; 89-B(9): 1261 - 1267. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Grainger TGF-{beta} and atherosclerosis in man Cardiovasc Res, May 1, 2007; 74(2): 213 - 222. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A M Hammad, H M Youssef, and M M El-Arman Transforming growth factor beta 1 in children with systemic lupus erythematosus: a possible relation with clinical presentation of lupus nephritis Lupus, September 1, 2006; 15(9): 608 - 612. [Abstract] [PDF]
|
|
|
|
 |
|
 T. Kawakita, E. M. Espana, H. He, R. Smiddy, J.-M. Parel, C.-Y. Liu, and S. C. G. Tseng Preservation and Expansion of the Primate Keratocyte Phenotype by Downregulating TGF-{beta} Signaling in a Low-Calcium, Serum-Free Medium Invest. Ophthalmol. Vis. Sci., May 1, 2006; 47(5): 1918 - 1927. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Zimmermann, J. Koenig, J. Zingsem, V. Weisbach, E. Strasser, J. Ringwald, and R. Eckstein Effect of Specimen Anticoagulation on the Measurement of Circulating Platelet-Derived Growth Factors Clin. Chem., December 1, 2005; 51(12): 2365 - 2368. [Full Text] [PDF]
|
|
|
|
 |
|
 T. Kawakita, E. M. Espana, H. He, A. Hornia, L.-K. Yeh, J. Ouyang, C.-Y. Liu, and S. C. G. Tseng Keratocan Expression of Murine Keratocytes Is Maintained on Amniotic Membrane by Down-regulating Transforming Growth Factor-{beta} Signaling J. Biol. Chem., July 22, 2005; 280(29): 27085 - 27092. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-A. Gomez, X. Molero, E. Vaquero, A. Alonso, A. Salas, and J.-R. Malagelada Vitamin E attenuates biochemical and morphological features associated with development of chronic pancreatitis Am J Physiol Gastrointest Liver Physiol, July 1, 2004; 287(1): G162 - G169. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Naito, T. Masaki, D. J. Nikolic-Paterson, C. Tanji, N. Yorioka, and N. Kohno Angiotensin II induces thrombospondin-1 production in human mesangial cells via p38 MAPK and JNK: a mechanism for activation of latent TGF-{beta}1 Am J Physiol Renal Physiol, February 1, 2004; 286(2): F278 - F287. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. M. Omer, J. B. de Souza, P. H. Corran, A. A. Sultan, and E. M. Riley Activation of Transforming Growth Factor {beta} by Malaria Parasite-derived Metalloproteinases and a Thrombospondin-like Molecule J. Exp. Med., December 15, 2003; 198(12): 1817 - 1827. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. G. Hobson, M. DeWing, H. S. Ho, B. M. Wolfe, K. Cho, and D. G. Greenhalgh Expression of Transforming Growth Factor {beta}1 in Patients With and Without Previous Abdominal Surgery Arch Surg, November 1, 2003; 138(11): 1249 - 1252. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Daddi, T. Suda, F. D'Ovidio, S. A. Kanaan, T. Tagawa, K. Grapperhaus, B. D. Kozower, J. H. Ritter, N. S Yew, T. Mohanakumar, et al. Recipient intramuscular cotransfection of naked plasmid transforming growth factor {beta}1 and interleukin 10 ameliorates lung graft ischemia-reperfusion injury J. Thorac. Cardiovasc. Surg., August 1, 2002; 124(2): 259 - 269. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Porreca, C. Di Febbo, G. Baccante, M. Di Nisio, and F. Cuccurullo Increased transforming growth factor-beta1 circulating levels and production in human monocytes after 3-hydroxy-3-methyl-glutaryl-coenzyme a reductase inhibition with pravastatin J. Am. Coll. Cardiol., June 5, 2002; 39(11): 1752 - 1757. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. YEVDOKIMOVA, N. A. WAHAB, and R. M. MASON Thrombospondin-1 Is the Key Activator of TGF-{beta}1 in Human Mesangial Cells Exposed to High Glucose J. Am. Soc. Nephrol., April 1, 2001; 12(4): 703 - 712. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. M. Coupes, S. Williams, I. S. D. Roberts, C. D. Short, and P. E. C. Brenchley Plasma transforming growth factor {beta}1 and platelet activation: implications for studies in transplant recipients Nephrol. Dial. Transplant., February 1, 2001; 16(2): 361 - 367. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Hoying, M. Yin, R. Diebold, I. Ormsby, A. Becker, and T. Doetschman Transforming Growth Factor beta 1 Enhances Platelet Aggregation through a Non-transcriptional Effect on the Fibrinogen Receptor J. Biol. Chem., October 22, 1999; 274(43): 31008 - 31013. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Erren, H. Reinecke, R. Junker, M. Fobker, H. Schulte, J. O. Schurek, J. Kropf, S. Kerber, G. Breithardt, G. Assmann, et al. Systemic Inflammatory Parameters in Patients With Atherosclerosis of the Coronary and Peripheral Arteries Arterioscler Thromb Vasc Biol, October 1, 1999; 19(10): 2355 - 2363. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
