0 (2007) 86–93

Clinical Biochemistry 4

Immunoassay of serine-phosphorylated isoform of insulin-like growth factor(IGF) binding protein (IGFBP)-1

Javad Khosravi a,c,⁎, Radha G. Krishna b, Umesh Bodani b, Anastasia Diamandi a,Najmuddin Khaja b, Bhanu Kalra b, Ajay Kumar b

a Diagnostic Systems Laboratories (Canada) Inc., Toronto, Ontario, Canada M5G 1L7b Diagnostic Systems Laboratories, Inc., Webster, TX, USA

c Department of Laboratory Medicine and Pathobiology, Faculty of Medicine, University of Toronto and Mount Sinai Hospital, Toronto, Ontario, Canada

Received 7 February 2006; received in revised form 29 June 2006; accepted 21 July 2006Available online 10 August 2006

Abstract

Objectives: Development of an ELISA for phosphorylated isoform of IGFBP-1. Serine phosphorylation is an important regulator of IGFBP-1bioactivity, but specific immunoassays for its measurement are currently lacking.

Design and methods: Assay design was based on a novel approach of first capturing the phosphorylated and non-phosphorylated IGFBP-1 byan anti-IGFBP-1 antibody and then selectively detecting the phosphorylated form by an anti-phosphoserine antibody. Method developmentinvolved pair-wise evaluation of the candidate antibodies and determinations of analytical performance and specificity. Specificity was monitoredby reactivity with dephosphorylated IGFBP-1, with antibodies against other phosphorylated residues that are not expressed, and by comparativeanalysis of sample containing different IGFBP-1 phosphorylation profile.

Results: Analytical evaluation demonstrated acceptable performance; detection limit 0.3 μg/L, dynamic range 1.56–100 μg/L; intra- and inter-assay CVs 2.1–8.6%; mean recovery (±SD) 97.8±9.2%, and mean recovery of sample dilution 93.4±6.0%. The phosphorylated and total IGFBP-1 medians in non-pregnant adult serum, which mostly contain the highly phosphorylated isoform, were 11.9 and 18.6 μg/L, respectively, and thesample values were tightly correlated (r=0.99). As expected, the corresponding medians in 1st trimester (17.4 and 63.0 μg/L) and 2nd trimester(30.9 and 75.8) samples with altered IGFBP-1 phosphorylation were significantly different (p<0.001). Similarly, a fraction (1.29%) of totalIGFBP-1 (13.3 mg/L) in amniotic fluids was found to be phosphorylated (0.172 mg/L). There was no reactivity with dephosphorylated IGFBP-1.

Conclusions: The present ELISA is highly specific for the phosphorylated isoform of IGFBP-1 and its availability should help expedite furtherinvestigations of IGFBP-1 phosphorylation.© 2006 The Canadian Society of Clinical Chemists. All rights reserved.

Keywords: Growth hormone; Insulin-like growth factors; Phosphorylation; IGFBP

Introduction

Insulin-like growth factors (IGF-I and IGF-II) belong to afamily of peptides that mediate a broad spectrum of growthhormone-dependent as well as-independent mitogenic andmetabolic actions essential for cell growth and development[1–4]. Unlike most peptide hormones, IGFs in circulation and

⁎ Corresponding author. Diagnostic Systems Laboratories (Canada) Inc.,311-101 College Street, MaRS Centre, South Tower, Toronto, Ontario, CanadaM5G 1L7.

E-mail address: jkhosravi@101college.net (J. Khosravi).

0009-9120/$ - see front matter © 2006 The Canadian Society of Clinical Chemistsdoi:10.1016/j.clinbiochem.2006.07.004

in other physiological fluids are associated with a group of high-affinity IGF binding proteins (IGFBPs) that specifically bindand modulate their bioactivity. Six structurally homologousIGFBPs with distinct molecular size, hormonal control, andtissue expression and functions have been identified [4–6].

IGFBP-1 is a 25-kDa protein expressed and secreted by avariety of cell types, including hepatocytes, ovarian granulosacells, and decidualized endometrium [7–9]. IGFBP-1 is presentin serum, is the predominant binding protein in amniotic fluid,and is one of the major IGFBP in fetal and maternal circulation[7,10,11]. In normal individuals, circulating IGFBP-1 concen-trations fluctuate within a few hours by 10-folds or more in

. All rights reserved.

87J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

response to changes in insulin concentrations [7]. Although thebiological relevance of IGFBP-1 has not been fully defined, itappears to play a significant role in multiple pathophysiologicalconditions including regulation of fetal growth and develop-ment [4–7,9–15].

IGFBP-1 is reportedly capable of both inhibition as well asaugmentation of the IGFs' actions [4,6,7]. The dual function-alities of IGFBP-1 have been partly explained by its post-translational serine phosphorylation, which alters IGFBP-1affinity for the IGFs by several folds [4,7,16], thereby affectingits capacity to regulate IGF bioavailability. Up to five IGFBP-1variants differing only in their degree of phosphorylation havebeen identified [16]. Various cell types such as Hep G2,decidual, and liver cells were found to secrete predominantlyphosphorylated forms, whereas amniotic fluid and fetal serumcontained substantial amounts of non-phosphorylated and lesserphosphorylated variants [4–7,16,17]. In contrast, in the sera ofnormal adults a predominantly phosphorylated form of IGFBP-1has been detected [18]. Dephophorylation of IGFBP-1 has beenproposed as a possible mechanism for increasing IGF bioavail-ability. The non-phosphorylated IGFBP-1 isoforms havereportedly four- to six-fold lower binding affinity for IGF-I ascompared with the phosphorylated variants [16]. The stimula-tion and inhibition of the IGF actions have been found inassociation with the non-phosphorylated and phosphorylatedIGFBP-1, respectively [4].

Despite the biological importance of altered phosphoryla-tion, which occurs in different biological fluids and in responseto different pathophysiological stimuli [4,7,16–18], immunoas-says for specific quantification of phosphorylated isoform ofIGFBP-1 are currently unavailable. In addition, immunoreac-tivity of IGFBP-1 is significantly altered in response to changesin its state of phosphorylation [18,19]. To overcome theproblem, we previously reported development of immunoas-says for total IGFBP-1 that is unaffected by altered state ofIGFBP-1 phosphorylation as well as immunoassay withpreferential (not absolute) reactivity for the non- or lesserphosphorylated variants of IGFBP-I [19]. Because of thelimitations of assay specificity, particularly for applicationsinvolving assessment of differential phosphorylation, we heredescribe, for the first time, the development of a novel ELISAfor specific determination of phosphorylated isoforms ofIGFBP-1. The distinguishing features of the proposed assaydesign are based on dual recognition partners, involving a well-characterized anti-IGFBP-1 capture antibody [19] combinedwith a generic antibody that specifically binds to thephosphorylated serine residues that might be expressed on thesurface of IGFBP-1. The analytical performance characteristicsand preliminary clinical data are presented.

Materials and methods

Samples

Serum samples from non-pregnant females (n=29, age 17–48 years), and from first (n=38, age 22–44 years) and second(n=29, age 23–42 years) trimester pregnancies were obtained

from Lenetix Medical Screening Laboratory Inc. (New York,New York). These specimens were residuals from routine orresearch tests samples. After collection, blood samples wereallowed to clot and were then separated. The residuals werestored at −20°C and used for the present studies within3 months. Amniotic fluids (n=20) from second trimesterpregnancies (15–18 week gestations) were provided by Dr. E. P.Diamandis (Mount Sinai Hospital, Toronto, Ontario). Thesamples were residuals from routine clinical test samples andwere stored at −70°C for less than 4 months before use. Allsamples were from individuals with informed consent.

ReagentsHorseradish peroxidase (HRP) was obtained from Scripps

Labs., San Diego, CA. Tetramethylbenzidine (TMB/H2O2)microwell peroxidase substrate system was from NeogenCorporation, Lexington, KY. Sulfosuccinimidyl 4-(N-maleimi-domethyl)cyclohexane-1-carboxylate (sulfo-SMCC) and 2-iminothiolane were purchased from Pierce, Rockford, IL.Enzyme immunoassay-grade alkaline phosphatase (ALP) wasobtained from Boehringer Mannheim, Indianapolis, IN. Allother chemical reagents were of highest quality and wereobtained from Sigma Chemical, St. Louis, MO or Amresco,Solon, OH. Microtitration strips and frames were products ofGreiner International, Germany.

AntibodiesFive different anti-IGFBP-1 mouse monoclonal antibodies

(mAb) and an affinity-purified goat polyclonal anti-IGFBP-1antibody were obtained from Diagnostic Systems Laboratories(DSL;Webster, TX). Among these antibodies, a mAb unaffectedby the state of IGFBP-1 phosphorylation or IGFBP-1 binding toIGF-I has been previously identified and incorporated into ahighly specific immunoassay for total IGFBP-1 that is nowcommercially available (DSL, Active Total IGFBP-1 ELISA).Mouse mAbs against the various phosphorylated amino acidresidues were purchased from commercial sources. Anti-phosphoserine antibody was obtained from BD Biosciences,San Jose, CA. Anti-phosphotyrosine was from Upstate labora-tories, Lake Placid, NY. Human IGFBP-I purified from humanamniotic fluid [20] was obtained from DSL (Webster, TX).

Phosphorylated IGFBP-1 (phIGFBP-1) ELISAThe assay buffer was 0.025 mol/L Tris–maleate, pH 6.1,

containing 9 g NaCl, 20 g bovine serum albumin (BSA), 0.5 mLTween-20, and 2.5 mL proclin 300 per liter. The compositionsof the coating buffer, blocking buffer, and wash solution, wereas described previously [21,22]. Procedures for antibodycoating to microwells as well as detection antibody conjugationto HRPO have been also described [21,22].

As previously reported, IGFBP-1 in non-pregnant adultsamples is known to be highly phosphorylated [18,19]. Becauseof unavailability of phIGFBP-1 reference material, serumsamples from non-pregnancy adults were assayed with thepresent phIGFBP-1 ELISA. A serum pool, prepared bycombining aliquots from 10 different samples with the highestphIGFBP-1 immunoreactivity, was then assayed with the DSL

88 J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

total IGFBP-1 ELISA that measures the phosphorylated as wellas the non-phosphorylated IGFBP-1 with equal potency [19].The obtained value was then assigned to the serum pool and theserum pool diluted appropriately into newborn calf serum toproduce the desired phIGFBP-1 standards. The quality controlsamples were serum samples with different concentrations ofphIGFBP-1.

phIGFBP-1 ELISA protocolThe phIGFBP-1 ELISA is based on conventional immu-

noassay protocols after determination of optimal assay condi-tions. The assay involves addition of standards, samples orcontrols (0.02 mL), and the assay buffer (0.10 mL) in duplicateto anti-IGFBP-1 antibody pre-coated wells, followed by 1 hincubation at room temperature with continuous shaking. Thewells are then washed five times and incubated with 0.10 mL/well of the detection anti-phosphoserine antibody conjugated toHRPO for 30 min at room temperature. After an additionalwashing step, the wells are incubated with 0.1 mL/well TMB/H2O2 substrate solution and incubated for 10 min as above.Stopping solution (0.1 mL of 0.2 mol/L sulfuric acid in dH2O) isthen added and absorbance is measured by dual wavelengthmeasurement at 450 nm with background wavelengthcorrection set at 620 nm. ELISA absorbance measurementswere performed with the Labsystems Multiskan Multisoftmicroplate reader (Labsystems, Helsinki, Finland).

phIGFBP-1 ELISA validation proceduresThe lower limit of detection (sensitivity) was determined by

interpolating the mean plus 2SD of 12 replicate measurementsof the zero calibrator (NBCS). The intra-assay CVs weredetermined by replicate analysis (n=12) of four samples atdifferent concentrations in one run and inter-assay CVs byduplicate measurement of samples in 9–12 separate assays.Recovery was assessed by adding 25 μL of high phIGFBP-1sample to 225 μL of three low phIGFBP-1 samples andanalyzing the supplemented and un-supplemented samples.Percent recovery was determined by comparing the amount ofadded phIGFBP-1 with the amount measured after subtractingthe endogenous phIGFBP-1 concentrations. Linearity wastested by analyzing three serum samples serially diluted (2- to16-fold) in the zero calibrator of the assay.

Other assaysTotal IGFBP-1 and non-phosphorylated IGFBP-1 were

assayed as previously reported [19], using DSL Total andNon-phosphorylated IGFBP-1 ELISAs. Both of these assaysare based on conventional methodology involving a commonanti-IGFBP-1 capture antibody combined with a polyclonal ormonoclonal for detecting total IGFBP-1 or non-phosphorylatedIGFBP-1 concentrations, respectively. As reported previously[19], the Total IGFBP-1 ELISA is unaffected by changes in thephosphorylation state of IGFBP-1, thus measuring the non-phosphorylated and differentially phosphorylated IGFBP-1isoforms with similar potency. The Non-PhosphorylatedIGFBP-1 ELISA has predominant specificity for the non- andlesser phosphorylated IGFBP-1 variants [19].

Dephosphorylation of IGFBPs

Dephosphorylation of IGFBP-1 was achieved by serumsample pre-treatment with alkaline phosphatase (ALP) asdescribed previously [19]. Briefly, 200 units of ALP dissolvedin 10 μL of 1 mol/L diethanolamine (DEA), pH 9.5, containing0.5 mmol/L MgCl2 was added to a 200-μL aliquot of thesample, mixed, and incubated at room temperature for 2 h. Theuntreated control aliquot received 10 μL of the DEA buffer andwas similarly incubated. ALP-treated and untreated sampleswere then analyzed.

Data analysisELISA data were analyzed with data reduction package

included with the instrumentation, using cubic spline(smoothed) curve fit. Descriptive statistics are presented asthe mean, median, and standard deviation unless otherwisespecified. Linear regression analysis was performed by theleast-squares method and correlation coefficients weredetermined by the Pearson method. Analysis of differencesin median concentrations of the various biomarkers indifferent sample groups was performed by Mann–WhitneyRank Sum Test. The plots and the indicated statisticalanalysis were performed by SigmaPlot and SigmaStat soft-ware (Systat Software Inc, Point Richmond, CA 94804-2028), respectively.

Results

Antibody evaluation and assay design

We evaluated a panel of anti-IGFBP-1 monoclonal andpolyclonal antibodies in pair-wise combination with genericantibodies recognizing phosphorylated amino acid residues.The primary objective was to assess identification of potentialantibody partners for specific quantification of the phosphory-lated isoforms of IGFBP-1. The effort identified a novelapproach to selective immunoassay of IGFBP-1 phosphoiso-forms, which demonstrated general applicability to immunoas-say of phosphoforms of other IGFBPs such as IGFBP-5 andIGFBP-3 (data not shown). In a two-step ELISA configuration,the best standard curve characteristics (signal-to-dose ratio,non-specific binding signal, and dynamic range) were obtainedby first capturing IGFBP-1 by a well characterized IGFBP-1 mAb [19] and then detecting the captured IGFBP-1phosphoisoforms by an HRPO-labeled anti-phosphoserineantibody. As per conventional assays [23–25], acceptableanalytical performance characteristics were obtained with acoating antibody concentration of 5 mg/L (500 ng/0.1 mL perwell), a detection antibody concentration of ∼0.1–0.25 mg/L(10–25 ng/0.1 mL per well), a sample size of 0.02 mL, a first-and second-step room temperature incubation of 1 h and30 min, respectively, and a 10-min substrate development step.A typical standard curve of the Phosphorylated IGFBP-1(phIGFBP-1) ELISA is shown in Fig. 1. The analyticalperformance characteristics of the assay are summarized inTable 1. The assay was unaffected by addition of haemoglobin

Fig. 1. phIGFBP-1 ELISA. Typical standard curve and assay specificity. Ahuman serum pool appropriately diluted in the standard matrix was assayedusing the detection anti-phosphoserine mAb or an anti-phosphotyrosine mAb asshown. The serum pool dephosphorylated by pre-treatment with alkalinephosphatase was also similarly diluted and assayed using the detection anti-phosphoserine mAb.

89J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

(up 1 g/L), triolein (up to 5 g/L), and 50 mg/L unconjugatedbilirubin to serum samples.

Specificity of phIGFBP-1 ELISASpecificity was assessed by evaluating assay response to an

anti-phosphotyrosine antibody as well as to the anti-phospho-serine detection antibody before and after IGFBP-1 depho-sphorylation by alkaline phosphatase [19]. As shown in Fig. 1,there was no detectable reactivity with the anti-phosphotyrosineantibody, and the expected binding to the anti-phosphoserinedetection antibody was completely diminished in response toIGFBP-1 dephosphorylation.

phIGFBP-1 in physiological fluids

Non-pregnancy samples. IGFBP-1 in non-pregnancy adultsamples is normally highly phosphorylated. Accordingly, in a setof non-pregnancy samples (n=29), the median phIGFBP-1(11.9 μg/L) by the present assay was not statistically different(p=0.074) from the total IGFBP-1median (18.6μg/L) quantifiedby the DSLTotal IGFBP-1 ELISA, but was significantly higher

Table 1phlGFBP-1 ELISA validation data

Assay parameter

Detection limit (μg/L) 0.30Standard range (μg/L) 1.56–100Intra-assay CV, % 2.1–8.6Inter-assay CV, % 4.0–7.3Recovery of additions, % 97.8±9.2Recovery after dilution, % 93.4±6.0

(p<0.001) than the concentration (0.81 μg/L) detected by theDSL Non-phosphorylated IGFBP-1 method (Fig. 2). Similarly,in comparative analysis, results by the phIGFBP-1 ELISAparalleled tightly those measured by the total assay (Fig. 3).Although significant correlation was also observed in compari-son of the phosphorylated vs. the non-phosphorylated IGFBP-1concentrations, the corresponding data showed considerablescattering around the regression line (Fig. 3).

Pregnancy samples. The phIGFBP-1 median measured in apanel of first (17.4 μg/L; n=38) and second (30.9 μg/L; n=29)trimester pregnancy serum samples were higher than themedian concentration (11.9 μg/L) detected in the above non-pregnancy samples by about 1.5- and 2.6-fold, respectively.The observed relative increases in the phIGFBP-1 medians inthe pregnancy sample groups vs. the non-pregnancy samples

Fig. 2. IGFBP-1 in non-pregnancy and pregnancy samples. Phosphorylated-IGFBP-1 (phIGFBP-1) as well as total and non-phosphorylated IGFBP-1 (non-phIGFBP-1) concentrations were measured in non-pregnancy (non preg., n=29)and in first trimester (1st trimester, n=38) and second trimester pregnancy (2ndtrimester, n=29) samples. The median and the 95% confidence intervals as wellas the p values for comparisons of the phIGFBP-1 median concentrations in thethree sample group vs. the corresponding medians obtained for the non-phIGFBP-1 and total IGFBP-1 concentrations are shown.

Fig. 3. Comparative analysis of phIGFBP-1 in non-pregnancy samples.Correlations of phIGFBP-1 measured by the present assay in non-pregnancyadult serum samples vs. total IGFBP-1 and non-phIGFBP-1 measured by DSLELISAs are shown. Values are the means of duplicate measurements.

Fig. 4. Comparative analysis of phIGFBP-1 in pregnancy samples. Correlationsof phIGFBP-1 measured by the present assay in the combined first and secondtrimester pregnancy samples vs. total IGFBP-1 and non-phIGFBP-1 measuredby DSL ELISAs are shown. Values are the means of duplicate measurements.

90 J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

appear to track a relatively parallel and significant increase(p<0.001) in the total IGFBP-1 medians (63.0 and 75.8 μg/L)quantified by the DSL Total IGFBP-1 ELISA (Fig. 2). In thesefirst and second trimester pregnancy samples, there were alsosignificant relative increases (p<0.001) in the correspondingnon-phosphorylated IGFBP-1 median concentrations (17.1 μg/L and 26.0 μg/L). Regression analysis of the phosphorylatedvs. total (r=0.61) or vs. the non-phosphorylated (r=0.41)IGFBP-1 concentrations in the combined first and secondtrimester pregnancy samples yielded relatively poor correla-tions and wide scattering of the regression data (Fig. 4). Thelatter is expected as the IGFBP-1 profile of normal adult serumis changed considerably during pregnancy when non-phos-phorylated and lesser-phosphorylated variants of IGFBP-1 arealso expressed [18,19].

Amniotic fluid. In amniotic fluid samples (n=20), thephIGFBP-1 median concentration (0.172 mg/L) was signifi-cantly (p<0.001) lower than the total (13.3 mg/L) or the non-phosphorylated (11.8 mg/L) medians (Fig. 5). This is alsoexpected as IGFBP-1 in amniotic fluid is predominantly non-phosphorylated [7,19].

Discussion

The reversible phosphorylation of proteins by post-transla-tional modification is an essential and almost universalmechanism of protein activation–deactivation, responsible forregulating nearly the entire cell signalling pathways andultimately all biological functions [26,27]. Accordingly,phosphorylation of IGFBPs has been recognized as animportant alternative mechanism by which bioavailability ofthe IGFs might be modulated [4,16,28–34]. In addition tonormal pregnancy, studies of IGFBP-1 have so far identifiedsignificant changes in its state of phosphorylation in subjectswith Larone syndrome [18,29] as well as in relation to pre-eclampsia [30], postnatal growth restriction [13], regulation ofbiosynthesis of collagen and glycosaminoglycan [30], and inrelation to the development of cardiovascular disease andvascular complications of type 2 diabetes [31]. PhosphorylatedIGFBP-1 concentrations are reportedly reduced in the cordblood of diabetic pregnancies [35], and its determination hasbeen proposed as a potential means of predicting cervicalripeness [36] or occurrence of preterm labor [37]. Recent

Fig. 5. IGFBP-1 in amniotic fluid. Phosphorylated-IGFBP-1 (phIGFBP-1) aswell as total and non-phIGFBP-1 concentrations was measured in amniotic fluid(n=20). The median and the 95% confidence intervals as well as the p values forcomparisons of the phIGFBP-1 median concentrations vs. the correspondingmedians obtained for the non-phIGFBP-1 and total IGFBP-1 concentrations areshown.

91J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

evidence appears to support differential regulation of decidualIGFBP-1 phosphorylation by steroid hormones and potentialeffect of the latter on enhanced expression of decidualmetalloproteinases that could specifically cleave IGFBP-1[38].

The analytical relevance of phosphorylation relates to itsconformational altering effects [39] and thus, the relativelevels of immunoreactivity that could be detected [18,19].Altered phosphorylation may result in significant variationsin the detectable immunoreactivity of IGFBP-1, which isparticularly important considering the variable expression ofthe various IGFBP-1 isoforms in different biological fluids aswell as potential changes in its circulating phosphorylationprofile [4,13,14,16–18,30,31,35–37]. The latter might be themain reason for the reported variability of up to 15-fold inthe mean concentrations of serum IGFBP-1 detected bydifferent IGFBP-1 antibodies [18,19]. Because of theimportance of accurate quantification of IGFBP-1, wepreviously reported development of an immunoassay fortotal IGFBP-1 that is unaffected by its state of phosphoryla-tion as well as an immunoassay with preferential specificityfor the non- or lesser-phosphorylated isoforms of IGFBP-1[19]. Although highly valuable for accurate determination oftotal IGFBP-1, the combined methodology is disadvantagedby its inability to accurately quantify phIGFBP-1 concentra-tions. Theoretically, the most obvious approach may involveselection of anti-IGFBP-1 antibodies that are highly specificfor its phosphorylated variants, but the approach in ourexperience has been difficult to implement due mainly to thesignificant effect of altered phosphorylation on immunor-eactivity [18,19]. The latter is consistent with the reportedlack of specificity of the available IGFBP-1 phosphoisoformsimmunoassays that are neither specific for the phosphory-lated or for the non-phosphorylated IGFBP-1 variants[19,35].

The present methodology involves an anti-IGFBP-1 specificcapture antibody that is unaffected by the state of IGFBP-1phosphorylation [19] combined with an antibody specifically

recognizing the serine phosphorylated residues that might beexpressed. This novel approach of isolating total IGFBP-1prior to detecting the phosphorylated moieties is capable ofaccurate quantification of phIGFBP-1 within the broadercontext of changes in its total concentrations. As presented inthis report, evaluation of the method demonstrated acceptableanalytical performance characteristics. Assay specificity wasbased on the design considerations confirmed by demonstrat-ing no reactivity with antibodies against other phosphorylatedresidues (e.g., phosphotyrosine) that are not expressed onIGFBP-1 [16] or by the diminished detectability of thephIGFBP-1 after its dephosphorylation by alkaline phospha-tase [19].

In accordance with the reported distribution of IGFBP-1phosphoisoforms in various biological fluids [18], the presentELISA detected the highest concentrations of phIGFBP-1 innon-pregnancy adult serum samples, while detecting relativelylow concentrations in amniotic fluids. Reflecting assayspecificity, the observed concordances between the phos-phorylated and the total IGFBP-1 in the non-pregnant adultsamples might be expected as such samples reportedlycontain the highly phosphorylated variants of IGFBP-1[18,19]. The latter was further confirmed by the detectionin these samples of considerably low levels of immunor-eactivity by the DSL Non-phosphorylated IGFBP-1 ELISA(Fig. 2). Accordingly in the non-pregnancy adult samples, thesum of the phosphorylated and non-phosphorylated IGFBP-1(y) was nearly identical to the corresponding total IGFBP-1(x), yielding y=0.98x−2.7 (r=0.99). In contrast, correlationsin pregnancy samples were relatively poor and demonstratedwide scattering of the corresponding measures of thephosphorylated vs. the total amounts or the non-phosphory-lated IGFBP-1. In these samples, combining the phosphory-lated and non-phosphorylated concentrations did not improvecomparison vs. the total concentrations of IGFBP-1(data notshown). This might be expected as the DSL Non-phosphory-lated IGFBP-1 ELISA has preferential specificity for the non-phosphorylated as well as for the lesser phosphorylatedisoforms of IGFBP-1 that are variably and differentiallyexpressed in pregnancy samples [18,19]. Detection of little orno immunoreactivity in amniotic fluid is also consistent withthe specificity of the present assay as IGFBP-1 in amnioticfluid is known to be predominantly non-phosphorylated[7,19] and as such would be undetectable by the presentELISA.

Immunoassay of IGFBP-1 is further complicated by acuteregulation of its circulating concentrations by insulin [7,19]. Innormal individuals, serum concentrations of IGFBP-1 fallrapidly after a meal with increases occurring during fasting.Because of potential difficulties in obtaining fasting samples,particularly during pregnancy, we propose that phIGFBP-1measures in non-fasting samples might be best interpreted inrelation to the corresponding total IGFBP-1 measured by areliable immunoassay. The proposed “ratio” determinationmight be then useful for correcting or normalizing the effectof the significant and acute daily fluctuations in the totalIGFBP-1 concentrations that occurs in relation to meals [7]. The

92 J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

recommended approach might be particularly differentiating incomparative assessment of conditions associated with alteredIGFBP-1 phosphorylation and where a divergence in thephosphorylated vs. the total IGFBP-1 concentrations might beexpected.

In summary, we describe the first report on development andvalidation of a highly specific and novel approach toimmunoassay of phIGFBP-1. This generally applicableapproach to direct immunoassay of phosphorylated isoformsof other IGFBPs or proteins of interest involves a two-stepimmunoreaction in which IGFBP-1 is first captured by an anti-IGFBP-1 antibody and the captured phosphoisoforms are thendetected by an antibody specific for the phosphorylated serineresidues. Availability of the method should facilitate investiga-tions of the pathophysiological roles and potential clinicalrelevance of IGFBP-1 phosphorylation.

References

[1] Rotwein P. Structure, evolution, expression and regulation of insulin-likegrowth factors I and II. Growth Factors 1991;5:3–18 [Review].

[2] Daughaday WH, Rotwein P. Insulin-like growth factors I and II. Peptide,messenger ribonucleic acid and gene structures, serum and tissueconcentration. Endocr Rev 1989;10:68–91.

[3] LeRoith D, Adamo M, Werner H, Roberts CT. Insulin-like growthfactors and their receptors as growth regulators in normal physiologyand pathological states. Trends Endocrinol Metab 1991;2:134–9[Review].

[4] Jones JI, Clemmons DR. Insulin-like growth factors and their bindingproteins: biological actions. Endocr Rev 1995;16:3–34.

[5] Cohen P, Rosenfeld RG. Physiologic and clinical relevance of the insulin-like growth factor binding proteins. Curr Opin Pediatr 1994;6:462–7[Review].

[6] Rajarm S, Baylink DJ, Mohan S. Insulin-like growth factor-bindingproteins in serum and other biological fluids: regulation and functions.Endocr Rev 1997;18:801–31.

[7] Lee PDK, Conover CA, Powell DR. Regulation and function of insulin-like growth factor binding protein-1. Proc Soc Exp Biol Med 1993;204:4–29.

[8] Suikkari A-M, Jalkanen J, Koistinen R, Butzow R, Ritvos O, Ranta T, et al.Human granulosa cells synthesize low molecular weight insulin-likegrowth factor-binding protein. Endocrinology 1989;124:1088–90.

[9] Giudice LC. Insulin-like growth factors and ovarian follicular develop-ment. Endocr Rev 1992;13:641–69.

[10] Rutanen E-M, BohnH, SeppalaM. Radioimmunoassay of placental protein12: levels in amniotic fluid, cord blood, and serum of healthy adults,pregnant women, and patients with trophoblastic disease. Am J ObstetGynecol 1982;144:460–3.

[11] Hills FA, English J, Chard T. Circulating levels of IGF-I and IGF-bindingprotein-1 throughout pregnancy: relation to birth weight and maternalweight. J Endocrinol 1996;148:303–9.

[12] Verhaeghe J, Billen J, Giudice LC. Insulin-like growth factor-bindingprotein-1 in umbilical artery and vein of term fetuses with signs suggestiveof distress during labor. J Endocrinol 2001;170:585–90.

[13] Kajantie E. Insulin-like growth factor (IGF)-I, IGF binding protein(IGFBP)-3, phosphoisoforms of IGFBP-1 and postnatal growth in very-low-birth-weight infants. Horm Res 2003;60:124–30.

[14] Sakai K, D'Ercole AJ, Murphy LJ, Clemmons DR. Physiologicaldifferences in insulin-like growth factor binding protein-1 phosphorylationin IGFBP-1 transgenic mice. Diabetes 2001;50:32–8.

[15] Rosenzweig SA. What's new in the IGF-binding proteins? GH IGF Res2004;14:329–36.

[16] Jones JI, D'Ercole AJ, Camacho-Hubner C, Clemmons DR. Phosphoryla-tion of insulin-like growth factor (IGF)-binding protein 1 in cell culture

and in vivo: effects on affinity for IGF-I. Proc Natl Acad Sci U S A1991;88:7481–5.

[17] Koistinen R, Angervo M, Leinonen P, Hakala T, Seppala M. Phosphoryla-tion of insulin-like growth factor-binding protein-1 increases in humanamniotic fluid and decidua from early to late pregnancy. Clin Chem Acta1993;215:189–99.

[18] Westwood M, Gibson JM, Davies AJ, Young RJ, White A. Thephosphorylation pattern of insulin-like growth factor-binding protein-1in normal plasma is different from that in amniotic fluid and changesduring pregnancy. J Clin Endocrinol Metab 1994;79:1735–41.

[19] Khosravi MJ, Diamandi A, Mistry J. Immunoassay of insulin-like growthfactor binding protein-I. Clin Chem 1997;43:523–32.

[20] Powell DR, Lee PDK, Shively JE, Eckenhausen M, Hintz RL. Method forpurification of an insulin-like growth factor-binding protein produced byhuman HEP-G2 hepatoma cells. J Chromatogr, B, Biomed Appl1987;420:163–70.

[21] Khosravi MJ, Papanastasiou-Diamandi A, Mistry J. Ultrasensitiveimmunoassay for prostate-specific antigen based on conventional colori-metric detection. Clin Biochem 1995;28:407–14.

[22] Khosravi MJ, Diamandi A, Mistry J, Lee PDK. Noncompetitive ELISAfor human serum insulin-like growth factor-I. Clin Chem 1996;42:1147–54.

[23] Khosravi J, Diamandi A, Mistry J, Krishna RG. The high molecular weightinsulin-like growth factor binding protein complex: epitope mapping,immunoassay, and preliminary clinical evaluation. J Clin EndocrinolMetab 1999;84:2826–33.

[24] Diamandi A, Mistry J, Krishna RG, Khosravi J. Immunoassay of insulin-like growth factor binding protein-3: new means to quantifying IGFBP-3proteolysis. J Clin Endocrinol Metab 2000;85:2327–33.

[25] Khosravi MJ, Diamandi A, Mistry J, Krishna RG, Khare A. Acid-labilesubunit of human insulin-like growth factor binding protein complex:measurement, molecular, and clinical evaluation. J Clin Endocrinol Metab1997;82:3944–51.

[26] Graves JD, Krebs EG. Protein phosphorylation and signal transduction.Pharmacol Ther 1999;82:111–21.

[27] Garcia BA, Shabanowitz J, Hunt DF. Analysis of protein phosphorylationby mass spectrometry. Methods 2005;35:256–64.

[28] Kelly KM, Oh Y, Gargosky SE, Gucev Z, Matsumoto T, Haw V, et al.Insulin-like growth factor binding proteins (IGFBPs) and their regulatorydynamics. Int J Biochem Cell Biol 1996;28:619–37.

[29] Westwood M, Gibson JM, Williams AC, Clayton PE, Hamberg O,Flyvbjerg A, et al. Hormonal regulation of circulating IGFBP-1phosphorylation status. J Clin Endocrinol Metab 1995;80:3520–7.

[30] Bankowski E, Sobolewski K, Palka J, Jawoski S. Decreased expressionof the insulin-like growth factor-I-binding protein-1 (IGFBP-1)phosphoisoform in pre-eclampsic Warton's jelly and its role in theregulation of collagen biosynthesis. Clin Chem Lab Med 2004;42:175–81.

[31] Heald AH, Siddals KW, Fraser W, Taylor W, Kaushal K, Morris J, et al.Low circulating levels of insulin-like growth factor binding protein-1(IGFBP-1) are closely associated with the presence of marcrovasculardisease and hypertension in type 2 diabetes. Diabetes 2002;18:2629–36.

[32] Schedlich LJ, Nilsen T, John AP, Jans DA, Baxter RC. Phosphorylation ofinsulin-like growth factor binding protein-3 by deoxyribonucleic acid-dependent protein kinase reduces ligand binding and enhances nuclearaccumulation. Endocrinology 2003;144:1984–93.

[33] Mishra S, Murphy LJ. Phosphorylation of insulin-like growth factor (IGF)binding protein-3 by breast cancer cell membranes enhances IGF-Ibinding. Endocrinology 2003;44:4042–50.

[34] Hou J, Clemmons DR, Smeekens S. Expression and characterization ofserine protease that preferentially cleaves insulin-like growth factorbinding protein-5. J Cell Biochem 2005;94:470–84.

[35] Loukovaara M, Leinonen P, Teramo K, Nurminen E, Anderssont S,Ruthanen EM. Effect of maternal diabetes on phosphorylation of insulin-like growth factor binding protein-1 in cord serum. Diabet Med2005;22:434–9.

[36] Nuutila M, Hiilesmaa V, Karkkainen T, Ylikorala O, Rutanen EM.

93J. Khosravi et al. / Clinical Biochemistry 40 (2007) 86–93

Phosphorylated isoforms of insulin-like growth factor binding protein-1 inthe cervix as a predictor of cervical ripeness. Obstet Gynecol 1999;94:243–9.

[37] Elizur SE, Yinon Y, Epstein GS, Seidman DS, Schiff E, Sivan E. Insulin-like growth factor binding protein-1 detection in preterm labor: evaluationof a bedside test. Am J Prenatal 2004;22:305–9.

[38] Kabir-Salmani M, Shimizu Y, Sakai K, Iwashita M. Posttranslationalmodifications of decidual IGFBP-1 by steroid hormones in vitro. Mol HumReprod 2005;11:667–71.

[39] Stultz CM, Levin AD, Edelman ER. Phosphorylation-induced conforma-tional changes in a mitogen-activated protein kinase substrate. J Biol Chem2002;277:47653–61.