D

Archives of Biochemistry and BiophysicsVol. 379, No. 2, July 15, pp. 209–216, 2000doi:10.1006/abbi.2000.1872, available online at http://www.idealibrary.com on

Evidence That the Interaction between Insulin-like GrowthFactor (IGF)-II and IGF Binding Protein (IGFBP)-4 IsEssential for the Action of the IGF-II-DependentIGFBP-4 Protease1

Xuezhong Qin,*,†,2 Dongwon Byun,*,† K.-H. William Lau,*,†,‡avid J. Baylink,*,†,‡ and Subburaman Mohan*,†,‡,§

*Musculoskeletal Disease Center, J. L. Pettis Memorial Veterans’ Medical Center, and Departments of †Medicine,‡Biochemistry, and §Physiology, Loma Linda University, Loma Linda, California 92357

Received October 26, 1999, and in revised form March 30, 2000

uwvAetttFt

A variety of human cell types, including human os-teoblasts (hOBs), produce an IGFBP-4 protease, whichcleaves IGFBP-4 in the presence of IGF-II. Recently,the pregnancy-associated plasma protein (PAPP)-Ahas been determined to be the IGF-II-dependent IG-FBP-4 protease produced by human fibroblasts. Thisstudy sought to define the mechanism by which IGF-IIenhances IGFBP-4 proteolysis. Addition of PAPP-A an-tibody blocked the IGFBP-4 proteolytic activity inhOB conditioned medium (CM), suggesting thatPAPP-A is the major IGFBP-4 protease in hOB CM.Pre-incubation of IGFBP-4 with IGF-II, followed byremoval of unbound IGF-II, led to IGFBP-4 proteolysiswithout further requirement of the presence of IGF-IIin the reaction. In contrast, prior incubation of thepartially purified IGFBP-4 protease from either hOBCM or human pregnancy serum with IGF-II did notlead to IGFBP-4 proteolysis unless IGF-II was re-added to the assays. To further confirm that the inter-action between IGF-II and IGFBP-4 is required forIGFBP-4 protease activity, we prepared IGFBP-4 mu-tants, which contained the intact cleavage site(Met135–Lys136) but lacked the IGF binding activity,by deleting the residues Leu72–His74 in the IGF bind-ing domain or Cys183–Glu237 that contained an IGFbinding enhancing motif. The IGFBP-4 protease was

1 This work was supported by grants from National Institutes ofHealth (R03 DE12142-01, R0-3 AR 45081-01, R0-1 AR45210 to X.Q.and R0-1 AR31062 to S.M.) and National Medical Technology Test-Bed, Loma Linda University.

2 To whom correspondence should be addressed at MusculoskeletalDisease Center, J. Pettis VA Medical Center (151), 11201 BentonStreet, Loma Linda, CA 92357. Fax: 909-796-1680. E-mail:

Xuezhong.Qin@med.va.gov.

0003-9861/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

nable to cleave these IGFBP-4 mutants, regardless ofhether or not IGF-II was present in the assay. Con-ersely, an IGFBP-4 mutant with His74 replaced by anla, which exhibited normal IGF binding activity, wasffectively cleaved in the presence of IGF-II. Takenogether, these findings provided strong evidence thathe interaction between IGF-II and IGFBP-4, ratherhan the direct interaction between IGF-II and IG-BP-4 protease, is required for optimal IGFBP-4 pro-

eolysis. © 2000 Academic Press

Key Words: IGFBP-4; human osteoblast; protease;IGF-II; PAPP-A.

The insulin-like growth factors (IGFs)3 play an im-portant role in promoting cell differentiation and pro-liferation in a variety of biological systems (1–4). Theactions of IGFs both in vitro (5–8) and in vivo (9–11)are modulated by IGF binding proteins (IGFBPs)which have high affinity and specificity for the IGFs.Among the six high-affinity IGFBPs, IGFBP-4 is apotent inhibitor of IGF actions in a variety of cell types(6–8, 12–14). Since IGFs, upon binding to IGFBP-4,are biologically inactive, the release of IGFs from theinactive IGFBP-4/IGF complex via proteolysis ofIGFBP-4 is important in the IGF physiology. In thisregard, it has previously been shown that an IGF-II-dependent IGFBP-4 protease was produced by varioushuman cells in vitro (15–18) and that the IGFBP-4

3 Abbreviations used: IGFs, insulin-like growth factors; IGFBPs,IGF binding proteins; hOB, human osteoblast; CM, conditioned me-

dium; PAPP-A, pregnancy-associated plasma protein-A.

209

pa

IrsnpbliIlw

twocIre(wFIosfic

DwPmJAf

wTwrqcwsL6r6

u2(

210 QIN ET AL.

proteolytic activity in human serum increased dramat-ically during pregnancy (19, 20). Recent studies fromour laboratory and others have demonstrated that theIGFBP-4 protease activity present in human osteoblast(hOB), human fibroblast conditioned medium (CM), orhuman pregnancy serum, each cleaved IGFBP-4 at asite between Met135 and Lys136 and that the proteol-ysis of IGFBP-4 by IGFBP-4 protease from all thesesources was significantly enhanced by the exogenouslyadded IGF-II (14, 20, 21). Recently, the identity of theIGF-II-dependent IGFBP-4 protease has been deter-mined by Lawrence et al. (22) and was shown to be the

reviously purified (23) and cloned (24, 25) pregnancy-ssociated plasma protein-A (PAPP-A).Results from in vitro studies suggest that the IGF-

I-dependent IGFBP-4 protease may play an importantole in modulating the bioavailability of IGFBP-4 and,ubsequently, the mitogenic activity of the IGFs in aumber of biological systems. For example, IGFBP-4roteolytic fragments, which exhibited little or no IGFinding activity, did not inhibit IGF induced cell pro-iferation (14, 21, 26). IGFBP-4 analogs, which exhib-ted normal IGF binding affinity but were resistant toGFBP-4 protease, inhibited the IGF-induced cell pro-iferation more potently than the protease sensitiveild type IGFBP-4 (14, 21, 26, 27).Studies on the regulation of the IGFBP-4 have led to

he observation that treatment of various types of cellsith the wild-type IGFs decreased the concentrationsf IGFBP-4 in the CM (28–30) whereas treatment ofells with the IGF-I analogs with reduced affinity forGFBPs was less effective (31). Incubation of rat neu-oblastoma cells or human fibroblasts with IGFs ex-rted little or no effect on the levels of IGFBP-4 mRNA32, 33), suggesting that the effect on IGFBP-4 by IGFsas not mediated through the biosynthesis of IGFBP-4.owlkes and Freemark (15) first demonstrated thatGF directly added to cell-free CM activates the prote-lysis of IGFBP-4. Enhancement of IGFBP-4 proteoly-is by IGF-II under cell-free conditions was later con-rmed by other investigators, using a number of humanell model systems (18, 31, 34, 35). Several potential

mechanisms have been proposed to account for theenhancement of IGFBP-4 proteolysis by IGFs. First,IGFs may bind directly to and inactivate a putativeIGFBP-4 protease inhibitor, leading to the activation ofthe IGFBP-4 protease activity. Second, IGFs may di-rectly interact with and activate the IGFBP-4 protease.Third, IGFs may bind directly to the IGFBP-4 andinduce a conformational change in the IGFBP-4, suchthat the cleavage site in IGFBP-4 is accessible to theproteolysis by the protease. In support of the latterpotential mechanism, previous studies have shownthat IGF-I or IGF-II analogs with a reduced affinity forIGFBP-4 have generally been shown to be less effective

in promoting IGFBP-4 proteolysis (18, 34, 36). Al- t

though the findings that the IGF-II analogs with re-duced affinity had a weaker effect on IGFBP-4 prote-olysis provided support for the above-mentioned thirdmechanism, studies using IGF-I analogs were not en-tirely consistent with this mechanism of action (34).

The goal of this study was to clarify the mechanismwhereby IGF-II enhances the IGFBP-4 proteolysis. Inorder to determine whether IGF-II enhances IGFBP-4proteolysis via interacting with IGFBP-4 or the IGFBP-4protease, we determined if pretreatment of IGFBP-4 orthe protease with IGF-II would lead to increasingIGFBP-4 proteolysis without further requirement ofthe presence of IGF-II in the proteolysis reaction. Inaddition, we also prepared specific IGFBP-4 analogswhich retained the intact proteolytic domain but con-ferred very little IGF-II binding affinity and deter-mined if the binding of IGF-II to IGFBP-4 is essentialfor the IGF-II-dependent IGFBP-4 proteolysis.

MATERIAL AND METHODS

Reagents. Serum samples from pregnant women were collectedin SoonChunHyang Hospital, South Korea, according to approvedresearch protocols. The samples were shipped on dry ice to theUnited States and stored at 280°C prior to use. Normal humanosteoblasts (hOBs) were isolated from calvarial bone specimens aspreviously described (37). Dulbecco’s modified Eagle’s medium(DMEM) was from Life Technologies, Inc. Bovine calf serum wasfrom Hyclone (Logan, UT). Bovine serum albumin (BSA) was pur-chased from Fluka (Buchs, Switzerland). Recombinant human IGF-Iand IGF-II were from Bachem, Inc. (Torrance, CA). Na125I was from

uPont NEN (Wilmington, DE). Ni–agarose and the pQE32 plasmidere products of Qiagen (Chatsworth, CA). IDA–agarose was fromierce (Rockford, IL). DEAE-Sepharose was from Pharmacia (Ala-eda, CA). Hydroxylapatite was purchased from Calbiochem (La

olla, CA). SDS–PAGE reagents were from Bio-Rad (Hercules, CA).ll other chemicals and reagents were of reagent grade and were

rom Sigma (St. Louis, MO).Preparation of recombinant IGFBP-4. The pQE32 plasmid vectoras used to prepare the test IGFBP-4 expression constructs (8).hese constructs led to the expression of IGFBP-4 peptides taggedith 6 3 His at the N-terminus. The wild type IGFBP-4 contained 5

esidues of the signal peptide and the entire mature IGFBP-4 se-uence and was designated the 63His-BP-4(-5/237). The peptidesontaining the N-terminal 182 residues and N-terminal 135 residuesere referred to as 63His-BP-4(-5/182) and 63His-BP-4(-5/135), re-

pectively. The full-length IGFBP-4 protein lacking the residueseu72, Met73, and His74 was produced and was designated the3His-BP-4(D72-74). The full-length IGFBP-4 protein with His74eplaced with Ala was designated 63His-BP-4(His74/Ala). The3His-BP-4(-5/237), 63His-BP-4(-5/182), 63His-BP-4(-5/135), and

63His-BP-4(His74/Ala74) expression constructs were prepared aspreviously described (8, 21). The 63His-BP-4(D72-74) was con-structed using the ExSite PCR-based site-directed mutagenesis kit(Stratagene) under modified conditions as described below. The for-ward primer, 59-TGTGTGCAGGGGCTTCTCCACCCC-39, and thereverse primer, 59-GGGCAAGGCGTGTGCATGGAGCTG-39, wereused for PCR amplification using the 63His-BP-4(-5/237) plasmid asa template and DNA polymerase, Pfu. The PCR was carried out

nder the following conditions: one cycle of 98°C for 3 min (hot start),2 cycles each of 98°C for 1 min (denaturation), 68°C for 1 minannealing), and 72°C for 16 min (extension). The PCR product was

reated with DpnI to digest the methylated parental DNA template,

y

CpcsrHbbMtespao3

2hat

t

211PROTEOLYSIS OF IGF BINDING PROTEIN-4

purified, self-ligated with T4 DNA ligase, and transformed into Esch-erichia coli XL 1-Blue cells.

Overexpression of the recombinant IGFBP-4 peptides in E. coliwas carried out as previously described (8). Recombinant 63His-BP-4(-5/237) and 63His-BP-4(His74/Ala) were sequentially purified byNi–agarose and IGF-I affinity chromatography (8). The 63His-BP-4(-5/182) was sequentially purified by Ni–agarose affinity, IGF-I–agarose affinity, and HPLC reverse-phase chromatography. The63His-BP-4(D72-74) was sequentially purified by Ni–agarose affin-ity and HPLC reverse-phase chromatography (this peptide did notbind to IGF-I agarose).

Western 125I-IGF ligand blot and IGFBP-4 immunoblot analyses.125I-IGF Western ligand blot analysis or IGFBP-4 immunoblot anal-sis was performed as previously described (38, 39).Partial purification of the IGF-II-dependent IGFBP-4 protease.alf serum-free CM from human calvarial osteoblast cultures wasrepared as previously described (21). The crude CM was pooled,oncentrated, dialyzed with 10 mM Hepes (pH 7.5) overnight, andtored at 280°C. Concentrated CM was applied to a DEAE-Sepha-ose 4B column (1.5 3 10 cm) that was preequilibrated with 10 mMepes (pH 7.5). After the column was washed with the equilibrationuffer, the proteins were eluted stepwise with the equilibrationuffer containing 0.14–0.5 M NaCl. The DEAE fractions (0.18–0.22

NaCl) were concentrated and applied to a Superose 12 gel filtra-ion column (1 3 30 cm) on the FPLC system. The proteins wereluted with 10 mM Hepes (pH 7.5) containing 100 mM NaCl, andixty 0.5-ml fractions were collected. Fractions 15–20 from the Su-erose 12 column were pooled and applied to an IDA–agarose Zn-ffinity column (0.5 3 2 cm), which was first equilibrated with 150 mlf filtered Zn-affinity buffer (10 mM Hepes, 50 mM NaCl, 0.01% Brij5, pH 7.5) containing 3 mM ZnCl2, and then the column was washed

with 100 ml of Zn-affinity buffer. The IGF-II-dependent IGFBP-4protease activity was eluted with 3 ml of Zn-affinity buffer with 50mM EDTA. The pooled fraction was dialyzed against 10 mM Hepes(pH 7.5) containing 50 mM NaCl. The resulting preparation ap-peared to be free of other protease activities.

To prepare partially purified IGFBP-4 protease from human preg-nancy serum, 0.5 ml of the third-trimester human pregnancy serumwas diluted 10-fold with 10 mM Hepes (pH 7.5). The sample wasloaded to a DEAE-cellulose column (5 ml resin). After the columnwas washed with 20 ml of Hepes buffer containing 100 mM NaCl,proteins were eluted with 10 ml of 300 mM NaCl. The 300 mM NaClfraction was loaded directly to a hydroxyapatite column (1 ml resin).After the column was washed with 150 mM phosphate (pH 7.5),bound proteins were eluted with 300 mM phosphate and desalted bydialysis against 10 mM Hepes (pH 7.5). The resulting partiallypurified protease preparation was stored at 280°C until use.

Mass spectrometric analysis. To obtain IGFBP-4 proteolytic frag-

TA

Partial Purification of IGF-II-Dep

Total protein(mg)

Concentrated CM 4800.0DEAE ion exchange 280.0Gel filtration 16.0Zn affinity 12.5

Note. The amount of IGFBP-4 degraded in the presence and absenhe excised bands (n 5 2) after IGF-II ligand blot analysis. One unit o

to cleave 1 ng of IGFBP-4 in the presence of a 2-fold molar excerepresentative of three independent purifications.

ments for mass spectrometric analysis, 2–6 mg of 63His-BP-4(-5/ i

37) was digested with the partially purified IGFBP-4 protease fromOB CM or human pregnancy serum in the presence of an equimolarmount to IGFBP-4 of IGF-II. The digested samples were subjectedo mass spectrometric analysis as previously described (20, 21).

RESULTS

Characterization and Partial Purification of theIGFBP-4 Protease from hOB CM

Inactivation of a putative IGFBP-4 protease inhibi-tor by IGF-II was proposed as one of the mechanismsby which IGF-II could induce IGFBP-4 proteolysis. Toevaluate this possibility, we subjected hOB CM or hu-man pregnancy serum to several column chromato-graphic steps. We reasoned that these various purifi-cation steps could effectively separate the putative in-hibitor from the IGFBP-4 protease. The IGF-II-dependent protease activity in hOB CM was present inthe fractions that eluted with 0.18 to 0.22 M NaCl froma DEAE-Sepharose 4B column, in the void volume fromSuperose 12 gel filtration column, and in the 50 mMEDTA-eluted fraction from the Zn-affinity column, re-spectively. After these three steps of purification, theIGF-II-dependent IGFBP-4 protease activity was en-riched by approximately 60-fold with a recovery of 16%(Table I). Addition of exogenous IGF-II not only en-hanced IGFBP-4 proteolysis catalyzed by the crudehOB CM but also increased the proteolysis of IGFBP-4mediated by the partially purified IGFBP-4 proteaseobtained from each of the purification steps (Fig. 1).

Recently, human pregnancy serum has shown to con-tain strong IGF-II-dependent IGFBP-4 protease activ-ity (20) and the IGF-II-dependent IGFBP-4 proteaseproduced by human fibroblasts in vitro was identical tothe previously identified and cloned PAPP-A in preg-nancy serum (22). Our recent studies revealed thatPAPP-A is the major if not sole IGFBP-4 protease inthe human pregnancy serum based on the finding thatimmunodepletion of PAPP-A from pregnancy serumwith PAPP-A antibody completely abolished subse-quent IGFBP-4 proteolysis (D. Byun et al., manuscript

I

dent IGFBP-4 Protease Activity

ecific activityits/mg protein)

Total activity(units)

Yield(%)

85 408,000 100779 218,120 53

4,780 76,480 195,100 63,750 16

of IGF-II was estimated by gamma counting of the radioactivity inFBP-4 protease activity is defined as the amount of protease needed

of IGF-II after a 12 h incubation at 37°C. Data shown here are

BLE

en

Sp(un

cef IGss

n preparation). To address the issue of whether

aomspsIptI

E

tttpttptt

C

212 QIN ET AL.

PAPP-A is the major IGFBP-4 protease produced byhOB, we determined if preincubation of hOB CM withPAPP-A antibody could abolish proteolysis of IGFBP-4.Consistent with a previous report (22), PAPP-A waspresent in the hOB CM as determined by immunoblotanalysis (data not shown). Addition of 1 mg of PAPP-Antibody to the protease assay blocked the proteolysisf IGFBP-4, while addition of an equal amount of nor-al rabbit IgG had no effect (Fig. 2). In addition, mass

pectrometric analysis of the mass of the IGFBP-4roteolytic fragments identified Met135–Lys136 as theingle cleavage site recognized by the partially purifiedGFBP-4 protease from both hOB CM and humanregnancy serum (Table II). These data demonstratehat PAPP-A also represents the major IGF-dependentGFBP-4 protease produced by hOB.

ffect of Pretreatment of IGFBP-4 or IGFBP-4Protease/PAPP-A with IGF-II on SubsequentIGFBP-4 Proteolysis

We next evaluated if the IGF-II-dependent activa-ion of IGFBP-4 proteolysis involves interaction of pro-ease or IGFBP-4 with IGF-II. In the first experiment,he partially purified IGFBP-4 protease from humanregnancy serum was incubated with IGF-II at a dosehat is sufficient to maximally stimulate IGFBP-4 pro-eolysis. The unbound IGF-II in the IGF-II/IGFBP-4rotease mixture was removed by ultrafiltration prioro addition of IGFBP-4. We would anticipate that par-

FIG. 1. Proteolytic activity of the partially purified IGF-II-depen-dent IGFBP-4 protease from hOB CM. Concentrated CM (1.5 mgprotein) and IGFBP-4 protease sequentially purified by DEAE (150ng protein), gel filtration (20.5 ng protein), and Zn affinity (20 ngprotein) were used to digest 150 ng of 63His IGFBP-4 (-5/237)protein in the presence of 100 ng of recombinant human IGF-II orvehicle (1 ml of 20 mM acetic acid). After incubation at 37°C for 12 h,samples were mixed with an equal volume of loading buffer, sepa-rated by 10% SDS–PAGE gel, and subjected to IGF-II ligand blotanalysis. A representative of three different experiments is shownhere.

ially purified protease preincubated with IGF-II

would cleave IGFBP-4, if the interaction betweenIGF-II and IGFBP-4 protease is essential for IGFBP-4proteolysis. Contrarily, pretreatment of the proteasepreparation with either IGF-II or vehicle control failedto cleave IGFBP-4 unless IGF-II was added during pro-tease assay (Fig. 3). Similar results were also obtainedusing crude human pregnancy serum (data not shown).In the second experiment, we pretreated rhIGFBP-4with an equimolar amount of IGF-II or vehicle control.After the pretreatment, the unbound IGF-II was re-moved by Ni–agarose pull-down as described in Fig. 4.In the absence of IGF-II, IGF-II/IGFBP-4 complex, butnot the IGFBP-4 pretreated with vehicle, was cleavedby the partially purified IGFBP-4 protease. Thesefindings are consistent with the premise that inter-action between IGF-II and IGFBP-4 is essential forthe IGFBP-4 proteolysis by the protease.

Proteolysis of the Wild-Type IGFBP-4 and IGFBP-4Analogs with Reduced IGF Binding Affinity

To further confirm if the interaction binding betweenIGF-II and IGFBP-4 is essential for the action of theIGF-II-dependent IGFBP-4 protease, we sought to de-termine if IGFBP-4 protease could cleave IGFBP-4 an-alogs that contained an intact proteolytic domain but adisrupted IGF binding domain. We would anticipatethat such IGFBP-4 analogs would not be cleaved if thebinding of IGF-II to IGFBP-4 is required for IGFBP-4proteolysis. A prerequisite for using this approach isthat the IGF binding domain and the proteolytic do-main in hIGFBP-4 must be structurally and function-ally separated. Our previous findings (8) revealed thatdeletion of a cleavage-site-containing region, residues121 to 142 in hIGFBP-4, had no effect on IGF-I and

FIG. 2. Effect of addition of PAPP-A antibody on IGFBP-4 proteaseactivity in hOB CM. A 10-ml amount of 100-fold concentrated hOB

M was incubated with 1 mg of purified anti-PAPP-A IgG or normalIgG at room temperature for 2 h prior to addition of IGF-II (50 ng)and IGFBP-4 (150 ng). After an additional 17 h of incubation at 37°C,the reaction mixture was subjected to IGF-II ligand blot analysis.

This experiment was repeated using different batches of hOB CM.

Bm6tC7

hlBdf4a

7tiB

Dfio I

213PROTEOLYSIS OF IGF BINDING PROTEIN-4

IGF-II binding activity, and that the IGF binding do-main is composed of a N-terminal motif (residues 72–91) and a C-terminal motif (residues 204–214), both ofwhich are separated from the identified cleavage site.These data strongly suggest that the domains for IGFbinding and proteolytic cleavage are not overlapping.Therefore, we prepared an IGFBP-4 analog, 63His-

P-4(D72-74), in which the N-terminal IGF bindingotif was disrupted, and a second IGFBP-4 analog,

3His-BP-4(-5/182), in which a C-terminal region con-aining the IGF binding enhancing motif was deleted.onsistent with the inability of the 63His-BP-4(D72-4) to bind to IGF-I agarose, this mutant did not ex-

TAB

Molecular Masses of Intact an

Molecular m

Observed

63His-BP-4(-5/135)a 16,826N-terminal fragment 16,828b (16,804)c

C-terminal fragment 11,346b (11,338)c

Intact 63His-IGFBP-4 28,148

a 63His-BP-4(-5/135) recombinant protein was prepared as a stfragment.

b Masses of IGFBP-4 fragments obtained by digesting of 63His-BPCM.

c Masses of IGFBP-4 fragments obtained by digesting of 63His-Bpregnancy serum.

FIG. 3. Proteolysis of IGFBP-4 by IGFBP-4 protease pretreatedwith IGF-II. The partially purified IGFBP-4 protease from the third-trimester pregnancy serum was incubated with excess of IGF-II (1mg) or vehicle at room temperature for 1 h in the presence of 100 ml

MEM with 0.5 mM CaCl2. Free IGF-II was removed by an ultra-ltration unit (30 kDa exclusion limit). After it was washed with 3 mlf DMEM/0.5 mM CaCl2, the sample volume was adjusted to 65 ml

with DMEM/0.5 mM CaCl2. A 6-ml amount of protease preparationwas then incubated with 150 ng IGFBP-4 in the presence of 50 ngIGF-II or vehicle for 9 h at 37°C. The digested samples were sub-jected to IGF-II ligand blot analysis. The data from this experiment

were confirmed in a different experiment.

ibit either IGF-I or IGF-II binding activity by the IGFigand blot analysis (Fig. 5). The binding of the 63His-P-4(-5/182) to IGF-I or IGF-II was dramatically re-uced compared with the wild type IGFBP-4. Resultsrom solution binding assay revealed that 63His-BP-(-5/182) binds to IGF-II with 50- to 100-fold reducedffinity compared with the wild type IGFBP-4 (10).In the presence of IGF-II, neither 63His-BP-4(D72-

4) nor 63His-BP-4(-5/182) peptide was cleaved, whilehe majority of wild-type IGFBP-4 was degraded underdentical conditions (Fig. 6). In contrast to the 63His-P-4(D72-74), another IGFBP-4 mutant with the

II

Fragment Forms of IGFBP-4

(Da)

Peptide sequenceExpected

16,816 63His-BP-4(Gly-5 to Met135)16,816 63His-BP-4(Gly-5 to Met135)11,388 IGFBP-4(Lys136 to Glu237)28,186 63His-BP-4(Gly-5 to Glu237)

dard to confirm the mass of the N-terminal IGFBP-4 proteolytic

5/237) with the partially purified IGFBP-4 protease from human OB

(-5/237) with the partially purified IGFBP-4 protease from human

FIG. 4. Proteolysis of IGFBP-4/IGF-II complex by partially purifiedIGFBP-4 protease from human pregnancy serum. A 3.2-mg amountof 63His-BP4(-5/237) was incubated with 0.8 mg rhIGF-II or vehiclefor 1 h at room temperature. After addition of 50 ml of Ni–agaroseresin presaturated with BSA, the reaction mixture was incubated atroom temperature with shaking for an additional 1 h. To remove freeIGF, the Ni–agarose resin was washed three times (1 ml each) withPBS containing 10 mg/ml BSA and then twice with PBS. TheGFBP-4 or IGFBP-4/IGF-II complex was eluted in with 50 ml of PBS

containing 350 mM immidazole. Approximately 50 ng of IGFBP-4/IGF-II complex was incubated with the partially purified protease inthe presence of 50 ng IGF-II or vehicle for 7 h at 37°C. The digestedsamples were subjected to IGF-II ligand blot analysis. The data from

LE

d

ass

an

-4(-

P-4

this experiment were confirmed in a different experiment.

214 QIN ET AL.

His74 substituted with Ala, which has previously beenshown to exhibit IGF-I and IGF-II binding activitysimilar to that of the wild-type IGFBP-4 (8), was effec-tively cleaved by the partially purified IGFBP-4 pro-tease from hOB CM under the same conditions (datanot shown).

DISCUSSION

The bioavailability of IGFs in the local tissues de-pends not only on the amount of IGF-II produced butalso on the amount of inhibitory IGFBPs, such asIGFBP-4. The activity of IGFBP-4 is in turn under thecontrol of the action of specific IGFBP-4 protease. Oneof the unique characteristics of the IGFBP-4 proteaseproduced by a number of human cell types in vitro isthat its IGFBP-4 proteolytic activity can be enhancedby the addition of the IGF-I or IGF-II in cell-free CM (8,14–18). In this study, we obtained several lines ofevidence, as discussed below, to support the hypothesisthat the interaction between IGF-II and IGFBP-4,rather than an interaction between IGF-II and theIGFBP-4 protease, is essential for IGFBP-4 proteoly-sis. In addition, we provided evidence for the first timethat PAPP-A is the predominant IGF-II-dependent

FIG. 5. IGF ligand blot analysis of the wild-type IGFBP-4 andIGFBP-4 analogs with disrupted IGF binding domain. Approxi-mately 150 ng of recombinant protein was loaded to each lane in A,C, D and 15 ng of peptide was loaded to each lane in B. (A) Coomassieblue staining. (B) Immunoblot analysis using polyclonal IGFBP-4antiserum. (C) 125I-IGF-I ligand blotting. (D) 125I-IGF-II ligand blot-ting.

IGFBP-4 protease produced by hOBs in vitro.

Various investigators have proposed that IGFs mayenhance IGFBP-4 proteolysis by inactivating a poten-tial IGFBP-4 protease inhibitor. In this regard,Fowlkes et al. (41) have provided evidence that addi-tion of IGFBP-3 to MC3T3-E1 cell culture conditionedmedium dramatically inhibited IGFBP-4 proteolysis.This inhibition can be reversed by addition of IGFs.Subsequently, they found that a heparin-binding,highly basic peptide sequence located in the C-terminalregion of IGFBP-3, -5, and -6 is contributable to theinhibition of IGFBP-4 proteolysis (42). Although themechanism by which IGFBP-3 inhibits IGFBP-4 pro-teolysis has not been clearly defined, IGFBP-3, themost abundant IGFBP in body fluids, may serve as asignificant inhibitor of IGFBP-4 proteolysis in vivo.Since IGFBP-3 is not present in significant amounts ifnot absent, in our partially purified IGFBP-4 protease(no detectable IGF binding activity at 46–48 kDa wasobserved), IGFBP-3 is unlikely to contribute to the lackof IGFBP-4 proteolysis in the absence of IGF-II. More-over, our data did not provide strong evidence for thepresence of other IGF-II inactivatable IGFBP-4 pro-tease inhibitors in hOB CM, since the partially purifiedIGFBP-4 protease did not negate the need of IGF-II toenhance IGFBP-4 proteolysis (Fig. 1), although thepossibility that a protease inhibitor may coelute withthe IGFBP-4 protease cannot be excluded.

Although previous studies demonstrate that PAPP-Ais the IGF-dependent IGFBP-4 protease produced byhuman fibroblasts and is also present in human OBCM (22), it has not been determined whether PAPP-Aalso predominantly contributes to the IGFBP-4 proteo-lytic activity in hOB CM. Our data showing that addi-tion of PAPP-A antibody produced in rabbit but notnormal IgG from nonimmunized rabbits to hOB CMblocked IGFBP-4 proteolysis provided strong evidencefor PAPP-A as the major IGFBP-4 protease producedby hOBs. The cleavage site recognized by the protease

FIG. 6. Proteolysis of 63His-BP-4(-5/237), 63His-BP-4(D72-74),and 63His-BP-4(-5/182) by the partially purified IGF-II-dependentIGFBP-4 protease from hOB CM. Approximately 150 ng of purifiedprotein was incubated with 20 ng of the partially purified IGFBP-4protease in the presence 100 ng IGF-II or vehicle. After a 17 h ofincubation at 37°C, the digested sample was separated by 8% SDS–PAGE and stained with Coomassie blue. A representative of three

different experiments is shown here.

1

1

1

1

1

1

1

215PROTEOLYSIS OF IGF BINDING PROTEIN-4

in crude human fibroblast CM (22), hOB CM (21), andcrude human pregnancy serum (20), and partially pu-rified IGFBP-4 protease from hOB CM and pregnancyserum (Table II) has been determined to be Met135and Lys136. In contrast to these findings, IGFBP-4protease/s produced by bovine smooth muscle cells(SMC) cleaved human IGFBP-4 at a major site ofLys120–His121 (27). The discrepancy could be ex-plained by difference in types of the IGFBP-4 proteasesproduced by different cell types (fibroblasts/osteoblastsvs SMC) or by different species (humans vs bovine).

Regarding the mechanism of IGF-II enhancement ofIGFBP-4 proteolysis, we have obtained two lines ofevidence in this study that strongly support the hy-pothesis that the interaction between IGF-II andIGFBP-4, rather than an interaction between IGF-IIand the IGFBP-4 protease, is essential for IGFBP-4proteolysis. First, the IGFBP-4 analog, 63His-BP-4(D72-74) which contained the proteolytic cleavage do-main but did not bind to IGFs, was completely resis-tant to cleavage by the partially purified IGF-II-depen-dent IGFBP-4 protease. It is possible that deletion ofthe residues Leu72 to His74 may affect the conforma-tion of IGFBP-4, such that the cleavage site was notaccessible to the protease. However, this possibility isremote, since substitution of His74 for Ala in this re-gion without affecting IGF-II binding (8) did not pro-vide protease resistance (data not shown). Moreover,the protease resistance conferred by a low IGF bindingmutant involving a deletion in the C-terminal regionfurther supports our contention that binding of IGF-IIto IGFBP-4 is required for optimal IGFBP-4 proteoly-sis (Fig. 6). Second, partially purified IGFBP-4 pro-tease pre-incubated with IGF-II did not cleave IGFBP-4unless IGF-II is readded to the assays (Fig. 3). On theother hand, IGF-II/IGFBP-4 complex was cleaved bythe protease (Fig. 4). These data indicate that enhance-ment of IGFBP-4 proteolysis by IGF-II is probablymediated through a direct interaction between IG-FBP-4 and IGF-II rather than between IGFBP-4 pro-tease and IGF-II.

Taken together, results from this study demonstratethat IGF-II is unlikely to enhance IGFBP-4 proteolysisthrough inactivating an IGFBP-4 protease inhibitor ordirectly interacting with the IGFBP-4 protease. Ourexperimental data showing that the IGF binding do-main is essential for the activity of IGF-II-dependentIGFBP-4 protease support the hypothesis that bindingof IGF-II to IGFBP-4 alters the conformation of IGFBP-4,such that the proteolytic domain is more accessible tothe IGFBP-4 protease.

The physiological role of the IGF-II-induced IGFBP-4proteolysis can only be speculated at this time. It isconceivable that the synthesis of the IGF-II-dependentIGFBP-4 protease/PAPP-A is low when IGFs are in

surplus with respect to the body’s need. In the absence

of IGFBP-4 protease, IGFBP-4 binds to IGF-II to forma biologically inactive complex in order to prevent ex-cess IGF-II exposure to the target tissues. As the de-mand for IGFs increases under certain physiological(e.g., pregnancy) or pathological (e.g., tumor) condi-tions, the synthesis of the IGF-II-dependent IGFBP-4protease/PAPP-A is triggered. The increased amount ofIGFBP-4 protease/PAPP-A in either circulation or localenvironment could lead to an increase in the proteoly-sis of IGFBP-4 bound to IGFs and, consequently, therelease of IGFs. Therefore, the inhibitory effect ofIGFBP-4 on IGF action is not contradictory to thestimulatory effect of IGFs on IGFBP-4 proteolysis.Rather, this seemingly contradictory phenomenon rep-resents the complexity and precision of the IGF regu-latory system as a result of evolution.

ACKNOWLEDGMENTS

The authors thank Drs. C. Kim, K. Shu, M. Yoo, and H. Lee(Departments of Endocrinology, Gynecology, and Obstetrics, Soon-ChunHyang University Hospital, South Korea) for providing us withthe human pregnancy serum, Joe Rung-Aroon for excellent technicalsupport, and Medical Media in J. L. Pettis VA Medical Center forillustrations. The authors also thank Ms. Carol Farrell for secre-tarial help.

REFERENCES

1. Canalis, E. (1993) Bone 14, 273–276.2. Rosen, C. J., Donahue, L. A., and Hunter, S. J. (1994) Proc. Soc.

Exp. Biol. Med. 206, 83–102.3. Clemmons, D. R. (1997) Cytokine Growth Factor Rev. 8, 45–62.4. Mohan, S., and Baylink, D. J. (1999) in Contemporary Endocri-

nology: The IGF System (Rosenfeld, R., and Roberts, C., Eds.),pp. 457–496. Humana Press, Inc., Totowa, NJ.

5. Campbell, P. G., and Novak, J. F. (1991) J. Cell. Physiol. 149,93–300.

6. Culouscou, J. M., and Shyoab, M. (1991) Cancer Res. 51, 2813–2819.

7. Mohan, S., Nakao, Y., Honda, Y., Landale, E., Leser, U., Dony,C., Lang, K., and Baylink, D. J. (1995) J. Biol. Chem. 270,20424–20431.

8. Qin, X., Strong, D., Baylink, D. J., and Mohan, S. (1998) J. Biol.Chem. 273, 23509–23516.

9. Bagi, C. M., Brommage, R., Deleon, L., Adams, S., Rosen, D., andSommer, A. (1994) J. Bone Miner. Res. 9, 1301–1312.

0. Miyakoshi, N., Richman, C., Qin, X., Baylink, D. J., and Mohan,S. (1999) Endocrinology 140, 4699–4705.

1. Richman, C., Baylink, D. J., Lang, K., Dony, C., and Mohan, S.(1999) Endocrinology 140, 4699–4705.

2. Mohan, S., Bautista, C. M., Wergedal, J. E., and Baylink, D. J.(1989) Proc. Natl. Acad. Sci. U.S.A. 86, 8338–8349.

3. Cheung, P. T., Smith, E. P., Shimasaki, S., Ling, N., and Cher-nausek, S. D. (1991) Endocrinology 129, 1006–1015.

4. Conover, C. A., Durham, S. K., Zapf. J., Masiarz, F. R., andKiefer, M. C. (1995) J. Biol. Chem. 270, 4395–4400.

5. Fowlkes, J., and Freemark, M. (1992) Endocrinology 131, 2071–2076.

6. Neely, E. K., and Rosenfeld, R. G. (1992) Endocrinology 130,

985–993.

1

1

2

2

2

2

2

2

2

2

22

3

3

3

3

3

3

3

3

3

3

4

4

216 QIN ET AL.

17. Durham, S. K., Kiefer, M. C., Riggs, B. L., and Conover, C. A.(1994) J. Bone Miner. Res. 9, 111–117.

8. Kanzaki, S., Hilliker, S., Baylink, D. J., and Mohan, S. (1994)Endocrinology 134, 383–392.

9. Kubler, B., Cowell, S., Zapf, J., and Braulke, T. (1998) Endocri-nology 139, 1556–1563.

0. Byun, D., Mohan, S., Kim, C., Suh, K., Yoo, M., Lee, H., Baylink,D. J., and Qin, X. (2000) J. Clin. Endocrinol. Metab. 85, 373–381.

1. Qin, X., Byun, D., Strong, D. D., Baylink, D. J., and Mohan, S.(1999) J. Bone Miner. Res. 14, 2079–2088.

2. Lawrence, J. B., Oxvig, C., Overgaard, M. T., Sottrup-Jensen, L.,Gleich, G. J., Hays, L. G., Yates, J. R., III, and Conover, C. A.(1999) Proc. Natl. Acad. Sci. U.S.A. 96, 3149–3153.

3. Lin, T. M., Halbert, S. P., Kiefer, D. J., Spellacy, W. N., and Gall,S. (1974) Am. J. Obstet. Gynecol. 118, 223–236.

4. Kristensen, T., Oxvig, C., Sand, O., Moller, N. P. H., and Sottrup-Jensen, L. (1994) Biochemistry 33, 1592–1598.

5. Haaning, J., Oxvig, C., Overgaard, M. T., Ebbesen, P., Kristensen,T., and Sottrup-Jensen, L. (1996) J. Biochem. 237, 159–163.

6. Chernausek, S. D., Smith, C. E., Duffin, K. L., Busby, W. H.,Wright, G., and Clemmons, D. R. (1995) J. Biol. Chem. 270,11377–11382.

7. Rees, C., Clemmons, D. R., Horvitz, G. D., Clarke, J. B., andBusby, W. H. (1998) Endocrinology 139, 4182–4188.

8. Conover, C. A. (1991) J. Clin. Invest. 88, 1354–1361.9. Hassager, C., Fitzpatrick, L. A., Spencer, E. M., Riggs, B. L., and

Conover, C. A. (1992) J. Clin. Endocrinol. Metab. 75, 228–233.

0. Myers, S. E., Cheung, P. T., Handwerger, S., and Chernausek,S. D. (1993) Endocrinology 133, 1525–1531.

1. Irwin, J. C., Dsupin, B. A., and Giudice, L. C. (1995) J. Clin.Endocrinol. Metab. 80, 619–626.

2. Ceda, G. P., Fielder, P. J., Henzel, W. J., Louie, A., Donovan,S. M., Hoffman, A. R., and Rosenfeld, R. G. (1991) Endocrinology128, 2815–2824.

3. Camacho-Hubner, C., Busby, W. H., McCusker, R. H., Wright, G.,and Clemmons, D. R. (1992) J. Biol. Chem. 267, 11949–11956.

4. Conover, C. A., Kiefer, M. C., and Zapf, J. (1993) J. Clin. Invest.91, 1129–1137.

5. Durham, S. K., Riggs, B. L., and Conover, C. A. (1994) J. Clin.Endocrionol. Metab. 79, 1752–1758.

6. Parker, A., Gockerman, A., Busby, W. H., and Clemmons, D. R.(1995) Endocrinology 136, 2470–2476.

7. Wergedal, J. E., Mohan, S., Lundy, M., and Baylink, D. J. (1990)J. Bone. Miner. Res. 5, 179–186.

8. Scharla, S. H., Strong, D. D., Mohan, S., Baylink, D. J., andLinkhart, T. A. (1991) Endocrinology 129, 3139–3146.

9. Honda, Y., Landale, E., Strong, D. D., Baylink, D. J., and Mohan,S. (1996) J. Clin. Endocrinol. Metab. 81, 1389–1396.

0. Fowlkes, J. L., Serra, D. M., Rosenberg, C. K., and Thrailkill,K. M. (1995) J. Biol. Chem. 270, 27481–27488.

1. Fowlkes, J. L., Thrailkill, K. M., George-Nascimento, C., Rosen-berg, C. K., and Serra, D. M. (1997) Endocrinology 138, 2280–

2285.