We have isolated, from human normal amniotic fluid collected in the weeks 16–18, the intact nonphosphorylated IGFBP-1 and five electrophoretically distinct phospho-isoforms and have determ
Trang 1factor binding protein-1 phosphorylation sites and
propensity to proteolysis of the isoforms
Lorenzo Dolcini1, Alberto Sala1, Monica Campagnoli1, Sara Labo`1, Maurizia Valli1, Livia Visai1,2, Lorenzo Minchiotti1, Hugo L Monaco3 and Monica Galliano1
1 Department of Biochemistry ‘A Castellani’, University of Pavia, Italy
2 Center for Tissue Engineering (C I T), University of Pavia, Italy
3 Biocrystallography Laboratory, Department of Biotechnology, University of Verona, Italy
Introduction
The insulin-like growth factor binding proteins
(IG-FBP) are six homologous molecules (IGFBP-1–6) that
play critical roles in a wide variety of important
physi-ological processes Upon binding they regulate the
availability of both insulin-like growth factors I and II
(IGF-I and -II) and their affinity for these ligands is
modulated by several mechanisms: attachment to the
extracellular matrix and post-translational modifica-tions, such as proteolysis, phosphorylation and glyco-sylation Additionally, several studies have revealed that IGFBPs, as well as their proteolytic fragments, have IGF-independent biological activities in cell adhe-sion and migration and in the regulation of the cell cycle and apoptosis [1–4] Furthermore, although these
Keywords
IGFBP; insulin-like growth factor binding
protein-1; mass spectrometry;
phosphorylation; proteolysis
Correspondence
M Galliano, Department of Biochemistry ‘A.
Castellani’, University of Pavia, viale
Taramelli 3b, 27100 Pavia, Italy
Fax: +39 0382 423108
Tel: +39 0382 987724
E-mail: galliano@unipv.it
(Received 16 March 2009, revised 27 July
2009, accepted 19 August 2009)
doi:10.1111/j.1742-4658.2009.07318.x
Insulin-like growth factor binding protein-1 (IGFBP-1) is the major secreted protein of human decidual cells during gestation and, as a modula-tor of insulin-like growth facmodula-tors or by independent mechanisms, regulates embryonic implantation and growth The protein is phosphorylated and this post-translational modification is regulated in pregnancy and repre-sents an important determinant of its biological activity We have isolated, from human normal amniotic fluid collected in the weeks 16–18, the intact nonphosphorylated IGFBP-1 and five electrophoretically distinct phospho-isoforms and have determined their in vivo phosphorylation state The unmodified protein was the most abundant component and mono-, bi-, tri-and tetraphosphorylated forms were present in decreasing amounts The phosphorylation sites of IGFBP-1 were identified by liquid chromatogra-phy–tandem mass spectrometry analysis of the peptides generated with trypsin, chymotrypsin and Staphylococcus aureus V8 protease Five serines were found to be phosphorylated and, of these, four are localized in the central, weakly conserved, region, at positions 95, 98, 101 and 119, whereas one, Ser169, is in the C-terminal domain The post-translational modifica-tion predominantly involves the hydrophilic stretch of amino acids repre-senting a potential PEST sequence (proline, glutamic acid, serine, threonine) and our results show that the phosphorylation state influences the propensity of IGFBP-1 to proteolysis
Abbreviations
IGF, insulin-like growth factor; IGFBP, insulin-like growth factor binding protein; LC-ESI-MS, liquid chromatography-electrospray ionization-mass spectrometry.
Trang 2proteins are all secreted in the bloodstream, the
func-tional role of IGFBP-1 [5], IGFBP-3 [2] and IGFBP-5
[2] inside the cells has been well documented
IGFBP-1, the subject of the present study, is
predom-inantly expressed in the liver and, in adult mammals,
the synthesis of the protein is upregulated in a number
of catabolic or stressful conditions, such as fasting and
diabetes [6] Additionally, it has recently been shown
that a portion of intracellular IGFBP-1 localizes to
mitochondria, where it acts as a prosurvival factor and
protects the liver from apoptosis [5] IGFBP-1
repre-sents a minor IGF binding protein in the circulation of
nonpregnant adults, but it is regarded as the most
relevant member of the IGFBP family during gestation
[7–10] Several studies indicate that IGFBP-1 regulates
embryonic growth as a local modulator of IGF
bio-availability and stimulates trophoblast migration
through binding of its C-terminal domain, which
con-tains an Arg–Gly–Asp sequence recognized by the
integrin family of cell surface receptors [11–14]
Furthermore, the protein has also been shown to play
crucial roles in ovarian, endometrial, trophoblast and
fetal–placental physiology and pathology and its level
in maternal as well as in fetal circulation and in the liver
increases under hypoxic conditions in the uterus,
result-ing in intrauterine growth retardation [15,16]
Its mature polypeptide chain consists of 234 amino
acids and, as all other IGFBPs, contains an N-terminal
and a C-terminal domain linked by a mid-region,
which has a less ordered structure [1,17] This central
domain is more variable in the different members of
the family and contains most of the sites involved in
post-translational modifications IGFBP-1 is subject to
phosphorylation on serine residues and backbone
cleavage [1,17]
In the serum of nonpregnant adults, the protein
occurs as a highly phosphorylated single species [18],
whereas differently phosphorylated IGFBP-1 isoforms
are present in the amniotic fluid [19] Earlier reports
have indicated the presence of a variable number of
isomers [19], probably reflecting sampling at different
gestational periods We have recently shown that
nor-mal amniotic fluid, collected in the weeks 16–18 of
gestation, contains the unmodified protein and five
electrophoretically distinct phosphoisoforms [20] A
previous study has shown that IGFBP-1 expressed in
Chinese hamster ovary cells is phosphorylated at
Ser101, 119 and 169 [21,22] and a recent paper
described the modification of Ser98, which was
revealed in the highly phosphorylated protein isoforms
found in hypoxia-treated cells [23]
Post-translational modification influences the
inter-action of IGFBP-1 with IGF-I [21,22,24] and has been
associated with gestational and fetal abnormalities [24– 26] However, neither the phosphorylation state of the protein isolated from normal human amniotic fluid nor the biological properties of the homogeneous IGFBP-1 phosphoisoforms have been examined so far
It has been shown that, in vitro, several metallo-proteases recognize IGFBP-1 as a substrate [27] and a modulating effect of phosphorylation on the backbone cleavage of the protein has been recognized [8] Recently, a specific IGFBP-1 protease activity has been described in a patient with multiple myeloma and identified as azurocidin [28]
Here we describe the purification on a preparative scale of the six isoforms present in human normal amni-otic fluid collected in the weeks 16–18 The availability
of sufficient amounts of the pure protein allowed a detailed identification of the amino acid residues involved in the IGFBP-1 post-translational modifica-tions occurring in vivo An important result is that three
of the five phosphorylation sites identified in the present work are in a region enriched in proline, glutamic acid, serine and threonine, as it is known that phosphate addition represents a mechanism for activating a latent PEST sequence [29] Amniotic fluid contains a specific IGFBP-1 protease activity, yielding a stable, functional and well-structured C-terminal domain [30] and we have examined the propensity of the different isoforms of IGFBP-1 to backbone cleavage in the presence of this partially purified specific protease
Results
Purification of IGFBP-1 isoforms The amniotic fluid proteins were separated by gel filtration and the low molecular mass components, eluting after the albumin peak, were pooled and fur-ther purified by anion exchange chromatography In a previous study [30], we have shown that amniotic fluid contains a metalloprotease that cleaves the majority of IGFBP-1, yielding a stable C-terminal fragment Thus,
to prevent the proteolytic process, 10 mm EDTA was added during storage of the fluid and in all buffers used throughout the gel filtration separation The pro-teins were then subjected to anion exchange chroma-tography (Fig 1) and the fractions obtained, resolved
by SDS gel electrophoresis, were transferred to poly(vinylidene difluoride) membranes by electroblot-ting and probed with anti-IGFBP-1 IgGs (data not shown) Immunoblotting showed that IGFBP-1 eluted
in six peaks (I–VI) as a single positive band with identical molecular mass of 30 kDa and that no fragmentation of the protein had occurred during the
Trang 3purification procedure The peaks were pooled and
further purified by gel filtration chromatography Six
homogeneous polypeptide chains with the same
N-ter-minal sequence, APWQCAPCSA, corresponding to
the first 10 residues of intact mature IGFBP-1, were
obtained Nondenaturing gel electrophoresis (Fig 2A)
and immunoblotting (Fig 2B) showed that the
electro-phoretic mobilities of the IGFBP-1 isoforms increased,
but not progressively, as expected on the basis of the
anion exchange chromatography elution profile Peak IV
migrated more anodically than the preceding
chro-matographic fraction and the last eluting fraction, VI,
displayed a very similar mobility to that of fraction V
The six proteins were submitted to treatment with alka-line phosphatase and, when analysed by isoelectric focusing, all acquired the same isoelectric point of the polypeptide chain eluting under peak I (data not shown) This form thus represents the unmodified protein, whereas the other fractions contain distinct phosphoiso-forms These results showed that the nonphosphorylated protein, as well as the five distinct phosphoisoforms, were obtained in homogeneous form by anion exchange chromatography Furthermore, the addition of the chelating agent was effective in preventing proteolytic cleavage, as only the intact proteins were obtained
Mass determination of the IGFBP-1 isoforms The molecular mass of the six IGFBP-1 isoforms was probed by liquid chromatography-electrospray ioniza-tion-mass spectrometry (LC-ESI-MS) (Table 1) The mass value determined deconvoluting the multiple charged ions of the protein eluting in peak I was
25 252 Da and, following reduction, 25 270 Da These values match the theoretical molecular mass of the intact, unmodified IGFBP-1 and the difference (+18 Da) measured for the reduced sample accounts for the presence of nine disulfide bonds involving the
18 cysteine residues of the polypeptide chain Analysis
of peak II yielded a molecular mass of 25 332 Da and the increase of 80 Da over the previous component is consistent with the presence of one phosphate group Peaks III and IV displayed the same molecular mass increase of 160 Da over the unmodified protein, corre-sponding to the presence of two phosphates Peaks V (+240 Da) and VI (+320 Da) contain the tri- and tetraphosphorylated protein, respectively (Fig 3) The relative amount of each form was determined
by measuring the corresponding peak area in the chro-matographic profile and were in good agreement with the absolute amounts obtained from the absorbance at
280 nm of the pooled fractions (assuming an absor-bance value of 1.42 for a 1 mgÆmL)1 solution of IGFBP-1 as calculated on the basis of the amino acid
Fig 1 Purification of IGFBP-1 isoforms The low molecular mass
fraction obtained from human amniotic fluid was resolved by
Q-Sepharose ion exchange chromatography The elution was
carried out with a linear gradient from 0 to 100% using 6.25 m M
Bistris-propane, pH 7.5 as buffer A and 6.25 m M Bistris-propane,
pH 9.5, 0.35 M NaCl as buffer B The peaks I–VI, positive when
probed with anti-IGFBP-1 IgGs, were pooled and further purified by
gel filtration.
A
B
Fig 2 Nondenaturing gel electrophoresis and western blot analysis
of IGFBP-1 isoforms IGFBP-1 isoforms isolated by ion exchange
chromatography from peaks I–VI were (A) resolved on a 17%
non-denaturing polyacrylamide gel and stained with Coomassie Brilliant
Blue (lanes 1–6) and (B) following western blot were probed with
anti-IGFBP-1 IgGs (lanes 1–6).
Table 1 Molecular mass of the IGFBP-1 peaks isolated by anion exchange chromatography.
IGFBP-1 isoforms
Measured mass (Da)
Theoretical mass (Da)
Dmass (Da)
Trang 4composition) The most abundant nonphosphorylated
form, eluting as peak I, accounted for 69.5% of the total
protein; the monophosphorylated isoform, eluting in
peak II, was 14.9%; the doubly modified forms, eluting
as peaks III and IV, accounted for 6.5 and 5.5%,
respec-tively; the triphosphorylated protein, peak V,
repre-sented 2.7%, and peak VI, corresponding to the form
with four phosphate groups, was 0.9% These results
show that amniotic fluid IGFBP-1 is modified by up to
four phosphate groups that produce five
graphically distinct isoforms The different
chromato-graphic behaviour of the doubly modified protein is
probably due to the alternative positions of the charged
groups, resulting in different isoelectric points
Identification of IGFBP-1 phosphorylation sites
The examination of the amino acidic sequence of
human IGFBP-1 with the protein pattern database
NetPhos [31] predicted 12 serines, four threonines and one tyrosine, which could potentially serve as phosp-hoacceptor sites Thus, the assumption that the phos-phate groups can have alternative distribution within the polypeptide chain suggested that a low level of modification at each site could be expected Further-more, it is known that the ionization efficiency of phosphopeptides is generally slightly lower than that
of their unmodified counterparts and only relatively large fragments, between six and 25 amino acids, are suitable for mass spectral analysis [32] Therefore, to enhance sequence coverage and ensure that each phosphorylation site was contained in at least one peptide of suitable size and hydrophobicity, the IGFBP-1 isoforms were cleaved with three different proteolytic enzymes Reduction of disulfide bonds was carried out before each enzymatic digestion and the resulting peptide mixtures were immediately submitted
to LC-ESI-MS⁄ MS analysis The HPLC separation
1000
25+24+
23+
22+ 21+ 20+ 19+
18+
17+
16+
15+
14+ 13+
25+24+
23+
22+21+20+
19+ 18+ 17+
16+
15+
14+ 13+
25+ 24+
23+
22+
21+
20+ 19+
18+
17+
16+
15+ 14+ 13+
25+
26+
27+
24+
23+ 22+ 21+ 20+
19+
18+
17+
16+ 15+
14+ 13+
25+
26+
27+
24+
23+ 22+
21+
20+
19+ 18+
17+ 16+
1200 1400
m/z
1600 1800 2000
25251 Da
25331 Da
25410 Da
25572 Da
25493 Da
E
D
C
B
A
Fig 3 Multicharged mass spectra of the IGFBP-1 phosphoisoform isolated from human amniotic fluid The mass values reported on the y-axis were obtained by deconvoluting the multiply charged pattern obtained for (A) nonphosphorylated protein eluting as peak I; (B) the mono-1-phosphorylated form, peak II; (C) the biphosphorylated chains eluting in peaks III and IV; (D) the triphosphorylated protein, peak V and (E) the tetraphosphorylated molecule eluting as peak VI.
Trang 5was performed under acidic conditions in order to
maximize positive charge ionization and the elution
gradient was chosen in order to reduce the loss of
small and hydrophilic peptides during the desalting
step Phosphoric acid was added to the sample
solu-tion because it significantly enhances the detecsolu-tion of
mono- and multiphosphorylated peptides following
RP-HPLC [33] The identity of the fragments was
assessed by manual inspection and by comparing the
experimental fragmentation pattern with the theoretical
values obtained using the protein prospector
(http://prospector.ucsf.edu/prospector/mshome.htm)
and peaks studio software (Bioinformatic Solution
Inc., Waterloo, Canada) In addition, as
phosphoryla-tion can reduce the efficiency of cleavage [34], the data
were also examined for incompletely digested
frag-ments The phosphorylated peptides were identified by
the presence of species with a theoretical peptide mass
increased by 80 Da (plus one phosphate group) or
multiples thereof, and in all cases, ions corresponding
to the loss of H3PO4 ()98 for single charged ions and
)49 for doubly charged ions) dominated the
fragmen-tation pattern A complete sequence coverage of the
protein was obtained (data not shown) and Table 2
lists the phosphorylated peptides observed The V8
fragments spanning residues 93–100 (AGSPESPE)
from the unmodified protein, at m⁄ z 773.3, and from
the monophosphorylated one, at m⁄ z 853.2, displayed
a mass difference of 80 Da, which is consistent with
the presence of one phosphate group Fragmentation
in the ion trap showed that this peptide was phosphor-ylated at Ser95 and, to a much lower extent, also at Ser98 The presence of the fragment ion at m⁄ z 755.3
is due to the loss of phosphoric acid (98 Da), the ion
at m⁄ z 296.1 (b3+P) is consistent with the modifica-tion of Ser95 and the ions at m⁄ z 638.2, (y5+P) and
at m⁄ z 442.2 (b5) are indicative of the modification at Ser98 (Fig 4B) Figure 4A shows the MS⁄ MS spec-trum obtained for the unmodified peptide The two peptides coelute and the doubly modified fragment was not observed The MS⁄ MS spectra of the V8 fragment 93–103 (AGSPESPESTE) from the monophosphory-lated isoform, at m⁄ z 1172.1 (+80 Da shift) is shown
in Fig 5A The loss of phosphoric acid (98 Da) from the precursor yielded a charged species at m⁄ z 1072.1 and the presence of a product ion at m⁄ z 642.0 (y5+P) is consistent with phosphorylation at Ser101 The fragment ions at m⁄ z 562.2 and at m ⁄ z 649.0 correspond to unmodified y5 and y6, respectively, and indicate that Ser95 is phosphorylated as well The modification of Ser98 could not be distinguished in this peptide Analysis of the V8 digest of the trip-hosphorylated protein showed the presence of peptide 93–103 carrying two phosphate groups MS⁄ MS spec-tra (Fig 5B) of the charged ion at m⁄ z 1250.0 dis-played fragment ions at m⁄ z 1152.1 and at m ⁄ z 1054.1, consistent with the loss of one and two phosphate groups, respectively The ion at m⁄ z 642.0 (y5+P) indicates the phosphorylation of Ser101 and the ion at m⁄ z 955.0 (y8+P) indicates that in the
Table 2 List of the phosphopeptides revealed by LC-ESI-MS ⁄ MS Modified residues are underlined.
Calculated mass (Da)
Measured mass (Da)
Dmass (Da) Chymotryptic peptides
Glu C peptides
Tryptic peptides
Trang 6biphosphorylated fragment, Ser95 and Ser101 are
modified The triphosphorylated peptide was not
observed These results show that the sequence
span-ning residues 93–103 contains three serine residues
whose adjacent amino acids form consensus sites for
phosphorylation The V8 fragment spanning residues
109–121, at m⁄ z 1515.3, was found in the unmodified
protein and the fragmentation pattern of its
phosphor-ylated counterpart, at m⁄ z 1595.2, displayed ions
accounting for the modification of Ser119 (data not
shown) The modification of Ser169 was detected in
the V8 fragment spanning residues 161–172, at m⁄ z
1330.3, which showed the expected fragmentation ions
(data not shown) The MS⁄ MS spectrum of the V8
fragment TSMDGE, found in all the IGFBP-1
sam-ples, and spanning residues 219–224 at m⁄ z 654.9
(Fig 6), showed that the molecular mass increase of
15.9 Da over the theoretical value was due to the
oxidation of methionine to sulfoxide, a common
post-translational modification to proteins occurring
in vivo The nonoxidized sequence was also present at
m⁄ z 639.1
These results lead to the conclusion that the IG-FBP-1 phosphoisomers isolated from normal amni-otic fluid contain five phosphoacceptor sites The phosphorylation of the five serine residues was detected in all the chromatographically distinct isomers A complete sequence coverage was obtained for each polypeptide chain and Table 2 lists the molecular mass of the peptides with phosphorylated residues One novel phosphorylation site on Ser101 was revealed, and the phosphoacceptor sites Ser95, Ser98, Ser119 and Ser169, previously described in cultured cells [22,23], were confirmed in our natural samples of human amniotic fluid IGFBP-1 In agree-ment with previous data [22,35], we did not find any evidence of threonine and tyrosine phosphor-ylation For this reason, and because it is well
Fig 4 LC-MS ⁄ MS ion spectra of IGFBP-1 peptide 93–100 (A) MS ⁄ MS spectrum of the singly charged 773.3 Da ion correspond-ing to the nonphosphorylated AGSPESPE fragment (B) MS ⁄ MS spectrum of the singly charged 853.2 Da ion corresponding
to the monophosphorylated AGSPESPE sequence.
Trang 7known that the phosphopeptides are not eluted
with high efficiency from RP-HPLC, a quantitative
analysis of each phosphorylation site could not be
performed
Chromatographic separation of the V8 peptides
In order to decide whether a major phosphoacceptor serine is present in the monophosphorylated protein, it
A
B
Fig 5 LC-MS ⁄ MS ion spectra of IGFBP-1
peptide 93–103 (A) MS ⁄ MS spectrum of
the singly charged 1172.1 Da ion
corre-sponding to the monophosphorylated
AGSP-ESPESTE fragment (B) MS ⁄ MS spectrum
of the singly charged 1250.0 Da ion
corresponding to the triphosphorylated
AGSPESPESTE peptide.
Fig 6 LC-MS ⁄ MS spectrum of the
IGFBP-1 peptide 2IGFBP-19–224 The asterisk marks the
oxidized methionine in the sequence and in
the internal ions at m ⁄ z 234.5 and 262.9.
Trang 8was digested with endopeptidase V8 from
Staphylococ-cus aureus and compared with the unmodified form
The peptide mixtures were resolved by RP-HPLC
chromatography and Fig 6 shows the superimposition
of the UV traces monitored at 214 nm The peaks were
manually collected and submitted to N-terminal
sequencing and LC-ESI-MS⁄ MS analysis The
chro-matographic profiles, as well as the peak area and
shape, did not differ in a significant manner (Fig 7)
The only major discrepancy was the presence of a
peak, indicated as 1, in the trace of the unmodified
protein, which was absent in the profile of the
monophosphorylated molecule, where an additional
one, indicated as peak 2, was present The N-terminal
sequence analysis showed that peak 1 contains the
sequence AGSPESPE (93–100) and that peak 2
corre-sponds to the fragment AGSPESPESTE (93–103)
Mass analysis confirmed these data and showed that
the latter was phosphorylated These results indicate
that the presence of the phosphate group prevents the
enzymatic cleavage Furthermore, as no other evident
peak shift was appreciable, it can be argued that the
monophosphorylated IGFBP-1 form is predominantly
modified in the sequence spanning residues 93–103
Proteolysis
In a previous paper [30], we described the isolation of
the C-terminal domain of IGFBP-1 and suggested that
the proteolytic cleavage could be ascribed to a specific
amniotic fluid metalloprotease Therefore, prior to
examining the effect of phosphorylation on the
proteo-lytic process, we prefractionated the amniotic fluid
proteins by gel filtration and analysed the fractions
obtained for their IGFBP-1 protease activity As shown
in Fig 8A, unmodified IGFBP-1 incubated with the fraction containing components with molecular mass above 50 kDa, remained almost unchanged, whereas in
Fig 7 HPLC elution profiles of the V8 peptides Trace A was
obtained from the nonphosphorylated IGFBP-1 and trace B from
the monophosphorylated protein The arrows indicate the major
differences between the two chromatograms (peaks 1 and 2).
A
B
C
Fig 8 IGFBP-1 proteolysis (A) Aliquots of the nonphosphorylated protein (1 lg in 20 lL of 20 m M Tris pH 7.5 containing 20 m M
CaCl 2 ) were incubated overnight at 37 C with 5 lL of the amniotic fluid fractions containing proteins eluting ahead (lane 1) and after (lane 2) the albumin peak in gel filtration chromatography; as a con-trol, the isolated C-terminal fragment and the untreated protein are shown in lanes 3 and 4, respectively Following SDS gel electropho-resis, the proteins were electroblotted and probed with anti-IGFBP-1 IgGs (B) Aliquots of the nonphosphorylated IGFBP-1 (1 lg in 20 m M Tris pH 7.5 containing 20 m M CaCl 2 ) were incubated overnight with no addition (lane 1), and with the addition of increas-ing amounts of the amniotic fluid fraction displayincreas-ing proteolytic activity: 2 lL (lane 2); 4 lL (lane 3); 8 lL (lane 4); and 16 lL (lane 5) Sample volumes were adjusted to 30 lL and incubations were terminated after 48 h by the addition of 30 lL of sample buffer The samples were then submitted to SDS gel electrophoresis The proteins were electroblotted and probed with anti-IGFBP-1 IgGs (C) Substrate in gel electrophoresis of the amniotic fluid protease-containing fraction Gelatin (lane 1) and IGFBP-1 (lane 2) were added to a 10% polyacrylamide solution before gel casting at a con-centration of 1 mgÆmL)1 The asterisk indicates the lysis area observed in IGFBP-1 zymography.
Trang 9the presence of proteins eluting after the albumin peak,
IGFBP-1 was partially cleaved, yielding two major
bands, one corresponding to the C-terminal domain
and a second with an apparent molecular mass of
20 kDa No fragmentation was observed when the
incu-bation was performed in the presence of 10 mm EDTA
(data not shown) The fraction containing the IGFBP-1
protease activity was then used to assess the dose
dependence of the proteolytic process Figure 8B shows
that by increasing the amount added to the protein the
intensity of the 30 kDa band is progressively reduced
The protease-containing fraction was then submitted to
zymography using gelatine and IGFBP-1 as substrates
in order to examine the specificity of the IGFBP-1
degrading activity The gelatine substrate zymogram
showed the presence of several areas of lysis,
represent-ing gelatine-degradrepresent-ing proteinase activity in the sample
On the contrary, only one area was evidenced in the
IGFBP-1 substrate zymogram (Fig 8C) These results
indicate that a specific protease is involved in the
prote-olytic cleavage occurring in amniotic fluid We then examined the effect of phosphorylation on this process Each isoform was incubated with the proteinase-containing fraction and the samples were then submit-ted to western blot analysis, together with identical aliquots of the untreated proteins (Fig 9A) The degradation process was evaluated on the basis of the ratio between the integrated areas of the intact protein
in the treated and untreated samples This value was determined for each isoform and the results (Fig 9B) show that the monophosphorylated IGFBP-1 was cleaved almost to the same extent as the unmodified protein and that the susceptibility to proteolytic degradation of the isoforms increased with the number
of phosphates linked to the polypeptide chains The different position of the groups bound to the biphosph-orylated protein also played a role in the cleavage process
Discussion
The purification procedure used in this study permitted the isolation of the unmodified as well as of the five phosphoisoforms of IGFBP-1 previously detected by proteomic analysis in human amniotic fluid collected
in weeks 16–18 of gestation [20] To obtain a sufficient amount of the biological fluid, several hundred individ-ual samples were pooled and, therefore, the prepara-tion was assumed to be normal as every individual difference would be below the detectability score The molecular mass of the nonphosphorylated protein was determined under both reducing and nonreducing con-ditions The values obtained matched the molecular mass deduced from the amino acid sequence and were consistent with the presence of nine disulfide bonds In addition, the molecular mass increments of the other isoforms defined the exact number of phosphate groups linked to the chromatographically distinct IGFBP-1 phosphoisomers isolated The characteriza-tion of the phosphorylacharacteriza-tion sites of IGFBP-I isoforms revealed that the post-translational modification involves five serine residues (numbered as Ser95, Ser98, Ser101, Ser119 and Ser169 in the mature polypeptide chain) In a previous study, we examined the phos-phorylation of IGFBP-1 isolated from amniotic fluid following the selective extraction of phosphopeptides
on a titanium dioxide (TiO2) cartridge and only three modified serines were located [36] Enrichment meth-ods are well suited when only minute amounts of the phosphoprotein are available, but a detailed analysis
of phosphoptides demands relatively high quantities of the homogeneous molecules All the residues identified
in the present study were recognized by the protein
A
B
Fig 9 Proteolysis of IGFBP-1 isoforms (A) Aliquots of each
IGFBP-1 isoform (1 lg in 20 lL of 20 m M Tris pH 7.5 containing
20 m M CaCl 2 ) were submitted to SDS electrophoresis and
immuno-blotting following incubation at 37 C overnight with the addition of
a 4 lL aliquot of the amniotic fluid fraction containing IGFBP-1
pro-tease activity As a control, identical aliquots of the six isoforms
diluted to the same concentration with the buffer containing 20 m M
CaCl2were incubated separately at 37 C overnight and analysed.
For each isoform, indicated with the peak number, the left lane
contains the untreated protein and the right lane the sample treated
with the protease-containing fraction (B) Histogram representation
of the susceptibility to proteolytic degradation of IGFBP-1 isoforms.
The y-axis shows the percentage of the ratio between the spot
area of the degraded and that of the untreated protein measured
using IMAGE J 1.37V software.
Trang 10pattern database NetPhos [31] as phosphoacceptor
sites, with a score always higher than 0.99 and, in
agreement with previous data [22,35], no modification
occurring at tyrosine and threonine residues was
found Although a previous study, using the protein
expressed in Chinese hamster ovary cells, identified
three serines (Ser101, Ser119 and Ser169) as the only
modified amino acids and excluded the involvement of
other residues [22], our results, showing the presence
of the tetraphosphorylated isoform, revealed a more
extensive post-translational modification of the
amni-otic fluid protein Because the process can be the
con-sequence of a combination of biological variables,
including different activity levels of both the
intracellu-lar protein kinases and the extracelluintracellu-lar phosphatases,
this discrepancy is probably due to the different source
of the examined molecules Moreover, it has been
shown that the phosphorylation of IGFBP-1 is
con-trolled by placental steroid hormones [37] and by
IGF-II [38] Both factors are probably regulated differently
in Chinese hamster ovary cells and in human decidual
cells A recent paper revealed the modification of
Ser98 in the hyperphosphorylated IGFBP-1 expressed
in hypoxia exposed cells and the finding was suggested
to be associated with the condition [23] However, our
data show that this phosphoacceptor site is also
modi-fied in protein purimodi-fied from normal amniotic fluid
The kinases responsible for in vivo phosphorylation of
IGFBP-1 are as yet unknown, but it has been shown
that IGFBP-1 is a substrate for several protein kinases
[31,35] and the residues adjacent to all the modified
se-rines form consensus sequences for post-translational
modification by casein kinase II
A quantitative determination of the modification
occurring at each site could not be achieved because of
the poor ionization efficiency and fast degradation of
phosphopeptides In particular, the detection of
multi-phosphorylated peptides is reduced owing to their high
propensity to adsorption to exposed surfaces [34]
Fur-thermore, nonspecific and partial enzymatic cleavages
caused the presence of several fragments containing the
same phosphorylation site However, the examination
of the peptides obtained following V8 proteolytic
cleav-age indicated that the monophosphorylated IGFBP-1
form is predominantly modified within the
AGSPES-PESTE sequence containing three of the five
phosphor-ylated residues identified in the present study It is
interesting to note that this polypeptide sequence is
rich in proline, glutamic acid, serine and threonine, a
so-called PEST region, which is typical of rapidly
metabolized proteins Briefly, the PEST hypothesis
suggests that protein regions containing a high local
concentration of the amino acids proline, glutamic acid,
serine, threonine, and, to a lesser extent, aspartic acid, are suitable targets for proteolytic degradation [29] Examination of the amino acid sequence of IGFBP-1 with the protein pattern database PESTfind (available
at http://www.at.embnet.org/toolbox/pestfind/), a com-putational tool designed to predict potential PEST sequences, confirmed that the region spanning residues 89–114 (HAAEAGSPESPESTEITEEELLDNFH) is potentially one of those This region is located 27 residues far from the proteolytic site that leads to the formation of the C-terminal fragment previously isolated and characterized from amniotic fluid [30] Proteolysis of IGFBPs is an important mechanism that controls IGF bioavailability [39,40] and previous studies have focused on the cleavage of IGFBP-1 by matrix metalloproteases present in amniotic fluid and conditioned medium from decidualized endometrial cells [8,25] The effect of IGFBP-1 phosphorylation
on its susceptibility to enzymatic cleavage has been examined in vitro and Gibson et al [8] showed that the highly phosphorylated protein is resistant to the prote-ase activity in decidual conditioned medium and to plasmin, whereas Kabir-Salmani et al [37] suggested that matrix metalloproteinase-9 selectively degrades phosphorylated IGFBP-1 Recently, it has been reported that the IGFBP-1 specific protease activity identified as azurocidin cleaves both phosphorylated and nonphosphorylated IGFBP-1 [28] These discrep-ancies probably reflect the specificity of any enzymatic activity and, although the protein has been recognized
as a potential physiological substrate for several matrix metalloproteases, little is known about the susceptibil-ity of IGFBP-1 to enzymatic cleavage Because the only proteolytic degradation of IGFBP-1 occurring without the addition of endogenous reactants so far described under normal conditions is that observed in the amniotic fluid [41], we performed the assays using
a partially purified fraction of amniotic fluid displaying IGFBP-1 specific protease activity Our data clearly show that the post-translational modification increases the susceptibility to cleavage by the amniotic fluid pro-tease and suggest that not only the degree, but also the position of the phosphate groups influence the process
As for the other IGFBPs, the proteolytic cleavage of IGFBP-1 occurs within the central domain and we have shown that the process generates a C-terminal domain (residues 141–234) carrying a phosphate group linked to Ser169 [30] It is possible that the charge repulsion between the phosphate groups in the mid-region and the one located in the C-terminal portion
of the protein may cause a conformational change that makes the cleavage site more exposed to the protease Thus, the post-translational modification might act as