(Received for publication, December 11, 1996)
From § Preclinical Research, Interrogation of the public expressed sequence
tag (EST) data base with the sequence of preproaprotinin identified
ESTs encoding two potential new members of the Kunitz family of serine
protease inhibitors. Through reiterative interrogation, an EST contig
was obtained, the consensus sequence from which encoded both of the novel Kunitz domains in a single open reading frame. This consensus sequence was used to direct the isolation of a full-length cDNA clone from a placental library. The resulting cDNA sequence
predicted a 252-residue protein containing a putative
NH2-terminal signal peptide followed sequentially by
each of the two Kunitz domains within a 170-residue ectodomain, a
putative transmembrane domain, and a 31-residue hydrophilic COOH
terminus. The gene for this putative novel protein was mapped by use of
a radiation hybrid panel to chromosome 19q13, and Northern analysis
showed that the corresponding mRNA was expressed at high levels in
human placenta and pancreas and at lower levels in brain, lung, and
kidney. An endogenous soluble form of this protein, which was
designated as placental bikunin, was highly purified from human
placenta by sequential kallikrein-Sepharose affinity, gel filtration,
and C18 reverse-phase chromatography. The natural protein
exhibited the same NH2 terminus as predicted from the
cloned cDNA and inhibited trypsin, plasma kallikrein, and plasmin
with IC50 values in the nanomolar range.
The Kunitz (1, 2), Kazal (2), Serpin (3), and mucus (4) families
of biological serine protease inhibitors play a vital role in the
spatial and temporal regulation of in vivo proteolysis. The
prototypical Kunitz inhibitor, bovine aprotinin (2), is a 58-amino acid
protein containing three intrachain disulfide bonds in a spacing that
is conserved in all family members (1, 2). Although the physiologic
function of aprotinin is uncertain, it is a potent inhibitor of several
serine proteases, and its potency against kallikrein and plasmin (5)
may be relevant to its clinical mode of action (5, 6), particularly in
the reduction of perioperative blood loss. A human functional homolog of aprotinin has not been identified, although several larger human
proteins containing one or more Kunitz domains are known. These
include: tissue factor pathway inhibitor
(TFPI),1 which contains three Kunitz
domains (7) and inhibits both factor Xa and the factor VIIa-tissue
factor complex (8); TFPI-2 (9), which contains two Kunitz domains (a
bikunin) and is a potent inhibitor of the factor VIIa-tissue factor
complex, factor XIa, and plasmin (10); and inter- The
full-length sequence of the bovine protein
preproaprotinin2 (National Center for
Biological Information (NCBI) sequence 162769) MKMSRLCLSVALLVLLGTLAASTPGCDTSNQAKAQRPDFCLEPPYTGPCKARIIRYFYNAKAGLCQTFVYGGCRAKRNNFKSAEDCMRTCGGAIGPWENL was used to query the data base of ESTs (dbEST) at the NCBI using the tBLASTn algorithm (19). This yielded cDNA sequences that when
translated, encoded proteins with a cysteine spacing that was similar
(R35464[GenBank]) or identical (R74593[GenBank]) to the spacing characteristic of the
Kunitz family of serine protease inhibitor domains, but which were
clearly different in their overall sequence from known human Kunitz
family members. The nucleotide sequences of these ESTs were then used
to re-query dbEST using BLASTn (19) and FASTA (20) to generate
overlapping nucleotide sequences that extended the sequence of the
putative cDNA. ESTs that overlapped these two sequences were
aligned using AssemblyLIGN (Ref. 21, Oxford Molecular Group, Campbell,
CA). This process was repeated by searching dbEST with the newly
identified sequences until the overlapping ESTs extended from a
putative ATG initiation site to a poly(A) at the 3 To obtain a
cDNA encoding the entire extracellular region of the bikunin, the
EST alignments were used to design a 5 A
human multiple tissue Northern blot (CLONTECH)
containing approximately 2 µg of poly(A)+ RNA/lane was
probed with the 780-bp bikunin PCR fragment that was labeled with
[32P]dCTP as described above. Hybridization and washing
conditions were according to the manufacturer's instructions.
The chromosomal location of the gene encoding placental
bikunin was determined by PCR amplification in conjunction with the Stanford G3 hybridization panel (Research Genetics, Huntsville, AL).
The following primers based on the cDNA sequence of placental bikunin were designed: sense, ATCCACGACTTCTGCCTGGTGTCG; antisense, GACAGTGGCACATTTCTTGAGG. These primers were used to amplify a 174-bp nucleotide fragment of human genomic DNA encoding a portion of the
NH2-terminal Kunitz domain. Amplification was achieved with the following conditions: 95 °C for 10 min (cycle 1); 95 °C for 30 min, 55 °C for 15 min, 72 °C for 30 min, (cycles 2-41);
72 °C for 5 min (cycle 42). Data were submitted to the Stanford RH
Mapping Center and analyzed using the program RHMAP.
A whole
frozen human placenta (Analytical Biological Services, Inc. Wilmington,
DE), weighing between 0.7 and 0.8 kg, was thawed to 4 °C, cut into
0.5-1.0-cm pieces on ice, and then washed with 600 ml of
phosphate-buffered saline. Tissue (240 g, wet weight, per run) was
added to 300 ml of 0.1 M Tris-HCl (pH 8.0) containing 0.1 M NaCl (buffer A), then homogenized in a Waring blender
(maximum speed for 2 min). This procedure was repeated until all the
tissue was processed. The supernatant, after centrifugation of this
homogenate at 4,500 × g for 60 min at 4 °C, was
collected, filtered through cheesecloth, and then applied to a
kallikrein-Sepharose affinity column that had been equilibrated in
buffer A at 4 °C. This column was made by attaching 70 mg of bovine
pancreatic kallikrein to 5.0 ml of CNBr-activated Sepharose (Pharmacia
Biotech Inc.) according to the manufacturer's instructions. After
loading, the column was washed with buffer A until the absorbance at
280 nm of the eluent decreased to zero. The column was further washed
with buffer A containing 0.5 M NaCl and then eluted with 3 volumes of 0.2 M acetic acid (pH 4.0). The flow rate was 2 ml/min throughout. Fractions containing kallikrein and trypsin
inhibitory activity were pooled, frozen, and concentrated by
lyophilization. The lyophilized sample was reconstituted in 1.0 ml of
0.1 M Tris-HCl (pH 8.0), containing 0.15 M
NaCl, and 0.01% Triton X-100 and applied in 200-µl aliquots to a
Superdex 75 10/30 column (Pharmacia) equilibrated and eluted (0.5 ml/min) with buffer A at room temperature. Fractions (0.5 ml)
containing resolved peaks of trypsin or kallikrein inhibitory activity
(see below) were pooled separately, adjusted to pH 2.5 by the addition
of trifluoroacetic acid, then applied directly to a Vydac
C18 reverse-phase column (5 µm, 0.46 × 25 cm) which had been equilibrated with 20% (v/v) acetonitrile in 0.1% (w/v) trifluoroacetic acid. Following a 20-ml wash with equilibration buffer,
the column was eluted with a linear gradient of 20-80% acetonitrile
in 0.1% trifluoroacetic acid over 50 min. The flow rate was 1 ml/min
throughout. Fractions (1.0 ml) containing trypsin and kallikrein
inhibitory activity were pooled, concentrated using a Speed-Vac
concentrator (Savant Instruments, Farmingdale, NY), and stored at
The purification of
placental bikunin was monitored with assays for the in vitro
inhibition of bovine trypsin and human plasma kallikrein. To monitor
trypsin inhibition, column fractions were preincubated with bovine
trypsin (Sigma) for 10 min, after which reactions were initiated at
25 °C by the addition of 50 µl of Tos-G-P-K-AMC substrate (Bachem
Bioscience Inc., King of Prussia, PA) to achieve the following final
component concentrations in a 0.17-ml reaction volume: trypsin (17.5 µg/ml); purification fraction (20 µl), Tos-G-P-K-AMC (33 µM) in 50 mM Hepes buffer (pH 7.5) containing
0.1 M NaCl, 2.0 mM CaCl2, 0.01%
(v/v) Triton X-100 (buffer B). To monitor plasma kallikrein inhibition,
aliquots (20 µl) of column samples were preincubated for 15 min at
25 °C with 7.0 nM final plasma kallikrein (Enzyme
Research Laboratories, South Lafayette, IN) in 50 mM
Tris-HCl (pH 8.0), containing 50 mM NaCl and 0.01% Triton
X-100 (buffer C). Reactions were then initiated with 66 µM final P-F-R-AMC (Sigma). Real time formation of
coumarin was determined fluorometrically (excitation = 370 nm,
emission = 432 nm) in 96-well microtiter plates (Perkin-Elmer) on
a Perkin Elmer LS-50B fluorometer equipped with a plate reader.
To determine IC50 values, active site concentrations of
trypsin, plasma kallikrein, and plasmin were determined by titration with p-nitrophenyl p NH2-terminal sequencing was
performed on a Hewlett-Packard model G1005A protein sequencing system
using Edman degradation and Version 3.0 sequencing methods as per the
manufacturer's instruction (24). Samples were loaded onto the
miniature biphasic reaction column and washed with 1 ml of 2%
trifluoroacetic acid prior to the initiation of Edman chemistry.
SDS-polyacrylamide gel electrophoresis was performed using 10-20%
Tricine-buffered polyacrylamide gels (Novex, San Diego, CA) according
to the manufacturer's specifications and developed by silver staining
with a Daiichi Silver Stain-II kit (Integrated Separation Systems,
Natick, MA).
To identify novel human EST sequences with homology to the Kunitz
family of serine protease inhibitors, the dbEST (17, 18) was queried
with the protein sequence of preproaprotinin using the tBLASTn
algorithm (19). Initially, two ESTs from human placenta (accession
numbers R35464[GenBank] and R74593[GenBank]) were identified. Translation of R74593[GenBank]
yielded a theoretical ORF of 108 residues which was flanked by stop
codons and contained six cysteine residues in a spacing characteristic
of Kunitz domains. A 110-residue ORF within R35464[GenBank] also contained a
Kunitz-like protein sequence; however, the second cysteine from the
NH2-terminal end was replaced by a phenylalanine.
Reinterrogation of dbEST with the nucleotide sequences of R35464[GenBank] and
R74593[GenBank] established that the 3
Nucleotide primers based on specific EST sequences (located in Fig.
1B) were used to amplify a cDNA fragment from human
placental cDNA with same size (780 bp) as predicted from the EST
contig. The PCR-derived fragment (not shown) encoded an ATG start site followed by a 240-residue open reading frame which was identical to the
EST consensus over a 222-residue stretch. Northern blot analysis using
this PCR fragment as a probe revealed high levels of expression of a
hybridizing mRNA in placenta and pancreas, with lower levels in
kidney, lung, brain, and heart and undetectable levels in skeletal
muscle (Fig. 2). The size of the mRNA (approximately 1.9 kilobases) was in reasonable agreement with that predicted by the
EST contig (Fig. 1). This novel ORF was designated placental bikunin in
accordance with its tissue abundance and sequence homology.
The PCR fragment was used to probe a placental cDNA library,
resulting in the isolation of a full-length placental bikunin cDNA
(Fig. 3A). The translated ORF within the
full-length clone contained 252 amino acids and was identical to the
EST consensus sequence over all but the first 15 residues. Analysis of
the translated sequence with the program SigCleave identified the first
27 amino acid residues following the ATG start site (defined herein as residues
The entire nucleotide sequence of placental bikunin (Fig.
3A) was searched using BLAST against the following data
bases: GenBank (current to 9/20/96), EMBL (release 47.0), GeneSeq DNA
and protein (release 20.0), PIR (release 49.0), and PatchX (release
49.0). No significant homologies or identities to the coding sequence were observed. However, a portion of the 3 Alignment of the Kunitz domains of placental bikunin with other Kunitz
domains showed that they contained the characteristic conserved spacing
of the six conserved cysteine residues as well as the conserved
..FXYXGCXGNXN.. motif
surrounding the fourth cysteine residue (Fig. 4).
Although the Kunitz domains within placental bikunin fragments 7-64
and 102-159 are clearly novel family members, they exhibited percent
identities with other human Kunitz domains which ranged between 29 and
50%. The amino acid residue COOH-terminal to the second cysteine
residue from the NH2 terminus strongly influences the
protease specificity of Kunitz domains (2). In each of the Kunitz
domains of placental bikunin this position is occupied by an arginine
residue, suggesting that the domains may have specificity toward
inhibition of trypsin-like serine proteases (2).
The chromosomal location of the gene encoding placental bikunin was
determined using primers based on the cDNA sequence encoding the
full-length protein. Accordingly, the closest linkage marker to the
fragment amplified from a Stanford G3 radiation hybrid panel was
identified as D19S228 on chromosome 19 (19q13). This analysis generated
an LOD score of 12.68. This agrees with the finding that 13 of the 29 ESTs used in the construction of our contig map have been mapped
independently within the UniGene set (27) to an unidentified transcript
close to this region between D19S224 and D19S408. These ESTs include:
N40851[GenBank], H16866[GenBank], R34808[GenBank], N57450[GenBank], R34701[GenBank], H39840[GenBank], H95233[GenBank], H39841[GenBank], N30199[GenBank],
N29508[GenBank], N26910[GenBank], H16757[GenBank], and N27732[GenBank].
To characterize this novel serine protease inhibitor, we purified the
naturally occurring protein from human tissue. Placenta was selected as
the source material for purification because of the relatively high
levels of placental bikunin mRNA expression in this tissue. The
possibility that placental bikunin might bind tightly to trypsin-like
serine proteases was exploited in these efforts through implementation
of affinity chromatography over a column of immobilized kallikrein as
the initial purification step following tissue homogenization.
Accordingly, a small fraction of the total trypsin and kallikrein
inhibitor activity present in the soluble fraction of a placental
extract bound this column and could be recovered subsequently by
elution at acidic pH. Gel filtration chromatography of this sample
yielded a peak of kallikrein and trypsin inhibitory activity which
eluted with an apparent molecular mass between 15 and 40 kDa based on
calibration of the column with molecular mass standards under identical
conditions (Fig. 5A). Reverse-phase
C18 chromatography (Fig. 5B) yielded four peaks
of inhibitory activity. The first peak of activity (peak 1) had the
largest number of units of activity, yet had the least amount of
protein based on its A215 nm, whereas peak 2 contained
TFPI-2, as determined by NH2-terminal sequence analysis.
SDS-polyacrylamide gel electrophoresis analysis of the fraction
containing peak 1, followed by silver staining (Fig. 6),
yielded a single major band with an apparent molecular mass of 24 kDa.
A summary of the purification of the native protein is shown in Table
I. Although determination of the fold purification and
yield was complicated by the presence of other protease inhibitors in
the crude homogenate, the purification scheme achieved at least a
4,700-fold purification of the inhibitor based on the specific activity
of the final preparation relative to the starting homogenate.
Table I.
Purification of a novel serine protease inhibitor from human placenta
Institute of Bone and Joint Disease and Cancer, and the
¶ Institute for Research Technologies,
ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
-trypsin inhibitor,
a plasma-associated bikunin (11). In addition, the following proteins
are known to contain a single Kunitz domain: COL
3/VI, the
(3)
chain of type VI collagen (12); HKI-B9, a human Kunitz inhibitor (13); the membrane-associated amyloid precursor proteins APP751
(14) and APP770 (15); the amyloid precursor-like proteins
(APLP) such as APLP2 (16). To identify novel human homologs of
aprotinin we employed a bioinformatic approach that exploited the
rapidly expanding human expressed sequence tags (ESTs) data base (17, 18). This resulted in the discovery of a novel human gene product designated as placental bikunin.
Identification of Novel Human Kunitz-containing ESTs
end. Alignment of
the sequences was used to assign the consensus base at each position
within the alignment while giving additional weight to those bases
defined as being in a region of high quality sequence.
PCR primer,
CACCTGATCGCGAGACCCC, based on the sequence of EST R34808[GenBank] and a 3
primer, CGAAGCTTCATCTCCGAAGCTCCAGACG (containing a HindIII site), based on the sequence of EST R74593[GenBank]. A 30-cycle PCR (95 °C
for 5 min, 1 cycle; 95 °C for 1 min, 55 °C for 30 s, 72 °C for 2 min, 30 cycles; 72 °C for 5 min, 1 cycle) using a GeneAmp PCR reagent kit (Perkin-Elmer) amplified a 780-bp fragment from
human placental cDNA (CLONTECH), which was then
ligated into the pCRII vector (Invitrogen, San Diego). This clone was
sequenced (FDA submission grade) on both strands by LARK Sequencing
Technologies (Houston) to confirm the presence of an open reading frame
(ORF) related to the EST consensus sequence. The PCR-derived cDNA
fragment was then used to probe a human placental cDNA library
(Unizap XR library, Stratagene, LaJolla, CA). Briefly, 2 × 106
plaques were plated on XL1 Blue cells onto NZY
plates (Northeast Laboratory, Waterville, ME) at 37 °C, overnight.
Plaques were transferred to nitrocellulose, denatured in 1.5 M NaCl, 0.5 M NaOH for 2 min; neutralized in
1.5 M NaCl, 0.5 M Tris (pH 8.0) for 5 min;
rinsed in 0.2 M Tris (pH 7.5), 2 × SSC (15 mM sodium citrate (pH 7.6) containing 150 mM
NaCl), then blotted onto Whatman 3MM paper and cross-linked by UV
irradiation with a Stratalinker (Stratagene). Filters were then
prehybridized at 42 °C for 3 h in 50% formamide containing
5 × SSPE (10 mM NaH2PO4 (pH
7.4), containing 1 mM EDTA and 150 mM NaCl),
5 × Denhardt's reagent, and 0.1% SDS. The 780-bp PCR fragment
was liberated from the pCRII plasmid by EcoRI (New England
Biolabs, Beverly, MA) digestion, gel purified on a 1% agarose gel, and
eluted with a gel extraction kit (Qiagen, Chatsworth, CA).
Approximately 40 ng of purified fragment was heat denatured (100 °C
for 5 min) and labeled with [32P]dCTP (Amersham Corp.)
with High Prime labeling reagent (Boehringer Mannheim) using the random
priming method. Unincorporated label was removed using Biospin columns
(Bio-Rad). The probe was again heat denatured as described above, added
to the filters, and hybridized at 42 °C overnight. Filters were then
washed twice for 10 min in 2 × SSC, 0.5% SDS at 25 °C,
followed by two 30-min washes in 1 × SSC, 0.1% SDS at 65 °C,
and exposed to Kodak XAR film overnight at
80 °C with an
intensifying screen. After three rounds of screening and plaque
purification, five independent clones were isolated. In vivo
excision in SOLR cells and DNA preparation were performed according to
the manufacturer's instructions (Qiagen). DNA was sequenced on both
strands by the dideoxynucleotide termination method (22) at the
sequencing facility at Yale University (New Haven, CT).
20 °C until needed.
-guanidinobenzoate HCl
(Sigma), as described (23). Natural placental bikunin was quantified by
titration against 2.2 nM bovine trypsin. Enzyme and
inhibitor were mixed in a total volume of 990 µl of the appropriate
buffer (see below) and incubated for 5 min at 37 °C. Reactions were
initiated by the addition of substrate. Initial component
concentrations were as follows: bovine trypsin activity, buffer B with
[E0] = 50 pM, inhibitor (eight
concentrations in the range 0-0.8 nM) and 30 µM Tos-G-P-K-AMC substrate (Km = 29 µM); human plasmin (American Diagnostica, Inc.,
Greenwich, CT), 50 mM Tris-HCl (pH 7.5), 0.1 M
NaCl, and 0.02% Triton X-100 with [E0] = 50 pM, inhibitor (six concentrations in the range 0-3
nM) and 500 µM Tos-G-P-K-AMC (Km = 726 µM); human plasma
kallikrein, buffer B with [E0] = 2.5 nM, inhibitor (eight concentrations in the range 0-12 nM) and 100 µM P-F-R-AMC
(Km = 457 µM). Hydrolysis of AMC-conjugated peptides was monitored on a Perkin-Elmer model LS50B
fluorometer (excitation = 370 nm, emission = 432 nm) over a
2-min period. Percent inhibition (%I) values were determined from the
equation: %I = 100 × [1-F0/F1], where
F0 and F1 are, respectively, the fluorescence in the presence and absence of inhibitor.
nucleotide sequence flanking the Kunitz domain encoded in R35464[GenBank] was identical to the 5
sequence flanking the
Kunitz domain encoded in R74593[GenBank]. This suggested that the two ESTs were
each part of a common cDNA that encoding a bikunin. Several
overlapping ESTs were obtained upon reinterrogation with R35464[GenBank] and
R74593[GenBank], which were in turn used to reinterrogate dbEST. In this
iterative fashion, a large number of overlapping ESTs were obtained
which could be aligned into an approximately 1,700-bp EST contig (Fig.
1A). Most of the ESTs comprising the contig
were of placental origin, although some were recovered from infant and
adult brain, breast, retina, olfactory epithelium, and ovary. A minimum
of four ESTs overlapped each position within the contig except for the
region between bp 800 and 900. This enabled the derivation of a
consensus sequence from which at least some sequencing or cloning
errors could be deconvoluted. For example, the stop codons flanking the
Kunitz domain in EST R74593[GenBank] were not evident in the consensus sequence,
and the Kunitz-like sequence corresponding to that observed in EST R35464[GenBank] contained the full complement of six correctly spaced cysteine
residues. The consensus nucleotide sequence was 1.6 kilobases in length
(Fig. 1B), extending to the start of a poly(A) tail, and
encoded a 5
ATG start site that was followed by a putative 248 ORF
encoding the two Kunitz domains.
Fig. 1.
Identification of a hypothetical human
cDNA encoding a protein with two Kunitz inhibitor domains.
Panel A, schematic overlay of related ESTs identified by an
EST walk. Starting ESTs were R35464[GenBank] and R74593[GenBank], which were identified through interrogation of dbEST with the sequence of preproaprotinin. Lengths of bars (with alternating shading for clarity) are
proportional to sizes in bp. Accession numbers are indicated to the
right of the respective EST bars. ESTs were derived from
human placental libraries unless indicated by either of the following
abbreviations to the left of the EST bars: AB,
adult brain; IB, infant brain; R, retina;
B, breast; O, ovary; OE, olfactory
epithelium. Panel B, schematic of the consensus nucleotide
sequence. Segments encoding the putative ATG start site, Kunitz
inhibitors (KID), hydrophobic stretch (H), and
putative stop codon and poly(A) tail (An) are
indicated, as are the location of primers used to isolate a 780-bp PCR
fragment.
[View Larger Version of this Image (36K GIF file)]
Fig. 2.
Northern blot analysis of the expression
pattern of mRNA encoding a novel bikunin. Northern blot
analysis was performed according to "Experimental Procedures."
Tissue sources of mRNA for each lane of the blot are indicated. The
amounts of poly(A)+ mRNA in each lane was found to be
comparable in control blots using a probe to the -actin mRNA
(not shown).
[View Larger Version of this Image (79K GIF file)]
1 to
27) as a likely signal peptide (Fig. 3A).
This was followed by a 225-amino acid mature protein sequence beginning with the sequence ADRER- and containing two tandem Kunitz-like inhibitor domains within residues 7-64 and 102-159, respectively. A
24-residue hydrophobic segment between residues 171 and 194 of this
mature protein sequence was followed by a hydrophilic tail of 31 residues. Analysis of the region surrounding the hydrophobic segment
using the method of Kyte and Doolittle (25, 26) highlighted the region
as a probable membrane anchor sequence. Several proline residues were
evident immediately NH2-terminal to the putative transmembrane domain. Two potential N-linked glycosylation
sites characterized by NXS/T motifs were evident at position
30 within the first Kunitz domain and at position 67 within the
38-amino acid segment separating the two Kunitz domains. These features are depicted schematically in Fig. 3B.
Fig. 3.
Nucleotide sequence and translation product
of a cDNA encoding full-length human placental bikunin. The
clone was isolated by colony hybridization as described in under
"Experimental Procedures." Panel A, the complete
cDNA sequence and its translation product are shown. Putative
signal peptide (italic symbols), membrane spanning segments
(underlined), and sites of N-glycosylation
(bold amino acid symbols) are highlighted. The
numbering of the protein sequence starts at the first amino acid
residue of the mature protein. Segments of the protein sequence
containing the NH2- (residues 7-64) and COOH-terminal
(residues 102-159) Kunitz domains are flanked by square
brackets. Within the nucleotide sequence, the ATG start site
(bold) and stop codon (asterisk) are highlighted. Panel B, schematic of the protein sequence depicting
full-length bikunin including location of the signal peptide
(SP), Kunitz inhibitor domains (KID),
proline-rich segment, putative transmembrane (TM) and
cytoplasmic (CYT) domains, and possible glycosylation sites
(CHO).
[View Larger Version of this Image (53K GIF file)]
-untranslated sequence in
the reverse direction was identical to a cDNA fragment found to be
expressed differentially in pancreatic cancer (accession number
Z36849[GenBank]).
Fig. 4.
Alignment of the amino acid sequences of
placental bikunin(7-64) and placental bikunin(102-159) with known
human Kunitz domains. Identities (%) between the known human
Kunitz domains and each of the placental bikunin (PB)
fragments are indicated. Conserved cysteine residues are
highlighted in bold.
[View Larger Version of this Image (30K GIF file)]
Fig. 5.
Purification of a novel serine protease
inhibitor from human placenta. Panel A, Superdex 75 gel
filtration. Fractions were monitored for
A280 nm (solid line) and inhibition of kallikrein () and trypsin (
). Panel B,
C18 reverse-phase high performance liquid chromatography.
Fractions were monitored for absorbance at 215 nm (solid
line) and inhibition of kallikrein (
) and trypsin (
).
Gradient elution (broken line) commenced at fraction 20. The
peaks of inhibitor activity are numbered in order of elution.
[View Larger Version of this Image (33K GIF file)]
Fig. 6.
Reducing SDS-polyacrylamide gel
electrophoresis of a novel serine protease inhibitor purified from
human placenta. Molecular size markers (lane 2)
included, from top to bottom: ovalbumin, 45 kDa;
carbonic anhydrase, 29 kDa; -lactoglobulin, 18.4 kDa; lysozyme, 14.3 kDa; bovine trypsin inhibitor, 6.2 kDa; and insulin, 3 kDa. Purified
placental serine protease inhibitor (0.5 µg) was applied to
lane 1. The Tricine-SDS-polyacrylamide gel was silver
stained as described under "Experimental Procedures."
[View Larger Version of this Image (55K GIF file)]
Step
Volume
Total
Protein
Unitsa
Units/mg
ml
mg
Placental
supernatant
1,800.0
75,060b
3,000,000
40.0
Kallikrein affinity
20.0
3.36b
16,000
4,880
Superdex 75
15.0
0.13b
3,191
24,546
C18
1.0
0.0005c
91
189,580
a
One unit is the amount of inhibitor that inhibits 50%
of trypsin activity assayed according to "Experimental Procedures."
b
Determined from the absorbance at 280 nm.
c
Determined by active site titration with trypsin under
stoichiometric conditions.
NH2-terminal sequence analysis of the final preparation
(Fig. 7) yielded an amino acid sequence that was
identical over the entire 50 cycles analyzed, to the NH2
terminus of mature placental bikunin as predicted by a SigCleave
analysis of the translated full-length cDNA (Fig. 3A).
High quality sequence extended well into the NH2-terminal
Kunitz domain. The cysteine residues within this sequence were not
detected because cysteine is recovered in low yield from the sequencer.
Interestingly, the asparagine at amino acid residue 30 of the sequence
was also not detected, as would be expected if it were glycosylated as
predicted. Residues 35 and 48 could not be determined presumably
because serine is recovered in low yield. The only other detectable
NH2-terminal sequence within the purified sample was
identical to the main sequence except that it commenced at residue 6 and represented only 6% of the total material sequenced. Peaks 3 and 4 recovered from C18 reverse-phase chromatography (Fig.
5B) remain unidentified. Similar efforts to isolate
placental bikunin from the membrane fraction following solubilization
with 1% (v/v) Triton X-100 did not yield detectable amounts of the
inhibitor (data not shown).
As might be predicted from the fact that placental bikunin contains Kunitz inhibitor domains, the natural form of the protein was a potent inhibitor of the following serine proteases: bovine trypsin (IC50 = 0.35 nM), human plasmin (IC50 = 2.5 nM), and human plasma kallikrein (IC50 = 8.0 nM).
We have described the identification of a novel human gene product containing two Kunitz inhibitor domains. This was designated as placental bikunin based on its sequence homology and the tissue abundance of its mRNA. The discovery of this protein was made possible through interrogation of the dbEST with preproaprotinin followed by implementation of an EST "walk" to establish a theoretical ORF. This ORF was used to direct the cloning of a corresponding full-length placental cDNA that encoded the protein. The gene encoding placental bikunin was assigned to chromosome 19q13, where it mapped adjacent to D19S228. Evidence that the placental bikunin gene directs the expression of a protein was obtained from the isolation of a functional serine protease inhibitor from human placenta which possessed the same NH2 terminus as predicted from the placental bikunin cDNA.
The isolated cDNA for placental bikunin encodes a protein that contains a signal peptide, a hydrophobic segment COOH-terminal to the two Kunitz domains, and a COOH-terminal hydrophilic tail. This suggests that the mature full-length protein encoded by the cDNA is targeted to the Golgi compartment following synthesis and may exist as a transmembrane protein. In this scenario, the 170-amino acid Kunitz-containing fragment would be exposed either to the extracellular milieu or lumen of a vesicular compartment depending on its trafficking pathway. Accordingly, NH2-terminal sequence analysis verified that the natural protein had undergone removal of the signal peptide predicted by the cDNA and that it had therefore likely been processed through the secretory pathway. The significance of the finding that the natural protein was recovered in the soluble fraction of placental homogenates without using detergents is open to speculation. Either the hydrophobic segment was present in the natural protein but did not act as a transmembrane domain, or it was absent from the protein as a consequence of alternate splicing or proteolytic processing. Proteolytic processing could have occurred physiologically as has been described for other Kunitz-containing proteins with transmembrane segments such as APP751 and APP770 (14, 15) or as a consequence of tissue homogenization.
Although the size of the natural protein was consistent with that predicted from the ORF within the cloned cDNA, it is unclear how much of the protein sequence encoded by the cDNA is actually present in the purified natural protein. The presence of the NH2-terminal Kunitz domain was clearly evident by NH2-terminal sequencing, and this domain alone could have accounted for the protease inhibitor activity of the natural protein. Because efforts to elucidate the entire covalent structure of the protein have been hampered by the small amounts of purified protein, we have resorted to an immunoblot evaluation of the presence of the predicted domains within the native protein. Preliminary results using antibodies directed against the NH2- and COOH-terminal Kunitz domains (not shown) indicate that only the NH2-terminal domain is present in the 24-kDa species following purification, whereas the antibody against the COOH-terminal domain reacted with a band at 6 kDa in the same preparation, indicating that the COOH-terminal Kunitz domain was present but not covalently attached to the larger protein.
Although placental bikunin seems to represent a low abundance protein as judged by its low recovery from human placenta, it is possible that the isolation procedure selected out a subpopulation of the inhibitor. For example, affinity chromatography over an immobilized protease would not have purified inhibitor that was bound to endogenous protease either prior to or as a consequence of tissue homogenization.
The physiological function of placental bikunin is a matter of speculation at present. Its potential for existence extracellularly perhaps in part as a cell surface-associated protein raises the possibility that the protein could play a role in the regulation of extracellular proteolytic cascades that are activated in close proximity to cell membranes. Based on the potency of the natural protein against plasma kallikrein, such pathways could include the sequelae of contact activation triggered by vascular injury, which include the processes of kinin formation from high molecular weight kininogen, coagulation via the intrinsic pathway, and complement activation. On the other hand, the protein could function in the regulation of fibrinolysis based on its potency against plasmin. The elucidation of the biological functions of placental bikunin will require further study of its post-translational processing, subcellular distribution, cellular expression pattern, and protease specificity.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) U78095[GenBank].
We acknowledge the following members of the research staff at the Bayer Research Center for important contributions: Carla Pellegrino (protein sequencing), John Kupcho (amino acid analysis), and David Murray and Rathin Das (Northern blot analysis).