From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
Role of Individual y-Carboxyglutamic Acid Residues of Activated Human Protein
C in Defining its In Vitro Anticoagulant Activity
By Li Zhang, Ashish Jhingan, and Francis J. Castellino
To evaluate the contributions of individual ycarboxyglutamic acid (gla) residues t o the overall Ca2+-dependent
anticoagulant activity of activated human protein C (APC),
we used recombinant (e) DNA technology t o generate protein
C (PC) variants in which each of the gla precursor glutamic
acid (E) residues (positions 6,7, 14, 16, 19,20,25,26, and 29)
was separately altered t o aspartic acid (D). In one case, a
gla26V mutation ([glaZIVIr-PC) was constructed because a
patient with this particular substitution in coagulation factor
IX had been previously identified. Two additional r-PC mutants were generated, viz, an r-PC variant containing a
substitution at arginine (R) 15 ([R15]r-PC), because this
particular R residue is conserved in all gla-containing blood
coagulation proteins, as well as a variant r-PC with substitution of an E at position 32 ([F31L, Q32EIr-PC). because gla
residues are found in other proteins at this sequence location. This latter protein did undergo ycarboxylation at the
newly inserted E32 position. For each of the 11 recombinant
variants, a subpopulation of PC molecules that were ycarboxylated at all nonmutated gla-precursor E residues has been
purified by anion exchange chromatography and, where
necessary, affinity chromatography on an antihuman PC
column. The r-PC muteins were converted t o their respective
r-APC forms and assayed for their amidolytic activities and
Caz+-dependent anticoagulant properties. While no significant differences were found between wild-type (wt) r-APC
and r-APC mutants in the amidolytic assays, lack of a single
gla residue at any of the following locations, viz, 7, 16,20, or
26, led t o virtual complete disappearance of the Ca2+dependent anticoagulant activity of the relevant r-APC mutant, as compared with its wt counterpart. On the other hand,
single eliminations of any of the gla residues located at
positions 6,14, or 19 of r-APC resulted in variant recombinant
molecules with substantial anticoagulant activity (80% t o
92%), relative t o wtr-APC. Mutation of gla residues at
positions 25 and 29 resulted in r-APC variants with significant
but low (24% and 9% of wtr-APC, respectively) levels of
anticoagulant activity. The variant, [R15L]r-APC, possessed
only 19% of the anticoagulant activity of wrt-APC, while
inclusion of gla at position 32 in the variant, [F31L, Q32glalrAPC, resulted in a recombinant enzyme with an anticoagulant activity equivalent t o that of wtr-APC.
o 1992by The American Society of Hematology.
H
into blood6 and/or its ability to interact productively with
inhibitors of plasminogen activators (PAI-l), thereby precluding their participation in inhibition of plasminogen
activation.' While complexes of PAI-1 and APC have been
observed,s the importance of this mechanism toward enhanced fibrinolysis in vivo has been questioned? More
recently, it has been proposed that APC inhibits an antifibrinolytic component derived from activation of prothrombin, thus promoting a fibrinolytic response.1° PS also
appears to be a cofactor for the fibrinolytic response of
PC."
The coding regions of the human PC gene are contained
in eight exonsI2that translate into a 42-amino acid residue
leader sequence, followed consecutively by a 155-amino
acid residue light chain and a heavy chain of 262-amino
acids. The catalytic triad of H211, D255, and S360 exists in
this latter domain region. A dipeptide, K156-Rl57, is
present in the translated protein that connects the two
chains, and its removal during normal processing is critical
to forming an activatable form of PC.*3,14Other processing
steps that generate mature plasma PC include cleavage of
the leader polypeptide,I2 y-carboxylation of the nine gla
precursor E residues, present at amino acid sequence
positions 6,7, 14, 16,19, 20, 25,26, and 29 of human PC,14,15
P-hydroxylation of D71,I6 and glycosylation of N97, N248,
N313, and N329.l' Based upon consideration of the positions of the introns in the gene, and amino acid sequence
homologies with other proteins, it is clear that PC contains
several domains. These include the y-carboxyglutamic acid
(g1a)-rich amino terminal polypeptide, followed by two
consecutive regions homologous to epidermal growth factor, an activation peptide, and the serine protease catalytic
region. Fully processed recombinant (r) human PC18J9and
r-APCZ0have been expressed in mammalian cells, accomplishments that allow structure-function investigations of
UMAN PROTEIN C (PC) is a member of the family
of vitamin K-dependent blood coagulation proteins
and exists in plasma as a zymogen. The activated form of
this protein (APC) functions to maintain blood fluidity by
functioning both as an anticoagulant and as a profibrinolytic agent. Activation of PC occurs at the endothelial cell
surface and results from limited proteolytic reactions catalyzed by thrombin/Ca2+in conjunction with the membrane
protein cofactor, thrombomodulin.' Much slower activation
of PC can occur in plasma, catalyzed by thrombin alone. In
this latter case, CaZ+inhibits the reaction.2 The mechanism
of the anticoagulant activity of APC is well established.
Here, APC, a serine protease, catalyzes specific limited
proteolytic events that lead to inactivation of coagulation
cofactors V (fv) and Va (fVa),3 as well as cofactors VI11
(fVIII) and VIIIa (fVIIIa).4 These inactivation reactions
are stimulated by Ca2+ and phospholipid (PL)? along with
a cofactor, protein S (PS).5 The fibrinolytic response of
APC involves several possible mechanisms, viz, its stimulation of release of plasminogen activators from vasculature
From the Department of Chemistry and Biochemistry, University of
Notre Dame, Notre Dame, IN.
Submitted December 24, 1991; accepted April 14, 1992.
Supported by Grant No. HL-19982 from the National Institutes of
Health and the Kleiderer-Pezold family endowed professorship (to
F.J.C.).
Address reprint requests to Francis J. Castellino, PhD, Department
of Chemistry and Biochemistry, University of Notre Dame, Notre
Dame, IN 46556.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked
"advertisement" in accordance with 18 U.S.C. section 1734 solely to
indicate this fact.
0 I992 by The American Society of Hematology.
0006-4971J9218004-0004$3.00/0
942
Blood, VOI 80, NO4 (August 15), 1992: pp 942-952
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
PROTEIN C MUTAGENESIS
r-PC and r-APC to be designed through site-directed
mutagenesis.
The posttranslational processing of appropriate precursor E residues to gla residues allows vitamin K-dependent
blood coagulation proteins to bind to Ca2+and PL, interactions that are essential to their proper functioning.2l Regarding PC, it is clear that incomplete y-carboxylation of
random gla-precursor E
as well as proteolytic
removal of the entire gla domain of plasma PC, leads to a
substantial diminution of the anticoagulant activities of
their activated forms, as shown by the reduction of the
APC-induced prolongation of activated partial thromboplastin times (APTT) of plasma. Because thromboembolic
complications in patients arising from PC deficiencies are
well-known clinical
and the increasing use of
polymerase chain reaction (PCR) technology has led to
identification of patients with defined amino acid mutations
in this protein,25we have embarked on a program aimed to
show the identity of specific amino acids that play major
roles in the anticoagulant activity of PC. To date, using
recombinant DNA technology, we have shown that double
mutations that prevent y-carboxylation at the E pairs
present at r-PC sequence positions 6 and 7,19 and 19 and
20,26residues conserved in all vitamin K-dependent coagulation proteins, greatly reduce the anticoagulant properties
of the resultingvariant r-APC molecules. More recently, we
have constructed, expressed, and purified r-PC molecules
containing individual mutations at all gla-precursor E
residues. This allowed us to identify which of these are
critical to the anticoagulant properties of APC. This report
is a summary of our findings.
MATERIALS AND METHODS
Proteins. Human 293 kidney cell-expressed wild-type (wt) r-PC
was generated and purified as described p r e v i o ~ s l y .Human
~~
plasma PC was donated by Enzyme Research Laboratories, Inc
(South Bend, IN).
All r-PC and variant r-PC activations were performed as described earlier,19 using the Agkisrrodon contortix contortix venom
activator of PC, Protac C, purchased from American Diagnostica
(New York, NY). The temporal progress of each activation was
monitored spectrophotometrically by the appearance of amidolytic
activity of r-APC toward the substrate, L-pyroglutamyl-L-prolyl-Larginine-p-nitroanilide (substrate ,52366; Helena Laboratories,
Beaumont, TX).
Murine monoclonal anti-PC, C3,27 was provided by Dr John
Griffin (La Jolla, CA) and used in Western analysis for screening
clones of transfected 293 cells for r-PC production. Murine
monoclonal antibody (MoAb), LI, a Ca2+-independent antibody
that recognizes a determinant in the protease domain of PC, was
generated in this laboratory2* and used to prepare immunoaffinity
chromatography columns to assist in the purification of some of the
variant r-PC molecules.
Restriction endonucleases were purchased from Fisher Scientific (Springfield, NJ) and BRL (Gaithersburg, MD). These enzymes were used according to the manufacturers' recommendations.
Genes for wt and mutant r-PC molecules. The cDNA coding for
wtr-PC was provided by Dr Earl Davie (Seattle, WA) in pUC119.
The changes that we incorporated into the cDNA for use with our
293 kidney cell expression system and the plasmid expression
vector used (pCIS2M) have been described in detail.19
943
The cDNAs coding for each of the r-PC mutants described were
constructed from the wt cDNA for human PC, or an r-PC mutant
cDNA, by site-directed mutagenesis using synthetic oligonucleotide primers. Screenings of positive transformants were accomplished by the gain or loss of suitable restriction endonuclease sites.
The presence of the proper nucleotide insertions in the relevant
cDNAs was evaluated by DNA sequence analysis before their
placement in the expression vectors. Insertion of the variant
cDNAs into the expression plasmid, pCIS2M, was accomplished as
for the wt-PC gene.19
Transfection in adenovirus-transformed human kidney 293 cells
(ATCC CRL. 1573). The 293 cell growth conditions, the procedures for their transfection with pCIS2M containing the cDNA of
interest, the selection and propagation of clones expressing wtr-PC
or the relevant r-PC mutants, and the harvesting of conditioned
cell media were as published previo~sly.'~
Purification of the recombinant proteins. Purification of wtr-PC
with selection for a subpopulation of PC molecules that contained
the maximal level of gla residues was conducted as described
earlier.19 Similar techniques were used for the purification of all
mutants, with minor operational modifications depending on the
elution behavior from the anion exchange columns and their
purities as shown by sodium dodecyl sulfate/polyacrylamide gel
electrophoresis ( D O ~ S O ~ / P A G EThe
) . ~following
~
general procedure sufficed in all cases to provide r-PC and r-PC mutants
containing gla residues at all available precursor E residues.
Benzamidine-HCl(5 mmol/L, final concentration) was added to
the conditioned media of the 293 cells. Approximately 1.5 L of this
solution was dialyzed against a buffer of 20 mmol/L Tris-HCl/lSO
mmol/L NaCI, pH 7.4 (column buffer), containing 5 mmol/L
benzamidine-HC1/4 mmol/L EDTA, and passed over a 5 mL
column of Pharmacia Fast Flow Q (FFQ) anion exchange resin
(Pharmacia, Piscataway, NJ), equilibrated with column buffer/4
mmol/L EDTA at 4°C. After washing the resin with 3 column
volumes of this buffer, followed by 3 column volumes of the same
buffer without EDTA, the r-PC, or r-PC variant, was eluted in one
step with column buffer/30 mmol/L CaC12, pH 7.4. At this stage,
the total sample volume was approximately 20 mL. Several such
samples were combined and, after dialysis against column buffer
containing 4 mmol/L EDTA, the dialysate was repassaged over a
column containing 3 mL of this same resin equilibrated with
column buffer at 4°C. The relevant r-PC or r-PC mutein was eluted
at a flow rate of 0.25 mL/min with a CaC12 gradient (start solution,
30 mL of column buffer; limit solution, 30 mL of column buffer/30
mmol/L CaC12, pH 7.4), as described.22 Fractions (1 mL) were
obtained and those containing r-PC, which were readily identified
by DodSOd/PAGE, were pooled and the CaCl2 removed by dialysis
against column buffer. Next, the protein was readsorbed to 3 mL of
a second FFQ column, equilibrated in column buffer at 4"C, after
which the r-PC was eluted at a flow rate of 0.25 mL/min with an
NaCl gradient (start solution 30 mL of column buffer; limit
solution, 30 mL of 20 mmol/L Tris-HC1/500 mmol/L NaCl, pH
7.4). r-PC-containing fractions (1 mL) were identified by DodS04/
PAGE and pooled. Some r-PC variants show slightly altered
column behavior from the wt-protein, and from each other, but use
of this procedure results in readily identifiable, highly homogeneous material. We have repeatedly observed for a variety of single
and double gla mutant proteins that the r-PC material at this stage
contained maximally y-carboxylated protein (at least seven residues of gla per mole of PC). Subpopulations of PC molecules
containing lower levels of y-carboxylation, the amounts of which
were considerable with some recombinant mutants, displayed very
different chromatographic behavior on each column. Generally,
incompletely y-carboxylated subforms were much more weakly
adsorbed to the first column.
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
944
In the few cases in which a small amount of non-r-PC material
was present after this last chromatography step, final purification
was accomplished using affinity chromatography with a column
(0.6 x 5 cm) of Sepharose-MoAb LI, equilibrated in column
buffer. After washing with this same buffer, the material that
eluted from the column upon treatment with 0.1 mol/Lglycine, pH
2.7, was dialyzed against 20 mmol/L Tris-HC1, pH 7.4. The
partially purified r-PC or r-PC variant was then reapplied to the
FFQ anion exchange column, which was equilibrated in this same
buffer. The column was then washed with a solution of 20 mmol/L
Tris-HC1/200 mmol/L NaC1, pH 7.4, and eluted with a buffer of 20
mmol/L Tris-HC1/500 mmol/L NaCl, pH 7.4. The highly purified
r-PC preparation was then dialyzed against 50 mmol/L NH4HC03,
followed by HzO.
Gla determinations. These experiments were performed by
amino acid analysis after alkaline hydrolysis. The hydrolyses were
conducted with 2.5 mol/L KOH for 20 hours at llO"C, and KOH
was removed from the samples as reported p r e v i o ~ s l yAmino
.~~
acid analysis by reverse-phase high performance liquid chromatography (HPLC) was performed after precolumn derivatization with
o-phthaldehyde/p-mercaptoethanol,using a 4.6 mm (ID) X 70 mm
Ultrasphere XL ODS column (3.0 p) with a precolumn (Beckman
Instruments, Fullerton, CA) step. The buffer used was 90% 0.1
mol/L NaOAc, pH 7.2/9.5% methanol/0.5% tetrahydrofuran. The
flow rate of the column was 1.5 mL/min. This chromatographic
step resulted in baseline resolution of gla (retention time, 0.8
minutes), D (retention time, 2.0 minutes), and E (retention time,
4.1 minutes). The ratios of gla/D and gla/E were used to obtain the
number of gla residues per mole of protein.
We used as reference standards both a commercial o-phthaldehyde-derivatized amino acid standard mixture as well as a peptide,
ANSFL(gla)(gla)LRSSL, synthesized by standard Fmoc chemistry
(the a-N-Fmoc-y,y'-ditBu-L-gla-OH
used in the peptide synthesis
for placement of gla in the peptide was chemically synthesized by
Dr Sushi1 Sharma in this laboratory). Use of this peptide for this
purpose allowed us to obtain recovery factors after alkaline
hydrolysis of the peptide and to determine very accurate conversion factors of peak areas to concentrations of gla residues in the
hydrolyzed samples. The gla/D ratio in this peptide of 2.0 (N is
converted to D as a result of hydrolysis), along with the concentration response factor for D from a commercial standard amino acid
solution, was used to obtain the concentration response factor for a
gla residue. Checks of the method were made by performing gla
residue analyses on human plasma PC and bovine plasma factor X.
Amino acid sequence analysis of various PC molecules. Automated amino terminal amino acid sequence analysis of each r-PC
sample was conducted as previously d e ~ c r i b e d .In
~ ~all cases,
protein sequencing was successful through 35 residues, steps that
were sufficient to obtain sequence information through the entire
gla-domain.
Amidolytic assays of r-APC and its variants. Optimal conditions
found for activation of r-PC and the variants described herein have
been described earlier.19 For amidolytic assay of the resulting
r-APCs, solutions were prepared with 20 to 400 pL of substrate
S2366 (stock solution, 4.2 mmol/L in HzO), 40 pL of Tris-HCI, pH
7.4 (stock solution, 1 mol/L), and 80 pL NaCl (stock solution, 1
mol/L). A quantity of H20 was added to adjust final volumes to
0.79 mL. The amidolytic reaction was accelerated as a result of
addition of 10 pL of enzyme and rates of p-nitroanilide release
were monitored by the increase of absorbancy at 405 nm. Km and
kcat values were calculated from Lineweaver-Burk plots of the
variation in initial velocities as a function of S2366 concentrations.
APTT assays. For this assay, the concentrations of wtr-APC,
human plasma APC, and all recombinant variants investigated
were adjusted so that each APTT assay was conducted at the same
ZHANG, JHINGAN, AND CASTELLINO
amidolytic activity of the particular APC under study. The concentration range of human plasma APC in the assay solutions was
0.025 to 0.4 pg/mL and was readjusted for the muteins, depending
on the clotting times observed.
APTT assays were conducted at 37°C with PC-deficient plasma
using the APTT assay kit (Sigma Diagnostics, St Louis, MO),
essentially as described earlier.I9 Controls in this assay were
performed in the absence of APC and with unactivated r-PCs in
place of their respective APCs. For the assay, a volume of 100 pL of
APTT reagent was added to the reaction vessel and allowed to
temperature-equilibrate at 37°C for 1minute. This was followed by
the addition of 10 pL of the desired concentration of the r-APC
under examination and 90 pL of PC-deficient human plasma. After
an additional 3 minutes of incubation, clotting was initiated by the
addition of 100 p L of a 25 mmol/L stock solution of CaC12,
pre-equilibrated at 37°C. Clotting times were measured with a
commercial fibrometer (Fisher Scientific, Springfield, NJ). Our
standard clot times (approximately 45 seconds) were essentially the
same as those reported by the manufacturer of the assay kit.
For calculation of the results, plots of the logarithm of clotting
time versus the logarithm of the dilution of the stock solutions of
the relevant APC were constructed for each APC sample. The
displacement of the sample line from the wtr-APC sample (the
dose-response curves for the different r-APC mutants were parallel) was used to calculate the relative potencies of each sample.
The specific activity of the wtr-APC sample in each assay was
obtained by comparison of the same graphs constructed for plasma
APC and wtr-APC. The specific activity of wtr-APC was calculated
using the value of 250 U/mg for human plasma APC'* and the
percent activity of wtr-APC relative to plasma APC. The specific
activities of each variant r-APC were calculated from the specific
activity of wtr-APC and from the above described direct determinations of the relative activities of wtr-APC and the variant r-APC of
interest.
Analyticalmethods. Our procedures for oligonucleotide synthesis, cDNA purification and sequencing, cell transfections, sitespecific mutagenesis, and Western analysis of the expressed PC
molecules have been previously described in detai1.*9,26
RESULTS
A number of strategically important recombinant gladomain variants of PC have been constructed, expressed,
and purified. These are listed in Table 1, along with the
oligonucleotides used for mutagenesis and the restriction
endonuclease sites used for screening positive bacterial
transformants. The translated amino acid sequence of the
gla-domain of human PC is illustrated in Fig 1, with
indications of the sequence positions of the amino acids
mutated in this investigation. After translation of each
cDNA in human kidney 293 cells, it was necessary not only
to purify the r-PC, but also to isolate the subpopulation of
r-PC molecules that contained gla residues at all precursor
E residues. We have found that some of the variants,
particularly [gla20D]r-PC, [glal6D]r-PC, and [gla26V]rPC, possessed only a small percentage (approximately 10%
to 20%) of maximally y-carboxylated r-PC antigen, whereas
in other variants, viz, [gla6D]r-PC and [gla7D]r-PC, a much
higher percentage (approximately 80% to 90%) of the total
r-PC antigen was y-carboxylated at all available precursor E
residues. Regardless of this circumstance, the purification
procedure that we used was very satisfactory for isolation of
maximally y-carboxylated subpopulations of r-PC variants,
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
PROTEIN C MUTAGENESIS
945
Table 1. Construction of Human r-PC Variants
Mutation
Primer
Screen
gla6D
gla7D
glal4D
R15L'
glal6Dt
glal9D
gla20D
gla25D
gla26V
gla29D
F31L, Q32gla
5'-TCCTTCCTGGAtGAaCTCCGTCACAGC
5'-CCTGGAGGAcCTCCGTCACAGC
5'-GCAGCCTGGAtCGcGAGTGCATAGA
5'-AGCCTGGAGCtcGAGTGCATAGAG
5'-CAGCCTGGAGCgCGAcTGCATAGAGGA
5'-GTGCATAGAcGAaATCTGT
5'-TGCATAGAGGAtATCTGTGAC
5'-TGTGACTTCGAtGAGGCgAAGGAAATT
5'-CTTCGAGGtGGCCAAGG
5'-AGGCCAAGGAtATcTTCCAAAATGT
5'-GCCAAGGAAATTcTagAAAATGTGGATG
-Sac I
-Sac I
+Nru I
+ Sac I
-Sac I
-Bgl II
-Bg/ II
-Sv I
+Bal I
+EcoRV
+Xba I
The lower case letters represent the mutagenic bases. The mutation
convention used is the normal amino acid and its sequence position in
PC, followed by the single letter code for the new amino acid placed in
that sequence position by mutagenesis. Gla refers to y-carboxyglutamic acid.
'The starting wt-cDNA was first mutated at the third base of the
codon for L8 to eliminate the interfering Sac I restriction endonuclease
site.
tThe starting cDNA was [R15L]r-PC.
and in all cases described herein isolation of appropriate
proteins was accomplished.
The first step of the purification used Ca2+-basedelution
from an FFQ anion exchange column, an example of which
is shown in Fig 2A, and provided PC molecules containing
at least seven residues of gla in the protein material eluted
with the Ca2+ gradient.19,22.26The second step of the
purification (Fig 2B) is based on elution from the same
chromatography resin with NaCl and results in removal of
small amounts of contaminating protein. The elution positions of the different variants described herein may differ
somewhat from each other and from the example provided
in Fig 2, but the major protein peaks observed are the
desired r-PC molecules. Finally, when needed (eg, [glal6D]rPC, [gla20D]r-PC, and [gla26V]r-PC), a further step on an
immunoaffinity column resulted in highly purified protein.
From our selected clones, the overall yields of the maximally y-carboxylated r-PC variants described varied considerably, with the lowest amount obtained for [glal6D]r-PC
of approximately 50 pg/L of conditioned cell culture
medium, to the highest amount obtained for [gla6D]r-PC of
c
0'648
.
5
2
184
(E
W
0
u 0.504
E
n
+
N
O
0
U
F r a c t i o n number
KI
$
I
n
-1220
0.09
6
18
30
42
F r a c t i o n number
54
-
I50
Fig 2. Purification of maximally y-carboxylated r-PC on FFQ anion
exchange chromatography at 4°C. The example provided is for
[glaGD]r-PC. (A) After batch elution with 30 mmol/L CaCI, of the
[glaGDIr-PC contained in 1.5 L of conditioned human kidney 293 cell
culture medium, three such preparations were combined. After dialysis against 20 mmol/L Tris-HCI/lW mmol/L NaC1/4 mmol/L EDTA,
pH 7.4, the resulting sample was reapplied t o the same column (3 mL),
equilibrated with a buffer of 20 mmol/LTris-HCl/l5Ommol/L NaC1/4
mmol/L EDTA, pH 7.4 at 4°C. The column was washed with this buffer,
followed by this buffer without EDTA, after which the indicated CaCl,
gradient was applied (60 mL, total volume). Fractions (1 mL) were
collected at a flow rate of 0.25 mL/min. The maior peak contains the
desired material. (B) The partially purified [glaGDIr-PC isolated in (A),
above, was dialyzed against a buffer of 20 mmol/L Tris-HCV150
mmol/L NaCI, pH 7.4 at 4°C. and reapplied t o a 3 mL column of FFQ
resin equilibrated in 20 mmol/L Tris-HCI/ 150 mmol/L NaCI, pH 7.4 at
4°C. The indicated gradient of NaCl was applied (60 mL, total volume)
and 1 mL fractions were collected. The major peak contains highly
purified [glaGDIr-PC.
approximately 500 pg/L of conditioned medium. For wtrPC, the usual yield of fully y-carboxylated material was
approximately 3 mg/L of conditioned medium.
DodS04/PAGE gels for the r-PC variants used in this
6
7
14
15
Fig 1. Primary structure of the gla-domain of
human PC. The region displayed spans from the
amino terminus of the PC light chain t o T37, which is
the last amino acid in the exon coding for the
gla-domain. y-Carboxyglutamic acid is represented
by Gla.
32
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
ZHANG. JHINGAN, AND CASTELLINO
946
investigation arc provided in Fig 3. In all cases, highly
homogcncous maximally y-carboxylated material has bccn
obtained. Thc two componcnts obscrvcd for cach PC
represent thc usual casc,1y.2?.?h
and arc rclatcd to variable
glycosylation of the four rclcvant N rcsiducs in thc protein..'? Wc dctcrmincd thc gla content of cach protein, and
the results listcd in Table 2 show that all of thc mutants
contained thc expected levels of gla. Additionally, aminoterminal amino acid sequence analysis for cach isolatcd
recombinant protein has been pcrformcd through 35 rcsiducs. In all cases, the desired mutation was prcscnt at thc
anticipated location, and no morc than a 5% rcpctitivc
yield of E was obscrvcd at any of thc scquencc positions
that should havc undergone y-carboxylation. Thcsc abscnccs of gla-prccursor E rcsiducs showcd that thc r-PC
variants that wc havc isolatcd and invcstigatcd containcd
the maximal lcvcls of gla rcsiducs rcsulting from processing
of thc appropriate prccursor E rcsiducs.
Wc havc convcrtcd cach r-PC variant to its rcspcctivc
APC by activation with thc snakc vcnom protcasc, Protac
C. This reaction does not rcquirc Ca?+,and it was cxpcctcd
that the gla-deficient r-PC mutants, dcspitc having probable variations in Ca?+binding propcrtics, would bc noncthcless fully convcrtcd to thcir rcspcctivc r-APC forms. While
we havc not as yet investigated possiblc diffcrcnccs in
activation kinetics of this reaction for thc r-PC variants, wc
did find that each r-PC was completely activatablc to its
corrcsponding r-APC. This includcd the [ RlSLIr-PC and
A
1 2 3 4 5 6 7
B 1 2 3 4 5 6 7 8
Fig 3. Nonreduced sodium dodecyl sulfate polyacrylamide gel
electrophoretograms of all purified r-PC variants used in this study.
The samples displayed are as follows. (A) Lane 1, human plasma PC;
2, 293 cell-expressed wtr-PC; 3, [glaGDlr-PC; 4, Igla7DIr-PC; 5,
Iglal4DIr-PC; 6, [RlSLIr-PC; 7, [glalGDlr-PC. (B) Lane 1, human
plasma PC; 2,293 cell-expressed wtr-PC; 3, IglalSDJr-PC;4, [glaZODIrPC; 5, [gla25D]r-PC; 6, [glaZGVlr-PC; 7, tgla29DIr-PC; 8, [F31L, 032glalrPC.
Table 2. y-Carboxyglutamic Acid and Contents of Various r-PC
Mutants
Gla (moleslmolel
Protein
Expected
Obtained
Bovine factor X
Human PC
wtr-PC'
[glaGDIr-PC
Igla7Dlr-PC
Iglal4Dlr-PC
[R 15Llr-PC
IglalGDIr-PC
[glalSDIr-PC
Igla20Dlr-PC
IgIa25Dlr-PC
[gla26Vlr-PC
[gla29D]r-PC
IF31L. 032glalr-PC
12.0
9.0
9.0
8.0
8.0
8.0
9.0
8.0
8.0
8.0
8.0
8.0
8.0
10.0
11.7 2 0.4
8.8 2 0.3
8.9 2 0.3
8.3 2 0.2
8.3 z 0.3
7.8 2 0.3
9.0 2 0.3
7.9 2 0.2
8.0 2 0.3
8.2 2 0.2
7.7 2 0.3
7.9 2 0.2
7.9 2 0.2
10.0 2 0.3
The mutation convention used is the [normal amino acid, its sequence position in PC, single letter code for the new amino acid placed
in that sequence position by mutagenesisl, followed by r-PC. Gla refers
to ycarboxyglutamic acid.
*Expressed in human kidney 293 cells.
[F31L, Q32Elr-PC mutcins. Additionally. we dctcrmincd
Km and kcat valucs, at pH 7.4 and 37°C. for wtr-APC and
rcprcscntativc variants that wcrc activc (Iglal9DIr-APC)
and inactive ([gla7D]r-APC) in A P T assays (vide infra).
For thcsc proteins, thc Km value for substratc S2366
rangcd from 1.1 to 1.4 mmol/L and thc kcat for this same
substratc rangcd from 170 to 182 s-I. This shows that thc
amidolytic activities arc csscntially thc samc, a fact that
would bc cxpcctcd from litcraturc p r c c ~ d c n t s . ~ ~ ~ ' ' ' ~ ~ ~
With thc availability of thc complctc sct o f single glamutated r-APC molcculcs, as wcll as somc othcr stratcgically mutated r-APC variants, the major goal of this
investigation was to assess thc rolc of cach gla rcsiduc as a
dctcrminant of thc in vitro anticoagulant activity of APC.
For this purpose, wc chosc to usc thc clinically rclcvant
A P T assay with PC-dcficicnt human plasma. Wc bclicvcd
it most important to diffcrcntiatc this specific function of
APC from its gcncral amidolytic activity. and thus chosc to
adjust thc APC solutions in thc assay such that thcy wcrc
prcsent at cqual amidolytic activities, which csscntially
corresponds to thc samc protein conccntrations in thc cascs
of thcsc particular mutations. As an cxamplc of thc data
obtained, an illustration of thc clotting timc data obtaincd
for wtr-APC and onc of thc gla variants, [gla25D]r-APC, is
provided in Fig 4. This typc of analysis has bccn madc for
cvcry gla mutant protein invcstigatcd. Thc resulting anticoagulant activitics of cach of thc recombinant variants arc
summarizcd in Table 3 and comparcd with wtr-APC, which
has bccn cxprcsscd in thc samc ccll linc. Thc rcsults clcarly
show that anticoagulant activity has bcen grcatly reduced
for recombinant variants that lack gla rcsiducs at any of thc
following scqucncc positions. viz, 7, 16,20, and 26. whcrcas
80% to 92% of thc anticoagulant activity of wtr-APC has
been observcd in recombinant variants that lack gla rcsiducs at any of thc following positions, viz, 6, 14, and 19. Thc
remaining two mutants containing substitutions at gla
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
PROTEIN C MUTAGENESIS
I
200
I
.
""
947
eluted at low pH, we examined the effect of this treatment
on the APTT activities of wtr-APC and two active mutants,
[gla6D]r-APC and [glal4D]r-APC. These samples were
adsorbed and eluted from the column as described, and
activated to their respective r-APCs. The specific activities
were 320 U/mg, 284 U/mg, and 290 U/mg, respectively.
This shows that this purification step does not affect the
functional property of interest.
I
DISCUSSION
I
....
I
I
0.1
1
I
5
Relative dilution
Fig 4. Assay of the overall anticoagulant activity of APC using the
APlT times with PC-deficient human plasma at 37°C. A stock solution
of wtr-APC was prepared and its amidolytic activity determined. A
similar stock solution of the desired r-APC mutant was prepared and a
minor adjustment of its concentration was made such that all
amidolytic activities were identical. A dose-response curve was
obtained for each sample and plotted as the APIT times (clotting
times) versus the relative dilution of the wtr-APC stock solution (stock
solution of wtr-APC = 1.0, final concentration approximately 0.3
pg/mL). The displacement of the lines provided the activity of the
r-APC sample relative to that of wtr-APC (the lines were parallel in all
cases). ( 0 )wtr-APC; (W) Igla25DIr-APC.
residues 25 and 29 showed significantly reduced levels of
anticoagulant activity, yielding values of 24% and 9%,
respectively, of wtr-APC. It has also been shown that R15 is
an important residue in allowing Caz+-dependentanticoagulant activity to be expressed, because only a small amount
of anticoagulant activity has been found, ie, 19% relative to
wtr-APC, consequent to the R15L mutation. Finally, inclusion of a gla residue at sequence location 32 did not
significantly influence the anticoagulant activity of the
resulting gla32-derived recombinant variant.
In addition, because some of the mutants have been
purified by adsorbtion to an MoAb affinity column and
Table 3. Anticoagulant Properties of r-APC Variants
Protein
Specific Activity
Wmg)
wtr-APC
Activity ( O h )
wtr-APC*
[glaGDJr-APC
Igla7DIr-APC
[gla14DI-APC
[R15L]r-APC
[gla 16DIr-APC
[gla1SDIr-APC
[gla20D]r-APC
Igla25Djr-APC
[gla26V]r-APC
[gla29D]r-APC
[F31L, Q32gIalr..APC
325 f 14
279 f 12
19 f 5
299 f 10
62 -+ 9
<6
260 2 11
<6
78 2 9
<6
29 f 6
319 f 13
100
86 f 4
622
92 2 3
19 f 3
<2
80 f 4
<2
24 f 3
<2
9 f 2
98 f 4
The mutation convention used is the [normal amino acid, sequence
position in PC, single letter code for the new amino acid placed in that
sequence position by mutagenesis], followed by recombinant r-APC.
Gla refers to y-carboxyglutamic acid.
*This protein and all muteins were expressed in human kidney 293
cells.
We have prepared and successfully expressed each of the
nine possible r-PC variants containing mutations at E
residues that would become y-carboxylated in the mature
protein. The fact that some of the recombinant mutants
contained a considerable amount of under-y-carboxylated
r-PC molecules may suggest that some recognition sites for
the y-carboxylase that catalyzes these reactions exist in the
gla-domain, an observation made
but which has
been ~ h a l l e n g e d ? ~However,
-~~
our investigation was not
designed to fully explore this question. In any case, a highly
homogeneous subpopulation of r-PC has been isolated for
each mutein. Each of these proteins contained the appropriate content of gla residues. Further, amino acid sequence
analysis has clearly shown that the mutation placed in the
corresponding cDNA was properly translated and that each
mutein was y-carboxylated at all available gla-precursor E
residues. Thus, the purified proteins were appropriate for
the planned investigations intended to show the importance
of individual gla residues of r-APC in its Ca2+-dependent
anticoagulant activity.
Each r-PC was converted to its corresponding r-APC as a
consequence of activation by Protac C, a reaction that does
not require Ca2+.37338
The amidolytic activity of the variant
r-APC preparations did not substantially differ from each
other or from wtr-APC, a result consistent with earlier
studiesls that showed that extensively under-y-carboxylated r-APC expressed in the presence of warfarin possessed a specific amidolytic activity similar to its wt counterpart. Each generated r-APC was then assayed for its
Ca2+-dependent anticoagulant activity. From the results
obtained, shown in Table 3, it is concluded that gla residues
at amino acid sequence positions 7, 16, 20, and 26 are
essential for the anticoagulant activity of APC, whereas
those present at amino acid sequence positions 6,25, and 29
are also of some importance to this same activity. The
muteins expressing the highest anticoagulant activity, ie,
80% to 92% of wtr-APC, were [gla6D]r-APC, [glal4D]rAPC, and [glal9D]r-APC, suggesting that gla at sequence
positions 6,14, and 19 in wtr-APC are the least important to
its specific function. These findings correlate well with our
previous studies showing that pairwise mutations of gla
residues present at amino acid sequence positions 6 and 7,19
as well as 19 and 20,26led to a virtual complete elimination
of the anticoagulant activity of each mutant r-APC. In the
former recombinant mutant, it is now clear that this
previously reported loss of activity was a result of mutation
at gla7, whereas in the latter case, mutation of gla20 alone
would have been sufficient for the observed loss of anticoagulant activity. Of significance, it also appears that R15, a
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
948
residue conserved in all vitamin K-dependent human coagulation proteins, is extremely important to the Ca2+dependent anticoagulant activity of APC (Table 3), because
only 19% of the activity of wtr-PC remained after alteration
of this amino acid to a noncharged variant containing an
L15 residue. Additionally, alteration of the Q32 in r-APC to
contain E32, which did in fact undergo y-carboxylation
during processing, led to a variant r-APC that contained full
anticoagulant activity. While positioning of gla residues in
other vitamin K-dependent blood coagulation proteins is
highly conserved up to amino acid sequence position gla29
(or its homologous residue), there are differences at sequence locations beyond this point. For example, human
factor IX contains additional gla residues at sequence
positions 33,36 and 40,39human prothrombin possesses an
additional gla32,4O and human factor VI1 contains gla35.41
Because inclusion of a gla at residue 32 of r-APC did not
affect its anticoagulant activity, it appears that APC represents a functioning vitamin K-dependent protein with the
minimum necessary levels of gla residues. Proteins containing additional gla residues may require such amino acids for
their specific functions and/or their specific cellular properties.
The results reported herein will be of great value in
relating mutations in the gla domain of APC, which are
being increasingly discovered with PCR technology in
symptomatic patients, with the thrombotic complications
observed in these same patients. As examples, in the recent
past, several gla mutations have been found in vitamin
K-dependent proteins in patients with clotting abnormalities. In factor IX-Nagoya 4, the nucleotide sequence of the
subject DNA was predictive of a translated protein with an
E21 (equivalent to E20 in PC) K mutation. This patient,
clinically classified as a severe B-type hemophiliac, possessed a Ca2+-dependentfactor IX coagulant (fIX:c) activity less than 1% of
This low level of residual
activity is analogous to the situation with [gla20D]r-APC,
wherein the gla-dependent activity of this latter protein in a
clotting assay was essentially eliminated despite a much
more conservative mutation at this same gla residue.
Similarly, in other severe hemophilia B patients with known
gla-domain variations, viz, factor IX-Seattle 343and factor
IX-Chongqing,4 DNA mutations that would be predicted
to translate into proteins with E27 (equivalent to E26 in
PC) K (fIX:c < 1%)and E27 (equivalent to E26 in PC) V
(fIX:c < 1%) alterations, respectively, have been found.
We generated an r-PC with this latter mutation, ie,
[gla26V]r-PC, and found that the corresponding [gla26V]rAPC also nearly completely lost its Ca2+-dependentactivity
in a clotting assay (Table 3). Thus, gla residues 20 and 26 in
human APC possess similar essential features to the corresponding residues 21 and 27 in human factor IXa, despite
the fact that factor IXa exhibits procoagulant activity,
whereas APC possesses anticoagulant activity. This suggests that the integrity of gla at these sequence positions is
of great importance to the Ca2+-dependentfunctions of
both proteins, probably related to common essential features of Ca2+binding, and this may be a uniform functional
property of other vitamin K-dependent coagulation pro-
ZHANG, JHINGAN, AND CASTELLINO
teins. One gla mutation in PC has been discovered, viz,
[gla20A] (along with [V34M]), in a patient with symptomatic type 2 PC deficien~y.4~
Assuming that the thrombotic
complication was due to the gla20A mutation, and not that
of V34M, this finding would be entirely predictable from
our data (Table 3), because the much more conservative
mutation, gla20D, led to nearly complete loss of the
Ca2+-dependentanticoagulant activity of the mutant r-APC.
Another symptomatic patient with a dysfunctional PC, PC
Yonago, has been identified, with an R15G mutation in the
g l a - d ~ m a i n .Our
~ ~ results with [R15L]r-APC, which possessed only 19% of the activity of wtr-APC, also are
predictive of a thrombotic state in a patient homogyzous for
this mutation.
A final factor IX mutation of interest to this investigation
has been found in a symptomatic patient with factor IX
HB9, wherein a gla33D replacement (fIXc = 4%) has
been di~covered.~’
Because this gla residue was found to be
of such importance for the clotting activity of factor IX, and
because PC does not contain a gla precursor E residue at
this location, we decided to substitute an E-residue for the
equivalent Q32 in a mutant r-PC to determine whether
r-APC activity would be influenced by this change. As seen
in Table 3, the anticoagulant activity of this mutein was
found to be approximately the same as that of wtr-APC.
Thus, a gla residue at this location does not appear to be of
similar importance to APC anticoagulant function as it is
for factor IX coagulant activity. In this r-APC variant, an
additional F31L mutation was incorporated due to the
nature of the oligonucleotide primer used that was found to
be highly suitable for screening bacterial transformants.
Because such a high level of anticoagulant activity was
obtained for this variant, we decided not to realter L31 to
its natural F residue. Additionally, L is present at this
position in human prothrombin and is obviously not harmful to expression of Ca2+-dependent properties of this
protein at that location.
It should be emphasized that we have intentionally
engineered very conservative gla mutations into r-APC
because our purpose was to discover which of the gla
residues in this protein were essential to its Ca2+dependent anticoagulant properties. We believe that when
such activity is lost with D-replacements for gla residues, it
is likely that any other substitution will also result in at least
this level of activity loss. As additional patients are discovered with other amino acid replacements for apparently
nonessential gla residues, these will be inserted and investigated in this manner, as was accomplished here with the
gla26V mutation.
When examining human subject data of these types, it is
extremely important to be able to decide whether the
mutation in itself is lethal to the particular activity involved
or whether the particular mutation caused activity loss
because of other direct or secondary effects, such as the
ability of a particular residue to influence the extent of
y-carboxylation of the protein. In the case of PC, it has been
shown that at least seven residues of gla are needed for
APC anticoagulant activity,18 and thus a nonspecific loss of
ability to y-carboxylate appropriate precursor E residues of
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
PROTEIN C MUTAGENESIS
this protein would have a deleterious effect on the Ca2+dependent anticoagulant activity of the resulting r-APC.
Thus, there is a need to conduct an investigation of this type
with recombinant mutant proteins that are y-carboxylated
at all nonmutated precursor E residues, a fact that we have
considered paramount in deciding to rigorously purify
subpopulations of r-PC that were maximally y-carboxylated. Were this investigation only to have considered total
mutant r-PC antigen as the source of APC, the results
would certainly not have been properly interpreted, because in virtually all cases a heterogeneous population of
incompletely y-carboxylated r-PC molecules was produced
by the transfected cells. This is likely also the case in plasma
of symptomatic patients with these mutations, thus complicating exact correlation of the thrombotic complication
with the observed amino acid mutation(s) in the protein.
An ancillary benefit of this work will be its assistance in
providing accurate and systematic correlations of this type
with patient plasmas. Further, these results will be invaluable for clinical assessment of patients who may be heterozygous for these mutations, because in these cases a wide
variety of functional and malfunctional molecules may be
present in plasma.
While elucidation of the exact mechanism(s) of the
variable anticoagulant effects of the r-APC mutants in
whole plasma was not the purpose of this study, such effects
are undoubtedly related to the relative propensities of the
muteins to inactivate coagulation cofactors fV,fVa, fVIII,
and fVIIIa, as well as their relative abilities to be stimulated
by protein S in plasma. It is also possible that the interaction of APC with its plasma inhibitor could be influenced by
mutations at gla residues. All of these functional properties
of APC are Ca2+-dependent,and it is of great interest to
this laboratory to determine in purified assays whether all
mutants exert their effects through the same functional
properties. This long-term effort is currently ongoing and
will be the subject of future communications.
Finally, with the assistance of the crystallographic coordinates of the Ca2+/bovine prothrombin fragment 1 complex,48a region of prothrombin that contains its gla domain,
it is possible to comment on the role of particular gla
residues in the binding of Ca2+ and to consider the
relationship between this binding and the Ca2+-dependent
functional properties of APC. Because bovine prothrombin
and human PC possess gla residues at equivalent locations
of the molecule, and a high degree of identity exists in the
remainder of the gla-domain, it is likely that Ca2+coordination in the gla domain is similar for both proteins.
Indeed, using a human prothrombin MoAb that recognizes
the Ca2+-induced conformational alteration common to
several vitamin K-dependent clotting proteins, including
PC, it has been found that the antigenic determinant for
this antibody is expressed on human prothrombin and
human PC. The epitope for this antibody is similarly
masked in both of the above proteins after complexation
with Ca2+, Mn2+,and Mg2+,49an observation that argues
for a similar conformational alteration in these proteins
after coordination with Ca2+.
A view of the crystallographic relationships between gla
949
side chains and bound Ca2+ in bovine prothrombin fragment 1is provided in Fig 5.48Other relevant amino acid side
chains are displayed, along with the peptide backbone of
the hexapeptide disulfide loop (residues 18 through 28 in
bovine prothrombin). The amino acid side chain, gla33, did
not show sufficient electron density to be observed in the
diffraction pattern. Interestingly, a total of seven atoms of
Ca2+ form a channel through the gla-domain, with only
three of these (Ca-1, Ca-6, and Ca-7) in good contact with
solvent. Ca-2 is partially exposed to solvent, and Ca-3, Ca-4,
and Ca-5 are not exposed to solvent. These latter three
Ca2+ ions are most probably involved with structural
maintenance of the complex. Gla8 in prothrombin (gla7 in
human PC), along with gla residues 17 (16), 21 (20), 27 (26),
and 30 (29), are involved in complex formation with Ca-2,
Ca-3, Ca-4, and Ca-5. All of these amino acid residues are
crucial to the Ca2+-dependent anticoagulant activity of
APC, thus showing the importance of the integrity of the
coordination of these Ca2+ ions to the structure and
concomitant function of APC. The fact that partial Ca2+dependent anticoagulant activity is observed with mutations of some of these gla residues to D may revolve around
the possibility that the side chain carboxyl group of D can
substitute as a coordination donor for Ca2+. If this is the
case, a reorientation of the geometry of the affected
complexes to accommodate the carboxyl group of D would
occur, and the partial loss of activity would not be surprising. It has been proposed that the Ca2+-inducedconformational change and binding to PL is dependent on the
presence of gla16 in human prothrombin (corresponds to
gla17 in bovine prothrombin and gla16 in human PC)>O The
importance of gla16 to the activity of APC is directly shown
in the current study, and likely correlates to the important
structural role proposed above for this residue. Similarly,
the essential role (Table 2) of gla20 (gla21 in bovine
prothrombin) in the Ca2+-dependent APC anticoagulant
activity of APC is also obvious from Fig 5 in that this residue
is highly important for coordination of Ca-5 and, perhaps,
Ca-6.
The nonessential character of gla14 in APC (gla15 in Fig
5) may be due to the fact that it only provides one
coordination site for Ca-7, and another nonessential gla
residue, gla19 (gla20 in Fig 5), provides the other two
coordination sites for this same Ca2+. Thus, it may be
concluded that these gla residues, and perhaps Ca-7, do not
play an essential anticoagulant function role in APC.
Because gla19 in APC also provides one donor site for Ca-6,
it is not surprising that some activity may be lost with a
substitution for this gla residue, and concomitant possible
weakening of the Ca-6/APC interaction. However, from
the fact that only a small loss in activity is observed with
[glal9D]r-APC, it is concluded that either the Ca-6/gla19
coordination site is relatively unimportant to the stability of
Ca-6 binding to APC, that the mutant D19 can satisfy part
of this function, and/or that Ca-6 is of lesser importance to
the anticoagulant activity of r-APC. The activity (24% of
wtr-APC) found despite the loss of gla25 (gla26 in Fig 5 ) in
APC can be explained if Ca-1 is important to this function
and the loss of the two possible donor sites provided to
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
ZHANG, JHINGAN, AND CASTELLINO
950
* .
.. , .-.i,
.
J
Fig 5. X-ray crystal structure of the bovine prothrombin fragment 1/Ca2+ complex. Only side chains of amino acids relevant t o this
investigation are displayed. The colors represent the following atoms: yellow, carbon; green, hydrogen; red, oxygen; blue, nitrogen; pink, sulfur;
white, calcium. The sequence positions of the amino acid side chains are indicated for bovine prothrombin (the corresponding amino acids and
sequence position in human PC are provided in parenthesis). yCarboxyglutamic acid is represented by y. The atoms depicted by the filled spheres
begin with the @-carbonof the amino acid side chain, and continue through the remainder of the side chain of that amino acid. Nonexchangeable
hydrogen atoms of the amino acid side chains are not displayed. The only peptide bonds illustrated are those contained in the hexapeptide
disulfide loop, encompassing residues 17 through 22. The solvent accessibility of each Ca2+ion is also outlined by the dotted area.
Ca-l/APC stabilization by gla25 weakens that interaction.
Because Ca-1 is also stabilized by two donor sites from the
essential gla29 (gla30 in Fig 5), it may be retained in the
complex (albeit more weakly), thus providing the lower
level of activity observed (Table 3). Similar considerations
may also govern the very low level of activity observed as a
result of mutation of gla29. Lastly, while R15 does not
provide sites for the coordination of Ca2+,the HH21 proton
of R15 (R16 in Fig 5) is sufficiently close (0.23 nm) to the
oxygen of the OE4 group of gla16 (gla17 in Fig 5) to form a
hydrogen bond, which may stabilize a conformational state
optimal for activity.
The fact that substitution for gla6 (gla7 in Fig 5) in r-APC
did not result in a significant activity loss (Table 3) requires
special consideration, because its apparent role in contributing to the stability of Ca-4 and Ca-5 would seemingly be
important. Only one of the two carboxyl groups of this gla
residue is involved as a donor site for these two Ca2+ions,
and it is possible that the carboxyl group of the substituted
D residue can provide this same kind of coordination.
Another explanation would be that loss of one of the seven
possible coordination sites for Ca-4 and two of the possible
five such sites from Ca-5, as a result of this mutation, did not
significantly affect the abilities of these two Ca2+ ions to
interact in a stable manner with the gla-domain.
In conclusion, we have assessed the individual roles of all
gla residues in APC that are important to the in vitro
anticoagulant activity of APC. In addition, we have attempted to correlate this information with known mutations in gla residues in symptomatic patients and with
structural features of the gla domain/Ca2+ complex.
Whereas this latter analysis has been derived from crystallographic data on the bovine prothrombin fragment 1/Ca2+
complex, the high degree of sequence homologies between
these domains of prothrombin and PC justifies the approach. Further, the fact that this crystal structure shows
such a strong correlation with the currently available
functional data allows firmly based hypotheses to be forwarded and tested regarding the role of various amino acids
in structure-activity relationships of APC.
ACKNOWLEDGMENT
The authors express their appreciationto Dr Alexander Tulinsky
for providing the coordinates of the bovine prothrombin fragment
1/Ca2+ complex to this laboratory and for many helpful discussions.
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
951
PROTEIN C MUTAGENESIS
REFERENCES
1. Esmon NL, Owen WG, Esmon CT: Isolation of a membranebound cofactor for thrombin-catalyzed activation of protein C. J
Biol Chem 257859,1982
2. Amphlett GW, Kisiel W, CastellinoFJ: Interaction of calcium
with bovine plasma protein C. Biochemistry20:2156,1981
3. Kisiel W, Canfield WM, Ericsson LH, Davie E W Anticoagulant properties of bovine plasma protein C following activation by
thrombin. Biochemistry 165824,1977
4. Vehar GA, Davie EW: Preparation and properties of bovine
factor VI11 (antihemophilicfactor). Biochemistry 19:401,1980
5. Walker FJ: Regulation of activated protein C by a new
protein. J Biol Chem 255:5521,1980
6. Comp PC, Esmon C T Generation of fibrinolytic activity by
infusion of activated protein C into dogs. J Clin Invest 68:1221,
1981
7. Taylor FB, Lockhart MS: A new function for activated
protein C: Activated protein C prevents inhibition of plasminogen
activators by releasate from mononuclear leukocytes-plateletsuspensions stimulated by phorbol ester. Thromb Res 35155,1985
8. Sakata Y, Loskutoff DJ, Gladson CL, Hekman CM, Griffin
JH: Mechanismof protein C-dependent clot lysis: Role of plasminogen activator inhibitor. Blood 68:1218, 1986
9. Weinstein RE, Walker FJ: Species specificity of the fibrinolytic effects of activated protein C. Thromb Res 63:123,1991
10. Bajzar L, Nesheim M: Activated protein C (APC) promotes
fibrinolysis by inhibiting prothrombinase catalyzed formation of an
antifibrinolytic component derived from prothrombin. Thromb
Haemost 65:1199,1991
11. de Fouw NJ, Haverkate F, Bertina RM, Koopman J, van
Wijngaarden A, van Hinsberg VW:The cofactor role of protein S
in the acceleration of whole blood clot lysis by activated protein C
in vitro. Blood 67:1189,1986
12. Foster DC, Yoshitake S, Davie EW: The nucleotide sequence of the gene for human protein C. Proc Natl Acad Sci USA
82:4673,1985
13. Foster DC, Davie E W Characterization of a cDNA coding
for human protein C. Proc Natl Acad Sci USA 81:4766,1984
14. Beckmann RJ, Schmidt RJ, Santerre RF, Plutzky J, Crabtree GR, Long G L The structure and evolution of a 461 amino
acid human protein C precursor and its messenger RNA, based
upon the DNA sequence of cloned human liver cDNAs. Nucleic
Acids Res 135233,1985
15. Kisiel W, Davie E W Protein C. Methods Enzymol 80:320,
1981
16. Drakenberg T, Fernlund P, Roepstorff P, Stenflo J: p-Hydroxyaspartic acid in vitamin K-dependent protein C. Proc Natl
Acad Sci USA 801802,1983
17. Kisiel W Human plasma protein C. J Clin Invest 64:761,
1979
18. Grinnell BW, Berg DT, Walls J, Yan SB: Trans-activated
expression of fully gamma-carboxylated recombinant human protein C, an antithrombotic factor. Biotechnology51189,1987
19. Zhang L, Castellino FJ: A y-carboxyglutamic acid variant
(y6D, y7D) of human activated protein C displays greatly reduced
activity as an anticoagulant. Biochemistry29:10828,1990
20. Erlich HJ, Jaskunas SR, Grinnell BW, Yan SB, Bang N U
Direct expression of recombinant activated human protein C, a
serine protease. J Biol Chem 264:14298,1989
21. Mann KG, Jenny RJ, Krishnaswamy S: Cofactor proteins in
the assembly and expression of blood clotting enzyme complexes.
Ann Rev Biochem 57:915,1988
22. Yan SCB, Pazzano P, Chao YB, Walls JD, Berg DT,
McClure DB, Grinnell BW: Characterization and novel purifica-
tion of recombinant human protein C from three mammalian cell
lines. Biotechnology8:655,1990
23. Coller BS, Owen J, Jesty J, Horowitz D, Reitman MJ, Spear
J, Yeh T, Comp P C Deficiency of plasma protein S, protein C, or
antithrombin 111 and arterial thrombosis. Atherosclerosis 7:456,
1987
24. Gladson CL, Scharrer I, Hach V, Beck KH, Griffin JH: The
frequency of type I heterogeneous protein S and protein C
deficiency in 141 unrelated young patients with venous thrombosis.
Thromb Haemost 5918,1988
25. Reitsma PH, Poort SR, Allaart CF, Briet E, Bertina RM:
The spectrum of genetic defects in a panel of 40 Dutch families
with symptomatic protein C deficiency type I: Heterogeneity and
founder effects. Blood 78890,1991
26. Zhang L, Castellino FJ: Role of the hexapeptide disulfide
loop present in the y-carboxyglutamicacid domain of protein C in
its activation properties and in the in vitro anticoagulant activity of
activated protein C. Biochemistry30:6696,1991
27. Heeb MJ, Schwartz P, White T, Lammle B, Berrettini M,
Griffin J: Immunoblottingstudies of the molecular forms of protein
C in plasma. Thromb Res 52:33, 1988
28. Zhang L: Structure-function relationships of the y-carboxyglutamic acid domain of human protein C. PhD Dissertation,
University of Notre Dame, Notre Dame, IN, 1991
29. Laemmli U K Cleavage of structural proteins during the
assembly of the head of bacteriophage T4. Nature 227680,1970
30. Kuwada M, Katayama K An improved method for determination of gamma carboxyglutamicacid in proteins, bone and urine.
Anal Biochem 117:259,1981
31. Chibber BAK, Urano S, Castellino FJ: Synthesis, purification and properties of a peptide that enhances the activation of
human [gluJlplasminogen by tissue plasminogen activator and
retards fibrin polymerization. Int J Peptide Protein Res 35:73,1990
32. Grinnell BW, Walls JD, Gerlitz B: Glycosylation of human
protein C affects its secretion, processing, functional activities, and
activation by thrombin. J Biol Chem 266:9778, 1991
33. Price PA, Fraser JD, Metz-Virca G: Molecular cloning of
matrix gla protein: Implications for substrate recognition by the
vitamin K-dependent y-carboxylase. Proc Natl Acad Sci USA
84:8335,1987
34. Jorgensen MJ, Cantor AB, Furie BC, Brown CL, Shoemaker
CB, Furie B: Recognition site directing vitamin K-dependent
y-carboxylationresides on the propeptide of factor IX. Cell 48:185,
1987
35. Ware J, Diuguid DL, Liebman HA, Rabiet M-J, Kasper CK,
Furie BC, Furie B, Stafford DW: Factor IX San Dimas. Substitution of glutamine for arg-4 in the propeptide leads to incomplete
y-carboxylation and altered phospholipid binding properties. J
Biol Chem 264:11401,1989
36. Huber P, Schmitz T, Griffin J, Jacobs M, Walsh C, Furie B,
Furie BC: Identification of amino acids in the y-carboxylation
recognition site of the propeptide of prothrombin. J Biol Chem
265:12467,1990
37. Exner T, Cotton B, Howden M: Detection of specific
proenzyme activatorsin snake venoms by a new immunoadsorbentchromogenic substrate method. Biochim Biophys Acta 832:351,
1985
38. Orthner CL, Bhattacharya P, Strickland D K Characterization of a protein C activator from the venom of Agkistrodon
contortriv contorhir. Biochemistry27:2558,1988
39. Kurachi K, Davie EW: Isolation and characterization of a
cDNA coding for human factor IX. Proc Natl Acad Sci USA
79:6461,1982
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
952
40. Degen SJF, MacGillivray RTA, Davie EW Characterization of the complementary deoxyribonucleic acid and gene coding
for human prothrombin. Biochemistry 222087,1983
41. Hagen FS, Gray CL, O’Hara P, Grant FJ, Saari GC,
Woodbury RG, Hart CE, Insley M, Kisiel W, Kurachi K, Davie
EW Characterization of a cDNA coding for human factor VII.
Proc Natl Acad Sci USA 83:2412,1986
42. Hamaguchi M, Matsushita T, Tanimoto M, Takahashi I,
Yamamoto K, Sugiura I, Takamatsu J, Ogata K, Kamiya T, Saito
H: Three distinct point mutations in the factor IX gene of three
Japanese CRM+ hemophilia B patients (factor IX BMNagoya 2,
factor IX Nagoya 3 and 4). Thromb Haemost 65:514,1991
43. Chen SH, Thompson AR, Zhang M, Scott CR: Three point
mutations in the factor IX genes of five hemophilia patients. J Clin
Invest 84:113,1989
44. Wang NS, Zhang M, Thompson AR, Chen SH: Factor IX
Chongqing: A new mutation in the calcium-binding domain of
factor IX resulting in severe hemophilia B. Thromb Haemost
63:24, 1990
45. Bovill EG, Tomczak J, Grant B, Pilimer E, Rainville I, Long
G L Association of two novel mutations in the gla-domain (Glu20
ZHANG, JHINGAN, AND CASTELLINO
to Ala and Val 34 to Met) with symptomatic type I1 protein C
deficiency. Thromb Haemost 65547,1991
46. Minuro J, Muramatsu S-i, Iijima K, Nakamura K, Yoshitake
S, Sakata Y, Matsuda M: Protein C Yonago: An amino acid
substitution of gly for arg-15 causing defective calcium-dependent
conformation. Thromb Haemost 65:1198,1991
47. Koeberl DD, Bottema CDK, Buerstedde JM, Sommer SS:
Functionally important regions of the factor IX gene have a low
rate of polymorphism and a high rate of mutation in the dinucleotide CpG. Am J Hum Genet 45:488,1989
48. Soriano-Garcia M, Padmanabhan K, deVos AM, Tulinsky
A The Ca2+ ion and membrane binding structure of the gladomain of Ca-prothrombin fragment 1.Biochemistry 31:2554,1992
49. Church WR, Boulanger LL, Messier TL, Mann KG: Evidence for a common metal ion-dependent transition in the 4-carboxyglutamic acid domains of several vitamin K-dependent proteins. J
Biol Chem 264:17882,1989
50. Borowski M, Furie BC, Furie B: Distribution of y-carboxyglutamic acid residues in partially carboxylated prothrombin. J Biol
Chem 261:1624,1986
From www.bloodjournal.org by guest on June 17, 2017. For personal use only.
1992 80: 942-952
Role of individual gamma-carboxyglutamic acid residues of activated
human protein C in defining its in vitro anticoagulant activity
L Zhang, A Jhingan and FJ Castellino
Updated information and services can be found at:
http://www.bloodjournal.org/content/80/4/942.full.html
Articles on similar topics can be found in the following Blood collections
Information about reproducing this article in parts or in its entirety may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests
Information about ordering reprints may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#reprints
Information about subscriptions and ASH membership may be found online at:
http://www.bloodjournal.org/site/subscriptions/index.xhtml
Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American
Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.