JOURNAL OF BONE AND MINERAL RESEARCH Volume 5, Number 8, 1990 Mary Ann Liebert, Inc., Publishers
Effect of Sodium Warfarin on Vitamin K-Dependent Proteins and Skeletal Development in the Rat Fetus RABAB FETEIH,',z MELISSA S. TASSINARI," and JANE B. LIAN3
ABSTRACT Sodium warfarin was administered daily to Sprague-Dawley rats from gestational day 8 to day 22 to examine the effects of this compound on the developing fetal skeleton and on the vitamin K-dependent bone and cartilage proteins. At a dose of 175 &kg of sodium warfarin there was a 43% mortality rate among the dams. Maternal prothrombin times and serum osteocalcin levels were slightly elevated but not significantly. In the surviving litters, fetal bone osteocalcin and y-carboxyglutamic acid were significantly reduced (50 and 57%, respectively, on gestational day 22) when compared to age- and/or weight-matched control pups. The high correlation of osteocalcin content in long bone (R = 0.64) and calvariae (R = 0.77) to fetal body weight observed in control fetuses was not seen in the warfarin-exposed pups. Examination of alizarinstained warfarin-exposed fetal skeletons for ossification centers showed no difference from controls. However, analysis of the tibia1 growth showed several changes compared to control that included (1) widened hypertrophic zones, (2) increased calcification of the hypertrophic zones, and (3) disorganization of the hypertrophic cells. These results suggest that the growth plate abnormalities seen with prenatal warfarin exposure relate to the inhibition of the vitamin K-dependent proteins of the skeletal system.
INTRODUCTION
S
is an anticoagulant that inhibits the vitamin K-dependent synthesis of y-carboxyglutamic acid (Gla) residues in proteins of liver,"' bone,(21and other tissues.") Warfarin induces the intracellular accumulation of liver-synthesized precursor proteins in the rat"' and bone-synthesized precursor proteins(2 in rat and chick that results in an inhibited synthesis of a processed extracellular Gla protein. The mechanism of this pharmacologic action of warfarin is an inhibition of the vitamin K epoxide reductase enzyme necessary for the regeneration of the active vitamin KH, required for the carboxylation reaction.") This provides the anticoagulant properties of warfarin for the proteins of the clotting cascade. Warfarin is neccessary for the treatment of thromboembolic disease and for patients with artificial heart values who require long-term anticoagulant the rap^.'^ s, ODIUM WARFARIN
Two vitamin K-dependent proteins have been characterized in the skeleton. Osteocalcin (bone Gla protein) is a 3 Gla residues per 5700 MW protein associated with hydroxyapatite crystals in the extracellular matrix and is present in bone in proportion to the In newborn rat pups, osteocalcin is about 0.2% of bone dry weight and increases to 2% in the adult rat skeleton,") accounting for one of the major noncollagenous proteins. A second vitamin K- dependent protein, the matrix Gla protein, is a 5 Gla residues per 11,ooOMW protein that predominates in embryonic bone and cartilage extracellular matrix but is also synthesized by lung, heart, and kidney.@' Its presence in bone is not correlated to mineral composition but remains at a constant level ( - 0.4 mg/g bone) in adult bone.I8) Prenatal warfarin exposure in the first trimester causes a fetal warfarin embryopathy in humans with distinct abnormalities of the skeleton. Consistent features of this syndrome include nasal hypoplasia with depression of the
'Department of Orthodontics, Harvard Medical School and Harvard School of Dental Medicine, Boston, MA, and 'Department of Cell Biology, University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655. 'Present address: Department of Orthodontics, Universit) of Riyadh, Saudi Arabia.
885
886
FETEIH ET AL.
nasal bridge and stippling of epiphyseal cartilage, primarily of the growth plate in fetal pups exposed to warfarin from in the axial skeleton, proximal femur, and c a l c a n e u ~ . ' ~ - " ~gestational day 8 at doses that did not significantly elevate Animal studies using milligram doses of warfarin or other maternal prothrombin time. Although bone Gla concentracoumarin derivatives have shown an increase in maternal tion and osteocalcin were significantly decreased (up to and fetal hemorrhage and stillbirth as a result of the im- Soyo) in all litters of warfarin-treated pregnant rats, none paired fetal coagulation system."*-'4) One study reported of the phenotypic features of the human fetal warfarin emdelayed development and decreased fetal weight and de- bryopathy were reproduced. Nonetheless, these studies layed ossification in mice injected with 1-4 mg sodium demonstrate for the first time a significant effect of warwarfarin during the period of organogenesis.("l None of farin on the synthesis of the vitamin K-dependent bone these studies, however, examined the biochemical changes proteins in fetal rats as a result of maternal exposure to in bone related t o the pharmacologic effects of warfarin. warfarin independent of its effect on the maternal coagulaTo understand better the bone abnormalities associated tion state. with the fetal warfarin embryopathy syndrome in humans and whether these are related to the two vitamin K-depenMATERIALS A N D METHODS dent proteins in bone, the present studies were carried out Animals in pregnant rats to determine the morphologic and associated biochemical effects of warfarin on the developing Timed pregnant Sprague-Dawley rats were obtained fetal skeleton. We report abnormalities in the development from Charles River Laboratories (Wilmington, MA) on
TABLE 1. REPRODUCTIVE OUTCOME A N D SERUM PARAMETERS IN WARFARIN EXPOSED (W) A N D CONTROL (C) LITTERS ~
Day
20 C
~
~~
Number of lifters (fetuses)
Litter size X + SD
Fetal weight gm f S D
10 (1 19)
11.9 f 2.3
2.38
f
Maternal serum prothrombin sec f S D
0.30
Mafernal serum osf eocaltin ng/ml + SD
Fetal serum osteocalcin ng/ml + S D
+
ND
13.6 f 0.4 n = 7
34.2
13.2 f 0.6 n = 6 13.5 f 0.9
32.0
2.8
n = 7
21
C
6 (83)
9.5 f 2.4
15 (182)
W
4.01
12.8 f 1.4
3.75
0.57
f
f
0.31
35.5
n = 6
22 C
9 (1 14)
5.22
11.8 f 0.8
5 (59)
W
5.46
12.7 f 3.0
f f
0.47 0.22
=
+ 3.1
n = 6
13.1 f 0.7 n = 7 13.4 f 0.4
35.1
+
4.60 f 1.00 n = 7 5.70 f 2.30
3.5
n = 7
37.7
n = 5 n
1.85 f 0.32 n = 6 2.05 f 0.41 n = 6
1.6
f
n = 6
*
4.9
n = 5
n = 5
number of animals for maternal analyses and number of pooled litter samples for fetal determinarions.
TABLE 2. NUMBER OF OSSIFICATION CENTERS IN WARFARIN EXPOSED (W)
AND
CONTROL (C) LITTERS (LSM
Fore limb
Day
20 C
0.01 f 0.01
21 C W
1.7 f 0.4 1.2f0.3
5.1 f 0.2 5.1f0.2
3.8 f 0.1 3.8f0.04
0.28 f 0.2 0.25f0.2
3.0 f 0.2 2.9*0.2
4.0 f 0.03 4.0f0.04
22 C W
4.7f0.5 4.3f0.7
6.0f0.02 6.0f0.03
4.0f0.01 4.0f0.01
3.5 f O . 2 3.5 f O . 2
4.6*0.1 4.6*0.1
4.7f0.1 4.7f0.2
(6) 2.6
aNurnbers in parentheses
=
f
0.3
Metacarpals (5)
2.6
f
0.2
Proximal phalanges (5)
Distal phalanges (5)
0.0
0.0
number of potential ossification sites; LSM
SE)
Hind limb
Cervical vertebral centra (7)a
Sternebrae
f
=
Metatarsals (5)
3.1
f
least square mean.
0.2
Proximal phalanges (5)
Distal phalanges (5)
0.0
0.0
0.0 0.0 2.8f0.3 2.8+0.4
2.6 f 0.1 2.5 f 0.1
5.0*0.04 4.9~t0.04
887
WARFARIN A N D SKELETAL DEVELOPMENT TABLE3. CRANIOFACIAL DIMENSIONS I N WARFARIN EXPOSED (W) A N D CONTROL (C) FETUSES (mm f SE)
Day
Mandibular length
Mandibular depth
20 C
3.5
3.5 f 0.10
7.1
f
0.10
1.6
21 C W
4.0 f 0.04 3.9 f 0.03
4.0 3.9
f
0.04 0.03
8.7 8.7
f
0.10 0.10
1.9 f 0.02 1.8 f 0.01
3.0 f 0.10 2.9 f 0.10
22 C W
9.2 9.1
4.9 4.8
f
0.10 0.10
9.7 f 0.10 9.6 f 0.10
2.1 2.1
3.3 f 0.02 3.3 f 0.03
f
f f
0.10
0.03 0.10
f
f
Maxillary length
f
Premaxillary length
0.02
f f
0.02 0.03
Snout length
2.5
f
0.04
All dimensions were adjusted for body weight.
'
.
.
FIG. 1. Proximal tibia1 growth plates of control (A) and warfarin-exposed day 21 rat pups (B and C). The mean width of the hypertrophic zone in the warfarin-exposed fetuses occupied a greater percentaege of the total cartilage length (18.3%) compared to the control (14.2%). A second warfarin-exposed pup from another litter showed a disruption of the normal columnar organization of the chondrocytes (C). ( x 45.)
gestational day 7. The day a vaginal plug was observed was designated day 1 of pregnancy. The dams were assigned to either control or warfarin-exposed groups. Sodium warfarin (EndoLabs, Wilmington, DE) was reconstituted with sterile saline (50 mg/ml), and beginning on gestational day 8 or day 10-22, the warfarin group received a daily subcutaneous injection of warfarin at a dose of 175 pg/kg body weight. The control group received a daily subcutaneous injection of saline. The stock solution was diluted to a concentration such that the dam received 0.5-0.6 ml volume for injection on the upper back. Litters were examined on days 20, 21, and 22. For each litter the number of fetuses
and resorptions were recorded, and then each fetus was removed, blotted dry, weighed, and examined for gross abnormalities. Individual maternal and fetal (pooled by litter) blood samples were obtained for assays of prothrombin time and osteocalcin levels. Fetuses from each litter were assigned at random for either whole skeletal examination or biochemical and histologic evaluation.
Skeletal and histologic analysis Fetuses for skeletal analysis were fixed in 95% ethanol and then stained with alizarin red and alcian blue."s' The
888
FETEIH ET AL.
FIG. 2. Longitudinal section of the fetal proximal (A) tibia and (B) femur of two severely affected day 22 fetuses exposed to warfarin (175 p g / k g ; x 45). Note the wide hypertrophic zone and the haphazard circular arrangements of the chondrocytes. A higher power view ( C ) shows the disruptive arrangements of the chondrocytes and the excessive pericelM a r calcification pattern compared to (D) a normal day 22 fetus that shows normal columnar arrangements of the chondrocytes and the vertical pattern of calcification ( x 110).
889
WARFARIN A N D SKELETAL DEVELOPMENT 21 1
3.0 i CALVARIA 2.5
-
2.0
-
1.5 -
6
3 20
3.0
1
21
22
I
LONGBONE
0
50
1
40
-
20
21
22
20
21
22
1
30 -
-
1.o
20
0.5
10 -
0.0
20
21
FETAL AGE , DAYS
22
0 .
FIG. 3. Effect of warfarin on Gla levels in calvaria (top) FETAL AGE , DAYS and long bone (bottom) of the rat fetus. Bone Gla levels (residues per lo00 glutamic acid residues) were measured FIG. 4. Effect of warfarin on osteocalcin levels in calafter 2 N KOH hydrolysis. Warfarin (175 pg/kg) caused a varia (top) and long bone (bottom) of the rat fetus. Radiosignificant reduction of Gla levels in calvariae on days 21 immunoassay measurement of osteocalcin (ng/mg dry (46.2% of control) and 22 (54% of control). A significant weight) from control and warfarin-exposed fetuses on gesreduction was also evident in the long bones on day 21 tational days 20, 21, and 22 are compared. Warfarin (175 (56.1% of control) and day 22 (53.9% of control) in the pg/kg) reduced the levels of osteocalcin significantly to warfarin-exposed fetuses. 76.9% of control on day 21 and 56.9% on day 22, and in the long bones osteocalcin levels were significantly reduced to 69% of control on day 21 and 50% on day 22 of control levels. number of ossification sites in the metacarpals, metatarsals, proximal and distal phalanges of the fore and hind limbs, sternebrae, and cervical vertebrae were also re- total length of cartilage present in the proximal epiphysis corded. The stained specimens were photographed, and and the length of the hypertrophic zone within the carticraniofacial dimensions were measured using the measure- lage end were measured using a calibrated reticle in the ments described by Lorente et a1."61 When appropriate, eyepiece of a microscope. measurements were corrected for body weight before statistical analysis. For morphologic analysis of the bone, the Biochemical assays right tibia and femur from assigned fetuses were fixed in 2% paraformaldehyde and embedded in plastic for serial Prothrombin times were measured on freshly collected sectioning (3 hm thickness). Sections were stained with 3 % plasma by the Children's Hospital Hematology Laboratory silver nitrate for mineral and counterstained with either he- utilizing the MLA Electra 700. maternal plasma was anamatoxylin and eosin or toluidine blue. For each section the lyzed individually, and fetal plasma was pooled by litter
890
FETElH ET AL. TABLE 4. SUMMARY OF LEVELS OF GLAIN LONG BONEAND CALVARIA MEAN
f
SE ( n ) ~~~
Long bone
Gla residuedl 000 glu
Day 20 C 21 C W
Yo reduction of control 22 C W Yo reduction of control
Gla nM/mg
1.80
f
0.22 (7)
0.34 f 0.08 (7)
2.21 1.27
f
0.21 (16) 0.09 (26)
0.35 0.12
f
f f
43.9 2.40 1.29
f
f
0.03 (16) 0.04 (25) 65
0.09 (17) 0.14 (7)
~
Calvaria
0.28 0.18
46.1
f
f
1.05
f
0.13 (7)
1.34 f 0.06 (16) 0.62 f 0.01 ( 1 5 )
0.28
f
0.07
0.31 0.10
f
0.01 (16)
* 0.02 (14)
53.8
0.05 (17) 0.03 (7) 52
Gla nM/mg
Glu
residues/1000 glu
67
1.50 0.08 (12) 0.81 f 0.11 (8) 46
0.38 0.16
f
f
0.04 (12) 0.04 (8)
57
Differences between control and warfarin values were all significant at p < 0.001
TABLE 5 . SUMMARY OF LEVELS OF OSTEOCALCIN IN LONG BONEAND CALVARIA DQY 20 C 21 C W To reduction of control 22 C W 070 reduction of control
Long bonesb
2.84
f
0.5 (9)
Calvaria b 6.70
f
0.76 (9)
6.08 f 0.41 (16) 4.20 f 0.91 (24)
13.00 + 0.80 (16) 10.00 f 0.71 (15)
31.0
23.1
20.32 9.97
f f
1.43 (17) 2.12 (8)a
40.13 22.83
50.0
i
+
1.95 (17) 2.71 (9)a
43.1
ap < 0.001 when compared to control values. bng/mg bone dry weight (LS mean f SE) (n).
before analysis. Osteocalcin was measured directly in maternal and fetal serum and in fetal bone (left tibia, femur, radius, and ulna) after extraction in 0.5 M EDTA by a radioimmunoassay as detailed elsewhere. ( I 7 ) In fetal bone this procedure removed 100% of the osteocalcin, as no more immunoreactive protein was detected in 0.5 M EDTA and 6 M guanidine HCI extracts. Purified rat osteocalcin used for standards and tracer and goat antirat osteocalcin antisera were prepared in these laboratories."" This antiserum recognizes normal and decarboxylated osteocalcin equally when assays are carried out in presence of 0.001 M EDTA."'] The linear curves ranged from 0.2 to 1.5 ng/ml. Gla content of the bones was measured after 2 N KOH hydrolysis of the whole bone as previously described,
with a limit of detection of 2 nM/ml.'l'l For analysis of bone Gla and osteocalcin, a pool of bone powder from three to four pups (from each litter) was made and 1-2 mg aliquots of this pool assayed in triplicate. All data were analyzed using the analysis of variance and covariance and Student's f-test using SAS (SAS Institute, Inc., Cary, NC).
RESULTS In preliminary studies, doses ranging from 150-200 pg/ kg were tested. At 150 pg/kg the dams showed no signs of external bleeding, the fetuses did not show any gross abnormalities, and fetal bone osteocalcin and Gla levels were not different from controls. The 185 and 200 Fg/kg doses were lethal to the dams, usually 5-7 days after the onset of dosing. At 175 pg/kg there was a maternal survival rate of 57% (26 of 46 dams). At this dose maternal death also occurred 5-6 days after the start of the warfarin injections. Mean maternal prothrombin time and serum osteocalcin levels were not significantly different from controls (Table I). This dose was selected for study of the effect of warfarin on the fetus. Mean litter size and fetal weights, although reduced for warfarin-exposed animals, were not significantly different from controls (Table I). The mean numbers of resorptions were also not significantly different from control litters. There was also no significant difference in the numbers of ossification centers examined between controls and warfarin-exposed fetuses (Table 2). First analysis of craniofacial dimensions showed significant decreases in measures of mandibular length and depth and maxillary length, but when these proportions were adjusted for fetal body weight no significant differences were found (Table 3). Histologic evaluation of long bones revealed several abnormalities occurring in the hypertrophic zone in the war-
891
WARFARIN AND SKELETAL DEVELOPMENT
TABLE6. CORRELATION OF BODYWEIGHT AND LEVEL OF OSTEOCALCIN IN BONEFOR CONTROL (C) VERSUS WARFARIN EXPOSED(W) FETUSES
Osteocalcin (long bone) ng/mg dry wt.
Weight Group
Age
~~~~
(gm) ~~
~
~
2.84
f
22
20.32
*
22
5.22
0.2
9.97
*
20
C W
f
Osteocalcin (calvaria) ng/mg dry wt.
Correlation coefficient
0.99 0.64
6.70 f 0.76 40.13 f 1.96
0.77 0.64
0.19
22.8
~
2.54 f 0.1 5.46 f 0.4
C
Correlation coefficient
0.5 1.4 2.1
f
2.7
0.19
TABLE7. CALCULATED ESTIMATION OF PERCENT GLA IN OSTEOCALCIN IN LONGBONESAND CALVARIA OF THE CONTROL RAT FETUS
Total Glaa nmoles/mg bone x + SD
Theoretical amountb of Gla in osteocalcin nmoles
% Glac
Long bones 20d 21 22
0.34 0.08 0.35 f 0.03 0.38 f 0.05
*
0.0015 0.0032 0.0107
0.44 0.91 2.81
Calvaria 20 21 22
0.28 f 0.07 0.31 f 0.01 0.38 f 0.04
0.0035 0.0068 0.021 1
1.26 2.21 5.56
aTota1 Gla measured after alkaline hydrolysis. bnmoles Gla = osteocalcin ng/5700 x 3 (Gla res/molecule). Osteocalcin was measured by RIA shown in Table 5. c % Gla = theoretical Gla in osteocalcin/total Gla. dFetal age in days.
farin-exposed pups (Fig. 1). The hypertrophic zone of the warfarin-exposed fetus was wider (Fig. 1B) compared to control (Fig. IA). At 21 days, total cartilage length was not different between control (1.68 f 0.24 mm) and warfarin 0.09 mm). In contrast, the width of the hygroup (1.65 pertrophic zone was significantly greater for the warfarin group (0.303 f 0.027 versus 0.240 + 0.030 mm for controls, p < 0.001). Marked disruption of the columnar arrangement of the hypertrophic chondrocytes (Fig. 1C) was also observed. Figure 2A and B shows a tibia and femur from two severely growth-retarded warfarin-exposed pups. Examination of the epiphyseal plates showed a very irregular arrangement of the hypertrophic cells (Fig. 2C) compared to control (Fig. 2D). There appeared to be a failure of the cells to progress to an organized pattern. This zone appeared heavily mineralized, but it was difficult to evaluate whether this represented an abnormal “overmineralization” or was a consequence of the lack of an organized columnar arrangement to allow the normal vertical intersepta1 calcification. The histologic evaluation therefore suggests that the effect of warfarin was not solely an effect o f retardation in development.
*
The morphologic defects in the development of bone were associated with biochemical effects of warfarin on the skeleton as seen by analysis of the bones for y-carboxyglutamic acid (Gla) in Fig. 3 and osteocalcin (Fig. 4) in both calvariae and long bones. On gestational day 21, Gla was decreased from 46 to 53% based on the number of residues per lo’ glutamic acid (Fig. 3). When Gla concentration was normalized to bone dry weight (Table 4) the decrease was even greater (65-67% of control). In contrast to serum osteocalcin values, which were not statistically different in the two groups (Table l ) , osteocalcin levels in calvariae of warfarin-exposed pups were decreased 23-43% by day 22 (Fig. 4). In long bones of these fetuses, osteocalcin was decreased from 25 to 50% (Fig. 4 and Table 5 ) . The osteocalcin concentration in both long bone and calvariae was highly correlated to the fetal body weight of controls only (Table 6). Since fetal bones contain two known vitamin K-dependent proteins, a calculation was carried out (Table 7) to determine quantitatively how much Gla in fetal bone is accounted for by the presence of immunoreactive osteocalcin. Table 7 shows that, at most, 3% of the Gla content in
FETEIH ET AL.
892
fetal long bone and 6% of the Gla content of fetal calvariae can be accounted for by the presence of osteocalcin. A large portion of the remainder of Gla-containing protein is likely to be the matrix Gla protein, although as yet unidentified proteins may also be present.'18) Since matrix Gla protein is present in much greater quantities than osteocalcin in embryonic bone and cartilage extracellular matrix, the large reduction in Gla content of the warfarin-exposed bones suggests that this protein was inhibited by the prenatal warfarin exposure. In one dam, prothrombin time was elevated 1.6 times. The fetuses of this litter were more severely affected than the fetuses of other warfarin-treated pups. The severely affected warfarin-exposed fetuses showed an absence of ossification centers in the proximal and distal phalanges of the fore and hind limb compared to control. In all these pups, the growth plates were widened and had overcalcified hypertrophic zones. In this litter, bone Gla was undetectable and osteocalcin was decreased 80% (Table 8). The marked decrease in bone osteocalcin in this litter was complemented by a corresponding decrease in serum osteocalcin, suggesting an inhibition of osteoblast synthesis and secretion of osteocalcin.
DISCUSSION Treatment of the pregnant rat with 175 pg/kg of sodium warfarin resulted in significantly decreased levels of the vitamin K-dependent proteins of fetal bone and morphologic alterations of the growth plate. This dose of warfarin did not cause an elevation of maternal prothrombin times, nor did it alter maternal serum osteocalcin levels except in one severely affected dam. These effects occurred independently of hemorrhagic conditions; therefore, the effects of warfarin on fetal development were not predictable from evaluation of maternal anticoagulation status. At the growth plate, the major effect was a disorganization of the normal columnar arrangement of hypertrophic cells. In the majority of the litters the zone was wider than in controls, suggesting a delay in the development of the growth plate.
In some litters there was a severe distortion of the columnar arrangement of the hypertrophic cells. Not only were the hypertrophic cells of the growth plate area distorted, but the arrangement of cells in the entire hypertrophic zone and extending into resting cartilage showed a lack of columnar arrangement and organization. This appearance is similar to a report of a human warfarin-affected fetus'") in which the distal phalanges showed an abnormal circular arrangement rather than columnar organization of chondrocytes. This is the first report of a warfarin effect on developing fetal bone in a mammalian animal. A previous study demonstrated that warfarin injection into chick embryos decreased Gla content to 70% of however, no obvious change in alizarin red-staining calcium detection of the skeleton were observed and histologic analyses were not reported. Postnatal rats maintained from birth up to 3 months in conjunction with high doses of vitamin K to prevent bleeding reduced Gla osteocalcin levels to 2% of control. After 3 months, premature closure of the epiphyseal growth plate (which does not normally fuse in rats) occurred.'2" Abnormal mineralization of the longitudinal septa of growth cartilage was observed. The experimental protocol is not comparable to the present studies, but the disarrangement of columnar chondrocytes observed in the fetal pups may relate to the postnatal observations. The effect of dicoumarol on ossification has also been examined in rats with an induced closed fracture of the metatarsals. Dicoumarol treatment caused a highly significant decrease in the amount of bone produced 12 days postfracture without affecting the total size of the callus.'") Our biochemical analyses indicate that both the osteocalcin and the y-carboxyglutamic acid concentrations of bone were significantly reduced, suggesting that the observed growth plate abnormalities may be related to inhibited synthesis of the vitamin K-dependent proteins of the skeletal system. Since the bone osteocalcin concentration in both control and warfarin-exposed embryo accounts for at most 6% of the total Gla-containing proteins in the bone (Table 6), the growth plate abnormalities may relate more to defective synthesis of the matrix Gla protein,
WARFARIN-EXPOSED LITTER TABLE8. DATACOLLECTED FROM A SEVERELY AFFECTED AT DAY22 OF GESTATION Control
Litter size Maternal prothrombin time (sec) Maternal serum osteocalcin (nglml) Fetal weight (x i SD) (gm) Fetal serum osteocalcin (ng/ml) Bone osteocalcin ng/mg dry bone weight Long bone (x f SD) Calvaria (x SD) Gla residues/1000 Glu Long bone (x f SD) Calvariae (x f SD)
12.7 13.1 35.1 5.46 4.60
f
20.30 40.10
f
2.40 1.50
3.0
Warfarin-exposed
3.5 0.47 1.00
6.0 21.3 20.5 2.11 f 0.04 0.75
f
1.40 1.90
4.10 2.30
f f
0.09 0.08
+ 0.7 f f f
* 0.10 f
0.10
0.00 0.00
893
WARFARIN AND SKELETAL DEVELOPMENT
a protein that predominates in fetal bone and appears to be more specific to cartilage tissues.(”) Osteocalcin levels in long bones and calvariae were found to be positively related to fetal weight in control animals. This correlation was not observed in warfarin-exposed animals. Serum osteocalcin levels did not reflect the marked inhibition of bone osteocalcin concentrations. The mean serum osteocalcin levels were higher in warfarin-exposed pups (Table 1) although not statistically significant. In contrast, bone osteocalcin levels were decreased up to 50% in warfarin-exposed fetuses compared to control. This disparity can be explained by the newly synthesized osteocalcin, which if insufficient in Gla residues will not bind to bone mineral and has been shown to produce higher levels of serum osteocalcin in postnatal warfarin-treated rats.(z41 Since our antibody recognizes both normal and decarboxylated forms,(”) the increased serum osteocalcin levels in warfarin-exposed pups may be comprised of Gla-deficient molecules. One dam had a severe response to the 175 pg warfarin dose. Maternal prothrombin time for this dam was elevated 1.6-fold and the maternal serum osteocalcin value was significantly less than control levels, indicating an inhibition of osteoblast synthesis of osteocalcin, which can occur in cells exposed to high doses of warfarin.(z51The surviving fetuses displayed severe growth retardation as seen by significant decreases in fetal weight and numbers of ossification sites. Levels of bone Gla were undetectable, and only 20% of the expected osteocalcin concentration was observed. Histologically, the growth plates were severely disorganized, with no alignment of the hypertrophic chondrocytes and abnormal pericellular mineralization. The incidence of appearance of this severely affected litter was 1/26 or 3.8%. The incidence of the human fetal warfarin embryopathy has been estimated at approximately 30% . I 9 ) It is noteworthy that none of the phenotypic features of the human syndrome were reproducible in the rat, especially in any pups of this severely affected litter. N o craniofacial disturbances were found (Table 3) in any of the warfarin-exposed litters. Recent studies of a child with a vitamin K deficiency coagulopathy syndrome has shown that decreased carboxylation of glutamic acid residues in the bone proteins resulted in developmental bone abnormalities. ( 2 6 1 In this particular syndrome the patient was never exposed to warfarin but, as a result of an abnormal vitamin K reductase enzyme, Gla synthesis in both blood clotting proteins and bone proteins was inhibited. The patient must be maintained on vitamin K daily, but this does not completely normalize his clotting time. An abnormal decarboxylated osteocalcin has been characterized in the serum of the patient. The patient was born with the same phenotype as fetal warfarin embryopathy, including stippling of the epiphysis, punctate calcifications in the vertebrae, hypoplastic nasal bridge, and shortened or stubby fingers. Thus, decreased synthesis of Gla-containing proteins in bone and cartilage, due either to warfarin exposure or a genetic disorder, appears to account for bone abnormalities during development.
ACKNOWLEDGMENTS We thank Dr. Herbert Kane (Boston University School of Medicine) for advice regarding statistical analysis of the data and Dr. Melvin Glimcher (Laboratory for the Study of Skeletal Disorders and Rehabilitation, Children’s Hospital, Boston) for helpful discussions. Supported by NIH grants AR 39166, AR 29633, and H L 32056.
REFERENCES 1. Suttie JW 1985 Vitamin K-dependent carboxylase. Annu Rev
Biochem 54:459-477. 2. Lian JB, Friedman PA 1978 The vitamin K-dependent synthesis of carboxyglutamic acid by bone microsomes. J Biol Chem 253:6623-6626. 3. Nishimoto SK, Price PA 1980 Secretion of the vitamin K-dependent protein of bone by rat osteosarcoma cells. Evidence for an intracellular precursor. J Biol Chem 255:6579-6583. 4. Rutherford SE, Phelan J P 1986 Thromboembolitic disease in pregnancy. Clin Perinatol 13:719-739. 5 . Iturbe-Alessio 1, del Carmen Fonseca M, Mutchinik 0, Santos MA, Zajarias A, Salazar E 1986 Risks of anticoagulant therapy in pregnant women with artificial heart valves. N Engl J Med 315:1390-1393. 6. Lian JB, Roufosse AH, Reit B, Glimcher MJ 1982 Concentration of osteocalcin and phosphoprotein as a function of mineral content and age in cortical bone. Calcif Tissue Int 34( SUppl 2): 82-87. 7. Lian JB, Carnes DL, Glimcher MJ 1987 Bone and serum concentrations of osteocalcin as a function 1.25-dihydroxyvitamin D , circulating levels in bone disorders in rats. Endocrinology 120:2123-2130. 8. Fraser JO, Price P A 1988 Lung, heart and kidney express high levels of mRNA for the vitamin K-dependent matrix Gla protein. Implications for the possible functions of matrix Gla protein and for the possible distribution of the gamma carboxylase. J Biol Chem 263:11033-11036. 9. Hall J , Pauli RM, Wilson KM 1980 Maternal and fetal sequelae of anticoagulation during pregnancy. Am J Med 68: 122- 140. 10. Shaul WL, Hall J G 1977 Multiple congenital anomalies associated with oral anticoagulants. Am J Obstet Gynecol 127: I91 - 198. 11. Pauli RM, Madden JD, Kranzler KJ, Culpepper W, Port R 1976 Warfarin therapy initiated during pregnancy and phenotypic chondrodysplasia punctata. J Pediatr 88506508. 12. Krauss AP, Perlow S, Singer K 1949 Danger of dicoumarol treatment in pregnancy. JAMA 139:758-762. 13. Kronick J, Phelps NE, McCallion DJ, Hirsh J 1974 Effect of sodium warfarin administered during pregnancy in mice. Am J Obstet Gynecol 118:819-823. 4. Hirsh J, Cade JF, Gallus AS 1970 Fetal effects of coumadin administered during pregnancy. Blood 36:623-627. 5. McLeod JM 1980 Differential staining of cartilage and bone in whole mouse fetuses by alcian blue and alizarin red S. Teratology 22299-301. 6 . Lorente CA, Tassinari MS, Keith DA 1981 The effects of phenytoin rat development: An animal model system for fetal hydantoin syndrome. Teratology 24: 169- 180.
894 17. Gundberg CM, Hauschka PV, Lian JB, Gallop PM 1984 Osteocalcin: Isolation, characterization and detection. Methods Enzymol 107516-544. IS. Loeser R, Wallin R 1990 Chondrocyte vitamin K-dependent carboxylation. In: Wright TM (ed) Transactions of the 36th Annual Meeting, Vol. 15. Orthopaedic Research Society, Illinois, p. 395. 19. Barr M Jr, Burdi AR 1976 Warfarin-associated embryopathy in a 17-week-old abortus. Teratology 14:129-134. 20. Hauschka PV, Reid ML 1978 Vitamin K dependence of a calcium-binding protein containing y-carboxyglutamic acid in chicken bone. J Biol Chem 253:9063-9068. 21. Price PA, Williamson MK, Haba T, Dell RB, Jee WSS 1982 Excessive mineralization with growth plate closure in rats on chronic warfarin treatment. Proc Natl Acad Sci USA 79: 7734-7738. 22. Dodds RA, Catterall A, Bitenky L, Chayen J 1984 Effects of fracture healing of an antagonist of the vitamin K cycle. Calcif Tissue Int 36:233-238. i3. ltawara Y, Price PA 1986 Developmental appearance of matrix Gla protein during calcification in the rat. J Biol Chem 23: 10828- 10832.
FETEIH ET AL. 24. Price PA, Williamson MK 1981 Effects of warfarin on bone. J Biol Chem 256:12754-12759. 25. Pan LC, Williamson MK, Price PA 1985 Sequence of the precursor to rat bone gammacarboxyglutamic acid protein that accumulates in warfarin-treated osteosarcoma cells. J Biol Chem 260:13398-13401. 26. Pauli RM, Lian JB, Mosher DF, Suttie JW 1987 Association of congenital deficiency of multiple vitamin K dependent coagulation factors and the phenotype of the warfarin embryopathy: Clues to the mechanism of teratogenicity of coumarin derivatives. Am J Hum Genet 41566-583.
Address reprint requests to:
Jane B. Lian Department of Cell Biology University of Massachusetts Medical School 55 Lake Avenue North Worcester, MA 01655 Received for publication December 11, 1989; in revised form March 2, 1990; accepted March 9, 1990.
Effect of Sodium Warfarin on Vitamin K-Dependent Proteins and Skeletal Development in the Rat Fetus RABAB FETEIH,',z MELISSA S. TASSINARI," and JANE B. LIAN3
ABSTRACT Sodium warfarin was administered daily to Sprague-Dawley rats from gestational day 8 to day 22 to examine the effects of this compound on the developing fetal skeleton and on the vitamin K-dependent bone and cartilage proteins. At a dose of 175 &kg of sodium warfarin there was a 43% mortality rate among the dams. Maternal prothrombin times and serum osteocalcin levels were slightly elevated but not significantly. In the surviving litters, fetal bone osteocalcin and y-carboxyglutamic acid were significantly reduced (50 and 57%, respectively, on gestational day 22) when compared to age- and/or weight-matched control pups. The high correlation of osteocalcin content in long bone (R = 0.64) and calvariae (R = 0.77) to fetal body weight observed in control fetuses was not seen in the warfarin-exposed pups. Examination of alizarinstained warfarin-exposed fetal skeletons for ossification centers showed no difference from controls. However, analysis of the tibia1 growth showed several changes compared to control that included (1) widened hypertrophic zones, (2) increased calcification of the hypertrophic zones, and (3) disorganization of the hypertrophic cells. These results suggest that the growth plate abnormalities seen with prenatal warfarin exposure relate to the inhibition of the vitamin K-dependent proteins of the skeletal system.
INTRODUCTION
S
is an anticoagulant that inhibits the vitamin K-dependent synthesis of y-carboxyglutamic acid (Gla) residues in proteins of liver,"' bone,(21and other tissues.") Warfarin induces the intracellular accumulation of liver-synthesized precursor proteins in the rat"' and bone-synthesized precursor proteins(2 in rat and chick that results in an inhibited synthesis of a processed extracellular Gla protein. The mechanism of this pharmacologic action of warfarin is an inhibition of the vitamin K epoxide reductase enzyme necessary for the regeneration of the active vitamin KH, required for the carboxylation reaction.") This provides the anticoagulant properties of warfarin for the proteins of the clotting cascade. Warfarin is neccessary for the treatment of thromboembolic disease and for patients with artificial heart values who require long-term anticoagulant the rap^.'^ s, ODIUM WARFARIN
Two vitamin K-dependent proteins have been characterized in the skeleton. Osteocalcin (bone Gla protein) is a 3 Gla residues per 5700 MW protein associated with hydroxyapatite crystals in the extracellular matrix and is present in bone in proportion to the In newborn rat pups, osteocalcin is about 0.2% of bone dry weight and increases to 2% in the adult rat skeleton,") accounting for one of the major noncollagenous proteins. A second vitamin K- dependent protein, the matrix Gla protein, is a 5 Gla residues per 11,ooOMW protein that predominates in embryonic bone and cartilage extracellular matrix but is also synthesized by lung, heart, and kidney.@' Its presence in bone is not correlated to mineral composition but remains at a constant level ( - 0.4 mg/g bone) in adult bone.I8) Prenatal warfarin exposure in the first trimester causes a fetal warfarin embryopathy in humans with distinct abnormalities of the skeleton. Consistent features of this syndrome include nasal hypoplasia with depression of the
'Department of Orthodontics, Harvard Medical School and Harvard School of Dental Medicine, Boston, MA, and 'Department of Cell Biology, University of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655. 'Present address: Department of Orthodontics, Universit) of Riyadh, Saudi Arabia.
885
886
FETEIH ET AL.
nasal bridge and stippling of epiphyseal cartilage, primarily of the growth plate in fetal pups exposed to warfarin from in the axial skeleton, proximal femur, and c a l c a n e u ~ . ' ~ - " ~gestational day 8 at doses that did not significantly elevate Animal studies using milligram doses of warfarin or other maternal prothrombin time. Although bone Gla concentracoumarin derivatives have shown an increase in maternal tion and osteocalcin were significantly decreased (up to and fetal hemorrhage and stillbirth as a result of the im- Soyo) in all litters of warfarin-treated pregnant rats, none paired fetal coagulation system."*-'4) One study reported of the phenotypic features of the human fetal warfarin emdelayed development and decreased fetal weight and de- bryopathy were reproduced. Nonetheless, these studies layed ossification in mice injected with 1-4 mg sodium demonstrate for the first time a significant effect of warwarfarin during the period of organogenesis.("l None of farin on the synthesis of the vitamin K-dependent bone these studies, however, examined the biochemical changes proteins in fetal rats as a result of maternal exposure to in bone related t o the pharmacologic effects of warfarin. warfarin independent of its effect on the maternal coagulaTo understand better the bone abnormalities associated tion state. with the fetal warfarin embryopathy syndrome in humans and whether these are related to the two vitamin K-depenMATERIALS A N D METHODS dent proteins in bone, the present studies were carried out Animals in pregnant rats to determine the morphologic and associated biochemical effects of warfarin on the developing Timed pregnant Sprague-Dawley rats were obtained fetal skeleton. We report abnormalities in the development from Charles River Laboratories (Wilmington, MA) on
TABLE 1. REPRODUCTIVE OUTCOME A N D SERUM PARAMETERS IN WARFARIN EXPOSED (W) A N D CONTROL (C) LITTERS ~
Day
20 C
~
~~
Number of lifters (fetuses)
Litter size X + SD
Fetal weight gm f S D
10 (1 19)
11.9 f 2.3
2.38
f
Maternal serum prothrombin sec f S D
0.30
Mafernal serum osf eocaltin ng/ml + SD
Fetal serum osteocalcin ng/ml + S D
+
ND
13.6 f 0.4 n = 7
34.2
13.2 f 0.6 n = 6 13.5 f 0.9
32.0
2.8
n = 7
21
C
6 (83)
9.5 f 2.4
15 (182)
W
4.01
12.8 f 1.4
3.75
0.57
f
f
0.31
35.5
n = 6
22 C
9 (1 14)
5.22
11.8 f 0.8
5 (59)
W
5.46
12.7 f 3.0
f f
0.47 0.22
=
+ 3.1
n = 6
13.1 f 0.7 n = 7 13.4 f 0.4
35.1
+
4.60 f 1.00 n = 7 5.70 f 2.30
3.5
n = 7
37.7
n = 5 n
1.85 f 0.32 n = 6 2.05 f 0.41 n = 6
1.6
f
n = 6
*
4.9
n = 5
n = 5
number of animals for maternal analyses and number of pooled litter samples for fetal determinarions.
TABLE 2. NUMBER OF OSSIFICATION CENTERS IN WARFARIN EXPOSED (W)
AND
CONTROL (C) LITTERS (LSM
Fore limb
Day
20 C
0.01 f 0.01
21 C W
1.7 f 0.4 1.2f0.3
5.1 f 0.2 5.1f0.2
3.8 f 0.1 3.8f0.04
0.28 f 0.2 0.25f0.2
3.0 f 0.2 2.9*0.2
4.0 f 0.03 4.0f0.04
22 C W
4.7f0.5 4.3f0.7
6.0f0.02 6.0f0.03
4.0f0.01 4.0f0.01
3.5 f O . 2 3.5 f O . 2
4.6*0.1 4.6*0.1
4.7f0.1 4.7f0.2
(6) 2.6
aNurnbers in parentheses
=
f
0.3
Metacarpals (5)
2.6
f
0.2
Proximal phalanges (5)
Distal phalanges (5)
0.0
0.0
number of potential ossification sites; LSM
SE)
Hind limb
Cervical vertebral centra (7)a
Sternebrae
f
=
Metatarsals (5)
3.1
f
least square mean.
0.2
Proximal phalanges (5)
Distal phalanges (5)
0.0
0.0
0.0 0.0 2.8f0.3 2.8+0.4
2.6 f 0.1 2.5 f 0.1
5.0*0.04 4.9~t0.04
887
WARFARIN A N D SKELETAL DEVELOPMENT TABLE3. CRANIOFACIAL DIMENSIONS I N WARFARIN EXPOSED (W) A N D CONTROL (C) FETUSES (mm f SE)
Day
Mandibular length
Mandibular depth
20 C
3.5
3.5 f 0.10
7.1
f
0.10
1.6
21 C W
4.0 f 0.04 3.9 f 0.03
4.0 3.9
f
0.04 0.03
8.7 8.7
f
0.10 0.10
1.9 f 0.02 1.8 f 0.01
3.0 f 0.10 2.9 f 0.10
22 C W
9.2 9.1
4.9 4.8
f
0.10 0.10
9.7 f 0.10 9.6 f 0.10
2.1 2.1
3.3 f 0.02 3.3 f 0.03
f
f f
0.10
0.03 0.10
f
f
Maxillary length
f
Premaxillary length
0.02
f f
0.02 0.03
Snout length
2.5
f
0.04
All dimensions were adjusted for body weight.
'
.
.
FIG. 1. Proximal tibia1 growth plates of control (A) and warfarin-exposed day 21 rat pups (B and C). The mean width of the hypertrophic zone in the warfarin-exposed fetuses occupied a greater percentaege of the total cartilage length (18.3%) compared to the control (14.2%). A second warfarin-exposed pup from another litter showed a disruption of the normal columnar organization of the chondrocytes (C). ( x 45.)
gestational day 7. The day a vaginal plug was observed was designated day 1 of pregnancy. The dams were assigned to either control or warfarin-exposed groups. Sodium warfarin (EndoLabs, Wilmington, DE) was reconstituted with sterile saline (50 mg/ml), and beginning on gestational day 8 or day 10-22, the warfarin group received a daily subcutaneous injection of warfarin at a dose of 175 pg/kg body weight. The control group received a daily subcutaneous injection of saline. The stock solution was diluted to a concentration such that the dam received 0.5-0.6 ml volume for injection on the upper back. Litters were examined on days 20, 21, and 22. For each litter the number of fetuses
and resorptions were recorded, and then each fetus was removed, blotted dry, weighed, and examined for gross abnormalities. Individual maternal and fetal (pooled by litter) blood samples were obtained for assays of prothrombin time and osteocalcin levels. Fetuses from each litter were assigned at random for either whole skeletal examination or biochemical and histologic evaluation.
Skeletal and histologic analysis Fetuses for skeletal analysis were fixed in 95% ethanol and then stained with alizarin red and alcian blue."s' The
888
FETEIH ET AL.
FIG. 2. Longitudinal section of the fetal proximal (A) tibia and (B) femur of two severely affected day 22 fetuses exposed to warfarin (175 p g / k g ; x 45). Note the wide hypertrophic zone and the haphazard circular arrangements of the chondrocytes. A higher power view ( C ) shows the disruptive arrangements of the chondrocytes and the excessive pericelM a r calcification pattern compared to (D) a normal day 22 fetus that shows normal columnar arrangements of the chondrocytes and the vertical pattern of calcification ( x 110).
889
WARFARIN A N D SKELETAL DEVELOPMENT 21 1
3.0 i CALVARIA 2.5
-
2.0
-
1.5 -
6
3 20
3.0
1
21
22
I
LONGBONE
0
50
1
40
-
20
21
22
20
21
22
1
30 -
-
1.o
20
0.5
10 -
0.0
20
21
FETAL AGE , DAYS
22
0 .
FIG. 3. Effect of warfarin on Gla levels in calvaria (top) FETAL AGE , DAYS and long bone (bottom) of the rat fetus. Bone Gla levels (residues per lo00 glutamic acid residues) were measured FIG. 4. Effect of warfarin on osteocalcin levels in calafter 2 N KOH hydrolysis. Warfarin (175 pg/kg) caused a varia (top) and long bone (bottom) of the rat fetus. Radiosignificant reduction of Gla levels in calvariae on days 21 immunoassay measurement of osteocalcin (ng/mg dry (46.2% of control) and 22 (54% of control). A significant weight) from control and warfarin-exposed fetuses on gesreduction was also evident in the long bones on day 21 tational days 20, 21, and 22 are compared. Warfarin (175 (56.1% of control) and day 22 (53.9% of control) in the pg/kg) reduced the levels of osteocalcin significantly to warfarin-exposed fetuses. 76.9% of control on day 21 and 56.9% on day 22, and in the long bones osteocalcin levels were significantly reduced to 69% of control on day 21 and 50% on day 22 of control levels. number of ossification sites in the metacarpals, metatarsals, proximal and distal phalanges of the fore and hind limbs, sternebrae, and cervical vertebrae were also re- total length of cartilage present in the proximal epiphysis corded. The stained specimens were photographed, and and the length of the hypertrophic zone within the carticraniofacial dimensions were measured using the measure- lage end were measured using a calibrated reticle in the ments described by Lorente et a1."61 When appropriate, eyepiece of a microscope. measurements were corrected for body weight before statistical analysis. For morphologic analysis of the bone, the Biochemical assays right tibia and femur from assigned fetuses were fixed in 2% paraformaldehyde and embedded in plastic for serial Prothrombin times were measured on freshly collected sectioning (3 hm thickness). Sections were stained with 3 % plasma by the Children's Hospital Hematology Laboratory silver nitrate for mineral and counterstained with either he- utilizing the MLA Electra 700. maternal plasma was anamatoxylin and eosin or toluidine blue. For each section the lyzed individually, and fetal plasma was pooled by litter
890
FETElH ET AL. TABLE 4. SUMMARY OF LEVELS OF GLAIN LONG BONEAND CALVARIA MEAN
f
SE ( n ) ~~~
Long bone
Gla residuedl 000 glu
Day 20 C 21 C W
Yo reduction of control 22 C W Yo reduction of control
Gla nM/mg
1.80
f
0.22 (7)
0.34 f 0.08 (7)
2.21 1.27
f
0.21 (16) 0.09 (26)
0.35 0.12
f
f f
43.9 2.40 1.29
f
f
0.03 (16) 0.04 (25) 65
0.09 (17) 0.14 (7)
~
Calvaria
0.28 0.18
46.1
f
f
1.05
f
0.13 (7)
1.34 f 0.06 (16) 0.62 f 0.01 ( 1 5 )
0.28
f
0.07
0.31 0.10
f
0.01 (16)
* 0.02 (14)
53.8
0.05 (17) 0.03 (7) 52
Gla nM/mg
Glu
residues/1000 glu
67
1.50 0.08 (12) 0.81 f 0.11 (8) 46
0.38 0.16
f
f
0.04 (12) 0.04 (8)
57
Differences between control and warfarin values were all significant at p < 0.001
TABLE 5 . SUMMARY OF LEVELS OF OSTEOCALCIN IN LONG BONEAND CALVARIA DQY 20 C 21 C W To reduction of control 22 C W 070 reduction of control
Long bonesb
2.84
f
0.5 (9)
Calvaria b 6.70
f
0.76 (9)
6.08 f 0.41 (16) 4.20 f 0.91 (24)
13.00 + 0.80 (16) 10.00 f 0.71 (15)
31.0
23.1
20.32 9.97
f f
1.43 (17) 2.12 (8)a
40.13 22.83
50.0
i
+
1.95 (17) 2.71 (9)a
43.1
ap < 0.001 when compared to control values. bng/mg bone dry weight (LS mean f SE) (n).
before analysis. Osteocalcin was measured directly in maternal and fetal serum and in fetal bone (left tibia, femur, radius, and ulna) after extraction in 0.5 M EDTA by a radioimmunoassay as detailed elsewhere. ( I 7 ) In fetal bone this procedure removed 100% of the osteocalcin, as no more immunoreactive protein was detected in 0.5 M EDTA and 6 M guanidine HCI extracts. Purified rat osteocalcin used for standards and tracer and goat antirat osteocalcin antisera were prepared in these laboratories."" This antiserum recognizes normal and decarboxylated osteocalcin equally when assays are carried out in presence of 0.001 M EDTA."'] The linear curves ranged from 0.2 to 1.5 ng/ml. Gla content of the bones was measured after 2 N KOH hydrolysis of the whole bone as previously described,
with a limit of detection of 2 nM/ml.'l'l For analysis of bone Gla and osteocalcin, a pool of bone powder from three to four pups (from each litter) was made and 1-2 mg aliquots of this pool assayed in triplicate. All data were analyzed using the analysis of variance and covariance and Student's f-test using SAS (SAS Institute, Inc., Cary, NC).
RESULTS In preliminary studies, doses ranging from 150-200 pg/ kg were tested. At 150 pg/kg the dams showed no signs of external bleeding, the fetuses did not show any gross abnormalities, and fetal bone osteocalcin and Gla levels were not different from controls. The 185 and 200 Fg/kg doses were lethal to the dams, usually 5-7 days after the onset of dosing. At 175 pg/kg there was a maternal survival rate of 57% (26 of 46 dams). At this dose maternal death also occurred 5-6 days after the start of the warfarin injections. Mean maternal prothrombin time and serum osteocalcin levels were not significantly different from controls (Table I). This dose was selected for study of the effect of warfarin on the fetus. Mean litter size and fetal weights, although reduced for warfarin-exposed animals, were not significantly different from controls (Table I). The mean numbers of resorptions were also not significantly different from control litters. There was also no significant difference in the numbers of ossification centers examined between controls and warfarin-exposed fetuses (Table 2). First analysis of craniofacial dimensions showed significant decreases in measures of mandibular length and depth and maxillary length, but when these proportions were adjusted for fetal body weight no significant differences were found (Table 3). Histologic evaluation of long bones revealed several abnormalities occurring in the hypertrophic zone in the war-
891
WARFARIN AND SKELETAL DEVELOPMENT
TABLE6. CORRELATION OF BODYWEIGHT AND LEVEL OF OSTEOCALCIN IN BONEFOR CONTROL (C) VERSUS WARFARIN EXPOSED(W) FETUSES
Osteocalcin (long bone) ng/mg dry wt.
Weight Group
Age
~~~~
(gm) ~~
~
~
2.84
f
22
20.32
*
22
5.22
0.2
9.97
*
20
C W
f
Osteocalcin (calvaria) ng/mg dry wt.
Correlation coefficient
0.99 0.64
6.70 f 0.76 40.13 f 1.96
0.77 0.64
0.19
22.8
~
2.54 f 0.1 5.46 f 0.4
C
Correlation coefficient
0.5 1.4 2.1
f
2.7
0.19
TABLE7. CALCULATED ESTIMATION OF PERCENT GLA IN OSTEOCALCIN IN LONGBONESAND CALVARIA OF THE CONTROL RAT FETUS
Total Glaa nmoles/mg bone x + SD
Theoretical amountb of Gla in osteocalcin nmoles
% Glac
Long bones 20d 21 22
0.34 0.08 0.35 f 0.03 0.38 f 0.05
*
0.0015 0.0032 0.0107
0.44 0.91 2.81
Calvaria 20 21 22
0.28 f 0.07 0.31 f 0.01 0.38 f 0.04
0.0035 0.0068 0.021 1
1.26 2.21 5.56
aTota1 Gla measured after alkaline hydrolysis. bnmoles Gla = osteocalcin ng/5700 x 3 (Gla res/molecule). Osteocalcin was measured by RIA shown in Table 5. c % Gla = theoretical Gla in osteocalcin/total Gla. dFetal age in days.
farin-exposed pups (Fig. 1). The hypertrophic zone of the warfarin-exposed fetus was wider (Fig. 1B) compared to control (Fig. IA). At 21 days, total cartilage length was not different between control (1.68 f 0.24 mm) and warfarin 0.09 mm). In contrast, the width of the hygroup (1.65 pertrophic zone was significantly greater for the warfarin group (0.303 f 0.027 versus 0.240 + 0.030 mm for controls, p < 0.001). Marked disruption of the columnar arrangement of the hypertrophic chondrocytes (Fig. 1C) was also observed. Figure 2A and B shows a tibia and femur from two severely growth-retarded warfarin-exposed pups. Examination of the epiphyseal plates showed a very irregular arrangement of the hypertrophic cells (Fig. 2C) compared to control (Fig. 2D). There appeared to be a failure of the cells to progress to an organized pattern. This zone appeared heavily mineralized, but it was difficult to evaluate whether this represented an abnormal “overmineralization” or was a consequence of the lack of an organized columnar arrangement to allow the normal vertical intersepta1 calcification. The histologic evaluation therefore suggests that the effect of warfarin was not solely an effect o f retardation in development.
*
The morphologic defects in the development of bone were associated with biochemical effects of warfarin on the skeleton as seen by analysis of the bones for y-carboxyglutamic acid (Gla) in Fig. 3 and osteocalcin (Fig. 4) in both calvariae and long bones. On gestational day 21, Gla was decreased from 46 to 53% based on the number of residues per lo’ glutamic acid (Fig. 3). When Gla concentration was normalized to bone dry weight (Table 4) the decrease was even greater (65-67% of control). In contrast to serum osteocalcin values, which were not statistically different in the two groups (Table l ) , osteocalcin levels in calvariae of warfarin-exposed pups were decreased 23-43% by day 22 (Fig. 4). In long bones of these fetuses, osteocalcin was decreased from 25 to 50% (Fig. 4 and Table 5 ) . The osteocalcin concentration in both long bone and calvariae was highly correlated to the fetal body weight of controls only (Table 6). Since fetal bones contain two known vitamin K-dependent proteins, a calculation was carried out (Table 7) to determine quantitatively how much Gla in fetal bone is accounted for by the presence of immunoreactive osteocalcin. Table 7 shows that, at most, 3% of the Gla content in
FETEIH ET AL.
892
fetal long bone and 6% of the Gla content of fetal calvariae can be accounted for by the presence of osteocalcin. A large portion of the remainder of Gla-containing protein is likely to be the matrix Gla protein, although as yet unidentified proteins may also be present.'18) Since matrix Gla protein is present in much greater quantities than osteocalcin in embryonic bone and cartilage extracellular matrix, the large reduction in Gla content of the warfarin-exposed bones suggests that this protein was inhibited by the prenatal warfarin exposure. In one dam, prothrombin time was elevated 1.6 times. The fetuses of this litter were more severely affected than the fetuses of other warfarin-treated pups. The severely affected warfarin-exposed fetuses showed an absence of ossification centers in the proximal and distal phalanges of the fore and hind limb compared to control. In all these pups, the growth plates were widened and had overcalcified hypertrophic zones. In this litter, bone Gla was undetectable and osteocalcin was decreased 80% (Table 8). The marked decrease in bone osteocalcin in this litter was complemented by a corresponding decrease in serum osteocalcin, suggesting an inhibition of osteoblast synthesis and secretion of osteocalcin.
DISCUSSION Treatment of the pregnant rat with 175 pg/kg of sodium warfarin resulted in significantly decreased levels of the vitamin K-dependent proteins of fetal bone and morphologic alterations of the growth plate. This dose of warfarin did not cause an elevation of maternal prothrombin times, nor did it alter maternal serum osteocalcin levels except in one severely affected dam. These effects occurred independently of hemorrhagic conditions; therefore, the effects of warfarin on fetal development were not predictable from evaluation of maternal anticoagulation status. At the growth plate, the major effect was a disorganization of the normal columnar arrangement of hypertrophic cells. In the majority of the litters the zone was wider than in controls, suggesting a delay in the development of the growth plate.
In some litters there was a severe distortion of the columnar arrangement of the hypertrophic cells. Not only were the hypertrophic cells of the growth plate area distorted, but the arrangement of cells in the entire hypertrophic zone and extending into resting cartilage showed a lack of columnar arrangement and organization. This appearance is similar to a report of a human warfarin-affected fetus'") in which the distal phalanges showed an abnormal circular arrangement rather than columnar organization of chondrocytes. This is the first report of a warfarin effect on developing fetal bone in a mammalian animal. A previous study demonstrated that warfarin injection into chick embryos decreased Gla content to 70% of however, no obvious change in alizarin red-staining calcium detection of the skeleton were observed and histologic analyses were not reported. Postnatal rats maintained from birth up to 3 months in conjunction with high doses of vitamin K to prevent bleeding reduced Gla osteocalcin levels to 2% of control. After 3 months, premature closure of the epiphyseal growth plate (which does not normally fuse in rats) occurred.'2" Abnormal mineralization of the longitudinal septa of growth cartilage was observed. The experimental protocol is not comparable to the present studies, but the disarrangement of columnar chondrocytes observed in the fetal pups may relate to the postnatal observations. The effect of dicoumarol on ossification has also been examined in rats with an induced closed fracture of the metatarsals. Dicoumarol treatment caused a highly significant decrease in the amount of bone produced 12 days postfracture without affecting the total size of the callus.'") Our biochemical analyses indicate that both the osteocalcin and the y-carboxyglutamic acid concentrations of bone were significantly reduced, suggesting that the observed growth plate abnormalities may be related to inhibited synthesis of the vitamin K-dependent proteins of the skeletal system. Since the bone osteocalcin concentration in both control and warfarin-exposed embryo accounts for at most 6% of the total Gla-containing proteins in the bone (Table 6), the growth plate abnormalities may relate more to defective synthesis of the matrix Gla protein,
WARFARIN-EXPOSED LITTER TABLE8. DATACOLLECTED FROM A SEVERELY AFFECTED AT DAY22 OF GESTATION Control
Litter size Maternal prothrombin time (sec) Maternal serum osteocalcin (nglml) Fetal weight (x i SD) (gm) Fetal serum osteocalcin (ng/ml) Bone osteocalcin ng/mg dry bone weight Long bone (x f SD) Calvaria (x SD) Gla residues/1000 Glu Long bone (x f SD) Calvariae (x f SD)
12.7 13.1 35.1 5.46 4.60
f
20.30 40.10
f
2.40 1.50
3.0
Warfarin-exposed
3.5 0.47 1.00
6.0 21.3 20.5 2.11 f 0.04 0.75
f
1.40 1.90
4.10 2.30
f f
0.09 0.08
+ 0.7 f f f
* 0.10 f
0.10
0.00 0.00
893
WARFARIN AND SKELETAL DEVELOPMENT
a protein that predominates in fetal bone and appears to be more specific to cartilage tissues.(”) Osteocalcin levels in long bones and calvariae were found to be positively related to fetal weight in control animals. This correlation was not observed in warfarin-exposed animals. Serum osteocalcin levels did not reflect the marked inhibition of bone osteocalcin concentrations. The mean serum osteocalcin levels were higher in warfarin-exposed pups (Table 1) although not statistically significant. In contrast, bone osteocalcin levels were decreased up to 50% in warfarin-exposed fetuses compared to control. This disparity can be explained by the newly synthesized osteocalcin, which if insufficient in Gla residues will not bind to bone mineral and has been shown to produce higher levels of serum osteocalcin in postnatal warfarin-treated rats.(z41 Since our antibody recognizes both normal and decarboxylated forms,(”) the increased serum osteocalcin levels in warfarin-exposed pups may be comprised of Gla-deficient molecules. One dam had a severe response to the 175 pg warfarin dose. Maternal prothrombin time for this dam was elevated 1.6-fold and the maternal serum osteocalcin value was significantly less than control levels, indicating an inhibition of osteoblast synthesis of osteocalcin, which can occur in cells exposed to high doses of warfarin.(z51The surviving fetuses displayed severe growth retardation as seen by significant decreases in fetal weight and numbers of ossification sites. Levels of bone Gla were undetectable, and only 20% of the expected osteocalcin concentration was observed. Histologically, the growth plates were severely disorganized, with no alignment of the hypertrophic chondrocytes and abnormal pericellular mineralization. The incidence of appearance of this severely affected litter was 1/26 or 3.8%. The incidence of the human fetal warfarin embryopathy has been estimated at approximately 30% . I 9 ) It is noteworthy that none of the phenotypic features of the human syndrome were reproducible in the rat, especially in any pups of this severely affected litter. N o craniofacial disturbances were found (Table 3) in any of the warfarin-exposed litters. Recent studies of a child with a vitamin K deficiency coagulopathy syndrome has shown that decreased carboxylation of glutamic acid residues in the bone proteins resulted in developmental bone abnormalities. ( 2 6 1 In this particular syndrome the patient was never exposed to warfarin but, as a result of an abnormal vitamin K reductase enzyme, Gla synthesis in both blood clotting proteins and bone proteins was inhibited. The patient must be maintained on vitamin K daily, but this does not completely normalize his clotting time. An abnormal decarboxylated osteocalcin has been characterized in the serum of the patient. The patient was born with the same phenotype as fetal warfarin embryopathy, including stippling of the epiphysis, punctate calcifications in the vertebrae, hypoplastic nasal bridge, and shortened or stubby fingers. Thus, decreased synthesis of Gla-containing proteins in bone and cartilage, due either to warfarin exposure or a genetic disorder, appears to account for bone abnormalities during development.
ACKNOWLEDGMENTS We thank Dr. Herbert Kane (Boston University School of Medicine) for advice regarding statistical analysis of the data and Dr. Melvin Glimcher (Laboratory for the Study of Skeletal Disorders and Rehabilitation, Children’s Hospital, Boston) for helpful discussions. Supported by NIH grants AR 39166, AR 29633, and H L 32056.
REFERENCES 1. Suttie JW 1985 Vitamin K-dependent carboxylase. Annu Rev
Biochem 54:459-477. 2. Lian JB, Friedman PA 1978 The vitamin K-dependent synthesis of carboxyglutamic acid by bone microsomes. J Biol Chem 253:6623-6626. 3. Nishimoto SK, Price PA 1980 Secretion of the vitamin K-dependent protein of bone by rat osteosarcoma cells. Evidence for an intracellular precursor. J Biol Chem 255:6579-6583. 4. Rutherford SE, Phelan J P 1986 Thromboembolitic disease in pregnancy. Clin Perinatol 13:719-739. 5 . Iturbe-Alessio 1, del Carmen Fonseca M, Mutchinik 0, Santos MA, Zajarias A, Salazar E 1986 Risks of anticoagulant therapy in pregnant women with artificial heart valves. N Engl J Med 315:1390-1393. 6. Lian JB, Roufosse AH, Reit B, Glimcher MJ 1982 Concentration of osteocalcin and phosphoprotein as a function of mineral content and age in cortical bone. Calcif Tissue Int 34( SUppl 2): 82-87. 7. Lian JB, Carnes DL, Glimcher MJ 1987 Bone and serum concentrations of osteocalcin as a function 1.25-dihydroxyvitamin D , circulating levels in bone disorders in rats. Endocrinology 120:2123-2130. 8. Fraser JO, Price P A 1988 Lung, heart and kidney express high levels of mRNA for the vitamin K-dependent matrix Gla protein. Implications for the possible functions of matrix Gla protein and for the possible distribution of the gamma carboxylase. J Biol Chem 263:11033-11036. 9. Hall J , Pauli RM, Wilson KM 1980 Maternal and fetal sequelae of anticoagulation during pregnancy. Am J Med 68: 122- 140. 10. Shaul WL, Hall J G 1977 Multiple congenital anomalies associated with oral anticoagulants. Am J Obstet Gynecol 127: I91 - 198. 11. Pauli RM, Madden JD, Kranzler KJ, Culpepper W, Port R 1976 Warfarin therapy initiated during pregnancy and phenotypic chondrodysplasia punctata. J Pediatr 88506508. 12. Krauss AP, Perlow S, Singer K 1949 Danger of dicoumarol treatment in pregnancy. JAMA 139:758-762. 13. Kronick J, Phelps NE, McCallion DJ, Hirsh J 1974 Effect of sodium warfarin administered during pregnancy in mice. Am J Obstet Gynecol 118:819-823. 4. Hirsh J, Cade JF, Gallus AS 1970 Fetal effects of coumadin administered during pregnancy. Blood 36:623-627. 5. McLeod JM 1980 Differential staining of cartilage and bone in whole mouse fetuses by alcian blue and alizarin red S. Teratology 22299-301. 6 . Lorente CA, Tassinari MS, Keith DA 1981 The effects of phenytoin rat development: An animal model system for fetal hydantoin syndrome. Teratology 24: 169- 180.
894 17. Gundberg CM, Hauschka PV, Lian JB, Gallop PM 1984 Osteocalcin: Isolation, characterization and detection. Methods Enzymol 107516-544. IS. Loeser R, Wallin R 1990 Chondrocyte vitamin K-dependent carboxylation. In: Wright TM (ed) Transactions of the 36th Annual Meeting, Vol. 15. Orthopaedic Research Society, Illinois, p. 395. 19. Barr M Jr, Burdi AR 1976 Warfarin-associated embryopathy in a 17-week-old abortus. Teratology 14:129-134. 20. Hauschka PV, Reid ML 1978 Vitamin K dependence of a calcium-binding protein containing y-carboxyglutamic acid in chicken bone. J Biol Chem 253:9063-9068. 21. Price PA, Williamson MK, Haba T, Dell RB, Jee WSS 1982 Excessive mineralization with growth plate closure in rats on chronic warfarin treatment. Proc Natl Acad Sci USA 79: 7734-7738. 22. Dodds RA, Catterall A, Bitenky L, Chayen J 1984 Effects of fracture healing of an antagonist of the vitamin K cycle. Calcif Tissue Int 36:233-238. i3. ltawara Y, Price PA 1986 Developmental appearance of matrix Gla protein during calcification in the rat. J Biol Chem 23: 10828- 10832.
FETEIH ET AL. 24. Price PA, Williamson MK 1981 Effects of warfarin on bone. J Biol Chem 256:12754-12759. 25. Pan LC, Williamson MK, Price PA 1985 Sequence of the precursor to rat bone gammacarboxyglutamic acid protein that accumulates in warfarin-treated osteosarcoma cells. J Biol Chem 260:13398-13401. 26. Pauli RM, Lian JB, Mosher DF, Suttie JW 1987 Association of congenital deficiency of multiple vitamin K dependent coagulation factors and the phenotype of the warfarin embryopathy: Clues to the mechanism of teratogenicity of coumarin derivatives. Am J Hum Genet 41566-583.
Address reprint requests to:
Jane B. Lian Department of Cell Biology University of Massachusetts Medical School 55 Lake Avenue North Worcester, MA 01655 Received for publication December 11, 1989; in revised form March 2, 1990; accepted March 9, 1990.
