| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
University of Nebraska Medical Center, Omaha, Nebraska 68198-6395, USA
1Correspondence: Department of Cell Biology and Anatomy, The University of Nebraska Medical Center, 600 South 42nd St., Omaha, NE 68198-6395. E-mail: throsenq{at}unmc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: neural tube defects • neural crest defects • orofacial defects • prevention
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3
4
5)
. This evidence
reveals an especially well-documented association between
periconceptional vitamin supplementation and a significant reduction in
the rate of occurrence of neural tube and neural crest defects
including failure of neural tube closure, conotruncal septation
anomalies in the heart, and orofacial defects. Folic acid
supplementation may provide protection against the occurrence of some
other defects as well, but the association is not as well demonstrated.
The cellular mechanism whereby folic acid confers this protection is
not obvious; however, in the absence of sufficient folate, the amino
acid homocysteine inevitably accumulates. It was hypothesized on the
basis of epidemiologic studies that homocysteine per se may
be a teratogenic agent in folic acid deficiency; indeed,
hyperhomocysteinemia is associated with the same set of defects even
when it is not caused by folic acid insufficiency (6
7
8)
.
Our recent experimental findings using the avian embryo model supported
this hypothesis: homocysteine induced neural tube, conotruncal, and
orofacial defects in a time- and concentration-dependent fashion; and
folic acid supplementation was effective in preventing these defects
only when such supplementation resulted in a reduction of homocysteine
(9)
. This experiment did not show the cellular or
molecular basis of the teratogenic effect of homocysteine.
In some experiments with adult cells, high concentrations of
homocysteine may be cytotoxic; by extrapolation, it may be predicted
that hyperhomocysteinemia might affect development adversely by killing
embryonic cells. However, our initial experiments with
hyperhomocysteinemia in the avian embryo model did not show evidence of
necrosis in association with the occurrence of neural tube closure
defects. The opposite was true: it was common to find duplication of
the notochord, each copy of which was of normal size or larger; in most
cases of abnormal neural tube closure, the spinal cord or brain
appeared to have a superabundance of cells that were organized
abnormally (9)
. We concluded, therefore, that the
induction of neural tube and related defects by homocysteine was not
the result of its killing cells, but that a more subtle mechanism must
be responsible.
Recently it was demonstrated that homocysteine has the ability to act
as an antagonist of the N-methyl-D-aspartate
(NMDA)2
type of glutamate receptor (10
11
12)
. Although no
previously published data specify a role for the activated NMDA
receptor in neural tube closure or neural crest migration, there is
evidence from later stages of development that the NMDA receptor is a
principal regulator of neuronal cell migration, cell-to-cell adhesion,
intracellular calcium flux, and programmed cell death
(13
14
15
16
17
18)
, processes central to normal neural tube closure
and neural crest migration. The NMDA receptor also behaves uniquely as
a growth factor during neuronal development (19)
.
Furthermore, epidemiologic and experimental studies had shown that
other NMDA receptor antagonists such as ethanol or certain
anticonvulsants are associated with congenital defects of neural crest
and neural tube derivatives (20
21
22
23
24
25)
. Therefore, we
developed the hypothesis that homocysteine, acting as an NMDA receptor
antagonist, could disrupt the orderly process of neural tube closure
and neural crest migration by reducing the activity of an NMDA receptor
(26)
.
We reasoned that the initial series of experiments should test this new
hypothesis physiologically: if activation of an NMDA receptor is
important to normal neural tube closure or neural crest migration, then
inhibition of NMDA receptor activity with selective antagonists will
result in abnormal development of neural crest derivatives and the
neural tube. When a set of compounds selected for their well-known
ability to act as highly specific antagonists of the NMDA receptor
(e.g., MK801) were given to early chicken embryos, the results showed
that the ability to disrupt neural crest/neural tube development was a
general property of NMDA receptor antagonists (26)
. If the
mechanism of homocysteine teratogenicity involves its ability to
inhibit the activity of the NMDA receptor, then the occurrence of
homocysteine-induced defects will be reduced by activation of the NMDA
receptor. The results described below show that the NMDA receptor
co-agonist glycine provided a highly significant level protection
against homocysteine-induced developmental abnormalities.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 26)
. In summary, embryos
in sets of 18–20 were given one of the following solutions (50 µl)
at three time points during early development: 0.9% saline vehicle; 5
µmol L-homocysteine thiolactone; or homocysteine thiolactone in
combination with one of a selected set of NMDA receptor agonists (see
below, Selected Agonists). We have shown previously that L-homocysteine
thiolactone has the same teratogenic effect as L-homocysteine per
se in keeping with its receptor-mediated (rather than metabolic)
mode of action (26)
, but the thiolactone form is
significantly more stable in solution (9)
. We have also
shown that a dose of 5 µmol of L-homocysteine thiolactone gives a
good ratio of surviving embryos to embryos with defects (9
, 26)
. With this dose, serum homocysteine rose from 10 up to 140
nmol/ml within 2 h and then fell rapidly in stage 35 embryos; when
supplementary folate kept the rise to 
40 nmol/ml, abnormal
development was prevented (9)
. Thus, a transitory peak of
40 - 140 nmol/ml is sufficient to disrupt development in our model
system. Concentrations in this range are achieved fairly frequently in
humans after a high methionine meal (27)
.
The time points selected for treatment were the prestreak stage (4 h of
incubation), the 3-somite stage (28 h), and the 19 somite stage (48–52
h). Solutions were delivered through a small access port just large
enough to admit the tip of a 50 µl pipette. This small opening was
not associated with the occurrence of the `windowing effect'
described by the Schoenwolf laboratory (28)
, who showed
that a larger opening in the shell induced spinal closure defects by
some mechanical influence; in this study and in previous reports by us,
neural tube defects were rare among embryos that were windowed and then
treated with vehicle control (<0.1%) (9
, 26)
. After
delivery of each bolus of experimental drug, the small access port was
covered with paraffin.
For each agonist (see below, Selected Agonists), the following were tested in a preliminary set of experiments: saline vehicle alone; 5 µmol homocysteine thiolactone; 0.05, 0.5, 5.0, 50, 500, or 5000 nmol of the agonist; each of the above doses of agonist with 5 µmol homocysteine thiolactone. Each of these experiments was repeated 3–9 times, using 18–20 embryos for each repetition; thus, each solution was given to 60–180 embryos. The treatment schedule described above was designed to expose the embryos to homocysteine during the processes of neurulation and gastrulation. All embryos were harvested upon the completion of neurulation, stage 20–21, or 76 h of incubation to maximize survival. Treatment groups were coded so that the evaluation of developmental abnormalities could be performed blindly.
Evaluation of developmental abnormalities
Stage 20–21 embryos were stained with the vital dye neutral red
and inspected in a dissecting microscope; all gross developmental
abnormalities were recorded. The abnormalities that were observed fell
into the following categories: severe torsion
(
270o) of the spinal cord without a neural tube
closure defect; orofacial defects including small (Fig. 1
) or absent eye and absent or abnormal branchial arches (Fig. 1)
; spinal
cord defects including closure defects in the thoracolumbar region
[i.e., spina bifida (9)
] or caudal regression; cranial
closure defect, typically exencephaly (Fig. 2
); or `multiple defects', when a given embryo showed an obvious defect
in two or more of the previous categories (Figs. 1
, 2)
. In a minority
of the embryos that were treated with homocysteine thiolactone, there
were blood-filled cysts among the caudal-most somites. These cysts were
not statistically associated with any of the neural crest/neural tube
abnormalities listed above, and it was not clear that their presence
signaled any true developmental abnormality.
|
|
Selected embryos were fixed in Carnoy's fluid prepared with methanol,
in which they may be stored indefinitely. To prepare them for
photography, embryos were placed in normal saline at room temperature
for 24 h, then the whole embryo was stained in toluidine blue for
3–5 min, rinsed in three changes of normal saline, and stored in 100%
glycerol for 24 h. They were photographed in fresh glycerol. This
method makes the embryos semitransparent and permits simultaneous
visualization of surface and internal structures (Figs. 1
and 2)
,
depending on the angle of the incident light.
Selected agonists
Two types of NMDA receptor agonist were selected. One set had
the capacity to activate the receptor at the glutamate binding site and
a second had the capacity to activate the receptor at the glycine,
co-agonist binding site. Glutamate site agonists selected were
L-glutamate and NMDA. Selected glycine site co-agonists were glycine,
D-cycloserine, and aminocyclopropane-carboxylic acid (ACPC). In each
case, the agonist was selected because it was the natural agonist
(glutamate or glycine) or was among the most highly characterized of
the many synthesized agonists available.
Statistical analysis
All data were entered into an electronic database using Excel
6.0 (Microsoft) and Prism 2.0 (GraphPad, Inc.) and were normalized to
the number of defects per 1000 embryos. Means were compared with a
single-factor analysis of variance, followed by Fisher post
hoc analysis (Statview 2.0, Abacus Concepts, Inc.), or with a
Student's t test developed for comparison of groups with
differing n (29)
:
 |
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
), but the most protective effect was provided by glycine at 50 nmol/day
(Fig. 3)
. Glycine was also the only NMDA receptor agonist that gave a
statistically significant protective effect at more than one
concentration (Fig. 3)
. For each agonist that afforded protection
against homocysteine-induced developmental abnormalities, the effect
was highly dose dependent (Fig. 4
).
|
|
After treatment with homocysteine at 5 µmol/day in the absence of any NMDA receptor agonist, the percentage of total defects that fell into each category was as follows: orofacial defects including obvious branchial arch abnormalities, 5.8%; torsion defects, 6.4%; cranial closure defects resembling exencephaly, 16.1%; spinal closure defects resembling spina bifida, 34.8%; and multiple defects, 36.7%. About half of the embryos in the multiple defects category had a cranial closure defect as one component. For each experiment, the total number of defects varied less widely than the number of defects within a given category. For example, if the number of cranial defects for a given experiment was lower than average, or perhaps even zero, it was likely that the number of spinal or multiple defects would be higher than average. It was not possible to predict with accuracy the proportion of embryos that would manifest a particular kind of defect. Possible sources of this variation could include differing origins of the eggs, seasonal variation in the robustness of the embryos, differences in the treatment techniques among different members of the research team, treatment at different times of the day, or differences in the state of maturity of the embryos at the time of treatment. None of these can be excluded with certainty as a contributor to the variability in defects, and the variability could be the result of the combined effects of all of them. Nevertheless, each of these conditions was controlled as much as possible, as described below.
Pathogen-free chicken embryos from SPAFAS were supplied with flock numbers; because of the duration of this study and the large number of embryos that were used, embryos were derived from different flocks. There were slight but statistically insignificant differences among the flocks in the survival rate as well as the overall rate of occurrence of defects; however, embryos from all flocks showed similar variation among defects per experiment. Likewise, there was an annual cycle of variation in embryo survival, with a midsummer acme and a midwinter nadir. This cycle affected treated and control embryos equally, and had no measurable effect on the rate of defects among survivors or variation among defects per experiment.
Regarding differences in the timing or technique of treatment, embryo incubations were always begun at exactly the same time of day, and the 4 h/28 h/52 h treatments were delivered with precision. During the 2 years these experiments were conducted, almost all treatments were performed by only two persons, and no persistent differences of any kind were detectable between their respective results.
We can confirm that the state of maturity of the embryo at the time of
homocysteine treatment may have an effect on the kind of defect that
results. For example, embryos that are treated only once prior to the
onset of neurulation show only cranial closure defects
(30)
, whereas embryos treated only late in the process of
neurulation show spinal defects as well (9)
. Variations in
embryo maturity at the time of treatment could have resulted from small
temperature variations that could not be controlled precisely, for
example, during delivery and storage.
In summary, the kinds of defects that occurred per experiment were more variable than the absolute number of defects, but the origin of this variability could only be estimated. Likewise, the reduction in developmental abnormalities that resulted from the use of NMDA receptor agonists was not specific to any category: there was no statistically provable relationship between any of the agonists and a change within any specific category of defect.
NMDA receptor agonists, including those used in this study, generally
are considered to be dangerous because they may induce epileptiform
seizures as well as apoptosis of fully differentiated neurons (see
Discussion below). Nevertheless, the agonists used in this study were
generally benign; only NMDA produced a statistically significant
increase in abnormal embryos when it was given at the rate of 500
nmol/day. In this case, 9.23 ± 3.4% of the surviving embryos
showed developmental abnormalities vs. 1.6 ± 0.2% of the embryos
that received vehicle only, but even this increase is small compared
with that resulting from 5 µmol/day of homocysteine, after which
38.59% of surviving embryos had abnormalities (see Fig. 3
). Note that
the same dose of NMDA given simultaneously with 5 µmol/day of
homocysteine had neither a positive nor a negative effect: 38.59% of
surviving embryos had neural tube or neural crest abnormalities after
treatment with homocysteine vs. 38.71% of surviving embryos with
neural tube or neural crest abnormalities after treatment with
homocysteine and NMDA.
Whereas glycine acted to restore normal development among the survivors
of homocysteine treatment, neither glycine nor any of the other NMDA
receptor agonists had any effect on survival among the treated embryos
whether given alone or with a teratogenic dose of homocysteine
(Fig. 5
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, who showed that homocysteine acts as a
competitive antagonist at the NMDA receptor glycine co-agonist site,
but as an agonist for the glutamate recognition site. In this way
glycine would be expected to compete with, and thus reverse the
inhibitory action of, homocysteine and possibly also unmask the ability
of homocysteine to act as an agonist.
In studies of the fully developed central nervous system, NMDA receptor
agonists can be toxic whereas NMDA receptor antagonists are protective
against these toxic effects (31)
. However, in the context
of neural tube development, NMDA receptor antagonists appear to be
toxic whereas agonists appear to be protective. During neural tube
closure and neural crest migration, the greater danger appears to be
presented by NMDA receptor antagonists; agonists generally did not
induce significant defects by themselves in the present study, but gave
a protective effect when delivered with homocysteine. NMDA receptor
antagonists may also disrupt normal brain development during early
postnatal life in the rat (32)
, but a protective role for
agonists has not been shown in that model.
The results of the present study indicate that an NMDA receptor or
NMDA-like receptor may regulate some key process in neural tube closure
and neural crest migration, but they do not show what that process may
be. However, it is biologically plausible that this unknown key process
could in fact be apoptosis or programmed cell death. Programmed cell
death is an essential, beneficial feature of early development of the
neural tube and the neural crest (33
34
35)
that is
regulated by the NMDA receptor in later stages of development, as
described above. A decrease in beneficial cell death results in neural
tube closure defects, for example, those that have been described in
p53 loss-of-function mutations in mice (36)
. Thus, it may
be hypothesized that antagonists of the NMDA receptor such as MK801 and
dextromethorphan (26)
or, in the present case,
homocysteine, may be teratogenic because they interfere with an
essential program for beneficial cell death. The phenotype resulting
from this decrease in beneficial cell death should include an abundance
of cells in the effected areas rather than the obvious deficits in the
effected areas that follow increased apoptosis in the neural tube
(e.g., related to retinoic acid-induced neural tube defects)
(37)
. Indeed, the exposed brains of exencephalic mice in
the p53 knockout model (36)
resemble those of
hyperhomocysteinemic chicken embryos (see ref 9,
Fig. 6).
NMDA receptor antagonists, like the agonists, may induce programmed
cell death in some developing neurons (32)
; thus, an
equilibrium between the yin and yang of agonist vs. antagonist may be a
key to normal development of the neural tube and neural crest,
beginning at a very early stage (as indicated by the present
investigation) and continuing into later stages, as shown in the
developing rat central nervous system (32)
.
The pharmacologic evidence given here implies the involvement of a
receptor that interacts with homocysteine and other NMDA receptor
antagonists, but the nature of this putative receptor is not yet known.
There are many possibilities. Functional NMDA receptors are present in
the early chicken embryo (38)
and there is evidence from
Western blots that the NR1 subunit protein is already present by stage
10, during the first 34 h of development (M. Thomas, personal
communication). The NMDA receptor is a heterooligomer composed of
NR1(a-h), NR2(A-D), and NR3
subunits (39
40
41
42)
; thus, there is the potential for
multiple combinations in the receptor complex. Alternatively, there is
the possibility that the NMDA agonist/antagonist effects reported here
may be mediated by an NMDA-like receptor that has not been described.
Investigations into these possibilities have been undertaken in this
and other laboratories.
Hyperhomocysteinemia is associated with increased embryo death, as
shown here and in an earlier study (9)
, and is associated
with spontaneous abortion and fetal death in human populations
(43
, 44)
. As is the case for the association between
hyperhomocysteinemia and neural tube defects, there is no agreement
regarding a mechanism for this increase in embryo and fetal death. The
results of the present study imply that at least in the case of the
avian embryo model, the mechanism for induction of neural tube defects
is different from the mechanism for embryo death, since the survival
rate of the embryos was unaffected when the rate of developmental
abnormalities among the survivors was decreased significantly. It is
possible that homocysteine may exert a lethal effect by acting on the
embryonic vasculature. There appear to be fewer extraembryonic vessels
after treatment of chicken embryos with homocysteine, and the branching
patterns are less complex (unpublished observations of the authors,
supported by observations by K. Eastep, personal communication). In
humans, the approximate equivalent is the growth of the placental
vasculature, which is also affected adversely by hyperhomocysteinemia
(43
, 44)
. Embryonic blood vessels expand by endothelial
outgrowth (45)
; consequently, their failure to grow and
the failure of the vasculature to arborize fully would result from
inhibition of endothelial growth. Since homocysteine has a negative
effect on the growth of endothelia in vivo (46)
and in vitro (47)
, we may hypothesize one
mechanism to explain the lethal effect of hyperhomocysteinemia:
inhibition of growth of the extraembryonic vasculature may starve the
embryo.
Inhibition of intraembryonic angiogenesis by the same mechanism also
could result in abnormal development of organs with an insufficient
blood supply. Indeed, disruption of vascular development can cause
major developmental defects. It has been shown that hemifacial
microsomia and microphthalmia may result from failed vascular
development in humans (48)
. This result is of special
interest to the present study, given the abnormalities of the eye and
branchial arches described above. It is less likely that failed
angiogenesis is a contributor to the neural tube closure defects
described here since the vasculature is insufficiently developed to
permit the blood to circulate effectively between the embryo and the
area vasculosa before stage 18, but neural tube closure is essentially
complete at stage 18 in the chicken embryo (49
; ref 50,
p. 604).
Studies of the effects of homocysteine on cells in vitro
have been criticized when they have used reduced homocysteine,
homocystine, or homocysteine thiolactone as these forms are extremely
scarce in vivo, where the vast majority of serum
homocysteine is bound to albumin (51)
. Although we have
used L-homocysteine or L-homocysteine thiolactone for the present and
earlier studies (9
, 26)
, the same criticism should not
apply here: unlike the case where homocysteine or its thiolactone is
added to culture medium in vitro and therefore may reach
cultured cells unbound, it is unlikely that homocysteine given to
embryos in ovo could reach the embryos without becoming
bound to albumin. Homocysteine, applied to the membranes surrounding
the embryo, may pass either through the albumin-rich milieu that
contains the early embryo or may at later stages enter the albumin-rich
embryonic serum via the extraembryonic vasculature (9)
.
Various forms of homocysteine as well as homocystine (52)
have been shown to have growth or other cellular effects in
vitro, and we do not know which or how many forms of homocysteine
may contribute to the teratogenic effects reported here and in earlier
studies of embryos in vivo (9
, 26)
. However,
there is a rapid exchange among the various forms of homocysteine
in vivo, and when the amount of albumin-bound homocysteine
increases, the other forms may increase consequently (53)
.
The data in this study support the hypothesis that homocysteine may induce abnormal development among neural tube and neural crest derivatives by acting as an antagonist of the NMDA receptor. However, about half of the homocysteine-induced abnormalities remained after application of the best protective agent in this study, glycine. This result could mean that some mechanism other than inhibition of an NMDA receptor is responsible for this residual damage or that the pharmacologic conditions needed for higher levels of protection have not yet been found. Indeed, the putative NMDA-like receptor that apparently is responsible for the protective effect appears to differ pharmacologically from previously described NMDA receptors; for example, the protective effects were not mimicked by the high efficacy partial agonist aminocyclopropane-carboxylic acid.
The fact that a large proportion of the homocysteine-induced
abnormalities appear to be due to its NMDA receptor antagonism presents
another possibility that needs to be explored experimentally.
Homocysteine may have the potential to interact with other teratogenic
NMDA receptor antagonists, additively or synergistically, to promote
abnormal development (54)
.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication February 5, 1999. Revised for publication April 1, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
