Cerebral Cortex, Vol. 10, No. 10, 981-991,
October 2000
© 2000 Oxford University Press
Dendritic Anomalies in Disorders Associated with Mental Retardation
Walter E. Kaufmann1,2,3,4,5 and
Hugo W. Moser2,3,5
1 Departments of Pathology, ,
2 Neurology, ,
3 Pediatrics, and ,
4 Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine and the ,
5 Kennedy Krieger Institute, 707 N. Broadway, Baltimore, MD 21205, USA
 |
Abstract
|
|---|
Dendritic abnormalities are the most consistent anatomical correlates
of mental retardation (MR). Earliest descriptions included dendritic
spine dysgenesis, which was first associated with unclassified
MR, but can also be found in genetic syndromes associated with
MR. Genetic disorders with well-defined dendritic anomalies
involving branches and/or spines include Down, Rett and fragile-X
syndromes. Cytoarchitectonic analyses also suggest dendritic
pathology in Williams and Rubinstein-Taybi syndromes. Dendritic
abnormalities appear to have syndrome-specific pathogenesis
and evolution, which correlate to some extent with their cognitive
profile. The significance of dendritic pathology in synaptic
circuitry and the role of animal models in the study of MR-associated
dendritic abnormalities are also discussed. Finally, a model
of genotype to neurologic phenotype pathway in MR, centered
in dendritic abnormalities, is postulated.
|
Introduction
|
|---|
Since the early 1970s, the application of dendritic labeling
techniques has shown that abnormalities of this postsynaptic
neuronal process are common in genetic and environmental conditions
associated with mental retardation (MR) [reviewed by Huttenlocher
and Kaufmann (Huttenlocher, 1991
; Kaufmann, 1996
)]. Although
these studies have been quite informative, their limitations
in terms of number of examined disorders, sample size and tissue
preservation have precluded definitive conclusions about the
nature and specificity of dendritic anomalies in MR. The recent
identification of the genetic bases of several prevalent MR-associated
disorders, combined with the accompanying generation of animal
models by transgenic technology, makes the elucidation of the
neuronal phenotype of these disorders even more compelling.
For the purposes of this review, we will define as ‘MR-associated disorders’ those conditions characterized by a non-progressive global cognitive deficit (e.g. Down syndrome, DS). We will focus on genetic syndromes, because they provide the potential for understanding the pathogenetic mechanisms of dendritic anomalies, and will also make some reference to childhood degenerative disorders that during their evolution display a cognitive profile that resembles MR (Kaufmann, 1996). The review includes a survey of disorders associated with dendritic anomalies, analyses of the relationships between dendritic pathology, cognitive phenotype and synaptic abnormalities, and the characterization of dendritic pathology in animal models relevant to MR. We conclude with a proposal for pathways linking specific gene mutations and dendritic phenotype.
|
Dendritic Spine Dysgenesis and Unclassified MR
|
|---|
The pioneering studies with Golgi impregnations by Huttenlocher
(Huttenlocher, 1970
, 1974
) and Purpura (Purpura, 1975a
,b
) established
the foundations for the assessment of dendrites in MR. The importance
of these investigations is underscored by the fact that standard
neuropathological methods, which had provided critical information
about the biological bases of many neurological diseases (e.g.
Parkinson disease) (Hornykiewicz, 1963
), had failed to disclose
specific abnormalities in many cases of MR (Freytag and Lindenberg,
1967
; Crome and Stern, 1972
; Jellinger, 1972
). The ‘rediscovered’
Golgi method demonstrated two fundamental abnormalities in the
cerebral cortex: reduction in number and length of dendritic
branches and the aberrant morphology and number of dendritic
spines (Huttenlocher, 1970
, 1974
; Purpura, 1974
, 1975a
, b
).
Most of the evaluations focused on pyramidal neurons, which
constitute the majority of the neurons in human neocortex (Crosby
et al., 1962
). In these cells, both apical and basilar branches
were noted to be shorter or less complex in the brains of persons
with unclassified MR. The most consistent and intriguing findings
involved the shape and length of dendritic spines. The latter
were not only sparse but also long and thin and, when compared
with controls of different ages, resembled immature dendritic
spine patterns (Purpura, 1974
). Following these initial descriptions,
this type of dendritic anomaly—referred to as spine ‘dysgenesis’—was
reported by Marin-Padilla in chromosomopathies associated with
MR (see Figure 1
) (Marin-Padilla, 1972
, 1974
, 1976
). Recent
investigations in children and adolescents with unclassified
MR have confirmed reduced density and spine dysgenesis involving
apical dendrites of the prefrontal cortex (von Bossanyi and
Becher, 1990
). However, the significance of dendritic spine
abnormalities in unclassified MR is not yet clear. Changes in
dendritic shafts are not necessarily present in every case.
Huttenlocher, in his 1991 review on the subject, emphasized
that in brains from adolescents with moderate to profound MR
dendrites branches appeared comparable to normal controls (Huttenlocher,
1974
, 1991
). Williams
et al. also found this to be the case
in a qualitative study of autistic individuals with MR (Williams
et al., 1980
). The extent to which age or degree of MR plays
a role in dendritic arborization reduction in MR is still unknown.
These initial investigations thus provided intriguing leads
that MR is associated with dendritic abnormalities, but did
not furnish definitive information with respect to their significance.
View larger version (28K):
[in this window]
[in a new window]
|
Figure 1. Dendritic spine abnormalities in Patau syndrome and DS. Drawings are from Golgi preparations depicting comparable segments of apical dendrites from layer V pyramidal neurons (motor cortex). (A–E) Different developmental stages in normal subjects (5th gestational month, 7th gestational month, neonatal period, 2nd postnatal month, 8th postnatal month, respectively). (F) A newborn with 13–15 trisomy. (G) An 18-month-old infant with DS (trisomy 21). Note the progressive increase in spine density, associated with a reduction in spine length, during normal development. Spines in Patau syndrome are not only sparse, but also longer than expected for a neonate. On the other hand, the infant with DS had shorter and thinner rather than long spines. Reprinted with permission (Marin-Padilla, 1972).
|
|
A strength of these pioneering studies resides in the combination
of post-mortem and biopsy material. The latter, by eliminating
the effects of post-mortem degradation, not only allowed optimal
conditions for the Golgi material but also provided the opportunity
for high-quality ultrastructural analyses. Purpura and collaborators
(Purpura
et al., 1982
) reported that varicosities in dendrites
of both pyramidal and non-pyramidal neurons were associated
with the spine changes described above and with disruption in
microtubule organization. By 3-D reconstructions of these microtubular
disarrays, these authors demonstrated that microtubular changes
underlie dendritic irregularities and that they represent aberrant
crosslinking of cytoskeletal elements (Bodick
et al., 1982
).
Even though these data on unclassified MR provided a strong
base for understanding the role of dendritic pathology in MR,
the lack of definition of etiology made the delineation of pathogenetic
mechanisms difficult. Many cases classified as ‘MR of
unknown cause’ two or three decades ago, can now be defined
in respect to etiology. For instance, the study by Williams
and colleagues (Williams
et al., 1980
) included a girl with
features compatible with Rett syndrome (RS) (Huttenlocher, 1991
).
The recent identification of the gene responsible for a large
proportion of RS cases (Amir
et al., 1999
) now permits accurate
diagnosis of this condition.
|
Genetic Disorders with Definitive Dendritic Involvement: DS, RS, Fragile-X Syndrome
|
|---|
Most of the studies of dendritic pathology in syndromes associated
with MR have focused on pyramidal neurons in the cerebral cortex.
The most comprehensively examined genetic disorders are DS and
RS. DS, the most common genetic condition associated with MR
(Moser, 1995
), is characterized by abnormal physical and neurologic
growth. In addition to malformations involving mainly the heart
and gastrointestinal tract, brain growth is delayed in DS. Brain
size and weight are reduced at birth, gyral pattern is immature,
neocortical laminar formation is irregular and myelination of
cortical fibers is delayed (Kemper, 1988
; Wisniewski, 1990
;
Golden and Hyman, 1994
; Wisniewski and Kida, 1994
). Several
neocortical regions have been examined by Golgi impregnations
and have revealed changes similar to those reported in unclassified
MR. The initial study of an 18-month-old infant with DS by Marin-Padilla,
the earliest to analyze dendritic spines in MR, demonstrated
spines that were sparse, small and had short stalks in the motor
cortex (Marin-Padilla, 1972
). This dendritic spine abnormality
contrasted with features displayed in a newborn with Patau syndrome
(D, 13–15 trisomy), which demonstrated the typical elements
of spine dysgenesis (Purpura, 1974
), decreased density, and
long and tortuous profile (Marin-Padilla, 1972
, 1974
) (see Figure
1
). A later and more detailed follow-up study by Marin-Padilla
(Marin-Padilla, 1976
) showed that the spine changes in DS are
not specific. Short spines were intermingled with the unusually
long spines described by Purpura (Purpura, 1974
). Interestingly,
the spine dysgenesis was associated with dendritic vacuolation
and neuronal necrosis, suggesting a degenerative process in
DS (Marin-Padilla, 1976
). At least three other investigations
have examined dendritic spines in DS neocortex and hippocampus.
Suetsugu and Mehraein (Suetsugu and Mehraein, 1980
) counted
spines in the cingulate cortex and hippocampus of seven DS subjects
without Alzheimer disease (AD)-like pathology. They found that,
compared with normal controls, spines were significantly reduced
in the middle and distal segments of apical dendrites. That
these findings in DS are specific is suggested by the fact that
similar analyses in a group of individuals with unclassified
MR did not show a decrease in dendritic spines (Suetsugu and
Mehraein, 1980
). The possibility that dendritic spine dysgenesis
in DS represents an early postnatal degenerative process was
raised by the study by Takashima and colleagues (Takashima
et al., 1981
), one of the largest studies in respect to sample
size, which analyzed neurons of the visual cortex in fetuses,
neonates and infants with DS. A decrease in dendritic spines
was observed only in newborns and older subjects with DS (Takashima
et al., 1981
). Another study which supports the unique nature
of spine pathology in DS is that of Ferrer and Gullotta (Ferrer
and Gullotta, 1990
). They examined CA1–CA3 pyramidal neurons
of the hippocampus in two older patients with DS but without
AD features. Spine counts were compared with normal controls
and two DS subjects with associated AD pathology. The number
of spines was reduced in all DS subjects, even though the changes
were more severe in those DS individuals with AD particularly
in CA1.
Dendritic length has also been evaluated in DS neocortex. In the previously cited study by Takashima and collaborators (Takashima et al., 1981), 14 fetuses and infants with DS were compared with normal age-matched controls. They found that shorter basilar dendrites were present only in those DS subjects who were older than 4 months. In a subsequent quantitative study of multiple dendritic branch parameters in eight infants and children with DS they showed that dendritic branching (measured as intersections using the Sholl's concentric circle) and length, in both apical and basilar dendrites, was greater than in controls in DS infants less than 6 months old (Becker et al., 1986). Subsequent to this age, there was a steady decrease in these measures to below normal in subjects older than 2 years. These changes seemed more dramatic in apical dendrites of layer III neurons (Becker et al., 1986). Additional cross-sectional studies have demonstrated marked reductions in dendritic branching, length and spine density in aged DS subjects (Takashima et al., 1989) [reviewed by Becker et al. (Becker et al., 1991)]. In these older individuals degenerative neuronal changes, such as those described by Marin-Padilla (Marin-Padilla, 1976), were also associated with dendritic abnormalities. In the last decade, these findings have been corroborated by two other reports on pyramidal (Schulz and Scholz, 1992) and non-pyramidal neurons (Prinz et al., 1997) from the parietal and motor cortices, respectively. Normal or increased branching in DS infants (Schulz and Scholz, 1992; Prinz et al., 1997) contrasts with reduced dendrites and degenerative changes in older DS children (Schulz and Scholz, 1992).
In contrast to DS, reductions in dendritic arborizations are present throughout life in RS (Armstrong et al., 1995). RS is a developmental disorder that affects females almost exclusively, with a large proportion of cases linked to mutations in the X chromosome gene MeCP2 (Amir et al., 1999). The latter codes for a transcriptional regulator protein, linked to transcriptional repression of methylated gene sequences (Stratling and Yu, 1999). MeCP2 mutations in RS would lead to altered functional domains of the protein (Amir et al., 1999; Wan et al., 1999). RS is characterized by an apparently normal perinatal period, followed by physical and neurological developmental arrest. Between ages 2 and 10 years (stages II and III) there is a regression of language and motor skills, seizures, and appearance of characteristic stereotypic movements. Severe neurologic impairment, including MR, stabilizes by late childhood (Naidu et al., 1995; Naidu, 1997). Initial neuroanatomic evaluations showed that cortical structure, including neuronal number and lamination, are relatively preserved in RS (Jellinger and Seitelberger, 1986; Jellinger et al., 1988). Cytoarchitectonic studies by Bauman and colleagues (Bauman et al., 1995) showed that while neuronal size is reduced there is an increase in neuronal cell packing density. Consistent with this increased neuronal packing, Armstrong and collaborators (Armstrong et al., 1995) demonstrated that basilar and, in some instances, apical branches of pyramidal cells from frontal, motor and inferior temporal cortices were significantly reduced when compared with normal controls. Relative preservation of posterior cortices is suggested by both volumetric neuroimaging (Reiss et al., 1993) and these post-mortem dendritic (Armstrong et al., 1995) studies. Basilar dendrites were more affected than apical ones, particularly in layer V neurons (Armstrong et al., 1995). A more recent investigation by the same group provided additional evidence for the specificity of the dendritic abnormalities in RS cerebral cortex. When compared with neurons from individuals with DS, dendritic arborizations from RS subjects showed the greatest reductions again in premotor, motor and inferior temporal cortices (Armstrong et al., 1998). These data underscore the severe nature of the dendritic tree disturbances in RS neocortex. Belichenko and collaborators (Belichenko et al., 1994), using the lipophilic dye labeling technique and confocal microscopy, were able to delineate the 3-D structure of dendrites from prefrontal, motor and middle temporal areas. They demonstrated that apical dendrites were asymmetric and reduced, and that dendritic spines were markedly decreased, with some segments devoid of these elements. Apparently these spine changes did not correspond to spine dysgenesis, since illustrations in this publication show rather short but otherwise unremarkable spines (Belichenko et al., 1994). Dendrites of hippocampal neurons have also been analyzed in RS. Armstrong et al. (Armstrong et al., 1995) showed reduced dendritic arborizations confined to neurons of layers II and IV of the subiculum, and none in CA1. These data are somewhat at variance with cytoarchitectonic evaluations that reported higher neuronal density throughout the pyramidal layer (including CA1) and, to a lesser extent, in the subiculum (Bauman et al., 1995).
Fragile-X syndrome (FraX) is the second most genetically determined form of MR [reviewed by Moser and Kaufmann and Reiss (Moser, 1995; Kaufmann and Reiss, 1999)]. Despite its high frequency, FraX has been studied less extensively with neuropathological techniques than DS or RS. Only two publications have addressed the issue of neuronal abnormalities. Hinton and colleagues (Hinton et al., 1991) extended the data on one subject previously reported by Rudelli et al. (Rudelli et al., 1985). Neocortical analyses of the three mildly to moderately mentally retarded adult FraX subjects showed that neuronal density in the posterior cingulate and anterior temporal regions was similar to controls. However, Golgi preparations showed long tortuous dendritic spines with prominent heads. The less than optimal quality of the Golgi impregnations precluded an analysis of the dendritic branch or spine density (Hinton et al., 1991). No subsequent studies have been reported in FraX subjects, although dendritic labeling by Golgi techniques has been applied to the mouse model of this condition, as described in a following section.
|
Genetic Disorders with Probable Dendritic Involvement: Williams Syndrome, Rubinstein-Taybi Syndrome
|
|---|
Cytoarchitectonical techniques have helped to delineate cortical
development in a large number of developmental disorders [reviewed
by Kaufmann (Kaufmann, 1996
)]. These approaches have been particularly
helpful in conditions affecting neuronal proliferation and migration,
in which disturbances in laminar organization and neuronal number
typically occur [reviewed by Barth and Kaufmann and Galaburda
(Barth, 1987
; Kaufmann and Galaburda, 1989
)]. Cytoarchitectonic
studies have also shown aberrant cytodifferentiation in neuronal
migration disorders, such as the presence of neurons of abnormal
size and orientation [reviewed by Rorke and by Crino and Eberwine
(Rorke, 1994
; Crino and Eberwine, 1997
)]. Similar findings have
been reported by Logdberg and Brun (Logdberg and Brun, 1993
)
in unclassified MR. In addition to these parameters, quantitative
cytoarchitectonics can provide information about neuronal cell
body size and cell packing density. In RS, for instance, there
is a reduction in neuronal soma size that is associated with
an increase in cell packing density (Bauman
et al., 1995
). Similar
features have been reported by the same investigators in the
hippocampus and other limbic-related regions in autism (Bauman
and Kemper, 1985
) [reviewed by Kemper and Bauman (Kemper and
Bauman, 1998
)]. The significance of increased neuronal density
has been underscored by parallel studies using the Golgi method
that showed a correlation between cell packing and reduction
in dendritic arborizations in these two conditions [discussed
for RS (Raymond
et al., 1996
; Kaufmann
et al., 1998
) and for
autism (Kemper and Bauman, 1998
)]. These data support the concept
that the finding of increased neuronal density by cytoarchitectonic
analyses can be interpreted as signifying a likely reduction
in length and complexity of dendritic trees.
Preliminary studies of cortical cytoarchitecture have been performed in two genetic disorders associated with MR: Williams syndrome (WS) and Rubinstein-Taybi syndrome (R-TS). WS is caused by a submicroscopic deletion on chromosome 7q11.23, which includes the elastin gene and the HPC-1/ syntaxin 1A (STX1A) gene that codes for a protein involved in the docking of synaptic vesicles [reviewed by Bellugi et al. and Botta et al. (Bellugi et al., 1999a; Botta et al., 1999)]. Patients with WS show a distinctive cognitive and social phenotype. While they demonstrate relatively preserved language and face processing abilities, they are typically impaired in visuospatial domains. In addition, they are hypersociable, with engaging personality and excessive sociability with strangers (Bellugi et al., 1999a,b). Neuropathological data on WS are limited to descriptions of associations with CNS malformations (Pober and Filiano, 1995) and AD-like changes (Golden et al., 1995). A single study of one patient by Galaburda and colleagues (Galaburda et al., 1994) reported several cytoarchitectonical anomalies, which included a reduction in columnar organization throughout the cortex, abnormal neuronal orientation and a generalized increase in cell packing density. These findings appeared to be more severe in posterior cortical regions, where there was a decrease in volume (Galaburda et al., 1994), as had previously been shown by quantitative neuroimaging (Jernigan et al., 1993). As the authors of both publications point out, this topographic distribution of severity of abnormalities is in general agreement with the WS cognitive phenotype. In conclusion, preliminary data indicate that in WS there is selective cortical hypoplasia associated with cytoarchitectonical features previously described in unclassified MR and DS (reduced laminar organization and abnormal neuronal orientation), and in autism and RS (increased neuronal cell packing density).
The second MR-related disorder in which cytoarchitectonics suggests dendritic abnormalities is R-TS. With an approximately similar incidence in the general population to RS (Moser, 1995), this condition is characterized by MR and a specific pattern of somatic abnormalities. The initial description by Rubinstein and Taybi (Rubinstein and Taybi, 1963) emphasized short stature, facial dysmorphia, broad thumbs and first toes, and MR. More recent work has demonstrated a wider spectrum of physical and neurologic abnormalities, which include deficits in expressive language and maladaptive behavior (Stevens et al., 1990). The genetic defect in R-TS (16p13.3) has been reported to involve cyclic AMP response element binding protein (CREB)-binding protein (CBP) (Petrij et al., 1995), a protein that is recruited by CREB to bind DNA and activates the basal transcription factor–enzyme complex (Kaufmann and Worley, 1999). Limited neuroimaging and neuropathologic investigations have shown an association between R-TS and several CNS malformations, such as agenesis of the corpus callosum (Stevens et al., 1990), Dandy-Walker malformation (Bonioli et al., 1989) and cortical clefts (Sener, 1995). The most comprehensive neuropathologic evaluation of a R-TS brain was carried out by Pogacar and collaborators (Pogacar et al., 1973). These authors reported one adult male case (33 years) with mild reduction in brain weight, preserved general cortical architecture, but decreased neuronal size and a marked increase (semi-quantitative) in cell packing density. As the latter findings closely resembled those reported in RS (Kaufmann et al., 1998), a reduction in dendritic arborizations appears also to be a feature of R-TS.
Two other relatively frequent genetic disorders, neurofibromatosis-1 and tuberous sclerosis (Bourneville disease, BD), usually present with mild MR or learning disorders (Harrison et al., 1999; Ozonoff, 1999), and display focal cytoarchitectonic abnormalities suggestive of dendritic pathology. Direct dendritic evaluations have been carried out only in BD. Cortical tubers, the hallmark lesion in BD, are foci of disrupted laminar cortical architecture containing large and disorganized cells. Ferrer and colleagues (Ferrer et al., 1984) and Machado-Salas (Machado-Salas, 1984) demonstrated that tubers consist of maloriented pyramidal neurons, with simplified structure and aberrant dendritic branches and spines. Moreover, large numbers of stellate neurons and astrocytes are also present in these clusters. Machado-Salas (Machado-Salas, 1984) even suggested neuronoglial formations with specialized contacts. Huttenlocher and Heydemann (Huttenlocher and Heydemann, 1984) confirmed most of these findings and examined the neocortex intervening between tubers. Despite its normal architecture and dendritic morphology, this adjacent cortex shows a decrease in dendritic branch length. These authors emphasized the similarity between these dendritic reductions and those found in several forms of MR. Table 1 summarizes the dendritic and cytoarchitectonic abnormalities found in genetic disorders with definitive and probable dendritic involvement, respectively.
|
Dendritic Abnormalities and Cognitive Profile
|
|---|
Proof of dendritic pathology as a distinctive feature and substrate
of MR requires the demonstration of its consistency across multiple
conditions associated with MR. The data reviewed above suggest
that either reductions in dendritic branch complexity or length
or changes in dendritic spines are consistent features in genetic
MR. More limited information is available about non-genetic
causes of MR; nevertheless, a few studies demonstrate cytoarchitectonic
and dendritic abnormalities similar to those in genetic disorders
[reviewed by DeLong (DeLong, 1993
)]. Two investigations have
shown reduction in dendritic arborizations (Cordero
et al.,
1993
; Benitez-Bribiesca
et al., 1999
) and spine dysgenesis (Benitez-Bribiesca
et al., 1999
), in moderate to severe protein-calorie malnutrition.
Iodine deficiency leads to MR, deaf-mutism and muscle hypertonia,
in addition to the somatic changes associated with cretinism
(DeLong, 1993
). Neuropathological evaluations have shown that,
in addition to moderate brain weight decreases, there is a reduction
in neuronal density and aberrant neuronal orientation, and a
decrease in dendritic branching involving both pyramidal and
non-pyramidal neurons (DeLong, 1993
; Yan
et al., 1989
). Studies
of fetuses from iodine-deficient mothers do not display such
abnormalities (Liu
et al., 1989
), suggesting that late phases
of neuronal development, particularly differentiation, are affected
most.
Are dendritic abnormalities a specific feature of MR? Huttenlocher (Huttenlocher, 1991) and others (Williams et al., 1978) in their studies of dendrites have emphasized the existence of confounding variables, such as cardiac malformations in chromosopathies. Even after these factors are taken into account, the description of similar changes in several metabolic disorders that affect primarily the cerebral cortex and present with a degenerative-dementing course (Della Giustina, et al., 1981; Takashima et al., 1985) support the concept that marked dendritic abnormalities are an index of major neuronal disruption. However, there are some differences between the findings in metabolic-degenerative disorders and those of genetic MR. First, neuronal migration defect or severe lamination disturbance are relatively common in the progressive metabolic disorders (Barth, 1987) but absent in the non-progressive genetic disorders such as DS. Second, in many metabolic disorders there is accumulation of storage material in neuronal somata and proximal dendrites and axons that leads to distinctive Golgi impregnation patterns (Williams et al., 1977; Purpura, 1978; Takashima et al., 1985). The dendritic aberrations in storage disorders such as gangliosidoses and neuronal ceroid lipofuscinoses are associated with abnormalities of the axon and the presynaptic domain, and include anomalous spatial configuration of cortical neurons (Purpura and Suzuki, 1976). Phenylketonuria (PKU) shows a yet different pattern. This aminoacidopathy, which leads to MR when untreated, has been characterized from the neuropathologic standpoint to be associated with changes in the white matter. There is no storage. However, Bauman and Kemper (Bauman and Kemper, 1982) with the aid of Golgi techniques have shown that gray matter pathology, particularly dendritic changes, is even more pronounced than that in the white matter. They found reductions in dendritic arborizations and spine dysgenesis that were similar to those reported in unclassified MR and DS. It appears therefore that metabolic-degenerative disorders show qualitative and quantitative alterations of dendritic morphology that correlate with cognitive impairment as in MR. However, these changes are dynamic and share with MR only the diffuse and severe magnitude of the dendritic pathology (Kaufmann, 1992).
If severe dendritic changes are a reflection of generalized cortical dysfunction, milder cognitive impairment should be associated with milder dendritic abnormalities. Two studies have partially addressed this issue. The first compared cortical areas with significant dendritic reductions in RS, in which there is profound cognitive impairment, with comparable samples from DS patients. All three premotor, motor and visual cortices in RS had reduced dendritic length compared with DS, with the greatest reductions affecting the frontal regions (Armstrong et al., 1998). As cognitive impairment is, in general, greater in RS than DS, these comparisons suggest a direct relationship between dendritic pathology and cognitive deficit. A more direct assessment was carried out by Yan et al. (Yan et al., 1989), who demonstrated a parallel between dendritic decreases and degree of MR in adult cretinism. Unfortunately, virtually no data about dendritic anomalies are available in individuals with learning disabilities. If severity and extent of dendritic pathology correlates with cognitive function, it is expected that learning disabled subjects would show mild dendritic abnormalities. In support of this hypothesis are the findings in three cases with FraX, with IQ in the 40–60 range. Hinton and collaborators (Hinton et al., 1991) reported the presence of dendritic spine abnormalities without cytoarchitectonic changes.
If dendritic abnormalities are a signature of global cognitive dysfunction, do they play a role in the selective cognitive deficits observed in many syndromes associated with MR? Neuropathologic and neuroimaging studies have already shown some correlations between cortical regional volume and specific impairment. For instance, in DS there is a reduction in temporal and frontal lobe volumes that, in general terms, is in agreement with the selective impairment in language in these patients (Kemper, 1988; Jernigan et al. 1993). Similar associations have been demonstrated for visuospatial impairment and parietooccipital volumes in WS (Jernigan et al., 1993). These topographical surveys, which require the evaluation of a large number of cortical regions, can be correlated with neuroimaging approaches, particularly when large blocks of cortex are measured. A more difficult task is the quantitative or semiquantitative study of cytoarchitectonics or Golgi-based dendritic evaluations of multiple cortical areas. To our knowledge, complete qualitative cytoarchitectonic surveys have only been reported for DS [multiple subjects (Kemper, 1988)], RS [three patients (Bauman et al., 1995)] and WS [one individual (Galaburda et al., 1994)]. Only the latter study attempted to relate pattern of cortical structure and cognitive profile, and no cytoarchitectonic investigation has yet reported measures of multiple cortical regions. In terms of dendritic arborizations, a single study of RS evaluated several cortical regions and neuronal populations within each area (Armstrong et al., 1995). Areas related to preparation (area 6, frontal premotor) and execution (area 4, frontal motor) of movements were affected in RS, whereas the occipital visual cortex (area 17) was relatively preserved. These findings are consistent with the RS phenotype, in which there is severe motor impairment (hypotonia, poor hand use, stereotypic movements, gait disturbances) with relative sparing of visual function (Naidu, 1997). A follow-up study confirmed and expanded the initial observations, but also showed preservation of the superior temporal cortex (Armstrong et al., 1998), a region involved in language processing. As language delay and regression are typical features of RS (Naidu, 1997), these later studies do not support the postulate that dendritic abnormalities reflect cortical dysfunction. More studies are needed to evaluate the relationship between dendritic abnormalities and cognitive profile in MR.
Another critical feature shown by several MR-associated disorders is the diminished maturational increment of cognition and sometimes-frank decline. This has been evaluated from the dendritic viewpoint in three of the main genetic syndromes: DS (Hayes and Batshaw, 1993), RS (Naidu, 1997) and FraX (Fisch et al., 1999a). Studies of DS have suggested dendritic changes as a basis for cognitive decline, but there is as yet no evidence for such a relationship for the other two disorders. In cross-sectional analyses of dendritic growth (length) in DS (Takashima et al., 1981; Becker et al., 1986), dendritic arborizations appeared to be overproduced initially, with a tendency to regress towards the end of the first year, a temporal pattern that parallels cognitive evaluations. In contrast, Armstrong and collaborators (Armstrong et al., 1995) did not find a relationship between dendritic length and age. It should be noted, however, that their regression analyses covered a wide age range (2.9–35 years), which could have masked the declining phase of cognition that occurs between ages 2 and 10 years.
|
Dendritic Abnormalities and Synaptic Circuitry and MR
|
|---|
The most obvious mechanism by which dendritic pathology in general,
and reductions in dendritic arborizations in particular, could
lead to neurologic impairment is a decrease in the cortical
postsynaptic surface. The latter should correlate with a reduction
in synaptic density, while spine dysgenesis may represent a
preferential reduction in excitatory synapses. These assumptions
have not yet been tested rigorously in genetic MR. To date no
study has compared dendritic labeling and synapse analysis in
MR, although some subjects have been independently examined
by one or the other approach. Electron microscopic (EM) studies
have been confined almost exclusively to DS, and with contradictory
results. There are reports of both decreased (Wisniewski
et al., 1985
) and increased (Cragg, 1975
) synaptic density as well
as of arrest in synaptic development (increased primitive synaptic
contacts) (Petit
et al., 1984
). Only a preliminary analysis
by P.R. Huttenlocher (personal communication) found reduced
synaptic density in the prefrontal cortex in RS. What are the
factors causing these discrepancies? They could be related to
the quality of the Golgi staining; long post-mortem interval,
agonal changes and aldehyde fixation interfere with adequate
dendritic impregnation (Williams
et al., 1978
; Buell, 1982
).
Hinton and colleagues (Hinton
et al., 1991
), in their study
of FraX neocortex, stated that ‘Golgi analysis was less
than optimal because of incomplete dendritic stain impregnation’,
and observations about dendritic spine morphology are the only
reliable data obtained in this investigation (Hinton
et al.,
1991
). In his review, Huttenlocher (Huttenlocher, 1991
) emphasized
age as an important factor. While dendritic labeling has been
performed in young subjects, synaptic counts have involved older
individuals. This is of particular significance, since some
studies suggest catch-up dendritic growth in unclassified MR
(Williams
et al., 1980
). The potential of correlating EM and
Golgi studies is suggested by Kaufmann
et al. (Kaufmann
et al.,
1995
, 1997a
,b
), who demonstrated in RS a parallel between dendritic
protein immunoreactivity, by immunoblotting and immunocytochemistry
(Figure 2
), and dendritic evaluations by Golgi (Armstrong
et al., 1995
, 1998
). Similar ‘validation’ of dendritic
abnormalities (Kaufmann, 1992
) has been carried out in HIV encephalopathy
(Masliah
et al., 1997
). Moreover, in RS dye labeling has also
corroborated findings by Golgi impregnations (Belichenko
et al., 1994
). These data suggest that combined dendritic labeling–EM
studies will aid the evaluation of the impact of dendritic changes
in MR.
View larger version (117K):
[in this window]
[in a new window]
|
Figure 2. MAP-2 immunoreactivity in RS. The marked reduction in MAP-2 immunostaining in somatosensory cortex is also seen, with variable intensity, in other neocortical regions. (A) MAP-2 immunostaining of somas and, predominantly, dendrites in a control subject. (B) An equivalent section from an RS patient displays smaller neuronal somas and few MAP-2 immunoreactive dendrites (arrowheads). Reprinted with permission (Kaufmann et al., 1998).
|
|
|
Animal Models of MR and Dendritic Pathology
|
|---|
Study of dendritic abnormalities in animal models of MR minimizes
confounding technical factors and permits evaluation of developmental
patterns. There is an extensive literature on dendritic morphology
and development in animal models relevant to MR [see the respective
section in the review by Huttenlocher (Huttenlocher, 1991
)].
Nonetheless, the validity of the animal models must be examined
with care. Considerable efforts have been made in generating
trisomic mice that model DS. Limited postnatal viability (Haydar
et al., 1996
) and incomplete characterizations (Holtzman
et al., 1996
) have precluded direct dendritic evaluations of these
DS models. Other problems include unexpectedly mild behavioral
phenotype, such as in the FraX mouse model (Fisch
et al., 1999b
).
This
FMR1 knockout mouse does, however, show spine dysgenesis
(Comery
et al., 1997
) that closely resembles that reported by
Hinton and colleagues (Hinton
et al., 1991
). Some experimental
models of several non-genetic causes of MR have succeeded in
reproducing the neuropathological and behavioral features of
the human disorder. Examples of these are PKU (Nigam and Labar,
1979
), protein-calorie malnutrition (Adaro
et al., 1986
; Diaz-Cintra
et al., 1990
), hypothyroidism (Ipina and Ruiz-Marcos, 1986
;
Nunez
et al., 1992
) and fetal alcohol exposure (Hannigan and
Berman, 2000
). These experimental paradigms are important because
they help overcome the limitations of human studies due to small
samples (e.g. PKU and cretinism) and suboptimal tissue processing
(e.g. FraX). Even more significantly, they provide information
about the dynamics of the process and potential remediation.
For instance, the study by Lacey (Lacey, 1985
) showed that dendritic
spine maturational delay and increased spine density can be
reversed by metabolic correction in PKU. On the other hand,
Diaz-Cintra and collaborators (Diaz-Cintra
et al., 1994
) found
that dendritic disturbances caused by prenatal malnutrition
(as in placental insufficiency) in the rat are not reversed
by postnatal nutritional rehabilitation. These approaches are
critically important for the field of genetic MR, since gene-based
and other therapies are beginning to be tested. If dendritic
abnormalities are the cause of MR, they should be evaluated
as outcome measures in experimental treatments. An example is
the positive influence of environmental stimulation on cognitive
development of children with DS (Ludlow and Allen, 1979
). The
morphological effects of sensory enrichment in rodent models
have reported by Greenough (Greenough, 1984
) and ourselves (Adaro
et al., 1986
) (Figure 3
). These studies demonstrate that environmental
stimulation leads to increases in length and complexity of dendritic
arborizations in several cortical regions, even in adult life.
|
From Genotype to Dendritic Phenotype
|
|---|
Dendritic development is a sequential process in which generation,
elongation, and retraction of dendritic branches and spines
are the result of the interaction between intrinsic genetic
programs and external modulators (e.g. neurotransmitters) [reviewed
by Kaufmann (Kaufmann, 1999
)]. Each morphologic stage is characterized
by the expression of a set of dendritic proteins (Petit
et al.,
1988
; Kaufmann
et al., 1997a
). For instance, early neuronal
differentiation (extension of primary branches) is associated
with the expression of the immature form of microtubule-associated
protein 2 (MAP-2), also termed MAP-2c, and of MAP-5. Dendritic
branching and terminal growth is linked to the adult form of
MAP-2 (MAP-2a/b), whereas late phases such as dendritic modeling
are associated with high levels of the high-molecular-weight
form of nonphosphorylated neurofilaments (see Figure 4
) (Kaufmann
et al., 1997a
). The role of neurotransmitters and growth factors
in the progression of dendritic development is beginning to
be disclosed (Figure 4
). While acetylcholine seems to affect
the dendritic expansion (Hohmann
et al., 1991
), activation of
some subtypes of glutamate receptors is linked to dendritic
pruning (Kaufmann, 1999
). On the other hand, neurotrophins modulate
dendritic arborizations in a layer-specific manner (McAllister
et al., 1997
).
View larger version (27K):
[in this window]
[in a new window]
|
Figure 4. Diagrammatic representation of dendritic development. Dendritic formation includes three distinctive stages, characterized by morphological and molecular features: early dendritic differentiation, dendritic expansion or growth, and dendritic remodeling and pruning. Several modulators that are displayed between parentheses have been identified to influence this process. In humans, a substantial portion of dendritic development occurs postnatally.
|
|
We postulate that the genetic defect in MR associated with dendritic
pathology disrupts signaling pathways and/or modulators of dendritic
development. The three genetic MR syndromes in which single
gene defects have been identified, FraX [reviewed by Kaufmann
and Reiss (Kaufmann and Reiss, 1999
)], RS (Amir
et al., 1999
)
and R-TS (Petrij
et al., 1995
), share a defect in the regulation
of gene expression. MeCP2 and CBP are involved in transcriptional
regulation while fragile-X mental retardation protein (FMRP)
modulates protein synthesis. The targets of these proteins are
still unknown; however, they most likely influence the expression
of regulatory proteins in signaling pathways rather than of
components of dendrites directly. Preliminary analyses of transcripts
potentially targeted by FMRP reveal that synthesis of MAP-2
in dendrites appears not be affected by the FMRP deficit (Steward
et al., 1998
). While it has been shown that FMRP binds transcripts
of itself as well as of other FMRP homologues, no evidence of
binding to transcripts of cytoskeletal or other structural proteins
has been presented (Ceman
et al., 1999
). The large number of
genes potentially regulated by either MeCP2 (Coy
et al., 1999
)
or CBP (Bading, 1999
) suggests a complex mechanism analogous
to the one we proposed for early genes (Kaufmann and Worley,
1999
). In the latter, we postulated that early genes that regulate
transcription have a ‘double effect’ by influencing
the expression of target (e.g. structural protein)
and regulatory
genes (e.g. transcription factor). Despite the large number
of potential targets, we hypothesize that FMRP, MeCP2 and CBP
will ultimately affect common signaling pathways. In ‘convergence’
points, we postulate there are proteins that are essential for
neuronal differentiation and maintenance, such as kinases and
the exclusively neuronal transcription factor NeuN. These ‘secondary’
genes may then modulate the quality and quantity of dendritic
proteins, manifested morphologically as dendritic branch and
spine structure. Figure 5
depicts this model of dendritic formation
and maintenance. Dendritic ‘regression’ in DS may
be the consequence of a different mechanism. Among the genes
overexpressed in trisomy 21 is the one coding for ß amyloid
protein. Amyloid deposition in the form of the Aß peptide
has been detected as early as at 21 gestational weeks in DS
(Teller
et al., 1996
), which coincides with initial cortical
dendritic growth (Huttenlocher, 1999
). Inflammatory responses
associated with Aß deposition in senile plaques in Alzheimer
disease include production of cytokines (Dickson, 1997
), which
may operate as negative modulators of dendritic growth and maintenance
in DS.
The identification of common (nodal) points in ‘dendritic’
cell signaling pathways may be of great importance, since the
regulation of these proteins by neurotransmitters and growth
factors provides the opportunity for therapeutic intervention.
Identification of the nature and timing of negative modulators
of dendritic growth opens the possibility of preventing deleterious
processes. The protracted dendritic development that characterizes
the human cortex provides the opportunity for modifying its
course, and the potential for interrupting or reversing abnormal
synaptic development.
|
Concluding Comments
|
|---|
Dendritic abnormalities are associated with many forms of MR,
and may be the only evident pathology in some of these disorders.
On the other hand, evaluations of presynaptic elements in MR
have not yet been carried out. More systematic studies of both
preand postsynaptic components in multiple cortical regions,
at different stages, are still needed to determine the impact
of dendritic abnormalities on synaptic structure and function.
As it has been demonstrated for neurodegenerative disorders,
establishment of genotype to phenotype molecular pathways, as
well as validation of putative animal models, depend on accurate
characterizations of neuropathologic features (Borchelt
et al.,
1998
). Recent advances in developmental neurobiology and genetics
of MR raise the possibility of more specific therapies for these
disorders. A better characterization and understanding of dendritic
abnormalities in MR will increase our capacity to accomplish
this goal.
|
Notes
|
|---|
Supported in part by NIH grants HD 24448, NS 35359 and HD 24061,
and grant IIRG-99-1608 from the Alzheimer's Association.
Address correspondence to Walter E. Kaufmann, MD, Department of Developmental Cognitive Neurology, Kennedy Krieger Institute, Room 522, 707 N. Broadway, Baltimore, MD 21205, USA. Email: wekaufma{at}jhmi.edu.
|
References
|
|---|
Top
Abstract
Introduction
