|  This series
of columns has reviewed how experience and repetition are critical
to learning. Exactly
how learning occurs at a molecular level within the
central nervous system has interested neuroscientists for decades, and
several important concepts have emerged. First, the basic pattern of neuronal
organization appears to be largely intrinsic to the developing brain. At
birth, the human brain already contains the majority of neurons it will
have in adulthood. In fact, there are many more neurons present at birth
than are actually needed, and approximately half will die through the process
of programmed cell death due to lack of use. The axiom “use them
or lose them” applies.
A second, related concept is that neuronal activity strengthens immature
synaptic connections between neurons, whereas inactive synapses weaken
and die away. This process, termed activity-dependent synaptic plasticity,
is thought to underlie learning. It requires structural modifications to
existing synaptic connections. During development, activity-dependent synaptic
plasticity is based on competition for limited amounts of growth factors
released by target neurons. More active synapses mature and strengthen.
Those that are less active receive insufficient levels of trophic factors
and undergo apoptosis.
The external environment exerts its effects on the brain through neuronal
activity. Persistent or repetitive neuronal activity leads to structural
modifications at exactly those synapses that are being used. Moreover,
in regions of the brain that are devoted to learning and memory, repetitive
firing at a synapse leads to a striking change in a neuron’s ability
to respond to action potentials. The synapse becomes more responsive to
future action potentials, as the amount of neurotransmitter required to
generate a postsynaptic action potential is reduced. This response, called
long-term potentiation, is thought to underlie learning and memory in brain
regions including the hippocampus, striatum, amygdala, and nucleus accumbens.
Thus two events occur over time with repetitive neuronal activity. The
number of synaptic connections between a neuron and its targets increases,
and the sensitivity of the individual synapse to neurotransmitters becomes
stronger. Both phenomena are thought to underlie the process by which we
learn and become fluent in various skills, and both phenomena require new
protein synthesis.
This column will review aspects of learning and memory over the next several
months. It will cover different forms of memory including declarative,
nondeclarative, and working memory. Experts in the field will summarize
evidence indicating that different brain regions are responsible for different
types of memory. Future columns will outline the molecular events that
take place within neurons that lead to the creation, storage, and retrieval
of memory. In addition, we will discuss what happens to cognitive processes
when mutations occur in genes encoding proteins important for memory formation.
Hundreds of proteins take part in these events, making it likely that many
developmental disorders and mental retardation syndromes result from mutations
that disrupt one or more of these proteins. It is also possible, although
this is more speculative, that certain psychiatric disorders are due to
disrupted signaling events involved in the formation and acquisition of
memory and its storage and retrieval. Specific phobias and anxiety disorders,
and the spectrum of stress disorders, are likely to involve abnormal processing
of memories. In the present column, we focus on one aspect of this field,
fragile X syndrome, and how discoveries of the genetic basis for this disease
have once again linked basic neuroscience with developmental neurobiology.
As mentioned above, activity-dependent synaptic plasticity involves modification
to existing synapses. Structural changes occur at both the pre- and postsynaptic
sites. For example, where before there was one synapse, two or more synapses
form as the neuronal connection responds to synaptic activity. This and
additional changes at the postsynaptic site mediate a stronger response
to the incoming signal.
How exactly do these structural changes occur? An individual neuron can
make contact with 1,000 other neurons, while many additional neurons may
synapse on that initial neuron. How does a postsynaptic neuron distinguish
among the thousands of potential sites on its dendritic arbor those that
require structural modification? This puzzle was partially solved when
it was shown that synaptic modification requires new protein synthesis.
Several events are required for new protein synthesis. A signal needs to
arrive at the neuron’s nucleus. This signal must be transmitted to
the nucleus after proteins called transcription factors have been activated.
Transcription factors must bind to promoters, which are the regulatory
regions on individual genes. Depending on which transcription factors are
activated, the promoter region either represses or enhances transcription
of that gene. When transcription is enhanced, messenger RNAs (mRNAs) are
rapidly transported to the cytoplasm for translation.
Earlier dogma suggested that translation occurred at ribosomes in the cytoplasm
around the nucleus, and therefore a considerable distance from the dendritic
spine where the initial synaptic input arrived, and where synaptic modification
was needed. But how would the newly synthesized synapse-related proteins “know” where
to go once synthesized? How are they transported to the correct synapse
where modifications are needed?
A new hypothesis was needed, and one soon emerged. Perhaps the idea that
new protein synthesis occurred only in the cell body of neurons, close
to the nucleus, was incomplete. If the mRNAs themselves were transported
to all the spines in the dendritic arbor, then they would be positioned
at all postsynaptic terminals, poised, as it were, to be translated into
protein after the arrival of the appropriate signal. This hypothesis offered
a solution to the problem of getting the newly synthesized proteins back
to only the specific spines where they were required.
Over the past several years, considerable data have accumulated in support
the idea that mRNAs are themselves transported to spines throughout the
dendritic arbor. This allows rapid, local, and selective translation of
proteins only at the spines where they are needed, so only a subset of
neuron’s synapses are modified in response to activation. Once again,
however, new questions emerged. How do mRNAs travel to the spines, and
how are they regulated? This is where the protein that is mutated in fragile
X syndrome comes in.
To understand the molecular biology of fragile X syndrome, it is useful
to review the changes in the brains of affected individuals. Overall, autopsy
analyses reveal few light microscopic changes. Higher-resolution analysis
with electron microscopy shows that the dendritic spines from fragile X
patients are abnormal. They are morphologically similar to the spines of
immature, developing brains. The dendritic spines in fragile X individuals
are long and thin compared with the short and broad spines of mature cortical
neurons (Fig. 1).
We know that neurogenesis and neuronal migration proceed normally in fragile
X individuals, suggesting that the pathology occurs later, during synaptic
maturation. Apparently, the protein affected by fragile X syndrome or proteins
that depend on this protein are necessary for proper synaptic growth and
maturation. We now turn to how this story unfolded over the past decade
through advances in molecular biology.
Fragile X syndrome is the most common form of inherited mental retardation.
It occurs with a frequency of approximately 1:4,000 in males and 1:8,000
in females. The fragile X mental retardation-1 (FMR1) gene was cloned more
than a decade ago—a major accomplishment. Characterization of the
mutation revealed a novel type of genetic mutation called a triplet repeat
expansion. This type of mutation was originally identified in Huntington
chorea. When the FMR1 gene was sequenced, and proved to be second example
of a triplet repeat expansion, genetic researchers realized that they had
discovered a new type of mutation. Since then, triplet repeat expansions
have been described in approximately a dozen neuropsychiatric disorders.
These mutations have in common the expansion of an unstable series of nucleotides.
In fragile X syndrome, three nucleotides (CGG, cytosine-guanine-guanine)
are abnormally repeated over and over again, and they may reach several
thousands of nucleotides in length in the most severely affected individuals.
The presence of a small number (between 5 and 50) of these triplet repeats
is normally present in the FMR1 gene. In some individuals, however, the
trinucleotide repeat expands to 200 repeated CGGs. When this happens, the
affected individual is a carrier of a premutation. Then, in the next generation,
a dramatic expansion of the repeated sequence among the offspring results
in the full-blown clinical syndrome. It is not clear how this expansion
occurs or how the premutation in the parent arises.
The consequences of this expansion, however, are dramatic. A chemical modification
called methylation occurs throughout the expanded region. In addition,
a complex folding to the secondary structure of the DNA molecule occurs.
These changes interfere with normal gene transcription. The enzyme that
needs to gain access to the DNA molecule to initiate transcription is unable
to do so. The net effect is that no message is made and no functional fragile
X mental retardation protein (FMRP) is translated.
The next question to interest researchers was how the absence of FMRP leads
to mental retardation. Once the gene for fragile X syndrome had been cloned,
researchers were able to sequence it and translate the nucleotide sequence
into its predicted amino acid sequence. They were then able to determine
whether there were any regions within FMRP that were homologous to known
proteins. The presence of preserved motifs with known function could suggest
a similar function for FMRP that could be tested in the laboratory. Indeed,
several amino acid domains were found that were highly homologous to known
motifs within other proteins. Three of these domains had previously been
shown to bind to RNA molecules, and the proteins that contain them are
therefore called RNA-binding proteins.
RNA has several functions within cells. The messages transcribed from DNA
consist of mRNA molecules. In addition, the ribosomal factories in the
cytoplasm that translate mRNA into protein contain RNA. These RNA populations
do not exist by themselves. In the case of mRNAs, a number of proteins
bind to them to chaperone them from one compartment to another, and to
facilitate their translation into protein. In the case of the ribosomes
themselves, RNA-binding proteins provide an organized structure for the
protein translation apparatus. In fact, ribosomes contain up to 80 different
proteins that come together in a complex that give the ribosomes their
form and function. Once again, these proteins have RNA-binding motifs that
allow them to associate with RNA molecules destined to be incorporated
into ribosomes, and form the scaffold that will accept mRNA molecules and
translate them into proteins.
A new series of experiments were designed to test whether FMRP could bind
to RNA and, if so, which type of RNA. Researchers showed that FMRP is able
to bind to synthetic polymers of RNA in vitro. Immunocytochemical analyses
showed that FMRP is predominantly cytoplasmic. Moreover, the protein has
two additional amino acid sequences that provide additional clues as to
its function. One is a nuclear localization signal present on proteins
that are at some point transported into the nucleus. A second amino acid
sequence, called a nuclear export signal, has the opposite effect. The
presence of both of these signals raised the intriguing possibility that
FMRP shuttles RNA messages from the nucleus to the cytoplasm, and then
returns to the nucleus to pick up a new RNA message.
At approximately the same time that this work was being done, a patient
with a very severe form of fragile X syndrome was discovered. Careful analysis
of his FMR1 gene revealed no triplet repeat expansion. Instead, researchers
found a point mutation that changed a single amino acid within one of the
putative RNA-binding domains. This mutation resulted in a protein that
was unable to bind with RNA molecules. This was strong evidence that the
ability of FMRP to bind to RNA was critical to its proper function. Together,
these observations suggested that FMRP normally binds to a subset of mRNAs.
The absence of FMRP disrupts normal translation of these target messages.
Subsequent experiments attempted to test this hypothesis by identifying
the subset of messages that bind with FMRP.
With a combination of biochemistry and a newer technology—microarray
analysis—two groups have now identified a series of mRNAs that associate
with FMRP. Their experimental approach was elegant and worth reviewing.
An antibody that recognizes FMRP had already been generated. It could be
used to pull-down, or immunoprecipitate, FMRP from a mixture of brain proteins
where FMRP normally resides. Under the right buffer conditions, any proteins
or mRNAs bound to FMRP would co-immunoprecipitate. Such experiments revealed
a number of messages that co-immunoprecipitated with FMRP. Exactly how
these messages were subsequently identified requires a brief description
of microarrays.
Microarrays are small chips to which thousands of known DNA sequences have
been robotically attached. The DNA sequences are complementary to the mRNAs
(thus called cDNAs). Chips are now available onto which 10,000 or more
cDNAs have been placed. Researchers can obtain chips for any tissue, including
brain-enriched microarrays. The exact location of each cDNA on the chip
is known and can be distinguished from any of the other attached sequences.
Researchers investigating the molecular basis of fragile X labeled immunoprecipitated
material containing the unknown mRNAs that were associated with the FMRP
complex. The mRNAs were labeled with fluorescent tags to make them visible
probes, and they were placed (in solution) over the microarray. Because
complementary nucleic acid sequences bind very tightly to each other, mRNAs
immunoprecipitated from the brain would bind to their complementary cDNAs
on the microarray. The cDNAs could now be fluorescently identified by their
position on the array. At this point, it becomes straightforward to determine
which of the thousands of brain-enriched messages were capable of binding
to FMRP. In this manner, a handful of mRNAs were identified. A particularly
interesting finding was that the bound messages all had a specific sequence
(called a G quartet) that was absolutely required for the message to bind
to FMRP.
One of the messages that was identified in this way encodes for a protein
called microtubule-associated protein MAP1B. One of the known functions
of this protein is to provide structural organization for the synapse.
For this reason, current thinking has it that the absence of FMRP leads
to a dysregulation of MAP1B expression at the synapse. The absence of MAP1B
at the synapse in turn is believed now to contribute to the structural
abnormalities seen with electron microscopy in individuals with fragile
X syndrome. Their inability to reorganize their synaptic architecture is
now believed to be the underlying basis for the cognitive abnormalities
characterizing this disorder.
top of page
|
