Download Darnell, JC, Warren, ST and Darnell, RB: The fragile X mental retardation protein, FMRP, recognizes G-quartets. Mental Retardation and Developmental Disabilities Research Reviews 10:49-52 (2004).

MENTAL RETARDATION AND DEVELOPMENTAL DISABILITIES
RESEARCH REVIEWS 10: 49–52 (2004)
THE FRAGILE X MENTAL RETARDATION
PROTEIN, FMRP, RECOGNIZES G-QUARTETS
Jennifer C. Darnell,1 Stephen T. Warren,2,3* and Robert B. Darnell1,3
1
The Rockefeller University, Laboratory of Molecular Neuro-Oncology New York, New York
2
Emory University School of Medicine, Departments of Human Genetics, Pediatrics and Biochemistry, Atlanta, Georgia
3
Howard Hughes Medical Institute
Fragile X mental retardation is a disease caused by the loss of function of a single RNA-binding protein, FMRP. Identifying the RNA targets
recognized by FMRP is likely to reveal much about its functions in controlling
some aspects of memory and behavior. Recent evidence suggests that one
of the predominant RNA motifs recognized by the FMRP protein is an
intramolecular G-quartet and that the RGG box of FMRP mediates this
interaction. Searching databases of mRNA sequence information, as well as
compiled sequences of predicted FMRP targets based on biochemical identification, has revealed that many of these predicted FMRP targets contain
intramolecular G-quartets. Interestingly, many of the G-quartet containing
RNA targets encode proteins involved in neuronal development and synaptic function. Defects in the metabolism of this set of RNAs, presumably in
the translation of their protein products, is likely to underlie the behavioral
© 2004 Wiley-Liss, Inc.
and cognitive changes seen in the disease.
MRDD Research Reviews 2004;10:49 –52.
Key Words: fragile X mental retardation protein (FMRP); RNA binding
protein; intramolecular G-quartet; in vitro RNA selection; translational control
I
n 2003, a workshop for the planning of future research on
fragile X mental retardation concluded that the number one
priority in the field is the identification of RNA targets to
which the FMRP protein binds. This prioritization clearly reflects both the successes and challenges facing work in this area.
The mystery of the fragile X syndrome lies in understanding how loss of function of one protein can lead to the complex
behavioral, cognitive, and phenotypic changes characteristic of
the disorder. The first clue in this mystery came with the
identification of the FMR1 gene in 1991 as the gene responsible
for fragile X syndrome (reviewed in [O’Donnell and Warren,
2002]. It subsequently became clear that the disorder was due, in
most patients, to expanded CGG triplet repeats in the 5⬘ untranslated region of FMR1 that led to hypermethylation of the
gene and consequently gene silencing [O’Donnell and Warren,
2002]. A single patient whose study has been of some importance in this story is a severely affected fragile X individual who
harbors a single point mutation within the gene (the I304N
mutation) in lieu of a triplet repeat mutation [DeBoulle et al.,
1993].
The second and more recent clues to the syndrome have
come from analysis of the function of the FMRP protein. This
work began with analysis of the predicted amino acid sequence
© 2004 Wiley-Liss, Inc.
of the protein, which revealed that the FMRP harbors several
RNA-binding motifs [Ashley et al., 1993; Gibson et al., 1993;
Siomi et al., 1993]. These include two tandem KH domains, so
named for their homology to the RNA-binding protein
hnRNP-K, and a less well-defined element consisting of repeats
of the amino acid element RGG (the RGG-box). Initial work
on hnRNP-K had revealed a role for this protein in the regulation of translational control of a specific RNA target, the
15-lipoxygenase mRNA expressed in erythrocytes [Ostareck et
al., 1997].
As FMRP is found predominantly in the cytoplasm of
neurons (and other cells) [Devys et al., 1993], a potential role for
FMRP in translational regulation was considered. FMRP was
found to localize to the polysome fraction of the cytoplasm
[Corbin et al., 1997; Feng et al., 1997a; Feng et al., 1997b], and
this localization was associated with functional polysomes, as it
was lost in the presence of sodium azide or sodium fluoride,
agents that cause ribosomes to run off translating mRNA [Feng
et al., 1997a]. Moreover, it was found that FMRP proteins
harboring the I304N mutation did not associate with polysomes,
establishing a link between the fragile-X mental retardation
syndrome and the role of the protein in mRNA translation
[Feng et al., 1997a].
The I304N mutation is located within the second FMRP
KH domain. Although the function for this amino acid has not
been precisely determined, X-ray crystallography of a related
KH domain bound to its RNA target, the Nova neuron-specific
RNA-binding protein, suggests a plausible role for the I304N
mutation [Lewis et al., 2000]. The Nova co-crystal revealed that
the KH domain functions to coordinate a specific RNA sequence, folding to allow the side chains of amino acid residues
to provide hydrogen bond donor and acceptor groups to RNA
molecules in lieu of the normal Watson–Crick hydrogen bonds.
In other words, the KH domain acts as a pseudo-second strand
of nucleic acid to specify sequence-specific RNA binding. The
*Correspondence to: Jennifer C. Darnell, Laboratory of Molecular Neuro-oncology,
The Rockefeller University, 1230, York Ave., New York, NY 10021. E-mail:
[email protected]
Received 9 May 2003; Accepted 1 July 2003
Published online in Wiley InterScience (www.interscience.wiley.com).
DOI: 10.1002/mrdd.20008
I304N mutation maps precisely to the
core of this RNA-binding pocket, suggesting that the mutation in FMRP may
interfere with sequence-specific RNA
binding. More recently, the crystal structures of two additional KH domains
bound to their cognate RNAs have been
solved and the analogous hydrophobic
amino acid to I304 in each case also lies at
the heart of the RNA-binding platform
[Ramos et al., 2003].
These results underscored the importance of identifying the RNA targets
to which FMRP binds. In 1993, Steve
Warren and colleagues identified the
FMR1 mRNA message as an in vitro
target of FMRP and further showed that
FMRP only bound some of a random
group of mRNAs produced from cDNA
clones, suggesting selective RNA binding [Ashley et al., 1993]. A few years later
this finding was extended to the 3⬘UTR
of myelin basic protein [Brown et al.,
1998]. Bob Denman and colleagues used
biotinylated FMRP to capture the RNAs
to which it bound, amplify them by differential display polymerase chain reaction, and determine their identities. In
addition to several unknown genes, prothymosin-a, Tip60a (a tat interactive protein), mitochondrial 26S rRNA, and a
piece of the FMR1 mRNA were identified as possibly binding to FMRP [Sung
et al., 2000]. The biological relevance of
these targets in vivo is unclear. The presumption based on these studies is that
FMRP binds to a subset of brain mRNAs
and that identifying these RNAs and understanding how FMRP acts on them
would provide important clues toward
understanding the nature of the disorder.
FMRP RNA TARGETS
Two parallel and synergizing biochemical approaches have been used to
identify FMRP RNA targets. In the first,
in vitro RNA selection was used to identify the preferred RNA target of the protein [Darnell et al., 2001]. This method
makes very few presumptions about what
RNA elements, whether they be structural or sequence elements, the protein
prefers to bind. The method aims to
identify the highest affinity FMRP targets, which arguably might be an important first step in identifying in vivo RNA
targets in much the same way that identifying the highest affinity protein–DNA
or protein–protein interactions have
been important first steps in identifying
protein function in other systems.
To undertake in vitro RNA selection with FMRP, a series of iterative
experiments were begun in which a pool
(library) of long (52 nucleotides) random
50
RNA sequences harboring PCR-amplifiable elements on their ends, were passed
over FMRP affinity columns. After extensive washing, RNA bound to protein
was specifically eluted with imidazole,
releasing His-tagged FMRP from the
column. The eluted protein was collected, and the RNA extracted, reverse
transcribed, and PCR amplified using the
fixed elements present on the ends of
each RNA in the pool. Because these
elements also contained a T7 RNA polymerase promoter on one end, the pool
could then be re-transcribed into a new
random library of slightly less complexity
(for example, complexity in the original
pool of 1015 different 52-mer RNAs was
reduced to ⬃1012 different RNAs after
this purification). This entire process was
then repeated, with re-purification of
these 1012 RNAs on a second column,
for a total of nine rounds of RNA selection. At the conclusion of this process, a
defined set of RNAs was present, approximately 95% of which bound to
FMRP directly and with high affinity
[Darnell et al., 2001].
The presumption based
on these studies is that
FMRP binds to a subset
of brain mRNAs, and
that identifying these
RNAs and understanding
how FMRP acts on them
would provide important
clues toward
understanding the nature
of the disorder.
A series of mutagenesis and binding experiments suggested several important points about the FMRP RNA targets. First, the domain that had selected
the RNA target was not the FMRP KH
domains, but the RGG box, suggesting
that within the context of the full-length
protein, this element may have the highest affinity for RNA. Second, the nature
of the RNA was such that it appeared to
form a complex tertiary structure termed
a G-quartet; in essence, a complex stemloop element that folds back on itself to
form non-canonical (Hoogsteen) base–
base interactions. Such complex elements
MRDD RESEARCH REVIEWS
●
have previously been identified as targets
for a number of DNA-binding proteins,
and in one case (a bacteriophage protein)
as the target for an RNA-binding protein
involved in translational regulation [Oliver et al., 2000]. G-quartet elements
form around a single small monovalent
cation (sodium or potassium), and the
requirement for FMRP binding to this
element was confirmed by the observation that FMRP bound to its selected
RNAs in the presence of potassium ions,
but not lithium ions, which are too small
to support the G-quartet structure [Darnell et al., 2001]. The net result of these
studies was identification of an RNA target with enough complexity that a bioinformatic screen of the genome database
for the presence of such structures led to
the identification of a limited set of targets, which were validated as discussed
below.
Concurrently, Herve Moine and
colleagues, in pursuing the finding that
FMRP binds to its own mRNA in vitro,
delineated the sequences within FMR1
message necessary for FMRP binding
[Schaeffer et al., 2001]. They found that
the FMRP-binding element folded into a
purine quartet that was bound by FMRP
with very high affinity. In addition, they
demonstrated that the interaction could
occur in a reticulocyte lysate, because
placement of the purine quartet near the
5⬘ end of a transcript repressed translation
of a reporter gene in an FMRP-dependent manner. This intriguing result
showing modulation of translation of an
mRNA in both a G-quartet and FMRPdependent manner further substantiates
the finding that FMRP recognizes Gquartet motifs in RNA targets and may
use this mechanism in vivo to control
translation of specific mRNAs [Moine
and Mandel, 2001].
A biologically relevant approach to
identifying FMRP targets came from a
combination of immunoprecipitation
and gene array analyses [Brown et al.,
2001]. One issue hampering the use of
immunoprecipitation in the study of
FMRP has been the identification of a
number of closely related proteins,
FXR-1 and FXR-2, both of which are
recognized by most “FMRP” antibodies.
Therefore the development of an
FMRP-specific antibody was an important first step in these experiments. This
antibody was found to be able to coprecipitate protein–RNA complexes from
mouse brain, although such a coprecipitation by itself was not specific enough to
directly identify FMRP targets. Thus
cDNAs were generated from such
RNA—protein precipitates and used to
THE FMRP PROTEIN RECOGNIZES G-QUARTETS
●
DARNELL
ET AL.
probe Affymetrix gene chips. To improve the signal to noise in these experiments, parallel arrays were probed with
samples from control immunoprecipitations, including Ips, from FMRP-null
mice brains. Serial subtraction led to the
identification of a small number (several
hundred) of RNAs that appeared to be
enriched in FMRP coprecipitations relative to both input RNA and abundance
in FMR1-null mouse brain immunoprecipitations [Brown et al., 2001].
To validate these targets, and address which RNAs might be directly
bound by FMRP (in vivo the FMRP
protein is present in a multiprotein complex that includes at least five other
RNA-binding proteins [Ceman et al.,
1999; Ceman et al., 2000; Ohashi et al.,
2002]), these proteins were analyzed for
the presence of G-quartet elements. Such
RNAs, as well as those identified by the
bioinformatics screen of the database for
the presence of G-quartets, led to 13
candidate FMRP targets that were further analyzed [Brown et al., 2001; Darnell et al., 2001].
The last level of validation of these
targets came from returning to the observation that FMRP is present on polysomes. This suggested that analysis of the
presence of these candidate RNAs on
polysomes as a function of FMRP (e.g.,
lymphoblastoid cell lines from pooled
normal individuals versus pooled fragile
X patients who express no FMRP)
would provide an important measure of
their likelihood as legitimate FMRP target RNAs. In fact, analysis of polysomes
prepared from the above cell lines led to
the finding that 10 of 13 candidate
FMRP RNA targets showed alteration
in their polysome distribution in the absence of FMRP [Brown et al., 2001].
About half of these RNAs showed increases and half decreases in their polysome association in the absence of
FMRP, suggesting that FMRP may normally be able to bind to specific mRNAs
to act to either facilitate or inhibit their
translation [Brown et al., 2001].
The small number of mRNAs
passing through these rigorous selection
criteria as candidate FMRP RNA targets
share a remarkably coherent biology.
Eleven of the set of 13 target RNAs
encode proteins whose function relates to
the biology of the synapse. These include
receptors and channels (the V1a receptor
and potassium channel Kv 3.1), proteins
involved in synaptic function (munc
13–2, NAP-22, sec-7 related guanine
nucleotide exchange factor, and the rab
6-binding protein), and three proteins
involved in neuritic extension and develMRDD RESEARCH REVIEWS
●
opment (MAP1B, semaphorin 3F, and
ID3) [Brown et al., 2001; Darnell et al.,
2001]. While work progresses on the further validation of these targets one in
particular deserves further mention.
MAP1B plays an important role in the
extension of dendrites and axons in the
developing neuron [Mandell and Banker,
1996; Takei et al., 2000]. It has recently
been shown that the Drosophila homolog
of FMRP, dFXR, acts as a translational
repressor of futsch, the fly homolog of
MAP1B [Zhang et al., 2001]. This is
consistent with the finding that MAP1B
mRNA is increased in the polysome
fractions of fragile X patient cells that
lack FMRP [Brown et al., 2001]. Interestingly, transcription of the rat MAP1B
message begins at one of three sites [Liu
and Fischer, 1996]. The G-quartet identified in MAP1B lies within 60 nucleotides downstream of the most 5⬘ start site
[Darnell et al., 2001]. Data suggesting
that any protein is a translational repressor if it binds an mRNA within this
distance of the cap site suggests that
FMRP binding to this site in vivo would
necessarily repress cap-dependent translation by steric hindrance [Stripecke et al.,
1994].
Recently, two other groups have
used additional approaches to identify
candidate RNA targets for FMRP.
Greenough, Eberwine and colleagues
used a technique called APRA (antibody-positioned RNA amplification) to
amplify RNAs in close proximity to the
epitope recognized by the anti-FMRP
antibody 1C3 in neuronal cultures [Miyashiro et al., 2003]. Primary rat hippocampal cells were fixed and permeabilized followed by application of the
antibody linked to random primers. Using previously established methods of in
situ transcription of cDNA on the cells
[Eberwine et al., 2001], a pool of probes
were obtained that were used to screen
cDNA arrays. The proteins encoded by
the RNA cargoes of FMRP identified by
this approach fall into many categories,
including cell signaling and communication molecules, those involved in cell
structure and motility, in the secretory
system, and in the regulation of transcription or translation (Table 1 [Miyashiro et
al., 2003]). Interestingly; the panel of targets was screened for G-quartet-like elements and a similar motif was found in 24
of 83 searched, although a stem surrounding the G-quartet was not included
as part of the query, nor was it stated how
many mRNAs in the unselected database
are positive with that particular query. It
should be noted that in both the Darnell/
Warren and Greenough/Eberwine searches
THE FMRP PROTEIN RECOGNIZES G-QUARTETS
●
for G-quartets, these queries are not likely
to be exhaustive. That is, until one can
better define exactly which G-quartets
FMRP recognizes, these queries are likely
to pick up only a subset of the actual Gquartet-binding sites present in mRNAs.
Conversely, putative G-quartets picked up
by sequence homology; but not tested for
FMRP binding in a potassium-dependent
manner, may not be true FMRP-binding
sites. This was shown by the significant
number of putative G-quartets picked up
by the RNABob program [Eddy; 2001;
Gautheret et al., 1990] cited in Darnell et
al. [2001] that did not bind FMRP when
tested. It is likely that they do not fold into
G-quartets because of competing structures
in the RNA or that there is sequence specificity in the loops of the motif.
Bagni et al. have recently shown
that FMRP associates with the dendritically localized, nontranslated RNAs BC1
in rodents and BC200 in primates [Zalfa
et al., 2003]. BC1 and BC200 may anneal with 3⬘UTR elements in some dendritically localized messages, including
MAP1B; ␣-CaM kinase II and Arc and
provide a mechanism through which
FMRP negatively regulates the translation of the messages. This would explain
their finding that these mRNAs are increased in their translation status in
FMR1 null mice. Although this is an
intriguing model, it has also been reported that FMRP does not bind BC200
[Kremerskothen et al., 1998]. Therefore,
it is difficult to fully evaluate this model
until the interaction between FMRP and
BC1 has been independently confirmed.
Taken together, these data provide
evidence for how FMRP may function
as an RNA-binding protein in health and
disease—to regulate the expression of
proteins that play critical roles in the regulation of the synapse itself. These observations fit neatly with a number of studies reviewed elsewhere in this volume,
suggesting that FMRP plays an important
role in the development and function of
the dendrite. f
REFERENCES
Ashley CT, Wilkinson KD, ReinesD, et al. 1993.
FMR-1 protein: conserved RNP family domains and selective RNA binding. Science
262:563–566.
Brown V, Jin P, Ceman S, et al. 2001. Microarray
identification of FMRP-associated brain
mRNAs and altered mRNA translational
profiles in Fragile X Syndrome. Cell 107,
477– 487.
Brown V, Small K, Lakkis L, et al. 1998. Purified
recombinant Fmrp exhibits selective RNA
binding as an intrinsic property of the fragile
X mental retardation protein. J Biol Chem
273:15521–15527.
Ceman S, Brown V, Warren ST. 1999. Isolation of
an FMRP-associated messenger ribonucleo-
DARNELL
ET AL.
51
protein particle and identification of nucleolin
and the fragile X-related proteins as components of the complex. Mol Cell Biol 19:7925–
7932.
Ceman S, Nelson R, Warren ST. 2000. Identification of mouse YB1/p50 as a component of
the FMRP-associated mRNP particle. Biochem Biophys Res Commun 279:904 –908.
Corbin F, Bouillon M, Fortin A, et al. 1997. The
fragile X mental retardation protein is associated with poly(A)⫹ mRNA in actively translating polyribosomes. Hum Mol Genet
6:1465–1472.
Darnell JC, Jensen KB, Jin P, et al. 2001. Fragile X
mental retardation protein targets G Quartet
mRNAs important for neuronal function.
Cell 107:489 – 499.
DeBoulle K, Verkerk A, Reyniers E, et al. 1993. A
point mutation in the FMR-1 gene associated
with fragile X mental retardation. Nat Genet
3:31–35.
Devys D, Lutz Y, Rouyer N, et al. 1993. The
FMR-1 protein is cytoplasmic, most abundant in neurons and appears normal in carriers
of a fragile X premutation. Nat Genet 4:335–
340.
Eberwine J, Kacharmina JE, Andrews C, et al.
2001. mRna expression analysis of tissue sections and single cells. J Neurosci 21:8310 –
8314.
Eddy S. 2001. http://www.genetics.wustl.edu/
eddy/software/.
Feng Y, Absher D, Eberhart DE, et al. 1997a.
FMRP associates with polyribosomes as an
mRNP, and the I304N mutation of severe
fragile X syndrome abolishes this association.
Mol Cell 1:109 –118.
Feng Y, Gutekunst CA, Eberhart DE, et al. 1997b.
Fragile X mental retardation protein–nucleocytoplasmic shuttling and association with somatodendritic ribosomes. J Neurosci
17:1539 –1547.
Gautheret D, Major F, Cedergren R. 1990. Pattern
searching/alignment with RNA primary and
52
secondary structures: an effective descriptor
for tRNA. Comput Appl Biosci 6:325–331.
Gibson T, Rice P, Thompson J, et al. 1993. KH
domains within the FMR1 sequence suggest
that fragile X syndrome stems from a defect in
RNA metabolism. Trends Biochem Sci 18:
331–333.
Kremerskothen J, Nettermann M, op de Bekke A,
et al. 1998. Identification of human autoantigen La/SS-B as BC1/BC200 RNA-binding
protein. DNA Cell Biol 17:751–759.
Lewis HA, Musunuru K, Jensen KB, et al. 2000.
Sequence-specific RNA binding by a Nova
KH domain: implications for paraneoplastic
disease and the fragile X syndrome. Cell 100:
323–332.
Liu D, Fischer I. 1996. Two alternative promoters
direct neuron-specific expression of the rat
microtubule-associated protein 1B gene.
J Neurosci 16:5026 –5036.
Mandell JW, Banker GA. 1996. Microtubule-associated proteins, phosphorylation gradients,
and the establishment of neuronal polarity.
Perspect Dev Neurobiol 4:125–135.
Miyashiro KY, Beckel-Mitchener A, Purk TP, et
al. 2003. RNA cargoes associating with
FMRP reveal deficits in cellular functioning
in Fmr1 null mice. Neuron 37:417– 431.
Moine H, Mandel JL. 2001. Biomedicine. Do G
quartets orchestrate fragile X pathology? Science 294:2487–2488.
O’Donnell WT, Warren ST. 2002. A decade of
molecular studies of fragile X syndrome.
Annu Rev Neurosci 25:315–338.
Ohashi S, Koike K, Omori A, et al. 2002. Identification of mRNA/protein (mRNP) complexes containing Puralpha, mStaufen, fragile
X protein, and myosin Va and their association with rough endoplasmic reticulum
equipped with a kinesin motor. J Biol Chem
277:37804 –37810.
Oliver AW, Bogdarina I, Schroeder E, et al. 2000.
Preferential binding of fd gene 5 protein to
MRDD RESEARCH REVIEWS
●
tetraplex nucleic acid structures. J Mol Biol
301:575–584.
Ostareck DH, Ostarecklederer A, Wilm M, et al.
1997. mRNA silencing in erythroid differentiation– hnRNP K and hnRNP E1 regulate
15-lipoxygenase translation from the 3⬘ end.
Cell 89:597– 606.
Ramos A, Hollingworth D, Pastore A. 2003. The
role of a clinically important mutation in the
fold and RNA-binding properties of KH motifs. RNA 9:293–298.
Schaeffer C, Bardoni B, Mandel JL, et al. 2001. The
fragile X mental retardation protein binds
specifically to its mRNA via a purine quartet
motif. EMBO J 20:4803– 4813.
Siomi H, Siomi M, Nussbaum R, et al. 1993. The
protein product of the fragile X gene, FMR1,
has characteristics of an RNA-binding protein. Cell 74:291–298.
Stripecke R, Oliveira C, McCarthy J, et al. 1994.
Proteins binding to 5⬘ untranslated region
sites: a general mechanism for translation regulation of mRNAs in human and yeast cells.
Mol Cell Biol 14:5898 –5909.
Sung YJ, Conti J, Currie JR, et al. 2000. RNAs
that interact with the fragile X syndrome
RNA binding protein FMRP. Biochem Biophys Res Commun 275:973–980.
Takei Y, Teng J, Harada A, et al. 2000. Defects in
axonal elongation and neuronal migration in
mice with disrupted tau and map1b genes.
J Cell Biol 150:989 –1000.
Zalfa F, Giorgi M, Primerano B, et al. 2003. The
fragile X syndrome protein FMRP associates
with BC1 RNA and regulates the translation
of specific mRNAs at synapses. Cell 112:317–
327.
Zhang YQ, Bailey AM, Matthies HJ, et al. 2001.
Drosophila Fragile X-Related Gene Regulates the MAP1B Homolog Futsch to Control
Synaptic Structure and Function. Cell 107:
591– 603.
THE FMRP PROTEIN RECOGNIZES G-QUARTETS
●
DARNELL
ET AL.