Content uploaded by Bernard W Balleine
Author content
All content in this area was uploaded by Bernard W Balleine
Content may be subject to copyright.
ORIGINAL RESEARCH ARTICLE
Sensorimotor gating abnormalities in young males
with fragile X syndrome and Fmr1-knockout mice
PW Frankland
1,2,3,6
, Y Wang
1,2,3
, B Rosner
2,5
, T Shimizu
1,2,3
, BW Balleine
3,4
, EM Dykens
2,5
, EM Ornitz
2,4
and AJ Silva
1,2,3,4,5
1
Department of Neurobiology, UCLA, Los Angeles, CA 90095, USA;
2
Department of Psychiatry, UCLA, Los Angeles,
CA 90095, USA;
3
Department of Psychology, UCLA, Los Angeles, CA 90095, USA;
4
Brain Research Institute, UCLA, Los
Angeles, CA 90095, USA;
5
Mental Retardation Research Center, UCLA, Los Angeles, CA 90095, USA
Fragile X syndrome (FXS) is the most common single gene (FMR1) disorder affecting cognitive
and behavioral function in humans. This syndrome is characterized by a cluster of
abnormalities including lower IQ, attention deficits, impairments in adaptive behavior and
increased incidence of autism. Here, we show that young males with FXS have profound
deficits in prepulse inhibition (PPI), a basic marker of sensorimotor gating that has been
extensively studied in rodents. Importantly, the magnitude of the PPI impairments in the fragile
X children predicted the severity of their IQ, attention, adaptive behavior and autistic
phenotypes. Additionally, these measures were highly correlated with each other, suggesting
that a shared mechanism underlies this complex phenotypic cluster. Studies in Fmr1-knockout
mice also revealed sensorimotor gating and learning abnormalities. However, PPI and learning
were enhanced rather than reduced in the mutants. Therefore, these data show that mutations
of the FMR1 gene impact equivalent processes in both humans and mice. However, since
these phenotypic changes are opposite in direction, they also suggest that murine
compensatory mechanisms following loss of FMR1 function differ from those in humans.
Molecular Psychiatry (2004) 9, 417–425. doi:10.1038/sj.mp.4001432
Published online 10 February 2004
Keywords: mouse model; FMRP; prepulse inhibition; startle
Introduction
Fragile X syndrome (FXS) is the most prevalent form
of inherited mental retardation, affecting about 1 in
4000 males.
1
In most instances, FXS is caused by large
expansions of a CGG trinucleotide repeat in the
promoter region of the fragile x mental retardation 1
(FMR1) gene.
2
The resulting absence, or reduced
expression, of the fragile X mental retardation protein
(FMRP) is responsible for a broad spectrum of
physical, behavioral and cognitive abnormalities.
3–5
FMRP is an RNA-binding protein that regulates local
protein synthesis required for synaptic maturation
and plasticity.
6
Targeted deletion of the Fmr1 gene
results in the loss of FMRP in mice.
7
These Fmr1-
knockout (Fmr1-KO) mice exhibit physical features of
FXS, such as enlarged testes,
7–9
indicating that these
mice provide a useful model of the syndrome. Here,
we examine the impact of the loss of FMRP on
cognitive and behavioral function in parallel studies
in mice and humans.
In addition to learning impairments, one of the
most common clinical features of FXS is heightened
sensitivity to sensory stimulation (or sensory defen-
siveness).
2,10
Altered sensitivity to sensory stimula-
tion might reflect underlying abnormalities in the
maturation of synaptic connections in sensory cir-
cuits.
6
To quantify changes in sensory processing, we
examined prepulse inhibition (PPI) of acoustic startle
in young males with FXS. PPI is a widely used
behavioral model of basic sensorimotor processing,
11
where a weak auditory prepulse attenuates subse-
quent responses to a loud startling noise.
Materials and methods
Human studies
Subjects
Young males diagnosed with the full
mutation (n¼10) were recruited through the local
chapters of the Fragile X Syndrome Association and
the FRAXA Research Foundation. Control subjects
(n¼7), matched for chronological age, puberty status,
ethnicity and family demographic variables (number
of siblings, parental age, income and education), were
recruited through advertising at UCLA. Two of the
Received 07 May 2003; revised 21 August 2003; accepted 10
September 2003
Correspondence: AJ Silva, PhD, Departments of Neurobiology,
Psychology and Psychiatry, 695 Charles E Young Drive South,
UCLA, Los Angeles, CA 90095-1761, USA.
E-mail: silvaa@ucla.edu;
EM Ornitz, MD, Department of Psychiatry and Brain Research
Institute, 760 Westwood Plaza, UCLA, Los Angeles, CA 90024,
USA. E-mail: eornitz@mednet.ucla.edu
6
Current address: Integrative Biology, Hospital for Sick Children
Research Institute, Toronto, Ontario, Canada M5G 1X8
Molecular Psychiatry (2004) 9, 417–425
&
2004 Nature Publishing Group All rights reserved 1359-4184/04
$
25.00
www.nature.com/mp
FXS subjects were brothers. One additional FXS
subject refused attachment of the recording
electrodes, and did not participate in the study. FXS
may be accompanied by hyperactivity, inattention,
irritability, anxiety, sensory sensitivities, aggression
and seizure disorders. These problems are
particularly severe during childhood, and
consequently many of the subjects (8/10) in the
present study were taking multiple neuropsychiatric
medications at the time of testing (see Table 2).
Subjects had been taking these medications between 5
months and 9 years. It was considered clinically
inadvisable to take them off their medication for the
purposes of this study. None of the control subjects
were on medication at the time of testing. Prior to the
experimental session, all subjects were
audiometrically screened for normal hearing.
Cognitive and behavioral assessment
Questionnaires (Child Behavioral Checklist, CBCL;
Autism Screening Questionnaire, ASQ) and semi-
structured interviews (Vineland Adaptive Behavior
Scales, VABS) were administered to parents to assess
cognitive and behavioral function. Adaptive behavior
(communication, socialization and daily living skills)
was characterized using the VABS.
12
Attention defi-
cits were characterized using the attention domain of
the CBCL.
13
The prevalence of behaviors associated
with autism (reciprocal social interaction, language
and communication, and repetitive and stereotyped
patterns of behaviors) was measured using the ASQ.
14
This questionnaire was administered only to parents
of FXS boys. In addition, the Kaufman Brief Intelli-
gence Test (K-BIT
15
) was administered to all children
to measure verbal and nonverbal intelligence.
Startle testing
Stimuli
General procedures for examination of PPI in
children have been previously described in detail.
16
Briefly, acoustic stimuli were presented binaurally
through Sony circumaural earphones. Startle stimuli
were 105 dB SPL (50 ms duration) white-noise bursts,
with 0 ms rise/fall times. Acoustic prepulses were
75 dB SPL (25 ms duration) 1000 Hz tones, with 4 ms
rise/fall times. On prepulse/startle trials, the prepulse
preceded the startle stimulus by 120 ms (onset to
onset). Ambient room noise was 30 dB SPL.
Recording Orbicularis oculi EMGs were recorded
bipolarly from the right eye with gold cup electrodes
(1.0 cm apart, edge to edge, as close to the margin of
the lower lid as possible, and the lateral electrode
0.6 cm medial to the outer canthus). EMGs were AC
amplified (Grass 15RX series Physiodata Amplifier
System) at a fixed gain (10 000) with filters set at half-
amplitude 30–1000 Hz. Vertical eye and lid
movements were recorded bipolarly (with filters set
at half-amplitude 0.01–30 Hz) with silver-silver
chloride electrodes placed above and below one eye
(EOG). All data were digitally recorded at 1000 Hz
and stored from 15 s before to 10.5 s after startle
stimulus onset.
Data processing On each trial, the rectified
orbicularis oculi EMG and the vertical EOG were
examined. Trials were rejected prior to data analysis if
there was EMG or EOG activity (indicating orbicularis
oculi contraction or lid movement) in the 20 ms
period following presentation of the startle stimulus.
FXS subjects had an average of 2.5 (range 0–8)
rejected trials per subject; control subjects had an
average of 1.3 (range 0–2) rejected trials per subject.
For accepted trials, with response onset between 20
and 80 ms, peak amplitude of the rectified EMG was
defined as the highest point within a window from
response onset to 105 ms following startle stimulus
onset, and was measured relative to a 200 ms
prestartle stimulus onset baseline. When there was
no measurable increase in EMG activity between 20
and 80 ms following the startle stimulus, the startle
response was taken to be zero.
Procedures Electrodes were applied and the subjects
were asked to sit quietly and watch a silent
videotaped movie. Prior to the first block of trials,
subjects were presented either with a single startle
only trial, or a series of six adaptation trials (startle
only trials increasing in intensity from 80 to 105 dB in
5 dB increments). Responses to these adaptation trials
were not used in subsequent data analysis. Following
this, subjects were presented with two types of trials:
those where only the startle stimulus was presented
and those where the lower intensity acoustic prepulse
preceded the startle stimulus by 120 ms (prepulse/
startle trials). The two trial types were presented in
eight blocks of four trials each. One FXS subject
completed only six blocks; all other subjects
completed all eight blocks. Within each block, each
type of trial occurred twice in a pseudorandom order.
The minimum intertrial interval was 44–63 s. For
each subject, EMG and EOG were continuously
monitored throughout the test session. Trials were
only initiated if (a) these physiological markers were
stable during an 8 s pretrial period, and (b) there were
no spontaneous blinks, eye movements or other
periocular activity during the 200 ms preceding
presentation of stimuli. Specific details of these test
conditions and procedures have been described
elsewhere.
16
Data analysis Startle response magnitude was taken to
be the peak EMG amplitude occurring within a 105 ms
window following the onset of the startle stimulus.
Using means of all usable trials, percent PPI was
calculated for each subject according to the following
formula: %PPI¼[1(response
prepulse þ startle stimulus
/res-
ponse
startle stimulus alone
)] 100). PPI differences were
initially analyzed with one-way analysis of variance
(ANOVA). Subsequent analysis of covariance
(ANCOVA) was used to control for (nonsignificant)
group differences in baseline startle and
Sensorimotor gating and fragile X syndrome
PW Frankland et al
418
Molecular Psychiatry
chronological age. Pearson r correlations were
calculated to examine the relation between PPI and
other cognitive and behavioral measures in both
control and FXS subjects. In addition, PPI was
calculated using two additional estimates of
response amplitude: the highest point of a 2 ms
running average and log transformation of the peak
EMG amplitude.
16
PPI calculations using these
measures yielded equivalent results.
Mouse studies
Mice
Male Fmr1 (B6.129-Fmr1
tm1Cgr
) mice were
obtained from Jackson Laboratories (Bar Harbor, ME,
USA). These mice had been backcrossed five
generations into the C57B6/J (B6) background. We
first bred these mice with normal B6 females, to
obtain female mice heterozygous for the Fmr1
mutation (Fmr1
þ /
). Second, we crossed these
Fmr1
þ /
mice with normal B6 males. Only the male
offspring (hemizygous Fmr1-KO and male WT
littermate controls) were used in experiments. Mice
were group housed (2–4 mice per cage), and had
continuous access to food and water. The colony was
maintained on a 12 : 12 h light : dark cycle, and all
testing was carried out during the light phase of the
cycle. At the commencement of testing, mice were
between 8 and 12 weeks of age. All animal care and
testing procedures were approved by the Animal
Research Committee at UCLA and were in accordance
with the NIH Principles of Laboratory Animal Care.
Acoustic startle experiments
Equipment
Startle testing was conducted in a MED-
ASR-310 startle testing system (MedAssociates, VT,
USA). Mice were placed in a Plexiglas cylinder
(3.2 cm internal diameter) for testing. Acoustic
startle stimuli and prepulse stimuli were delivered
via a high-frequency speaker, placed at a distance of
15 cm from the testing cylinder. Background white
noise was generated by a standard speaker. The
testing cylinder was mounted on a sensor platform.
A piezoelectric accelerometer, attached to the base of
the sensor platform, detected and transduced all cage
movements, and these were digitized and stored by a
computer. The startle amplitude was taken to be the
maximal response occurring up to 100 ms following
presentation of the startle stimulus. The sound levels
for background noise and startle/prepulse stimuli
were calibrated with a digital sound level meter. The
speakers, testing cylinder and sensor platform were
housed within a sound-attenuated chamber.
Prepulse inhibition (experiment 1) Prepulse
inhibition was initially tested using identical
stimulus parameters to the human PPI studies.
Following a 5-min acclimation period where no
stimuli were delivered, WT (n¼11) and Fmr1-KO
(n¼10) mice were presented with a series of six
adaptation trials. These consisted of startle stimuli
(50 ms duration white-noise bursts, with 0 ms rise/fall
time) of increasing intensity (80–105 dB). Next, mice
were presented with a series of 32 trials: on half of
these trials, the startle stimulus only was presented
(105 dB, 50 ms duration, white-noise burst, with 0 ms
rise/fall time); on the remaining trials, a prepulse
(75 dB, 25 ms duration, 1000 Hz tone, with a 4 ms rise/
fall times) preceded the startle stimulus by 120 ms
(onset to onset). Background noise levels were
maintained at 50 dB throughout testing, and trials
were spaced 60 s apart.
Prepulse inhibition (experiment 2) PPI was also
examined in a second group of mice using an
alternative set of procedures. WT (n¼15) and Fmr1-
KO (n¼14) mice were initially given a habituation
session to acclimate them to the testing environment.
In this session, mice were presented with 80 startle
stimuli, delivered at a fixed intertrial interval of 15 s.
The startle stimulus was a 40 ms, 120 dB noise burst
with a rise/fall time of less than 1 ms. Background
noise levels were maintained at 65 dB.
One day following this, prepulse inhibition was
tested. Following an acclimation period of 5 min,
mice were presented with a total of 20 noise bursts
(40 ms duration, 120 dB, o1 ms rise/fall time). In the
prepulse inhibition phase, mice were presented with
a total of 90 trials. Three prepulse intensities were
tested: 70, 75 and 80 dB. Prepulses were 20 ms in
duration with a rise/fall time of less than 1 ms. For
each prepulse intensity, there were three types of
trial: prepulse alone, prepulse/startle stimulus and
startle stimulus alone. In the prepulse/startle stimu-
lus trial, the onset of the prepulse preceded the onset
of the startle stimulus by 100 ms. Background noise
levels were maintained at 68 dB throughout testing,
and the trials were spaced 15 s apart.
Approximately 1 week later, mice were given a
startle threshold test session. Following an acclima-
tion period of 5 min, mice were presented with a total
of 99 trials at a fixed intertrial interval of 15 s. There
were 11 trial types: no stimulus (NS), and 10 types of
trials where startle stimuli at a range of intensities
were presented (75–120 dB; 5 dB increments). The
startle stimuli were 40 ms noise bursts with a rise/fall
time of less than 1 ms. The 11 trial types were
presented in a pseudorandom order such that each
trial type was presented once within a block of 11
trials. Background noise levels were maintained at
65 dB throughout the test session. Startle threshold
was defined as the minimal intensity at which
responding was significantly greater than in the NS
trials.
Instrumental conditioning
Training
Apparatus and general methods have been
previously described.
17
.WT(n¼7) and Fmr1-KO
(n¼8) mice were first given two magazine training
sessions in which food pellets were freely delivered
on a random time 60-s schedule into a recessed
Sensorimotor gating and fragile X syndrome
PW Frankland et al
419
Molecular Psychiatry
magazine with the levers retracted. Mice were then
trained to press a lever positioned to the left of the
food magazine with the pellet reward delivered such
that a 20-s period had to time out before the next lever
press was rewarded (ie a fixed time 20-s schedule).
This training was continued until 100 rewards were
earned. At no time was lever pressing explicitly
shaped by the experimenter.
Devaluation After the initial acquisition phase
(above), the same groups of mice were given two 30-
min sessions of training on the lever with the reward
delivered on a random five schedule. During this
period, one WT and one Fmr1-KO mouse stopped
lever pressing and were, therefore, not included in the
devaluation study. Following this training, the WT
(n¼6) and Fmr1-KO (n¼7) mice were given two brief,
10-min extinction tests. Prior to one test, they were
allowed to eat the food pellets for 1 h prior to a test on
the lever. Prior to the other test they were allowed to
drink 10% sucrose for 1 h. The two tests were
conducted in a counterbalanced order with three
WT and four Fmr1-KO mice being tested after pellet
prefeeding first and after sucrose prefeeding second
whereas the remainder received the opposite test
order.
Omission The procedures for omission training
closely followed those reported previously.
18
After
training in which responding on each of two levers,
one each side of the central magazine, delivered a
pellet reward, WT (n¼9) and Fmr1-KO (n¼11) mice
were shifted to a situation in which pellets continued
to be delivered by lever pressing, but, in addition,
0.01 ml of 10% sucrose was also delivered freely on a
fixed time 20-s schedule. Responding on one lever,
the omission lever, delayed the delivery of the sucrose
by resetting the 20-s timer with each response,
whereas responding on the other lever had no effect
on sucrose delivery. For five WT and six Fmr1-KO
mice the left lever was the omission lever, whereas for
the remainder it was the right lever. Omission training
was conducted in one, 30 min session each day for 4
days.
Data analysis Percent PPI was calculated for
each mouse according to the following
formula: %PPI¼[1(response
prepulse þ startle stimulus
/res-
ponse
startle stimulus alone
)] 100). ANOVAs were used to
analyze phenotypic differences in PPI and
instrumental learning. For the PPI studies, an
ANCOVA was used to control for group differences
in baseline responding. Where appropriate Newman–
Keuls post hoc comparisons were used.
Results
Prepulse inhibition is disrupted in males with FXS
PPI was examined in young males (aged 8–17 years
old) diagnosed with the full mutation. Control
subjects were matched for chronological age, puberty
status, ethnicity and family demographic variables
(number of siblings, parental age, income and educa-
tion). Stimuli were presented to subjects via head-
phones, and EMG responses were recorded from the
lower eyelid (orbicularis oculi muscle). Two types of
trials were presented: those in which only the startle
stimulus (a 105 dB, 50 ms white noise burst) was
presented (startle only trials) and those in which a
lower intensity acoustic prepulse (a 75 dB, 25 ms,
1000 Hz tone) preceded the startle stimulus by 120 ms
(prepulse/startle trials). Startle responding was sub-
stantially reduced in the prepulse/startle trials in
control subjects (43.8710.0% PPI). In contrast, there
was almost no PPI in FXS subjects (1.678.5%)
(F(1,15)¼10.3, Po0.01) (Figure 1a and b). Audio-
metric screening confirmed that FXS subjects could
hear the prepulse at intensities much lower than the
75 dB used during testing. These data indicate that
sensorimotor gating was almost completely abolished
in FXS subjects.
Examination of the startle alone trials showed that
there was a nonsignificant trend for increased base-
line responding in FXS subjects (F(1,15)¼1.31,
P¼0.27) (Figure 1c). In addition, FXS subjects were
marginally (nonsignificantly) older than the control
subjects (Table 1). To eliminate these possible con-
founds from our analysis, we conducted a second
analysis including these factors as covariates (ANCO-
VA). Importantly, PPI deficits in FXS subjects re-
mained significant after accounting for differences in
baseline startle (F(1,14)¼8.6, Po0.05) and chronolo-
gical age (F(1,14)¼12.7, Po0.01). It should be noted
that while PPI appears to be abolished in FXS
subjects, it is possible that the temporal character-
istics of the PPI are altered in FXS subjects. Several
FXS subjects exhibited prepulse facilitation, rather
than inhibition. In normal subjects, facilitation occurs
at shorter (o30 ms) and longer (41400 ms) prepulse–
startle stimulus intervals.
11
Future studies are
required to evaluate this possibility.
Eight out of the 10 FXS subjects were on medica-
tions (or combinations of medications) at the time of
testing (Table 2). Within each class of drug, PPI levels
were distributed throughout the range of levels for our
FXS sample (Figure 1d), suggesting that PPI deficits
are independent of possible drug effects. While
ethical considerations have limited the study of
neuropsychiatric medications on PPI in humans, the
data that are available suggest that drug effects are not
responsible for the PPI deficits in our sample of FXS
patients. Five FXS subjects were taking stimulants (or
dopamine agonists). Previous studies have shown
that stimulants have mixed effects, either reducing
(dextroamphetamine,
19
bromocriptine
20
) or having no
effect (amphetamine,
20
bromocriptine
21
) on PPI. Four
FXS subjects were taking anxiolytics. Anxiolytics
have been shown to have no effect (diazepam
22
)orto
reduce PPI (midazolam
23
). The latter effect, however,
was associated with sedation and severe reduction in
baseline startle responding. Neither alpha-adrenergic
drugs, such as clonidine,
22
or atypical antipsychotic
Sensorimotor gating and fragile X syndrome
PW Frankland et al
420
Molecular Psychiatry
drugs, such as risperidone
24
and clozapine,
25
affect
PPI in normal adults. Finally, anticonvulants, includ-
ing carbamazine, have been found to increase, rather
than reduce, PPI in normal subjects.
26
At very least,
these data suggest it is unlikely that medications can
account for the PPI deficits in the FXS patients in this
study.
Deficits in PPI predict severity of cognitive and
behavioral pathology in FXS
FXS is associated with mental impairment (ranging
from learning disabilities to mental retardation),
anxiety and emotional problems, autistic-like beha-
viors, attention deficits and hyperactivity.
2
To char-
acterize the range of cognitive and behavioral deficits
in our population of young males with FXS, we
measured IQ (K-BIT test
15
), and administered struc-
tured interviews and questionnaires to primary
caregivers. These were used to evaluate the severity
of deficits in attention (Child Behavioral Checklist
13
)
and adaptive behavior (Vineland Adaptive Beha-
vior
12
), as well as the severity of the autistic
phenotype (Autism Screening Questionnaire
14
). Com-
pared to control children, young males with FXS had
lower IQ, and deficits in attention and adaptive
behavior. Furthermore, FXS subjects scored high on
the autistic scale, with 40% exceeding the diagnostic
cutoff for autism (Table 1). The severity of these
cognitive and behavioral deficits is typical for young
males with FXS,
2
indicating that our sample is
representative of the FXS population as a whole.
PPI varied within our FXS population (50.8 to
þ 35.6% PPI) (Figure 1b). Therefore, we next asked
whether PPI magnitude was related to the severity of
the cognitive and behavioral symptoms. Remarkably,
PPI magnitude was highly predictive of the severity of
Figure 1 PPI is disrupted in young males with FXS.
(a) Mean PPI is shown for control (closed bar; n¼7) and FXS
(open bar; n¼10) subjects. PPI was significantly reduced in
FXS subjects. (b) Scatterplot showing range of PPI scores in
control (closed circles) and FXS (open circles) subjects.
(c) Mean startle responding on startle only trials for control
(closed bar) and FXS (open bar) subjects. There was a
nonsignificant trend for higher baseline responding in FXS
subjects. (d) Relationship between medication and PPI scores
in FXS subjects. For each class of medication, PPI scores were
evenly distributed throughout the range of PPI scores for FXS
subjects. The horizontal lines correspond to the means for each
drug class. (e) Scatterplot of IQ vs PPI magnitude for FXS
subjects. Lower IQ scores were associated with reduced levels
of PPI (Po0.05). (f) Scatterplot of adaptive behavior scores vs
PPI magnitude for FXS subjects. Maladaptive behavior was
associated with reduced PPI (Po0.05). (g) Scatterplot of autism
scores vs PPI magnitude for FXS subjects. Subjects scoring
higher on the autism scale were likelier to have reduced PPI
levels (Po0.05). (h) Scatterplot of attention scores vs PPI
magnitude for FXS subjects. Attention deficits were associated
with reduced PPI (Po0.05). Data are means (7SEMs).
Table 1 Chronological age of control (n¼7) and FXS (n¼10)
subjects, and mean scores for each of the four core cognitive
and behavioral measures
Control Fragile X
Age (years) 11.4 (71.2) 13.2 (70.7)
IQ 119.9 (73.0) 50.2 (73.8)
Adaptive behavior 100.9 (73.4) 33.5 (72.9)
Autism N/A 19.7 (71.7)
Attention 27 (1.2) 12.9 (71.6)
IQ was measured using the Kaufman Brief Intelligence Test
(K-BIT). Adaptive behavior was assessed using the Vineland
Adaptive Behavior Scale (VABS). In this scale, lower scores
reflect less adaptive behavior. Attention deficits were
characterized using the attention domain of the Child
Behavioral Checklist (CBCL). Higher scores reflect a higher
incidence of attention problems. The prevalence of autistic-
like behaviors was measured using the Autism Screening
Questionnaire (ASQ). This questionnaire was only admi-
nistered to parents in FXS subjects. Four (out of 10) of the
FXS subjects had scores exceeding the diagnostic cutoff for
autism of 22. Data are means (7SEMs).
Sensorimotor gating and fragile X syndrome
PW Frankland et al
421
Molecular Psychiatry
the cognitive/behavioral phenotype in FXS subjects:
each of the four core measures (IQ, attention, adaptive
behavior and autism) was highly correlated with PPI
(Figure 1e–h). This relationship was only true for FXS
subjects. In control subjects, PPI magnitude (range 0.1
to þ 79.9%) was unrelated to IQ, attention and
adaptive behavior scores (data not shown).
Fmr1-KO mice exhibit enhanced PPI
Because of the limitations of studying the implica-
tions of these sensorimotor gating deficits in FXS
patients, we examined PPI in Fmr1-KO mice. PPI is
especially suited for translational studies since
equivalent procedures and stimuli can be used in
human and animal subjects. Furthermore, the brain
circuitry and neurochemistry underlying PPI have
been extensively characterized in rodents,
27
and there
is a large degree of homology between measures of PPI
in rodents and humans.
28
In contrast to FXS subjects,
Fmr1-KO mice showed enhanced PPI using identical
stimulus parameters to those in the human studies
(F(1,19)¼4.7, Po0.05) (Figure 2a). To confirm the
enhanced PPI of the mutant mice, we also tested PPI
with three different prepulse intensities (70, 75 and
80 dB). Consistent with previous reports,
9,29
PPI was
increased in the Fmr1-KO mice at each of these three
prepulse intensities (F(1,27)¼40.8, Po0.001) (Figure
2b). While the threshold for startle was similar in
Fmr1-KO and WT mice, at higher intensities the
startle response magnitude was reduced in Fmr1-KO
mice (Stimulus Intensity Genotype interaction:
F(9,243)¼15.5, Po0.001) (Figure 2c), as previously
observed when the Fmr1 mutation is maintained in a
predominantly C57B6/J background.
9
Most impor-
tantly, an ANCOVA indicated that increased PPI in
Fmr1-KO mice was not related to group differences in
baseline startle (F(1,26)¼25.9, Po0.001). Although
moderated by genetic background, increased PPI and
reduced baseline startle has been observed consis-
tently in Fmr1-KO mice.
9,29
These results demonstrate
that mutations of the FMR1 gene have profound
effects on sensorimotor gating in both humans and
mice. However, whereas PPI is reduced (or even
absent) in young males with FXS, it is increased in
Fmr1-KO mice.
Learning is facilitated in Fmr1-KO mice in two tests
of instrumental conditioning
In addition to sensorimotor gating problems, FXS is
associated with learning impairments. However, pre-
vious studies have shown that learning may be
normal,
8,30,31
mildly impaired,
7,30–33
or even im-
proved
34
in Fmr1-KO mice. However, there are many
different types of learning,
35
and the Fmr1-KO
mutation may affect some forms more than others.
To explore this possibility, we tested the Fmr1-KO
mice in instrumental conditioning. This form of
learning has not been extensively explored in these
mice, and is especially suited for teasing apart
components of complex behaviors.
36
Mice were
initially trained to press a lever for food pellet reward.
The rate of acquisition of lever pressing was similar in
WT and Fmr1-KO mice (F(1,13)o1, P40.05), indicat-
ing that basic motor and motivational processes
necessary for instrumental learning are normal in
these mice (Figure 2d).
In contrast, Fmr1-KO mice showed enhanced
performance in outcome devaluation, which is com-
monly used to assess how well an animal learns the
relationship between lever pressing (the action) and
the specific food reward (or outcome). During acqui-
sition (above), mice learned to press a lever for a food
pellet reward. At the end of this training, both WT
(48.776.6 presses/min) and Fmr1-KO (51.377.2
presses/min) mice responded at similar rates. On
the test day, mice were first given free access to either
food pellets or a sucrose solution until sated.
Immediately following this, mice were placed back
in the instrumental chamber and allowed to press the
Table 2 Summary of medications taken by FXS subjects at the time of testing
Subject Percent
PPI
No
medications
Stimulants Anxiolytics Alpha-adrenergics Anticonvulsants Antipsychotics
1 35.6 Carbamazepine
2 31.5 Dextroamphetamine Venlafaxine Guanfacine
3 20.2 Adderall Clonazepam
4 11.1 None
5 9.7 None
6 7.5 Methylphenidate
7 9.6 Risperidone
8 11.7 Adderall Lorazepam Clonidine Divalproex
9 27.4 Adderall Clonazepam Lamotrigine
10 50.8 Guanfacine
The number of medications per subject ranged from 0 to 4 and included stimulants, anxiolytics, alpha-adrenergics,
anticonvulants and antipsychotics. All medications are generic names except Adderall, which is the brand name for a
mixture of several dextroamphetamine and amphetamine salts. Venflaxine is an antidepressant frequently used to treat
anxiety as in the case of subject 2. Risperidone was used to treat severe irritability and aggressive behavior in subject 7.
Sensorimotor gating and fragile X syndrome
PW Frankland et al
422
Molecular Psychiatry
lever. Lever pressing in this extinction session was
not reinforced. Prefeeding mice food pellets, but not
sucrose, resulted in reduced lever pressing during the
extinction test session in both WT (F(1,11)¼15.0,
Po0.01) and Fmr1-KO (F(1,11)¼53.1, Po0.001) mice.
As reduced responding is specific to the outcome that
has been devalued, this indicates that mice can
encode the relationship between lever pressing and
a specific food reward, and not just reward in general.
However, Fmr1-KO mice were more sensitive to this
devaluation procedure, reducing responding (after
being prefed pellets) to a greater degree compared to
WT littermate controls (Lever Genotype interaction:
F(1,11)¼7.4, Po0.05) (Figure 2e). Post hoc compar-
isons confirmed that the difference in responding for
valued vs devalued outcome was greater in Fmr1-KO
than WT mice (Po0.05).
Fmr1-KO mice also showed enhanced performance
in a complex omission procedure.
18
This task has two
components: first, mice were trained to press two
levers for food pellet reward. Both WT and Fmr1-KO
mice showed similar rates of lever pressing at the end
of acquisition (data not shown). After earning 100
food pellet rewards on both levers, a new contingency
was introduced: although both levers continued to
deliver food pellets, pressing on one of the two levers
(the omission lever) delayed delivery of an additional
(freely delivered) sucrose reward. Therefore, in order
to gain access to the sucrose, mice must learn to
withhold responses on the omission lever. Both WT
(F(1,18)¼7.2, Po0.01) and Fmr1-KO (F(1,18)¼19.6,
Po0.001) mice reduced responding on the omission
lever. However, Fmr1-KO mice suppressed respond-
ing on the omission lever to a greater degree
compared to their WT littermate controls (Le-
ver Genotype interaction: F(1,18)¼6.6, Po0.01)
(Figure 2f); that is, they demonstrated greater ability
to selectively withhold pressing the lever that delayed
access to the highly valued sucrose reward. Post hoc
comparisons confirmed that the difference in re-
sponding on omission vs noncontingent levers was
greater in Fmr1-KO than WT mice (Po0.05).
In both the devaluation and omission procedures,
enhanced performance in the Fmr1-KO was charac-
terized by more selective responding. Since general
changes in motivation (eg, frustration) would be
expected to have nonspecific effects on instrumental
performance, this suggests that the Fmr1-KO were
Figure 2 Summary of behavioral experiments in Fmr1-KO
mice. (a) PPI was enhanced in Fmr1-KO (n¼10) compared
to WT (n¼11) mice. In this experiment, identical stimulus
parameters to the human studies were used. (b) In a second
set of mice, PPI was examined using prepulses at three
different intensities (70–80 dB). PPI was enhanced in
Fmr1-KO mice (open bars; n¼14) compared to WT controls
(closed bars; n¼15) at each of the prepulse intensities
tested. (c) Acoustic startle threshold curve tested in the
same WT (closed circles) and Fmr1-KO (open circles) mice.
Mean response amplitudes (7SEM) are plotted for trials
where no startle stimulus was presented (NS) and for trials
where startle stimuli (75–120 dB) were presented. While
the threshold for startle was similar in Fmr1-KO and WT
mice, at higher intensities the magnitude of startle
response was reduced in Fmr1-KO mice. (d) Acquisition
of lever pressing for a food pellet reward is similar in WT
(closed circles) and Fmr1-KO (open circles) mice. Perfor-
mance is presented as rate of lever pressing per pellet
reward. (e) In the outcome devaluation test, mice were
either sated on sucrose (valued group; closed bars) or food
pellets (devalued group; open bars). Immediately following
this, they were tested on the lever in a nonreinforced
extinction session. Both WT and Fmr1-KO mice reduced
responding when pellets were devalued relative to the
sucrose. However, Fmr1-KO mice were more sensitive to
the devaluation treatment compared to WT controls. (f) In
the omission test, responding on the omission lever (open
bars), but not the other, noncontingent, lever (closed bars),
delays the delivery of an additional sucrose reward. Both
WT and Fmr1-KO mice reduced responding on the
omission lever. However, Fmr1-KO mice suppressed
responding on the omission lever to a greater degree
compared to WT controls. Data are means (7SEMs).
Sensorimotor gating and fragile X syndrome
PW Frankland et al
423
Molecular Psychiatry
able to learn action–outcome associations more
efficiently than WT controls.
Discussion
Our PPI studies indicated that individuals with FXS
have dramatic impairments in sensorimotor gating.
The severity of these PPI deficits is greater than those
typically seen in other disorders, including schizo-
phrenia.
28
Most importantly, deficits in this simple
and well-understood phenomenon were highly pre-
dictive of cognitive and behavioral pathology in FXS,
suggesting that disturbances in sensorimotor gating
may reflect a disturbance in a core mechanism that is
responsible for normal cognitive and behavioral
functioning. Therefore, exploration of these PPI
deficits may be especially useful in understanding
disease mechanism, and PPI may represent a straight-
forward and quantitative behavioral measure that can
be used to track therapeutic progress of experimental
treatments in the future.
This study is the first to examine the impact of a
single gene mutation on PPI in humans and mice side
by side. PPI is especially suited for these sorts of
translational studies since identical stimuli and
similar procedures may be used in humans and mice.
However, whereas PPI was severely disrupted in
individuals with FXS, it was enhanced in the Fmr1-
KO mice. While these data support a role for FMRP in
sensorimotor processing in both humans and mice,
the lack of correspondence between mouse and
human phenotypes has important implications for
the use and interpretation of the Fmr1-KO mouse
model. Indeed, whereas patients showed severe
cognitive deficits, we found enhanced performance
in two different learning tasks in Fmr1-KO mice.
While there is relative sparing of some aspects of
cognitive function in FXS, no skills are improved
compared to control subjects.
2
Similarly, whereas
anxiety problems are prevalent in FXS patients,
Fmr1-KO mice exhibit reduced anxiety.
8,32
Higher
cognitive function is influenced by complex interac-
tions between multiple genes. Therefore, in mouse
models of inherited cognitive disorders, even subtle
differences in compensatory mechanisms between
humans and mice may amplify, blunt or alter the
disease phenotype.
37
For example, FMR1 and the
fragile X-related genes, FXR1 and FXR2, encode a
family of functionally similar RNA-binding pro-
teins.
38
Although FXR1 and FXR2 expression levels
are not altered in FXS patients or Fmr1-KO mice,
39,40
greater redundancy among these proteins in mice may
attenuate the impact of the FMR1 mutation. In this
regard, it is of particular interest that PPI is disrupted
in Fxr2-KO mice, and learning is more severely
affected in these mice compared to Fmr1-KO mice.
41
Alternatively, in FXS FMRP can be normally ex-
pressed during early embryonic development (before
hypermethylation leads to transcriptional silencing of
FMR1), whereas in Fmr1-KO mice the protein is
absent throughout development. Therefore, it is
possible that transient exposure to FMRP during
development may have a more devastating impact
on later cognitive function.
Acknowledgements
These studies were supported by FRAXA Research
Foundation grants to PWF and AJS, and to EMD and
EMO, and by support from the Virginia Friedhofer
Charitable Trust to EMO. We thank Sheena Josselyn
and Anna Matynia for comments on earlier versions
of this manuscript.
References
1 Turner G, Webb T, Wake S, Robinson H. Prevalence of fragi-
le syndrome. Am J Med Genet 1996; 64: 196–197.
2 Verkerk AJ et al. Identification of a gene (FMR-1) containing a CGG
repeat coincident with a breakpoint cluster region exhibiting
length variation in fragile X syndrome. Cell 1991; 65: 905–914.
3 Hagerman RJ, Hagerman PJ. Fragile X Syndrome: Diagnosis,
Treatmen, and Research, Vol xii Johns Hopkins University Press:
Baltimore, 2002 540pp.
4 Kooy RF, Willemsen R, Oostra BA. Fragile X syndrome at the turn
of the century. Mol Med Today 2000; 6: 193–198.
5 O’Donnell WT, Warren ST. A decade of molecular studies of fragile
X syndrome. Annu Rev Neurosci 2002; 25: 315–338.
6 Greenough WT et al. Synaptic regulation of protein synthesis and
the fragile X protein. Proc Natl Acad Sci (USA) 2001; 98: 7101–
7106.
7 Bakker CE et al. Fmr1 knockout mice—a model to study fragile X
mental retardation. Cell 1994; 78: 23–33.
8 Peier AM et al. (Over)correction of FMR1 deficiency with YAC
transgenics: behavioral and physical features. Hum Mol Genet
2000; 9: 1145–1159.
9 Nielsen DM, Derber WJ, McClellan DA, Crnic LS. Alterations in
the auditory startle response in Fmr1 targeted mutant mouse
models of fragile X syndrome. Brain Res 2002; 927: 8–17.
10 Miller LJ et al. Electrodermal responses to sensory stimuli in
individuals with fragile X syndrome: a preliminary report. Am J
Med Genet 1999; 83: 268–279.
11 Geyer MA, McIlwain KL, Paylor R. Mouse genetic models
for prepulse inhibition: an early review. Mol Psychiatry 2002; 7:
1039–1053.
12 Sparrow SS, Balla D, Cicchetti DV. Vineland Adaptive Behavior
Scales. American Guidance Service: Circle Pines, MN, 1984.
13 Achenbach TM. Manual of the Child Behavior Checklist.
University of Vermont: Burlington, VT, 1991.
14 Berument SK, Rutter M, Lord C, Pickles A, Bailey A. Autism
screening questionnaire: diagnostic validity. Br J Psychiatry 1999;
175: 444–451.
15 Kaufman AS, Kaufman NL. Kaufman Brief Intelligence Test.
American Guidance Service: Circle Pines, MN, 1990.
16 Ornitz EM et al. Prepulse inhibition of startle and the neurobiol-
ogy of primary nocturnal enuresis. Biol Psychiatry 1999; 45: 1455–
1466.
17 Carvalho OM, Silva AJ, Balleine BW. Evidence of selective
learning deficits on tests of Pavlovian and instrumental condition-
ing on alpha-CaMKII-T286 mutant mice. Int J Comp Psychol 2001;
14: 161–174.
18 Dickinson A, Squire S, Varga Z, Smith JW. Omission learning after
instrumental pretraining. Quart J Exp Psychol Comp Physiol
Psychol 1998; 51B: 271–286.
19 Hutchison KE, Swift R. Effect of
D-amphetamine on prepulse
inhibition of the startle reflex in humans. Psychopharmacology
(Berl) 1999; 143: 394–400.
20 Swerdlow NR et al. Dopamine agonist effects on startle and
sensorimotor gating in normal male subjects: time course studies.
Psychopharmacology (Berl) 2002; 161: 189–201.
Sensorimotor gating and fragile X syndrome
PW Frankland et al
424
Molecular Psychiatry
21 Abduljawad KA, Langley RW, Bradshaw CM, Szabadi E. Effects of
bromocriptine and haloperidol on prepulse inhibition of the
acoustic startle response in man. J Psychopharmacol 1998; 12:
239–245.
22 Abduljawad KA, Langley RW, Bradshaw CM, Szabadi E. Effects of
clonidine and diazepam on prepulse inhibition of the acoustic
startle response and the N1/P2 auditory evoked potential in man.
J Psychopharmacol 2001; 15: 237–242.
23 Schachinger H, Muller BU, Strobel W, Langewitz W, Ritz R.
Midazolam effects on prepulse inhibition of the acoustic blink
reflex. Br J Clin Pharmacol 1999; 47: 421–426.
24 Mackeprang T, Kristiansen KT, Glenthoj BY. Effects of antipsycho-
tics on prepulse inhibition of the startle response in drug-naive
schizophrenic patients. Biol Psychiatry 2002; 52: 863–873.
25 Graham SJ, Langley RW, Bradshaw CM, Szabadi E. Effects of
haloperidol and clozapine on prepulse inhibition of the acoustic
startle response and the N1/P2 auditory evoked potential in man.
J Psychopharmacol 2001; 15: 243–250.
26 Pouretemad HR, Thompson PJ, Fenwick PB. Sensorimotor gating
in patients with non-epileptic seizures. Epilepsy Res 1998; 31:
1–12.
27 Geyer MA, Krebs-Thomson K, Braff DL, Swerdlow NR. Pharma-
cological studies of prepulse inhibition models of sensorimotor
gating deficits in schizophrenia: a decade in review. Psychophar-
macology 2001; 156: 117–154.
28 Braff DL, Geyer MA, Swerdlow NR. Human studies of prepulse
inhibition of startle: normal subjects, patient groups, and pharma-
cological studies. Psychopharmacology (Berl) 2001; 156: 234–258.
29 Chen L, Toth M. Fragile X mice develop sensory hyperreactivity to
auditory stimuli. Neuroscience 2001; 103: 1043–1050.
30 Van Dam D et al. Spatial learning, contextual fear conditioning and
conditioned emotional response in Fmr1 knockout mice. Behav
Brain Res 2000; 117: 127–136.
31 Dobkin C et al. Fmr1 knockout mouse has a distinctive strain-
specific learning impairment. Neuroscience 2000; 100: 423–429.
32 Qin M, Kang J, Smith CB. Increased rates of cerebral glucose
metabolism in a mouse model of fragile X mental retardation. Proc
Natl Acad Sci (USA) 2002; 99: 15758–15763.
33 D’Hooge R et al. Mildly impaired water maze performance in male
Fmr1 knockout mice. Neuroscience 1997; 76: 367–376.
34 Fisch GS, Hao HK, Bakker C, Oostra BA. Learning and memory in
the FMR1 knockout mouse. Am J Med Genet 1999; 84: 277–282.
35 Squire LR, Knowlton B, Musen G. The structure and organization
of memory. Annu Rev Psychol 1993; 44: 453–495.
36 Balleine BW, Dickinson A. Goal-directed instrumental action:
contingency and incentive learning and their cortical substrates.
Neuropharmacology 1998; 37: 407–419.
37 Watase K, Zoghbi HY. Modelling brain diseases in mice: the
challenges of design and analysis. Nat Rev Genet 2003; 4: 296–307.
38 Jin P, Warren ST. New insights into fragile X syndrome: from
molecules to neurobehaviors. Trends Biochem Sci 2003; 28: 152–
158.
39 Feng Y et al. FMRP associates with polyribosomes as an mRNP,
and the I304N mutation of severe fragile X syndrome abolishes
this association. Mol Cell 1997; 1: 109–118.
40 Tamanini F et al. Differential expression of FMR1, FXR1 and FXR2
proteins in human brain and testis. Hum Mol Genet 1997; 6: 1315–
1322.
41 Bontekoe CJ et al. Knockout mouse model for Fxr2: a model for
mental retardation. Hum Mol Genet 2002; 11: 487–498.
Sensorimotor gating and fragile X syndrome
PW Frankland et al
425
Molecular Psychiatry
