Download Premature stop codons involved in muscular dystrophies

Gene Therapy (2004) 11, 619–627
& 2004 Nature Publishing Group All rights reserved 0969-7128/04 $25.00
www.nature.com/gt
RESEARCH ARTICLE
Premature stop codons involved in muscular
dystrophies show a broad spectrum of readthrough
efficiencies in response to gentamicin treatment
L Bidou1, I Hatin1, N Perez2, V Allamand3, J-J Panthier4 and J-P Rousset1
1
CNRS UMR 8621, Institut de Génétique et Microbiologie, Université Paris-Sud, Orsay Cedex, France; 2GENETHON, CNRS UMR
8115, Evry, France; 3INSERM U582-Institut de Myologie, IFR14, Groupe Hospitalier Pitié-Salpétrière, Paris Cedex, France; and
4
UMR955 INRA/ENVA de Génétique Moléculaire et Cellulaire, Ecole Nationale Vétérinaire d’Alfort, 7, Maisons-Alfort Cedex, France
The suppression levels induced by gentamicin on premature
stop codons, caused by primary nonsense mutations found
in muscular dystrophy patients, were assessed using a very
sensitive dual reporter gene assay. Results show that: (i) the
effect of gentamicin on readthrough is similar in cultured cells
and in vivo in murine skeletal muscle; (ii) a wide variability of
readthrough efficiency is obtained, depending on the mutation tested; (iii) due to the complexity of readthrough
regulation, efficiency cannot be predicted by the nucleotide
context of the stop codon; (iv) only a minority of premature
stop codons found in patients show a significant level of
readthrough, and would thus be amenable to this
pharmacological treatment, given our present understanding
of the problem. These results probably provide an
explanation for the relative failure of clinical trials
reported to date using gentamicin to treat diseases due
to premature stop codons, and emphasize that
preliminary assays in cell culture provide valuable information concerning the potential efficiency of pharmacological
treatments.
Gene Therapy (2004) 11, 619–627. doi:10.1038/sj.gt.3302211
Published online 19 February 2004
Keywords: muscular dystrophy; premature stop codons; readthrough; gentamicin; dual reporter gene; skeletal muscle
Introduction
While gene therapy is potentially a method of choice in
the treatment of genetic diseases, alternative therapeutic
methods independent of the transfer of genetic material
offer an attractive option. In particular, pharmacological
approaches aimed at modification of gene expression,
rather than the gene itself, constitute an ideal strategy for
a subset of genetic diseases in which the molecular defect
is known and can be circumvented. This is the case for
premature stop codon (PTC) diseases, in which the
primary cause of the disease is a nonsense mutation in
the gene coding sequence: about 5–10% of patients for
the major genetic diseases, though the frequency may be
higher for some syndromes or in specific human
populations. Aminoglycosides have long been known
to affect translational accuracy. In particular, they
promote nonsense mutation suppression in bacteria
and yeast.1 These antibiotics interact with the highly
conserved decoding center of ribosomal RNA, and allow
normal tRNAs to recognize incorrect codons, including
stop codons.2 In 1985, Burke and Mogg3 initially
suggested that some aminoglycosides presented the
same properties in cultured mammalian cells. They
demonstrated that paromomycin and G-418 could
partially restore the synthesis of an entire protein from
a mutant gene interrupted by a premature stop codon.
Correspondence: J-P Rousset, CNRS UMR 8621, Institut de Génétique et
Microbiologie, Université Paris-Sud, Orsay Cedex 91405, France
Received 30 June 2003; accepted 10 November 2003
Bedwell and co-workers4,5 have also shown that G-418
and gentamicin can restore the expression of the cystic
fibrosis transmembrane conductance regulator (CFTR) in
a bronchial cell line with a nonsense mutation in the
CFTR gene. Similarly, a Hurler syndrome fibroblast cell
line heterozygous for a stop mutation showed a
significant increase in alpha-L-iduronidase when cultured in the presence of gentamicin.4,5 The strength of
this approach was first demonstrated in vivo by the work
of Barton-Davis and co-workers, who demonstrated that
gentamicin could partially restore expression of fulllength dystrophin in mdx mice with the X-linked
muscular dystrophy mutation (Mdmmdx), a C–T transition
at position 3185, resulting in a termination codon
replacing a glutamine codon in the dystrophin gene.6
Negamycin, a dipeptide antibiotic with similar stop
codon-misreading activity, has also been shown to
restore dystrophin synthesis in the mdx mouse skeletal
muscle.7 More recently, Bedwell and co-workers demonstrated that gentamicin is effective in a transgenic mouse
model of cystic fibrosis.8 Although controversial results
were obtained with mdx mice,9 a very recent report
showed a high level of dystrophin reexpression in a
DMD patient treated with gentamicin.10
Overall, these observations suggest that antibiotic
treatment is a promising alternative to gene therapy for
patients with premature stop codons due to nonsense
mutations. However, it is well known from experiments
conducted both in cultured mammalian cells and in yeast
that the stop codon context plays an important role in
readthrough efficiency. First, the three stop codons are
Gentamicin and muscular dystrophies
L Bidou et al
620
not equally susceptible to readthrough in basal conditions: UAA is, on average, less sensitive to basal
suppression than UAG and UGA.11 Secondly, several
studies demonstrated the critical influence of both
upstream and downstream sequences, which determine
the overall efficiency of translational termination in
yeast.12,13 Less is currently known on the influence of
the sequence context on basal and antibiotic-induced
readthrough levels in mammalian cells. Results from
different laboratories have demonstrated that the antibiotic effect is dependent on the nucleotides surrounding
the stop mutation; they also suggest that the nucleotide
immediately following the stop codon (numbered þ 4)
has a major role in readthrough.14,15
In this study, we analyzed the gentamicin susceptibility of individual primary stop mutations concerning
Duchenne muscular dystrophy (DMD) and congenital
muscular dystrophy (CMD) patients, using a very
sensitive dual reporter gene system.16 This experimental
model closely corresponds to the physiological situation,
in which a given mutation is embedded in a unique
nucleotide context, and differs between patients. Several
reasons indicated an investigation of gentamicin: its
effect on readthrough is the best documented to date,
and it is known to present a much lower toxicity than
other suppressive molecules like G418.3 Moreover, this
antibiotic is currently used in human therapy against
bacterial infections, and its pharmacokinetics properties
and side effects are well characterized.17 However, one
cannot exclude that a long-term use of gentamicin may
induce other types of side effects due to generalized
readthrough of normal termination codons.
DMD, the most common X-linked fatal genetic
disease, is a degenerative disorder of striated muscle
caused by reduction or absence of the protein dystrophin, and affects approximately 1 in 3500 newborn
males. The classical form of CMD is a severe and
heterogeneous neuromuscular disease caused by mutations in the LAMA2 gene encoding the a2 chain of
laminin.18–20 At least 10% of DMD patients and 20% of
CMD patients have a nonsense mutation resulting in
premature translation termination.18,21
Our results show that a very broad variation is
observed in both basal and gentamicin-induced readthrough efficiency in different mutations. They also
demonstrate that the nucleotide following the stop codon
is hardly the unique determinant of readthrough efficiency, and that the critical parameters might be different
in the absence or presence of the antibiotic. Additionally,
using electrotransfer of plasmid DNA into murine skeletal
muscle, we demonstrated that both the basal readthrough
level and the effect of gentamicin are comparable in
cultured cells and in vivo. Our results strongly suggest that
the precise quantification of readthrough levels, using
reporter systems such as those described here, is a
prerequisite in the identification of patients with mutations displaying a gentamicin response, justifying their
inclusion in potential clinical trials.
Results
Quantification of readthrough levels in NIH3T3 cells
A dual gene reporter system16 was used to quantify the
effect of gentamicin on stop mutation readthrough in
Gene Therapy
culture cells (Figure 1). Sequences to be analyzed,
spanning 27 nucleotides centered on the different stop
mutations (Table 1), were inserted in-frame between the
lacZ and luc coding sequences; every translating ribosome would thus result in b-galactosidase synthesis, but
only those reading through the stop codon would yield
active luciferase. Readthrough efficiency was estimated
by calculating the ratio of luciferase/b-galactosidase
activity. This dual gene reporter system provides an
internal control permitting calibration of individual
experiments for the overall expression level of each
construct (vector stability, transfection efficiency, transcriptional and translational rates). Both reporter proteins are expressed from the same message, eliminating
in particular potential variation in mRNA stability
between the different targets analyzed. This is especially
important in the comparison of constructs harboring
nonsense mutations, since transcripts containing a
premature stop codon have been shown to be preferentially degraded by the nonsense-mediated mRNA decay
pathway (NMD).22
To establish the relative b-galactosidase and luciferase
activities, expressed in equimolar amounts, we also
constructed control plasmids in which each stop mutation was replaced by the sense codon found in the wildtype gene.
In all, 18 DMD mutations and eight CMD mutations
were analyzed (Table 1). The molecular defect in the
Mdmmdx murine mutation, causing premature termination of the 427 kDa dystrophin polypeptide, was included since Mdmmdx/Y mice are used as a model for
human dystrophy.23 In addition, we tested the platinum
(Tyrcp) mutation, an A–T substitution, changing a lysine
residue at position 489 of tyrosinase to a termination
codon. This tyrosinase truncation results in misrouting of
the enzyme to the melanocyte periphery; thus, Tyrcp/
Tyrcp homozygous mice have a white coat color, like
albino homozygotes.24 Readthrough is expected to
restore normal pigmentation in platinum mice. Finally,
Figure 1 Description of the dual gene reporter system. For quantification
in NIH3T3 cells, ORF1 and ORF2 correspond to b-galactosidase and
firefly luciferase coding sequences, respectively, and their expression is
driven by the SV40 promoter (pAC99 vectors). For in vivo quantification,
ORF1 and ORF2 correspond to Renilla and Firefly luciferase, respectively,
and their expression is driven by the hCMV promoter (pCRFL vectors).
Oligonucleotides including the stop mutation to be analyzed, surrounded
by four codons in 50 and four codons in 30 , were cloned in-frame between
ORF1 and ORF2. Readthrough efficiency was determined by comparing
the firefly luciferase to b-galactosidase (or Renilla luciferase) ratio obtained
for each nonsense mutation, to the same ratio obtained with an in-frame
control corresponding to the same sequence in which the stop codon was
replaced by the amino acid naturally present in the protein.
Gentamicin and muscular dystrophies
L Bidou et al
621
Table 1 Basal and gentamicin-induced readthrough for DMD and CMD mutations
Readthrough atb gentamicin (mg/ml) ¼
Stop targeta
Wild-type
denomination amino acid Nucleotidic context of stop mutation
1593d
60d
988c
MDX
1240c
1143d
2726d
1437c
2125d
STOP PLATI
3149d
STOP DYS
651d
1417d
2522d
931d
2264d
673d
1326c
967c
3085c
744c
STOP LAM
3190d
1549c
1967d
3381d
2098d
145d
319d
E
Q
Q
Q
Q
Q
E
Q
Q
Q
W
L
Q
E
Q
Q
R
C
R
R
R
R
R
R
R
R
S
GCT ACA GAT ATG TAA TTG ACA AAG AGA
GGC CTG ACA GGG TAA AAA CTG CCA AAA
CAA TGT TGG TGC TAA CCT GGA GTC ACA
TTG AAA GAG CAA TAA AAT GGC TTC AAC
AAA CTT CCA GAA TAA TTT GAA GGA AAG
GAT CAC ATG TGC TAA CAG GTC TAT GCC
GAA AGG CTC CTA TAA GAC TCC AAG GGA
ACA TCG ATA TGC TAG AAT TGT CAA CAT
ACA GAA GCT GAA TAG TTT CTC AGA AAG
AAG AAG AAG AAG TAG AAG CAA CCC CAG
ATG GAT ATC CTG TAG ATT ATT AAT TGT
GAG GAC ACA ATG TAG GAA GTC TTT TCC
TTT GCC CGG TGT TAG GAT AAT TTA GTC
ATC CAA TCT GAT TAG ACA AGT CAT GAG
CAG GAT TTG GAA TAG AGG CGT CCC AGT
GCC AGA GAG AAA TAG CTA CAG ACA ATT
TTG CCC CTG CGC TAG GGA ATT CTC AAA
GTC ACC ACC ACT TAG CCA TCA CTA ACA
AGA ACT GTG ACC TGA GAA GAC TTC TTG
GGC TGT GTT CCC TGA AAC TGC AAT TCT
AGT ATT CCG TTC TGA GGT TGC ATC AGA
CCT AGG CAC AGG TGA GTT AAC GGC ACT
AGG CCC TGG AAC TGA GGG GCG TTC AAC
TAT GAT ACG GGA TGA ACA GGG AGG ATC
TTC TGC ACG TGC TGA CCT GGA GCC ACG
TTT GCT CAG TTT TGA AGA CTA AAC TTT
AAA AAC AAA TTT TGA ACC AAA AGG TAT
GAC CGA CAA GGG TGA TTT GAC AGA TCT
CTG AGC TGG GTC TGA CAA TCA ACT CGT
AGC CCA TTT CCT TGA CAG CAT TTG GAA
0
600
0.010%
0.006%
0.005%
0.006%
0.020%
0.006%
0.008%
0.010%
0.080%
0.040%
0.040%
0.030%
0.012%
0.020%
0.012%
0.370%
0.090%
0.080%
0.020%
0.010%
0.050%
0.040%
0.050%
0.110%
0.110%
0.080%
0.033%
0.730%
0.310%
0.060%
0.05%
0.05%
0.07%
0.08%
0.11%
0.15%
0.17%
0.05%
0.06%
0.10%
0.12%
0.13%
0.22%
0.30%
0.30%
0.92%
1.21%
1.90%
0.11%
0.14%
0.19%
0.20%
0.23%
0.46%
0.60%
0.70%
0.76%
1.40%
1.74%
2.65%
Increase factor between
doses 0 and 600
5
8
13
13
6
25
21
5
1
3
3
4
18
15
25
2
13
24
6
14
4
5
5
4
5
9
23
2
6
44
a
Mutations are primary nonsense mutations, and are named by the position of the modified amino acid in the protein sequence, followed by a
‘d’ or a ‘c’ indicating, respectively, DMD or CMD mutations. The stop mutation present, of mdx mouse, is indicated as ‘MDX’. ‘STOP LAM’
and ‘STOP DYS’ correspond to the natural termination codon of laminin and dystrophin genes. ‘STOP PLATI’ indicates the platinum coat
color mutation.
b
Readthrough efficiency has been measured as described in Figure 1. Each value of readthrough efficiency (in the absence or presence of
gentamicin) corresponds to the mean of four to six independent experiments. Standard deviation errors did not exceed 20%.
the natural stop codons of the dystrophin and laminin a2
chain (LAMA2) genes were also included as controls.
Transfected NIH3T3 cells were cultured for 2 days in the
absence or presence of 200, 300, 400, 500 and 600 mg/ml
of gentamicin. Preliminary experiments with nontransfected cells demonstrated that no signs of cytotoxicity
were detectable after 48 h of treatment with up to
600 mg/ml gentamicin. Indeed, following transfection,
surviving cells showed similar morphology and comparable growth rate in the presence or absence of gentamicin
(data not shown). Similar experiments were performed
with several constructs, using C2C12 myoblasts instead
of NIH3T3 cells as recipient, and identical results were
obtained either in the absence or presence of gentamicin
(data not shown).
Figure 2 shows dose–response curves obtained with
five nonsense mutations. For all of these, a linear
response was obtained for doses up to 600 mg/ml. In
addition, very different slopes were obtained for each
mutation. For 20 other targets, readthrough level was
similarly increased when higher concentrations of
gentamicin were used (data not shown). For this reason,
only dosages of 0 and 600 mg/ml were used in further
experiments. Results obtained at doses 0 and 600 are
shown in Table 1 and analyzed below. Even for targets
Figure 2 Gentamicin-induced readthrough on five representative nonsense
mutations 3381d (filled circle), 2522d (open circle), 1417d (triangle),
2726d (filled square) and 1143d (open square). Sequences surrounding the
stop codon are presented in Table 1. These sequences were cloned in the
pAC99 dual reporter vector. To obtain dose–response curves, NIH3T3 cells
were electroporated with the different constructs and grown for 2 days in
the absence or presence of various concentrations of gentamicin. After
protein extraction, luciferase and b-galactosidase activity were measured.
Termination readthrough is expressed as the luciferase/b-galactosidase
ratio of the test construct, normalized to the in-frame control.
Gene Therapy
Gentamicin and muscular dystrophies
L Bidou et al
622
that exhibited a low increase effect (o4), the difference
between basal and gentamicin-induced readthrough
level was statistically significant (P-value ranging from
0.00197 to 0.00451). The variation between readthrough
levels detected in the absence or presence of gentamicin
was reflected in the level of increase observed for each
mutation (Table 1). A wide variability in responsiveness
to gentamicin was observed between the different
mutations, but was not dependent on the identity of
the stop codon. Increase factors between readthrough
levels at doses 0 and 600 ranged from 1 (target 2125d) to
44 (target 319d).
Figure 3 shows the readthrough levels in the absence
or presence of gentamicin, for all mutations tested. For
basal readthrough, values ranged from 0.005 to 0.73%,
the vast majority of the mutations yielding less than 0.1%
of readthrough. Previous studies reported that the UAA
stop codon was a better terminator in comparison with
UAG and UGA.15 We confirmed this observation; on the
wide number of mutations studied here, the average
basal readthrough was 0.006% for UAA codons, compared to 0.041% for UAG and 0.055% for UGA. These
differences were highly significant (P ¼ 0.00089 when
Figure 3 Comparison between basal and gentamicin-induced readthrough
level. The distribution of readthrough levels driven by different stop codons
in the absence of gentamicin (filled bars) or in the presence of 600 mg/ml of
gentamicin (open bars). Experiments were performed as indicated in
Figure 2. Each value corresponds to the mean of four to six independent
experiments. Standard deviation errors did not exceed 20%.
Gene Therapy
comparing UAA versus UAG in a Mann–Whitney
nonparametric test and P ¼ 0.00036 when comparing
UAA versus UGA). By contrast, the readthrough level
was similar for the UAG and UGA stop codons
(P ¼ 0.242). Nevertheless, some of the TAG mutations in
less favored contexts exhibited equivalent or even lower
gentamicin-induced readthrough levels than those displayed by TAA mutations surrounded by leakier
sequences (for example, compare TAA 1240c with TAG
1437c, TAG 651d, TAG 2522d and to TGA 967c) (Table 1).
Like for the basal level, a very broad distribution was
observed for gentamicin-induced readthrough levels,
ranging from 0.05 to 2.65%. UAA stop codons also gave
lower average readthrough values in the presence of the
antibiotic: 0.077% compared to 0.22% for UAG (P ¼ 0.032)
and 0.53% for UGA (P ¼ 0.00076). Here again, some of
the TGA mutations in less favored contexts exhibited
equivalent or even lower gentamicin-induced readthrough levels than those corresponding to TAA mutations surrounded by leakier sequences (for example,
compare TAA 2726d with TAG 1437c, TAG 2125d and
TGA 1326c, among others) (Table 1). We examined
whether the mutations that gave the highest basal
readthrough were particularly highly or poorly responsive to the gentamicin treatment; however, as illustrated
in Table 1, no correlation was found between the
readthrough efficiency and the increase factor that
reflects the responsiveness to the antibiotic (correlation
coefficient ¼ 0.28).
The nucleotide immediately following the stop codon
(position þ 4) is thought to have a major role in dictating
readthrough levels. Indeed, the presence of a C residue at
this position has often been reported to be correlated
with maximal levels of readthrough, in the absence or
presence of antibiotics.14,15 We thus analyzed the impact
of the þ 4 nucleotide on readthrough level on the entire
data set. A significant lower readthrough value was
observed only for A þ 4 compared to C þ 4, in the
presence of gentamicin (P ¼ 0.023). Figure 4 shows the
percentage of readthrough level in function of the
identity of the þ 4 nucleotide for different TGA mutations, in the presence of gentamicin. Obviously, there was
no systematic correlation between the þ 4 nucleotide and
the readthrough level. For example, a C at this position
was associated either with a very high (2.65% for ‘319d’)
or a moderate (0.6% for ‘1549c’) readthrough level. The
same phenomenon was observed when a G was present
at position þ 4 (from 1.66% for the TGAg target to 0.1%
for the 1326c sequence). Similar variations are observed
with A or T at this position and for the four nucleotides
in the absence of gentamicin.
Similar analyses have been carried out for TAA and
TGA mutations and lead to similar conclusions; no
nucleotide can be clearly associated with a particularly
low or high readthrough level (see Table 1).
Finally, analyses of all other nucleotide positions in the
cloned segment did not identify any major determinant
involved in basal or gentamicin-induced readthrough.
In vivo quantification of readthrough levels
We assessed the validity of the cell culture approach by
investigation of basal and gentamicin induced readthrough in vivo. For this, we used plasmid DNA
electrotransfer into C57BL/6 mouse skeletal muscle.25
Using the lacZ-luc dual gene reporter system, we were
Gentamicin and muscular dystrophies
L Bidou et al
623
Figure 4 Impact of nucleotide in þ 4 position on gentamicin-induced
readthrough. Distribution of readthrough levels driven by constructs with
a TGA nonsense mutation in the presence of 600 mg/ml of gentamicin. For
each construct, the nucleotide located immediately in 30 of the stop codon
( þ 4 position) is indicated above the histogram. Data are from Table 1,
except for mutations indicated by an asterisk, which have been constructed
to complete the schema (see Materials and methods section).
unable to detect any significant b-galactosidase activity.
To improve the sensitivity of our reporter vector, we
replaced the SV40 promoter of the pAC vector by the
strong CMV promoter, and the lacZ gene by the Renilla
luc gene encoding a very sensitive luciferase marker (see
Materials and methods and Figure 1), yielding the
pCRFL vector. As an initial validation, we introduced
the well-characterized leaky termination codon context
of the Tobacco Tumour Virus in pCRFL.26 This sequence
has been previously shown to promote a very high
readthrough efficiency (2–3%) in cultured cells.27,28 We
chose to use the same antibiotic dose (34 mg/kg) as that
in previously reported in vivo studies, since it showed no
toxic effects.6 Preliminary results showed that a significant readthrough level was obtained in murine skeletal
muscle (4%), and that gentamicin increased this 1.5-fold.
We then constructed a pCRFL reporter vector harboring
the 319d mutation (pCRFL319d), which exhibited the
highest gentamicin-induced readthrough efficiency and
the highest induction factor (0.06–2.65%) in cultured cells
assays (Table 1). Due to the changes introduced in the
reporter vector, we were able to detect the very low basal
readthrough level promoted by the 319d mutation. As
illustrated in Figure 5, both readthrough efficiencies and
the gentamicin effect were comparable in NIH3T3 cells
and, in vivo, increasing from 0.12% in the absence to 4.1%
in the presence of antibiotic in mouse skeletal muscle.
Discussion
Over the last few decades, significant progress has been
accomplished in the comprehension of the physiopathological mechanisms which result in muscular dystrophies; linking this knowledge to therapeutic
developments has so far proven difficult. Recently, the
Figure 5 Comparison of gentamicin effect in cultured cells and in vivo.
Vectors carrying the 319d mutation were either electroporated in NIH3T3
cells (pAC constructs) or in C57BL/6 mice (pCRFL constructs).
Readthrough efficiencies obtained in the absence (black histogram) or in
the presence of 600 mg/ml of gentamicin (gray histogram) were determined
as indicated in Materials and methods. For in vivo studies, 50 mg of
pCRFL plasmid DNA was used to determine readthrough levels into TA
muscles of 6-week-old male C57BL/6 (n ¼ 4 per group). Plasmid DNA was
injected in the left TA muscle and six transcutaneous electric pulses were
applied. Groups of mice received a daily intramuscular dose of 34 mg/kg
gentamicin or PBS during the next 5 days.
possibly most encouraging therapeutic approach for a
subset of muscular dystrophies has emerged with the
report by Barton-Davis et al,6 showing restoration of
dystrophin levels to 10–20% of those normally found in
the skeletal muscle of mdx mice, following subcutaneous
injections of gentamicin, an aminoglycoside long known
for its suppression of premature termination codons.
Aminoglycoside antibiotics are active in the codon–
anticodon recognition of aminoacyl tRNAs during
translation.3,29,30 Significantly, 10–20% of the mutations
identified in DMD and DMC are nonsense mutations
resulting in premature stop codons.18,21
In the analyses conducted to date, the role of the stop
codon nucleotide context on readthrough efficiency has
been investigated by directed mutagenesis of target
sequences inserted into reporter genes.14,15,28 In consequence, the analyzed sequences are related, which may
preclude the analysis of more complex effects possibly
involving a larger nucleotide context. In fact, a combinatorial analysis performed in yeast showed that the
context modulating readthrough extends at least six
nucleotides after the stop codon.13 The gentamicin
response of individual PTC found in patients might thus
vary considerably, depending on the position of the
mutation in the coding sequence. These informations
may permit a more accurate prediction of the response
of a patient to antibiotic treatment, and help to establish
the appropriate concentration of the drug in a potential
clinical trial.
Readthrough in vivo
To date, the potential of gentamicin to induce readthrough of nonsense mutations has been tested by two
Gene Therapy
Gentamicin and muscular dystrophies
L Bidou et al
624
complementary methods. First, a precise quantification
of readthrough was obtained using reporter systems,
either by transfection of cultured cells or through
expression of synthetic RNAs in rabbit reticulocyte
lysates.5,15 Secondly, the re-expression of a mutated gene
after gentamicin treatment has been assessed in animal
models, such as Mdxmdx or CFTR mutant mice.6,8
However, the correlation between readthrough levels
obtained in vitro or ex vivo and the re-expression
observed in vivo remained uncertain, since several
processes might modulate the primary molecular effect
of the antibiotic on readthrough, in particular mRNA
stabilization and protein accumulation.
In order to compare real readthrough levels obtained
in cultured cells and in vivo, we used electrotransfer of a
dual reporter gene in murine skeletal muscle, and
compared the readthrough efficiencies obtained in vivo
with those obtained in cultured NIH3T3 fibroblast cells.
Remarkably, almost identical results were obtained in
cultured cells and in mice. This result demonstrated, for
the first time, that no potentiation of treatment was
induced in muscles. This validates our approach,
allowing rapid and accurate quantification of genuine
readthrough levels.
Effect of stop codon identity
In basal conditions, targets with a UGA or UAG codon
showed, on average, higher translational readthrough
than those with a UAA. This was consistent with the
relative termination efficiency previously described for
the three stop codons (UAA4UAG4UGA).14,15 This
hierarchy was also observed in the presence of 600 mg/
ml of gentamicin, UGA and UAG, with readthrough
levels of 0.53 and 0.22%, respectively, and UAA only
0.076%. Nevertheless, in the presence or absence of
gentamicin, this order might be completely changed due
to the influence of the embedded stop condon context.
Overall, although UAA was more often associated with
low readthrough efficiency, in both the absence and
presence of the antibiotic, the identity of the stop codon
was not sufficient to predict either the basal readthrough
level, or the response of a given mutation to gentamicin.
Effect of the þ 4 nucleotide
This position has been extensively examined and
described as the major determinant of readthrough
efficiency.12,13,15,26,31,32 A C residue following either a
UAA or a UGA stop codon has been reported to be
usually correlated with high levels of readthrough, either
in the presence or absence of an aminoglycoside
antibiotic. UAG gave more erratic results, with U or C
being associated with high or low readthrough, possibly
reflecting the difference in the surrounding context
between the experimental systems used.14,15 The contribution of the þ 4 nucleotide to gentamicin-induced
readthrough appeared moderate in our collection of
mutations. In particular, a C at this position could be
associated with some of the highest readthrough levels,
but also with very moderate ones. In addition, a G, which
was previously correlated with low readthrough efficiencies, appeared to be associated either with low or
high readthrough levels in this study.
In conclusion, the impact of the þ 4 nucleotide seemed
much less critical than previously reported, and was
largely dependent on the surrounding context. Apparent
Gene Therapy
divergences between our results and some previous
observations most probably proceed from the fact that
we analyzed a number of different contexts, rather than
related sequences derived from a unique core by
mutagenesis.
In practice, this implies that the identification of the
nucleotide immediately downstream of the stop codon is
not sufficient to predict readthrough efficiency of a given
stop mutation, either in the absence or in the presence of
gentamicin.
Different rules govern basal and induced readthrough
Another inference drawn from our data is that no
correlation could be found between the efficiency of
stop codon readthrough in basal conditions and the
responsiveness of the stop codon to gentamicin treatment. In other words, sequences that mediate more
efficient termination at dose 0 were not systematically
the least (or the most) sensitive to the aminoglycoside.
Symmetrically, mutations displaying an intrinsic high
readthrough level could be poorly or highly responsive
to gentamicin. Thus, factors influencing gentamicininduced readthrough might be, at least partially, distinct
from factors that determine basal translational termination efficiency.
Relationship between stop codon readthrough and
gene re-expression
As already noted, the mouse Mdxmdx mutation gave a
particularly low level of readthrough upon gentamicin
treatment, although re-expression of dystrophin in
treated mice was reported to reach 10–20%.6 This
observation strongly suggests that other levels of
gentamicin action are involved in the phenotypic
reversion of the mdx mutation previously reported. A
good candidate is the NMD pathway, which might
potentiate the antibiotic effect by stabilization of the
message read by ribosomes.22 Similarly, other mechanisms like protein stability and accumulation might
collectivelly participate to potentiate the primary translational effect of gentamicin. Finally, gentamicin may also
affect gene expression at other levels than translation. In
particular, aminoglycoside antibiotics are known to affect
splicing, raising the possibility that they might be able to
induce exon skipping.33 None of these potential effects is
taken into account in our analyses, since we used dual
reporter systems allowing accurate quantification of the
translational readthrough event at the molecular level,
with no interference with other levels of control.
Potential use of gentamicin in treatment of genetic
diseases due to nonsense mutation
Overall, the readthrough levels obtained were very low:
85% of the stop codon contexts gave less than 1%
readthrough after gentamicin treatment, and the highest
values were lower than 3%. These results thus strongly
suggest that only a subset of DMD and CMD patients
carrying stop mutations would potentially benefit of a
gentamicin treatment aiming at suppressing premature
termination codons. They may also explain the failure or
relative difficulty to obtain significant improvement of
symptoms in clinical trials already reported,34–36 since
the patients included in these trials were not chosen on
the criteria of the response of their mutation. Similarly,
Gentamicin and muscular dystrophies
L Bidou et al
our efforts to correct the albino phenotype associated
with the platinum (Typcp) coat color mutation by local or
systemic treatment with gentamicin at several stages
during embryogenesis, in pups and in adults failed (data
not shown).
In conclusion, our opinion is that no prediction of the
efficiency of readthrough – and thus of a potential benefit
– can be drawn from the nucleotide context of the
mutation; individual tests of each particular mutation
might therefore provide useful information before submitting a patient to a potential clinical trial. These
observations emphasize that more remains to be learned
about the sequences and mechanisms promoting readthrough and translation termination in general, in order
to allow the rational use of drugs for the treatment of
nonsense mutation-associated diseases.
Materials and methods
Stop mutation targets
Both DMD and Congenital Muscular Dystrophy (CMD)
mutations were included, and are identified in the text
by their code followed either by the letter c or d (for
CMD and DMD, respectively). Sequences were either
retrieved from databases21 or provided by several
laboratories involved in a collaborative project coordinated by the ‘Association Française contre les Myopathies’. TGAg and 1952T targets used in Figure 4 have
been designed specially to complete analyses on nucleotide at the þ 4 position.
TGAg:
1952T:
50 TGG GGA ACA CAA TGA GAA TTA CAG
CCA 30
50 TTT GCC CGG TGT TGA TAT AAT TTA
GTC 30
Reporter plasmid cloning
Complementary oligonucleotides corresponding to
the sequences in Table 1 were annealed and ligated in
the pAC99 reporter plasmid, as previously described.37
This plasmid contains a unique Msc1 cloning site
between lacZ and luc genes. For a given test construct,
b-galactosidase and luciferase coding sequences are
separated by an in-frame stop codon corresponding to
the mutation found in a patient, flanked by four codons
in 50 and 30 , respectively. For each mutation, a control
plasmid was constructed, in which the stop codon was
replaced by the corresponding sense codon found at this
position in the wild-type allele, allowing expression of
100% of the fusion protein.
After transformation of Escherichia coli strain DH5a, all
target DNA sequences were assessed by sequencing.
Plasmids were purified using the Qiagen Midiprep Kit
(Courtaboeuf, France) according to the manufacturer’s
specifications.
For in vivo injections, plasmids were purified using the
Qiagen Megaprep endotoxin-free columns, and solubilized in NaCl (0.9%) endotoxin-free (Sigma).
Construction of the dual luciferase reporter vector
pCRFL
The PstI/HpaI DNA fragment from the p2luci vector38 (a
gift of Dr Atkins) containing the Renilla luciferase and
Firefly luciferase genes was cloned into the pBC SK( þ )
vector (Stratagene) at the PstI/HincII restriction site. At
the BglII site of this first construct, a BglII, MscI, HpaI, and
ClaI polylinker (GCCATGGAGGCCCCGGGG ATC) was
added. To remove the AUG translational start codon of
the Firefly luciferase gene, the sequence delineated by
HpaI/ClaI restriction sites was replaced by the Firefly
luciferase sequence contained within the MscI/ClaI sites
from the pAC99 vector.37 Finally, the NotI fragment
containing the 2 luciferase fusion genes was cloned at the
NotI restriction site on the pCDNA3 vector from
Invitrogen Company. In this construct, the expression
of the protein fusion Renilla luciferase/Firefly luciferase
was under the control of the CMV promoter. This dual
reporter vector was named pCRFL.
625
Cell culture and transfection
NIH3T3 cells were cultured in Dulbecco’s modified
Eagle’s medium supplemented with 7% fetal calf serum
(Gibco BRL), and incubated at 371C in humiditysaturated 6.5% CO2. Cells were electroporated with
18 mg of each pAC (lacZ-luc) plasmid construct, as
described,39 and plated immediately in culture medium.
Culture medium was replaced 7 and 30 h later with fresh
medium at a concentration of gentamicin (GIBCO BRL)
varying from 0 to 600 mg/ml. At 2 days after transfection,
cells were harvested, spun 5 min at 41C and lysed by
repeated pipetting in 150 ml of cold luciferase assay
buffer.39 For each construct, at least four independent
transfection experiments were performed.
Electric pulsed delivery in murine skeletal muscle
C57BL/6 males, 6-weeks old (IFFA Credo, France),
were anesthetized during the entire procedure by
intraperitoneal injection of ketamine (30 mg/kg)
and xylazine (10 mg/kg). One or two legs were shaved,
and 50 mg of DNA from the pCRFL319d (eight mice)
or the in-frame control plasmid (eight mice) in 0.9%
endotoxin-free NaCl (25–35 ml) was injected in the left
Tibialis anterior (TA) muscle, percutaneously and
longitudinally, using a Hamilton syringe with a 30-gauge
needle.
Immediately after injection, transcutaneous electric
pulses were applied using two stainless steel plate
electrodes placed on either side of the hind limb as
previously described,40 with some modifications.
Briefly, six square-wave electric pulses were generated
by an electropulsator (ECM-830 BTX, San Diego,
CA, USA) with an output voltage of 300 V/cm, a pulse
length of 15 ms, and a frequency of pulse delivery
of 2 Hz. Electric field strengths (in V/cm) are reported
as the ratio of applied voltage to the distance between
the electrodes.
In vivo gentamicin treatment
At 15 min after electrotransfer, four mice having received
the pCRFL319d and four mice having received the inframe control vector were injected with 17 mg of
gentamicin in each muscle. Gentamicin (17 mg) was then
injected each day for the next 3 days. The gentamicin was
from the same batch as that used for the cell culture
experiments. The remaining mice were injected with
phosphate-buffered saline.
Gene Therapy
Gentamicin and muscular dystrophies
L Bidou et al
626
Muscle extracts and determination of luciferase gene
expression
At the indicated times after DNA injection, mice were
euthanized, the Tibialis anterior was dissected and
rapidly frozen in liquid nitrogen for further analysis.
Muscle extracts were prepared as described.25 Briefly,
muscle extracts were prepared by homogenizing the
muscle tissue with an Ultra-Turrax homogenizer (Bioblock Scientific, Illkirch, France) in 500 ml of Promega
lysis buffer containing protease inhibitors (Sigma).
Extracts were frozen-thawed twice and immediately
cleared by centrifugation (5000 g for 15 min at 41C).
Enzyme assays
Luciferase and b-galactosidase activities were measured
from the same crude extract as described.39 For in vivo
tests, dual luciferase detection was achieved using the
dual detection kit from PROMEGA, following the
manufacturer’s instructions.
Stop codon readthrough was calculated by dividing
the ratio of luciferase to b-galactosidase activity obtained
with a given test construct by the ratio obtained from the
corresponding control construct.
Statistical analysis was performed using the Mann–
Whitney nonparametric test, which investigates the
difference between two independent samples carrying
different values of an individual subject. This test is an
alternative to the independent group t-test, when the
assumption of normality or equality of variance is not
met. This, like many nonparametric tests, uses the ranks
of the data rather than their raw values to calculate the
statistic (Richard lowry, 2000, Inferential statisticshttp://faculty.vassar.edu/lowry/ch11a.html).
Acknowledgements
We are especially indebted to Maryse Godon for help
with plasmid preparations. We are grateful to Laetitia
Baudry for animal experiences and Jerome Poupiot for
technical assistance. We wish to thank the ‘Association
Française contre les Myopathies’ for providing DMD
mutations. This work was supported by the ‘Association
Française contre les Myopathies’ (contract 7757) and by
the ‘Association pour la Recherche sur le Cancer’
(contract 4699).
References
1 Martin R et al. Aminoglycoside suppression at UAG, UAA and
UGA codons in Escherichia coli and human tissue culture cells.
Mol Gen Genet 1989; 217: 411–418.
2 Wimberly BT et al. Structure of the 30S ribosomal subunit. Nature
2000; 407: 327–339.
3 Burke JF, Mogg AE. Suppression of a nonsense mutation in
mammalian cells in vivo by the aminoglycoside antibiotics G-418
and paromomycin. Nucleic Acids Res 1985; 13: 6265–6272.
4 Bedwell DM et al. Suppression of a CFTR premature stop
mutation in a bronchial epithelial cell line. Nat Med 1997; 3:
1280–1284.
5 Howard M, Frizzell RA, Bedwell DM. Aminoglycoside
antibiotics restore CFTR function by overcoming premature
stop mutations (see comments). Nat Med 1996; 2: 467–469.
Gene Therapy
6 Barton-Davis ER et al. Aminoglycoside antibiotics restore
dystrophin function to skeletal muscles of mdx mice (see
comments). J Clin Invest 1999; 104: 375–381.
7 Arakawa M et al. Negamycin can restore dystrophin in mdx
skeletal muscle. Acta Myol 2001: 154–158.
8 Du M et al. Aminoglycoside suppression of a premature stop
mutation in a Cftr/ mouse carrying a human CFTR-G542X
transgene. J Mol Med 2002; 80: 595–604.
9 Dunant P, Walter MC, Karpati G, Lochmuller H. Gentamicin fails
to increase dystrophin expression in dystrophin-deficient
muscle. Muscle Nerve 2003; 27: 624–627.
10 Politano L et al. Gentamicin administration in Duchenne patients
with premature stop codon. Preliminary results. Acta Myol 2003;
22: 15–21.
11 Tate WP, Poole ES, Mannering SA. Hidden infidelities of the
translational stop signal. Prog Nucleic Acid Res Mol Biol 1996; 52:
293–335.
12 Bonetti B, Fu LW, Moon J, Bedwell DM. The efficiency of
translation termination is determined by a synergistic interplay
between upstream and downstream sequences in Saccharomyces
cerevisiae. J Mol Biol 1995; 251: 334–345.
13 Namy O, Hatin I, Rousset JP. Impact of the six nucleotides
downstream of the stop codon on translation termination.
EMBO Rep 2001; 2: 787–793.
14 Howard MT et al. Sequence specificity of aminoglycosideinduced stop codon readthrough: potential implications for
treatment of Duchenne muscular dystrophy. Ann Neurol 2000; 48:
164–169.
15 Manuvakhova M, Keeling K, Bedwell DM. Aminoglycoside
antibiotics mediate context-dependent suppression of
termination codons in a mammalian translation system. RNA
2000; 6: 1044–1055.
16 Stahl G, Bidou L, Rousset JP, Cassan M. Versatile vectors to study
recoding: conservation of rules between yeast and mammalian
cells. Nucleic Acids Res 1995; 23: 1557–1560.
17 Kahlmeter G, Dahlager JI. Aminoglycoside toxicity – a review of
clinical studies published between 1975 and 1982. J Antimicrob
Chemother 1984; 13 (Suppl A): 9–22.
18 Allamand V, Guicheney P. Merosin-deficient congenital
muscular
dystrophy,
autosomal
recessive
(MDC1A,
MIM#156225, LAMA2 gene coding for alpha2 chain of
laminin). Eur J Hum Genet 2002; 10: 91–94.
19 Helbling-Leclerc A et al. Mutations in the laminin alpha 2-chain
gene (LAMA2) cause merosin-deficient congenital muscular
dystrophy. Nat Genet 1995; 11: 216–218.
20 Tome FM et al. Congenital muscular dystrophy with merosin
deficiency. C R Acad Sci III 1994; 317: 351–357.
21 Roberts RG, Gardner RJ, Bobrow M. Searching for the 1 in
2,400,000: a review of dystrophin gene point mutations. Hum
Mutat 1994; 4: 1–11.
22 Culbertson MR. RNA surveillance. Unforeseen consequences for
gene expression, inherited genetic disorders and cancer. Trends
Genet 1999; 15: 74–80.
23 Ryder-Cook AS et al. Localization of the mdx mutation within
the mouse dystrophin gene. EMBO J 1988; 7: 3017–3021.
24 Beermann F et al. Misrouting of tyrosinase with a truncated
cytoplasmic tail as a result of the murine platinum (cp)
mutation. Exp Eye Res 1995; 61: 599–607.
25 Perez N et al. Tetracycline transcriptional silencer tightly controls
transgene expression after in vivo intramuscular electrotransfer:
application to interleukin 10 therapy in experimental arthritis.
Hum Gene Ther 2002; 13: 2161–2172.
26 Skuzeski JM, Nichols LM, Gesteland RF, Atkins JF. The signal for
a leaky UAG stop codon in several plant viruses includes the
two downstream codons. J Mol Biol 1991; 218: 365–373.
27 Berteaux V, Rousset JP, Cassan M. UAG readthrough is not
increased in vivo by Moloney murine leukemia virus infection.
Biochimie 1991; 73: 1291–1293.
Gentamicin and muscular dystrophies
L Bidou et al
627
28 Cassan M, Rousset JP. UAG readthrough in mammalian cells:
effect of upstream and downstream stop codon contexts reveal
different signals. BMC Mol Biol 2001; 2: 3.
29 Palmer E, Wilhem JM, Sherman F. Phenotypic suppression of
nonsense mutants in yeast by aminoglycoside antibiotics. Nature
1979; 277: 148–150.
30 Singh A, Ursic D, Davies J. Phenotypic suppression and
misreading Saccharomyces cerevisiae. Nature 1979; 277: 146–148.
31 Li G, Rice CM. The signal for translational readthrough of a
UGA codon in Sindbis virus RNA involves a single cytidine
residue immediately downstream of the termination codon.
J Virol 1993; 67: 5062–5067.
32 Pedersen WT, Curran JF. Effects of the nucleotide 30 to an amber
codon on ribosomal selection rates of suppressor tRNA and
release factor-1. J Mol Biol 1991; 219: 231–241.
33 Schroeder R, Waldsich C, Wank H. Modulation of RNA function
by aminoglycoside antibiotics. EMBO J 2000; 19: 1–9.
34 Clancy JP et al. Evidence that systemic gentamicin suppresses
premature stop mutations in patients with cystic fibrosis. Am J
Respir Crit Care Med 2001; 163: 1683–1692.
35 Wagner KR et al. Gentamicin treatment of Duchenne and Becker
muscular dystrophy due to nonsense mutations. Ann Neurol
2001; 49: 706–711.
36 Wilschanski M et al. A pilot study of the effect of gentamicin on
nasal potential difference measurements in cystic fibrosis
patients carrying stop mutations. Am J Respir Crit Care Med
2000; 161: 860–865.
37 Namy O, Duchateau-Nguyen G, Rousset JP. Translational
readthrough of the PDE2 stop codon modulates cAMP
levels in Saccharomyces cerevisiae. Mol Microbiol 2002; 43:
641–652.
38 Grentzmann G et al. A dual-luciferase reporter system
for
studying
recoding
signals.
RNA
1998;
4:
479–486.
39 Bidou L et al. In vivo HIV-1 frameshifting efficiency is directly
related to the stability of the stem-loop stimulatory signal. RNA
1997; 3: 1153–1158.
40 Mir LM et al. High-efficiency gene transfer into skeletal muscle
mediated by electric pulses. Proc Natl Acad Sci USA 1999; 96:
4262–4267.
Gene Therapy