Functional characterization and targeted correction of ATM mutations identified in Japanese patients with ataxia-telangiectasia


  • Communicated by Michel Goossens


A recent challenge for investigators studying the progressive neurological disease ataxia-telangiectasia (A-T) is to identify mutations whose effects might be alleviated by mutation-targeted therapies. We studied ATM mutations in eight families of Japanese A-T patients (JPAT) and were able to identify all 16 mutations. The probands were compound heterozygotes in seven families, and one (JPAT2) was homozygous for a frameshift mutation. All mutations—four frameshift, two nonsense, four large genomic deletions, and six affecting splicing—were novel except for c.748C>T found in family JPAT6 and c.2639-384A>G found in family JPAT11/12. Using an established lymphoblastoid cell line (LCL) of patient JPAT11, ATM protein was restored to levels approaching wild type by exposure to an antisense morpholino oligonucleotide designed to correct a pseudoexon splicing mutation. In addition, in an LCL from patient JPAT8/9, a heterozygous carrier of a nonsense mutation, ATM levels could also be partially restored by exposure to readthrough compounds (RTCs): an aminoglycoside, G418, and a novel small molecule identified in our laboratory, RTC13. Taken together, our results suggest that screening and functional characterization of the various sorts of mutations affecting the ATM gene can lead to better identification of A-T patients who are most likely to benefit from rapidly developing mutation-targeted therapeutic technologies. Hum Mutat 33:198–208, 2012. © 2011 Wiley Periodicals, Inc.


Ataxia-telangiectasia (A-T; MIM# 208900) is an autosomal recessive neurodegenerative disorder characterized by progressive cerebellar degeneration, ocular apraxia and telangiectasia, increased cancer risk, immunodeficiency, sensitivity to ionizing radiation (IR), chromosomal instability, and cell cycle abnormalities [Boder and Sedgwick, 1958; Gatti, 2001]. A-T is caused by mutations in the ATM gene (MIM# 607585) that usually encodes a 13 kb transcript that produces a 370 kDa protein [Gatti et al., 1988; Lange et al., 1995; Savitsky et al., 1995]. Intranuclear ATM protein is low or absent in most A-T patients, despite the presence of relatively normal levels of ATM transcripts. ATM is activated by autophosphorylation after binding with the MRN (Mre11-Rad50-Nbs) complex at sites of DNA double strand breaks [Bakkenist and Kastan, 2003; Kozlov et al., 2006], and subsequently phosphorylates hundreds of downstream target proteins involved in cell cycle checkpoints, DNA repair, and apoptosis [Bolderson et al., 2009; Matsuoka et al., 2007; Shiloh 2006]. ATM also appears to play a critical role in resolving chronic inflammation [Westbrook and Schiestl, 2010].

A-T patients are usually compound heterozygotes, carrying two distinct mutations. Mutations occur throughout the entire gene without hot spots. Founder effects are commonly observed in many ethnic isolates [Birrell et al., 2005; Campbell et al., 2003; Cavalieri et al., 2006; Gilad et al., 1996a; Laake et al., 1998; McConville et al., 1996; Mitui et al., 2003, 2005; Telatar et al., 1998a, b] wherein patients often carry mutations in a homozygous state. We have previously shown [Du et al., 2007, 2009, 2011; Lai et al., 2004] that accurately analyzing the functional consequences of mutations in individual A-T patients enables the grouping of patients into “mutation categories” that are most likely to be corrected by future customized mutation-targeted therapies.

The aims of the present study were to: (1) characterize the ATM mutations in Japanese A-T (JPAT) families; and (2) identify which JPAT patients might be candidates for personalized mutation-targeted therapy. We report that three of eight JPAT families examined are potential candidates for mutation-targeted therapy based on partial restoration of functional ATM protein production.

Materials and Methods

Cell Lines

Lymphoblastoid cell lines (LCLs) [Svedmyr et al., 1975] or activated T-cells [Minegishi et al., 2006] were established from affected members of eight Japanese A-T families, including three sibling pairs (JPAT4/5, 8/9, and 11/12). The families came from different geographical regions. Clinical descriptions of patients from these families have been reported previously [Morio et al., 2009].

Short Tandem Repeat (STR) Haplotype Analysis

Standardized STR (short tandem repeat/microsatellite) genotyping for the ATM gene region was performed as previously described [Mitui et al., 2003]. Briefly, we used four fluorescently labeled microsatellite markers located within a 1.4 cM region of chromosome 11q22-q23: D11S1819, NS22, D11S2179, and D11S1818. Markers NS22 and D11S2179 are located within the ATM gene, in introns 45 and 62, respectively [Udar et al., 1999; Vanagaite et al., 1995]. Allelic sizes were detected with an ABI 3730 DNA analyzer (Applied Biosystems Inc, Carlsbad, CA) and standardized to a reference sample (CEPH 1347-02).

Identification of Mutations

Total RNA was isolated from patient-derived T-cell lines using RNeasy (QIAGEN, Valencia, CA), and cDNA was synthesized using random primers and the Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA). The entire ATM coding region was divided into eight overlapping fragments (Regions 1–8) ranging from 1,500 to 1,800 bps [Du et al., 2008]. These regions were PCR amplified and then sequenced using 19 different primers. Mutations on the cDNA level were confirmed in genomic DNA (gDNA) by sequencing relevant exon and intron boundaries. Mutation analysis is based on the same ATM reference sequence used for ATM mutations in the Leiden Open Variation Database (; NCBI reference sequence:NM_000051.3).

Maximum Entropy Scores and Search for Exonic Splicing Enhancers (ESEs)

The strength of the 5′ and 3′ splice sites (ss) was determined by calculating and comparing the wild-type and mutant 5′ and 3′ ss using the Maximum Entropy software available at [Eng et al., 2004; Mitui et al., 2009; Yeo and Burge, 2004]. We scanned for putative binding motifs for serine/arginine-rich (SR) proteins using the ESEfinder software available at [Cartegni et al., 2003; Smith et al., 2006].

Long-Range PCR and Breakpoint Regions for Genomic Deletions

To amplify large gDNA fragments, 500 ng of gDNA was used as template, followed by 35 cycles of 95C for 1 min, 68oC for 10 min, and extension at 72°C for 15 min using EX Taq polymerase according to the manufacturer's protocol (Takara Bio Inc, Shiga, Japan). Fragments containing large genomic deletions (LGDs) were isolated from agarose gels and sequenced to determine the breakpoints.

Multiplex Ligation-dependent Probe Amplification (MLPA)

A total of 100 ng of gDNA was used as starting material for the SALSA MLPA P041 and P042 ATM kits (MRC-Holland, Amsterdam, Netherlands, [Schouten et al., 2002]. The P041 probe mix contained probes for 33 of the 65 exons as well as three probes for exon 1. The P042 ATM probe mix contained probes for the remaining ATM exons. Both probe mixtures also contained probes for control genes. After hybridization, ligation, and amplification, according to the instructions of the manufacturer, 1 μl of PCR product was mixed with 0.2 μl of ROX-500 labeled internal size standard, separated on an ABI Prism 3100 Avant automatic sequencer (Applera, Norwalk, Connecticut, CA), and analyzed using the GeneScan software ver.3.1. For MLPA data analysis, we used Coffalyser MLPA DAT software developed by MRC-Holland. For each probe, a range from 1 ± 0.2 was considered as a normal exon dosage, while a deletion was determined as being between 0.3 and 0.7.

Antisense Morpholino Oligonucleotide (AMO) Design and Treatment

A 25-mer antisense morpholino oligonucleotide (AMO) was designed to target the 5′ aberrant splice site of a pseudoexon mutation in pre-mRNA of JPAT11/12. The AMO-J11 sequence was: CCTGGAAAAATACTTACAATTAAAC. AMO748C (ATTCACACACTCGAATTCGAAAGTT) and AMO4956GC (CTTGGATAACTGCAACAAATTGACA) were designed to target wild-type sequences to determine potential regulatory elements at the site of a mutation(s). AMOs were synthesized by Gene-Tools (Philomath, OR). Treatment of LCLs with AMOs was performed as previously described [Du et al., 2007]. Cells were suspended in 5% FBS/RPMI medium and the AMO was added directly to medium at the concentrations indicated. Endo-Porter (Gene-Tools) was added to the medium to assist in intracellular incorporation of the AMO. Cells were collected after 48 hr for RNA analysis, and after 84 hr for ATM protein detection. Vivo-AMO was also used to treat JPAT 11 to enhance cellular delivery (Gene-Tools).

Irradiation Induced ATM-Ser1981 Foci Formation (IRIF)

Immunostaining of nuclear foci of ATM-Ser1981was performed as described [Du et al., 2007, 2009]. In brief, LCLs were first treated with the relevant compounds for 4 days before being irradiated with 2 Gy and then incubated at 37°C for 30 min. Next, the cells were fixed with 4% paraformaldehyde and then permeabilized on cover slips. The cover slips were blocked for 1 hr and incubated with mouse anti-ATM pSer1981 for 1 hr (1:500; Cell Signaling Technology, Danvers, MA). After a second blocking, cells were stained with Alexa Fluor 488 anti-mouse IgG (1:150; Invitrogen) for 1 hr and mounted onto slides.

Flow Cytometry Analysis of ATM-Ser1981 Autophosphorylation (FC-ATM-pSer1981)

FC-ATM-pSer1981 was used to verify the restoration of Ser1981 autophosphorylation by readthrough compounds (RTCs) [Du et al., 2009; Nahas et al., 2009]. The cells were treated for 4 days with RTCs, resuspended in PBS, and irradiated with 10 Gy. After 1 hr, the cells were fixed and permeabilized using FIX & PERM (Invitrogen). The cells were then incubated with 1 μl of mouse ATM-s1981 antibody (Cell Signaling Technology) for 2 hr at room temperature. After this time, cells were washed and resuspended in 100-μl PBS with Alexa Fluor 488 anti-mouse IgG (Invitrogen) for 45 min, and then washed and resuspended in PBS with 0.2% paraformaldehyde, before being analyzed using a FACSCalibur (BD, Franklin Lakes, NJ).

Western Blotting

Nuclear extracts were prepared by following the NE-PER protocol (Thermo Fisher Scientific, Rockford, IL). Proteins were separated on a 7.5% SDS-polyacrylamide gel. Western blots were prepared as described [Du et al., 2007], and probed with anti-ATM (Novus Biologicals, Littleton, CO), -SMC1, or -KAP1 antibodies (Novus Biologicals).


Mutation Analysis

We initially screened our A-T patients for two previously reported Japanese mutations, c.4776(IVS33)+2T>A and c.7883_7887delTTATA [Ejima and Sasaki 1998; Fukao et al., 1998]. Neither of these mutations was detected.

STR genotyping of the ATM genomic region was performed for 11 JPAT patients, but since parental gDNAs were unavailable, we could only verify that one patient was homozygous for all markers (JPAT2): [S1819, 131; NS22, 165; S2179, 143; S1818, 162] [Mitui et al., 2003]. As a result, we set out to directly sequence the entire ATM coding region after PCR amplifying eight partially overlapping fragments from patients’ cDNA [Du et al., 2008]. We identified 12 of the 16 expected mutations (75%) and confirmed them upon sequencing gDNA (Table 1). Only one patient (JPAT2) was homozygous, suggesting that most JPAT patients do not result from consanguineous marriages. The 12 mutations included four frameshifts (counting the homozygous JPAT2 twice), two nonsense, and six splice variants (Table 1). The remaining mutations (4/16; 25%) were four LGDs, which we identified after performing long-range PCR using gDNA as template. Fourteen mutations were novel; two had been previously reported: c.748C>T in JPAT6 [Teraoka et al., 1999] and c.2639-384A>G in JPAT11/12 [Sobeck 2001]. All mutations resulted in the absence of ATM protein (Supp. Fig. S1 and data not shown).

Table 1. Mutations of Eight Japanese Families
Ex/IntPatientcDNA changeGenomic DNA mutationConsequence
  1. a

    Bolded mutations have not been reported previously.

  2. b

    aFirst allele.

  3. c

    bSecond allele.

  4. d


  5. e

    Nucleotide numbering is based on +1 being the A of the first translation start codon in exon 4 (NCBI reference sequence: NM_000051.3).

IVS6JPAT1ac.186_331del146 (deletes exon 6)c.331+5G>A (5′ss 9.81>3.58)Aberrant splicing (IV)
9JPAT6bc.663_901del239 (deletes exon 9)c.748C>T (R>X)Aberrant splicing (III)
10JPAT8/9ac.902_1065del164 (deletes exon 10)c.902-19_1065+869del1052Large genomic deletion
IVS19JPAT11/12ac.2639_2640ins58c.2639-384A>G (5′ss 0.36>8.54)Aberrant splicing (II)
IVS19JPAT3ac.2639_2838del200 (deletes exon 20)c.2639-19_2639-7del13 (3′ss 8.8>3.4)Aberrant splicing (IV)
20JPAT8/9bc.2877C>Gc.2877C>G (Y>X)Nonsense (TAG)
35JPAT4/5ac.4910_5005del96 (deletes exon 35)c.4956GC>TT (LQ>FX)Aberrant splicing (III)
38JPAT1bc.5415G>Ac.5415G>A (W>X)Nonsense (TGA)
IVS48JPAT11/12bc.6808_7515del708 (deletes ex 49-52)c.6807+272_7516-275del5350Large genomic deletion
60JPAT10ac.8419_8584del166 (deletes exon 60)c.8419-643_8507del732Large genomic deletion
61JPAT10bc.8585_8671del87 (deletes exon 61)c.8585-1G>C (5′ss 10.2>2.0)Aberrant splicing (IV)
IVS63JPAT3bc.8851_9697del847c.8852-2kbdel17kb (CRAT [B] mutation?)Large genomic deletion

Splicing Mutations

The six splicing mutations identified were analyzed by using Maximum Entropy software (MaxENT) to estimate the strength of the splice sites [Yeo and Burge, 2004] and type of splice defect [Eng et al., 2004]. The mutations found are described below, and diagrams for potential splicing mechanisms are shown in Figure 1.

Figure 1.

ATM splicing mutations. Genomic mutations causing splicing mutations were analyzed for changes in splicing scores calculated by Max ENT. Classification of splicing mutations is reported accordingly to Eng et al. [2004]. See text for additional details.

  • (1)c.331+5G>A (IVS6): This mutation changed the MaxENT score of the 5′ ss from 9.8 to 3.6. A shorter PCR product compatible with exon 6 skipping was observed at the cDNA level in patient JPAT1 using primers for exons 4 and 7 (Figs. 1 and 2A, lane 3).
  • (2)c.748C>T: cDNA from patient JPAT6 showed skipping of exon 9 (Figs. 1and 2B, lane 5). This allele with substitution c.748C>T predicted an amino acid change from Arg to a stop codon (CGA >TGA). Given that c.748C>T did not affect the scores for consensus splice sites, nor affect an ESE site, we hypothesized that it affected an as yet unknown splicing regulatory element. To test this idea further, we designed an AMO targeting the wild-type sequence at the site of the mutation in order to block the interaction between any regulatory molecule(s) and the wild-type sequence. Wild-type cells treated with increasing concentrations of AMO748C (Fig. 2G) showed skipping of exon 9, supporting idea model that the region around nucleotide 748 most likely contains a regulatory splicing motif.
  • (3)c.2639-384A>G (IVS19): The c.2639-384A>G variant in patient JPAT11/12 creates a novel splice acceptor site within IVS19 (Fig. 1), thereby creating a cryptic splice and “pseudo-exon” of 58 bp is created in intron 19 (Fig. 2C, lanes 5 and 6). This results in a frameshift and a predicted secondary premature stop codon.
  • (4)c.2639-19_2639-7del13 (IVS19): In Figure 2D (lane 3), the PCR products from JPAT3 cDNA showed a normal and an additional prominent lower band (783 bp and 583 bp, respectively). Sequencing of the 583-bp band revealed skipping of exon 20. gDNA sequencing identified a 13 nt deletion in intron 19 at position c.2639-19_2639-7. The 3′ MaxENT score changed from 8.8 to 3.4 (Fig. 1).
  • (5)c.4956GC>TT: In family JPAT4/5, we identified a c.4956GC>TT substitution within exon 35 (p.LQ1652_1653FX) that leads to skipping of exon 35 without affecting an ESE or canonical splice sites (Figs. 1 and 2E, lanes 3 and 4). Exposing wild-type LCLs to increasing concentrations of AMO4956GC, targeting the mutation site, revealed skipping of exon 35 (Fig. 2H); these results suggest that nucleotide 4956 is part of a regulatory protein binding site, which when disrupted influences the aberrant splicing observed in JPAT4/5.
  • (6)c.8585-1G>C (IVS60): JPAT10 harbors the IVS60-1G>C mutation that changed the MaxENT score of the 3′ ss from 10.2 to 2.0, resulting in a skipping of the exon 61 (Fig. 2F, lane 4). Interestingly, the second allele of this patient was a splicing mutation that is predicted to result in exon 60 skipping (Fig. 2F, lane 4). We sequenced gDNA for exons 59–62 but failed to find a mutation that would account for the skipping of exon 60 (however, see additional results on JPAT10 below).
Figure 2.

Effect of splicing mutations on cDNA. Agarose gel images of PCR products showed aberrant spliced products. Patient cDNA were used as templates for PCR amplifications in the regions displaying splicing mutations. M (lane 1) is 1 kb plus ladder (Invitrogen), wild-type cDNA was used as control (lane 2). (A) Skipped exon 6 in JPAT1 (lane 3). (B) Skipped exon 9 in JPAT6 (lane 5) and skipped exon 10 in JPAT8 (lane 6). (C) Pseudoexon of JPAT11 and JPAT12 (lanes 5 and 6). (D) Skipped exon 20 in JPAT3 (lane 3). (E) Skipped exon 35 in JPAT4 and JPAT5 (lanes 3 and 4). (F) Skipped exons 60 and 61 in JPAT10 (lane 4). (G) AMO-treated wild-type lymphoblastoid cell line (LCL) produced alternative spliced product that skipped exon 9. JPAT6, carrying the c.748C>T mutation, showed a skipped exon 9 product (lane 6), (H) AMO 4956GC treated wild-type LCL produced alternative spliced product that skipped exon 35. JPAT4 that has 4956GC>TT mutation showing skipped exon 35 products as a control (lane 6). See text for additional details.

Large Genomic Deletions (LGDs)

  • (1)c.902-19_1065+869del1052 (del ex10): Two siblings (JPAT8/9) yielded an abnormal 369-bp fragment when cDNA was amplified from exon 9 to 11 (Fig. 3A, cDNA gel, lanes 3 and 4). When this band was isolated and sequenced, we found a deletion of exon 10. No mutation was observed in exons 9–11, ruling out a conventional splicing mutation. Using long-range PCR to amplify the genomic region from exon 9 to 11, we obtained a 3.3 kb fragment (Fig. 3A, gDNA gel lanes 3 and 4), whose sequence revealed a 1,052-bp deletion from IVS9-19 to IVS10+869; this deletion included exon 10 (164 bp).
  • (2) c.6807+272_7516-275del5350 (del ex49-52): Two siblings (JPAT11/12) showed an abnormal PCR fragment of 1.1 kb when cDNA was amplified from exon 48 to 53 (Fig. 3B, cDNA gel). The sequence of the PCR product showed a deletion of exons 49–52. A long-range PCR performed on gDNA using primers for exons 48 and 53 produced a 1.1 kb band instead of the expected 6.4 kb (Fig. 3B, left). Sequencing of the 1.1 kb band revealed a 5,350-bp genomic deletion that starts in intron 48 and ends in intron 52.
  • (3)c.8419-643_8507del732 (del ex 60): In patient JPAT10, we suspected that skipping of exon 60 might reflect an LGD. We amplified the gDNA surrounding exons 59–61 and found a 732-bp genomic deletion extending from IVS59-643 to nucleotide 89 of exon 60 (Fig. 3C).
  • (4)c.8851-2kbdel17kb (del ex64-65): When mutation screening failed to identify a second pathogenic mutation in JPAT3, we were prompted to search for an LGD mutation with Multiplex Ligation-dependent Probe Amplification (MLPA). We observed a significant decrease in peak height for the final exons 64 and 65, indicative of a deletion carried in heterozygous state (Fig. 4A). Previous studies have demonstrated two LINE-1 sequences between IVS63 and downstream of exon 65, as well as a 17 kb genomic deletion in the ATM gene of A-T patients with Costa Rican, Dutch, and Brazilian backgrounds [Broeks et al., 1998; Coutinho et al., 2004; Mitui et al., 2003; Telatar et al., 1998b].
Figure 3.

Large genomic deletions (LGDs). (A) Schematic representation of cDNA showing deletion of exon 10 and agarose gel image of PCR products (using primers GFw and GRev) (left): Lane 1 is 1 kb plus ladder (Invitrogen), lane 2 is wild-type control, lane 3 (JPAT8) and lane 4 (JPAT9) showing deletion of exon 10, lane 5 is JPAT control. Agarose gel image (right) for genomic DNA (gDNA) PCR products (using primers EX9Fw and EX11Rev) showing deletion of 1 kb in JPAT8/9. Schematic representation of DNA shows LGD, as well as repetitive elements within the region (at bottom). Sequence data with junction sequences are shown on right. (B) Schematic representation of cDNA change in JPAT11/12 between exon 48 and exon 53, which are analyzed by PCR (using primers FATFw and FATRev). Agarose gel image (left) for cDNA shows aberrant spliced products of JPAT11 (lane 5). Lane 1 is 1 kb plus ladder, lane 2 is wild-type, lanes 3 and 4 are JPAT control. Agarose gel image (right) for gDNA shows deletion of 5.3 kb (using EX Taq polymerase with LREX48Fw and LREX53Rev primers). Schematic representation of gDNA (at bottom) shows large deletion. Sequence data with junction sequences are shown on right. (C) Schematic representation of JPAT10 cDNA shows deletion of exon 60. Agarose gel image (left) shows aberrant spliced products. Agarose gel image (right) for gDNA PCR products (using primers EX59Fw and EX61Rev) show reduced size in JPAT10 (3.6 kb) compared to wild type (4.3 kb) and schematic representation of gDNA showing deletion at c.8269-643del732, which includes the first 89 bp of exon 60. Sequences are shown at right.

Figure 4B summarizes the locations of primers, LINE-1 sequences, and an LGD for this region.

Figure 4.

Analysis of LGD in patient JPAT3. (A) Multiplex Ligation Probe Amplification (MLPA) analysis using MLPA P041 (A) and P042 (B) kits. For each of the two analyses normalized peak area histograms of JPAT3 are shown. The dotted lines indicate the normal exon dosages between 0.8 and 1.2. The heavy black arrows indicate the exon probes of decreased signal, corresponding to the genomic deletion of exons 64 and 65. (B) PCR products depicting a LGD between two LINE-1 sequences. Lane 1 is 1 kb plus ladder, lane 2 is wild-type control, lane 3 is JPAT3, lane 4 is homozygous CRAT [B] patient with a very similar mutation, lane 5 is a heterozygous individual with CRAT [B] mutation, lane 6 is JPAT8. Schematic diagram (below) shows the relative locations for four different primers at the region of the LGD.

We used two sets of primers: Primer set #1 (P1Fw and P4Rev) was 23 kb apart, flanking the 17 kb deletion. Because of the nature of our PCR conditions, no PCR product was anticipated from the wild-type allele, while the mutant allele should yield a 6 kb fragment. Primer set #2 (P2Fw and P3Rev) was placed within the 17 kb deletion, which should have produced a 2.4 kb fragment from only the wild-type allele [Telatar et al., 1998b]. Figure 4B (lane 2) shows that wild-type gDNA produced the 2.4 kb fragment, while CRAT [B] (a Costa Rican patient homozygous for a 17 kb deletion) produced the 6 kb fragment (lane 4). A CRAT [B] heterozygote produced both the 2.4 kb fragment and the 6 kb product from the deletion (Fig. 4B, lane 5). The CRAT [B] band pattern was also observed in the gDNA of JPAT3, suggesting the presence of an LGD between two LINE-1 sequences (Fig. 4B, lane 3). Available breakpoints and surrounding sequences were analyzed using Repeat Masker software to search for flanking repetitive elements [Babushok and Kazazian, 2007; Kazazian and Goodier, 2002; Telatar et al., 1998b] (see Fig. 3). Because the breakpoint was in a highly homologous repeat sequence, the ends could not be accurately determined. The other Japanese patient (JPAT8) who did not have a deletion in this region showed a pattern identical to the wild type (Fig. 4B, lane 6).

The STR haplotypes for JPAT3, CRAT [B], and BRAT3 differed. JPAT3: S1819 [131,133]; NS22 [173,175]; S2179 [137,137]; S1818 [160,168]; CRAT [B]: S1819 [131]; NS22 [171]; S2179 [141]; S1818 [160] [Mitui et al., 2003]; BRAT 3: S1819 [133]; NS22 [155]; S2179 [147]; S1818 [146] [Mitui et al., 2003]. These results suggested that the c.8851-2kbdel17kb mutations in the three patients were not ancestrally related.

Correction of Type II Pseudoexon Splicing Mutation using an AMO

In family JPAT11/12, we identified a type II splicing mutation [Eng et al., 2004] c.2639-384A>G, which created a cryptic acceptor splice site resulting in the inclusion of 58 bp of intronic sequence (Figs. 2C and 5A). We designed AMO-J11 to target the cryptic 5′ splice site (Fig. 5A) [Du et al., 2007; Eng et al., 2004]. The LCL of JPAT11 was treated with AMO-J11 for 4 days followed by RT-PCR analysis. Mutant splicing was almost completely abrogated in an AMO dose-dependent manner and normal transcript was restored (Fig. 5B). Nuclear extracts from treated JPAT11 cells also showed a full-length ATM protein (data not shown). In order to enhance the delivery and efficiency of the AMO, we also designed a structurally modified AMO referred as “Vivo-AMO” [Morcos et al., 2008; Moulton and Jiang, 2009]. Notably, a significant amount of functional ATM protein was induced by 0.5 μM Vivo AMO-J11 (Fig. 5C). However, “Vivo-AMO” started to show possible cytotoxicity at 0.8μM (Fig. 5C, lane 4).

Figure 5.

Correction of a splicing mutation in JPAT11 using AMO-J11. (A) Schematic representation of inclusion of pseudoexon (type II) and location of designed AMO. (B) JPAT11 cells were treated with 0, 10, 20, and 40 μM AMO for 4 days and RNAs were isolated and analyzed for corrected splicing products. (C) JPAT11 cells were treated with 0.2, 0.5, and 0.8 μM of Vivo-AMO-J11 for 4 days and nuclear lysates were isolated for western blotting. KAP1 antibody was used as a loading control.

Correction of Nonsense Mutation in JPAT8 using RTCs

The JPAT8/9 siblings lack ATM protein because they carry an LGD and a nonsense mutation (c.2877C>G, p.Tyr959X). Functional ATM protein is inducible with compounds that readthrough premature termination codons [Du et al., 2009] even when the LCL carries the nonsense mutation in a heterozygous state [Lai et al., 2004]. We treated JPAT8 LCL with the readthrough compound RTC13 for 4 days, and measured ATM autophosphorylation after DNA damage induced by IR. An aminoglycoside RTC, G418, served as a positive control. Both G418 and RTC13 successfully induced ATMpS1981 autophosphorylation as shown by (A) FC-ATMpSer1981 and (B) IRIF ATMpSer1981 assays (Fig. 6), thereby demonstrating a potential therapeutic approach for these siblings, despite the presence of a nonsense mutation in only one allele.

Figure 6.

Readthrough compounds (RTCs) restored ATMpSer1981 autophosphorylation in JPAT8 LCL with a nonsense mutation. Cells were treated with readthrough compounds RTC13 and G418 for 4 days and then analyzed for ATMpSer1981. (A) ATMpSer1981 autophosphorylation level measured by FC-ATMpSer1981. Delta FI (fluorescence intensity) reflects the difference in FI between nonirradiated and irradiated cells. The data represent one of three independent experiments and results were consistent. (B) ATMpSer1981 foci formation by IRIF assay. The data are an average of two independent experiments.


Whereas a high (0.5%) coefficient of inbreeding had been recorded for the Japanese population [Pattison, 2004], the ATM mutation spectrum identified in our JPAT patient cohort included almost no homozygous mutations and no founder mutations. The c.4776+2T>A and c.7883_7887del5 mutations reported by Ejima and Sasaki [1998] and Fukao et al. [1998] were not observed. In fact, all mutations detected in the JPAT families—four frameshift, two nonsense, four LGDs, and six affecting splicing—were new except for a previously identified c.748C>T splicing mutation in an Irish–American family [Teraoka et al., 1999] and the c.2639-384A>G mutation [Sobeck, 2001]. It is noteworthy that only three of the six splicing mutations involved canonical sites (c.331+5G>A, c.2639-19_2639-7del13, c.8585-1G>C). In the other mutations: (1) c.748C>T (exon 9) and 4956GC>TT (exon 35) presented as nonsense mutations in gDNA, but at the cDNA level caused skipping of the exon in which they were located; and (2) JPAT11/12 had a deep-intronic mutation (c.2639-384A>G) in intron 19 that seemed to activate a cryptic acceptor splice site resulting in the insertion of a 58-bp pseudoexon in the transcript (Figs. 1, 2C, and 5).

Nonsense mutations frequently alter the splicing of the exon containing them, an observation that has been termed nonsense-associated altered splicing (NAS) [Valentine, 1998]. In most cases of NAS, the mutation disrupts an ESE critical for exon inclusion [Liu et al., 2001]. AMOs usually act by covering/masking a mutated site in the pre-mRNA, but cannot be used to correct splicing mutations at canonical sites. Thus, AMOs are most effective in correcting type II and IV splice mutations [Eng et al., 2004]. Most relevant here, AMOs can be applied to the functional analysis of ATM mutations. By using AMOs designed to bind the wild-type exon 9 and exon 35 sequences located around the mutation sites, we were able to induce alternative splicing (Fig. 2G and H). These results suggest that unknown regulatory elements are probably located near or at the sites of the mutation and are necessary for modulating normal splicing events in these exons. Furthermore, as described earlier, the deep-intronic mutation (c.2639-384A>G) in intron 19 is among the most attractive candidates for AMO therapy (Fig. 5) [Du et al., 2007].

LGDs and duplications within the ATM gene account for approximately 2% of reported mutations [Cavalieri et al., 2006, 2008; Coutinho, et al., 2004; Ejima and Sasaki, 1998; Gilad et al., 1996b; Mitui et al., 2003; Telatar et al., 1998b]. Few LGDs have been found in a homozygous state in A-T patients [Mitui et al., 2003]. The recent introduction of the MLPA technique has greatly improved the detection of genomic rearrangement mutations, including LGD and duplications in the ATM gene [Cavalieri et al., 2008].

Using MLPA, we identified the c.8851-2kbdel17kb mutation in JPAT3. While it was possible that this LGD mutation was ancestrally related to the previously reported CRAT [B] [Telatar et al., 1998a] and BRAT3 [Mitui et al., 2003] mutations, this now seems unlikely, since all three alleles are carried on different STR haplotypes. Given that homology between repetitive sequences is thought to underlie the formation of some genomic deletions and duplications [Kazazian and Goodier, 2002; Telatar et al., 1998a], we carried out an in silico analysis of DNA sequences flanking genomic deletions and were able to identify several repetitive sequences. The presence of microhomology (GGA in JPAT11) suggests that the recently described microhomology-mediated break-induced replication FoSTeS/MMBIR mechanism could be responsible for generating these deletions [Hastings et al., 2009].

Readthrough of PTCs was first described almost 50 years ago when it was noticed that certain aminoglycosides, such as streptomycin and gentamicin, “suppressed” the mutated phenotype of auxotrophy in strains of Escherichia coli suggesting that the drug interfered with accurate translation of the RNA code into protein [Davies et al., 1965]. Later, crystallography studies elegantly demonstrated that aminoglycosides bind to the internal loop of helix 44 (the decoding site) of the 16 S ribosomal RNA [Dibrov et al., 2010; Lynch et al., 2003]. More recently, we identified two new nonaminoglycoside small molecules with readthrough activity on both ATM and dystrophin genes [Du et al., 2009].

Based on the mutation spectrum of ATM, it is estimated that approximately 30% of ATM mutations in A-T patients are potentially treatable by mutation-targeted therapy using either RTCs or AMOs. This includes A-T patients who are compound heterozygotes, since RTCs and AMOs restore significant amounts of ATM protein even when only one allele is targeted. We identified three such examples amongst the eight Japanese families. In fact, in vitro, we were able to correct (1) abnormal splicing for JPAT11/12 (Fig. 5A), using a custom-designed AMO to mask the cryptic splice site created by the pseudoexon mutation (c.2639-384A>G); (2) a nonsense mutation in cells from JPAT8/9 using RTCs (Fig. 6). Thus, our results demonstrate that mutation-targeted treatment of cells carrying poorly understood DNA variants can extend our understanding of the consequences of such changes and may also have important therapeutic potential.


S.C. received a fellowship from “Associazione Gli Amici di Valentina.” We thank Drs. Shareef Nahas, Hailiang Hu, and Mark Ambrose for helpful discussions.