A mouse model for nonsense mutation bypass therapy shows a dramatic multiday response to geneticin

Abstract

Aminoglycosides can bypass nonsense mutations and are the prototypic agents for translational bypass therapy (TBT). Initial results demonstrate the need for more potent drugs and an in vivo model system for quantitative assessment of TBT. Herein, we present an in vivo system for evaluating the efficacy of premature stop codon management therapies: in vivo quantitative stop codon management repli-sampling TBT efficacy assay (IQSCMaRTEA). Application of IQSCMaRTEA reveals that geneticin is much more efficacious in vivo than gentamicin. Treatment with geneticin elicits a multiday response, and residual F9 antigen can be detected after 3 weeks. These data demonstrate the utility of IQSCMaRTEA for evaluating drugs that bypass nonsense mutations. In addition, IQSCMaRTEA may be helpful for testing inhibitors of nonsense-mediated decay, as stop codon management therapy will sometimes require inhibition of nonsense-mediated decay and translational bypass of the nonsense mutation. Furthermore, geneticin, its metabolites, or better tolerated analogues should be evaluated as a general treatment with multiday response for severe genetic disease caused by nonsense mutation.

Aminoglycosides can cause extensive misreading of the mRNA code in vitro (1) and bypass nonsense mutations as demonstrated in Escherichia coli (2), tetrahymena (3), wheat embryo (4), yeast (5), cultured animal cells (6) and human cells (4, 7–9). Barton-Davis et al. (10) described gentamicin-mediated translational bypass therapy (TBT) in the dystrophin gene in a mouse model for muscular dystrophy. Two other mouse models (6, 11) have been described, and clinical trials have been initiated (12, 13). The clinical studies have been difficult to perform and disappointing to date, suggesting the need for a better animal model to explore the dose–response relationships of compounds that mediate TBT. No in vivo animal system was heretofore available for direct and quantitative functional assessment of TBT efficacy.

We describe such an in vivo system: in vivo quantitative stop codon management repli-sampling TBT efficacy assay (IQSCMaRTEA). Three mutant human minigenes are used as models, two with nonsense mutations with and without dramatic nonsense-mediated decay (NMD) and one with a missense mutation.

Hemophilia is the oldest known hereditary bleeding disorder affecting ≈1 in 10,000 males. There are ≈20,000 hemophilia patients in the United States. Each year, ≈400 babies are born with this disease, ≈80% with hemophilia A and 20% with hemophilia B (HB). Approximately 70% of hemophilia patients have severe hemophilia, including those with nonsense mutations. TBT has the potential to provide an inexpensive prophylaxis in hemophilia and greatly improve or eliminate symptoms from any severe genetic disease caused by a nonsense mutation (15% of severe inherited genetic disease). However, NMD and sequence context effects complicate the practical application of TBT (14, 15).

IQSCMaRTEA allows (i) direct quantitation of TBT products (antigen levels), (ii) measurement of functional activity of the TBT products (enzymatic activity), (iii) repeat sampling over one or multiple experiments, and (iv) analysis of nonsense mutations with and without dramatic NMD, enabling assessment of two dimensions of stop codon management therapy: nonsense bypass and inhibition of NMD. Application of IQSCMaRTEA reveals a multiday response to geneticin.

Results and Discussion

Human F9 minigenes with the R338X or R29X nonsense mutations, G381E missense mutation, or wild-type sequence were placed in HB knockout mice that lack the murine endogenous F9 gene (Table 1). The complete IQSCMaRTEA system contains five mouse strains with knockout of the mouse F9 gene and (i) a human wild-type minigene, (ii) no human minigene, (iii) a human R338X minigene, (iv) a human R29X minigene, or (v) a human G381E minigene. The nonsense and missense mutations are known to cause HB in humans.

Table 1.

Five mouse models used for direct quantitative assessment of TBT efficacy

Mouse model	F9 genotype		Hemophilia phenotype	Drug response
	Murine endogenous gene	Human transgene		Gentamicin	Geneticin
WT-HB (WT-7)*	No	Yes	No	No	No
HB	No	No	Yes	No	No
R338X (LS-64)†	No	Yes	Yes	Yes	Dramatic
R29X (ES-13)‡	No	Yes	Yes	No	No
G381E (cCH-6)§	No	Yes	Yes	No	No

All mouse models do not have a functional murine F9 gene.

*Transgenic for the human wild-type F9 gene.

^†Late stop codon.

^‡Early stop codon.

^§Missense mutation.

Human F9 cDNA (or variants thereof) is interrupted by a 300-bp segment of intron 1. Expression is driven by the murine transthyretin promoter, and a simian virus 40 polyadenylation signal is added at the 3′ end as described (16). mRNA levels in R338X and R29X are 18% and 1.4% of a wild-type human F9 minigene, respectively (16).

To demonstrate that the observed reduction in steady-state levels of human F9 R29X mRNA is caused by NMD, transient transfection assays were performed in HepG2 cells with the same constructs used for the generation of the transgenic mice with WT-HB (wild type) and R29X human minigene models. Consistent with previous in vivo data (16), real-time PCR quantification of total relative human F9 mRNA levels demonstrated that R29X exhibited ≈7% of wild-type mRNA levels (P < 0.05; Fig. 1A, solid bars).

Fig. 1.

Low steady-state mRNA levels for R29X are caused by NMD. (A) Real-time PCR quantification of total relative human F9 mRNA levels demonstrated that R29X exhibited ≈7% of wild-type mRNA levels (*, P < 0.05, vs. WT) but were restored above normal when emetine (an NMD inhibitor) was used. (B) Rates of transcription for WT and R29X [expressed as a ratio of RNA levels after transcription in the presence (+rNTP) or absence (−rNTP) of rNTPs] were similar, indicating that changes in stability rather than the rate of transcription are responsible for the apparent reduction in human F9 steady-state mRNA levels for R29X and R338X. All experimental values were derived from at least three independent experiments. Mock-transfected HepG2 cells showed negligible levels of human F9 mRNA (data not shown).

In contrast, with the NMD inhibitor emetine (17, 18), levels of steady-state human F9 mRNA were observed for R29X that were indistinguishable from wild type (P > 0.6; Fig. 1A, open bars), indicating that inhibition of NMD by emetine resulted in restoration of R29X steady-state mRNA levels. To further confirm that the substantially lower steady-state levels of R29X human F9 mRNA are caused by changes in stability rather than a reduced rate of transcription, nuclear run-on assays using nuclei from the transfected HepG2 cells were performed. Using real-time quantitative PCR (qPCR), the levels of human F9 mRNA were measured after transcription in the presence of ribonucleotides (rNTPs; +rNTP reaction), and the basal human F9 mRNA levels (in the absence of rNTPs; −rNTP reaction) in the nuclei of the transfected HepG2 cells were measured. This process allowed the determination of the rate of transcription (fold increase), expressed as a ratio of (+rNTP)/(−rNTP). The data showed that the ratio of (+rNTP)/(−rNTP) was not statistically different between R29X constructs and wild type (P > 0.3; Fig. 1B), indicating that R29X was transcribed in a similar fashion with the wild type (in the presence of rNTPs). Collectively, these results are consistent with a hypothesis that the difference in steady-state mRNA levels in the R29X mice is caused by NMD of the message and not by changes in transcription.

Translational Bypass Therapy: Gentamicin.

Gentamicin is the prototypic drug for TBT. The half-life of human F9 in mouse plasma is ≈2.2 h (19). Gentamicin has a half-life of <2 h in mouse plasma (6, 10). Thus, a TBT response from this drug was expected to peak within a few hours and decay rapidly thereafter. The response to gentamicin was small, but significant, in the R338X model. The maximal effect of gentamicin treatment was an antigen level in R338X mice of 2.82 ± 0.42 ng/100 μl versus 2.06 ± 0.16 ng/100 μl in controls (Wilcoxon–Mann–Whitney test, P = 0.006) (Table 2). No significant increase of either antigen level or coagulant activity was observed in R29X or G381E mice.

Table 2.

F9 antigen level and coagulant activity after gentamicin treatment (34 mg/kg s.c.) after 2 days of pretreatment

Mice	F9 antigen, ng/100 μl	F9 coagulant activity, %
R338X (n = 6)	2.82 ± 0.42*	3.38 ± 0.82†
R29X (n = 9)	1.85 ± 0.15	2.73 ± 0.56
HB (n = 4)	2.06 ± 0.16	2.6 ± 0.55

Untreated R338X mice, treated and untreated HB mice, and treated and untreated R29X mice have indistinguishable baseline F9 antigen and F9 coagulant activities. Time elapsed was 3, 9, and 48 h after last treatment. No differences were observed in antigen level or coagulant activity between these three time points.

*P = 0.0017, R338X antigen level compared with R29X antigen level.

^†R338X, R29X, and HB coagulant activities are not significantly different.

Gentamicin dosages were varied from 17 to 170 mg/kg (Table 3). No dramatic dosage effect was found.

Table 3.

Translational bypass of R338X and gentamicin dosage with 2 days of pretreatment

Gentamicin dosage, mg/kg per day	F9 antigen level, ng/100 μl		F9 coagulant activity, %, 9 h
	2 h	9 h
0	2.25 ± 0.01 (3)	2.21 ± 0.18 (4)	2.5 ± 0.32 (4)
17	2.33 ± 0.06 (3)	2.78 ± 0.42 (2)	3.25 ± 0.64 (2)
34	2.38 ± 0.04 (3)	2.66 ± 0.59 (2)	3.65 ± 1.2 (2)
68	2.31 ± 0.04 (3)	3.2 ± 0.2 (2)	3.3 ± 0.28 (2)
170	2.43 ± 0.12 (3)	3.19 ± 0.07 (2)	3.15 ± 0.07 (2)

The numbers in parentheses indicate the number of mice. Time indicated is time elapsed after last treatment.

Translational Bypass Therapy: Geneticin.

Dramatic increases in antigen and coagulant activity were observed with geneticin-treated R338X mice, whereas there was no significant increase in the R29X or G381E mice. Lack of response in the R29X mice may reflect NMD and/or sequence context effects. Geneticin dosages were varied from 7 to 42 mg/kg, and the maximal response occurred in R338X mice at a dosage of 28 mg/kg (Table 4).

Table 4.

Translation bypass of R338X and geneticin dosage

Geneticin dosage, mg/kg per day	Time point after 2 days of pretreatment, h	F9 antigen, ng/100 μl	F9 coagulant activity, %
0	2	1.71 ± 0.21 (2)	3.45 ± 0.21 (2)
7	2	2.28 ± 0.68 (2)	4.55 ± 0.64 (2)
14	2	5.79 ± 4.1 (4)	6.18 ± 2.48 (4)
28	2	8.75 ± 5.53 (19)	6.94 ± 3.11 (16)
28	9	27.28 ± 30.03 (23)*	8.03 ± 5.56 (16)
42	9	13.57 ± 13.72 (4)	4.98 ± 2.58 (4)

The time points range from 1.5 to 3 h with an average of 2 h. The numbers in parentheses indicate the number of mice.

*P = 0.011 Wilcoxon-Mann-Whitney test compared with F9 antigen 2 h after 28 mg/kg per day of treatment.

Surprisingly, after two or seven daily injections of 28 mg/kg, the peak response in R338X mice occurred 9–48 h after the last injection (Fig. 2).

Fig. 2.

F9 antigen and coagulant after R338X and R29X mice were pretreated for 2 days with 28 mg/kg per day s.c. geneticin. Substantial interindividual variation occurs with peak responses in R338X 9–48 h after the last injection. Control mice are either HB or R338X or R29X mice without geneticin. These values are similar. (A) Logarithm of antigen responses after two daily treatments. R338X, black dots; R29X, green diamonds; HB control, red triangles; WTHB, blue cubes. The numbers between 100 and 1,000 are antigen levels observed in R338X. (B) R338X F9 antigen and coagulant activity after 7 days of single daily injections of geneticin. Antigen (ng/100 μl), blue diamonds; coagulant activity (%), pink squares.

Peak response is on average ≈5% of the wild-type human positive control (WT-HB). Large interindividual variation in antigen was found (Fig. 2A), as reported (10, 11). Individual mice tend to have consistently lower or higher responses over time. Herein the total number of blood samples at each time point is: 1.5 h, 5; 3 h, 15; 4.5 h, 7; 9 h, 23; 24 h, 17; 48 h, 9; 72 h, 5; 6 days, 3; 2 weeks, 7; 3 weeks, 10. It is of note that seven mice in the study had responses consistently 2-fold or higher than the average and 15 mice had responses consistently 50% higher than the average.

At 9 h, the average increase in F9 antigen was 24-fold over background, and the coagulant activity, which had a higher residual background, was elevated 4.8-fold. Interestingly, the antigen level at 9 h was significantly higher than at 3 h after the last (second) injection (Fig. 2A; P = 0.035, Wilcoxon–Mann–Whitney test). Residual antigen could be detected at 3 weeks, and residual coagulant could be detected at 6 days. Even after seven consecutive daily injections, antigen and coagulant were enhanced at 9 h relative to 3 h after the last injection (43.9 ± 18.7 vs. 22.8 ± 14.4 ng/100 μl) (Fig. 2B). Relative to background, antigen at 9 h was elevated 24-fold, whereas coagulant at 9 h was elevated only 2.6-fold, reflecting a higher endogenous background and/or some antigen-positive molecules with bypassed amino acids that have reduced or no activity.

Our study provides a direct comparison of the efficacy of geneticin and gentamicin for in vivo TBT. The results are consistent with previous in vitro data that geneticin provides more efficient bypass than gentamicin (6, 7, 20–22). However, this in vitro finding is not universal: Xi et al. (9) found little, if any, significant difference between geneticin and gentamicin when ochre, amber, and opal nonsense mutations were compared in vitro.

Pharmacokinetic Behavior of Geneticin in Hemophilia Mouse Plasma and Liver.

Geneticin concentration in plasma and liver were determined by using an HPLC-tandem mass spectrometric assay developed and validated in the Analytical Pharmacology Core Facility at the City of Hope National Medical Center (Fig. 3).

Fig. 3.

The half-lives of geneticin in HB mouse model plasma samples (A) and liver samples (B). Drug concentrations were measured by tandem mass spectrometry. Samples were initially purified on HPLC. The sensitivity is ≈1,000-fold greater than assays using antibodies (6).

Aminoglycosides, including geneticin and gentamicin, are hydrophilic drugs that are mostly excreted unchanged. The initial half-life of geneticin in the inbred mouse strain (C57BL/6J) is shorter (15 min) (Fig. 3A) than reported for 129/SvJ (6). The elimination half-life of geneticin is 80 h in the liver, the organ in which F9 is produced (Fig. 3B).

The Toxicity of Geneticin.

There were no obvious adverse effects of treatment. Activity levels, food consumption, fur consistency, and a crude hearing test (response to faint clapping) were similar to untreated controls. Gross and microscopic examination of tissues from killed sentinel mice did not reveal any differences between treated mice and controls. Microscopic analysis of the kidneys and liver did not reveal any abnormality. In addition, the LD₅₀ of s.c. geneticin injection in C57BL/6J hemophilia mice used in this study was determined to be 1,300 mg/kg (data not shown), implying that the observed responses at 28 mg/kg per day occur at a reasonable safety threshold.

Caveats.

Although the endogenous mouse F9 promoter may be more physiological for the human F9 minigene, a strong promoter was chosen because the goal is a quantitative assay system with enzymatic activity and antigen that can be assayed from small samples of blood containing a protein with rapid turnover. The chosen transthyretin promoter has the advantage of generating relatively abundant mRNA, allowing multiple sampling in an experiment.

The toxicity of geneticin in humans is unknown to our knowledge, but the one available animal study (in beagles) raises concern about toxicity. La Roca et al. (23) reported mouth and genital ulceration in beagles 7–14 days after a single oral dose of 50 mg/kg (2.5% estimates LD₅₀). Also, emesis, loose stools, and decreased food consumption were observed. In our study, the LD₅₀ for s.c. doses in the mouse (1,300 mg/kg in mouse) was roughly similar to the LD₅₀ in beagle, and the optimal response was observed at 1.3% of the LD₅₀ without any obvious toxicity. In cell culture, the LD₅₀ analysis found that geneticin was ≈30- to 60-fold more potent than gentamicin in human fiberblasts, human myoblasts, hybridoma cells, and HepG2 hepatoma cells (24).

It is possible that the toxicity of geneticin may limit its clinical utility. Because geneticin is very similar in structure to gentamicin (25), high-frequency hearing loss and some renal toxicity may be anticipated in at least some patients with subchronic or chronic daily use (25). However, the following requires further evaluation: (i) the toxicity of single-dose geneticin every few days may be lower than the toxicity of daily doses; (ii) the level of toxicity acceptable for a drug that may dramatically ameliorate an untreatable, lethal disease is substantially higher than the dose of an antibiotic used to treat infections in normal individuals when alternative antibiotics are available; (iii) megalin mediates aminoglycoside accumulation; this receptor represents a recently discovered drug target to prevent aminoglycoside induced nephrotoxicity (26); (iv) a potent metabolite of geneticin may be efficacious at low doses with little toxicity; and (v) analogs of geneticin may be of similar efficacy with reduced toxicity.

The mechanisms responsible for the dramatically increased in vivo response to geneticin relative to gentamicin are unclear. This difference may reflect: (i) generally superior efficacy of geneticin in TBT, as suggested by some in vitro studies (6, 7, 20–22, 27); (ii) an idiosyncratic result caused by sequence context effects (27, 28), or (iii) more rapid metabolic inactivation of gentamicin in this mouse strain, limiting the dose achieved within the liver.

Possible Sequence Context Effects.

The R338X mouse model responded dramatically when geneticin was administered, whereas the R29X model did not.

In addition to the effect of NMD, it is possible that flanking sequences play a role in observed differences between R338X and R29X. Both nonsense mutations of R338X and R29X are at UGA, but in different contexts (CTT UGA TCT and GCA UGA GAA, respectively). Transfection studies with reporter constructs suggest that termination efficiency is influenced by the identity of the stop codon (UGA > UAG > UAA) and flanking sequence (29, 30). The strongest sequence context effect results from the nucleotide 3′ of the stop codon (31). At this position, purines create a more effective termination environment than pyrimidines. Approximately 90% of the most highly expressed genes in mammals carry a purine in this position (31). Howard et al. (30) reported that the nucleotide immediately downstream from the stop codon has a significant impact on the efficiency in the order of C > U > A > = G by geneticin, gentamicin, and paromomycin treatment. Manuvakhova et al. (27) also reported that geneticin readthrough efficiency varies by stop codon with UGA > = UAG > UAA. For UAG and UGA, efficiency varies with the 3′ nucleotide with U > C, G > A, and C > A, G > U, respectively (27). Uhlenbeck et al. (28) reported that tetrannucleotide AUGA has >10-fold higher readthrough than AUGC, AUGG, or AUGU.

Implications.

The data herein have implications for development of clinically useful nonsense bypass (15). In addition to testing the in vivo efficacy of nonsense bypass, these mouse models may be helpful for testing inhibitors of NMD, as a combination of geneticin and an inhibitor of NMD may produce a good response in the R29X model.

The enhanced response of R338X to geneticin at 9 h versus 3 h occurs by an unknown mechanism. Shifting of geneticin from different in vivo compartments is possible. An intriguing possibility is the presence of a more active metabolite, perhaps one that competes with geneticin and is therefore unmasked by the decay of geneticin. Additionally, the reason for the large interanimal variation in response is unclear. The mice are 129/SvJ back-crossed to C57BL/6J for seven or eight generations, so individual mice are expected to share ≈99% of their genes. The interanimal variation is unlikely to result from injection technique because a given mouse responds in a similar manner in different experiments (i.e., high responders in one experiment are high responders in subsequent experiments).

TBT has the potential to greatly improve symptoms from any severe genetic disease caused by a nonsense mutation. For hemophilia A/B patients, infusion of either recombinant or plasma-derived clotting factor may be the treatment of choice. However, the treatment is expensive and not generally available in developing countries. As with many severe genetic diseases, 2% or more of normal activity can substantially alleviate symptoms in patients with hemophilia (12). Stable, low-cost, nonsense bypass drugs could provide affordable therapy for families that do not have access to protein replacement therapy.

Additionally, TBT has the potential to benefit patients with cancer predisposition syndromes and other adult-onset genetic deficiencies, by delaying or preventing the onset of disease. TBT may also bypass somatic mutations, with implications for cancer chemotherapy. Finally, demonstrated efficacy of TBT for nonsense mutations may stimulate the development of compounds that bypass other types of mutations, thereby extending the range of TBT.

It may be many years before cure by classical gene therapy becomes a routine practical reality. Meanwhile, one might envision that the identification of a nonsense mutation by a clinical molecular genetic laboratory will be followed by TBT for that specific mutation using a drug chosen from a battery of a few small molecules that are efficacious for the treatment of that specific mutation in any of the 20,000 or so human genes. These drugs may be readily available in developing countries that do not have the medical infrastructure or capital for expensive protein-based therapies.

The interindividual response variation in R338X mice and the lack of response in R29X mice illustrate current limitations of TBT. A thumbnail calculation illustrates the potential of geneticin or analogues: one in 200 infants born annually has a severe genetic deficiency disease. If 1/7 of these babies have a nonsense mutation and even 10% of these respond clinically to geneticin or analogues, then, annually worldwide, 12,000 babies may be helped in their first and subsequent years. This thumbnail calculation suffers from multiple caveats including: (i) humans may not respond in the same manner; (ii) the toxicity of nonsense bypass may be unacceptably high long term; (iii) it is possible that tolerance to the nonsense bypass will develop; (iv) the response for any given nonsense mutation will depend on the activities of the mix of amino acids inserted and affected by the sequence context (31); (v) the response rate could be greater if an inhibitor of NMD was discovered; and (vi) the response rate could be lower if the response of R338X is atypical.

In conclusion, IQSCMaRTEA may be useful for quantitative testing of the in vivo efficacy of improved nonsense bypass drugs by directly measuring enzyme antigen and activity. In addition to testing the in vivo efficacy of nonsense bypass drugs, IQSCMaRTEA may be helpful for testing inhibitors of NMD, as a combination of a nonsense bypass drug, and a NMD inhibitor may produce a good response. The model demonstrates that geneticin produces a significant multiday in vivo response. Geneticin can mediate bypass of the R338X nonsense mutation in TBT mice, achieving measurable human F9 antigen for 3 weeks.

Materials and Methods

Breeding strategy (Fig. 4), animal care, HB mice genotyping, and detection of the pharmacokinetic behavior of geneticin are described in supporting information (SI) Text.

Fig. 4.

Breeding strategy. Transgenic mice (human F9 wild type, R29X, R338X, or G381E) were bred with hemophilia mice (murine F9 knockout) to obtain mice with a disrupted mouse F9 gene and a human F9 transgene. Arrows show human F9 wild type, R29X, R338X, or G381E.

Scheme for Construction of HB Gene Knockout and Human F9 Transgenic Gene Knock-In Mice.

Mouse models of hemophilia were C57B/6J mice with knockout of the murine endogenous F9 gene (HB mice) and transgenic for a human F9 gene with R29X (ES-13, early stop codon), or R338X (LS-64, late stop codon) nonsense changes, G381E (cCH-6) missense change [i.e., a substitution at a generic amino acid known to produce severe hemophilia or the human wild-type gene (WT-HB, WT-7)] (Table 1) (16). Both nonsense mutations of R338X and R29X create UGA stop codons, and their upstream and downstream flanking nucleotide sequences are: R29X, GCA UGA GAA; R338X, CTT UGA TCT. Mouse F9 (X-linked) knockout (HB), R29X, and R338X mice do not produce murine or human F9 protein. Assay background levels are 1.3 ± 0.34 ng/100 μl and 2.84 ± 0.49% for the antigen and coagulation assays, respectively.

The G381E mouse model produces very little human F9 protein, consistent with mutations at the same residue in canine HB (32) and human HB. All four mouse models have severe hemophilia disease (Table 1) and are genotypic and phenotypic models for HB patients. The mouse model hemophilia phenotype shows: hemorrhagic swelling of the top of the feet or of the footpads; pale footpads; spontaneous bleeding; excellent survivorship if no injury occurs; occasional sudden death during growth caused by internal hemorrhage or after normal fighting with cage mates and splenomegaly, indicating anemia caused by blood loss. Wild-type mice show normal coagulant activity.

Cell Culture and Transfection.

HepG2 cells (human hepatocellular carcinoma cell line) were cultured in DMEM-F12 medium supplemented with 10% FBS, 100 units/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine (all from Invitrogen, Carlsbad, CA). For transfection, 1 or 2.5 × 10⁶ cells were transfected and seeded in culture medium or medium containing emetine (25 μg/ml; Sigma, St. Louis, MO) using an electroporator (Amaxa, Gaithersburg, MD) with 5 μg or 10 μg of either plasmid and 1 μg of GFP control vector (pmaxGFP), according to the manufacturer's protocol (H-22 protocol; Amaxa). Six hours posttransfection, cells were harvested, and total RNA was isolated by using the RNeasy Mini kit (Qiagen, Valencia, CA) with DNaseI treatment (Qiagen). Cells treated with emetine were harvested 20 h posttransfection. The RNA concentration was determined spectrophotometrically.

Human F9 mRNA Quantitation by Real-Time qPCR.

Total RNA (40 ng) isolated from transfected cells was reverse-transcribed by random hexamers using the SuperScript III First Strand Synthesis system (Invitrogen). cDNA was then diluted from 10- to 500-fold, and 5 μl was used in a qPCR with gene-specific primers for human F9 (forward, 5′-TTCGATCTACAAAGTTCACCATCTATAAC and reverse, 5′-AAACTGGTCCCTTCCACTTCAG) or primers for human β-actin (Applied Biosystems, Foster City, CA). The probe for human F9 or human β-actin had a 5′-FAM group (human F9 probe; 5-FAM-AATCTCTACCTCCTTCATGGAAGCCAGCA). The qPCR for human F9 contained 1X TaqMan Master Mix (Applied Biosystems) and 900 nM each primer and 200 nM probe, in a total volume of 25 μl. The qPCR for human β-actin contained 1× TaqMan Master Mix and 3 μM each primer and 2 μM probe, in a total volume of 25 μl. The qPCR was performed in triplicate by using an Applied Biosystems SDS-7500 real-time PCR machine. For each gene (human F9 or human β-actin) qPCR, a standard curve was generated within the reverse transcription reaction using human liver total RNA (spanning from 200 to 1 ng starting material; Clontech, Mountain View, CA). The relative quantity of RNA was calculated as described (17), using human β-actin.

PCR-Based Nuclear Run-On Assay.

Nuclei from transfected HepG2 cells were isolated with the Nuclei EZ Prep Nuclei isolation kit (Sigma), according to the manufacturer's protocol. The total nuclei were split into two equal aliquots, and an equal volume of the 2× reaction buffer [containing 10 mM Tris, pH 8/5 mM MgCl₂/0.3 M KCl/5 mM DTT/40 units of RNasin (Promega, Madison, WI)]. Five hundred micromolars of each ribonulceotide (rATP, rGTP, rUTP, and rCTP; Promega) was added to one aliquot (+rNTP reaction), and an equal volume of nuclease-free water was added to the other nuclear aliquot (−rNTP reaction). The transcription reaction was performed for 30 min at 30°C, and the nuclear RNA was isolated with a RNeasy Mini kit (Qiagen) with DNaseI treatment (Qiagen). The RNA concentration was determined spectrophotometrically. Thirty nanograms of RNA was reverse-transcribed, and the human F9 mRNA amount was calculated by using real-time qPCR, as described for total mRNA. The rate of transcription was determined as the ratio of human F9 mRNA after transcription (+rNTP) over the amount of human F9 before transcription (−rNTP), using identical amounts of starting RNA material.

TBT Materials and Method for Injection.

Gentamicin (liquid) was ordered from American Pharmaceutical Partners (Schaumburg, IL), and Geneticin (dissolved in water before use) was from Invitrogen. Mice received a s.c. injection performed with a 0.33-mm × 13-mm gauge needle under dorsal skin in a volume of ≈0.1 ml.

Method for Blood Sampling and Processing.

Mice were warmed under a heat lamp for 10 min to facilitate blood extraction. Blood was taken after tail clipping and was anticoagulated with 0.105 M buffered trisodium citrate at a ratio of 1:9 (anticoagulant blood). Three hundred microliters of blood was typically obtained. After centrifugation at 10,000 × g for 10 min, the platelet-poor plasma was removed. Plasma was stored at −70°C until testing was performed. Thirty microliters of plasma was used for antigen-level quantification. The remaining plasma was used for quantitating coagulant activities or drug levels. Coagulant activity was carried out in a central reference laboratory after transfer of plasma on dry ice. Testing was performed in such a way that the technologist was “blinded” to the identity of individual samples.

Detection of F9 Antigen and Coagulant Activity.

F9 antigen levels were determined by ELISA (Human F9 ELISA Lot F9 EIA kit; Enzyme Research Laboratories, South Bend, IN) according to the manufacturer's protocol, using a human standard curve (Enzyme Research Laboratories).

F9 coagulant activity of mouse plasma was detected as a percentage with one-stage clotting assays by using automated activated partial thromboplastin time (aPTT) from Dade Behring (Deerfield, IL; Sysmex CA-1500 Coagulation Analyzer). The percent activity of a clotting factor was determined by the degree of correction of the aPTT when a diluted plasma sample was mixed with a known factor deficient plasma. Normal pooled human plasma was used as a standard and is considered to give 100% correction. Reagents included the following: Dade Behring aPTT reagent; Dade Behring calcium chloride, 0.02 M; Dade Behring standard human plasma ORKL13; Dade Behring factor-deficient plasma (depending on assay); Dade Behring CA System Buffer, and preservative-free distilled or deionized water. The sensitivity of the one-stage assays was 0.01 unit/ml (1%). Samples were not diluted.

Monitoring Drug Toxicity.

Before and after drug injection, activity, food consumption, and gross hearing were tested. Sentinel mice, including those treated in more than one experiment, were killed for gross and microscopic analysis.

Pharmacokinetic Behavior of 28 mg/kg Geneticin in Plasma and Liver of Hemophilia Mice.

The mouse models used were R338X, R29X, and HB. Mice were treated with daily s.c. injections of 28 mg/kg for 2 days, then according to the experiment time points, plasma was extracted, and the liver was harvested at death. Geneticin plasma and liver levels were determined by using an HPLC-tandem mass spectrometric assay developed and validated in the Analytical Pharmacology Core Facility at the City of Hope National Medical Center. An assay with a sensitivity of 10 nM was developed.

Statistics.

The responses at different time points were compared by using the Wilcoxon–Mann–Whitney Test implemented by StatXact (CYTEL Software, Cambridge, MA). The Wilcoxon–Mann–Whitney Test is an exact test for differences in mean that is almost (96%) as powerful as ANOVA when the ANOVA assumptions of normality and homogeneity of variances are met, but does not require those assumptions. ANOVA was typically not valid for these data because of nonhomogeneity of variances.

The data from human F9 mRNA quantitation by real-time qPCR and PCR-based nuclear run-on assay were analyzed with InStat 3.0 software (GraphPad, San Diego), using an unpaired, two-tailed t test with or without the Welch correction.

Abbreviations

TBT

translational bypass therapy

IQSCMaRTEA

in vivo quantitative stop codon management repli-sampling TBT efficacy assay

NMD

nonsense-mediated decay

qPCR

quantitative PCR

rNTP

ribonucleotide

HB

hemophilia B.

Note Added in Proof:

The recent developments of massively parallel sequencing and gene identification by NMD inhibition (GINI) can efficiently identify nonsense mutations in tumors (33), thereby enhancing the potential of TBT as a novel approach to cancer chemotherapy.