Dystrophin restoration in skeletal, heart and skin arrector pili ... - Nature

Gene Therapy (2010) 17, 432–438 & 2010 Macmillan Publishers Limited All rights reserved 0969-7128/10 $32.00 www.nature.com/gt

SHORT COMMUNICATION

Dystrophin restoration in skeletal, heart and skin arrector pili smooth muscle of mdx mice by ZM2 NP–AON complexes A Ferlini1, P Sabatelli1,2, M Fabris1, E Bassi1, S Falzarano1, G Vattemi3, D Perrone4, F Gualandi1, NM Maraldi5, L Merlini1, K Sparnacci6, M Laus6, A Caputo7, P Bonaldo7, P Braghetta7 and P Rimessi1 1

Department of Experimental and Diagnostic Medicine, Section of Medical Genetics, University of Ferrara, Ferrara, Italy; 2IGM-CNR, Unit of Bologna c/o IOR, Bologna, Italy; 3Department of Neurological Sciences and Vision, Section of Clinical Neurology, University of Verona, Verona, Italy; 4Department of Biology and Evolution, University of Ferrara, Ferrara, Italy; 5Department of Human Anatomical Sciences, University of Bologna and Laboratory of Cell Biology, IOR, Bologna, Italy; 6Department of Environmental and Life Sciences INSTM, University of Eastern Piedmont, Alessandria, Italy and 7Department of Histology, Microbiology, and Medical Biotechnology, University of Padua, Padua, Italy

Potentially viable therapeutic approaches for Duchenne muscular dystrophy (DMD) are now within reach. Indeed, clinical trials are currently under way. Two crucial aspects still need to be addressed: maximizing therapeutic efficacy and identifying appropriate and sensible outcome measures. Nevertheless, the end point of these trials remains painful muscle biopsy to show and quantify protein restoration in treated boys. In this study we show that PMMA/N-isopropilacrylamide+ (NIPAM) nanoparticles (ZM2) bind and convey antisense oligoribonucleotides (AONs) very efficiently. Systemic injection of the ZM2–AON complex restored dystrophin protein synthesis in both skeletal and cardiac muscles of mdx mice, allowing protein localization in up to 40% of muscle fibers. The mdx exon 23 skipping level was up to 20%, as measured by the RealTime assay, and dystrophin restoration was confirmed by both reverse transcription-PCR and

western blotting. Furthermore, we verified that dystrophin restoration also occurs in the smooth muscle cells of the dorsal skin arrector pili, an easily accessible histological structure, in ZM2–AON-treated mdx mice, with respect to untreated animals. This finding reveals arrector pili smooth muscle to be an appealing biomarker candidate and a novel low-invasive treatment end point. Furthermore, this marker would also be suitable for subsequent monitoring of the therapeutic effects in DMD patients. In addition, we demonstrate herein the expression of other sarcolemma proteins such as a-, b-, g- and d-sarcoglycans in the human skin arrector pili smooth muscle, thereby showing the potential of this muscle as a biomarker for other muscular dystrophies currently or soon to be the object of clinical trials. Gene Therapy (2010) 17, 432–438; doi:10.1038/gt.2009.145; published online 12 November 2009

Keywords: dystrophin; antisense; nanoparticles; arrector pili; smooth muscle

Duchenne muscular dystrophy (DMD) is an inherited X-linked degenerative muscular disorder predominantly caused by frame-disrupting mutations arising from gross rearrangements in the dystrophin gene.1 DMD patients are affected by both severe skeletal muscle wasting and dilated cardiomyopathy. The milder allelic form of the disease, Becker muscular dystrophy, is due to in-frame mutations that preserve a shortened but functional protein. Recently, novel therapeutic approaches for DMD have emerged, some of which have thus far only been tested in animal models. Others, however, such as drugs which induce stop-codon reversion2 (PTC124) and antisense oligoribonucleotide (AON)-mediated exon skipping,3–5 Correspondence: Professor A Ferlini, Department of Experimental and Diagnostic Medicine, Section of Medical Genetics, University of Ferrara, Via Fossato di Mortara, 74, Ferrara, FE 44100, Italy. E-mail: [email protected] Received 27 July 2009; revised 23 September 2009; accepted 25 September 2009; published online 12 November 2009

are already undergoing clinical trials.6 These novel therapeutic options for DMD represent the first step toward a possible cure for this devastating disorder. However, several major obstacles, namely optimization of nontoxic effective doses, improvement of the delivery system, distribution to all affected tissues, achievement of a sustainable therapeutic effect and development of an administration strategy suitable for lifelong treatment,7–10 still remain to be overcome. We previously showed that inert PMMA T1 nanoparticles (NPs) conjugated with AON are capable of inducing dystrophin restoration in body-wide muscles in the mdx animal model.11 Although encouraging in terms of dystrophin-positive fiber percentage, indeed we obtained a functional effect with one-eightieth of the normal AON dose regimen, the efficiency of this novel compound was relatively unsatisfactory. Therefore, to obtain a more efficient compound, we designed and prepared a novel type of cationic core-shell NP (termed ZM2), made up of a predominantly PMMA core and a random copolymer shell consisting of units derived from

Nanoparticle–AON dystrophin rescue A Ferlini et al

N-isopropil-acrylamide+ (NIPAM) and reactive methacrylate-bearing cationic groups (Supplementary Figure S1, a and b). These novel NPs were found to be nontoxic in vitro, even following repeated administrations. The cytotoxicity of ZM2 NP was assayed in HeLa cells following incubation with increasing amounts of NPs (see Supplementary Materials and methods and Supplementary Figure S2), and no significant reduction in cell viability was observed up to 1 mg ml1, as compared with untreated cells (P40.05). Moreover, adsorption of the M23D (+0718) AON 20 O-methyl full-length phosphorothioate (20 OMePS)11 onto ZM2 NPs proved to be a highly reproducible process at a loading value of 90 mg of AON per mg of NPs (see Supplementary Materials and methods, and Result). We initially divided 6-week-old mdx mice into two groups: group 1 (four animals) received 225 mg intraperitoneal injections of naked M23D AON (naked AON)

and group 2 (four animals) received intraperitoneal injections of 225 mg of M23D AON conjugated with 2.5 mg of ZM2 NPs. A further four mdx mice were injected with 2.5 mg of ZM2 NPs dissolved in sterile unpreserved saline solution (0.9% sodium chloride); four non-injected mdx mice were used as controls. The animals were injected weekly for 7 weeks (7.5 mg kg1 per injection) and killed 1 week after the final administration. The total amount of M23D AON received by each animal was 52.5 mg kg1 (1575 mg per animal). Immunohistochemical analysis subsequently demonstrated that the ZM2–AON complex induces dystrophin restoration very efficiently in the skeletal muscles of the quadriceps, gastrocnemius and diaphragm as well as in the cardiac muscle (Figure 1). Seven-micron-thick frozen transverse sections of skeletal and cardiac muscles from untreated, naked AON-treated, ZM2–AON-treated mdx mice and wild-type mice were labeled with a rabbit

433

Figure 1 Immunohistochemical findings in mdx skeletal and cardiac muscles. At 1 week after the seventh injection, the 4 mdx mice from groups 1 and 2 were killed; the quadriceps, diaphragm, gastrocnemius and heart were isolated, blotted dry, trimmed of external tendon, snap-frozen in liquid N2-cooled isopentane, and stored at 80 1C until further processing. As controls, four untreated mdx mice and four mdx treated with ZM2-NP alone were also killed. All immunohistochemical procedures were carried out as previously described.11 Dystrophin appears to be properly expressed at the membrane of muscle fibers of ZM2–AON (M23D)-treated mdx mice. The intensity and localization of labeling in rescued dystrophin-positive muscle fibers is reminiscent of normal muscle. Dystrophin traces are also visible in mice treated with the same dose of naked AON. No dystrophin labeling was detected at the sarcolemma in the mdx either untreated or treated only with ZM2NP. Bars indicate the magnifications. Gene Therapy


434

polyclonal anti-dystrophin antibody (H-300, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and revealed with an anti-rabbit fluorescein isothiocyanate-conjugated secondary antibody (Dako, Glostrup, Denmark). All samples were double-labeled with a rat monoclonal antibody for laminin-2 (a2-chain, 4H8-2 Alexis Biochemical, Farmingdale, NY, USA), followed by anti-rat tetramethylrhodamino isothiocyanate-conjugated antibody (Dako) as internal control. As counting positive fibers is considered a fundamental parameter for evaluating the efficiency of dystrophin restoration, we compared two independent methods to provide a more precise quantification of dystrophin positivity in muscles. The first method used was a manual system, the AnalySIS program (Soft Imaging System, Muenster, Germany),11 which counts dystrophin-positive fibers with a labeling that covers at least 50% of the sarcolemma perimeter. The second method was a novel, semiautomated and semiquantitative analysis system, which calculates the ratio between dystrophin and laminin a-2-chain-positive areas in double-labeled samples using novel software developed in cooperation with Nikon (NIS-Elements 3.0 AR imaging program, Nikon, Firenze, Italy). The mean value of the dystrophin/laminin a2-chain (fluorescein isothiocyanate/tetramethylrhodamino isothiocyanate) area ratio

was calculated following subtraction of the background. To provide a graphical representation of the above, mean values for the fluorescein isothiocyanate/tetramethylrhodamino isothiocyanate area ratio of treated and untreated mdx mice were normalized to wild-type values and expressed as a percentage of dystrophin-positive fibers over the total number of fibers (laminin-a2 positive) (Figure 2a and Supplementary Figure S1). The two different counting approaches gave quite similar results. In fact, dystrophin positivity in ZM2– AON-treated mdx mice was quantified as 18% by manual count and 16% by the semiautomated system in the diaphragm, respectively, 26 and 19% in the gastrocnemius, 40 and 35% in the quadriceps and 3 and 6% in the heart (Figure 2a and Supplementary Figure S1). Dystrophin rescue was found to be significantly improved in mice treated with ZM2–AON complex with respect to those treated with naked AON, which showed a maximum of 11–13% dystrophin-positive fibers in the quadriceps (Figure 2a and Supplementary Figure S1). Exon-specific RealTime PCR assays revealed an exon 23-skipping percentage of 4–21% in skeletal muscles and 3% in the heart (Figure 2b): double the % detected by using the T1-AON compound in skeletal muscles, but not in the heart.11 Exon 23 skipping was confirmed by reverse transcription-PCR in all muscles except for the

Figure 2 (a) Immunohistochemical quantification of dystrophin-positive fibers in skeletal muscles (quadriceps, gastrocnemius, diaphragm) and heart of ZM2–AON- and naked AON-treated mdx mice evaluated by manual counting. Bars represent the percentage of dystrophinpositive fibers in ZM2–AON- (black bars) and naked AON-treated (gray bars) mouse muscles. At least 5000 muscle fibers from each muscle, obtained from three different levels of tissue blocks for each sample, were studied; data were analyzed using the Mann–Whitney test and the significance was calculated versus untreated mdx mice. Asterisks indicate statistically significant Po0.05. (b) RealTime reverse transcription (RT)-PCR analysis (ESRA) for detecting exon 23 skipping. Percentage of exon 23 skipping (y axis) was determined by RealTime PCR in cardiac and skeletal muscles (quadriceps, gastrocnemius, diaphragm, x axis) from treated mdx animals in comparison with untreated mice. Histograms represent the percentage of exon 23-skipped transcript (2DDCT values) calibrated on exons 22 and 25, and normalized to wildtype (WT) mice. Black bars: naked AON-treated mice; gray bars: ZM2–AON-treated mice; white bars: mdx mice. (c) Nested RT-PCR analysis of dystrophin mRNA from the quadriceps, diaphragms, gastrocnemius and hearts of animals in groups 1 and 2 and untreated mice (NT). Primary and nested PCRs were performed as previously described.11 Full-length dystrophin transcript was detected in all cDNA samples, and a faint shorter transcript arising from AON-induced exon 23 skipping was observed in all skeletal muscles but not the heart of ZM2– AON-treated animals and in diaphragm of naked-AON-treated mice. M: molecular Weight Marker IX (Roche, Indianapolis, IN, USA). (d and e) Dystrophin immunoblot using DYS2 antibody. For WT sample, the total protein loaded was 1/15 of the quantity in the other lanes. (d) Restored expression of the protein in quadriceps muscle from ZM2–AON- and naked AON-treated mice. Quadriceps from WT and untreated mdx mice were used as positive and negative controls, respectively. (e) dystrophin protein is clearly visible in the diaphragm from both naked AON- and ZM2–AON-treated mice and was also detectable, though faintly, in the heart of ZM2–AON-treated mice. WT mouse diaphragm was used as positive control. Gene Therapy


heart, as expected considering the small amount of skipping identified by exon-specific RealTime PCR assay (Figure 2c). Western blotting analysis showed the presence of high molecular weight dystrophin protein in skeletal muscles (quadriceps and diaphragm) from ZM2–AON-treated mdx mice (Figures 2d and e). As expected from the immunohistochemical and transcriptional data, dystrophin protein in the heart is visible, albeit poorly, under western

blot (Figure 2e). Interestingly, a lower but detectable dystrophin restoration was clearly visible under immunohistochemical analysis (Figure 1), although poorly so by western blotting (Figures 2d and e), at the same dose of naked AON. Moreover, the total amount (52.5 mg kg1) of AON injected as ZM2–AON was reduced in comparison with previous studies on naked AON delivery (120–600 mg kg1).7,8

435

Figure 3 Immunolabeling of the smooth muscle arrector pili in treated and untreated mdx mice. The sections of skin biopsies taken from the region of the scruff of the neck of wild-type (WT), ZM2–AON-treated mdx and untreated mdx mice were labeled with anti-dystrophin antibody (rabbit polyclonal, Santa Cruz Biotechnology) and incubated with a tetramethylrhodamino isothiocyanate-conjugated secondary antibody. All samples were double-labeled with an anti-desmin antibody (goat polyclonal 1:10, Santa Cruz Biotechnology) and incubated with anti-goat fluorescein isothiocyanate-conjugated antibody (1:100, Sigma, St Louis, MO, USA). All samples were observed with a Nikon Eclipse 80i fluorescence microscope. Dystrophin (red) is clearly visible in the arrector pili smooth muscle of WT mice, absent in untreated mdx, and rescued in treated mdx. High magnification of treated mdx arrector pili shows an intense and homogeneous labeling of smooth muscle cells (lower panel). Desmin (green) colocalizes with dystrophin and was expressed in all animals. Bar, 200 mm. Gene Therapy


436

Regarding PMO–AONs, optimization of naked AON dose regimens has previously been addressed, and some dystrophin restoration was detected following intravenous administration of a very low dose of PMO–AONs (total of 20 mg of PMO per kg).12 Although PMO and 20 OMePS molecules possess different kinetics, and studies on specific molecules will be required to optimize dose regimens, our data support the fact that even with low doses of 20 OMePS, AON dystrophin restoration can be achieved. Subsequently, we tested whether dystrophin restoration in treated animals could be detected in the smooth muscles of the skin arrector pili, as dystrophin expression in human arrector pili smooth muscle has been previously documented and found to be absent in DMD patients,13–15 although it had never been studied in the mdx animal model. We demonstrated that dystrophin is indeed expressed in the arrector pili smooth muscle of wild-type mice but absent in the mdx, and that ZM2– AON-treated mdx mice clearly show protein rescue in this smooth muscle (Figure 3), unlike mice treated with an identical dose of naked AON. Furthermore, we confirmed the expression of dystrophin in human arrector pili from a healthy subject, as well as its absence in a DMD patient (Figure 4a). In addition, we explored whether other sarcolemma proteins, namely a-, b-, d- and g-sarcoglycans, laminin-a2-chain, dysferlin and caveolin3 were expressed in the arrector pili smooth muscle of human skin. Accordingly, all skin samples were doublelabeled with anti-desmin antibody (goat polyclonal, Santa Cruz Biotechnology) and revealed with an antigoat fluorescein isothiocyanate-conjugated antibody. All sarcoglycans were found to be strongly expressed and detectable in control human skin arrector pili smooth muscle (Figure 4b), whereas the other sarcolemma proteins analyzed (dysferlin, caveolin-3 and laminin-a2chain) were not detected (data not shown). Although expression of sarcoglycans in smooth muscle has been well documented, especially in blood vessels,7 their expression in the smooth muscle of skin arrector pili has not previously been reported. The evidence suggests that this novel ZM2–AON compound greatly enhances dystrophin restoration in skeletal muscles and slightly enhances it in the cardiac muscles of mdx mice, thereby confirming that these NPs represent a very promising vehicle for systemic delivery

Figure 4 Immunolabeling of dystrophin on normal human and Duchenne muscular dystrophy (DMD) skin smooth muscle arrector pili. (a) The sections of skin biopsies from a healthy human subject and the DMD patient were labeled with the anti-dystrophin antibody (rabbit polyclonal, Santa Cruz Biotechnology) and incubated with tetramethylrhodamino isothiocyanate (TRITC)-conjugated secondary antibody. To identify smooth muscle cells of arrector pili, samples were double labeled with an anti-desmin antibody and revealed with a fluorescein isothiocyanate (FITC)conjugated secondary antibody. Dystrophin was intensely expressed at the sarcolemma of smooth muscle cells of the arrector pili of the healthy subject, but absent in the DMD patient. (b) Immunolabeling of a-, b-, g- and d-sarcoglycans in normal human skin. Anti-a-, b-, g- and d-sarcoglycan antibodies (1:5, Novocastra, NewCastle upon Tyne, UK) were revealed with a TRITC-conjugated antibody. All samples were double-labeled with an anti-desmin antibody and a FITC-conjugated antibody (1:100, Sigma). Smooth muscle cells of the arrector pili, identified by desmin labeling, show an intense labeling with anti-a and -b antibodies. Bar, 20 mm. Gene Therapy

of AONs. The observed dystrophin rescue was due to a high skipping percentage (10–20%), remarkable when one considers the low AON dose delivered, and to the expression of high molecular weight dystrophin. The new dystrophin protein was correctly localized at the


sarcolemma in up to 40% of the fibers, as detected by immunohistochemical analysis. The combination of slow release and depot effects, together with protection from degradation/sequestration and the high AON-binding capacity of this novel ZM2NP could be responsible for its efficacy and efficiency. A possible explanation of this depot-release behavior could be provided by considering the expanded nature of the shell in water, as AON species form ion pairs with cationic groups placed within the outer layer or well inside the shell. Thus, a fast release kinetic could correspond to the release of the AON species adsorbed onto the outer shell, whereas the slow desorption process may be associated to their release from the inner part of the shell. Hence, the possibility of varying shell chain density by appropriately adjusting the synthetic parameters of the NP could provide an additional tool for fine-tuning the depot characteristics of the AON–NP complex. Remarkably, we obtained dystrophin restoration with a very low dose of naked 20 OMePS AON. Optimal dose regimens of AONs for therapeutic use and their lowest nontoxic effective dose are still being debated,5,9,11,12 but, interestingly our results suggest that lower doses of naked 20 OMePS AONs administered for longer periods (in our study 7 weeks) might be sufficient to induce low but detectable dystrophin restoration, similar to that reported for PMOs, at least in the mdx animal model, thereby prompting further research in this direction. NPs are thus confirmed as a very promising delivery system for AONs, especially due to the fact that, intriguingly, the skin arrector pili muscle cells were shown to be susceptible to exon skipping and dystrophin restoration in ZM2–AON-treated mice. Therefore, if confirmed for other types of treatment and with different dose regimens, skin smooth muscle could represent a very promising, less invasive enrichment or even surrogate end point for verifying dystrophin restoration and monitoring the efficacy of treatment, not only in animal models but also in clinical trials.16–19 This would consequently facilitate approval procedures for the latter by greatly alleviating the negative impact that compulsory muscle biopsies have on perception of treatment among patients and their families. In addition, arrector pili smooth muscle could be considered as a lowinvasive biomarker in other muscular dystrophies such as sarcoglycanopathies. In summary, our results show that the ZM2–AON complex effectively restores dystrophin synthesis in striated muscles at low doses of AON. Further studies into the half-life, elimination and administration of NPs are required to evaluate their eligibility as innovative nontoxic delivery systems for AON-mediated therapy in DMD.

Conflict of interest The authors declare no conflict of interest.

Acknowledgements The Telethon Italy grants GGP05115, GUP07011 and GGP09093 (to AF) are acknowledged. Thanks are also due to Prof A Medici (Department of Chemistry,

University of Ferrara) to the Industria Chimica Emiliana (ICE Reggio Emilia) Grant (to AF), and to TREAT-NMD Network of Excellence of EU FP7 n. 036825 (to LM and Telethon-Italy). We are also grateful to the ISS National AIDS Program grants, in support of the Nanoparticle technology platform, awarded to AC, LT and ML. The ZM2–AON compound and its effects have been patented at the University of Ferrara, Industrial Liaison Office, patent IP number TO2009A000782.

437

References 1 Aartsma-Rus A, Van Deutekom JC, Fokkema IF, Van Ommen GJ, Den Dunnen JT. Entries in the Leiden Duchenne muscular dystrophy mutation database: an overview of mutation types and paradoxical cases that confirm the reading-frame rule. Muscle Nerve 2006; 34: 135–144. 2 Hamed SA. Drug evaluation: PTC-124—a potential treatment of cystic fibrosis and Duchenne muscular dystrophy. IDrugs 2006; 9: 783–789. 3 Aartsma-Rus A, van Ommen GJ. Antisense-mediated exon skipping: a versatile tool with therapeutic and research applications. RNA 2007; 13: 1609–1624 (Review). 4 Hoffman E. Skipping toward personalized molecular medicine. N Engl J Med 2007; 357: 2719–2722. 5 Du L, Gatti R. Progress toward therapy with antisense-mediated splicing modulation. Curr Opin Mol Ther 2009; 11: 116–123. 6 van Deutekom J, Janson AA, Ginjaar IB, Frankhuizen WS, Aartsma-Rus A, Bremmer-Bout M et al. Local dystrophin restoration with antisense oligonucleotide PRO 051. N Engl J Med 2007; 357: 2677–2686. 7 Lu Q, Rabinowitz A, Chen YC, Yokota T, Yin H, Alter J et al. Systemic delivery of antisense oligoribonucleotide restores dystrophin expression in body-wide skeletal muscles. Proc Natl Acad Sci USA 2005; 102: 198–203. 8 Alter J, Lou F, Rabinowitz A, Yin H, Rosenfeld J, Wilton SD et al. Systemic delivery of morpholino oligonucleotide restores dystrophin expression bodywide and improves dystrophic pathology. Nat Med 2006; 12: 175–177. 9 Heemskerk H, de Winter CL, de Kimpe SJ, van Kuik-Romeijn P, Heuvelmans N, Platenburg GJ et al. In vivo comparison of 20 -Omethyl phosphorothioate and morpholino antisense oligonucleotides for duchenne muscular dystrophy exon skipping. J Gene Med 2009; 11: 257–266. 10 Popplewell L, Trollet C, Dickson G, Graham IR. Design of phosphorodiamidate morpholino oligomers (PMOs) for the induction of exon skipping of the human DMD gene. Mol Ther 2009; 17: 554–561. 11 Rimessi P, Sabatelli P, Fabris M, Braghetta P, Bassi E, Spitali P et al. Cationic PMMA nanoparticles bind and deliver antisense oligoribonucleotides allowing restoration of dystrophin expression in the mdx mouse. Mol Ther 2009; 17: 820–827. 12 Malerba A, Thorogood FC, Dickson G, Graham IR. Dosing regimen has a significant impact on the efficiency )of morpholino oligomer-induced exon skipping in mdx mice. Hum Gene Ther 2009; 20: 955–965. 13 Marbini A, Gemignani F, Bellanova MF, Guidetti D, Ferrari A. Immunohistochemical localization of utrophin and other cytoskeletal proteins in skin smooth muscle in neuromuscular diseases. J Neurol Sci 1996; 143: 156–160. 14 Marbini A, Marcello N, Bellanova MF, Guidetti D, Ferrari A, Gemignani F. Dystrophin expression in skin biopsy immunohistochemical. Localisation of striated muscle type dystrophin. J Neurol Sci 1995; 129: 29–33. 15 Tanveer N, Sharma MC, Sarkar C, Gulati S, Kalra V, Singh S et al. Diagnostic utility of skin biopsy in dystrophinopathies. Clin Neurol Neurosurg 2009; 111: 496–502. Gene Therapy


438 16 Gleason R, Dye WW, Wilson E, Humphrey JD. Quantification of the mechanical behavior of carotid arteries from wild-type, dystrophin-deficient, and sarcoglycan-delta knockout mice. J Biomech 2008; 41: 3213–3218. 17 Prentice R. Surrogate and mediating endpoints: current status and future directions. J Natl Cancer Inst 2009; 101: 216–217.

18 Mandrekar S, Sargent DJ. Clinical trial designs for predictive biomarker validation: one size does not fit all. J Biopharm Stat 2009; 19: 530–542. 19 McShane L, Hunsberger S, Adjei AA. Effective incorporation of biomarkers into phase II trials. Clin Cancer Res 2009; 15: 1898–1905.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Gene Therapy