ARTICLE Coagulation & its Disorders Ferrata Storti Foundation Variable readthrough responsiveness of nonsense mutations in hemophilia A Lluis Martorell,1,2 Vicente Cortina,3 Rafael Parra,4 Jordi Barquinero1,* and Francisco Vidal2,5,6* 1 Gene and Cell Therapy Laboratory, Vall d'Hebron Institut de Recerca (VHIR), Universitat Autònoma de Barcelona (UAB); 2Congenital Coagulopathies Laboratory, Banc de Sang i Teixits (BST); 3Vall d'Hebron Core Laboratory (Section of Thrombosis and Haemostasis), Hospital Vall d’Hebron; 4Banc de Sang i Teixits-Hospital Vall d’Hebron; 5Molecular Haematologica 2020 Diagnosis and Therapy Unit, VHIR-UAB and 6CIBER de Enfermedades Cardiovasculares Volume 105(2):508-518 (CIBERCV), Barcelona, Spain *JB and FV contributed equally to this work as co-senior authors. ABSTRACT R eadthrough therapy relies on the use of small molecules that enable premature termination codons in mRNA open reading frames to be misinterpreted by the translation machinery, thus allowing the gen- eration of full-length, potentially functional proteins from mRNA carrying nonsense mutations. In patients with hemophilia A, nonsense mutations potentially sensitive to readthrough agents represent approximately 16% of the point mutations. The aim of this study was to measure the readthrough effect of different compounds and to analyze the influence of premature termination codon context in selected nonsense mutations causing hemo- philia A. To this end, primary fibroblasts from three patients with hemo- philia A caused by nonsense mutations (p.W1586X, p.Q1636X and p.R1960X) and Chinese hamster ovary (CHO) cells transfected with 12 dif- ferent plasmids encoding mutated F8 (p.Q462X, p.Q1705X, p.Q1764X, Correspondence: p.W274X, p.W1726X, p.W2015X, p.W2131X, p.R1715X, p.R1822X, p.R1960X, p.R2071X and p.R2228X) were treated with gentamicin, JORDI BARQUINERO/FRANCISCO VIDAL PEREZ geneticin, PTC124, RTC13 or RTC14. Responses were assessed by analyz- jordi.barquinero@vhir.org/fvidal@bst.cat ing not only F8 mRNA expression and FVIII biosynthesis (FVIII antigen by ELISA, western blot and immunofluorescence) but also the FVIII activity Received: November 23, 2018. (by chromogenic assay). In the patients’ fibroblasts, readthrough agents nei- Accepted: June 11, 2019. ther stabilized F8 mRNA nor increased FVIII protein or activity to Pre-published: June 13, 2019. detectable levels. In CHO cells, only in five of the 12 F8 variants, readthrough treatment increased both FVIII antigen and activity levels, which was associated with a reduction in intracellular accumulation of doi:10.3324/haematol.2018.212118 truncated forms and an increase in full-length proteins. These results pro- vide experimental evidence of genetic context dependence of nonsense sup- Check the online version for the most updated pression by readthrough agents and of factors predicting responsiveness. information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/105/2/508 Introduction ©2020 Ferrata Storti Foundation Hemophilia A (HA) is an X-linked disorder caused by molecular defects in the Material published in Haematologica is covered by copyright. coagulation factor VIII gene (F8). Nonsense mutations represent approximately All rights are reserved to the Ferrata Storti Foundation. Use of 16% of point mutations leading HA (http://www.factorviii-db.org). These patients published material is allowed under the following terms and conditions: usually have severe HA and a high risk of FVIII inhibitor development.1 HA predis- https://creativecommons.org/licenses/by-nc/4.0/legalcode. poses patients to recurrent bleeds, primarily into the joints and soft tissues, with Copies of published material are allowed for personal or inter- the severity of the disease inversely correlated to FVIII:C activity. Therapy is based nal use. Sharing published material for non-commercial pur- on the prevention of bleeding using FVIII replacement (either recombinant or plas- poses is subject to the following conditions: ma-derived), with an annual estimated cost of >150,000 euros in severely affected https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for com- patients.2 mercial purposes is not allowed without permission in writing Although major advances have been reported with gene therapy in both HA3 and from the publisher. hemophilia B (HB),4 it has yet to be approved for clinical use and will not, therefore, be commercially available for the majority of patients in the immediate future. Recently, readthrough therapy has emerged as a potential strategy for the treat- 508 haematologica | 2020; 105(2) Variable nonsense suppression in Hemophilia A Table 1. Molecular and clinical data of patients with hemophilia A included in the study. Patient Gene Nucleotide Amino Nucleotide Phenotype FVIII:C % Inhibitors FVIII:Ag % region change acid change context [HGVS] (-1Stop+4) #1 Exon 14 c.4757G>A p.W1586X TTAGG Severe <1 Yes nd #2 Exon 14 c.4906C>T p.Q1636X ATAAA Severe <1 nd nd #3 Exon 18 c.5878C>T p.R1960X TTGAT Moderate 4 No nd #4 Exon 18 c.5879G>A p.R1960Q (missense) Mild 10-15 No nd FVIII:Ag: FVIII antigen levels; FVIII:C: Factor VIII coagulant activity; HGVS: Human Genome Variation Society nomenclature; nd: not determined. ment of hereditary diseases caused by nonsense muta- pating patients and controls provided informed consent in accor- tions. It is based on the use of small molecules that cause dance with the Declaration of Helsinki. The study was approved ribosomes to ignore premature termination codons (PTC) by our institutional Research Ethics Committee. during translation, thus allowing the generation of full- The genetic characteristics of each patient and their plasma length, potentially functional, proteins from the mutant FVIII:C activities at the time of diagnosis are summarized in mRNA. PTC are estimated to occur in approximately 10- Table 1. 15% of the patients with monogenic disease (homo or heterozygotes), who may potentially benefit from Generation of F8 variants harboring premature readthrough agents (RTA). However, this approach has termination codon mutations important limitations, including the toxicities of some of All F8 B-domain deleted (F8BDD) cDNA variants were designed these compounds (which require long-term administra- according to our hypotheses and purchased from GeneArt tion), the usually very low levels of readthrough that can (Thermo Fisher Scientific, Waltham, MA, USA). All the mutations be obtained, and the fact that an amino acid different from studied are shown in Figure 1A. that originally encoded may be introduced at the PTC site, with potentially detrimental consequences in terms of Cell lines and Chinese hamster ovary-cellular model protein function. Nonetheless, clinical trials or pilot stud- Human hepatocarcinoma cell line Huh-7 was kindly provided ies of RTA have already been carried out in patients with by Dr. J. Quer (VHIR, Barcelona) while CHO-S cells were pur- cystic fibrosis,5,6 Duchenne muscular dystrophy7,8 or hemo- chased from Thermo Fisher Scientific. Both cell lines were cul- philia,9 although often with inconclusive results. tured in Dulbecco's Modified Eagle's Medium supplemented with The aminoglycosides gentamicin and geneticin, as well as 10% fetal bovine serum, penicillin and streptomycin (all from non-aminoglycoside compounds, such as PTC124 Biowest, Nuaillé, France). For CHO transfection, cells were passed (Ataluren),10 RTC13 and RTC14,11 have been used as RTA three times and seeded at 4-5 x 104 cells in 12-well plates to based on their ability to suppress PTC and to allow the gen- achieve a confluence of 50-80%. The day after, the cells were eration of full-length proteins with at least partial function- transfected with 1.5 mg of the WT or mutated F8BDD plasmids ality. However, the efficiency of RTA therapy varies accord- using 1.5 mL lipofectamine-LTX (Thermo Fisher Scientific) per ing to not only the PTC type and sequence context,12-14 but well, following the manufacturer’s recommendations. also to the specific agent that is used, which can influence the binding of specific aminoacidyl-tRNA to the PTC and Readthrough agent treatment thus the amino acid added to the nascent peptide, with con- Patient-derived skin fibroblasts (3.5x103 cells/cm2) were seeded sequences for the functionality of the recoded protein.15 in 6-well plates and grown to 70-80% confluence before RTA Here, we present an analysis of the effect of five RTA (gen- treatment, while CHO cells were treated after 8 hours of transfec- tamicin, geneticin, PTC124, RTC13 and RTC14) on F8 tion. The cells were treated with gentamicin, geneticin (Thermo mRNA expressed in primary skin fibroblasts from three Fisher Scientific), PTC124 (Selleckchem Co., Houston, TX, USA), patients with HA as well as in a Chinese hamster ovary RTC13 or RTC14 (ID: 5735019 and ID: 5311257) (ChemBridge. (CHO)-cell-based model of HA. Our aim was to assess the San Diego, CA, USA) using a range of concentrations reported to readthrough effect of these RTA on the FVIII activity, in induce PTC readthrough in vitro (50-100 mg/mL for gentamicin and addition to FVIII:Ag levels, and the influence of the molec- geneticin, 10 mM for PTC124, RTC13 and RTC14).17,18 ular context, including type of stop codon, adjacent sequences, and the amino acid originally encoded by the F8 mRNA analysis wild-type (WT) protein at the mutated site. Total RNA was extracted using the RNeasy mini kit followed by on-column DNase I treatment (Qiagen. Hilden, Germany). Single-stranded cDNA was generated with the high capacity Methods cDNA reverse transcription kit (Thermo Fisher Scientific) using 500 ng of total RNA and random primers in a final volume of Patients and isolation of skin fibroblasts 25 mL, as previously described.19 The cDNA obtained was used to Four patients with HA caused by either nonsense mutations quantify F8 mRNA expression. (p.W1568X, p.Q1636X and p.R1960X) or a missense mutation (p.R1960Q), diagnosed at the Hemophilia Unit of the Vall FVIII Ag levels of F8BDD variants d'Hebron University Hospital and genetically characterized at the Human FVIII antigen levels (FVIII:Ag) were determined in cul- Congenital Coagulopathies Laboratory of the Blood and Tissue ture supernatants using the IMUBIND® factor VIII ELISA kit Bank of Catalonia (BST)16 were selected for this study. All partici- (Sekisui Diagnostics, Lexington, MA, USA). haematologica | 2020; 105(2) 509 L. Martorell et al. Table 2. Summary of the in silico analysis of the predicted readthrough effect at the protein level using the SIFT and Polyphen-2 softwares. F8 Mutation Mutation Affected Predicted Predicted SIFT Polyphen-2 Sequence Expected FVIII:C FVIII:Ag (% variant HGVS* Legacy† FVIII MW (kDa)/ AA change prediction prediction context/RT effect (fold before/after domain (%)‡ efficiency# on protein increase) RT Score Predicted Score Predicted [0-1] [1-0] 0 WT WT N.A 166.19 N.A N.A N.A N.A N.A A TGA C N.A N.A N.A 1 W274X W255X A1 30.92 W/W/(98) 1 SAME AA 0 SAME AA T TGA C/ WT 3.7 1.55 / 24.6 Highest variant (15.87) W/R/(1.2) 0 DAMAGING 1 DAMAGING W/C/(0.4) 0 DAMAGING 1 DAMAGING 2 Q462X Q443X A2 52.8 Q/Q/(64) 1 SAME AA 0 SAME AA T TAG C/ Mainly WT 2.6 3.65 / 18.7 Q/K/(34) 0.26 TOLERATED 0.99 DAMAGING High variant (5.12) Q/L/(0.5) 0.02 DAMAGING 0.99 DAMAGING 3 Q1705X Q1686X A3 91.66 Q/Q/(64) 1 SAME AA 0 SAME AA T TAG A/ Mainly WT 1 1.25 / 4.5 Low variant (3.6) Q/K/(34) 0.37 TOLERATED 0.99 DAMAGING Q/L/(0.5) 0.37 TOLERATED 0.397 BENIGN 4 R1715X R1696X A3 92.85 R/W/(98) 0.05 DAMAGING 1 DAMAGING A TGA C/ Mainly non 1.2 0.6 / 6.7 High functional (11.17) variant R/R/(1.2) 1 SAME AA 0 SAME AA R/C/(0.4) 0 DAMAGING 1 DAMAGING 5 W1726X W1707X A3 94.21 W/W/(98) 1 SAME AA 0 SAME AA C TGA G/ WT 1 0.75 / 5.0 Low variant (6.67) W/R/(1.2) 0 DAMAGING 1 DAMAGING W/C/(0.4) 0 DAMAGING 1 DAMAGING 6 Q1764X Q1745X A3 98.51 Q/Q/(64) 1 SAME AA 0 SAME AA T TAG C/ Mainly 3.3 1.6 / 13.4 High WT variant (8.37) Q/K/(34) 0.67 TOLERATED 0.062 BENIGN Q/L/(0.5) 0.13 TOLERATED 0.012 BENIGN 7 R1822X R1803X A3 105.24 R/W/(98) 0.04 DAMAGING 0.887 DAMAGING T TGA A/ Mainly non 1.3 0.4 / 5.3 Medium functional (13.25) variant R/R/(1.2) 1 SAME AA 0 SAME AA R/C/(0.4) 0.01 DAMAGING 0.84 DAMAGING 8 R1960X R1941X A3 121.5 R/W/(98) 0 DAMAGING 1 DAMAGING T TGA T/ Mainly non 1 1.1 / 9.55 Medium functional (8.68) variant R/R/(1.2) 1 SAME AA 0 SAME AA R/C/(0.4) 0 DAMAGING 1 DAMAGING 9 W2015X W1996X A3 127.9 W/W/(98) 1 SAME AA 0 SAME AA T TGA C/ WT 2.9 0.5 / 15.65 Highest variant (31.3) W/R/(1.2) 0.01 DAMAGING 1 DAMAGING W/C/(0.4) 0 DAMAGING 1 DAMAGING 10 R2071X R2052X C1 134.11 R/W/(98) 0 DAMAGING 1 DAMAGING C TGA C/ Mainly non 1.2 0.35 / 7 High functional (20) variant R/R/(1.2) 1 SAME AA 0 SAME AA R/C/(0.4) 0 DAMAGING 1 DAMAGING 11 W2131X W2112X C1 140.99 W/W/(98) 1 SAME AA 0 SAME AA G TGA C/ WT 2.7 2.85 / 13 High variant (4.56) W/R/(1.2) 0 DAMAGING 1 DAMAGING W/C/(0.4) 0 DAMAGING 1 DAMAGING 12 R2228X R2209X C2 151.85 R/W/(98) 0 DAMAGING 1 DAMAGING T TGA C/ Mainly non 1.1 0.6 / 3 Highest functional (5) variant R/R/(1.2) 1 SAME AA 0 SAME AA R/C/(0.4) 0 DAMAGING 1 DAMAGING *HGVS: Human Genome Variation Society numbering, where codons are numbered with codon +1 coding for the first residue (Met) of the 19-residue signal peptide (this is -19 in Legacy numbering). †Legacy numbering, where codon +1 refers to that encoding the first AA of the mature FVIII protein (in HGVS numbering, it is codon +20). ‡According to Roy et al.15. #According to nucleotide context rule (-1-Stop-+4). AA: amino acid. N.A: not aplicable. RT: readthrough. WT: wild type. 510 haematologica | 2020; 105(2) Variable nonsense suppression in Hemophilia A A B C Figure 1. F8 mutations studied and detection of F8 mRNA levels. (A) Schematic representation of the distribution of premature termination codons (PTC) across the F8BDD cDNA generated by site-directed mutagenesis for the Chinese hamster ovary (CHO) model or naturally occurring in the hemophilia A (HA) patients. The black arrow represents the F8 cDNA (5’ to 3’), numbers below the arrow correspond to the nucleotide position, while the gray bar represents the BDD-FVIII protein, and numbers below correspond to the amino acid position according to the Human Genome Variation Society (HGVS) nomenclature. The distribution of mutations in F8 mRNA analyzed in CHO model, patients' fibroblasts (in gray boxes) or both cellular models (black lined gray box) are also shown. BD-L: BDD-linker; FVIII-HC: heavy chain; FVIII-LC: light chain. (B) F8 mRNA levels detected by quantitative real-time polymerase chain reaction in the fibroblasts of HA-patients or a normal control. GAPDH: glyceraldehyde 3-phosphate dehydrogenase; Control: fibroblasts of a HB patient; Q1636X, W1586X and R1960X HA patients fibroblasts harboring these nonsense mutations; and R1960Q: HA patient fibroblasts harboring this missense mutation. CT: untreated cells; GN: geneticin 100 mg/mL; GT: gentamicin 100 mg/mL; PTC: PTC124 10 mM; RTC13: RTC13 10 mM; CHX: cycloheximide 1 mg/mL (n=3). (C) Time course of F8BDD mRNA levels detected by BDD-specific semi-quan- titative polymerase chain reaction in the CHO-based HA-model. F8BDD: WT variant; p.Q1705X, p.W1726X and p.R1960X: F8BDD variants harboring PTC; CHO: non transfected CHO cells; Huh-7: human hepatocellular carcinoma cell line (a cell line that expresses high levels of endogenous F8 but used here as a negative control of the F8BDD-specific amplification); MW: 100-bp DNA ladder. P values: *P<0.05, **P<0.01 and *** P<0.001. (n=3). haematologica | 2020; 105(2) 511 L. Martorell et al. Figure 2. Readthrough agents (RTA) can increase FVIII antigen levels in the supernatants of some transfected Chinese hamster ovary (CHO) cells. Human FVIII:Ag levels analyzed by ELISA in the supernatants of CHO cells transiently transfected with the F8 variants, showing the increase in FVIII:Ag levels before and after RTA- treatment versus the F8 wild-type (WT) variant. BDD-FVIII: WT; CT: untreated controls; GN100: 100 mg geneticin/mL and GT100: 100 mg gentamicin/mL. *P<0.05; **P<0.01; ***P<0.001. (N=3). FVIII immunodetection of F8BDD variants taining mRNA, as reported in previous studies.21,22 Chinese hamster ovary lines transfected with the F8BDD vari- Although some of the RTA increased F8 mRNA levels in ants were stained with several anti-FVIII antibodies for the the fibroblasts of HA patients, the differences were not immunodetection of FVIII in cell homogenates (Online statistically significant: p.W1586X: 59.89%±5.55 (CT) Supplementary Table S1). versus 86.01%±11.82 (100 mg geneticin/mL); p.Q1636X: 46.23%±9.19 (CT) versus 64.68%±12.41 (10 mM RTC13), FVIII activity of F8BDD variants and p.R1960X: 57.09%±3.82 (CT) versus 79.21%±5.66 The FVIII activity (FVIII:C) of the supernatants of cultured (100 mg gentamicin/mL) (Figure 1B). A possible explana- fibroblasts or CHO cells transfected with F8BDD variants, treated tion for this could be that none of the nonsense mutations or not with RTA, was assessed by the chromogenic method using analyzed in patients' fibroblasts met the PTC rule. the COAMATIC FVIII kit (Chromogenix, Werfen, Barcelona, In addition, we used CHO cells, which is the most com- Spain). The protocol was modified from Yatuv et al.20 monly used cell line for commercial production of FVIII,23 to analyze F8 mRNA levels to measure the effects of RTA. Results Time course qRT-PCR analysis using F8BDD-specific primers showed that the CHO cell model expressed high F8 mRNA levels after readthrough agent treatment levels of F8 mRNA without any presence of NMD over In the fibroblasts of HA-patients harboring nonsense time (Figure 1C). mutations, F8 mRNA levels measured by quantitative real-time-polymerase chain reaction (qRT-PCR) were Analysis of the human FVIII:Ag levels in the culture <60% (p.Q1636X: 46.23%±9.19; p.W1586X: media 59.89%±5.55; p.R1960X: 57.09%±3,81) of those detected Although fibroblasts ectopically express F8 mRNA, they in control fibroblasts from healthy individuals or from the do not synthesize FVIII protein.24 For this reason, and HA patients caused by the missense mutation. Treatment because none of the nonsense mutations analyzed in with the protein synthesis inhibitor cycloheximide, which patients' fibroblasts met the PTC rule, all protein studies also inhibits nonsense-mediated decay (NMD), restored were performed in the CHO cellular model. To determine the levels of PTC-containing transcripts to normal values, the ability of the already established RTAs geneticin and which suggested a role for NMD in our HA patients har- gentamicin to increase FVIII production, FVIII:Ag levels in boring nonsense mutations (Figure 1B). We then analyzed the culture supernatants of the treated CHO cells were the ability of RTA to suppress PTC and stabilize PTC-con- analyzed by ELISA (Table 2 and Figure 2). Compared to 512 haematologica | 2020; 105(2) Variable nonsense suppression in Hemophilia A Figure 3. Gentamicin increased FVIII biosynthesis in some transfected Chinese hamster ovary (CHO) cells. Representative images of the immunofluorescence detection of FVIII in transiently transfected CHO cells stained with antibodies against either FVIII heavy chain or FVIII light chain (both in green). Gentamicin treatment (GT50: 50 mg/mL) increased FVIII labeling using the anti-FVIII light chain antibody in CHO cells transiently transfected with variants harboring the p.W274X, p.Q462X, p.Q1764X, p.W2015X and p.W2131X premature termination codons. the controls (CT), these two drugs, each at 100 mg/mL, sig- WT F8BDD variant (Figure 3). However, after treatment of nificantly increased the human FVIII:Ag levels of the vari- the transfectants with 50 mg gentamicin/mL, staining for ants p.W274X (CT=1.55 vs. 24.6% with gentamicin), light chain was visible in cells transfected with variants p.Q462X (CT=3.65 vs. 18.75% with gentamicin), p.W274X, p.Q462X, p.Q1764X, p.R1960X, p.W2015X, p.Q1764X (CT=1.6 vs. 13.45% with geneticin), p.W2015X and p.W2131X (Figure 3). FVIII production in cell lysates (CT=0.5 vs. 15.9% with geneticin) and p.W2131X was also analyzed, using an anti-FVIII antibody raised (CT=2.85 vs. 13% with geneticin). In the p.R1960X vari- against FVIII heavy chain (Nt). Before RTA treatment, ants, FVIII:Ag levels increased significantly only in cells truncated FVIII accumulated intracellularly in cells trans- treated with 100 mg geneticin/mL (CT=1.1 vs. 9.55%). fected with the different constructs harboring PTC, which Smaller, non-significant, increases were observed for the is, in addition to a faulty degradation, one of the possible p.Q1715X, p.R1715X, p.W1726X, p.R18822x, p.R2071X consequences of the inability of FVIII to fold properly.23 In and p.R2228X variants. contrast, treatment with some of the RTA reduced the amount of intracellular accumulation (Figure 4 and Online Immunodetection of FVIII after readthrough agent Supplementary Figure S1). In the p.Q462X, the highest treatment reduction in FVIII accumulation was observed with To establish the mechanism by which the previously geneticin, at the two tested concentrations. The p.Q1764X used RTA increased FVIII production, protein levels were and p.W2015X variants were also responsive to geneticin, analyzed in transfected CHO cells by immunofluores- but the highest reduction in intracellular accumulation of cence using antibodies generated against either heavy FVIII was achieved with RTC13. The p.R1822X showed a (N-terminal) or light (C-terminal) chains (Online reduction in intracellular accumulation only with the high- Supplementary Table S1). Before RTA treatment, light chain er concentration of gentamicin tested. Other variants also staining was detected only in cells transfected with the showed a qualitative reduction (p.Q1705X for geneticin haematologica | 2020; 105(2) 513 L. Martorell et al. and gentamicin; R1960X for geneticin, gentamicin but also Rescue of FVIII:C activity after readthrough agent PTC124, and p.W2131X for geneticin, and RTC13) or only treatment a minimal qualitative reduction (p.W1726X and Finally, we focused in the RTA-mediated effect on the p.R2228X) in the intracellular accumulation of truncated protein functionality rather than on the protein levels. To FVIII (Online Supplementary Figure S1). this end, we analyzed the FVIII:C activity by means of a A B Figure 4. Readthrough agents (RTA) reduced intracellular accumulation of FVIII in transfected Chinese hamster ovary (CHO) cells. Representative images of the immunodetection of FVIII heavy chain in cell homogenates of transiently transfected CHO cells. (A) Western blot of CHO cells transfected with either wild-type (WT) BDD-FVIII (166.19 kDa), the p.Q462X (52.8 kDa), p.Q1764X (98.51 kDa) (top panel), p.W2015X (127.9 kDa), or the p.R1822X (105.24 kDa) variants (bottom panel). The intracellular accumulation of BDD-FVIII or the truncated proteins before and after RTA-treatment is shown. (B) Densitometry of the bands in (A). CT: Untreated cells; GN50: 50 mg geneticin/mL; GN100: 100 mg geneticin/mL; GT50: 50 mg gentamicin/mL; GT100: 100 mg gentamicin/mL; PTC: PTC124 10 mM; RTC13 and RTC14: 10 mM; Mock: untransfected cells; MW: Pageruler Plus prestained protein ladder; Refacto®: recombinant human BDD-FVIII (Pfizer); LR: loading reference used for densitometry normalization. 514 haematologica | 2020; 105(2) Variable nonsense suppression in Hemophilia A chromogenic assay using not only the well established readthrough response was that of p.W274X, in which a RTA geneticin and gentamicin, but also other RTA such as 3.7-fold increase in activity was obtained after treatment PTC, RTC13 and RTC14. RTA treatment resulted in a sig- with 100 mg gentamicin/mL. Smaller increases in FVIII:C nificant dose-dependent increase in FVIII:C in only five of activity were measured at a lower dose of gentamicin (50 the 12 variants (Table 2 and Figure 5). The strongest mg/mL) or geneticin (50 and 100 mg/mL). Increases in A B C Figure 5. Readthrough agents (RTA) increase FVIII coagulant activity in the supernatants of transfected Chinese hamster ovary (CHO) cells. (A) Fold increase of FVIII:C (relative to untreated controls) in the supernatants of cultured CHO cells transiently transfected with variants harboring a QX premature termination codons (PTC) (see legend to Figure 1). (B) Fold increase of FVIII:C in the supernatants of cultured CHO cells transiently transfected with variants harboring a WX PTC (see leg- end to Figure 1). (C) Fold increase of FVIII:C in the supernatants of cultured CHO cells transiently transfected with variants harboring a RX PTC (see legend to Figure 1). CT: untreated controls; GN50: 50 mg geneticin/mL; GN100: 100 mg geneticin/mL; GT50: 50 μg gentamicin/mL; GT100: 100 mg gentamicin/mL; PTC: PTC124 10 mM; RTC13 and RTC14: 10 μM. *P<0.05; **P<0.01; ***P<0.001. (N=3). haematologica | 2020; 105(2) 515 L. Martorell et al. FVIII:C after exposure to 100 mg geneticin/mL were also NMD is absent, as cells were transfected with plasmid observed in the p.W2015X and p.W2131X variants (2.9- containing the F8 cDNA, but lacking the exon-exon junc- and 2.7-fold, respectively) and in the p.Q462X and tion complexes (EJC) which are necessary for PTC-con- p.Q1764X variants (up to 2.6- and 3.3-fold, respectively). taining mRNA to be targeted by NMD, thus providing a These variants were also responsive, although to a lesser clean tool to amplify and detect eventual readthrough extent, to treatment with 100 mg gentamicin/mL but the activity of RTA. While cDNA models do not mimic the in increases in FVIII:C were smaller. Variants responsive to vivo situation (in which NMD is present), they can be either gentamicin or geneticin showed a slight response to used for assessing potential readthrough effects and to other RTA, particularly to PTC124 and RTC13, but these identify new RTA.10,11,17,25 In practice, to show a increases were not statistically significant. readthrough effect in vivo or in primary cells, RTA will probably have to be combined with NMD inhibition. In silico analysis of the premature termination codons Nonetheless, the CHO-based HA model allowed us to Variable responses to RTA-induced nonsense suppres- demonstrate that the effectiveness of readthough therapy sion related to the sequences adjacent to the PTC have is mutation-specific and that in vitro assays may be useful been reported. We therefore analyzed the PTC type and for classifying patients that can likely benefit from RTA. the context effect in both the patients' fibroblasts and the It has recently become clear that different types of 12 F8BDD variants (Tables 1 and 2, and Online endothelial cells, including liver sinusoidal endothelial Supplementary Table S2). In patients' fibroblasts respon- cells (LSEC), constitute the main source of FVIII produc- sive to RTA, we would expect at least a partial restoration tion in humans,24,26-29 perhaps to a greater extent than of the F8 mRNA levels; however, none of the PTC hepatocytes. This may explain why, although ectopically sequence context filled the PTC rule, so we cannot expect expressed at the mRNA level, none of the fibroblasts any mRNA stabilization. In contrast, in the CHO model, from HA patients analyzed showed a significant stabi- we could evaluate the readthrough effect on the produc- lization of F8 mRNA levels. However, as the PTC tion of full-length protein and on the restoration of sequence context does not fill the PTC rule, we cannot FVIII:C activity. The results suggested a TGA≥TAG hier- draw any conclusions as to the effectiveness of the RTA archy of PTC suppression sensitivity and a positive RTA used. Furthermore, since fibroblasts do not produce readthrough effect on both FVIII:Ag levels and FVIII:C FVIII,24 we could not expect to detect FVIII protein. activity for sequences in which a C was located at posi- Despite the limitations of this study, which include the tion +4 and a T at position −1 as present in variants relatively low number of mutations analyzed, the intrin- pW274X, pQ462X, pQ1764X and pW2015X. sic variability in the transfection efficiencies, and the low Furthermore, the pW2131X variant also showed respon- levels of FVIII:C achieved (in the low range of detection siveness even when a G is present at position -1 instead by FVIII:C assays, i.e. 0.5-3%), the observed increases in of a T. In contrast, variants pR1715X, pR2071X and the FVIII:C activity in some mutations suggest a potential pR2228X, even containing a TGA and a T at position –1 phenotypic improvement, from severe to moderate/mild and a C at position +4 were unable to significantly HA. Thus, RTA treatment significantly rescued FVIII pro- increase FVIII:C activity. Similarly, the pR1822X variant duction in some of the F8BDD variants, as measured by was also unresponsive to RTA, despite the fact that it FVIII:Ag levels. This increase was associated with a qual- contains a TGA and a T at position –1. itative reduction in intracellular accumulation of FVIII Although positions −5 (A, C>G) and +8 (G>C) were also observed in the western blots, and a concomitant increase reported to influence readthrough, our data show that T, in light-chain FVIII detection by immunofluorescence. G and A at position −5 as well as T or C at position +8 Moreover, data obtained in the FVIII:C analyses high- promoted PTC suppression, as measured by FVIII:C lev- lighted the importance of the amino acid incorporated els; other tested combinations may have positive effects upon RTA treatment in the partial restoration of FVIII:C measured by FVIII:Ag levels. Whether positive effects are activity. In our study, after drug treatment, only variants associated with yet other nucleotide combinations could that natively encoded tryptophan (Trp=W) or glutamine not be determined due to the limited number of F8 vari- (Gln=Q) and had a cytosine at position +4, regardless of ants tested (Online Supplementary Table S2). the PTC type, were able to significantly increase FVIII:C in the culture supernatants, while this activity was not increased in variants natively encoding arginine (Arg=R), Discussion even when a cytosine at position +4 was present, likely because the RTA may promote the incorporation of The present study represents a comprehensive analysis amino acids that do not necessarily restore FVIII function- of factors influencing the responsiveness to RTA of a ality. series of nonsense mutations that are representative of Nonsense suppression was observed in only five of the those causing HA. Readthrough therapy is not expected 12 transfected F8BDD variants (p.Q462X, p.Q1764X, to restore the synthesis and activity of the recoded pro- p.W274X, p.W2015X and p.W2131X). This apparent dis- teins to curative levels in most patients harboring non- crepancy between protein levels and restored activity can sense mutations. However, in the case of hemophilia, be explained by the context-dependent ability of RTA to even a relatively small percentage of restoration may be restore the full-length recoded protein13 and by the pre- very valuable. Thus, in patients with severe HA (<1% of ferred aminoacyl tRNA that will recognize each PTC in FVIII:C), a slight increase in activity to 2-5% could signif- RTA-treated cells.15 Accordingly, upon secretion, only full- icantly improve the bleeding phenotype as reported for length protein that has incorporated the native amino FIX production in severe HB patients treated with gen- acid (or another one with similar chemical properties) is tamicin.9 expected to have a greater chance of restoring maximum Our study uses a CHO-based HA model in which or partial levels of FVIII:C activity. Additionally, the phy- 516 haematologica | 2020; 105(2) Variable nonsense suppression in Hemophilia A logenetic conservation of this amino acid, or whether effective RTA. By taking into account the functionality of deleterious missense mutations have been described in the restored protein generated by RTA, our cellular model, this codon, are additional important factors to be consid- or other similar models, provide a method for testing the ered. effectiveness of new RTA.31 Under our experimental conditions, the ability of RTA Although we evaluated only a small number of F8BDD to increase C-terminal (light chain) positive staining or variants, and these results cannot be generalized, our FVIII:Ag levels did not always correlate with the ability of analysis of the molecular environment of these PTC pro- RTA to restore FVIII:C activity. This suggests that the vides valuable information on additional factors involved amino acid incorporated at the PTC and its structural sim- in eventual responses to RTA treatment. From these ilarity with the natively encoded amino acid can be criti- results we can conclude that: 1) responsiveness is greater cal for the restoration of FVIII:C, but not necessarily for in PTC with the consensus sequence T-Stop-C; 2) in TGA that of FVIII:Ag. According to the PTC context rule,13 the (opal) than in TAG (amber) PTC; and 3) it will be higher hierarchy of RTA responsiveness is T-TGA-C>T-TAG-C> when the mutated original amino acid is tryptophan or T-TAA-C; thus, a stronger RTA effect at TGA (opal) is glutamine as opposed to arginine, in which, despite the expected in PTC originally encoding tryptophan and argi- efficient intracellular reduction of truncated protein and nine. However, we found that PTC generated at triplets extracellular increase of FVIII:Ag levels, the increase natively encoding arginine and fulfilling the PTC context observed in FVIII:C activity was minimal. rule led to production of a non-functional protein, In conclusion, the results obtained with our model rep- although this situation can also result in a gain of func- resent a proof-of-concept that RTA are capable of sup- tion, as reported in a cellular model of readthrough in pressing PTC in some nonsense mutations causing HA, hemophilia B.30 Conversely, according to the PTC context albeit with different efficiencies and different functional rule, a PTC generated at triplets encoding glutamine is consequences depending on the sequence context. This predicted to be less responsive to RTA treatment. model, linked to a functional assay based on the determi- However, in our study, RTA allowed partial restoration of nation of FVIII:C activity, and the application of the PTC FVIII:C levels in 2 of 3 recoded FVIII mutants originally rule described here, allowed for the first time the identifi- encoding glutamine (pQ462X and pQ1764X). A similar cation of five RTA-responsive nonsense mutations in HA result was reported in a study of cystic fibrosis in which and may facilitate the classification of nonsense mutations patients harboring the Y122X mutation had a better than reported in the HA database into categories based on the expected response to RTA, even though the PTC was predicted levels of responsiveness to RTA, thus promoting caused by a TAA codon, followed by a C nucleotide at patient selection in readthrough therapy clinical trials.32 position +4, which was previously reported to have the lowest RTA sensitiveness.13 However, both the Acknowledgments readthrough effect due to the presence of cytosine at The authors thank the participating patients and their families. position +4, and the incorporation of the natively encod- We also thank Yvonne Richaud-Patin and Dr. Angel Raya ed tyrosine might have been facilitated by the RTA treat- (CMRB, Spain) for their help with the fibroblast isolation. ment,15 which would account for the clinical benefit observed. Funding Our work points out the importance of not only evalu- This project was supported by two grants from the Instituto de ating RTA-mediated protein biosynthesis, but also the Salud Carlos III (PI11/03024 and PI11/03029) / e-Rare-2 functional activity of the recoded protein generated. This (HEMO-iPS) that were cofinanced by the European Regional is particularly important in studies of the potential clinical Development Fund (FEDER), and the Europe Aspire 2013 benefit of RTA therapy and the search for safer and more Award in Hemophilia Research. References tamicin. J Clin Invest. 2007;117(3):683-692. 447(7140):87-91. 6. Sermet-Gaudelus I, Renouil M, Fajac A, et 11. Du L, Damoiseaux R, Nahas S, et al. 1. Gouw SC, van den Berg HM, Oldenburg J, al. In vitro prediction of stop-codon sup- Nonaminoglycoside compounds induce et al. F8 gene mutation type and inhibitor pression by intravenous gentamicin in readthrough of nonsense mutations. J Exp development in patients with severe hemo- patients with cystic fibrosis: a pilot study. Med. 2009;206(10):2285-2297. philia A: systematic review and meta- BMC Med. 2007;5:5. 12. Manuvakhova M, Keeling K, Bedwell DM. analysis. Blood. 2012;119(12):2922-2934. 7. Finkel RS, Flanigan KM, Wong B, et al. Aminoglycoside antibiotics mediate con- 2. Escobar MA. Health economics in Phase 2a study of ataluren-mediated dys- text-dependent suppression of termination haemophilia: a review from the clinician’s trophin production in patients with non- codons in a mammalian translation system. perspective. Haemophilia. 2010;16:29-34. sense mutation Duchenne muscular dystro- RNA. 2000;6(7):1044-1055. 3. Rangarajan S, Walsh L, Lester W, et al. phy. PLoS One. 2013;8(12):e81302. 13. Floquet C, Hatin I, Rousset J-P, Bidou L. AAV5-Factor VIII Gene Transfer in Severe 8. Bushby K, Finkel R, Wong B, et al. Ataluren Statistical analysis of readthrough levels for Hemophilia A. N Engl J Med. 2017; treatment of patients with nonsense muta- nonsense mutations in mammalian cells 377(26):2519-2530. tion dystrophinopathy. Muscle Nerve. reveals a major determinant of response to 4. George LA, Sullivan SK, Giermasz A, et al. 2014; 50(4):477-487. gentamicin. PLoS Genet. 2012; Hemophilia B Gene Therapy with a High- 9. James PD, Raut S, Rivard GE, et al. 8(3):e1002608. Specific-Activity Factor IX Variant. N Engl J Aminoglycoside suppression of nonsense 14. Dabrowski M, Bukowy-Bieryllo Z, Med. 2017;377(23):2215-2227. mutations in severe hemophilia. Blood. Zietkiewicz E. Translational readthrough 5. Linde L, Boelz S, Nissim-Rafinia M, et al. 2005; 106(9):3043-3048. potential of natural termination codons in Nonsense-mediated mRNA decay affects 10. Welch EM, Barton ER, Zhuo J, et al. eucaryotes--The impact of RNA sequence. nonsense transcript levels and governs PTC124 targets genetic disorders caused by RNA Biol. 2015;12(9):950-958. response of cystic fibrosis patients to gen- nonsense mutations. Nature. 2007; 15. Roy B, Leszyk JD, Mangus DA, Jacobson A. haematologica | 2020; 105(2) 517 L. Martorell et al. Nonsense suppression by near-cognate 21. Bedwell DM, Kaenjak A, Benos DJ, et al. Tuddenham EG, Lawn RM. Distribution of tRNAs employs alternative base pairing at Suppression of a CFTR premature stop factor VIII mRNA and antigen in human codon positions 1 and 3. Proc Natl Acad Sci mutation in a bronchial epithelial cell line. liver and other tissues. Nature. U S A. 2015;112(10):3038-3043. Nat Med. 1997;3(11):1280-1284. 1985;317(6039):726-729. 16. Vidal F, Farssac E, Altisent C, Puig L, 22. Keeling KM, Brooks DA, Hopwood JJ, et al. 28. Olgasi C, Talmon M, Merlin S, et al. Gallardo D. Rapid hemophilia A molecular Gentamicin-mediated suppression of Patient-Specific iPSC-Derived Endothelial diagnosis by a simple DNA sequencing pro- Hurler syndrome stop mutations restores a Cells Provide Long-Term Phenotypic cedure: identification of 14 novel mutations. low level of alpha-L-iduronidase activity Correction of Hemophilia A. Stem Cell Thromb Haemost. 2001;85(4):580-583. and reduces lysosomal glycosaminoglycan Rep. 2018;11(6):1391-1406. 17. Jung ME, Ku J-M, Du L, Hu H, Gatti RA. accumulation. Hum Mol Genet. 29. Suzuki Y, Kakisaka K, Matsumoto T, et al. Synthesis and evaluation of compounds 2001;10(3):291-299. Orthotopic liver transplantation for that induce readthrough of premature ter- 23. Kumar SR. Industrial production of clotting haemophilia A may not always lead to a mination codons. Bioorg Med Chem Lett. factors: Challenges of expression, and phenotypic cure of haemophilia A: A case 2011;21(19):5842-5848. choice of host cells. Biotechnol J. report. Haemophilia. 2018;24(6):e420-e422. 18. Yu H, Liu X, Huang J, et al. Comparison of 2015;10(7):995-1004. 30. Branchini A, Ferrarese M, Campioni M, et readthrough effects of aminoglycosides 24. Turner NA, Moake JL. Factor VIII Is al. Specific factor IX mRNA and protein and PTC124 on rescuing nonsense muta- Synthesized in Human Endothelial Cells, features favor drug-induced readthrough tions of HERG gene associated with long Packaged in Weibel-Palade Bodies and over recurrent nonsense mutations. Blood. QT syndrome. Int J Mol Med. 2014; Secreted Bound to ULVWF Strings. PLoS 2017;129(16):2303-2307. 33(3):729-735. One. 2015;10(10):e0140740. 31. Baradaran-Heravi A, Balgi AD, 19. Martorell L, Corrales I, Ramirez L, et al. 25. Benhabiles H, Gonzalez-Hilarion S, Amand Zimmerman C, et al. Novel small mole- Molecular characterization of ten F8 splic- S, et al. Optimized approach for the identi- cules potentiate premature termination ing mutations in RNA isolated from fication of highly efficient correctors of codon readthrough by aminoglycosides. patient’s leucocytes: assessment of in silico nonsense mutations in human diseases. Nucleic Acids Res. 2016;44(14):6583-6598. prediction tools accuracy. Haemophilia. PLoS One. 2017;12(11):e0187930. 32. Schmittgen TD, Zakrajsek BA, Mills AG, et 2015;21(2):249-257. 26. Hollestelle MJ, Thinnes T, Crain K, et al. al. Quantitative reverse transcription-poly- 20. Yatuv R, Dayan I, Baru M. A modified chro- Tissue distribution of factor VIII gene merase chain reaction to study mRNA mogenic assay for the measurement of very expression in vivo--a closer look. Thromb. decay: comparison of endpoint and real- low levels of factor VIII activity (FVIII:C). Haemost. 2001;86(3):855-861. time methods. Anal Biochem. 2000; Haemophilia. 2006;12(3):253-257. 27. Wion KL, Kelly D, Summerfield JA, 285(2):194-204. 518 haematologica | 2020; 105(2)