EBioMedicine 46 (2019) 342–355 Contents lists available at ScienceDirect EBioMedicine journal homepage: www.ebiomedicine.com Research paper Age-related metabolic changes limit efficacy of deoxynucleoside-based therapy in thymidine kinase 2-deficient mice Cora Blázquez-Bermejo a,b, David Molina-Granada a,b, Ferran Vila-Julià a,b, Daniel Jiménez-Heis a, Xiaoshan Zhou c, Javier Torres-Torronteras a,b, Anna Karlsson c, Ramon Martí a,b, Yolanda Cámara a,b,⁎ a Research Group on Neuromuscular and Mitochondrial Disorders, Vall d'Hebron Institut de Recerca, Universitat Autònoma de Barcelona, Barcelona, Spain b Biomedical Network Research Centre on Rare Diseases (CIBERER), Instituto de Salud Carlos III, Madrid, Spain c Division of Clinical Microbiology, Department of Laboratory Medicine, Karolinska Institute, Karolinska University Hospital, 141 86 Stockholm, Sweden a r t i c l e i n f o a b s t r a c t Article history: Background: Thymidine kinase 2 (TK2) catalyses the phosphorylation of deoxythymidine (dThd) and Received 4 June 2019 deoxycytidine (dCtd) within mitochondria. TK2 deficiency leads to mtDNA depletion or accumulation of multiple Received in revised form 5 July 2019 deletions. In patients, TK2 mutations typically manifest as a rapidly progressive myopathy with infantile onset, Accepted 15 July 2019 leading to respiratory insufficiency and encephalopathy in the most severe clinical presentations. TK2- Available online 24 July 2019 deficient mice develop the most severe form of the disease and die at average postnatal day 16. dThd+dCtd ad- ministration delayed disease progression and expanded lifespan of a knockin murine model of the disease. Keywords: TK2 Methods: We daily administered TK2 knockout mice (Tk2KO) from postnatal day 4 with equimolar doses of dThd Deoxynucleoside therapy +dCtd, dTMP+dCMP, dThd alone or dCtd alone. We monitored body weight and survival and studied different mtDNA depletion variables at 12 or 29 days of age. We determined metabolite levels in plasma and target tissues, mtDNA copy Encephalomyopathy number in tissues, and the expression and activities of enzymes with a relevant role in mitochondrial dNTP anab- Thymidine olism or catabolism. Deoxycytidine Findings: dThd+dCtd treatment extended average lifespan of Tk2KO mice from 16 to 34 days, attenuated growth retardation, and rescued mtDNA depletion in skeletal muscle and other target tissues of 12-day-old mice, except in brain. However, the treatment was ineffective in 29-day-old mice that still died prematurely. Bioavailability of dThd and dCtd markedly decreased during mouse development. Activity of enzymes catabolizing dThd and dCtd increased with age in small intestine. Conversely, the activity of the anabolic enzymes decreased in target tissues during mouse development. We also found that administration of dThd alone had the same impact on survival to that of dThd+dCtd, whereas dCtd alone had no influence on lifespan. Interpretation: dThd+dCtd treatment recruits alternative cytosolic salvage pathways for dNTP synthesis, suggest- ing that this therapy would be of benefit for any Tk2 mutation. dThd accounts for the therapeutic effect of the combined treatment in mice. During the first weeks after birth, mice experience marked tissue-specific metabolic regulations and ontogenetic changes in dNTP metabolism-related enzymes that limit therapeutic efficacy to early developmental stages. Fund: This study was funded by grants from the Spanish Ministry of Industry, Economy and Competitiveness, the Spanish Instituto de Salud Carlos III, the Fundación Inocente, Inocente, AFM Téléthon and the Generalitat de Ca- talunya. The disclosed funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. © 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http:// creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Abbreviations: CDA, cytidine deaminase; CNS, central nervous system; dCK, Thymidine kinase 2 (TK2), a mitochondrial enzyme encoded by nu- deoxycytidine kinase; dCMP, deoxycytidine monophosphate; dCtd, deoxycytidine; dN, clear DNA, catalyses a limiting step in mitochondrial deoxynucleoside deoxynucleoside; dNMP, deoxynucleoside monophosphate; dNTP, deoxynucleoside triphosphate; dThd, deoxythymidine; dTMP, deoxythymidine monophosphate; KO, (dN) salvage. This pathway recycles dNs from endogenous metabolism knockout; mtDNA, mitochondrial DNA; nDNA, nuclear DNA; PBS, phosphate buffered and diet by using a set of kinases that catalyse equivalent reactions in saline; Thy, thymine; TK1, thymidine kinase 1; TK2, thymidine kinase 2; TP, thymidine the cytosol and mitochondria. TK2 phosphorylates the pyrimidine dNs, phosphorylase; Tymp, thymidine phosphorylase gene; WT, wild-type. deoxythymidine (dThd) and deoxycytidine (dCtd), to deoxythymidine ⁎ Corresponding author at: Laboratori de Patologia Neuromuscular i Mitocondrial, Institut de Recerca Hospital Universitari Vall d'Hebron-CIBERER, Passeig Vall d'Hebron, monophosphate (dTMP) and deoxycytidine monophosphate (dCMP), 119-129, 08035 Barcelona, Spain. respectively. These deoxynucleoside monophosphates (dNMPs) are E-mail address: yolanda.camara@vhir.org (Y. Cámara). consecutively phosphorylated to deoxynucleoside triphosphates https://doi.org/10.1016/j.ebiom.2019.07.042 2352-3964/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 343 Research in context Evidence before this study Thymidine kinase 2 (TK2) is a key enzyme for mitochondrial pyrim- idine dNTP synthesis. Mutations in TK2 reduce availability of dTTP and dCTP and interfere with normal mitochondrial DNA (mtDNA) replication leading typically to a severe and rapidly-progressive myopathy. Supplementation with precursors in the form of deoxynucleosides (dNs), specifically deoxythymidine (dThd) and deoxycytidine (dCtd), activates the synthesis of dTTP and dCTP through alternative enzyme pathways and effectively counteracts the defect in a knockin mouse model of the disease. These results prompted the compassionate use of dThd plus dCtd for the treatment of patients suffering TK2 deficiency with signifi- cant positive results so far. Added value of this study Here, we use an alternative mouse model completely devoid of TK2 activity (knockout mouse) with a fatal encephalomyopathy phenotype, similar to that observed for the knockin mice. We con- firm that administration of dThd+dCtd, delays disease progres- sion, doubles life expectancy and counteracts mtDNA depletion in most target tissues of young knockout mice. In addition, we ob- Fig. 1. Schematic representation of the main enzyme pathways involved in mitochondrial pyrimidine dNTPs metabolism. Proteins involved in the de novo synthesis pathway are serve how administration of dThd alone leads to the same thera- depicted in blue. Kinases involved in the deoxynucleoside salvage pathway are depicted peutic effect in terms of survival. However, important tissue- in red. Enzymes participating in catabolism of dNs are depicted in purple. Nucleoside specific metabolic regulations predict different outcomes for the transporters from cytosol to mitochondria are represented in green. Abbreviations: CDA, combined or individual therapy depending on the target tissue. cytidine deaminase; cdN, cytosolic deoxyribonucleotidase; dCTD, dCMP deaminase; dCK, deoxycytidine kinase; ENT1, equilibrative nucleoside transporter 1; mdN, mitochondrial Both TK2-deficient mouse models die prematurely in spite of the deoxyribonucleotidase; NDPK, nucleotide diphosphate kinase; NMPK, nucleotide treatment. In this study, we deeply characterize metabolic re- monophosphate kinase; TK1, thymidine kinase 1; TK2, thymidine kinase 2; TP, sponse to dNs-treatment, and identify decreased activity of sal- thymidine phosphorylase; TS, thymidylate synthase; RNR, ribonucleotide reductase. vage kinases and increased deoxynucleoside catabolism as key factors limiting therapeutic efficacy to early developmental stages in mice. Implications of all the available evidence with mild limb weakness and slow progression to respiratory insuffi- ciency [11]. No effective therapeutic option was available for TK2 deficient pa- Two different TK2-deficient mouse models have been generated to tients before dTMP and dCMP were proposed, followed by the ob- investigate the molecular mechanism of this disease, a knockin model servation that the active compounds in this treatment are dThd for a common pathogenic mutation, p.His121Asn in humans (p. and dCtd. In the most severe cases, TK2-deficient patients die dur- His126Asn in mice) [1], and a knockout mouse [34]. In patients, TK2 de- ing early infancy. After recent experience of compassionate use of ficiency predominantly causes myopathy, but both mouse models man- dThd and dCtd in TK2 deficiency, a cohort of patients is being ifest major central nervous system (CNS) involvement, with treated with positive effects on survival, motor and respiratory development of fatal encephalomyopathy resembling the most severe function. Hence, it is at the utmost importance to understand cases of the disease in humans [11,12]. how dThd and dCtd exert their metabolic effect and what factors Therapies designed to feed dN salvage have proven to delay disease may be lessening their therapeutic potential. progression and increase survival. The first pre-clinical studies on the Our study will be crucial for the design of a next-coming clinical knockin model intended to bypass the TK2 defect with administration trial for TK2-deficiency treatment and will surely be of interest in of dTMP and dCMP. This treatment augmented mtDNA levels in the tar- expanding dN-based therapies for the treatment of other mito- get tissues and prolonged survival with dose-related efficacy, although chondrial DNA depletion syndromes. phenotype rescue was incomplete and mice still died prematurely be- fore adulthood [10]. Based on these results, dNMPs were first used to treat a few patients [6]. Later studies identified dNs as the active agents (dNTPs) and incorporated into mitochondrial DNA (mtDNA) during rep- following dNMPs administration. dNMPs are charged compounds and lication [32] (Fig. 1). Thus, mutations in TK2 result in mitochondrial in the absence of a known transporter, they cannot be internalized by deoxythymidine triphosphate (dTTP) and deoxycytidine triphosphate cells. In addition, dNMPs are rapidly dephosphorylated to dNs in vivo (dCTP) insufficiency, dNTP pool imbalance, and consequent mtDNA de- by nucleotidases and phosphatases [4,35]. In a later study using the pletion or accumulation of multiple mtDNA deletions [2,11,26,29]. same mouse model, López et al. found that co-administration of dThd The clinical presentation of TK2 deficiency is mainly myopathic. and dCtd yielded similar results [16]. These positive data served as a Based on the age at onset and severity, patients can be classified into proof-of-concept that dNs could be effective therapeutic agents for three main groups, from the most severe and common to the milder TK2 deficiency, and prompted their use as compassionate therapy. So and less frequent manifestations: 1) infantile-onset myopathy with oc- far, the treatment has provided clinically significant benefits, especially casional neurological involvement, severe mtDNA depletion, and early in infantile- and childhood-onset cases, and has led to improvement or death; 2) childhood-onset myopathy with mtDNA depletion and mod- stabilization of respiratory function in late-onset cases after a few years erate to severe progression of weakness; and 3) late-onset myopathy or even months of therapy [6]. 344 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 Despite the promising pre-clinical results, dN-treated mice still die (angiogenin 1 [Ang1, Mm00833184_s1] single copy gene) copy number prematurely [16]. As there is no currently available alternative therapy quantification. Two different standard curves with cloned 16S rRNA and for patients with TK2 deficiency, it is crucial to determine what ele- Ang1 amplicons were used for absolute quantification of mtDNA ments may act to lessen the efficacy of dNs. Here, using the knockout and nDNA. Detection was done using an ABI PRISM 7900HT Sequence model generated by Zhou et al. [34], we perform a deep analysis of the Detection System (Applied Biosystems). mechanisms regulating dNTP homeostasis and identify factors that should be key for leveraging the therapeutic potential of dNs. 2.4. Determination of metabolite levels in mouse plasma and tissues 2. Materials and methods dThd, dCtd, and their degradation products (thymine [Thy] and deoxyuridine [dUrd], respectively) were measured in plasma and tissue 2.1. Ethics statement samples by liquid chromatography coupled to tandem mass spectrom- etry (LC-MS/MS), using an LC-MS/MS system (Acquity UPLC-XevoTM This study was conducted in accordance with the rules established TQ Mass Spectrometer, Waters, Milford, MA, USA). Frozen tissues by the Generalitat de Catalunya for the Care and Use of Laboratory Ani- were ground in liquid nitrogen and homogenized in lysis buffer mals. The protocol was approved by the Ethics Committee for Animal (50 mM Tris-HCl, pH 7.2; 1% Triton X-100; 2 mM phenylmethylsulfonyl Experimentation of the Vall d'Hebron Research Institute (Permit Num- fluoride; 0.02% 2-mercaptoethanol; 100 μM tetrahydrouridine (THU); ber: 02/17). and 100 μM 6-amino-5-bromouracile [6A5BrU]) by mechanical disrup- tion using a plastic pestle. Deuterated dThd (1 μM; Cambridge Isotope 2.2. Mice housing, treatment administration, tissue collection and Laboratories) was used as an internal standard. Homogenates were cen- endpoints trifuged at 20,000g at 4 °C for 30 min, and supernatants were collected. Protein concentration was measured by the Bradford method. Diluted The study was carried out with TK2 knockout mice (Tk2KO) (C57BL/6 plasma and tissue homogenates were deproteinised by ultrafiltration genetic background) previously generated and characterized as de- (10 kDa Amicon Ultra filters, Millipore) at 14,000 g at 4 °C for 30 min scribed [34]. Mice were housed and bred in a standard controlled envi- and stored at 20 °C until further use. Samples and standards were ronment under a 12-h light-dark cycle with ad libitum access to water loaded into 96-well plates for injection in the LC-MS/MS system. For and regular rodent chow diet. Tk2KO mice develop a fatal encephalomy- all determinations, the stationary phase was an Acquity UPLC BEH C18 opathy and die at around two weeks of age. We monitored body weight, column (100 × 2.1 mm, 130 Å pore, 1.7 mm particle, Waters). For deter- general behaviour and appearance, and survival on a daily basis. For eth- minations, 7.5 μl of sample was injected into the LC-MS/MS system and ical reasons, we euthanized mice in the survival studies when they were resolved at 0.5 ml/min through a binary gradient elution using a saline too weak to move in the cage and thereby obtain necessary food and buffer (20 mM ammonium acetate, pH 5.6) and acetonitrile as follows: drink. time 0–1.1 min, isocratic 100% saline buffer; 1.1–5 min, gradient from 0 Treatment was administered daily by single-dose oral gavage to 13.6% acetonitrile; 5–5.1 min, gradient from 13.6 to 100% acetonitrile; starting from postnatal day 4. Solutions of dThd, dCtd, dTMP, and 5.1–6.1 min, isocratic 100% acetonitrile; and 6.1–7.2 min, isocratic 100% dCMP were prepared in PBS, filter-sterilized, and stored at 4 °C (−20 saline buffer. Detection of the eluate components was performed using °C for long-term storage) until administered to Tk2KO mice and wild- multiple reaction monitoring, in positive electrospray mode with the type littermates (Tk2WT). dThd and dCtd were administered, either following m/z transitions: 255.0 N 133.8 (deuterated dThd), 242.8 combined or individually, each at a dose of 400 mg/kg/day (165 mM N 127.1 (dThd), 228.8 N 113.1 (dUrd), 227.9 N 112.1 (dCtd), and 126.7 dThd and 176 mM dCtd), whereas dTMP and dCMP were co- N 110.0 (Thy). Identification of all compounds was based on the reten- administered at an equimolar dose of 620 mg/kg/day. A group of control tion time and specific ion transitions. Calibration curves made with mice was administered PBS solution (vehicle). aqueous multistandards were processed in parallel, and concentrations Mice were sacrificed at 12 or 29 days of age (except those included were obtained from interpolation of the peak areas using TargetLynx in the survival analysis) 1 h after treatment administration (coinciding software (Waters). with maximum bioavailability of the administered treatments in blood, as assessed in our preliminary pharmacokinetic studies). The 2.5. RNA isolation and analysis of expression health status of treated Tk2KO mice degenerated gradually, showing the first signs of weakness a few days before death. Hence, to prevent RNA was isolated from frozen tissues ground in liquid nitrogen with dehydration and malnutrition at early stages of the disease, we pro- TRIzol Reagent (Invitrogen), resuspended in RNAase-free water, and vided all litters with accessible semi-solid food from day 15. quantified using a NanoDrop spectrophotometer. A 5-μg amount of Circulating peripheral blood (mice at 12 days of age) or cardiac RNA was treated with DNaseI (Ambion DNAse-free) and reverse tran- puncture blood (mice at 29 days of age) was collected in EDTA scribed to cDNA by using the High-Capacity cDNA Archive Kit (Applied (Microvette capillary tubes, SARSTEDT) and processed (3500 g at 4 °C Biosystems). Quantitative real-time PCR reactions were carried out in for 5 min) to obtain plasma. Samples were stored at −20 °C until further 384-well plates with TaqMan Universal PCR Master Mix II, with use. Tissues (brain, liver, and skeletal muscle [gastrocnemius]) were col- UNG (Applied Biosystems). Predesigned TaqMan Gene Expression lected, immediately frozen in liquid nitrogen and stored at −80 °C until Assays (Applied Biosystems) were used for detection of thymidine phos- further use. The small intestine lumen was cleansed with abundant sa- phorylase (Tymp; Mm01301808_m1), cytidine deaminase (Cda, line solution prior to freezing in liquid nitrogen. Mm01341706_m1), thymidine kinase 1 (Tk1, Mm01246403_g1) and deoxycytidine kinase (Dck, Mm00432794_m1) mRNA. Each sample 2.3. mtDNA copy number determination was analysed in triplicate, and the average value was used as the mRNA measure. mRNA levels of cyclophilin A (Ppia; Mm02342430_g1) Total DNA was isolated from frozen tissues ground in liquid nitrogen were used as the reference for normalization among samples. (QIAamp DNA Mini Kit, Qiagen) and dissolved in 10 mM Tris-HCl (pH 8). For each sample, 1:3 dilutions and duplicates were assessed. 2.6. Enzyme activity determinations Quantitative real-time PCR reactions were performed in 384-well plates using the TaqMan Universal PCR Master Mix II, with UNG (Applied Catabolic activities thymidine phosphorylase (TP) and cytidine de- Biosystems). Custom-designed sets of TaqMan primers and probes aminase (CDA) were adapted from Martí et al. [19]. Briefly, frozen tis- were used for relative mtDNA (16S rRNA gene [30]) versus nDNA sues were ground in liquid nitrogen and homogenized in lysis buffer C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 345 (50 mM Tris-HCl, pH 7.2; 1% Triton X-100; 2 mM phenylmethylsulfonyl 2.8. Data availability fluoride [PMSF]; 0.02% 2-mercaptoethanol) by mechanical disruption with a plastic pestle. Homogenates were centrifuged at 20,000g at 4 °C The datasets generated during the experimental procedures and for 30 min, and supernatants collected for protein concentration deter- analysis are available from the corresponding author on reasonable mination by the Bradford method. A 100-μg amount of protein was in- request. cubated at 37 °C for 1 h with 10 mM dThd in a total volume of 100 μl reaction buffer (0.1 M Tris-arsenate, pH 6.5) for TP activity or with 3. Results 5 mM dCtd in a total volume of 100 μl reaction buffer (0.2 M Tris-HCl, pH 8; 100 μM 6-amino-5-bromouracile [6A5BrU]) for CDA activity. A 3.1. Daily administration of pyrimidine deoxynucleosides extends life ex- blank in which dThd or dCtd was omitted, was also processed for each pectancy of a mouse model devoid of TK2 enzyme activity sample. The reaction was stopped by addition of 1 ml of 0.55 M HCl to all samples and 10 mM dThd or dCtd to all blanks. Samples were centri- We treated TK2 knockout (Tk2KO) pups and wild-type littermates fuged at 20,000g at 4 °C for 10 min. Supernatants were injected into an (Tk2WT) with dThd plus dCtd (each at 400 mg/kg/day), an equimolar Acquity UPLC system (Waters, Milford, MA, USA), and eluted at dose of the corresponding dNMPs, dTMP plus dCMP (each at 620 0.5 ml·min-1 with a saline buffer eluent (20 mM ammonium acetate, mg/kg/day), or phosphate buffered saline (PBS, vehicle) by daily oral ga- pH 5.6) and an organic eluent (methanol gradient grade), according to vage starting at postnatal day 4. Previous studies have shown that un- the following gradient: time 0–1.1 min, isocratic 100% saline buffer; treated TK2KO mice die prematurely at around two weeks of age [34]. 1.1–5 min, gradient from 0 to 13.6% methanol; 5–5.1 min, gradient We obtained similar results after daily oral PBS administration. PBS- from 13.6 to 100% methanol; 5.1–6.1 min, isocratic 100% methanol; treated Tk2KO mice exhibited pronounced weight loss starting at 14 to and 6.1–7.2 min, isocratic 100% saline buffer. The column used was an 15 days of age and died shortly after (median, 16 days of age; mean Acquity UPLC BEH C18 column (100 × 2.1 mm, 130 Å pore size, 1.7 ± SD, 16.3 ± 2.6 days of age; n = 20) (Fig. 2). Treatment with either py- μm particle size [Waters]). Eluate absorbance was monitored at rimidine dNMPs (dTMP+dCMP) or the corresponding dNs (dThd 267 nm, and the Thy (product of TP reaction) or dUrd (product of CDA +dCtd) delayed disease progression and strongly attenuated growth reaction) peaks were identified and quantified using a multistandard retardation (Fig. 2). Survival of Tk2KO mice treated with dThd+dCtd or calibration curve. dTMP+dCMP was extended up to 34 or 32 (median) days of age Anabolic activities thymidine kinase 1 (TK1) and deoxycytidine ki- (mean ± SD, 31.9 ± 6.9 days of age; n = 12; mean ± SD, 34.1 ± nase (dCK) were measured using a radiochemical method [24]. Briefly, 6.0 days of age; n = 11, respectively) (Fig. 2A). There were no adverse tissue homogenates were prepared as for TP and CDA determinations effects in Tk2WT littermates treated up to 41 days with either PBS or in lysis buffer (50 mM Tris-HCl, pH 7.2; 1% Triton X-100; 2 mM PMSF; dThd+dCtd (up to 51 days with dTMP+dCMP). 0.02% 2-mercaptoethanol). Homogenates were centrifuged at 20,000g These results are equivalent to those obtained with an alternative for 30 min at 4 °C, and supernatants collected for protein concentration TK2-deficient model, the p.His126Asn knockin mouse, exposed to sim- determination by the Bradford method. A 120-μg amount of extract was ilar treatments [1,10,16]. incubated for 1 h at 37 °C in a reaction buffer (50 mM Tris-HCl, pH 7.6; 5 mM MgCl2; 5 mM ATP; 10 mM NaF; 2 mM DTT; 0.5 mg/ml BSA) con- 3.2. dThd+dCtd administration rescues mtDNA copy number in skeletal taining appropriate [3H]-labelled substrates (Moravek, Inc.) and inhibi- muscle only at early ages tors. For TK1 activity, 25 μM [methyl-3H]-Thd, 1 mM (for brain determinations) or 100 mM (for liver determinations) dCtd, 100 μM We analysed Tk2KO mice and Tk2WT littermates at two time-points, THU, and 100 μM 6A5BrU were used. For dCK activity, 25 μM [3H]- at 12 days of age, close to the stage when untreated or PBS-treated dCtd, 10 mM dThd, 100 μM THU, and 100 μM 6A5BrU were used. THU Tk2KO naturally die [34], and 29 days of age, close to the maximum and 6A5BrU were added to prevent degradation of the substrates. As survival achieved with treatment. At day 12, Tk2KO mice showed a sig- TK2 can phosphorylate both substrates and interfere with specific de- nificantly reduced mtDNA copy number in skeletal muscle (gastrocne- termination of TK1 and dCK activity, we used an excess of competitive mius), and particularly, in brain (Fig. 3). Treatment with dThd+dCtd substrates (dCtd or dThd respectively) to inhibit TK2-dependent phos- rescued mtDNA depletion only in skeletal muscle. On day 29, we were phorylation in the two reactions. In preliminary assays, we determined only able to study Tk2WT and treated Tk2KO animals because PBS- the inhibitory capacity of competitive substrates of each enzyme and treated Tk2KO mice do not survive beyond 20 days of age. At this stage, calculated the final values accordingly. After incubation, the reactions mtDNA depletion was obvious in all analysed tissues of treated Tk2KO were spotted on DEAE Filtermat filters (Perkin Elmer) and left to air- animals (liver, brain, and skeletal muscle) with the exception of small dry. Filters were washed twice for 5 min in 1 mM ammonium formate, intestine, indicating that the treatment had failed to rescue mtDNA and for 5 additional min in water, and then left to air-dry. Dried filters copy number in any case. were coated with MeltiLex scintillation wax (Perkin Elmer). The reac- Furthermore, dThd+dCtd treatment elevated mtDNA copy number tion product retained in the filters was detected and quantified with a in several tissues of Tk2WT mice. This increase was significant at MicroBeta2 system (Perkin Elmer). Results are expressed in pmols of 12 days of age in liver and at 29 days in skeletal muscle (Fig. 3). the product (dTMP or dCMP) formed per mg of protein per min. Activity measurements were done in triplicate and quantified using a calibration 3.3. dThd and dCtd bioavailability markedly decreases during mouse curve. development 2.7. Statistical analysis We collected blood and tissues 1 h after dThd+dCtd administration and determined concentrations of the two dNs and their first catabolic Statistical analysis was performed with GraphPad Prism 6 software. products, thymine (Thy) and deoxyuridine (dUrd), respectively. As The specific tests used for each assay are indicated in the Results section was expected, plasma concentrations of both dNs were significantly or the figure legends. For statistical purposes, undetectable values were higher after treatment, with dCtd reaching half the levels of dThd considered zero. Mice sacrificed for plasma and tissue collection, or mice (Fig. 4). Nevertheless, administration of the same doses of dThd+dCtd that died due to experimental manipulation were considered as cen- resulted in much lower plasma levels of these compounds in 29-day- sored observations in Kaplan-Meier statistical analyses. Detailed infor- old mice (mean ± SD of dThd: Tk2WT-dThd+dCtd 40.6 ± 17.0; Tk2KO- mation on animals included in the survival study is provided in dThd+dCtd 25.4 ± 9.0; mean ± SD of dCtd: Tk2WT-dThd+dCtd 7.5 Supplementary Table 1. ± 9.8; Tk2KO-dThd+dCtd 2.9 ± 1.7) than in the 12-day-old littermates 346 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 Fig. 2. Co-administration of dThd + dCtd increases life expectancy and attenuates disease progression in Tk2KO mice. Tk2KO mice (solid lines) and Tk2WT (dashed lines) littermates were daily administered PBS (vehicle; 10 μl/g), 400 mg/kg of dThd+dCtd, or an equimolar dose of 620 mg/kg of dTMP+dCMP. (A) Survival proportions, Kaplan-Meier curve (Tk2WT-PBS, n = 47; Tk2KO-PBS, n = 58; Tk2WT-dThd+dCtd, n = 55; Tk2KO-dThd+dCtd, n = 48; Tk2WT-dTMP+dCMP, n = 14; Tk2KO-dTMP+dCMP, n = 11. Mantel-Cox test: Tk2KO-PBS versus Tk2WT-PBS, p b .001; Tk2KO-dThd+dCtd versus Tk2KO-PBS, p b .001; Tk2KO-dTMP+dCMP versus Tk2KO-PBS, p b .001; Tk2KO-dTMP+dCMP versus Tk2KO-dThd+dCtd, non-significant differences). Censored observations appear as black ticks in the graph (B) Weight progression for female and male mice (Males: Tk2WT-PBS, n = 30; Tk2KO-PBS, n = 32; Tk2WT-dThd+dCtd, n = 40; Tk2KO-dThd+dCtd, n = 36; Tk2WT-dTMP+dCMP, n = 7; Tk2KO-dTMP+dCMP, n = 3. Females: Tk2WT-PBS, n = 30; Tk2KO-PBS, n = 25; Tk2WT-dThd+dCtd, n = 35; Tk2KO-dThd +dCtd, n = 32; Tk2WT-dTMP+dCMP, n = 7; Tk2KO-dTMP+dCMP, n = 8. Results represent mean values for each time point; error bars indicate SD). (mean ± SD of dThd: Tk2WT-dThd+dCtd 207.9 ± 29.3; Tk2KO-dThd high in small intestine, likely due to active absorption of these com- +dCtd 190.2 ± 54.5; mean ± SD of dCtd: Tk2WT-dThd+dCtd 90.9 ± pounds after oral administration (Fig. 5). 11.2; Tk2KO-dThd+dCtd 82.6 ± 27.1) (Fig. 3). Likewise, although the In general, we found no marked genotype-related differences, ex- age-related differences were not as marked, we observed the same cept for a trend in which dThd and dUrd concentrations were slightly trend for Thy and dUrd plasma levels (Fig. 4). lower in treated Tk2KO mice than in Tk2WT (dThd from 29d Tk2WT- We also determined dThd, dCtd, Thy, and dUrd levels in the 4 differ- dThd+dCtd vs Tk2KO-dThd+dCtd: plasma p b .05, brain and liver p b ent tissues (brain, liver, skeletal muscle, and small intestine). Given the .01; dUrd from 29d Tk2WT-dThd+dCtd vs Tk2KO-dThd+dCtd: plasma presence of dNs-dedicated transporters in cell membranes [33], we ex- and brain p b .05, liver p b .01. P-values obtained with the Mann- pected that the intracellular levels would largely mirror the concentra- Whitney U test) (Figs. 4 and 5, and Supplementary Fig. 1). tion of these compounds in plasma. In fact, dThd+dCtd administration triggered an increase in the levels of all compounds in all tissues of 3.4. Expression and activity of key metabolic enzymes undergo marked on- 12-day-old mice analysed at only 1 h after oral administration (Fig. 5 togenetic changes and Supplementary Fig. 1). Again, levels of the therapeutic dNs and those of their first catabolic products were significantly lower in tissues In the absence of TK2 activity, phosphorylation of administered from older animals. dThd and dCtd concentrations were particularly dThd and dCtd necessarily depends on the catalytic activity of C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 347 Fig. 3. Co-administration of dThd + dCtd rescues mtDNA depletion in targeted tissues of Tk2KO mice only at an early age. mtDNA and nDNA levels were determined by quantitative real- time PCR in mice aged 12 (12d) or 29 days (29d). Results indicate the ratio of mtDNA to nDNA in brain, liver, skeletal muscle (gastrocnemius), and small intestine of Tk2KO (KO; triangles) and Tk2WT (WT; dots) mice, treated with either 400 mg/kg/day of dThd+dCtd (+; blue) or PBS (−; green). Tissues were collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range (mice for each group in brain, n = 7–18; liver, n = 7–18; skeletal muscle, n = 7–18 and small intestine, n = 5–8). P-values obtained with the Mann-Whitney U test: *p b .05; **p b .01; ***p b .001; ns, non-significant differences. Age-related statistical significance lines and p-values are in red. alternative enzymes. Cytosolic deoxypyrimidine kinases, namely An important factor that regulates the bioavailability of adminis- deoxycytidine kinase (dCK) and thymidine kinase 1 (TK1), catalyse tered dNs is the activity of catabolic enzymes responsible for their deg- equivalent reactions to TK2 outside mitochondria. We investigated ex- radation. Thymidine phosphorylase (TP) is the main enzyme catalysing pression of these key anabolic enzymes in tissues of PBS-treated or degradation of dThd in humans and mice (although in the latter, dThd is dThd+dCtd-treated Tk2WT and Tk2KO littermates at 12 and 29 days of also substrate for uridine phosphorylase [8]). Cytidine deaminase (CDA) age. is the enzyme catalysing dCtd deamination, the first step in the catabolic Downregulation of TK1 expression has already been described pathway of this dN [9]. We performed a preliminary longitudinal study during normal mouse development [7]. Accordingly, we found de- of TP and CDA activities in untreated wild-type mice up to 70 days of creased Tk1 mRNA expression in all analysed tissues from 29-day- age. Up to 30 days of age, we observed no major differences over time old mice relative to that of 12-day-old mice (except for small intestine in brain and skeletal muscle (very low activity for both enzymes), but where Tk1 mRNA levels remained unchanged in both age groups) there was a clear increase in the activity of both enzymes in small intes- (Fig. 6A). Interestingly, we found that Tk1 mRNA levels were tine and a decrease in liver (Supplementary Fig. 4). Hence, we decided sustained specifically in brain of treated 29-day-old Tk2KO mice. Be- to focus on liver and small intestine for a deeper study in all treated an- cause we lacked a working antibody that specifically recognized imal groups. mouse TK1 protein, we were unable to determine TK1 protein levels. We confirmed that Tymp mRNA expression (Fig. 7A) and TP activity We also found that 12-day-old Tk2KO-PBS mice had slightly increased (Fig. 7B), were markedly higher in small intestine of older mice. This ob- (in liver) and decreased (in skeletal muscle) Tk1 mRNA levels. How- servation is in agreement with the high Thy concentration found in in- ever, these changes were not confirmed when TK1 enzyme activity testine of 29-day-old mice (Fig. 5). CDA activity was higher than TP was tested in liver and brain (Fig. 6B). In liver, TK1 activity was re- activity at early stages of development, but in small intestine it also in- duced in older animals, whereas in brain, it was particularly low al- creased with age. Upregulation of both TP and CDA enzyme activity ready at 12 days of age. and mRNA levels in small intestine suggests this tissue may be impor- Dck mRNA expression had decreased in liver of all 29-day-old mice tant for limiting the efficacy of orally administered dThd and dCtd. regardless of their genotype, although the decrease was greater in In liver, we also observed an upregulation of Tymp mRNA levels treated Tk2KO mice (Fig. 6A). Enzyme activity (Fig. 6B) determinations (Fig. 7A) and TP activity (Fig. 7B) at 29 days of age, specifically in treated in brain and liver correlated with steady state dCK protein levels Tk2KO animals. Conversely, Cda mRNA expression (Fig. 7A) and activity assessed by western blot (Supplementary Figs. 2 and 3) and confirmed (Fig. 7B) were downregulated in liver from older mice. Some genotype- a considerable dCK reduction in both tissues during normal mouse de- related changes in Cda mRNA expression were also revealed with either velopment. Interestingly, this decreased was not observed in brain increases (in brain) or decreases (in skeletal muscle), specifically in 29- from Tk2KO mice. day-old treated Tk2KO mice (Fig. 7A). Nonetheless, as untreated or PBS- 348 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 Fig. 4. Bioavailability of administered dThd + dCtd descreases with age. Concentrations of dThd, dCtd and their first degradation products (dUrd and Thy) in plasma of mice aged 12 (12d) or 29 days (29d) by LC-MS/MS. Results indicate metabolite concentrations (μM) in plasma of Tk2KO (KO; triangles) and Tk2WT (WT; dots) mice, treated with either 400 mg/kg/day of dThd +dCtd (+; blue) or PBS (−; green). Blood was collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range (12d: Tk2WT-PBS, n = 13; Tk2WT-dThd+dCtd, n = 13; Tk2KO-PBS, n = 17; Tk2KO-dThd+dCtd, n = 6. 29d: Tk2WT-PBS, n = 10; Tk2WT-dThd+dCtd, n = 11; Tk2KO-dThd+dCtd, n = 6). P-values obtained with the Mann-Whitney U test: ***p b .001; ns, non-significant differences. Age-related statistical significance lines and p-values are in red. treated Tk2KO are not viable after 20 days of life [34], we cannot exclude and dUrd for TP activity. We observed a similar relative increase a contribution of chronic daily dThd+dCtd administration to these for dThd and Thy concentration in brain and skeletal muscle of observations. dThd-treated mice. However, in small intestine and liver, either dThd or Thy concentrations were equivalent in some groups 3.5. dThd alone is responsible for the therapeutic effect to those achieved after combined administration of dThd+dCtd (Supplementary Fig. 5). In contrast, acute oral administration of the Even though the two dNs were administered at the same dose, we same dose of dCtd alone resulted in concentrations similar to those systematically observed lower bioavailability for dCtd than for dThd found in animals receiving the combined therapy (Supplementary (Figs. 4 and 5). This observation prompted us to investigate the contri- Fig. 6). bution of each dN to the therapeutic effect. To this end, we administered Analysis of mtDNA copy number in brain, liver and small intestine dThd or dCtd to independent groups of Tk2KO and Tk2WT littermates. Ad- evidenced a better response after administration of dThd alone than ministration of 400 mg/kg/day dCtd had no effect on the lifespan of when combined with dCtd (Fig. 9B and Fig. 3). In general, all groups of Tk2KO (median, 15 days of age; mean ± SD, 17.7 ± 1.5 days of age; n mice experienced a greater increase in mtDNA copy number after = 6), but administration of the same dose of dThd increased life expec- dThd administration, including Tk2WT mice (Fig. 9B). tancy to a median of 33 days (mean ± SD, 34.2 ± 2.6 days of age; n = Importantly, mtDNA copy number nearly normalized in brain of 12- 12), similar to the values obtained after combined administration of day-old Tk2KO animals treated with dThd only, whereas the combined dThd+dCtd at the same dose (Fig. 8A). In addition, the body weight treatment induced no positive response in this tissue. However, in skel- evolution in dThd-treated Tk2KO was equivalent to that observed with etal muscle of mice at the same age, a better response was achieved the combined treatment (Fig. 8B). after combined dThd+dCtd administration (Fig. 9B). These findings In light of these results, we proceeded to investigate bioavailability suggest that complex tissue-specific metabolic regulations may affect and mtDNA copy number in a group of animals given only dThd (400 the outcome of dN-based therapies. mg/kg/day). We found that plasma dThd concentration was around At a later stage, 29-days of age, liver, brain and especially skeletal two-fold higher in these animals than in those administered dThd muscle, experience a marked increase in mtDNA copy number as part +dCtd at the same dose (Figs. 4 and 9). A mild increase in dUrd plasma of their normal development (as indicated by the increase observed in concentration occurred in parallel to the increased bioavailability of Tk2WT mice). At this timepoint, no treatment was able to normalize dThd (Fig. 9), likely because of substrate competition between dThd mtDNA levels. C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 349 Fig. 5. dThd and dCtd concentrations in tissues are lower in older mice. dThd and dCtd concentrations in brain, liver, skeletal muscle (gastrocnemius), and small intestine determined by LC-MS/MS. Results are expressed in nmols per mg of protein in tissues of mice aged 12 (12d) or 29 days (29d). Tk2KO (KO; triangles) and Tk2WT (WT; dots) mice, treated with either 400 mg/kg/day of dThd+dCtd (+; blue) or PBS (−; green). Tissues were collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range. Mice in each group, n = 5. P-values obtained with the Mann-Whitney U test: *p b .05; **p b .01; ns, non-significant differences. 4. Discussion [34] as was shown here, leads to the conclusion that dNs must recruit al- ternative pathways for their ultimate transformation into dNTPs. This Previous studies had shown that dN supplementation delays disease would imply that this type of therapy may be effective for any TK2 path- progression and expands survival in a knockin mouse model for the p. ogenic mutation whether or not there is any residual enzyme activity. His121Asn mutation [10,16]. However, kinetic studies revealed that Although we observed a notable effect on survival, which doubled the mutant protein retains its capacity for dThd phosphorylation, sug- with dThd+dCtd treatment, Tk2KO mice still exhibited growth retarda- gesting that increased substrate concentration might stimulate residual tion and prematurely died before adulthood (median survival, enzyme activity [31]. The fact that dThd+dCtd administration has a 34 days). Thus, the therapy delays disease progression but fails to similar effect on a knockout model devoid of any TK2 enzyme activity completely rescue the phenotype in mice. In an attempt to uncover 350 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 Fig. 6. dN anabolic enzyme mRNA expression and activity decrease during mouse development. We measured mRNA levels and enzyme activity of cytosolic salvage kinases involved in dThd and dCtd activation (TK1 and dCK, respectively) in various targeted tissues from mice aged 12 (12d) or 29 days (29d). Tissues were collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range. P-values obtained with the Mann-Whitney U test: *p b .05; **p b .01; ns, non-significant differences. (A) Relative Tk1 and Dck mRNA levels in brain, liver, skeletal muscle (gastrocnemius), and small intestine, determined by real-time PCR with specific TaqMan probes. Ppia mRNA level was used for normalization between samples. Results are represented as the fold change related to the mean value in 12-day-old PBS-treated Tk2WT mice. Mice in each group, n = 5. (B) TK1 and dCK enzyme activities in brain and liver, determined using a biochemical method with specific radiolabelled substrates. Results are represented as pmol of product formed in 1 min from 1 mg of protein in tissue homogenate. Mice in each group, n = 6. Age-related statistical significance lines and p-values are in red. factors that could be critical in limiting the effectiveness of dNs, we iden- His126Asn knockin mouse model yielded similar results on survival tified ontogenetic changes in the expression of related catabolic enzymes [17], so reduced efficacy of deoxynucleosides with aging is likely due to and necessary cytosolic salvage kinases. dThd and dCtd bioavailability factors other than simply absorption at the level of small intestine. The was clearly reduced in older mice. The activity of CDA and principally activity of the required cytosolic deoxynucleoside kinases (TK1 and TP, which catalyse the first degradation step of dCtd and dThd, respec- dCK) was also markedly altered with age, especially in brain and liver tively, was markedly increased in small intestine of older mice, a decisive where both activities were significantly reduced, which could also con- tissue for the bioavailability of orally administered substances. Accord- tribute to the age-related loss of effect. Overall, our findings regarding ingly, in tissues of older mice, the decrease in dThd and dCtd levels was these limiting factors are in agreement with the observations on a TK2 generally larger than that of their respective catabolic products, in agree- His126Asn knockin mice after IP and oral dThd+dCtd administration ment with higher degradation activities. Other routes of administration, [16,17]. Finally, ontogenetic changes in other factors, such as dN trans- such as intraperitoneal (IP) injection leads to higher bioavailability of porters involved in intestinal absorption and cellular internalization, dThd+dCtd. However, IP administration of dThd+dCtd to a TK2 could further lessen the therapeutic potential of dNs in older mice. C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 351 Fig. 7. dNs catabolic enzyme mRNA expression and activity is upregulated in small intestine of older mice. mRNA levels and enzyme activity of the main catabolic enzymes involved in dThd and dCtd degradation (TP and CDA, respectively) in various target tissues of mice aged 12 (12d) or 29 days (29d). Tissues were collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range. P-values obtained with the Mann-Whitney U test: *p b .05; **p b .01; ns, non-significant differences. (A) Relative Tymp and Cda mRNA levels of in brain, liver, skeletal muscle (gastrocnemius), and small intestine determined by real-time PCR with specific TaqMan probes. Ppia mRNA level was used for nor- malization between samples. Results are represented as the fold change relative to the mean value in 12-day-old PBS-treated Tk2WT mice. Mice in each group, n = 5. (B) TP and CDA en- zyme activities, determined by a biochemical method coupled to UPLC analysis in liver and small intestine. Results are represented as nmol of product formed in 1 h from 1 mg of protein in tissue homogenate. Mice in each group, n = 5. Age-related statistical significance lines and p-values are in red. In addition, important tissue-specific differences were evident al- of the combined therapy on survival in mice actually depends on dThd ready at early ages. As an example, while mtDNA was repopulated in alone. Non-dividing cells are highly dependent on deoxynucleoside sal- skeletal muscle of young treated Tk2KO mice in response to combined vage due to downregulation of de novo dNTP synthesis pathways administration of dThd and dCtd, the same treatment triggered no de- (Fig. 1). Unlike dCK, which is ubiquitously expressed throughout cell tectable increase in mtDNA copy number in brain, another targeted tis- cycle, TK1 expression is highly linked to S-phase during cell cycle sue. dNs can cross the blood-brain barrier through the activity of [13,22]. Thus, mtDNA synthesis demands for dCTP are possibly lower dedicated transporters [23,28]. Accordingly, our results showed a than for dTTP in the absence of functional TK2 activity. In fact, adminis- marked increase in brain dN concentrations already 1 h after adminis- tration only of dThd led to a higher mtDNA repopulation index in most tration, suggesting that lack of effectiveness depends on downstream tissues of 12-day-old mice, including brain, in which levels were much events, such as the extremely low TK1 activity detected in this tissue closer to normalization than after dThd plus dCtd co-administration. which makes it more dependent on TK2 activity [7,27]. This effect is consistent with the finding of higher plasma and tissue Interestingly, although TK2 is responsible for both dThd and dCtd dThd concentration after administration of dThd alone, which is likely phosphorylation in mitochondria, we found that the therapeutic effect due to interactions of both these pyrimidine dNs with transporters 352 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 Fig. 8. dThd is responsible for extended survival and slowed disease progression. in Tk2KO mice. Tk2KO mice (solid lines) and Tk2WT (dashed lines) littermates were daily administered PBS (vehicle; 10 μl/g; green lines), 400 mg/kg of dThd+dCtd (blue lines), 400 mg/kg of dCtd (orange lines) or 400 mg/kg of dThd (red lines). PBS and dThd+dCtd treated mice were included also in Fig. 1. (A) Survival proportions, Kaplan-Meier curve (Tk2WT-dThd, n = 10; Tk2KO-dThd, n = 12; Tk2WT-dCtd, n = 9; Tk2KO-dCtd, n = 6. Mantel-Cox test: Tk2KO-PBS versus Tk2KO- dThd p b .001; Tk2KO-dThd versus Tk2KO-dThd+dCtd no significant differences; Tk2KO-dThd+dCtd versus Tk2KO-dCtd p b .001; Tk2KO-PBS versus Tk2KO-dCtd; ns, non-significant differences). Censored observations appear as black ticks in the graph (B) Growth progression for female and male mice (Males: Tk2WT-dThd, n = 9; Tk2KO-dThd, n = 12; Tk2WT-dCtd, n = 8; Tk2KO-dCtd, n = 3. Females: Tk2WT-dThd, n = 10; Tk2KO-dThd, n = 5; Tk2WT-dCtd, n = 3; Tk2KO-dCtd, n = 4. Results represent mean values for each time point; error bars indicate SD). involved in their absorption [21]. The positive mtDNA synthesis re- Taken together, our results would seem to argue against any need sponse in brain provides evidence that the extremely low TK1 levels for dCtd, as it apparently provided no benefits over dThd alone in in this tissue can still be stimulated by high concentrations of substrate, mice. However, the dCtd increase after treatment was smaller than (ie, dThd), a particularly relevant finding from the standpoint of thera- that of dThd already at early stages of development. Thus, we cannot peutic effectiveness, and is another indication that reduced availability rule out the possibility that a positive response might be triggered in of dNs with age may be an important limitation to this therapy. Tk2KO mice by increasing dCtd availability. Furthermore, although In contrast to what occurs in humans, who mainly manifest TK2 de- higher dThd availability led to better results in brain, it had a detrimen- ficiency as a purely myopathic condition [11,25], mice show marked tal effect on mtDNA copy number in skeletal muscle of young Tk2KO CNS involvement [1,3,34]. Thus, the ability of dThd to repopulate mice, indicating that demands for the different dNTP species may vary mtDNA in brain appeared to be a key factor defining the effectiveness between the various target tissues. A similar observation has been of this treatment in mice. Nevertheless, survival after dThd administra- made by López-Gómez et al. in liver from older TK2-deficient mice tion alone was equivalent to survival after the combined therapy. Up- [17]. Such tissue-specific metabolic requirements and homeostasis regulation of catabolic enzymes, and particularly of TP, at a later age make it difficult to define the contribution of each particular tissue to leads to lower dThd concentrations in plasma and lower mtDNA repop- the therapeutic intervention and to the overall disease phenotype. In ulation in all target tissues, which possibly leads to ultimately equiva- addition, other mouse models for dNTP metabolism-related defects dis- lent effects on survival than after the combined therapy of dThd+dCtd. play phenotypes that are remarkably different from those in humans C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 353 354 C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 [5,18], suggesting considerable divergence in dNTP homeostasis be- applied in clinical use in TK2-deficient patients [6], and they will be of tween the two species. All these observations prevent us from anticipat- interest in the design of an upcoming clinical trial. ing whether dCtd administration may be dispensable in humans, where muscle is actually the main tissue affected. Acknowledgements Other issues should also be considered when analysing factors with a potential impact on therapy efficacy. It is possible that a lack of TK2 ac- We wish to thank the patients and their families for their collabora- tivity could trigger some adaptive changes that would be detected as tion in this study. We also thank Dr. José Luis Martínez Moreno and José genotype-linked differences. Nonetheless, we detected scarce differ- Julián Martínez Padilla for their helpful insight when developing the ences between genotypes at day 12 of age, and on day 29, the observa- project, Dr. Liya Wang for crucial technical advice on enzyme activity tions were challenged by the absence of PBS-treated Tk2KO mice, which determinations, and Dr. Michio Hirano and Dr. Carlos López (Columbia did not survive beyond day 20. Also, animals were exposed to daily University, NY) for fruitful cooperative discussions throughout the treatment starting from postnatal day 4. Thus, we cannot exclude that project. our observations in Tk2KO mice on day 29 may be due to the combined effect of chronic exposure to treatment and genotype adaptations. For Funding sources example, we observed higher Dck and Cda mRNA levels in brain of Tk2KO. Liver TP activity, and consequently Thy levels, were also higher, This study was funded by grants from the Spanish Ministry of Indus- specifically in Tk2KO mice. In agreement with this observation, we try, Economy and Competitiveness [grant BFU2014-52618-R and found that concentrations of dThd and dUrd, both substrates of TP, SAF2017-87506-R to YC], the Spanish Instituto de Salud Carlos III [grant were lower in various tissues of treated Tk2KO (Figs. 4 and 5 and Supple- PI15/00465 and grant PMP 15/00025 to RM, co-financed with ERDF], mentary Fig. 1). the Fundación Inocente, Inocente [grant 2017 to YC], and AFM Téléthon Garone et al. reported an increase in TK1 activity after chronic dTMP [grant 19,965 to YC]. JT was funded by a fellowship granted by the and dCMP administration [10]. We found that Tk1 mRNA levels, which Generalitat de Catalunya (PERIS program, SLT002/16/00370). The normally drop with age, were sustained in brain from treated Tk2KO disclosed funders had no role in study design, data collection and anal- mice (Fig. 6A). This interesting result suggests that the high dThd levels ysis, decision to publish, or preparation of the manuscript. could stabilise TK1 by increasing TK1 transcription. In fact, although TK1 undergoes specific proteolysis in mitotic exit, dThd binding to TK1 has Declaration of interests been shown to stabilise the enzyme during growth arrest in HeLa cells [14]. However, our activity measurements did not corroborate our RM and the Vall d'Hebron Research Institute (VHIR) have filed pat- mRNA observations, which suggest a participation of posttranslational ent applications covering potential use of deoxythymidine and degradation mechanisms that would counteract upregulation of TK1 deoxycytidine treatment for TK2 deficiency in humans. RM, YC, CB, JT, transcription [13]. Furthermore, it is important to note that 29-day-old VHIR and The Biomedical Network Research Centre on Rare Diseases Tk2KO mice are at a late stage of the disease, so some of the effects ob- (CIBERER) have filed patent applications covering potential use of served may be associated with this circumstance. deoxynucleoside treatment for POLG deficiency and other mtDNA rep- TK2 deficiency reduces dTTP and dCTP levels, interfering with lication defects in humans. VHIR and CIBERER have licensed pending mtDNA replication. By activating alternative salvage pathways, the patent applications related to these technologies to MODIS Therapeu- treatment provided enough substrates to counteract mtDNA depletion tics, Inc., and VHIR and CIBERER may be eligible to receive payments re- in most tissues from young Tk2KO mice. In addition, increased dThd con- lated to the development and commercialization of the technologies. centration also led to increased mtDNA levels in treated Tk2WT mice. In Any potential licensing fees earned will be paid to VHIR and CIBERER Tk2WT, expansion of the mitochondrial dTTP pool counts on functional and are shared with all inventors mentioned above through VHIR and TK2 activity in addition to cytosolic salvage. In fact, the increase in CIBERER policies on distribution and objectivity in research. RM serves mtDNA levels was higher in some tissues of Tk2WT than of Tk2KO mice as a paid consultant to MODIS Therapeutics, Inc. and has equity in this (Fig. 9). These results suggest that basal dNTP levels, actually limit company. RM, YC, CB, VHIR and The Biomedical Network Research Cen- mtDNA replication in normal physiological conditions, at least in some tre on Rare Diseases (CIBERER) have filed patent applications covering tissues. potential use of deoxynucleoside treatment as a way to increase mtDNA Unbalanced dNTP pools may have detrimental consequences for copy number. DNA replication fidelity [20]. Mutation frequencies in nuclear and mito- chondrial genome seemed not affected in transgenic Tk2KO mice overex- Authors contribution pressing cytosolic DmdNK, which lead to a 200-fold and 5-fold expansion of dTTP and dCTP pools respectively [15]. Nevertheless, mu- C. Blázquez-Bermejo, R. Martí, and Y. Cámara designed the study; C. tagenesis and other dN therapy-related safety issues will necessarily be Blázquez-Bermejo, D. Molina-Granada, F. Vila-Julià, D. Jiménez-Heis J. addressed in further research. Torres-Torronteras, R. Martí and Y. Cámara performed experiments, col- Despite the significant differences between humans and mice re- lected and analysed the data; X. Zhou and A. Karlsson generated the garding dNTP homeostasis, we believe the results of this study provide TK2-knockout mouse model, actively participated in project discussions insight into the pathomechanisms underlying TK2 deficiency and offer and contributed to data analysis; C. Blázquez-Bermejo and Y. Cámara useful information on key factors that could limit the efficacy of dNs- wrote the manuscript; All contributing authors participated in the crit- based therapies. These data are of value, as dNs are currently being ical reading of the manuscript; Y.Cámara supervised the project, had full Fig. 9. Higher dThd concentration correlates with higher mtDNA copy number in targeted tissues. Concentration of metabolites in plasma and mtDNA copy number in tissues of mice aged 12 (12d) or 29 days (29d) and of Tk2KO (KO; triangles) or Tk2WT (WT; dots) mice, treated with either 400 mg/kg/day of dThd+dCtd (+; blue), dThd (dThd, red), or PBS (−; green). Tissues and blood were collected 1 h after oral gavage. Scatter plots represent the median (horizontal line) with interquartile range. P-values obtained with the Mann-Whitney U test: *p b .05; **p b .01; ***p b .001; ns, non-significant differences. (A) Concentration (μM) of dThd, Thy, dCtd and dUrd in plasma, assessed by LC-MS/MS (12d:Tk2WT-PBS, n = 13; Tk2WT-dThd+dCtd, n = 23; Tk2WT-dThd, n = 8; Tk2KO-PBS, n = 17; Tk2KO-dThd+dCtd, n = 14; Tk2KO-dThd, n = 5. 29d: Tk2WT-PBS, n = 10; Tk2WT-dThd+dCtd, n = 11; Tk2WT-dThd, n = 10; Tk2KO-dThd+dCtd, n = 8; Tk2KO-dThd, n = 9). (B) mtDNA and nDNA levels were determined by quantitative real-time PCR. Results indicate the ratio of mtDNA to nDNA in brain, liver, skeletal muscle (gastrocnemius), and small intestine (mice in each group in brain, n = 5–10; liver, n = 5–10; skeletal muscle, n = 5–10 and small intestine, n = 5–10). C. Blázquez-Bermejo et al. / EBioMedicine 46 (2019) 342–355 355 access to data in the study and had final responsibility for the decision to [16] Lopez-Gomez C, Levy RJ, Sanchez-Quintero MJ, Juanola-Falgarona M, Barca E, Garcia- Diaz B, et al. Deoxycytidine and Deoxythymidine treatment for thymidine kinase 2 submit for publication. deficiency. Ann Neurol 2017;81:641–52. [17] Lopez-Gomez C, Hewan H, Sierra C, Akman HO, Sanchez-Quintero MJ, Juanola- Appendix A. Supplementary data Falgarona M, et al. Bioavailability and activities of cytosolic kinases modulate re- sponse to deoxynucleoside therapy in TK2 deficiency. Submitted 2019;2019. [18] Lopez LC, Akman HO, Garcia-Cazorla A, Dorado B, Marti R, Nishino I, et al. Unbal- Supplementary data to this article can be found online at https://doi. anced deoxynucleotide pools cause mitochondrial DNA instability in thymidine org/10.1016/j.ebiom.2019.07.042. phosphorylase-deficient mice. Hum Mol Genet 2009;18:714–22. [19] Marti R, Lopez LC, Hirano M. Assessment of thymidine phosphorylase function: mea- surement of plasma thymidine (and deoxyuridine) and thymidine phosphorylase References activity. Methods Mol Biol 2012;837:121–33. [20] Mathews CK. DNA precursor metabolism and genomic stability. FASEB J 2006;20: [1] Akman HO, Dorado B, Lopez LC, Garcia-Cazorla A, Vila MR, Tanabe LM, et al. Thymi- 1300–14. dine kinase 2 (H126N) knockin mice show the essential role of balanced [21] Pastor-Anglada M, Urtasun N, Perez-Torras S. Intestinal nucleoside transporters: deoxynucleotide pools for mitochondrial DNA maintenance. Hum Mol Genet function, expression, and regulation. Compr Physiol 2018;8:1003–17. 2008;17:2433–40. [22] Rampazzo C, Miazzi C, Franzolin E, Pontarin G, Ferraro P, Frangini M, et al. Regulation [2] Alston CL, Schaefer AM, Raman P, Solaroli N, Krishnan KJ, Blakely EL, et al. Late-onset by degradation, a cellular defense against deoxyribonucleotide pool imbalances. respiratory failure due to TK2 mutations causing multiple mtDNA deletions. Neurol- Mutat Res 2010;703:2–10. ogy 2013;81:2051–3. [23] Redzic ZB, Malatiali SA, Grujicic D, Isakovic AJ. Expression and functional activity of [3] Bartesaghi S, Betts-Henderson J, Cain K, Dinsdale D, Zhou X, Karlsson A, et al. Loss of nucleoside transporters in human choroid plexus. Cerebrospinal Fluid Res 2010;7 thymidine kinase 2 alters neuronal bioenergetics and leads to neurodegeneration. (2). Hum Mol Genet 2010;19:1669–77. [24] Rylova SN, Mirzaee S, Albertioni F, Eriksson S. Expression of deoxynucleoside kinases [4] Camara Y, Gonzalez-Vioque E, Scarpelli M, Torres-Torronteras J, Caballero A, Hirano and 5′-nucleotidases in mouse tissues: implications for mitochondrial toxicity. M, et al. Administration of deoxyribonucleosides or inhibition of their catabolism as Biochem Pharmacol 2007;74:169–75. a pharmacological approach for mitochondrial DNA depletion syndrome. Hum Mol [25] Saada A, Shaag A, Mandel H, Nevo Y, Eriksson S, Elpeleg O. Mutant mitochondrial Genet 2014;23:2459–67. thymidine kinase in mitochondrial DNA depletion myopathy. Nat Genet 2001;29: [5] Dalla Rosa I, Camara Y, Durigon R, Moss CF, Vidoni S, Akman G, et al. MPV17 loss 342–4. causes deoxynucleotide insufficiency and slow DNA replication in mitochondria. [26] Saada A, Ben-Shalom E, Zyslin R, Miller C, Mandel H, Elpeleg O. Mitochondrial PLoS Genet 2016;12:e1005779. deoxyribonucleoside triphosphate pools in thymidine kinase 2 deficiency. Biochem [6] Dominguez-Gonzalez C, Madruga-Garrido M, Mavillard F, Garone C, Aguirre- Biophys Res Commun 2003;310:963–6. Rodriguez FJ, Donati MA, et al. Deoxynucleoside therapy for thymidine kinase 2 [27] Saada A, Shaag A, Elpeleg O. mtDNA depletion myopathy: elucidation of the tissue (TK2) deficient myopathy. Ann Neurol 2019;86:293–303. specificity in the mitochondrial thymidine kinase (TK2) deficiency. Mol Genet [7] Dorado B, Area E, Akman HO, Hirano M. Onset and organ specificity of Tk2 deficiency Metab 2003;79:1–5. depends on Tk1 down-regulation and transcriptional compensation. Hum Mol [28] Spector R, Eells J. Deoxynucleoside and vitamin transport into the central nervous Genet 2011;20:155–64. system. Fed Proc 1984;43:196–200. [8] el Kouni MH, el Kouni MM, Naguib FN. Differences in activities and substrate speci- [29] Tyynismaa H, Sun R, Ahola-Erkkila S, Almusa H, Poyhonen R, Korpela M, et al. Thy- ficity of human and murine pyrimidine nucleoside phosphorylases: implications for midine kinase 2 mutations in autosomal recessive progressive external chemotherapy with 5-fluoropyrimidines. Cancer Res 1993;53:3687–93. ophthalmoplegia with multiple mitochondrial DNA deletions. Hum Mol Genet [9] Fanucchi MP, Watanabe KA, Fox JJ, Chou TC. Kinetics and substrate specificity of 2012;21:66–75. human and canine cytidine deaminase. Biochem Pharmacol 1986;35:1199–201. [30] Walker UA, Auclair M, Lebrecht D, Kornprobst M, Capeau J, Caron M. Uridine abro- [10] Garone C, Garcia-Diaz B, Emmanuele V, Lopez LC, Tadesse S, Akman HO, et al. gates the adverse effects of antiretroviral pyrimidine analogues on adipose cell func- Deoxypyrimidine monophosphate bypass therapy for thymidine kinase 2 defi- tions. Antivir Ther 2006;11:25–34. ciency. EMBO Mol Med 2014;6:1016–27. [31] Wang L, Saada A, Eriksson S. Kinetic properties of mutant human thymidine kinase 2 [11] Garone C, Taylor RW, Nascimento A, Poulton J, Fratter C, Dominguez-Gonzalez C, suggest a mechanism for mitochondrial DNA depletion myopathy. J Biol Chem 2003; et al. Retrospective natural history of thymidine kinase 2 deficiency. J Med Genet 278:6963–8. 2018;55:515–21. [32] Wang L. Mitochondrial purine and pyrimidine metabolism and beyond. Nucleosides [12] Gotz A, Isohanni P, Pihko H, Paetau A, Herva R, Saarenpaa-Heikkila O, et al. Thymi- Nucleotides Nucleic Acids 2016;35:578–94. dine kinase 2 defects can cause multi-tissue mtDNA depletion syndrome. Brain [33] Young JD. The SLC28 (CNT) and SLC29 (ENT) nucleoside transporter families: a 30- 2008;131:2841–50. year collaborative odyssey. Biochem Soc Trans 2016;44:869–76. [13] Hu CM, Chang ZF. Mitotic control of dTTP pool: a necessity or coincidence? J Biomed [34] Zhou X, Solaroli N, Bjerke M, Stewart JB, Rozell B, Johansson M, et al. Progressive loss Sci 2007;14:491–7. of mitochondrial DNA in thymidine kinase 2-deficient mice. Hum Mol Genet 2008; [14] Ke PY, Hu CM, Chang YC, Chang ZF. Hiding human thymidine kinase 1 from APC/C- 17:2329–35. mediated destruction by thymidine binding. FASEB J 2007;21:1276–84. [35] Zimmermann H, Zebisch M, Strater N. Cellular function and molecular structure of [15] Krishnan S, Paredes JA, Zhou X, Kuiper RV, Hultenby K, Curbo S, et al. Long term ex- ecto-nucleotidases. Purinergic Signal 2012;8:437–502. pression of Drosophila melanogaster nucleoside kinase in thymidine kinase 2- deficient mice with no lethal effects caused by nucleotide pool imbalances. J Biol Chem 2014;289:32835–44.