Abstract
Background/Aim: GS 9219 is a double prodrug of antiproliferative nucleotide analog 9-(2-Phosphonylmethoxyethyl)guanine (PMEG), with potent in vivo efficacy against various hematological malignancies. This study investigates the role of adenosine deaminase-like (ADAL) protein in the intracellular activation of GS-9219. Materials and Methods: A cell line resistant to 9-(2-Phosphonylmethoxyethyl)-N6-cyclopropyl-2,6-diaminopurine (cPrPMEDAP), an intermediate metabolite of GS-9219, was generated and characterized. Results: The resistant cell line was cross-resistant to cPrPMEDAP and GS-9219, due to a defect in the deamination of cPrPMEDAP to PMEG. Mutations in the ADAL gene (H286R and S180N) were identified in the resistant cells that adversely-affected its enzymatic activity. Introduction of the wild-type ADAL gene re-sensitized resistant cells to both cPrPMEDAP and GS-9219. Conclusion: The ADAL protein plays an essential role in the intracellular activation of GS-9219 by catalyzing the deamination of cPrPMEDAP metabolite to PMEG. Mutations affecting the activity of ADAL confer resistance to both GS-9219 and its metabolite cPrPMEDAP.
In addition to their well-established broad-spectrum antiviral activity, some of the acyclic nucleoside phosphonate analogs are potent antitumor agents (1). 9-(2-Phosphonylmetho-xyethyl) guanine (PMEG) is one such example as it exhibits potent antiproliferative effects in multiple in vitro and in vivo models. In cells, PMEG is transformed to an active phosphorylated metabolite PMEG diphosphate (PMEGpp) that acts as a potent inhibitor of nuclear DNA polymerases α, δ and ε, blocking DNA synthesis and repair via effective incorporation and chain termination (2, 3). GS-9219 was designed as a novel double prodrug of PMEG consisting of the N6-substituted prodrug of PMEG, 9-(2-phospho-nylmethoxyethyl)-N6-cyclopropyl-2,6-diaminopurine (cPrPMEDAP), and two symmetrical ethyl-alanine phosphonoamidate groups that mask the charged hydrophilic phosphonate moiety. The N6-cyclopropyl prodrug moiety allows for a specific intracellular activation, limiting the plasma exposure to the parent compound PMEG that can adversely affect renal function in vivo. The phosphono-amidate prodrug moieties increase the delivery of PMEG into lymphoid cells and tissues (4-6).
The intracellular metabolic activation of GS-9219 is shown in Figure 1. GS-9219 is hydrolyzed by intracellular proteases to generate cPrPMEDAP. Serine protease cathepsin-A has been shown to proteolytically cleave GS-9219 in an in vitro enzymatic assay (6). In a second metabolic step, cPrPMEDAP is deaminated to PMEG. The adenosine deaminase-like (ADAL) protein, also named N6-methyl-AMP aminohydrolase or abacavir monophosphate deaminase, is able to deaminate cPrPMEDAP in vitro (6, 7). Subsequently, PMEG is phosphorylated by intracellular kinases to the active metabolite PMEGpp. The first phosphorylation to PMEGp is catalyzed by guanosine monophosphate kinase whereas the non-specific nucleoside diphosphate kinase phosphorylates PMEGp to PMEGpp (8-10). PMEGpp competes with the natural substrate dGTP for incorporation by DNA polymerases (2, 11). The incorporation of PMEGpp into newly-synthesized DNA during chromosomal replication leads to DNA chain termination and induction of apoptosis. Thus, the inhibition of DNA synthesis is the primary mechanism of action of GS-9219 in replicating cells (6). Furthermore, GS-9219 inhibits DNA repair in chronic lymphocytic leukemia (CLL) cells from patients and activates cell signaling pathways leading to the induction of apoptosis (3).
In domestic dogs, the incidence rate of lymphomas is 5% to 7% of all diagnosed canine neoplasms and accounts for 85% of canine hematopoietic tumors (12). Canine lymphomas are commonly treated with conventional multi-agent chemotherapy, e.g. cyclophosphamide/doxorubicin/vincristine/prednisone (CHOP) to achieve remission in 60% to 90% of the dogs for 6 to 12 months, depending on the protocol used (13). Following relapse, several second-line therapies are attempted to achieve remission but success rates are low and remissions only last for a few months (14-17). Recurrent or refractory disease remains problematic and is an area of unmet medical need. GS-9219 exhibits potent in vivo antiproliferative activity in a canine model of spontaneously occurring non-Hodgkin's lymphoma (6). In this model, an antitumor response was observed in 79% of previously-untreated or relapsed/refractory dogs (18). GS-9219 is currently being developed as an antineoplastic agent for veterinary applications under the designation VDC-1101 (http://vet-dc.com/gs-9219-gilead-sciences.html). Investigating the importance of enzymes involved in the activation of GS-9219 and the impact of mutations on their enzymatic activity will help to understand the development of resistance in cancer treatment.
In the present study, we examined the importance of ADAL-dependent deamination in the conversion of GS-9219 to PMEG by generating a cell line resistant to the intermediate metabolite cPrPMEDAP. We investigated the enzymatic activity and substrate specificity of ADAL mutants identified in the resistant cells and performed structural modeling of ADAL to analyze the molecular mechanism of the nucleotide deamination step. To our knowledge, this is the first report characterizing functionally relevant mutations in ADAL, an enzyme that is emerging as an important pharmacological target for the activation of therapeutically relevant nucleoside and nucleotide prodrugs such as the antineoplastic agent GS-9219 and the antiretroviral agent abacavir.
Materials and Methods
Chemicals and reagents. GS-9219, cPrPMEDAP, PMEG, adefovir [9-[2-(phophonylmethoxyethyl)adenine (PMEA)], cidofovir [(S)-1-(3-hydroxy-2-phosphonylmethoxypropyl)cytosine (HPMPC)] and its cyclic form (cHPMPC) were synthesized at Gilead Sciences, Inc. (S)-9-(3-hydroxy-2-phosphonylmethoxypropyl)adenine (HPMPA) and 9-(2-phosphonylmethoxyethyl)-2,6-diaminopurine (PMEDAP) were provided by the Institute of Organic Chemistry and Biochemistry, Academy of Sciences of Czech Republic. Cytarabine (Cytosar®) was from Pfizer Belgium (Puurs, Belgium), cisplatin (Platosin®) and fludarabine (Fludarabini Phosphas®) from Teva Pharma Belgium (Wilrijk, Belgium), and gemcitabine (Gemzar®) from Eli Lilly Benelux (Brussels, Belgium). [14C]GS-9219 (57 mCi/mmol) was synthesized by Moravek Biochemicals (Brea, CA, USA). Doxorubicin was purchased from Sigma-Aldrich (St. Louis, MO, USA). N6-Methyl-AMP (Me-AMP) and abacavir-monophosphate (ABC-MP) were synthesized from the corresponding nucleosides according to Yoshikawa et al. (19) and purified by chromatography on POROS 50 HQ columns (Applied Biosystems, Foster City, CA, USA). N6-Methyl-dAMP (Me-dAMP) was synthesized by methylation of dAMP (Calbiochem, San Diego, CA) with trimethyl phosphate, according to Tanabe et al. (20). All other chemicals were from Sigma-Aldrich (St. Louis). Calf intestine phosphatase was from New England Biolabs (Ipswich, MA, USA).
Cell culture and resistance selection. C33A cervical carcinoma cells were purchased from the American Type Culture Collection (Manassas, VA, USA) and were cultured in Dulbecco's Modified Eagle's Medium (DMEM) with 2 mM GlutaMax, antibiotics (all from Life Technologies, Grand Island, NY, USA) and 10% fetal bovine serum (Hyclone, Logan, UT, USA). C33A cells were exposed to increasing concentrations of cPrPMEDAP, starting at 0.6 μM. As soon as the cells recovered from the growth inhibition by cPrPMEDAP, the drug concentrations were increased two-fold. After approximately 45 passages, a cell line was isolated capable of growing in the presence of 164 μM cPrPMEDAP and was designated C33A-Res.
Cytotoxic and cytostatic assays. The cytotoxic effect of tested compounds was determined using a CellTiter-Glo luminescent cell viability assay (Promega, Madison, WI, USA). Briefly, 10,000 cells/well were seeded in 96-well plates and compounds were added in a dose-dependent manner starting at 300 μM or 1,000 μM at five-fold dilutions. After five-day incubation, CellTiter-Glo reagent was added and the luminescence was analyzed on a Victor Luminescence Reader (Perkin-Elmer, Waltham, MA, USA). Data analysis was performed with XLfit (ID Business Solutions, Alameda, CA, USA) using non-linear regression analysis. The CC50 value for each tested compound and cell line was defined as a concentration reducing cell viability by 50% compared to non-treated cells. Cytotoxicity assays were also performed on C33A-Res cell line transiently transfected with various ADAL forms. For these studies, 1×107 cells were seeded in T75 flask 24 h before transfection and 20 μg plasmid DNA was transfected using the TransIt transfection reagent (Mirrus Bio, Madison, WI, USA) according to the manufacturer's protocol. Transfected cells were harvested after 24 h and seeded at 5,000 cells/well in 96-well plates. After an additional 24 h incubation, tested compounds were added and samples were analyzed for the cytotoxic effect as described above.
Cytostatic activities of cPrPMEDAP and other nucleoside phosphonates, as well as additional control anti-proliferative compounds, were determined from their effect on cell growth. Exponentially growing cells were seeded onto 96-well plates at a density of 2.5×103 cells/well and incubated with serial dilutions of tested compounds for seven days. Cell counts were determined using a Z1 Coulter Counter apparatus (Beckman Coulter, Fullerton, CA, USA). The concentration of drug that inhibited cell proliferation by 50% (IC50) was determined from a plot of cell count (%) versus compound concentration.
Analysis of GS-9219 metabolism. Cells (6×107) seeded in polyamine-coated T75 flasks (BD Biosciences, San Jose, CA, USA) were incubated in duplicates with 10 μM [14C]GS-9219 for 24 h. Cells were washed three times with cold phosphate-buffered saline (PBS), harvested using 0.5% trypsin EDTA (Life Technologies) and pelleted by centrifugation. Cell pellets were resuspended in 70% methanol, stored at −80°C for 24 h and centrifuged at rcf 18,883 ×g for 10 min to remove denatured proteins. Methanolic extracts were lyophilized, resuspended in 100 μl water and a small sample aliquot was analyzed by scintillation counting to calculate total amount of radiolabeled metabolites. One set of the duplicates was treated with 1 U/ml of calf intestine phosphatase (New England Biolabs) for 30 min at 37°C to dephosphorylate the metabolites. Samples were separated by high-performance liquid chromatography (HPLC) on a 5 μm ODS3, 150×4 mm Prodigy column (Phenomenex, Torrance, CA, USA) using a gradient of 0 to 70% acetonitrile in 25 mM K2HPO4 (pH 6) and 5 mM tetrabutylammonium bromide. Individual separated metabolites were quantified using an on-line radiomatic detection unit (Waters, Medford, MA, USA).
Cloning, mutagenesis and sequencing of ADAL. The cloning of the wild-type human ADAL gene was described previously (21). An expression plasmid with a short isoform of ADAL (GenBank accession number NM_001012969) was from OriGene Technologies (Rockville, MD, USA). Genes encoding mutant human ADAL and wild-type canine Adal (GenBank accession number: XM_544649) were generated by gene synthesis (Genscript, Piscataway, NJ, USA). All genes were cloned into the pCMV-3flag vector (Agilent Technologies, Santa Clara, CA, USA). Empty and enhanced green fluorescent protein (EGFP)-containing pCMV-3flag DNA plasmids were used as controls. For sequencing, RNA from 1×106 C33A cells was extracted with Qiagen RNeasy kit (Qiagen, Valencia, CA, USA) and converted to cDNA with the High Capacity RNA-to-cDNA kit (Applied Biosystems, Foster City, CA, USA). Adal was amplified by PCR using forward (5’-CTG CAG GAA TTC GAT ATC CGC CAT GGG AAT AGA GGC-3’) and reverse (5’-AAT CCT CGA GGT CGA CAA TAT GTA ACA CTC TGG GCT TCA GG-3’) primers and subsequently cloned to pCMV-3flag vector (Stratagene, La Jolla, CA, USA). Sequencing was performed either directly on Adal PCR products or on individual Adal-containing constructs by Elim Biopharmaceuticals (Hayward, CA, USA) using following Adal-specific primer sets: 5’-CAT CAG CTT ACT AGT AGC CCT GAA-3’, 5’-GAA AAC TTG GAC ATT GAT GTT AGG-3’, 5’-ACT CCT GGA TCT GCT TCC TGA C-3’, 5’-TGA TCT GTA CTG ATG ATA AGG GTG T-3’, 5’-CAG CTG GTA CTC TTG AGA AAG GT-3’, 5’-TTG GAA TCT CTG AAA GAT GCA A-3’, 5’-CCT AAC ATC AAT GTC CAA GTT TTC-3’, 5’-CAA AGT TCT TTT CTT TCC CTT GTC-3’, T3 promotor (5’-AAT TAA CCC TCA CTA AAG GG-3’) and T7 reverse (5’-GCT AGT TAT TGC TCA GCG G-3’). Sequence alignment was carried out with Vector NTI (Life Technologies).
Generation of stable Adal-expressing cell lines in C33A mutant background. Adal plasmid DNA was transfected into C33A-Res cells (20 μg/107 cells) using Mirus Bio TransIt reagent (Madison, WI, USA) according to the manufacturer's protocol. Pools of stably-transfected cells were selected in the presence of 1 mg/ml geneticin (Life Technologies). Expression of respective Adal proteins in the generated stable cell lines was verified using anti-Flag western blot analysis as described below.
Western blot analysis. Cells were washed with PBS, harvested using 0.5% trypsin/EDTA and pelleted. Subsequently, 1×107 cells were lysed in 30 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 1 mM PMSF, 1% Triton X-100, 10% glycerol and 1× protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN, USA) and incubated on ice for 10 min. Lysates were centrifuged for 10 min at rcf 21842 ×g and 4°C. The cellular extracts were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (4 to 12% Bis-Tris gels; NuPAGE; Life Technologies) and transferred to a nitrocellulose membrane (Life Technologies). The membrane was blocked with 3% bovine serum albumin (Roche Diagnostics) in 40 mM Tris-HCl pH 7.4, 150 mM NaCl and 0.1% Tween (TBS-T). Blots were incubated with M2 anti-flag antibody (Sigma-Aldrich) or antibody to α-tubulin (Sigma-Aldrich) for 1 h and washed three times for 10 min with TBS-T. Membranes were incubated with a secondary goat anti-mouse IgG coupled to horseradish peroxidase (Southern Biotech, Birmingham, AL, USA) for 1 h and washed three-times for 10 min with TBS-T. Protein bands were detected with the PIERCE ECL Western Blotting Kit (Thermo Fisher Scientific, Rockford, IL, USA) and Amersham Hyperfilms (GE Healthcare, Piscataway, NJ). Exposed Hyperfilms were scanned and the protein bands were quantified with ImageJ (22).
Recombinant deaminase expression and purification. The baculovirus-based expression in insect cells and purification of recombinant human ADAL proteins were described previously (21). Briefly, wild-type or mutant ADAL genes or the spleen tyrosine kinase dead (SYK KD) mutant gene were sub-cloned into the baculovirus expression shuttle vector pFASTBacHT (Life Technologies) and the sequence was verified. Baculovirus genomic DNA (bacmid) was produced in the Escherichia coli strain DH10Bac. Recombinant baculovirus stocks were generated by transfection of SF9 insect cells with the isolated bacmids. Hi5 insect cells were maintained at a density of 1×106 cells/ml and infected with the respective baculovirus at an optimal multiplicity of infection (MOI) for soluble protein expression.
For the generation of crude cell extracts, the insect cells were harvested 48 h after infection and lysed. The total protein concentration in the crude lysates was determined and the expression level of the His-tagged ADAL protein was quantified by anti-His western blotting.
To purify the ADAL wild-type and S180N mutant enzymes, 1×109 Hi5 insect cells were infected with corresponding recombinant baculovirus, harvested by centrifugation 48 h after infection and lysed by microfluidization. The cleared homogenate (40 min at rcf 30,000 ×g) was loaded onto a 5-ml Ni-NTA column (Qiagen) and the eluted fractions were subjected to SDS-PAGE analysis. Fractions containing the desired protein were pooled and loaded onto a 1-ml Q HP column (Amersham Biosciences AB, Uppsala, Sweden). Collected fractions were analyzed by SDS-PAGE and the protein-containing fractions were pooled. The final protein preparation showed the purity of >95% based on SDS-PAGE analysis and its identity was verified by mass spectroscopy. Native gel filtration confirmed the presence of a 42 kDa protein, indicating primarily a monomeric form.
ADAL activity in cell extracts. ADAL activity was determined in reaction buffer [50 mM HEPES pH 7.5, 250 mM NaCl, 2.5 mM 1,4-Dithioerythritol (DTT)] with 15 μl cell extract in a total reaction volume of 150 μl. [14C]O6-nPrPMEG (25 μM) was used as a highly efficient ADAL substrate. Following incubation for 30 min at 37°C, 25 μl reaction aliquots were quenched with 80 μl 100% methanol, stored at −80°C for 24 h and centrifuged at rcf 18,883 × g for 10 min to remove denatured proteins. Methanolic extracts were lyophilized, resuspended in 70 μl water and separated by HPLC on a 5 μm ODS3, 150×4 mm Prodigy column (Phenomenex) using a gradient of 0 % to 70% acetonitrile in 25 mM K2HPO4 (pH 6) and 5 mM tetrabutylammonium bromide. The product of the hydrolysis (PMEG) was quantified using an on-line radiomatic detection unit (Waters) and normalized against the expression level of His-tagged proteins.
In vitro enzymatic ADAL assay. The enzymatic deamination assay was described previously (7). Briefly, the deamination assay was run in 50 μl reaction mixtures containing 1 μg/ml recombinant wild-type or S180N ADAL enzyme, 50 μM substrate (N6-Methyl-AMP, N6-Methyl-dAMP, abacavir-monophosphate or cPrPMEDAP), 50 mM PIPES (pH 6.8) and 2 mM DTT. The reactions were carried out at 37°C for 20 min and stopped by the addition of 50 μl 10% trichloroacetic acid (TCA). After 10 min incubation on ice, the samples were centrifuged and TCA was removed from the supernatant using extraction with tri-n-octylamine/1,1,2-trichlorotrifluoroethane mixture (1:4, v/v). The aqueous phase was then separated by centrifugation. The acid-soluble extract was analyzed in a Waters HPLC system (996 PDA Detector, PDA Software Millenium32, version 3.05, 616 Pump with 600S Controller) equipped with 150×4 mm Supelcosil™ LC-18T 3 μm reverse-phase column. A gradient of 0 to 30% acetonitrile in 50 mM potassium dihydrogen phosphate and 3 mM tetrabutylammonium hydrogen sulfate, pH 5.1 was used at a flow rate of 0.75 ml/min. Peaks of substrates (N6-Methyl-AMP, N6-Methyl-dAMP, abacavir-monophosphate, and cPrPMEDAP) and corresponding products (IMP, dIMP, carbovir-monophosphate and PMEG, respectively) were detected by UV absorption, identified with the aid of an external standard spectra library and quantified by standard integration method.
Computational modeling of ADAL proteins with substrate. A homology model of the human ADAL active site was built from an X-ray crystal structure of murine adenosine deaminase (Ada) containing 2-deaza-adenosine in the active site (pdb code: 1ADD, 2.4Å resolution). The active sites of ADAL and Ada exhibit 50% amino acid sequence identity. Residues 19, 61, 62, 69, 103, 106, 155, 183, and 269 from the Ada crystal structure were mutated to reflect the sequence of the active site of human ADAL and 2-deaza-adenosine molecule was manually converted to cPrPMEDAP. The mutated residues and the modified substrate were then submitted to an extensive conformational search while keeping the rest of the protein fixed, using Macromodel as implemented in Maestro version 9.1 (Schrödinger, Portland, OR, USA). To evaluate the spatial changes in the active site of ADAL caused by the S180N mutation, a corresponding S183N mutation was introduced into the modified Ada sequence. Finally, a conformational search of the asparagine side chain in the absence of the cPrPMEDAP substrate was carried out to identify its preferred orientation within the active site.
Results
Selection and phenotyping of parental and cPrPMEDAP-resistant cell lines. Selection for cPrPMEDAP resistance was achieved by exposure of the parental C33A cells (C33A-WT) to increasing concentrations of the compound, starting at the half-maximal cytotoxic concentration (CC50) for cPrPMEDAP. After 45 passages in the presence of cPrPMEDAP, the selected cells (C33A-Res) were expanded and subjected to phenotypic drug sensitivity assays against PMEG, cPrPMEDAP and GS-9219 (Table I). In the parental cell line CC33A-WT, GS-9219 exhibited higher potency (CC50=0.62 μM) compared to cPrPMEDAP and PMEG (CC50=3.0 μM and 1.2 μM, respectively). In contrast, the cPrPMEDAP-selected C33A cells were highly-resistant to both cPrPMEDAP and GS-9219 (CC50>300 μM and 181.9 μM, respectively), but showed only a minor, 4.6-fold decrease in sensitivity to PMEG (CC50=5.6 μM). In addition, doxorubicin, a potent antineoplastic agent with a different mechanism of action, showed similar activity both in C33A-WT and C33A-Res cells. The lack of a major change in the susceptibility of C33A-Res cells to PMEG and doxorubicin together with the high-level cross-resistance against cPrPMEDAP and its prodrug GS-9219 are suggestive of a perturbation in the intracellular pharmacology of cPrPMEDAP.
The cytostatic activity of cPrPMEDAP and other nucleoside phosphonates as well as additional control compounds was compared between the C33A-WT and C33A-Res cells using a cell counting method following a 7-day treatment with the indicated compounds (Figure 2). The C33A-Res cells showed selective high-level resistance against cPrPMEDAP (>200-fold) and only a minor reduction in susceptibility to PMEG (3.1-fold) and PMEDAP (4.9-fold), a close analog of cPrPMEDAP. None of the other tested compounds exhibited loss of activity in C33A-Res cells. This result is consistent with observations from the cytotoxicity assays suggesting a specific change in the intracellular metabolism of cPrPMEDAP. The cPrPMEDAP resistance remained stable in the absence of selecting agent for at least 10 additional passages (data not shown).
Metabolism of GS-9219 in C33A-WT and C33A-Res cell lines. To investigate whether a defect in the downstream conversion of cPrPMEDAP to PMEG might be responsible for the drug-resistance phenotype, we compared the intracellular metabolism of [14C]GS-9219 in the parental C33A-WT and C33A-Res cell lines. In the C33A-WT cell line, we detected all major metabolites of GS-9219 including the active metabolite PMEGpp (6, 18). Intact prodrug GS-9219 was not detected in cell extracts, indicating a fast conversion to cPrPMEDAP-Ala, the proximal proteolytic cleavage product of GS-9219 (Figure 3A). The detection of cPrPMEDAP indicated efficient hydrolysis of cPrPMEDAP-Ala. Although we barely measured any detectable amounts of PMEG in C33A-WT cells, we observed significant levels of PMEGp and PMEGpp, indicating rapid phosphorylation of PMEG to PMEGp and subsequently PMEGpp. In contrast to the parental C33A cell line, cPrPMEDAP-Ala and cPrPMEDAP were the only metabolites detected in the C33A-Res cell line. The lack of detectable levels of PMEG, PMEGp and PMEGpp suggests a defect in the conversion of cPrPMEDAP to PMEG.
To simplify the metabolic analysis of GS-9219 activation, we treated the extracted metabolites with calf intestine phosphatase to convert PMEGp and PMEGpp to PMEG. As expected, we detected cPrPMEDAP-Ala, cPrPMEDAP and PMEG in the de-phosphorylated extract from the C33A-WT cell line with PMEG being the predominant metabolite (Figure 3B). In contrast to the parental C33A cell line, cPrPMEDAP was the predominant metabolite in C33A-Res cell line with no detectable PMEG following the dephosphorylation of cell extract. This further supports the conclusion that the conversion of cPrPMEDAP to PMEG is impaired in the C33A-Res cell line.
Restoration of the sensitivity of C33A-Res cells by ADAL expression. Previous studies have shown that a recombinant human ADAL protein can catalyze the enzymatic conversion of cPrPMEDAP to PMEG in vitro (6, 7). To further confirm that the conversion of cPrPMEDAP to PMEG is impaired in the C33A-Res cell line, we transiently transfected an expression plasmid encoding the wild-type ADAL protein into these cells. We then measured the cytotoxicity of GS-9219, cPrPMEDAP, PMEG and doxorubicin in these transfected cells and compared it to the cytotoxicity of the same compounds in C33A-Res and C33A-WT cell lines both transfected with the control empty vector. Introduction of the ADAL protein into the C33A-Res cell line resulted in >15-fold sensitization of the cells to both GS-9219 and cPrPMEDAP treatment as compared to the same cells transfected with the control empty expression plasmid (Table II). This effect was specific to cPrPMEDAP and GS-9219 as ADAL expression did not affect the sensitivity of C33A-Res cells to PMEG or doxorubicin. Taken together, these findings indicate that the resistance phenotype of the C33A-Res cell line was overcome by the expression of the ADAL protein. In addition, it can be concluded that the phosphorylation of PMEG is not impaired in the C33A-Res cell line as only the sensitivity to GS-9219 and cPrPMEDAP, but not to PMEG, is changed in the C33A-Res cells expressing the ADAL protein.
Genotypic analysis of ADAL gene in C33A-Res cells. We investigated whether any mutation might have occurred in the ADAL gene as a result of the cPrPMEDAP resistance selection process that could help explain the lack of cPrPMEDAP conversion to PMEG in the C33A-Res cell line. We sequenced the entire cDNA from the ADAL gene generated by the RT-PCR amplification of RNA isolated from both the C33A-WT and C33A-Res cell lines. The population sequencing analysis detected the presence of two mutations in the ADAL mRNA amplified from the C33A-Res cells, a serine to asparagine substitution at the amino acid residue 180 (S180N) and a histidine to arginine substitution at the position 286 (H286R). Subsequent clonal cDNA sequencing confirmed the presence of both mutations with the majority of ADAL clones (11/15) containing the single H286R mutation (Table III). Interestingly, we identified one clone that contained both S180N and H286R mutations.
Population sequencing analysis of genomic DNA isolated from C33A-WT cells indicated that the parental cell line is homozygous, encoding Ser180 and His286 in both alleles of the deaminase (data not shown). While clonal sequencing of ADAL cDNA from C33A-WT cells also showed absence of the S180N mutation, one cDNA clone was detected with the pre-existing H286R mutation that likely expanded during the resistance selection process in the presence of cPrPMEDAP. Mutations in genomic DNA at both the 180 and 286 residues of ADAL started emerging as early as seven passages after the initiation of selection in the presence of cPrPMEDAP (data not shown).
Metabolic activity profiling of ADAL variants. To evaluate the influence of the two identified ADAL mutations on the enzyme activity, we generated cell lines stably-expressing different ADAL variants in the C33A-Res cell line background and analyzed the intracellular metabolism of [14C]GS-9219. In the C33A-Res cell line stably transfected with the empty expression vector, only cPrPMEDAP-Ala and cPrPMEDAP, but no PMEG were detected after incubation with [14C]GS-9219 and dephosphorylation of cellular extracts (Figure 4A and B). In contrast, the C33A-Res cell line stably-expressing wild-type ADAL protein efficiently converted cPrPMEDAP to PMEG, indicating that the introduction of wild-type Adal repaired the deamination defect in these cells (Figure 4A and B). Without the cell extract de-phosphorylation, PMEGp and PMEGpp were detected in the C33A-Res cells expressing wild-type ADAL but not in C33A–Res cells (data not shown). On the other hand, the stable cell line C33A-Res expressing the S180N mutant ADAL showed only minimal restoration of the deaminase activity, with very limited amounts of PMEG detected in the de-phosphorylated extracts from cells incubated with [14C]GS-9219 (Figure 4A and B). In further contrast, no traces of PMEG were detected in extracts from the C33A-Res cell line stably expressing the ADAL H286R mutant. Immunoblot analysis confirmed the expression of the respective Adal variant in all stably transfected cell lines. The expression level of ADAL S180N in cells was comparable to that of wild-type ADAL cells whereas the expression level of the ADAL H286R-bearing cells was approximately 30% of that of wild-type ADAL-bearing cells when determined by a quantitative immunoblot analysis using anti-Flag antibody (Figure 4C). It should also be noted that the levels of cPrPMEDAP-Ala and cPrPMEDAP in the C33A-Res cell lines expressing the wild-type, S180N or H286R variants of ADAL were comparable indicating that the selection of stably transfected cells did not affect the upstream metabolism of GS-9219. Together, these data indicate that the S180N mutant ADAL retains some residual enzymatic activity while the H286R mutant is enzymatically inactive.
Due to the complete lack of ADAL activity, the C33A-Res cells represent an attractive tool for the profiling of any ADAL variant of interest. For example, sequence databases contain a cDNA of a short isoform of ADAL, encoding a truncated polypeptide with only 267-amino-acid residues (GenBank accession number NM_001012969). Upon stable transfection into C33A-Res cells, the truncated form was not able to restore the conversion of cPrPMEDAP to PMEG indicating that it is enzymatically inactive (Figure 4B). In contrast, the expression of a full-length canine orthologous Adal gene can fully reconstitute the metabolic conversion of cPrPMEDAP to PMEG in the C33A-Res cells (Figure 4B). This result indicates that the canine Adal is capable of effectively recognizing cPrPMEDAP as a substrate for deamination, an observation consistent with the potent in vivo antitumor efficacy of GS-9219 in dog models (6).
Enzymatic activity of ADAL wild-type and mutant variants. To further extend the observations from cells transfected with various forms of ADAL, we compared the enzymatic activity of the S180N and H286R mutant ADAL proteins with that of the wild-type ADAL protein. His-tagged ADAL proteins were expressed side-by-side with the unrelated SYK KD protein in Hi5 insect cells using a baculovirus expression system. The protein expression levels were determined by a quantitative anti-His immunoblot (Figure 5A) and used to calculate specific enzymatic activity of individual ADAL proteins. The conversion of O6-nPrPMEG, a highly effective ADAL substrate, to PMEG was measured to determine the ADAL activity in the cellular extracts. Cell extracts from the S180N mutant ADAL cells showed a 2.6-fold reduced specific activity compared to those from wild-type ADAL (458±63 nmol mg−1 min−1 versus 1,170±12 nmol mg−1 min−1; Figure 5B). In contrast, the ADAL activity in extracts from cells expressing H286R mutant was 50.7-fold lower (23±9 nmol mg−1 min−1), exhibiting a similar range of activity as the negative control SYK KD extract (46±5 nmol mg−1 min−1). These results are consistent with observations from the intracellular metabolism experiments and confirm the differential effect of the two mutations on the ADAL enzymatic activity.
Because the S180N mutant displayed measurable enzymatic activity in contrast to the enzymatically-inactive H286R mutant, we further investigated the catalytic properties of the S180N mutant. The recombinant wild-type and S180N ADAL proteins were expressed in insect cells and purified (Figure 5C). The identity of both proteins was verified by mass spectrometry (data not shown) and their specific enzymatic activity was determined with natural substrates N6-Methyl-dAMP (Me-dAMP) and N6-Methyl-AMP (Me-AMP), as well as cPrPMEDAP, and the known ADAL nucleotide analog substrate ABC-MP. Consistent with previous observations (21), the enzyme kinetic profiling showed that cPrPMEDAP is a less effective substrate for the wild-type Adal (specific activity of 133±25 nmol mg−1 min−1) relative to the other substrates Me-dAMP, Me-AMP and ABC-MP (specific activities 1003±221 nmol mg−1 min−1, 711±42 nmol mg−1 min 1 and 834±100 nmol mg−1 min−1, respectively). In comparison with the wild-type ADAL protein, the specific activities of the S180N mutant ADAL enzyme were 4.0-, 5.0-, and 2.6-fold lower for Me-dAMP, Me-AMP, and ABC-MP, respectively (Figure 5D). In contrast, the specific activity of the S180N mutant ADAL protein towards cPrPMEDAP was substantially more reduced (17.3-fold) relative to the wild-type ADAL enzyme. These findings indicate that the S180N mutation has a much more profound effect on the deamination of cPrPMEDAP compared to the other tested substrates.
Potential structural implications of ADAL mutations. Alignment of the Adal sequences from the protein regions surrounding S180 and H286 residues indicates that both amino acid positions are well conserved across Adal proteins from multiple evolutionary distinct species (Figure 6A), suggesting their importance for the structure and/or function of the Adal enzyme.
Since no X-ray structure of Adal from any species has yet been determined, we aligned the sequences of murine adenosine deaminase (Ada) of known structure and human ADAL (data not shown). The alignment showed that while both proteins show only 18% identity in their overall amino acid sequence, which precludes modeling the complete structure of human ADAL with confidence, the active site comprised of 22 residues is 50% identical. The comparison also showed that while the protein region surrounding the ADAL S180 residue (A183 residue in Ada) is highly conserved between the two proteins and close to the catalytic Zn2+ ion, the region surrounding ADAL H286 is not conserved and is located further away from the active site.
To understand how cPrPMEDAP interacts with ADAL and what impact the S180N mutation might have on the substrate-enzyme interaction, we built a homology model of the active site of human ADAL from the crystal structure of Ada containing 2-deaza-adenosine (23). For the purpose of clarity in this structural section, residue numbering corresponds to the original Ada crystal structure.
Figure 6B shows 2-deaza-adenosine in the active site of Ada. Half of the residues lining the active site are conserved (represented as purple sticks) and include three histidines and an aspartate coordinated to the catalytic Zn2+ ion. The other 11 residues forming the active site (represented as orange sticks) differ in the two enzymes. The latter were mutated to the corresponding residues in ADAL, and 2-deaza-adenosine was converted manually to cPrPMEDAP as described in the Materials and Methods section, to generate a model of the ADAL active site with the cPrPMEDAP substrate bound (Figure 6C).
In the Ada X-ray structure (Figure 6B), the heterocyclic base of 2-deaza-adenosine is held in place by three hydrogen bonds to the protein, while the sugar moiety interacts with aspartate-19 and catalytic histidine-17. Phenylalanine-61 shapes the bottom of the pocket and is mutated to cysteine in the modeled ADAL active site, creating a pocket to accommodate the cyclopropyl group of cPrPMEDAP. Aspartate-19 in Ada mutates to asparagine in ADAL, providing a perfect hydrogen bond complement for the monoprotonated phosphonate group of cPrPMEDAP. Two additional hydrogen bonds are formed between the phosphonate oxygen atoms and the side chain hydroxyl and backbone NH moieties of ADAL threonine-103. Finally, the loss of the hydrogen bond between glutamate-217 and the 6-amino group of the substrate heterocyclic base is offset by a new hydrogen bond formed between the glycine-184 carbonyl oxygen and the 2-amino group of cPrPMEDAP.
In the model of the wild-type ADAL active site, serine-183 represented by a purple mesh surface is located above glycine 184, and makes attractive van der Waals contacts with the heterocyclic base of cPrPMEDAP (represented by a green mesh surface in Figure 6D). The ADAL S183N mutation effectively decreases the size of the active site pocket, causing a steric collision between the cPrPMEDAP substrate (represented by a green mesh surface in figure 6E), and the larger side chain of asparagine-183 (represented by a blue mesh surface), potentially reducing the substrate binding to the active site, in agreement with our experimental results. Alternatively, for the ADAL S183N mutant protein to be able to catalyze the reaction, the position of some of the other active site residues would have to shift to accommodate the substrate, which may result in an alteration of the substrate recognition motif, impacting the enzyme activity.
In contrast to S183N, the ADAL H286R mutation (corresponds to Ada H306R) appears to be located outside of the active site pocket and therefore may only indirectly affect the substrate binding or perhaps impact proper folding or structural stability of the protein. Due to the limitations of the ADAL homology model, no specific conclusions could have been made regarding the structural impact of H286R on the activity of the ADAL enzyme.
Discussion
The N6-substituted acyclic nucleotide analog cPrPMEDAP is an intermediate metabolite in the intracellular activation of the prodrug GS-9219 that exhibits potent antitumor activity in dogs with various spontaneous hematological malignancies. We previously reported that the initial intracellular activation of GS-9219 involves enzymatic and chemical hydrolysis to an intermediate metabolite cPrPMEDAP that is subsequently converted to the parent cytotoxic guanine nucleotide PMEG via oxidative deamination (6). In the current study, we demonstrated that ADAL mutations emerged during resistance selections with cPrPMEDAP affecting the enzymatic activity of the ADAL protein. Susceptibility of the resistant cell line to cPrPMEDAP and GS-9219 was restored by the introduction of fully enzymatically-active wild-type ADAL. Enzymatic profiling of the ADAL mutant proteins showed that the ADAL mutants converted cPrPMEDAP less efficiently than their natural substrates N6-Me-AMP and N6-Me-dAMP. Structural analysis of the ADAL mutations based on homology modeling further revealed that the substrate binding pocket might be altered and thus affected the enzymatic activity of the ADAL protein. These results strongly suggest that ADAL might be an important factor in the acquisition of resistance to cPrPMEDAP and its prodrug GS-9219.
Acquired resistance of canine lymphomas to first and second line chemotherapy usually decreases the chances of achieving remission and leads to a mortality of relapsed dogs. Second line chemotherapy success rates only range within 20% to 50% for a median duration of two to three months for complete remission (14-17). Several mechanisms are described to cause resistance of tumor cells to chemotherapy. In dogs, overexpression of the permeability glycoprotein 1, also known as multidrug resistance protein 1 (MDR1) or ATP-binding cassette sub-family B member 1 (ABCB1), is commonly seen and confers resistance to chemotherapeutics including doxorubicine, prednisone and vincristine (24-27).
Resistance to the pharmacological effects of N6-substituted purine nucleotide analogs can occur at different levels of their activation cascade. Membrane efflux pumps such as the ATP-binding cassette transporters can induce resistance by decreasing the intracellular levels of nucleoside analogs and their metabolites (28). In particular, ABCC4 and ABCC5 (formerly known as MRP4 and MRP5, respectively) have been implicated in the resistance to N6-substituted nucleotide analogues (29, 30). In addition, Mertlíková-Kaiserová et al. identified mutations in guanylate kinase as a mechanism for resistance to PMEG and cross-resistance to cPrPMEDAP (9). The cPrPMEDAP resistant C33A-Res cell line that we generated showed only a minor, 4.6-fold reduced sensitivity to PMEG relative to the parent C33A cell line indicating that the resistance of the selected C33A cells to cPrPMEDAP is likely driven by a defect in the deamination. The minor decrease in the sensitivity to PMEG could be due to the up-regulation of efflux proteins such as the ATP-binding cassette transporters ABCC4 and ABCC5 or other resistance pathways, However, resistance due to defects upstream of PMEG activation, in particular mutations in ADAL, has not been described yet.
Previous biochemical in vitro studies with native and recombinant enzymes have demonstrated that N6-Me(d)AMP aminohydrolase (also named adenosine deaminase-like enzyme; ADAL) can catalyze the deamination of cPrPMEDAP to PMEG (21). We showed that the expression of the wild-type ADAL protein in the cPrPMEDAP-resistant cell line restored the sensitivity to cPrPMEDAP and GS-9219 together with their ability to convert cPrPMEDAP to PMEG and its downstream phosphorylated metabolites. Nucleoside and nucleotide deaminases are present in virtually all eukaryotic cells and maintain proper balance in the intracellular metabolism by controlling various pathways for the salvage of nucleosides and nucleotides. In particular, ADA plays an important role in the purine metabolism by deaminating adenosine and 2’-deoxyadenosine. ADA deficiency or mutations in ADA that affect its activity are associated with severe combined immunodeficiency disease (31). ADA, together with adenosine monophosphate deaminase (AMPD) and more recently identified Adal belong to an adenyl deaminase family of enzymes (32) that are involved in the metabolism of multiple pharmacologically important nucleoside and nucleotide analogs. For example, the lack of stability of many adenosine nucleoside analogs due to ADA-mediated deamination prompted the design of ADA-resistant antineoplastic agents such as cladribine or fludarabine (33). On the other hand, ADA is important for the effective conversion of antitumor nucleoside nelarabine to its parent active metabolite arabinosyl-guanine (33, 34). Similarly, in addition to catalyzing the oxidative deamination of cPrPMEDAP, ADAL mediates the catabolic activation of abacavir, a nucleoside HIV reverse transcriptase inhibitor, via the deamination of abacavir-monophosphate to carbovir-monophosphate (35).
Sequencing analysis of cPrPMEDAP-resistant cells identified two distinct ADAL mutations (S180N and H286R). We demonstrated that the S180N mutation retained some residual deaminase activity towards cPrPMEDAP while the H286R mutant was devoid of any enzymatic activity. The serine 180 and the histidine 286 residues in the ADAL protein as well as the surrounding residues are highly conserved between human and dog. Therefore similar resistance mechanisms in human and dog might occur.
Analysis of a structural model of ADAL active site generated based on the known X-ray structure of murine Ada indicates that the Ser 180 residue is located within the active site of the ADAL enzyme. In addition, the computational analysis suggested that the S180N mutation reduces the volume of and/or changes the charge distribution within the ADAL active site cavity. This fits with the observation that the deamination of natural substrates N6-Me-AMP and N6-Me-dAMP as well as that of abacavir-monophosphate was only minimally affected by the S180N mutation compared to cPrPEMDAP. Multiple examples of mutations affecting the structure and/or function of enzymes from the adenyl deaminase family have been described in the past, primarily in the context of ADA-specific severe combined immunodeficiency disease (36-40). These include mutations R101W/Q, R211H, P297Q and L304R that are located in close proximity of the ADA active site pocket and have either a direct structural effect on the size and shape of the active site pocket or affect the coordination of the zinc atom (23, 41), thereby causing the loss of ADA activity. The S180N mutation in the ADAL protein might have a similar impact on the structural integrity of the active site pocket, but could also directly affect the binding of the substrate, with more significant effect on the binding of acyclic nucleotide analogs such as cPrPMEDAP compared to the substrates with cyclic sugar moieties.
In contrast to S180N ADAL mutation within the active site, understanding the effects of the H286R ADAL mutation outside of the active site remains elusive, mainly due to inability of the computational tools to analyze its impact based on currently available information. Wilson et al. suggested that some of the point-mutations in Ada do not directly affect the substrate binding, but instead appear to modify the protein tertiary structure around the active site pocket that is composed of an eight-stranded parallel α/β barrel (41). The study concluded that any point mutation disrupting the alignment of the β strands would render the Ada enzyme catalytically-inactive. A similar mechanism might explain the profound effect of the H286R mutation on the activity of the ADAL enzyme. At this point, the determination of the ADAL protein crystal structure would be necessary to fully understand the interactions of ADAL enzyme with its substrates as well as the structural and functional implications of the two identified ADAL mutations.
In conclusion, the present study shows the importance of the ADAL protein in the intracellular metabolism of N6-substituted nucleotide analogs, particularly cPrPMEDAP and its pharmacologically relevant prodrug GS-9219. Specific mutations in human ADAL emerging from drug resistance selection affect the enzyme deamination activity and confer resistance to both cPrPMEDAP and GS-9219 due to deficient intracellular conversion to the parent PMEG. Our study is the first to identify ADAL mutations that affect its enzymatic function and further investigations are needed to fully understand the potential implications of these findings for the veterinary applications of GS-9219.
Acknowledgements
The Authors would like to thank Gabriel Birkus for helpful scientific and technical discussions and Anita Camps for excellent technical assistance. This work was supported by grants from the K.U.Leuven, Program Financing (PF/10/018) and the Geconcerteerde Onderzoeksactie (GOA).
Footnotes
-
Conflicts of Interest
The Authors Frey, Cannizzaro, Han, Fung, Hung, Liu, Geleziunas and Cihlar are employees and stock holders of Gilead Sciences, Inc. All the other co-authors declare no conflicts of interest.
- Received February 28, 2013.
- Revision received April 11, 2013.
- Accepted April 12, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved