751
Matrix metalloproteinase–activated doxorubicin prodrugs inhibit HT1080 xenograft growth better than doxorubicin with less toxicity
Charles F. Albright, Nilsa Graciani, Wei Han, MMP-selective conjugates were given to mice with Eddy Yue, Ross Stein, Zhihong Lai, HT1080 xenografts and the distribution of doxorubicin Melody Diamond, Randine Dowling, was determined. These studies showed that MMP- Lisa Grimminger, Shu-Yun Zhang, selective conjugates were preferentially metabolized in Davette Behrens, Amy Musselman, HT1080 xenografts, relative to heart and plasma, leading Robert Bruckner, Mingzhu Zhang, Xiang Jiang, to 10-fold increases in the tumor/heart ratio of doxorubi- Daniel Hu, Anne Higley, Susan DiMeo, cin. The doxorubicin deposited by a MMP-selective Maria Rafalski, Sandya Mandlekar, Bruce Car, prodrug, compound 6, was more effective than doxorubi- Swamy Yeleswaram, Andrew Stern, cin at reducing HT1080 xenograft growth. In particular, Robert A. Copeland, Andrew Combs, compound 6 cured 8 of 10 mice with HT1080 xenografts Steve P. Seitz, George L. Trainor, Rebecca Taub, at doses below the maximum tolerated dose, whereas Pearl Huang, and Allen Oliff doxorubicin cured 2 of 20 mice at its maximum tolerated dose. Compound 6 was less toxic than doxorubicin at this Bristol-Myers Squibb Pharmaceutical Research Institute, Wallingford, Connecticut efficacious dose because mice treated with compound 6 had no detectable changes in body weight or reticulo- cytes, a marker for marrow toxicity. Hence, MMP- activated doxorubicin prodrugs have a much higher Abstract therapeutic index than doxorubicin using HT1080 xeno- Matrix metalloproteinase (MMP)–activated prodrugs were grafts as a preclinical model. [Mol Cancer Ther 2005;4(5): formed by coupling MMP-cleavable peptides to doxorubicin. 751–60] The resulting conjugates were excellent in vitro sub- strates for MMP-2, -9, and -14. HT1080, a fibrosarcoma cell line, was used as a model system to test these Introduction prodrugs because these cells, like tumor stromal fibro- Doxorubicin is an anthracycline natural product that is blasts, expressed several MMPs. In cultured HT1080 frequently used to treat breast cancer, liver cancer, soft- cells, simple MMP-cleavable peptides were primarily tissue sarcomas, and non-Hodgkin’s lymphoma (reviewed metabolized by neprilysin, a membrane-bound metallopro- in ref. 1). Doxorubicin kills tumor cells by causing teinase. MMP-selective metabolism in cultured HT1080 topoisomerase II–stimulated DNA strand breaks and cells was obtained by designing conjugates that were potentially other mechanisms (reviewed in ref. 2). Like good MMP substrates but poor neprilysin substrates. To other cytotoxic drugs, doxorubicin doses are limited by determine how conjugates were metabolized in animals, unwanted toxicity to nontumor tissues. Myelosuppression is the dose-limiting toxicity for doxorubicin, although stomatitis, mucositis, and alopecia are also frequently Received 1/7/05; revised 2/9/05; accepted 3/2/05. observed. In addition to these typical chemotherapeutic The costs of publication of this article were defrayed in part by the toxicities, doxorubicin causes cardiomyopathy that payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to depends on the cumulative dose of doxorubicin. indicate this fact. Matrix metalloproteinases (MMP) are a family of >20 Note: N. Graciani is currently at Wyeth Pharmaceutical, 500 Arcola Road, enzymes (reviewed in ref. 3). MMPs are synthesized as Collegeville, PA 19426. E. Yue and S. Yeleswaram are currently at inactive proenzymes that become activated by proteolysis. Incyte Pharmaceuticals, Route 141 and Henry Clay Drive, Wilmington, DE 19880-0400. R. Stein is currently at the Department of Neurology, For instance, the secreted enzyme pro-MMP-2 is proteo- Harvard Medical School, Cambridge, MA 02139. Z. Lai, M. Diamond, lyzed and activated by the membrane-anchored MMP-14 S-Y. Zhang, R.A. Copeland, P. Huang, and A. Oliff are currently at (4). Once activated, MMPs can cleave a variety of GlaxoSmithKline, Collegeville, PA 19426. L. Grimminger is currently at Johnson & Johnson, Malvern, PA 19355. A. Musselman is currently at extracellular matrix proteins, such as collagen, laminin, Merck Research Laboratories, Merck & Co., West Point, PA 19486. fibronectin, and elastin, which are potentially important M. Zhang is currently at Neurocrine Biosciences, Inc., 1055 Science Center Drive, San Diego, CA 92130. A. Stern is currently at Ensemble for tumor angiogenesis, invasion, and metastasis. Addi- Discovery Corp., 99 Erie Street, Cambridge, MA 02139. R. Taub is tionally, MMPs cleave a variety of other proteins, such currently at Roche Pharmaceuticals, Nutley, NJ 07110. as growth factor receptors and cell adhesion molecules, Requests for reprints: Charles F. Albright, Bristol-Myers Squibb which may be important for tumor growth and survival Pharmaceutical Research Institute, 5 Research Parkway, Wallingford, CT 06492. Phone: 203-677-5937; Fax: 203-677-7569. (reviewed in refs. 5, 6). MMPs can be inactivated by tissue E-mail: [email protected] inhibitors of MMPs, a family of four proteins, which bind Copyright C 2005 American Association for Cancer Research. tightly to MMPs.
Mol Cancer Ther 2005;4(5). May 2005
Downloaded from mct.aacrjournals.org on September 29, 2021. © 2005 American Association for Cancer Research. 752 MMP-Activated Doxorubicin Prodrugs
Increasing the therapeutic index of doxorubicin could because these cells grow as xenografts, are sensitive to benefit cancer treatment by allowing greater concentrations doxorubicin, and express several different MMPs (14, 15). of doxorubicin in tumors and thereby potentially greater Furthermore, HT1080 cells are derived from a fibrosarcoma tumor growth inhibition. Alternatively, increasing the and were, therefore, hoped to mimic stromal fibroblasts, a therapeutic index of doxorubicin could allow reduced major constituent of many human tumors. Using this toxicities and thereby permit the use of other cytotoxics that strategy, prodrugs were identified that were excellent would be prevented because of overlapping toxicities. One in vitro substrates for MMP-2, -9, and -14. Optimization of way to improve the therapeutic index of doxorubicin is to conjugate structure led to prodrugs that were selectively create prodrugs that are selectively activated in the tumor metabolized by MMPs in HT1080 cultures and preferen- environment. MMPs are attractive enzymes to activate tially deposited doxorubicin in HT1080 xenografts relative prodrugs in the tumor environment because MMPs are to heart tissue. The preferential doxorubicin deposition led intimately connected with tumorigenesis. In particular, to an improved therapeutic index, relative to doxorubicin, preclinical studies implicate MMPs in tumor progression as shown by improved HT1080 xenograft growth inhibition (reviewed in ref. 6). For instance, overexpression of MMP- with reduced toxicity. 14 in mouse mammary tissue causes mammary tumors (7), whereas loss of MMP-2 or -9 reduced the lung colonization by B16-LBL6 melanoma cells and Lewis lung carcinoma Materials and Methods cells (8, 9). Additionally, MMP expression in human Compound Synthesis tumors frequently correlates with disease progression All peptides were synthesized following the same (reviewed in ref. 6). For instance, one study found MMP- protocol. An example synthesis of compound 1 is provided. 2 and -14 expression in 90% of breast carcinomas and a The peptide acid was synthesized on the solid phase strong correlation between tumor cell membrane staining starting with commercially available Fmoc-Leu-Wang resin and lymph node metastases (10). Although there are strong (0.40 g, 0.25 mmol, Advance Chemtech, Louisville, KY). connections between MMPs and tumor progression, pub- The synthesis was done on an ABI 433A peptide lished studies do not determine the absolute or relative synthesizer (Applied Biosystems, Foster City, CA) using MMP activity in the tumor environment. Such information standard Fmoc protocols. The completed peptide on resin is clearly critical to the design of MMP-activated prodrugs was N-acetylated with acetic anhydride. The desired because the MMP enzyme activity must be sufficient to peptide was cleaved from the resin with 90% trifluoroacetic activate efficacious levels of prodrug. acid in water for 2 hours. After solvent removal, the Preclinical testing of MMP-activated prodrugs is com- peptide was dissolved in H2O:CH3CN and freeze dried. plicated by differences between mouse and human tumors. Product was confirmed by mass spectrometry (ES m/e Many human tumors, including breast cancers, are com- calculated for C21H36N4O6 [M-H] , 439.54; found, 439.3). posed of roughly equal numbers of tumor cells and stromal Analytic high-performance liquid chromatography on a cells, including stromal fibroblasts, inflammatory cells, and Metachem Monochrome C18 reverse-phase column (50 endothelial cells (reviewed in ref. 11). The interplay of these 4.6 mm, MetaChem Technologies, Torrance, CA) showed cell types is thought to be crucial for MMP activation in that crude peptide to be 85% pure. To this intermediate (19.9 mg, different cell types express and activate specific MMPs 40 Amol) dissolved in dimethylformamide (0.2 mL) in a (reviewed in ref. 6). For instance, stromal fibroblasts small amber vial was added benzotriazol-1-yl-oxytripyrro- express MMP-2, yet tumor cells are the source of MMP-14 lidinophosphonium hexafluorophosphate (20.8 mg, 40 Amol). that convert pro-MMP-2 to active MMP-2. Unfortunately, Doxorubicin hydrochloride (18.6 mg, 32 Amol, Aldrich, mouse tumor models do not accurately reflect the cell types Milwaukee, WI) was added as a suspension in dimethyl- observed in human tumors because both xenografts and formamide (0.1 mL) followed by diisopropylethylamine transgenic tumor models are composed primarily of (0.0139 mL, 80 Amol). The reaction was stirred for 2 hours. rapidly growing tumor cells with relatively few stromal Solvent was removed under vacuo. The sample was cells. dissolved in H2O:CH3CN and purified using a Dynamax This study discovered MMP-activated doxorubicin pro- C18 reverse-phase column (41.4 250 mm, Varian, Palo drugs that had a greater therapeutic index than doxorubi- Alto, CA) with a linear gradient from 30% to 50% cin using HT1080 cells as a model system. In the absence of acetonitrile, 0.05% ammonium acetate over 20 minutes information about enzyme activity levels, prodrugs were with a flow rate of 45 mL/min. Fractions were pooled and identified that were efficiently cleaved by MMP-2, -9, and freeze dried to afford the purified peptide-doxorubicin -14 because these enzymes were frequently overexpressed conjugate (ES m/e calculated for C48H63N5O16 [M-H] , in human tumors, affected tumor progression in preclinical 965.06; found, 964.6). Caution: Doxorubicin is a highly toxic models, and were very active in vitro. Because these MMPs compound. Appropriate measures should be taken to avoid may have insufficient activity for prodrug activation in vivo, exposure to the compound. prodrug design was further constrained so that prodrugs Enzyme Assays were likely be activated by other MMPs, with the exception Doxorubicin prodrugs were incubated with MMP-2, -9, of MMP-11 (12), based on the known MMP substrate or -14 or neprilysin to compare the competency of these specificities (13). HT1080 cells were used as a model system conjugates as substrates for these enzymes. Recombinant
Mol Cancer Ther 2005;4(5). May 2005
Downloaded from mct.aacrjournals.org on September 29, 2021. © 2005 American Association for Cancer Research. Molecular Cancer Therapeutics 753
MMP-2, -9, and -14 were purified from HT1080 fibrosarco- intrinsic fluorescence as above. To determine the amount ma cells, insect cells, and Escherichia coli, respectively, as of metabolism that was due to MMPs, cultures were described previously (12, 16, 17). MMP-2 and -9 were incubated with 0.15 mmol/L marimastat, a broad-spec- activated with 4-aminophenylmercuric acetate for 2 hours trum MMP inhibitor (21). This concentration of marima- at 37jC. Unreacted 4-aminophenylmercuric acetate was stat was sufficient to inhibit MMPs based on in vitro removed using Bio-Spin 6 columns (Bio-Rad, Hercules, assays and the conversion of pro-MMP-2 to MMP-2 in CA). Neprilysin was purified from rat kidney as described HT1080 cultures (data not shown). To determine the (18). Conjugates (1 Amol/L) were incubated with either 10 fraction of metabolism due to neprilysin, cultures were nmol/L MMP-2, 2 nmol/L MMP-9, or 5 nmol/L MMP-14 incubated 100 nmol/L phosphoramidon. in 50 mmol/L HEPES (pH 7.5), 10 mmol/L CaCl2, 0.1% ProdrugMetabolisminMice Brij-35 at 37jC. At various times up to 60 minutes, an HT1080 tumors were transplanted into naive mice using aliquot of the reaction was removed and injected into a trocar needles. Dry conjugate powders were dissolved Novapak C18 reverse-phase high-performance liquid chro- in N,N-dimethylacetamide and diluted with 5% dextrose matography column (Waters, Milford, MA). The reaction to yield a final dosing solution of 10% N,N-dimethylace- mixture was eluted with a 27% to 63% acetonitrile gra- tamide, 4.5% dextrose. Mice were then given the desired dient containing 0.1% trifluoroacetic acid. Doxorubicin- dose using 0.1 mL dosing solution. Drugs were given i.v. containing compounds were detected and quantified using into the tail vein. At the indicated times, three mice per f the intrinsic fluorescence of doxorubicin (excitation wave- time point were anesthetized with CO2. Blood ( 1 mL) length, 480 nm; emission wavelength, 580 nm). Alternately, was collected by cardiac puncture in a syringe containing conjugates (1 Amol/L) were incubated with 10 nmol/L 0.1 mL sodium citrate to prevent coagulation and cells were neprilysin in 50 mmol/L Tris-HCl (pH 7.5), 10 mmol/L removed by centrifugation. The resulting plasma was CaCl2, 0.1% Brij-35 and processed as above. The substrate transferred to another tube and frozen in liquid nitrogen. peak area was used to determine the kcat/Km for each Tumor and heart tissue were removed, frozen in liquid j conjugate under first-order conditions (substrate << Km): nitrogen, and stored at 80 C. For analysis, tissues were thawed on ice, weighed, and minced with scissors, and kcat=Km ¼ lnðS0=StÞ=ðt ðEÞÞ cold, citrated mouse plasma was added (Cocalico Biolog- ical, Reamstown, PA). Plasma was added at 5 mL/g for where S0 is the initial substrate concentration, St is the tumor tissue and 10 mL/g for heart tissue. Slurries were substrate concentration at various times, t is time, and E is homogenized for 1 minute and 0.5 mL was transferred to enzyme concentration (19). an Eppendorf tube. Silver nitrate (0.1 mL, 33%) was ProdrugMetabolisminCulturedHT1080Cells added to the homogenate to improve the recovery of To measure the rate of prodrug metabolism in cell doxorubicin and the mixture was vortexed briefly. Aceto- culture, actively growing HT1080 fibrosarcoma cells nitrile (0.05 mL) was then added, vortexed briefly, and (American Type Culture Collection, Manassas, VA) were mixed for 15 minutes. This mixture was centrifuged for plated at 60% confluence. Two hours after plating, the cell 5 minutes. The supernatant was transferred to another medium (DMEM, 10% fetal bovine serum) was removed tube, dried under a nitrogen stream, and stored at 80jC. and replaced with DMEM containing 0.1% bovine serum Doxorubicin-containing compounds were separated by albumin and 40 nmol/L phorbol 12-myristate 13-acetate. reverse-phase high-performance liquid chromatography Phorbol 12-myristate 13-acetate was added to increase the as above and quantified using a doxorubicin standard expression of MMP-9 and conversion of pro-MMP-2 to curve. Control experiments showed that this method active MMP-2 (20). After 16 hours, doxorubicin prodrug extracted 80% doxorubicin-containing compounds and that (500 nmol/L) was added to the medium. Aliquots of metabolism of doxorubicin-containing compounds during medium (0.1 mL) were removed at the indicated times, analysis was negligible. 0.3 mL acetonitrile was added to each aliquot, and the Inhibition ofHT1080 XenograftGrowth sample was mixed and then centrifuged for 5 minutes. HT1080 tumors were transplanted into naive mice. After Solvent was evaporated from clarified supernatant (0.3 mL) 5 to 7 days, tumors had reached a median size of 75 to 100 using a nitrogen stream. Samples were then vortexed with mm3. Mice were then grouped into cohorts of 10 mice such 0.06 mL acetonitrile. Buffer A [0.06 mL; 14 mmol/L that each cohort had the same median tumor size and a sodium phosphate, 0.5 mmol/L triethylamine (pH 4.2)] similar tumor size distribution. Treatment was then was then added, vortexed, and centrifuged. Supernatant initiated as described in Results and tumors were measured (0.06 mL) was mixed with 0.06 mL buffer A and 0.1 mL every 3 days with calipers. was injected onto a Novapak C18 column at a flow rate of Reticulocyte Measurements 1 mL/min and eluted with a linear gradient from Three mice per condition were anesthetized with CO2 75% buffer A, 25% acetonitrile to 50% buffer A, 50% 3 days after the last treatment because control experiments acetonitrile over 12 minutes. Control experiments showed showed that this time corresponded to the nadir of the that doxorubicin and doxorubicin conjugates were recov- reticulocyte decrease. Blood (0.5 mL) was collected by ered with f80% yield using this method. Doxorubicin- cardiac puncture in a syringe containing EDTA to prevent containing compounds were detected based on their coagulation. RBC and reticulocyte concentrations were
Mol Cancer Ther 2005;4(5). May 2005
Downloaded from mct.aacrjournals.org on September 29, 2021. © 2005 American Association for Cancer Research. 754 MMP-Activated Doxorubicin Prodrugs
determined the same day using an Advia 120 Hematology of compounds to humans. Although the necessary solubil- Analyzer. ity would depend on clinical usage, a target solubility of Lethal Dose Determination 1 mg/mL was chosen to guide compound design. Increasing doses of drugs were given i.v. to five mice per Prodrug Cleavage by MMP-2, -9, and -14 In vitro dose. Animals were observed for 30 days for signs of severe To determine the effect of prime-side length on MMP toxicity and euthanized if required according to Dupont cleavage rate, peptides of various lengths were attached to Pharmaceuticals Animal Care and Use Committee guide- doxorubicin and the catalytic efficiencies for cleavage by lines. MMP-2, -9, and -14 were determined (Table 1). For this analysis, simple collagen-like peptide sequences (PLGfL) Results were used because most MMPs cleave peptides containing Design ofMMP-Activated Doxorubicin Prodrugs this sequence (reviewed in ref. 13). A conjugate with the one prime-side residue, compound 1, was not detectably Doxorubicin prodrugs were formed by linking the cleaved by MMP-2 or -9. A conjugate with two prime-side COOH terminus of a peptide to the amino group of residues, compound 2, was efficiently cleaved by MMP-9, doxorubicin (Fig. 1). The resulting prodrug is not cytotoxic modestly cleaved by MMP-2, and poorly cleaved by MMP- because the peptide prevents the prodrug from entering 14. Conjugates with three prime-side residues, compound cells (22). MMP cleavage occurs in the middle of the 3, or four prime-side residues, compound 4, were cleaved illustrated hexapeptide generating a tripeptide doxorubicin efficiently by MMP-2, -9, and -14. The absolute k K conjugate (Fig. 1). The tripeptide doxorubicin conjugate cat/ m must be further cleaved by extracellular proteases to form values for efficiently cleaved conjugates, such as com- Leu-Dox before doxorubicin-containing compounds can pounds 3 and 4, were close to values for the most efficient k K efficiently penetrate cells. Leu-Dox may be converted to MMP substrates. In particular, cat/ m values for the best reported fluorogenic peptide substrates for MMP-2, -9, doxorubicin either inside or outside cells. Leucine was 1 1 chosen as the COOH-terminal residue because Leu-Dox and -14 are 629, 87, and 1,590 mmol/L s , respectively was more efficiently converted to doxorubicin than other (24–26). Metabolism ofConjugates in Cultured HT1080 Cells amino acid doxorubicin conjugates (23) and leucine allowed efficient MMP-2, -9, and -14 cleavage (data not The metabolism rate of conjugates with two or more shown). The peptide was capped to prevent aminopepti- prime-side residues in cultured HT1080 cells was deter- dase degradation and increase solubility. Sufficient aque- mined because in vitro enzymatic digestions showed that ous solubility was required to allow the i.v. administration these conjugates were substrates for MMP-2, -9, or -14. For this analysis, conjugates were incubated with cultured HT1080 cells, aliquots of culture supernatant were removed after various times, and the composition of doxorubicin-containing compounds was determined. Using compound 2 as an example, the parental conjugates disappeared linearly with time for several hours and the major metabolite was L-Dox (Fig. 2A). Other metabolites, suchasLL-Dox,whichwouldbeformedbyMMP cleavage of compound 2, were only detected in trace amounts. To determine why these expected metabolites were not detected at higher levels, compounds were cleaved by MMP-2 in vitro and then added to cultured HT1080 cells. Analysis of these culture supernatants showed that post-MMP cleavage metabolites were rapidly converted to L-Dox (data not shown). The kinetics of conversion of these post-MMP cleavage metabolites to L- Dox were consistent with the trace amounts of these metabolites detected following incubation of the parental compounds. To allow comparisons between compounds, the fraction processed was defined as the fraction of compounds present in the culture supernatant that could result from MMP cleavage. The fraction processed per hour was calculated by fitting a straight line to the fraction processed values measured during the first 6 hours after Figure 1. Structure of doxorubicin (A) and compound 6, a representative incubation of the compound with cultured HT1080 cells. prodrug, with expected enzymatic cleavage products (B). The MMP Although this method is considered the most rigorous, f cleavage site is indicated by . Residues on the NH2-terminal side of the cleavage site are designated P1, P2, etc., whereas residues on the COOH- very similar values would be obtained by simply terminal side of the cleavage site are designated P1V,P2V, etc. See Results considering the formation of L-Dox. Analysis of com- for further description. pounds 2 to 4 showed a dramatic effect of prime-side
Mol Cancer Ther 2005;4(5). May 2005
Downloaded from mct.aacrjournals.org on September 29, 2021. © 2005 American Association for Cancer Research. Molecular Cancer Therapeutics 755
Table 1. In vitro properties of selected prodrugs