Pharmacological Characterization of SDX-7320/Evexomostat: A Novel Methionine Aminopeptidase Type 2 Inhibitor with Anti-tumor and Anti-metastatic Activity

The cytoplasmic metalloproteases methionine aminopeptidase types 1 and 2 (METAP1 and METAP2) are fundamental in cellular function across prokaryotes and eukaryotes, primarily responsible for N-terminal methionine cleavage of proteins during de novo biosynthesis (1). A limited number of METAP2-selective substrates have been reported to date: thioredoxin-1, cyclophilin A, eukaryotic elongation factor 2, RAB37, GAPDH, 14–3-3γ and SH3 binding glutamic acid rich-like protein (2–4). Inhibiting the aminopeptidase activity of METAP2 causes these substrates to retain their initiator methionine, thereby affecting post-translational modifications, protein folding, cellular localization, and stabilization (2, 3, 5, 6).

Fumagillin is a natural product isolated from the fungus Aspergillus fumigatus Fresenius that has been shown to irreversibly bind selectively to METAP2 (but not METAP1) and inactivate its aminopeptidase activity (1). Fumagillin-bound METAP2 is stabilized and accumulates intracellularly, exerting additional pharmacodynamic effects including inhibition of MAPK activity (7, 8). Therefore, inhibition of METAP2 elicits multiple downstream effects leading to various physiological responses (9–11) including inhibition of tumor growth (12), metastasis (13), and angiogenesis, due to endothelial cell-cycle arrest in the late G₁ phase (14, 15). METAP2 has been reported to be overexpressed in many malignancies including breast, prostate, colorectal, esophageal, lung, lymphomas mesothelioma, and cholangiocarcinoma (16–20).

Multiple fumagillin-based METAP2 inhibitors have been evaluated clinically as anti-cancer compounds. TNP-470 is a fumagillin derivative with 50-fold higher potency in vitro than the natural product, for which synergistic activity was observed in preclinical studies when combined with cytotoxic chemotherapeutics (21–23). Despite promising clinical efficacy, TNP-470 suffered from dose-limiting neurological side effects, presumably due to penetration of the small molecule across the blood–brain barrier (24). Other challenges to its clinical development included poor aqueous solubility and a pharmacokinetic (PK) profile which necessitated frequent intravenous dosing due to its short half-life (9, 25).

Polymer–drug conjugation is a well-documented approach that presents potential benefits with respect to the limitations associated with drugs like TNP-470 (26). As a hydrophilic polyacetal linked to a fumagillol derivative, the polymer–drug conjugate (PDC) XMT-1107 reached phase I clinical studies in patients with late-stage cancer (27). However, high doses were required to elicit pharmacodynamic effects and development was halted (28). A fumagillol analogue derived from TNP-470, conjugated to a hydroxypropylmethacrylamide (HPMA) copolymer backbone displayed favorable preclinical characteristics in both neurobehavioral tests as well as tumor models; however, it did not undergo clinical evaluation (29). Importantly, poly(HPMA) with other pendant small molecules has not been found to induce signs of toxicity nor immunogenicity (e.g., PK1, FCE28068; ref. 30). Alternate non–fumagillol-derived PDCs using the HPMA copolymer backbone have been investigated clinically at average daily doses >1.5 g/m² (e.g., AP5280; ref. 31). With respect to these PDCs, total drug doses in excess of 1.0 g/m²/day were typically required for clinical anti-tumor efficacy.

On the basis of the encouraging anti-tumor efficacy profile of TNP-470 and other fumagillin-based inhibitors of METAP2, a novel fumagillol-derived small molecule was sought which when conjugated to a high molecular weight polymeric system would yield a prodrug suitable for clinical development. This report describes the structure, synthesis, and pharmacological characterization of SDX-7320 (INN: evexomostat), a PDC of the novel, selective METAP2 inhibitor SDX-7539, which displays anti-tumor and anti-metastatic properties with an improved safety profile versus previous small molecule fumagillin analogues.

The chemical structure of SDX-7320 is shown in Fig. 1, along with the pharmacologically active small molecule METAP2 inhibitor SDX-7539. The preparation and characterization of SDX-7320 and SDX-7539 are described in the Supplementary Data and in Supplementary Fig. S1 and Supplementary Table S1, along with the related compounds SDX-9246, SDX-9280, SDX-9178, TNP-470, and CKD-732. The benzenesulfonic acid salt form of SDX-7539 (SDX-9402) was used for in vitro experiments.

Cathepsin isoforms were purchased from the suppliers shown in the Supplementary Data. Cathepsin isoforms and human neutrophil elastase (HNE) were tested for their ability to catalyze the release of SDX-7539 from SDX-7320 in vitro, using manufacturer-recommended assay conditions, as described in the Supplementary Data.

Enzyme activity was measured at room temperature using a coupled assay system (R&D Systems) containing the fluorogenic substrates Met-Gly-Pro-AMC (for METAP1) or Met-AMC (for METAP2). Reactions for METAP2 contained 10 μg/mL hMETAP2, 5 μg/mL DPPIV, and 1 μmol/L Met-AMC, while METAP1 reactions contained 4 μg/mL METAP1, 1 μg/mL DPPIV, and 1 μmol/L Met-Gly-Pro-AMC in 50 mmol/L HEPES, 0.1 mmol/L CoCl₂, 100 mmol/L NaCl, pH 7.5. Test compounds were diluted in DMSO, then diluted in 96-well plates containing enzymes, followed by addition of the fluorogenic substrate. Fluorescence was measured over time using excitation and emission wavelengths of 350 nm and 440 nm.

Details of the expression of METAP2, co-crystallization with SDX-7539, data reduction and refinement are provided in the Supplementary Data. The final refined structure was deposited in the PDB under code 8OXG.

Human umbilical vein endothelial cells (HUVEC; from the ATCC) were cultured in EBM-2 medium with EGM-2 (Lonza #CC-3156 and CC4176, respectively) containing 2% FBS and 2X gentamicin. Cells were plated into 96-well plates (5,000 cells/well) and incubated overnight. Test compounds were initially diluted in DMSO, then diluted into culture medium listed above containing 2% FBS, 2X gentamicin (final DMSO = 0.1%). After 72 hours, 37°C Promega Substrate CellTiter 96 Aqueous One Solution Reagent was added to each well. After 3 hours at 37°C, the OD at 490 nm was measured, then corrected for baseline by subtracting the average A₄₉₀ of the wells with highest concentration of compound added. Inhibitory potency was calculated using GraphPad's non-linear regression analysis [“log(inhibitor) vs. response − Variable slope (four parameters)”].

All animal procedures were carried out under the institutional guidelines of each organization and were approved by their respective Institutional Animal Care and Use Committees.

Evaluation of the neurobehavioral safety of SDX-7320 in male experimentally naïve CD [Crl:CD(SD)] rats (approximately 8 weeks of age at the start of the study) was conducted under good laboratory practice (GLP) conditions following single subcutaneous doses (ten/group) of 3, 10 or 30 mg/kg. Assessments of neurobehavioral effects and general toxicity were based on mortality and functional observation battery (FOB) evaluations (32), clinical observations, and body weights. FOB evaluations were conducted pre-dose (day −1) and at 12 and 36 hours post-dose. A detailed description of the FOB and the results can be found in the Supplementary Data.

The A549 human NSCLC cell line was obtained from ATCC. Female athymic nude mice (Hsd:Athymic Nude-Fox-n1^nu from Harlan), were inoculated subcutaneously in the right hind flank with 0.1 mL of a 50% RPMI1640/50% Matrigel (BD Biosciences) mixture containing a suspension of A549 tumor cells (approximately 5 × 10⁶ cells/mouse). At time of inoculation, the mice were 5–6 weeks old. Twelve days following inoculation, tumors were measured using a digital caliper. Sixty mice with tumor volumes of 107–180 mm³ (mean tumor volume of 142 mm³) were randomized into groups of ten mice/group. Tumor volume and body weight were recorded twice weekly. All test agents were dosed intravenously (tail vein). SDX-7320 was dosed every four days (Q4D × 7) while TNP-470 and SDX-7539 were dosed every other day (QOD × 10). All mice were sacrificed 29 days after initiating dosing.

Female C57Bl/6j mice that had undergone surgical ovariectomy at 4 weeks of age to mimic the post-menopausal state were obtained from Jackson Laboratory. EO771 tumor cells [a model of triple-negative breast cancer (TNBC), obtained from CH3 Biosystems] were implanted under ether anesthesia into the fourth mammary fat pad when mice were approximately 18 weeks of age. Treatment with vehicle (5% mannitol/water, s.c., Q4D) or SDX-7320 (24 mg/kg, s.c., Q4D × 4) was initiated when tumors were approximately 20 mm³. All mice were euthanized on day 15 when tumor and other tissues were excised, and a portion of tumor tissue was fixed in buffered formalin and processed for histology and IHC (see Supplementary Data for IHC details).

MDA-MB-231 cells were obtained from ATCC. The effect of SDX-7320 on tumor growth, angiogenesis, and metastatic invasion was evaluated in tumors grown in fertilized White Leghorn eggs as described in the Supplementary Data.

Female C57/Bl6 mice (CrTac:C57BL/6NTac) were obtained from Taconic. The B16-F10 mouse melanoma tumor cell line was from ATCC. Cultures were maintained in DMEM (Gibco, Invitrogen), supplemented with 5% FBS, and housed at 37°C in a humidified 5% CO₂ atmosphere. Beginning 24 hours after tumor cells were injected via tail vein, drug treatments were administered intravenously as follows (8 mice/group): vehicle (saline), TNP-470 (30 mg/kg), and SDX-7320 (25 mg/kg). Vehicle and SDX-7320 were dosed every fourth day starting on day 1 for three treatments (Q4D × 3) while TNP-470 was dosed every other day for six treatments (Q2D × 6). All animals were euthanized on day 15 and lungs were excised. Excised lungs were placed under a stereomicroscope and visible B16-F10 lung metastases were counted and recorded. Statistical analysis was performed with GraphPad Prism v6.0 software. Differences in lung metastasis counts were confirmed using a Student t test (two-tailed, two-sample equal variance).

Female athymic nude mice were obtained from Charles River and implanted with subcutaneous, slow-release estrogen (90-day, 0.72 mg β-estradiol; Innovative Research of America). Three days after receiving the estrogen pellets, mice were injected in the fourth mammary fat pad with 2 × 10⁷ BT-474 cells (a model of hormone receptor⁺/HER-2⁺ breast cancer), suspended in 50% Matrigel/50% RPMI1640). When tumors exceeded 95 mm³, they were eligible for randomization (8 mice/group). SDX-7320 was prepared fresh for each dose (administered s.c./Q4D × 9) in 5% mannitol/water. At necropsy, tumors were excised, and a portion snap-frozen and stored at −80°C until shipped for RNA sequencing. Methods for RNA isolation, sequencing, and analysis are described in the Supplementary Data.

Data for this study were generated at multiple contract research organizations. Reports containing raw data as well as derived data supporting the findings of this study are available from the corresponding author upon reasonable request. X-ray crystallography data are available from the Protein Data Bank (PDB) under code 8OXG. Gene expression data are available in the GEO database under accession number GSE230687.

The PDC SDX-7320 features a poly(methacrylamide) backbone functionalized with 2-hydroxypropyl groups and a glycyl-l-phenylalanyl-l-leucyl-glycyl- (GFLG) linker bearing the novel fumagillin-derived METAP2 inhibitor SDX-7539 (Fig. 1). The small molecule SDX-7539 is released from the tetrapeptide linker by enzymatic cleavage (see below). After conjugation of SDX-7539 to the copolymer via aminolysis of an active ester, the resultant PDC was purified by ultrafiltration and lyophilized to give SDX-7320 as an amorphous powder on 100-gram scale. Unconjugated SDX-7539 was not detected by HPLC analysis (LOD 0.02%). Small molecule SDX-7539 loading was measured to be 19% by weight using HPLC. SDX-7320 average molecular weight (MW) was determined by size exclusion chromatography with refractive index detection to be 29 kDa (PDI 1.49) relative to a calibration curve established using poly(methyl methacrylate) standards. Aqueous solubility of SDX-7320 was greater than 200 g/L (25°C). SDX-7320 was observed to have a monomodal particle size distribution with Z-average 10 nm (PDI 0.23) as measured by dynamic light scattering in 5% aqueous mannitol.

The tetrapeptide linker “GFLG” has been previously used to create PDCs and is a substrate for lysosomal proteases such as cathepsin B (CTSB; 47). Using a standard set of conditions, multiple cathepsin isoforms were profiled for their ability to catalyze the release of SDX-7539 from SDX-7320 in vitro (Supplementary Table S2). The results showed that several cathepsin isoforms were capable of releasing SDX-7539 from SDX-7320, and cathepsin S (CTSS) was the most active isoform against SDX-7320. Additional cathepsin isoforms with catalytic activity against SDX-7320 included L, H, C, K, and B (Fig. 2). However, differences in each enzymes’ pH optima, requirement for pre-activation with CTSL as well as specific activity limit the direct comparison of catalytic efficiency between the enzymes for cleavage of SDX-7320.

The binding of SDX-7320, SDX-9402 (the salt form of SDX-7539) and previously studied fumagillin-class METAP2 inhibitors to METAP2 was measured in vitro by competition ELISA (33). The binding of CKD-732 was less potent than SDX-7539 while SDX-9246 (N-acetyl analogue of SDX-7539) was similar to SDX-9402 (Supplementary Fig. S2). Importantly, both the PDC SDX-7320 and the small molecule SDX-9178 (lacking the spiroepoxide warhead present in SDX-7539) were inactive in this assay. These results support the established role of the spiroepoxide in binding of small molecule fumagillol analogues to METAP2 and also demonstrate that SDX-7539, when conjugated to the polymeric backbone, is restricted from binding to METAP2.

Off-target activity of the salt form of SDX-7539 (SDX-9402) was assessed against a panel of receptors/transporters and enzymes (n = 38 and 6, respectively). No effect was observed at any of the receptors or enzymes studied (up to a concentration of 10 μmol/L SDX-7539; Supplementary Tables S3 and S4).

The interaction of SDX-7539 with METAP2 was studied further by creating co-crystals of SDX-7539 and METAP2, followed by X-ray crystallography. The structure of the co-crystals was resolved to 1.6 Å and demonstrated covalent modification of active site histidine-231 (Fig. 3; Supplementary Table S5). There is a high degree of overlap between the structure of SDX-7539 bound to METAP2 and the reported structure of human METAP2 co-crystallized with TNP-470 (Supplementary Fig. S3).

Inhibition of aminopeptidase enzyme activity in vitro demonstrated both potency and selectivity of SDX-7539 for METAP2 over METAP1 (Fig. 4; Supplementary Fig. S4). SDX-7539 was equipotent compared to TNP-470 against METAP2, while neither compound inhibited METAP1 (up to 1000 nmol/L, Supplementary Fig. S4). Both of these fumagillin-class METAP2 inhibitors exhibited time-dependent shifts in IC₅₀, a hallmark of irreversible enzyme inhibitors. The PDC SDX-7320 was inactive against METAP2 (up to 10 μg/ml) in agreement with findings from competition binding assays (Fig. 4G). Note that the apparent inhibition observed at 10 μg/mL SDX-7320 was also observed in the absence of METAP2 in the enzyme assay. The analogue SDX-9178, which lacks the spiroepoxide warhead, was inactive against METAP2, underscoring the importance of this functional group for inhibition of METAP2 (Fig. 4H).

Cytostatic inhibition of endothelial cell growth is a hallmark of fumagillin and its analogues (12, 34). SDX-7539 and SDX-7320 both inhibited the growth of HUVECs (Fig. 5). SDX-7539 inhibited growth of HUVECs with sub-nanomolar potency while TNP-470 was approximately 3-fold less potent than SDX-7539. The PDC SDX-7320 inhibited the growth of HUVECs, albeit with >500-fold reduced potency relative to the small molecule inhibitors. As a prodrug, the decreased potency of SDX-7320 is likely due to the requirement for cellular uptake followed by lysosomal processing to release the small molecule METAP2 inhibitor SDX-7539. The small molecule analogue SDX-9178 (which lacks METAP2 binding and enzyme inhibition) did not affect cell growth (up to 1,000 nmol/L), indicating the requirement for METAP2 inhibition within this chemical series to decrease the growth of HUVECs (Fig. 5).

The pharmacokinetic profile of the small molecule SDX-7539 was assessed in rats, following intravenous bolus as the unconjugated small molecule and as the PDC SDX-7320 (Supplementary Fig. S5). The terminal elimination half-life for SDX-7539 after intravenous administration ranged from 10–15 minutes as the unconjugated small molecule, whereas after dosing the PDC SDX-7320, the half-life of SDX-7539 was calculated to be approximately 10 hours, most likely due to the additional processing required for enzymatic release of SDX-7539 from the PDC.

A primary driver for conjugating the small molecule METAP2 inhibitor SDX-7539 to a high MW polymer was to limit CNS exposure and minimize CNS-related toxicity. To assess the impact of polymer conjugation on CNS exposure, levels of small molecules (SDX-7539, TNP-470) in plasma and cerebrospinal fluid (CSF) were measured after intravenous dosing of either SDX-7320, SDX-7539 or TNP-470 at equimolar small molecule doses in rats. Because TNP-470 undergoes rapid metabolism in vivo, producing an active metabolite (M-IV), the M-IV metabolite was measured as opposed to the parent molecule. CSF levels of SDX-7539 (measured near plasma T_max) following intravenous administration of SDX-7320 were >50-fold lower compared with when an equivalent amount of free (unconjugated) SDX-7539 was administered, and 17-fold lower than the active metabolite of TNP-470 (Supplementary Table S6).

To assess the CNS side-effect profile of SDX-7320 in vivo, the neurobehavioral FOB was conducted in rats under GLP conditions, at 12 and 36 hours after a single subcutaneous dose of SDX-7320 (3, 10 and 30 mg/kg). No signs of abnormal CNS-related behavioral toxicity (e.g., tonic movements, startle-response, hind-limb splay) were observed in rats up to 36 hours following treatment with SDX-7320 (Supplementary Table S7). A decrease in body weight gain relative to vehicle-treated animals was found following dosing with 10 and 30 mg/kg SDX-7320 (36 hours postdose). This finding was not considered adverse due to the slight magnitude of the changes (<5%), and is not unexpected given the known effects of this drug class on adipose tissue and body weight (12). There were no other SDX-7320–related changes during the study.

Salient features of ADME and IND-enabling toxicology profiles will be presented in subsequent reports, along with the first in human clinical results from a phase I safety study of evexomostat/SDX-7320.

Anti-tumor activity of SDX-7320 was initially evaluated in A549 xenografts, a model of NSCLC and was compared with the small molecule METAP2 inhibitors SDX-7539 and TNP-470, dosed at equimolar small molecule doses. SDX-7320, administered intravenously (6, 60 mg/kg, s.c., Q4D), caused a significant, dose-dependent inhibition of A549 tumor growth, relative to vehicle-treated controls (Fig. 6A). Tumor growth inhibition in response to 6 mg/kg SDX-7320 (i.v., every 4D, which delivers the equivalent of 1.2 mg/kg SDX-7539 at each dose) matched what was observed with the small molecule SDX-7539 (IV, 37 mg/kg, every 2D) and exceeded the efficacy observed in response to TNP-470 (i.v., 30 mg/kg, every 2D), highlighting the potential advantage of polymer conjugation from a dose-effect and dosing schedule perspective.

To further investigate the potential anti-tumor effects of SDX-7320, EO771 allografts, a model of TNBC was used. SDX-7320 (24 mg/kg SC, Q4D) significantly inhibited tumor growth (Fig. 6B), and histologic analysis of dissected tumors showed reduced staining for Ki67 and CD31, indicating anti-proliferative and anti-angiogenic activity, respectively (Fig. 6C and D).

SDX-7320 was then tested in a model of human TNBC using MDA-MB-231 tumors grown in chick embryos with paclitaxel serving as a positive control. SDX-7320 (15 and 90 μg, every 4D) was administered twice, beginning two days after implantation of tumor cells. The results demonstrated dose-dependent inhibition of tumor growth relative to vehicle-treated tumors, with the higher dose exceeding the efficacy of paclitaxel (2.1 μg, every other day, total of 4 doses, Fig. 6E). The effect of SDX-7320 on angiogenesis was assessed two days prior to termination, and a significant decrease in blood vessel density was observed (Fig. 6F).

Because the chick embryo tumor model has been used to determine whether a drug possesses anti-metastatic activity (35), the anti-metastatic effects of SDX-7320 were determined in this model. Quantitation of human DNA in the lower chorioallantoic membrane was used as a measure of metastasis, and the results showed that SDX-7320 significantly inhibited metastasis of MDA-MB-231 tumors (Fig. 6G).

Building upon the findings in the chick embryo model of metastasis, the activity of both SDX-7320 and TNP-470 were investigated in a syngeneic B16-F10 melanoma lung metastasis assay. Drugs were administered beginning one day after tumor cells were introduced via tail vein injection. SDX-7320 significantly inhibited the outgrowth of B16-F10 melanoma cells in the lungs relative to vehicle-treated animals, and an identical degree of inhibition was observed with TNP-470 albeit at a dose that was six times greater than the dose of SDX-7320 (comparison is based upon small molecule equivalents; Fig. 6H).

To gain additional mechanistic insight into the anti-tumor effects of SDX-7320, BT-474 cells were utilized in an orthotopic xenograft model of HR⁺/HER-2⁺ breast cancer. SDX-7320 elicited significant, dose-dependent inhibition of BT-474 tumor growth (6, 12 mg/kg; Fig. 6I). Next RNA sequencing (RNA-seq) was used to compare gene expression in tumors from vehicle-treated versus SDX-7320–treated mice (12 mg/kg). Principal component analysis (PCA) of the RNA-seq results showed that gene expression in vehicle samples and in SDX-7320–treated samples each clustered with their own groups, however expression differences across principal component one indicated one vehicle sample may have significantly different gene expression than the other samples within this group (Supplementary Fig. S6). A number of genes were identified with significant differences in expression between vehicle-treated and SDX-7320–treated tumors (Fig. 6J). Pathway analysis conducted on the set of genes with significant differences in expression between the two groups showed that several pathways important in tumor biology including “Hallmark Inflammatory Response” as well as “Hallmark Hypoxia” were significantly altered in tumors from SDX-7320 versus vehicle-treated mice (Fig. 6K). The expression of several hypoxia-regulated genes including the master regulator of the hypoxic response, HIF1α, were significantly reduced in tumors from SDX-7320–treated mice (Fig. 6L). This suggests that inhibiting the adaptive response of BT-474 tumors to hypoxia is one mechanism by which SDX-7320 inhibits tumor growth, or alternatively that normalization of tumor vasculature following inhibition of METAP2 led to the attenuation of hypoxia (36). Genes related to inflammation were significantly affected in tumors from SDX-7320–treated mice, including IL6, chemokine genes CXCL1, CXCL3, CXCL8, and CCL20, as well as the chemokine receptors CCR7 and CXCR4 which were all significantly suppressed (Fig. 6L; Supplementary Data DEG file).

The ability of the mycotoxin fumagillin, and semi-synthetic derivatives such as TNP-470, to act as irreversible inhibitors of METAP2, with consequent anti-angiogenic and anti-proliferative potential, is well established (37). TNP-470 itself has been extensively investigated and has displayed potent anti-tumor activity in multiple preclinical models as well as in the clinical setting (24, 25, 38). A poor PK profile and limited tolerability of TNP-470 have driven the exploration of structural modifications based on the fumagillol pharmacophore, retaining the irreversible nature of METAP2 inhibition (1). SDX-7539 is a novel small molecule fumagillol derivative bearing a 4-aminocyclohexyl carbamate with trans stereochemistry across the cyclohexane ring whose half-life is 4-fold longer than that of TNP-470, and does not possess the N-oxidation potential nor the hydrolytic ester lability in vivo of some second-generation analogues (e.g., CKD-732; refs. 39, 40; Fig. 1). SDX-7539 is a potent and selective inhibitor of METAP2, based on results from in vitro binding, enzyme, and cell-based assays. X-ray structure determination of METAP2-SDX-7539 co-crystals yielded strong similarities to the published structures of METAP2-Fumagillin and METAP2-TNP-470, confirming the nucleophilic attack of the histidine-231 imidazole on the spiroepoxide, forming a covalent bond in the enzyme active site (41).

Polymer conjugation has been investigated for several decades, mainly with the aim of enhancing drug delivery and improving the therapeutic index of pharmacologically active small molecules (26). Covalent conjugation to various water soluble polymers using TNP-470 as starting material has been reported wherein facile nucleophilic displacement of the TNP-470 chloride heteroatom was utilized to generate anti-angiogenic PDCs bearing a 2-O-acetyl-linked carbamoylfumagillol attached to poly(vinyl alcohol), or bearing 2-aminoethylglycyl-carbamoylfumagillol attached to poly(HPMA) via a tetrapeptide linker (29, 42). Although these PDCs showed promising preclinical activity in their respective indications, neither proceeded to human clinical evaluation. The PDC SDX-7320 is based on a non-immunogenic, biocompatible, hydrophilic p(HPMA) polymer backbone (43, 44) and utilizes a GFLG tetrapeptide linker to conjugate the selective SDX-7539 METAP2 inhibitor (Fig. 1).

In the genesis of the high MW polymer conjugate SDX-7320 and its active component SDX-7539, objectives were: to ameliorate the dose-limiting CNS toxicity of historical fumagillin derivatives such as TNP-470 by reducing the amount of small molecule penetrating the blood–brain barrier; to provide a sustained PK profile of the active moiety suitable for convenient administration; and to harness potential preferential tumor accumulation of a novel, potent METAP2 inhibitor via the enhanced permeability and retention effect (EPR) effect. This report describes the characteristics of SDX-7320 and its pharmacological profile as a novel anti-tumor agent.

In terms of physical properties described above, the active pharmaceutical ingredient SDX-7320 has high aqueous solubility (>200 mg/mL), and excellent long-term chemical stability has been observed at ambient conditions (> 48 months). On a weight basis, the SDX-7539 payload was determined to be approximately 20%, and the SDX-7320 average MW was approximately 30 kDa, with a particle size distribution measured to be 10 nm Z-average (PDI < 0.25) in aqueous solution. Accordingly, SDX-7320 is a high drug loading, hydrophilic PDC which has the potential to take advantage of preferential macromolecular passive tumor accumulation while remaining largely within the approximate glomerular filtration threshold (about 30–50 kDa) suitable for renal clearance (45). Importantly, free, unconjugated SDX-7539 was not detected in SDX-7320 as isolated, and based on the relative activities of the PDC versus the small molecule, METAP2 inhibition is not achieved until SDX-7539 is released enzymatically.

The tetrapeptide linker “GFLG” has been previously used to construct PDCs and can be hydrolyzed by lysosomal proteases such as cathepsin B (CTSB; 46). However, as endopeptidase activity is required, the proximal structure of the drug being cleaved is clearly fundamental to isoform specificity and efficiency: only select cysteine cathepsin isoforms were capable of releasing SDX-7539 from SDX-7320, with cathepsin S (CTSS) exhibiting the highest activity in vitro against SDX-7320.

In the case of SDX-7320 a major objective of polymer conjugation was to improve the therapeutic index relative to previous fumagillin-based METAP2 inhibitors such as TNP-470 by conjugating the pharmacologically active moiety SDX-7539 to a high molecular weight prodrug backbone, thereby limiting CNS exposure. Analysis following intravenous administration of TNP-470, SDX-7539, or the PDC SDX-7320 at equimolar small molecule doses demonstrated a marked reduction in CSF levels of SDX-7539 when dosed as the PDC compared with the unconjugated small molecule or to TNP-470 (Supplementary Table S6). In addition, no significant behavioral abnormalities were observed in rats following a single doses of the SDX-7320 PDC in a GLP functional observational battery (Supplementary Table S7).

As a macromolecular system structurally based on poly(HPMA) with a MW ∼ 30 kDa, SDX-7320 is purported to benefit from two passive tumor–, and tumor microenvironment–targeting processes: the EPR effect and extravasation through leaky vasculature and inflammatory cell-mediated sequestration (ELVIS). The EPR effect is a well-described, albeit mechanistically controversial, concept which may enable appropriately sized nanoparticles to preferentially accumulate, relative to small molecules, in solid tumors via diffusion through the complex, immature neovasculature with reduced escape due to limited lymphatic drainage (47, 48). ELVIS is a more recently described phenomenon in which certain HPMA-based copolymer systems have been reported to target tumor-associated macrophages (49, 50).

Polymer conjugation is anticipated to be most beneficial when conjugating inhibitors with high potency against the target of interest, in order to limit the total dose of PDC. Unexpected reductions in efficacious doses of the SDX-7320 PDC were observed compared to either the released small molecule SDX-7539 alone or other small molecule METAP2 inhibitors, in xenograft and in syngeneic models of cancer (Fig. 6A and H). For example, in the B16-F10 melanoma model (Fig. 6H) when SDX-7320 was dosed at 25 mg/kg (delivering approximately 5 mg/kg SDX-7539 per dose) every four days, it was equally efficacious as TNP-470 dosed at 30 mg/kg every other day. After 15 days of treatment, the effect on lung metastases in response to each agent was virtually identical despite an 83% reduction in total dose of small molecule METAP2 inhibitor delivered with the PDC. Polymer conjugation likely contributes to the reduced total drug dose required for efficacy by prolonging METAP2 inhibition, secondary to increased half-life of the inhibitor, and possibly by enhancing tumor uptake, due to the EPR and ELVIS effects. To the best of our knowledge, this is the first instance that reductions in small molecule–equivalent efficacious doses following polymer conjugation have been reported, as typically the total dose of small molecule–equivalent increases following polymer conjugation, rendering SDX-7320 unique in this respect.

The mechanism by which SDX-7320 inhibits tumor growth is multifactorial, extending beyond its anti-angiogenic effects, to include inhibition of metastases. Analysis of gene expression data from BT-474 tumors treated with SDX-7320 identified multiple pathways that were affected, including hypoxia and inflammatory responses which may contribute to inhibition of tumor growth following METAP2 inhibition.

Potential limitations of this PDC include tumor penetration that is dependent upon hemodynamics (i.e., EPR/ELVIS effects) and on passive uptake, which could limit PDC tumor exposure and subsequent oncologic efficacy. In addition, the requirement for cathepsin-mediated processing of the PDC to release the active moiety raises the potential for inter-subject variability based upon levels of these enzymes. This report represents the initial disclosure of the structure, synthesis, and preclinical pharmacological profile of the PDC SDX-7320, and its active payload SDX-7539. As a polymer-based prodrug, SDX-7320 has displayed multiple improved characteristics compared with historical fumagillin-derived small molecule analogues such as TNP-470. This METAP2 inhibitor, a next-generation nanomedicine, has completed a phase I clinical dose-escalation safety trial in solid tumors (NCT02743637) and is currently being investigated in multiple phase Ib/II clinical proof-of-concept studies in patients with TNBC and hormone receptor–positive breast cancer (NCT05570253, NCT05455619).

P. Cornelius reports patents 10,646,463, 11,273,142, and 11,612,577 issued; patents for 18/036,565, PCT/US2023/064550, PCT/US2023/072023, and PCT/US2023/080080 pending; in addition, P. Cornelius is an employee of SynDevRx, Inc and holds stock options in SynDevRx, Inc. B.A. Mayes reports patents for PCT/US2023/064550 and PCT/US2023/080080 pending; in addition, B.A. Mayes is an employee of SynDevRx, Inc. and has ownership of stock options in SynDevRx. J.S. Petersen reports patents 9173956, 9433600, 9750737, 9757373, 10010544, 10588904, 11304944, 9320805, 9585909, 9730955, 9895449, 10159692, 10722532, 9969722, and 10287277 issued. P.J. Dufour is an employee SynDevRx, Inc, with ownership and options on equity. A.J. Dannenberg reports personal fees from SynDevRx during the conduct of the study. J.M. Shanahan reports patents 9,173,956, 9,433,600, 9,750,737, 9,757,373, 10,010,544, 10,588,904, 11,304,944, 10,646,463, and 11,273,142 issued; in addition, J.M. Shanahan is a full-time employee of SynDevRx, and holds both stock and stock options. B.J. Carver reports patents for 18/036,565 and PCT/US2023/064550 pending; in addition, B.J. Carver is a full-time employee of SynDevRx, Inc., and holds both stock and stock options in SynDevRx, Inc. No disclosures were reported by the other authors.

P. Cornelius: Conceptualization, data curation, supervision, visualization, methodology, writing–original draft, writing–review and editing. B.A. Mayes: Conceptualization, data curation, formal analysis, supervision, visualization, methodology, writing–original draft, writing–review and editing. J.S. Petersen: Conceptualization, formal analysis, supervision, investigation, methodology, project administration, writing–review and editing. D.J. Turnquist: Formal analysis, supervision, investigation, writing–review and editing. P.J. Dufour: Data curation, software, formal analysis, investigation, visualization, methodology, writing-review and editing. A.J. Dannenberg: Writing–review and editing. J.M. Shanahan: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing. B.J. Carver: Conceptualization, resources, supervision, funding acquisition, project administration, writing–review and editing.

The authors would like to acknowledge the seminal work of Dr. Judah Folkman in the field of angiogenesis, Dr. Donald Ingber, MD, PhD in discovering the antiangiogenic activity of fumagillin, Dr. Daniel Von Hoff for discussions in the development of SDX-7230, and Ronit Satchi-Fainaro, PhD for her early work on polymer conjugates. The following are acknowledged for executing preclinical experiments and data collection: SBH Sciences, Neosome Life Sciences (Sara Little), QPS, Vivisource, Apredica, MPI Research, TD2, Inovotion, Wax-It Histology Services, Azenta/GeneWiz, Watershed AI, Pharm-analyt GmbH, Peak Proteins (Dr. Stephen Moss). In addition, the following are acknowledged for performing chemical synthesis and analytical development: Organix, Inc. (Dr. Paul Blundell, Dr. Mario Gonzales, Dr. Jie Li), Solid Form Solutions Ltd., PCI, Inc., Lyophilization Services of New England, Inc., Omm Scientific, Inc. (Dr. Donald Stewart), Boston Analytical, Jordi Labs, LLC, HepatoChem, Inc., Patheon, Custom NMR Services (Dr. Jin Hong), and CuriRx. The authors would like to thank Andrea Gwosdow and Becca Kranz (Gwosdow Associates Science Consultants) for their assistance in preparation of the draft and are grateful to Prof. Bruce Zetter, PhD and Dr. Rakesh Jain for providing feedback on the manuscript. The graphical abstract was created with BioRender.com.