We're Not “DON” Yet: Optimal Dosing and Prodrug Delivery of 6-Diazo-5-oxo-L-norleucine

Glutamine is the most abundant amino acid in blood, with serum concentration of several hundred micromolar. Several recent reviews detail glutamine utilization in multicellular organisms (1–3). Rapidly proliferating healthy cells (GI epithelium, lymphocytes) or cells under physiologic stress have increased demand for glutamine. Glutamine is transported into cells by one of multiple amino acid transporters (e.g., ASCT2 and BOAT2), several of which are thought to be upregulated in cancer cells (2). Once inside the cell, glutamine is used in multiple metabolic processes (Fig. 1): (i) It is hydrolyzed to glutamate and ammonia by glutaminase (“glutaminolysis”; ref. 4); (ii) it is subject to glutamine amidotransferases that catalyze the use of its amido group as a building block for nucleosides, amino acids, and hexosamine sugars (2); and (iii) it is used directly to charge tRNA for protein synthesis. Glutamate, produced from glutamine by glutaminase and glutamine amidotransferase activities, may be further metabolized to alpha ketoglutarate and provide a carbon skeleton source for the mitochondrial tricarboxylic acid cycle (TCA cycle). In addition to generating reducing equivalents and high-energy intermediates through mitochondrial respiration, the TCA cycle also exports carbon skeletons for synthesis of lipids, nucleotides, and reducing equivalents, particularly in rapidly proliferating cells (4, 5). Glutamine-derived glutamate is also involved in the synthesis of the reducing equivalent glutathione, vital to maintaining cellular redox status.

Many tumors become largely dependent on glutamine to provide carbon and nitrogen building blocks needed for proliferation. Warburg noted in the 1920s that in the presence of adequate oxygen, tumors increase glucose uptake and ferment much of it to lactate (6). In cancer model systems, Eagle and colleagues (7) first demonstrated tumor cells in culture require supplementation with exogenous glutamine for efficient proliferation. It was subsequently shown that when deprived of glutamine tumor cells undergo apoptosis (8). As interest in cancer metabolism has grown, glutamine utilization by cancer cells and its genetic regulation have become areas of intense interest (1–3). The most well-characterized oncogene to regulate glutamine metabolism is MYC (9), which enhances glutaminase expression, upregulates glutamine transporters, and enhances glutamine utilization in energy production and biosynthesis (1). Other pro-tumorigenic regulators such as KRAS and mTOR, as well as tumor suppressors (p53, VHL) have also been associated with alterations in glutamine metabolism (5, 10).

Tumor glutamine dependence has been targeted with selective glutaminase inhibitors with some success. Several allosteric inhibitors, including BPTES [bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide], compound 968, and CB-839 (Calithera) have shown robust activity in cell culture experiments and promising single-agent preclinical activity (11–13). CB-839 has proceeded into clinical studies. Although target engagement was clearly observed (14), single-agent antitumor activity was minimal; combination trials are now underway with promising initial results (15, 16).

Perhaps a reason for the lack of robust clinical effect of selective glutaminase inhibitors is that glutamine metabolism in tumors is more complex than initially hypothesized. Tumor cells are highly adaptable and alter nutrient uptake and metabolic networks to resist single-agent glutaminase inhibition (17, 18). Therefore short-term cell culture and preclinical studies may not adequately predict the metabolic response of tumors with longer-term drug exposure. In addition, in vitro studies rarely account for the effects of stromal cells or the microenvironment on nutrient availability to tumor. Indeed, it was recently shown that cells in the microenvironment of several tumor types upregulate glutamine production, thereby enabling tumor cells to escape glutaminase inhibition (19). All of these studies recommend combination therapy as a means to improve efficacy and avoid tumor resistance to single-agent glutaminase inhibition or a broader approach to inhibition of glutamine utilization.

6-diazo-5-oxo-L-norleucine (DON; Fig. 2A) is the best-studied broadly active glutamine antagonist, having multiple supporting biochemical, preclinical and clinical evaluations. DON was originally isolated from fermentation broth of a Streptomyces in the 1950s (20). Biochemical studies on DON identified a two-step, mechanism-based mode of inhibition across multiple glutamine-using enzymes. First, DON binds competitively to the glutamine active site, then a covalent adduct is formed irreversibly inhibiting the enzyme (21). Importantly, DON's diazoketone group is stable under physiological conditions because of the electron-withdrawing carbonyl group stabilizing the diazo dipole. As a result, DON acts as a reactive electrophile only when protonated at the α-position under certain conditions (e.g., in the proximity of the active-site serine residue in glutaminase), triggering the release of nitrogen (N₂; ref. 22). Thus DON serves as a selective mechanism-based inactivator of glutamine-using reactions rather than a non-specific reactive intermediate. DON inhibits glutamine-using enzymes, including glutaminase at low micromolar levels (21) as well as multiple glutamine amidotransferases (23) involved in de novo purine and pyrimidine synthesis (24–28), coenzyme synthesis (29), amino acid synthesis (30, 31), and hexosamine production (ref. 32; Fig. 1). The kinetics of inhibition and inactivation have been described for some, though not all, of DON's target enzymes. At far higher concentrations DON also serves as a substrate and an inhibitor of several amino acid transporters and transglutaminases (28), as well as a number of amino acid synthesis reactions more relevant in prokaryotic systems (33).

In early preclinical cancer models, DON inhibited the growth of multiple cancer cell lines in culture (34), prevented tumor growth, and increased survival in several murine cancer models including murine sarcomas, carcinomas, and leukemias (28, 35). Of note, the most effective dosing regimens in some early rodent studies were daily low-dose therapy (36). On the basis of these data, there was interest in taking DON into human studies as an antitumor agent.

The first-in-human study of DON was published in 1957. Magill and colleagues (37) evaluated DON as a single agent in 63 patients. The majority of patients had advanced inoperable solid tumors, including lung, breast, colon, and genitourinary tumors; a smaller number had leukemia or lymphoma. DON was given as a single agent by either intravenous, intramuscular, or oral routes at doses ranging from 0.2 to 0.6 mg/kg/dose once daily (IV/IM) or 0.2 to 1.1 mg/kg/dose (oral; conversion 7.4–40.7 mg/m²/dose; Table 1). Several smaller groups of patients also received multiple smaller daily doses (e.g., 0.2 mg/kg/dose q4–6 hours) or less frequent infusions at higher mg/kg dosing (1.6 mg/kg IV q4 days). Duration of DON treatment varied on the basis of toxicities but three-quarters of patients were treated for two or more weeks.

Magill's early study was not designed or powered to show therapeutic efficacy and clinical trial terminology was less standardized than it is today; nonetheless there were hints of activity among the patients treated (Table 1). Seven of 47 (15%) patients who had at least 2 weeks' duration of DON exposure were reported to have partial response to treatment. These included 2 patients with breast cancer and 2 patients with bronchogenic carcinoma with decrease in metastases. An advanced sigmoid adenocarcinoma patient, a uterine cancer patient, and an advanced Hodgkin's lymphoma patient also experienced transient partial responses of metastatic disease. Four additional patients (two breast tumor and two “other carcinoma” patients) showed reversal of hypercalcemia of malignancy in response to DON. In 24 patients, there was neither demonstrable regression nor progression of the disease, and DON therapy was deemed to have “no effect” on this subset. Twelve patients had disease progression.

Following the report by Magill and colleagues (37), several other groups evaluated daily dosing of DON in humans against a variety of malignant diseases (Table 1; refs. 38–40). Included among these was one pediatric study in which children with acute leukemia were treated with the combination of DON (daily 0.25 mg/kg) and 6-mercaptopurine (6-MP). Forty-two percent of patients achieved complete remission with the combination therapy, an advantage over 6-MP alone.

In the 1980s interest in the antitumor effects of DON were rekindled by its robust activity against several human tumor xenografts in the newly introduced nude mouse model. DON was reported to be effective against human lung, colon, and breast tumor xenografts, inducing tumor regression at high intermittent dosing regimens (25–100 mg/kg IP every 4 days; ref. 41). Mechanistic studies showed DON treatment led to decreased cell proliferation and S phase blocks (42). Animal toxicity studies were performed by the National Cancer Institute using azotomycin, which is metabolized to DON. These studies established target organs of the gastrointestinal tract and bone marrow; effects on the heart, kidneys, and liver became more pronounced with increased drug exposure (20, 43). Because the highest DON exposures could be achieved with intermittent doses and this dosing regimen was effective at inducing tumor regression in the nude mouse model, an intermittent dosing paradigm was employed in further clinical trials.

Five phase I dose-escalation studies of DON were carried out and published in the 1980s (43–47). Patients received DON intravenously in doses ranging from 50 mg/m²/dose to 600 mg/m²/dose (approximately 1.35–16.2 mg/kg equivalent) generally two to three times weekly every 3 to 4 weeks. Four studies examined adult patients with a range of treatment refractory cancers, and one study focused on a pediatric population. Among the five phase I studies, no objective responses to therapy were noted in any adult solid tumor group with measurable disease. Among the pediatric patients treated with DON as a single agent, a handful of both hematologic and solid tumor patients had studies suggesting stable disease to partial responses (43).

The subsequent phase II studies of DON at intermittent high doses carried out between the 1980s and 2000s generally had disappointing outcomes (Table 1; refs. 48–51). Four of these studies used DON as a single agent in advanced or refractory adult solid tumors. These studies used maximum-tolerated dosing regimens based on the phase I trials ranging from 50 mg/m²/dose daily for 5 days every 3 weeks, given to sarcoma patients, to 200 mg/m²/dose twice weekly for two weeks with 1 week off, given to patients with colorectal cancer. There were few clinical responses even in the more well-defined oncology populations, with the exception of data published as an abstract more recently on concurrent therapy with DON and PEGylated glutaminase (52). On the basis of these collective experiences, it was concluded that intermittent high doses were ineffective and too toxic (see below), and DON was abandoned as an anticancer therapy.

In studies from the 1950s using daily doses of DON, the most common toxicity was direct toxicity to the gastrointestinal mucosa (Table 1). Mucositis was observed in up to 83% patients, in spite of preventive treatments, and diarrhea in 48% patients (37). Nausea and vomiting (30% patients), GI bleeding (16% patients), and abdominal pain (8% patients) also occurred. One third of patients also had hematologic changes (leukopenia/thrombocytopenia) but these included patients with underlying bony metastases or disease involving the marrow. Mucositis and diarrhea were more severe in patients receiving multiple daily divided doses compared with patients receiving the same total dose in one daily infusion, though sample sizes in the divided dose regimens were small. Similar side effects were reported across the early studies using low daily DON doses (37–40). In the DON and 6-MP combination study the improved outcomes with combination therapy were offset by an increase in toxicity, especially mucositis and other gastrointestinal symptoms.

When DON was administered on intermittent high dose schedules, nausea and vomiting were the major dose-limiting toxicities (Table 1; refs. 48–52)—the stomatitis/mucositis/diarrhea observed on earlier daily dosing regimens was less frequently observed even though the total dose administered was much greater. Moderate to severe nausea and vomiting was observed in one study using DON doses between 300 and 550 mg/m² given by rapid IV infusion (45). In these patients, DON peak plasma concentration (C_max) was found to be over 20 μg/mL (45). Longer infusion times of high intermittent doses of DON in two additional studies resulted in peak plasma concentrations measured at less than 10 μg/mL and slightly milder side effects (46). By contrast, low daily dosing of DON in earlier studies (0.2–0.3 mg/kg IV) resulted in peak plasma concentrations of only 400 to 500 ng/mL measured by an indirect microbiological assay (37). This dose schedule primarily resulted in oral mucositis and some diarrhea, with less nausea and vomiting reported.

Taken together, these observations suggest that low daily or divided dose exposures produced low plasma concentrations and a characteristic local GI toxicity, whereas high intermittent doses lead to more typical chemotherapy-induced nausea and vomiting. Of note, at the time many of the clinical studies on DON were done, 5HT₃ receptor antagonists were not routinely in clinical practice as antiemetics, although several of the phase II studies and a pediatric phase I study pretreated patients with prochlorperazine or chlorpromazine and observed decreased severity of nausea and vomiting (43, 48, 51). Other non–dose-limiting toxicities of DON observed in the high intermittent dosing studies include reversible myelosuppression and transient hypocalcemia (43, 44, 47).

Our review of the clinical observations suggests that the low dosing paradigm of the early studies was sufficient to inhibit glutamine utilization. In the trials using low daily dosing stomatitis and diarrhea were consistently observed, reflecting local glutamine starvation of the gastrointestinal mucosa and inhibition of cell cycle in rapidly proliferating intestinal epithelial cells. Of note, these side effects are not reported with CB-839, the glutaminase inhibitor currently in human trials (14), suggesting that full blockade of glutamine utilization is dependent on enzymes beyond glutaminase. Evidence for low dose DON inhibition of glutamine amidotransferases was directly observed in humans in a study of DON for gout (53). In this study, DON was administered at low daily intramuscular doses of only 5 to 20 mg and inhibited glutamine utilization as measured by radiotracer. These doses were more than an order of magnitude lower than those used in subsequent cancer trials in the 1980s (45, 50, 51), supporting the notion that high intermittent dosing was not an optimal schedule to achieve blockade of glutamine metabolism with DON.

Targeting cancer metabolism may be complementary to classic cytotoxic approaches and many inhibitors of “metabolic targets” are in preclinical studies or in early clinical trials (54). As is suggested for DON optimal dosing of metabolic inhibitors will differ compared from the cytotoxic agents typically used with intermittent dosing. Constant target inhibition with overall low dose therapy is thought to be important for avoiding resistance mechanisms that can occur more readily when agents are started and stopped intermittently. Currently approved metabolic inhibitors are dosed clinically on a daily to BID basis, allowing for constant target inhibition at doses below the maximally tolerated dose (55).

Given DON's breadth of activity and promising efficacy in treating glutamine-dependent tumors, particularly with daily low doses, it is of clinical importance to identify a strategy to deliver it preferentially to the tumor while minimizing its peripheral exposure, and thus toxicity to non-tumor tissues. Prodrugs have proven to be a valuable strategy to alter tissue distribution and improve delivery of active agents to target tissues (56). We have developed a dual prodrug strategy to mask DON's carboxylic acid and amino groups with addition of promoieties (Fig. 2A). The resulting prodrugs circulate intact and inert in plasma and preferentially release DON at the target tissue, permitting dose reductions and improved therapeutic index. We have developed two types of DON prodrugs—those that preferentially deliver DON to brain and those that preferentially deliver DON to peripheral tumors.

We were particularly interested in delivering DON to the CNS as many brain tumors (e.g., gliomas, medulloblastoma) have amplification of the MYC or MYCN oncogenes, potentially rendering them glutamine dependent (57). The brain-targeted prodrugs, exemplified by Prodrug 1 (Fig. 2A) were designed to introduce bulky hydrophobic groups to enhance the cLogP and cellular permeability of DON, while maintaining plasma stability such that the peripheral exposure to DON would be limited (11, 58). Following infusion in swine, Prodrug 1 delivered DON preferentially to CSF and brain, resulting in a 15-fold enhanced CSF/plasma ratio and 9-fold enhanced brain/plasma ratio when compared directly with equimolar doses of DON (Fig. 2B, I and II; ref. 58). A second class of prodrugs was designed to enhance tumor delivery by using promoieties that are cleaved by enzymes enriched in tumor. This approach is exemplified by Prodrug 2, designed to take advantage of HDAC and Cathepsin L overexpression by tumors (59–61). When directly compared with DON, Prodrug 2 shows a greater than 50-fold enhanced tumor cell/plasma ratio (Fig. 2C). Importantly, when incubated with GI tissue homogenates, Prodrug 2 did not release significant amounts of DON (Fig. 2D), confirming that DON delivery is preferential to tumor cells and suggesting that the prodrug approach should decrease gut exposure to DON, and thereby reduce GI toxicity.

We have evaluated the in vivo tolerability and efficacy of DON prodrugs using a low daily dosing paradigm, reminiscent of the original studies in humans (37) and similar to dosing paradigms employed with other metabolic inhibitors (55). Mice bearing EL-4 lymphoma were treated with vehicle or DON prodrug (1 mg/kg p.o. × 5 days followed by 0.3 mg/kg, p.o. × 9 days). As clearly demonstrated, Prodrug 3 (62) robustly induced tumor elimination while mice succumbed to vehicle-treated tumors (Fig. 2E). No overt side effects or weight changes were observed with prodrug treatment, in agreement with the observation that DON prodrugs do not release significant DON in gut tissue (Fig. 2D).

DON prodrugs may also avoid the resistance to glutaminase inhibition and metabolic adaptation to single-agent therapy that has been a concern in trials of the selective glutaminase inhibitor CB-839. In contrast, DON broadly inhibits glutamine utilization, including glutaminase, glutamine amidotransferases (used in de novo purine and pyrimidine synthesis, coenzyme synthesis, and hexosamine synthesis), and glutamine synthetase (Fig. 1). Thus DON prodrug inhibition may be more difficult for tumor and stromal cells to adapt to and circumvent. DON's breadth of inhibition, tolerability at low daily doses, and potential utility in combinations can now be exploited with the prodrug approach in further clinical oncology studies.

With the growing appreciation of the complexity of targeting glutamine metabolism in cancers and the promising preclinical results with the novel DON prodrugs described above, we believe that clinical studies of these DON prodrugs are warranted. Dosing in humans should be trialed at low daily doses—the prodrug doses that we have tested in mice of 0.3 to 1 mg/kg are equivalent to less than 0.1 mg/kg in humans, which is below the lower limit of tolerable doses evaluated in the original 1957 human study (37, 63). Unlike the early DON clinical trials, however, modern DON prodrug clinical studies should aim to include oncology patients whose tumors have the best chance of benefiting from therapy targeting tumor glutamine dependence.

We propose that investigations of DON prodrugs be prioritized in three clinical areas. First, many CNS tumors are in need of new therapeutic options that cross the blood–brain barrier. Preclinical evidence suggests that subtypes of brain tumors (e.g., glioblastoma, group III medulloblastoma) have MYC or MYCN amplification, are glutamine-avid, thus should benefit therapeutically from glutamine antagonists (64, 65). Brain-targeted DON prodrugs have been shown to deliver DON preferentially to the CSF to enhance drug activity in a tissue usually inaccessible to chemotherapy. In this way brain-targeted DON prodrugs offer an advantage when compared to CB-839, which is reported to have limited CNS availability (66).

Second, non-CNS tumors with evidence of glutamine dependence are good candidates for treatment with DON prodrugs. Multiple means exist to identify such populations. Mutations in genes established to drive glutamine dependence (e.g., MYC or MYCN, KRAS) as well as more recently identified genetic mutations affecting glutamine utilization (e.g., VHL, KEAP1, ASS1) can be used for rational patient selection (67–72). Alternatively, or in combination, new technologies to evaluate tumor metabolism in vivo may be employed. Positron emission-tomography (PET)–based imaging with ¹⁸F-glutamine, evaluation of the tumor environment with magnetic resonance spectroscopy, and metabolomics or tracer profiling of tumor biopsy at diagnosis can identify increased glutamine uptake or dependence in vivo (64, 65, 73–75). These techniques may also prove valuable in clinical trials to establish makers of blockade of glutamine utilization after DON prodrug administration.

Finally, we propose that DON prodrugs be evaluated in combination with either standard-of-care therapies or in rationally directed new pairings. For example, in an early clinical trial, DON was successful when combined with 6-MP in pediatric acute leukemias (1, 40). In addition DON prodrugs should be combined with newer targeted therapies. Glutamine metabolism promotes tumor resistance to mTOR inhibitors in certain tumor contexts (54, 64); the combination of a DON prodrug and an mTOR inhibitor will be interesting to study. Finally, it is known that aberrant glutamine metabolism by a tumor can lead to more-wide ranging effects on the local microenvironment or even whole organism that play into the broader effects of cancer (3, 19). Therefore, DON prodrugs should be evaluated for synergy with tumor cell non-autonomous therapeutics, including immunotherapy agents.

By inhibiting a wide range of metabolic pathways using glutamine that are critically upregulated in cancer cells, DON has been shown to be a selective but broadly targeted therapy. To deliver the active drug to compartments like the CNS and minimize GI exposure, we have synthesized and conducted thorough preclinical investigations of targeted DON prodrugs. Now with an improved understanding of how to dose these agents and what tumor types are best suited for broad inhibition of glutamine metabolism, these novel glutamine antagonists are well positioned for future success in the clinic.

J.J. Vornov is a vice president of Medical Affairs at Medpace. R. Rais and B.S. Slusher have ownership interest (including stock, patents, etc.) and are the consultant/advisory board members for Dracen Pharmaceuticals. No potential conflicts of interest were disclosed by the other author.

We apologize to authors whose work we could not cite due to space limitations. We thank Eric Raabe, Jonathan Powell, and Frank Smith for critical reading and comments. We thank members of Johns Hopkins Drug Discovery for helpful discussion and figure preparation. We gratefully acknowledge Jennifer Fairman's assistance with the preparation of Fig. 1. This work was supported by NIH T32CA060441 (to K.M. Lemberg), an Alleghany Health Network grant (to R. Rais), and NIH P30MH075673, R01 CA193895 and Alex's Lemonade Stand Foundation, Children's Cancer Foundation, Alleghany Health Network, and TEDCO Maryland Innovation Initiative grants (to B.S. Slusher).