|Home | About | Journals | Submit | Contact Us | Français|
Efforts to improve cancer care in the developing world will benefit from the identification of simple, inexpensive, and broadly applicable medical modalities based on emergent innovations in treatment, such as targeting mechanisms of tumoral immune tolerance. In this report we offer preclinical evidence that the low-cost, anti-inflammatory agent ethyl pyruvate (EP) elicits a potent immune-based antitumor response through inhibition of indoleamine 2,3-dioxygenase (IDO), a key tolerogenic enzyme for many human tumors. Consistent with its reported ability to interfere with NFκB function, EP blocks IDO induction both in vitro and in vivo. Antitumor activity was achieved in mice with a non-cytotoxic dosing regimen of EP previously shown to protect against lethality from sepsis. Similar outcomes were obtained with the functional EP analog 2-acetamidoacrylate. EP was ineffective at suppressing tumor outgrowth in both athymic and Ido1-deficient mice, providing in vivo corroboration of the importance of T cell-dependent immunity and IDO targeting for EP to achieve antitumor efficacy. While EP has undergone early phase clinical testing, this was done without consideration of its possible applicability to cancer. Our findings that IDO is effectively blocked by EP treatment deepens emerging links between IDO and inflammatory processes. Further, these findings rationalize oncological applications for this agent, by providing a compelling basis to reposition EP as a low cost immunochemotherapy for clinical evaluation in cancer patients.
Mounting evidence argues that immune escape through the establishment of dominant, immune tolerance is a fundamental hallmark of tumor progression (1), and it is likely that successful immunotherapy will require the implementation of strategies to overcome this barrier through targeting of tolerogenic determinants protecting the tumor (2). One promising target for pharmacologic intervention in this regard is the tryptophan catabolizing enzyme indoleamine 2,3-dioxygenase (IDO) (3, 4). Two closely related IDO isoforms encoded by the IDO1 and IDO2 genes have been identified. IDO2 appears to be expressed in a more restricted range of tissue types than IDO1 (5, 6), but the functional and physiological relevance of the IDO2 isoform has yet to be clearly delineated while genetic evidence clearly supports the importance of IDO1 in tumoral immune escape (7, 8). In numerous clinical studies, IDO upregulation in cancer patients has been associated with a less favorable prognosis (9), while in various animal models of cancer, systemic blockade of IDO activity with small molecule inhibitors suppresses the outgrowth of tumors and cooperates with chemotherapy, radiotherapy, or cancer vaccines to trigger regression of tumors that are otherwise recalcitrant to treatment (7, 10, 11). These encouraging outcomes have sparked interest in further discovery and development of inhibitors of IDO signaling to evaluate as cancer therapeutics (12–16). One approach to facilitating this process is to consider whether any existing agents might leverage this immunologic mechanism to permit repositioning for cancer treatment.
In earlier studies of IDO dysregulation in cancer, we demonstrated that, in addition to the well established JAK/STAT signaling requirement, NFκB signaling is also essential for IDO induction in oncogenically transformed skin epithelial cells (10). In a skin carcinogenesis model, we subsequently demonstrated that IDO is critical for inflammation-based tumor promotion (8). Given the central involvement of NFκB signaling in both cancer and inflammation, we speculated that relieving IDO-mediated tumor tolerance may be a key mechanism whereby clinical agents that interfere with NFκB signaling might exert an immunotherapeutic effect in cancer. Among clinically evaluated anti-inflammatory agents that have been shown to inhibit NFκB signaling, ethyl pyruvate (EP) is particularly notable as a simple, inexpensive, non-toxic food additive that, via intraperitoneal or intravenous routes of administration, displays in vivo efficacy in mouse models of sepsis and other inflammatory disorders (17). Here we report preclinical evidence that EP can induce robust antitumor immune responses through its ability to inhibit the in vivo expression of IDO.
Chemicals were purchased from the following vendors, ethyl pyruvate (EP) Aldrich cat. #W245712, 2-acetamidoacrylic acid (2-AA) Fluka cat. #00190, 6-Amino-4-(4-phenoxyphenylethylamino)quinazoline (QNZ) Biomol cat. #EI352, 1,1-Dimethylethyl)-9-fluoro-3,6-dihydro-7H-benz[h]-imidaz[4,5-f]isoquinolin-7-one (DBI) Calbiochem cat. #420099, methyl-thiohydantoin tryptophan (MTH-Trp or Necrostatin 1) Biomol cat. #AP309. Stock solutions were prepared in DMSO for use in cell based assays.
Myc + Ras-transformed keratinocytes from a Bin1-deficeint mouse (Bin1−/− MR KECs) previously described (10) and B16-F10 mouse melanoma cells (ATCC, Manassas, VA USA) were cultured in DMEM (Mediatech, Manassas, VA USA) supplemented with 10% FBS (Hyclone, Logan, UT USA) and 1% penicillin-streptomycin (Mediatech, Manassas, VA USA) at 37°C and 5% CO2. U937 human monocytic cells (ATCC, Manassas, VA USA) were cultured in RPMI 1640 (Mediatech, Manassas, VA USA) supplemented with heat inactivated 10% FBS (Mediatech, Manassas, VA USA), 55 µM β-mercaptoethanol and 1% penicillin-streptomycin (Mediatech, Manassas, VA USA) at 37°C and 5% CO2. Recombinant human interferon-γ (IFNγ) from R&D Systems cat.# 285-IF was used at a final concentration of 100ng/ml and lipopolysaccharides from Escherichia coli 0111:B4 (LPS) from Sigma cat.# L2630 was used at a final concentration of 100 ng/ml.
C57BL/6 mice and athymic NCr-nu/nu (nude mice) were obtained from NCI-Frederick. Congenic, homozygous Ido1-null mice on the C57BL/6 strain background (Ido1-KO) described previously (18), were a kind gift from Dr. Andrew Mellor (Medical College of Georgia). All studies involving mice were approved by the institutional animal use committee of the Lankenau Institute for Medical Research.
Assessment of IDO activity in U937 cells by LC/MS/MS analysis of kynurenine in the medium at 24 hours postinduction was analyzed by HPLC coupled to electrospray ionization tandem mass spectroscopy (LC/MS/MS) analysis as described (19), using a Varian 320-MS triple quadrupole mass spectrometer system. Quantitation of kynurenine was based on analysis of two daughter ions.
The transient transfection procedure was adapted from (20) for electroporation of U937 cells. Specifically, 2 µg of huIDOpro1245-luc with the luciferase gene controlled by the human IDO1 promoter (−1207 to +38) adapted from (21), 1 µg CMV-β-galactosidase (to normalize for transfection efficiencies) and 2 µg pcDNA3.1. or CMV4-IκBα-SR expressing the (S32A/S36A) super-repressor mutant of IκBα (22). Total plasmid DNA in each transfection was made up to 25 µg with pUC19 carrier plasmid DNA in a total of 50µl 0.1×TE and was added to 0.3ml complete media containing 5 ×106 U937 cells. Electroporation was carried out at 960 mF and 220 volts using a Gene Pulser (Bio-Rad) using disposable electroporation cuvettes with a 0.4 cm electrode gap (Denville Scientific, South Plainfield, NJ, USA). After electroporation, cell were left to recover for 48 hours then they were stimulated with a combination of IFNγ+LPS for 24 hours and then harvested for luciferase and β-galactosidase activity as previously described (10).
Northern blot analysis of IDO was conducted using a human full length cDNA probe essentially as described (23). Western blot analysis was performed using standard methods. Antibodies to IDO1 (clone 10.1 (10)), actin (1–19; sc-1616) (Santa Cruz Biotechnology, Santa Cruz, CA, USA), were used as described (24) or as recommended by the vendor. Immunoprecipitation of Ido1 protein from mouse lung tissue with purified rabbit polyclonal antibody (25) followed by Western blot-based detection with rat monoclonal antbody (clone mIDO-48; BioLegend) was performed as described (25).
IDO was induced in 8–10 week old C57BL/6 mice by intrapulmonary delivery of lipopolysaccharides (LPS) from Escherichia coli (0111:B4) (Sigma cat. # L2630). Mice anesthetized by inhalation of isoflurane were instilled intranasally with 25 µg of LPS in 25 µl sterile saline. EP was administered by intraperitoneal injection at 40 mg/kg in 500 µl sterile modified Ringer’s ethly pyruvate solution (in which ethyl pyruvate substitutes for lactate) (26) immediately prior to LPS challenge and with repeated dosing at 6 and 24 hours post-challenge. Mice were euthanized at 24 hours post challenge by cervical dislocation and lung tissues were weighed and frozen on dry ice. For analysis of kynurenine, frozen lung tissue samples were homogenized in PBS (1:4 w/v) and subjected to three rounds of freeze thaw lyses. Deproteinated lysates were analyzed by HPLC coupled to electrospray ionization tandem mass spectroscopy (LC/MS/MS) analysis as described (19), using a Varian 320-MS triple quadrupole mass spectrometer system. Quantitation of kynurenine was based on analysis of two daughter ions.
Tumor graft experiments were carried out in 8–10 week old mice injected subcutaneously with 1 × 106 Bin1−/− MR KEC cells or 1 × 105 B16-F10 melanoma cells. Treatment with either vehicle alone or EP or 2-AA in modified Ringer’s solution (26) injected intraperitoneally at 40 mg/kg b.i.d. (twice a day) was initiated at day 7 following initial tumor cell engraftment. Tumor growth was monitored by performing caliper measurements of orthogonal diameters and the estimated tumor volume was calculated based on the formula for determining a prolapsed elliptoid (d2 × l/0.52) where d is the shorter of the two orthogonal measurements. Graphing and statistical analysis of the data was performed using using Graph Pad Prism 4 software (Graph Pad Software Inc., San Diego, CA).
Ethyl pyruvate (EP) is a stable aliphatic ester of pyruvate that has been reported to impair nuclear translocation and DNA binding by NFκB through covalent modification of the p65 RelA subunit at Cys38 (27, 28). To evaluate the effects of EP on IDO expression, we employed U937 monocytic cells as a model system in which IDO is upregulated by exposure to interferon-γ (IFN-γ) and lipopolysaccharide (LPS). Under assay conditions in which the cells are maintained in medium supplemented with low endogenous endotoxin serum, we found that LPS is required in conjunction with IFN-γ to substantially induce IDO gene expression and enzymatic activity that is otherwise marginally induced with either factor alone (Fig. 1A). At concentrations of EP that interfere with NFκB activity (27), we found that EP could suppress the level of kynurenine (the product of tryptophan catabolism by IDO), in the medium by nearly 90% after LPS+IFN-γ treatment (Fig. 1A). This degree of IDO inhibition was comparable to that achieved with the unrelated NFκB inhibitory compound, QNZ (29) as well as with the pan-JAK inhibitory compound DBI (30). In all three instances, the inhibition of IDO activity by these compounds correlated with suppression of IDO1 protein expression (Fig 1B). By way of comparison, these compounds were at least as effective at inhibiting IDO activity as the bioactive, competitive inhibitor MTH-Trp (10), which did not suppress the level of IDO1 protein expression (Fig. 1).
Titration of EP in this cell-based assay yielded an EC50 of ~2.2 mM (Supplemental Fig. S1), a value consistent with previous studies of this agent as an NFκB inhibitor (27). When EP was titrated against purified recombinant IDO1 enzyme in the same manner as in the cell-based assay, no inhibitory activity was observed (Supplemental Fig. S1). Western blot analysis of LPS+IFNγ-stimulated U937 cells demonstrated that EP acted by blocking the induction of IDO1 protein expression within the effective dose range (Supplemental Fig. S2). We confirmed the absence of cytotoxicity at the levels of EP exposure evaluated in these studies by flow cytometric analysis of propidium iodide-stained cells (Supplemental Fig. S3). A substantial increase in sub-G1 cells was observed after LPS+IFNγ stimulation, but this was not exacerbated by exposure to EP. Northern blot analysis revealed that the primary impact of EP treatment on IDO1 expression was to decrease the level of induced message (Fig. 2). Following the removal of IFNγ+LPS from the culture medium there was no demonstrable contribution of newly synthesized message to the IDO1 mRNA pool as demonstrated by treatment with actinomycin D, and the level of IDO1 mRNA declined precipitously past 4 hours irrespective of the presence or absence of EP (Fig. 2A). As opposed to the lack of evidence for destabilization of IFNγ+LPS-induced IDO1 mRNA, evaluation of the effect of EP on IFNγ+LPS-stimulated IDO1 promoter activity in a transcriptional reporter assay supported the conclusion that EP interferes with new message synthesis (Fig 2B). Expression of the mutated IκBα ‘super-repressor’ protein (31), which interferes with so-called ‘canonical’ NFκB signaling, had a comparable effect to that of EP in suppressing IDO1 promoter activity in this reporter assay (Fig 2B), consistent with the interpretation that EP blocks IDO induction through its ability to target p65 RelA which is a major component of the canonical NFκB signaling pathway (22).
Lung has long been recognized as a tissue in which high levels of IDO activity can be induced in response to exposure to bacterial lipopolysaccharides (LPS) (32, 33). Based on these data, we have developed a pharmacodynamic (PD) assay whereby IDO is induced in the lungs of mice in response to pulmonary exposure to LPS (34–37). In isolated lung tissue, the kynurenine level, which is reflective of IDO activity, was elevated by ~4-fold at 24 hours following LPS administration (Fig. 3). Mice with a homozygous disruption of the Ido1 gene displayed a lower baseline level of kynurenine in the lungs than did wild type mice and also exhibited no significant elevation in kynurenine in response to LPS exposure, demonstrating the specificity of this assay for assessing Ido1 enzyme activity (Fig. 3A). In this PD assay, administration of EP at a dose level previously identified as sufficient to interfere with in vivo NFκB activity (34–37) as well as produce a positive survival outcome in a mouse sepsis model (26), suppressed kynurenine elevation by 83% to less than 1.5-fold above baseline (Fig. 3A). Western blot analysis confirmed the expectation that the reduction in kynurenine elevation by EP correlated with the suppression of Ido1 protein induction in lung tissue from EP-treated animals (Fig. 3B). Taken together, our results indicate that, at concentrations where it has been shown to mediate anti-inflammatory effects, EP inhibits IDO expression in vitro and in vivo.
Compounds that directly interfere with IDO-mediated tryptophan catalysis have demonstrated antitumor activity. The ability of EP to block IDO activity by suppressing its expression suggested that it might be capable of exerting similar antitumor effects. We have tested this prediction in two tumor models, Myc+Ras-transformed keratinocytes from Bin1-deficient mice (Bin1−/− MR KECs) and B16-F10 melanoma cells, both of which have previously been used to examine the antitumor activity IDO inhibitors (7, 10, 14, 38).
Our earlier work had identified IDO as a gene that is dysregulated as a result of homozygous deletion of the Bin1 tumor suppressor gene in MR KEC cells and had shown that an IDO-dependent immune escape mechanism renders these Bin1−/− MR KECs more aggressively tumorigenic in syngeneic, immunocompetent animals than their wild type counterparts (10). The Bin1−/− MR KEC cell line also serves as a model in which IDO induction is NKkB-dependent (10). Ectopic expression of the IκBα super-repressor in Bin1−/− MR KECs effectively suppressed the ability of these cells to form tumors (Fig. 4A), confirming the requirement for NFκB signaling to support IDO-dependent outgrowth in this tumor model. EP treatment likewise suppressed tumor growth when administered at the same dose level that blocked IDO activity and expression in the PD assay (Fig. 4B). Thus, as predicted, EP treatment can suppress the outgrowth of tumors in this model and this correlates with its ability to inhibit the NFκB-dependent expression of IDO.
Because the Bin1−/− MR KECs represent a rather specialized tumor model, we extended these studies to evaluate EP treatment in the widely used B16-F10 melanoma isograft tumor model as well. Unlike the Bin1−/− MR KECs, IDO is not expressed detectably in B16-F10 tumor cells themselves but rather has been found to be expressed in antigen-presenting cells within the tumor-draining lymph nodes of the host animal (39). B16-F10 cells form highly aggressive, poorly immunogenenic tumors that have been shown to be resistant to a variety of immunotherapeutic strategies, however, direct inhibitors of the IDO enzyme can elicit robust single agent responses in this model (14, 38). Comparing final mean tumor volumes between EP-treated and control animals at the ~4 week endpoint, we found that EP treatment of B16-F10 challenged mice caused significant suppression of tumor growth (Fig. 5A) which equates to a %T/C of 5.9% (a T/C ratio of < 42% is indicative of efficacy according to standard NCI criteria (40)). This finding corroborates our observations made in the Bin1−/− MR KEC model regarding the antitumor activity of EP.
To independently test the proposed biological basis for EP antitumor activity, we examined the effects of a functional analog, 2-acetamidoacrylate (2-AA), that has also been reported to inhibit nuclear translocation and DNA binding by p65 RelA (28). Although appearing to be structurally dissimilar, 2-AA has been postulated to actually mimic the enol tautomer of EP which may represent the biologically active form (28). In support of EP and 2-AA having a shared mechanism of action, we found that 2-AA also inhibited IDO expression in U937 cells (Fig. 1A). Furthermore, when administered to B16-F10 tumor-bearing mice, 2-AA suppressed tumor outgrowth as effectively as EP (Fig. 5B). Thus, the key predictions that a distinct compound sharing the NFκB-dependent mechanism of action of EP would both inhibit the induction of IDO activity and suppress tumor growth were effectively borne out.
As a therapeutic class, IDO enzyme inhibitors require intact T cell function to suppress tumor outgrowth in mice (7, 10, 14, 16, 38). Therefore, if the biological consequences of EP treatment are primarily mediated through its ability to block IDO induction, it should exhibit similar requirements as well. To evaluate the importance of T cell-dependent immuntiy to the antitumor activity of EP, athymic ‘nude’ mice that are deficient in mature T cells were challenged with B16-F10 tumors. In the context of these mice, EP treatment had no discernible impact on tumor outgrowth and tumor growth rate in both EP untreated as well as treated mice was somewhat accelerated relative to tumor growth rate in wild type mice (Fig. 5C). Because IDO is expressed only in normal host cells in the B16-F10 model and not the tumor cells themselves, Ido11-KO mice were employed to genetically evaluate the direct relevance of Ido1 blockade to the mechanism of action of EP. As predicted, EP treatment was ineffective at suppressing the outgrowth of B16-F10 tumors in Ido1-KO mice (Fig. 5D), although its antitumor efficacy appears not to have been completely abolished as was previously observed with direct IDO inhibitors (14, 38). These findings support the concept that EP exerts its antitumor effects in a T cell-dependent manner primarily mediated through its ability to block the induction of IDO.
The finding that EP elicits IDO-directed, immune-based antitumor responses reinforces emerging concepts about the important role of inflammatory processes in supporting cancer pathophysiology. Prompted by observations of Fink and colleagues in rodent models of ischemia/reperfusion injury, hemorrhagic shock and sepsis (26, 35, 41), EP treatment has been investigated for several years in a variety of animal models of acute and chronic inflammatory disorders (17), while clinical testing thus far has been limited to the evaluation of EP as an intravenous agent for the prevention of single and multiple organ dysfunction in patients undergoing cardiopulmonary bypass surgery (42). Early studies of IDO have noted its elevation in response to bacterial infections or LPS exposure (32, 43), and more recently it has been reported to be a mediator of endotoxic shock-associated lethality (44). It is intriguing to speculate that diseases as seemingly disparate as cancer and sepsis may be linked at some underlying level of pathophysiology through a shared dysregulation of IDO. In this light, the interaction of cancer cells with the host may bear some similarity to unresolved infections that result in the clinical manifestations of sepsis. Survival following the onset of severe sepsis as modeled in the mouse can be dramatically improved by the administration of EP (26), and it will be important to assess whether this benefit is also linked to the ability of EP to inhibit IDO.
Many studies have suggested that correcting imbalances in NFκB signaling in cancer may have important benefits in the context of both the tumor cell and the inflammatory tumor microenvironment, but the concept of correcting immune escape via this signaling pathway has received relatively little attention. Given the centrality of NFκB as a signal transduction node, the degree to which EP antitumor activity was found to rely specifically upon IDO targeting in the host might be considered somewhat surprising. However, this outcome aligns with a concept we have termed ‘tolerance addiction’, proposed as a result of previous studies of IDO inhibitory compounds (14, 38). As noted previously, B16-F10 melanomas are illustrative of a class of tumors that appear to preferentially use IDO as an immune escape mechanism, such that, once a tumor is established, continued IDO activity must be sustained to maintain the immunoprivileged state of the tumor. In this context, acute disruption of IDO activity, as with a pharmacological agent, causes an immunological unmasking that promotes rejection. However, if upregulation of IDO activity is not an available option (as in the IDO-deficient animal), alternate immune escape mechanisms can apparently be accessed by the developing tumor in which case IDO-targeting compounds are ineffectual. The specific target of EP, p65 RelA, is an important component of canonical NFκB signaling but plays no apparent role in the non-canonical pathway (22). This specifically implicates canonical NFκB signaling as the regulatory pathway controlling IDO expression, a conclusion that is further supported by the demonstration that the NFκBα (IkBα) super-repressor, which also selectively interferes with canonical NFκB signaling, also effectively suppresses induction of IDO promoter activity. These results appear to run counter to evidence that non-canonical NFκB signaling is important for IDO induction mediated through GITR signaling (45). Our data do not, however, necessarily contradict these findings, but rather indicate that the regulation of IDO expression is likely to be complex and that the relative importance of canonical versus non-canonical NFκB signaling in controlling this process may be contextual.
Although our findings are consistent with published evidence of EP as an NFκB inhibitor, they do not rule out alternative mechanisms that may be germane to its in vivo effects. EP has also been reported to exert anti-inflammatory effects through ROS scavenging (46) and through blocking HMGB1 release (26), and it is not inconceivable that elevated ROS or HMGB1 release may support dysregulated expression of IDO. One group has recently reported that EP can exert antitumor activity in a liver metastasis model and has suggested that EP may produce anti-inflammatory and pro-apoptotic effects responsible for its antitumor activity though mechanistic validation was lacking (47). From our studies it is clear that direct cytotoxicity is not sufficient to account for the antitumor efficacy of EP against B16-F10 tumors in vivo, as no evidence of antitumor activity was observed when this compound was administered to athymic, tumor-bearing mice. Furthermore, it is clear from the loss of EP efficacy in Ido1-deficient mice that the relevant regulatory pathway targeted by EP directs an immune escape mechanism that is predominantly orchestrated through the elevation of IDO activity.
While direct inhibition of the IDO enzyme is presently being explored by many groups as an interventional approach, EP may offer an alternative, low-cost, readily accessible tool to indirectly block IDO for therapeutic purposes. It is likely that IDO inhibitors will prove most effective when combined with other cancer treatment modalities (7, 10, 11), and EP, as a safe and inexpensive food additive, could readily be evaluated as an adjuvant to standard of care treatments with minimal risk of adverse side effects. Given accumulating evidence that elevated IDO activity may have a pathophysiological role in other diseases such as chronic infections and autoimmune disorders (25, 48), EP may find other clinical applications as an IDO inhibitory strategy as well. Fink and colleagues have described a simple formulation to administer EP (49) by substituting it for lactate in Ringer’s lactate solution that is usually given intravenously for fluid resuscitation after blood loss or as a conduit for drug delivery. Insofar as IDO inhibitors have been shown to cooperate with different types of cancer therapy in mouse tumor models, we suggest the same Ringer’s formulation as a route to administer EP with standard i.v. chemotherapeutics. Repositioning EP for an oncology study in this manner would be a straightforward strategy to clinically evaluate EP’s potential as a cutting edge immunochemotherapeutic agent that could address the acute need in developing countries for simple, low cost advances in cancer treatment.
We thank Erika Sutanto-Ward for excellent technical assistance and Dr. Laura Mandik-Nayak and Dr. Lisa Laury-Kleintop for helpful discussions. This work is supported by the Lankenau Hospital Foundation (to A.J.M. and G.C.P.). A.J.M. is the recipient of a Lance Armstrong Foundation Grant. G.C.P. is the recipient of National Institutes of Health Grants CA109542, CA82222, and CA100123. Additional support is thankfully acknowledged from the Dan Green Family Foundation, Lankenau Hospital Foundation, and Main Line Health System.