Role Of Dioxygenase Enzymes in Overcoming The Blockage of Kynurenine Pathway
IDO2 and TDO2 are other enzymes with roles in the KYN metabolism pathway. IDO2 is a very poor producer of KYN and has been suggested to function differently from IDO1 [27; 28]. TDO2 is generally known to only be expressed in the liver, kidney, brain, and with very low lung expression in normal tissue [29]. Consistent with these reports, we detected very low to non-detectable IDO2 and TDO2 proteins expressed in our CS and CR lung cancer cells as shown in Fig. 2A and Supplemental Figure S2A. However, IDO2 and TDO2 may be part of a compensatory/mutualistic mechanism induced when IDO1 is inhibited, providing another potential reason for the lack of efficacy presented in the Phase 3 trial (ECHO-301) of epacadostat in combination with pembrolizumab for the treatment of melanoma [30]. As evidenced in Fig. 2A and S2A, knocking down IDO1 led to a marked increase in TDO2 expression in CR cells. To determine whether inhibition of IDO1 may also induce IDO2 and TDO2 expression in vivo, C57BL/6 mice were inoculated with either Lewis lung carcinoma mouse cells (LLC) or its platinum-resistant counterpart (LLC-CR) cells. It is noteworthy that LLC-CR cells exhibited higher basal IDO1 activity compared to LLC cells, which are sensitive to the cisplatin [7]. Here, mice were treated with IDOi at 200 mg/kg P.O. once a day for 15 days or with methylcellulose (control group (CTRL)). At harvest, tumor tissues were collected and assayed for IDO1/TDO2 and IDO2 protein expressions. IDO1 expression was elevated in mouse allografts with LLC-CR, along with very low to non-detectable TDO2 expression in the control group, as anticipated. However, in LLC-CR mice treated with the selective IDO1 inhibitor, we observed a significant increase in TDO2 expression (Figure. 3A and right panel). We did not observe changes in IDO2 expression during IDO1 inhibitions (Supplemental Figure S3A). We then created CRISPR-edited Ido1 knockout in LLC-CR (LLC-CRSG) cells. The LLC-CRSG2 clone was selected due to the complete knockout efficiency for IDO1 (Supplemental Figure S3B). Again, we detected significantly enhanced TDO2 expression but did not observe significant changes in IDO2 expression in IDO1 knockout conditions (Fig. 3B). To further identify a potential compensatory mechanism in CR cells and determine if TDO2 inhibition may lead to an increase in IDO1 expression in vivo, we administered a selective TDO2 inhibitor (LM10) instead of IDOi under the same conditions (200 mg/kg P.O. once a day for 15 days). Blocking TDO2 led to an increase in IDO1 expression and activity (higher KYN production) (Supplementary Figure S3C&D) with an increase in IDO2 expression in LLC-CR (Supplementary Figure S3C).
Nevertheless, the question remained whether compensatory increases in TDO2 or IDO2 have functional consequences via higher KYN production. We then assayed for cellular KYN in human cells with IDO1 knockdown (shIDO) and IDO1 mouse-knockout (LLC-CRSG2) cells where TDO2 was inhibited (LM10). Extracellular KYN concentrations were significantly suppressed to levels lower than in CS cells with significantly increased TRP concentrations, suggesting that IDO2 may not play an important role in affecting KYN secretion in our CR cell models (Fig. 3C (pink) and Supplemental Figure S4A). To further investigate the observed dual role of IDO1 and TDO2 enzymes in our CR models, we first examined IDO1 and TDO2 enzyme inhibition in cell culture by means of the novel IDO1/TDO2 dual inhibitor AT-0174. In an enzyme inhibition colorimetry assay to determine KYN activity, murine LLC cells transfected with human-IDO1 (LLC-hIDO1) revealed an inhibitory potency (IC50) at IDO1 of 0.17 µM. The IC50 of the TDO2 enzyme inhibition in murine glioma cells (GL261) transfected with human-TDO2 (GL261-hTDO2) was 0.25 µM (Fig. 3D). Blocking IDO1/TDO2 inhibited the colony-forming ability only in CR cell models further demonstrating the dependence of CR cancer cells on these enzymes for survival (Figs. 3E and Supplementary Figure S4B). In all, our data support the mutualistic role of IDO1/TDO2 in CR tumors and the AT-0174 has a strong equipotency at both enzymes.
Decreased Tumor Burden Upon IDO1 or Dual IDO1/TDO2 Inhibition in Syngeneic and Humanized Murine Models of Platinum-Resistant NSCLC
To determine whether our in vitro findings on cancer cell growth and survival can be translated in vivo, we employed humanized mice (NSG-hu CD34+), modeling a human immune system, to determine whether blocking the KYN pathway can suppress CR tumor growth. Human CS cell line “A” or CR cell line “ALC” were subcutaneously implanted into the right flank of humanized NSG-hu CD34+ mice and orally treated with AT-0174 (170 mg/kg, 1/day) or with IDOi (200mg/kg, 1/day) for 15 days. IDOi suppressed tumor growth in ALC allografts (Fig. 4A-green line; right panel) with a significant reduction in total tumor weight (Fig. 4B; right panel). However, when comparing IDO1 inhibition to dual inhibition (red line), it was evidenced that dual inhibition was more potent than selective IDO1 inhibition alone in reducing tumor growth and tumor weight of CR tumors (Fig. 4A & B-red).
An increase in TDO2 expression was observed in LLC-CR inoculated mice treated with either IDOi or with AT-0174 is shown in Fig. 3A (right panel). To further examine the mechanistic role of IDO1 and TDO2 in CR tumors, we implemented a syngeneic mouse model. Mice bearing LLC or LLC-CR tumors were treated with AT-0174 or IDOi using the same regimen as the humanized model described above. Again, selective IDO1 inhibition suppressed tumor growth in CR allografts (Fig. 4C-green line; right panel), with a significant reduction in total tumor weight (Fig. 4D; right panel) when compared with LLC mouse tumors. However, blocking both IDO1 and TDO2 with the dual inhibitor AT-0174 induced a greater reduction in tumor size (Fig. 4C-blue), and tumor weight in LLC-CR allograft mice (Fig. 4D). Moreover, tumor formation of CRISPR Ido1-KO mice (LLC-CRSG2) was significantly slower than control LLC-CR, and treatment with AT-0174 further suppressed LLC-CRSG2 tumor growth and weight (Fig. 4E). Together, our data strongly support the presence of a compensatory mechanism wherein cisplatin-resistant NSCLC tumors can employ IDO1 and/or TDO2 activation to overcome single enzyme pharmacological blockade as therapy.
Assessment Of Tumor Immune Cell Profiles And KYN/TRP Levels After IDO1 Or Dual IDO1/TDO2 Inhibition In Syngeneic And Humanized Mouse Models Of Platinum-Resistant NSCLC
As mentioned, anti-tumor immunity is reliant on checks and balances between immune effector cells and immunosuppressive cells. To determine whether inhibiting the KYN pathway can restore anti-tumor immunity in CR tumor models, we analyzed tumor-infiltrating lymphocytes (TIL) using flow cytometry (Supplemental Figure S1B for gating strategy in mice). When compared to mice-bearing CS tumors, CR tumors in both syngeneic (LLC-CR) and humanized (ALC) models showed an increase in TIL-Treg and tumor-infiltrating MDSC populations further confirming the immunosuppressive role of KYN in CR tumors and suggesting that CR tumors may have adapted to evade immune surveillance (Fig. 5A & C; black dots, right panels). Treg population in CR tumors was significantly reduced by IDO1 inhibition (Fig. 5A & C; green dots, right panel), and very significantly reduced with dual IDO1/TDO2 inhibition in both CR mouse models (Fig. 5A & C; blue & red dots, right panel), leading to increased NKG2D expression on NK and CD8+ T populations (Fig. 5A & C; blue & red dots, left panels). It is also important to recognize that IDO1 inhibition but not dual inhibition in CS tumors (LLC and A) resulted in a slightly increased Treg and MDSC population (Fig. 5A & C, green dots, right panel). These data were consistent with a minor increase in TDO2 protein expression shown in Fig. 3A.
Our data further confirm that KYN not only plays a critical role in the reprogramming of naïve T cells to Tregs but also in impairing NK and T-effector cells’ ability to mount anti-tumor immune responses by modulating the NKG2D receptor in vivo. Serum KYN concentrations were significantly decreased in CR mice after IDO1 inhibition (Fig. 5B & D green bars, left panel), and more intensified after dual inhibition (Fig. 5B & D blue & red bars, left panel). Concomitantly, higher serum TRP levels were found in mice treated with a dual inhibitor when compared to the IDO1 inhibitor alone, confirming a sustained inhibition of TRP catabolism by TDO2 in vivo (Fig. 5B & D; blue & red bars, right panel).
Effects Of IDO1/TDO2 Inhibition In Combination With PD1 Blockage On CR Tumor Growth And Immune Cell Profile
Previous studies reported higher expression of PD-L1 (program death receptor ligand-1) in many solid tumors, including NSCLC, upon treatment with platinum chemotherapeutic agents [24; 25; 31]. To ascertain that PD-L1 protein expression is increased in CR compared to CS cells, we analyzed baseline PD-L1 protein expression in our model. Increased PD-L1 expression (Fig. 6A & B) was observed in all of our CR cell models (human and mouse), thus providing a rationale for combining the use of dual IDO1/TDO2 inhibition with PD1 blockade (anti-PD1 antibody) which we subsequently examined in the LLC-CR syngeneic mouse model. Intraperitoneal injection of anti-PD1 antibody (10 mg/kg, every 3 days) in combination with AT-0174 resulted in greater suppression of CR tumor growth than PD1 blockage alone, or AT-0174 treatment alone (Fig. 6C). Significantly lower tumor weights were found in all of the treatment groups, with IDO1/TDO2 inhibition plus PD1 blockade yielding the greatest reductions (Fig. 6C; right panel). This treatment combination did not affect tumor growth and weight in a sensitive LLC mouse model, demonstrating the applicability of dual inhibition + anti-PD1 blockade in CR tumors (Supplemental Figure S5).
CR tumor-bearing mice with combination treatment resulted in significantly decreased KYN in serum and significantly elevated TRP (Fig. 6D; orange). Correlating with these changes, combination treatment also significantly increased the NKG2D frequency of NK+ cells and CD8+ T cells. Immunosuppressive cell populations (Tregs and MDSCs) were highest in high KYN and low TRP serum conditions, and decreased in all treatment groups, compared to no treatment, with the most significant decrease seen with AT-0174 + PD1 blockade (Fig. 6E; orange). Next, we analyzed the survival benefit of combination treatment on syngeneic mice bearing CR tumors. Comparing dual IDO1/TDO2 inhibition (AT-0174) with IDO1 inhibition when each was given alone, we observed that dual inhibition increased median survival (36 days) more than selective IDO1 inhibition (32 days) (Fig. 6F). As depicted in Fig. 6F, the median survival of mice treated with AT-0174 or anti-PD1 was 36 and 35 days, respectively; which increased to 50 days with the combination AT-0174 + PD1 antibody treatment.
Collectively, our results show that the AT-0174 increases the antitumor effect of PD1 blockade on cisplatin-resistant, PD-L1-expressing, NSCLC in vivo. These data suggest platinum-resistant tumors have an IDO1- and TDO2-dependent metabolic impact (higher KYN/lower TRP) on the tumor microenvironment by promoting immunosuppression (higher Treg and MDSC frequencies, lower % of NKG2D NK+ and CD8+ cells).