Development of a novel lactate dehydrogenase A inhibitor with potent antitumor activity and immune activation

Abstract Lactate accumulation in the tumor microenvironment was shown to be closely related to tumor growth and immune escape, and suppression of lactate production by inhibiting lactate dehydrogenase A (LDHA) has been pursued as a potential novel antitumor strategy. However, only a few potent LDHA inhibitors have been developed and most of them did not show potent antitumor effects in vivo. To this end, we designed new LDHA inhibitors and obtained a novel potent LDHA inhibitor, ML‐05. ML‐05 inhibited cellular lactate production and tumor cell proliferation, which was associated with inhibition of ATP production and induction of reactive oxygen species and G1 phase arrest. In a mouse B16F10 melanoma model, intratumoral injection of ML‐05 significantly reduced lactate production, inhibited tumor growth, and released antitumor immune response of T cell subsets (Th1 and GMZB+CD8 T cells) in the tumor microenvironment. Moreover, ML‐05 treatment combined with programmed cell death‐1 Ab or stimulator of interferon genes protein (STING) could sensitize the antitumor activity in B16F10 melanoma model. Collectively, we developed a novel potent LDHA inhibitor, ML‐05, that elicited profound antitumor activity when injected locally, and was associated with the activation of antitumor immunity. In addition, ML‐05 could sensitize immunotherapies, which suggests great translational value.


| INTRODUC TI ON
Metabolic reprogramming of tumors creates spatial and temporal heterogeneity in the TME, which profoundly affects the infiltration and phenotype of immune cells during tumor progression. 1,2 Among the different types of metabolic remodeling, the first recognized feature of metabolic reprogramming is aerobic glycolysis in cancer. Tumor cells prefer to use glycolysis and generate large amounts of lactate, which is secreted outside of cells even under adequate oxygen supply, well-known as the Warburg effect. 3 Lactate is the product of glycolysis, which promotes cell proliferation and neovascularization. 4 Importantly, high levels of lactate lead to acidosis of the TME and induce immunosuppression through multiple mechanisms, including the inhibition of monocyte differentiation to dendritic cells and dendritic cell activation, 5 recruitment of myeloidderived suppressor cell-like immunosuppressive cell infiltration, 6,7 and the inhibition of T cells and natural killer cells. 8,9 Therefore, inhibition of lactate production in the TME has been pursued as a novel antitumor strategy or as an adjuvant for immunotherapy.
Lactate dehydrogenase A catalyzes the production of lactate in glycolysis, 10 which is overexpressed in most types of cancers, 11,12 and its depletion led to the inhibition of tumor growth in multiple tumor models. [13][14][15] Moreover, the absence of LDHA in bone marrow cells triggers antitumor immunity, and increased glycolysis in tumor are characteristics of immune system resistance to adaptive T cell therapies, 16,17 suggesting the important impact of LDHA on the TME and immunotherapy. Therefore, inhibition of LDHA activity in tumors could improve the immune microenvironment and initiate antitumor immunity.
To date, several LDHA inhibitors have been developed. Oxamate is a pyruvate analogue that was developed as the first LDHA inhibitor; however, it has a very low effect on LDHA activity with IC 50 of approximately 800 μM. 18 Since 2010, compounds such as FX-11 19 ( Figure 1), N-hydroxyindole-2-carboxylates, 20 dicarboxylic acids, 21 2-thio-6-oxo-1,6-dihydropyrimidines, 22 2-amino-5-aryl-pyrazines, 23 and 3-hydroxy-2-mercaptocyclohex-2-enones 24 have been reported as LDHA inhibitors with IC 50 at micromolar or submicromolar levels. Recently, more potent LDHA inhibitors have been developed with IC 50 at nanomolar levels, such as quinoline 3-sulfonamides (GSK2837808A), 25 GNE-140, 26 and pyrazole-based compound 63 27 ( Figure 1). However, most LDHA inhibitors did not show potent in vivo antitumor effects, which could relate to their inhibitory activity, metabolic stability, or route of administration. Furthermore, it remains unclear whether and how LDHA inhibitors improve the immunosuppressive TME to release antitumor immunity. To this end, in the current study we designed novel LDHA inhibitors and tested their antitumor activity both in vitro and in vivo, as well as their influence on the tumor immune microenvironment.

| Lactate measurement
Tumor cells were plated into 96-well plates (1.5 × 10 4 /well) and then treated with serial doses of compound or 1% DMSO as a vehicle control in serum-free medium. After 6 h, the cell supernatants were measured by lactate assay according to the instruction kit (A019-2; Nanjing Jiancheng). The EC 50 was calculated by GraphPad software as the concentration-inhibition rate concentration curve. Kit (#103020-100) was uesd to detect extracellular acidification rate.

| Extracellular acidification rate assay
Prehydrated probe plates were replaced with XF Calibrant (pH 7.4, #100840-000) and incubated together with cell plates. After setting the program, the probe plates were placed in the instrument for 20 min to equilibrate, and then the cell plates were placed and assayed in the XF-96 analyzer (Seahorse Bioscience) using Wave software. The measurement of the extracellular acidification rate was expressed as mpH/min.

| Statistical analysis
All data were expressed as mean ± SEM, and unpaired Student's ttest, or one-way or two-way ANOVA followed by Tukey's multiple comparisons test as indicated. The statistical analyses were undertaken using GraphPad Prism (version 8.0.1) software. Other methods are shown in the Appendix S1.

| Design of novel LDHA inhibitor ML-05
GNE140 is a potent LDHA inhibitor possessing good molecular (LDHA IC 50 = 3 nM) and cellular activity; however, it did not show antitumor activity in the MIA PaCa-2 xenograft model even at the dose of 400 mg/kg. 26 Genentech previously reported another series of diketone compounds as LDHA inhibitors. 24 Among them, compound 31 was optimized as the most potent analogue (LDHA IC 50 = 6 nM), however, this compound was found unstable in plasma due to its ester bond. To increase the in vivo stability, ML-01, an amide analogue of compound 31, was prepared and tested. However, it displayed a significant decrease of activity as compared to its parent compound 31 ( Figure 2A). We envisioned that replacing the phenyl ring with various aromatic groups might increase the π-cation

| ML-05 inhibited lactate production and glycolysis in cultured cells
Human-derived cancer cells were used to test the inhibitory activity of ML-05 on lactate production. ML-05 dose-dependently inhibited the production of lactate in A375, CAL51, and MIA PACA2 cells with comparable potency with GNE140 after 6 h of treatment ( Figure 3A-C). Furthermore, ML-05 was tested on the inhibition of lactate production in murine cancer cells. Similarly, ML-05 dosedependently inhibited the lactate production in B16F10, EMT6, and Panc02 cells with comparable potency with GNE140 after 6 h of treatment ( Figure 3D-F). These data indicated that ML-05 could reduce lactate production in human-derived cells and murine-derived cells with comparable potency.
Melanoma is a highly glycolytic tumor, in which lactate induces an immunosuppressive TME and correlates with metastasis. 28,29 Inhibition of LDHA activity exerts antitumor effects by promoting T cell infiltration. 17 Figure 4F). Concordantly, the association of cyclin A and CDK2 is required for G 1 /S transition 31 and thus their decrement by ML-05 treatment was associated with G 1 phase arrest. Collectively, these results indicated that ML-05 has a pan-inhibitory effect on cancer cell proliferation, which was associated with decrease in ATP production and increase in ROS production and G 1 phase arrest.

| ML-05 inhibited growth of B16F10 melanoma in immunocompetent mice
ML-05 did not inhibit the growth of B16F10 allograft tumor in immunocompetent C57BL/6 mice by oral administration or i.p. injection, similar to GNE140 ( Figure S1A-D). Considering the intratumoral inhibition of lactate production might alleviate local immunosuppressive TME, we tested the antitumor effect of ML-05 by intratumoral injection. Interestingly, ML-05 could significantly inhibit the growth of B16F10 allograft tumor with a TGI rate of 60% and reduce the tumor weight, whereas GNE140 only showed modest antitumor activity without statistical difference on tumor growth or tumor weight ( Figure 5A,B), and neither impacted on the body weight of mice ( Figure 5C). The reduction of lactate production in tumor tissues was confirmed by measuring the lactate level after 6 h of ML-05 treatment before mice were killed ( Figure S1E). To further confirm whether the antitumor effect of ML-05 is through the inhibition of LDHA activity, we used LDHA siRNA by intratumoral injection in the B16F10 allograft tumor model, to observe whether local knockdown of LDHA expression could exert antitumor effects. Consistently, local knockdown of LDHA exerted significant tumor inhibition with a TGI rate of 40% ( Figure 4D), which was less than the TGI rate of ML-05, suggesting that ML-05 inhibited the tumor growth mainly through the inhibition of LDHA.
We next asked whether the antitumor effect of ML-05 acted through the improvement of TME, as ML-05 was also able to directly inhibit cancer cell proliferation. Thus, we used nude mice implanted with B16F10 tumor cells, which are deficient in T cells.
Interestingly, the antitumor effect of ML-05 was largely diminished in nude mice, and GNE140 also did not show significant inhibition of tumor growth or tumor weight in nude mice, nor on the body weight of mice ( Figure 5E

| ML-05 increased Th1 and cytotoxic CD8T cells in TME
Given the antitumor effect of ML-05 was related to T cells and adaptive immunity, we further investigated how ML-05 impacts on tumor infiltrating immune cell subsets, which could account for the antitumor effect. The T cell subsets were gated, as shown in Figure 6A,B, and we found ML-05 treatment did not induce significant changes in CD3 + T, CD4 + T, or CD8 + T cells ( Figure 6C-E). In a further analysis of the CD4 + T subsets, we found ML-05 but not GNE140 induced a significant increase in Th1 cells ( Figure 6F), but neither had a significant influence on Th17 or Treg cells ( Figure 6G,H). For the CD8 + T cell subsets, we found the levels of GMZB + CD8 T cells were dramatically increased by ML-05 but not GNE140 treatment ( Figure 6I), with no significant change for IFNγ + CD8 + T cells ( Figure 6J). These results suggested the increase of Th1 and GMZB + CD8 T cells in the TME under ML-05 treatment could play a critical role in mediating its antitumor effect. In contrast, GNE140 did not show any impact on Th1 or GMZB + CD8 T cells in the TME, which could explain its marginal antitumor effect in vivo.

| ML-05 enhanced antitumor effects of PD-1 Ab and STING agonist
As ML-05 could increase Th1 and GMZB + CD8T cells in B16F10 TME, we next determined whether it could enhance the efficacy of immunotherapy. Programmed cell death-1 is a well-known immune checkpoint, and PD-1 Ab has achieved great success in clinic patients bearing with "hot tumors". However, a large proportion of "cold tumors" are refractory to anti-PD-1 therapy, and some studies have reported that the excessive lactate could induce an immunosuppressive TME. 33,34 Therefore, we tested whether ML-05 and PD-1 Ab have a combined antitumor effect. In the B16F10 tumor model, PD-1 Ab was given i.p. twice a week, and ML-05 was given i.t. daily. As shown in Figure 7A, PD-1 Ab did not show significant inhibition of B16F10 tumor growth or weight, whereas the combination of ML-05 and PD-1 Ab exerted a much more dramatic inhibition on tumor growth and weight than monotherapy ( Figure 7B), without impacting on body weight ( Figure 7C).
Innate immunity is essential for immune surveillance and activating the innate immune signaling pathways is now emerging as an important strategy in cancer immunotherapy. Among them, STING agonist is a hot topic based on its potent activation on antitumor immune response. To explore the combined effect of ML-05 with STING agonist, we used a STING agonist (diABZI) developed by GSK. 35 In the B16F10 tumor model, diAZBI was given i.p. and ML-05 was given i.t. We found a slightly higher inhibition of tumor growth in the combined group compared to monotherapy ( Figure 7D). Tumor weight was much lower in the combined treatment group compared monotherapy ( Figure 7E), without impacting on body weight ( Figure 7F).

| DISCUSS ION
Glycolysis produces large amounts of lactic acid, which exacerbates the immunosuppressive TME and promotes tumor progression.
Lactate dehydrogenase A is the main catalyzing enzyme for lactate production, and there were positive correlations between LDHA and tumor progression. [36][37][38] Thus, inhibiting LDHA activity and lactate production is expected to promote the antitumor immune response or sensitize immunotherapy. GNE140 shows the most potent activity on LDHA inhibition, but did not show potent antitumor activity in vivo. 26 Similar to GNE140, ML-05 did not dramatically inhibit B16F10 tumor growth in vivo by systemic administration. Therefore, we hypothesized that targeting lactate production in local tumor tissue might be suitable to reverse the immunosuppressive TME. Intriguingly, this concept was confirmed by the intratumoral injection of ML-05, which showed a dramatic inhibition on B16F10 tumor growth in vivo. These results suggested that local treatment with an LDHA inhibitor could be an appropriate antitumor approach for future development.
Although we found the antiproliferative effect of ML-05 in cultured cancer cells, this effect was not sufficient to produce pronounced tumor inhibition in vivo, as shown in nude mice. Notably, the antitumor effect was observed in immunocompetent mice, and GMZB + CD8 T cells, which are two important antitumor immune subsets. 39 In contrast, GNE140 did not show significant antitumor activity nor any effect on T cell infiltration in tumors, which might relate to its metabolic or other issues; the exact mechanism needs to be further explored. Notably, highly glycolytic tumors could dampen T cell antitumor activity by limiting glucose use and promoting lactate accumulation. 9,17,40,41,42 Lactate dehydrogenase A could affect the immune response of T cells through the PI3K-Akt-Foxo1 pathway, 43 and high lactate levels inhibited cytokine production and impaired the lytic function of cytotoxic T cells. 44 Our results substantiated that the inhibition of LDHA could restore the antitumor immunity of Th1 and GMZB + CD8 T cells in melanoma, although it did not have a significant impact on total T cells, CD4 + T cells, or CD8 + T cells. Lactate could polarize macrophages to an M2-like state but inhibit the polarization to an M1-like state that favors tumor growth. 45 However, there were no significant changes in the frequencies of M1-like or M2-like macrophages, which might relate to their dynamic changes in the TME under ML-05 treatment; however, we might not have captured the changes in macrophages at the right time. Thus, we tested the influence of ML-05 on macrophages in primary macrophages, and found that ML-05 could reverse the polarization of M2 and M1 induced by lactate. Therefore, the reactivation of macrophage function by ML-05 might also contribute to antitumor activity, which could further promote the activation of Th1 and GMZB + CD8 T cells. Given the complex and dynamic changes of immune subsets in the TME under ML-05 treatment, further study is required to fully determine the impact of ML-05 on immune activation. Furthermore, we asked whether LDHA inhibitor could sensitize immunotherapy. As is well-known, B16F10 is a "cold tumor" that is resistant to PD-1/L1 therapy, thus we tested whether ML-05 could sensitize the antitumor effect of PD-1 Ab in the B16F10 tumor model. Interestingly, the combined use of ML-05 and PD-1 Ab produced much higher inhibition of B16F10 tumor growth, suggesting the ML-05 could turn "cold tumors" to "hot tumors" and has the potential to expand the patient population that benefits from immune checkpoint therapy. This finding is concordant with the knockout of LDHA in mice with enhanced response to CTLA-4 Ab. 47 In addition, activating the innate immunity is emerging as a novel antitumor strategy; among them, STING stands out as a promising new target. [48][49][50] Herein, we tested the combination of ML-05 with a STING agonist, diABZI. Although there was only a slight increase in the inhibition of tumor growth compared to monotherapy, the tumor weight of the combined group decreased remarkably, suggesting the tumor tissues became more "inflamed" with combined therapy. These results indicated the LDHA inhibitor could sensitize both innate and adaptive immunotherapy.
In summary, we developed a novel LDHA inhibitor, ML-05, which showed potent inhibition on lactate production, ROS production, and cell cycle arrest. ML-05 could release antitumor immunity through increasing the function of Th1 and GMZB + CD8 + T cells in vivo and potentially macrophages, suggesting it has the potential to turn "cold tumors" into "hot tumors". Importantly, inhibition of LDHA in the local TME could sensitize immunotherapies, like PD-1 Ab and STING agonist, which might expand the number of patients that would benefit from their clinical application; inhibition of LDHA demonstrated great translational value. These findings underscore that targeting LDHA in the local TME is a feasible antitumor strategy, especially for glycolytic-dependent tumors, and provide a basis for development of LDHA inhibitors in the future.

ACK N OWLED G M ENTS
The authors gratefully acknowledge the financial support by the

D I SCLOS U R E
The authors declare no conflicts of interest.

F I G U R E 7
Combined antitumor effect of ML-05 with programmed cell death-1 (PD-1) Ab or stimulator of interferon genes (STING) agonist. (A-C) B16F10 tumor model in C57 mice was treated with ML-05 (50 mg/kg/day) and PD-1 Ab (10 mg/kg, twice a week) alone or in combination. (D,E) B16F10 tumor model in C57 mice was treated with ML-05 (50 mg/kg/day) and diamidobenzimidazole (diABZI; 0.5 mg/kg, day 1, 4, 7) alone or in combination. Tumor volume was measured every 3 days, and the tumor growth was compared by two-way ANOVA. Tumor weight was compared with vehicle group by one-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001. ****p < 0.0001 compared to vehicle group