The Voltage-Gated Hv1 H+ Channel Is Expressed in Tumor-Infiltrating Myeloid-Derived Suppressor Cells

Myeloid-derived suppressor cells (MDSCs) are key determinants of the immunosuppressive microenvironment in tumors. As ion channels play key roles in the physiology/pathophysiology of immune cells, we aimed at studying the ion channel repertoire in tumor-derived polymorphonuclear (PMN-MDSC) and monocytic (Mo-MDSC) MDSCs. Subcutaneous tumors in mice were induced by the Lewis lung carcinoma cell line (LLC). The presence of PMN-MDSC (CD11b+/Ly6G+) and Mo-MDSCs (CD11b+/Ly6C+) in the tumor tissue was confirmed using immunofluorescence microscopy and cells were identified as CD11b+/Ly6G+ PMN-MDSCs and CD11b+/Ly6C+/F4/80−/MHCII− Mo-MDSCs using flow cytometry and sorting. The majority of the myeloid cells infiltrating the LLC tumors were PMN-MDSC (~60%) as compared to ~10% being Mo-MDSCs. We showed that PMN- and Mo-MDSCs express the Hv1 H+ channel both at the mRNA and at the protein level and that the biophysical and pharmacological properties of the whole-cell currents recapitulate the hallmarks of Hv1 currents: ~40 mV shift in the activation threshold of the current per unit change in the extracellular pH, high H+ selectivity, and sensitivity to the Hv1 inhibitor ClGBI. As MDSCs exert immunosuppression mainly by producing reactive oxygen species which is coupled to Hv1-mediated H+ currents, Hv1 might be an attractive target for inhibition of MDSCs in tumors.


Introduction
The tumor microenvironment (TME) is composed of a complex mixture of tumorassociated fibroblasts, infiltrating immune cells, endothelial cells, extracellular matrix proteins (ECM), proteases and signaling molecules, as well as cytokines [1]. Interactions between these cellular and acellular constituents of the microenvironment play critical roles in cancer development and the response to therapeutics [1]. The immune system plays a crucial role in the regulation of progression of multiple tumor types [2]. Tumors develop strategies to escape the immune attack in order to survive. At early stages of tumor development, the immune cells efficiently remove the transformed cells; however, this function becomes ineffective as cancer progresses [2].
Myeloid-derived suppressor cells (MDSCs) represent heterogeneous, phenotypically immature myeloid cells that play a tumor-promoting role by maintaining a state of immunologic anergy and tolerance [3]. Activated MDSCs secrete chemokines, cytokines, and enzymes, which suppress local T-cell activation and viability [3]. In addition, MDSCs can suppress the anti-cancer effect of T cells through deprivation of nutrients, such as L-arginine and L-cysteine in the TME. A key characteristic of MDSCs is their generation of reactive oxygen (ROS) [4] and reactive nitrogen species (RNS) in the TME [5]. Consequently, the T cell receptor becomes oxidized and loses its ability to recognize foreign antigens. Moreover, they influence the chemotactic activity of T cells and this impairs the recruitment of cytotoxic CD8 + T cells to the TME [3,6]. MDSCs boost a group of T cells, the regulatory T cells (T regs), that are important for immune tolerance [7,8]. Two distinct subsets of MDSCs, polymorphonuclear (PMN-MDSCs) and monocytic (Mo-MDSCs), can be identified, each promoting tumor progression by different mechanisms and to a different extent [9]. Selective elimination of PMN-MDSCs is sufficient to induce the activation and proliferation of systemic and intra-tumor CD8 + T cells [10]. Owing to their versatile immunosuppressive effects, MDSCs represent an attractive but somewhat elusive potential therapeutic target.
LLC, a murine tumor model for non-small cell lung carcinoma [11], is known to host a very low number of CD4 + and CD8 + T lymphocytes and a large number of myeloid cells including macrophages, neutrophils, and dendritic cells. Because of this, it is considered an immunologically "cold" tumor [12]. It has been observed that MDSCs accumulate in large numbers both in the spleen [13,14] and in the blood [15] of LLC tumor-bearing mice. Compared to other murine cell lines of lung carcinoma, LLC tumors are characterized by a significantly lower infiltration of CD3 + T lymphocytes, in particular, the anti-tumoral cytotoxic CD8 + subgroup. This "cold" environment not only drives MDSCs into the tumor but renders immune checkpoint inhibitors like anti-PD1 and anti-PD-L1 to be almost completely ineffective in slowing cancer progression [16]. Therapies inhibiting MDSCs enhance the therapeutic effects of anti-PD1 antibodies [17].
Ion channels and transporters are well-known regulators of effector functions of T and B lymphocyte such as cytokine production, proliferation, differentiation, and cytotoxicity [18]. However, little is known about the expression of ion channels and their activity in MDSCs. Until recently, two types of ion channels have been described in MDSCs: P2X7R and TRPV1. In a murine neuroblastoma model, P2X7R is expressed in both spleen Moand PMN-MDSCs, but with different localization and function. Mo-MDSCs produce more arginase-1, TGF-β1, ROS, and upon activation with ATP, they secrete more CCL2, a tumorpromoting cytokine, than their neutrophil-like counterparts [19]. TRPV1 activation with cannabidiol stimulates the recruitment and activation of MDSCs and this exerts a protective function in a murine hepatitis model [20]. Very recently, the functional expression of the Hv1 proton channel was described in in vitro differentiated murine MDSCs [21].
The Hv1 proton channel is built up by a voltage sensor domain, an intracellular C-and N-terminal domain and the proton transport occurs through the voltage sensor domain since it lacks a classical pore [22]. Hv1 is activated, among others, by membrane depolarization, the pH gradient across the plasma membrane and temperature [23]. The Hv1 proton channel has been described in inflammatory cells such as granulocytes [24,25] macrophages [26], eosinophils [27], B cells [28], plasmacytoid DCs [29] and to a less extent, in T cells [30,31]. The Hv1 proton channel has been shown to regulate intracellular pH of tumor cells by mediating outward H + fluxes, thereby contributing to the acidification of the TME and the enhanced survival and mobility of tumor cells [32,33]. Pharmacological blocking of Hv1 induces intracellular acidification, which leads to apoptosis [34].
While acidic TME promotes survival and proliferation of cancer cells, it impairs the function of effector T cells [35]. Hv1 proton channel has been shown to support ROS production through NADPH oxidase activity in immune cells of innate and adaptive immunity [36]. However, so far, there are no data about the Hv1 proton channel in tumor-associated inflammatory cells, including MDSCs. In the present study, we aimed to investigate the Hv1 proton channel in MDSCs associated with Lewis lung carcinoma. Our results show that both Mo-MDSCs and PMN-MDSCs isolated from the tumor express Hv1 proton channel at RNA and protein level. Using whole-cell patch clamping, we demonstrated proton currents in both tumor-derived MDSCs, which could be blocked by the proton channel inhibitor 5-chloro-2-guanidinobenzimidazole (ClGBI).

Identification, Isolation, and Functional Characterization of Tumor-Associated Mo-and PMN-MDCs
An increased number of MDSCs is correlated with bad/worse prognosis in multiple tumor types [3]. First, we aimed to visualize the presence and localization of MDSC subsets within the TME of the LLC lung tumors induced in mice. MDSCs were identified by the co-expression of CD11b (myeloid cell-specific marker) and either Ly6G (granulocytic marker, PMN-MDSCs) or Ly6C (monocytic marker, Mo-MDSCs) using immunofluorescence ( Figure 1). PMN-MDSC-like cells represented an abundant group of cells and formed extensive aggregates in the tumor tissue ( Figure 1A). The phenotypically Mo-MDSCs were scattered throughout the tumor sections and, however, were less abundant than PMN-MDSCs ( Figure 1B). Additionally, we observed blood vessel staining with the Ly6C antibody, similar to recently published data [37].
immunity [36]. However, so far, there are no data about the Hv1 proton channel in tumor-associated inflammatory cells, including MDSCs. In the present study, we aimed to investigate the Hv1 proton channel in MDSCs associated with Lewis lung carcinoma. Our results show that both Mo-MDSCs and PMN-MDSCs isolated from the tumor express Hv1 proton channel at RNA and protein level. Using whole-cell patch clamping, we demonstrated proton currents in both tumor-derived MDSCs, which could be blocked by the proton channel inhibitor 5-chloro-2-guanidinobenzimidazole (ClGBI).

Identification, Isolation, and Functional Characterization of Tumor-Associated Mo-and PMN-MDCs
An increased number of MDSCs is correlated with bad/worse prognosis in multiple tumor types [3]. First, we aimed to visualize the presence and localization of MDSC subsets within the TME of the LLC lung tumors induced in mice. MDSCs were identified by the co-expression of CD11b (myeloid cell-specific marker) and either Ly6G (granulocytic marker, PMN-MDSCs) or Ly6C (monocytic marker, Mo-MDSCs) using immunofluorescence ( Figure 1). PMN-MDSC-like cells represented an abundant group of cells and formed extensive aggregates in the tumor tissue ( Figure 1A). The phenotypically Mo-MDSCs were scattered throughout the tumor sections and, however, were less abundant than PMN-MDSCs ( Figure 1B). Additionally, we observed blood vessel staining with the Ly6C antibody, similar to recently published data [37].  Staining with CD11b-specific antibody marks all myeloid cells (green), while co-staining with Ly6G (A) and Ly6C (B) antibodies (red) reveals PMN-MDSCs and Mo-MDSCs (arrows), respectively. Labels in the top right corner of the panels indicate the images obtained with filter settings specific for individual fluorophores or the merged image. Arrowheads mark blood vessels stained with the Ly6C antibody. Boxed areas in the left images are shown in higher magnification to the right panels. Scale bars are 100 µm for the low and 50 µm for higher magnification images.
Next, MDSCs were isolated from the TMEs of LLC-tumor-bearing mice according to Materials and Methods and MDSC subfractions were sorted using flow cytometry ( Figure 2A). First, we gated for all viable myeloid cells using a combination of morphology, singlet, and viability gates in combination with the myeloid marker CD11b + (Figure 2(A1-A4)). Within this population, we identified PMN-MDSCs as Ly6G + cells (Figure 2(A5)), and Mo-MDSCs were identified as CD11b + /Ly6G − /F4/80 − /MHCII − /Ly6C + cells (Figure 2(A6,A7)). We determined the relative proportion of these two populations within the CD11b + cells in the tumor and detected a predominance of PMN-MDSCs (60%) compared to Mo-MDSCs (around 10%) ( Figure 2B). The remaining 30% of myeloid cells include tumor-associated dendritic cells (TADC), macrophages (TAM), and MDSCs differentiating into TAMs. However, even these MDSC cell surface markers have overlapping expression patterns with other cell types such as monocytes and granulocytes and to date, no specific marker combinations have been described that unequivocally identify MDSC sub-populations [38,39].   The main characteristic of MDSCs Is their potent immune suppressive nature. To determine if purified MDSCs display immunosuppressive potency, we set up a polyclonal T cell proliferation suppression assay. Under these experimental conditions, Mo-MDSCs had suppressive capacity when they were co-cultured with murine splenocytes in 1:1 and 1:2 ratios, whereas PMN-MDCSs failed to demonstrate immunosuppressive properties in accordance with other studies [9]. Mo-MDSCs were the dominant immunosuppressive population of these MDSCs that suppressed both CD8 + and CD4 + T cell proliferation (Figures 3 and 4).

Detection of Hv1 Proton Channel in Tumor Derived MDSCs
A recent study reported the expression of the voltage-gated proton channel Hv1 in in vitro differentiated MDSCs [21]. Motivated by this, we tested the expression of the Hv1 transcript in PMN-and Mo-MDSCs isolated from the LLC tumor using qPCR ( Figure 5). Albeit both MDSCs subpopulations expressed the Hv1 gene, we found that the expression level in PMN-MDSCs was higher than in Mo-MDSCs. Western blot analysis using an antibody specific for Hv1 showed the same pattern, a higher Hv1 protein level in PMN-MDSCs compared to Mo-MDSCs.

Expression of Hv1 in MDSCs Infiltrating LLC Tumors
To gain more insight into the in situ expression of Hv1 in MDSCs infiltrating the LLC tumors, we performed immunofluorescence staining of LLC tumor sections using the Hv1-specific antibody validated in WB analysis. Figure 6A shows strong Hv1

Expression of Hv1 in MDSCs Infiltrating LLC Tumors
To gain more insight into the in situ expression of Hv1 in MDSCs infiltrating the LLC tumors, we performed immunofluorescence staining of LLC tumor sections using the Hv1-specific antibody validated in WB analysis. Figure 6A shows strong Hv1 immunofluorescence that overlaps with the expression of the myeloid marker CD11b. As expected, based on the WB data, the Hv1 signal was much stronger in PMN-MDSCs ( Figure 6B) compared to Mo-MDSCs ( Figure 6C). PMN-MDSCs in the tumor showed focal distribution whereas the Mo-DSCSs are more sparsely distributed, similar to the images in Figure 1.

Ion Currents in MDSCs
MDSCs obtained by cell sorting of the tumoral mass were analyzed using single-cell electrophysiology (patch-clamp) for the expression of whole-cell ion currents using Figure 6. Immunofluorescence staining of LLC cryosections to detect expression of Hv1 by myeloid cells (A-C). Boxed areas in the 1st column are shown in higher magnification to the right. Scale bars, 100 µm for the lower and 50 µm for the higher magnification images. (A) CD11b (red) marks all myeloid cells accumulated in the LLC tumor (4th column); the Hv1 signal is in green (3rd column), whereas the CD11b/Hv1 co-expressing myeloid cells are in yellow in the merged images (2nd column). (B) Hv1 + /Ly6G + MDSCs (arrow) cells in an LLC tumor section. L6yG + (red) marks PMN-MSDCs (4th column); the Hv1 signal is in green (3rd column), whereas the L6G + /Hv1 co-expressing myeloid cells are in yellow in the merged images (2nd column). (C) Hv1 + /Ly6C + cells (arrow) in a tumor section. Ly6C + (red) marks Mo-MSDCs (4th column); the Hv1 signal is in green (3rd column), whereas the Ly6C + /Hv1 co-expressing myeloid cells are in yellow in the merged images (2nd column). Ly6C additionally marks the blood vessels (arrowheads).

Ion Currents in MDSCs
MDSCs obtained by cell sorting of the tumoral mass were analyzed using single-cell electrophysiology (patch-clamp) for the expression of whole-cell ion currents using various intra-and extracellular solution combinations. Initially, we used K + -based intracellular and Na + -based extracellular solutions, which allow the recording of voltage-gated K + and Na + currents (see Materials and Methods). When either PMN-or Mo-MDSCs were subjected to 15 ms-long voltage steps from −100 mV holding potential to +50 mV test potential, we could not detect any classical voltage-gated ion currents in the outward direction ( Figure S1). Ion currents over a wider range of membrane potentials and depolarization durations were studied using voltage ramps. In these experiments ( Figure S1), we did not see inward currents characteristic of the presence of voltage-gated Na + or Ca 2+ channels. We did not optimize further the ion concentrations and voltage protocols for recording Na + and Ca 2+ currents, so we cannot exclude the possibility that a more detailed analysis would report some currents that were not readily seen during the initial characterization of the MDSC currents. On the other hand, we detected a voltage-gated outward current that activated at depolarized membrane potentials and was sensitive to the extracellular pH ( Figure S1). The presence of the proton current was observed more clearly when the recording solutions lacked conventional permeating cations and contained reduced Cl − concentration to eliminate outward currents other than the proton current (using NMDGbased solutions) and also rich in non-volatile buffers in order to keep both pH i and pH e stable [40]. Figure 7 shows the instantaneous I-V curves obtained using a voltage ramp protocol (from −60 mV to +150 mV), while keeping the intracellular pH constant at 6.2 and changing the extracellular pH from 5.7 to 7.4 in 3 steps. The Hv1 channel is characterized by a phenomenon called ∆pH-dependent gating: a change in the intracellular pH or in the extracellular pH strongly modulates the voltage at which the channel opens (threshold voltage, V thr ) [41]. Figure 7A,B show several features characteristic of Hv1. First, the smaller the pH gradient across the membrane [∆pH e-i = (pH e − pH i )], the more depolarized the V thr . Second: the larger the pH gradient, the larger the currents are at identical membrane potentials. The qualitative and quantitative analysis of the latter phenomenon is in Figure 7C-F. Plotting the current of each sweep at +145 mV as a function of the sequentially numbered sweep numbers (the time interval between the sweeps is 15 s) shows that changing the extracellular pH induces rapid and reversible effects on the current amplitude with currents being gradually smaller as the extracellular pH becomes more acidic, for both PMN-MDSCs ( Figure 7C) and Mo-MDSCs ( Figure 7D). Even if the capacitance measurements suggest that Mo-MDSCs are bigger than PMN-MDSCs (1.83 ± 0.14 (n = 40) vs. 3.13 ± 0.14 pF (n = 39), mean ± SEM, p < 0.0001) ( Figure S2), the H + current density in PMN-MDSCs ( Figure 7E) was~3 times bigger compared to Mo-MDSCs at pH e = 7.4 ( Figure 7F). At every pH e value, except 5.7, the current density on PMN-MDSCs was significantly larger than on Mo-MDSCs.
The V thr shift can be quantitatively inferred from the current-voltage relationship shown in Figure 8. The families of whole-cell currents in Figure 8A-D were obtained in a single PMN-MDSC upon applying 2000 ms-long step voltage depolarizations in 10 mV increments. The intracellular pH was maintained at pH i = 6.2 and the extracellular pH ranged from pH e = 7.4 ( Figure 8A) to pH e = 5.7 ( Figure 8D). Comparison of the topmost traces in Figure 8 panels A-D, obtained at +100 mV, indicates that the larger the pH gradient, the larger the current and the quicker its activation kinetics. Although larger depolarizations caused currents with faster activation kinetics, 2 s long pulses were not long enough to obtain saturation of the current. This can be observed both in PMN-MDSCs ( Figure 8A-D) as well as in Mo-MDSCs ( Figure S3) and it is a common feature of Hv1 currents [40,42]). Figure 8E,F show the normalized peak currents as function of the membrane potential. The V thr values were determined using the statistical criteria explained in detail in the Materials and Methods and in [42] and indicated by arrows in Figure 8E for PMN-MDSCs and in Figure 8F for Mo-MDSCs. The V thr values are shifted to more depolarized membrane potentials as pH e became more acidic while keeping the intracellular pH at pH i = 6.2. For statistical analysis, the V thr values were individually determined on a cell-by-cell basis and plotted as a function of the extracellular pH in Figure 9A, and as a function of ∆pH in Figure 9B. The numerical values of V thr for PMN-MDSCs were −16.7 ± 3.1 mV at ∆pH e-i = 1.2; +20 ± 5 mV at ∆pH e-i = 0.2; +15.0 ± 6.5 mV at ∆pH e-i = 0; and +63.3 ± 3.3 mV at ∆pH e-i = −0.5. For Mo-MDSCs, the V thr was −14.4 ± 5.5 mV for ∆pH e-i = 1.2; +40.0 ± 10.8 mV at ∆pH e-i = 0.2, +38.0 ± 9.2 mV for ∆pH e-i = 0; and +57.5 ± 7.5 mV at ∆pH e-i = −0.5 ( Figure 9A). These values were used to indicate the V thr in the current-voltage relationships ( Figure 8E,F) as colored arrows. The linear regression analysis of the V thr -∆pH e-i relationship did not deviate from the "rule of forty", i.e., (~40 mV shift per one unit ∆pH change) for either PMN-or Mo-MDSCs [41] ( Figure 9B); however, the V thr -∆pH e-i is shifted to depolarized potentials for the currents recorded in Mo-MDSCs ( Figure 9A). This is indicated by a slight upward shift in the V thr -∆pH e-i relationship obtained for Mo-MDSCs (dashed line in Figure 9B) versus that for PMN-MDSCs (solid line in Figure 9B), i.e., at identical pH gradients the thresholds are more positive for Mo-MDSCs as compared to PMN-MDSCs. current (using NMDG-based solutions) and also rich in non-volatile buffers in order to keep both pHi and pHe stable [40]. Figure 7 shows the instantaneous I-V curves obtained using a voltage ramp protocol (from −60 mV to +150 mV), while keeping the intracellular pH constant at 6.2 and changing the extracellular pH from 5.7 to 7.4 in 3 steps. The Hv1 channel is characterized by a phenomenon called ΔpH-dependent gating: a change in the intracellular pH or in the extracellular pH strongly modulates the voltage at which the channel opens (threshold voltage, Vthr) [41]. Figure 7A,B show several features characteristic of Hv1. First, the smaller the pH gradient across the membrane [ΔpHe-i = (pHe − pHi)], the more depolarized the Vthr. Second: the larger the pH gradient, the larger the currents are at identical membrane potentials. The qualitative and quantitative analysis of the latter phenomenon is in Figure 7C-F. Plotting the current of each sweep at +145 mV as a function of the sequentially numbered sweep numbers (the time interval between the sweeps is 15 s) shows that changing the extracellular pH induces rapid and reversible effects on the current amplitude with currents being gradually smaller as the extracellular pH becomes more acidic, for both PMN-MDSCs ( Figure 7C) and Mo-MDSCs ( Figure 7D). Even if the capacitance measurements suggest that Mo-MDSCs are bigger than PMN-MDSCs (1.83 ± 0.14 (n = 40) vs. 3.13 ± 0.14 pF (n = 39), mean ± SEM, p < 0.0001) (Figure S2), the H + current density in PMN-MDSCs ( Figure 7E) was ~3 times bigger compared to Mo-MDSCs at pHe = 7.4 ( Figure 7F). At every pHe value, except 5.7, the current density on PMN-MDSCs was significantly larger than on Mo-MDSCs.  High H + selectivity is a prominent property of the Hv1 channels [43]. The H + selectivity of the currents in PMN-MDSCs was estimated by determining the reversal potential (Erev) from the analysis of the whole-cell tail currents. The cells were depolarized to +100 mV for 500 ms to activate the current, followed by repolarizations to various membrane potentials (from -60 to +100 mV) to obtain the tail currents. Figure 10A-D shows a representative set of the tail current experiments at pHi 6.2 and at the indicated pHe values ranging from 5.7 to 7.4. The membrane potential at which the tail current reversed its polarity was considered as the reversal potential. Figure 10E shows selected traces at higher time and amplitude resolutions to illustrate the determination of Erev. As expected for a H + current, the Erev approached ~0 mV at ΔpHe-i = 0 (−1.3 ± 4.3 mV, mean ± SEM, n = 11). The Erev values were plotted against ΔpHe-i in Figure 10F and a straight line was fit to the data points. The slope of the best linear regression line indicates that Erev shifts −42 mV for every one-unit shift in the extracellular pH ( Figure 10F). However, the slope is different from a perfectly selective H + conductance (−59.16 mV/ΔpH) as predicted by the Nernst equation. Mo-MDSCs' currents were too low for a reliable tail current analysis. High H + selectivity is a prominent property of the Hv1 channels [43]. The H + selectivity of the currents in PMN-MDSCs was estimated by determining the reversal potential (E rev ) from the analysis of the whole-cell tail currents. The cells were depolarized to +100 mV for 500 ms to activate the current, followed by repolarizations to various membrane potentials (from −60 to +100 mV) to obtain the tail currents. Figure 10A-D shows a representative set of the tail current experiments at pH i 6.2 and at the indicated pH e values ranging from 5.7 to 7.4. The membrane potential at which the tail current reversed its polarity was considered as the reversal potential. Figure 10E shows selected traces at higher time and amplitude resolutions to illustrate the determination of E rev . As expected for a H + current, the E rev approached~0 mV at ∆pH e-i = 0 (−1.3 ± 4.3 mV, mean ± SEM, n = 11). The E rev values were plotted against ∆pH e-i in Figure 10F and a straight line was fit to the data points. The slope of the best linear regression line indicates that E rev shifts −42 mV for every one-unit shift in the extracellular pH ( Figure 10F). However, the slope is different from a perfectly selective H + conductance (−59.16 mV/∆pH) as predicted by the Nernst equation. Mo-MDSCs' currents were too low for a reliable tail current analysis.
ClGBI is an apolar small molecule, which can cross the plasma membrane and block the Hv1 channel from the cytosolic side [44]. The sensitivity to guanidine derivatives, and particularly to ClGBI, is often used in the literature as a pharmacological argument for the identification of Hv1 currents [21,34,42,45]. Accordingly, we tested the sensitivity of the whole-cell currents to ClGBI at 200 µM concentration, which was reported to block~80% of the Hv1 current in a reversible manner [44]. Figure 11A shows that 200 µM ClGBI blocked almost completely the whole cell current in a PMN-MDSC and that the block was reversible, the current returned to the control upon washing the recording chamber with the ClGBI-free extracellular solution. The development of the block was very fast whereas >30 episodes in a ClGBI-free solution were needed to wash-out the effect ( Figure 11B). While PMN-MDSCs were robust enough to withstand repeated 2000 ms-long depolarizations from −80 mV to +100 mV ( Figure 11A), we were able to do pharmacological experiments in Mo-MDSCs using repeated application of a voltage-ramp protocol where long exposure to depolarized test potentials can be avoided ( Figure 11C). Regardless of the voltage protocol used (i.e., step depolarization vs. voltage ramp), the application of 200 µM ClGBI reduced the magnitude of the current significantly,~80% reduction in PMN-MDSCs and 75% in Mo-MDSCs ( Figure 11D). ClGBI is an apolar small molecule, which can cross the plasma membrane and block the Hv1 channel from the cytosolic side [44]. The sensitivity to guanidine derivatives, and pharmacological experiments in Mo-MDSCs using repeated application of a voltage-ramp protocol where long exposure to depolarized test potentials can be avoided ( Figure 11C). Regardless of the voltage protocol used (i.e., step depolarization vs. voltage ramp), the application of 200 µM ClGBI reduced the magnitude of the current significantly, ~80% reduction in PMN-MDSCs and 75% in Mo-MDSCs ( Figure 11D).

Discussion
Our paper demonstrates for the first time that murine MDSCs obtained directly from tumor tissue express the Hv1 H + channel both at the mRNA and at the protein level and that the properties of the whole-cell current in tumor-derived MDSCs recapitulate the hallmarks of Hv1 currents recorded in various cells and in cells expressing Hv1 heterologously. These hallmarks are the voltage-dependent activation, ~40 mV shift in the activation threshold of the current per unit change in the extracellular pH and high H +

Discussion
Our paper demonstrates for the first time that murine MDSCs obtained directly from tumor tissue express the Hv1 H + channel both at the mRNA and at the protein level and that the properties of the whole-cell current in tumor-derived MDSCs recapitulate the hallmarks of Hv1 currents recorded in various cells and in cells expressing Hv1 heterologously. These hallmarks are the voltage-dependent activation,~40 mV shift in the activation threshold of the current per unit change in the extracellular pH and high H + selectivity as reviewed extensively by [43]; and the sensitivity to the guanidine derivative ClGBI [44].
A key novelty of our study is that the expression of Hv1 was shown in MDSCs obtained from tumors. As described in the introduction, the LLC tumor of mice was a good candidate for the isolation of MDSCs. LLC is considered an immunologically "cold" tumor [12] which is characterized by high MDSC infiltration and an immunosuppressive microenvironment [46]. Although MDSCs are in the focus of tumor immunology and are intensively investigated there is no consensus regarding the phenotypic definition of these cells. Usually, in mice, Mo-MDSCs are defined as CD11b + Ly6C + and PMN-MDSCs as CD11b + Ly6G + Ly6C low , but these markers are commonly defining other subsets of myeloid cells as well [47]. Taking into consideration the limitation of the identification of these cells by cell surface markers, we demonstrated the presence of both PMN-MDCSs and Mo-MDSCs in LLC tumors induced in mice using immunofluorescence (Figure 1). Moreover, Mo-MDSCs can be distinguished from tumor-associated macrophages (TAMs) because of their lower expression of F4/80 [48]. This was utilized in flow cytometric separation and specific enrichment MDSCs for electrophysiological investigations. We found tumor-derived PMN-MDSCs to be the most abundant subset in LLC, being detected by flow cytometry~6 times more than Mo-MDSCs ( Figure 2B). This is common for this kind of tumors [49] and similar proportions have been reported in pancreatic ductal adenocarcinoma (PDAC) [50] and autoimmune diseases like autoimmune arthritis as well [51].
Since the definition of MDSCs via membrane markers is not straightforward, it is common practice to verify the identity of the cells with a functional study. This is usually achieved by demonstrating the suppression of T cell proliferation to avoid confusion with phenotypically similar monocytes and neutrophils [48]. In our hands, tumor-derived Mo-MDSCs were able to suppress T-cell proliferation, while PMN-MDSCs, although more abundant, did not suppress T-cells, at least at the MDSC/splenocyte ratios we used. This insufficient anti-proliferative phenotype of PMN-MDSCs agrees with other studies performed with LLC tumors in mice [9] and it has been observed in PDAC [52], autoimmune arthritis [51], and MDSCs accumulating in transplanted organs in humans [53]. However, PMN-MDSCs may promote tumor growth independent of the inhibition of T cell proliferation by directly inhibiting cytolytic T cell activation and indirectly influencing other myeloid cells and NK cells [54,55]. PMN-MDSCs are the major source of immunosuppressive mediators like ROS and RNS, which suppress TCR signaling and modulate cytokine secretion [56]. Additionally, PMN-MDSCs impair recruitment of cytolytic T cells [57] and contribute the tumor progression by secreting MMPs and factors that promote tumor angiogenesis [58][59][60]. Moreover, a recent electrophysiological study strongly supports that the cells we classified as PMN-MDSCs are different from neutrophils. Using electrophysiological assays, Immler and co-workers have shown that neutrophil polymorphonuclear leukocytes (PMN) functionally express voltage-gated Kv1.3 K + channels [61], in clear contrast to our whole-cell records, where Kv1.3 or any other voltage-gated K + current was missing.
Several lines of evidence support that whole-cell Hv1 currents were recorded in tumorderived Mo-and PMN-MDSCs in our study. First, the currents were slowly activating, rapidly deactivating, and with no sign of inactivation, which is characteristic of Hv1. Moreover, the currents were recorded using intra-and extracellular solutions that lacked (K + , Na + ) or contained negligible concentration (Cl − ) of conventional permeating ions; thus, the contribution of other conductances to the whole-cell current, that could mimic the behavior of Hv1, are minimized.
Second, the whole-cell currents in both MDSC types were sensitive to the pH gradient across the membrane and the membrane potential. The threshold voltage for the activation of the currents shifts along the voltage axis when changing ∆pH,~40 mV per unit change in the extracellular pH, closely mirroring what has already been described for proton currents in various cells [43,62] including bone-marrow-derived MDSCs [21]. On the other hand, V thr of the current varies among different cell types [42,63]. For example, at identical pH gradients (∆pH = 1.2) and recording solutions, the Hv1 current in human chorion-derived mesenchymal cells activate at~10 mV more positive membrane potential than hHv1 expressed in HEK-293 cells [42], whereas the V thr values in both types of MDSCs in this study are~10 mV more negative than that of hHv1 in HEK-293. This more negative threshold potential may facilitate the opening of Hv1 at membrane potentials in the physiological range typically observed for non-excitable cells. Nevertheless, the V thr values determined in our study are more positive than the reversal potentials of the H + currents obtained at a wide range of ∆pH values (compare Figures 9 and 10), thus, allowing the Hv1 channel to conduct protons solely in the outward direction, similar to other cells [43]. We also found that the V thr in PMN-MDSCs is depolarized as compared to Mo-MDSCs under symmetrical pH conditions (pH e~p H i ), even if this did not affect the overall V thr-∆pH relationship of -40 mV/∆pH [64]. The V thr of~+40 mV in symmetric solutions in PMN-MDSCs is qualitatively similar to what has been determined for the Hv1 current in murine neutrophil granulocytes (~+50 mV, [65]), that are closely related to PMN-MDSCs. Moreover, specific mutations generated in the hHv1 channel drastically modify V thr , without influencing the V thr -∆pH relationship [64]. We do not know whether our observations originate from technical errors mainly due to the extremely low ion currents in Mo-MDSCs or from a translational or post-translational difference between Hv1 in PMNand Mo-MDSCs.
Third, the Hv1 current in PMN-MDSCs is fairly H + -selective, since the E rev -∆pH relationship resembles the theoretical relationship obtained for H + from the Nernst equation. The slope of the E rev -∆pH relationship in rat alveolar epithelial cells [62] and canine myocytes [66] is similar to the theoretical slope calculated from the Nernst equation for H + (~−59 mV/∆pH); however, slopes in Jurkat (−47 mV/∆pH, [30]) and in MDSCs in our study (−42 mV/∆pH) are shallower. A similar discrepancy has been observed in murine microglia as well, where proton depletion, as a consequence of the proton current passing through Hv1 channels, was suggested to account for the shallower slope [67]. In addition, the small currents in MDSCs can be easily contaminated by non-specific leak even if leak corrections are applied: any contribution of leak to the whole cell current shifts the reversal potentials to depolarized potentials. The complications originating from incomplete leak subtraction ruled out the reliable determination of the reversal potential in Mo-MDSCS where currents are very small, in many cases less than 100 pA even under optimal ∆pH and membrane potential combinations.
Fourth, the currents in both PMN-MDSCs and Mo-MDSCs were sensitive to ClGBI, the guanidine derivative small molecule inhibitor Hv1 [44]. Even if the selectivity of ClGBI towards Hv1 has not been assessed yet, it is widely used as an indicator of the presence of the Hv1 current in various cell types [42,66,68]. We showed that ClGBI at 200 µM concentration reversibly blocks~80% of the whole cell currents in both PMN-and Mo-MDSCS. Based on the block percentage and assuming a sigmoidal dose-response function with a Hill coefficient of 1, the single-point estimate of the IC 50 is~50 µM, which is consistent with the reported potency of ClGBI in inhibiting Hv1 [44,66]. This pharmacological clue strengthens our conclusion that these currents correspond to proton currents mediated by Hv1 in MDSCs.
Fifth, the electrophysiological data are strongly supported by molecular biology where the mRNA transcript of Hv1 was identified in MDSCs using RT-qPCR along with the Hv1 protein itself in Western blots. The human Hv1 proton channel protein has two isoforms, a long, full-length isoform and a short one, which lacks an N-terminal region due to alternative splicing [69]. The antibody used in the present study recognizes both isoforms, which is confirmed in the CH12 mouse B cell lymphoma cell line used in our study as a positive control, and similar to human B cell lymphomas described previously [69]. However, we detected only the long form both in PMN-MDCS and Mo-MDSCs isolated from the LLC tumor. The short form might not be expressed, or it is in a negligible amount, under the detection limit, suggesting that the long isoform of Hv1 may contribute to the function of tumor-derived MDSCs.
Although the Hv1 proton channel has been well characterized in several immune cell types, to our knowledge, there is no information about Hv1 expression in tumor-associated inflammatory cells. Our study, for the first time, detected within a tumor a high number of Hv1 + myeloid cells that are consistent with the cell surface maker phenotype of PMNand Mo-MDSCs (see above). Our electrophysiology results are consistent with the data described for in vitro produced MDSCs where Hv1 H + currents of similar magnitude (between 200 pA and 1 nA at +130 mV) to our study were reported using patch-clamp in a mixed MDSC population [21]. The MDSCs used by Alvear-Arias et al. were obtained by induction of the differentiation of bone marrow-derived myeloid precursors by GM-CSF [21] whereas in our study, MDSCs were isolated directly from LLC tumors induced in mice. Our study suggests that in vitro differentiated MDSCs may serve as useful tools to understand Hv1-dependent regulation of T cell function in cancer as the channel phenotype of these cells is similar to the tumor-derived MDSCs. Moreover, we also showed that both PMN-and Mo-MDSCs display Hv1-mediated H + currents, albeit to a different extent, so the ion channel phenotype of the two MDSCs subtypes is similar, at least in mice.
What can be the functional consequence of the Hv1-mediated H + currents in MDSCs? Neutrophils, which are closely related to MDSCs, express a functional Hv1 and the Hv1 H + currents contribute to the counterbalancing positive charge efflux required for the maintenance of ROS production [36,70]. As ROS production is also a hallmark of MDSCs' immunosuppression [4], the functional expression of Hv1 in MDSCs and its sensitivity to Hv1 inhibitors seems logical. Consistent with this, Hv1-mediated H + currents were shown in MDSCs using electrophysiology ( [21] and this study), and blocking Hv1 using ClGBI and Zn 2+ inhibited ROS production and alleviated the inhibition of T cell proliferation by MDSCs [21]. Longer than 2 h exposure of MDSCs to Hv1 inhibitors induced significant cell death which raised some ambiguity regarding the specificity of the effect of ClGBI application. This, and the application of a reversible blocker (ClGBI) in a short-term preincubation to MDSCs followed by wash-out, makes the interpretation of the data on T cell/MDSCs co-cultures complex and difficult. In our hands, ClGBI applied alone, in the absence of MDSCs, inhibited T cell proliferation, which ruled out T-cell/MDSC co-culture experiments in the presence of ClGBI. The potential side effects of currently available Hv1 inhibitors argue for the development of more specific and higher affinity Hv1 inhibitors.
The proton efflux through the Hv1 proton channel may also contribute to the acidic milieu in the TME, which is well tolerated by tumor cells but impairs the tumor suppressive ability of T cells, NK cells [35]. Thus, modulating the acidic tumor microenvironment by Hv1 inhibition may facilitate the tumor-suppressive effect of immune cells in cancer therapy. However, recently it has also been shown that the increase of intracellular acidity in activated T cells due to the lack of Hv1 proton channel reduces the effector function of T cells [31], which must also be considered to determine the overall outcome of Hv1-targeted cancer therapy.

Cells
CH12 B cell lymphoma cells were grown in RPMI 1640 medium supplemented with 10% FBS, 1% GlutaMAX, 1% penicillin-streptomycin. The Lewis lung carcinoma cell line (LLC) was a kind gifts from László Nagy (Department of Biochemistry and Molecular Biology, Faculty of Medicine, University of Debrecen, Hungary). Cells were grown in RPMI 1640 medium supplemented with 10% FBS, 1% GlutaMAX, and 1% penicillin-streptomycin.

Real Time Quantitative PCR (RT-qPCR)
Total RNA was isolated from the sorted cells using Trizol reagent (Life Technologies/Thermo Fisher Scientific, Waltham, MA, USA), treated with RNAse-free DNase 1 (Thermofisher, AM2222) for 30 min at 37 • C, followed by inactivation of the DNase at 75 • C for 10 min. The DNase-treated RNAs were submitted to cDNA synthesis. cDNAs were synthesized using a High-Capacity cDNA Reverse Transcription Kit (ThermoFischer Scientific, Waltham, MA, USA, cat #4368814) following the manufacturer's instruction.
Quantitative PCR was performed by Applied Biosystems Step One Plus platform, 95 • C for 10 min, 40 cycles of 95 • C 15 sec and 60 • C 45 sec, using Light Cycler 480 SYBR Green I. Master mix (Roche, Basel, Switzerland). Gene expression was quantified by the comparative threshold cycle method and normalized to mouse 18S expression as a housekeeping gene. All PCR reactions were performed in duplicate. Values are expressed as means ± SD. The following primers were used: mouse Hv1 (forward: TCGTGCTTGCT-GAACTCCTCCT and reverse: GGCAAAGCTCATGTAGTGGAACG); mouse 18S RNA: (forward: GGGAGCCTGAGAAACGGC and reverse: GGGTCGGGAGTGGGTAATTTG).

Western Blot
The sorted cells were lysed in 2xLaemmli buffer supplemented with proteinase (Sigma/Merck KGaA, Darmstadt, Germany, cat #P8340) and phosphatase inhibitor (Sigma, cat #P5726) cocktail followed by a denaturation step at 98 • C for 5 min. Fifteen µg protein per sample was run on 7.5% polyacrylamide gel and transferred to a nitrocellulose membrane (Bio-Rad Laboratories, Hercules, CA, USA). The non-specific binding sites were blocked by incubating the nitrocellulose membrane in 5% w/v non-fat dry milk in Tris-Buffered Saline (TBS) supplemented with 0.2% v/v Tween (TBST) for 1 h. Next, the nitrocellulose membrane was incubated with primary antibodies (Hv1 (#13) and β-actin (#14), see Table 1) diluted in 2.5% w/v non-fat dry milk in TBST overnight at 4 • C. After washing for 3 × 7 min in TTBS, the nitrocellulose membrane was incubated with donkey anti-rabbit IgG-HRP linked (#17) secondary antibody for detection of Hv1 protein.
Sheep anti-mouse IgG-HRP linked (#15) secondary antibody was used for the detection of actin. The chemiluminescence signal was detected using an Azure c300 Gel Imaging System (Azure Biosystems, Dublin, CA, USA). CH12 B cell lymphoma lysate was used as a positive control. The Western blot analysis was performed on sorted cells from 3 different experiments. The relative Hv1 protein level was calculated using Azure Spot Pro Analysis software and expressed as Hv1/actin ratio.

Immunofluorescence
Frozen sections were fixed in methanol at −20 • C for 15 min, washed, blocked in 3% bovine serum albumin in PBS salt solution, and incubated overnight at 4 • C with the antibodies diluted in blocking solution. The antibodies used for staining cryosections are listed in Table 1 (#3, 5, 7, 13).
After washing for 3 × 5 min in 1×PBS, sections stained for Hv1 were additionally incubated for 1 h at room temperature with an appropriate secondary antibody (#16).
Sections were stained with DAPI solution (Invitrogen, 1 µg/mL) to visualize nuclei and mounted with Fluoromount G (eBioScience, San Diego, CA, USA) mounting media. The specificity of the secondary antibody was verified by omitting the primary antibody from the staining procedure. Sections were examined using a LSM800 microscope and Zen 2.3 SP1 software (Zeiss, Jena, Germany).

Electrophysiology and Pharmacology
Electrophysiology measurements were carried out using the patch-clamp technique in voltage-clamp mode. Whole-cell currents were recorded from murine PMN-and Mo-MDSCs using a Multiclamp 700B amplifier connected to a DigiData 1440A digitizer (Molecular Devices, Sunnyvale, CA, USA). Micropipettes were pulled from GC 150 F-15 borosilicate capillaries (Harvard Apparatus, Kent, UK) resulting in 3-to 5-MΩ resistance in the bath solution. The standard extracellular solution used to study Hv1 at pH e = 7.4 contained 180 mM HEPES, 75 mM N-Methyl-D-Glucamine (NMDG), 15 mM glucose, and 3 mM MgCl 2 (titrated with CsOH), whereas in the extracellular solutions at pH e = 6.4/6.2/5.7, 180 mM HEPES buffer was substituted with 180 mM MES (2-(N-morpholino)ethanesulfonic acid, titrated with CsOH or HCl). The standard intracellular solution at pH i = 6.2 contained 180 mM MES, 75 mM NMDG, 15 mM glucose, 3 mM MgCl 2, and 1 mM EGTA (ethylene glycol-bis(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid, titrated with CsOH). To explore the presence of other voltage-gated currents, we used a Na + -based extracellular solution at pH e = 7.35 containing 145 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 2.5 mM CaCl 2 , 5.5 mM glucose, and 10 mM HEPES and a K + -based intracellular solution at pH i = 7.22 containing 140 mM KF, 2 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES, and 11 mM EGTA. The pH of the solutions was checked before every experiment, and all salts and components of the solutions were purchased from Sigma-Aldrich Budapest, Hungary. A custom-built gravity-driven perfusion system was used to provide the necessary solution exchange around the cells.
The guanidine derivative Hv1 blocker 5-cholor-2-guaninidbenzimidazole (ClGBI, Sigma Aldrich Kft. Budapest, Hungary, S517038) was kept in DMSO at a stock concentration of 100 mM and suitably diluted in the standard extracellular solution when needed.

Electrophysiology Data Acquisition and Analysis
Voltage ramps (2000 ms long from −60 mV to +150 mV, every 15 s) were used to demonstrate qualitatively the dependence of the activation threshold of the Hv1 currents at various pH e values. Traces were filtered (lowpass boxcar, 25 smoothing points), off-line leak-corrected manually point-by-point. Linear regression line was fit to the data points below the activation threshold of the H + current (between 50 ms to 330 ms, corresponding to −60 mV and −30 mV) and the fitted parameters were used to subtract the non-specific leak [42]. The leak-corrected currents between +145 mV and +146 mV were extracted, averaged, and considered as the peak current. The average currents of two or three stable traces at a given pH e condition defined one data point. Currents are either shown as absolute values or expressed as current density obtained by dividing the currents measured in pA with the cell capacitance in pF to yield pA/pF.
The current-voltage (I-V) relationships and the activation threshold voltage of the currents (V thr ) were determined using 2 s long step depolarizations from a holding potential of −80 mV to +100 mV in +10 mV increments. The protocol was applied every 15 s; the sampling rate was 5 kHz. For the I-V curves, every trace was filtered (lowpass boxcar, 25 smoothing points) and leak-corrected manually. Peak currents were calculated as the average of the last 18 points (i.e., between 2051.1 and 2051.9 ms) at the end of the depolarizing pulses. For the V thr determination, leak correction was performed using the first 5 (pH e 7.4), 7 (pH e 6.4 and 6.2), and 10 peak currents of the I-V relationship (i.e., between −80 and −50/−30/+10 mV) and the SD was calculated using the first 5 values in the I-V (i.e., between −80 and −50 mV). The V thr was selected as the membrane potential at which the current was above 2 × SD [42].
For recording tail currents, the Hv1 current was fully activated using 500 ms long single-step depolarizations from a holding potential of −80 mV to +100 mV. The tail currents were recorded upon stepping back from this potential in 20 mV decrements to −60 mV, and the currents were recorded for 250 ms at the back-step potentials. The protocol was applied every 15 s with a sampling rate of 20 kHz. The traces were leak-corrected manually and filtered (lowpass boxcar, 25 smoothing points).
To test the presence of voltage-gated K + currents, 15 ms long depolarization steps were applied to +50 mV from a holding potential of −100 mV every 15 s, with a sampling rate of 20 kHz. Voltage ramps, as specified above, were also used to study the presence of voltage-gated ion currents in MDSCs over an extended membrane potential range and depolarization duration in physiological salt solutions.
The pClamp 10.5, 10.7, and 11.2 software packages were used to acquire the data. The pClamp 10.7 and 11.1 software packages (Molecular Devices Inc., Sunnyvale, CA, USA) were used to analyze the data. Statistical analyses were performed with GraphPad Prism 8.4.3 (GraphPad Software, Inc., San Diego, CA, USA).