Adenosine deaminase inhibition suppresses progression of 4T1 murine breast cancer by adenosine receptor‐dependent mechanisms

Abstract The activity of a cell‐surface ecto‐adenosine deaminase (eADA) is markedly increased in the endothelial activation and vascular inflammation leading to decreased adenosine concentration and alterations in adenosine signalling. Depending on the specific pathway activated, extracellular purines mediate host cell response or regulate growth and cytotoxicity on tumour cells. The aim of this study was to test the effects of adenosine deaminase inhibition by 2′deoxycoformycin (dCF) on the breast cancer development. dCF treatment decreased a tumour growth and a final tumour mass in female BALB/c mice injected orthotopically with 4T1 cancer cells. dCF also counteracted cancer‐induced endothelial dysfunction in orthotopic and intravenous 4T1 mouse breast cancer models. In turn, this low dCF dose had a minor effect on immune stimulation exerted by 4T1 cell implantation. In vitro studies revealed that dCF suppressed migration and invasion of 4T1 cells via A2a and A3 adenosine receptor activation as well as 4T1 cell adhesion and transmigration through the endothelial cell layer via A2a receptor stimulation. Similar effects of dCF were observed in human breast cancer cells. Moreover, dCF improved a barrier function of endothelial cells decreasing its permeability. This study highlights beneficial effects of adenosine deaminase inhibition on breast cancer development. The inhibition of adenosine deaminase activity by dCF reduced tumour size that was closely related to the decreased aggressiveness of tumour cells by adenosine receptor‐dependent mechanisms and endothelial protection.


| INTRODUCTION
Breast cancer is a major cause of deaths in women. 1 Despite advances in early detection and therapy in the past few years, malignant breast cancer stands for the poor prognosis. 1 Therefore, new therapeutic approaches are needed. Adenosine deaminase could be a potential therapeutic target as several studies have shown its increased activity in serum and tumour tissues in breast cancer 2,3 and other malignant cancers. 4,5 Adenosine deaminase (ADA, E.C. 3.5.4.4) is an enzyme that catalyses the irreversible deamination of both 2′deoxyadenosine (dAdo) and adenosine (Ado). 6 There are two isoenzymes of ADA in human tissues, ADA1 and ADA2. 7 The ADA1 isoenzyme is ubiquitous and has a similar affinity for both substrates (dAdo/Ado deaminase ratio of 0.75). ADA1 Km for Ado is 5.2 × 10 −5 mol/L and its optimal pH is 7.0-7.5. Therefore, ADA1 is highly efficient in biological sites where the pH is neutral even if substrate concentration is low. 8 Moreover, ADA1 could interact with membrane proteins and exist as an ectoenzyme (eADA), deaminating Ado and dAdo in extracellular space. 9 ADA2 is not ubiquitous and coexists with ADA1 only in human monocytes-macrophages, being the main ADA isoenzyme found in human serum. The Km for Ado of ADA2 is much higher (200 × 10 −5 mol/L) and it has a weak affinity for dAdo (dAdo/Ado deaminase ratio of 0.25). ADA2 has an optimum pH of 6.5, making it efficient in deaminating high adenosine concentrations in the slightly acidic environment, for example during inflammation. 8 The significance of ADA in the breast cancer development seems to be particularly important as its activity regulates the pool of intraand extracellular adenosine, a key modulator of a cell function via adenosine receptor-dependent 10 and independent mechanisms. 11 It has been shown that both ADA isoenzymes were elevated in tumour tissues of patients with breast cancer correlating with tumour grade, size and lymph node involvement. 12,13 On the other hand, only ADA2 activity was increased in serum of those patients. 12 Substantial differences in ADA activity have been also revealed between malignant and benign breast neoplastic tissues. 14 Interestingly, malignant breast tissue revealed a much higher activity of ADA1 than ADA2. While, no differences were found in serum activities of both isoenzymes between malignant and benign breast neoplasm.
Such differences in tissue activities of ADA indicate the diverse cellular origin of both isoenzymes. Cancer cells could be a potential source of ADA, but this has not been comprehensively studied. It has been also speculated that elevated ADA activity in serum and breast cancer tissues may originate from other sources than tumour cells. The increased activity of tumour ADA could be a reflection of ADA-rich immune cell accumulation in malignant breast tumours. 6,15 As we reported previously that activated endothelial cells also exhibit strongly increased eADA activity, ADA in tumour could also derive from actively proliferating endothelial cells during tumour angiogenesis. 16,17 Because of the unclear data regarding the source of ADA in tumour development, this work focuses on cellular origin of its activity.
It has been proposed that high ADA activity in tumour microenvironment may be a compensatory mechanism against a toxic accumulation of its substrates due to increased purine and pyrimidine metabolism in cancerous tissues. High ADA activity might give a selective advantage to the cancer cells producing a high amount of hypoxanthine, a substrate for the salvage pathway by the activity of hypoxanthine guanine phosphoribosyl transferase (HGPRT). 18 Therefore, the inhibition of adenosine degradation deserves a special attention in the cancer therapy. Up to date, a potent anticancer effect of nonspecific adenosine deaminase inhibitor, erythro-9-(2hydroxy-3-nonyl)adenine (EHNA), was indicated against malignant pleural mesothelioma, but the effect of a specific ADA inhibition on the breast cancer development has never been studied. 19 Therefore, the aim of this work was to evaluate the antitumour potential of a specific inhibitor of total ADA, 2′deoxycoformycin (dCF) in mouse breast cancer models and human breast cancer cell lines as well as to estimate the activities of both ADA isoenzymes in malignant and benign breast neoplastic cells, immune cells and endothelial cells.

| MATERIALS AND METHODS
An expanded Methods section is available in the Supporting information online.

| Orthotopic murine breast cancer model
According to the experimental protocol ( Figure 1A), 24 female BALB/ c mice at 8 weeks of age were randomly divided into four groups, F I G U R E 1 Adenosine deaminase inhibition by deoxycoformycin suppresses progression of 4T1 murine breast cancer. The design of the experiment (A). The computed tumour volume measured with a calliper every 2 d during the experiment starting from the 14th d after orthotopic injection of 3 × 10 5 4T1 cancer cells in PBS (n = 7) and dCF-treated (n = 7) BALB/c mice (B). Tumour mass of PBS-(4T1) and dCFtreated (4T1 + dCF) mice weighed on 28 d of the experiment (C). Representative images of tumours visualized by USG of 4T1 (D) and dCF+4T1 (E) mice. White arrows point to tumour border. Black arrows indicate the necrosis. The area of tumours analysed using USG (F). The area of tumour necrosis measured as a percentage of total tumour area analysed using USG (G). Spleen mass of BALB/c control mice treated with PBS (control, n = 5) or 0.2 mg/kg dCF (dCF, n = 5) every 3 d for 28 d of the experiment and mice 28 d after orthotopic injection with 4T1 cancer cells treated with PBS (4T1) or dCF (dCF+4T1) (H). The activity of vascular adenosine deaminase measured in the tissue homogenate (intra-and extracellular ADA) (I) and on the surface of the vessel (ecto-ADA, eADA) (J) in descending thoracic aorta of mice studied. The activity of tumour total adenosine deaminase (tADA) measured in tissue homogenate in 4T1 and 4T1 + dCF mice groups (K). Data are presented as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by two-way ANOVA followed Holm-Sidak post hoc test (B), one-way ANOVA followed Holm-Sidak post hoc test (H-J) or by Student's t test (C, F, G, K) phosphate-buffered saline (PBS)-treated mice uninjected with 4T1 cancer cells (control, n = 5), deoxycoformycin (dCF)-treated mice uninjected with 4T1 cancer cells (dCF, n = 5), PBS-treated mice injected with 4T1 cancer cells (4T1, n = 7) and dCF-treated mice injected with 4T1 cancer cells (4T1 + dCF, n = 7). Sterile PBS or dCF (0.2 mg/kg body weight in sterile PBS) was injected intraperitoneally, every 72 hours starting from 1th day of tumour inoculation.
The bodyweight of each mouse was determined before each dCF/ PBS injection.
The 4T1 tumour cells suspension diluted in sterile PBS was subcutaneously injected (0.15 mL, 3 × 10 5 cells/mouse) in the right armpit. Mice uninjected with 4T1 cells (control, dCF) received adequate volume of sterile PBS. The tumour was detected palpably after 2 weeks of induction. The weight of each mice and the tumour size were measured every 2 days starting from 14th day of tumour inoculation. The tumour was measured with a calliper and its volume was calculated using following formula: V (mm 3 ) = (a × b 2 )/2, where a and b represented maximum and minimum diameter, respectively.
Twenty-eight days after the inoculation of cancer cells, mice were weighed and anaesthetized with a ketamine-xylazine (100 mg/ kg/10 mg/kg) by an intraperitoneal injection. Subsequently, animals underwent two-dimensional USG analysis of tumours, when tumour area and tumour necrosis area were analysed. Venous blood and heparinized plasma were collected and immediately frozen in liquid nitrogen. Then, tumour and spleen were removed and weighed. Thoracic aorta was collected, and perivascular adventitia was removed.

| Determination of vascular extracellular adenosine deaminase activity
Purified fragments of mice thoracic aorta were opened longitudinally by an incision along its ventral aspect and were incubated with 50 μmol/L adenosine in 1 mL of HBSS by immersing aortic fragments in the incubation medium. Samples were collected after 0, 5, 15 and 30 minutes of incubation in 37°C and directly analysed with high-performance liquid chromatography (HPLC). Adenosine and inosine concentrations were measured by reversed-phase HPLC as described earlier. 16 The rate of adenosine to inosine deamination was calculated from a linear phase of the reaction and expressed as the inosine increase over the time normalized for the weight of wet tissue [μmol/min/g tissue].  20 The results were expressed as the inosine increase over the time (μmol/min/g tissue).

| Determination of arginine analogues in mice plasma
The concentration of asymmetric dimethyl L-arginine (ADMA), symmetric dimethyl L-arginine (SDMA), N-monomethyl L-arginine (L-NMMA) and L-arginine was measured using previously published method as described in the Supporting information online. 21

| Determination of nucleotides and metabolites in mice blood and blood morphology
To determine nucleotide and their metabolites concentration, frozen blood was extracted with 1.3 mol/L HClO 4 (1:1 v/v) and centrifuged at 20 800 g (10 minutes, 4°C). Supernatants were neutralized with 3 mol/L K 3 PO 4 and centrifuged (20 800 g, 10 minutes, 4°C). The concentration of nucleotides in supernatants was measured by HPLC as described earlier. 22 Blood morphology was analysed as described in the Supporting information online.

| Determination of the effect of dCF on total (intra-and extracellular) ADA activity in cell cultures
After reaching confluence at 6-well culture plates, H5V and 4T1 cell monolayers were rinsed with PBS and 500 μL of cold deionized H 2 O was added to each well. Then, plates were frozen for 20 minutes at

ADA2 activities in cell cultures
After reaching confluence at 24-well culture plates, H5V, 4T1, E0771 LA (less aggressive murine breast cancer cell line originally KUTRYB-ZAJAC ET AL. | 5943 obtained from C57Bl/6J mice), E0771 MA (more aggressive murine breast cancer cell line originally obtained from C57Bl/6J mice), PM (murine peritoneal macrophages isolated as describe earlier 16 ), human breast cancer cell lines (MDA-MB-231, T47D and MCF-7), HAEC (human primary aortic endothelial cells), SC (human monocyte/ macrophage cell line) were rinsed with PBS and 1 mL HBSS was added to each well. Then, to measure total adenosine deaminating activity, 50 μmol/L adenosine was added and samples were collected after 0, 5, 15 and 30 minutes of incubation at 37°C and analysed with HPLC as described earlier. 16 To measure ADA2 activity, above assay was conducted in the presence of 10 μmol/L EHNA. ADA1 activity was calculated by subtracting the activity of ADA2 from a total adenosine deaminating activity. Cell residue was dissolved in 0.5 mol/L NaOH, and protein concentration was measured with a Bradford method according to the manufacturer's protocol. The results of cell-surface ADA1 and ADA2 activities were expressed as the inosine increase over the time (μmol/min/g of protein).

| Determination of total (intra-and extracellular) ADA1 and ADA2 activities in cell cultures
After reaching confluence at 6-well culture plates, H5V, 4T1,

| Determination of the effect of dCF on intracellular nucleotide and metabolites in cultured cells
After reaching 80% of confluence at 24-well culture plates, H5V and 4T1 cells were serum-starved overnight and treated for 24 hours with dCF (0, 5, 50, 150, 500 nmol/L) in a serum-free medium. After

| Adenosine deaminase inhibition by dCF suppressed progression of 4T1 murine breast cancer in vivo
To determine the effect of dCF on breast cancer progression in vivo, low dose of dCF (0.2 mg/kg) was administered intraperitoneally every 72 hours starting from the 1st day after orthotopic 4T1 cell injection into BALB/c mice ( Figure 1A). Tumour volume was measured every 48 hours starting from the 14th day after cancer cell implantation. Based on our previously published study, this treatment protocol maintained the activities of plasma adenosine deaminase and vascular ecto-adenosine deaminase at level lower than 20% throughout the duration of the experiment. 16 4T1 cancer cell implantation and dCF treatment did not affect mouse weight ( Figure S1). However, dCF treatment significantly decreased tumour growth ( Figure 1B) and final tumour mass (Figure 1C). The area of tumours measured using two-dimensional USG on the last day of the experiment was lower after dCF treatment ( Figures 1D-F), while the percentage area of tumour necrosis was higher ( Figure 1G).

| dCF had a minor effect on immune system stimulation exerted by 4T1 cell implantation
4T1 cancer cell orthotopic injection and dCF administration did not change red blood cell parameters and platelet count (Table 1) in blood morphology. However, 4T1 cancer cell inoculation significantly stimulated immune system. The spleen mass ( Figure 1H) was six times higher in 4T1-injected mice than in controls. Similarly, white blood cell count (WBC) reflected stimulated immune system after 4T1 cell implantation ( Table 1). The composition of WBC populations was also affected in 4T1-injected mice, primarily in terms of a significant increase in the total percentage of granulocytes as well percentage of their immature forms (Table 1). dCF treatment had a minor effect on immune system stimulation exerted by 4T1 cell implantation, including spleen mass or WBC parameters and did not change the composition of WBC populations. Moreover, 4T1 cancer cells implantation and not dCF administration increased blood concentration of ATP and ADP as well as ATP/NAD ratio that could derive from circulating cancer cells or increased count of blood cells, especially immune cells (Table 2).

| dCF decreased total and cell-surface adenosine deaminase activity
To evaluate the effect of 28-day-long dCF treatment on vascular adenosine deaminase activity in thoracic aorta of 4T1-injected and control mice, total (intracellular and extracellular) ADA activity was measured in tissue homogenate, while extracellular ADA activity  Similarly, dCF decreased about 60% of total adenosine deaminase activity in tumour homogenates ( Figure 1K).  Figure 2A) that is highly released from activated endothelial cells. 23 Other metabolites such as symmetric dimethyl L-arginine (SDMA) ( Figure 2B) and N-monomethyl L-arginine (L-NMMA) ( Figure 2C) that are produced to a lesser extend or do not inhibit eNOS remained unchanged. 4T1 cancer cell implantation also did not affect plasma L-arginine concentration (Figure 2D), which results in a higher ADMA/L-arginine ratio ( Figure 2E), another plasma marker of endothelial dysfunction that strongly correlates with endothelial vasodilator function. 24 dCF treatment reduced both ADMA concentration and ADMA/L-arginine ratio induced by 4T1 cancer cell injection (Figures 2A, 2E).

| dCF counteracted endothelial dysfunction stimulated by 4T1 cell orthotopic implantation
While, dCF administration alone in control mice not injected with cancer cells did not affect endothelial function based on these parameters.

| Intravenous injection of 4T1 murine breast cancer cells stimulates the increase in vascular adenosine deaminase activity that precedes endothelial dysfunction
Intravenous (iv.) injection of 4T1 cancer cells was administered according to the protocol ( Figure 3A) to verify the endothelial-T A B L E 1 Blood morphology in analysed experimental groups of mice 28 d after orthotopic inoculation of 4T1 tumour cells The concentration of nucleotides and their catabolites in venous blood of BALB/c mice 28 d after orthotopic injection of PBS (control, n = 5) or 0.2 mg/kg dCF (dCF, n = 5) or after orthotopic injection of 4T1 cancer cells treated with PBS (4T1, n = 7) or 0.2 mg/kg dCF (4T1 + dCF, n = 7) every 3 d for 28 d of the experiment. Data are presented as mean ± SEM, *P < 0.05 vs control, **P < 0.05 vs dCF by one-way ANOVA followed Holm-Sidak post hoc test.
protective effects of dCF. No changes were observed in the blood morphology and blood nucleotide concentration after cancer cell iv. injection (Tables S1-S4). The increase in vascular total and extracellular ADA activity was observed after 2 days of 4T1 cell injection ( Figure 3B). While, 21 days after cancer cell administration, this change was compensated ( Figure 3C). dCF inhibited ADA activity by about 50% as compared to its activity after 2 days of cancer cell injection, without affecting its activity compared to control values ( Figure 3B). After 21 days of 4T1 cell administration, dCF effectively inhibited vascular ADA activity by about 70% compared to both PBS-(control) and 4T1-injected (4T1) mice (Figure 3C). Endothelial function parameters were not affected 2 days of cancer cell injection ( Figure 3D) but both, plasma ADMA concentration and ADMA/L-arginine ratio, were increased when measured 21 days after 4T1 cell administration ( Figure 3E). dCF treatment prevented the increase of these parameters ( Figure 3E).

| dCF decreased activity of total and ectoadenosine deaminase on the surface of murine endothelial cells and breast cancer cells
The efficiency of adenosine deaminase inhibition by dCF was also evaluated in murine 4T1 breast cancer cells ( Figure 4A) and murine H5V heart endothelial cells ( Figure 4B). Low concentrations of dCF

| dCF decreased an invasive phenotype of 4T1 cells in vitro
The F I G U R E 4 Murine endothelial cells express much higher activity of intracellular and cell-surface adenosine deaminase activity than breast cancer cells and deoxycoformycin decreases these activities in both types of cells. The activity of total adenosine deaminase (tADA = intra-and extracellular ADA) in cell extracts and ecto-adenosine deaminase (eADA) on the cell surface of murine breast cancer cell line (4T1, A) and murine heart endothelial cell line (H5V, B). Data are presented as mean ± SEM, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 by one-way ANOVA followed Holm-Sidak post hoc test

| dCF decreased migration and invasion of human breast cancer cells in vitro
To confirm the role of dCF on human breast cancer development,  Figure 6B).

| Endothelial and immune cells express higher adenosine deaminase activity than cancer cells in murine and human specimen
To investigate the cellular origin of ADA in carcinogenesis, we measured its activity in different types of murine and human cells Murine cells are devoid of ADA2 isoform; therefore, its analysis was excluded from experimental studies on the mouse model. 25 However, the measurement of both ADA1 and ADA2 activities could be particularly important in human cells. As ADA2, unlike ADA1, is resistant to the inhibition by EHNA, we used this inhibitor to differentiate the sum of ADA1 and ADA2 activity (assay without EHNA) from ADA2 activity (assay with EHNA). ADA1 activity was then estimated subtracting ADA2 from total adenosine deaminating activity. In this study, we confirmed that all analysed mouse cells do not have both intra-and extracellular ADA2 activity ( Figure 7B-F).
While, human cells expressed some activities of ADA2 isoform only intracellularly ( Figure 7B-F). The dominant population of human cells that was responsible for the origin of ADA2 were monocytes/macrophages ( Figure 7G), but also less aggressive cancer cells (MCF-7) showed slight activity of ADA2 ( Figure 7C).The remaining ADA1 activity inside the cells and the entire eADA activity in analysed human cells were catalysed by ADA1 ( Figure 7C-G).

| DISCUSSION
The Several lines of evidence highlight the importance of adenosine as a crucial regulatory autocrine and paracrine factor that accumulates in the tumour microenvironment. 26 The dominant pathway leading to high extracellular adenosine levels is the extracellular phosphohydrolysis of ATP by ecto-nucleotidases, CD39 and CD73. 27 ATP at high concentrations accumulates in the tumour microenvironment as a danger signal and a proinflammatory mediator. 26 In turn, adenosine regulates its receptors on immune cells that are altered in tumours, which thereby switches immune surveillance and host defence to promotion of cancer transformation and growth. 28 Adenosine pathway also regulates cancer growth and dissemination by interfering with cancer cell proliferation, apoptosis and metastasis, but depending on adenosine receptor subtype stimulation on neoplastic cells, it has disparate effects. 28 In most cancers, the activation of A1, A2a and A2b receptors induces tumour cell proliferation, while A3 receptor stimulation limits this process. In turn, the studies on A2a, A2b and A3 receptors revealed that adenosine stimulates apoptosis through all these receptors in caspase-dependent or independent manner. 28  which is a key point in the cell movement during tumour growth. 35 Moreover, the effect of the dCF on tumour invasion and migration was also clearly visible in more invasive human breast cancer cells, MDA-MB-231 and T47D. It has been shown previously, that this could be a receptor-independent effect mediated by adenosine via inhibition of AMP-activated protein kinase. 36 However, based on our study, the significant role of adenosine receptors in these processes should also be noticed. Other antitumour mechanism of dCF could be related with the modulation of angiogenesis, an important aspect in tumour growth. Even though adenosine is known as a proangiogenic factor, 37 dCF treatment increased the percentage of tumour necrosis while reducing tumour size in 4T1 mouse model, suggesting down-regulation of tumour vascularization. As the effect of dCF on endothelial cell migration in vitro was minor, other antiangiogenic mechanisms should be considered, for example the inhibition of macrophage induced angiogenesis. 38 As there are only a few reports on ADA activity in breast cancer patients and unclear data about its origin, we also investigated the sources of ADA activity in murine and human cells engaged in tumour development. The main ADA isoform found in all analysed cells, including endothelial cells, immune cells and breast cancer cells was ADA1. We have demonstrated endothelial cells and monocytes/ macrophages as a major source of intra-and extracellular ADA activity in both species. As rodents are devoid of ADA2 isoform, we further investigated, which populations of human cells expressed ADA2 activity. 25 Human monocytes/macrophages showed the highest ADA2 activity, which was found only inside the cells. There was no ADA2 activity on the surface of all analysed cell types. Therefore, circulating monocytes and tumour-associated macrophages are most likely responsible for the increased serum ADA2 activity in patients with breast cancer. 12 It has been previously reported that the most of adenosine deaminase that is released from monocytes and macrophages is found in plasma as a soluble form. 39 Moreover, the role of ADA2 activity in human carcinogenesis could be particularly important, as it induces M2 macrophage polarization that is a critical point in reprogramming the immunosuppressive microenvironment and then promoting tumour progression. 40 Although the mice are lacking ADA2 activity, we demonstrated that ADA1 inhibition could be an effective antitumour therapy associated with the endothelial protection and reduction in the invasive potential of tumour cells. Based on our in vitro characteristic of the rates of adenosine deamination in different cell populations, the low-dose dCF treatment should be able to exert above anticancer effects, by complete inhibition of the ADA activity in cancer cells and only mild suppression of its activity in endothelial and immune cells, without any toxic effects. While our studies on the antitumour potential of ADA activity inhibition have been performed in mice that lack ADA2, which plays a key role in the abolishment of immune surveillance leading to cancer progression, dCF is known to affects both ADA isoenzymes. Therefore, beneficial effects we found in mice will not be depreciated in humans due to ADA2 activity.

| CONCLUSIONS
The present study highlights low-dose adenosine deaminase inhibition as a potential therapeutic strategy for breast cancer. Despite confusing reports on the role of adenosine in the breast cancer development, 41 we determined the effects of suppressed degradation of adenosine on improved function of endothelial cells and their barrier function in breast cancer models as well as on cancer cell migration, invasion, adhesion and transmigration through the endothelial cell layer. Furthermore, this study introduces possible mechanisms of the control of breast cancer growth by adenosine deaminase inhibitor via adenosine receptor-dependent and independent processes.

CONFLI CT OF INTEREST
Not declared.