Tumour suppressor PTEN activity is differentially inducible by myo‐inositol phosphates

Abstract Tumour evolution and efficacy of treatments are controlled by the microenvironment, the composition of which is primarily dependent on the angiogenic reaction to hypoxic stress. Tumour angiogenesis normalization is a challenge for adjuvant therapy strategies to chemo‐, radio‐ and immunotherapeutics. Myo‐inositol trispyrophosphate (ITPP) appears to provide the means to alleviate hypoxia in the tumour site by a double molecular mechanism. First, it modifies the properties of red blood cells (RBC) to release oxygen (O2) in the hypoxic sites more easily, leading to a rapid and stable increase in the partial pressure of oxygen (pO2). And second, it activates the endothelial phosphatase and tensin homologue deleted on Chromosome 10 (PTEN). The hypothesis that stable normalization of the vascular system is due to the PTEN, a tumour suppressor and phosphatase which controls the proper angiogenic reaction was ascertained. Here, by direct biochemical measurements of PTEN competitive activity in relation to PIP2 production, we show that the kinetics are complex in terms of the activation/inhibition effects of ITPP with an inverted consequence towards the kinase PI3K. The use of the surface plasmon resonance (SPR) technique allowed us to demonstrate that PTEN binds inositol derivatives differently but weakly. This method permitted us to reveal that PTEN is highly sensitive to the local concentration conditions, especially that ITPP increases the PTEN activity towards PIP3, and importantly, that PTEN affinity for ITPP is considerably increased by the presence of PIP3, as occurs in vivo. Our approach demonstrates the validity of using ITPP to activate PTEN for stable vessel normalization strategies.

get conditioned to such stress and participate in hypoxic tumour expansion and aggressiveness. [7][8][9] Angiogenesis is the first response to hypoxia by the HIF-1 and -2 transcription in the tumour cells, leading to the synthesis of proangiogenic factors responsible for typical pathologic tumour angiogenesis, which is the first hallmark of cancer and from which other hallmarks result. 8,9 Although aiming to compensate for the low partial pressure of oxygen inside the tumour, hypoxic stress-induced angiogenesis is inefficient due to the overproduction by the tumour cells of vascular endothelial growth factor (VEGF) along with other factors. [10][11][12][13] The newly formed pathological vessels not only are unable to sustain the blood flow but also favour the dissemination of hypoxia-and acidresistant tumour cells. 10 Moreover, it has been widely documented that hypoxia-conditioned factors signal the recruitment of immunosuppressive cells or modulate the phenotype of immune cells in the microenvironment, making them more tolerant to tumour cells. 14,15 Consequently, the challenge is to make the tumour vessels functional to instal a blood flow that allows for the delivery of treatment molecules 11,16 and allows the red blood cells to increase the pO 2 , thus, alleviating hypoxia. 17,18 It is therefore necessary to disrupt the equilibrium of haemoglobin-mediated O 2 release, allowing the VEGF pathway to be downregulated as it is responsible for turning on and maintaining the proangiogenic state. The pO 2 increase in the vessels, due to the dissociation of the red blood cell carried-oxyhaemoglobin, provides the signals for primary hypoxia alleviation/vessel normalization. 19,20 As shown, such an effect can be reached transiently by antiangiogenic strategies aiming to eliminate the proangiogenic factors overexpressed in the tumours. The limitations were mainly due to the poorly applicable controls of such active molecules that lead to the destruction of the vessels and the deleterious selection of aggressive tumour stem-like cell populations. 10,21,22 The direct modification of the properties of haemoglobin (Hb) by an allosteric effector makes it possible to shift its dissociation curve, allowing for the easier and complete release of O 2 in equilibrium with the pO 2 external to the erythrocytes. [23][24][25] Myo-inositol tris pyrophosphate (ITPP) was shown to operate this pO 2 increase upon the direct modification of the RBCs. 24,26 Its use in the angiogenesis-based treatment of hypoxia-dependent diseases such as cancer 19 or heart failure 27,28 has shown that its long-term administration led to stable vessel normalization. This has resulted in the hypothesis being propounded according to which its action is dependent on phosphate-linked pathways. 19 The most significant is the action of the phosphatase and tensin homologue mutated on chromosome 10 (PTEN), mutated in multiple advanced cancers. 29,30 PTEN is the main tumour suppressor the activation of which controls the PI3K/AKT/mTOR pathway in the cytoplasm and the nucleic p53 suppressive activity. 31 Moreover, the significance and potential of PTEN as a key suppressor of tumour growth is underlined by its major role in orchestrating angiogenesis progression. More precisely, PTEN controls the NOTCH4-mediated regulation of the tip-stalk organization of endothelial cells to build normal vessels. 32 In view of the necessarily active PTEN in normal vessels, it makes it a highly potent candidate for a vessel normalization target. 20,32,33 Besides the effect of ITPP-charged RBCs on angiogenesis in vitro, resulting directly from the release of oxygen under shear stress, 26 the ITPP molecule does activate the endothelial PTEN. 19 In vivo, upon ITPP injection, both a rapid pO 2 increase inside the tumour, as well as a long-term bioavailability of ITPP in the blood, were observed in the treatment protocols. This indicates that such an approach may lead to stable vessel normalization, offering an adjuvant strategy for the treatment of the numerous hypoxia-dependent diseases. 19 Tumour reaction analysis demonstrated that ITPP-mediated vessel normalization in tumours can modify the microenvironment so strongly that the cells expressing PD-L1 and its level of expression were downregulated. Likewise, CTLA4 and CD47 were reduced, while PD1-expressing cells increased. 20 The Tregs were less present and the M1 macrophage phenotype increased. NK cell entry into the tumour mass was shown to be regulated by the level of PD-L1 expression on the tumour vessel endothelial cells, which is induced in hypoxia but down-regulated upon hypoxia alleviation and PTEN activation. 20 As immunotherapy is the actual main purpose for cancer treatment, strategies that favour the immune response are actively searched for, with a huge need for methods that would be independent of the inhibitory monoclonal antibodies specific for either PD1 or PD-L1 that unfortunately do not treat the causal mechanisms of their induction; thus, this limits the impact of their use. 34 Our approach allows for the modification of the tumour microenvironment by taking advantage of the Hb allosteric properties and their enhancement by ITPP for a complete O 2 release in order to, on the one hand, alleviate hypoxia in the pathological sites and, on the other hand, for its long-term action on endothelial cell PTEN activation to maintain the vessels in their normal state, ensuring a prolonged blood flow efficacy. Consequently, the oxygen level inside the tumour is continuously maintained in physioxia. This breaks the vicious circle in which pathologic angiogenesis is responsible for keeping a tumour hypoxic, demonstrating that PTEN activity in the endothelial cells is critical. 5,35 Remarkably, such an approach presents the main advantages of PTEN expression and activity being independent of the properties of tumour cell. This strategy is addressed to the endothelial cells that perform angiogenesis, which are external normal cells recruited by hypoxia-responding tumour cells. Indeed, vessel normalization by ITPP only deals with the normal endothelial PTEN to activate it.
Consequently, ITPP treatment allows for tumour therapy regardless of the possible mutations of the PTEN tumour suppressor gene.
Moreover, considering the numerous ways in which PTEN can exert its tumour suppressor effect, ranging from the direct long form of soluble PTEN interacting with PTEN − cells to the exosome-specific microRNA (miRNA) modulation, the proposed strategy of its activation possesses a huge potential for improving the efficacy of antitumour approaches. 36 In view of this, the present work was undertaken to decipher whether the direct recognition of ITPP by PTEN occurs, if such an interaction can be specific compared to ITPP with its metabolite, the bisphosphate bis-pyrophosphate (BPBPP) and the inositol hexaphosphate (IHP) with the non-cyclic form of phosphate residues. The biochemical assessment of the modulation of PTEN activity by those molecules was compared to the similar modulation of PI3K, the kinase that produces PIP3 to activate the proliferation pathway, which showed an interdependency of both the phosphatase and the kinase with regards to the action of ITPP in a highly concentration-dependent manner.
Moreover, the biological reaction of endothelial cells to the process of hypoxia/reoxygenation evidenced that the activation of PTEN is greatly assisted by ITPP and involves the AKT activation pathway.
Direct evidence of PTEN/ITPP recognition and binding was found here using the plasmon resonance technique to assess the binding characteristics of possible PTEN substrates. It allowed the direct but weak binding of ITPP and BPBPP to be displayed by the enzyme, as compared to the natural substrate PIP3, and a high increase of this binding when PIP3 and ITPP interact with PTEN.
Based on the present data, ITPP and BPBPP enhance PIP3 binding by PTEN specifically, and such an activation effect is highly concentration-dependent.
When experiments were performed in conditions mimicking the intracellular environment where PTEN is in the presence of PIP3, we observed a much more efficient association for ITPP which was able to bind strongly to a PTEN-PIP3 preformed complex, similar to the in situ situation. The data indicate that ITPP provides a potential mean to induce PTEN activation and control cell growth in hypoxic pathologic environments.

| Effect of ITPP treatment on the activation of endothelial PTEN
Frozen sections of murine melanoma tumours treated as previously described 19 were labelled by anti-PTEN (rabbit IgG) (Cell Signalling) and mouse anti-CD31 (rat monoclonal IgG2a) (eBiosciences) before detection by tetramethylrhodamine isothiocyanate and fluorescein isothiocyanate secondary antibodies, respectively. Nuclei were stained with bis-Benzimide H 33258 (Sigma-Aldrich).

| PTEN phosphatase activity-mediated phosphate production assessment towards ITPP, BPBPP and IHP
The malachite green-molybdate reagent was used for a colorimetric assay of pure PTEN phosphatase activity (MAK308, Sigma-Aldrich). The colorimetric assay was used to examine PTEN

| PTEN activity towards ITPP, BPBPP and IHP assessed by PIP2 production
PTEN biochemical activity towards ITPP, BPBPP and IHP was assessed by an ELISA competition assay using the lipid Phosphatase Activity Assay to quantify PTEN activity (ELISA kit K-4700, Echelon Biosciences, Salt Lake City, UT) following the manufacturer's protocol, designed to quantify the phosphatase activity of PTEN by the detection of the enzyme product, PI(4,5)P 2 . Reagents are added to a PI(4,5)P 2 -coated microplate, the concentration of which is detected by the PI(4,5)P 2 detector protein added for competitive binding and, in turn, detected by a peroxidase-
The enzyme, the PIP3 standards, and the controls were mixed and incubated with highly specific PIP3 binding and then transferred to a PIP3-coated microplate for competitive binding with the PIP3 detector protein, further detected by a peroxidase-linked secondary detector and quantitatively evidenced by a colorimetric reaction to estimate the amount of PIP3 produced. The quantitative colorimetric signal at 560 nm was inversely proportional to the amount of PI(3,4,5)P 3 produced by PI3K. Recombinant PI3K was allowed to react with PI(4,5)P 2 (5 × 10 −6 M) to produce PI(3,4,5)P 3 in the presence or not of ITPP, BPBPP or IHP (concentration range 0.1 × 10 −6 M to 100 × 10 −6 M).

| Real-time interaction analysis by surface plasmon resonance
The binding of different analytes (PIP3, ITPP, BPBPP, IHP) and their mixtures to recombinant PTEN immobilized on CM5 sensor chips (GE Healthcare) was analysed by surface plasmon resonance (SPR) using a BIAcore T200 system. The PTEN immobilization level corresponds to approximately 5500 RU. All the binding experiments were carried out at 25°C with a flow rate of 30 μL/min in HBS running buffer.
To determine the affinity of the analysed ligands and their mixtures to PTEN, at least five different concentrations of each as well as a sample buffer blank were passed over the ligand-immobilized chip surface (association phase), followed by dissociation with running buffer (HBS). Details of each set of interactions are provided in the table below ( Table 1)

| Statistical analysis
The data are presented as the mean ± SEM of at least three independent experiments The analysis was performed using GraphPad Prism software. Student's t-test or anova was used to assess the statistical significance. The data were presented as means ± SEM.

| Activation of PTEN upon in vivo treatment by ITPP
Microscopic analyses of the vasculature status in B16F10 melanoma tumour-bearing animals, 19 as presented in Figure 1A Figure 1B. After the ITPP treatment, according to the previously described protocol, 19,28 PTEN protein distribution was distinct from CD31 ( Figure 1C, D), which clearly indicated a PTEN activation state. 40,41 This effect correlates with the restructuring of the vessel morphology, which appears normalized upon ITPP treatment ( Figure 1C), as opposed to chaotic in the untreated tumours ( Figure 1A).

| Effect of myo-inositol phosphate derivatives on PTEN hydrolytic activity towards PIP3
3.2.1 | PTEN phosphatase activity towards ITPP, BPBPP and IHP Malachite green molybdate/orthophosphate complex formation measured after the PTEN hydrolytic reaction on PIP3 was performed in the presence or absence of myo-inositol phosphate derivatives as possible modulators of the enzyme. 42 The formed complex was measured at 620 nm. The amount (nmol) of free phosphate released in each well was determined by linear regression analysis against a standard phosphate curve (Figure 2A).

| Effect of myo-inositol phosphate derivatives on PTEN phosphatase activity towards PIP3 and PIP2 production
The PTEN-mediated hydrolysis of PIP3 into PIP2, measured by a competitive ELISA assay assessing the binding of anti-PIP2 antibodies, is presented on Figure 3. This method was used to avoid the potential participation of phosphate residues from myoinositol derivatives in the above-described green molybdate phosphate    Figure 3D reveals the different mode of action of BPBPP, which shows a partial inhibitory effect. A total inhibitory effect appears for a 100 × 10 −6 M concentration of BPBPP.

| Endothelial cell PTEN modulation and phospho-Akt activation upon hypoxia/reoxygenation stress, the effects of ITPP and BPBPP
Since the effects of ITPP and its metabolite BPBPP appeared to operate on the activity of PTEN and PI3K differently, the elucidation of their mode of action required the investigation of their effect on the PTEN downstream signalling pathway. In order to assess whether ITPP and/or BPBPP act directly on the PTEN phosphatase regulatory pathway or primarily by PI3K inhibition, the regulation of the active form of Akt, that is phospho-Akt, was estimated by the p-Akt/ Akt ratio as a function of the reaction to the pO 2 variation, as shown in Figure 4A, B.
In normoxia, no clear effect of ITPP or BPBPP was observed, while in hypoxia, the endothelial cells from murine lungs MLuMEC,

| PIP3 binding characteristics to PTEN in vitro
To examine whether PIP3 binds to PTEN and how the analysed To evaluate the impact of ITPP/BPBPP/IHP and check its potential to modulate PIP3 and PTEN interaction, we analysed it using SPR, as described in the Materials and Methods section (Table 1).
An injection of PIP3 as a mixture with ITPP or BPBPP or IHP (see Table 1, #3) revealed a significant change in PTEN-PIP3 interactions ( Figure 5B-E). First, in the case of the PIP3 and ITPP (20 mM) mixture, the association phase was clearly affected and seemed to be a twostage process. What is more, the formed complex was highly stable and almost no dissociation was observed ( Figure 5B). This very slow dissociation rate of 5-8 × 10 −6 1/s, approximately 100 times slower than the pure PIP3-PTEN complex, resulted in a calculated K D in the nanomole range (20-30 nanomoles). Despite the visible similarities in the curvature of SPR sensorgrams for the interaction of PIP3 and PTEN in the presence of BPBPP, no clear kinetic data was obtained ( Figure 5C). As for IHP, there was no PIP3-PTEN interaction in the said mixture ( Figure 5D).
To determine the stage on which ITPP or BPBPP or IHP impact the PIP3-PTEN interaction, we pre-injected 20 mM of ITPP or BPBPP over an immobilized PTEN, followed by an injection of PIP3 as the analyte ( Table 1).
The pre-injection of 20 mM ITPP or BPBPP and subsequent analysis of the PTEN-PIP3 interaction did not bring any important changes in the curves and K D . (Figure 6A-D).
It is worth mentioning that a direct interaction between PTEN and ITPP/BPBPP can be detected in our condition but only above 20 mM ( Figure 6). Neither ITPP nor BPBPP were able to bind PTEN at a concentration below 20 mM; the interaction remained very weak with a K D below instrument detection level even in higher concentrations ( Figure 6E, F). IHP did not bind to PTEN at any investigated concentration ( Figure 6G).
Notably, when we investigated the binding of ITPP or BPBPP to

| DISCUSS ION
The complex and diverse molecular mechanisms by which the PTEN tumour suppressor is able to act make it a highly valuable target for therapeutic strategies. 30,33,35,36,43 Its actions are crucial as far as cell growth control is concerned, due to its upstream position in the growth activation pathway of PI3K/Akt/mTOR in the cytoplasm, as well as in tuning the p53 tumour suppressor in the nucleus. This goes together with its exocrine/paracrine regulatory effects on pathologic cells and the cells of the microenvironment, like the endothelial cells. Thus, PTEN activation presents a huge regulatory potential, which raises the need for a deeper knowledge to be gained of the PTEN reactions, mainly in view of its possible modes of activation. 31,35,36 In the case of PTEN, being the main controller of angiogenesis, its active form in the endothelial cells is crucial for tumour angiogenesis normalization 32 which, in turn, is vital to alleviate the reduced level of oxygen tension in the treatment of hypoxia-dependent diseases, such as cancer, diabetes and cardiac and neurodegenerative diseases. 18 As demonstrated by us, such activation is possible by myoinositol trispyrophosphate derivatives in the cancer, 19 but also in cardiac disease 28 Therefore, a biophysical approach using the plasmon resonance technique was necessary to directly analyse the affinity and recognition properties of PTEN with the phosphoinositide derivatives and their effects on its natural hydrolytic reaction towards PIP3. This led to the evidence that ITPP and its metabolites are, individually, weak ligands for PTEN, but they work relatively efficiently and get to reach strong binding characteristics when they are present in association with PIP3. This allows them to be utilized in vivo, especially in the case of ITPP whose K D upon interaction with PTEN was not measurable by SPR, but its range reached a nanomolar level after PTEN had reacted with PIP3.
Indeed, this reaction explains the activation of PTEN upon in vivo treatment by ITPP since it reproduces the in situ conditions. In the cell, PTEN is a phosphatase that maintains the PIP3 to PIP2 balance for activation signal control, thus, it is always in the presence of PIP3. Consequently, in natural in situ conditions, PTEN, in the F I G U R E 6 (A-D) PIP3 binding to PTEN is not affected by ITPP, BPBPP and IHP when they are preinjected. PTEN was immobilized on a CM5 chip and its interaction with (A) ITPP, (B) BPBPP and (C) IHP, followed by PIP3 and analysed using surface plasmon resonance (BIAcore T200). All the details have been explained in the Materials and Methods section. (E-H) ITPP, BPBPP and IHP interactions with PTEN analysed using SPR. PTEN was immobilized on a CM5 chip and its interaction with (E) ITPP, (F) BPBPP and (G) IHP and analysed by surface plasmon resonance using a BIAcore T200. All the details have been explained in the Materials and Methods section.
presence of PIP3, is capable of binding ITPP with a high affinity and this, in turn, promotes an increased affinity of PTEN for PIP3, as demonstrated herein.
Thus, the activity of PTEN in the presence of ITPP favours the production of PIP2, thus, reducing the action of PI3K. Moreover, our biochemical data show that ITPP is an inhibitor of PI3K, making its application in cancer cell growth control very promising. Moreover, these data also confirm the potential of in vivo ITPP injections for angiogenesis normalization, as PTEN is a vital molecule whose activity may ensure proper and organized angiogenesis.