Mafosfamide, a cyclophosphamide analog, causes a proinflammatory response and increased permeability on endothelial cells in vitro

Post-transplantation cyclophosphamide (PTCy) has decreased GVHD incidence. Endothelial damage in allo-HCT is caused by multiple factors, including conditioning treatments and some immunosupressants, and underlies HCT-complications as GVHD. Nevertheless, the specific impact of PTCy on the endothelium remains unclear. We evaluated the effect of mafosfamide (MAF), an active Cy analog, on endothelial cells (ECs) vs. cyclosporine A (CSA), with known damaging endothelial effect. ECs were exposed to MAF and CSA to explore changes in endothelial damage markers: (i) surface VCAM-1, (ii) leukocyte adhesion on ECs, (iii) VE-cadherin expression, (iv) production of VWF, and (v) activation of intracellular signaling proteins (p38MAPK, Akt). Results obtained (expressed in folds vs. controls) indicate that both compounds increased VCAM-1 expression (3.1 ± 0.3 and 2.8 ± 0.6, respectively, p < 0.01), with higher leukocyte adhesion (5.5 ± 0.6, p < 0.05, and 2.8 ± 0.4, respectively). VE-cadherin decreased with MAF (0.8 ± 0.1, p < 0.01), whereas no effect was observed with CSA. Production of VWF augmented with CSA (1.4 ± 0.1, p < 0.01), but diminished with MAF (0.9 ± 0.1, p < 0.05). p38MAPK activation occurred with both compounds, being more intense and faster with CSA. Both drugs activated Akt, with superior MAF effect at longer exposure. Therefore, the cyclophosphamide analog MAF is not exempt from a proinflammatory effect on the endothelium, though without modifying the subendothelial characteristics.


INTRODUCTION
Cyclophosphamide (Cy) is an antineoplastic and immunomodulatory drug playing an active and well-established role in conditioning for allogeneic hematopoietic cell transplantation (allo-HCT) [1]. This agent, used before transplantation to avoid the engraftment rejection by inhibiting the host immune system, causes an increment of graft-versus-host disease (GVHD) risk following allogeneic T cell infusion in mouse models [2]. Nowadays, enough evidence indicates that the administration of posttransplant cyclophosphamide (PTCy) suppresses alloreactive T cells in vivo, and in consequence, graft rejection and GVHD are reduced [3,4].
The use of Cy few days after allo-HCT (days 3 and 4, postinfusion) is associated with better engraftment, low GVHD rates, and non-relapse mortality in haploidentical HCT. When this strategy is applied in combination with reduced intensity conditioning, it is effective in patients up to 75 years old, being equivalent to younger patient results [5]. The effect of PTCy has a potent toxic action on the most reactive T lymphocytes, without altering the less reactive ones that are still present to perform their regulatory function [6,7]. PTCy administration causes a "chimeric tolerance" where the recipient tolerates better the cells received from the donor (chimeric) [3,8].
GVHD is one of the most important early complications after allo-HCT, with high morbidity and mortality rates [9]. According to previous studies [10][11][12], GVHD and other early HCT complications, are related to the endothelial damage associated with HCT. The endothelium plays a crucial role in many physiological functions including the regulation of vasomotor tone, hemostasis, leukocyte trafficking, angiogenesis, and permeability [13]. Due to its location, the endothelial cells (ECs) are exposed to several physiological and pathological stimuli [14,15], and constitute the first barrier for many different conditions, including the drugs used during HCT. ECs may adapt to these conditions but, depending on the intensity and duration, they may become dysfunctional and constitute a real harm to the host. Therefore, the endothelium may be partly responsible for the development of HCT-related complications, including GVHD [16,17].
Endothelial cell activation and damage may be due to different factors, such as the conditioning regimens, cytokines and chemokines produced by the injured tissues, the engraftment itself, the allogenicity, and also some drugs used during the transplant procedure, like granulocyte colony-stimulating factor (G-CSF) [18] and immunosuppressant therapies [19,20]. Cyclosporine A (CSA) is an immunosuppressive drug used, combined with other compounds, as the standard GVHD prophylaxis in allo-HCT [21,22]. However, it has been demonstrated that CSA has a clear adverse effect on the endothelium [19,23,24].
The use of PTCy combined with other immunosuppressive drugs is effective at reducing the GVHD incidence after allo-HCT [25,26]. With this evidence, the present study aimed at comparing, in vitro, the effect of a Cy analog with direct activity, mafosfamide (MAF) [27,28], and CSA, due to its well-known proinflammatory and prothrombotic effect, on the endothelium. A better understanding of the action of Cy and other immunosuppressant compounds on the endothelium is important to find successful treatment strategies to prevent endothelial damage and dysfunction in association with HCT.

MATERIALS AND METHODS Experimental design
ECs were exposed to MAF or CSA. Each condition was compared to a control group (C), without drugs. Drug concentrations were selected based on previous studies: MAF (2 µg/mL) [28][29][30] and CSA (400 ng/mL) [19,31]. Cells were exposed to the compounds, separately, for 48 h or 7 days depending on the experimental approach. MAF is an active analog of Cy. It was used as an alternative to Cy since the later requires hepatic metabolism to be active and it would not be suitable to explore its effects in vitro.
The effects of the drugs on ECs were assessed by measuring changes in the expression of vascular cell adhesion molecule 1 (VCAM-1) on the cell surface; the adhesion of leukocytes on the cell monolayer, after being exposed to circulating citrated blood under flow conditions (300 s −1 , 10 min); presence of vascular endothelial cadherin (VE-cadherin) on the cell surface; and production of von Willebrand factor (VWF), using immunofluorescence (IF). Activation of intracellular signaling proteins was evaluated with specific antibodies to phosphorylated p38MAPK and phosphorylated Akt.

Leukocyte adhesion analysis
ECs were cultured on 18 × 18 mm 2 coverslips in six-well plates (VWR) in medium containing the compounds (48 h). Cells-coated coverslips were perfused with citrated whole blood from healthy donors in a parallel-plate perfusion chamber (shear rate of 300 s −1 , 10 min). After perfusion, ECs were fixed (4% paraformaldehyde), permeabilized (0.025% Triton X-100), incubated with a monoclonal antibody to CD45 (Ab40763, Abcam, Cambridge, United Kingdom) (4°C, ON) followed by a secondary antibody conjugated with Alexa Fluor-488 (Thermo Fisher Scientific/Invitrogen) (RT, 1 h). Nuclei and cell membranes were stained with DAPI and wheat-germ agglutinin (Molecular Probes, Eugene, OR, USA), respectively. Images were captured by a microscope (DM4000 B) equipped with fluorescence filters. Leukocyte attachment was quantified (ImageJ Fiji). Results were expressed as fold increases of the adhered leukocytes with respect to controls.

Statistical analysis
Data are reported as Mean ± SEM. Statistical analysis was performed with raw data following a normal distribution using the Student's t test for unpaired samples, comparing results from each group of experiments to controls. Results were considered statistically significant when p < 0.05. N = 5 for each experimental approach.

RESULTS
Proinflammatory effect of mafosfamide and cyclosporine A on endothelial cells VCAM-1 expression and adhesion of leukocytes to the cell surface under flow conditions were evaluated to investigate the proinflammatory effect of the drugs under study. Levels of VCAM-1 significantly increased in the presence of MAF (fold increase: 3.1 ± 0.3 vs. control, p < 0.01) and CSA (fold increase: 2.8 ± 0.6 vs. control, p < 0.01) (Fig. 1).
An increment of the percentage of leukocytes attached to EC monolayer was observed in the presence of MAF and CSA, but it was higher and statistically significant only with MAF (fold increases: 5.5 ± 0.6 (p < 0.05) and 2.8 ± 0.4, respectively, vs. control).

Mafosfamide altered the endothelial cell monolayer integrity
The expression of VE-cadherin was analyzed to proof the EC monolayer integrity. Levels of VE-cadherin were significantly decreased in response to MAF (fold increase: 0.8 ± 0.1 vs. control, p < 0.01). No effect on VE-cadherin expression was noticed in the presence of CSA (fold increase: 1.0 ± 0.1 vs. control) (Fig. 2).
The production of VWF by endothelial cells was enhanced in response to cyclosporine A but not to mafosfamide Expression of VWF in ECs exposed to MAF was slightly, though significantly, reduced (fold increase: 0.9 ± 0.1 vs. control, p < 0.05), whereas it was significantly higher when the cells were grown in the presence of CSA (fold increase: 1.4 ± 0.1 vs. control, p < 0.01) (Fig. 3).
Cyclosporine A and mafosfamide increased the phosphorylation of p38MAPK and Akt Levels of p38MAPK phosphorylation were significantly higher when EC monolayers were exposed to both drugs separately for 15 min in comparison to the control. In presence of MAF, the phosphorylation level of p38MAPK was in crescendo with the exposure time (fold increases: 1.2 ± 0.2 and 1.8 ± 0.3 (p < 0.05) at 5 and 15 min, respectively, vs. control). In presence of CSA, p38MAPK expression was also higher and faster when the exposure time was increased (fold increases: 1.6 ± 0.3 and 2.1 ± 0.3 (p < 0.05) at 5 and 15 min, respectively, vs. control) (Fig. 4).
The expression of phosphorylated Akt was superior but not significant when EC monolayers were exposed to both drugs separately for different times. In response to MAF, the increment of phosphorylated Akt levels was observed when the exposure time was increased (fold increases: 1.1 ± 0.1 and 1.4 ± 0.3 at 5 and 15 min, respectively, vs. control). In response to CSA, phosphorylated Akt expression was slightly more activated at shorter exposure time (fold increases: 1.2 ± 0.2 and 1.1 ± 0.2 at 5 and 15 min, respectively, vs. control) (Fig. 4).

DISCUSSION
GVHD remains a relevant complication after allo-HCT, in which endothelial activation and damage plays a crucial role. The use of PTCy for GVHD prevention has significantly decreased the incidence of clinically relevant GVHD. We aimed at exploring the effect of MAF, an active analog of Cy, on the endothelium comparatively with CSA, with a well-known damaging effect. The results from the present study demonstrate that both compounds induced a proinflammatory effect on the endothelium, promoting leukocyte adhesion on the cell surface, and activation of the inflammation related protein p38MAPK and the cell-stress related protein Akt. Contrarily to CSA, MAF decreased VE-cadherin expression, and had a minimal effect on the production of the proadhesive protein VWF. These results indicate that MAF increases cell permeability but does not induce a prothrombotic phenotype on the endothelium. Allo-HCT remains the most extended immunotherapy for various hematological and non-hematological diseases [33,34]. Some changes in allo-HCT process provoked a better response in the last years, mainly, considering the adjustment of treatment intensity in conditioning regimen and the T lymphocyte attenuation of the graft and patient. These advances in the field of HCT have been essential to lead to an increase in the recipient age, reduce HCT complications and expand the donor availability for transplantation including non-totally compatible donors [35]. In these cases, the excessive alloreactivity caused by host and donor T cells leads to significant risks of graft rejection and severe GVHD [36,37]. While the total depletion of T lymphocytes reduces the risk of severe GVHD, the immunity to infections and the malignancy is seriously compromised [1]. Therefore, attenuation of lymphocytes without completely suppressing them can result in a decrease in the most prevalent complications and can also reduce the relapse of the disease [5]. The control of the alloimmune response of the graft towards the receptor leads to a reduction of both the GVHD incidence and the immunosuppression duration. All these facts together allow a more efficient immune reconstitution and also a better control of the disease.
Cy, which is an alkylating compound, was introduced as an antitumor agent in 1958 [30]. It is used to treat different malignant diseases and autoimmune disorders and included as part of the preparative regimens in allo-HCT [38]. PTCy prevents GVHD by inducing alloreactive T cell dysfunction and suppression [39]. However, its direct effect on the endothelium has not been explored before. In this regard, the use of Cy in the setting of allo-HCT, as part of the conditioning regimen, has been associated with rates of cardiac toxicity that range from 1 to 17% [40,41]. It is feasible to assume that the early stages of cardiovascular injury may be related to endothelial damage [42], potentially caused by the administration of Cy and other compounds during the HCT.
To explore the potential effect of Cy on the endothelium, we exposed ECs in vitro to an active analog such as MAF. MAF was used in the context of autologous bone marrow transplantation in the past, for ex vivo chemical purging to eliminate leukemic cells from suspensions of healthy bone marrow [43,44]. However, it also exhibited a cytotoxic effect on non-tumoral cells, compromising the engraftment [45]. To overcome MAF toxicity, amifostine was used as a protective compound [46,47], exhibiting a dual role since it induced sensitization of the leukemia cells while protecting the normal marrow progenitor cells [48]. In the present study, we have exposed ECs to MAF at a concentration that did not induce toxicity.
Cy, or the metabolites resulting from its hepatic metabolism [6], seem to be connected to endothelial damage [49]. Zeng et al. [50] Fig. 4 Differential activation of the intracellular signaling proteins p38MAPK and Akt in ECs exposed to mafosfamide or cyclosporine A, separately. ECs were exposed to mafosfamide and cyclosporine A at different times (5 and 15 min). Images are representative of western blotting experiments. The differential expression of p38MAPK and Akt was measured and expressed by bar diagrams as fold increase vs. control (dashed line) (N = 5, being *p < 0.05 vs. control).

Control
Mafosfamide reported elevated circulating ECs in mice treated with Cy and methotrexate, suggesting that both chemotherapeutic drugs could be responsible for endothelial injury during HCT process.
In agreement with those results, we demonstrate that MAF induces a proinflammatory effect on the endothelium, with increased expression of adhesion receptors at the cell surface and phosphorylation of both p38MAPK and Akt, related to inflammation and cell stress. The present results are in concordance with similar findings [51,52]. Regarding VE-cadherin, an adhesion molecule localized at ECs junctions that regulates the vascular migration and permeability [53,54], its expression is downregulated by MAF. As a consequence, vascular permeability may be increased, which may help the extravasation of circulating cells through the endothelial monolayer. This process may promote stem cell homing during HCT [55]. On the other hand, our results showed that MAF treatment decreased VWF production by ECs, finding also reported by Lau et al. [56]. Therefore, the prothrombotic effect previously attributed to CSA [19] was not found with MAF.
CSA is a calcineurin inhibitor used as an immunosuppressant drug to prevent GVHD [57]. It has been extensively administered to inhibit the release of proinflammatory cytokines, the activation and proliferation of lymphocytes and the consequent GVHD development [58,59]. Nonetheless, there is evidence demonstrating that CSA promotes endothelial dysfunction, with increased expression of adhesion receptors at the cell membrane and enhanced leukocyte attachment [19,31,60], aggravating the effects of the irradiation induced in HCT on the endothelium [61]. In addition, Carmona et al. [19] also reported a prothrombotic effect of CSA, enhancing platelet deposition. Our present results, showing increased production of VWF in response to CSA, correlates with higher platelet adhesion. Therefore, CSA causes both a proinflammatory and prothrombotic phenotype in ECs. Of note, calcineurin inhibitors have been associated with the development of TA-TMA, which is an allo-HCT related endotheliopathy [62][63][64].
In summary, our present results indicate that MAF and CSA, both drugs used separately, activate the EC monolayer conferring a proinflammatory phenotype. However, they differ in other in vitro effects on ECs: CSA exhibits a prothrombotic phenotype, while MAF does not modify the production of the proadhesive protein VWF; also, only MAF alters endothelial tight junctions. Nevertheless, there is not a clinical correlation with these laboratory findings since patients receiving PTCy do not show a clinical phenotype of capillary leak syndrome. Moreover, despite having demonstrated the deleterious effect towards the endothelium of both compounds separately, it is not likely that they produce a synergistic effect on patients, since when both are administered concomitantly there are not higher rates of HCT endothelial related complications. The widespread adoption of PTCy, as another immunosuppressant compound used in allo-HCT, has shifted the high risk of GVHD in the last decade. MAF, an analog of Cy, is not exempt from endothelial activation, but appears to be less aggressive than CSA. Further studies are needed to clarify the precise mechanisms of action of Cy.