Eryptosis Indices as a Novel Predictive Parameter for Biocompatibility of Fe3O4 Magnetic Nanoparticles on Erythrocytes

Fe3O4 magnetic nanoparticles (Fe3O4-MNPs) have been widely used in clinical diagnosis. Hemocompatibility of the nanoparticles is usually evaluated by hemolysis. However, hemolysis assessment does not measure the dysfunctional erythrocytes with pathological changes on the unbroken cellular membrane. The aim of this study is to evaluate the use of suicidal death of erythrocytes (i.e. eryptosis indices) as a novel predictive and prognostic parameter, and to determine the impact of Fe3O4-MNPs on cellular membrane structure and the rheology properties of blood in circulation. Our results showed that phosphatidylserine externalization assessment was significantly more sensitive than classical hemolysis testing in evaluating hemocompatibility. Although no remarkable changes of histopathology, hematology and serum biochemistry indices were observed in vivo, Fe3O4-MNPs significantly affected hemorheology indices including erythrocyte deformation index, erythrocyte rigidity index, red blood cell aggregation index, and erythrocyte electrophoresis time, which are related to the mechanical properties of the erythrocytes. Oxidative stress induced calcium influx played a critical role in the eryptotic activity of Fe3O4-MNPs. This study demonstrated that Fe3O4-MNPs cause eryptosis and changes in flow properties of blood, suggesting that phosphatidylserine externalization can serve as a predictive parameter for hemocompatibility assay.


Results
Fe 3 O 4 -MNPs synthesis and characterization. Fe 3 O 4 -MNPs were prepared by aging ferrous hydroxide gels at elevated temperatures. On the basis of TEM micrographs, Fe 3 O 4 -MNPs displayed a spheroid-like shape with a relatively uniform size. Meanwhile, its size distribution analysis confirmed that the diameter of Fe 3 O 4 -MNPs centered around 72.6 ± 0.57 nm (n = 1200), while 58.17% Fe 3 O 4 -MNPs were between 60 and 90 nm in diameter (Fig. 1A,B). Meanwhile, the average particle size of Fe 3 O 4 -MNPs was measured using DLS. Fe 3 O 4 -MNPs dispersed in water had a hydrodynamic size of 88.78 nm (Figure  Figure S 3A and S 4, after 3 h incubation the Fe 3 O 4 -MNPs at the concentration of 100, 800 and 1,600 μ g/ml resulted in 1.78%, 4.08% and 7.81% erythrocyte lysis, respectively. When the incubation time was extended to 24 h, the nanoparticles at the concentration of 100, 200, and 400 μ g/ml led 4.44%, 6.81% and 12.34% erythrocyte lysis (Figure S 3A,B and S 4). These results indicated that the hemolytic rate was affected by Fe 3 O 4 -MNPs in a dose-and time-dependent manner. Accordingly, the images of erythrocytes collected after incubation with Fe 3 O 4 -MNPs for 24 hours showed that hemolysis was visible (seen by unaided eye) when Fe 3 O 4 -MNPs reached a concentration greater than 400 μ g/ml (Figure S 3C).

Dose-and time-dependent hemolytic activity of Fe 3 O 4 -MNPs. As shown in
Moreover, the blood gas analysis showed that the standard bicarbonate, TCO 2 , and HCO 3 − were increased significantly after incubation with Fe 3 O 4 -MNPs, suggesting a decreased affinity of CO 2 (Table  S1). However, most of parameters, including pH, K + , Ca 2+ , Na + , Mg 2+ , pO 2 , and pCO 2 , changed slightly after treating with Fe 3 O 4 -MNPs at 200 μ g/ml for 48 hours.

Impact of Fe 3 O 4 -MNPs on RBCs morphology.
The externalization of phosphatidylserine can be quickly and reliably detected by fluorescent annexin V conjugates 37 , while no study has reported the use of annexin V for evaluation of Fe 3 O 4 -MNPs caused eryptosis. To evaluate Fe 3 O 4 -MNPs induced injury, phosphatidylserine exposure was quantified by flow cytometry and fluorescent imaging. When RBCs were exposed to 200 μ g/ml Fe 3 O 4 -MNPs, the number of annexin V-positive erythrocytes increased gradually after 6, 12, 24, and 48 h incubation ( Fig. 2A,C). As shown in Fig. 2A, the percentage of PS-displaying cells almost reached 40% at 48 h. And the percentage of PS-displaying cells was increasing along with the increase of Fe 3 O 4 -MNPs concentration (Fig. 2B,D). After exposure with Fe 3 O 4 -MNPs at the concentration of 25 μ g/ml for 24 h, 95% of the erythrocytes still had an intact cellular membrane; however, the percentage of phosphatidylserine exposing cells was statistically higher than that of control group.
Additional experiments were performed to elucidate the differences between hemolysis and eryptosis caused by Fe 3 O 4 -MNP exposure. The hemolytic grade was lower than 5% after incubation with the Fe 3 O 4 -MNPs at a dose of 100 μ g/ml for 24 hours. The corresponding percentage of PS-displaying erythrocytes was 15.1 ± 1.85%. Consistent with the notion of excellent hemocompatibility, Fe 3 O 4 -MNPs displayed remarkably low hemolytic activity. However, erythrocytes with annexin V positive were three times more than hemolytic erythrocytes (Fig. 2E,F). Annexin V binding is indicative of alterations in the cell surface characteristics, phospholipid organization, and membrane integrity. As a result, the injured cells are recognized, phagocytized, and degraded by macrophages 38 .  3D, red arrow) were much more apparent after 48-hour exposure. Subsequently, an obvious perturbation and curvature of the cellular membrane occurred after Fe 3 O 4 -MNPs exposure. These morphological characteristics reflected the changes occurring at metabolic levels, indicating that after incubation with Fe 3 O 4 -MNPs erythrocytes were more prone to Phorbol 12-myristate 13-acetate (PMA)-induced THP-1 cells phagocytosis (Fig. 3E,F). With phosphatidylserine externalization and membrane blebbing, RBCs deformability (represented by elongation index) decreased in a dose-dependent manner. When the concentration reached 50 μ g/ml, the elongation indices in most shear rates were significantly different from those of untreated group (Fig. 3G). Meanwhile, the concentration of cytosolic 2,3-Disphosphoglycerate, an indicator of RBCs oxygen delivering capacity, also showed a reduction in a dose-dependent manner (Fig. 3H).

Eryptosis induced by Fe
Reactive oxygen species (ROS) is known as an important trigger for eryptosis and ROS antagonist N-acetylcysteine (NAC), an antioxidant agent 39 , can block ROS-mediated eryptosis. Incubation with Fe 3 O 4 -MNPs resulted in significant increases in ROS production and increases in phosphatidylserine exposure. NAC efficiently inhibited Fe 3 O 4 -MNPs-induced ROS production and phosphatidylserine externalization (Fig. 4A,B,E). Ca 2+ influx is an important signaling mechanism leading to eryptosis. Since Fe 3 O 4 -MNPs exposure also led to high cytosolic Ca 2+ levels, blocking Ca 2+ entry by Ca 2+ -free Ringer solution remarkably decreased phosphatidylserine exposure ( Fig. 4C-E). It was observed that the inhibitory effect of Ca 2+ -free Ringer solution in combination with NAC was more effective compared to the use of NAC or Ca 2+ -free alone (Fig. 4F).
Mature erythrocytes lack the capacity to store Ca 2+ , and as a result, the Ca 2+ entry caused by Fe 3 O 4 -MNPs exposure can be readily blocked by Ca 2+ -free Ringer solution. However, Ca 2+ -free Ringer solution could not completely block phosphatidylserine externalization. The percentage of phosphatidylserine exposure in cells in Ca 2+ -free Ringer solution was 13.36% and 20.23% in the presence of Ca 2+ (Fig. 4B). Furthermore, hexavalent (VI) chromium, an eryptotic trigger causing Ca 2+ influx, resulted in observable Ca 2+ entry when exposed for 6 h. However, a detectable increase in cytosolic calcium appeared after Fe 3 O 4 -MNP exposure for more than 12 h, with the cells volume unaltered indicating by forward scatter (Fig. 4G-J). In addition, N-acetylcysteine inhibited ROS production, even with Ca 2+ entry Scientific RepoRts | 5:16209 | DOI: 10.1038/srep16209 (Fig. 4H,J). Altogether, these findings indicated that oxidative stress played a pivotal role in Fe 3 O 4 -MNPs induced eryptosis and the initial process of eryptosis involves calcium-independent phosphatidylserine externalization, while Ca 2+ entry also played an important role in this programmed cell death. In Ringer solution, ATP was consumed rapidly, regardless of whether erythrocytes were treated with Fe 3 O 4 -MNP incubation or not (Fig. 4I) and it was consistent with previous report 40 . Values represent means ± SEM, n = 9, **p < 0.01, ***p < 0.001.  . We found that Fe ion concentrations were higher than that in in vitro experiment until 6 h (Figure S 5). As compared with the control group, the Fe 3 O 4 -MNPs treated rats showed no obvious damage or inflammation in the major organs (i.e. Liver, spleens, kidneys, hearts, and lungs). The aggregation of Fe 3 O 4 -MNPs was mainly found in the livers and spleens, consistent with the normal distribution pattern of substances injected intravenously 41 . NAC administration did not change the distribution or the aggregation of Fe 3 O 4 -MNPs in the major organs (Fig. 5).
Compared with the control groups, Fe 3 O 4 -MNP injection significantly increased ROS production, consistent with the in vitro results (Fig. 6A,C). The percentage of PS-displaying erythrocytes was less than 1% in the control groups. With Fe 3 O 4 -MNP injection, this percentage increased to 2.48%, meaning that erythrocyte degradation and regeneration was accelerated roughly 2-fold ( Fig. 6B,D). NAC significantly inhibited Fe 3 O 4 -MNP-induced increases in ROS and phosphatidylserine exposure, although they were still higher than those of the control groups ( Fig. 6A-D). These results suggested that Fe 3 O 4 -MNPs injection in rats induced eryptosis in vivo via ROS production and NAC could alleviate the oxidative stress to some extent.
The majority of the clinical chemistry parameters were within the normal ranges and showed no difference between treated-and the control groups, suggesting limited systemic toxicity caused by the doses investigated in this study ( During hematology analysis, the majority of hematology markers relevant to erythrocytes displayed significant difference between the control group and the nanoparticle injected group, such as RBC counts, hemoglobin, mean corpuscular volumes, mean corpuscular hemoglobin, and red cell distribution widths (Figure S 7). The values of all of these markers were within the normal range except hemoglobin ( Figure S 7). Decreases of RBC counts, hemoglobin, and mean corpuscular hemoglobin indicated an excessive degradation of erythrocytes. There was also a decrease in mean corpuscular volumes and red cell distribution widths, which reflected a more uniform size and indicated that there were an enormous number of erythrocytes to be generated. However, NAC administration did not reverse changes of these erythrocyte-relevant markers. These results suggested that Fe 3 O 4 -MNP injection might lead to a mild to moderate risk for anemia.
The impact of nanoparticles on the flow properties of blood was evaluated by hemorheology. The hematocrit and plasma viscosities were not statistically significant. However, the indices related to the mechanical properties of erythrocytes differed significantly between the control and the Fe 3 O 4 -MNPs groups, such as erythrocyte electrophoresis time, RBC aggregation index, erythrocyte deformation index, erythrocyte rigidity index, and the viscosity of whole blood. The changes in RBC aggregation index and erythrocyte electrophoresis were inhibited by NAC administration (Fig. 7). These findings demonstrated that Fe 3 O 4 -MNPs treatment led to a decrease in deformability and changes in hemorheology, which would contribute to microcirculation disturbance and tissue damage by directly blocking capillaries and potential thrombogenesis 42 . Altogether, the results in vivo suggested that although there was a risk of anemia or thrombus formation, there was no obvious systemic toxicity risk after the administration of nanoparticles.

Discussion
Developing an assessment system, in addition to hemolysis, with sensibility, predictability, and reliability is critical to detect the cytotoxicity of Fe 3 O 4 -MNPs 43 . Currently, evaluation of the hemocompatibility of nanoparticles mainly relies on hemolysis analysis, which is not delicate enough to reflect the full spectrum of erythrocyte injuries. However, determination of the hematologic toxicity of Fe 3 O 4 -MNPs on alternations of cellular membrane structure and function of erythrocytes remains unexplored. Thus, a predictive parameter beyond hemolysis testing is highly desirable. In this study, we developed a reliable system with eryptotic indices to evaluate the erythrocyte compatibility with Fe 3 O 4 -MNPs and to determine hazards (e.g. anemia and thrombogenesis) associated with the MNPs even at safe doses. In the present study, our data demonstrated that eryptosis analysis was more sensitive in measuring hemocompatibility of nanoparticles for erythrocyte injuries, compared with currently used chemolysis for hemolysis. Currently, the safe and non-cytotoxic concentration is up to 100 μ g/ml based on hemolysis standard 44 . Even when the concentration of nanoparticles rises to up to 400 μ g/ml, the hemolysis rate of modified magnetic nanoparticles is less than 2% 43 . However, our data showed that Fe 3 O 4 -MNP caused phosphatidylserine externalization was statistically higher in the treated group than in the non-treated group when the dosage of nanoparticles is 25 μ g/ml in vitro. Previous studies have shown that phosphatidylserine externalization evaluation is more sensitive than hemolysis testing [30][31][32] . Eryptosis is presumably a physiological protective mechanism to eliminate injured or defective erythrocytes to forestall hemolysis and the release of hemoglobin 14,45 . Phosphatidylserine acts as a key role in cell cycle signaling, specifically in relationship to apoptosis. When exposed to Fe 3 O 4 -MNP in vitro, the percentage of phosphatidylserine exposed on the cell surface reached over 3-fold greater than the hemolysis rate in this study. Then the phagocytosis and subsequent degradation of cells with exposed phosphatidylserine are inevitable.
In vivo, accelerated eryptosis leads to erythropenia, anemia and impairment of microcirculation. Normally, the loss of eryptotic erythrocytes is compensated by stimulation of erythropoiesis 37 . However, the results showed that Fe 3 O 4 -MNPs exposure caused excessive eryptosis could not be fully compensated by erythropoiesis. This is because numbers of erythrocytes and the hemoglobin concentration were significantly lower in the Fe 3 O 4 -MNP treated groups compared with the non-treated group. Particularly, for hemoglobin concentrations, Fe 3 O 4 -MNP administration resulted in a sharp decline beyond the normal ranges, suggesting a significant risk for anemia. In hemorheologcal analysis, markers relevant to the mechanical properties of erythrocytes decreased with Fe 3 O 4 -MNP treatment. These outcomes can lead to a risk of ischemia or hypoxia caused by microcirculation blood hypoperfusion (Fig. 8) 46 . Interestingly, N-acetylcysteine administration could not alleviate the acceleration of eryptosis and erythrocyte degradation, although it may be useful to ameliorate defects in some mechanical properties, such as whole blood viscosity (WBV), RBC aggregation index (RAI), and erythrocyte electrophoresis time (EI).
Furthermore, evaluations based on eryptosis can better predict the fate of the erythrocyte in circulation and can also reflect the risks posed by damaged erythrocytes that disrupt circulation. Phosphatidylserine is restricted to the cytoplasmic leaflet of the plasma membrane. However, it was shifted to the outer cellular membrane induced by Fe 3 O 4 -MNPs exposure, where it is recognized by specific phagocytic receptors, such as T-cell immunoglobulin mucin receptor 4 (TIM4). Therefore, cells with exposed PS were rapidly engulfed and degraded by macrophages 47,48 . Systematic administration of Fe 3 O 4 -MNPs induced 2% of erythrocytes' phosphatidylserine externalization, which was lower that in vitro. It is attributed to one remarkable fact that the Annexin-V-FLUOS labeled erythrocytes represented only a fraction of eryptotic cells, since phosphatidylserine externalization is recognized and rapidly engulfed and degraded by macrophages. In fact, the percentage was over 2-fold greater and statistically higher than those of phosphatidylserine externalization in the control group. Membrane blebbing accounts for nearly 20% loss of membrane surface area, leading to a shape transformation and a reduction in deformability. Excessive blebbing accelerated the aging processes of erythrocytes 49 . ROS production resulted in the degradation of band 3 and spectrin and the opening of cation channels. It is critical for maintaining RBCs morphology, plasticity, and osmotic stability 50 . The increased cytosolic Ca 2+ affected the skeleton flexibility and stability, intracellular ion balance, and facilitates phosphatidylserine externalization 51 . Membrane blebbing, phosphatidylserine externalization, shape transformation and reduction in deformability contributed to the phagocytic uptake and mechanical properties changes, which were reflected by hematology and hemorheology alternation after the treatment of Fe 3 O 4 -MNPs. Furthermore, microvesicles generated by blebbing and externalized phosphatidylserine have procoagulant activity suggesting a hazard of thrombin generation 15,52 . Therefore, the acceleration of eryptosis induced by Fe 3 O 4 -MNPs should not be overlooked and warrants further investigation. Overall, some hallmarks of eryptosis may represent more sensitive and predictive parameter of toxicity compared with hemolysis.
Recently, the interaction between nanoparticles and erythrocytes has been investigated 25,53 . However, the possible mechanism underlining the Fe 3 O 4 -MNP induced eryptosis was still not clear. There are reports that the increases of intracellular ROS might cause the potential toxicity in eryptosis 37,54 . Oxidative stress induces activation of Ca 2+ -permeable nonselective cation channels. With Ca 2+ influx, erythrocytes suffered from PS externalization , cellular membrane blebbing, the loss of deformability, and clearance by macrophages. Our results confirmed that ROS production was the original and pivotal trigger of eryptosis. N-acetylcysteine, a ROS inhibitor, effectively blocked the eryptotic processes. Ca 2+ influx is also a powerful trigger of eryptosis. Inhibition of Ca 2+ entry relieved the phosphatidylserine externalization, consistent with previous studies 14,51,55 . However, the loss of membrane phosphatidylserine asymmetry was calcium-independent. The significant increase in cytosolic calcium concentrations occurred after ROS production, while Ca 2+ influx was remarkably influenced by oxidative stress. Consequently, with ROS generation and Ca 2+ entry promoted the eryptosis caused by Fe 3 O 4 -MNPs. There is mounting evidence that the impacts of nanoparticle exposure on erythrocytes include a series of subtle changes in ionic balance, energy metabolism, physiology and rheological properties, in addition to hemolysis. As a result, there are often conflicting data between hemolysis testing in vitro and blood-compatibility in vivo. The injured erythrocytes progress to eryptosis, lose capability of oxygen delivering, impeding microcirculation, and are cleared from circulating blood at last. These changes, as correlated with some specific clinical disorders, can be detected by eryptotic indices in vitro. Consequently, eryptotic indices seem to be reasonable prediction metrics for predicting how nanoparticles impact erythrocytes and blood circulation in vivo (Fig. 8).

Conclusions
In summary, we reported that 25 ug/ml Fe 3 O 4 -MNPs caused significant damage to erythrocytes in in vitro experiments and 12 mg/kg Fe 3 O 4 -MNPs lead to apoptosis of circulating erythrocytes in vivo. Erythrocyte injury of Fe 3 O 4 -MNPs can be divided into the early and later phages, eryptosis and hemolysis. Eryptosis does not induce acute hemolyzation characterized by hemoglobin releasing; it leads pathological alternations on cellular membrane and erythrocyte dysfunction. In this study, we demonstrated that Fe 3 O 4 -MNPs cause programmed cell death in erythrocytes with pathological changes on cellular membrane, abnormal cytosolic calcium levels, and oxidative stress and changes in the mechanical property of erythrocytes in vitro and in vivo. This study indicates that phosphatidylserine exposure, the index of eryptosis, can serve as a sensitive and reliable predictor for erythrocyte injury, while monitoring the nanotoxicity of the nanoparticles when systemically administrated. In addition, these metrics provide potential in determining the hazards of new types of nanoparticles or other biomaterials for clinical applications.

Erythrocyte preparation. This study was approved by the Medical Ethics Committee of The Second
Hospital affiliated with The Third Military Medical University. All procedures were conducted in accordance with the approved guidelines of the Medical Ethics Committee of The Second Hospital affiliated with The Third Military Medical University. All human subjects involved gave written informed consent. Leukocyte-free erythrocytes from healthy donors were used shortly after collection and were provided by the Chongqing Blood Centre. Hematocrit was adjusted to 0.4% with Ringer solution (125 mM NaCl, 32 mM HEPES, 5 mM glucose, 5 mM KCl, 1 mM MgSO4, 1 mM CaCl 2 , pH7.4), and all incubation condition of erythrocytes were 37 °C, 5% CO 2 , with 95% humidity. Calcium-free Ringer solution was made by using 1 mM EGTA as a substitute for 1mM CaCl 2 . Cell culture and cell uptake assay. THP-1 cells were obtained from the American Type Culture Collection. 2 × 10 6 cells were pre-activated by PMA 10 ng/ml for 24 h. Differentiated THP-1 cells were washed by complete RPMI-1640 medium and cultured for another 48 h before phagocytosis 58 . Erythrocytes were incubated with Fe 3 O 4 -MNPs (200 μ g/ml) or with Ringer solution (negative control) for 12 h, then labeled by DiI and washed by PBS. Differentiated THP-1 cells were incubated with labeled erythrocytes for 120 min. After incubation, THP-1 cells were washed and counterstained with DAPI 59 . Finally, the images were obtained by CLSM z-stack scanning and analyzed using 3D reconstruction. Measurement of intracellular ROs, Ca 2+ , 2,3-DPG, ATP, and RBC deformability. Intracellular ROS, Ca 2+ were monitored using fluorescent probes DCFH-DA and Fluo-3 AM, respectively. The changes in fluorescence were measured using FACS Calibur in fluorescence channel FL-1. The geometric means of fluorescence intensity were analyzed by Flowjo software. Intracellular 2,3-DPG and ATP concentrations were detected using Human 2,3-DPG ELISA Kit and ATP assay kit, respectively. The RBC deformability was monitored using an erythrocyte deformability analyzer (LBY-BX, China).

Toxicity of Fe 3 O 4 -MNPs and protective effects of N-acetylcysteine in vivo. The toxicity of
Fe 3 O 4 -MNPs and the protective effects of N-acetylcysteine were investigated in a rodent model. The animal received concentration of 12 mg/kg/injection of Fe, a dosage generally considered to be safe 12 . All procedures were performed in accordance with protocols approved by the Animal Management Rules of the Ministry of Health of the People's Republic of China (Document NO. 55,2001). Female CD ® IGS Rats, aged 8 weeks, weighing 190 ± 10 g were obtained from Beijing Vital River Laboratories (China). Twenty-eight rats were randomly assigned to the following four groups (G) (n = 7/G): G1, the control group in which animals were injected with 1 ml saline; G2, the control + N-acetylcysteine group in which animals were injected with 1 ml saline with N-acetylcysteine (2 g/L) administered in distilled water; G3, the Fe 3 O 4 -MNPs group in which Fe 3 O 4 -MNPs were injected at a dosage of 12 mg/kg of Fe by intravenous injection; G4, the Fe 3 O 4 -MNPs + N-acetylcysteine group in which animals received Fe 3 O 4 -MNPs at a dosage of 12 mg/kg of Fe by intravenous injection and were administered N-acetylcysteine (2 g/L) in distilled water. Rats were intravenously treated with saline or Fe 3 O 4 -MNPs every other day for a total of three treatments. N-acetylcysteine was administered during the experimental period.
At day 6, animals were anaesthetized using isoflurane. Blood samples were collected and used to measure ROS, phosphatidylserine exposure, hematology analysis, blood serum biochemistry and hemorheology analysis. The normal ranges of biochemistry and hematology data of healthy female CD ® IGS Rats were obtained from Charles River Laboratories (http://www.criver.com/files/pdfs/rms/cd/rm_ rm_r_cd_rat_clinical_pathology_data.aspx). The major organs were stained with hematoxylin and eosin (H&E). The distribution of Fe 3 O 4 -MNPs and the morphology of organs were observed under a light microscope (Olympus BX63, Japan).
In order to investigate pharmacokinetic characteristics of Fe 3 O 4 -MNPs, we measured blood Fe ions concentration after Fe 3 O 4 -MNPs injection using ICP-OES. Eight Rats were randomly assigned to two groups (n = 4/G): G'1, the control group in which animals were injected with 1 ml saline; G'2, the Fe 3 O 4 -MNPs group in which Fe 3 O 4 -MNPs were injected intravenously at a dosage of 12 mg/kg of Fe. Blood samples were obtained from G'1 group and G'2 group at 5, 10, 20, 30, 60, 120, 180, 240, 360, 480, 720, and 1440 min after the last injection. Blood Fe ions concentration was measured using iCAP 6000 SERIES (Thermo SCIENTIFIC, USA) after acid hydrolysis. statistical analysis. All data were expressed as mean ± SEM (standard error of the mean). The results were analyzed using GraphPad Prism 5 software (GraphPad Software Inc., CA). A p-value < 0.05 was considered statistically significant.