SNTA1 Gene Rescues Ion Channel Function in Cardiomyocytes Derived from Induced Pluripotent Stem Cells Reprogrammed from Muscular Dystrophy Patients with Arrhythmias

Patients with cardiomyopathy of Duchenne Muscular Dystrophy (DMD) are at risk of developing life-threatening arrhythmias, but the mechanisms are unknown. We aimed to determine the role of cardiac ion channels controlling cardiac excitability in the mechanisms of arrhythmias in DMD patients. To test whether cardiac dystrophin mutations lead to defective NaV1.5–Kir2.1 channelosomes and arrhythmias, we generated iPSC-CMs from two hemizygous DMD males, a heterozygous female, and two unrelated controls. Two Patients had abnormal ECGs with frequent runs of ventricular tachycardia. iPSC-CMs from all DMD patients showed abnormal action potential profiles, slowed conduction velocities, and reduced sodium (INa) and inward rectifier potassium (IK1) currents. Membrane NaV1.5 and Kir2.1 protein levels were reduced in hemizygous DMD iPSC-CMs but not in heterozygous iPSC-CMs. Remarkably, transfecting just one component of the dystrophin protein complex (α1-syntrophin) in hemizygous iPSC-CMs restored channelosome function, INa and IK1 densities and action potential profile. We provide the first demonstration that iPSC-CMs reprogrammed from skin fibroblasts of DMD patients with cardiomyopathy have a dysfunction of the NaV1.5-Kir2.1 channelosome, with consequent reduction of cardiac excitability and conduction. Altogether, iPSC-CMs from patients with DMD cardiomyopathy have a NaV1.5-Kir2.1 channelosome dysfunction, which can be rescued by the scaffolding protein α1-syntrophin to restore excitability.


Introduction 56 57
Null mutations in the Dp427 isoform of the dystrophin gene result Duchenne Muscular 58 Dystrophy (DMD). 1 This inheritable X-linked disease affects primarily adolescent males 59 causing progressive skeletal muscle deterioration, with negative effects in the central 60 nervous system. 2 Muscular dystrophies are also characterized by cardiac muscle 61 involvement, 3 which usually starts with an abnormal ECG. 4 Eventually, most patients with 62 DMD will develop cardiomyopathy by 20 years of age. 5 Many will be at a high risk for 63 arrhythmia and sudden cardiac death (SCD), which contributes considerably to the 64 morbidity and mortality of the disease. 6 However, diagnosis and prevention of arrhythmia 65 is challenging in DMD patients. 7

67
The mechanisms responsible for arrhythmias and SCD in patients with DMD 68 cardiomyopathy are poorly understood. The dystrophin associated protein complex 69 (DAPC) is involved in mechanoprotection of the plasma membrane. 8 The DAPC acts also 70 as a putative cellular signaling complex that forms a scaffold for numerous signaling and 71 membrane ion channel proteins. [9][10][11] Absence of dystrophin in DMD has the potential to 72 alter trafficking, localization and function of DAPC associated proteins in skeletal and 73 cardiac muscle. 12 For example, the expression and function of ion channels are defective 74 in ventricular cardiomyocytes of the mdx mouse model. 10,[13][14][15][16] Absence of dystrophin in 75 young mdx mice affects the function of NaV1.5, leading to cardiac conduction defects. 10 76 similar action potential parameter changes were obtained at 2 Hz (Suppl. Table 1,270 Suppl. Figures 1a-f and 2). 271 272

Conduction velocity is impaired in DMD iPSC-CM monolayers 273 274
The reduced dV/dtmax at the single cell level suggested that conduction velocity (CV) may 275 be compromised in iPSC-CMs monolayers from affected individuals. Hence, we 276 conducted optical mapping experiments using the voltage-sensitive fluorescent dye 277 FluoVolt TM in control, DMD, and female iPSC-CM monolayers paced at various 278 frequencies ( Figure 4a). CV in dystrophin-deficient iPSC-CM monolayers was 50% slower 279 than control monolayers paced at 1 Hz (27 ± 2 cm/s and 29 ± 4 cm/s in hemizygous Male 280 1 and Male 2 cells, respectively, versus 56 ± 3 cm/s in control cells, Figure 4b-c). CV of 281 Control 2 monolayers was 42 ± 5 cm/s (Suppl. Figure 3). Remarkably, CV in the 282 heterozygous female monolayers was even slower (18 ± 3 cm/s). In all three groups, the 283 CV restitution curve displayed slightly slower velocities at higher frequencies ( Figure 4d). 284 Most important, in the female monolayer (Figure 4e), slower and more heterogeneous 285 patterns of electrical wave propagation were accompanied by focal discharges in the form 286 of trigeminy (Figure 4e,left), which often triggered unidirectional block and reentry ( Figure  287 4e, right, and Suppl. Video 1 and 2). Altogether, the data presented in Figures 3 and 4  288 provide a direct mechanistic explanation for the conduction abnormalities and arrhythmias 289 seen on the ECGs of at least two of the patients (see Figure 1). In all three iPSC-CMs 290 from affected individuals, the reduced CV occurred in the absence of measurable 291 changes in connexin43 (Cx43) protein (Suppl. Figure 4). We did not detect any significant 292 differences in Cx43 expression among control, heterozygous, and hemizygous 293 iPSC-CMs in these monolayer experiments. 294 295

Sodium current is down-regulated in DMD iPSC-CMs 296 297
Sodium channels determine the upstroke velocity of the cardiac action potential and 298 consequently play a key role in the conduction of the cardiac electrical impulse. 34 Here 299 we compared the sodium current (INa) density in the DMD male and female iPSC-CMs 300 versus each of the controls. In Figure 5a and b, the peak inward INa density in hemizygous 301 iPSC-CMs was significantly decreased (-14 ± 1 pA/pF for Male 1 cells and -15 ± 1 pA/pF 302 for Male 2 cells) compared to both Control 1 (-27 ± 3 pA/pF) and Control 2 iPSC-CMs Suppl. Figure 1g,Suppl. Table 4 and 5). Importantly, the INa density in 304 heterozygous female cells was also dramatically reduced (-11 ± 1 pA/pF). Altogether, 305 except for peak sodium current density, statistical comparisons in terms of biophysical 306 properties of INa (half maximal activation, slope factor, reversal potential) for DMD vs 307 Control 1 (Suppl. Table 4), DMD vs Control 2 (Suppl. Table 5) and Control 1 vs Control 2 308 (Suppl. Table 6) showed no differences among any of the groups. Also, as shown in 309 Suppl. Figure 5, cell capacitance in all the patient-specific cells was similar to control, 310 indicating that cell size was similar in all groups. 311

312
The above data indicate that dystrophin deficiency reduces the INa density, which may be 313 considered one of the main causes for the cardiac conduction defects reported in DMD 314 patients. 35,36 The absence of dystrophin might also affect other ionic currents. For 315 instance, the L-type calcium current (ICa,L) is increased in cardiomyocytes from adult mdx 316 mice. 13,37 In addition, as previously suggested, ICa,L density is increased in iPSC-CMs 317 from DMD patients. 35 However, under our experimental conditions, ICa,L was unaltered in 318 hemizygous and heterozygous DMD iPSC-CMs (Suppl . Tables 4-6  Apart from the well-described regulation of NaV1.5 channels by the DAPC, 10,17 there is 325 evidence that this protein complex also regulates Kir2.1 inward rectifying potassium 326 channels in mdx cardiomyocytes. 14 Moreover, a pool of NaV1.5 channels co-localizes with 327 Kir2.1 forming protein complexes with scaffolding proteins at the cardiomyocyte lateral 328 membrane and intercalated disc, where they modulate each other's surface expression. 11, 329 19, 20 To test whether, in addition to INa, the inward rectifier potassium current is also 330 affected in iPSC-CMs from DMD patients, we compared Ba 2+ -sensitive potassium 331 currents (IK1). In Figure 5c and d, IK1 density measured at -120 mV was significantly 332 reduced in Male 1 (-1 ± 0.3 pA/pF) and Male 2 (-1.2 ± 0.3 pA/pF) iPSC-CMs compared to 333 Control 1 (-3.2 ± 0.5 pA/pF). IK1 density of Control 2 cells was -2.6 ± 0.6 pA/pF (Suppl. 334 Figure 1i). Changes in IK1 were highly variable in heterozygous cells, and the difference 335 with control was not significant, likely due to the variability of expression of dystrophin 336 ( Figure 2e)  Consistent with what has been described for mdx mice, 10 both hemizygous DMD 348 iPSC-CMs showed increased SCN5A expression (Suppl. Figure 7a, top), also like human 349 cardiac tissue from a Becker MD (BMD) individual (Suppl. Figure 7a, bottom). Similarly, 350 KCNJ2 gene expression was up-regulated in both hemizygous DMD cell lines, as well as 351 the BMD individual (Suppl. Figure 7b). This suggests that the increase in cardiac SCN5A 352 and KCNJ2 mRNA levels might be a general compensatory phenomenon in DMD 353 patients. On the other hand, consistent with the unaffected ICaL, neither CACNA1C nor 354 CaV1.2 were modified in either male or female DMD iPSC-CMs compared to control 355 (Suppl. Figure 7c). 356

357
To test whether the decreased IK1 and INa in both DMD iPSC-CMs were due to reduced 358 NaV1.5 and Kir2.1 protein levels, we performed Western blot experiments with total 359 protein lysates of iPSC-CMs monolayers. In Suppl. Figure 8a- and INa in DMD iPSC-CMs was due to reduced membrane protein levels, we conducted 363 protein biotinylation assays (Suppl. Figure 8c-d). Biotinylated NaV1.5 was significantly 364 lower than control in the Male 2 cell line only. Biotinylated Kir2.1 was significantly reduced 365 the hemizygous cells, consistent with the reduction in IK1. Altogether, the results 366 presented thus far support the idea that, the absence of dystrophin in the DMD iPSC-CMs, 367 resulted in reduced abundance of NaV1.5 protein in the whole cell and possibly reduced 368 trafficking of both NaV1.5 and Kir2.1 to the cell membrane, as predicted from our previous 369 work. 19-21 370 371 The data in iPSC-CMs from the heterozygous female are more challenging. NaV1.5 total 372 protein levels and biotinylated NaV1.5 channels were not different from control (Suppl. 373 Figure 8), but the INa density in single iPSC-CMs was even smaller than in DMD 374 iPSC-CMs. This, together with the lack of significance in the changes of IK1 density, total 375 Kir2.1 protein level, and biotinylated Kir2.1, lead us to conclude that the large variability 376 in the expression of dystrophin significantly influenced the overall results in the 377 heterozygous cells. In the heart, the dystrophin-associated protein α1-syntrophin (SNTA1) acts as a scaffold 383 for numerous signaling and ion channel proteins that control cardiac excitability. 33, 38, 39 384 α1-syntrophin is a PDZ domain protein that co-localizes and forms a macromolecular 385 complex ("channelosome") with Kir2.1 and NaV1.5 at the sarcolemma. 17, 19 39 11 Since α1-386 syntrophin has been shown to modify INa and IK1 by enhancing membrane NaV1.5 and 387 Kir2.1 membrane levels, 19 we hypothesized that even in the absence of dystrophin, 388 increasing α1-syntrophin should restore normal electrical function in the DMD iPSC-CMs. 389 Therefore, we stably transfected SNT1A gene via piggyBac transposon-based 390 mammalian cell expression system in Male 1 cells verifying an increase in syntrophin 391 expression (Figure 6a and b). As illustrated in Figure 6c, α1-syntrophin expression 392 increased the Kir2.1 and NaV1.5 protein levels in the membrane fraction as indicated by 393 co-localization with wheat germ agglutinin (WGA) compared to controls transfected with 394 GFP. In Figure 7a  internal PDZ-like binding domain localized at the N-terminus that also interacts with 434 α1-syntrophin. 10,19 Changes in the INa and IK1 might alter cardiac conduction and increase 435 the probability of premature beats like those seen in the ECG from the DMD patient. 10 We 436 proved here that in addition to reduced INa, iPSC-CMs from DMD patients also have 437 iPSC-CMs, including dV/dtmax, AP amplitude and overshoot (Table 1) INa reduction coincided with IK1 reduction in both hemizygous DMD iPSC-CMs, supporting 474 the idea that both channels require PDZ-mediated interaction with components of the 475 DAPC to modulate reciprocally their proper expression. 10,18 It is likely that the reduced IK1 476 in the DMD iPSC-CMs contributed to the reduced dV/dtmax, although the MDP in the 477 iPSC-CMs from the two dystrophic patients was like control. In this regard, it is important 478 to note that the relationship between MDP and INa availability is highly nonlinear in such 479 a way that a very small reduction in MDP is expected to result in substantial reduction in 480 sodium current during the action potential upstroke. 46 Regardless, the biotinylation 481 experiments demonstrated that Kir2.1 levels at the membrane were significantly lower in 482 both DMD iPSC-CMs with respect to the control. The elevated SCN5A and KCNJ2 mRNA 483 levels excluded the possibility that a decrease in gene expression was responsible for the 484 protein loss, and therefore, to smaller INa and IK1 densities in the DMD iPSC-CMs. This 485 somehow contrasts with reports in mdx 5cv mouse hearts, where the NaV1.5 mRNA levels 486 remained unchanged with a strong reduction in the NaV1.5 protein levels. 10 As such, the 487 reduction in the NaV1.5 and Kir2.1 protein levels could be related to ubiquitylation and 488 proteasome degradation as suggested previously in studies in dystrophin-deficient mdx 5cv 489 mice. 47 However, our results in DMD iPSC-CMs strongly suggest that disruption of the 490 DAPC due to lack of dystrophin significantly impairs ion channel expression and 491 function. 10,15,16 Specifically, we demonstrate that the decrease in ion channel current Previous reports indicate that although heterozygous DMD females, have negligible 503 skeletal muscle symptoms, they are not free of cardiac involvement. 48 For example, the 504 clinical expression of the X-linked DMD cardiomyopathy of heterozygous females 505 increases with age. 48 The female patient represented in this study suffered from a relative 506 severe phenotype, characterized by skeletal myopathy and cardiomyopathy, which could 507 be explained by a malignant mutation disrupting the N-terminal of the dystrophin gene. 508 One could assume that one gene of dystrophin should produce enough dystrophin to 509 preserve function in multinucleated skeletal muscle of females. 49 Unexpectedly, we found 510 that INa density in iPSC-CMs from the heterozygous female was even more reduced 511 compared to hemizygous iPSC-CMs. Interestingly, the QRS duration was significantly 512 prolonged on the ECG from the heterozygous female compared to the hemizygous 513 patient (see Figure 1), suggestive of a more dramatic loss-of-function effect on NaV1.5 in 514 heterozygous females. Probably this is related to the heterogeneity seen in 515 immunostaining studies where some heterozygous female cells express normal 516 dystrophin levels while others show absence or very low expression likely due to random 517 X-inactivation of the WT allele. 23 Because of random inactivation of one of the X 518 chromosomes, heterozygous females should constitute a mosaic of 2 or more cell types 519 dramatically differing in the extent of dystrophin expression. Thus, it would not be 520 surprising that females with DMD are more prone to suffer arrhythmias because of spatial 521 electrical inhomogeneity due to variable expression of the mutant allele. The 522 heterogeneity in dystrophin expression has been also observed in canine carrier models 523 of X-linked dystrophy, which exhibit a cardiac mosaic pattern, where dystrophin in each 524 myocyte is either fully expressed or absent. 50  We have derived data from experiments conducted in iPSC-CMs from patients who carry 550 independent dystrophin mutations and two unrelated controls, which may be a potential 551 limitation of our study. The original study design included siblings for each DMD cell line. 552 However, getting more experimental groups from the same family was not possible. 553 Nevertheless, both DMD lines lack dystrophin, which gives credence to the idea that loss 554 of dystrophin is important to the shared electrophysiological phenotype independently of 555 the specific mutation. Further, we show new insight into how heterozygous DMD females 556 might show a wide range of cardiac involvement, ranging from asymptomatic to severely 557 impaired electrical cardiac function, particularly the highly reduced INa leading to slowing 558 of conduction velocity, which is reflected on the ECG from the female patient. Thus, 559 together with the structural alterations, the electrophysiological changes may contribute 560 to left ventricular dysfunction in female DMD patients. 54 However, the impact of the finding 561 that the female carrier of the mutation presents a decrease in INa is somehow mitigated 562 by the fact that since she carries a different mutation, it is difficult to define how the 563 reduction of the INa in the female carrier compares with the reduction observed in the 564 affected individuals. 565 566 iPSC-CMs still show significant differences with adult ventricular cardiomyocytes and are 567 still far from recapitulating chamber-specific and layer specific electrical phenotypes of 568 the normal or dystrophic heart. In addition, we cannot generalize our results to patients 569 with different dystrophic gene mutations, such as those underlying Becker muscular 570 dystrophy, which lead to partially truncated dystrophins and may retain specific functional 571 properties of full-length dystrophin. However, enrolling a Becker MD patient was not 572 possible. Also, our syntrophin-mediated rescue experiments were limited to the Male 1 573 iPSC-CMs line. While caution should be exerted when attempting to extrapolate to the 574 other two DMD cell lines, it is important to note that the functional defects in the NaV1.5-575 Kir2.1 channelosome were very similar in the iPSC-CMs from all three patients, which 576 gives credence to our interpretation. 577 578

Data Availability 579
Authors will make materials, data and associated protocols promptly available to readers 580  Israel) supported this study by caring for and enrolling the female patient. We thank Dr. 608 Giovanna Giovinazzo and the staff of the CNIC Pluripotent Cell Unit for their help in 609 processing the iPSCs used in the syntrophin-mediated rescue experiments. We are 610 grateful to patients who despite having a lethal disease agreed to undergo skin biopsy for 611  parentheses. *P < 0.05; **P < 0.01; and ***P < 0.001. 918  Fig. 1

. Electrophysiological analysis in Control 2 (human foreskin-derived BJ iPSC-CMs). (a) Representative AP trace of ventricular-like control BJ iPSC-CMs obtained at 1 Hz of pacing.
Inset. First derivative with respect to time (dV/dt). (b-f) Action potential properties. Recordings at 1 and 2 Hz were similar to those obtained from the healthy donor patient derived-iPSC-CMs (Control 1). (g-i) Current traces, I/V curves, and normalized current densities for NaV1.5, CaV1.2, and Kir2.1 ion channels, respectively. Data obtained from the control BJ iPSC-CMs (Control 2) were similar to the other control iPSC-CMs.

Generation of iPSCs
Cell lines were generated using Sendai virus CytoTune-iPS 2.0 Sendai reprogramming kit (Thermo Fisher) for transfection of Yamanaka's factors: OCT4, KLF4, c-Myc, and SOX2, as described. 1,2 Subsequently, iPSCs were cultured on Matrigel (Corning)-coated 6-well plates with mTeSR1 medium (Stemcell Technologies) at 37°C with 5% CO2. iPSCs were passaged every 5 days at a ratio of 1:6 by mechanical dissociation using 1 mL/well of Versene solution (Invitrogen) following incubation at 37 °C for 7 min. DMD iPSCs were transported from Israel in dry ice to Michigan and to CNIC where they were differentiated to iPSC-CMs and used for the initial (Michigan) and syntrophin rescue (CNIC) studies. All iPSCs were tested for pluripotency before starting cardiomyocyte differentiation protocols. All of cells correlated well with the expression status of the pluripotency factors.
Differentiation markers were also assessed.

CMs purification using MACs negative selection
The directed differentiation method used here does not generate a completely pure iPSC-CM population. Hence, the following purification steps preceded any characterization or experiments. iPSC-CMs ≥30 days in culture were washed with DPBS (Gibco) and dissociated using 1 mL of 0.25% Trypsin/EDTA per well. Next, 2 mL of EB20 media was added per well of dissociated cells, each well was triturated and then The flow through or iPSC-CMs fraction was triturated and 1 mL of the total suspension was placed in a 1.5-mL Eppendorf tube to count the iPSC-CMs using a Millipore Scepter with Sensor tips (60 µm), this 1 mL was added back to the iPSC-CMs suspension total.
Next, the purified (98-99%) iPSC-CMs were centrifuged, the supernatant aspirated, and then resuspended in media for plating. Plating. The purified iPSC-CMs fractions were resuspended in EB20 media with 5 µM of ROCK inhibitor to 200-300k cells/200-300 µL volume and plated as monolayers on 22 mm × 22 mm cut Matrigeldiluted in DMEM/F12 media) PDMS. The plate was transferred to the incubator at 37°C and 5% CO2 for 2 hours. Next, 3 mL of EB20/ROCK inhibitor media was added to each well. After 2 days, iPSC-CMs were washed with 3 mL DPBS with Ca 2+ and Mg 2+ (Gibco) followed by addition of 3 mL of RPMI +B27 media; media was changed every 3 days. The highly purified iPSC-CMs were in monolayer culture on Matrigel-PDMS for at least 7 days after plating to induce maturation. Then monolayers were dissociated with 0.25% Trypsin/EDTA and re-plated onto Matrigel-coated micropatterned PDMS. All iPSC-CM selection materials were purchased from Miltenyi Biotec, except for culture media which was mixed in the laboratory. All the tests carried out in this study were performed using at least 3 separate cardiomyocyte differentiations.

Micropatterning on PDMS (adapted from ref 7 ).
Micropatterned area was 1 cm × 1 cm total, each island was 100 µm length × 15 µm width and islands were spaced 80 µm from each other. Preparing PDMS stamps. The surface of stamps was cleaned with scotch tape followed by sonication in 70% ethanol/milli-Q water for at least 20 min. In a sterile hood, they were allowed to dry and then, incubated with 250 μL at room temperature for at least 1 h.
Preparing PDMS substrates in 6-well plates. 18 mm PDMS circles were sonicated in 70% ethanol for 20 min and transferred to a 6-well plate after shaking excess EtOH off. When ready for microprinting, the culture dish was UVO treated with the lid off for 9 min.
Microprinting. While UVO is performed on PDMS circles, the Matrigel solution from the PDMS stamps was aspirated. After UVO was completed, dried stamps were inverted onto each PDMS circle and removed one by one after ~2 min. Later, the micropatterned PDMS plate was incubated with pluronic-F127 overnight at room temperature. Single cell re-plating. Before re-plating iPSC-CMs, micropattern plates were cleaned with 3× PSA (Penicillin-Streptomycin-Amphotericin B solution; Thermo Scientific) diluted in PBS (Gibco) for 1 h, and exposed to UV light for 15 min. iPSC-CMs were dissociated from monolayers using trypsin 0.25% with EDTA for 8-10 min and adding RPMI media containing 10% FBS after dissociation. Next, dissociated iPSC-CMs were transferred through a 70 μm filter into a 50-mL conical tube. The iPSC-CM suspension was centrifuged at 700 RPM for 3 min. Subsequently, iPSC-CMs were re-suspended in warm RPMI/B27+ (with insulin) media supplemented with 2% FBS and 5 μM ROCK inhibitor (re-plating media). Finally, 30k iPSC-CMs in 350 μL re-plating media were placed in the center of the micropatterned area. After 5 h, 2 mL of re-plating media was added very gently. Plate was returned to the incubator and media change was performed at days 1 and 3 after re-plating. iPSC-CMs were on micropatterns at least 4 days prior to patch-clamping experiments.

Electrophysiology
Standard patch-clamp recording techniques were used to measure action potentials, INa, ICaL, and IK1 3,8 . All the experiments were performed at room temperature (22°C-25°C), except for the AP that were recorded at 37°C.
Voltage-clamp experiments were controlled with a Multiclamp 700B amplifier and a Digidata 1440A acquisition system (Molecular Devices). Data were filtered at 5 kHz and sampled at 5-20 kHz. Activation curve data were fitted to a Boltzmann equation, of the form g = gmax / (1 + exp (V50 − Vm) / k), where g is the conductance, gmax the maximum conductance, Vm is the membrane potential, V50 is the voltage at which half of the channels are activated, and k is the slope factor.

Optical Mapping
iPSC-CMs were plated as monolayers at a density of ~50k iPSC-CMs in RPMI/B27+ media. After 7 days in culture, media was removed and each iPSC-CMs monolayer was washed with Hank's balanced salt solution with Ca 2+ and Mg 2+ added (HBSS ++ , Thermo Scientific) to remove remaining media. Next, iPSC-CMs were incubated with the FluoVolt membrane potential probe (F10488; Thermo Scientific) diluted in HBSS ++ , as reported before 11 . After a 30-minute incubation time, iPSC-CMs were washed with HBSS ++ and then heated at 35°C before optical mapping recordings. All iPSC-CMs monolayers displayed pacemaker activity, and the spontaneous and paced APs were recorded using a charge-coupled device camera (200 fps, 80 × 80 pixels; Red-Shirt Little Joe) with the appropriate emission filters and light-emitting diode illumination 12 . The recorded videos were filtered in both the time and the space domain, and CV was measured as described previously 3,13 .

Statistics
Statistical analyses were performed with Prism 8 (GraphPad Software). Values were first tested for normality (Shapiro-Wilk test) before statistical evaluation. Nonparametric Mann-Whitney rank test (two-tailed) was used. Multiple comparisons were analyzed using