Pulsatility Protects the Endothelial Function during Extracorporeal Membrane Oxygenation


 Background

Pulsatile flow has been proved to protect vital organ function and microcirculation during extracorporeal membrane oxygenation (ECMO). Studies revealed that pulsatile shear stress plays a vital role in the microcirculatory function and integrity. The objective of this study was to investigate how pulsatility affects wall shear stress and microcirculation during ECMO.
Methods

Using the i-Cor system, we compared the effects of pulstile or non-pulsatile flows in a canine ECMO model, with hemodynamic parameters and peak wall shear stress (PWSS) calculated. Serum concentrations of syndecan-1 and heparan sulfate were measured at different time points during ECMO. Pulstile shear stress experiments were also validated in endothelial cells exposed to different magnitude of pulsatility, with cell viability, the expressions of syndecan-1 and endothelial-to-mesenchymal tranformation (EndMT) markers analyzed.
Results

The pulsatile flow generated more surplus hemodynamic energy and preserved higher PWSS during ECMO. Serum concentrations of syndecan-1 and heparan sulfate were both negatively correlated with PWSS, and significantly lower levels were observed in the pulsatile group. In addition, non-pulsatility triggered EndMT, with EndMT related genes up-regulated, and endothelial cells exposed to low pulsatility had the lowest possibility of EndMT.
Conclusion

The maintenance of the PWSS by pulsatility during ECMO contributes to the beneficial effects on glycocalyx integrity and microcirculatory function. Moreover, pulsatility prevents EndMT in endothelial cells, and low pulsatility exhibits the best protective effects. The augmentation of pulsatility may be a future direction to improve the clinical outcome in ECMO.


Introduction
Extracorporeal membrane oxygenation (ECMO), a life-saving approach, is essentially important for the treatment of patients with severe cardiorespiratory failure, including those extremely critical COVID-19 patients [1], however, overall survival rate for extracorporeal life support (ECLS) is increasing sluggishly over the past ten years [2].While ECLS rescues the cardiorespiratory sytem, microcirculatory malperfusion easily occurrs when prolonged ECLS is executed as the continuous non-physiological ow impairs the microcirculatory function [3].Microcirculatory disturbance has been identi ed as an independent risk factor for mortality during ECMO [4].It is generally acknowledged that microcirculatory malperfusion affects the endothelial integrity, which leads to hypoxia, edema, acidosis, and in ammatory responses.
Currently, the standard circuit consists of a membrane oxygenator and centrifugal pump, of which the priming volume is smaller, with lower incidence of blood component damage [5].However, one of its main drawbacks is the typically non-pulsatile ow, of which the non-physiological nature leads to malperfusion to peripheral tissue and microcirculation [6].Recently, several pumps that can produce pulsatile ow for ECLS circuits have been developed [3,7,8].The novel i-Cor system (Xenios AG, Heilbronn, Germany) consists of a diagonal blood pump that can provide pulsatile ow triggered by electrocardiogram and has been applied in Europe for several years [9,10].Ündar`s group used the i-Cor system in animal models with both ow patterns and had reported that the renal function and vascular hemodynamics were better in the pulsatile group as compared to the non-pulsatile group [7].
The advantages of pulsatile ow have been well-established in the mechanical circulatory support system, including conventional cardiopulmonary bypass, ECMO, and ventricular assist devices.Pulsatile ow generates more optimal hemodynamic energy and extra-pressure [11], which preserved microcirculatory perfusion [12], and subsequently alleviates systemic in ammatory responses [13].
Endothelial dysfunction has been proved to be a crucial determinant of microcirculatory disturbance.The microvascular endothelial cell facilitates the microcirculatory exchanges of biological molecules, participating in various pathophysiological processes like in ammation, angiogenesis, vascular permeability, and so on [14].
The glycocalyx, formed by proteoglycans, glycosaminoglycan chains, and membrane glycoproteins, coats the inner surface of endothelium and provides protective layers, directly affecting the microcirculatory intergrity and function [15].Extensive research has proved that glycocalyx is also a mechanosensor [16], and a stable wall shear stress balances between biological synthesis and degradation in glycocalyx [17,18].Assmann et al. [19] revealed that non-pulsatile ow in cardiopulmonary bypass decreases peak wall shear stress (PWSS) and causes elevations of syndecan-1 and heparan sulfate, both of which are the by-products for glycocalyx degradation.However, the correlation between shear stress and microcirculation during ECMO is poorly understood.In this study, we investigated the relationship between shear stress and microcirculatory function, which was detected by glycocalyx degradation by-products, in an attemp to determine whether pulsatile ow under ECMO improves microcirculation and preserves endothelial function.

Study design
To evaluate the pulsatile ow as a mediator of shear stress under ECMO, both in vitro and in vivo models were developed.For the in vivo model, we developed a canine model with cardiogenic shock, which was supported by pulsatile or non-pulsatile mode of ECLS established by the i-Cor system.Pulsatile ow and continuous ow were compared in canine models with respect to hemodynamics, shear stress, and endothelial integrity.The correlation between shear stress and syndecan-1 level was also studied.For the in vitro model, the Flexcell system (Flexcell Inc., McKee-sport, PA, USA), a bio uid mechanics device was employed.Pulmonary microvascular endothelial cells (PMVECs) were cultured under the Flexcell system with continuous ow or pulsatile ow with different shear stresses.After a dynamic cultivation, the effect of ow patterns on PMVEC viability, expression of syndecan-1, and possiblity of endothelial to mesenchymal transformation (EndMT) were investigated.

ECMO settings
The experimental ECMO settings consisted of a console (i-Cor, Xenios AG, Heilbronn, Germany), a diagonal pump that can easily generated pulsatility, a gas blender (Sechrist, Whitewright, TX, USA) and a membrane oxygenator (Medos Medizintechnik AG, Stolberg, Germany).To prime the ECLS circuit, 200 mL Ringer`s solution was employed, with a pump rate set at 70 beats per minute (bpm) during ECMO.PaO 2 ranging from 10 kPa to 16 kPa, and PaCO 2 from 4.0 kPa to 6 kPa were maintained for the oxygen / air ow, as suggested by previous publications [10].Switching from continuous ow to pulsatility could be easily achieved via the i-Cor system.Triggered and synchronized by electrocardiogram (ECG), the pulsatile ow runs in an equivalent fashion (1:1) if the heart rate (HR) was lower than 100bpm.Intravenous verapamil (1.5mg) was administered in case of tachycardia (HR > 120 bpm) to maintain a pulsatility at a parallel ratio.

Anesthesia and surgical procedures
Sixteen beagles was obtained from the Laboratory Animal Center of the Southern Medical University with a mean body weight about 10 kg.Brie y, general anesthesia was induced by midazolam (10mg, intravenous), followed by the procedure of tracheotomy with an endotracheal tube properly inserted.We established the venous line via the left internal jugular vein.Intra-operative anesthesia was maintained by the intravenous administration of fentanyl (150 µg/kg) and the inhalation of 2.5% sevo urane through the endotracheal tube, which was then connected to a mechanical ventilator for experimental animals (HX-300S, Taimeng Inc., Chengdu, China).The ventilator supported at a rate of 22 times/minute, and the tidal volume was set at 10mL/kg/respiration.The right common carotid artery and the right jugular vein were dissected, unfractionated heparin bolus (100 U/kg) was then administered intravenously, with an activated clotting time (ACT) ranging from 180 to 240 seconds.The right common carotid artery and the right jugular vein were successfully cannulated with the 8-Fr and 10-Fr cannulas (Medtronic Inc., Minneapolis, MN, USA), respectively.A venous-arterial (V-A) ECMO was initiated at a ow rate of 130 mL/kg/min for both perfusion modes.
Animal models and monitoring ECMO was initiated under normal cardiac condition for the rst 20 minutes.Animal models of cardiogenic shock was made through ventricular brillation, which was created by the 4-V alternating current externally.Once the cardiogenic shock model was made, hemodynamic variables including heart rate, arterial pressure, oxygen saturation, blood gases, blood samplings, ACT, and urinary output were monitored (Truwave and Vigilance II, Edwards Lifesciences, LLC, USA).Hemodynamic data and blood samplings were collected at 6 time points: baseline, before ECMO, 15 minutes after ECMO, 1 hour after ECMO, 3 hours after ECMO, and 6 hours after ECMO.After 6 hours of ECLS, the circuit was discontinued.Phenylephrine shots were used during ECLS in case of hypotension systolic, which was de ned as pressure below 30 mmHg.At the end of the experiment, animals were euthanized by potassium chloride (3 mEq/kg) under general anesthesia.The experimental ECMO circuit was shown in Fig. 1.

Pulsatility assessment
Shepard`s model was used to quantify pulsatility as described previously [12], and the following parameters were assessed: Energy-equivalent pressure (EEP)= QPdt/ Qdt, in which Q is the blood ow (mL/s), P is the instantaneous pressure (mmHg), and t is the time (s); Surplus hemodynamic energy (SHE) = 1,332 (EEP − MAP) (ergs/cm 3 ), in which MAP is the mean arterial pressure [12].

Shear stress assessment
To assess wall shear stress during pulsatile ECMO, blood ow velocity (v), radius of the femoral arteries (R), and viscocity of the whole blood (η) were measured initially.The radius and velocity of the vessels were accrued using an ultrasound machine for animals (P6-VET, Dawei, Jiangsu, China).The M-mode was utilized for determining radius of the femoral artery, of which the inner radius in the end-diastolic phase was detected.Mean value of the velocities in three successional cardiac cycles was recorded.The velocity and radius of the femoral arteries were examined at the aforementioned six time points.Viscosity of the whole blood (η) was deternined at each time point by a hemorheology meter (LG-R-80E, Steellex Inc., Beijing, China).
To more precisely quantify pulsatile ow, the Womersley principle, which de ned the ow by pulsatile pressure gradient, was applied [20].The peak wall shear stress (PWSS) was used and was calculated as follows: In this equation, η is the viscosity of the whole blood, R is the radius, n is the harmonics, N is the maxium of harmonics, is the n-th harmonic component of the axial blood velocity, =, and are the Bessel function of order 0 and 1, is the Womersley value, is the density, 2n is the circular frequency, and is the fundamental frequency.

Measurements of serum syndecan-1 and heparan sulfate
Blood samples were taken at six time points according to the experiment protocol.Serum concentrations of syndecan-1 and heparan sulfate were detected using the enzyme-linked immunosorbent assay (ELISA) kits (Renjiebio Co., Shanghai, China).
PMVECs isolation and culture 20 male SD rats ranging from 100 to 150g were obtained from the Laboratory Animal Center of the Southern Medical University.The rats were sacri ced after general anesthesia by intraperitoneal administration of pentobarbital (30mg/kg).Lung tissues were separated and collected in the absence of pleura and large vessels.Tissues were then cut into slices and were kept in culture asks.Tissues were cultivated at 5% CO2 in room temperature for 4 days, with 15% fetal bovine serum added.PMVECs were then isolated, and we changed the mediums every two days.Rat PMVECs were cultured in line with the proven techniques in our laboratory, as described previously [21].

In vitro ow shear stress experiments
Pulstile shear stress experiments were conducted in vitro using the Flexcell apparatus (Flexcell™ Inc., McKee-sport, PA, USA), which applied uid shear stresses to PMVECs under various conditions.PMVECs were seeded on a 6-well Flexcell plates and incubated for 2 days.Seeded PMVECs (1×10 5 cells per well) were deprived from FBS and exposed to continuous ow or pulsatile ow, with frequency set at 1 Hz.Different degrees of pulsatility were applied, low pulsatility, intermittent pulsatility, and the high pulsatility, which were de ned as pulsatile ow at 5 dyne/cm 2 , 10 dyne/cm 2 , and 20 dyne/cm 2 , respectively.The ow rate was maintained at about 2mL/min.PMVECs were cultured and treated with various shear stress settings for 6 hours.

Cell viability assay
At the end of the ow experiment, PMVECs viability was assessed using the CCK-8 kits.PMVECs were placed in the 96-well plate under humidi ed environment (5% CO 2 /95% air) at 37℃ overnight.10µL CCK-8 was added to the wells, where PMVECs were additionally incubated 3 hours at 37℃.The absorbance at 450 nm re ected cell viability and was detected using the Microplate Reader (Enspire, PerkinElmer, MA, USA).

RNA isolation and quantitative RT-PCR
PMVECs were lysed with the Trizol reagent (ThermoFisher Scienti c, MA, USA), and total RNAs were extracted with the RNeasy Mini Kit (Qiagen, Cary, NC, USA) after the completion of shear stress experiments.The extracted RNAs were reversed transcribed using the Oligo (dT) primers and samples were prepared by mixing complementary DNA, speci c primers, and power-SYBR Mix (Yeason Biotech Co., Shanghai, China).Quantitative RT-PCR was performed using the LightCycler 480 (Roche, Basel, Swizerland) according to the manufacturer`s instructions.Each experiment was performed thrice.Expressions of syndecan-1 mRNA, EndMT markers (ACTA2, Snail1), and PECAM-1 were measured.Expression levels of each mRNAs were measured with the comparative cycle threshold (ΔΔCT) approach.
Level of the non-pulsatile condition was set to be 1.Expressions of genes were normalized to GADPH as a housekeeping gene.Primers used in the present study were listed in Table 1.

Statistics
Statistical analyses were conducted with IBM SPSS Statistics version 16.0 software (SPSS Inc., Chicago, IL, USA).Continuous variables were displayed as mean ± standard deviation (SD).Wilcoxon signed-rank test was used to compare differences at different time points, and the Spearman rank correlation coe cient was applied for the analysis of correlation.Comparisons for multiple time points and different ow settings were performed using the repeated measures of analysis of variance (ANOVA).All hypotheses in the present study were two-sided and all p values below 0.05 were considered to be statistically signi cant.

Pulsatility assessment
All circuits ran uneventfully for 6 hours, without major adverse effects.Figure 2 displays the waveforms of femoral arteries during ECLS in non-pulsatile or pulsatile fashion, in which the diastolic enhancement was seen.As shown in Table 2, the hematocrit, hemoglobin, and platelet counts decreased over time, while leukocyte counts and the level of lactic acid increased during ECLS, however, no between-group differences were noted.After the initiation of ECMO, MAP decreased in both groups, while MAP levels were signi cantly higher in the pulsatile group at between 3 hours and 6 hours (Table 3).Similarly, higher levels of EEP were observed in the pulsatile group at between 1 hour and 6 hours during ECMO, at all time points.Accordingly, the pulsatile ow generated more SHE than did the non-pulsatile ow at between 1 hour and 6 hours, at all time points, based on Shepard's model (Table 3).

Shear stress evaluation
Based on Womersley`s theory, Fig. 3A shows the PWSS values of the femoral artery during ECLS.After the commencement of ECMO, the PWSS values decreased gradually in both groups, nonetheless, the pulsatile group had higher PWSS levels than did the non-pulsatile group from 15 min to 6 hours during ECLS.PWSS nadirs occurred at 6 hours during ECLS in both groups, reaching averages of 3.125 ± 0.83 and 14.88 ± 3.18 dyne/cm 2 , respectively.The PWSS values were positively correlated with EEP (r = 0.70, p < 0.01) and SHE (r = 0.73, p < 0.01), as shown in Fig. 3B and 3C.

Serum concentrations of syndecan-1 and heparan sulfate
The baseline serum levels of syndecan-1 in the pulsatile group and non-pulsatile group achieved average values of 1.69 ± 0.80 and 1.68 ± 0.79 µg/dL, respectively.The serum syndecan-1 concentration increased after the commencement of ECMO, and reached its summit at 6 hours during ECMO (Fig. 4A).The values at 6 hours were approximately 6.5 and 13.3 times of the baseline levels in these two group, respectively.
The syndecan-1 levels in the pulsatile group at 6 hours were signi cantly lower as compared with the non-pulsatile group (p < 0.01).Similarly, the serum levels of heparan sulfate also increased after ECLS and peaked at 6 hours, equivalent to 1.4 and 5.1 times ot the baseline levels in the pulsatile group and non-pulsatile group, respectively, and again, the pulsatile group had lower heparan sulfate levels than did the non-pulsatile group (p < 0.05), as shown in Fig. 4B.

Responses of endothelial cells to pulsatility
We investigated the cell viability responses of PMVECs following incubations of non-pulsatile or pulsatile ows (low, intermittent, or high pulsatility), and we found no difference in cell viability among these cultural conditions using the CCK-8 assay (Fig. 5A).We further investigated the endothelial glycocalyx related gene expressions under these cultural conditions.Using static condition as the reference, exposure to non-pulsatile ow signi cantly up-regulated the mRNA expression of syndecan-1 in PMVECs as compared to pulsatile ows (p < 0.01).In additonal, PMVECs that exposed to low pulsatility (5 dyne/cm 2 ) had lower syndecan-1 expressions as compared to high pulsatility (20 dyne/cm 2 )(p < 0.05), as shown in Fig. 5B.
Moreover, we studied the possibility of phenotypic alteration of EndMT in PMVECs under various pulsatile conditions (Fig. 5C-D).Compared to pulsatile conditions, the non-pulsatile ow signi cantly upregulated EndMT-related genes including ACTA2 and Snail-1.No differences of ACTA2 expressions were reached among low, intermittent, and high pulsatility, however, the high pulsatility group had higher expressions of Snail-1 than did the low pulsatility group (Snail-1 expression 2.1 ± 0.4 fold increase over non-pulsatility, p < 0.01) and the intermittent pulsatility group (Snail-1 expression 1.7 ± 0.5 fold increase over nonpulsatility, p < 0.05).Finally, we analyzed the expression of PECAM-1, the indicator of endothelial phenotype (Fig. 5E).On the contrary, lower level of PECAM-1 expression was obseved in the non-pulsatile group (54 ± 11% of the static control), as compared to low pulsatile or intermittent conditions.Higher expression of PECAM-1 was seen in PMVEC exposed to low pulsatility (84 ± 5% of the static control), as compared to the high pulsatility (64 ± 15% of the static control, p < 0.05).

Discussion
The hemodynamic advantages of pulsatile ow have been generally acknowledged in the mechanical circulatory support systems, however, whether pulsatile ow improves microcirculation and preserves endothelial integrity has to be con rmed.In the present study, we found that the pulsatile ow generates more SHE and maintains higher PWSS during ECMO as compared to the non-pulsatile ow.The levels of syndecan-1 and heparan sulfate, which are negatively correlated with PWSS, were signi cantly higher in the non-pulsatile group, indicating that PWSS during ECMO has bene cial effects on endothelial integrity.
Moreover, our ndings show that non-pulsatility facilitates EndMT, upregulating EndMT related genes, whereas low pulsatility exerts the best protective effects.
ECMO is a life-saving device that rescues the cardiorespiratory sytem, however, the conventional nonpulsatile ow during prolonged ECLS may result in microcirculatory disturbance, which is detrimental to the outcome for ECMO [4].It has been con rmed that the alteration of ow pattern during ECLS declines ow velocities and PWSS [19].The endothelial glycocalyx, a mechanosensor, receives the signals of hemodynamic and shear stress changes, affecting the synthesis and biological functions of glycocalyx [22].It has been reported that the degradation of glycocalyx treated with non-pulsatile ow is signi cantly higher as compared to pulsatile ow [23].A relatively low PWSS also promotes the glycocalyx degradation and thus impairs the microcirculatory integrity [24,25].Wang's group [26] recently reported that wall shear stress is closely relevant to glycocalyx shedding during cardiopulmonary bypass, and the non-pulsatile ow contributes to the decomposition of glycocalyx.In this study, we found that the PWSS during ECMO is positively correlated with SHE, the extra energy produced by pulsatile ow, whereas negative correlations were also observed in PWSS with glycocalyx biomarkers, suggesting that the pulsatile ow inhibits the endothelial glycocalyx degradation as it maintains some PWSS during ECMO.
Syndecan-1 and heparan sulfate are components on the surface of the glycocalyx network, and were frequently applied to indicate the glycocalyx integrity.When mast cells are activated by in ammation, ischemia, or hypoxia, matrix metalloprotease is upregulated [27], resulting in the cleavage of syndecan-1 [28].Chou et al. [29] found that the expression of matrix metalloprotease genes can be also upregulated due to the alteration of ow-related shear stress.We found that not only serum levels of syndecan-1, but also mRNA expressions in cellular level were signi cantly higher under non-pulsatile ow as compared to the pulsatile ow.Besides shear stress changes, the increase of syndecan-1 can be induced by various conditions [16].Our ndings, however, show that the endothelial cells directly sense the changes in pulsatility, yielding the changes in the expressions of glycocalyx-related genes correspondingly.Moreover, our ndings also indicate that the acute changes of pulsatility and subsequent activation of in ammation during ECMO cause an acute release of syndecan-1 and heparan sulfate, stored in the glycocalyx network.
It is well-acknowdedged that the biological activities of endothelial cells behave differently to the magnitude of ow stress [30,31].Cell alignment and elongation are seen when the magnitude of ow shear stress increases [32,33].Faure et al. [34] reported that endothelial elongation and orientation were observed with the increase of pulsatile wall shear stress, contributing to subsequent phenotypic changes and transcriptional difference in endothelial cells.Hellmann et al. [35] also found that the endothelial cell morphology, integrity, and expressions of typical endothelial markers could be remained in pulsatile shear stress up to 8.6 dyne/cm 2 .In the present study, we observed that PMVECs exposed to low pulsatility preserves glycocalyx integrity and had lower phenotypic transformation, as the syndecan-1 and EndMT related genes expressions were signi cantly in PMVECs exposed to high pulsatility.These results agree well with our previously study that used pulsatile ow during cardiopulmonary bypass in pediatric patients undergoing congenital cardiac surgeries, and the results showing that low pulsatility has better hemodynamic pro les, organ protective effects, and better oxidative status [12].
EndMT is an endothelial phenotypic alteration relevant to various cardiovascular diseases and mechanical microenvironment.The pathological process of EndMT starts when the cell-cell interactions of the endothelial cells are deprived, with a series of subsequent changes in biological behaviors, such as the separation from the monolayer, migration into the interstitial space, and the lost of endothelial markers [36].This change in phenotype as well as the transcriptional difference occurs in endothelial cells in their response to bio uid ow [37].Several studies have shown that the development of EndMT is induced by the disturbance of shear stress and is often associated with in ammatory activation and tissue degeneration [34,38].In our observations, over-expressions of ACTA2 and Snail-1, the EndMTrelated genes, were seen in PMVECs exposed to non-pulsatile ow, while relatively higher expression of PECAM-1 were observed in PMVECs exposed to low pulsatility, suggesting that low pulsatility has lower tendency of phenotypic transformation and protects the endothelium.
The mechanical circulatory support devices depend not only on the nature of the pump itself but also on the pattern of the ow (pulsatile or non-pulsatile) delivered by both of the pump and the left ventricle [31].
In patients with long-term mechanical circulatory support, the pulsatility is associated both with the left ventricular contractivity and the pump itself [39,40].In addition to hemodynamics, pulsatility in these devices seems to improve hemocompatibility, prevent bleeding complications, avoid von Willebrand Factor de ciency, and reduce systemic in ammation [31,41].
In fact, in those critically ill COVID-19 patients with severe respiratory failure, ECLS typically lasts for weeks or even months, thus devices that provide more effective microcirculatory perfusion is optimal.Flow modi cation allowing for an augmentation of pump pulsatility in ECMO is now under investigation.
In addtion to the i-Cor system, other recent reported pulsatile ow generators included the Medos Deltastream DP3 system (Medos Medizintechnik AG, USA) [42] and the K-Beat system by Inamori's group [8].Overall, the development of the future ECMO devices will require an aviailability of pulsatility to reduce the complication of microcirculatory malperfusion.

Conclusion
The pulsatile ow produces more SHE and preserves more effective PWSS during ECMO.The levels of syndecan-1 and heparan sulfate are negatively correlated with PWSS, and are signi cantly lower under the pulsatile ECLS circuit, indicating that the pulsatility during ECMO protects the glycocalyx and the endothelial integrity.Moreover, pulsatility prevents EndMT of the endothelial cells, and low pulsatility has the best protective effects.Our data indicate that the modi cation of pulsatility may be a therapeutic strategy to improve the outcome in ECMO.The experimental ECMO circuit.Abbreviations: ECMO, extracorporeal membrane oxygenation.
approval was obtained from the Institutional Animal Care and Use Committee of Sun Yat-sen Memorial Hospital (Reference No. SYSU-IACUC-2020-B0402).The experimental setting was established in accordance with the Guide for the Care and Use of Laboratory Animals by the National Institute of Health (NIH) of the United States (revised in 1985).

Figure 4 The
Figure 4

Figure 5 Responses
Figure 5

Table 2
Blood parameters of the ECMO circuits.