Lymphoedema conditions disrupt endothelial barrier function in vitro

Abstract

Lymphatic vessel contractions generate net antegrade pulsatile lymph flow. By contrast, impaired lymphatic vessels are often associated with lymphoedema and altered lymph flow. The effect of lymphoedema on the lymph flow field and endothelium is not completely known. Here, we characterized the lymphatic flow field of a platelet-specific receptor C-type lectin-like receptor 2 (CLEC2) deficient lymphoedema mouse model. In regions of lymphoedema, collecting vessels were significantly distended, vessel contractility was greatly diminished and pulsatile lymph flow was replaced by quasi-steady flow. In vitro exposure of human dermal lymphatic endothelial cells (LECs) to lymphoedema-like quasi-steady flow conditions increased intercellular gap formation and permeability in comparison to normal pulsatile lymph flow. In the absence of flow, LECs exposed to steady pressure (SP) increased intercellular gap formation in contrast with pulsatile pressure (PP). The absence of pulsatility in steady fluid flow and SP conditions without flow-induced upregulation of myosin light chain (MLCs) regulatory subunits 9 and 12B mRNA expression and phosphorylation of MLCs, in contrast with pulsatile flow and PP without flow. These studies reveal that the loss of pulsatility, which can occur with lymphoedema, causes LEC contraction and an increase in intercellular gap formation mediated by MLC phosphorylation.

1. Introduction

The lymphatic system is a subsystem of the circulatory system that regulates fluid homeostasis, immune cell trafficking and the absorption of dietary fats [1]. The initial uptake of interstitial fluid, antigens, proteins, lipids and inflammatory cells into lymphatic capillaries, also called initial lymphatics, is regulated by a pressure differential between the interstitium and lymphatic capillaries. This transient pressure differential further drives lymph to the precollector lymphatic vessels. Transport of lymph through the precollectors and larger collecting lymphatic vessels is primarily propelled by local smooth muscle cells (SMCs) at relatively low flow rates and secondarily by skeletal muscle contractions, respiration and external compressions, but not by a central pump [2]. Lymph flow direction in precollectors and collecting vessels is aided by sequential contraction of SMCs and one-way bi-leaflet valves to maintain net antegrade flow [3]. Consequently, the lack of SMC contractions and valvular dysfunction can affect lymph transport and may cause lymphoedema.

Lymphedoema is the accumulation of fluid in the interstitium and is generally caused by lymph transport insufficiency [4] as a consequence of impaired lymphatic function. Lymphoedema is classified into primary lymphoedema, which is an inherited or congenital condition resulting from malformation of the lymphatic system, while secondary lymphoedema is more common and results from injury or damage to the lymphatic system. The initial stages of lymphoedema are not completely understood, although ultrasonography, histology and microscopy studies have elucidated the physiological and structural changes of collecting vessels during the onset of lymphoedema [5,6]. Initially, in secondary lymphoedema, the collecting lymphatic vessels transition from the normal to the ectasis type, a dilated vessel with a greatly enlarged lumen. The vessel transitions from ectasis to the contraction type, where the lymphatic vessel lumen is partially reduced due to thickening of the smooth muscle layer. Finally, there is progression to the sclerosis type with a narrowed vessel lumen, the loss of elasticity and lymphatic contractility. In primary lymphoedema, the ectasis and sclerosis types have been observed, while the contraction type has not been observed [5].

Endothelial cells (ECs) lining the inner layer of blood and lymphatic vessels are exposed to mechanical stresses, including wall shear stress (WSS), pressure and hoop stress, induced by the fluid flow. In the blood circulatory system, ECs exposed to disturbed fluid flow experience a constantly changing WSS direction and are more susceptible to atherosclerosis [6,7]. In hypertensive patients, the increase in mean blood arterial pressure yields an increase in hoop stress and causes an increase in endothelial reactive oxygen species, inflammation and apoptosis [8]. In vitro and in vivo, blood vessel endothelial cells (BECs) respond differently to pulsatile and steady blood flow, with pulsatile flow (PF) conferring a protective phenotype clinically to arterial BECs [9–12]. For example, patients under cardioplegic arrest for cardiac surgery who received pulsatile cardiopulmonary bypass blood (CPB) flow instead of steady CPB flow experienced a reduction in endothelial dysfunction and renal replacement therapy [12]. PF increases basal release of endothelium-derived nitric oxide (eNOS) and decreases vascular resistance in comparison to steady blood flow [13].

During the different degenerative stages of collecting lymphatic vessels in regions of lymphoedema, lymph flow rates and flow characteristics change, and consequently the mechanical stresses experienced by the lymphatic endothelial cells (LECs) change. In lymphoedema patients, increased lymphatic flow rates can occur, and pulsatility can be affected [14,15]. Flow in normal lymphatic vessels is pulsatile, while predominantly steady under lymphoedema conditions [2,16–19]. It is unclear how the loss of pulsatility that may occur during lymphoedema can affect LECs and further exacerbate lymphoedema. A possible consequence of steady lymph flow is impairment of the lymphatic barrier function, which could increase lymphatic permeability and may further contribute to lymphoedema [20]. In BECs, barrier function is regulated by WSS with PF downregulating permeability in comparison to steady flow (SF) [21]. The disruption of BEC barrier function and consequent hyperpermeability is influenced by several mechanisms, including intercellular gap formation caused by the contractility of actomyosin stress fibres [22]. Disturbed flow in atherosclerosis prone arteries can induce EC hyperpermeability via cytoskeletal tension [23,24]. Actomyosin-driven cytoskeletal tension is mediated by F-actin stress fibre formation and phosphorylation of myosin light chain (MLC). In non-muscle cells, the movement of conventional myosin is controlled by phosphorylation of two regulatory MLCs [25]. MLC may be mono- or di-phosphorylated (pMLC and ppMLC) at Ser19 or Thr18/Ser19 sites by several mediators of MLC phosphorylation, especially myosin light chain kinase (MLCK). pMLC and ppMLC play distinct roles in blood vessel endothelial barrier disruption [26]. pMLC is constitutively present in the endothelial cytosol during normal conditions and becomes localized with actin stress fibres during inflammation. ppMLC is scarcely detected in BECs under normal conditions; however, it increases during inflammation and predominantly localizes in peripheral stress fibres contributing to cellular tension.

Apart from WSS, it is unclear how changes in pressure and pulsatility could further exacerbate lymphoedema. As lymphoedema is often associated with changes in transmural pressure and pulsatility, lymphoedematous conditions may influence the LEC phenotype via pressure mechanotransduction [2,27]. In the absence of flow, steady pressure (SP) increases cause BEC barrier function disruption due to MLC phosphorylation [28,29]. Conversely, BECs exposed to pulsatile pressure (PP) show decreased or no change in permeability [30]. Given that BEC barrier function can be regulated by pressure and WSS independently, it is possible that fluid mechanics stresses cause changes in pMLC and ppMLC quantity and localization, playing a role in lymphatic barrier regulation.

Herein we have used a CLEC2-deficient mouse model to research the ectasis lymphatic vessel type where the vessel is distended, and the pressure and lymph flow rate can increase, and is a lymphatic vessel type relevant for both primary and secondary lymphoedema. This mouse model experiences an adverse pressure gradient that affects lymph flow and drainage due to a haemostatic defect where the thoracic duct drains into the venous system [31]. In the early stages of development, the adverse pressure gradient obstructs lymph flow and affects functional valve development, a flow-dependent process [32]. The reduction of lymph flow causes lymphoedema. The lymphatic flow fields downstream of the inguinal lymph node of a normal and a CLEC2-deficient mouse were characterized in vivo. The in vivo lymph flow characteristics were recreated in vitro to address the role of normal pulsatile lymphatic flow and quasi-steady lymphoedema flow conditions on LEC barrier function. Furthermore, the effect of SP and PP was decoupled from the fluid flow to elucidate the role of pressure on LEC barrier function. We report the differential response in phenotype and morphology of LECs exposed to physiologically relevant SF, PF, atmospheric pressure (AP), PP and SP.

2. Material and methods

2.1. Cell culture

Human dermal LECs (2–9; PromoCell, Heidelberg, Germany) were cultured in EGM-2 MV cell medium (Lonza, Basel, Switzerland) at 37°C in 5% CO₂. For parallel plate flow chamber (PPFC) experiments, cells were grown to confluence on 75 × 38 mm² glass slides (Fisher Scientific, MA, USA) coated with 0.2% gelatine (Eastman, NY, USA). For pressure chamber experiments, cells were cultured either on six-well cell culture plates (TrueLine, NJ, USA) for mRNA and protein isolation or on no. 1 glass coverslips (Chemglass Life Sciences, NJ, USA) coated with 0.2% gelatine for immunofluorescence (IF). All cell experiments were started at least 48 h after cells reached 100% confluence.

2.2. Mouse experiments

All animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee. The Clec2^–/– and Prox1-GFP BAC transgenic mice were obtained from the mutant mouse regional resource centres [33,34]. CLEC2-deficient mice have been previously reported to develop lymphoedema [31,33,35]. The Prox1-GFP (normal) mouse weighed approximately 14.5 g, while the Clec2^–/– (lymphoedema) mouse weighed about 25 g due to interstitial fluid accumulation because of lymphoedema. Lymph flow was measured and characterized in the collecting vessel downstream of the inguinal lymph node. The lymphatic flow field was analysed over a 25 s period for the normal mouse and 1 s for the lymphoedema mouse and although differing in length of time, each time period captured the representative lymphatic flow characteristics for the region of interest. Fluorescent microparticles were injected into the inguinal lymph node, transported by the lymph flow, imaged with epifluorescence microscopy and tracked using ImageJ software as previously described [17]. Different segments within the regions of interest were selected for further analysis, including vessel wall contractility and the centre plane average velocity U̅ (figure 1). The centre plane average velocity was determined as follows

$$\overline{U}={\displaystyle \frac{{\int }_{-R}^{+R}u(r)r\hspace{0.17em}\mathrm{d}r}{{\int }_{-R}^{+R}r\hspace{0.17em}\mathrm{d}r},}$$

where u(r) is the velocity at a distance r from the centre and is the lymphatic vessel radius. Particles with velocities that deviated more than 10% from the Poiseuille solution were removed from the flow field except near the valves, when the overall velocity was close to zero, or the flow field was changing directions and the non-Newtonian effects were evident. The potential inclusion of particles from neighbouring planes can introduce an at most 10% error into the flow field without considering the uncertainty arising from particle lag effects, which was estimated to be at most 1%. This inherent deviation is exaggerated in the shape of the velocity profile due to the nonlinear nature of the parabolic profile.

Figure 1.

Collecting lymphatic vessels downstream of the inguinal lymph node of (a) normal and (b) CLEC2-deficient lymphoedema mouse model. The field of view of the normal mouse includes a normal bi-leafleat valve, while the lymphoedema mouse has an incompetent malformed valve. White dashed rectangular regions are highlighted to denote areas for further analysis.

2.3. Pulsatile pressure system

An in-house pressure chamber system was designed and assembled to provide a 1 Hz PP condition to cells (figure 2a). An airtight 3.8 l commercial pressure cooker pot (01341, Presto, WI, USA) was modified and retrofitted with pressure sensors and solenoid valves. An in-house LabVIEW software script along with a data acquisition card (USB 6343; National Instruments, TX, USA) controlled two solenoid valves (ET-2-12; Clippard, OH, USA) via solid-state relays (SSR-40; TWTADE, Suzhou Taishun Electronics, Suzhou, China) to regulate in- and outflow of a humidified 5% CO₂/95% air gas mix to create a sinusoidal PP condition inside the pressure chamber. The waveform is generated by a momentary influx of pressurized 5% CO₂/95% air gas mix into the pressure chamber combined with a slower flow rate efflux via a 0.6 mm orifice. The shape of the waveform is sensitive to the influx rate and length of time of the pressurized gas and the diameter of the exit orifice. A Pegas 4000 gas mixer (Columbus Instruments, OH, USA) uses thermal mass flow controllers to supply an accurate 5% CO₂/95% air gas mix. A pressure transducer (PX26-GV and SEK002; Honeywell, NC, USA) controlled by an Arduino Uno controller (Arduino, MA, USA) continuously monitored and served to regulate the pressure inside the pressure chamber. The humidifier and pressure chamber were placed inside an incubator maintained at 37°C with humidified 5% CO₂ gas. The experimental volume where the cells were located was more than 95% water to decrease the volume of gas needed and to ensure 100% relative humidity (figure 2).

Figure 2.

Diagram of the (a) pulsatile and (b) SP systems. The gauge pressure within the pressure chamber was monitored via a pressure transducer to ensure the desired pressure values and waveforms.

2.4. Steady pressure system

The PP system was modified to generate a constant pressure condition (figure 2b). A syringe pump (PhD Ultra; Harvard Apparatus, MA, USA) in coordination with two solenoid valves periodically infused 37°C 5% CO₂/95% air gas or withdrew gas from inside the pressure chamber to maintain a constant pressure. A water reservoir inside the pressure chamber maintained constant 100% relative humidity for both the PP and SP systems.

2.5. Pressure and flow waveform conditions

LECs were exposed to either a normal pulsatile or steady lymphoedema fluid flow waveform in a PPFC for 48 h. The waveforms were generated using the LabVIEW software and outputted by the accompanying DAQ card (PCI-6035E, National Instruments, TX, USA). The voltage signal from the DAQ card regulated the angular velocity of the peristaltic pump (530U Watson Marlow, Falmouth, UK) to generate the desired waveform. A representative normal pulsatile waveform was based on our previous in vivo measurements [17], while the steady lymphoedema waveform characteristics are based on the Clec2^–/– mouse lymph flow measurements herein. All pressure measurements are reported as gauge pressure. At the inlet of the PPFC, the pressure was measured with a pressure transducer (PX26-005GV; Omega Engineering, CT, USA) and the flow rate was measured using an ultrasound sensor (ME5PXL Transonic Systems, Inc., Ithaca, NY, USA) [36]. The LECs were in the test section of the PPFC with width (w) and height (h) dimensions of 54 and 1 mm, respectively. The cross-sectional dimensions yield an aspect ratio of 1 : 54 creating a quasi-uniform flow field in the spanwise direction, such that all ECs experience equal WSS. By applying the continuity equation at the inlet where the velocity (V_inlet) is measured and using the cross-sectional areas at the inlet (A_inlet) and the test section (A_TS), the test section velocity (V_TS) can be determined as follows, V_TS = V_inlet × A_inlet/A_TS. Using V_TS, the WSS experienced by the LECs in vitro can be approximated with the Poiseuille relationship as follows, WSS = 3µV_TS/(h/2), where the dynamic viscosity (µ) of the cell medium is 7.8 × 10⁻⁴ m² s⁻¹. The pulsatile fluid flow waveform exposed LECs to a 1 Hz waveform of −0.14–0.49 dyn cm⁻² WSS and an approximate 9.5–11.7 cmH₂O pressure range with a mean pressure value of 10.55 cmH₂O (figure 3). The steady flow waveform exposed LECs to a relatively constant WSS of 0.3–0.35 dyn cm⁻² and an approximate pressure range of 10.56–10.78 cmH₂O with a mean of 10.69 cmH₂O. Although the selected WSS values for the different experiments are physiologically relevant, the WSS waveforms reflect the presence or absence of pulsatility and not specific WSS values, since significant inter- and intra-animal variations can occur even in the absence of conditions affecting the lymphatic system. In the absence of flow, LECs on well plates, cell culture inserts, coverslips and glass slides are exposed to AP (0 cmH₂O), SP (10.51–10.58 cmH₂O, mean 10.54 cmH₂O) or 1 Hz PP (8.8–12.2 cmH₂O, mean = 10.59 cmH₂O) conditions. The pressure sensor for the no-flow conditions did not account for the hydrostatic pressure due to the cell medium height, which accounts for, at most, an additional 0.4 cmH₂O of pressure.

Figure 3.

(a) Pressure measurements for LECs exposed to either atmospheric, cyan line, no-flow steady, black line, no-flow pulsatile, black star, SF, blue circle or PF, red square, conditions. (b) The LECs were exposed to either no-flow, cyan line, SF blue circle or PF, red square, WSS conditions. The no-flow pulsatile and no-flow SP conditions experienced a WSS magnitude of 0 dyn cm⁻² due to the absence of flow.

2.6. Endothelial permeability assay under pressure without flow

LECs were seeded on cell culture inserts (Corning Incorporated, NY, USA) with 1.0 µm sized pores and placed in 24-well plates (Trueline, NJ, USA). Cells were then exposed to either the AP, SP or PP condition for 48 h. Then, the medium of the upper chamber was replaced with 150 µl of medium with 10 kDa FITC-Dextran (TCI, OR, USA) at a 1 : 40 dilution. Cells were incubated for 1 h at 37°C, inserts were discarded and then 100 µl of medium was transferred to a 96-well plate in triplicate, and fluorescence was measured with a plate reader (BioTek, VT, USA). Experiments were repeated using 10 µM of the MYLK inhibitor ML-7 (Abcam, MA, USA). Permeability was approximated as follows:

$$\mathrm{Permeability}={\displaystyle \frac{(C_B\times V)}{(C_T\times A\times t)}.}$$

where A is the permeable surface area of the insert which cells are attached to, C_T is the concentration of FITC-Dextran in the upper chamber at time equal 0, C_B is the concentration of FITC-Dextran in the bottom chamber at the end of the assay, V is the volume in the bottom chamber at the end of the assay and t is the length of time of the assay [37].

2.7. Quantitative reverse transcription polymerase chain reaction

Reverse transcription and quantitative real-time polymerase chain reaction (qRT-PCR) were conducted using the CFX Connect Real-Time Polymerase Chain Reaction Detection System (Bio-Rad Laboratories, CA, USA). Total RNA was extracted from cells with the RNeasy Plus Mini Kit isolation reagents (QIAGEN, Hilden, Germany). Extracted RNA was reverse transcribed to complementary DNA using iScript gDNA Clear cDNA Synthesis Kit (Bio-Rad Laboratories, CA, USA) and amplified with the iQ SYBR Green Supermix (Bio-Rad Laboratories, CA, USA) to identify relative expression of MYH9, MYH10, FLNB, MYL9, MYL12a and MYL12b mRNA and normalized to GAPDH. The primer sequences are shown in table 1.

Table 1.

Primers used in qRT-PCR.

gene	locus ID	forward primer (5′−3′)	reverse primer (3′−5′)	fragment (bp)
GAPDH	NM_002 046.1	TGTAGTTGAGGTCAATGAAGGG	ACATCGCTCAGACACCATG	143
MYL9	NM_181 526.1	TGATGGCTTCATTGACAAGGAG	TCGTCCACTTCCTCATCTGT	126 or 288
MYH9	NM_002 473.1	CAAGACCGAGAAGATCAATCCA	CCACGTACGCCAGATACTG	97
MYH10	NM_001 256 012.3	CAGGGATGAGCAGAATGAAGA	CTTTGAAGCTACAGCAAGCG	112
MYL12B	NM_033 546.1	CCTAACGCTCTTCGCTGTC	TTGGATGTTGCACGCTGA	144
MYL12A	NM_006 471.1	ATTTCACCATGTTCCTCACCA	GCTCTCTCAAGTAATCTTCCTGT	132
FLNB	NM_001 164 317.4	ACTTCGTGGTAGAATCCATTGG	GTCGTTGTACTCAATCTTTGCC	89

2.8. Western blot analysis

Protein was extracted by lysing the cells for 1 h in 10% RIPA (Cell Signaling Technology, Danvers, MA, USA) containing MG132 protease inhibitor (Tocris Bioscience, Bristol, UK), Halt protease inhibitor (Life Technologies, Carlsbad, CA) and Halt phosphatase inhibitor (Life Technologies, Carlsbad, CA) at 4°C and centrifuged at 15 000g. Twenty micrograms of protein was loaded onto 10% SDS-PAGE gels (Bio-Rad Laboratories, CA, USA) and transferred to polyvinylidene difluoride (PVDF) membranes (Invitrogen, Carlsbad, CA, USA). PVDF membranes were incubated in a blocking buffer of tris-buffered saline (TBS) with 0.1% Tween-20 and 5% non-fat dry milk for 1 h at room temperature and probed overnight at 4°C with the primary antibody (table 2). The membrane was washed and incubated for 1 h at room temperature with HRP-conjugated antibody (table 2). The protein signal was detected with the Femto-Chemiluminescent substrate (Thermo Fisher Scientific, MA, USA), imaged with the ChemiDoc Touch imager (Bio-Rad Laboratories, CA, USA) and analysed with Image Multi-Gauge Software (Fujifilm, Tokyo, Japan). All measurements were normalized to the levels of β-actin or β-tubulin.

Table 2.

Antibodies used in Western blotting.

antibody target	vendor	catalogue no.	working dilution
beta-tubulin	Santa Cruz Biotechnology	166 729	1 : 1000
beta-actin	Life Technologies	AM4302	1 : 20 000
phospho-MLC 2 (Ser19)	Cell Signaling Technology	3675S	1 : 200
phospho-MLC 2 (Thr18/Ser19)	Cell Signaling Technology	3674S	1 : 200
filamin B	Cell Signaling Technology	12979S	1 : 1000
anti-mouse HRP	Thermo Fisher Scientific	31430	1 : 5000
anti-rabbit HRP	Thermo Fisher Scientific	31460	1 : 5000
anti-MYL9/MYL12A/B	Santa Cruz Biotechnology	SC-48414	1 : 1000

2.9. Immunofluorescence

Fixed cells were permeabilized with 0.1% Triton X in TBS for 15 min and then incubated in 10% w/v bovine serum albumin–TBS blocking buffer for 1 h. Cells were incubated with a primary antibody (table 3) blocking buffer solution overnight at 4°C. LECs were washed with TBS three times for 5 min on an orbital shaker. TBS was used instead of PBS, as PBS is suboptimal for antibodies that target phosphorylated proteins. Cells were incubated with a secondary antibody blocking buffer solution for 1 h at room temperature. For actin staining, LECs were incubated with a 1 : 40 Texas-Red phalloidin (Invitrogen, MA, USA) blocking buffer solution for 1 h and washed with TBS three times for 5 min. For nuclear staining, LECs were incubated with a 1 : 10 000 Hoechst (33342, MilliporeSigma, MA, USA) blocking buffer solution for 10 s, washed with TBS three times for 5 min and mounted with Fluoroshield (Sigma-Aldrich, MO, USA) mounting medium. In some cases, the brightness and contrast of IF images were adjusted to emphasize the localization of a protein of interest. Therefore, IF results herein are only used to determine the localization of proteins, while Western blot results provide protein quantification results. The Pearson correlation coefficient to quantify the colocalization of proteins was determined with ImageJ software using the JACoP plugin.

Table 3.

Antibodies used in IF.

antibody target	vendor	catalogue no.	working dilution
phospho-MLC 2 (Thr18/Ser19)	Cell Signaling Technology	3674S	1 : 200
phospho-MLC 2 (Ser19)	Cell Signaling Technology	3675S	1 : 200
VE-cadherin	Novus Biologicals	MAB9381	1 : 500
anti-rabbit 649	Novus Biologicals	NBP1-72962	1 : 2000
anti-mouse GFP	Invitrogen	A28175	1 : 2000

3. Results

3.1. Lymphoedema causes loss of pulsatility and increased mean velocities

Figure 4a shows the measured velocity vector field for the normal mouse at t = 16.2 s and t = 21.4 s, corresponding to the minimum and maximum average velocity time points within the 25 s data acquisition period, respectively. At the minima, flow moves from right to left in the retrograde direction, with higher velocity near the vessel core and the highest velocity is −190.4 µm s⁻¹ near the valve. At the maxima, the flow is moving in the antegrade direction with the flow accelerating through the valves and reaching a high velocity of 349.5 µm s⁻¹. The presence of the valve yields a spatially accelerating flow field with a slow-moving recirculation zone next to the valve. By contrast, the vessel affected by lymphoedema is greatly distended with flow accelerating up to 1418.5 µm s⁻¹ on the right side, where the lumen area decreases by 19% (figure 4b). The incompetent valve does not affect the flow enabling a more uniform flow field in the radial and longitudinal directions, while the valve in the normal vessel introduces a spatial non-uniformity in the flow field. Vessel contractility of three different segments within the normal vessel demonstrated higher frequency smaller amplitude contractions of up to 8.2% superposed over less frequent larger amplitude contractions (figure 4c,e). By contrast, two different segments within the lymphoedema vessel showed imperceptible contractility of the vessel wall (figure 4d,f). The segment mean velocity, U̅, for the normal vessel has maxima and minima of 44.7, 55.1, 71.7 µm s⁻¹ and −45.1, −51.2, −52.1 µm s⁻¹, respectively, for segments 1, 2 and 3, respectively (figure 4g). The similarity in synchronized mean velocities between the two lymphangions is likely caused by the incomplete closure of the bi-leaflet valve commonly observed at this site for these experiments. The pulsatility index (PI) was used to determine the level of deviation from the lymph segment mean velocity by calculating PI = |(maximum of U̅ - minimum of U̅)/mean of U̅ | for the time periods considered. Although yielding net antegrade flow, the values of U̅ fluctuate between positive and negative values due to the combination of anterograde and retrograde flow throughout the cycle of the normal vessel yielding PI values of 9.3, 8.7 and 8.7 for segments 1, 2 and 3, respectively, of the normal mouse. By contrast, the lymphoedema vessel lacks the retrograde component observed in the normal vessel and U̅ fluctuates between 277.7–283.5 µm s⁻¹ and 276.9–282.7 µm s⁻¹ yielding lower PI values of 0.003 for both segments 1 and 2 (figure 4h).

Figure 4.

The lymph velocity vector field was calculated for the (a) normal mouse at the minimum and maximum mean velocity time points of 16.2 s and 21.4 s, respectively, and for the (b) lymphoedema mouse at a representative mean velocity at the time point 0.5 s, inside lymphatic vessels distal from the inguinal lymph node. Schematics of the lymphangion highlight segments within the (c) normal and (d) lymphoedema mice that were interrogated to determine the normalized diameter, D/D_max, where the local diameter is divided by the maximum lymphatic diameter within the region of interest for the (e) normal and (f) lymphoedema mice. Similarly, the segment mean velocity, U̅, was determined for the (g) normal and (h) lymphoedema mice.

3.2. Steady pressure and steady flow upregulate mRNA expression of myosin light chains MYL9 and MYL12B

LECs were exposed for 48 h to AP, SP and PP conditions without flow, and SF and PF conditions to determine the role of pulsatile mechanical stresses on MLC mRNA expression. As MLCs are encoded by several genes, the expression of MYL9, MYL12A and MYL12B mRNA was quantified. The mRNA expression of MYL9 and MYL12B mRNA was significantly upregulated in response to SP and SF compared to AP, PF and PP (figure 5). MYL9 mRNA expression increased by 1.9 and 2.3 fold (p < 0.001), while MYL12B increased by 1.3 (p < 0.01) and 1.2 (p < 0.001) fold in response to SP and SF, respectively. MYL12A was significantly upregulated 2.1-fold (p < 0.01) exclusively by SP.

Figure 5.

mRNA expression of (a) MYL9, (b) MYL12A, (c) MYL12B, (d) MYH9, (e) MYH10 and (f) FLNB for LECs exposed to PF, AP, SP, PP and SF. Data are normalized to GAPDH and expressed as mean ± s.e.m. fold of AP. n = 3–6 ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, n.s., non-significant.

The mRNA expression of myosin heavy chains MYH9 and MYH10 was also quantified, yet no significant expression changes were observed in response to any condition. As actin cross-linkers contribute to actomyosin-mediated cellular stiffening, mRNA expression of the actin cross-linker FLNB was examined. FLNB is an isoform of FLNA and FLNA was previously shown to be upregulated in response to pressure [29]. However, no FLNB mRNA expression change was measured in response to any experimental condition. Taken together, these results suggest that MYL9 and MYL12B are sensitive to SP and SF yet are not influenced by the lower atmospheric SP, PF or PP.

3.3. Steady pressure and steady flow promote myosin light chain phosphorylation

The effect of AP, SP and PP conditions without flow, and SF and PF conditions on MLC protein levels and phosphorylation was also tested. MLC protein levels did not significantly change after exposure to any condition (figure 6a). As an increased MLC phosphorylation state is a hallmark of endothelial hyperpermeability and inflammation, the single and double phosphorylation states of MLC were assessed (figure 6b,c). The quantity of pMLC in LECs exposed to SP and SF was significantly upregulated by 1.3-fold (p < 0.05) for each condition, respectively, in comparison to AP. By contrast, pMLC was not significantly upregulated for LECs exposed to PP and PF. Following a similar trend, both SP and SF caused significant increases of 10.8 and 9.6 fold (p < 0.05), respectively, in the quantity of LEC ppMLC compared to AP, while no significant changes were observed due to the PP and PF conditions. Therefore, SP and SF increase pMLC and ppMLC, suggesting a contractile phenotype with increased permeability.

Figure 6.

Protein levels of (a) MLC, (b) pMLC and (c) ppMLC for LECs exposed to PF, AP, SP, PP and SF. Data are normalized to β-actin or β-tubulin and expressed as mean ± s.e.m. fold of AP. n = 3–9, ANOVA, *p < 0.05, **p < 0.01, n.s., non-significant.

3.4. Steady pressure and steady flow promote phosphorylated myosin light chain to colocalize with peripheral actin network and disrupt barrier function

Colocalization of phosphorylated MLC with the peripheral actin network disrupts barrier function [26]. To determine whether fluid mechanics stresses modulate MLC phosphorylation and actin network formation in the cell periphery, LECs were exposed to the five different experimental conditions, and protein localization was assessed via IF and staining (figures 7–10). Under exposure to AP, pMLC was observed throughout the cytoplasm consistent with previous reports of BECs (figures 7 and 8) [26]. These cells were also characterized by a continuous presence of VE-cadherin at the junctions. Consistent with previous studies, the actin stain revealed prominent circumferential actin filaments and negligible stress fibre formation [38] (figure 8). ppMLC was sparsely distributed and localized in the periphery (figures 9 and 10). LECs exposed to PP showed central actin stress fibres (figures 8 and 10). pMLC colocalized in the cytoplasm with the central stress fibres, while ppMLC was sparsely present in the periphery, similar to LECs exposed to AP.

Figure 7.

SP and SF promote pMLC peripheral distribution. Epifluorescence images of LECs exposed to AP, PP, SP, SF and PF for 48 h. Nuclei were detected with DNA stain, while VE-cadherin and pMLC were detected with antibodies. White arrows indicate endothelial intercellular gap formation. Brightness and contrast were adjusted to emphasize cell morphology and protein localization. Scale bar: 75 µm.

Figure 10.

SP and SF promote ppMLC and actin peripheral distribution. Epifluorescence images of LECs exposed to AP, PP, SP, SF and PF for 48 h. Nuclei and actin were detected with DNA and F-actin stains, respectively, while ppMLC was detected with antibodies. Brightness and contrast were adjusted to emphasize cell morphology and protein localization. Scale bar: 75 µm.

Figure 8.

SP and SF promote pMLC and actin peripheral distribution. Epifluorescence images of LECs exposed to AP, PP, SP, SF and PF for 48 h. Nuclei and actin were detected with DNA and F-actin stains, respectively, while pMLC was detected with antibodies. Brightness and contrast were adjusted to emphasize cell morphology and protein localization. Scale bar: 75 µm.

Figure 9.

SP and SF promote ppMLC peripheral distribution. Epifluorescence images of LECs exposed to AP, PP, SP, SF and PF for 48 h. Nuclei were detected with DNA stain, while VE-cadherin and ppMLC were detected with antibodies. White arrows indicate endothelial intercellular gap formation. Brightness and contrast were adjusted to emphasize cell morphology and protein localization. Scale bar: 75 µm.

LECs exposed to both SP and SF showed similar morphological responses. Actin assembly was observed primarily in the peripheral cortex with minimal central stress fibre assembly. pMLC and ppMLC both colocalized with peripheral actin. Disruption of VE-cadherin protein at the junctions highlighted intercellular gaps. The peripheral pMLC and ppMLC localization and gap formation suggest significant cellular tension.

Similar to the PP condition, PF caused pMLC colocalization with central actin stress fibres. ppMLC colocalized with central stress fibres as well as peripheral stress fibres. The continuous presence of VE-cadherin at the junctions did not suggest cellular tension or gap formation characteristic of a cytoskeletal tension-driven permeability increase.

Further analysis of the IF results with Pearson correlations showed weak colocalization of pMLC and ppMLC with VE-cadherin for all conditions, while greater colocalization of pMLC and ppMLC with actin (table 4). Overall, SP and SF induced intercellular gap formation, while PP and PF did not exhibit intercellular gap formation, suggesting that pulsatility protects LEC barrier function.

Table 4.

Pearson correlation coefficient for IF colocalization of MLC with actin or VE-cadherin.

MLC phosphorylation state	colocalization target	AP	PP	SP	SF	PF
pMLC	VE-cadherin	−0.34	−0.124	0.034	0.138	−0.131
ppMLC	VE-cadherin	0.184	0.078	0.047	0.119	0.078
pMLC	actin	0.35	0.675	0.867	0.828	0.867
ppMLC	actin	0.389	0.587	0.821	0.791	0.349

3.5. Exposure of lymphatic endothelial cells to steady pressure, but not pulsatile pressure, causes barrier function loss that is rescued by myosin light chain kinase inhibitor ML-7

An in vitro transwell assay was used to determine the functional change of permeability, via flux of FITC-labelled dextran, in LEC monolayers after exposure to 48 h of AP, SP and PP conditions without flow (figure 11). SP induced a significant increase of 1.8 fold (p < 0.001) in LEC permeability compared to AP. LECs exposed to PP showed no significant change in permeability compared to those exposed to AP. As increased EC permeability is often the result of MLC phosphorylation via MLCK, each experimental condition was simultaneously performed using the MLCK inhibitor ML-7. LECs treated with ML-7 exposed to AP, SP and PP showed no significant change in permeability. These data suggest that SP, but not PP or AP conditions, caused increased permeability that is rescued by MLCK inhibition.

Figure 11.

SP increased the permeability of LECs. LECS were cultured on transwell chambers exposed to AP, SP or PP for 48 h in the absence or presence of the MLCK inhibitor ML-7. Permeability was measured via flux of FITC-labelled dextran. Data are expressed as mean ± s.e.m. fold of AP. n = 4–8, ANOVA, ***p < 0.001, n.s., non-significant.

4. Discussion

Pulsatility is inherent in the blood circulatory and lymphatic systems. A consequence of the loss of pulsatility in some blood vascular beds is BEC barrier function disruption. Here, we demonstrated that lymph flow affected by lymphoedema can be quasi-steady, LECs are responsive to physiologically relevant pressure conditions in the presence and absence of flow under steady and pulsatile conditions, and loss of pulsatility increases permeability via an actomyosin-driven mechanism. The results suggest that the loss of pulsatility that can occur due to lymphoedema may disrupt LEC barrier function and may further contribute to interstitial oedema.

Normal lymph flow is pulsatile, while lymphoedema can promote quasi-steady lymph flow with the loss of pulsatility, as noticed in the in vivo results herein, where vessel contractility was nearly absent, and flow was relatively steady [2,17]. In vitro, the absence of pulsatility in the SP and SF conditions caused upregulation of MLCs MYL9 and MYL12B mRNA expression, suggesting that the stress due to pressure is sufficient to upregulate mRNA expression since both the flow and no-flow conditions experience approximately the same pressure. Although the pressure values in this study are lower than the pressure values in studies examining the response of BECs to pressure, LECs can respond to mechanical stimuli an order of magnitude lower than BECs [39,40]. Furthermore, mesenteric collecting vessels can discriminate between pressure changes of 1–9 cmH₂O [41,42]. The MYL9 and MYL12B mRNA upregulation due to the loss of pulsatility suggests that transcriptional regulatory mechanisms can discriminate between SP and PP. It is not surprising that LECs can discriminate between PP and SP, since LECs respond differently to pulsatility in other mechanical stimuli. For instance, lymphatic collecting vessel contraction occurs when LECs sense PF, while contractions are inhibited by SF even at similar mean WSS values [43]. This response is impaired when LECs are denuded, demonstrating that LECs are the primary mechanosensors.

Not only can LECs sense the loss of pulsatility, but also differentiate between WSS and pressure. MYL12A mRNA was upregulated by SP but not SF, suggesting that WSS and pressure influence expression independently, while MYH9, MYH10 and FLNB mRNA expressions remained unchanged. Although WSS and pressure regulated the transcriptional expression of MYL9, MYL12A and MYL12B, the total amount of MLC protein did not significantly change after exposure to any of the experimental conditions. However, post-translational modification of MLC showed that the amount of mono- and di-phosphorylated MLC was significantly increased after exposure to SP and SF. This concurs with previous reports of increased MLC phosphorylation in BECs exposed to SP [28,29]. PP and PF did not induce significant MLC phosphorylation.

In ECs, stress fibres consist of polymerized actin and myosin filaments that can regulate cell contractility by the level of MLC phosphorylation [44]. ECs respond to undisturbed blood flow by forming stress fibres that align with the longer cell axis and preserving barrier function, while disturbed blood flow does not promote stress fibre formation [45] and coincides with barrier function disruption [46–48]. The LECs exposed to the PF condition displayed stress fibre formation that tended to align with the longer cell axis, while LECs exposed to SF formed less stress fibres. In the venous system, similar results have been observed where lower pulsatile venous blood flow caused BECs to form less stress fibres [49]. SF and SP promoted peripheral actin bundle formation and colocalization with pMLC and ppMLC along with VE-cadherin disruption and gap formation. Likewise, bovine pulmonary artery ECs exposed to steady WSS showed preferential peripheral colocalization of ppMLC and actin [50]. In BECs, thrombin induced MLC di-phosphorylation and actin bundle formation at the periphery, VE-cadherin junction interruption, and a decrease in transendothelial electrical resistance, a metric of endothelial barrier disruption [26]. LECs under PP and PF maintained VE-cadherin at the junctions.

During the different stages of lymphoedema, lymphatic vessels can experience loss of pulsatility, a condition shown here to promote pMLC, ppMLC and actin filament colocalization in the cell periphery. Increased cytoskeletal tension via MLC phosphorylation in ECs can induce hyperpermeability [51], potentially exacerbating interstitial oedema under lymphoedema where the loss of pulsatility is possible. In the absence of flow, permeability was increased with the loss of pulsatility agreeing with findings from BECs where SP increased permeability via MLC phosphorylation [28,29]. The permeability increase due to SP exposure was rescued via MLCK inhibition, suggesting an MLCK-dependent mechanism. By contrast, LECs exposed to PP, with a similar mean pressure value as the steady condition, did not exhibit the permeability increase, increased MLC phosphorylation state, or intercellular gap formation observed in LECs exposed to SP.

SP induced peripheral re-organization of actin, while pulsatility caused stress fibre formation throughout the LECs. A potential mechanotransduction pressure pathway mediating actin reorganization is the fibroblast growth factor-2 (FGF-2) pathway. The secretion of FGF-2 by BECs is elevated in patients with pulmonary hypertension where blood pressure in the pulmonary artery is increased in this PF environment [52]. Exposure to PP without flow causes endothelial tyrosine phosphorylation of the FGF-2 receptor and actin filament re-organization [53]. In conditions with steady elevated pressure and without flow, endothelial FGF-2 binding affinity increases, although in a nonlinear fashion [54]. The importance of FGF-2 on the endothelial cytoskeleton was confirmed after incubation of BECs with FGF-2 and the consequent re-organization and disruption of the actin cytoskeleton [55]. The FGF-2 signalling pathway is a potential mechanotransduction pressure pathway that deserves to be further explored, since it is responsive to pressure and is involved in proliferation and vessel remodelling, hallmarks of advanced lymphoedema.

It is important to consider that the effects of lymphoedema in different lymphatic vessels are highly heterogeneous and some vessels may experience increased flow rates and loss of pulsatility in the earlier stages, and near absence of flow and pulsatility in the latter stages of the disease [2,14,15,19,56,57]. The findings of this study emphasize the sensitivity of LEC barrier function and hyperpermeability to changes in flow and pressure. One obvious consequence of LEC hyperpermeability is reduced lymph transport due to a decrease in the lymph pressure gradient necessary for propelling lymph. In a state of increased LEC permeability, lymph can exit the collecting lymphatic vessels exacerbating interstitial oedema and potentially disrupting the transport of antigens and antigen-presenting cells to lymph nodes, possibly weakening an adaptive immune response [58]. Induced LEC hyperpermeability in mice caused inflammation in local tissue due to antigen leakage [59]. Thus, increased LEC hyperpermeability could explain symptoms and clinical complications observed in lymphoedema patients. The regulation of permeability in collecting vessels and its role in lymphatic function requires further study.

Although the results herein highlight the importance of pulsatility in the LEC phenotype, there are limitations. The LECs were grown on glass slides, which are not representative of tissue stiffness values, and it is known that pMLC colocalization with actin stress fibres and barrier function disruption in BECs can vary when cultured on softer substrates in response to inflammatory mediators [60]. Furthermore, the presence of a soft substrate that could deform under pressure would allow the LECs to experience strain akin to an in vivo vessel environment. Although the results for the PP and PF conditions agreed, and the results for the SP and SF tended to also be similar, it could be beneficial to build a chamber where the flow and pressure conditions could be tested, and the permeability could be measured for all conditions. However, flow chambers are not ideal for static experiments, since the Hele–Shaw fluid flow conditions require very large width-to-height aspect ratio channels that limit the amount of stagnant cell medium available for the cells. Permeability values were approximated as detailed in Frost et al. [37] 2019 using measurements from only two time points, which can miss nonlinear dynamics present in transendothelial transport. Transendothelial transport undersampling more closely represents leakage across the endothelium. Velocity measurements in the lymphatic vessels may be skewed and appear irregular as also observed in Pujari A et al. (2020) [17], if out of plane particles are included in the velocity measurements due to the finite depth of focus of the microscope objectives in combination with the highly three-dimensional flow field generated by the valves.

While the findings of this study need to be corroborated in vivo, these observations suggest potential mechanical and pharmaceutical therapeutics to ameliorate the effects of lymphoedema. Mechanical therapeutics could provide direct or indirect pulsatile stimulus to restore endothelial barrier function (figure 12). Although inelastic bandages, graduated compression stockings, multi-layered wraps and pneumatic compression devices have been shown to effectively regulate volume in lymphoedema [61,62], these treatments are not without potential complications and are not designed to promote the natural pulsating lymph flow that may restore partial lymphatic function. This is especially relevant in the advanced stages of lymphoedema where contractions of lymphatic collecting vessels and leg muscular contractions fail to generate pressure gradients capable of propelling lymph [2]. MLCK could be considered a potential therapeutic target. Inhibition of MLCK has attracted interest in treating several vascular conditions [63]. Interestingly, the total deletion of MLCK in rat ECs caused no observed deleterious effects [64]. Despite this, due to the importance of MLCK in numerous other cell types, the inhibition of MLCK is associated with significant side effects; therefore, MLCK inhibitors have not been approved for use in human patients. In the future, targeted drug delivery of an MLCK inhibitor to LECs or a pulsatile mechanical stimulus may serve as a potential therapy for lymphoedema patients.

Figure 12.

(a) LECs were exposed to perpendicular (pressure) and/or tangential (WSS) stress(es). (b) Stress fibre distribution depended on LEC exposure to AP, PP, SP, SF or PF. The loss of pulsatility in steady conditions promoted peripheral distribution of stress fibres, endothelial barrier disruption and increased permeability.

Ethics

All animal experiments were approved by the University of Pennsylvania Institutional Animal Care and Use Committee.

Data accessibility

This article has no additional data.

Authors' contributions

J.D.H.: conceptualization, data curation, formal analysis, investigation, methodology and writing—original draft; S.F.: investigation; R.B.G.: investigation; A.P.: data curation, formal analysis and software; D.T.S.: conceptualization and investigation; M.L.K.: funding acquisition, project administration, resources and supervision; J.M.J.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, software, supervision, validation, writing—original draft and writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported by the National Heart, Lung, and Blood Institute (NHLBI) grants no. K25 HL107617 (to J.M.J.), T32 HL007439 (to D.T.S) and R01 HL103432 (to M.L.K.), and National Institute of General Medical Sciences (NIGMS) grant no. T32 GM135096 (to J.D.H.).