Variation in wall shear stress in channel networks of zebrafish models

Abstract

Physiological functions of vascular endothelial cells (ECs) vary depending on wall shear stress (WSS) magnitude, and the functional change affects the pathologies of various cardiovascular systems. Several in vitro and in vivo models have been used to investigate the functions of ECs under different WSS conditions. However, these models have technical limitations in precisely mimicking the physiological environments of ECs and monitoring temporal variations of ECs in detail. Although zebrafish (Danio rerio) has several strategies to overcome these technical limitations, zebrafish cannot be used as a perfect animal model because applying various WSS conditions on blood vessels of zebrafish is difficult. This study proposes a new zebrafish model in which various WSS can be applied to the caudal vein. The WSS magnitude is controlled by blocking some parts of blood-vessel networks. The accuracy and reproducibility of the proposed method are validated using an equivalent circuit model of blood vessels in zebrafish. The proposed method is applied to lipopolysaccharide (LPS)-stimulated zebrafish as a typical application. The proposed zebrafish model can be used as an in vivo animal model to investigate the relationship between WSS and EC physiology or WSS-induced cardiovascular diseases.

1. Introduction

Vascular endothelial cells (ECs), which are metabolically active monolayers, form the inner surface of blood vessels. ECs are in direct contact with blood flowing inside the vessel [1]. Wall shear stress (WSS) is the frictional force generated by the blood flow on the vessel wall, and the EC layer is continually exposed to various WSS conditions. The EC layer reacts to the variation of WSS via mechanotransduction by changing gene regulation and cell phenotype [2,3]. EC reaction to WSS determines the physiology or pathology of various cardiovascular systems [1]. Atherosclerosis, the main cause of heart attack and stroke [4], is predominantly observed in low-WSS regions, such as bifurcated or curved blood vessels [5]. Low WSS triggers phenotypic changes of ECs. As a result, the corresponding blood vessel enters a prothrombotic state which results in easy formation of platelet aggregates and fibrin plugs and leads to the early stage of atherosclerosis [6]. WSS is actually an essential stimuli for embryonic cardiogenesis [7], wound-healing angiogenesis [8] and leucocyte–EC interaction, which is an important process in inflammation [9].

The effects of WSS on ECs have been investigated using various in vitro and in vivo models [10]. Microchannels embedded with an EC monolayer have been widely used in in vitro studies [11]. Booth et al. [12] created multiple channels, providing a wide range of WSS conditions (0–86 dyne cm⁻²), to analyse permeability, transendothelial electrical resistance, morphometry and protein expression of EC layer depending on the WSS magnitude. Ploppa et al. [13] used an in vitro channel to investigate the effect of WSS on the rolling and adhesion of leucocytes when ECs or leucocytes are activated through lipopolysaccharide (LPS) treatment. Although these in vitro studies provided considerable insights, natural physiological environments are difficult to be fully simulated in these artificial channels [14]. For in vivo studies, pigs [15], rabbits [16,17] and mouse [18] have been used as animal models to reveal the roles of WSS on EC functions or WSS-related vascular diseases. Two types of studies have been conducted using these animal models. The first type compares physiological phenomena at several vessels under different WSS conditions [19]. The second type controls the inherent WSS condition through surgical manipulation using arteriovenous fistula [20], vascular graft [21], ligation [22] and perivascular devices [23]. The first type does not provide reproducible and controllable WSS conditions [12]. Both types of studies commonly require post-mortem examination to investigate the physiological variations in ECs. However, post-mortem examination has technical limitations in studying temporal variation of WSS-related pathology. The unavailability of a suitable experimental model because of these technical limitations is one of the main obstacles in studying the relationships between WSS magnitude and pathologic phenomena in the EC layer.

Zebrafish (Danio rerio), which is a tropical freshwater fish, has been receiving considerable attention as a cardiovascular disease (CVD) model for the studies of embryological development [24] or pathology of circulatory vascular diseases [25] because genetic manipulation is easy to treat in zebrafish [26]. Moreover, the optical clarity of zebrafish enables real-time monitoring of developing pathologies. Zebrafish has been used to investigate the roles of WSS in vascular development [27] and cardiogenesis [7]. In our previous study [28], we investigated the relationship between low WSS and cholesterol deposit using hypercholesterolemic zebrafish models. Results were in good agreement with that of clinical studies, validating the feasibility and usefulness of zebrafish as a model to study WSS-related physiologies. However, isolating the effects of WSS is difficult because the relationship between WSS magnitude and cholesterol deposit was investigated using several different blood vessels.

In this study, a new experimental technique which can control WSS at a caudal vein of zebrafish is proposed. The accuracy and reproducibility of the proposed method are validated using an equivalent circuit model. The typical use of the proposed method was followed wherein the number of rolling cells in blood vessels of the LPS-stimulated zebrafish model is investigated under controlled WSS conditions. The caudal vein of a zebrafish is selected as the target vessel because previous studies already reported physiological events, including lipid accumulation, lipid uptake through macrophage and variation of EC permeability in the vessel [28–30].

2. Material and methods

2.1. Zebrafish and blood vessels

Transgenic (fli1a:EGFP) zebrafish whose ECs express green fluorescent protein (GFP) are provided by the Korea Zebrafish Organogenesis Mutant Bank (Daegu, Korea). Zebrafish larvae are raised at room temperature of 28 ± 0.5°C under a 14 h L : 10 h D cycle. Five days post fertilization (dpf) zebrafish are used in this study. Intersegmental vessels (ISVs) are vertically aligned as shown in the angiogram (figure 1a) [31]. Normally, 22 ISVs are positioned between the right end of the swim bladder and the end of the tail in the stage. The 22 ISVs are classified into four groups and named as ISV1, ISV2, ISV3 and ISV4 (figure 1a). The ISV1, ISV2 and ISV3 have five ISVs, respectively. The remaining seven ISVs are allocated to the ISV4. The main artery and vein are also divided into four regions using three partitions, as denoted by dotted vertical lines in figure 1a. The three partitions are located at the position of the first ISV of the three ISV groups (ISV2–4). The four regions are named in regular order from the first artery region (A1) to the fourth vein region (V4). Each region (A1–V4) is divided into several parts, depending on the position of ISVs. For example, vein 2 (V2) is divided into five parts (V2(1)–V2(5)) by five ISVs in the group ISV2.

Figure 1.

(a) Angiogram of 5 days post-fertilization (dpf) zebrafish. Intersegmental vessels (ISVs) are classified into four groups and named as ISV1, ISV2, ISV3 and ISV4. The first three ISV groups (ISV1–3) have five ISVs, respectively. The remaining seven ISVs are allocated to the ISV4. The main artery and vein are also divided into four regions using three dotted vertical lines. The three lines correspond to the position of the first ISV of the ISV2–4. The four regions are named from the first artery region (A1) to the fourth vein region (V4). A1–3 and V1–3 regions are divided into five parts by five ISVs at each region. The fourth region (A4 and V4) is divided into seven parts through the seven ISVs in the ISV4. Each part in A1, V2 and ISV3 is only represented for simple illustration. Five white circles near the ISV2 indicate injection sites of agarose solution for blocking five ISVs in the ISV2. (b) Schematic of blood-vessel network. Black arrows indicate the direction of blood flow. Q_A1(1) and Q_A3(1) represent the flow rate at A1(1) and A3(1). The schematic is drawn using Power point 2013 (Microsoft). (Online version in colour.)

2.2. Blocking blood flow in intersegmental vessels to change wall shear stress

To block blood flow in an ISV, 0.5% low-melting agarose solution is injected into tissues next to the middle of the target ISV. As an example, the injection sites for blocking five ISVs in ISV2 are represented in figure 1a (five white circles near ISV2). The needle with 1 ml syringe is used to put melted agarose in a capillary. Zebrafish is anaesthetized with a short exposure to 0.02% Tricaine before the injection; 2–4 nl agarose solution is injected next to the target ISV through the capillary with monitoring through a stereomicroscope (Zeiss Stemi 200-C, Germany). The glass capillaries (tip diameter: 5–10 µm, Hilgenberg, Germany) are enclosed in a manipulator (Marzhauser Wetzlar, Germany) connected to a micro-injection pump (FemtoJet; Eppendorf, Germany). The injected agarose gel compresses the blood vessels, blocking the blood flow in the vessel.

2.3. Micro-particle image velocimetry

A 5 dpf zebrafish is anaesthetized with 0.02% Tricaine and placed in a chamber (Chamlide TC; Live Cell Instrument, Korea) with 7% methylcellulose. A modified fluorescence microscope (Zeiss Axiovert 200, Germany) fitted with a 40× objective lens (NA = 0.6) is employed to acquire blood flow images and to observe the ECs of blood vessels. Each zebrafish is illuminated through a fluorescence lamp (X-Cite 120 Q; Lumen Dynamics, Ontario, Canada) with a shift-free filter to identify ECs expressing GFP. Flow images are consecutively acquired using illuminating halogen lamp at the centre plane of the vessel for 3 s, which corresponds to six to nine cardiac cycles. Blood flow images are captured at different frame rates ranging from 125 frames to 500 frames s⁻¹ using a high-speed camera (Fastcam SA1.1; Photron, USA), depending on the flow rate in each vessel. The effective pixel size is approximately 0.77 µm.

The captured flow images are assessed using a fast Fourier transform-based cross-correlation particle image velocimetry (PIV) algorithm. The PIV technique detects the displacements of tracer particles in each interrogation window by searching the peak location in the corresponding cross-correlation map [32]. RBCs in the blood are used as tracer particles in this study. The size of each interrogation window is 48 × 8 pixels, and adjacent interrogation windows are overlapped by 50%. A recursive correlation method with multiplication mode [33] is used, and the Gaussian peak fit is employed to increase measurement accuracy. In addition, the background image obtained by averaging sequential images is subtracted from the raw images [34]. PIV analysis is conducted using the PIVview-2C (PIVTEC, GmbH) software.

2.4. Identification of cell-free layer

The cell-free layer (CFL) is formed near the wall region of blood vessels because of the Fåhraeus−Lindqvist effect [35,36]. Thus, to assess WSS accurately, the CFL thickness must be calculated. In this study, the thickness of the CFL (T_CFL) is measured indirectly using digital image processing of blood flow images, because the surrounding tissues and organs make it hard to distinguish boundaries of individual RBCs and inner vessel wall. T_CFL is evaluated using the following equation:

2.1

where D_O, D_R and T_EC represent the outer diameter of the blood vessel, diameter of the RBC-rich region and the thickness of EC layer, respectively. In this evaluation, s.d. map is used to distinguish the RBC-rich regions [37]. Pixels in the RBC-rich regions have high s.d. values because RBC movement generates intensity variations. An iterative thresholding method is adopted to the s.d. maps to make binary images [38]. The binary images are used to measure D_R. D_O is measured from the captured GFP images of blood vessels. In this study, the thickness (T_EC) is regarded as 2 µm based on our previous study [28]. The experimental procedure for measuring CFL thickness is explained in detail in the electronic supplementary material, S1.

2.5. Wall shear stress measurement

The WSS acting on blood vessels is evaluated by assuming a linear velocity profile in the CFL. The assumption of Poiseuille flow which is conventionally used in cardiovascular flows cannot be applied to the blood vessels of zebrafish, because the velocity profile in a small vessel is known to be more blunt than the parabolic shape [39]. Several studies injected artificial tracer particles for PIV in circulatory systems to enhance measurement accuracy, but recent studies still use RBCs as tracer particles because the effect of the artificial particles in each animal models are not fully known [40,41]. Namgung et al. [42] suggested measuring WSS in small blood vessels (29.5–67.1 µm) with assuming a linear velocity profile in the CFL. They validated the measurement accuracy of the suggested technique through in vitro experiment using a micro-tube. The measured WSS values were compared with those predicted using the force balance law and pressure drop (ΔP) along the tube, measured by a pressure transducer (MP 100 System, CA, USA). As a result, the WSS values measured using the assumption of linear velocity profile are agreed within ±20% with that predicted using the measured pressure drop. The R²-value for the linear regression was about 0.76. Al-khazraji et al. [43] adopted the same assumption (linear velocity profile in the CFL) to measure mass flow rates in rat skeletal muscle arterioles (21–115 µm). The measured mass flow rates were reasonably accurate with 0.6 ± 3.2% error when they investigated mass balance law at bifurcations.

In this study, WSS (τ_w) in the main vein is estimated using the mean velocity profile averaged for 3 s because the blood flow in veins is relatively steady compared with the blood flow in arteries (figure 2a). The WSS in the main arteries is determined using phase-averaged velocity profiles obtained at each arterial regions (A1–4). The phase-averaging method is conducted after each cardiac cycle has been divided into 10 phases [28]. τ_w is calculated using the following equation:

2.2

Figure 2.

(a) Temporal variations of RBC mean velocity for 2.5 s. (b) Velocity profiles at V1(3) and V4(4). Red square dot: RBC mean velocity at the edge of RBC-rich region. Blue circle dot: mean RBC velocity measured using micro-PIV technique. Blue dashed line: polynomial fitting curve. Red solid line: velocity profile of plasma in CFL. (Online version in colour.)

where μ is the dynamic viscosity of plasma, and V_edge represents RBC velocity at the edge of the RBC-rich region (red square dots in figure 2b). Given that plasma is a Newtonian fluid, the dynamic viscosity can, therefore, be assumed as 1.2 cp because the dynamic viscosity of icefish plasma is similar to the viscosity of human plasma [44,45]. The V_edge is calculated by extrapolating the RBC velocities (blue circles in figure 2b) through polynomial curve fitting. The RBC velocities are measured through a micro-PIV technique.

2.6. Resistance in equivalent circuit model of blood vessels

The blood vessel networks are modelled as sets of resistance in electrical circuit by assuming that the vessels are rigid networks having circular cross section. The Reynolds number is small in blood vessels of zebrafish [28] that means blood flow in the vessels is governed by a Stokes equation as follows:

3.1

The term ρδu/δt of the unsteady Stokes equation is negligible because the characteristic time of the blood flow (0.333 s), which is calculated from the heart rate (≒3 Hz), is larger than the value (0.016 s) determined by dividing the characteristic length (≒16 µm, mean diameter of arteries) by the characteristic velocity (≒1 mm s⁻¹, mean velocity of RBCs in artery). The deformability of oval zebrafish RBCs [46] is assumed to be negligible.

The flow rate (Q) has a linear relation with the pressure drop (ΔP) along the blood vessels because the governing equations for blood flows in a zebrafish, equation (3.1) and continuity equation, are a linear equation [47]. The relation is calculated as follows:

3.2

where R is the resistance of blood vessels. Based on the relation between τ_w and ΔP in a circular pipe flow [48], R is calculated as follows (electronic supplementary material, S2):

3.3

where ū, l and D denote the mean velocity, length and diameter of the blood vessel, respectively. The equation (3.3) is valid regardless of the shape of velocity profile. The mean velocity of RBCs () is used instead of ū in this study. The reasonability of the substitution is described in the electronic supplementary material, S3. Among the 44 parts of the main artery and vein, R for the centre parts of A1–4 and V1–4 is calculated using equation (3.3). τ_w, , l and D are determined using PIV results and GFP images of the blood vessels. R for the other parts are estimated through polynomial interpolation of the R value for the centre parts. The resistance in various ISVs is derived using equation (3.2) and the division ratio of flow rate between main arteries and the ISVs at each bifurcation point. The flow rate in each blood vessel is calculated by multiplying and the cross-sectional area.

2.7. Lipopolysaccharide-stimulated zebrafish model

To create an LPS-stimulated zebrafish model, 0.05 mg ml⁻¹ LPS (Ultrapure LPS, E. coli 0111:B4; InvivoGen, USA) is injected into a vein next to the swim bladder of zebrafish anaesthetized with a short exposure to 0.02% Tricaine. The injection is conducted using the equipment described in §2.2.

2.8. Fluorescent microscopy

A modified Zeiss Axio Observer.Z1 epifluorescence microscope fitted with a 20× (Plan-Neofluar; NA = 0.5) objective lense and a Roper Scientific CoolSnap HQ CCD camera is used to capture GFP images of ECs (figure 1). An XBO 75 W/2 Xenon lamp (75 W, Osram) and GFP (EX BP470/40, BS 495, EMBP 525/50) filter sets are used for fluorescence imaging.

3. Results

3.1. Effect of blocking blood flow

Figure 1a illustrates an angiogram of a 5 dpf zebrafish. ISVs, an artery and a vein are divided into four parts using three partitions (dotted lines in figure 1a) as described in §2.1. Blood flow directions inside blood vessels are represented in figure 1b. The ISVs carry a portion of blood from the artery to vein [49,50]. Blood vessels where blood is flowing are visualized by grey images. The images are obtained by visualizing the s.d. values of light intensity in sequential flow images (figure 3a,b). The position of blood vessels with blood flow is bright because RBC movement increases the s.d. of light intensity values. To block blood flow in certain ISVs, agarose gel is injected near the ISVs. Figure 3b shows typical results showing the effect of blocking blood flow in five ISVs in ISV2. Blood flow variation after agarose injection is depicted in figure 3b. The effect of blocking blood flow on the inlet flow rate entering the main artery (Q_A1(1)) is shown in figure 3c. The control in figure 3 indicates a zebrafish model without blood flow blocking, in other words, normal zebrafish. The flow rates in the region A1(1) at different times are normalized with the mean value of Q_A1(1) for the control model. The reason for this normalization is described in the electronic supplementary material S3. The normalized Q_A1(1) for the control model sustains the value for 1 h after the blood flow has been blocked in the ISV2. Figure 3d represents the variation in the ratio of the flow rate Q_A3(1) in A3(1) to the flow rate Q_A1(1) in A1(1). The Q_A3(1)/Q_A1(1) value indicates the portion of the maintained arterial flow rate as the blood flows from A1(1) to A3(1). By blocking the blood flow in the ISV2, the value increased and then decreased at 60 min after blocking.

Figure 3.

Effect of blocking blood flow in the ISV2. Blood flow is visualized using a standard deviation projection method (a) before and (b) after blocking blood flow in the ISV2. (c) Temporal variation of normalized Q_A1(1) for 1 h after blocking the blood flow in the ISV2. The control indicates zebrafish model without the blocking procedure (normal zebrafish). The value of Q_A1(1) is normalized by the mean value for the control model. (d) The ratio of Q_A3(1) to Q_A1(1) for 1 h after blocking the ISV2. Data shown in (c) and (d) are values of mean ± s.d. (n = 5, *p < 0.05, **p < 0.001, by one-way ANOVA/Tukey post hoc test). (Online version in colour.)

3.2. Equivalent circuit model

Q_A3(1)/Q_A1(1) values are measured as the blood flow of each ISV groups (ISV1–4) is blocked, and the results are validated using an equivalent circuit model. The model consists of 66 electrical resistors with a constant current source (figure 4a). The 66 resistor represent 22 ISVs and 44 parts of the main artery and vein (figure 1). Among the 44 resistors for the main artery and vein, the eight resistors for the centre parts in the four regions (A1–4, V1–4) are represented in figure 4b. The resistance value was normalized with the mean resistance value for A1(3). Variation of the normalized resistance along 22 ISVs is depicted in figure 4c. The resistance values of ISVs are much higher than that for each part of the artery and vein.

Figure 4.

(a) Equivalent circuit diagram composed of 66 resistors that represent each part of blood vessel. Variation of normalized resistance of (b) A1(3)–A4(4), V1(3)–V4(4) and (c) 22 ISVs. The data in (b) are determined experimentally (n = 5) and represented by values of mean ± s.d. *p < 0.05, **p < 0.01 by one-way ANOVA/Tukey post hoc test, compared with the value of A4(4). (c) The values for ISVs are represented in the order of their position from the left to the right. The resistance values are normalized by the mean resistance value of A1(3). (d) Q_A3(1)/Q_A1(1) in five different kinds of zebrafish models (control and ISV1–4). ISV1–4 represent zebrafish models whose blood flow in the ISV1–4 regions is blocked. Triangle and inverted triangle denote the values determined by the electrical circuit model and experiment (n = 3 ∼ 5, *p < 0.05, **p < 0.01 by one-way ANOVA/Tukey post hoc test). (Online version in colour.)

Q_A3(1)/Q_A1(1) values for five different zebrafish models are compared in figure 4d. ISV1–4 on the x-axis represents the zebrafish models whose ISV1–4 regions are blocked, respectively. Black triangles denote the values determined by using the equivalent circuit model, and red inverted triangles denote experimental results. In the equivalent circuit model, the value of Q_A3(1)/Q_A1(1) in the ISV1–4 models is calculated by the increasing resistance value of the blocked ISV region infinitely. The modelling results are not exactly the same as that of the experimental results (error < 3.5%). However, the general tendency of the Q_A3(1)/Q_A1(1) values for the five zebrafish models (control, ISV1–4) is similar. In both the modelling and the experimental results, Q_A3(1)/Q_A1(1) has the highest value when blood flow in the ISV2 region is blocked (ISV2 models).

3.3. Variation of wall shear stress according to blocked intersegmental vessels

The normalized inlet flow rate (Q_A1(1)) and WSS in the part of the target vessel (V3(3)) for five different zebrafish models are compared (figure 5). The Q_A1(1) values in the five different models did not exhibit statistically significant differences. However, the WSS values at V3(3) are varied according to blocked vessels (ISV1 ∼ 3). The values of WSS are not significantly different between ISV3 and ISV4. The WSS at the V3(3) has the highest value in the ISV2 model.

Figure 5.

(a) The normalized Q_A1(1) and (b) WSS values at V3(3) are measured experimentally in five different zebrafish models. Flow rate at A1(1) is normalized by the mean value of the control models. All data shown are values of mean ± s.d. (n = 3 ∼ 5, *p < 0.05, **p < 0.01, by one-way ANOVA/Tukey post hoc test).

3.4. Effect of wall shear stress variation on the number of rolling cells

The ISV2 model is used to demonstrate the effect of altered WSS on the endothelial response because the model makes the biggest variation in WSS magnitude compared with the control model. Figure 6a shows four consecutive frames of a video clip captured at part of the V3 region of main vein. White arrows indicate two rolling cells inside the vein. To observe the cells more clearly, the background image obtained by averaging the sequential images is subtracted from the original images. The rolling cells are defined as the cells that roll along the vessel wall with a slower velocity than that of RBCs flowing through the vessel [51]. The number of rolling cells at V3 for 3 min (from 27th to 30th minute after blocking blood flow) in LPS-stimulated zebrafish models (figure 6b) is higher than the values in normal zebrafish (Control). However, the number of rolling cells does not increase in ISV2 models after injecting LPS.

Figure 6.

Effect of WSS variation at V3 on the number of rolling cells. (a) Four consecutive frames of a video clip captured at a part of the V3. White arrows indicate two rolling cells inside the vein. (b) The number of rolling cells at V3 for 3 min (from 27th to 30th minute after blocking blood flow) in normal zebrafish (control), LPS-stimulated zebrafish models (control + LPS injection) and LPS-stimulated ISV2 models (ISV2 + LPS injection). All data shown are values of mean ± s.d. (n = 5, *p < 0.01 by one-way ANOVA/Tukey post hoc test).

4. Discussion

In this study, the WSS magnitude acting on the caudal vein is controlled by blocking blood flow in four groups of the ISVs. Blood from the heart enters into the main artery through A1 [52]. A portion of the blood moves towards the tail, and the rest moves to the main vein directly through ISVs, as shown in figure 1b [53]. The main vein, which is nearly parallel to the main artery, delivers the blood back to the heart.

The portion of blood moving towards the tail is enlarged to increase WSS in the caudal vein. The increase of the blood portion is done by blocking the bypass flow passing through the ISVs, and the effect is examined by comparing Q_A3(1)/Q_A1(1) values. The increase of WSS in the caudal vein caused by the increase of Q_A3(1)/Q_A1(1) is experimentally confirmed (figure 5b). The descending order of Q_A3(1)/Q_A1(1) and WSS values at V3(3) for five different zebrafish models (control and ISV1–4) is similar (figures 4d and 5b).

Agarose gel is injected to compress tissues surrounding a certain ISV region to block blood flow in the corresponding vessels. It is known as a biocompatible material, and agarose has been used as an injectable implant material [54]. Agarose gel has also been used in bioengineering for growth of tissues [55,56] and in drug delivery systems [57]. Previous studies [54,58] confirmed the safety and biocompatibility of the agarose gel for four to eight months after injection into rat tissue. Scarano et al. [59] injected 2.5% agarose gel into the lip of 62 patients for tissue augmentation and proved that the gel is a reliable and predictable material during 3 years of clinical use. The injection procedure and equipment discussed in §2.2 are similar to those used in previous studies [60–62]. The injection procedure successfully blocked blood flows with about 90% probability. The injection procedure of agarose gel does not change the inlet flow rate (Q_A1(1)), as shown in figures 3c and 5a. In approximately 80% cases, the high values of Q_A3(1)/Q_A1(1) caused by blocking a group of ISVs starts to decrease after 30 min from the blocking (figure 3d). The amount of decrease is about 13.5% in the period from 30 min to 1 h after the blocking procedure. The recovery of blood flows in the blocked ISVs is mainly attributed to decrease of the normalized flow rate, Q_A3(1)/Q_A1(1) (see electronic supplementary material, S4). The water-soluble agarose gel injected near the ISVs seems to be dissolved by water in the surrounding tissues [63]. Thus, the WSS condition controlled by the agarose gel is maintained up to 30 min. The duration time of the controlled WSS condition in this study could be increased by replacing the agarose gel with a non-water-soluble material. However, since the prolonged compression of ISVs can cause vascular inflammation by triggering mechanotransduction pathways, this situation may not be desirable. Three WSS conditions that are different from the value in normal zebrafish (control model) are made by blocking different groups of ISVs. The magnitude of controllable WSS ranged from 0.8 dyne cm⁻² to 2.3 dyne cm⁻² at V3 in the proposed zebrafish model. This WSS range has been widely used to study the relationship between WSS and leucocyte behaviours [13,64].

In this study, the equivalent circuit model is used to verify the accuracy and reproducibility of the experimental results. The value of the normalized flow rate (Q_A3(1)/Q_A1(1)) instead of the WSS values is used for verification, because the circuit model cannot anticipate the exact velocity profile. The descending order of Q_A3(1)/Q_A1(1) that are measured in the five different zebrafish models (control and ISV1–4) is reproduced through circuit modelling. A constant current source is employed in the circuit model, because there is no statistically meaningful difference between the values of inlet flow rate (Q_A1(1)) in five different zebrafish models (figure 5a). The diameter of each blood vessel is considered as a constant value because the diameter of V3 does not show notable variation during 30 min, the duration for controlled WSS as shown in electronic supplementary material, S5.

The effect of altered WSS magnitude on rolling of leucocytes is demonstrated by counting rolling cells in the V3 region when we inject LPS into blood vessels. The rolling is a well-known behaviour of the leucocytes activated by LPS [13,51,65]. Endothelial responses to WSS magnitude have been reported as an important step to decide the rolling behaviour of leucocytes [13,66,67]. Ploppa et al. [13] activated neutrophils with LPS (100 ng ml⁻¹ and 10 ng ml⁻¹) inside a flow chamber and monitored the rolling motion of the neutrophils. LPS treatment increased the number of rolling and adhesion leucocytes, and the amount of rolling and adhesion leucocytes decreased when WSS in the chamber was increased to higher than 1 dyne cm⁻². The general trend of rolling leucocytes according to WSS magnitude in the study of Ploppa et al. is in good agreement with the results of this study (figure 6b).

Based on these results, the proposed zebrafish model can be used to investigate the effects of WSS on ECs or WSS-related diseases. The zebrafish model has a strong potential to disclose unknown CVD physiology. In addition, zebrafish models can overcome the technical limitations of previous experimental models because the optical clarity enables observation of the developmental progression of CVDs for different WSS conditions in vivo. The present results demonstrate that the proposed zebrafish model can be used in drug discovery studies for WSS-related diseases. In vivo monitoring of the effects of drug treatment on the haemodynamics and pathology of WSS-related diseases is now possible at a low cost and in a straightforward manner [68].

5. Conclusion

A novel method that enables the placement of a wide range of WSS on blood vessels of zebrafishes is proposed in this study. The inlet flow rate in the tail blood vessel (Q_A3(1)) increases, as blood flow in the ISV1 or ISV2 is blocked using agarose gel injection. ISVs in four different regions (ISV1–4) are blocked individually to compare the effect of blocked vessels on the variation of WSS. The increase of Q_A3(1) augments WSS acting on the caudal vein. The experimental results are validated using an equivalent circuit model. The number of rolling cells inside V3 of the LPS-stimulated zebrafish model decreases when WSS is increased by blocking the ISV2. The present results imply that zebrafish can be used as a new platform to study WSS-related physiology. The zebrafish model is a promising tool for discovering suitable drugs and useful for probing cellular functions of ECs in molecular biology research.

Ethics

All experimental procedures were approved by the Animal Care and Ethics Committee of POSTECH (POSTECH-2015–0058-R1), and the experiments were conducted in accordance with the approved guidelines.

Authors' contributions

W.C., J.D. and S.J.L. designed the experiments. W.C., H.M.K. and S.P. performed the experiments. W.C. and E.Y. analysed the data. W.C., E.Y., J.D. and S.J.L. wrote the paper. All authors participated in completing the manuscript and gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant (no. 2008-0061991) and the Industrial Technology Innovation Program (No. 10048358) funded by the Ministry Of Trade, Industry & Energy (MI, Korea).