You are on page 1of 10

Molecular Simulation

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/gmos20

Motion of a tumour cell under the blood flow at


low Reynolds number in a curved microvessel

L.L. Xiao, X.J. Song & S. Chen

To cite this article: L.L. Xiao, X.J. Song & S. Chen (2021) Motion of a tumour cell under the
blood flow at low Reynolds number in a curved microvessel, Molecular Simulation, 47:1, 1-9, DOI:
10.1080/08927022.2020.1856377

To link to this article: https://doi.org/10.1080/08927022.2020.1856377

Published online: 15 Dec 2020.

Submit your article to this journal

Article views: 182

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=gmos20
MOLECULAR SIMULATION
2021, VOL. 47, NO. 1, 1–9
https://doi.org/10.1080/08927022.2020.1856377

Motion of a tumour cell under the blood flow at low Reynolds number in a curved
microvessel
a
L.L. Xiao , X.J. Songa and S. Chenb
a
School of Mechanical and Automotive Engineering, Shanghai University of Engineering Science, Shanghai, People’s Republic of China; bSchool of
Aerospace Engineering and Applied Mechanics, Tongji University, Shanghai, People’s Republic of China

ABSTRACT ARTICLE HISTORY


Investigation of the flowing behaviours of circulating tumour cells (CTCs) under the blood flow is of Received 18 July 2020
fundamental importance for understanding haematogenous metastasis. In this study, the motion of a Accepted 18 November 2020
CTC at low flow rates in microvessels were simulated by dissipative particle dynamics combined with
KEYWORDS
a spring-based network model to characterise cell deformation. The effects of vessel curvature, cell Cell deformation; cell
deformability and the presence of RBCs on the motion of an initially adherent tumour cell were adhesion; curved
investigated. The results suggested that at low Reynolds number, the viscous force plays a dominant microvessel; RBCs
role and the curved vessel would initiate the formation of more simultaneous bonds. And, the shear
force acted on the softer tumour cell would induce large contact area and further stimulate the
formation of more ligand–receptor bond numbers. Moreover, to investigate the non-Newtonian
nature of the blood flow on tumour cell motion, the blood was regarded as a suspension of RBCs.
With the presence of RBC suspensions, the collision between the tumour cell and RBCs increases the
drag force and promotes the disassociation of the CTC from the vessel wall.

1. Introduction
curved vessels is that a secondary flow may develop on the
A common way for cancer cell metastasis is the bloodstream, cross-section around the bend and the axial velocity is skewed
since blood vessels are often nearby the malignant tumour. towards the inner wall or outer one, which is attributed to the
Once enters the bloodstream, the cell is called the circulating balance between viscous and inertial competitions [7,8]. Such a
tumour cell (CTC). The circulation of the blood plays a signifi- flow pattern may be contributed to the occurrence of blood
cant role in determining where CTCs travel. As the principal diseases. It has been reported that the higher shear stress/
step in cancer cell metastasis, cell adhesion has attracted rate gradient at the positive curvature would mediate cell
much attention in the past few decades. Previous studies accumulation in the curved vessels [9]. The 2D numerical
showed that CTCs may hijack neutrophils for arrest in the studies on tumour cell adhesion in curved microvessels
endothelium of distant organs by a ‘two-step adhesion’ revealed that the intercellular interaction between travelling
hypothesis that includes initial neutrophil tethering on the tumour cells and localised shear stresses and their gradients
endothelium and subsequent CTCs being captured by tethered would also affect the cell adhesion significantly [10,11]. In
neutrophils [1,2]. CTCs may also imitate the behaviour of neu- these studies, the tumour cell was considered as rigid and
trophils by directly adhering to the vascular endothelium [1]. the blood was assumed to be homogeneous Newtonian fluid.
In this study, neutrophil-mediated CTC adhesion was not con- As a matter of fact, cell deformability plays a key role in cell
sidered due to the complicated molecular interactions includes adhesion [12,13] and neglecting it may significantly overpre-
collision and bond formation between the neutrophils and dict the flow resistance and drag force [14]. In our previous
CTCs. A major conceptual development is the recognition study on the tumour cell adhesion in a straight microvessel,
that CTC adhesion to vascular endothelium is often mediated the blood was regarded as a suspension of RBCs and the results
via a surprisingly small number of receptor–ligand bonds [3]. showed that the haematocrit, RBC aggregation as well as vessel
The adhesion process highly depends upon the local microen- size would also be the major factors for cell adhesion at low
vironment, including the geometric features of the vasculature, flow rates.
local haemodynamic factors, such as the levels of haematocrit, In this study, dissipative particle dynamics method com-
wall shear rate and residence time [1]. A series of in vivo exper- bined with a spring-based network cell model was employed
iments showed that tumour cells prefer to aggregate near the to carry out 3D simulations of the motion of a rolling tumour
bend in a curved microvessel and the bifurcations in microvas- cell in a curved microvessel. Then, the effects of curvature,
culature serving the major sites of tumour cell adhesion [4,5]. tumour cell deformability as well as intercellular interaction
Most of the previous studies focused on cells’ motion in between RBC and CTC on the motion of a rolling CTC were
straight tubes with uniformly sized rectangular or cylindrical investigated. Finally, summary and conclusions were
cross-section [6]. The main difference between straight and presented.

CONTACT L.L. Xiao xiaoll_sues@sues.edu.cn


© 2020 Informa UK Limited, trading as Taylor & Francis Group
2 L.L. XIAO ET AL.

2. Models and methods area, local area and volume; Atot and V are the instantaneous
membrane area and the cell volume; Atot tot
0 and V0 are their
The blood flow is simulated by using dissipative particle
respective specified total area and volume values. A, A0 are
dynamics, which has been employed in our previous study
the instantaneous and initial local area.
[15,16]. Details of DPD method have been given in previous
Nodal forces are derived from the total energy as follows:
studies [17–19]. In a word, each DPD particle represents a
soft lump of atoms and interacts with surrounding particles Fmembrane
i = −∂E{(ri )}/∂ri (6)
through three simple pairwise additive forces: conservative
force, dissipative force and random force. Particle motion The macroscopic properties of a cell are based on a linear
obeys Newton’s law. Numerical models characterised the analysis of a hexagonal system built with equilateral triangles
behaviours of cell deformation and CTC adhesion are specifi- [23]. The linear shear modulus of the WLC-POW model is
cally described in the following sections. √
3 kB T x0 1 1
G= ( 3− 2+ )
4plmax x0 2(1 − x0 ) 4(1 − x0 ) 4
√
2.1. Cell membrane model 3kp (m + 1)
+ , x0
4l0m+1
In simulations, the cell membrane is discretised into a series of
triangular elements. The adjacent nodes of the triangular grids = l0 /lmax (7)
are connected by elastic springs. A spring-based network The linear area dilation modulus is defined as
model endowed with in-plane and bending energy as well as
constraint of surface area and volume has been introduced K = 2G + ktot
area + karea (8)
by Boey et al. [20] to describe RBC initially. Then Pivkin
The Young’s modulus Y for the two-dimensional sheet can be
and Karniadakis [21] provided a systematic coarse-grained
expressed through the shear and area-compression moduli as
procedure to reduce the number of degrees of freedom dra-
follows:
matically. This coarse-grained model was further improved
by Fedosov et al. [22], yielding accurate mechanical response. 4KG
Y= (9)
The total energy of the network is defined as K+G
E({ri }) = Ein-plane + Ebending + Earea + Evolume (1) And the Poisson’s ratio ν is given by
where ri expresses the vertex coordinates and the triangular K−G
n= (10)
elements is connected by worm-like chains (WLC) to model K+G
the nonlinear elasticity of the cell membrane. Such that the Based on the incompressibility assumption
in-plane elastic energy for WLC-POW model is given by ktot + k .. Gis set, so Y  4G and n  1.
area area
 kB Tlmax 3x2 − 2x3 kp
 The relationship between bending modulus kbend and the
Ein-plane = l l
+ (2) macroscopic membrane bending rigidity kc is derived for the
all edges
4p 1 − xl (m − 1)lm−1
case of a spherical shell in the Helfrich bending energy, as follows:
where xl = l/lmax [ (0, 1), lmax is the maximum spring exten- 2
sion, p the persistence length, kB T the Boltzmann temperature kbend = √ kc (11)
3
of the system. kp is a POW force coefficient and m is a specified
exponent, here we set it to 2 [22].
The bending energy to resist the wrinkle and distortion of
membrane is defined as 2.2. CTC adhesion model
 CTC adhesion under flow in microvssels is mediated by bind-
Ebending = kbend [1 − cos (uab − u0 )] (3)
all triangle adjacents
ing between cell surface receptors and ligands distributed on
the surface of the endothelial cells lined in the vessel wall.
where kbend is a bending modulus; uab and u0 are the instan- The adhesion model provides rules of formation and dis-
taneous and spontaneous angles formed between the outer normal sociation of bonds between receptors and ligands, which is
vectors of two adjacent triangles a, b sharing the common edge. improved by adding the probabilistic model developed by
The area and volume restraint energy used to constrain the Hammer and Apte [24]. The probabilities of bond formation
variations of the surface area and volume are presented as fol- and disassociation in a time interval Dt are given by
lows: 
1 − e−kon Dt , lb , don
2  karea (A − A0 )2 Pon = , Poff
ktot
area (A
tot
− Atot
0 ) 0, lb ≥ don
Earea = + (4) 
2Atot
0 all triangles
2A0 1 − e−koff Dt , lb , doff
= (12)
0, lb ≥ doff
kvolume (V − V0tot )2
Evolume = (5) where kon and koff are the association rate and off-rate, respect-
2V0tot
ively. At a given time instance, two random numbers j1 and j2
where ktot
area , karea and kvolume are constraint constants for global uniformly distributed on [0, 1] are generated. An existing bond
MOLECULAR SIMULATION 3

is allowed to break if j1 , Poff and left unchanged otherwise. A Table 1. Simulation parameters for cell membrane and plasma
new bond is allowed to from according to the relationship- Parameter Simulation Physical values
Pon . j2 under the condition that the distance between a recep- Blood plasma density (r) 3 1.0 × 103 kg/m3 [27]
tor and a free ligand is less than the reactive distance. A free ligand Blood plasma viscosity (m) 19.3 1.2 × 10−3 Pa s[27]
Temperature (T ) 0.0828 310 K
refers to that it is not bound to any receptors. Reversely, a pre- Membrane Young’s modulus for RBCs (YR ) 369 18.9 mN/m[28]
existing bond may break with off-rate koff or if its length exceeds Membrane Young’s modulus for CTCs (YT ) 80.5–8050 4.16 − 416 mN/m[25]
the rupture distance doff . The rates kon and koff are computed as Membrane bending modulus for RBC 5.364 2.8 × 10−19 J[22]
R
(kbend )
2 Membrane bending modulus for CTC 69.28 3.6 × 10−18 J[25]
son (lb − l0b ) T
(kbend )
kon = k0on exp ( − ), Time step (Dt) 0.001 0.001 ms
2kB T
(13) tot
Global area constraint constant (karea ) 5 × 104 2.58 × 10−3 N/m
soff (lb − l0b )
2 Local area constraint constant (karea ) 100 5.2 mN/m
koff = k0off exp ( ) Volume constraint constant (kvolume ) 5 × 104 3.35 × 10−3 N/m
2kB T Unstressed on rate (kon0
) 10 104 s−1[29]
0
Unstressed off-rate (koff ) 0.02 20 s−1[30]
where k0on and k0off are the unstressed reaction rates at the distance On strength (son ) 10 0.5 mN/m[31]
lb = l0b between a receptor and a ligand with the equilibrium Off strength (soff ) 1 0.05 mN/m[31]
Association length (don ) 0.1 0.1 µm
spring length l0b . The effective on and off strengths son and soff Disassociation length (doff ) 0.1 0.1 µm [32]
define a decrease or an increase of the corresponding rates within Spring constant (ks ) 2 × 104 1 × 10−3 N/m[33]
the interaction lengths don and doff . Bonds are assumed to behave Equilibrium spring length(l0b ) 0.025 0.025 µm [31]
Receptor density (nr ) 4.6 4.6/µm2
as elastic springs following the Hooke’s law Ligand density (nl ) 4 4/µm2

F(lb ) = ks (lb − l0b ) (14)


where ks is the spring constant. The binding force can be derived studies [10,11]. The tumour cell is of spherical shape with a
as Fbond = F(rij )r̂ij . diameter of 9 µm, constructed by 1170 particles, without con-
i
sidering the nucleus for simplicity in this study. The CTC is
filled with a Newtonian fluid with a viscosity of
2.3. Model and physical units scaling 6 × 10−3 Pa s. The parameters of plasma and CTC are listed
in Table 1. Mechanical properties of the simulated CTC have
To relate the model’s non-dimensional units to the physical
been discussed in our previous studies [15,16]. Calculations
parameters, a scaling procedure has been presented by Fedo-
based on the experimental measurements [25] give a surface
sov et al. [22], which ensures that the modelling system is con-
membrane tension of 6.76 ×10−6 N/m in the breast cancer
sistent with the physical system. The length scale is given by
cell. The average Young’s modulus for malignant breast cells
DP0 measured by atomic force microscopy ranges from 300 to
LS = (15) 600 Pa at different loading rates [26]. The Young’s modulus
DM
0
computed by the average cell stiffness multiplied by the mem-
where the superscripts M and P denote ‘model’ and ‘physical’. brane thickness based on a two-dimensional sheet-based cell
The energy scale is provided as follows: membrane model ranges from 2.1 to 5.4 μN/m. The DPD par-
ameters of interactions among different types of particles are
Y P DP0 2
ES = ( ) (16) presented in Table 2. No-slip boundary condition was imposed
Y M DM
0 near the wall and a periodic boundary condition was applied
The force scale is defined by along the flow. Bounce-back reflection was exerted on the sur-
face of the cell membrane to ensure its impenetrability. To
Y P DP0 drive the flow, a uniform body force was applied to all particles
NS = (17)
Y M DM
0 along the centreline of microvessel, which is equivalent to the
pressure gradient DP/L = rf , where ΔP is the pressure drop
The scaling between model and physical times is defined as fol-
over the tube length and ρ is the suspension’s mass density.
lows:
Y M DP0 hP
tS = (18) 4. Results and discussion
0 h
Y P DM M

4.1. Fluid flows in a curved microvessel


In this section, the simulation of a Poiseuille flow in the curved
3. Simulation set-up and modelling parameters microvessel was first performed. It is well known that in a flow
The motion of a rolling tumour cell under plasma flow in through a curved microvessel, secondary flow arises because of
microvessels with a diameter of 14 µm and the total length Table 2. DPD parameters.
of 224 µm was simulated. The curved vessel is of symmetrical Interaction a γ rc
geometry, containing a straight segment and a negatively Plasma–plasma 4 9.15 1.5
curved segment of π/3 bending angle, with the inner curvature Plasma–membrane, wall–membrane, plasma–wall 2 9.15 1.5
radius of 28 µm, and a positively bent segment with the inner Cytoplasm–cytoplasm, cytoplasm–membrane 4 20.0 1.5
Membrane–membrane 100 20.0 0.75
curvature radius of 42 µm, same as the model in the previous
4 L.L. XIAO ET AL.

the radial pressure gradient driving the fluid in the cross-section the receptor–ligand bond formation. When the flow was
to move radially. In this DPD model, an external body force is turned on, the plasma began to flow and the adherent tumour
50 m/s2, which is equivalent to a pressure gradient of cell rolled along the wall. The history of the simulated CTC
DP/L = rfx = 0.05 Pa/mm[10]. The analytic solution for the motion within the curved microvessel is presented in Figure
profile of flow velocity along the diameter of the cross-section 2(a). It can be found that the cell translates and rotates during
is vx (r) = rfx (R2 − r2 )/4m and the Reynolds number is its rolling. Figure 2(b) shows the variation of number of
Re = rum D/m = 2.98 × 10−3 , which is the typical value for ligand–receptor bonds and the trajectories of x and y centroid
the arterioles with a diameter of 15 µm [34]. And the shear of the CTC. The adherent CTC is nearly spherical at the stage
rate is g = um /D = 18 s−1 , which facilitates cell margination of initial tethering. Under the shear force upon by the fluid, the
[35] and subsequent cell adhesion. The adherent CTC was cap- CTC rotates clockwise and deforms to make a flat contact area
tured at the postcapillary venule by tracing the individual trajec- with the vessel wall. With the increase in contact area, the bond
tories of tumour cells in blood flow with relatively low velocity number grows up significantly and the rolling motion tempor-
of <1 mm/s [5], corresponding to a shear rate of 36 s−1. arily stops. As can be seen from Figure 2(b), the cell moves
The distribution of velocity contours for the cross-section slowly and is arrested by tethering of receptors. Under the
for the Poiseuille flow in the curved microvessel is illustrated influence of binding force and shear force, some new recep-
in Figure 1(a). And the axial velocity, obtained from the tor–ligand bonds form in the front end and several formed
cross-section extracted at the microvessel summit, is axisym- bonds break in the rear end. Then the arrested cell deforms
metric, as shown in Figure 1(b). Here, a dimensionless number, into a ‘tear drop’ shape (see position ‘A’ in Figure 2(a)). Sub-
Dean number De, can be introduced to identify the effects of sequently, the cell rolling commences and the contact area
curvature on the flow structure, which describes the relative decreases, indicated by the reduced bond number in the period
importance of centrifugal to viscous force. It is defined as
√ of time from t=150 ms to t=200 ms. The strong adhesion
De = Re R/Rc [36], where Rc is the minimum curvature radius makes the CTC travels slowly in the straight part of the curved
of the microvessel centreline. In this study, De is much smaller microvessel (see the slope of the curve of variation in CTC cen-
than 1. Thus, the viscous force plays a dominant role and no sec- troid position). When the cell enters into the negative curva-
ondary flow is developed, as shown in Figure 1(c). ture vessel (after turning point ‘A’), a few more bonds start
to form. But compared with the bond number fluctuation
between 12 and 32 in the straight segment, the bond number
4.2. Motion of a CTC flowing in a curved microvessel
for the adherent CTC in the negative curvature vessel ranges
and in a straight microvessel
from 5 to 20. Consequently, the cell moves a bit quicker
Initially, CTC with Young’s modulus of 41.6 µN/m was placed (between point ‘A’ and point ‘B’). Once the CTC moves
near the wall with a distance of 50 nm to facilitate binding by away from the negative curvature vessel, all the bonds are

Figure 1. (Colour online) The distribution of velocity magnitude contours for the cross-section of the flow without CTC in a curved microvessel (a), the profile of axial
velocity (b) and vectors of the radial velocity on the cross-section normal to the centreline, extracted at the microvessel summit (c).
MOLECULAR SIMULATION 5

Figure 2. (Colour online) Trajectory of the tumour cell in a curved microvessel (a) and the variation of number of receptor–ligand bonds formed on the surface of CTC
and the trajectories of x and y centroid of the CTC (b).

ruptured and the shear force is large enough to cause rapid dis-
sociation. Then the cell moves freely and no cell adhesion
occurs in the positive curvature vessel at all (after point ‘B’).
Tumour cell migrates in the flow under the combination of
hydrodynamic force and adhesive force. The hydrodynamic
force is defined by the cumulative force generated by the inter-
actions with the plasma and the adhesive force is the sum of
the spring forces arising from all bonds. To investigate the
effects of these two forces on the cell moving along the micro-
vessel wall, the forces are divided into the wall-directed force
and drag force in flow direction. Figure 3 shows the variation
of wall-directed force and drag force exerted on the rolling
tumour cell. As seen from Figure 3(a), the initially adhered
cell subjects to a weaker hydrodynamic wall-directed force
or even a lift force (illustrated by negative values) and a strong
adhesive force when the bond number increases greatly. The
wall-directed binding force is up to nearly 3000 pN and the
averaged adhesive force is nearly 500 pN. In comparison, the
adhesive force for the tethered leucocyte obtained in previous
numerical studies varies from about 100 to 750 pN for shear
rates 100–500 s−1 [37]. But the hydrodynamic drag (shear
force) is larger than the axial component of the binding
force, as shown in Figure 3(b), thus the cell rotates slowly
along the wall and the contact area enlarges. Under the consist-
ent streamwise forces, the number of ruptured bonds located
in the rear end of tumour cell is larger than that of formed
bonds in the front part. Consequently, the wall-directed bind-
ing force decreases largely. Once the cell approaches to the
negative curvature vessel, the rate of bond formation is greater
Figure 3. (Colour online) The hydrodynamic force and total adhesive force (a)
than that of bond breakage. So, the bond number experiences towards the wall (b) and in the flow direction, acted on the tumour cell in the
an upward trend and the axial component of binding force curved microvessel.
6 L.L. XIAO ET AL.

Figure 4. (Colour online) Snapshots of the tumour cell every other 400 ms in the straight microvessel (a), the variation of number of receptor–ligand bonds formed on
the surface of CTC and the trajectories of x centroid of the CTC (b).

becomes positive, which promotes cell adhesion. At this stage, −72 µm, as shown in Figure 6. Though the bond number
negative hydrodynamic drag force occurs, which may be experiences a small jump for the cell at the following adhesion
attributed to the reverse flow generated at the downstream stage in the straight microvessel, the wall-directed force is so
of tumour cell [10]. When the cell comes to the positive curva-
ture vessel, the negative hydrodynamic drag force grows larger
and the wall-directed hydrodynamic force is so weak, which
leads to rapid dissociation.
To evaluate the effect of the curvature, the motion of a CTC
in straight microvessl was studied. To drive the flow, a body
force same as that in the model of curved microvessel was
applied to the flow. Figure 4(a) shows the history of the CTC
movement in a straight microvessel with adhesion. It also
can be observed that the transient deformation of the initial
adherent CTC into a tear-drop shape, which agrees well with
previous studies [14]. Cell motion is characterised by several
steps including rolling, and short stay during which the CTC
is adherent, which can be obviously seen from the x displace-
ment of the CTC and the variation of the formed bond number
in Figure 4(b). The bond number experiences a fluctuation and
the adherent CTC would roll slowly when the bond number
reaches to the peak value while it would move fast during
which the bond number drops to 0. Cell motion is affected
by the combination of the cell deformation, hydrodynamic
force and the stochastic bond kinetics. This cannot be pre-
dicted in Yan’s studies [10,11], in which the cell deformation
is neglected. Similar to that in the straight part of the curved
microvessel, the initially tethered tumour cell moves under
the strong adhesive force and hydrodynamic drag (shear
force), as shown in Figure 5. The initially spherical cell
deforms to tear-drop shape under the shear force, which
stimulates the formation of more receptor–ligands in the
front end of the tumour cell. But under the persistent stream-
wise force and hydrodynamic lift force, the bond formation
rate is lower than bond breakage rate and the bond number Figure 5. (Colour online) The hydrodynamic force and total adhesive force (a)
towards the wall (b) and in the flow direction, acted on the tumour cell in the
decreases gradually when the cell moves at x=−78 to straight microvessel.
MOLECULAR SIMULATION 7

Figure 6. (Colour online) Comparison of bond number for tumour cell adhesion
in the straight microvessel and curved microvessel.

weak that the cell finally detaches from the vessel wall. By con-
trast, for the cell moving at the negative curvature vessel part,
the number of new formed bonds in the front end is higher
than that of ruptured bonds in the rear end. Therefore, the
number of the simultaneous bonds formed at the bend
location is much higher than that for the straight microvessel.
It indicates that the bend structure facilitates cell adhesion by
narrowing the gap between the leading end of cell and vessel
wall and then increasing bond formation rate.

Figure 7. (Colour online) The variation of number of receptor–ligand bonds vs x


centroid position (a) and the trajectories of x and y centroid of the CTC (b) in
terms of cell deformability.
4.3. Effects of CTC deformability on its motion in the
curved microvessel
than 50. Under the strong adhesive force, the cell experiences
The deformation during cell rolling is of importance in durable arrest and remain motionless.
adhesion owing to their direct effect on contact area correlated
with the number of receptor–ligand bonds. To evaluate the
effects of CTC deformability on its deformation and adhesion 4.4. Effects of RBC suspension on the motion of CTC in
in the curved microvessel, three values of elastic modulus the curved microvessel
Y = 416, 41.6and 4.16 mN/mwere employed. At the initial The simulated systems mentioned above simply regard the
time, there is no difference for the number of formed recep- blood as Newtonian fluid, which ignores the particulate nature
tor–ligand bonds between the cells with different deformabil-
ities, as shown in Figure 7(a). Once the fluid is driven by the
external force, the cell deforms to respond to the hydrodyn-
amic force. Under the shear force, the adherent cell is extended
along the flow direction. Then the bond number increases with
decreasing elastic modulus due to the enhanced contact area.
For the stiffest cell (Y = 416 mN/m), the number of formed
bonds during rolling is the smallest, reaching up to only 10, as
shown in Figure 7(a). The binding force is so weak to over-
come the shear force, and consequently, all the bonds are bro-
ken up rapidly at about x=−77 μm (marked by an arrow in
Figure 7(b)). Then the cell moves as a free cell and the cell vel-
ocity jumps up, illustrated in Figure 7(b), indicating the hydro-
dynamic force is dominant. Compared with the cell of
Y = 41.6 mN/m, the stiffer cell peels off the vessel wall in
advance. For the softest cell (Y = 4.16 mN/m), it deforms lar-
gely under the shear force initially to induce large contact area Figure 8. (Colour online) Variation of number of receptor–ligand bonds formed
so that the number of formed bonds increases sharply to more on the surface of CTC and the trajectories of x and y centroid of the CTC.
8 L.L. XIAO ET AL.

Figure 9. (Colour online) Snapshots of the flow of the RBC suspension and the CTC when the CTC is released at the initial time (a) and moves to the position labelled by
‘A’–‘E’ in Figure 8, corresponding to (b)–(f).

of the blood. As in such small curved microvessel, the RBCs, as 5. Conclusions


the main blood constituents, are the same as the vessel in
The effects of the curvature, cell deformability and the pres-
diameter. They are not dispersed uniformly and behave like
ence of RBCs on the motion of a single tumour cell in the
elastic objects suspended in plasma and further affect the
curved microvessel were investigated by using dissipative par-
motion of the CTC. In this section, the effects of blood flow
ticle dynamics and the probabilistic adhesive dynamics model.
with haematocrit of 0.1 on the motion of CTC are investigated.
Under the adopted driven body force, the Dean number is
It has been found that RBCs not only enlarge the stream-
much smaller than 1 so that no secondary flow occurs at the
wise forces, impairing the WBC binding, but also cause an
cross-section and the Reynolds number is such low that the
average wall-directed force, which is expected to enhance
viscous force dominates the flow. The curvature has a signifi-
binding [38]. As the CTC attaches to the vessel initially, and
cant impact on the ligand–receptor formation. Simultaneous
the RBCs located in the vessel centre move faster and flow
bond number would increase at the curved segment of the
past the CTC, which can be observed in Figure 9(b). It
vessel. Under the influence of shear force, increasing cell
seems that the RBC suspension has a small impact on the num-
deformability largely enlarges the contact area so that the
ber of formed receptor–ligand bonds when the CTC travels in
number of formed bonds increases sharply. As a result, the
the straight segment of the curved vessel, as shown in Figure 8.
cell experiences durable arrest and remain motionless if the
But obviously, the intercellular interaction increases the drag
binding force is greater than the shear force. Considering the
force and it takes the CTC about 110 s to roll to the turning
blood as a suspension of RBCs, the collision between the
point A. Compared to the case without RBCs, the velocity of
CTC and RBCs increases the drag force, reducing the contact
the CTC is almost twice as fast. When the CTC cross over
time so that the adherent CTC moves quickly away from the
the first turning point of the curved vessel, the intercellular
vessel wall.
force exerted by the RBCs promotes the peeling of the CTC
off the vessel wall. Though more wall-directed force generated
by the collision between RBCs and the CTC are applied on the Acknowledgements
CTC, the time to contact the vessel wall is so short. As a result,
This work is supported by the National Natural Science Foundation of
the formed bond number decreases rapidly when CTC passes China (Grant Nos. 11902188,11872283), Shanghai Sailing Program
through the negative curvature part of the vessel, as plotted in (19YF1417400). The grants are gratefully acknowledged.
Figure 9.
When the CTC reaches close to x=−50, the number of
bonds decreases to zero, suggesting that there is no cell Disclosure statement
adhesion (see Figure 9(c)) and as result, the CTC moves No potential conflict of interest was reported by the author(s).
gradually away from the vessel wall. Subsequently, the CTC
recovers to spherical shape and translates in the centre of
the vessel and no margination occurs in the curved vessel, Funding
as shown in Figure 9(d–f). It has been demonstrated that This work was supported by National Natural Science Foundation of
the cell margination depends on cell relative size to the China: [Grant Number 11872283, 11902188]; Shanghai Sailing Program:
vessels. When the discrepancy in the radiuses between vessel [Grant Number 19YF1417400].
and tumour cell exceeds the thickness of red blood cell,
tumour cell tends to margination [39]. It is necessary to
investigate the vessel size on the cell adhesion in the curved ORCID
microvessel in the future work. L.L. Xiao http://orcid.org/0000-0002-7177-1086
MOLECULAR SIMULATION 9

References [20] Boey SK, Boal DH, Discher DE. Simulations of the erythrocyte
cytoskeleton at large deformation. I. Microscopic Models Biophys
[1] Wirtz D, Konstantopoulos K, Searson PC. The physics of cancer: J. 1998 Sep;75(3):1573–1583.
the role of physical interactions and mechanical forces in metastasis. [21] Pivkin IV, Karniadakis GE. Accurate coarse-grained modeling of
Nat Rev Cancer. 2011 Jun 24;11(7):512–522. red blood cells. Phys Rev Lett. 2008 Sep 12;101(11):118105.
[2] Liang S, Slattery MJ, Dong C. Shear stress and shear rate differen- [22] Fedosov DA, Caswell B, Karniadakis GE. Systematic coarse-grain-
tially affect the multi-step process of leukocyte-facilitated mela- ing of spectrin-level red blood cell models. Comput Method Appl
noma adhesion. Exp Cell Res. 2005 Nov 1;310(2):282–292. M. 2010;199(29–32):1937–1948.
[3] Zhu C, Bao G, Wang N. Cell mechanics: mechanical response, cell [23] Dao M, Li J, Suresh S. Molecularly based analysis of deformation of
adhesion, and molecular deformation. Annu Rev Biomed Eng. spectrin network and human erythrocyte. Mat Sci Eng C-Bio S.
2000;2:189–226. 2006 Sep;26(8):1232–1244.
[4] Zhang L, Zeng M, Fu BM. Inhibition of endothelial nitric oxide [24] Hammer DA, Apte SM. Simulation of cell rolling and adhesion on
synthase decreases breast cancer cell MDA-MB-231 adhesion to surfaces in shear-flow – general results and analysis of Selectin-
intact microvessels under physiological flows. Am J Physiol Heart mediated neutrophil adhesion. Biophys J. 1992 Jul;63(1):35–57.
Circ Physiol. 2016 Jun 1;310(11):H1735–H1747. [25] Guo HL, Liu CX, Duan JF, et al. Mechanical properties of breast
[5] Guo P, Cai B, Lei M, et al. Differential arrest and adhesion of tumor cancer cell membrane studied with optical tweezers. Chin Phys
cells and microbeads in the microvasculature. Biomech Model Lett. 2004;21(12):2543–2546.
Mechanobiol. 2014 Jun;13(3):537–550. [26] Li QS, Lee GY, Ong CN, et al. AFM indentation study of breast can-
[6] Dabagh M, Gounley J, Randles A. Localization of rolling and firm- cer cells. Biochem Biophys Res Commun. 2008 Oct 3;374(4):609–
adhesive interactions between circulating tumor cells and the 613.
microvasculature wall. Cell Mol Bioeng. 2020;13(2):141–154. [27] Skalak R, Chien S. Handbook of bioengineering. New York:
[7] Ye T, Phan-Thien N, Lim CT, et al. Red blood cell motion and defor- McGraw-Hill; 1987.
mation in a curved microvessel. J Biomech. 2017 Dec 8;65:12–22. [28] Suresh S, Spatz J, Mills JP, et al. Connections between single-cell
[8] Donghe S, Nan X, Di J, et al. Multi-relaxation time lattice boltz- biomechanics and human disease states: gastrointestinal cancer
mann simulation of inertial secondary flow in a curved microchan- and malaria. Acta Biomater. 2005 Jan;1(1):15–30.
nel. Chin Phys B. 2013;22(11):114704. [29] Schwarz US, Alon R. L-selectin-mediated leukocyte tethering in
[9] Liu Q, Mirc D, Fu BM. Mechanical mechanisms of thrombosis in shear flow is controlled by multiple contacts and cytoskeletal ancho-
intact bent microvessels of rat mesentery. J Biomech. 2008 Aug rage facilitating fast rebinding events. Proc Natl Acad Sci USA. 2004
28;41(12):2726–2734. May 4;101(18):6940–6945.
[10] Yan WW, Liu Y, Fu BM. Effects of curvature and cell-cell inter- [30] Alon R, Chen SQ, Puri KD, et al. The kinetics of L-selectin tethers
action on cell adhesion in microvessels. Biomech Model and the mechanics of selectin-mediated rolling. J Cell Biol. 1997 Sep
Mechanobiol. 2010 Oct;9(5):629–640. 8;138(5):1169–1180.
[11] Yan WW, Cai B, Liu Y, et al. Effects of wall shear stress and its gra- [31] Dembo M, Torney DC, Saxman K, et al. The reaction-limited kin-
dient on tumor cell adhesion in curved microvessels. Biomech etics of membrane-to-surface adhesion and detachment. Proc R Soc
Model Mechanobiol. 2012 May;11(5):641–653. Ser B-Bio. 1988 Jun 22;234(1274):55–83.
[12] Rejniak KA. Investigating dynamical deformations of tumor cells in [32] Marshall BT, Sarangapani KK, Wu JH, et al. Measuring molecular
circulation: predictions from a theoretical model. Front Oncol. elasticity by atomic force microscope cantilever fluctuations.
2012;2:111. Biophys J. 2006 Jan;90(2):681–692.
[13] Deng Y, Papageorgiou DP, Chang H-Y, et al. Quantifying shear- [33] Fritz J, Katopodis AG, Kolbinger F, et al. Force-mediated kinetics of
induced deformation and detachment of individual adherent sickle single P-selectin ligand complexes observed by atomic force
red blood cells via combined experimental-computational studies. microscopy. Proc Natl Acad Sci USA. 1998 Oct 13;95(21):12283–
Biophys. J.. 2019;116(2):360–371. 12288.
[14] Pappu V, Doddi SK, Bagchi P. A computational study of leukocyte [34] Popel AS, Johnson PC. Microcirculation and Hemorheology. Annu
adhesion and its effect on flow pattern in microvessels. J. Theor. Rev Fluid Mech. 2005 Jan 1;37:43–69.
Biol.. 2008 Sep 21;254(2):483–498. [35] Nash GB, Watts T, Thornton C, et al. Red cell aggregation as a fac-
[15] Xiao LL, Liu Y, Chen S, et al. Numerical simulation of a single cell tor influencing margination and adhesion of leukocytes and plate-
passing through a narrow slit. Biomech Model Mechanobiol. 2016 lets. Clin. Hemorheol. Microcirc.. 2008;39(1–4):303–310.
Apr 15;15(6):1655–1667. [36] Helin L, Thais L, Mompean G. Numerical simulation of viscoelastic
[16] Xiao LL, Lin CS, Chen S, et al. Effects of red blood cell aggregation dean vortices in a curved duct. J Non-Newtonian Fuid Mech.
on the blood flow in a symmetrical stenosed microvessel. Biomech 2009;156(1–2):84–94.
Model Mechanobiol. 2020 Feb;19(1):159–171. [37] Pappu V, Bagchi P. 3D computational modeling and simulation of
[17] Hoogerbrugge PJ, Koelman JMVA. Simulating microscopic hydro- leukocyte rolling adhesion and deformation. Comput Biol Med.
dynamic phenomena with dissipative particle dynamics. Europhys 2008 Jun;38(6):738–753.
Lett. 1992 Jun 1;19(3):155–160. [38] Isfahani AHG, Freund JB. Forces on a wall-bound leukocyte in a
[18] Espanol P. Hydrodynamics from dissipative particle dynamics. Phys small vessel due to red cells in the blood stream. Biophys. J..
Rev E. 1995 Aug;52(2):1734–1742. 2012;103(7):1604–1615.
[19] Groot RD, Warren PB. Dissipative particle dynamics: bridging the [39] Takeishi N, Imai Y, Yamaguchi T, et al. Flow of a circulating tumor
gap between atomistic and mesoscopic simulation. J Chem Phys. cell and red blood cells in microvessels. Phys Rev E. 2015;92
1997 Sep 15;107(11):4423–4435. (6):063011.

You might also like