(Bio)physics of early embryo

morphogenesis

Single cell mechanics Multicellular system

Chii Jou Chan, Hiiragi Lab

Lautenschläger Summer School, EMBL, 18.07.19
 Outline

•  Historical perspectives
•  Basic concepts of mechanobiology
•  How do mechanical forces shape the tissue?
•  From tissue architecture to cellular functions
•  Approaches to study morphogenesis
•  Conclusion and outlook
Some brief history…
 transformations of human heads
Hypothesizing the ‘laws of growth’ drawn by artist Albrecht Dü rer (Dü rer,
1528; Fig. 366 from Thompson,
1917a). (C) Thompson’s comparative
D’Arcy Thompson illustrations of chimpanzee (left) and
human (right) cranial ontogenetic
(1860-1948) shape changes (Figs 405 and 406
REVIEW Development (2017) 144, 4284-4297 doi:10.1242/dev
C from Thompson, 1917a).
(D,E) Evolution of the horse skull from
•  Growth
Hyracotheriumgenerates
Fig.
(Eocene) to diverse
the
2. Biological shapes,
modern horse, represented as a
biological forms andand
transformations
coordinate transformation
the ‘laws of g
to the
(A) Geometric shape changes durin
same scale of magnitude. (F) Diagram
hyacinth leaf growth, which follows
of the Cartesian coordinates
•  Quantitative specific
projecting shape
ways to in e.g. foll
set of spatial
outlines of skullsrules,
understand morphogenesis
the lineageparticular ratios of radial
from Hyracotherium and tangen
to the
growth
A tovelocities (Fig. 359 from Tho
(morphological
modern horse.
1917a). (B)space)
H indicate
Geometric morphometri
progressive changes in morphology
describing
through evolution (Figsshape alterations
401 and 402 in hyen
•  Tissue during
shape
from Thompson, development
can be
1961). from juvenile to
(Tanner et al., 2010). (C) The ‘laws
understood growth’ininterpreted
terms of as acell-
broad set o
level physics developmental mechanisms transla
genetic information and physical for
the environment into individual biolo
shapes and contributing to their div
Images in C created by Keith Chan
copyright Alexmit (www.fotosearch.
animal images are illustrations from
(1898).
Genes F

Developmental mechanisms
(‘laws of growth’)
Environment Morphological variation
Abzhanov Dev 2017
 The birth of Physical Biology…
Johannes Holtfreter
(1901-1992)
Cells have selective affinity

Cells move with directionality

Cell shape change can drive (filopodia, polarity)
morphogenesis (e.g. bottle cells)

Harris Science 1980

Cells movement is mechanically integrated

 Physical model for tissue patterning
Allan Turing (1912 – 1954)
cross-reactions between
two diffusing species
(morphogens) could
generate large-scale
spatial-temporal patterns

Reaction-diffusion model

•  Short range activator

•  Long range inhibitor

Turing Phil. Trans. R. Soc. 1952

Inaba Science 2012
Examples of early embryo development

EPI

Cortical tension
Adhesion force
Fluid pressure
Apical domain
 Examples of early embryo development

•  Genes encodes the assembly kit, but cellular behaviour drives changes in tissue
morphology on a macroscopic scale

•  Mechanobiology combines physics, biology and engineering to study how

physical forces and changes in cell/tissue mechanics contribute to development
Basic concepts of Mechanobiology
 Some key mechanical players in development
Actomyosin contractility Cell-cell adhesion

Purse-string-like actomyosin belt

Pulsed contractions of actomyosin network

drive apical constriction
Can influence contact area
Martin Nature 2008 and cell fate through signaling!
 P H2B-mCherry
18 h.p.f. 19.5 h.p.f. 21 h.p.f. 23 h.p.f.

Some key mechanical players in development

Extracellular matrix Lumen pressure
Lifeact-GFP
H2B-mCherry Zebrafish inner ear
Lifeact-GFP memb-mCherry

14 h.p.f. 14.5 h.p.f. 16.5 h.p.f. 17.5 h.p.f. 22.5 h.p.f.

Hoijman et al. Nat Comm 2015

60

Lumen axis length (µm)

30 50
25
Lumen volume

40 dx (ML)
Mouse embryonic AP
lung
(1×103 µm3)

dy (AP)
20 dx (DV)
30
15
20 ML
10
5 10 DV
0 0
17 18 19 20 21 22 23 17 18 19 20 21 22 23
Time (h.p.f.) Time (h.p.f.)

Figure 1 | Inner ear hollowing dynamics. (a) Dorsal confocal live imaging of otic vesicle from 13.5 to 23 h.p.f. showing lumen formation and expans
in coronal planes. Membranes and nuclei are stained by co-injection of membrane-GFP and H2B-mCherry mRNAs, respectively. Scale bar, 20 mm.
(b) Dynamics of F-actin-polarized distribution during lumen formation and expansion. Embryos are Tg(actb1:lifeact-GFP). Schematic representations
F-actin localization are shown below (F-actin in green, membranes in black). Scale bar, 20 mm. (c,d) Quantification of lumen volume (c) and length
each axis (d) during expansion of the cavity, obtained from data acquisition of 4D movies. Data are mean±s.e.m. (n ¼ 5). In (d), a linear regression for
data for each axis is shown. r ¼ 0.98 (AP) and 0.99 (ML and DV). (e) 3D lumen drawing showing its biaxial ellipsoidal shape at 23 h.p.f. The radii of the
ML and DV axes are shown.

NATURE COMMUNICATIONS | 6:7355 | DOI: 10.1038/ncomms8355 | www.nature.com/naturecommunications

Proag Development 2019 & 2015 Macmillan Publishers Limited. All rights reserved. Li et al. Dev Cell 2018
 Pulling
compressed in one direction and extended in the orthogonal direction,

Basic concepts of tissue rheology

Micromanipulators that pull on a cell or tissue sample deform it by
1. Elastic response
while the volume remains constant. In this figure and others, thick
•  Underaspirating
a given it through
shear the aperture (gray
stress
outlined arrows indicate forces. Forces are expressed in Newtons (N),
of a pipette with circularthe
rectangle), or
stresses in Newtons per square meter or Pascals (Pa). rectangular cross-section. At equilibrium, the shape of the sample
flows at
partainserted
constant rate,entrance
into the pipette whichresults
is proportional
from the balance ofto its
r a given stress
A (gray rectangle), B the material deforms surface
viscously tension
dissipatesand the pressure
theThe of
appliedaspiration (Evans
energy. and Yeung,
1989; Tinevez
and et al.,
fluid-like 2009).
at sample
long surface
timescale...tension can then be
Tissuesproportionally
ntaneously are elastictoatthe short
load timescale
to Fa constant
Stress σ =
strain
•  When inferred
the stress
from theis removed,
applied aspiration the flow
pressure and ceases
the radii of an
the
A
elastically
(definedstores
byAthe A applied energy.
stiffness)
a sample(defined by viscosity and surface
r e
n the stress is removed, the sample recovers
Deformation ε=
L–L 0remains in inside
its and outside thestate.
deformed pipette (Fig. 1Ca).
L 0
Suchtension)
surface tension measurements provide absolute values with
diately to its original shape.
Force F precision of a few percent: on cell aggregates in vitro, the biological
variability dominates the experimental uncertainty (Guevorkian
C et al., 2010). These experiments require a few minutes and are
therefore suitable for morphogenetic events much slower than
minutes, but not for quicker processes.
s ion
en L In one key example in the living mouse embryo, a quantitative
T
map in space and time of each cell surface tension using pipette
Compression aspiration revealed that the first morphogenetic movement of the
embryo, compaction, is mediated by increased actomyosin
Sh contractility, rather than by increased cell-cell adhesion as
ea
L0 r commonly thought (Maître et al., 2015; Fig. 1Cb). Pipette
aspiration has also been used at larger scales to measure the
surface tension of a group of sea urchin cells (Mitchison and Swann,
1954) and of mouse eggs (Larson et al., 2010), and to uncover the
role of tissue tension to ensure vertebrate 3D body shape (Porazinski
et al., 2015).

Manipulation using light

viscosity) but also on forces and stresses, in particular the material’s
Stress
surface tension = stiffness
and internal pressure atxdifferent
strainspatial scales.
Microplates are parallel plates that apply uniaxial compression on a
large surface area of the sample; one of these plates, either soft itself
or linked with a soft spring, acts as a force transducer. The sample
surface tension can then be inferred from the applied pressure and
the profile shape of a single cell or aggregate (Fig. 1Aa). Surface
tensions between tens of µN/m and tens of mN/m have been
measured in vitro on cell aggregates (Norotte et al., 2008; Mgharbel
In epithelia, cell-cell contacts are often supported by adherens
Stress
junctions, which can= be
viscosity
considered asxone-dimensional
strain ratestructures.
Tension at adherens junctions is in this case a line tension (expressed
in Newtons). To measure this tension, two methods use light to
perturb the cell-cell junction: optical tweezers and laser ablation; the
latter can also be applied at tissue scales to measure stresses. Thanks
to the penetration of light, both methods can be applied within the
living organism.
 From cell mechanics to tissue morphogenesis

•  Feedback loops across multiple scales

•  Involves biochemical signaling (which can change cell mechanics)

•  Self-organisation -> no intrinsic hierarchy, non-linear process

•  Emergent properties (e.g. collective cell jamming, stiffness sensing)

Ideal gas Law:

Collective cell organization can
PV = n R T only be described by mesoscale
physical principles!
How do mechanical forces
shape the tissue?
 Changes in tissue shape during development
RT I C L E S
Mouse
A R T I C Lembryo compaction
E SA R T I C LES
Maitre NCB 2015
a a b b b
γcm
γcm γcm
Pc Pc
Pc
Rp Rp
γR
Young–Laplace equation Young–Laplace equation
γcm Rc cmp
Rc
θi γ = P /2 (1/R
θ i cm Young–Laplace
c p – γ
1/R c ) = Pcequation
cm /2 (1/Rp – 1/Rc )
γcm Rc
γcc γcc
θθie θi θe γcm = Pc /2
θ iYoung–Dupré (1/R
equation p – 1/Rcequation
Young–Dupré )
α = γcc /2γcm = cos(θα /2)
= γ /2γ = cos(θe /2)
γcm γcc γcm
e cc cm
3:01 3:01 5:27 5:27 10:16 10:16
θe θi Young–Dupré equation
α =d γcc /2 d θe180
γ = cos( /2)
c
γcm 700 c 700 180cm e 600Increase
e in actomyosin
600 f
5:27 contractility
1.0 f 1.0
Surface tension γcm (pN µm–1)

Surface tension γcm (pN µm–1)

3:01

Surface tension (pN µm–1)

Surface tension (pN µm–1)

600 600 500leads to compaction of the embryo

Compaction parameter

Compaction parameter
θe θ 500 0.8 0.8
e
γcm γcm
Contact angle (°)

Contact angle (°)

500 500
120 120 400 400

α = γcc /2γcm

α = γcc /2γcm
700 400 400 d 180 θi θi e300 600 0.6 0.6
Surface tension γcm (pN µm–1)

Fluid-to-solid transition leads to tail elongation of zebrafish 300

Surface tension (pN µm–1)

300 300 0.4 0.4
600 60 60 200 200 500
200 200 θe
γcm
Contact angle (°)

0.2 γ /2 0.2
500 100 100 100 100 γcc /2 cc
120 400
0 0 0 0 0 0 0 0
400 60 120 60 180 120 0:00 180
3:00 6:00 9:00
0:00 12:00
3:00 6:00 9:00 θ i 6:00 9:00
0:00 12:00
3:00 0:00 12:00 0:0012:00
3:00 6:00 9:00 3:00 6:00 9:00
0:00 12:00
3:00
Contact angle θ i (°) Contact angle θ iTime
(°) post-division (hh:mm)
Time post-division (hh:mm) 300
Time post-division (hh:mm)
Time Time post-division (hh:mm)
post-division (hh:mm) Time pos
300
60 200
200 1 Spatiotemporal
Figure Figure map
1 Spatiotemporal
of tensions during
map ofcompaction.
tensions during
(a) Diagram
compaction. scale(a) bar,
Diagram
10 µm scale
(Supplementary
bar, 10 µmVideo(Supplementary
2). (c) Surface
Videotension
2). (c) cmSu
of the tension mapping
of the method.
tension mapping
Using a method.
micropipette
Usingof aradius
micropipette
Rp , a ofa radius
functionRpof, acontact
a function
angle ✓ofi measured
contact angle on n =
✓i 466
measured
blastomeres
on n = pool
466
100 Pc is applied
pressure pressure
to thePc issurface
applied
of to
blastomeres
the surface of of
curvature
blastomeres
1/Rc .of curvature
from 28 embryos
1/Rc . from 100of 11
representative
28 embryos representative
independentofexperiments. γccPears
11 independent /2
The surface tension The cmsurface
is calculated
tension using
cm is the
calculated
Young–Laplace
using the equation.
Young–Laplace
correlation
equation.
R = 0.517
correlation
(P < 0.001).
R = 0.517 (d–f) (PTime
< 0.001).
course (d–f)
of contact
Time cour
ang
From0 cm , the external
Fromand cm ,internal
the external
contact
andangles
internal and0 ✓i , respectively)
(✓e contact angles (✓e and ✓✓ii, and
respectively)
✓e (d), tensions (d), 0
✓i andcm✓e and tensions
cc /2 (e) and cm and
compaction
cc /2 (e)parameter
and compac ↵
60
of adjacent cells, ofwe120
adjacent
use thecells, 180use the
Young–Dupré
we equation 0:00
Young–Dupré
to calculate3:00 the 6:00
equation to Mean 9:00
calculate 12:00
± s.d.the
over 2
Mean
h of ±n s.d.
= 466 0:00 3:00 6:00 9:00 12:0
over
contacting
2 h of n blastomeres
= 466 contacting
pooled blastom
from
interfacial tension interfacial
cc and thetension
compactioncc and
parameter
the compaction
↵. (b) Representative
parameter ↵. (b) embryos.
Representative
Statisticsembryos.
source data
Statistics
are shown
sourcein Supplementary
data are shownTable
in Suppleme
1. Imag
Contact
images of tensionimages angle
mapping θ
of tension (°)
experiments.
i mappingTimeexperiments. Time
post-divisionTime post-division
(hh:mm); in b are
post-division (hh:mm)
(hh:mm); in b of
representative are11 Time
representative
independent ofpost-division
Mongera
experiments.
11 independent (hh:mm)
Natureexperiments
2018
Changes in tissue shape during development
Drosophila ventral furrow formation
 Role of mechanics in tissue patterning
Differential contractility also
Differential adhesion hypothesis leads to cell sorting
(Steinberg 1964)

Current Biology

Review

B
Trophectoderm fate
Germ band extension Apical domain
•  Cell sorting by
thermodynamic principles
•  Lowest interfacial free
energy cluster together
Higher contractility
Inner cell mass fate
Anterior midgut
Maitre Nature 2016
How does the tissue maintain
its size and pattern?
 Cells can ‘sense’ global tension
Mechanosensing: Active response of cells to external forces through
force-induced changes in biochemical signaling

Zebrafish epiboly

Campinho NCB 2013

 not follow the long-axis-rule at all. In such cases, division In the Drosophila wing disc epithelium, symmetric cell

…and orient their division accordingly

orientation is determined by the polarizing activity of
biochemical signals originating from the environment
(reviewed in Morin and Bellaı̈che, 2011). In addition,
externally applied forces have also been suggested to
Drosophila fly imaginal disc
1 Differential proliferation 2 Tension anisotropy
divisions preferentially align with the proximal-distal (PD)
axis, thus elongating the organ along this axis (Baena-López
et al, 2005). This preferential cell division orientation is
determined by the Fat-Dachsous pathway, which promotes
3 Cell shape anisotropy and division orientation

Low Tension
proliferation anisotropy

P
Tension builds up Cell shape changes
P D P

P High
proliferation

Mao et al. 2013, Legolf et al. 2013

Figure 1 Differential rates of cell division between distal (green) and proximal (red) regions of the Drosophila wing disc epithelium (1) give
rise to global patterns of tension anisotropy within the tissue (2). This tension anisotropy promotes cell elongation along the main axis of
tension, thereby controlling the orientation of cell division via cell shape anisotropies in proximal regions of the wing disc (3); D, distal;
P, proximal.

Tissue geometry
& 2013 European Molecular Biology Organization The EMBO Journal VOL 32 | NO 21 | 2013 2783
-> global tension
-> cell shape change
-> division axis orientation
-> smart way to relieve tissue stress!

Gudipaty Nature 2017

controlling the size of multicellular tissue cysts. To test our quantitatively, we start with a minimal model tha
Fluid pressure can regulate embryo size
predictions, we introduce an in vitro experimental system for the
study of oscillations in synthetic cysts and show how our general
proliferation and osmotic influx rate (Jo=KPosm
independent of the tension, and that the materia
theory is consistent with our experimental observations.
CT strengthens
Then, we may write a set of equations for the evo
CT increases Junctional leakage
tight junction
of the shell R(t), the thickness h(t), the hydrostati
upon mitosis

RESULTS Cellular level the wall stress σ(t) that couples cell proliferation, a
(cell shape,
Theory tension) to the generation of tension in the tissue wall,
Cavity expansion
To describe the growth and dynamics of a cyst, a multicellular cyst, release, and reads
Cavity as:
collapse
P ~ 150 Pa

we assume it has the shape of a simple spherical shell filled with fluid
at a different pressure relative to the environment, as shown in d
4p ðR2 hÞ ¼ Jc (cell volume conse
Fig. 2A. Cell proliferation changes the volume of the wall occupied Cell stretching dt
by cells, and any osmotic imbalance or active pumping creates an
inward flux of solvent that increases thelevel
Tissue radius of the shell while dR
Cavity size Cortical Tension

(lumen size,
¼ KðPosm $ High $ Q (lumen volume co
PÞ Pressure
stressing its wall. When the wall stress
pressure)
is greater thanLow
a
Pressure
critical Tight dt
junction
maturation

Size Control
Early Blastocyst
Late Blastocyst

Rupture stress
(mitosis)

Tissue stiffness
(junctional maturation)
Chan Nature 2019
 Mechanical cues can regulate gene expression
echanosensitivity
Matrix elasticity directs stem cell lineage specification

•  Stem cells can differentiate into

neuronal, muscle, bone
depending on substrate
stiffness

•  Actomyosin contractility is
essential

•  Mechanical microenvironment
can influence its fate

Engler, et al. Cell, 2006.

Engler Cell 2006
 Mechanotransduction in development

A B
Compression of
stomodeal cells Germ band extension Apical domai

Nuclear
β-catenin

Twist

Anterior midgut
Desprat Dev Cell 2008
 How do skin cells make evenly spaced hairs and feathers?
bmp2

Turing instability + mechanotransduction!

Uniform traction when adequately resisted Dermis COMPRESSES epidermis
leads to an array of aggregates force causes translocation of β-catenin
[field of dermal cells from above] [cross-section of tissue bilayer]
RESISTANCE

NONE

membrane

MANY
TRACTION

ONE
nucleus

Fig. 4. Movement of b-catenin to the nucleus in the forming primordium is mechanically trig-
gered and upstream of the primordia gene expression program. (A) b-catenin localization and2018
Shyer, Science
Lumen size can influence mouse blastocyst patterning
SOX2
CDX2

Reduced lumen size leads

to more cells allocated to
the inside of the embryo

Chan Nature 2019

 Mechanics can shape the signaling landscape

Tissue buckling -> shape of morphogen gradient -> cell differentiation

Figure 1. Forming a Morphogen Gradient by Tissue Folding
(A) Schematic of morphogen gradient model: morphogens are secreted by source cells and form a graded concentration profile in the target tissue, where cells
express different target genes (X, Y, and Z) and ultimately adopt different cell fates dependent on the morphogen concentration.
(B) Tissue folding leads to villi formation in the gut. Shh molecules are shown in blue. As the villi grow more acute, the maximal Shh concentration at the tip
Shyer,
increases; the Shh concentration ultimately exceeds a high threshold (dotted lines) above which the formation of the villus cluster in the underlying Cell,
mesenchyme is 2015
Approaches to study
morphogenesis
eins (kon, koff in panel f), and the activity successfully described complex features of cell monolayers
hat pull on the filaments1 µ(F
Theoretical
m in panel f).
models d(multi-scale!) such as 10
wave
µm propagation in the absence
30 µmof inertia .
76,133

c 1 µm 10 µm 30 µm
e f g h
eb II Clutch model f c 3D vertex model,
g d Potts model h
Myosin
Actin
Myosin II
v Actin kon/koff σ
Adaptors γab
v df
kon/koff
AdaptorsCrosslinks γab
df v
Integrins Myosin II F
kon/koff Crosslinks γl
Substrate
Integrins Actin Myosin II F
kon/koff
Fn γl p
Substrate
10 µm Actin 30 µm 100 µm
Fn
Mechanical
µm structures and modelsg of cells10and h
µmtissues at different length scales. a, Aµm
30
Continuum model
focal adhesion of an unroofed HeLa cell
100stained
µm
phalloidin (cyan) and imaged with total internal reflection fluorescence microscopy/platinum-replica transmission electron microscopy
correlative . b, An osteosarcoma η (red) was imaged using stimulate
f light–electron microscopy cell (U2OS) plated onlength
fibronectin.
hActin
134
Mechanical structures and models of cells
g and tissues at different scales. a, A focal adhesion of an unroofe
Polymer
depletionphalloidin
microscopy network
and aand model
focalimaged
adhesion marker ζ c, A monolayer
kon/koff (cyan) with total(paxillin,
internalgreen) was imaged
reflection using confocal
fluorescence microscopy.
microscopy/platinum-replica
σ of epitheliae
transmission
using a membrane-targeted 134γ (green). d, An immunofluorescent image of a small intestinal organoid
green fluorescent protein η (red) showing
ctin df correlative light–electron microscopy . ab b, An osteosarcoma cell (U2OS) plated on fibronectin. Actin was imam
proliferation (EdU, red), differentiation (Krt20, green) and nuclei (blue). e, Clutch model of F-actin linkage to fibronectin ζ(Fn) via integr
ks depletion microscopy and a focal adhesion marker (paxillin, green) was imaged using confocal σ G. microscopy. c, A mo
network model. g, 3D vertexkon /koff h, Continuum model. Credit: courtesy of J. Taraska and
model. v K. Schack
ΔP (a) and Jacquemet and J. Ivask
n II using a membrane-targeted green fluorescent protein (green).
γ ab d, An immunofluorescent image of a small intestinal
adapted from
F ref.!135, Bio-protocol
df (d).
proliferation γl
(EdU, red), differentiation (Krt20, green) and nuclei (blue). e, Clutch model of F-actin linkage to fibrone
Crosslinks
ctin network model. g, 3D vertex model. h, Continuum model. Credit: courtesy of J. Taraska and v K. SchackΔP(a) and G. Jac
Myosin II F p
adapted from
shear component 5 ref.!normal
. The 135
, Bio-protocol
component(d).indicates the extent to
γl extensions called filopodia or lamellipodia, and sphe
hich a material is tensed or compressed at a given point, whereas the blisters called blebs. Cells migrate using lamellipod
Actin
ear component indicates forces parallel to the surface. blebs, depending on a diversity
p of intrinsic and ext
ndBesides
tissuesthe
a shear ability
at differenttolength
component generate
5
. The active
normal
scales. stresses,
a, A a key feature
component
focal adhesion thattheincluding
indicates
of an unroofed extent toexpression
HeLa cell of fluorescent
extensions
stained with adhesion proteins, density
called filopodia or lamea
fferentiates
nal which tissues
reflection fromispassive
afluorescence
material tensed soft materials is at
or compressed
microscopy/platinum-replica thea given
self-propul-
point,
transmission of the ECM,
whereas
electron the confinement,
blisters
microscopy topology
called
(grey) blebs.and cortical
Cells con
migrate
Approach depends on the level of abstraction required
nveosteosarcoma
ability
shearofcomponent
eachcell
individual cell. Cells
indicates
(U2OS) plated forcesachieve
parallelself-propulsion
on fibronectin. to the(red)
Actin surface. migrate,
wasbyimaged using cells adhereemission
blebs,
stimulated their protrusions
depending to the surr
on a diversity
models of cells and tissues at different length scales. a, A focal adhesion of an unroofed HeLa cell stained with fluorescent of
 of the velocity of the cell and, therefore, of the movement of its
ariety of models neighbours have. However,recently beenis betterimplemented to explain of the energy function but also by a
Two commonly used models
88
dynamics of monolayers captured
by imposing an active feedback mechanism inspired by flocking
rich
1/μ phenomenology
is the friction models that tends toof
137
aligncell monolayers
cell polarity with cell velocity.
79,121–123,136
. Among The equation of motion then reads:
VIEW ARTICLE | INSIGHT
specified explicitly Given that cell velocity is in part due to the force exerted by the
chcontributions,
models, or vertex
neighbours, models
this feedback treat
mechanism the cellcouples
effectively monolayer
the as a close-
of an energy func- movement ofVertex
| of polygonal objects that represent constituent Cellular Potts Model dx k
Model
VIEW ARTICLE INSIGHT
ked mosaic adjacent cells with an alignment rate J (refs ).
86,87

to compute such
= μFkNAT
+ v0
s//doi.org/10.1038/s41567-018-0194-9
en 82,83
. These models study the dynamic evolution of the 2D
cell i in terms of
dt
Analogy to Ising model in magnets
work
K (P −P ) formed by cell vertices and by connections between
J
2
Pi i i0

m. 3 | TheVertex modelsxk of each vertex k is determined by the equa-

position vk where nk is a unit vector indicating
nticof motion
ty where A is a i0
v is a constant. In the simplest appro
ariety modulus.
ofThismodels have recently been implemented to explain of the 0energy function but also by a self-pro
stood in terms
trich
fluctuations.
of
phenomenology
The of dcell monolayers 79,121–123,136
. Among The dynamicsofismotion
equation governedthen byreads:
random rotati
, with a preferred
vi
xk Cell k
of the velocity of the cell and, therefo
. Expansion ofvertex models treat =
hPK, which
models, theμFcell k monolayer as a close-
Area Ak
Ai Perimeter Pk

ked
i describes
r term inmosaic
P , which of polygonal objects dt that represent constituent neighbours88. However,dx k dynamics of m
cell adhesion
i
and
by imposing an active =feedback
μFk + v0 n mech
k
sell82,83 . These
adhesion tends models study the dynamic evolution of the 2D
Cell i
d t
ere Fkformedis the force acting on the andvertex and 1/μ isbetween the friction models137 that tends to align cell p
Area Ai

work s cortical tension

by cell vertices
Perimeter Pi
by connections
ween
mulation ofthe vertexcell and its substrate. F may be specified explicitly Given
m. The position
ved from particle
xk of each vertex k isk determined by the equa- where nk that cell vector
is a unit velocity is in part
indicating thedue
pol
acell ofsum
(SPV) motion
models of .contractile, adhesive and osmotic contributions, or
vertices, SPV
86,87
v0 is neighbours,
a constant. In this feedbackapproximatio
the simplest mechanis
ained implicitly
each cell obtained Illustrationby computing
of vertex and SPV models. In thethe gradient
absence of an energy func-
of self-propulsion, movement
dynamics of adjacent
is governed by randomcellsrotational
with an alig
diff
nnlytex of the monolayer. A common x
on describing the cell dynamics is determined by changes in cell area and perimeter. With
d k
= μFk approach to compute such of the velocity of the cell and, therefore, of th
models but the self-propulsion, two adjacent cells align their velocity vectors according to
by the gradient an alignment rate J.
rgy function is to express dt the energy of a given cell i in terms of neighbours88. However, dynamics of monolaye
area
nal changes Aiinand pro- its
someperimeter Pi: dynamic signatures such as the abil-
cells also have distinctive by imposing an active feedback mechanism in
re F is the force acting on the vertex and 1/μ is the friction models137 that tends to align cell polarity
The substrate stiff- ity to rotate in swirls69. The dynamics of these swirls is determined
ned by thek binding by intrinsic and extrinsic factors such as cell adhesion, division and J
weenEthe cell and =∑ = ∑ K F(Amay
itsEsubstrate. −Abe) specified
+ K (Pi −explicitly Given that cell velocity is in part due to the f
Nonetheless, trac- confinement70. 2
the flow of actin in
monolayer Pi 0 )2
A
k
The renewed interest in cell monolayers—an
i i experimental sys-
i 0 P
sum of contractile, adhesive and osmotic contributions, or neighbours, this feedback mechanism effec
ction generated by tem that dates back more than one icentury—originates from thei
developmenti of new technologies
i to measure not only velocity
ained implicitly by computing the gradient of an energy func- movement of adjacent cells with an alignmentv
and deformation fields but also tractions and intercellular stresses
cally identical, cells (Box 1)25. Force mapping has unveiled phenomenological principles
of the monolayer. A common approach to compute such
The first Pterm= preferred
represents perimeter
an area elasticity where Ai0 is a
in terms of shape, of cell organization such as the alignment of the cell body with
e to a free edge are iothe direction of maximum stress5,71,72. This phenomenon, called
rgy function is
K areato express
= area the energy of a given cell i in terms of
ferred cell
to be cuboidal and Aiandelastic
K is modulus
an area elastic modulus. This
rate large traction plithotaxis, implies that cells organize in sheets so as to minimize
intercellular shear stress. A
rea A and its perimeter P : Plithotaxis
i provides a mechanim for cells
tributionKto =the
i
lamellipodia
perimeter elastic
cell energy modulus
can be understood in terms of
i a preferred mechanical direction during
pro- to migrate collectively in
ithin these two cat- Pi
wound healing and cancer invasion73. J
incompressibilityE and
= resistance
cells at the leading
K (A −A to) height
+ K (P fluctuations. The
The organization of cells in sheets and clusters has also been
E = ∑ −P )2 ∑
an their immediate shown to enable collective sensing of both chemical 2 and mechani-
 R1
0 In
Contact manipulation
Time (min)
Development (201
ablat
Atomic force microscopy Parallel plate compression morp
Ba Bb variety
12 of tissues and orga et al.
10
femtosecond lasers like cellus
those

Force (µN)
8 Pure water
Force

(Rauzi
6 et al., 2008) or
0.05 M NaCl pulsed(Fig.
UV
* can
4 be employed.
0.1 M NaCl
0.15 M NaCl singl
0.2 M NaCl
0 2
In the Drosophila0.25 M NaClembryo ablat
an
Turgor Turgor 0.3 M NaCl
Time (min)
0
ablation polar
–1 0 1experiments
2 3 4 5 have revea
Indentation depth (µm) Sl
Beauzamy Biophy J. 2015 morphogenetic movements and d
tissu
12 et al., 2000; Farhadifar et al., 2007
Ca 10 Cb gene
cell junctions drives cell Cam inter
* *
Force (µN)

*
8 Pure water
6 Rp ∆P 0.05 M NaCl (Fig. 2Ba; Rauzi et al., 2008). of w
R 0.1 M NaCl
4 0.15 M NaCl single-cell embryo, anisotropystres of c
0.2 M NaCl
2 0.25 M NaCl ablation Surface
produces cortical
tension or elasticity flows,
large
0.3 M NaCl
0
polarity (Mayer et al., 2010). once
Fig. 1. Contact manipulation. (A) Pushing: microplates. (Aa) A force applied
–1 0 1 2 3 4 5
Indentation
on a cell or aggregate depthcompresses
sample (µm) Slightly enlarging the scale
it with a known applied pressure (P). Wou to
Image analysis determines the sample profile tissue level,is characterized
shape, which a laser canbysimultan whic
 module F-actin

Optical and magnetic manipulation

Microdroplet z
∆P Optical tweezer FRET biosensor
Micropip

Pressure
t0 t1 sensor
t2 T
θ Ft Pre
Droplet-based sensor

High stiffness T Tension

2D epithelium

Force inference
Bambardekar PNAS 201

Magnetic tweezer Bb B Optogenetics

Spatio-temporal control of forces in vivo

displacement (µm)
Brillouin microscopy

Optogenetics

Vertex
Blue laser
Magnetic tweezer

t: 10 s t: 68 s
Time (s)
Inhibition of apical constriction
Magnetic liposomes
blocks mesoderm invagination

Cb
Desprat Dev Cell 2008
C Ex vivo reduced systems
 Bb
Laser ablation to probe tissueBbtension

displacement (µm)
Ba

displacement (µm)
Vertex
T

Vertex
T

t: 0 s t: 10 s t: 68 s
Time (s)

Ca Cb
t: 68 s
Time (s)

Ellipse axis
length (µm)
Stress

Cb Time (s)
t: 0 s t: 1 s t: 30 s
Ellipse axis
ight. (A) Optical tweezers. Principle of the optical trap. A transparent particle (with a higher refraction index
Stress length (µm)
[red; color scale indicates the light intensity (I )] is subject to two forces: a gradient (Aa) and a scattering force
pulls the particle along the optical axis (z) and in the transverse plane (x, y), towards the highest intensity re
ht panel). The black arrows indicate the refraction of light. (Ab) The scattering force (Fscat; unfilled arrow) push
on of incident light. The black arrows indicate the scattering of light. An effective optical trap occurs when Fg
can also be trapped directly. Trapping a cell-cell interface and moving it (shown in three successive picture
derm) can be used to determine tension (T ) by considering the balance between T and the trapping force (
•  Directionality, relative tension
θ is the angle that the interface makes with respect to Ft. (B) Subcellular laser ablation. (Ba) Inset shows netwo
•  Also as
r (red arrowhead), way of
vertices tothe
induce tensioncell-cell
manipulated in vivointerface (yellow circles) move Campinho
apart, indicating that
Time (s) NCB 2019
 Migration
0.55 High
Visual sensors
Microdroplets Bb
ted In a tissue
Anti-E-cad

Quantifying cell scale stresses during zebrafish development

Cell 1 Cell-generated forces that emerge during neural rod formation in zebrafish embryo

quantified by computationally reconstructing the deformations of embedded PLL-Cy3

Cell 2 beads. The beads were microinjected into the developing neural plate at the tailbud sta

ntifying cell scale stresses during zebrafish development hours post fertilization (hpf)) and time-lapse confocal microscopy was used to visualize

escent Adhesion dimensional shape changes of the beads in the neuro-epithelium (Fig. 4).
cules molecules
generated Anisotropic
forces that emerge during neural rod formation in zebrafish embryos stress were σ (nN µm )
n –2

tifiedBead sensors reconstructing the deformations of embedded PLL-Cy3 PAAm

by computationally Mongera Nature 2018
RET tension sensor module consists of two fluorophores, a donor (blue) and an acceptor (green), linked with a
ases
s. Thewith
beadspulling
were forces (arrows)
microinjected intoacting on it. (Ab)
the developing Energy
neural platetransfer efficiency
at the tailbud (light gray arrows in Aa) sharply de
stage (10
Ac) The FRET efficiency-force relationship is experimentally calibrated in vitro. Red line indicates a fitting cu
so post fertilizationmolecules
DE-cadherin (hpf)) and time-lapse
can measureconfocal microscopy
their tension. was
FRET used to visualize
efficiency three-
(color bar) in border cells of Drosophila
made shape
nsional of fluorocarbon oil (light
changes of the beadsorange) is coated by surfactant
in the neuro-epithelium (Fig. 4). molecules, fluorescent molecules and adhes
olated drop is measured in solution (left panel). When the liquid drop is inserted between cells in a tissue (right
age analysis determines the local mean curvature of the droplet surface (H ), from which the anisotropic comp
be determined according to the equation σ = 2γH. (Bb) Liquid drops (red) injected between tooth mesenchy
he anisotropic stresses (color bar) are mapped onto the three-dimensional shape of the liquid drop (enlarged
t al. (2014) and (Ba, Bb) Campàs et al. (2014).

cted. Finally, the measurement exerted on the drop (Fig. 3Ba). Provided that the drops ar
Treber bioRxiv, 2019
n the spring is entirely stretched enough compared with tissue size, they can be left in place f
 CellFIT: A Cellular Fo
Force inference
Laplace (non-invasive)
equation, is
Ca Cb Cc
*
0,41]. 23 Within *
hr APFany given cell, these asTit approaches the TJ. For any particular TJ to
1 P1
hanically analogous, but opposite in the adjacent cell edges must Dpijsatisfy
~cthe ij =r force
ij b
pressures that might be present. Here P
15.2 h 16.8 h
expansion force as a positive effective 3
P2 of
*
T2 33.6 h X
^rmnA
where
simply intracellular pressure. Cell the radius
T 3 Timecurvature cmnr ij , ~0, mu
the
Tension
o occur sufficiently slowly that membrane
viscous curvature,
where Pressure
the unit is ^considered
vectors rmnA are constructed posittan
he cell can be considered quasi-static 0.4along the
Cc
pared to VFM. As a result of this convex Cd into angle cell j.
at which the membrane Tensionbound
and n approaches the Ath triple junction and p
formulation is not appropriate
chanical observation techniques. The (A)
for cell
Internalmembrane
birefringence.
the junction, behaves
(Aa) A tissue
and the summation rather
under
is carried an
out
ecoil or other rapid motions where
s of larger and smaller stresses (σ1 and connect
σ2). Thistocan beTJ.detected using polarized li
icant. Cells may also carrynecessarily bowed thatwhen pressure
The cWhich mn values
toolare differ
the
to use? corre
tractions,
ts of combined linear and quarter-wavelength membrane polarizers andFor
tensions. illuminated.
illustrativeThe light beF
purposes,
etween themselves and a subjacent sides as
or
ccording to the refractive indices of the sample.
a consequence
such cell Spatial
edges, but of
variations
there wind action.
of retardance
•  could
Range beof more,
force as thea
T
, but these mechanical effects are
esent
following
study. equation: difference,
retardance∼c d (σ 1 −σ the
and2 )/λ greater
(where
rosettes c is
[44]. thethe •  tension
Length scale
photo-elastic in thea
constant
ontrol (left) and vertically stretched (right) Drosophila
Equation 3 wing discs.
may seem • oversimplified,
Invasive vs. not
Compressive stres
tak
ysis
0.4deals only with given
in-plane
1.6 geometry.
forces,
–0.2 Furthermore,
0.15 In fact as the
non-invasive riggi
released Tension
by tissue stretching.
it considers them to be in (B)
governed External Pressure
into birefringence.
proper account. (Ba) By changing
it is complete, the ma as
patterns
ge tensionsare obtained.
and
the sail considering
(Bb)pressures.
intracellular Measurement increases Figand
of forces during to its
2C the curvature
represent
growth ofa avery small
chickpea redrer
Aa) A tissuelabel
asterisks undertheanisotropic
root tip. stress
(C) Forcein axes 1 and
labeled
inference. A,2 has
as
(Ca) different
suggested
Input imagebyrefractive
the
of small indices
epithelialredacrocircle
cell s
where
with other recent studies of cell sheets there is no pressure difference
ed using
).t (Cb) polarized
Image light.
analysis In the
yields set-up
cell shapes,shown,
region
vertex aofsample
interestofisand
positions thickness
made d PLOS
smaller,
connectivity;
Brodland is the
placedlength
they
One areo
2014
hamper future enhancement or
have negligible curvature but still carry ten
Conclusion and outlook
 Physics can contribute to developmental biology
Statistical physics
•  Quantitative approach
Collective cell migration as unjamming transition
•  Conceptual framework

Soft
a
matter physics
o
Tissue rheology, wetting-dewetting
Tissue as liquid crystals? b
z to
z
x
y y
Bi Nat. Phys. 2015
w
to
Critical phenomenon in clonal dynamics w
ra
b –270 min –160 min –30 min 0 min 23
Same fractal dynamics as
for aerosol and polymer
d
1

(μm h–1)
branching network

Speed
th
ti
0
Saw Nature 2017 Rulands Nat. Phys. 2018 D
 Many challenges remain…

•  Robustness and precision in symmetry, size, shape?

•  From 2D to 3D
•  Crosstalk between fluid flow and tissue mechanics

Role of mechanics in evolution

Do mechanical cues from environment

exert evolutionary pressure?

Stoddard Science 2017

 Other processes guided by physical principles…

•  Morphogen gradient (James Sharpe)

•  Oscillations
•  Bioelectricity (e.g. regeneration)
•  Plants
…and the list is not exhaustive…

References:

•  Biological physics of developing embryo (G. Forgacs, S. Newman)

•  Physical biology of the cell (R. Phillips et al.)
•  Introduction to cell mechanics and mechanobiology (C.R. Jacobs et al.)
•  Kavli Institute for Theoretical Physics
(https://www.kitp.ucsb.edu/activities/morpho19)
•  iBiology (https://www.ibiology.org)
 Mechanotransduction
Nucleus as a mechanosensor

Wang, et al. Nature Reviews:

MCB, 2009.

•  (1) A local force applied to integrins through the ECM is concentrated at focal adhesions an
channelled to actin filaments, which is bundled by a-actinin and made tense by myosin II, w
•  Physical transduction:
generates prestress.tension from ECM can be transmitted to the nucleus
-> cell
•  functions as a single
(2) Actin filaments mechanically
are coupled to MTs andcoupled system
to IFs via plectin 1. Plectin 1 also connects IFs w
nesprin 3 on the outer nuclear membrane.
•  (3) Nesprin 1 and nesprin 2 connect actin to the inner nuclear membrane protein SUN1; nes
•  Chemicalconnects
transduction: focal
plectin 1 to adhesion
SUN1 and SUN2. proteins can alter their binding kinetics and
shuttle• between
(4) The the
Sun cytoplasm andtonucleus
proteins connect to modulate
lamins that form nucleartranscriptional activityto
scaffold, which attaches
•  Chromatin and DNA.
•  This whole process takes milliseconds, as opposed to the seconds taken by chemical diffus