Developmental Biology 433 (2018) 254–261

Contents lists available at ScienceDirect

Developmental Biology
journal homepage: www.elsevier.com/locate/developmentalbiology

Original research article

Linking stem cell function and growth pattern of intestinal organoids MARK
a,⁎,1 a,b,1 a c,d
Torsten Thalheim , Marianne Quaas , Maria Herberg , Ulf-Dietrich Braumann ,
Christiane Kernerb, Markus Loefflere, Gabriela Austb, Joerg Gallea
a
Interdisciplinary Centre for Bioinformatics, Leipzig University, Haertelstr. 16-18, 04107 Leipzig, Germany
b
Department of Surgery, Research Laboratories, Leipzig University, .Liebigstr. 19, 04103 Leipzig, Germany
c
Faculty of Electrical Engineering and Information Technology, Leipzig University of Applied Sciences (HTWK), Wächterstraße 13, 04107 Leipzig, Germany
d
Fraunhofer Institute for Cell Therapy and Immunology (IZI), Perlickstraße 1, 04103 Leipzig, Germany
e
Institute for Medical Informatics, Statistics and Epidemiology, Leipzig University, Haertelstr. 16-18, 04107 Leipzig, Germany

A R T I C L E I N F O A BS T RAC T

Keywords: Intestinal stem cells (ISCs) require well-defined signals from their environment in order to carry out their
Intestinal stem cell specific functions. Most of these signals are provided by neighboring cells that form a stem cell niche, whose
Organoid culture shape and cellular composition self-organize. Major features of this self-organization can be studied in ISC-
Computational modeling derived organoid culture. In this system, manipulation of essential pathways of stem cell maintenance and
Wnt
differentiation results in well-described growth phenotypes.
Notch
We here provide an individual cell-based model of intestinal organoids that enables a mechanistic
explanation of the observed growth phenotypes. In simulation studies of the 3D structure of expanding
organoids, we investigate interdependences between Wnt- and Notch-signaling which control the shape of the
stem cell niche and, thus, the growth pattern of the organoids. Similar to in vitro experiments, changes of
pathway activities alter the cellular composition of the organoids and, thereby, affect their shape. Exogenous
Wnt enforces transitions from branched into a cyst-like growth pattern; known to occur spontaneously during
long term organoid expansion. Based on our simulation results, we predict that the cyst-like pattern is
associated with biomechanical changes of the cells which assign them a growth advantage. The results suggest
ongoing stem cell adaptation to in vitro conditions during long term expansion by stabilizing Wnt-activity.
Our study exemplifies the potential of individual cell-based modeling in unraveling links between molecular
stem cell regulation and 3D growth of tissues. This kind of modeling combines experimental results in the fields
of stem cell biology and cell biomechanics constituting a prerequisite for a better understanding of tissue
regeneration as well as developmental processes.

1. Introduction
Self-renewal of the intestinal tissue is substantially impaired in
several developmental diseases (Markel et al., 2008) and its efficiency
decreases during ageing (Martin et al., 1998). Moreover, the intestinal
tissue is not capable of macroscopic re-growing after resection leading
to short bowel syndrome (Donohoe and Reynolds, 2010). Tissue
engineering applying in vitro expanded intestinal stem cells (ISCs)
has been envisioned as a possible therapeutic approach.
In order to enable such approaches long-term in vitro culture of ISCs
has been tried for decades. Sato et al. (2009) successfully established an
intestinal organoid culture that enables long term tissue expansion. Mouse
organoids require specific culture supplements (EGF, Noggin, R-spondin:
ENR) for expansion. Under standard ENR conditions organoids grow in a
branched pattern, resampling crypt- and villi morphology of the intestine.
Moreover, they contain all types of cells present in the mouse intestine,
including ISCs, Paneth cells (PCs), goblet cells (GCs) and enterocytes (ECs).
Organoids obtained from mouse intestinal tumors grow in a cystic pattern
without Noggin and R-spondin and show a quantitatively different cellular
composition (Sato et al., 2009).
The establishment of intestinal organoid culture together with the
development of sophisticated mouse reporter systems provided new in-
sights into ISCs organization in the last years (Merker et al., 2016). For
instance, Farin et al. (2012) demonstrated a link between the molecular
regulation of Wnt, the cellular composition of the organoids and their
growth pattern. In particular, they showed that increased Wnt-activity
changes the growth pattern of the organoids from branched to cyst-like,
while in parallel the number of PCs per organoid bud increases. Moreover,

Abbreviations: ISC, intestinal stem cell; PC, Paneth cell; GC, goblet cell; EC, enterocyte; ENR, EGF- Noggin- R-spondin; NET, model polymer network; PSET, parameter set
⁎
Corresponding author.
1
Equal contribution.

https://doi.org/10.1016/j.ydbio.2017.10.013
Received 30 May 2017; Received in revised form 28 September 2017; Accepted 13 October 2017
Available online 30 November 2017
0012-1606/ © 2017 Elsevier Inc. All rights reserved.
T. Thalheim et al. Developmental Biology 433 (2018) 254–261

they found that PC-derived Wnt-3 is required for intestinal organoid
culture, and that deletion of Wnt-3 can be rescued by adding exogenous
Wnt-3a. Notably, Wnt-activity in ISCs can be bistable. This bistability
originates from a positive feedback loop of the TCF/β-catenin complex
(Schuijers et al., 2015). Switches between the states are associated with
lineage specification, where high Wnt-activity is specific for ISCs and PCs
and low Wnt-activity for ECs and GCs. The absolute level of Wnt-activation
can be modulated e.g. by changing R-spondin 1 (in the following R-
spondin) concentrations in the culture medium. R-spondin suppresses
Rnf43 and Znrf3, which are repressors of Frizzled (Wnt-) receptors (Farin
et al., 2016). Experiments by Yin et al. (2014) manipulating the activity of
the Wnt- and the Notch- pathway demonstrated that the cellular composi-
tion of organoids depends specifically on culture supplements and that
nearly pure ISC organoids can be grown.
Despite the experimental progress, a comprehensive mechanistic
model explaining ISC self-organization in organoids is currently
missing. There are only a few computational approaches to describe
organoid growth (Pin et al., 2015; Langlands et al., 2016). In 2012, we
have introduced a first individual cell-based model approach (Buske
et al., 2012) which builds on our intestinal crypt model (Buske et al.,
2011). This computational model allowed for simulations of the self-
organization of the ISC niche within an organoid. Here, we provide an
extension of this model that enables a mechanistic explanation of how
alterations of the Wnt- and Notch- pathways impact on the organiza-
tion of the ISC-niche and organoid growth pattern. For this purpose,
we transferred regulatory principles of Wnt- and Notch- signaling, that
we have successfully applied to describe intestinal crypts (Thalheim
et al., 2016), to our organoid model. Thereby, we link for the first time
evolving 3D morphology of intestinal organoids to molecular regulation
of these pathways in individual cells.
2. Methods
2.1. The model of lineage specification
Our model is based on a set of simple principles of self-organization
of the ISC-niche under homeostasis which are described in detail in
(Buske et al., 2011, 2012). A sketch of these principals is shown in
Fig. 1A. We assume that ISCs require Wnt- and Notch-activity for self-
renewal in agreement with experimental findings (van de Wetering
et al., 2002; Fre et al., 2005). If Wnt-activity decreases, modeled ISCs
specify into absorptive ECs, while decreasing Notch-activity forces
them to specify into secretory PCs. Notch-ligands are provided by both
secretory lineages, PCs (Sato et al., 2011) and GCs (Stamataki et al.,
2011), while Wnt is provided by PCs only (Sato et al., 2011; van Es
et al., 2012). Thus, in the absence of exogenous Wnt, ISC-maintenance
requires PCs in the direct neighborhood. Therefore, two essential
parameters of the model are the minimal numbers of Wnt-secreting
neighbors NW and of Notch-ligand-presenting neighbors NN that are
required for ISC maintenance. Outside the niche, ECs require NN,EC
Notch-ligand-presenting neighbors in order to prevent specification
into GCs.
Allowing ISCs to specify into PCs independent of their environment
would result in a permanently expanding ISC-PC system, i.e. an
expanding ISC-niche, capable of self-maintaining. Because such ex-
pansion is not seen in vitro under ENR conditions, we restricted ISC-
specification into PCs to those ISCs that are located in highly convex
areas of the tissue (Buske et al., 2012), where the local mean curvature
C is larger than 1/R0 (Fig. 1B). Here, R0 is the radius of a spherical
organoid in which ISCs can no longer specify into PCs. Enrichment of
ISCs and PCs in highly convex areas is indeed observed for branched
organoid growth (Sato et al., 2009). Moreover, the assumed principle
provides an explanation, why prolonged culture of flat intestinal tissue
was not successful (Booth and Potten, 2000). Key extensions of the
lineage specification model compared to our original approach are
explained in the Results section.
2.2. Biomechanical properties of the organoids
We described the biomechanical properties of the organoids by the
properties of a closed but permanently reorganizing layer of semi-
flexible polymers (Buske et al., 2012). This model allowed for simple
control of the expansion and shape of the layer by adjusting two
parameters: the elastic modulus of the polymers KL and the bending

Fig. 1. The branched growth pattern. A) Regulatory circuit assumed in the model. Lineage specification (black arrows) is controlled by external signals and thus is reversible. For self-
maintenance, ISCs require at least NW and NN neighbors that secret Wnt-3 (blue arrow) and provide Notch-ligands (brown T bar), respectively. ECs require at least NN,EC neighbors that
provide Notch-ligands (brown T bar) to suppress GC-specification. PC-NET interactions induce local curvature that is required for PC-specification (orange arrows). B) Slice of a
simulated organoid (viable cells): PCs are specified in regions with high mean curvature (C > 1/R0). C) Phase contrast image of an organoid grown under ENR conditions (scale bar:
200 µm). D) Snapshots of an in silico growing organoid (SC: red, PC: green, GC: yellow, EC: blue; NN=3, PSET I). Numbers indicate days (d) in culture. E) Cell type composition of the
organoid shown in (D) at selected time points. F) The sphericity of organoids (PSET I) and of their lumen decrease over time indicating the formation of branches. Organoid sphericity
reaches a plateau after about 10 days as the cell composition stabilizes (n = 6 organoids, mean ± SD).

255
T. Thalheim et al.
modulus KC of the surface of the polymer layer, consisting of triangles
formed as a consequence of polymer linkage. Thereby, these para-
meters underlie two essential confinements. Firstly, one has to choose
KL < KL,0, at least for a subset of the polymers, to ensure that the
organoids behaves like a viscous material (on the time scale of the cell
cycle τ). For these values of KL the expanding cell population is capable
of stretching the net such that the triangles become larger than a
threshold size, new polymers are introduced and the local polymer
stress becomes relaxed, i.e. cell growth can overcome the contractile
force of the layer. Thereby, KL,0 depends on the sensitivity of the cells to
contact inhibition of growth, i.e. for KL > KL,0 the system becomes
completely contact inhibited under the given threshold of volume
conservation. Importantly, for KL < KL,0 the organoid can grow with a
speed depending on the actual value of KL. Secondly, we choose KC
such that the resulting forces are small enough to ensure that the
polymer net will not expand by itself, in order to form a flat layer. This
means that KC is always smaller than KC,0(KL).
All cells are assumed to adhere to the polymer network (NET). Cells
that loose contact undergo apoptosis. The apoptotic cells remain in the
organoid lumen, as observed in vitro. To facilitate bud formation, we
assume PC-specific interactions with the NET that favor formation of local
tissue curvature. We assume PCs to adhere stronger to the NET and to
weaken the elastic modulus KL of the polymers they are attached to. Based
on these assumptions, a positive feedback loop is given between the local
occurrence of PCs and local tissue curvature (Fig. 1B). To enable robust and
efficient simulations of bud formation, we consider a model extension
compared to Buske et al. (2012). We support bud formation by introducing
an apical network of the cells. This net interconnects neighbor cells by semi-
flexible polymers. Details can be found in Appendix A.
The velocity of reorganization of the NET under the effective force
of polymer-polymer- and cell-polymer-interactions is controlled by
friction constant γM. Details about the model dynamics can be found in
Appendix B. The parameter sets (PSETs) applied in the simulations are
given in Appendix C.
2.3. Mouse organoids
Preparation, culture, 3D imaging, and growth analysis of normal
intestinal organoids of C57BL/6 mice were performed as described
(Keysselt et al., 2016). About 50 ng/ml mEGF (Thermo Fisher
Scientific, Waltham, US), 20% Nog-Fc-conditioned medium and 10%
Rspo1-Fc-conditioned medium (1.1 μg/ml) were added to the organoid
culture medium (ENR). The Rspo1-Fc-expressing and the Nog-Fc-
expressing cell lines were kind gifts from Calvin Kuo (Stanford, CA,
USA) and Gijs R van den Brink (Amsterdam, Netherlands), respec-
tively. Wnt-3a conditioned medium was produced using stably trans-
fected L cells (ATCC® CRL-2647™). After 7 days of primary culture in
ENR the organoids were passaged 24 h before starting experiments.
Organoids were further cultured for 6 days at different R-spondin
(25%, 10%, 1.25% and 0.0%) and Wnt-3a (25%, 0.0%) concentrations.
Growth of organoids was recorded by 3D Life Scanning microscopy
consisting of a CKX41 microscope (Olympus, Hamburg, Germany) and an
incubation system (ibidi, Martinsried, Germany). Z-stack series were taken
for 6 days every 4 h. All-focus images were generated using the Extended
Focus Imaging projection tool of the CellSens software (Olympus).
For fluorescence imaging, cultures were fixed (2% paraformalde-
hyde, 40 min at room temperature), permeabilized (0.2% Triton X-
100/PBS, 30 min at 4 °C), blocked (Universal Blocking Reagent,
BioGenex, 1 h at 4 °C), and incubated with rabbit anti-Lysozyme EC
3.2.1.17 (A0099, Dako, Glostrup, Denmark; 1:200, overnight at 4 °C)
and rat anti-E-cadherin DECMA-1 (U3254, Sigma; 1:100, overnight at
4 °C) antibodies. Binding of the primary antibodies was detected using
goat-anti-rabbit Alexa Flour 546 (A11071, Moleculare Probes) and
goat-anti-rat Alexa Flour 488 (A11006, Molecular Probes) antibodies,
respectively. Nuclei were stained with DAPI (Sigma). Images were
captured using LSM 700 (Zeiss).
256
Developmental Biology 433 (2018) 254–261
3. Results
3.1. The branched growth pattern
Organoids cultured in ENR typically grow in a branched pattern
with a well-defined inner lumen (Fig. 1C). Applying the above
described model, simulated organoids can reproduce this pattern
(Fig. 1D, PSET I). In the growing organoids, the cell type fractions
are self-organized. They change over time, until they reach a dynamic
equilibrium after about 10 days. At this stage simulated organoids,
even slowly growing, contain already several hundred cells (Fig. 1E).
Their sphericity, i.e. the ratio between the surface area of a sphere with
the same volume as the object analyzed and the surface area of the
object itself, reaches a plateau. (Fig. 1F).
3.2. Wnt- activity controls organoid cell composition and growth
pattern
In our intestinal crypt model, we assumed that an increase of the
Wnt-activity increases the probability of ISCs to specify into a secretory
cell lineage (Thalheim et al., 2016). This assumption is consistent with
the observations that: i) increased Wnt-activity increases PC-numbers,
ii) mutant cells with high Wnt-activity preferentially differentiate into
PCs (Farin et al., 2012), and that iii) an attenuation of Wnt-activity
rescues the PC-accumulation associated with Notch-blockade (Tian
et al., 2015). Developing our extended organoid model, we adhered to
and detailed this regulatory principle. We assume that ISC mainte-
nance involves Notch-signaling because it lowers the Wnt-signal after
activation to ISC specific levels (Tian et al., 2015). Increased Wnt-
activity in ISCs, e.g. achieved by increasing R-spondin, thus, requires
increased Notch-signaling in these cells. Fig. 2A illustrates this
principle. As we consider an additive Notch-activation by all PC- and
GC-neighbors, this increased Notch-signaling can be induced by an
increased number of such neighbors. The minimum number of them
enabling ISC maintenance at a defined Wnt-regulation scenario is given
by NN. Note that, while additional PC-neighbors increase the Notch-
signal, they do not increase the Wnt-signal in our model. We assume
that a single PC in direct contact with an ISC (Farin et al., 2016)
provides sufficient Wnt-3 for Wnt-receptor saturation on ISCs. Further
details can be found in Appendix D.
Fig. 2 B/C show simulated organoids growing under different levels
of Wnt-activation, characterized by different number NN. Low Wnt-
activation (NN = 1) leads to organoids with only a few elongated
branches. Thereby, ISCs and PCs are confined to the tip of these
protrusions. Higher Wnt-activation first leads to branched growth
pattern (NN = 2) and eventually to more or less spherical growth
pattern (NN = 3). Our simulated growth pattern are in good agreement
with in vitro findings by Farin et al. (2012) and our own in vitro data
on growth pattern under increasing R-spondin concentrations
(Fig. 2D). The in silico growth pattern which is most similar to 10%
R-spondin culture, the standard for our in vitro culture, is the pattern
referring to NN = 2. Here, the PC fraction is about 10% and the ISC
fraction about 25% (Fig. 2E) which is slightly above in vitro data (Farin
et al., 2012; Yin et al., 2014).
3.3. Elastic properties of the organoids affect their growth pattern
Next, we studied the impact of polymer elasticity on the organoid
growth for all considered Wnt-activation levels (NN = 1–3, Fig. 3).
Thereby, we noted, that: i) specification of PCs facilitates expansion
while at the same time ii) it reduces the number of proliferative cells.
Thus, in all simulations the speed of expansion reached is much below
the maximum value given for completely unrestricted growth (τ = 10 h,
equal to the growth time T, see: Appendix C). Fig. 3A shows results for
a high elastic module of the NET polymers: KL > KL,0 (PSET I). For this
setting, only the NET attached to PCs, can re-organize, because only the
T. Thalheim et al. Developmental Biology 433 (2018) 254–261

Fig. 2. Wnt- activity controls organoid cell composition and growth pattern. A) Scheme of the regulatory principles linking R-spondin concentrations, Wnt-activation levels and the
minimal number of PCs (NN) required for ISC maintenance. Higher R-spondin leads to a higher Wnt-signal. Accordingly, a higher number of PCs is required to increase Notch-activation
that lowers the Wnt-signal to ISC levels. Note that the Wnt-signal does not change with R-spondin in ECs and GCs. The distributions of exogenous Wnt-concentrations provided by 1–3
PCs are illustrated by bell-shaped curves. B) Snapshot of simulated organoids after 30 (NN = 1) and 15 (NN = 2,3) days of in silico culture (PSET I). C) Cross section through the
organoids shown in (B), viable cells only. Elongated protrusions are seen for NN = 1, while for NN = 3 organoids grow more or less spherical. The basolateral polymer network is shown in
grey. D) Fluorescence images of organoids after 6 days in culture at different R-spondin concentrations (merged images, red: Lysozyme, green: E-cadherin, blue: DAPI, scale bars:
50 µm). E) Size and composition of the organoids shown in (B-C).

polymers attached to them are sufficiently soft: KL,PC < KL,0 (see: re-organize. Under these conditions organoid expansion becomes
Section 2). Under these conditions higher R-spondin leads to faster strongly accelerated. A maximum speed is reached for NN = 2 (green
expansion, because more PCs are specified (see: Fig. 2E). Decreasing symbols in Fig. 3B). In this case, only a few PCs are specified and a high
KL below KL,0 for all polymers (Fig. 3B, PSET II), the entire NET can fraction of cells proliferates. The observed growth pattern, however, do

Fig. 3. Organoid growth behavior depends on biomechanics. A-B) Normalized number of cells within a growing organoid for high and low elastic modulus KL (PSET I and II). For high
KL (A), increasing NN accelerates growth. For low KL (B), maximum growth is observed for NN = 2. These organoids do not form branches (see insert, NN = 2, ca. 600 cells). C-D)
Increasing NN decreases the ratio between ECs and ISCs for both high (C) and low (D) KL. E) In vitro growth behavior of the normalized cross sectional area A of organoids. Without R-
spondin the organoids stop growth after about 4–5 days. At low R-spondin (1.25%) a tendency for accelerated growth is seen compared to higher R-spondin (10% and 25%). F)
Quantitative model adaptation to the experimental results for NN = 1–3. The open symbols are results for ηM = 8.6 Ns/m (PSET I, NN = 1) the solid ones for ηM = 0.86 Ns/m (PSET III,
NN=1–3). A-F): n = 10 organoids per PSET, mean ± SD. The black line in C and F indicates the average organoid growth curve obtained in Keysselt et al. (2016) for 10% R-spondin.

257
T. Thalheim et al.
no longer correspond to normal organoid growth under ENR, as no
branches are formed (see: insert) and the organoid lumen collapses
(Appendix E).
Fig. 3 C,D show the time-dependent ratio between ECs and ISCs for
high and low KL, respectively. In both cases the fraction of ECs in the
organoids increases with decreasing R-spondin (decreasing NN) being
consistent with progressive differentiation at lower Wnt-activity (Farin
et al., 2016). Assuming low KL (KL < KL,0, for all polymers, PSETII), the
ratio decreases for NN≥ 2 (stronger niche expansion), but increases for
NN = 1 (stronger EC expansion, no contact inhibition) compared to the
case of high KL.
3.4. Model adaptation to in vitro data
We analyzed the growth behavior of the cross sectional area A of
organoids under different R-spondin concentrations (Fig. 3E). Without
R-spondin the organoids stop growing after 4–5 days in agreement
with former reports (Farin et al., 2012). Organoids grown at 1.2% R-
spondin grow very fast and frequently burst at growth times larger than
5 days, having accumulated a large amount of cell debris in their
lumen. This is indicative of a high fraction of ECs that finished the life
cycle. Organoids at 10% or 25% R-spondin grow slightly slower than
those at 1.2% and did not burst. For a direct comparison between in
vitro and in silico growth, we quantified the size of the in silico growing
organoids with the same method used for in vitro organoids (Fig. 3F),
starting at day 2. We achieved quantitative agreement for fast NET-
reorganization, i.e. assuming a lower polymer friction constant γM.
(PSET III) compared to PSET I. Thereby, simulated growth curves for
NN = 1 corresponds to those observed in vitro curves for 0% R-
spondin. Simulated growth curves for NN = 2 and 3 are very similar and
correspond to in vitro growth at 10% and 25% R-spondin. There is no
model scenario corresponding to the fast growth at 1.25% R-spondin
although the growth pattern under this condition is similar to simula-
tion results for NN = 0 (see: Fig. 2D). This difference might originate in
shorter cell cycle times of ECs progenitors compared to ISCs (Barker
et al., 2008) not yet considered in the model.
In summary, if Wnt is provided by PCs only, our model reproduces
growth pattern, speed, and composition as observed for intestinal
organoids under different R-spondin concentrations.
3.5. The cyst-like growth pattern
As already mentioned, Wnt-3 secreted by PCs is essential for
organoid growth. Exogenous Wnt-3a can rescue growth of organoids
derived from Wnt-3 knock-out mice (Farin et al., 2012). However, if
added to standard ENR organoid culture, it induces cyst-like growth
(Fig. 4A). Thus, exogenous Wnt-3a appears to suppress organoid
budding. Notably, organoids grown in ENR culture medium supple-
mented with Wnt-3a spontaneous revert to normal growth after some
time, i.e. start budding. We here ask, how organoid budding is
(temporarily) suppressed by Wnt-3a.
Wnt-3a induces cytoskeleton reorganization in epithelia
(Shibamoto et al., 1998). This suggests that formation of a cyst-like
organoid is associated with altered biomechanical properties of the
intestinal epithelium. Such changes are known to occur during forma-
tion of adherence junctions associated with terminal differentiation
(Chartier et al., 2011; Harris et al., 2014). In our crypt model, terminal
differentiation is induced if the Wnt-activity falls below a threshold.
Here, we assume that under sufficient exogenous Wnt-3a this state is
not reached and associated changes in biomechanics, among them
stiffening of the polymers, are missing. The scheme given in Fig. 4B
illustrates the assumption. Actually, we set in cyst-like organoids the
elastic module KL of all NET polymers to values below KL,0.
Accordingly, the entire NET can be re-organized bearing the effects
discussed above (see: Fig. 3B,D). In addition, we increased the surface
bending modulus KC to enforce formation of spherical lumens. More
258
Developmental Biology 433 (2018) 254–261
details are given in the Appendix E.
Regardless of these assumptions, our model does not allow to
simulate the experimentally observed flattening of individual cells
(Fatehullah et al., 2013). We mimic the resulting organoid shape by
a formal image transformation that increases the radius of the
simulated organoid from (R) to (aR2/4 + R). Here, the parameter a
is the ratio between the osmotic pressure inflating the organoid and the
effective elastic module of the organoid Eeff (Keysselt et al., 2016). Eeff
describes the effective elastic response of a quiescent and relaxed
organoid. In our model, it depends on the properties of the individual
polymers of the NET, their linkage and on all other cell-cell interac-
tions, including those related to the apical net. For low pressure or a
high effective elastic module, the parameter a is small and the
respective change of the radius can be neglected.
Fig. 4C shows an in silico growing organoid and its transformed
images in a cyst-like growth scenario (NN = 2, PSET IV). The nearly
spherical organoid expands faster than branched organoids but is still
composed of all 4 cell types (Fig. 4D). It shows a higher PCs fraction
than branched organoids of the same size in agreement with experi-
mental findings (Yin et al., 2014). Increasing exogenous Wnt above
Wnt3a(T) (Fig. 4B), this change of the lineage specification would
amplify and eventually a pure ISC-PC-organoid would remain. PSET IV
refers exogenous Wnt slightly below Wnt3a(T).
In summary, cyst-like growth observed under exogenous Wnt-3a can
be described assuming changes of the elastic properties of the epithelium
being associated with hindered terminal differentiation. Similar results
are expected in case of intrinsic Wnt-activation, e.g. following mutations
in the adenomatous polyposis coli (APC) gene (Barker et al., 2008).
3.6. Spontaneous transitions of organoid growth
Recently, we demonstrated that the cyst-like growth pattern of
organoids derived from mismatch-repair deficient mice enrich with
mice age and time in culture (Keysselt et al., 2016). We proposed a
model of progressive tissue transformation and suggested that the
changes from branched towards cyst-like growth of the organoids are
associated with changes in their elasticity as well as differentiation. The
proposed changes in cell behavior are illustrated in Fig. 5A. Whether
the changed behavior is associated with mutation or epi-mutations
remains an open question.
In vitro, we always observed either ‘global’ transitions of the entire
organoid from cyst-like to branched growth (Fig. 5B) or ‘local’
transitions to cyst-like growth, where only local buds grow fast, while
the other parts of the organoid do not (Fig. 5C). A possible reason for a
global transition from cyst-like to branched is a global decrease of e.g.
exogenous Wnt-3a (if provided) or R-spondin, due to ongoing con-
sumption (Schuijers et al., 2015). In contrast, local transitions from
branched to cyst-like growth are more likely to be associated with cell
intrinsic changes. We simulated such a scenario assuming an intrinsic
activation of the Wnt-pathway in 10% of the ISCs and inheritance of
this state to the entire progeny. We call these cells cyst-like cells.
Related simulation results are shown in Fig. 5D; local buds can form
that increase much faster in size than the rest of the organoid.
As the progeny of cyst-like cells, i.e. the clones with intrinsic
activated Wnt, will expand faster, they eventually will overgrow the
entire organoid. In simulations of competing normal (PSET III) and
cyst-like (PSET V) cells we explicitly followed the expansion of the cyst-
like clones. Fig. 5E provides simulation results on the fraction of
different populations of cyst-like cells in the organoids. The fraction of
cyst-like ISCs approximately doubles within 4 days of culture. Note,
that in a few simulations (< 10%) the cyst-like clones vanished.
In summary, our model predicts that cyst-like growth assigns the
cells a competitive growth advantage leading to accumulation of ISCs
with intrinsically activated Wnt in the culture. This agrees well with the
experimental finding of an accumulation of cyst-like growing organoids
in long term culture (Keysselt et al., 2016).
T. Thalheim et al. Developmental Biology 433 (2018) 254–261

Fig. 4. The cyst-like growth pattern. A) Phase-contrast images of organoids cultured under 25% Wnt-3a (day 6 of culture, scale bar: 200 µm). B) Scheme explaining the regulatory
changes associated with exogenous Wnt-3a. Decreasing Wnt-signaling, the elastic properties of the organoid change from soft towards stiff between Wnt-state I and II (insert) and
eventually the cells differentiate terminally (state III). Under sufficient exogenous Wnt-3a, state II is not reached and the cells remain soft. Above, a threshold concentration Wnt3a(T)
also lineage specification is hindered, cells reach low Wnt-states less frequently and the fraction of SCs and PCs increases. C) Snapshots of slices of a simulated cyst-like growing organoid
(viable cells, PSET IV). Area extension is mimicked by an image transformation (orange arrows). The actually simulated organoid is shown below. Color coding of the cell types as in
Figs. 1 and 2. D) Cell numbers of the growing organoid in (C).

Fig. 5. Transitions between cyst-like and branched growth pattern. A) Scheme illustrating the consequences of a transition into an intrinsically activated Wnt-state (pink line). As under
exogenous Wnt the cells do not terminally differentiate and the organoid remains in a soft state. B) Phase contrast images showing an organoid during a global transition from cyst-like
into branched growth. The organoid first shrinks in size and only afterwards starts budding again. C) Local induction of cyst-like growth. B,C): Δt = 16 h, scale bars: 200 µm. D) Two
examples of simulated organoids 2 and 4 days after transition of about 10% of the ISCs into an intrinsic activated Wnt-state (cyst-like ISCs: orange, PCs: olive, GCs: beige, ECs: light
blue, PSET V, reference: PSET III). E) Simulated time-dependent fraction of cyst-like cells within such organoids. Shown are averages over 40 organoids (mean ± SD).

259
T. Thalheim et al.
4. Discussion
Understanding the regenerative capabilities of organs requires a
multi-scale approach including e.g. transcriptional regulation that
controls stem cell maintenance and differentiation, 3D tissue self-
organization based on stem cell behavior, organ function relying on
tissue interaction and immune function and possibly even more.
Individual cell-based approaches have been shown to be capable to
bridge many of these scales (Wolkenhauer et al., 2014). Although many
aspects of the complex interactions involved are still not understood,
such models allow to develop hypotheses about the impact of selected
processes on the regeneration of the whole system (Drasdo et al., 2014;
Buske et al., 2011). Applying such models, we focused on questions
regarding stem cell self-organization as intrinsic versus extrinsic
regulation (Buske et al., 2012; Thalheim et al., 2016) and stem cell
heterogeneity (Krinner et al., 2010; Hamidouche et al., 2017) in the
last years. In the present study, we focus on the link between stem cell
lineage specification, growth pattern and clonal expansion.
In vitro expansion of stem cells for therapeutic use has come into
focus more than 20 years ago. However, the risks associated with their
rigorous expansion to achieve sufficient cell numbers have been
addressed only rarely. In vitro expansion of stem cells is accompanied
by stem cell ageing, and sometimes even by stem cell transformation.
Studying long term expansion of ISCs, we have recently quantified
transitions between different growth phenotypes and have shown that
the percentage of organoids with cyst-like growth becomes enhanced
with increasing culture time (Keysselt et al., 2016). We suggested that
this is a hallmark of ISC adaptation to in vitro conditions. The most
probable molecular origin of this adaptation is stabilization of high
Wnt-signaling. Recent findings by Nalapareddy et al. (2017) show that
in vivo Wnt-activity is rather decreased during ageing. Thus, if actually
Wnt-associated, the observed adaptation phenomenon has to be
considered as a pro-tumorigenic effect.
Simulation of such in vitro adaptation requires to describe the
biomechanics of the organoids. Sophisticated modeling techniques in
the field still suffer from missing experimental data (Okuda et al.,
2015). Regarding the intestine, experimentalists mainly focused on
questions on how lineage specification and growth are effected by
elasticity properties of the environment (Gjorevski et al., 2016). We
here provide an individual cell-based model of intestinal organoids
which suggests that expansion of intrinsically Wnt-activated ISCs is
accelerated due to changes in biomechanics. The assumption that
biomechanical properties are thereby inherited to different lineages is
in agreement with the observation that expanding cyst-like organoids
show a heterogeneous cell composition (Fatehullah et al., 2013). The
prediction made by the model, that the clones of these ISCs will
overgrow the entire culture, can be tested experimentally applying
clonal labeling techniques.
As every model, our individual cell-based model has several
limitations. For example, the model does not allow to account for
changes of the inflating pressure and for changes of the cell shape being
associated with transition between branched and cyst-like growth.
Explicit simulation of an inflating pressure might be essential in order
to stabilize the shape of large cyst-like organoids. Explicit simulation of
non-spherical cell shape would enable a better mechanical character-
ization of the organoids. This would require model types as suggested
by Odenthal et al. (2013) thereby touching the current limits of
computational effort. Performing simulation series with the model
introduced here, such that statistical analysis can be applied, already
takes enormous CPU time. Thus, further development will require
more efficient calculations in particular of the interactions between the
cells and the polymer network (NET).
The simulation results presented here rise the question whether
quantitative 1:1 modeling of organoids is possible. We compared
growth curves of in vitro and in silico growing organoids and
demonstrated that in vitro growth curves can be fitted with our model.
260
Developmental Biology 433 (2018) 254–261
However, these fits are not unique. Starting with the reference
parameter set, a fit of in vitro growth curves can be achieved e.g. by
adapting friction coefficients and/or cell cycle properties. More reliable
fits would require detailed information about cell type specific elasticity
and viscosity. Alternatively, more data on the time-dependent compo-
sition and shape of growing organoids could help to narrow down the
physiological parameter ranges of the model to avoid overfitting.
Accordingly, the actual potential of the model to predict clonal
development in intestinal organoids is still limited and requires
additional experiments both in vitro and in silico.
5. Summary
Our model suggests that growth pattern of intestinal organoids are
controlled by PCs specification. These cells provide the signals required
for ISC maintenance and in parallel control the shape of the ISC niche.
ISC maintenance under increasing Wnt-activity requires increasing
Notch-signaling. The related changes in PC-specification change the
cell composition of the organoid and thereby its growth pattern.
Exogenous Wnt or intrinsically activated Wnt can suppress terminal
differentiation in the organoids resulting in changed biomechanics and
cyst-like growth pattern. As this growth pattern assigns cells a growth
advantage, clones of ISCs with intrinsically activated Wnt will expand
in the culture.
Acknowledgements
The computations were performed using a Bull cluster at the Center
for Information Services and High Performance Computing (ZIH) at
the TU Dresden.
Funding
This study was supported by the Bundesministerium für Bildung
und Forschung (grant: INDRA, grant number: BMBF
031A312).Author's contribution
TT: performed all simulations, analyzed and interpreted related
data, contributed to writing. MQ: performed all experiments, contrib-
uted to writing. GA and MH: supported experimental design and data
interpretation, contributed to writing. CK: supported mouse experi-
ments, UDB: supported data interpretation. ML: contributed to data
interpretation, financial support, JG: supervised the study, analyzed
and interpreted data and wrote the main part of the manuscript.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the
online version at doi:10.1016/j.ydbio.2017.10.013.
References
Barker, N., van de Wetering, M., Clevers, H., 2008. The intestinal stem cell. Genes Dev.
22 (14), 1856–1864.
Buske, P., Przybilla, J., Loeffler, M., Sachs, N., Sato, T., Clevers, H., Galle, J., 2012. On
the biomechanics of stem cell niche formation in the gut–modelling growing
organoids. FEBS J. 279 (18), 3475–3487.
Buske, P., Galle, J., Barker, N., Aust, G., Clevers, H., Loeffler, M., 2011. A comprehensive
model of the spatio-temporal stem cell and tissue organisation in the intestinal crypt.
PLoS Comput. Biol. 7 (1), e1001045.
Booth, C., Potten, C.S., 2000. Gut instincts: thoughts on intestinal epithelial stem cells. J.
Clin. Investig. 105 (11), 1493–1499.
Chartier, N.T., Lainé, M.G., Ducarouge, B., Oddou, C., Bonaz, B., Albiges-Rizo, C.,
Jacquier-Sarlin, M.R., 2011. Enterocytic differentiation is modulated by lipid rafts-
dependent assembly of adherens junctions. Exp. Cell Res. 317 (10), 1422–1436.
Donohoe, C.L., Reynolds, J.V., 2010. Short bowel syndrome. Surgeon 8 (5), 270–279.
Drasdo, D., Bode, J., Dahmen, U., Dirsch, O., Dooley, S., Gebhardt, R., Ghallab, A.,
Godoy, P., Häussinger, D., Hammad, S., Hoehme, S., Holzhütter, H.G., Klingmüller,
U., Kuepfer, L., Timmer, J., Zerial, M., Hengstler, J.G., 2014. The virtual liver: state
of the art and future perspectives. Arch. Toxicol. 88 (12), 2071–2075.
Farin, H.F., Van Es, J.H., Clevers, H., 2012. Redundant sources of Wnt regulate intestinal
T. Thalheim et al.
stem cells and promote formation of Paneth cells. Gastroenterology 143 (6),
1518–1529, (e7).
Farin, H.F., Jordens, I., Mosa, M.H., Basak, O., Korving, J., Tauriello, D.V., de Punder,
K., Angers, S., Peters, P.J., Maurice, M.M., Clevers, H., 2016. Visualization of a short-
range Wnt gradient in the intestinal stem-cell niche. Nature 530 (7590), 340–343.
Fatehullah, A., Appleton, P.L., Näthke, I.S., 2013. Cell and tissue polarity in the intestinal
tract during tumourigenesis: cells still know the right way up, but tissue organization
is lost. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 368 (1629), 20130014.
Fre, S., Huyghe, M., Mourikis, P., Robine, S., Louvard, D., Artavanis-Tsakonas, S., 2005.
Notch signals control the fate of immature progenitor cells in the intestine. Nature
435 (7044), 964–968.
Gjorevski, N., Sachs, N., Manfrin, A., Giger, S., Bragina, M.E., Ordóñez-Morán, P.,
Clevers, H., Lutolf, M.P., 2016. Designer matrices for intestinal stem cell and
organoid culture. Nature 539 (7630), 560–564.
Hamidouche, Z., Rother, K., Przybilla, J., Krinner, A., Clay, D., Hopp, L., Fabian, C.,
Stolzing, A., Binder, H., Charbord, P., Galle, J., 2017. Bistable epigenetic states
explain age-dependent decline in mesenchymal stem cell heterogeneity. Stem Cells
35 (3), 694–704.
Harris, A.R., Daeden, A., Charras, G.T., 2014. Formation of adherens junctions leads to
the emergence of a tissue-level tension in epithelial monolayers. J. Cell Sci. 127 (Pt
11), 2507–2517.
Keysselt, K., Kreutzmann, T., Rother, K., Kerner, C., Krohn, K., Przybilla, J., Buske, P.,
Löffler-Wirth, H., Loeffler, M., Galle, J., Aust, G., 2016. Different in vivo and in vitro
transformation of intestinal stem cells in mismatch repair deficiency. Oncogene 36
(19), 2750–2761.
Krinner, A., Hoffmann, M., Loeffler, M., Drasdo, D., Galle, J., 2010. Individual fates of
mesenchymal stem cells in vitro. BMC Syst. Biol. 4, 73.
Langlands, A.J., Almet, A.A., Appleton, P.L., Newton, I.P., Osborne, J.M., Näthke, I.S.,
2016. Paneth cell-rich regions separated by a cluster of Lgr5+ cells initiate crypt
fission in the intestinal stem cell niche. PLoS Biol. 14 (6), e1002491.
Markel, T.A., Crisostomo, P.R., Lahm, T., Novotny, N.M., Rescorla, F.J., Tector, J.,
Meldrum, D.R., 2008. Stem cells as a potential future treatment of pediatric
intestinal disorders. J. Pediatr. Surg. 43 (11), 1953–1963.
Martin, K., Kirkwood, T.B., Potten, C.S., 1998. Age changes in stem cells of murine small
intestinal crypts. Exp. Cell Res. 241 (2), 316–323.
Merker, S.R., Weitz, J., Stange, D.E., 2016. Gastrointestinal organoids: how they gut it
out. Dev. Biol. 420 (2), 239–250.
Nalapareddy, K., Nattamai, K.J., Kumar, R.S., Karns, R., Wikenheiser-Brokamp, K.A.,
Sampson, L.L., Mahe, M.M., Sundaram, N., Yacyshyn, M.B., Yacyshyn, B., Helmrath,
M.A., Zheng, Y., Geiger, H., 2017. Canonical Wnt signaling ameliorates aging of
intestinal stem cells. Cell Rep. 18 (11), 2608–2621.
Odenthal, T., Smeets, B., Van Liedekerke, P., Tijskens, E., Van Oosterwyck, H., Ramon,
H., 2013. Analysis of initial cell spreading using mechanistic contact formulations for
a deformable cell model. PLoS Comput. Biol. 9 (10), e1003267.
Okuda, S., Inoue, Y., Adachi, T., 2015. Three-dimensional vertex model for simulating
multicellular morphogenesis. Biophys. Physicobiol. 12, 13–20.
261
Developmental Biology 433 (2018) 254–261
Pin, C., Parker, A., Gunning, A.P., Ohta, Y., Johnson, I.T., Carding, S.R., Sato, T., 2015.
An individual based computational model of intestinal crypt fission and its
application to predicting unrestrictive growth of the intestinal epithelium. Integr.
Biol. 7 (2), 213–228.
Sato, T., Vries, R.G., Snippert, H.J., van de Wetering, M., Barker, N., Stange, D.E., van
Es, J.H., Abo, A., Kujala, P., Peters, P.J., Clevers, H., 2009. Single Lgr5 stem cells
build crypt-villus structures in vitro without a mesenchymal niche. Nature 459
(7244), 262–265.
Sato, T., van Es, J.H., Snippert, H.J., Stange, D.E., Vries, R.G., van den Born, M., Barker,
N., Shroyer, N.F., van de Wetering, M., Clevers, H., 2011. Paneth cells constitute the
niche for Lgr5 stem cells in intestinal crypts. Nature 469 (7330), 415–418.
Shibamoto, S., Higano, K., Takada, R., Ito, F., Takeichi, M., Takada, S., 1998.
Cytoskeletal reorganization by soluble Wnt-3a protein signalling. Genes Cells 3 (10),
659–670.
Stamataki, D., Holder, M., Hodgetts, C., Jeffery, R., Nye, E., Spencer-Dene, B., Winton,
D.J., Lewis, J., 2011. Delta1 expression, cell cycle exit, and commitment to a specific
secretory fate coincide within a few hours in the mouse intestinal stem cell system.
PLoS One 6 (9), e24484.
Schuijers, J., Junker, J.P., Mokry, M., Hatzis, P., Koo, B.K., Sasselli, V., van der Flier,
L.G., Cuppen, E., van Oudenaarden, A., Clevers, H., 2015. Ascl2 acts as an R-
spondin/Wnt-responsive switch to control stemness in intestinal crypts. Cell Stem
Cell 16 (2), 158–170.
Thalheim, T., Buske, P., Przybilla, J., Rother, K., Loeffler, M., Galle, J., 2016. Stem cell
competition in the gut: insights from multi-scale computational modelling. J. R. Soc.
Interface 13 (121), (20160218).
Tian, H., Biehs, B., Chiu, C., Siebel, C.W., Wu, Y., Costa, M., de Sauvage, F.J., Klein, O.D.,
2015. Opposing activities of Notch and Wnt signaling regulate intestinal stem cells
and gut homeostasis. Cell Rep. 11 (1), 33–42.
van de Wetering, M., Sancho, E., Verweij, C., de Lau, W., Oving, I., Hurlstone, A., van der
Horn, K., Batlle, E., Coudreuse, D., Haramis, A.P., Tjon-Pon-Fong, M., Moerer, P.,
van den Born, M., Soete, G., Pals, S., Eilers, M., Medema, R., Clevers, H., 2002. The
beta-catenin/TCF-4 complex imposes a crypt progenitor phenotype on colorectal
cancer cells. Cell 111 (2), 241–250.
van Es, J.H., Sato, T., van de Wetering, M., Lyubimova, A., Nee, A.N., Gregorieff, A.,
Sasaki, N., Zeinstra, L., van den Born, M., Korving, J., Martens, A.C., Barker, N., van
Oudenaarden, A., Clevers, H., 2012. Dll1+ secretory progenitor cells revert to stem
cells upon crypt damage. Nat. Cell Biol. 14 (10), 1099–1104.
Wolkenhauer, O., Auffray, C., Brass, O., Clairambault, J., Deutsch, A., Drasdo, D.,
Gervasio, F., Preziosi, L., Maini, P., Marciniak-Czochra, A., Kossow, C., Kuepfer, L.,
Rateitschak, K., Ramis-Conde, I., Ribba, B., Schuppert, A., Smallwood, R.,
Stamatakos, G., Winter, F., Byrne, H., 2014. Enabling multiscale modeling in
systems medicine. Genome Med. 6 (3), 21.
Yin, X., Farin, H.F., van Es, J.H., Clevers, H., Langer, R., Karp, J.M., 2014. Niche-
independent high-purity cultures of Lgr5+ intestinal stem cells and their progeny.
Nat. Methods 11 (1), 106–112.