1
2
3
4
5
6
7
Excample of an output of transcriptional profiling study using Illumina 
sequencing performed in our lab. Shown is just a tiny fragment of the complete 
list, copmprising about 7K genes revealing differential expression in the studied 
mutant. 
8
9
10
11
Excample of an output of transcriptional profiling study using Illumina 
sequencing performed in our lab. Shown is just a tiny fragment of the complete 
list, copmprising about 7K genes revealing differential expression in the studied 
mutant. 
12
13
One of such recent and very useful tools is Gorilla software, freely available at 
http://cbl‐gorilla.cs.technion.ac.il/.
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
According to experimental evidence for the system under study, the hormone IAA, the
peptide TDIF, and the microRNA MIR165/6 are able to move among the cells. In the case
of TDIF and MIR165/6, the mobility is defined as diffusion and is given by the following
equation:
g(t+1)T[i]= H(g(t)[i]+ D (g(t)[i+1]+g(t)[i‐1] – N(g(t)[i]))‐b) (2),
where g(t)T[i] is the total amount of TDIF or MIR165 in cell (i). D is a parameter that
determines the proportion of g that can move from any cell to neighboring ones and is
correlated to the diffusion rate of g. b is a constant corresponding to a degradation term.
H is a step function that converts the continuous values of g into a discrete variable that
may attain values of 0, 1 or 2. N stands for the number of neighbors in each cell.
Boundary conditions are zero‐flux. In the case of IAA, the mobility is defined as active
transport dependent on the radial localization of the PIN efflux transporters and is
defined by the equation:
iaa(t+1)T[i]=Hiaa(iaa(t)[i]+Diaa(pin(t)[i+1])(iaa(t)[i+1])+Diaa(pin(t)[i‐1])(iaa(t)[i‐
1])−N(Diaa)(pin(t)[i])(iaa(t)[i])−biaa) (3),
where Diaa is a parameter that determines the proportion of IAA that can be
transported among cells. The transport depends on the presence of IAA and PIN in the
cells and biaa corresponds to a degradation term. As in equation 2, H is a step function
that converts the continuous values to discrete ones and N stands for the number of
neighbors in each cell. Boundary conditions for IAA motion are also zero‐flux.
29
The proposed model considers data that we identified and evaluated through an
extensive search (up to January 2012). It takes into account molecular
interactions, hormonal and expression patterns, and cell‐to‐cell communication
processes that have been reported to affect vascular patterning in the bundles of
Arabidopsis. The model components and interactions are graphically presented in
the figure above. In the network model, nodes stand for molecular elements
regulating one another’s activities. Most of the nodes can take only 1 or 0 values
(light gray nodes in the figure), corresponding to “present” or “not present,”
respectively. Since the formation of gradients of hormones and diffusible
elements may have important consequences in pattern formation, mobile
elements TDIF and MIR, as well as members of the CK and IAA signaling systems,
can take 0, 1 or 2 values (dark gray nodes in the figure above) Benitez and
Hejatko, submitted.
30
31
In comparison to the model shown on slide 21, the final version of the model 
contains the predicted interactions (dashed lines).
32
33
34
The initial conditions specify the initial state of some of the network elements (figure above) and are the following :
I) In the procambial position (central compartment), CK is initially available and there is an initial and sustained IAA input or self‐upregulation. This 
condition is supported by several lines of evidence. Also HB8, a marker of early vascular development that has been found in preprocambial cells, is 
assumed to be initially present at this position. These conditions are not fixed, however. After the initial configuration, all the members of the CK and 
IAA signaling pathways, as well as HB8, can change their states according to the logical rules. 
II) In the xylem and phloem positions, it is assumed that no element is initially active except for the CK signaling pathway and TDIF, both in the phloem 
position.The level of expression for a given node is represented by a discrete variable g and its value at a time t+1 depends on the state of other 
components of the network (g1, g2, ..., gN) at a previous time unit. The state of every gene g therefore changes according to: 
gn(t+1)=Fn(gn1(t),gn2(t),..., gnk(t)) (1).
In this equation, gn1, gn2,…, gnk are the regulators of gene gn and Fn is a discrete function known as a logical rule (logical rules are grounded in
available experimental data, for example see slide 20). Given the logical rules, it is possible to follow the dynamics of the network for any given initial
configuration of the nodes expression state. One of the most important traits of dynamic models is the existence of steady states in which the entire
network enters into a selfsustained configuration of the nodes state. It is thought that in developmental systems such self‐sustained states correspond
to particular cell types.
According to experimental evidence for the system under study, the hormone IAA, the peptide TDIF, and the microRNA MIR165/6 are able to move
among the cells. In the case of TDIF and MIR165/6, the mobility is defined as diffusion and is given by the following equation:
g(t+1)T[i]= H(g(t)[i]+ D (g(t)[i+1]+g(t)[i‐1] – N(g(t)[i]))‐b) (2),
where g(t)T[i] is the total amount of TDIF or MIR165 in cell (i). D is a parameter that determines the proportion of g that can move from any cell to
neighboring ones and is correlated to the diffusion rate of g. b is a constant corresponding to a degradation term. H is a step function that converts the
continuous values of g into a discrete variable that may attain values of 0, 1 or 2. N stands for the number of neighbors in each cell. Boundary
conditions are zero‐flux. In the case of IAA, the mobility is defined as active transport dependent on the radial localization of the PIN efflux
transporters and is defined by the equation:
iaa(t+1)T[i]=Hiaa(iaa(t)[i]+Diaa(pin(t)[i+1])(iaa(t)[i+1])+Diaa(pin(t)[i‐1])(iaa(t)[i‐1])−N(Diaa)(pin(t)[i])(iaa(t)[i])−biaa) (3),
where Diaa is a parameter that determines the proportion of IAA that can be transported among cells. The transport depends on the presence of IAA
and PIN in the cells and biaa corresponds to a degradation term. As in equation 2, H is a step function that converts the continuous values to discrete
ones and N stands for the number of neighbors in each cell. Boundary conditions for IAA motion are also zero‐flux.
Using the logical rules, equations 1–3, and a broad range of parameter values (not shown here), it is possible fully to reproduce the results and 
analyses reported in the following sections (see the figure above for the simulation time course).
35
Another representation of the distinct expression profiles in the individual 
vascular bundle compartments (phloem, procambium and xylem).
36
37
38
39
40
Hypertrophic scars (HS) area fibroproliferative disorder of the skin, which causes aesthetic and 
functional impairment. However, the molecular pathogenesis of this disease remains largely 
unknown and currently no efficient treatment exists. MicroRNAs (miRNAs) are involved in a 
variety of pathophysiological processes, however the role of miRNAs in HS development remains 
unclear. To investigate the miRNA expression signature of HS, microarray analysis was performed 
and 152 miRNAs were observed to be differentially expressed in HS tissue compared with normal 
skin tissues. Of the miRNAs identified, miRNA‐21 (miR‐21) was significantly increased in HS 
tissues and hypertrophic scar fibroblasts (HSFBs) as determined by reverse 
transcription‐quantitative polymerase chain reaction analysis. It was also observed that, when 
miR‐21 in HSFBs was blocked through use of an antagomir, the phenotype of fibrotic fibroblasts 
in vitro was reversed, as demonstrated by growth inhibition, induction of apoptosis and 
suppressed expression of fibrosis‐associated genes collagen type I α 1 chain (COL1A1), COL1A2 
and fibronectin. Furthermore, miR‐21 antagomir administration significantly reduced the 
severity of HS formation and decreased collagen deposition in a rabbit ear HS model. The total 
scar area and scar elevation index were calculated and were demonstrated to be significantly 
decreased in the treatment group compared with control rabbits. These results indicated that 
the miR‐21 antagomir has a therapeutic effect on HS and suggests that targeting miRNAs may be 
a successful and novel therapeutic strategy in the treatment of fibrotic diseases that are difficult 
to treat with existing methods.
miRNA expression signature profiling in hypertrophic scars (HS). (A) Volcano plot presenting 
differentially expressed miRNAs between HS and paired (non‐scar, obtained from donor sites 
during scar resection) NS tissue. miRNA microarray expression profiling from three paired HS and 
NS tissues was performed. Differentially expressed miRNAs were identified by fold change and a 
P‐value calculated using Student's t‐test. The threshold set to identify up and downregulated 
genes was a fold change ≥2 and P<0.05. Red dots indicate points‐of‐interest that exhibit 
large‐magnitude fold‐changes (x‐axis; log2 of the fold change) and high statistical significance 
(y‐axis; ‐log10 of the P‐value). (B) Hierarchical clustering showing differentially expressed miRNAs 
from HS samples compared with paired NS tissues. Each row represents one miRNA and each 
column represents one tissue sample. The relative miRNA expression is depicted according to 
the color scale. Red indicates upregulation and green indicates downregulation. N1‐3 represents 
NS tissue samples, whereas H1‐3 represents HS tissue samples. The differentially expressed 
miRNAs were clearly separated into clusters. miRNA, microRNA; hsa‐miR, human microRNA; HS, 
hypertrophic scar; NS, normal skin.
41
In the root, several functional and anatomical units could be recognized.
Along the longitudinal axis, the root meristem forms a distal root tip, including
stem cell niche, columella and lateral root cap, proximal meristem with a
population of rapidly dividing cells and elongation zone where cells leaving the
root meristem undergo rapid elongation and mature.
42
43
44
GENIST algorithm
The MATLAB source code for GENIST is publically available at 
https://github.com/madeluis/GENIST.
For the detailed description of the procedure, see de Luis Balaguer et al., 2017, SI 
(https://www.pnas.org/content/114/36/E7632/tab‐figures‐data) 
45
GENIST algorithm
The MATLAB source code for GENIST is publically available at 
https://github.com/madeluis/GENIST.
For the detailed description of the procedure, see de Luis Balaguer et al., 2017, SI 
(https://www.pnas.org/content/114/36/E7632/tab‐figures‐data) 
GENIST block diagram. GENIST is implemented in MATLAB, and is composed of 
two consecutive steps, clustering and GRN inference. Clustering is performed 
based on a spatial dataset. Each resulting cluster is
independently processed by the GRN inference step, based on a temporal 
dataset.
46
47
48
PAN subnetwork in the QC inferred with the 12 developmental time points of the Arabidopsis root. (A) Optimal configuration (combination of signs— activation or repression—
of the regulations that were inferred with undefined signs, which best fits the data in the simulations of the equations) of the subnetwork of PAN and its downstream targets. (B 
and C) Resulting expression values of PAN and its downstream targets, over time, after simulating the optimal configuration of the model. Simulations were run for 5 d and plots 
are shown until all factors reached steady states in the WT and pan mutant simulations. (B)Model simulated with the fitted equation parameters. (C)Model simulated with the 
PAN‐associated parameters set to zero to simulate a pan mutant situation. (D) Normalized expression values of PAN and its predicted downstream targets in Col‐0 wild type and 
in pan mutant. Statistically significant changes of expression between the mutant and the wild type, *q < 0.05.
In the WT simulation, all targets reached steady states by day 1 with subtle changes of expression during the transients (time length until expression values reach their steady 
states). On the contrary, the pan mutant simulation showed that EIN3 and WIP4 presented high expression values during the transients and reached steady states at later stages 
(days 3 and 4, respectively). These delayed responses and initial activations of EIN3 and WIP4 reflect the prediction that these genes are indirectly affected by PAN. Further, the 
dynamics of our simulations depict that BRAVO, NTT, and WIP4 are, in our equations, connected through feedback loops. During the transient phase of the mutant simulation, 
NTT and BRAVO show an exponential decay, which is consistent with the prediction that they activate each other in the absence of PAN. However, their steady states are not 
immediately reached since they are activated by WIP4 and EIN3. Conversely, WIP4, which is repressed by a decaying NTT, shows high levels of expression.
49
50
51