Abstract
The viscoelastic properties of tissues influence their morphology and
cellular behavior, yet little is known about changes in these
properties during brain malformations. Lissencephaly, a severe cortical
malformation caused by LIS1 mutations, results in a smooth cortex.
Here, we show that human-derived brain organoids with LIS1 mutation
exhibit increased stiffness compared to controls at multiple
developmental stages. This stiffening correlates with abnormal
extracellular matrix (ECM) expression and organization, as well as
elevated water content, measured by diffusion-weighted MRI. Short-term
MMP9 treatment reduces both stiffness and water diffusion levels to
control values. Additionally, a computational microstructure mechanical
model predicts mechanical changes based on ECM organization. These
findings suggest that LIS1 plays a critical role in ECM regulation
during brain development and that its mutation leads to significant
viscoelastic alterations.
Subject terms: Soft materials, Mechanisms of disease, Pluripotent stem
cells, Neurogenesis
__________________________________________________________________
Brain tissue mechanics influence development and disease. Here, the
authors show that LIS1 mutations increase stiffness in human brain
organoids due to ECM alterations and that targeted ECM modulation
restores mechanical properties.
Introduction
Tissue mechanics is crucial in shaping tissue growth, function, and
disease progression. However, this field is understudied in human brain
developmental diseases due to limited access to human brain tissues and
suboptimal animal models^[52]1–[53]6. It has been suggested that the
tissue composition, the dynamic cellular processes that occur in the
developing brain, and tissue mechanics play a role in shaping brain
structure^[54]7–[55]19. The organization, shape, and amount of
structural extracellular matrix (ECM) in the tissue provide tissues
with their mechanical forces, thereby affecting cellular behavior
during and after development^[56]20,[57]21. ECM has been proposed to be
a crucial element in the structural organization of the developing
brain and the formation of human brain folds^[58]22,[59]23.
Lissencephaly, characterized by the absence of cortical convolutions,
provides insights into human brain fold formation. LIS1 mutations,
prevalent in lissencephaly patients, impact the scaffold protein LIS1,
affecting cytoplasmic dynein, RNA interactions, splicing, and gene
transcription^[60]24–[61]26. Having established that LIS1 influences
the physical characteristics of embryonic stem cells in our earlier
research^[62]27, we sought to investigate the unknown aspect of how
these properties are affected at the tissue level. Studying this
disease in mouse models has been useful but limited since their cortex
naturally lacks convolutions, and human brain organoids, which usually
lack folds, have the potential to form them, as we have previously
demonstrated^[63]16. The mutant mice exhibited deficits in neuronal
migration and hippocampal pathology^[64]28–[65]36. Here, we used human
pluripotent stem cell-derived organoids to study how biomechanical
changes are involved in cortical malformation development. We found
that brain organoids mutated for LIS1 are stiffer than control
organoids and unraveled a substantial ECM disorganization phenotype in
the disease. Using an interdisciplinary approach, we applied data
obtained from rheological tests, MRI, and ECM composition and structure
characterization to develop a computational model. This model
successfully predicts mechanical changes associated with differential
ECM localization and integrity in the developing brain.
Results
Mutant LIS1 brain organoids are stiffer
To investigate whether the cortical structural abnormalities observed
in cases of lissencephalic pathologies are linked to mechanical
abnormalities, we performed rheological tests on cortical organoids
(corticOs). These organoids were generated from two types of cell
lines: control human embryonic stem cells (hESCs) and isogenic lines
with a LIS1 heterozygous mutation introduced using CRISPR/Cas9 genome
editing techniques^[66]16.
CorticOs were generated using a protocol for self-organizing cortical
tissue (Supplementary Fig. [67]1a). A series of characterizations on
different days indicated that the early ectoderm-like organoids were
expressing different neural progenitors on days 9 and 18; and by day 60
corticOs contained post-mitotic neurons and astrocytes (Supplementary
Fig. [68]1b–e). To enrich the basal radial glial progenitors
population, which is thought to play an important role in the
development of the human cortex^[69]37, we added hLIF from day 35.
Basal radial glial progenitors were detected in 96-day-old corticOs
(Supplementary Fig. [70]1f, g).
Changes in mechanics can alter the structure, development, and function
of cells that make up a tissue, such as the brain^[71]38. To determine
the mechanical effects of LIS1 mutations, we employed micropipette
aspiration (MPA) rheology and performed creep test measurements of
brain organoids from LIS1^+/^− and control organoids at multiple
developmental ages (days 9, 18, 35, and 70). The aspiration dynamics of
the organoids into the pipette under a constant negative pressure were
recorded and analyzed (Fig. [72]1a). All organoid measurements,
regardless of their developmental age, shared a characteristic response
to the applied load: organoids stretched elastically the moment suction
was applied, followed by a gradual aspiration into the pipette over
five to ten seconds, and approached a steady-state finite deformation
(Supplementary Fig. [73]2a).
Fig. 1. LIS1 mutation leads to the stiffening of brain organoids, which are
solid–like viscoelastic tissues.
[74]Fig. 1
[75]Open in a new tab
a Under constant suction pressure, organoids are continuously aspirated
into the pipette and gradually approach a steady-state deformation. The
organoids’ creep compliance is well-fitted by the standard linear solid
(SLS) viscoelastic model (a’). b–e Averaged creep compliance
measurements (symbols) are fitted by the SLS model (curves) at the
specified conditions. Symbols and error bars correspond to the mean and
standard error of the mean. b’–e’ SLS fits the instantaneous (k[0]) and
steady-state (k[st]) stiffness, and response time (τ) are plotted.
Symbols represent individual organoids recorded, while the bar graphs
and error bars correspond to the mean and standard deviation. The
analysis included the following number of organoids, each measured
separately and fitted independently: b Day-9: n[control] = 6,
n[10F] = 7, n[9G] = 7. c Day-18: n[control] = 4, n[10F] = 5. d Day-35:
n[control] = 8, n[10F] = 8, and e Day-70: n[control] = 7, n[10F] = 9.
Statistical significance is evaluated via one-way ANOVA test in b’ and
via two-tailed independent student’s t test in c’–e’. * To strengthen
our findings, we tested an additional LIS1^+/^− ESCs line annotated as
9G, generated in the same approach discussed in the methods. However,
throughout the manuscript, we consistently used the 10F cell line.
The mechanical behavior of organoids was analyzed using the standard
linear solid (SLS) model. In its Maxwell representation, it consists of
an elastic element (spring k[1]) that is connected in parallel to a
second elastic element (spring k[2]) positioned in series with a
viscous element (dashpot µ) (Fig. [76]1a’). We calculated the creep
compliance function, J(t), to measure time-dependent
deformability^[77]39, using parameters like the aspirated fraction
length, pipette radius, applied pressure, and a geometrical factor. The
organoid mechanics were quantitatively characterized by fitting the SLS
creep compliance function. We assessed the instantaneous stiffness
k[0], steady-state stiffness k[st], and the viscoelastic transition
time scale τ. This model showed high accuracy in representing organoid
behavior, as evidenced by the high R-square values in our fits (See
Methods for a full description of the model and its calculation).
CorticOs stiffness, as estimated by k[0] and k[st], ranged over
hundreds of pascals, indicating that it is as soft as cream cheese
(Fig. [78]1b–e)^[79]40. Notably, this range aligns with the lower
spectrum of brain tissue stiffness, which typically spans from 0.1 to 2
kilopascals^[80]41. With time, the corticOs stiffen, likely due to
continuous ECM deposition and/or fibrillation. We found that LIS1^+/^−
mutations increased the stiffness of corticOs as early as 9 days after
their aggregation and that they remain stiffer than controls up to the
latest tested time point, day 70. However, no significant difference in
the viscoelastic transition time scale τ is observed (Fig. [81]1b’–e’).
These findings indicate that, when subjected to a physiologically
relevant load across multicellular length scales, the corticOs exhibit
characteristics of a solid-dominant viscoelastic behavior that becomes
stiffer with progressing developmental stages. Overall, our data
suggest that changes to the biomechanics of LIS1^+/^− organoids appear
early in development and continue over a considerable period.
Cortical LIS1^+/^− organoids express abnormal ECM and more Lamin A
To delineate the molecular changes that are associated with the
stiffening of the LIS1^+/^− cortices, we analyzed the proteomic
signature of control and LIS1^+/^− corticOs. On day 35, we extracted
proteins from the LIS1^+/^− mutated corticOs, WIBR3 control, and a
PX335 control. The PX335 control line was created with the empty Cas9
nickase plasmid used in creating the original LIS1 mutant,
electroporated into the parental WIBR3 line to produce a second control
CRISPR corticOs. A total of 7842 proteins were identified and
quantified, of which 429 were differentially expressed (DE) in the
mutation, based on the threshold criteria of Log2Fold change ≥|0.5|, at
least 1 peptide per protein, and ANOVA p value < 0.05 in both PX335 and
WIBR3 control vs. LIS1^+/^− comparisons (Supplementary Data [82]1a). We
then conducted a Metascape analysis^[83]42 to examine the pathways
differing between the two control lines and the mutated corticOs, and
identified proteins associated with the ECM as most affected
(Supplementary Data [84]1b, Fig. [85]2a). In addition, previous studies
indicated that basal radial glial cells might be involved in cortical
gyrification (review^[86]43). Therefore, we have chosen to conduct an
additional analysis in 105-day-old corticOs, following the appearance
and expansion of that progenitor population (Supplementary Fig. [87]1f,
g). On day 105, there were 526 DE proteins between the LIS1^+/^− and
control corticOs (Fig. [88]2b, Supplementary Data [89]2a, b). The top
affected pathway of the DE proteins common identified by
[90]www.geneanalytics.com^[91]44 was the superpath “collagen-containing
extracellular matrix”, which was consistent even when using more
stringent sorting (Log2Fold change ≥|1|.
Fig. 2. Matrisome composition of corticO.
[92]Fig. 2
[93]Open in a new tab
a Metascape analysis on the proteomics data from 35-day-old corticOs. b
Heatmap of top DE matrisomal proteins in LIS1^+/^− and control
105-day-old cortical organoids. Cell color expresses normalized reads
following logarithmic transformation and Z-score normalization. c
Western blot analysis indicated the elevation of Lamin A/C and the
reduction in γH2AX in 105-day-old LIS1^+/^− corticOs. d Quantification
of western blot results (mean ± SD, two-tailed unpaired student’s t
test, α = 0.05, * means p value < 0.1 and ** means p value < 0.01). The
analysis included: a Day-35, for each genotype N [corticOs] = 4,
n[corticOs] = 10–12, b, c Day-105 proteomics and WB: N[corticOs] = 4;
n[corticOs] = 6–8.
In line with the observed stiffening of the corticOs in the MPA
procedure, we also observed increased expression of the LMNA protein in
the LIS1^+/^− corticOs (Fig. [94]2b). Lamins are intermediate filament
proteins of the nucleus that provide structural stability. Lamin A
expression is strongly correlated with tissue stiffness, whereas other
members of the nuclear lamina family, Lamin B, and Lamin C, are not
affected by the physical properties of the cells^[95]45. The roles of
LMNA are not strictly structural as it also affects chromatin
organization, gene regulation, cell differentiation, and signaling
pathways, including the Wnt/β-catenin pathway, TGFβ, and Notch^[96]46.
The 30% increase in LMNA protein levels indicated by the proteomics was
recapitulated by Western blot analysis, whereas the levels of Lamin B
were unchanged (Fig. [97]2c, d). It was further hypothesized that the
rise in Lamin A would reduce the double-strand breaks in the tissue due
to an increased nuclear protective shield. This was supported by the
Western blot quantification, which showed a significant reduction in
the double-strand breaks marker γ-H2AX in the LIS1^+/^− 105-day-old
corticOs. These changes in chromatin and lamins expression may
potentially affect tissue stiffness; however, in this work, we focused
on the ECM. Overall, the increased levels of Lamin A, together with the
changes observed in ECM-related proteins, suggest that a mutation in
LIS1 results in adverse biomechanical abnormalities to the brain
organoids.
In addition, RNA-seq analysis of the same 105-day-old corticOs cohort
revealed a total of 2061 DE genes between the LIS1^+/^− and control
samples (Supplementary Fig. [98]3a, Supplementary Data [99]3). Pathway
analysis of DE genes indicated that here, too, the collagen-associated
pathway was the most affected GO-term in the mutation (Supplementary
Fig. [100]3b). In addition, we observed no significant increase in the
expression of the LMNA gene. Accordingly, genes involved in generating,
processing, or regulating collagens, such as several non-fibrillar
collagens, including COL12A1, COL14A1, COL16A1, and COL24A1 were
affected in LIS1 mutant corticOs. Other collagen coding genes were
downregulated in the mutation, including different α-chains of collagen
type IV and COL5A3^[101]27,[102]44,[103]47,[104]48.
Hippocampal LIS1 organoids exhibit increased ECM
Mutations in LIS1 substantially impact the organization and structure
of the cerebral cortex, and some abnormalities have been noted in the
hippocampus^[105]49. The Lis1 mouse models display pronounced
hippocampal abnormalities, but the human hippocampal pathophysiology
has not been extensively characterized^[106]28–[107]36. We generated
control and LIS1^+/^− mutated hippocampal organoids (hippOs). The
hippOs were generated by exposing the 18-day-old aggregates to a
temporal BMP4 and WNT activation (Supplementary Fig. [108]4a)^[109]50.
After 70 days, the tissue contained hippocampal-like ZBTB20^+ and
LEF1^+ cells, SOX2^+, PAX6^+ and HOPX^+ progenitors, NeuN^+ and MAP2^+
neurons, and GFAP^+ astrocytes (Supplementary Fig. [110]4b–h).
Using Mass spectrometry, we explored differences in the protein content
of control and mutant hippOs^[111]16 (Supplementary Fig. [112]4i). A
total of 5068 proteins were identified and quantified (Supplementary
Data [113]4), of which 1178 were DE in the mutation, based on the
threshold criteria of p < 0.05, Log2Fold change ≥|0.5|, and >1 peptide
per protein. The proteome confirmed a considerable reduction of the
mutant LIS1 protein and revealed that LIS1^+/^− organoids were highly
enriched with ECM-related proteins (Supplementary Fig. [114]4i). DE ECM
proteins included an increased level of structural proteins in the
mutant, including several collagen types. These excessive collagens
undergo efficient hydroxylation in the mutant organoids, as assessed by
the post-translational modifications (PTM) analysis extracted from the
mass spectrometry data (Supplementary Fig. [115]4j, Supplementary
Data [116]5). Increased levels of the enzymes involved in these
modifications (P3H1-3 and PLOD-3) in the mutant samples further support
this. Co- and post-translational hydroxylation of proline residues are
essential for maintaining the stability of the triple helical collagen
structure, which plays a crucial role in brain development by
influencing the properties of the ECM, as well as interactions between
cells and their nearby collagens, thus impacting cell
behavior^[117]51–[118]53.
In addition to the enrichment in structural proteins, we also observed
increased levels of ECM remodeling proteins, such as MMP-14 and LOX, as
well as serine protease inhibitors such as SERPINA3, SERPINE2,
SERPINB6, and others. These observations were also supported by Gene
Set Enrichment Analysis^[119]54, which highlighted the enrichment of
genes associated with ECM, collagen fibril organization, and more
(Supplementary Fig. [120]4k). Overall, these findings suggest that
brain organoid models representing the human hippocampus also
demonstrated pathology linked to the mutation. This aligns with the
limited human data available and is in line with observations from
previous mouse models.
Rescue treatment with MMP9 softened the LIS1-mutated organoids
The stiffening effect of LIS1^+/^− mutations can be explained by the
altered regulation of ECM secretion and remodeling, as demonstrated by
our mass-spec data. To examine whether the stiffness is derived from
excessive structural ECM proteins, we treated day 18 corticOs with the
catalytic subunit of MMP9. This zinc-dependent endopeptidase can cleave
multiple ECM and non-ECM fibers broadly expressed in the
brain^[121]55,[122]56. The activity of the MMP9 catalytic domain was
tested first at different concentrations by an ELISA activity assay
(Supplementary Fig. [123]5a). CorticOs were then immersed with 500 nM
MMP9 catalytic domain for 10 min and submitted for MPA rheology
(Fig. [124]3a). Despite the proteolytic treatment, the organoids
maintained an SLS-like aspiration dynamics and the creep compliance
functions indicated a significant increase in deformability
(Fig. [125]3b). Indeed, MMP proteolytic treatment decreased both
[MATH: k0
:MATH]
and
[MATH:
kst
:MATH]
by ~15% in non-mutated organoids and ~50% in LIS1^+/^− mutated
organoids relative to non-treated organoids. Notably, the response time
[MATH: τ :MATH]
remained invariant to ECM digestion (Fig. [126]3b’).
Fig. 3. The effects of LIS1^+/^− mutation and ECM proteolysis on corticOs
mechanics and structural organization.
[127]Fig. 3
[128]Open in a new tab
a Day-18 control and LIS1^+/^− mutated organoids were treated with
500 nM MMP9 catalytic domain for 10 min at 37 °C before being submitted
for MPA creep test measurement and MRI imaging. Created in BioRender.
Solomonov, I. (2025) [129]https://BioRender.com/bncn17s. b Similar to
the non-treated organoids, the creep compliance of MMP9-treated
organoids of both genotypes was fitted to the linear viscoelastic SLS
model with a high goodness of fit (R-square (R^2 > 0.99)). Symbols and
error bars correspond to mean and SEM across, organoids. b’, Mean ± SD
of the fitted SLS viscoelastic elements, k[0,] k[st], and τ show
greater softening of the mutated corticOs by ECM proteolysis.
Statistical significance is evaluated via one-way ANOVA test with the
following number of organoids, each measured separately and fitted
independently: n[control] = 4, n[10F] = 5, n[controlMMP9] = 5,
n[10FMMP9] = 4. c A representative DW-MRI of eight LIS1^+/^− corticOs
(left) and the calculated ADC map. One central slice is shown. Three
out of six b-values (the degree of diffusion weighting) are shown for a
central slice. d Estimated maximal likelihood position of the ADC
(apparent diffusion coefficient) distributions of control, LIS1^+/^−,
and MMP9-treated LIS1^+/^− corticOs show the effect of altered
genotypes rescued by ECM proteolysis. Error bars correspond to SD
across consecutively repeated scans. e The relative deviations of the
ADC values of LIS1^+/^− and MMP9-treated LIS1^+/^− corticOs from
control organoids (n = 5) are plotted. The potential impact of
longitudinal drift was eliminated by calculating each scan’s relative
deviation, which was then averaged.
In addition, it has been previously shown that mutations cause
biomechanical changes at the cellular level^[130]16. So far, we found
that LIS1^+/^− organoids were stiffer (
[MATH: k0
:MATH]
and
[MATH:
kst
:MATH]
) compared to control organoids and that this stiffness was reduced by
MMP9 treatment. However, MPA tests with 12.5 µm pipettes, in which
cell-level mechanics can be examined, revealed that cell contractility
interference by myosin ATPase inhibition of blebbistatin also resulted
in softened k[o] and k[st] parameters in 18-day-old LIS1^+/^− organoids
(Supplementary Fig. [131]5b–d). LIS1^+/^− organoids exhibit prolonged
energy dissipation, likely due to increased ECM deposition. These
findings confirm that the LIS1 mutation enhances stiffness and friction
at both the cell and tissue levels. Combined, the MMP9 and blebbistatin
treatments implicated both ECM and contractile forces involved in
tissue stiffening of the LIS1^+/^− organoids.
The effects of LIS1^+/^− mutations and ECM proteolysis on organoid
mechanics are likely associated with structural remodeling. To test
this directly, we employed diffusion-weighted magnetic resonance
imaging (DW-MRI) to compare differences in the structural organization
of day 18 control, LIS1^+/^−, and MMP-treated LIS1^+/− corticOs.
Notably, DW-MRI was previously shown to provide noninvasive means for
identifying changes in ECM organization^[132]57, extracellular free
water component^[133]58, stiffness, and other structural
parameters^[134]59. To examine the sensitivity of DW-MRI, groups of
control, LIS1^+/^−, and MMP-treated LIS1^+/^− containing multiple
corticOs were placed in separate wells and scanned (Fig. [135]3c-left).
Taking advantage of the ultra-high magnetic field (15 T MRI scanner),
we reached a high imaging resolution of 100 × 100 μm^2 in-plane and
200 μm slice thickness voxels. Apparent diffusion coefficient (ADC)
maps were calculated (Fig. [136]3c-right), and the maximal likelihood
position was estimated for each group (Fig. [137]3d). To improve
statistical significance, five consecutive scans were performed and
analyzed by a matched t-test; the matching was effective (P < 0.0001).
The ADC of LIS1^+/^− organoids exhibited a statistically significant
8.9 ± 2.2% increase in comparison to control organoids (adjusted p
value: 0.0007, n = 5) (Fig. [138]3e). In contrast, the ADC of
MMP-treated LIS1^+/^− organoids was not significantly different from
controls. To support our findings, we performed three additional
experiments that showed similar trends (Supplementary Fig. [139]6a–d).
These results demonstrate that LIS1^+/^− mutations contribute to
increased organoid stiffness through altered ECM regulation, as
evidenced by mass-spec data and the significant impact of MMP9
proteolysis on LIS1^+/^− corticOs. DW-MRI further validates these
findings by highlighting structural changes in ECM organization and
stiffness in mutated and treated organoids.
MMP9 treatment reversed numerous abnormal gene expression patterns in
LIS1^+/^− organoids
We enquired whether some of the DE genes in LIS1^+/^− would reverse in
response to the treatment and its impact on the levels of stiffness and
diffusion in the mutated organoids shortly after the treatment. To
emphasize potential rescue mechanisms, our attention was directed
toward genes that exhibited differential expression between LIS1^+/^−
and control groups prior to treatment but not afterward (Supplementary
Data [140]6). A heatmap of a subset of these genes is shown in
Fig. [141]4a. These results indicate that at least part of the response
is mediated through immediate (~10 min) changes in gene expression. One
possible mechanism for this quick recovery is a stiffness-sensitive
expression of micro-RNAs (miRs).
Fig. 4. Rescued genes and inverse interaction between mRNA and miRs in
LIS1^+/^− organoids.
[142]Fig. 4
[143]Open in a new tab
a A heatmap showing the top rescued genes and their expression before
and after the MMP9 treatment, N = 4*, n = 12–14. b Top KEGG pathways
identified to be most influenced by the inverse relation between miRs
and mRNA expression. Negative numbers in the bar plot indicate that
mRNA levels were reduced and that their targeting miRs increased. In
contrast, positive values on the x-axis indicate an increase in mRNAs
in the LIS1^+/^− samples while their targeting miRs were reduced
compared to the control. c Small RNA-seq integrated with target
matrisome-related genes showing the opposite expression pattern of miRs
and their predicted targeted mRNAs. *Two outlier groups from the
non-treated LIS1^+/^− samples were excluded from the analysis.
Previous studies have shown that changes in the stiffness of a
substrate used to grow cells affected the expression of a large group
of miRs, and changes in miRs can affect the physical properties of
cells and tissues^[144]60–[145]62. In addition, our group recently
reported that LIS1 is involved in regulating gene expression at several
levels, including gene transcription, RNA splicing, and regulation of
miRs. These changes were either dependent or independent of the
Argonaute complex^[146]27. Thus, we also conducted a small
RNA-sequencing analysis, which revealed 274 DE miRs (Supplementary
Fig. [147]7, Supplementary Data [148]7). These results suggest that at
least some of the ECM expression abnormalities are linked to altered
regulation of these genes’ expression by miRNA.
Using our existing database of the miRs and mRNA expression of corticOs
on day 105, we found that, indeed, the most significant interaction
between DE mRNAs and miRs in the LIS1^+/^− mutation was related to
abnormal expression of genes associated with the ECM and the miRs that
are predicted to target and regulate their expression. Interestingly,
when DE miRs identified in small RNA-seq were paired with DE mRNAs to
identify regulatory mechanisms supported by both expression profiles,
KEGG pathways enrichment analysis
([149]http://microrna.gr/miRPathv3)^[150]44,[151]47,[152]48 pointed to
the “ECM-receptor signaling” pathway (p < 0.0001) (Fig. [153]4b,
Supplementary Data [154]8). Targeted genes highlighted by this miR-mRNA
comparison included COL4A3/4/5, COL5A3, ITGA1, COL1A1, THB2, COL24A1,
and COL6A6 (Fig. [155]4c). These results suggest that at least some of
the ECM expression abnormalities are linked to altered regulation of
these genes’ expression by miRs.
Collagen organization and a microstructure mechanical model
To further characterize the ECM abnormality, control and mutated
organoids were immunostained for several of the abnormally expressed
structural ECM components in mutated and control organoids. While these
immunostainings did not record the collagen enrichment, they revealed
substantial disorganization of type 4 and type 3 collagen fibers in 9-
and 18-day-old LIS1^+/^− organoids. Whereas an apparent ring-like
structure was observed in the control samples, the signal in the
LIS1^+/^− samples was diffuse and punctate (Fig. [156]5a, b).
Fig. 5. Collagen organization and a microstructure mechanical model.
[157]Fig. 5
[158]Open in a new tab
a–b Immunostainings of COL4A1 and COL3A1 on a day 9 and b day 18, in
control and LIS1^+/^− organoids with respective normalized intensity
quantification of staining showing no difference between control and
mutant organoids (Two-tailed independent student’s t test, α = 0.05,
Day 9: N[control] = 3, N[LIS1+/−] = 3; Day 18–N[control] = 5,
N[LIS1+/−] = 5). Scale bars represent 50 µm. c–d Sholl analysis of
COL4A1 signal in c day 9 (Mann–Whitney U test for the signal data
between control and LIS1 groups, p value = 0.0057) and d day 18
(Mann–Whitney U test for the signal data between control and LIS1
groups, p value = 3.28e^−^47) control and LIS1^+/^− organoids
represented as distribution from the center of the organoids
highlighting the circular arrangement of collagen deposition in the
control compared to an abnormal collagen distribution in the LIS1^+/^−
organoids (Mann–Whitney U test, α = 0.05). The Sholl analysis included:
Day 9–N[control] = 2, N[LIS1+/−] = 3; Day 18–N[control] = 3,
N[LIS1+/−] = 3. e–f Simulation snapshots for 1% and 20% uniaxial
compression strain in the control case. The control case is described
by ECM removed within a localized circular region in the center of the
system, as motivated by the ring of ECM. g, h Simulation snapshots for
1% and 20% uniaxial compression strain in LIS1^+/^− mutant case. The
mutant case is described by a randomly diluted ECM. i Organoid
stiffness (in simulation units) as a function of occupation probability
p, which indicates the amount of ECM. Here, N = 64, N[c] = 32, and
R = 10.
These changes were further recorded in Sholl analysis, measuring the
frequency of signals at each normalized distance from the center of
each organoid on days 9 and 18. On day 9, collagen type 4 forms a
ring-like structure near the outer borders of the organoids in control
samples but rather scattered in the mutant (p value = 0.0057)
(Fig. [159]5c). By day 18, the ring becomes more internalized in the
control organoids but remains more widely and randomly spread in the
mutated organoids (p value = 3.28e^−^47) (Fig. [160]5d). Thus, across
different developmental time points, the LIS1^+/^− organoids continue
to exhibit a disorganized distribution of COL4A1 rather than a
circularly organized collagen as observed in the controls. These
findings suggest that the LIS1 protein is essential for the proper
organization and localization of the ECM. We compared the LIS1 mutant
organoids with the control ones using additional immunostainings, and
they had a greater number of SOX2^+ and PAX6^+ progenitors
(Supplementary Fig. [161]8a, b). However, they did not differ in the
population of pVim^+ radial glia or HOPX^+ basal radial glia
(Supplementary Fig. [162]8c, d). We did observe an increase in the
number of pHH3^+ mitotically active cells and the number of cells
exiting the cell cycle (KI67^−/EdU^+), which may be attributed to the
roles of LIS1 in cell cycle progression; but no changes were noted in
the population of KI67^+ proliferating cells and EdU^+ cells entering
the S-phase. Furthermore, there were no significant differences in the
number of apoptotic cells (Supplementary Fig. [163]9).
Finally, given the changes in the amount and spatial organization of
the ECM between the control case and the LIS1^+/^− case, as well as the
differences in stiffness between the two cases, we devised a more
detailed mechanical model beyond the SLS model to help provide a
quantitative link between the amount and organization of the ECM to
brain stiffness. More specifically, we considered a computational model
that includes both ECM fibers and cells and computed the stiffness of
such a model. For simplicity, we investigated a two-dimensional model,
or a cross-section, of the brain organoid. We will address this
simplification in terms of what results we anticipate will carry over
to three dimensions below.
The ECM was modeled as a triangular network of fibers, each consisting
of both stretching energy and bending energy, with the latter encoding
the semiflexibility of the collagen fibers^[164]63,[165]64. In
contrast, the cells are modeled as individual triangles randomly
inserted in the lattice with some area stiffness^[166]65
(Fig. [167]5e–h). While the shapes of the cells in the computational
model do not reflect the actual cellular shape, they do encode a key
role in the mechanics of the cells. A line of edges on the lattice in
one of the three principal directions represents a collagen fiber. To
illustrate the disordered structure of the ECM, each edge in the
triangular lattice was occupied independently and at random with
occupation probability p, which is also a measure of the fraction of
edges in the network. In other words, the larger the p, the more ECM
and vice versa, and so changing this p parameter encodes the amount of
ECM.
Moreover, given the observation of more patterned ECM in the
control case, we also explored an ECM of a similar amount of ECM
(similar p). However, the ECM is removed only within a localized region
(Fig. [168]5e, f). We then compared patterned ECM with randomized ECM
in terms of influencing brain organoid stiffness. We note that such a
model has already qualitatively captured nontrivial mechanical features
of a fiber network with embedded particles, such as the phenomenon of
compression stiffening^[169]65,[170]66 (See Methods for a detailed
description of the model and accompanying numerical calculations).
In control samples (Fig. [171]5e, f), as the brain organoid tissue is
compressed with an increasing amount of uniaxial strain, the cells and
fibers begin to distort; therefore, the energy of the tissue increases.
We observed that the stiffness of the tissue, measured in simulation
units, decreased as more ECM was removed from the localized circular
region (Fig. [172]5i), as one goes from the gray box to the gray X.
Notably, the decrease in stiffness is nonlinear, and the ECM provides
the tissue with structural support and enhanced stiffness.
However, in the mutant case (Fig. [173]5g, h), where the ECM does not
appear as organized and so the ECM randomly occupies edges on the
network, we found that the tissue stiffness decreased nonlinearly as
more ECM is removed (See Fig. [174]5i) as one goes from the green box
to the green X. Interestingly, when the patterned ECM stiffness is
compared to the random ECM stiffness for the same amount of ECM, the
patterned ECM stiffness is not as large as the random ECM stiffness, at
least for smaller amounts of ECM, i.e., smaller p. We argue that to
compare the experiments to our computational model qualitatively, one
can go from a smaller amount of ECM (lower p) on the patterned curve to
a higher amount of ECM (higher p) on the random curve as indicated on
(Fig. [175]5i) by going from the gray box to the green box. For
instance, if we begin at p = 0.85 fraction of edges occupied on the
patterned ECM curve to 0.95 fraction of edges on the random ECM curve,
we find a relative increase in the average stiffness of ~54%.
Moreover, this microstructure mechanical model qualitatively supports
the results from the MMP9 treatment, in which there was a greater
decrease in organoid stiffness for the mutant case compared to the
control. As decreasing the occupation probability also reduces the
length of ECM fibers, the same trend is observed in our microstructure
mechanical model, as indicated by going from the green box to the green
X versus going from the gray box to the gray X. Note that this
treatment observation is more specific to the parameters at hand. These
results were further tested under two additional cases of the parameter
space, with more cells and with more stretchable fibers (Supplementary
Fig. [176]10a, b, respectively).
Overall, the computational model supports our findings that an
upregulation of ECM enhances stiffness nonlinearly, which is not
unexpected. However, our finding that the patterning of the same amount
of ECM—diluted from a localized region as compared to randomly—affects
organoid tissue stiffness is rather nontrivial in living tissue. The
patterning creates weak spots, which, in turn, create a weaker living
material. Clearly, the computational model provides a quantitative
understanding of the trends observed in the experiments, including the
MMP9 treatment findings. Finally, the computational model can predict
brain organoid stiffness for decellularized material, presumably
leading to compression softening^[177]65,[178]67. The model can also
provide predictions for the trend in any stiffness change for other
mutated brain organoids should the organization of the ECM be altered
in a different way from the mutant studied here.
Discussion
Here, we studied the effects of LIS1 haploinsufficiency on ECM
composition and organization and how these changes translate into
modified physical properties of human brain-like organoids at different
developmental time points. Our findings highlight LIS1 as an important
regulator of ECM dynamics during human brain development. In both
hippocampal and cortical organoid models, LIS1^+/^− mutations have been
shown to significantly influence ECM composition and organization,
leading to observable changes in tissue stiffness. Mass spectrometry
analyses confirmed the enrichment of ECM-related proteins, particularly
collagens, in LIS1^+/^− organoids, indicating a mutation-driven
alteration in ECM secretion and remodeling. These proteomic changes
were complemented by biomechanical assessments using MPA rheology,
revealing increased stiffness in LIS1^+/^− organoids. Intriguingly, the
application of MMP9, a zinc-dependent endopeptidase, on cortical
organoids demonstrated a notable reduction in stiffness, especially in
LIS1^+/^− organoids, suggesting the reversibility of these changes
through ECM proteolysis. Furthermore, DW-MRI provided noninvasive
insights into the structural organization alterations in these
organoids, reinforcing the link between LIS1 mutations, ECM
dysregulation, and brain tissue mechanics. These results collectively
highlight the critical impact of LIS1 on ECM regulation and its
subsequent influence on brain development and structural integrity.
The ECM is not merely a passive substance between cells but an active
source of signaling and regulating the stem cell niche^[179]68–[180]70.
It consists of molecules secreted by cells, serving as a structural
scaffold while also harboring water reserves, growth factors,
morphogens, and various bioactive molecules that engage neighboring
cells and influence signaling processes^[181]71,[182]72. Past studies
identified that the ECM plays a pivotal role in the development of the
nervous system. For example, the synchronized movement of the neural
plate and the mesoderm relies on the interplay between two ECM
components, laminin and fibronectin^[183]73. Furthermore, the ECM is
highly abundant within neuronal progenitors, and evidence shows that
ECM enrichment is most pronounced in progenitors, which is uniquely
prevalent in the developing human brain compared to the mouse
brain^[184]22,[185]23. Ex vivo experimental changes in the ECM affected
the structure of human embryonic brain sections^[186]20,[187]21. Our
analyses revealed an abundance of matrisome-related proteins in
LIS1^+/^− hippocampal and cortical organoids, whereas the composition
of these two types of organoids exhibited distinctions. Previous
studies have indicated that brain regions have ECM content
variability^[188]74. The observed proteomic changes in the corticOs
partially correlated with corresponding transcriptomics data (mRNA and
small RNA), underscoring the roles of LIS1 in post-transcriptional
regulation^[189]16,[190]24,[191]27. LIS1 is an RNA-binding protein
involved in post-transcriptional regulation and governs the physical
properties of embryonic stem cells^[192]27.
The LIS1 mutation affected the physical properties of the brain-like
organoids at multiple developmental stages; the LIS1 mutant organoids
were stiffer. The nuclei embedded within this more rigid tissue
exhibited modulated parameters. LIS1^+/^− brain organoids exhibited
increased levels of the nuclear lamina proteins, Lamin A/C, which are
known to scale with increased
stiffness^[193]39,[194]45,[195]75–[196]77. In a correlative manner, the
levels of DNA damage, indicated by γH2AX, decreased. The decrease in
γH2AX levels in LIS1 mutants is significant, as γH2AX marks DNA damage
and recruits repair machinery to maintain genomic stability. Reduced
γH2AX suggests impaired DNA damage response, potentially hindering
effective repair and leading to mutation accumulation. This may
exacerbate developmental defects in highly dynamic tissues like the
developing brain. The elevated levels of the ECM and increased rigidity
are echoed in the ADC values. ADC values reflect the free water content
of the tissue, which, in the case of brain-like tissue, is correlated
with ECM^[197]78. ADC values decrease in pathologies that involve cell
swelling (edema) and can increase in chronic phases of stroke or other
diseases involving necrotic cell death^[198]78,[199]79. A monkey model
employed to simulate a developmental brain disorder known as maternal
immune activation, considered a model for autism spectrum disorders,
displayed a significant elevation in the presence of extracellular free
water within the gray matter of the cingulate cortex^[200]58.
The physical measurements encompassing rheology and MRI appear to have
greater sensitivity than the protein-based assays involving mass
spectrometry and specific collagen immunostainings. This heightened
sensitivity is evident in the fact that we failed to observe a
substantial rise in matrisome proteins through proteomics analysis or,
in particular, collagens through immunostaining for the early time
points. However, immunostaining unveiled an unanticipated spatial
arrangement of COL4A1 and COL3A1. While in control samples, they formed
a circular structure, in LIS1 mutant samples, they appeared scattered.
To consolidate our findings, we employed a computational model that
takes into account the mechanics of cells, the ECM, and ECM
organization. The control and the LIS1 mutant differed in the amount of
ECM and ECM organization and, hence, altered brain organoid stiffness,
with the control case being less stiff than the mutant case in both the
experiments and the computational model. Moreover, both the experiments
and the computational model showed that effectively chopping up ECM
enzymatically via MMP9 treatment led to a decrease in stiffness, with a
pronounced decrease in the mutant case. Indeed, the mechanics and
structure of brain organoids are intertwined as one helps determine the
other. Computational models such as the one presented here, as well as
other mechanical models^[201]80–[202]84 are, therefore, key to
understanding the structure of brain organoids in both healthy and
diseased states.
Moreover, multiscale, computational modeling tying the chromatin scale
to the brain organoid or tissue scale to unravel the multiscale
mechanics-structure relationship is on the horizon^[203]85. Indeed,
examining brain organoids from a material point of view, as
demonstrated here, provides perspectives about their structure and will
help unravel the intricate mechanics-structure-function relationship in
the developing brain and morphogenesis more generally.
Our study reveals the critical role of tissue mechanics in brain
development, demonstrating how viscoelastic properties are altered in
lissencephaly caused by LIS1 mutations. Using human brain organoids, we
uncovered key mechanical and molecular changes, such as increased
stiffness and abnormal ECM organization, providing insights into
previously understudied aspects of the disease. Combining rheological
measurements, MRI imaging, molecular analyses, and computational
modeling, we showed that short-term MMP9 treatment can reverse
stiffness and water diffusion abnormalities, highlighting the
therapeutic potential of targeting ECM dysregulation. These findings
establish tissue mechanics as a vital factor in brain malformations and
a promising target for intervention. Our research sets a paradigm for
studying mechanical principles in brain disorders, advancing
mechanobiology in neurodevelopmental and neurological diseases, while
offering innovative therapeutic targets.
Methods
Cell lines
An NIH-approved hESC line NIHhESC-10-0079, WIBR3 (W3), was used in this
study. Isogenic mutant cell-line clones were previously generated by
CRISPR-Cas9-mediated heterozygous deletion in the LIS1 gene^[204]16.
Cell lines were regularly checked for mycoplasma contamination. The
pX335 plasmid, an empty Cas9 nickase plasmid used in creating the
original LIS1^+/^− cell line, was electroporated into the parental
WIBR3 line to produce a second control. The PX335 line was used for the
proteomic analysis on 35-day-old organoids. Finally, a single colony
was isolated from the WIBR3 line through subcloning and was used as an
additional control for the rheology experiment on day 35.
Generation of hESC-derived corticOs and hippOs
hESCs were cultured in naïve media^[205]86 until 70% confluency and
then dissociated and aggregated in low-adhesion wells (Day = 0).
Aggregates were primed towards a neurogenic fate by application of the
TGF-β- and WNT- signaling inhibitors, SB-431542 and IWR-1, respectively
(as specified^[206]50,[207]87). These molecules facilitate
neuroectoderm fate by inhibiting the mesodermal-promoting Nodal/Activin
pathway^[208]88; while preventing premature neuronal differentiation by
inhibiting the WNT pathway via AXIN2 stabilization^[209]89,[210]90. On
the 19^th day, aggregated neural precursors were introduced to
conditions promoting either hippocampal or neocortical formation. In
corticogenesis, the region of the dorsal pallium gives rise to the
neocortex. In contrast, the medial pallium generates the hippocampus,
positioned between the neocortex and the midline cortical hem. The
cortical hem secretes dorsalising patterning morphogens such as WNTs
and BMPs, which promote medial pallium formation. These also suppress
the expression of FGF8 from the competent cortical primordium, which
defines a more dorsal neocortical identity^[211]91. Timely exposure
(72 h) to BMP4 and the WNT agonist CHIR-99021 was shown to be
sufficient for inducing hippocampal fate by inhibiting GSK3β and
activating the WNT pathway^[212]89,[213]90, and was therefore used to
generate hippocampal organoids (hippOs). For developing cortical
organoids (corticOs), aggregates were grown in a serum-free floating
culture where they preferentially expressed FGF8. FGF8 then promotes
self-organizing rostral-caudal polarity. After 35 days, corticOs were
treated with hLIF to induce bRG progenitor proliferation^[214]37.
Organoids were not embedded or exposed to Matrigel.
Immunohistochemistry
Ectoderm-like organoids were examined using immunohistochemistry 9 and
18 days after aggregation. Organoids were washed in PBS for 10 min at
RT thrice, fixed for 30 min in 4% PFA, and washed again in PBS. Samples
were dehydrated overnight at 4 °C in 30% sucrose in PBS, embedded in
OCT blocks, and sliced into 16 µm cryosections. Antigen retrieval was
conducted in a water bath heated to 90 °C in 10 mM sodium citrate
buffer pH = 6 for 20 min and then chilled at RT. Tissues were
permeabilized and blocked in a blocking solution (10% normal donkey
serum in PBST [0.1% Triton X-100]) for 3 h and then incubated with
primary antibodies for 48 h at 4 °C. After three washes, slides were
presented with the secondary antibodies from the relevant species to
fit the primary antibody for 1:30 h at RT at 1:200 concentration, after
which they were incubated with 1:5000 4’,5-diamidino-2-phenylondole
(DAPI) for 5 min. All the antibodies used in the study, their
dilutions, and catalog numbers are listed in the supplementary
information (Supplementary Table [215]1).
Image analysis and quantification
Immunostained slides were imaged with Andor Dragonfly spinning disk
confocal microscope system HC FLUOTAR ×25/0.95 W VISIR lens, and
stitched to get complete images of the whole organoid sections using
the Andor: Oxford Instruments Fusion Shell software. The images were
then analyzed using the Spots Analysis of the Oxford Imaris software to
determine the percentage of the total number of cells that expressed a
particular cell-type-specific marker and compare this parameter between
control and mutant organoids. The total number of cells is considered
the same as the total number of DAPI^+ nuclei. Markers with nuclear
expression, such as KI67, pHH3, EdU, SOX2, PAX6, HOPX, and pVim, were
determined by setting up a colocalization filter under the Spots
analysis.
For markers like COL3A1 and COL4A1, the mean intensity of the signal in
the total area of the organoid section was calculated in Fiji ImageJ
software to compare their levels of expression. These values were then
compared using statistical tests in GraphPad Prism to determine
statistically significant differences, if any.
Sholl analysis quantification
The distribution of COL4A1 signal in day 9 and day 18 corticOs was
determined using the Sholl analysis plugin
([216]https://imagej.net/imagej-wiki-static/Sholl) in Fiji ImageJ. This
method assesses signal complexity and spatial distribution relative to
the organoid’s center. A thresholded image of the organoid section was
used, and a straight line of uniform thickness was manually drawn
across the widest diameter of the section, determining the geometric
center. The plugin then generated a series of concentric circles,
centred on this point, at predefined radial increments. Each circle
represents a distance bin for quantifying COL4A1 signal density. The
software quantified the number of COL4A1-positive signals intersecting
each concentric circle, starting from the centre and moving radially
outward toward the periphery of the organoid section. This process was
repeated across multiple organoid sections to ensure robust sampling.
Accordingly, a Sholl profile was produced and exported as a
distribution table. The data were further analysed statistically and
visualised in GraphPad Prism, where a nonlinear regression curve was
fitted to the distribution to model spatial signal decay or clustering
patterns.
Proteomics and western blot
Sample preparation
Organoids were placed on ice and washed twice with 1× PBS for 5 min. In
the third wash, organoids were placed in 100 mM Tris-HCL, pH = 7.5.
Proteins were extracted using a lysis buffer (5% SDS 100 mM Tris
pH = 7.5). The lysates were transferred into soft tissue homogenizing
CK14 tubes (Bertin Corp), placed in a homogenizer shaker at 400 bps for
2 sec, and then put on ice for 20 min. Samples were centrifuged at
12,000 × g at 4 °C for 20 min and then placed on ice. The uppermost,
transparent liquid was transferred to a new Eppendorf tube and sent
immediately to the De Botton Protein Profiling Institute of the Nancy
and Stephen Grand Israel National Centre for Personalised Medicine,
Weizmann Institute of Science, for a ‘global quantification of protein’
mass spectrometry. This discovery experiment aimed to quantify as many
proteins as possible using label-free methods. Samples included control
and LIS1^+/^− mutated hippOs and corticOs on days 70 and 105,
respectively. Leftovers were aliquoted and used for Western blot
verifications.
Cell lysates were subjected to in-solution tryptic digestion using the
suspension trapping (S-trap) method as previously described^[217]92.
Briefly, lysates were incubated at 96 °C for 5 min, followed by six
cycles of 30 s of sonication (Bioruptor Pico, Diagenode, USA). Protein
concentration was measured using the BCA assay (Thermo Scientific,
USA). From each sample 20 µg of total protein were reduced with 5 mM
dithiothreitol and alkylated with 10 mM iodoacetamide in the dark. Each
sample was loaded onto S-Trap microcolumns (Protifi, USA) according to
the manufacturer’s instructions. After loading, samples were washed
with 90:10% methanol/50 mM ammonium bicarbonate. Samples were then
digested with trypsin (1:50 trypsin:protein ratio) for 1.5 h at 47 °C.
The digested peptides were eluted using 50 mM ammonium bicarbonate.
Trypsin (1:50 trypsin:protein ratio) was added to this fraction and
incubated overnight at 37 °C. Two more elutions were made using 0.2%
formic acid and 0.2% formic acid in 50% acetonitrile. The three
elutions were pooled together and vacuum-centrifuged to dryness.
Samples were resuspended in H[2]O with 0.1% formic acid and subjected
to solid phase extraction (Oasis HLB, Waters, Milford, MA, USA)
according to manufacturer instructions and vacuum-centrifuged to
dryness. Samples were kept at −80 °C until further analysis.
Liquid chromatography
ULC/MS grade solvents were used for all chromatographic steps. Dry
digested peptides were dissolved in 97:3% H[2]O/acetonitrile + 0.1%
formic acid. Each sample was loaded using split-less nano-Ultra
Performance Liquid Chromatography (nanoElute Bruker Daltonics,
Germany). The peptides were separated using an Aurora column (75 μm
ID × 25 cm, IonOpticks) at 0.3 µL/min. Peptides were eluted from the
column into the mass spectrometer using the following gradient: 2% to
27% B in 100, then back to initial conditions.
Each sample was loaded using split-less nano-Ultra Performance Liquid
Chromatography (10 kpsi nanoAcquity; Waters, Milford, MA, USA). The
mobile phase was: A) H[2]O + 0.1% formic acid and B) acetonitrile +
0.1% formic acid. The samples were desalted online using a
reversed-phase Symmetry C18 trapping column (180 µm internal diameter,
20 mm length, 5 µm particle size; Waters). The peptides were then
separated using a T3 HSS nano-column (75 µm internal diameter, 250 mm
length, 1.8 µm particle size; Waters) at 0.35 µL/min. Peptides were
eluted from the column into the mass spectrometer using the following
gradient: 4% to 33% B in 155 min, 33% to 90% B in 5 min, maintained at
90% for 5 min and then back to initial conditions.
Mass Spectrometry
The nanoUPLC was coupled online to a Time-of-flight mass spectrometer
(timsTOF Pro, Bruker, Daltonics, Germany). Data was acquired in
data-dependent acquisition with ion mobility mode (data-dependent
acquisition (DDA) PASEF^[218]93 using a 1.1 sec cycle-time method with
10 MS/MS scans. For ion mobility 1/K0 range was 0.60–1.60 Vs/cm^2,
Energy Start in PASEF CID was set to 20.0 eV, and Energy End was set to
59.0 eV. Other parameters were kept as the default parameters of the
DDA PASEF method.
For hippocampal organoids proteomics: The nanoUPLC was coupled online
through a nanoESI emitter (10 μm tip; New Objective; Woburn, MA, USA)
to a quadrupole orbitrap mass spectrometer (Q Exactive HF, Thermo
Scientific) using a FlexIon nanospray apparatus (Proxeon).
Data was acquired in DDA mode, using a Top10 method. MS1 resolution was
set to 120,000 (at 200 m/z), mass range of 375-1650 m/z, AGC of 3e6 and
maximum injection time was set to 60 msec. MS2 resolution was set to
15,000, quadrupole isolation 1.7 m/z, AGC of 1e5, dynamic exclusion of
45 sec and maximum injection time of 60 msec.
Data processing and analysis
The raw data was processed with FragPipe v17.1. The data was searched
with the MSFragger search engine v3.4 against the human (Homo sapiens)
protein database as downloaded from Uniprot.org, appended with common
lab protein contaminants. Enzyme specificity was set to trypsin, and up
to two missed cleavages were allowed. Fixed modification was set to
carbamidomethylation of cysteines, and variable modification was set to
oxidation of methionines and protein N-terminal acetylation. The
quantitative comparison was calculated using Perseus v1.6.0.7. Decoy
hits were filtered out, and only proteins that had at least two valid
values after logarithmic transformation in at least one experimental
group were kept. For statistical calculations, missing values were
replaced by random values from a normal distribution using the
Imputation option in Perseus (width 0.3, downshift 1.8). A Student’s t
test, after the logarithmic transformation, was used to identify
significant differences between the experimental groups across the
biological replica. Fold changes were calculated based on the ratio of
geometric means of the different experimental groups.
Raw data was processed with MaxQuant v1.6.6.0^[219]94. The data was
searched with the Andromeda search engine against the human (Homo
sapiens) protein database as downloaded from Uniprot
([220]www.uniprot.com), and appended with common lab protein
contaminants. Enzyme specificity was set to trypsin, and up to two
missed cleavages were allowed. Fixed modification was set to
carbamidomethylation of cysteines, and variable modifications were set
to oxidation of methionines, asparagine and glutamine deamidation, and
protein N-terminal acetylation. Peptide precursor ions were searched
with a maximum mass deviation of 4.5 ppm and fragment ions with a
maximum mass deviation of 20 ppm. Peptide and protein identifications
were filtered at an FDR of 1% using the decoy database strategy
(MaxQuant’s “Revert” module). The minimal peptide length was 7 amino
acids, and the minimum Andromeda score for modified peptides was 40.
Peptide identifications were propagated across samples using the
match-between-runs option checked. Searches were performed with the
label-free quantification option selected. The quantitative comparisons
were calculated using Perseus v1.6.0.7. Decoy hits were filtered out,
and only proteins that were detected in at least two replicates of at
least one experimental group were kept. For statistical calculations,
missing values were replaced by random values from a normal
distribution using the Imputation option in Perseus (width 0.3,
downshift 1.8). A Student’s t test, after logarithmic transformation,
was used to identify significant differences between the experimental
groups across the biological replica. Fold changes were calculated
based on the ratio of geometric means of the different experimental
groups. Additionally, raw data was searched for PTMs using the GPTMD
module of the MetaMorpheus v0.0.311 algorithm^[221]95. The data were
searched against the same database described above.
Micropipette aspiration
MPA was performed using a manometer setup as reported
previously^[222]39. The organoids were placed, immersed in media, on a
glass coverslip mounted on top of an inverted fluorescence microscope
stage (Nikon Eclipse, Ti-U). The pipettes (Pipette brand and model)
were washed in media with serum to lubricate the inner surface walls
and decrease friction. Creep tests were performed over several seconds
under 0.5-to-3 kPa suction-generated load consistent with the
microenvironmental elasticity of healthy and sclerotic brain tissues.
For each organoid, the pipette inlet was brought into contact and creep
test was performed by applying 1-to-3 kPa constant suction, which was
measured relative to atmospheric pressure via a pressure transducer
(Validyne, Northridge CA, USA). Time-lapse imaging of the aspiration
dynamics into the pipette was recorded over ~ten sec and included ~two
sec before applying suction. We used pipettes with 0.15-to-0.5 mm inner
radii to measure tissue-level mechanics, thus probing integrated
multicellular and ECM contributions. Specifically, inner diameter
0.3 mm (B100-30-7.5HP, SUTTER INSTRUMENTS) was used on Day-9 and Day-18
corticOs, 0.5 mm (B100-50-7.5HP, SUTTER INSTRUMENTS) for Day-35, and
1 mm (BF200-100-10, SUTTER INSTRUMENTS) for 70-day-old coricOs.
This creep test response to applied load is characteristic of the
minimal linear viscoelastic SLS model. In its Maxwell representation,
it consists of an elastic element (spring k[1]) that is connected in
parallel to a second elastic element (spring k[2]) positioned in series
with a viscous element (dashpot µ) (Fig. [223]1a’).
The effective time-dependent deformability of the organoids is given by
the creep compliance function J(t), which we obtain over a range of
small deformations using the half-space model relationship^[224]96:
[MATH: Jt=2<
/mn>πLt3ϕ<
/mi>RpP :MATH]
1
The length of the aspirated fraction (L(t)) is scaled by the pipette
inner radius. RpP is the applied pressure (relative to atmospheric
pressure), and
[MATH: ∅ :MATH]
= 2.1 is a geometrical pipette wall factor. The organoids exhibited a
creep compliance function that was consistent with the SLS model,
allowing us to quantitatively evaluate the three viscoelastic
parameters by fitting^[225]97:
[MATH:
JSLS(t)=1kst1−k<
/mrow>0−kstk0e−t<
mi>τ :MATH]
2
In this physically intuitive representation, the instantaneous
stiffness k[0] = k[1] + k[2] measures the elastic resistance to abrupt
impact. The steady-state stiffness that determines the long-term
restoring forces is k[st] = k[1]. The time scale for the transition
from elastic stretching to steady-state deformation is estimated by:
[MATH: τ=μk1+k2k1
k2
:MATH]
3
Satisfyingly, the R-square goodness of fit of all organoid measurements
was high (typically > 0.98), confirming the choice of the minimal
linear viscoelastic SLS model (Supplementary Fig. [226]2b–f).
Consistently, the goodness of fit of the mean creep compliance
functions averaged across multiple organoids per condition was > 0.99
(Fig. [227]1b–e).
Blebbistatin treatment
The engaging pressure of the pipette with the organoids was about
1-1.5 kPa. Then, we increased the suction pressure to start aspiration
and continued to follow the creep length until it reached the steady
state. For control, the suction pressure required for aspiration was
1.5 kPa on average, and for mutated organoids, the average pressure
required to initiate aspiration was 2 kPa. Organoids were treated with
Myosin ATPase inhibitor blebbistatin (Racemic Blebbistatin, Sigma
Aldrich Catalog no. 203389) at a concentration of 50 µM and were
incubated in the incubator for 2 hr. For the blebbistatin-treated
mutated organoids, we used an average of 1.2 kPa. A pipette of
25-micron in diameter was used for all three conditions.
ELISA activity assay
The MMP9 catalytic domains were incubated in a 96-well plate with
different concentrations of Mca-PLGL-Dpa-AR-NH2 Fluorogenic MMP
Substrate (ES001, R&D SYSTEMS), a substrate that is cleaved by MMP9.
The peptide substrate contains a highly fluorescent 7-methoxy coumarin
group that is quenched when cleaved. Active peptidases that can cleave
an amide bond between the fluorescent group and the quencher group
cause an increase in fluorescence, which can confirm their
function^[228]98. Each well was supplemented with 20 nM of either MMP9,
a TNC digestion buffer (50 mM TRIS-HCl pH = 7.5, 150 mM NaCl, 5 mM
CaCl[2], and 0.05% Brij), and the substrate at a dilution series
(1–100 µM). Two control conditions were ‘TNC-only’ and a 10µM ‘TNC +
substrate’ condition. While at 37 °C, fluorescence was captured by a
plate reader (Synergy H1) every 50 seconds for an overall duration of
40 minutes (Supplementary Fig. [229]5).
MMP9 catalytic domain treatment
On day 18, control and LIS1^+/^− organoids were treated with 500 nM of
MMP9 catalytic domain diluted in the organoids’ original media and
incubated for 10 min at 37 °C, 5% CO[2]. The treated media was then
replaced with fresh 100 µl of ‘SA1 media’, and plates were held in the
incubator until the organoids were tested.
Diffusion MRI–data analysis
The diffusion-weighted imaging (DWI) MRI dataset was collected to
assess the differences between brain organoid groups. Based on the
collected dataset, the ADC maps were calculated using a
mono-exponential fit. The images with the highest b-value (the degree
of diffusion weighting) were used to segment and identify the voxels of
the brain organoid tissue. Supplementary Fig. [230]5a shows the ADC
maps and the contour of the segmented voxels. A normalized distribution
of the ADC values in the identified voxels for each type, consisting of
100 bins, was defined. A denoising of the distribution profile was then
performed, removing high-frequency components using FT. The
distribution of the diffusion in the brain organoid tissue results in
an asymmetric profile; therefore, the maximal likelihood position was
defined as the center of points with 2/3I[max]. intensity (Imax was
found based on the denoised distribution, and the two points had
2/3I[max] intensity from both sides of the distribution by
interpolation). The same steps were repeated for each organoid type and
scan. Supplementary Fig. [231]5a–c shows the ADC values distributions
and the estimated maximal likelihood positions. Note that the ADC
values increased as a function of the repeated scans. This can be due
to the long scan duration during which the organoids were out of their
regular environment. However, the deviation of LIS1^+/^− compared to
Control was preserved over the scan duration. Three additional
experiments were performed with different subsets of the organoid
types. Supplementary Fig. [232]5d shows a similar observed trend in the
repeated experiments. The ages of the organoids in the different
experiments were: experiment 1, 19 days; experiment 2 & 3, 18 days;
experiment 4, 16 days. The lower ADC values in Exp.#4 can be due to
different scan parameters (with a slice thickness of 100 μ compared to
200 μ; see the scan parameters summarized below.
MRI scanner details: Horizontal Bruker Biospec 15.2-T USR preclinical
MRI scanner with an Avance IIIHD console.
Container used for the MRI scans: In experiments #1–3, the organoids
were placed in a container with separated wells for each type, allowing
for scanning all types in one scan. Exp. #4 was scanned with a single
Eppendorf in which the organoids of one kind were placed and scanned,
and then the scan was repeated with the second type. The medium used
was PBS.
RF coil details: 1H MRI CryoProbe 2-element array kit for the mouse
head.
The DWI scan parameters:
* Exp.#1: TR/TE 400/13.7 ms, FOV 12 × 12 mm^2, in-plane resolution
100 × 100 μ^2, slice thickness 200 μ, 20 slices, ten averages, scan
duration 48 minutes. The scan was repeated five times. The scan
included six b-values - 0, 200, 400, 600, 1000, 1200 s/mm^2.
* Exp. #2: TR/TE 400/13.7 ms, FOV 18 × 12 mm^2, in-plane resolution
100 × 100 μ^2, slice thickness 200 μ, nine slices, ten averages,
scan duration 48 minutes. The scan included six b-values - 0, 200,
400, 600, 1000, 1200 s/mm^2.
* Exp.#3: TR/TE 400/13.7 ms, FOV 12 × 12 mm^2, in-plane resolution
100 × 100 μ^2, slice thickness 200 μ, 20 slices, 20 averages, scan
duration 1h36m0s0ms. The scan included six b-values - 0, 200, 400,
600, 1000, 1200 s/mm^2.
* Exp.#4: TR/TE 300/12.7 ms, FOV 12.6 × 2.4 mm^2, in-plane resolution
100 × 100 μ^2, slice thickness 100 μ, 60 slices (3D acquisition),
one average, scan duration 1h 12m 0s 0ms. The scan included four
b-values–0, 200,600,1000 s/mm^2.
RNA and miR sequencing
Total RNA from 105 days old corticOs (N = 4; n = 6–8 per genotype) and
from 18 days old MMP9-treated and untreated organoids (N = 4, of which
two outlier groups from the non-treated LIS1 +/− samples were excluded
from the analysis; n = 12–14) was extracted using the Direct-zol RNA
Miniprep Plus kit (Zymo Research) following the manufacturer’s
protocols. RNA concentration and integrity were measured using Nanodrop
(Thermo Scientific). Samples were then sent to the DNA Link Sequencing
Lab in S. Korea, where RNA-seq libraries were prepared using TrueSeq
standard mRNA library kit or small RNA-seq libraries. Libraries were
sequenced in NovaSeq6000. The raw data was processed using the Weizmann
Institute bioinformatics pipeline. DE genes were analyzed using
Metascape^[233]42 and GeneAnalytics^[234]44.
Computational model
The energy functional for the mechanics of the microstructure of a
brain organoid with N nodes and N[C] number of cells:
[MATH:
E=ks2
∑<ij>pijlij<
/mrow>−l02+<
/mo>kb2∑<ijk>pijpj<
mi>kθij<
mi>k−π2+∑n=1
NCλn<
msup>An−A02 :MATH]
4
Here, k[s] denotes the individual ECM fiber stretching stiffness, while
k[b] denotes the individual ECM fiber bending stiffness to give a
persistence length. Moreover, p[ij] represents the edge occupation
probability, or the fraction of edges in the network that are occupied.
We set p[ij] = p for convenience in the main body of the manuscript. As
l[ij] quantifies an edge length in the network, the first term in the
equation above captures the stretching energy. As θ[ijk] denotes the
angle between two edges, the second sum in the equation above
quantifies the semiflexibility of the fiber network, while the third
term encodes the bulk stiffness of N[c] cells in each area A[n] and
stiffness λ[n] = λq[n], with q[n] either zero or one as the number of
cells is set and then the triangles in the lattice are randomly
occupied with cells. Therefore, λ quantifies the individual bulk cell
stiffness, which is, for simplicity, the same for all cells. Note that
k[s]/(k[b]l[0]^2) = R and is the dimensionless ratio comparing the
individual ECM fiber stretching stiffness to bending stiffness.
To measure the stiffness K of this composite system, once the model
parameters are chosen, we impose periodic boundaries in the horizontal
direction and cut, delete the periodic edges in the vertical direction,
and then compress the system in the vertical direction by applying
strain γ. We then compute the energy E of the system, the first
derivative of the energy to obtain the strain, and, finally, the second
derivative of the energy of the system to compute the material
stiffness K, which includes a 1/A[0] prefactor. We make such
measurements for different amounts of randomly positioned ECM (mutant)
as well as for patterned ECM (control) via the removal of ECM from
localized circular regions near the center.
Reporting summary
Further information on research design is available in the [235]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[236]Supplementary Information^ (24.7MB, pdf)
[237]41467_2025_59252_MOESM2_ESM.pdf^ (98.8KB, pdf)
Description of Additional Supplementary Files
[238]Supplementary Data 1^ (220.8KB, xlsx)
[239]Supplementary Data 2^ (78.8KB, xlsx)
[240]Supplementary Data 3^ (1.5MB, xlsx)
[241]Supplementary Data 4^ (21.3KB, xlsx)
[242]Supplementary Data 5^ (125KB, xlsx)
[243]Supplementary Data 6^ (243.4KB, xlsx)
[244]Supplementary Data 7^ (381.1KB, xlsx)
[245]Supplementary Data 8^ (39.6KB, xlsx)
[246]Reporting Summary^ (4MB, pdf)
[247]Transparent Peer Review file^ (652.3KB, pdf)
Source data
[248]Source Data^ (1.8MB, xlsx)
Acknowledgements