Access to this full-text is provided by Wiley.
Content available from Advanced Functional Materials
This content is subject to copyright. Terms and conditions apply.
RESEARCH ARTICLE
www.afm-journal.de
Tumor-On-A-Chip Model Incorporating Human-Based
Hydrogels for Easy Assessment of Metastatic Tumor
Inter-Heterogeneity
Cátia F. Monteiro, Inês A. Deus, Inês B. Silva, Iola F. Duarte, Catarina A. Custódio,*
and João F. Mano*
The coordinated migration of invasive tumor cells is a complex and dynamic
mechanism driven by diverse cellular and molecular events. Unfortunately,
the inherent heterogeneity within tumors raises multiple challenges in
deciphering key biomarkers and novel therapeutic approaches to prevent
tumor metastasis. Here, a microengineered tumor-on-a-chip system
incorporating human platelet lysate hydrogels is proposed to recreate the
early metastatic process of tumor invasion and drug response. By co-culturing
human bone marrow mesenchymal stem cells with two tumor cell lines with
distinct metastatic capability, the developed model can emulate the 3D tumor
microarchitecture and inter-heterogeneity regarding its intrinsic metastatic
ability. The recreated microenvironment supports tumor and stromal cell
movement, evidencing the synergistic tumor-stromal cell and cell-extracellular
matrix interactions of an invading tumor. Through gene and protein
expression analysis and exometabolomic profiling, this tumor-on-a-chip
platform provides evidence for the role of a dynamic environment as a key
regulator of tumor metastatic ability. Additionally, the effect of doxorubicin
treatment on tumor invasiveness and biomarker profile highlights the
suitability of the established models for therapy assessment. Overall, this
study presents a tumor-on-a-chip model useful to pursue mechanistic studies
on early metastatic events in a fully human-derived microenvironment, while
contributing with fundamental insights into biomolecular profiling.
1. Introduction
Cancer aggressiveness is primarily attributed to the rapid growth
and metastatic potential of tumor cells and is often associated
C. F. Monteiro, I. A. Deus, I. B. Silva, I. F. Duarte, C. A. Custódio,
J. F. Mano
CICECO –Aveiro Institute of Materials
Department of Chemistry
University of Aveiro
Campus Universitário de Santiago, Aveiro -, Portugal
E-mail: catarinacustodio@ua.pt;jmano@ua.pt
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/./adfm.
© The Authors. Advanced Functional Materials published by
Wiley-VCH GmbH. This is an open access article under the terms of the
Creative Commons Attribution-NonCommercial-NoDerivs License,
which permits use and distribution in any medium, provided the original
work is properly cited, the use is non-commercial and no modifications
or adaptations are made.
DOI: 10.1002/adfm.202315940
with drug resistance and poor prognosis.
Characterized by the coordinated migra-
tion of tumor cells, metastasis is a complex
mechanism of multi-step cascade events in-
volving tumor growth, invasion, intravasa-
tion, extravasation, and colonization into
distant organs.[]Despite the increasing ef-
fectiveness of therapeutic approaches for
primary tumors, the high ability of some
tumors to disseminate and generate mi-
crometastasis at early tumor stages is a piv-
otal concern in cancer research, accounting
for the majority of cancer-related deaths.[]
The poor recapitulation of tumor patho-
physiological hallmarks, including the in-
trinsic inter- and intra-tumoral heterogene-
ity, has hindered a comprehensive un-
derstanding of key cellular and molec-
ular features of the metastatic cascade.
Recent advances in bioengineered D
in vitro tumor models relying on cell
spheroids, patient-derived organoids,
extracellular matrix (ECM)-mimicking
hydrogels, and porous scaolds have
undoubtedly contributed to unraveling
intricate biomolecular signatures of tu-
mor progression.[,]Several studies have
reported the suitability of tumor spheroids as high-throughput
platforms to screen and validate new therapies since they closely
recapitulate tumor biochemical gradients (oxygen, nutrients,
pH).[,]Furthermore, embedding tumor spheroids in polymeric
hydrogels has sought to provide the physical confinement en-
countered in vivo, allowing the study of more complex mecha-
nisms like the ones required for tumor growth and invasion. De-
spite the widespread application of reconstituted basement mem-
brane and type I collagen hydrogels in this context, we previ-
ously demonstrated the feasibility of methacryloyl platelet lysate
(PLMA) hydrogels as an alternative approach for studying tumor
progression in a fully human-derived matrix.[,]Besides support-
ing tumor spheroid growth and invasion,[]the tumor-stromal
cell crosstalk was recapitulated,[]featuring the recognized pro-
tumorigenic activity of human bone marrow mesenchymal stem
cells (hBM-MSCs) upon recruitment by tumor cells.[]Indeed,
accumulating evidence has revealed the key role of the multicel-
lular stromal population, known to undergo fundamental mor-
phological and functional changes to prompt tumorigenesis.[]
Adv. Funct. Mater. 2024,34, 2315940 (1 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 1. Concept of a pathologically relevant tumor-on-a-chip model to study tumor aggressiveness. a) Schematic illustration of the in vivo tumor
pathophysiology and representation of the cellular and microenvironmental aspects involved in the modulation of the migratory ability of an invasive
tumor: tumor inter-heterogeneity (patient-to-patient variability of tumor genotype and phenotype) and features of the surrounding microenvironment,
including tumor-stromal cell biomolecular signaling, ECM remodeling, tumor cell invasiveness ability, fluid flow-induced shear stress, and tumor metas-
tasis. b) Design of the micro-compartmentalized tumor-on-a-chip for the incorporation of a hydrogel (PLMA-based hydrogel) in a tissue channel (yellow),
separated by a perfusable channel (pink) through a micropatterned rail (dark gray). The side view representation demonstrates the recreation of the
tumor microenvironment and its features by co-culturing a tumor spheroid surrounded by dispersed hBM-MSCs. c) Overview of the tumor-on-a-chip
experiments timeline. Samples cultured under dynamic conditions were connected to a continuous fluid perfusion system (syringe pump) at day . All
microfluidic chips, cultured in static or dynamic settings, were prepared similarly.
However, important physical metastatic drivers are still missing
to fully recapitulate the microenvironment that cells experience
during tumor progression.[]
Microphysiological systems have emerged as a promising tool
to provide fundamental insights into the biochemical and me-
chanical mechanisms underlying tumor progression and drug
resistance.[]While allowing intricate tumor-stromal cell inter-
actions, these microfluidic devices have the potential to address
the need of incorporating dynamic flow to study events such as i)
tumor neovascularization,[]ii) tumor cell intravasation[, ]and
extravasation,[]iii) immune cell infiltration,[ ]and (iv) drug de-
livery strategies.[,]By integrating the spheroid culture in mi-
crofluidic chips, the eect of the flow on tumor size and drug
sensitivity has also been addressed.[]In addition, some reports
have demonstrated the feasibility of tumor-on-a-chip to inves-
tigate organ-specific dissemination and colonization of tumor
cells in metastatic sites.[,]Despite significant insights, most of
these studies relied on tumor-endothelial cell co-culture to study
late metastatic steps, disregarding the importance of having a D
spheroid-like tumor mass and an ECM-mimicking matrix with
relevant biomechanical properties to study the early steps of the
metastatic cascade during which tumor cells acquire invasive-
ness abilities.
Motivated by current limitations, a microengineered
biomimetic model was herein developed in an attempt to
mimic the cellular and molecular mechanisms underlying the
distinct invasion and metastatic ability of tumor cells (Figure 1a).
Leveraging the ability of PLMA-based hydrogels to support
tumor spheroid invasion in a humanized microenvironment,
two osteosarcoma cell lines were used to produce spheroids with
dierent metastatic capabilities (MG-: low-metastatic, and
B: high-metastatic) and co-encapsulated with hBM-MSCs
in PLMA hydrogels under dynamic conditions. The so-called
tumor-on-a-chip model provided a suitable in vivo-like D
microenvironment to recreate: i) the eect of the fluid flow
in tumor invasion, ii) the dierent tumor invasiveness ability,
iii) tumor-stromal cell crosstalk, iv) cell-ECM interaction, and
v) tumor invasion-associated biomarkers. We found that the
Adv. Funct. Mater. 2024,34, 2315940 (2 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
continuous perfusion improved the invasiveness ability of the
metastatic tumor cells. Furthermore, by gene and protein expres-
sion analysis, we discovered that the dynamic culture provided
an appropriate microenvironment to recreate the biomarker
profile of tumors with high metastatic ability, where lower levels
of collagen and higher levels of vascular endothelial growth
factor (VEGF) expression are verified. Notably, we were also
able to characterize the exometabolomics profile of the estab-
lished models. The tumor responsiveness to an anti-cancer drug
(doxorubicin (DOX)) highlighted the potential of these models
for drug screening purposes. Altogether, we demonstrated the
suitability of the established tumor-on-a-chip models to recreate
the tumor’s distinct metastatic ability and biomarker profile.
2. Results
2.1. Design and Production of a Microfluidic Device Tailored
for Tumor Invasiveness Studies
The microphysiological model herein developed was specifically
tailored to recreate the D microarchitecture and the cellular and
microenvironmental aspects involved in the modulation of the
migratory ability of an invasive tumor (Figure ). The device was
fabricated with two PDMS layers: a microstructured bottom layer
with patterned channels and an upper layer for fluid and hydro-
gel inlet and outlet (Figure Sa, Supporting Information). The
bottom layer consists of a single fluid channel separated from a
tissue channel by a micropatterned rail protruding from the bot-
tom of the device (Figure b; Figure Sb, Supporting Informa-
tion). The tissue channel was designed envisioning the encapsu-
lation of physiologically relevant tumor spheroids and the study
of its invasiveness ability within the hydrogel (for detailed repre-
sentation and dimensions, see Figure S and Table S, Support-
ing Information).
To demonstrate the feasibility of the designed device to gen-
erate a spatially confined D scaold to mimic the tissue ECM,
the PLMA precursor solution was injected into the empty device
through the tissue channel inlet, photopolymerized with light,
and perfused with cell culture medium (Figure 2a,b; Movies S
and S, Supporting Information). A controlled injection of the hy-
drogel solution along the tissue channel was possible due to the
inclusion of the microfabricated rail outlining the tissue channel,
which is higher in curvature to allow the complete fill and hydro-
gel confinement without spillage to the media channel (Figure
Sb, Supporting Information). As microscopically observed in
the device cross-section, the surface tension generated was suf-
ficient to prevent the hydrogel from overflowing into the media
channel, a phenomenon known as capillary pinning, while allow-
ing maximization of the contact area between the hydrogel and
the media for an improved molecular diusion (Figure Sd,Sup-
porting Information).[]
2.2. PLMA Hydrogel Permeability in the Tumor-On-A-Chip
To address the suitability of the PLMA hydrogel to recreate tis-
sues’ permeability, as well as study the influence of a continuous
fluid flow in that feature, permeability assays were performed.
The diusive transport of both nutrients and oxygen is expected
to be ensured by the high porosity of the PLMA hydrogel, even
at the hydrogel-air surface (Figure S, Supporting Information).
To further confirm this permeability within this human-derived
hydrogel and assess the importance of having a continuous fluid
flow, the penetration of kDa fluorescein isothiocyanate (FITC)-
labeled dextran, was analyzed in static and dynamic conditions.
The FITC-dextran diusion monitored for h by confocal mi-
croscopy indicated that the hydrogel structural properties allow
an appropriate diusion of nutrients (Figure c). As expected
for a highly permeable material, FITC-dextran diusion through
the hydrogel is higher in dynamic culture conditions, highlight-
ing the importance of a dynamic environment to provide proper
nutrient availability (Figure d; Figure Sa, Supporting Informa-
tion).
Dense ECM is also a barrier to drug transport through the ex-
travascular spaces and has been associated with increased drug
resistance. To evaluate the diusion of DOX into the PLMA hy-
drogels, a similar permeability assay was performed, taking ad-
vantage of the intrinsic fluorescence of DOX. The fluorescence
intensity measurement indicates the diusion and accumula-
tion of the anthracycline drug within the hydrogel (Figure e,f;
Figure Sb, Supporting Information). Interestingly, an increase
in hydrogel fluorescence intensity above the one corresponding
to DOX in the fluid channel suggests a DOX entrapment inside
the hydrogel. Actually, limited drug availability for tumor cells is
not only related to drug diusivity inside the ECM but also to its
binding to ECM components.[,]In this context, it is hypothe-
sized that DOX interacts with PLMA components, primarily with
human serum albumin, the predominant protein in PLMA. Hu-
man serum albumin demonstrates a high anity for binding to
DOX in blood plasma and has been extensively investigated as a
potential nanocarrier for delivering drugs to tumor tissues.[,]
2.3. Modeling Tumor Cell Invasiveness Ability in a
Tumor-On-A-Chip
Aiming the development of a relevant tumor model to study
the cellular and molecular mechanisms underlying the dier-
ent aggressiveness of tumor cells, a low-metastatic and a high-
metastatic model were established. For that, two OS cell lines
with distinct metastatic ability, MG- (low-metastatic) and B
(high-metastatic), were co-cultured with hBM-MSCs. OS tumor
spheroids were assembled by self-agglomeration in ultra-low ad-
hesion plates. A cell density of cells per spheroid gener-
ated robust tumor microtissues of a suitable size for in-chip cul-
ture. Moreover, those spheroids developed a necrotic core by day
of formation (Figure S, Supporting Information), enabling
a faithful recreation of poor-vascularized and hypoxic regions
of in vivo tumors. Indeed, these regions are of particular in-
terest to better study the importance of that microenvironment
on the early events involved in the metastatic cascade and drug
sensitivity.[]
To further mimic the tumor microenvironment, tumor
spheroids and hBM-MSCs were suspended in PLMA hydrogel
and injected into the tissue channel of the microfluidic chip. Tu-
mor and stromal cells were transduced with lentivirus expressing
red (MG--RFP and B-RFP) and green (hBM-MSCs-GFP)
Adv. Funct. Mater. 2024,34, 2315940 (3 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 2. Molecule permeability in the designed tumor-on-a-chip platform. a) Sequential steps of model preparation: PLMA-based hydrogel precursor
solution injection through tissue channel inlet, followed by hydrogel photo-polymerization and fluid perfusion (Movie S, Supporting Information). b)
Time-lapse images of cell culture medium (colored red) diusion inside the PLMA hydrogel, perfused at μLmin
−for min (Movie S, Supporting
Information). Confocal imaging of the (c,d) FITC-Dextran ( kDa) and (e,f) DOX permeability inside the PLMA hydrogel over min, under static
and dynamic conditions. c,e) Time-lapse images reveal the rail (right side) and the hydrogel (left side), separated by a dashed line, throughout the
molecule diusion experiment. d,f) Relative molecule concentration in the hydrogel (tissue channel width of μm) after min of static and dynamic
incubation. Scale bar: μm.
fluorescent proteins, respectively, in order to visualize their
organization in the established co-culture models and assess
their crosstalk during the days of culture. High-resolution con-
focal fluorescence imaging revealed the successful spatial or-
ganization of the cell populations, recapitulating the high cell
density mass occupied by the forming tumor and the neigh-
boring stromal cells, toward the development of a suitable tu-
mor microenvironment (Figures S and S, Supporting Infor-
mation). Throughout the days of culture, both tumor cell types
invaded the PLMA hydrogel and directly interacted with the sur-
rounding stem cells, although B cells showed higher inva-
siveness and metastatic ability compared with MG- (Figure 3a;
Figure S, Supporting Information). These findings corroborate
the distinct metastatic ability of these tumor cell lines as re-
ported elsewhere[,]demonstrating that the established D co-
culture model is a relevant platform to faithfully recapitulate tu-
mor inter-heterogeneity aggressiveness and tumor-stromal cell
interactions.
Adv. Funct. Mater. 2024,34, 2315940 (4 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 3. Tumor cell invasiveness ability and proliferation in the tumor-on-a-chip model. a) Representative confocal images of the device cultured for
days under static and dynamic conditions, including the DOX treatment condition, containing a region rich in hBM-MSCs () and a central region
where the tumor spheroid is located (). Low-metastatic (MG-) and high-metastatic (B) tumor cell lines were transfected with RFP (orange) and
hBM-MSCs were transfected with GPF (green). Scale bar: μm. b,c) Quantification of the (b) viability and (c) proliferation of the whole cellular
population in the models. d) Tumor cell viability of the (i) low-metastatic and (ii) high-metastatic model in each culture condition. Luciferase-transfected
tumor cells were used for this quantification. Fold change in tumor cell viability between static and dynamic control conditions is represented. Data are
presented as mean ±SD (n≥). Statistical significance between low- and high-metastatic models and within each model is represented with a solid or
dashed line, respectively. *p<.; **p<.; ***p<.; **** p<..
Adv. Funct. Mater. 2024,34, 2315940 (5 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
2.3.1. The Influence of the Dynamic Environment on Tumor
Metastatic Ability
Aiming to evaluate the role of the dynamic fluid cues on tumor
cell proliferation and metastatic ability, thetumor-on-a-chip mod-
els were cultured under static and dynamic conditions. By fluo-
rescence imaging, no evident dierence in tumor invasion was
found in the low-metastatic model over the days of culture, con-
trasting with the clearly enhanced invasion of B cells (high-
metastatic) cultured under continuous cell culture medium per-
fusion (Figure a;FiguresS and S, Supporting Information).
This increased invasiveness capacity, evident as early as day of
the assay, allowed the tumor cells to extensively invade the PLMA
hydrogel and colonize the hydrogel-fluid interface, proliferating
and migrating through the perfused channel.
Live/dead staining of the low-metastatic model revealed a
lower cell death and higher cell viability near hydrogel borders,
whereas morphology analysis by nucleus and actin filaments
staining demonstrated the great interconnected network estab-
lished between tumor and stromal cells (Figures S and S,Sup-
porting Information). Unfortunately, it was not possible to obtain
meaningful microscopy images for live/dead and morphological
staining in the high-metastatic model. The high cell density and
metabolic activity of the B cells revealed to be a challenge to
properly stain the actin filaments of cells inside the hydrogel.
This also resulted in calcein metabolization mainly by the cells
in the hydrogel periphery and perfused channel.
In order to support these imaging findings, the metabolic ac-
tivity and proliferation of the models were quantified. Results
showed a slight increase in cell viability and proliferation of
the low-metastatic model when cultured in a dynamic setting
(Figure b,c). To understand if these findings were related to tu-
mor proliferation, tumor cells were transduced with firefly lu-
ciferase protein and this protein was quantified after days in
culture (Figure d). Higher luciferase expression (.-fold) in
the dynamic low-metastatic model revealed that the continuous
fluid flow prompted tumor cell proliferation, although no evi-
dent dierence was found in terms of tumor cell invasion. Re-
garding the high-metastatic model, a very significant increase in
the metabolic activity and proliferation of the entire model was
verified. Moreover, an .-fold increase in luciferase expression
strengthens the information provided by fluorescence imaging,
evidencing the higher ability of this aggressive tumor cell line
(B) to migrate and proliferate in a meaningful microenviron-
ment.
2.3.2. Dynamic Tumor-On-A-Chip as an Effective Model for Drug
Screening
Once confirmed the significance of a dynamic culture, we sought
to study the tumor-on-a-chip models’ responsiveness to DOX,
a standard-of-care anti-cancer drug known to act on prolifera-
tive cells inducing cell cycle arrest and cell death. To this end,
after days in culture, both models were treated with . μ
of this chemotherapeutic drug for h and then maintained in
culture for an additional days with fresh cell culture medium
(Figure c). This DOX concentration was chosen based on the
IC-value obtained for the free-spheroid culture in our previous
study.[]As demonstrated by luciferase expression quantification,
MG- cell viability was not significantly aected by DOX treat-
ment (Figure d). Nevertheless, the overall cell metabolic activ-
ity and proliferation ability on the model was slightly decreased
when compared to the control, as also observed by live/dead
staining, suggesting that DOX may have aected the cell viabil-
ity, particularly that of hBM-MSCs. Such drug sensitivity could
be triggered by an accumulation of DOX in the tumor neighbor-
hood, potentially explaining the partial death of stromal cells lo-
calized closer to the tumor cells. Indeed, the drug resistance ex-
hibited by MG- tumor cells can be related to the development
of drug resistance mechanisms dependent on reduced uptake or
increased eux.[,]
Regarding the high-metastatic model, the tumor cells exhib-
ited increased sensitivity to DOX, noticeable as soon as h af-
ter completing the DOX treatment (Figure ; Figure S,Sup-
porting Information). This could be attributed to the higher
metabolic activity of the B tumor cells, potentially contribut-
ing to an increased DOX metabolism. The tumor spheroid com-
pactness is also hypothesized to be associated with this drug sen-
sitivity, as increasing evidence has been directly correlating the
spheroid compactness and ECM deposition with their therapeu-
tic resistance.[]In fact, MG- cell aggregation resulted in a very
compact spheroid maintained during the days of culture, while
B cells generated tight aggregates which rapidly spread inside
the PLMA hydrogel, losing their aggregated structure.
2.4. Tracking Cell Movement During Tumor Invasion
Live imaging analysis of the dynamic models was performed by
confocal microscopy using fluorescent protein-expressing cells
to evaluate the cellular distribution in the hydrogel and charac-
terize tumor cell invasiveness profile and stromal cell spatial ar-
rangement (Figure 4a; Movies S and S, Supporting Informa-
tion). Images acquired every hour throughout the first days of
culture revealed that PLMA-encapsulated hBM-MSCs were able
to stretch inside the hydrogel within h, and directly interacted
with the tumor cells. Regarding the tumor invasion, both tumor
spheroids were able to invade the surrounding ECM-like matrix.
Interestingly, high-metastatic tumor cells (B) cultured in dy-
namic conditions were able to reach the fluidic channel as early as
day of culture, rapidly migrating throughout that channel, con-
tributing to an increased in-gel tumor cell migration compared
with the static culture (Figures a and a,b).
Aiming to further investigate stromal cell migration during
tumor invasion, hBM-MSCs movement was monitored during
the days of live imaging (Figure c,d). Results revealed that
the movement speed and distance of hBM-MSCs in the low-
metastatic model increased on day , being significantly more
motile in this model compared with the high-metastatic model.
Normalized migration cell tracks demonstrated that hBM-MSCs
were able to migrate in all directions in the low-metastatic model
whereas, in the high-metastatic model, they were restricted to the
unidirectional movement, possibly due to the rapid in-gel inva-
sion of the tumor cells in parallel to the tissue channel direc-
tion. Moreover, the D representation of the stromal cell paths
revealed that some hBM-MSCs migrated toward the tumor, while
other cells moved away from the tumor. The greater tendency of
Adv. Funct. Mater. 2024,34, 2315940 (6 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 4. Tracking of tumor and stromal cell movement during tumor invasion. a) Time-lapse images of the h confocal live imaging experiments
(Movies S and S, Supporting Information) revealing RFP-expressing tumor cell invasion into the surrounding microenvironment and GFP-expressing
hBM-MSCs stretching inside the PLMA hydrogel, around the tumor spheroid. i,ii) Close proximity of tumor and stromal cells at h of culture reveals
their juxtacrine interaction during tumor progression. Scale bar: μm. b) Representative images of dierent regions of the high-metastatic tumor
model at days of culture demonstrating B-RFP tumor cells migrating along the bottom of the microfluidic chip (region ) (vertical dark regions
correspond to the microstructured rail, middle region is the tissue channel, and lateral regions correspond to the medium channel). Tumor cell migration
through the outlet side of the perfusable channel reveals tumor metastization inside the chip (region ), which remains viable (region ). Calcein AM:
viable cells, PI: dead cells. Scale bar: μm. c) Quantification of the hBM-MSCs movement (i) distance and (ii) speed every h and over the h
of the live imaging experiment. d) Representation of the hBM-MSCs (green) movement along the live imaging in the (i,ii) low-metastatic and (iii,iv)
high-metastatic models. (i,iii) Normalized cell tracking paths of hBM-MSCs (each color represents a single cell). The x-andy-axes correspond to the
cell movement in parallel or perpendicular to the tissue channel. (ii, iv) D representation of the stromal cell paths, being arrowhead indicating the
h of live imaging. The confocal image corresponds to h of the experiment, and tumor cells were colored in grey to facilitate cell path observation.
Orange paths represent the hBM-MSCs that migrated toward the tumor or whose migration did not contribute to their movement away from the tumor.
Light orange paths represent cells moving away from the tumor. Scale bar: μm. Data are presented as mean ±SD (n≥). *p<.; **p<.;
***p<.; **** p<..
Adv. Funct. Mater. 2024,34, 2315940 (7 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
hBM-MSCs to move in the opposite direction to the tumor in
the high-metastatic model can be attributed to the search for re-
gions with higher nutrient availability, as metastatic tumor cells
are known to rely on an increased glucose-dependent glycolytic
metabolism.[]These findings highlight that the established tis-
sue compartment allows the maturation of a stromal microenvi-
ronment suitable to recapitulate tumor invasion ability while of-
fering an appropriate environment for hBM-MSCs to exhibit an
elongated phenotype, as evidenced by their well-defined actin fil-
aments forming lamellipodia-based protrusions (Figure S,Sup-
porting Information).
2.5. Influence of the TME and Tumor Metastatic Ability on
Biomarker Expression
The clinical heterogeneity of a tumor is characterized by the
dierential expression of specific biomarkers involved in dis-
ease progression, metastasis, and resistance to chemotherapeu-
tic drugs.[]To further validate the potential of this tumor-on-
a-chip model to emulate in vivo tumor aggressiveness, the gene
and protein expression of key biomolecular markers involved in
tumor invasiveness and metastatic ability were evaluated. The
influence of the dynamic environment and DOX treatment on
those biomarkers was also characterized.
The quantification of the transcripts for VEGFA and COLA
in the low-metastatic model revealed that the dynamic environ-
ment induced a decrease in VEGFA and an increase in COLA
expression, while an opposite eect was verified for the high-
metastatic model (Figure 5a). In fact, when comparing the es-
tablished models in each culture condition (static vs dynamic),
it was verified that only the dynamic culture oered a suit-
able microenvironment to reproduce the increased matrix de-
position and neovascularization signaling of low-metastatic and
high-metastatic tumors, respectively (Figure b).[]VEGF pro-
tein quantification by ELISA revealed an increase in VEGFA se-
cretion over time, which is in line with the VEGFA gene expres-
sion profile (Figure c; Figure S, Supporting Information). In
addition to comparing static versus dynamic conditions, we as-
sessed the impact of DOX treatment on the expression of these
biomarkers. Although a non-significant increase was verified in
VEGFA gene expression for both models, ELISA analysis re-
vealed that DOX prompted a higher VEGFA production. Regard-
ing the COLA, drug treatment induced a decreased expression
in the low-metastatic model and an increased expression in the
high-metastatic model. These findings suggest that DOX exerts
a tumor phenotype-dependent deregulatory eect. Furthermore,
the invasive capacity of tumor cells in the surrounding microenvi-
ronment is largely enhanced by extracellular space remodeling, a
mechanism where matrix metalloproteinases (MMPs) play a crit-
ical role.[]On this line, MMP- and − gene expression analy-
sis demonstrated that the dynamic environment prompted a de-
crease in MMP- expression in both models and an increase in
MMP- expression in the high-metastatic model (Figure a). It
is worth noticing that MMP- gene expression was not detected
in the low-metastatic model. Moreover, tumor models’ compari-
son revealed a higher MMP- and lower MMP- gene expression
in the low-metastatic model (Figure b). Regarding the eect of
drug treatment, data show that DOX reduced the expression of
the MMPs with higher transcription levels in the dynamic control
(Figure a).
Ezrin staining of the herein established low-metastatic model
revealed an increased expression of this protein in the dy-
namic setting compared with the static, while a reduced expres-
sion was verified in the DOX-treated model (Figure d). In the
high-metastatic model, the same was not verified in the hydro-
gel region. Instead, ezrin expression was observed only in the
cells on the hydrogel periphery or the ones that migrated to
the perfused channels, except for the dynamic control culture
(Figure S, Supporting Information).
2.6. Exometabolomics Profiling
Having demonstrated that the dynamic environment induced a
phenotypical and genotypical profile characteristic of in vivo tu-
mor aggressiveness, H NMR exometabolomic analysis was ex-
plored to study the metabolic profile of the developed in vitro
models and to evaluate the influence of fluid flow and drug treat-
ment on specific biomarkers. First, a detailed examination of the
consumed and secreted metabolites that define the metabolic
state of each model (low- versus high-metastatic), cultured in
static conditions, was carried out (Figure 6a). Exometabolomic
analysis of the euent collected from the devices on days
and indicated the consumption of glucose and several amino
acids and the secretion of lactate, consistent with the well-known
reliance of tumor cells on high glycolytic activity. Additionally,
changes in metabolite consumption/secretion detected over time
emphasize the impact of the tumor invasive events and cell-cell
crosstalk in the metabolomic profile of cellular populations in the
tumor microenvironment. In particular, the overflow of lactate
that characterizes the high glycolytic flux was attenuated from
day to day of culture.
Since the dynamic environment was already demonstrated to
oer the ideal conditions to recreate the tumor aggressiveness
and its respective biomarkers, the metabolite secretion pattern of
dynamic models was studied (Figure a). It is important to note
that the metabolite consumption in the dynamic setting was not
analyzed since the dierences in the levels of consumed metabo-
lites were not significant, due to the continuous supply of culture
medium. Among the secreted metabolites, the results showed
that the higher aggressiveness of the high-metastatic model cul-
tured in dynamic conditions was translated into an increased gly-
colytic flux, displaying higher levels of lactate compared with the
low-metastatic model. Alongside lactate, an increased secretion
of acetate and formate was also observed in the metastatic model.
The tumor metabolic response to DOX was assessed by mea-
suring lactate secretion in both models at day ( h after start-
ing DOX) and at day ( days after finishing DOX treatment),
to characterize the temporal dependence of cell drug sensitiv-
ity (Figure b). While a significant increase in lactate between
non-treated and DOX-treated was observed in the low-metastatic
model at day , the production of this metabolite stagnated at day
, possibly as a response of tumor cell resistance to DOX. On the
contrary, in the high-metastatic DOX-treated model, a significant
decrease in lactate secretion was observed on day and accen-
tuated on day , which corroborates the high sensitivity of B
cells to DOX treatment.
Adv. Funct. Mater. 2024,34, 2315940 (8 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 5. Gene and protein biomarkers expression depending on tumor metastatic ability. a) Relative gene expression of VEGFA, COLA, MMP, and
MMP in the (i) low-metastatic and (ii) high-metastatic models for the dierent culture conditions, normalized to the static culture. b) Comparison
of low- and high-metastatic relative gene expression of (i) VEGFA, (ii) COLA, and (iii) MMP, for each culture condition. c) ELISA quantification of
VEGFA protein secretion at days , , and . Statistical significance between dierent culture conditions, for the same time-point, is represented with
a solid line. Unless otherwise stated, time-dependent significance in each culture condition and model is statistically dierent. d) Confocal imagesof
Ezrin expression (green) in the low-metastatic model, at days of culture. Actin filaments and nuclei are stained in red and blue, respectively. Scale bar:
μm. Data are presented as mean ±SD (n≥). *p<.; **p<.; *** p<.; ****p<., ns: not significant.
Adv. Funct. Mater. 2024,34, 2315940 (9 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Figure 6. Exometabolomic profiling of the established tumor-on-a-chip models. a) Heatmap representing the variations in metabolite consump-
tion/secretion in low- and high-metastatic models, at days and of static and dynamic control cultures. b) Boxplots of concentration levels (scaledto
unit variance) of lactate in the (i) low- and (ii) high-metastatic models, at day ( h after starting DOX treatment) and day ( h after finishing DOX
treatment). Data are presented as mean ±SD (n≥). *p<.; **p<.; *** p<.; ****p<..
3. Discussion
The coordinated migration of invasive tumor cells, a well-known
hallmark of cancer, is driven by a variety of cellular and molecular
events closely related to the complexity and dynamic nature of its
surrounding microenvironment.[]Despite the undeniable con-
tribution of D in vitro tumor models to unravel tumor biomark-
ers, the bioengineering of relevant models to investigate tumor
cell invasion mechanism, while recapitulating the microenviron-
ment biophysical cues and tumor-stroma cell crosstalk, remains
in high demand. Envisioning the recreation of the D tumor mi-
croarchitecture and inter-heterogeneity in a dynamic setting, a
microfluidic device was tailored to allow the synergistic interac-
tion of cellular and biophysical components of the tumor mi-
croenvironment (ECM, stromal cells, and fluid flow).
Merging the biomaterial science advancements with the un-
questionable potential of organ-on-a-chip technology, human
PLMA-based hydrogels demonstrated to be suitable for integra-
tion into the designed microfluidic chip. The integration of an
ECM-mimicking hydrogel recapitulates tissues’ tridimensional-
ity, enabling to study more complex physiological and patholog-
ical events compared with devices integrating a porous mem-
brane sandwiched between vertically stacked compartments for
vasculature-tissue interface studies.[,]While protecting cells
from constant shear forces, hydrogels’ porosity enables the es-
tablishment of an interstitial fluid flow known to generate a gra-
dient of soluble factors, modulating tumor invasion and drug
response.[,–]In fact, PLMA hydrogel permeability experi-
ments confirmed its ability to recapitulate the natural diusion
of nutrients from neighboring vascular vessels to the tumor or
tissue-resident cells, especially under dynamic conditions. Ad-
ditionally, a DOX permeability assay suggested an interaction
between the hydrogel matrix components and DOX – a phe-
nomenon that has been implicated in the limited drug availabil-
ity for tumor cells in vivo.[,]Importantly, the mechanical sta-
bility of the PLMA hydrogel over the conventionally used phys-
ically crosslinked hydrogels (reconstituted basement membrane
and collagen) highlights its feasibility for conducting physiolog-
ical studies under prolonged flow conditions spanning several
days.
As a highly heterogeneous disease, cancer can exhibit a wide
range of phenotypes closely correlated with its metastatic capa-
bility, even when originating from the same tissue.[]Such tu-
mor inter-heterogeneity raises multiple challenges in decipher-
ing key biomarkers, signaling pathways, and novel therapeutic
approaches to prevent tumor metastasis. By co-culturing hBM-
MSCs with two OS cell lines expressing distinct metastatic abil-
ity, this study recreates the tumor inter-heterogeneity at the cel-
lular and molecular levels. Here, the significantly higher inva-
siveness reported for the B cells was observed, particularly
when cultured under continuous fluid perfusion.[,]In these
conditions, this highly metastatic cell line was even able to col-
onize the hydrogel-fluid interface, proliferating and migrating
through the perfused channel – a behavior that can be attributed
to its higher potential to generate spontaneous metastasis
in vivo.[]
Adv. Funct. Mater. 2024,34, 2315940 (10 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
Although MSCs have a pivotal role in the homeostasis of sev-
eral tissues, these cells are known to be recruited to the in vivo
tumor microenvironment and educated by tumor cells to medi-
ate the process of tumor progression and metastasis.[,]As re-
vealed by confocal microscopy live imaging, the incorporation of
hBM-MSCs dispersed around the tumor allowed the recreation
of the tumor-stromal cell juxtacrine interactions. The paracrine
crosstalk between tumor cells and nearby hBM-MSCs, previously
demonstrated in a static tri-culture OS model[]was herein re-
flected by the hBM-MSCs movement inside the PLMA hydro-
gel. Regarding therapy response, the high-metastatic tumor cells
demonstrated an increased DOX sensitivity, a cell behavior that
is corroborated by in vivo studies that reveal perivascular localiza-
tion of DOX and limited drug penetration into more compact re-
gions of the tumor, including hypoxic and adjacent regions.[]In-
deed, clinical studies have been combining DOX with new drug
carriers and/or extracorporeal devices to circumvent this perme-
ability issue and improve DOX pharmacokinetics.[,]These ob-
servations further emphasize that the micro-compartmentalized
chip is capable of mimicking the spatial arrangement of tumor
cells and stromal elements, oering a relevant D environment
to study cell-cell and cell-ECM interaction as well as eective ther-
apy assessment.
The development of tumor models that emulate in vivo tu-
mor aggressiveness is expected to advance our understanding of
fundamental mechanisms such as chemoresistance, biomarker
expression, and metabolic reprogramming. Our results demon-
strate that cells engage in metabolic reprogramming during tu-
mor cell invasion and interaction with stromal cells, being highly
dependent on the microenvironmental conditions. Gene and pro-
tein expression analysis revealed an inverse correlation between
the expression of collagen and VEGFA. Such results were con-
sistent with previous studies linking low-metastatic tumors to
abnormal secretion of collagen by stromal and tumor cells,[]
and high-metastatic tumors to increased levels of VEGFA, a key
mediator of tumor neovascularization and therefore of tumor
metastasis.[]Moreover, MMP- and MMP- gene expression
was found to be correlated with the metastatic capability of the
established models.[,,]Exometabolomic analysis conducted
in the static culture models showed the cell’s metabolic activity
is modulated over time. The emergence of lactate overflow por-
trays a shift toward enhanced glycolytic activity that features the
active role of tumor-invasive events in the tumor microenviron-
ment metabolic interplay. Indeed, the higher expression of gly-
colytic enzymes, including lactate dehydrogenase A, was found
to be particularly correlated with increased aggressiveness in dif-
ferent cancers[,]and in xenograft models.[ ]While limited
information on metabolite consumption was obtained for the
dynamic culture, metabolite secretion data showed significantly
higher levels of acetate and formate in the high-metastatic model,
attributed to the increased proliferative and invasiveness capac-
ity of this aggressive model.[–]Finally, lactate secretion read-
outs at dierent time-points after DOX treatment demonstrated
that this dynamic model can provide key insights into the time-
dependent drug sensitivity, a factor often underrated in treat-
ment performance evaluation.[]The intensified response from
the high-metastatic model is related to the high DOX cytotoxi-
city observed in this model, which hampered cell proliferation
and tumor growth, a behavior verified in an in ovo metabolomics
phenotyping of invasive breast cancer.[]The opposite eect ob-
served for the low-metastatic model was previously reported for
a DOX-resistant cell line, suggesting that MG- cells were able
to develop a mechanism of defense supported by an enhanced
glycolytic flux.[]
Although further studies are needed to address the metabolic
features of physiological and pathological models in dynamic
conditions, the gene and protein expression as well as the
metabolite secretion here reported emphasized the crucial role
of the dynamic environment to better mimic tumor aggressive-
ness in D in vitro tumor models. By closely resembling the gene
and secretome profile of a metastatic tumor and its response to a
standard-of-care drug, the established models allowed the recre-
ation of the metabolic states of tumors with distinct aggressive-
ness. In this context, metabolomic profiling can be a uniquely
valuable strategy to identify new biomarkers of tumor survival
mechanisms.
Encouraged by the potential of microphysiological systems to
recreate physiological and pathological processes, the herein es-
tablished tumor-on-a-chip model incorporating human PLMA
hydrogel represents a significant advance toward a more com-
prehensive analysis of tumor inter-heterogeneity as well as the
deconstruction of fundamental tumor hallmarks. Despite that,
further improvements can be made toward a more biologically
complex and predictive model. Considering the active tumor-
endothelial cell crosstalk during tumor progression, the incor-
poration of endothelial cells in the perfused channels can be an
important step to more closely mimic the role of vasculature
in tumor invasion and metastasis. Besides acting as a cellular
barrier during tumor invasion, endothelial cells are modulated
by tumor cells to support tumor invasion.[]Taking advantage
of the dynamic culture, future studies should also investigate
the role of immune cells in these early metastatic events, and
eventually explore them as therapeutic agents.[,]From another
perspective, coupling the established metastatic tumor model to
a downstream device recapitulating one of the most preferable
metastatic sites of the tumor in a study is of utmost interest to
unravel the mechanisms and implications of secondary tumor
formation. Actually, only a few studies have been investigating
the preference of tumor cells to migrate to dierent tissues.[,]
4. Experimental Section
Microdevice Design and Fabrication:The bottom layer of the tumor-on-
a-chip microfluidic device was designed in Autocad software (AutoDesk,
USA), comprising a central channel with an inlet where the hydrogel pre-
cursor solution is introduced, and an adjacent channel for fluid perfusion
that surrounds the hydrogel channel (channels’ dimensions and respec-
tive volumes are represented in Figure S, Supporting Information, and
described in Table S, Supporting Information). Hydrogel and fluid chan-
nels are separated by microfabricated rails protruding from the bottom
of the device. The master mold was fabricated through micro-milling on
an acrylic plate and then the tumor-on-a-chip device was fabricated us-
ing the standard soft-lithography microfabrication technique. PDMS base
(Sylgard , Dow Corning) was thoroughly mixed with curing agent at a
: ratio (w/w), poured on the top of the master mold, and polymerized
overnight in the oven at °C. To fabricate the final PDMS microfluidic
devices, the PDMS positive mold was deeply cleaned, and the surface was
fluorinated using air plasma ( W, min, Diener Electronic, Germany)
followed by trichloro(H,H,H,H-perfluorooctyl)silane (Sigma–Aldrich,
USA) vapor in a desiccator for h. The PDMS mixture was poured on the
Adv. Funct. Mater. 2024,34, 2315940 (11 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
positive molds, thermally cured at °C for h, and peeled o. For the up-
per layer of the microdevice, a ≈ mm layer of PDMS was produced con-
taining lateral input ports for tubing. Biopsy punches of . and . mm
were used to make a hole for hydrogel injection and two holes to serve as
solution reservoirs for analysis purposes and in-chip bubble traps to pre-
vent bubble entrapment inside the fluid channel, respectively. The bottom
and upper layers of the microfluidic device were treated with air plasma
( W, min) and aligned for permanent bonding. The . mm holes were
closed using squared thin layers of PDMS bonded using the same plasma
treatment. TwoTygon tubes (. mm inner diameter, Darwin Microfluidics,
France) were sealed to the top edges of the device with PDMS, and used
to connect the fluid perfusion channel to the pumping system for continu-
ous nutrient/drug supply, as well as cell metabolic/wasteproduct recovery.
Before use, the fully assembled microdevice was flushed and immersed in
% ethanol for min, dried in the oven at °C, and stored under sterile
conditions until further use.
SOLIDWORKS Flow Simulation CAD software (SOLIDWORKS,
USA) was used to perform the computational fluid dynamics modeling
and determine the laminar flow velocity profile in the perfusable channel.
Synthesis of Methacryloyl Platelet Lysates:Human methacryloyl platelet
lysates (PLMA) were synthesized as previously reported.[,,]Briefly, hu-
man platelet lysates (STEMCELL Technologies, Canada) were chemically
modified with methacrylic anhydride (% (MA), Sigma–Aldrich, USA) un-
der constant stirring, at room temperature (RT). Afterward, PLMA was di-
alyzed against deionized water, sterilized with a low protein retention filter
(. μm, Sigma–Aldrich, USA), and lyophilized (LyoAlfa , Telstar, USA).
Chemically modified PLMA was stored at °C until further use.
Hydrogel Preparation, Characterization, and Injection in the Microflu-
idic Chip:PLMA hydrogel precursor solution at % (w/v) was pre-
pared by dissolving lyophilized PLMA in a sterile solution of .%
(w/v) lithium phenyl-,,-trimethylbenzoylphosphinate (LAP, Biosynth
AG, Switzerland) photoinitiator in dulbecco’s phosphate buered saline
(dPBS, Sigma–Aldrich, USA). Cylindrical hydrogels ( mm in diameter and
. mm in height) were prepared for mechanical characterization by pho-
topolymerizing the PLMA solution with light irradiation for s, using a
curing lamp at standard power mode (– nm, VALO Cordless-LED
Curing Light, Ultradent Products, USA). Compression testing was per-
formed using the Instron Series Universal Testing System (Instron,
USA) equipped with a N load cell. The slope of the linear region (–
% of strain) of the strain–stress curve was used to calculate the hydrogel
Young’s modulus.
For in-chip hydrogel polymerization, the PLMA precursor solution was
injected through the hydrogel inlet on the upper layer in order to fill the
channel and photopolymerized as above described. The exterior and in-
ternal porosity of the in-chip polarized hydrogel was observed by scanning
electron microscopy (SEM, Hitachi SU-, Hitachi, Japan) coupled with
a standard SEM coolstage (−– °C, Deben, UK) and measured using
ImageJ software.
Dextran and Doxorubicin Diffusivity:PLMA hydrogel was photopoly-
merized inside the microfluidic device as previously described and the
hydrogel inlet was covered with a squared thin layer of PDMS previously
treated with air plasma ( W, min) and ethanol sterilized. The in-chip hy-
drogel was hydrated overnight at °C, in static conditions, with Minimum
Essential Medium Alpha (𝛼-MEM, Thermo Fisher Scientific, USA) sup-
plemented with sodium bicarbonate (. g mL−, Sigma–Aldrich), %
(v/v) heat-inactivated fetal bovine serum (FBS, Thermo Fisher Scientific,
USA) and % (v/v) antibiotic/antimycotic ( units mL−of peni-
cillin, μgmL
−of streptomycin and μgmL
−of Amphotericin B,
Thermo Fisher Scientific, USA).
Fluorescein isothiocyanate-dextran kDa (FITC-Dex, μgmL
−,
Sigma–Aldrich, USA) and doxorubicin (DOX, μ, Biosynth AG,
Switzerland) solutions were prepared in 𝛼-MEM cell culture medium
and perfused at μLmin
−for h through the fluid channel using a
syringe pump (PHD ULTRA Syringe Pumps, Harvard Apparatus, USA)
and Tygon tubing. The FITC-Dex and DOX diusivity were screened by
confocal imaging (LSM , Carl Zeiss, Germany), maintaining laser
intensity and z-axis coordinates. Fluorescence intensity was quantified
using the “Plot profile” function of ImageJ software, normalized to
the time-point min, and the data were analyzed in GraphPad Prism
. software.
Cell Culture:MG- cell line (European Collection of Authenticated
Cell Cultures, ECACC, UK) and human bone marrow mesenchymal stem
cells (hBM-MSCs, American Type Culture Collection, ATCC, USA) were cul-
tured in 𝛼-MEM cell culture medium. B cell line (American Type Culture
Collection, ATCC, USA) was cultured in Minimum Essential Medium with
Earle’s Salts (MEM, BioConcept, Switzerland) supplemented with sodium
bicarbonate (. g mL−, Sigma–Aldrich, USA), -bromo-′-deoxyuridine
(. mg mL−, TCI Chemicals, China), % (v/v) heat-inactivated FBS
and % (v/v) antibiotic/antimycotic. All cells were maintained in a humid-
ified incubator with % COat °C (standard cell culture conditions)
and passaged at ≈% confluence. The medium was replaced every –
days.
Generation of Fluorescence Protein- and Luciferase-Expressing Cells:The
tumor cell lines (MG- and B) were stably transfected with lentivirus
expressing red fluorescent protein (CMV-RFP, Cellomics Technology, USA)
or firefly luciferase protein (CMV-Luc, Kerafast, USA). hBM-MSCs cells
were transfected with lentivirus expressing green fluorescent protein
(CMV-GFP, Vigene Biosciences, USA). Prior to transfection, cells were cul-
tured overnight in adherent -well plates. The cells were then transfected
at a multiplicity of infection (MOI) of for h in the appropriate cell cul-
ture medium containing polybrene ( μgmL
−, Sigma–Aldrich, USA). Af-
ter days in standard cell culture conditions, protein-expressing cells were
selected with puromycin ( μgmL
−, STEMCELL Technologies, Canada)
in the appropriate cell culture medium.
Cell Culture in the Microfluidic Chip:For tumor spheroid formation,
MG-(-Luc or -RFP) or B(-Luc or -RFP) cells were detached with Try-
pLE Express (Gibco, Thermo Fisher Scientific, USA) and re-suspended in
the appropriate cell culture medium (𝛼-MEM and MEM, respectively).
Spheroids at a density of cells were generated in -well round-
bottom ultra-low attachment plates (Corning, Thermo Fisher Scientific,
USA), centrifuged at g for min, and incubated for h at stan-
dard cell culture conditions. Spheroids were imaged by optical contrast
microscopy using an inverted light microscope (Primostar, Carl Zeiss, Ger-
many), and the images were processed using the Image Processing tool
of ZEN Image software. An open-source MATLAB-based high-throughput
image analysis software, SpheroidSizer, was used for the quantification of
the spheroid size. This software uses an adapted active contour algorithm
to measure the spheroid size automatically or manually, providing infor-
mation on the spheroid area and minor/major axis.[]The spheroid di-
ameter and circularity were then calculated using the following equations
(Equations –):
Diameter =×Area
𝜋()
Circularity =𝜋×Area
Perimeter()
where the Perimeter was obtained as:
Perimeter =𝜋×
MajorAxis
+MinorAxis
()
For microfluidic chip cell culture, hBM-MSCs cells were detached with
TrypLE Express, centrifuged at g for min, and resuspended in %
(w/v) PLMA precursor solution (prepared as previously described) at a
density of million cells mL−. Each tumor spheroid was then resus-
pended in hBM-MSCs-containing PLMA solution and injected in the cen-
ter channel of the microdevice through the hydrogel inlet. The spheroid
was placed approximately in the center of the channel by manually rotat-
ing the microfluidic device before polymerizing the hydrogel with light. The
hydrogel inlet was then covered as previously mentioned, rendering a fully
enclosed device. The chips were then filled with the appropriate cell culture
medium as follows: 𝛼-MEM for the non-metastatic model (incorporating
Adv. Funct. Mater. 2024,34, 2315940 (12 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
MG- spheroids) and 𝛼-MEM/MEM (:) for the metastatic model (in-
corporating B spheroids). This denomination will be used from now
on to facilitate the understanding of the reader.
For static chip culture, the devices were incubated for days inside Petri
dishes with humidified chambers to hamper cell culture medium evapo-
ration. For dynamic culture, the microfluidic chips inlet was connected to
a / multi-syringe pump (PHD ULTRA Syringe Pumps, Harvard Appa-
ratus, USA) and the outlet to a closed reservoir (falcon tube) using Ty-
gon tubing. The pump speed was set at μLmin
−and the devices were
maintained in normoxic conditions (% COand °C) for days. Con-
ditioned media were recovered at days , , and of culture and used for
protein expression and exometabolomics analysis. On static conditions,
the medium was recovered from the in-chip reservoirs and totally replen-
ished with a new cell culture medium; on dynamic chips, the conditioned
medium was recovered from the outlet reservoirs without aecting inlet
perfusion. For DOX drug sensitivity analysis, dynamic chips on the third
day of culture were injected with a cell culture medium containing . μ
DOX for h. For protein quantification and exometabolomics purposes,
static and dynamic acellular chips were cultured using the same condi-
tions, and the conditioned medium was recovered as mentioned for cellu-
lar chips. Recovered conditioned media were stored at − °C until further
use.
Cell Viability and Proliferation Assay:On the th day of culture, the
chips were disconnected from the inlet and outlet, and the in-chip reser-
voirs were uncovered to facilitate pipetting. The cells were recovered from
the microfluidic devices by digesting the PLMA hydrogel and detaching the
cells with .% trypsin/EDTA (Gibco, Thermo Fisher Scientific, USA), for
h. The cell suspension was pipetted through the . mm holes and trans-
ferred to a -well plate. The channels were flushed with the appropriate
cell culture medium to remove the remaining cells, and cell proliferation
was assessed using CellTiter-Glo D Cell Viability Assay (Promega, Madi-
son, USA), according to the manufacturer’s instructions. Briefly, CellTiter-
Glo D reagent was added to cell suspension at a : ratio, the samples
were vigorously mixed for min and incubated for min at RT. Lumi-
nescence was measured in -well flat-bottom opaque white plates using
a Synergy HTX microplate reader (BioTek Instruments, Winooski, USA).
The samples were stored at − °C for posterior DNA quantification us-
ing the Quant-iT PicoGreen dsDNA Assay Kit (Thermo Fisher Scientific,
USA). Samples were thawed at °C, a standard curve was obtained with
a dsDNA solution and the manufacturer protocol was followed using a
-well flat-bottom opaque black plate. After min of incubation, fluo-
rescence was measured in a microplate reader at an excitation/emission
wavelength of / nm.
Tumor Cell Viability Assay:In order to assess the viability of tumor
cells, luciferase-expressing tumor cells were used for in-chip culture. The
cells were recovered from the chips as above-mentioned and their via-
bility was assessed with ONE-Glo Luciferase Assay kit (Promega, Madi-
son, USA), following the manufacturer’s instructions. An equal volume of
reagent (relatively to trypsin+medium) was added to each well and, af-
ter min of incubation at RT, the luminescence was measured in -well
flat-bottom opaque white plates using the microplate reader.
Real-Time Imaging and Cell Tracking:RFP-tagged tumor cells and
hBM-MSCs-GFP cells were used to monitor cells in real-time, in dynamic
conditions. A confocal microscope coupled with an incubation stage and
chamber (LSM , Carl Zeiss, Germany) was used to capture time-lapse
images every hour, for h, and at and days of culture. hBM-MSCs
cell tracking was measured using the manual tracking plugin of the Im-
ageJ software, and the speed of cell migration was calculated for each h
period.
Immunofluorescence Staining:After completion of cell culture exper-
iments, the chips were disconnected and the in-chip reservoirs were un-
covered to allow the injection of the washing and staining solutions, taking
advantage of the gravitational flow throughout the channels.
Fluorescence live/dead staining was performed on tumor spheroids
before encapsulation ( days of formation) and after days of culture
inside the microfluidic chips. The samples were incubated for h in a
solution of : of Calcein AM solution ( mg mL−in dimethyl sul-
foxide, Thermo Fisher Scientific, USA) and : of Propidium Iodide
( mg mL−in distilled water, Thermo Fisher Scientific, USA) in cell cul-
ture medium, at standard cell culture conditions. Tumor spheroids were
observed under a widefield microscope (Axio Imager M, Carl Zeiss, Ger-
many), and the tumor-on-a-chip models were visualized under a confocal
microscope.
Microfluidic devices used for morphology assessment or immunocyto-
chemistry analysis at the end of the assays ( days of culture) were washed
with PBS and fixed in a % formaldehyde (Sigma–Aldrich, USA) solution
in PBS for h. Before staining, chips were permeabilized with .% (v/v)
Triton-X (Sigma–Aldrich, USA) for min and blocked with % FBS in
PBS for h at RT. The samples were incubated with primary mouse mon-
oclonal anti-human Ezrin antibody (: in % FBS (v/v) in PBS, C,
Santa Cruz Biotechnology, USA) at °C for days, followed by PBS wash-
ing for h, and then incubated with the secondary antibody goat anti-
mouse Alexa Fluor (: in % FBS (v/v) in PBS, Thermo Fisher Sci-
entific, USA) at °C for days. Afterward, the samples were counterstained
for the actin filaments with : Phalloidin-iFluor reagent (ab,
Abcam, UK) solution in PBS for h at RT, followed by nucleus staining
with : DAPI (′,-diamidino--phenylindole, Thermo Fisher Scientific,
USA) solution in PBS for min at RT. After washing with PBS, the chip
models were observed under a confocal microscope.
Real-Time Polymerase Chain Reaction (RT-PCR):In order to compare
the relative gene expression levels between the dierent conditions, the
total RNA was isolated from a group of samples (n=, technical repeats
≥) using a column-based kit (PureLink RNA Mini Kit, Thermo Fisher
Scientific, USA) according to manufacturer’s specifications. Briefly, after
days in culture both in static or dynamic conditions, the chips were dis-
connected and the in-chip reservoirs were uncovered as above mentioned.
After washing with PBS, the Lysis Buer was introduced inside the chip to
lyse the cells and mechanically weaken the hydrogel structure, facilitating
hydrogel removal from the chip. Samples of lysed cells/hydrogel from a to-
tal of chips were combined and stored at − °C until further use. To as-
sure complete hydrogel disintegration and cell lysis, prior to RNA isolation,
the samples were homogenized at rpm for min in a homogenizer
(AT-MD-, Falc Instruments, Italy) coupled to a potter (Thermo Fisher
Scientific, USA). For total RNA isolation, -mercaptoethanol (% purity,
Alfa Aesar, Germany) was added to the Lysis Buer and a -gauge syringe
was used for solution homogenization. One volume of % ethanol was
added to one volume of cell homogenate and transferred to a spin car-
tridge with a collection tube. After several washes, RNA was eluted with
RNase-free water in a collection tube. RNA purity and quantity were an-
alyzed with a Take Micro-Volume plate (BioTek Instruments, Winooski,
USA) and a microplate reader, and only samples with a / purity
ratio higher than . were used for cDNA synthesis. First-strand cDNA
synthesis was performed from ng of isolated total RNA using the In-
vitrogen SuperScript IV VILO Master Mix with ezDNase enzyme (Thermo
Fisher Scientific, USA) and the MicroAmp Reaction Tube with Cap (Applied
Biosystems, Thermo Fisher Scientific, USA), following manufacturer’s in-
structions.
The Real-Time Polymerase Chain Reaction (RT-PCR) analysis was per-
formed in a QuantStudio Real-Time PCR System (Applied Biosystems,
Thermo Fisher Scientific, USA) using the TaqMan Fast Advanced Master
Mix (Applied Biosystems, Thermo Fisher Scientific, USA), according
to the manufacturer’s recommendations. TaqMan gene expression
assays (Applied Biosystems, Thermo Fisher Scientific, USA) were car-
ried out with human-specific primers for COLA (Hs_m),
VEGFA (Hs_m), MMP (Hs_m) and MMP
(Hs_m). S (Hs_m) was used as the endogenous
housekeeping control. QuantStudio Design and Analysis software v..
(Applied Biosystems, Thermo Fisher Scientific, USA) was used to analyze
amplification profiles and determine the relative expression levels through
the comparative Ctmethod. The expression levels of each target gene
were normalized to that of the housekeeping gene.
Enzyme-Linked Immunosorbent Assay (ELISA):The concentration of
human vascular endothelial growth factor (VEGF) in the chips conditioned
media from days , , and of culture was assessed by ELISA quantifica-
tion assay (ELISA MAX Deluxe Set Human VEGF, BioLegend, USA), ac-
cording to manufacturer protocol. Absorbance was measured at nm
Adv. Funct. Mater. 2024,34, 2315940 (13 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
in a microplate reader and the concentration of VEGF was calculated from
the standard curve.
NMR-Based Exometabolomics Data Acquisition and Processing:Ex-
ometabolomics analysis was performed on media samples (n≥) col-
lected from the tumor-on-a-chip models at days , , and of culture, in
static or dynamic settings, using H-Nuclear Magnetic Resonance (H-
NMR) spectroscopy. To precipitate interfering proteins, two volumes of
cold methanol (.%, Fisher Scientific, UK) were added to one volume of
thawed medium, followed by min at − °C and centrifugation ( g
for min). The supernatants were then collected and dried in a vacuum
concentrator (CentriVap Cold Trap, Labconco, USA). Dried samples were
stored at − °C until further analysis.
For NMR analysis, dried extracts were resuspended in μL of deuter-
ated phosphate buer ( m, pH .) prepared in % deuterium
oxide (DO, CortecNet, France), containing . m of -(trimethylsilyl)
propionic acid (TSP-d, Sigma Aldrich), and μL of each sample was
transferred to a mm NMR tube (Wilmad, Merck, Germany). The samples
were analyzed in a Bruker Avance III HD NMR spectrometer (Univer-
sity of Aveiro, Portuguese NMR Network) operating at . MHz for
H observation, at °C. Standard D H spectra with water presatura-
tion (pulse program “noesyprd”, Bruker library) were recorded with k
points, . Hz spectral width, a s relaxation delay, and scans.
Spectral processing was performed in TopSpin .. (Bruker BioSpin, Rhe-
instetten, Germany), comprising cosine multiplication (ssb ), zero-filling
to k data points, manual phasing and baseline correction, and calibra-
tion to the TSP-dsignal (𝛿= ppm). Metabolite identification was per-
formed by matching D spectral information to reference spectra avail-
able in Chenomx (Edmonton, Canada) and BBIOREFCODE--- (Bruker
BioSpin, Rheinstetten, Germany).
NMR Spectra Integration and Analysis:Quantitative measurements
of metabolic variations were carried out through spectral integration
of representative metabolite signals, using Amix-Viewer .. (Bruker
BioSpin, Rheinstetten, Germany). The percentage of variation (%var) of
each metabolite was calculated relative to the respective acellular con-
trol, along with the eect size (ES) and statistical significance (p-value).
Metabolites with |%var| >% and |ES| >. were expressed in a heatmap
generated using the GraphPad software. Moreover, UV-scaled data (ob-
tained by subtraction of the signal area in each spectrum by the average
area in all spectra, and divided by the standard deviation) were used to
represent lactate variations in the form of boxplots.
Statistical Analysis:All experiments were carried out with a minimum
of three independent devices for each experimental group (n≥). For
comparisons between several groups, the statistical significance was as-
sessed via One-way or Two-way ANOVA, with Turkey’s and Sidak’s multiple
comparison tests, respectively. Otherwise, the unpaired Student’s t-test
was used. Data were statistically analyzed using GraphPad Prism Soft-
ware and expressed as mean ±standard deviation (SD) or as box-plots
representing the th–th percentiles, median value, minimum value,
and maximum value. Dierences were considered statistically significant
when the p-value was less than . and was represented by: *p<.,
**p<., *** p<., and ****p<..
Supporting Information
Supporting Information is available from the Wiley Online Library or from
the author.
Acknowledgements
This work was developed within the scope of the project
CICECO-Aveiro Institute of Materials, UIDB//
(DOI ./UIDB//), UIDP// (DOI
./UIDP//) & LA/P// (DOI
./LA/P//), financed by national funds through the
FCT/MCTES (PIDDAC). The authors would like to acknowledge the
European Research Council for the Advanced Grant Agreement number
H-ERC-AdG– for the project REBORN and for the Proof-
of-Concept Grant Agreement number ERC--PoC- for
the project HumanINK. This work was also supported by the Foun-
dation for Science and Technology through the individual contract
..CEECIND of Dr. Catarina A. Custódio and the doctoral grant
SFRH/BD// of Cátia F. Monteiro. The NMR spectrometer
is part of the National NMR Network (PTNMR), partially supported
by Infrastructure Project No. (co-financed by FEDER through
COMPETE , POCI, and PORL and FCT through PIDDAC).
Conflict of Interest
The authors declare no conflict of interest.
Data Availability Statement
The data that support the findings of this study are available from the cor-
responding author upon reasonable request.
Keywords
human protein-derived hydrogels, metastasis, microfluidics, osteosar-
coma, platelet lysate, tumor heterogeneity, tumor-on-a-chip
Received: December ,
Revised: March ,
Published online: April ,
[] C. D. Paul, P. Mistriotis, K. Konstantopoulos, Nat. Rev. Cancer 2017,
17, .
[] C. A. Klein, Nat. Rev. Cancer 2020,20, .
[] C. F. Monteiro, C. A. Custódio, J. F. Mano, Adv. Ther. 2018,2, .
[] J. Rodrigues, M. A. Heinrich, L. M. Teixeira, J. Prakash, Trends Cancer
2021,7, .
[] S. Kim, E. M. Kim, M. Yamamoto, H. Park, H. Shin, Adv. Healthcare
Mater. 2020,9, .
[] S. J. Han, S. Kwon, K. S. Kim, Cancer Cell Int. 2021,21, .
[] C. F. Monteiro, S. C. Santos, C. A. Custódio, J. F. Mano, Adv. Sci. 2020,
7, .
[] C. F. Monteiro, C. A. Custódio, J. F. Mano, Acta Biomater. 2021,134,
.
[] L. P. Ferreira, V. M. Gaspar, R. Henrique, C. Jerónimo, J. F. Mano,
Biotechnol. J. 2017,12, .
[] X. Li, Q. Fan, X. Peng, S. Yang, S. Wei, J. Liu, L. Yang, H. Li, Cell Death
Discov. 2022,8,.
[] D. Wirtz, K. Konstantopoulos, P. C. Searson, Nat. Rev. Cancer 2011,
11, .
[] A. Sontheimer-Phelps, B. A. Hassell, D. E. Ingber, Nat. Rev. Cancer
2019,19, .
[] Y. Nashimoto, R. Okada, S. Hanada, Y. Arima, K. Nishiyama, T. Miura,
R. Yokokawa, Biomaterials 2020,229, .
[] V. S. Shirure, Y. Bi, M. B. Curtis, A. Lezia, M. M. Goedegebuure, S. P.
Goedegebuure, R. Aft, R. C. Fields, S. C. George, Lab Chip 2018,18,
.
[] E. A. Adjei-Sowah, S. A. O’Connor, J. Veldhuizen, C. Lo Cascio, C.
Plaisier, S. Mehta, M. Nikkhah, Adv. Sci. 2022,9, .
[] A. Boussommier-Calleja, Y. Atiyas, K. Haase, M. Headley, C. Lewis, R.
D. Kamm, Biomaterials 2019,198, .
[] H. Kim, H. Chung, J. Kim, D. Choi, Y. Shin, Y. G. Kang, B. Kim, S. Seo,
S. Chung, S. H. Seok, Adv. Sci. 2019, .
Adv. Funct. Mater. 2024,34, 2315940 (14 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
www.advancedsciencenews.com www.afm-journal.de
[] K. Haase, G. S. Oeddu, M. R. Gillrie, R. D. Kamm, Adv. Funct. Mater.
2020,30, .
[] C. Eilenberger, M. Rothbauer, F. Selinger, A. Gerhartl, C. Jordan, M.
Harasek, B. Schädl, J. Grillari, J. Weghuber, W. Neuhaus, S. Küpcü, P.
Ertl, Adv. Sci. 2021,8, .
[] J. Aleman, A. Skardal, Biotechnol. Bioeng. 2019,116, .
[] F. Conceição, D. M. Sousa, J. Loessberg-Zahl, A. R. Vollertsen, E.
Neto, K. Søe, J. Paredes, A. Leferink, M. Lamghari, Mater. Today Bio.
2022,13, .
[] B. Gumuscu, J. C. T. Eijkel, in C-BMicroarrays (Eds.: P. Ertl, M. Roth-
bauer), Springer, New York, NY 2018, pp. .
[] M. W. Dewhirst, T. W. Secomb, Nat. Rev. Cancer 2017,17, .
[] Q. Chang, O. I. Ornatsky, I. Siddiqui, R. Straus, V. I. Baranov, D. W.
Hedley, Sci. Rep. 2016,6, .
[] S. C. Santos, C. A. Custódio, J. F. Mano, Adv. Healthcare Mater. 2018,
7, .
[] E. N. Hoogenboezem, C. L. Duvall, Ad v. Dr ug D eliv. Rev. 2018,130,
.
[] Z. Chen, F. Han, Y. Du, H. Shi, W. Zhou, Sig. Transduct. Target Ther.
2023,8,.
[] S. U. Lauvrak, E. Munthe, S. H. Kresse, E. W. Stratford, H. M. Namløs,
L. A. Meza-Zepeda, O. Myklebost, Br. J. Cancer 2013,109, .
[] L. Ren, A. Mendoza, J. Zhu, J. W. Briggs, C. Halsey, E. S. Hong, S. S.
Burkett, J. J. Morrow, M. M. Lizardo, T. Osborne, S. Q. Li, H. H. Luu,
P. Meltzer, C. Khanna, Oncotarget 2015,6, .
[] C. Mao, F. Li, Y. Zhao, W. Debinski, X. Ming, Theranostics 2018,8,
.
[] L. Marchandet, M. Lallier, C. Charrier, M. Baud’huin, B. Ory, F.
Lamoureux, Cancers 2021,13, .
[] M. Cortini, F. Macchi, F. Reggiani, E. Vitale, M. V. Lipreri, F. Perut, A.
Ciarrocchi, N. Baldini, S. Avnet, Cancers 2023,15, .
[] J. Zhao, X. Huang, Z. Xu, J. Dai, H. He, Y. Zhu, H. Wang, Mol. Med.
Rep. 2017,16, .
[] J. Winkler, A. Abisoye-Ogunniyan, K. J. Metcalf, Z. Werb, Nat. Com-
mun. 2020,11, .
[] E. Goggins, Y. Mironchik, S. Kakkad, D. Jacob, F. Wildes, Z. M.
Bhujwalla, B. Krishnamachary, Cancer Biol. Ther. 24, .
[] G. Gonzalez-Avila, B. Sommer, D. A. Mendoza-Posada, C. Ramos, A.
A. Garcia-Hernandez, R. Falfan-Valencia, Crit. Rev. Oncol. Hematol.
2019,137, .
[] D. Huh, D. C. Leslie, B. D. Matthews, J. P. Fraser, S. Jurek, G. A.
Hamilton, K. S. Thorneloe, M. A. McAlexander, D. E. Ingber, Sci.
Transl. Med. 2012,4, ra.
[] S. Jalili-Firoozinezhad, F. S. Gazzaniga, E. L. Calamari, D. M.
Camacho, C. W. Fadel, A. Bein, B. Swenor, B. Nestor, M. J. Cronce,
A. Tovaglieri, O. Levy, K. E. Gregory, D. T. Breault, J. M. S. Cabral, D.
L. Kasper, R. Novak, D. E. Ingber, Nat. Biomed. Eng. 2019,3, .
[] C.-W. Chi, Y.-H. Lao, A. H. R. Ahmed, E. C. Benoy, C. Li, Z. Dereli-
Korkut,B.M.Fu,K.W.Leong,S.Wang,Adv. Healthcare Mater. 2020,
9, .
[] D. B. Chou, V. Frismantas, Y. Milton, R. David, P. Pop-Damkov, D.
Ferguson, A. MacDonald, Ö. V. Bölükba¸sı,C.E.Joyce,L.S.Moreira
Teixeira, A. Rech, A. Jiang, E. Calamari, S. Jalili-Firoozinezhad, B. A.
Furlong, L. R. O’Sullivan, C. F. Ng, Y. Choe, S. Marquez, K. C. Myers,
O. K. Weinberg, R. P. Hasserjian, R. Novak, O. Levy, R. Prantil-Baun,
C. D. Novina, A. Shimamura, L. Ewart, D. E. Ingber, Nat. Biomed. Eng.
2020,4, .
[] S. Bersini, J. S. Jeon, G. Dubini, C. Arrigoni, S. Chung, J. L. Charest,
M. Moretti, R. D. Kamm, Biomaterials 2014,35, .
[] J. Kim, K.-T. Lee, J. S. Lee, J. Shin, B. Cui, K. Yang, Y. S. Choi, N. Choi,
S. H. Lee, J.-H. Lee, Y.-S. Bahn, S.-W. Cho, Nat. Biomed. Eng. 2021,5,
.
[] L. Brady, M. Kriner, I. Coleman, C. Morrissey, M. Roudier, L. D. True,
R. Gulati, S. R. Plymate, Z. Zhou, B. Birditt, R. Meredith, G. Geiss, M.
Hoang, J. Beechem, P. S. Nelson, Nat. Commun. 2021,12, .
[] A. J. Primeau, A. Rendon, D. Hedley, L. Lilge, I. F. Tannock, Clin. Can-
cer Res. 2005,11, .
[] P. C. Lyon, M. D. Gray, C. Mannaris, L. K. Folkes, M. Stratford, L.
Campo, D. Y. F. Chung, S. Scott, M. Anderson, R. Goldin, R. Carlisle,
F. Wu, M. R. Middleton, F. V. Gleeson, C. C. Coussios, Lancet Oncol.
2018,19, .
[] J. Lee, M.-K. Choi, I.-S. Song, Pharmaceuticals 2023,16, .
[] W. Yu, L. Chen, Y.-Q. Yang, J. R. Falck, A. M. Guo, Y. Li, J. Yang, Cancer
Chemother. Pharmacol. 2011,68, .
[] E. I. Deryugina, J. P. Quigley, Matrix Biol. 2015,94, .
[] T.-L. He, Y.-J. Zhang, H. Jiang, X. Li, H. Zhu, K.-L. Zheng, Med. Oncol.
2015,32, .
[] S. Gao, D.-N. Tu, H. Li, J.-X. Jiang, X. Cao, J.-B. You, X.-Q. Zhou,
Biomed. Pharmacother. 2016,81, .
[] J. Meiser, S. Tumanov, O. Maddocks, C. F. Labuschagne, D. Athineos,
N. Van Den Broek, G. M. Mackay, E. Gottlieb, K. Blyth, K. Vousden, J.
J. Kamphorst, A. Vazquez, Sci. Adv. 2016,2, .
[] J. Meiser, A. Schuster, M. Pietzke, J. Vande Voorde, D. Athineos,
K. Oizel, G. Burgos-Barragan, N. Wit, S. Dhayade, J. P. Morton, E.
Dornier, D. Sumpton, G. M. Mackay, K. Blyth, K. J. Patel, S. P. Niclou,
A. Vazquez, Nat. Commun. 2018,9, .
[] S. Bose, V. Ramesh, J. W. Locasale, Trends Cell Biol. 2019,29,.
[] A. Iliou, A. Panagiotakis, A. F. Giannopoulou, D. Benaki, M.
Kosmopoulou, A. D. Velentzas, O. E. Tsitsilonis, I. S. Papassideri,
G. E. Voutsinas, E. G. Konstantakou, E. Gikas, E. Mikros, D. J.
Stravopodis, Int. J. Mol. Sci. 2020,21, .
[] G. R. Howard, T. A. Jost, T. E. Yankeelov, A. Brock, PLoS Comput. Biol.
2022,18, .
[] I. W. Achkar, S. Kader, S. S. Dib, K. Junejo, S. B. Al-Bader, S. Hayat,
A. M. Bhagwat, X. Rousset, Y. Wang, J. Viallet, K. Suhre, A. Halama,
Metabolites 2020,10, .
[] R. M. Maria, W. F. Altei, H. S. Selistre-de-Araujo, L. A. Colnago, Bio-
chemistry 2017,56, .
[] C. B. X. Huang, T.-Y. Tu, Front. Onco l. 2023,13, .
[] R. D. Leone, J. D. Powell, Nat. Rev. Cancer 2020,20, .
[] C. P. Miller, W. Shin, E. H. Ahn, H. J. Kim, D.-H. Kim, Trends Biotechnol.
2020,38, .
[] A. Guzman, V. S. Alemany, Y. Nguyen, C. R. Zhang, L. J. Kaufman,
Biomaterials 2017,115, .
[] W. Chen, C. Wong, E. Vosburgh, A. J. Levine, D. J. Foran, E. Y. Xu, J.
Visualized Exp. 2014,8, .
Adv. Funct. Mater. 2024,34, 2315940 (15 of 15) © The Authors. Advanced Functional Materials published by Wiley-VCH GmbH
