Fabrication of oxygen‐carrying microparticles functionalized with liver ECM‐proteins to improve phenotypic three‐dimensional in vitro liver assembly, function, and responses

Oxygen and extracellular matrix (ECM)‐derived biopolymers play vital roles in regulating many cellular functions in both the healthy and diseased liver. This study highlights the significance of synergistically tuning the internal microenvironment of three‐dimensional (3D) cell aggregates composed of hepatocyte‐like cells from the HepG2 human hepatocellular carcinoma cell line and hepatic stellate cells (HSCs) from the LX‐2 cell line to enhance oxygen availability and phenotypic ECM ligand presentation for promoting the native metabolic functions of the human liver. First, fluorinated (PFC) chitosan microparticles (MPs) were generated with a microfluidic chip, then their oxygen transport properties were studied using a custom ruthenium‐based oxygen sensing approach. Next, to allow for integrin engagements the surfaces of these MPs were functionalized using liver ECM proteins including fibronectin, laminin‐111, laminin‐511, and laminin‐521, then they were used to assemble composite spheriods along with HepG2 cells and HSCs. After in vitro culture, liver‐specific functions and cell adhesion patterns were compared between groups and cells showed enhanced liver phenotypic responses to laminin‐511 and 521 as evidenced via enhanced E‐cadherin and vinculin expression, as well as albumin and urea secretion. Furthermore, hepatocytes and HSCs exhibited more pronounced phenotypic arrangements when cocultured with laminin‐511 and 521 modified MPs providing clear evidence that specific ECM proteins have distinctive roles in the phenotypic regulation of liver cells in engineering 3D spheroids. This study advances efforts to create more physiologically relevant organ models allowing for well‐defined conditions and phenotypic cell signaling which together improve the relevance of 3D spheroid and organoid models.


| INTRODUCTION
The need to create physiologically realistic in vitro liver models has resulted in the emergence of technologies such as organ on a chip, three-dimensional (3D) printing, and rotational culture methods, yielding protocols to create improved human preclinical models (Khanna et al., 2021).However, 3D cellular aggregates present certain drawbacks.First, there is no vascular network formation; therefore, nutrient and waste transport limits the maximum size of 3D tissues grown in vitro (Mansouri & Leipzig, 2021).Second, in vivo niche-based cues such as extracellular matrix (ECM) components are missing, which slows initial aggregation and cell activation resulting in loss of function and phenotype in long-term cultures (Nagata et al., 2020).Unlike other organs, the liver has a unique makeup with a proportionately small ECM in relation to its volume, mainly consisting of collagen, fibronectin, and laminin (Martinez-Hernandez & Amenta, 1993).Collagen represents 60% of human liver ECM molecules and constitutes mostly fibrillar collagens such as type I and III collagen (Nyström, 2021), providing tensile strength to the organ.Whereas non-collagenous proteins such as fibronectin and laminin are vital to maintaining basement membrane and functional integrity (Arriazu et al., 2014).Fibronectin is a multifunctional adhesive glycoprotein that originally synthesized by liver cells and abundantly present in liver tissue.This glycoprotein is directly involved in regulating cellular behavior such as cell survival and proliferation (Aziz-Seible, 2011).
Laminin is another major noncollagenous adhesive glycoprotein present in the hepatic perisinusoidal space (space of Disse).It includes specific combinations of α, β, and γ chains giving rise to functional diversity within a common structural framework.Within this family the distribution and expression of α5, β1, and β2 laminin chains in the mammalian liver has been widely reported (Lou & Leung, 2018).
ECM-based molecules (peptides or whole proteins) have been widely incorporated into biomaterials to promote the formation and function of various types of spheroids and organoids (Saydé et al., 2021;Tibbitt & Anseth, 2009).For example, human intestinal organoids were generated from pluripotent stem cells in a synthetic hydrogel based on a four-armed, maleimide-terminated poly(ethylene glycol) macromer functionalized with arginine-glycine-aspartate (RGD) adhesive peptides (Cruz-Acuña et al., 2017).The authors showed that organoids encapsulated in this scaffold show high viability as well as expression levels of pluripotency, endoderm, and epithelial junction markers compared to nonmodified gels.Similarly, Lin et al. (2014) prepared a hydrogel based on poly(ethylene glycol)tetra-norbornene (PEG4NB) functionalized by bioactive peptides (e.g., fibronectin-derived arginine-glycine-aspartate-serotonin [RGDS]) to improve cell-matrix interactions.They observed elevated urea secretion and Cytochrome P450 3A4 (CYP3A4) enzymatic activities, as well as upregulated mRNA levels of multiple hepatocyte genes (e.g., CYP3A4, bile salt export pump [BESP] and sodiumtaurocholate cotransporting polypeptide [NTCP]) of two human hepatoma-derived cell lines (Huh7 and HepG2) encapsulated in these gels.Despite these reports, new methodologies are needed for synthetic cellular microenvironments to reduce mass transport limitations, especially for metabolically active cell types/tissues such as the liver.Various methodologies have been formulated to promote oxygen transport and maintain viability as well as the function of cells in 3D culture (Powers et al., 2002;Takayama et al., 2013).In our recent work, we created fluorinated (PFC) chitosan microparticles (MPs) to overcome common limitations of spheroids, such as inadequate oxygen supply and ultimate loss of cell/organ-specific functions over long-term cultures (Mansouri et al., 2022).Our data suggested that PFC-conjugated MPs offer a simple, affordable, and direct approach for improving mass transport of nutrients within spheroids and other engineered tissues.
In the present study, we extended our PFC-MP approach to yield improved hepatic 3D cell culture models using ECM compo-

| Studying oxygen levels in MPs
We used the ruthenium complex Tris(4,7-diphenyl-1,10- (10 mg mL −1 ) and sonicated thoroughly before getting injected into the channel.A light-emitting diode (LED, 4 µW cm −2 ) was used as the excitation source (peak wavelength 455 nm), and a long wave pass filter (Thorlabs, FMP1, cutoff wavelength 600 nm) was used to minimize unwanted wavelengths.We compared the oxygen level within the chamber using two gases including nitrogen (N 2 ) and oxygen (O 2 ), to represent 0% and 100% oxygen, respectively.N 2 gas was initially injected into the channel, followed by imaging, and then the gas source was switched to O 2 to present a saturated state.For each condition, eight frames per second were obtained for up to 60 s.
The fluorescent intensity of unquenched images (I 0 ) was divided by quenched images (I) on a frame-by-frame basis to obtain a Stern-Volmer image.

| Preparation of liver spheroids incorporated with PFC-MPs
The creation of spheroids and incorporation with MP followed procedures recently published (Mansouri et al., 2022).We obtained the immortalized human hepatocellular carcinoma cell line, HepG2  (Mansouri et al., 2022).Briefly, HepG2 and HSC cells were mixed at a 4:1 ratio, then added to MPs to achieve a desired final cell-to-particle ratio (150:1) and transferred to the 96-well cell repellant plates (CELLSTAR ® Plates with Cell-Repellent Surface, cat#: 650970) at a seeding density of 20,000 cells/well.To form cell aggregates, the well plate was centrifuged at 300 XG for 5 min.We utilized a high cell density to generate larger spheroids to improve their physiological relevance.Cells were overlaid with a serum-free and defined hepatocyte culture media (Corning ® , cat# 355056).

| Measurement of oxygen concentration within spheroids
For oxygen characterization, RuDPP was added to cell suspensions (2 µg mL −1 ) before spheroid formation.After 5 days of culture, the spheroids were live imaged using a laser scanning confocal microscope (Olympus FV1000) to capture Z-stack images throughout the spheroids.µ-Slide 8 well glass bottom slides from ibidi (cat# 80827) were used to image spheroids (excitation: 455 nm, emission: 610 nm).
The imaging was performed by taking z-stacks at 10 µm intervals

| Surface functionalization of MPs using liver adhesive ligand
The protocol for the covalent bonding of proteins on PFC-MPs was similar to that reported in previous studies (Wilkinson et al., 2013).

| Histologic and immunohistochemical examinations
For immunofluorescence staining of 3D liver spheroids, spheroids were fixed for 30 min in 4% w/v paraformaldehyde in PBS.Then

| Data analysis
Data processing and displaying were performed using GraphPad Prism 5 (GraphPad Software).Mean and standard deviation were displayed for all data sets.Statistical significance was determined using either Student's t-test or two-way analysis of variance (ANOVA) followed by Tukey's post hoc test.Unless otherwise stated, a p < 0.05 was considered significant for all statistical analyses.
Groups with significant differences are shown with alphabetic letters to signify the ranking of means.Groups indicated with the same letters were not found to be significantly different.

| Characterization of PFC-MPs
It has been well established that oxygen-carrying properties of PFC-MPs stem from the PFC substitutions, (Jägers et al., 2021;Riess, 2005); therefore, in this report we focused on studying the composition of microfluidic generated MPs while confirming the presence of PFC molecules (Figure 1). Figure 1b shows the TGA spectrum of crosslinked non-PFC-MPs and PFC-MPs.The weight loss in both spectra begins at approximately 90°C and continues until 130°C, which is likely associated with loss of absorbed water in the crosslinked MPs.Based on these data both modified and unmodified MPs were able to absorb water correlating to approximately 25% via hydrophilic moieties in chitosan (Sunarti et al., 2019).Another major mass loss stage is observed between 280°C and 400°C, which is related to thermal decomposition of chitosan (Timur & Paşa, 2018).
Overall, around 43.1% w/w of PFC-MPs and 52.82% w/w of non-PFC-MPs were lost when heated above 400°C, suggesting higher thermal stability of PFC-MPs compared to non-PFC-MPs.
Carbon-fluorine (C-F) bonds in PFC molecules are strongly polarized, which accounts for their thermal stability (Wang & Liu, 2020).Next, XPS was used to provide a quantitative analysis from the outer 10 nm of MP surfaces.XPS spectra showed carbon (C1s at 284.7 eV) and oxygen (O1s at 531.5 eV) as well as fluorine (F1s at 688.6 eV) on the surfaces of PFC-MPs (Figure 1c), suggesting the presence of PFC molecules on the surfaces of MPs.This is expected since the particles were crosslinked while still in the oil phase.We tried Fouriertransform infrared spectroscopy (FTIR) as well (Figure S2); however, due to PFC abundance being lower than the detection threshold (2 wt%), we could not reliably detect C-F bonds.

| Assessing oxygen-carrying properties of MPs using RuDPP
In our previous studies, PFC-conjugated chitosan hydrogels were tested for oxygen transport properties using commercial phosphorescent dots (Wijekoon et al., 2013) and needle sensors (Li et al., 2014).Our investigations showed these hydrogels enhanced oxygen transport in aqueous conditions and allowed oxygen tensions to reach higher equilibrium states compared with controls not integrating PFCs.In this study, we were interested in understanding the impact of our PFC-MPs on oxygen tensions using RuDPP in finer spatial detail and at a cellular level.RuDPP has a long fluorescence lifetime and a long Stokes shift.It quenches in the presence of oxygen and thus is a suitable probe dye for determining dissolved oxygen (Jiang et al., 2017).A mixture of MP solution and RuDPP in PBS was injected into a sealed, custom-built chamber, and exposed to gases (Figure 2a).We measured the fluorescence intensity after purging with pure N 2 and then pure O 2 gases to create oxygen partial pressures at the extremes of 0% oxygen and 100% oxygen in solution.A solution without any MPs was used as a control.As shown in Figure 2b, when switching from pure N 2 to pure O 2 , the fluorescent intensity decreased due to oxygen quenching of RuDPP.
We also observed a decrease in fluorescence intensity in the presence of MPs, which can be attributed to increased dissolved oxygen concentration.The sensitivity of the optical oxygen sensor with (I) and without (I 0 ) a quencher (here oxygen) was quantified in term of the ratio I 0 /I 100 , as shown in Figure 2c at equilibrium.Our results indicated that the solution containing PFC-MPs exhibited the highest sensitivity to oxygen with the maximum quenching ratio of I 0 /I 100 = 1.1.Following saturation, we then stopped oxygen flow and studied the release kinetics.We observed that the control group showed complete RuDPP recovery within 3 min, whereas PFCmodified MPs slowed the RuDPP recovery process to 10 min (Figure 2d).This can be explained by these MPs' ability to uptake oxygen and then release it gradually into a low-oxygen tension aqueous environment.These results confirm that PFC-MPs can act as both an oxygen-absorbing material as well as a reservoir to release oxygen gradually in an aqueous environment.This finding is in agreement with our earlier work with PFC-modified hydrogels, where we examined their ability to deliver and sustain biological levels oxygen for enhancing cellular functions essential in wound healing (Akula et al., 2017;Patil et al., 2016).Most recent evidence from another research group shows the potential of PFC nanoemulsions as artificial oxygen carriers (Lambert & Janjic, 2021).In this study, PFC nanoemulsions exhibited five times higher dissolved O 2 concentration compared to water, as measured by a submerged O 2 probe.

| Oxygen levels within live spheroids cultured with/without PFC-MPs
In our prior research, we quantified the degree of cell death in spheroids cultured with PFC-MPs by monitoring the release of lactate dehydrogenase (LDH) into the culture medium (Mansouri et al., 2022).We observed a decrease in LDH levels from whole spheroids, suggesting elevated oxygen tensions due to the presence of our fluorinated oxygenating MPs.However, measuring LDH Although the working concentration of dye in our system is low, we were able to utilize this concentration to directly image live intracellular oxygen levels.To study cellular oxygen levels throughout spheroids we made 3D images of z-stacks and looked at the fluorescence intensity throughout the whole live spheroid as shown in Figure 3a.We observed cells at higher oxygen tensions (shown in black) in spheroids cultured with PFC-MPs, demonstrating that MPs improve oxygen levels in the spheroids.PFCs have the ability to transport oxygen from the surrounding environment, and their weak molecular-level interactions allow for oxygen release in low-oxygen tension environments (Fathollahipour et al., 2018).This enables PFC-MPs to facilitate the diffusion of oxygen from the external environment (media) via enhanced gradients, thereby replenishing local oxygen levels within the spheroids.It's important to note that our MPs do not generate oxygen; instead, they serve as a transport enhancer that reduces the gradient from the environment to the interior of the spheroids, as we have previously reported (Mansouri et al., 2022).
Quantitative cellular measurements corresponding O 2 % were performed via a calibration curve based on RuDPP in a controlled oxygenated environment (Figure S1C).To perform this, we focused on individual cells and the intensity of those cells along a line drawn in a central confocal plane of spheroids.As shown in Figure 3b, the addition of MPs increased oxygen levels from 7% to 10%, emphasizing the validity of our approach.In related work, partial pressure of oxygen (PO 2 ) levels in human colorectal spheroids with a diameter of approximately 600 μm were measured using electron spin resonance (ESR) microscopy, and PO 2 values were reported in the range of 50-60 mmHg which is equivalent to 6.6%-7.9%oxygen at inner regions (Hashem et al., 2015).In vitro experiments are usually performed in incubators that maintain a PO 2 of approximately 142 mmHg (18.7% oxygen), whereas cells in our body do not experience a PO 2 greater than approximately 100 mmHg (13% oxygen).Despite lower oxygen tensions in vivo, cells can tolerate these levels and easily survive because of short transport distances from supply via blood vessels (Tse et al., 2021).When it comes to avascular 3D cell aggregates, especially larger ones, we expect lower PO 2 due to greater distances and mass transport limitations.The Hypoxyprobe™ (pimonidazole hydrochloride) method is often used to study hypoxia responses in spheroids (Godet et al., 2022); however, the method is an indirect endpoint test with no quantitative analysis in terms of oxygen tensions.Using RuDPP is advantageous for live and kinetic imaging since the fluorescence lifetime of RuDPP is largely insensitive to pH, ion concentrations, and cellular contents, making it suitable for cell culture applications (Gerritsen et al., 1997).
Previous work has shown that MCF-7 cells are labeled with RuDPP at a concentration of 20 µg mL −1 maintained 90% viability after 24 h, (Breitkopf et al., 2014) confirming its biocompatibility.

| Covalent immobilization of liver ECM adhesive ligands to PFC-MPs
Next, we modified MP surfaces by covalently binding ECM proteins to overcome the limitations associated with simple physical adsorption of ECM proteins, including undesired conformational changes and release from surfaces over time (Custódio et al., 2010).In this study, biofunctionalization with the ECM proteins plasma fibronectin, laminin-111, 511, and 521 to the surface of PFC-MPs was accomplished by a coupling reaction between the carboxyl residues of proteins to the free amine residues of chitosan using EDC as a crosslinker and NHS ester as an enhancer of coupling (Figure 4a).The EDC/NHS coupling reaction is a selective method for preservation of biological activity of the protein and has previously been applied in the production of protein-functionalized chitosan derivatives (Chen et al., 2014;Ho et al., 2005;Taylor et al., 2015).
Since the pH of the reaction solution is critical for maximizing the amination reaction, we tested different pH levels keeping all other conditions the same, to determine the favored pH to maximize attachment.The recommended pH conditions for the carbodiimide reaction via EDC/NHS is in the range of pH 4.5-5.5 (Kuo et al., 1991).However, due to the presence of amine groups, chitosanbased materials become protonated at a pH below pK a 6.0 (Pilipenko et al., 2019).Between a pH of 5.9 and 6.8 we achieved ligand conjugation on MPs ranging from 2 to 3.5 µg mL −1 , as quantified by BCA kit (Figure 4b).We can attribute the poorer yield at the lowest pH (5.9) to high chitosan protonation, resulting in a slower reaction.
Poor yield at the highest pH (6.8) might be due to the formation of macroscopic aggregates resulting in a reduced availability of surface area for reactions.The formation of aggregates at pH > 6.5 where chitosan particles are weakly charged and less stable, was reported previously (Buschmann et al., 2013;Germershaus et al., 2008).We further characterized the uncoated and coated MPs modified at different pH levels via XPS. Figure 4c presents the atomic compositions of MPs modified at different pHs.This result shows an increase in atomic nitrogen percentages (N1s%) as a reliable marker to detect additional amide groups (N-C = O) from amino acids in added ECM proteins compared to nonmodified MPs.Interestingly, we observed an increase from 0.9% to 6.0% at pH = 6.2, which confirms the data obtained from colorimetric quantification.Similarly, Huang et al. (2007) evaluated the grafting efficiency of laminin (derived from Engelbreth-Holm-Swarm mouse sarcoma basement membrane) on poly(lactic-co-glycolic acid) (PLGA) film using XPS and observed an increase in N1s from 0.5% to 1.1% after modification with laminin.We opted to focus on only N1s because C1s and O1s peaks can originate from the MP composition itself and might lead to misleading data.arginase activity, which occurs via an ATP-independent cycle (Ware & Khetani, 2017).Despite the fact that HepG2 cells do not produce urea through typical metabolic processes, they do express arginase, which can convert arginine into ornithine and urea (Mavri-Damelin et al., 2008).To characterize these critical cellular functions, we collected culture medium and measured albumin and urea production at specific time points (Figure 5b).

| Influence of specific ECM proteins on liver-specific functions
This experiment showed that the laminin isoforms −511 and −521 boosted albumin and urea production compared with the other groups (p < 0.01), providing evidence that these two isoforms can more closely reflect the in vivo composition of cells present in the liver.This response can be attributed to the role of integrins in facilitating cell-ECM interactions, thus allowing cells to anchor and perform cell phenotypic functions.Specifically, the binding of integrins to ECM molecules initiates the formation of focal adhesions, which in turn triggers intracellular signaling pathways that ultimately lead to the transcriptional activation of genes responsible for regulating the cell cycle, morphology, and function (Hoshiba et al., 2007).Consistent with this interpretation laminin-511 and −521 coated surfaces were proven to improve culture matrices for hepatic specification and differentiation of human pluripotent stem cells (hPSCs), (Kanninen et al., 2016) as well as human embryonic stem cells (hESC) (Cameron et al., 2015).
We also considered laminin-111 to represent a variation of laminin not found abundantly in the adult liver to compare the difference between liver-specific isotypes to nonspecific ones.
Laminin-111 is expressed during embryogenesis and plays an important role in assembling early basement membranes (Horejs et al., 2014).We did not find any significant differences (p > 0.05) between laminin-111 conjugated and noncoated MPs in terms of liver-specific functions probed, highlighting the importance of proteins native to the mature in vivo liver environment for benefitting cell culture in vitro.Our result is consistent with an experiment conducted by Klaas et. al (2021) where they cultured HepG2 cells on 3D nano-and microfiber structures based on gelatin, preincubated with laminin-111 and observed no significant changes in cell morphology, viability, and production of proteins between coated with laminin-111 and uncoated 3D gelatin scaffolds.isoforms in liver tumors.They showed that cancer cells could specifically interact with these ligands through integrin α3β1 and α6β4, which are widely expressed in hepatocellular carcinoma (Kikkawa et al., 2008;Nishiuchi et al., 2006).Laminin-111 primary binding integrins are α6β1 (Cho & Mosher, 2006) and α7β1 (Nishiuchi et al., 2006), while fibronectins main integrin receptors are α5β1, α8β1, αVβ1 (Takada et al., 2007).HepG2 cells are known to express α2, α6, β1, and β4 integrin subunits (Kawakami-Kimura1 et al., 1997) and very low levels of the fibronectin receptors, presenting the possibility that they may respond to laminin-511 and 521, but not fibronectin and laminin-111, simply because they do not express sufficient levels of the appropriate integrin receptors.

| Visualizing cell-cell and cell-ECM interactions
Equally important in the liver to integrin activation is cell-cell adherins formation.Hepatocytes in the liver arrange with one surface in direct contact with blood and the other surface forming tight junctions with adjacent cells.For every eight hepatocytes, one HSC helps to maintain the storage of vitamin A and protects hepatocytes in case of injury (Lee et al., 2021).Thus, we studied physical linkage between cells to explore the effect of modified MPs on E-cadherin expression levels.E-cadherin, a cell-cell adhesion molecule in epithelial tissues, is localized on the surfaces of cells and allows physical linkage between neighboring cells (Benton et al., 2009).Our data showed an increased expression of E-cadherin in spheroids cultured with laminin-511 and −521 modified MPs as compared with other groups (p < 0.01).Importantly, we observed high expression of E-cadherin at the edges, which is associated with more heterotypic cell-cell contacts (Figure S4).The degree of E-cadherin contact has been shown to affect proliferation and liver-specific function (Brieva & Moghe, 2001).Taken together, analysis of vinculin and E-cadherin expression suggests that the phenotype of cells was maintained better in groups cultured with laminin-511 and −521 and may indeed play a role in modulating hepatocytes functions.Another important aspect of adherens-type junctions is the stabilization of the interactions between adjacent cells, which is a crucial factor in forming and maintaining 3D cellular aggregates.In our previous study with liver spheroids, (Mansouri et al., 2022), we showed that HSCs This is likely because HSCs secrete many biomolecules, cytokines, and growth factors leading to better shape and stability.In the liver, HSCs do not constitute direct cell adhesions with liver sinusoidal endothelial cells (LSECs), whereas they establish adherens junctions with hepatocytes through thorn-like microprojections or spines (Wake, 2006) with E-cadherin as the main adhesion molecule.

| Visualizing cell arrangements
In vivo, ECM molecules, including fibronectins, collagens, and laminins not only affect signaling properties and cell function but also modulate overall tissue architecture with direct consequences on cell arrangements and tissue organization (Akhavan et al., 2012).To better understand the effects of specific ECM proteins on the cell organization, we used immunohistochemistry staining for HepG2 and HSC cells in whole-mount spheroids to visualize the impact of ECMmodified MPs on arrangements (Figure 7).When HepG2(C3A) subclone cells are cultured in a 3D environment, they offer several advantages, such as the ability to restore important liver cell functions like cytochrome P450 expression (Štampar et al., 2021).
Additionally, research has demonstrated that intercellular signaling between hepatocytes and HSCs can significantly enhance the enzymatic activity of CYP450 (Lee et al., 2013).Therefore, we confirmed that the HepG2 cells utilized in our model did indeed express CYP450, validating the choice of CYP antibody staining for positive identification.The staining patterns for hepatocytes and HSCs in our study were similar to what reported by Wong et al. (2011) and 3D HepaRG/HSC cultures (Leite et al., 2016).
Although HepG2 cell distribution was homogenous, we observed distinct localization of HSCs toward the periphery of spheroids, generating a distinct outer layer of similar thickness in all groups.This also correlates with our E-cadherin staining results (Figure 6b).The contacting area of hepatocytes is known to present collagen fibrils, (Urushima et al., 2021) resulting in a membrane thickening which we also observed in our images (the purple ring at the edge of all groups).
We observed that groups treated with laminin-511 and 521 resulted in localization of some HSCs toward the center of spheroids, perhaps This study has certain limitations, one of which pertains to the origin of the cells used.We employed cells from the HepG2 line, which are a recognized liver cell line and are frequently used as a preliminary in vitro model for human hepatocytes (Guo et al., 2011).
However, we acknowledge the fact that HepG2 cells have lower metabolic functions than primary hepatocytes, therefore, they are less suitable for studying toxicity of drugs that are dependent on metabolic conversion.In addition, metabolomic analyses may be nents.Cells typically begin to form ECM approximately 3-4 days after culture initiation(Malakpour-Permlid et al., 2021), which implies that there is no ECM present during the initial stages of spheroid formation.To provide ECM-based ligand signaling, we modified the MPs by incorporating ECM molecules to encourage early assembly and enable cells to attach and transition to synthetic activities more quickly.First, we examined if PFC-MPs play a role in driving oxygen into spheroidal aggregates using optical sensing.Then, we covalently tethered ECM proteins, including laminin-111, laminin-511, laminin-521, and fibronectin on the surface of PFC-MPs followed by coculturing these particles with two immortalized human liver cells including HepG2 and HSC.We characterized liver-specific synthetic functions and cell adhesion patterns, which allowed us to collect evidence that different ECM proteins presented from PFC-MPs have distinctive roles in the physiological regulation of liver cells in vitro 3D culture.2| EXPERIMENTAL SECTION2.1 | Microparticle synthesis and characterizationPFC-MPs were synthesized using a T-shaped polydimethylsiloxane (PDMS) droplet generator as previously described(Mansouri et al., 2022).Briefly, using oil (span80, Sigma Aldrich, cat# 1338-43-8) as a continuous phase in the two side channels and polymer solution as the dispersed phase in the middle channel, we produced a steady stream of homogenous PFC-MPs.These MPs were passed through a glass tube and polymerized using the photoinitiator Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP,0.1 wt% in distilled water, Sigma-Aldrich, cat# 85073-19-4) included in the polymer phase, under the irradiation of a lamp (irradiance: 2 µw/cm 2 ).The composition of MPs was characterized via X-ray photoelectron spectroscopy (XPS) with a scanning monochromated Al Kα (117.40 eV; 25.0 W; spot size, 100 μm).The take-off angle between the sample surface and analyzer was 45.0°, and the X-ray beam collected C1s, N1s, F1s, and O1s elemental information.We also tested the MPs using thermogravimetric analysis (TGA) to study their thermal stability.5 mg of PFC-MPs were weighed and placed in an open pan (platinum 100 µL) attached to a microbalance (TGA550, TA Instruments).The sample was heated at 10°C min −1 from 25°C to 700°C under dry nitrogen in standard mode with a ramp test type.
phenanthroline) ruthenium (II) dichloride (Santa Cruz, cat#sc-216023B), or RuDPP, as an oxygen sensing molecule.RuDPP shows a large Stokes shift by emitting orange light (610 nm peak) when excited with blue light (455 nm peak).We used a custom-made chamber with two inlets for sample and gas (Figure S1), a CCD camera (Hamamatsu ORCA-R2/model: C10600-10B), and HCImage Live software (version 4.8.3.0,Hamamatsu Corporation) to measure light intensity.1 mg of RuDPP was dispersed in 1X phosphatebuffered saline (PBS) solution (pH = 7.4) with or without PFC-MPs using 4X magnification.The fluorescent signals for spheroids cultured with/without PFC-MPs were quantitatively and spatially assessed using ImageJ (National Institutes of Health).To correlate fluorescent intensity to the concentration of oxygen, a calibration curve was generated for RuDPP relating fluorescent intensity versus various oxygen concentrations (O 2 %).Additional details are provided in the Supporting Information document.
The activation of proteins was achieved by mixing 10 μg mL −1 of each respective ECM protein solution for 30 min at room temperature in a solution containing 2 mM N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide (EDC) (Sigma Aldrich, cat# E7750) in ultrapure water (18.2MΩ) and 5 mM N-hydroxysuccinimide esters (NHS) (Sigma Aldrich, cat#130672) in 0.1 M 2-(N-morpholino) ethanesulfonic acid hemisodium salt (MES) buffer (Sigma Aldrich, cat#M3671).PFC-MPs were also dispersed in 0.1 M MES and their pH was adjusted to four different values, including 5.90, 6.20, 6.50, and 6.80 by addition of 1 M NaOH.Afterward, activated proteins were added to the particles to react with the free amino groups available on MP surfaces.The mixture was kept under mild shaking for 5 h at room temperature, and dialyzed for 48 h using SnakeSkin™ Dialysis Tubing (3.5 K MWCO, Fisher Scientific, cat#PI68035) against distilled water included 0.05% triton 100X (Sigma Aldrich, cat#X100-100ML) and then freeze-dried.The amount of protein conjugated to MPs was measured by Bicinchoninic acid (BCA) assay (Thermofisher, cat# PI23235) using manufacturer's directions, as well as via XPS surface analysis, as described above.For simplicity, we picked MPs decorated with laminin-111 as a model to study efficiency of protein grafting.After spheroids were cultured with functionalized MPs, levels of albumin were determined using a commercially available kit (QuantiChrom™ BCP Albumin, BioAssay Systems, cat# DIAP-250) as a measure of liver cells' function.The experiment was performed according to the manufacturer's instructions.Briefly, the medium was collected on Days 5 and 10.A 20 µL aliquot was used to measure albumin secretion into the medium.200 μL of working reagent was added, and after 5 min incubation at room temperature, the optical density at 590-630 nm (peak absorbance at 610 nm) was read by a plate reader (Tecan Infinite M200, Tecan Austria GmbH).Urea was also measured by the QuantiChrom™ Urea Assay Kit (BioAssay Systems, cat# DIUR-100,) using a similar procedure to the albumin kit. 5 μL of sample was mixed with 200 μL of working reagent, incubated for 20 min, and absorbance was read at 520 nm by plate reader.
Studying the arrangement of fluorinated (PFC) chitosan molecules after microparticle (MP) formulation.(b) Thermal degradation study of MPs using thermogravimetric analysis (TGA) to assess their stability.The shaded box shows temperature rang of interest (100°C-150°C).(c) Characterization of PFC-modified and non-PFC-modified MPs using X-ray photoelectron spectroscopy (XPS).The resulting spectrums showed a sharp peak related to PFCs, suggesting availability of PFC groups on the surface.*Refers to fluorine spectrum located at 688.6 eV.activity levels is not a reliable indicator of hypoxic regions inside spheroids.Therefore, we implemented more direct methods to study radial oxygen levels within spheroids to better understand the effects of MPs on oxygen levels in our spheroid model.Thus, we used RuDPP and fluorescent microscopy in new experiments, building upon our earlier validation experiments that were conducted without cells.For the spheroid system, we mixed cell suspension with different amounts of RuDPP to find an optimum concentration required for the formation of spheroids with RuDPP (FigureS1A).
Illustration of custom-built oxygen sensing measurement setup with CCD camera, excitation light, selective filter, and flow chamber.(b) Fluorescence emission spectra at different oxygen concentrations (0% and 100%) excited at 455 nm measured by CCD camera equipped with a telecentric lens.(c) Relative fluorescent intensity of Stern-Volmer equation over time obtained by dividing values of unquenched and quenched fluorescence intensities (I 0 /I), representing the oxygen concentration in the chamber.(d) MP oxygen release kinetics following saturation with 100% oxygen.The intensity of each condition was obtained by averaging the intensity over the channel.MP, microparticle; PFC, fluorinated chitosan molecules.

Following
the confirmation of successful covalent tethering of ECM to MPs, we cocultured modified MPs with liver cells to form spheroids.To study the potential impact of different ECM proteins on spheroid formation and liver-specific functions, we combined hepatocytes and stellate cells with the ECM-modified MPs(Figures S3 and 5a).In culture, viable hepatocytes synthesize and excrete albumin and urea into the surrounding medium.The secretion of albumin reflects the synthetic function of liver cells, whereas the production of urea in HepG2 cells is at low levels and is attributed to the degradation of urea-containing compounds through F I G U R E 3 Visualizing oxygen within liver spheroids using RuDPP as an oxygen sensing dye, which quenches fluorescence emission in the presence of sufficient oxygen.(a) Reconstruction of a threedimensional (3D) spheroid and intensity variance throughout the spheroid cultured for 5 days.PFC-MPs resulted in lower fluorescence intensity compared to the control, which did not have any MPs.(b) Measurement of intensity of individual cells along a circular line in a central plane of spheroids with quantitative data.The average intensity of bright dots around each line was quantified and converted to oxygen percentage (O 2 %) via a calibration curve.Data represented as mean ± SD. n = 3 independent spheroids, one-way ANOVA, and Tukey's post hoc.ANOVA, analysis of variance; MP, microparticle; PFC, fluorinated chitosan molecules; RuDPP, ruthenium complex Tris(4,7-diphenyl-1,10-phenanthroline) ruthenium (II) dichloride; SD, standard deviation.F I G U R E 4 EDC/NHS coupling reaction for bioconjugation of ECM proteins to MPs.(a) Schematic of EDC/NHS reaction between protein carboxyl residues (COOH) and free amine groups (NH 2 ) from PFC-oxygenating MPs in varied pHs.(b) Quantification of covalently tethered protein (laminin 111) via BCA assay kit.Data represented as mean ± SD, n = 4. Statistical analysis via one-way ANOVA with Tukey's post hoc testing with A as the highest mean and C with the lowest mean.(c) Atomic composition of modified and non-modified PFC-MPs measured via XPS.The nitrogen (N1S) percentage correlates to the concentration of proteins attached to the surface.ANOVA, analysis of variance; BCA, bicinchoninic acid; ECM, extracellular matrix; EDC, N-ethyl-N′-(3-dimethylaminopropyl) carbodiimide; MP, microparticle; NHS, N-hydroxysuccinimide esters; PFC, fluorinated chitosan molecules; SD, standard deviation; XPS, X-ray photoelectron spectroscopy.F I G U R E 5 (a) Schematic of experimental procedure for formation of enhanced spheroids using ECM-modified oxygenating MPs and two liver cell lines.(b) Quantification of albumin and urea secretion into the media over the culture period (10 days).Data represented as mean ± SD. n = 3 independent spheroids.Statistics via two-way ANOVA followed by Tukey's post hoc testing.ANOVA, analysis of variance; ECM, extracellular matrix; MP, microparticle; PFC, fluorinated chitosan molecules; SD, standard deviation.
Improved function in response to laminin-modified MPs further strengthened our hypothesis that presented ECM ligands can be differently sensed and translated by cells.To investigate ECM activation events, we used immunofluorescence staining to probe for vinculin and cadherin expression of cells in spheroids.Vinculin is a common component of activated focal adhesion complexes when cells are actively engaging ECM ligands with integrin receptors.Cadherins are essential constituents of adherens junctions, important in the formation and maintenance of cell-cell contacts (Peng et al., 2010).

Figure
Figure 6a shows representative confocal microscope images of liver spheroids cultured with various modified MPs and stained for vinculin.We observed high levels of vinculin expression in the laminin-511 group suggesting better integrin-mediated focal adhesion kinase (FAK) activation and focal adhesion formation.In connection to our findings, the Mitaka group investigated the deposition of laminins containing the α5 chain in the liver cancer environment and concluded that laminin-511 and −521 are major play a central role in compactness and formation of stable liver spheroids.Improved E-cadherin expression prompted us to determine if we could use MPs modified with laminin 511 or 521 instead of HSCs to induce the cells form compact spheroids.Our data suggested that although these MPs lead to smaller-size spheroids and more cell-cell junctions compared to HepG2-only spheroids (FigureS5), they failed to achieve the compactness caused by HSCs.
because PFC-laminin 511/521 MPs stimulate cells to change position relative to their neighbors toward contact strengthening.This finding lends support to prior findings and confirms that interactions between cells and the ECM at integrin-based adhesion sites allow cells to sense their physical surroundings and adjust mechanisms of migration and anchorage(Cavalcanti-Adam et al., 2007).Thus, besides organization, our co culture system provides a similar physicochemical structure that more closely mimics in vivo tissue counterparts.F I G U R E 6 (a) Immunofluorescent staining of fixed spheroids composed of HepG2s:HSCs (4:1), 20k cells initially per spheroid cultured with different oxygenating MPs for 10 days stained with anti-Vinculin (for focal adhesions, yellow) and E-Cadherin (for adherens, purple) using confocal microscopy on central plane.(b) Quantitative analysis of images using ImageJ.IntDen = (sum of pixel values in selection) × (area of one pixel).Statistics via one-way ANOVA and Tukey's post hoc testing with A as the highest mean and D with the lowest mean, p < 0.001.Mean ± SD, n = 2 spheroids with S = 3. ANOVA, analysis of variance; HSCs, hepatic stellate cells; MP, microparticle; SD, standard deviation.In this study, we demonstrated a new strategy for improving cellular assembly and functions inside human cell-based spheroids via synergistically promoting biophysical and biochemical cues using engineered MPs.Our MPs integrated covalently attached PFC groups and ECM adhesive ligands on the surface to enhance oxygen tensions internally while simultaneously improving cell/tissue architecture and functions, respectively.We confirmed the value of PFC-MPs for dissolving and releasing oxygen to enhance oxygen tensions at the cellular level using a RuDPP oxygen-sensitive dye with microscopybased fluorescence sensing.We next observed that cells in assembled liver spheroids responded best to MPs presenting laminin-511 and −521 ECM proteins, which are more prevalent in the mature liver as compared to laminin-111.These laminin isoforms encouraged enhanced phenotypic liver spheroid formation with upregulation of E-cadherin and vinculin expression, as well as greater albumin and urea secretion as compared to MPs presenting other ECMs and the controls.HSCs also arranged in native liver type arrangements when laminin 511/521 conjugated MPs were used as compared to laminin-111, fibronectin, and control groups; providing evidence that ECM proteins have distinct roles in the phenotypic regulation of mature liver-derived cells.
necessary in the future to gain a deeper understanding of the mechanisms underlying the improvement in albumin and urea secretion following the presentation of laminin 511 and 521 ligands.

F
I G U R E 7 Cell arrangement in spheroids using HepG2:HSC (4:1), 20k cells initially per spheroid cultured with different oxygenating MPs.(a) Representative confocal microscopic images of central plane of spheroids stained with CYP3A4 (for HepG2 cells, green), GFAP (for HSC cells, purple), and Hoechst 33342 (for nuclei, blue).Arrows highlight the localization of HSCs.(b) Quantitative analysis of images using ImageJ.Statistics via one-way ANOVA and Tukey's post hoc testing with A as the highest mean and C with the lowest mean, p < 0.001.Mean ± SD, n = 2 with S = 3. ANOVA, analysis of variance; HSCs, hepatic stellate cells; MP, microparticle; SD, standard deviation.