WO2023242189A1 - Collagen visualization on microfluidic device - Google Patents

Abstract

The invention relates to a method for generating an in-vitro cell culture model of cell development or cell differentiation, comprising a) providing a microfluidic chamber through which a stream of cell culture medium can be conducted; b) providing a plurality of primary cells embedded in a synthetic macroporous hydrogel, the macroporous hydrogel comprising polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase, c) applying a flow of cell culture medium to said plurality of stem cells inside of the microfluidic chamber, thereby subjecting said cells to fluid shear stress in order to trigger functional maturation and collagen secretion. Another aspect of the invention relates to a microfluidic device, comprising a microfluidic chamber that comprises a macroporous hydrogel wherein cells are present. The macroporous hydrogel is composed of polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase.

Description

Collagen Visualization on Microfluidic Device The present application claims benefit of the priorities of EP22178671.8 filed 13 June 2022, and EP22183705.7 filed 7 July 2022, both of which are incorporated herein by reference. Field The invention relates to a device, compositions and methods for generating and using biosystems that mimic osteogenesis in humans. The invention provides compositions that can be used to generate macroporous hydrogels comprising mammalian cells, in microfluidic devices that allow the application of biophysical conditions that lead to differentiation of cells similar to the development of tissues in the human body.

Bone development — or osteogenesis — is a complex process that involves changes in cellular behavior and extracellular matrix (ECM) organization induced by an intricate cellular signaling network and various physical factors. At an early stage of osteogenesis, osteoid — a matrix comprised mostly of collagen type I — is formed by a subset of osteoblasts. Mineralization of this matrix causes these cells to become embedded and differentiate into mature osteocytes that form a complex three-dimensional (3D) cellular network within the lacuno-canalicular network (LCN) system by reorganization of their cytoskeletal architecture. It has been shown that fluid shear stress (FSS) in the LCN porosities caused by mechanical loads on bone is in the range of 0.8–2.0 Pa and fluid flow enhances matrix mineralization in 3D perfusion culture. In osteogenesis imperfecta (OI) or brittle bone disease, skeletal deformity and bone fragility are caused by mutations in collagen type I encoding genes affecting collagen quantity and structure. Since many genes are involved in OI, at least 20 different animal models have been established. These models, however, are limited in their translation for human therapy due to interspecific differences. Considering the high cost and ethical constraints of animal models, there is an urgent need for in vitro models representing early stages of bone development. In the last decade, microfluidic technology has found various applications in the biomedical field, especially in tissue engineering in the form of ‘organs-on-a-chip’. These tools provide exquisite control over environmental cues, such as mechanical stimuli and cell-cell interactions. In a study by Xu et al. (L. Xu, X. Song, G. Carroll, L. You, Integrative Biology 2020, 12, 303), a microfluidic chip was fabricated to study osteocyte-osteoclast interaction in 2D under physiological FSS. In contrast to standard dynamic cell culture, microfluidic cell culture requires much fewer cells and reagents while enabling real-time analysis of transient cellular response to drug treatments by high-resolution live imaging. Given these merits, microfluidic organ-on-chip biosystems hold the potential to revolutionize the fields of disease modeling and in vitro drug discovery. Efforts to replicate the complex bone-like tissue environment within (micro-)fluidic systems have been actively sought. In a recent study, Nasello et al. (G. Nasello, P. Alamán-Díez, J. Schiavi, M. Á. Pérez, L. McNamara, J. M. García-Aznar, Frontiers in bioengineering and biotechnology 2020, 8, 336) investigated the effect of seeding density of primary human osteoblasts on cell differentiation in a 3D collagen type I matrix on a microfluidic chip. The authors showed that a high cell seeding density (1×10⁶ ml^-1) promotes osteogenic differentiation although this study was merely conducted in static culture. In another study, Sun et al. (Q. Sun, S. Choudhary, C. Mannion, Y. Kissin, J. Zilberberg, W. Y. Lee, Bone 2017, 105, 245) reported the reconstruction of a 3D osteocyte network using primary human osteoblastic cells within a scaffold made of biphasic calcium phosphate microbeads (20–25 μm) on a microfluidic perfusion system. Although the interstitial spaces between the microbeads promote cell outgrowth, the microbeads are non-degradable. Very recently, Bahmaee et al. (H. Bahmaee, R. Owen, L. Boyle, C. M. Perrault, A. A. Garcia-Granada, G. C. Reilly, F. Claeyssens, Frontiers in bioengineering and biotechnology 2020, 8, 1042) combined porous polymerized high internal phase emulsion (polyHIPE) with a microfluidic chip for long- term 3D perfusion culture of mesenchymal progenitor cells. Using silk fibroin scaffolds under loading in a spinner flask bioreactor, Akiva et al. (A. Akiva, J. Melke, S. Ansari, N. Liv, R. van der Meijden, M. van Erp, F. Zhao, M. Stout, W. H. Nijhuis, C. de Heus, Advanced Functional Materials 2021, 31, 2010524) recently reported the in vitro differentiation of top-seeded human mesenchymal stem cells (hMSC) into a functional 3D co-culture of osteoblasts and osteocytes mimicking woven bone formation on a larger scale. To date, however, existing on-chip models often lack key architectural cues as well as fluid mechanics inside an osteoid tissue and fail to obtain a functional 3D cellular network. Furthermore, most conventional polymeric biomaterials are not enzymatically degradable and thus non-permissive for cell-matrix remodeling through proteolysis. Based on the above-mentioned state of the art, the objective of the present invention is to provide means and methods to enable applying the biophysical conditions underlying connective tissue differentiation and bone formation in an in-vitro system. This objective is attained by the subject-matter of the independent claims of the present specification, with further advantageous embodiments described in the dependent claims, examples, figures and general description of this specification. Any patent document or scientific publication mentioned in the present specification is to be deemed incorporated herein by reference in its entirety. Summary of the Invention Herein, the inventors report a 3D microfluidic perfusion culture that combines a synthetic void- forming hydrogel with a dynamic 3D multicellular environment and FSS to closely resemble the early stage of bone formation on a chip. This macroporous hydrogel enables rapid formation of a 3D cellular network within 24 h as well as facile visualization of cell-secreted collagen fibers due to its synthetic nature. Optimization of gel stiffness, biodegradability and permeability allows for successful integration with a microfluidic chip and time-lapsed imaging of interstitial fluid flow under perfusion culture. Furthermore, the experiments reported herein demonstrate flow-enhanced maturation of 3D bone cellular networks and matrix mineralization after 13 days using physiological FSS as calculated by a computational fluid dynamics (CFD) model. Altogether, the invention provides an in vitro tool to mimic early bone formation in a synthetic environment, opening avenues for the study of human bone (patho-)physiology in the future. In one aspect, the invention relates to a method for generating an in-vitro cell culture model of cell development or cell differentiation, said method comprising a. providing a microfluidic chamber through which a stream of cell culture medium can be conducted; b. providing a plurality of mammalian primary cells embedded in a macroporous hydrogel, the macroporous hydrogel comprising polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase, c. applying a flow of cell culture medium to said plurality of primary cells inside of the microfluidic chamber, thereby subjecting said cells to fluid shear stress in order to trigger functional maturation and matrix secretion. A related aspect of the invention relates to a method for imaging or assaying cell development, cell differentiation and / or collagen secretion. An in-vitro cell culture model of cell development or cell differentiation by a method according to the preceding aspect is used to grow cells, and to and visualize cell differentiation and collagen secretion by optical methods inside the microfluidic chamber. Another aspect of the invention relates to a microfluidic device, comprising a microfluidic chamber. The microfluidic chamber comprises a macroporous hydrogel wherein primary cells, particularly stem cells or osteoblasts, are present. The macroporous hydrogel is composed of polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase. The microfluidic chamber comprises an inlet port and an outlet port, the inlet port being connectable to a cell culture medium influx, and the outlet port allow cell culture medium outflow to leave the chamber. Terms and definitions For purposes of interpreting this specification, the following definitions will apply and whenever appropriate, terms used in the singular will also include the plural and vice versa. In the event that any definition set forth below conflicts with any document incorporated herein by reference, the definition set forth shall control. The terms “comprising”, “having”, “containing”, and “including”, and other similar forms, and grammatical equivalents thereof, as used herein, are intended to be equivalent in meaning and to be open-ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items, or meant to be limited to only the listed item or items. For example, an article “comprising” components A, B, and C can consist of (i.e., contain only) components A, B, and C, or can contain not only components A, B, and C but also one or more other components. As such, it is intended and understood that “comprises” and similar forms thereof, and grammatical equivalents thereof, include disclosure of embodiments of “consisting essentially of” or “consisting of.” Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit, unless the context clearly dictates otherwise, between the upper and lower limit of that range and any other stated or intervening value in that stated range, is encompassed within the disclosure, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the disclosure. Reference to “about” a value or parameter herein includes (and describes) variations that are directed to that value or parameter per se. For example, description referring to “about X” includes description of “X.” As used herein, including in the appended claims, the singular forms “a”, “or” and “the” include plural referents unless the context clearly dictates otherwise. "And/or" where used herein is to be taken as specific recitation of each of the two specified features or components with or without the other. Thus, the term "and/or" as used in a phrase such as "A and/or B" herein is intended to include "A and B," "A or B," "A" (alone), and "B" (alone). Likewise, the term "and/or" as used in a phrase such as "A, B, and/or C" is intended to encompass each of the following aspects: A, B, and C; A, B, or C; A or C; A or B; B or C; A and C; A and B; B and C; A (alone); B (alone); and C (alone). Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art (e.g., in cell culture, molecular genetics, nucleic acid chemistry, hybridization techniques and biochemistry). Standard techniques are used for molecular, genetic, and biochemical methods (see generally, Sambrook et al., Molecular Cloning: A Laboratory Manual, 4th ed. (2012) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. and Ausubel et al., Short Protocols in Molecular Biology (2002) 5th Ed, John Wiley & Sons, Inc.) and chemical methods. Amino acid residue sequences are given from amino to carboxyl terminus. Capital letters for sequence positions refer to L-amino acids in the one-letter code (Stryer, Biochemistry, 3^rd ed. p.21). Lower case letters for amino acid sequence positions refer to the corresponding D- or (2R)-amino acids. Sequences are written left to right in the direction from the amino to the carboxy terminus. In accordance with standard nomenclature, amino acid residue sequences are denominated by the conventional single letter code. The term photoinitiator in the context of the present specification relates to any compound capable of initiating a polymerization reaction when triggered by electromagnetic radiation, particularly by light in the visible or near infrared or near UV spectrum. Non-limiting examples of photoinitiators useful to practice the current invention include single photon initiators such as Li-phenyl-2,4,6trimethylbenzoylphosphinate, 2-hydroxy-^ƍ-(2-hydroxyethoxy)-2- methylpropiophenone (Irgacure 2959), Eosin Yellow + triethanol amine, bisacylphosphineoxide (BAPO) salts such as BAPO-ONa and BAPO-OLi; TPO (diphenyl(2,4,6-trimethylbenzoyl)phosphine oxide) lithium or sodium, VA-086 (2,2'-azobis[2- methyl-N-(2-hydroxyethyl)propionamide]; CAS No. 61551-69-7), and two-photon initiators exemplified by P2CK (3,3’-((((1E,1’E)-(2-oxocyclopentane-1,3- diylidene)bis(methanylylidene))bis(4,1-phenylene))bis(methylazanediyl))dipropanoate), G2CK (sodium 2,2’-((((1E,1’E)-(5-methyl-2-oxocyclohexane-1,3- diylidene)bis(methanylylidene))bis(4,1-phenylene))bis(methylazanediyl))diacetate), DAS ^WHWUDSRWDVVLXP^ ^^^ƍ-(1,2-ethenediyl)bis[2-(3-sulfophenyl)diazenesulfonate]). See also Benedikt et al., J. Polymer Science Part A Polymer Chemistry (2015) Vol.54, 473-479; and Tromayer et al., Polymer Chemistry (2018) 9, 3108-3117. The term “primary cells” relates to cells that are directly isolated from living tissues or organs of an organism. Primary cells are obtained through methods such as enzymatic digestion or mechanical dissociation, allowing them to retain their original characteristics and behaviour. Primary cells are distinct from established cell lines, which are cell populations that have been immortalized through continuous culturing and can undergo unlimited divisions. In contrast, primary cells have a limited lifespan and can undergo a finite number of divisions before entering a state of replicative senescence or cell death. Examples of primary cells include human stem cells, human mesenchymal stem cells, osteoblasts, osteocytes, primary human fibroblasts, epithelial cells, endothelial cells, hepatocytes, neurons, and many others. A linker molecule amenable to cleavage by an extracellular endopeptidase is a linker for which it is known that an endopeptidase exists that can cleave the linker under suitable conditions, as may be the concentration of the linker and enzyme, aqueous conditions and physiological pH. The skilled person is readily able to identify such endopeptidase and a suitable substrate that can serve as a linker as specified herein. Detailed Description of the Invention A first aspect of the invention relates to a method for generating an in-vitro cell culture model of cell development, or cell or tissue differentiation. This method comprises the steps: a. A microfluidic chamber comprising an inlet and an outlet facilitating a stream of cell culture medium is provided. Exemplary chamber designs are shown in WO2016076795A1 and in Shin et al. (Nature Protocols volume 7, 1247–1259 (2012)). The chamber needs to be fitted with windows allowing inspection of its interior by microscopy. b. A plurality of mammalian primary cells, particularly stem cells, osteoblasts or osteocytes, embedded in a macroporous hydrogel are provided inside the chamber. The macroporous hydrogel may comprise polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase. In certain particular embodiments, the extracellular endopeptidase is a subtype of matrix metalloproteinase (MMP). c. A flow of cell culture medium is applied to the plurality of primary cells inside of the microfluidic chamber, thereby subjecting said cells to fluid shear stress in order to trigger functional maturation and matrix secretion. In certain embodiments, the macroporous hydrogel is a synthetic macroporous hydrogel. In certain embodiments, the macroporous hydrogel is an injectable synthetic macroporous hydrogel. In certain embodiments, the mammalian primary cells are derived from a human subject. In certain particular embodiments, the mammalian primary cells are human stem cells. In more particular embodiments, the mammalian primary cells are human mesenchymal stem cells. In certain particular embodiments, the mammalian primary cells are human osteoblasts. In certain particular embodiments, the mammalian primary cells are human osteocytes. In certain particular embodiments, the mammalian primary cells are mammalian stem cells. In certain embodiments, the primary cells are selected from the group of mesenchymal stem cells, fibroblasts, osteoblasts, osteocytes, neurons, and epithelial cells. In certain embodiments, the cell culture medium may comprise pharmacologically active molecules, such as growth factors, or drug molecules. In certain particular embodiments, the medium may comprise a drug candidate molecule under evaluation. This aspect of the invention may be regarded as the essential part of a method to assay or study such drug candidate molecules, or to evaluate their efficacy in addressing bone formation or other collagen-related disease. In certain embodiments, the plurality of stem cells in a macroporous hydrogel is generated in a gelation step, by polymerization-induced phase separation process from an aqueous precursor solution comprising the stem cells. In particular embodiments, the aqueous precursor solution comprises a. a first polymer susceptible to in situ crosslinking, and optionally, a crosslinking agent; and b. a second polymer not susceptible to crosslinking, (particularly not susceptible to photo- crosslinking) The first polymer is miscible with the second polymer when the first polymer is not crosslinked. The mixture comprised of the first polymer and the second polymer undergoes phase separation, separating said first and said second polymer into separate phases once the first polymer is crosslinked. The first polymer In certain embodiments, the first polymer is susceptible to photo-crosslinking, and the precursor comprises a photoinitiator. In certain other embodiments, the first polymer is susceptible to thiol-Michael addition. Polymers susceptible to thiol-Michael addition have been described, inter alia, in Lutolf et al., Nature biotechnology 2003, 21, 513. In certain embodiments, the first polymer is a multi-arm vinylsulfone-modified poly(ethyleneglycol) (PEG) or a norbornene-functionalized polyvinyl alcohol, and a thiol crosslinking agent is present in the composition. The thiol crosslinking agent can be a dithiol agent such as PEG-2-SH, or a di-cysteine peptide; DOWHUQDWLYHO\^^PDFURWKLROV^PD\^EH^HPSOR\HG^^ZKLFK^FRPSULVH^^^^-SH groups per polymer, such as four-arm PEG thiols, PVA macrothiols and hyaluronan macrothiols. The use of macrothiols such as 4-arm PEG thiols can significantly increase the crosslinking efficiency. Without wanting to be bound by theory, the inventors assume that such macrothiols substantially decrease the formation of intramolecular loops during in situ crosslinking with multi-arm vinylsulfone- modified PEG, thereby enabling the creation of low-defect hydrogels at low polymer contents. The second polymer In certain embodiments, the second polymer is selected from the group comprised of dextran sulfate, chondroitin sulfate, sulfated alginate, dextran, mannuronan, hyaluronan, alginate. In certain particular embodiments, the second polymer is selected from the group comprised of dextran, dextran sulfate, hyaluronan and chondroitin sulfate. In certain particular embodiments, the second polymer is selected from dextran derivatives such as dextran and dextran sulfate. The concentrations and M_w of this second polymer may be tuned to change the pore size and porosity of the macroporous hydrogels. In certain particular embodiments, the second polymer is selected from the group of high Mw polysaccharides exemplified by hyaluronan and alginates. These polymers provide unique mechanical properties (viscoelasticity) to the hydrogel matrices with a half-time (t_1/2) for stress relaxation of 1 – 300 s, particularly 5 – 30 s. In certain embodiments, the precursor solution comprises a. as first polymer, a vinylsulfone-modified poly(ethyleneglycol) (PEG) and as crosslinker, a crosslinker selected from the group comprising i. a crosslinker comprising an endopeptidase recognition oligopeptide and two thiol moieties capable of reacting with the vinylsulfone modified PEG, ii. a dithiol-PEG or tetrathiol-PEG; iii. a mixture of peptide crosslinker and a di-thiol-PEG or tetrathiol-PEG crosslinker with tuned degradation rates; iv. a cysteine-containing RGD peptide for cell attachment; b. as second polymer, dextran and/or hyaluronan (HA). Precursor solutions comprising the crosslinker with an endopeptidase recognition oligopeptide and two thiol moieties are of particular utility to generate biodegradable matrices. Precursor solutions comprising the crosslinker with dithiol-PEG or tetrathiol-PEG are of particular utility to generate non-degradable matrices. In certain particular embodiments, the vinylsulfone-modified PEG is a four-arm-PEG- vinylsulfone. The inventors use a di-Cys peptide crosslinker when introducing specific MMP-sensitivity to the matrix to enable cell-matrix remodelling. Non-MMP-sensitive peptides can be used as a non-degradable control. PEG di-thiol may be used to make non-degradable matrices at significantly lower costs. In certain embodiments, the vinylsulfone-modified poly(ethyleneglycol) (PEG) is present in the precursor solution at 1.5% to 3.0% (w/v). In certain particular embodiments, the vinylsulfone- modified poly(ethyleneglycol) (PEG) is present at 1.8% to 2.5% (w/v). Any concentration given in this specification, unless explicitly stated otherwise, is to be deemed to be given as weight (mass) per volume, in other words a 1% concentration of a polymer is a concentration of 1 g of polymer in a 100ml solution. In certain embodiments, hyaluronate (HA) is present in the precursor solution at 0.15% to 1.0%, and dextran is present at 0.2-2.5%. The inventors have found that the HA provides viscoelastic properties to the crosslinked matrix. In particular embodiments, the viscosities of the precursor solution are in the range of 50 - 10000 mPa.s. In more particular embodiments, the viscosities of the precursor solution are in the range of 100-5000 mPa.s. In even more particular embodiments, the viscosities of the precursor solution are in the range of 200 - 2000 mPa.s. Any viscosity value is determined according to the parameters as reported in Fig.2. High viscosities of the solutions may slow down the cross-linking and yield unstable gels, and make it difficult to fill the device according to the invention. In certain particular embodiments, dextran is present at 0.5 to 2%. In other particular embodiments, dextran is present at 0.5-1%. In more particular embodiments, HA is present at 0.25 – 0.8% or at 0.25-0.5%, and dextran is present at 0.5-1%. In even more particular embodiments, the vinylsulfone-modified poly(ethyleneglycol) (PEG) is present at 1.8% to 2.5%, HA is present at 0.25-0.5% and dextran is present at 0.5-1%. In particular embodiments, the M_w of HA is in the range of 200 – 2000 kDa. In more particular embodiments, the M_w of HA is in the range of 1000-2000 kDa. In particular embodiments, the M_w of dextran is in the range of 5-1000 kDa. In more particular embodiments the M_w of dextran is in the range of 40-500 kDa. In other particular embodiments, the M_w of HA is in the range of 1000-2000 kDa and the M_w of dextran is in the range of 40-500 kDa. According to some embodiments, the HA component may be removed through enzymatic digestion with hyaluronidase (HAse) to enhance gel permeability for perfusion culture. The concentration of HAse may be from 0.2-2 mg/ml, particularly from 0.5-1 mg/ml. The duration of treatment may range from 15-60 min. In particular embodiments, the duration of treatment ranges from 15-60 min at 37 °C. The gelation step may be initiated by irradiation with light; useful photoinitiators are known in the art and include, but are not limited to those mentioned above (see Terms and Definitions). No light is necessary to initiate crosslinking in the PEG thiol-Michael system, the reaction can be initiated by temperature at 37 °C. The pH in the system used in the experimental part is 7.4. The literature reports conditions from pH 7.0 to 9.0. The more alkaline the composition, the faster the crosslinking proceeds. This has to be weighed, however, against the tendency of accessible thiol groups to decrease at high pH due to disulfide formation. In the thiol-Michael system, the gelation can either be triggered by a temperature increase to 37 °C or by addition of base such as triethanolamine (TEOA) buffer, pH 7.4-8.5. In certain embodiments, the flow of cell culture medium is adjusted to generate (in other words: generates) a fluid shear stress (FSS) of 0.5-3.0 Pa, particularly of 1.2 to 2.50 Pa. This value refers to the fluid mechanics inside the porous matrix. The FSS inside the microfluidic device of the invention depends on pore geometry, dimensions, porosity and flow rate. Typical flow rates range from 5 – 500 μl/min. The chip dimensions used in the work underlying the present invention are described in the example part of this specification. One exemplary device which may be modified with the macroporous hydrogel as disclosed herein is described in WO2016076795A1, also published as US10711234B2, incorporated by reference herein in its entirety. In certain embodiments, the mammalian cells are selected from primary cells and cell culture cells. In particular embodiments, the mammalian cells are stem cells. In other particular embodiments, the mammalian cells are human primary osteoblasts. A related aspect of the invention provides a method for imaging or assaying cell development, cell differentiation and / or collagen secretion, comprising generating an in-vitro cell culture model of cell development or cell differentiation by a method according to any one of the preceding embodiments, and visualizing collagen secreted by cells. This method can be used for example to test drug candidates or other investigatory compounds with regard to their ability to modulate disease processes for which the generation of bone or other connective tissue, and in particular, the generation and structure of collagen under conditions of shear stress, is relevant. This provides an important opportunity to avoid costly and ethically problematic animal research. Another aspect of the invention relates to a microfluidic device that comprises a microfluidic chamber. The microfluidic chamber in turn comprises a macroporous hydrogel into which primary cells, such as stem cells or osteoblasts, are embedded. The macroporous hydrogel comprises, or essentially consists of, polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase. This device allows for the study of cell-material interactions by live-cell imaging and on-chip visualization of cell-secreted collagen by means of different optical imaging techniques. The microfluidic chamber comprises an inlet port and an outlet port, the inlet port being connectable to a cell culture medium influx, and the outlet port allow cell culture medium outflow to leave the chamber. The simplest form to generate a chamber for practicing the method of the invention is to situate the macroporous hydrogel between an inlet and an outlet. In this simplest form, a chamber with inlet and outlet is expected to provide the minimal functionality required for the generation of FSS, even though such simple setup might cause some technical difficulties when injecting the hydrogel precursor, since the medium channels are expected to be blocked relatively easily. Moreover, the flow through the gel in such a simple chip would vary a lot depending on the position along the channel since the pressure difference would be larger compared to a more complex geometry, wherein two fluid streams are provided on opposite sides of a chamber, the two streams creating a pressure gradient between them. With such more complex geometry, it is however also possible to create a “direct flow” through the gel by using both inlets of one medium channel and the two ports on the other channel as outlets. In this case, it would not be necessary to use two flows. This is also how the dynamic cell culture study was performed. In certain embodiments, the microfluidic chamber comprising the macroporous hydrogel is situated between a first stream of cell culture medium and a second stream of cell culture medium, the first stream being applied at a higher flow rate than the second stream, thereby generating a pressure gradient between the first and second stream. One exemplary setup suitable for practicing the invention is described in Shin et al. (Nature Protocols volume 7, 1247–1259 (2012)). As mentioned above, it is not necessary to use two parallel streams to cause a pressure difference, and it can simply be done with two inlets in one channel. This setup is very useful to inject the hydrogel into the chip and still leaving the medium channels accessible to flow. In certain embodiments of any aspect of the invention described in here, both of the method according to the invention, or the microfluidic device according to the invention, the linker molecules amenable to cleavage by an extracellular endopeptidase comprise a matrix metalloproteinase-sensitive peptide. One particular embodiment thereof that has proven useful in the hands of the inventors is a matrix metalloproteinase-sensitive peptide that comprises SEQ ID NO 03 (GPQGIWGQ). In particular embodiments, the matrix metalloproteinase-sensitive peptide is a di-cysteine peptide that is or comprises SEQ ID NO 01 (KCGPQGIWGQCK) or SEQ ID NO 04 (GCRD- GPQGIWGQ-DRCG). The dissolution of peptide SEQ ID NO 04 in PBS (pH 7.4) may result difficult, but this can be solved by using 0.2 M triethanolamine (TEOA) buffer (pH 7.4-8.0). SEQ ID NO 01 is labile at least to MMPs 1, 2 and 9. See Patterson and Hubbel, Biomaterials 31 (2010) 7836e7845, which provides further possible sequences, and which is incorporated herein by reference. In particular embodiments, any di-cysteine sequence used herein for crosslinking is N- acetylated and C-amidated. In certain embodiments of any aspect of the invention described in here, both of the method according to the invention, or the microfluidic device according to the invention, the polymer chains comprise a peptide molecule capable of promoting adherence of cells. In particular embodiments thereof, the peptide molecule capable of promoting adherence of cells is a peptide comprising a fibronectin-derived arginyl-glycyl-aspartic acid motif. In more particular embodiments, the peptide is or comprises SEQ ID NO 02 (GRCGRGDSPG) and SEQ ID NO 05 (CGRGDSP). This RGD peptide only has 1 C moiety, and is added to react with only 10%, at most, of vinylsulfone residues. It is not expected to contribute to the crosslinking. Another peptide that has a similar functionality is SEQ ID NO 05 (CGRGDSP). This peptide has better water- solubility than SEQ ID NO 02. The inventors have additional data to emphasize the importance of RGD motifs for cell network formation. This finding is important for bone bioengineering applications since some publications appear to suggest that RGD sites are not needed for 3D neuronal cell network formation (Broguiere et al. Biomaterials 2019, 200, 56). The porous architecture attained by the invention allows fast 3D cellular network formation within short time, particularly within 12-48 h. The ability to tune gel compositions facilitates the generation of tunable porous architecture and morphologies of cellular networks with predictable functions. Wherever alternatives for single separable features are laid out herein as “embodiments”, it is to be understood that such alternatives may be combined freely to form discrete embodiments of the invention disclosed herein. The invention is further illustrated by the following examples and figures, from which further embodiments and advantages can be drawn. These examples are meant to illustrate the invention but not to limit its scope. Description of the Figures Fig.1 shows synthesis and characterization of macroporous PEG hydrogels by polymerization-induced phase separation (PIPS). a) Illustration of in situ void formation by PIPS at 37 °C: upon addition of a matrix metalloproteinase (MMP)- degradable di-cysteine crosslinker (SEQ ID NO 01), 4-arm poly(ethylene glycol) vinyl sulfone (4-PEG-VS) is crosslinked in the presence of dextran and hyaluronic acid (HA), leading to in situ pore formation. RGD peptide: SEQ ID NO 005. b) Time- lapsed confocal microscopy images during gelation showing PIPS between rhodamine-labeled 4-PEG-VS (2.5%), HA (0.5%) and dextran (1%), scale bar: 10 μm. c) Storage modulus (G’) of macroporous hydrogels with varying 4-PEG-VS concentration and MMP-degradable peptide or PEG di-thiol (2.0 kDa) as non- degradable crosslinker after 60 min of crosslinking at 37 °C. Data represented as mean ± standard deviation (SD), n=3. d) Time-sweep plots showing G’ evolution during in situ crosslinking with varying HA concentration at 37 °C: 2.5% 4-PEG- VS, PEG di-thiol, -SH:-ene=4:5. Fig.2 shows the viscoelasticity of macroporous PEG hydrogel matrices as determined by frequency sweep measurements on a rheometer. Temperature = 37 °C, oscillatory strain = 5%, angular frequency = 0.1-100 rad/s. These parameters were used for any viscosity measurement reported herein, unless stated otherwise explicitly. Fig.3 shows the characterization of the porous architecture of void-forming PEG hydrogels (2% 4-PEG-VS, 0.5% HA, peptide crosslinker: SEQ ID NO 01 or SEQ ID NO 04) in function of dextran concentrations and molecular weights. a) Confocal microscopy images of rhodamine-labeled PEG hydrogels formed with varying dextran (40 kDa) concentrations without physical constraints, scale bars: 10 μm. b-c) Quantification of pore radius and porosity of hydrogels formed with varying dextran concentrations. Data represented as boxplots with whiskers indicating min and max values, n=3 images (b) and mean ± SD, n=3 (c). d) Confocal microscopy images of rhodamine-labeled PEG hydrogels (2% 4-PEG-VS, 0.5% HA, peptide crosslinker: SEQ ID NO 01 or SEQ ID NO 4) formed with 1% low M_W (40 kDa) and high M_W (500 kDa) dextran without physical constraints, scale bars: 10 μm. e-f) Quantification of pore radius and porosity of hydrogels formed with low M_W (40 kDa) and high M_W (500 kDa) dextran. Data represented as boxplots with whiskers indicating min and max values, n=3 images (e) and mean ± SD, n=3 (f). Fig.4 shows the characterization of the porous architecture of void-forming PEG hydrogels on a microfluidic chip. a) Confocal microscopy images of rhodamine- labeled PEG hydrogels formed with 1% low M_W (40 kDa) and high M_W (500 kDa) dextran, scale bars: 10 μm. b-c) Quantification of pore radius and pore connectivity of hydrogels formed with low M_W (40 kDa) and high M_W (500 kDa) dextran on chip. Data represented as boxplots with whiskers indicating min and max values, n=3 images. d) Distribution of pore radii (top) and pore connectivity (bottom) in PEG hydrogels with low and high M_W dextran on a microfluidic chip. Fig.5 shows static human mesenchymal stem cell (hMSC) culture within degradable and non-degradable void-forming PEG hydrogels (2.0% 4-PEG-VS, 0.5% HA). a) Illustration of the formation of 3D cellular networks in degradable gels (left) and the degeneration of cellular networks in non-degradable gels (right). b) The impact of gel composition on cell viability on day 2 (cell density: 5×10⁵ ml^-1). Data represented as mean ± SD, n=3. c) The impact of gel composition on mean cell area after 4, 14 and 21 days of 3D culture (cell density: 5×10⁵ ml^-1). Data represented as mean ± SD, n=3. d) Representative maximum intensity projections (MIPs) of actin-nuclei stained cellular networks in degradable and non-degradable g^{els after 8 days of culture (cell density: 3.5×106} _ml ^-1 _{), scale bars: 50 μm. e) 3D} view of actin-nuclei stained embedded cells as shown in d), scale bars: 200 μm. Fig.6 shows a histological analysis of static hMSC culture within MMP-degradable and non-degradable PEG gels: 2.0% 4-PEG-VS, 0.5% HA, and cell density at 3.5×10⁶ ml^-1. a) Microscopy images of osteogenic markers on day 30, including cell morphology determined by confocal microscopy (MIP, scale bar: 100 μm), collagen fiber secretion determined by picrosirius-polarization microscopy (scale bar: 100 μm), matrix mineralization determined by Alizarin red staining (scale bar: 100 μm) and osteocalcin expression by immunohistostaining (MIP, scale bar: 50 μm). b) Quantification of collagen content by fiber hue depending on the matrix degradability at day 8 and day 30, color indicates fiber thickness from green (thin, immature) to red (thick, mature) (n=3). c) Quantification of osteocalcin expression by fluorescence intensity. Data represented as mean ± SD, n=3. Fig.7 shows static human osteoblast (hOB) culture within degradable and non- degradable void-forming PEG hydrogels (2.0% 4-PEG-VS, 0.5% HA). a) Cell viability of hOBs after embedding following 2 days of osteogenic culture. Data represented as mean ± SD, n=3. b) The impact of gel composition on cell morphology on day 2 (cell density: 3×10⁶ ml^-1). Scale bars: 100 μm. c) Immunofluorescence staining of cell proliferation marker Ki-67, scale bars: 100 μm. d) The impact of gel composition and culture duration on cell number. Data represented as mean ± SD, n=3. e) The impact of gel composition and culture duration on cell proliferation. Data represented as mean ± SD, n=3. f) Semi- automatic labelling of cell bodies and dendrites by NeuriteQuant plugin in ImageJ for quantitative comparisons of cell morphologies. Scale bars: 100 μm. g) Quantification of mean dendrite length per cell using NeuriteQuant in degradable and non-degradable hydrogels. Data represented as mean ± SD, n=3. h) Quantification of mean cell area in degradable and non-degradable hydrogels. Data represented as mean ± SD, n=3. Fig.8 Visualization of fluid flow, characterization of gel permeability and computational fluid dynamics simulation. a) Microfluidic setup to visualize fluid flow through PEG gels, Q₁ and Q₂ denote volumetric flow rate, P₁ and P₂ represent positions of image acquisition. b) Time-lapsed fluorescence microscopy images of FITC-dextran (500 kDa) tracer perfusing through PEG gels (MMP-degradable, dextran M_w = 40 kDa) on chip in response to a pressure gradient in position P1, scale bar: 200 μm. c) Changes in normalized fluorescent intensity in position P₁. Data represented as mean ± SD, n=3. Fig.9 shows the permeability of macroporous PEG hydrogels with 1% low M_w (40 kDa) and high M_w (500 kDa) dextran. a) Microfluidic setup for permeability test depending on a gravity-driven fluid flow through the PEG gels, h₁ and h₂ denote measured heights, h_d is the difference to calculate the changes in pressure drop over time. b) Calculated pressure change over time (symbols) and fitted exponential functions (lines) across 40 kDa and 500 kDa dextran PEG hydrogels (n=3) and reference hydrogels made of 2 mg/ml collagen type I (n=3). Data represented as mean ± SD. c) Resulting Darcy’s permeability of the 3 hydrogels obtained from pressure change over time. Fig.10 Schematic representation of microfluidic 3D cell culture by combining a commercial AIM Biotech chip with hMSCs embedded within a macroporous PEG hydrogel. Fig.11 Illustration of a CFD model to simulate FSS within macroporous PEG gels on chip. a) Global multiphasic CFD model of microfluidic device with the whole scaffold (porous media domain) in it and local CFD model of the subsection whose struts geometry is constructed from confocal images of fluorescently labelled PEG gel formed inside the microfluidic chip. b) Pressure distribution within the porous media (homogenized scaffold domain) under the applied flow rate of 10 μl min^-1 per inlet. Fig.12 shows CFD simulation of the local mechanical environments showing the FSS distribution and average FSS (^_^) within 4 subsections (x-y-z: 20 x 20 x 30 μm) under an applied flow rate of 10 μl min^-1 per inlet. Fig.13 Functional adaptation of 3D hMSC culture on chip in response to low FSS (20 μl min^-1) or high FSS (200 μl min^-1) on day 13 in PEG hydrogels (MMP-degradable, cell concentration: 1×10⁶ ml^-1). Control: static culture on chip. a) 3D rendering of confocal microscopy images of actin-nuclei-stained cells, scale bars: 150 μm. b) Confocal microscopy images of actin-nuclei-stained cells (MIP) showing subcellular morphological features, scale bars: 10 μm. c) Phase contrast microscopic images of Alizarin red (ARS) stained samples to determine the level of matrix mineralization, scale bar: 500 μm. d-f) Cell viability as a function of culture conditions. Data represented as mean ± SD, n=3. e) Mean cell area as a function of culture conditions. Data represented as mean ± SD, n=3. f) The impact of FSS on matrix mineralization as determined by color threshold. Data represented as mean ± SD, n=3. Fig.14 shows confocal images of actin-nuclei-stained human osteoblasts following 2 days cultivation in MMP-degradable PEG hydrogels, emphasizing the importance of RGD motifs for 3D bone cellular network formation. Scale bars: 100 μm. Fig.15 shows on-chip cultivation of hMSCs with high (1×10⁶ ml^-1) and low (5×10⁵ ml^-1) seeding density inside MMP-degradable PEG hydrogels on day 7. a) Phase contrast images showing difference in cell number in the hydrogel channel, scale EDU^^ ^^^^ ^P^^ b–c) Confocal microscopy images of actin-nuclei staining (MIP) showing K06&^QHWZRUN^^VFDOH^EDU^^E^^^^^^^P^^F^^^^^^P^ Fig.16 shows fast formation of 3D cellular networks from hMSCs on a commercial chip on day 2. Gel composition: 2% 4-PEG-VS, 1% dextran (500 kDa), 0.5% HA, crosslinker PEG-2-SH (3.4 kDa), thiol:vinyl sulfone = 4:5. Scale bars, 100 μm. Fig.17 shows the effect of dextran concentration (500 kDa) in MMP-degradable PEG hydrogels on human osteoblast morphology on day 2 of osteogenic culture, scale bars: 100 μm. Examples Synthesis and Characterization of Void-Forming PEG Hydrogels An injectable synthetic PEG void-forming hydrogel was designed to generate 3D living cellular networks from hMSCs on a microfluidic chip to mimic an osteoid-like environment in early osteogenesis. PEG hydrogels were formed by thiol-Michael crosslinking (see M. P. Lutolf, F. E. Weber, H. G. Schmoekel, J. C. Schense, T. Kohler, R. Müller, J. A. Hubbell, Nature biotechnology 2003, 21, 513) between 4-arm poly(ethylene glycol) vinylsulfone (4-PEG-VS) and di-thiol crosslinkers in the presence of dextran and hyaluronic acid (HA) for in situ pore formation by polymerization-induced phase separation (PIPS, Fig. 1a). Fibronectin-derived arginylglycylaspartic acid (GRCGRGDSPG SEQ ID NO 02 or CGRGDSP SEQ ID NO 05 ) was added to promote cell attachment, whereas a matrix metalloproteinase (MMP)-sensitive di- cysteine peptide (N-&^^.&*34*Ļ,:*4&.^ SEQ ID NO 01; or GCRD-GPQGĻIWGQ-DRCG, SEQ ID NO 04; Ļ^LQGLFDWHV^FOHDYDJH^VLWH) and PEG di-thiol (PEG-2-SH, M_W=2.0 kDa or 3.4 kDa) were used as degradable and non-degradable crosslinker, respectively. Using fluorescently labelled 4-PEG-VS, dynamic in situ pore formation was evidenced by time-lapsed confocal microscopy. After 1 h of crosslinking, spatial patterns of binodal nucleation were observed in the hydrogel (Fig.1b). PIPS and interconnected porosity were only observed in the gels with the addition of dextran. Rheological analysis (Fig. 1c) revealed that increasing the concentration of 4-PEG-VS from 2.0% to 2.5% yielded significantly stiffer gels for both degradable and non-degradable groups with storage moduli (G’) ranging from ca.20 Pa to 500 Pa. Depending on the gel composition, gelation started after 3-5 min by in situ crosslinking at 37 °C, whereby MMP-degradable gels crosslinked faster than their non-degradable counterparts. Lutolf et al. (ibid.) reported that the crosslinking kinetics of Michael-addition PEG hydrogels can be controlled by the presence of charged amino acid residues in proximity to the cysteine of the crosslinker that modulates the pKa of the thiol group. Since thiolates rather than thiols are the reactive species in a thiol- Michael reaction, a decrease in pKa resulting from the addition of positively charged amino acids in proximity to the cysteine results in an increase in polymerization rate. In contrast to PEG-2-SH, the MMP-sensitive peptide crosslinker comprising positively charged lysine residues on each end next to the cysteine accelerated the crosslinking with 4-PEG-VS. By varying the concentration of HA and thereby the viscosity of the gel precursor, the crosslinking kinetics (Fig. 1d) as well as the stiffness of the macroporous hydrogel could be tuned. High viscosity has been suggested to prevent the phases from collapsing into microspheres before the structures are stabilized by crosslinking in PIPS. (Broguiere et al., Biomaterials 2019, 200, 56) Our findings show that the inclusion of HA accelerated the crosslinking when its concentration was increased from 0.25% to 0.50%. However, higher concentration of HA (0.83%) reduced both G’ and loss modulus (G”), indicating that crosslinking was less efficient for highly viscous formulations. To integrate the macroporous hydrogel with a 3D microfluidic perfusion cell culture, we reason that an intermediate stiffness (2.2% 4-PEG-VS and 0.5% HA) is stable enough to be casted into the gel chamber of the microfluidic device while being sufficiently permissive to allow for rapid cell spreading and 3D cellular network formation. Interestingly, a rheological test evidenced the viscoelasticity in the PEG hydrogel matrices for both degradable and non-degradable groups, presumably due to the HA content (Fig.2). This property may facilitate fast cell-matrix remodeling and 3D cellular network formation. Since not only matrix stiffness but also architectural cues such as porosity are crucial for successful generation of 3D cellular networks inside hydrogels, the porosity after PIPS was analyzed by confocal microscopy imaging. We investigated the impact of dextran inclusion on pore architecture in PEG hydrogels. As shown in Fig. 3a-c, the increase of dextran concentration from 0% to 0.5%, 1% and 2% led to a significant increase in pore radius and porosity. We reason that macroporous hydrogels with pore sizes larger than 10 μm are extremely useful for solute transport and microfluidic perfusion culture. Furthermore, the increase of dextran M_w from 40 kDa to 500 kDa (Fig. 3d-f) resulted in an increase of pore radius. But the resultant porosity in 500 kDa dextran group seems to be smaller than that of 40 kDa group. Furthermore, we assessed the effect of physical constraints inside a microfluidic channel on pore formation (Fig. 4a). For quantification, confocal images were used as an input for an algorithm (Vandaele et al., Soft Matter 2020, 16, 4210) to obtain pore radius (Fig.4b, d) and pore connectivity (Fig.4c, d). Despite spatial constraints, the pore radius for high M_w dextran group was larger than that of low M_w dextran group. Moreover, pore connectivity was increased as the increase of dextran M_w from 40 kDa to 500 kDa. These findings show that described PEG hydrogels are void-forming even when casted on a chip with physical constraints, thereby enabling microfluidic perfusion cell culture. Rapid Formation of 3D Bone Cellular networks and Functional Maturation in a Static Culture To assess whether the void-forming hydrogel is permissive for in vitro generation of 3D cellular networks, hMSCs were embedded inside MMP-degradable and non-degradable PEG gels. MMPs are known to be crucial for hMSC differentiation as well as for osteoblast survival. Therefore, we reason that only when cells are able to remodel their surrounding matrix through proteolysis by MMPs, they form an interconnected 3D cellular network with long-term stability (Fig.5a). Cell viability on day 2 was above 80% in both degradable and non-degradable gels (Fig.5b). Notably, ultrafast cell spreading as well as 3D cellular network formation within these macroporous gels were observed as early as after 1.5 h and 24 h, respectively. Cell morphology analysis indicated that only when the gels were composed of MMP-sensitive peptide motifs and therefore degradable by cells, hMSCs could remodel their pericellular environment and maintain a 3D cellular network for at least 35 days. Quantification of mean cell area further evidenced the permissiveness of degradable gels (Fig.5c). The average cell area in the degradable gels was significantly larger compared to non-degradable gels. By increasing the cell seeding density to 3.5×10⁶ ml^-1, an extensive 3D cellular network was formed in the degradable gels on day 8. By contrast, the extent of cellular network formation was significantly less in the non-degradable gels. Fig. 5d and Fig. 5e shows the maximum intensity protections (MIPs) and 3D renderings of actin-nuclei stained cells, respectively. When grown in a more permissive environment, hMSCs formed multiple dendritic processes connecting them with neighboring cells. As the osteogenic differentiation proceeded, the cell- laden hydrogels became increasingly opaque. As such, histological sections of the same cultures on day 8 and day 30 (Fig. 6) were prepared to further compare cell morphology, collagen fiber secretion, matrix mineralization and osteocalcin expression (marker for mature osteoblasts). Similar as in the non-sectioned samples, the results show that degradable gels facilitated pronounced network formation on day 8 that remained stable for at least 30 days whereas non-degradable gels did not. Picrosirius red-polarization imaging (Rittié, in Fibrosis, Springer, 2017) revealed the presence of cell-secreted collagen fibers on day 8, especially in the MMP-degradable gels due to its permissiveness for cell-matrix remodeling. Collagen type I is the major ECM protein secreted by osteoblasts and therefore the main component in osteoid. Importantly, collagen secretion could be assessed within a macroporous PEG hydrogel due to its synthetic nature considering all detectable collagen fibers should be produced by the embedded cells. Therefore, this hydrogel holds great potential to be used as a 3D matrix for imaging and assessing cell-secreted collagen in human diseases such as rare bone diseases and fibrosis, which is unachievable in conventional proteinaceous hydrogels such as collagen type I and gelatin derivatives. Next, collagen content and maturity were quantified based on the fiber hue method as described elsewhere. (L. Rich, P. Whittaker, Journal of morphological sciences 2017, 22, 0) Fig. 6b shows that more green and yellow color corresponding to low fiber thickness and immature collagen was present in the non-degradable gels. In contrast, cells within the MMP- degradable gels produced more mature collagen fibers as indicated by the larger proportion of red and orange color. Moreover, the red color content significantly increased from day 8 to day 30 (p<0.01). Alizarin red staining further indicated more pronounced matrix mineralization especially in close proximity to embedded cells within the degradable gels compared to non- degradable ones. Compared to day 8, an increase in mineral deposition on day 30 implies the 3D osteogenic differentiation of hMSCs into a mature bone cell phenotype. Osteocalcin, a mature marker for osteoblasts, was predominantly expressed in MMP-degradable gels after cultivation for 30 days. In contrast, only limited expression of osteocalcin was observed in the non-degradable gels (Fig. 6c). Together, these results suggest that the MMP-degradable macroporous PEG gels are extremely permissive for the formation of 3D bone cell networks and subsequent osteogenic differentiation under a static culture. Since cell seeding density has been shown to impact the differentiation of primary human osteoblasts within a collagen type I hydrogel, we further investigated the effect of two cell seeding density (low - 5×10⁵ ml^-1 and high - 1×10⁶ ml^-1) on 3D cellular network formation within PEG gels. Actin-nuclei staining after 7 days of cultivation in osteogenic medium revealed that single hMSCs at high density exhibited a dendritic morphology and formed a 3D cellular network, whereas at low seeding density, cells appeared elongated and failed to form intercellular connections. Since higher cell seeding density enhanced cell network formation notably without compromising gel stability, a seeding density of 3×10⁶ ml^-1 was eventually selected for on-chip integration. Based on the success of hMSC cultures, we further tested the suitability of macroporous PEG hydrogels for 3D osteogenic culture of primary human osteoblasts (hOBs). hOBs were embedded at a cell density of 3.0×10⁶ ml^-1 in either MMP-degradable or non-degradable hydrogels and cultured in an osteogenic medium (Fig.7). Cell viability after embedding was above 90% in both groups and remained high after 2 days of culture (Fig. 7a). Similar to hMSCs, a difference in cell morphology between degradable and non-degradable hydrogels was observed (Fig.7b). To assess the impact of gel degradability on cell proliferation, hOBs were stained for the cell proliferation marker Ki-67 (Fig.7c). Even though the proportion of Ki- 67 positive cells was similar in both gels after embedding (Fig.7e), an increase in cell number within degradable gels compared to non-degradable gels was observed on day 2 of culture (Fig.7d). Moreover, the proportion of proliferating cells decreased in the degradable gels from day 0 to day 2, implying the progress of osteogenic differentiation. Segmentation of cells into cell bodies and dendrites using the ImageJ plugin NeuriteQuant (Fig. 7f) and subsequent quantification of dendrite length, show that hOBs within degradable PEG gels have longer dendrites (Fig.7g) even though the mean cell area is similar in both conditions (Fig.7h). In summary, our data shows the presence of an MMP-sensitive crosslinker within macroporous PEG hydrogels seems to be beneficial for hOBs to form 3D cell networks and exhibit a morphology resembling the osteocyte networks in vivo. Next, the effect of fibronectin-derived RGD motifs for 3D cellular network formation was investigated. Broguiere et al. showed the formation of 3D neuronal networks does not request adhesive motifs in a non-degradable macroporous PEG hydrogel. The results show that hOBs formed a 3D cellular network after 2 days cultivation in RGD-functionalized adhesive hydrogels, but not in non-adhesive hydrogels (Fig.14). Fig.15 depicts the effect of cell seeding density on 3D cellular network formation. The higher cell concentration seems to promote 3D cell-cell contacts using an actin-nuclei staining. Fig. 16 shows the feasibility of the formation of interconnected 3D cellular networks from hMSCs on a commercial microfluidic chip on day 2. Fig.17 shows the effect of dextran concentration on the morphology of embedded hOBs after 2 days of osteogenic culture. Higher concentrations of dextran (1.0%) – yielding larger pore sizes – allow for extensive cell spreading and network formation whereas in hydrogels with smaller pores created by the addition of 0.2% dextran cells appear more round with limited spreading. Fluid Flow Visualization and Permeability Quantification For dynamic microfluidic cell culture, gel permeability is a critical property to determine the level of FSS. Flow visualization showed that interstitial liquids could pass through the macroporous PEG gels when applying a pressure gradient on a microfluidic setup (Fig.8a). The pressure difference across an acellular MMP-degradable gel was created by applying dissimilar flow rates (Q₁>Q₂) to the two medium channels. The perfusion of different FITC- dextran tracer molecules (70 kDa and 500 kDa) was visualized by fluorescence microscopy. Time-lapsed microscopy images in position P₁ indicate that fluorescence intensity increased gradually for both tracers over time. The intensity of tracer molecules increased gradually until it reached a plateau after 4–5 min (Fig.8c). Yet, no statistical difference in the intensity change was observed between the two groups. We thus reason that pores within PEG gels were large enough for both sizes of tracer molecules to diffuse through. These results correlate with the pore size quantification (Fig.3^^WKDW^SRUH^VL]HV^LQ^3(*^JHOV^DUH^LQ^WKH^RUGHU^RI^^P^^ZKLFK^DUH^ much larger than the sizes of tracers. To quantify gel permeability, a microfluidic setup (Fig.9a) was employed to create a pressure difference across the hydrogel caused by different volumes of medium added to each syringe barrel. The change in pressure difference over the hydrogel was monitored and an exponential GHFD\^DFFRUGLQJ^WR^ǻP(t^ ǻP(0)×e^-ct was fitted as depicted in Fig.9b. The values for exponent coefficient c and calculated Darcy’s permeability K are summarized in Fig. 9c and Table 1. The addition of 500 kDa dextran increased the permeability of the macroporous hydrogel compared to 40 kDa dextran. Importantly, the permeability is comparable to that of a reference hydrogel made of collagen type I. We thus envisage that these PEG gels hold great potential for future on-chip 3D perfusion culture studies where mechanistic understanding of cell-matrix interactions is gained using a chemically defined macroporous hydrogel. Table 1. Exponent coefficient and resulting permeability of various hydrogels: PEG hydrogels (MMP-degradable, 2.0% 4-PEG-VS, 0.5% HA, n=3) formed with 40 kDa or 500 kDa dextran and a reference hydrogel made of 2 mg ml^-1 collagen type I (n=3). Hydrogel Exponent coefficient c [s^-1] Darcy’s permeability K [m²] PEG + dex40k 0.032 7.61×10^-14 PEG + dex500k 0.125 2.97×10^-13 Collagen I 0.160 3.83×10^-13 Effect of FSS on Bone Cellular network Maturation After imaging the interstitial fluid flow inside the porous gels on chip, we further investigated how controlled delivery of FSS impacts the functional maturation of bone cellular networks in a dynamic culture. A commercial microfluidic chip was used to embed hMSCs in the central channel within the macroporous PEG gel (Fig.10). Cells were mechanically stimulated for 2×10 min per day with either a low flow rate (Q_low=10 μl min^-1 per inlet) or high flow rate (Q_high=100 μl min^-1 per inlet), while cells cultivated under static conditions were selected as the control. We applied a multiscale and multiphase computational fluid dynamics (CFD) model (F. Zhao et al., Biomechanics and modeling in mechanobiology 2019, 18, 1965.) to estimate the FSS within the macroporous hydrogel. Using the permeability values of macroporous PEG gels, a global model on the microfluidic chip (Fig. 11a) was generated to estimate the pressure gradient across the hydrogel region by applying two different flow rates (Qlow = 10 μl min^-1 and Q_high = 100 μl min^-1 per inlet). According to CFD calculations, the pressure drop over a hydrogel subsection in the center of the channel at Q_low was 196 Pa (Fig.11b). This value was used for defining the loading conditions of a local model as illustrated in Fig.12. After solving the local CFD model of 4 discretized subsections, the FSS distribution in each subsection is shown in Fig.12. The average FSS values (^_^) for Section 1-4 was 2.12 Pa, 1.38 Pa, 1.68 Pa and 1.79 Pa, respectively. The overall average FSS at Q_low was 1.74 Pa. Considering the FSS is proportional to the applied flow rate, the average FSS at Q_high was 17.43 Pa. These results imply that the majority of the FSS values inside the porous geometry at Q_low is within the physiological range for bone tissues^[4a], whereas it is far exceeded at Q_high. 3D bone cellular networks after 13 days of cultivation are displayed, showing the impact of FSS on cellular network morphology (Fig.13a), single cell morphology (Fig.13b), and matrix mineralization (Fig.13c), respectively. A live/dead assay (Fig.13d) showed a comparable cell viability for static and low FSS culture, whereas cell viability was significantly reduced under high FSS. Indeed, laminar flow at physiological levels has been shown to have an anti-apoptotic effect on MSCs and osteocytes. It is likely that local shear stresses at Q_high were far beyond the physiological range and thus caused apoptosis. These findings correlate with the CFD calculation that the average FSS IJ_a at Q_high was 17.43 Pa which exceeds the physiological range (0.8–2.0 Pa) by more than 8 times. Local shear stresses within smaller pores were even higher and therefore resulted in more apoptotic cells compared to static and low FSS groups where the average FSS IJ_a was estimated to be 1.74 Pa. It is important to note that these results were calculated based on the confocal images of the porosity in acellular gels without HAse treatment. A major limitation of this CFD model is attributed to the dynamic cell-matrix remodeling and resultant changes of pore geometry over time. To more accurately quantify the FSS, a real-time confocal image- based CFD model is warranted in future investigations. Actin-nuclei staining on day 13 (Fig. 13a-b) shows that hMSCs cultivated under dynamic conditions maintained their 3D interconnected network morphology over time. In static culture, only cells close to the medium channels remained interconnected and lost their protrusions in the center of the gel channel, presumably due to the lack of nutrient supply and mechanical stimuli. Quantification of mean cell area confirmed enhanced cell-cell contacts as shown in Figure 13e. For the loaded cells, up to 10 μm long stress fibers were observed. However, a detailed cell morphometry analysis is warranted to investigate how bone cells undergo architectural changes in response to mechanical stimulation by FSS. Finally, the level of matrix mineralization was analyzed by Alizarin red staining (Fig. 13c, f). Interestingly, only hMSCs following 13 days culture with low FSS were notably stained, implying the highest amount of mineral deposition. It is widely appreciated that matrix mineralization is coordinated by osteoblasts in bone formation. Our results imply that hMSCs stimulated with low FSS (i.e., average value = 1.74 Pa) differentiated into osteoblastic phenotypes and induced matrix mineralization. Furthermore, it has been shown that high levels of FSS (>1.0 Pa) can cause apatite crystals to be less organized, thereby inhibiting mineralization which might explain the results observed herein in addition to the lower cell viability for the high FSS group. Together, these findings from dynamic cell culture show that mechanical stimulation of hMSCs by perfusion with low FSS induced osteogenic differentiation, maintained 3D cellular network, and promoted matrix mineralization. Discussion This work provides a novel in vitro platform for the investigation of early-stage bone formation using a synthetic macroporous hydrogel in which an interconnected 3D bone cellular network can form rapidly within 1–2 days. Within the MMP-degradable matrices, single hMSCs and hOBs sense the porous architecture to form an interconnected cellular network in 3D and then differentiate into an osteoid-like tissue. Unlike animal-derived Collagen or Matrigel that suffer from batch-to-batch variations, macroporous PEG hydrogels are chemically defined. In addition, the in situ PIPS process allows for the formation of interconnected pores in the presence of living cells, which is unachievable with other types of macroporous hydrogels formed by emulsification, porogen leaching and particle annealing. The established tool could be used in the future to investigate the mechanisms of cell-matrix interactions as well as matrix defects in musculoskeletal disorders such as OI. Given the initially unmineralized nature and low stiffness, this void-forming PEG hydrogel closely mimics the properties of an osteoid tissue. Therefore, it holds great potential to embed patient-derived cells for disease phenotyping and drug screening towards personalized in vitro models and treatments. Compared to traditional bioreactors, much fewer cells (1×10⁴ instead of 1×10⁶ per sample^[13]) and lower quantity of reagents are needed in an on-chip culture, making it a cost-efficient in vitro tool and offering the promise to replace animal experiments in the spirit of 3Rs principle. In order to further investigate cellular phenotype, future studies on the expression of osteocytic markers such as DMP-1 or sclerostin are warranted. Moreover, changes in dendritic cell morphology in response to FSS, fluid dynamics within the microfluidic device containing a 3D cellular network as well as cell-secreted ECM requires future investigations through long-term on-chip cultures. Conclusion In summary, the inventors have developed a 3D microfluidic perfusion culture by interfacing injectable synthetic void-forming hydrogels with FSS for the study of early bone development in vitro. Using a macroporous hydrogel, physiologically relevant mechanical stimuli can be delivered to the 3D environments to promote cell maturation and matrix mineralization over time. Furthermore, the synthetic nature of this hydrogel enables visualization and quantification of cell-secreted proteins such as collagen fibres as a potential biomarker for studying the (patho)physiological conditions in bone formation. Therefore, this organ-on-a-chip technique represents a promising in vitro tool to dissect the fundamental mechanisms of a variety of human bone diseases at cellular and molecular level for the replacement of animal models in the future. Experimental Section Hydrogel Preparation: Rhodamine-labeled 4-PEG-VS and 4-PEG-VS were synthesized according to the protocol by Broguiere et al. (ibid.) detailed below. Unless otherwise noted in the figure caption, final concentrations were 2.2% 4-PEG-VS, 1% dextran (low M_w: 40 kDa or high M_w: 500 kDa), 0.5% HA and a thiol/ene ratio of 0.8 between 4-PEG-VS and crosslinker and 0.07 between 4-PEG-VS and RGD. RGD and crosslinker stock solutions were prepared directly before mixing the precursor solution and kept on ice to prevent rapid oxidation of thiol groups. For cell culture, hMSCs (Lonza, PT-2501) were mixed into the HA stock solution to obtain final cell concentrations of 5×10⁵–5×10⁶ ml^-1. Once the gel precursor solution was prepared, it was rapidly cast into either a custom-made round poly(dimethylsiloxane) (PDMS) mold (h=0.5 mm, d=5mm, 15 μl hydrogel solution) on a confocal dish (VWR, 734-2905) for static cell culture or into the central gel channel of a 3D cell culture chip (AIM Biotech, DAX-1, 10 μl hydrogel solution) inside the corresponding chip holder (AIM Biotech, HOL-1 or HOL-2) for static and dynamic culture according to the manufacturer’s protocol. Gels were crosslinked DW^^^^Û&^DQG^^^^&2₂ for at least 60 min. After crosslinking, gels were washed 1–2 times with phosphate buffered saline (PBS, Gibco, 10010-015, pH 7.4), to remove non-reacted RGD. On- chip samples undergoing HAse treatment were subsequently incubated with 1 mg ml^-1 HAse (Sigma-Aldrich, 37326-33-3) in PBS for 30–60 min before washing twice with PBS. Characterization of Porosity: For time-lapsed imaging of PIPS, Rhodamine-labeled 4-PEG-VS was used when preparing the hydrogel precursor solution in order to image the 4-PEG-VS (2.5% w/v) phase in the gel. The hydrogel was crosslinked directly in a confocal dish in a custom-made PDMS mold inside a confocal microscope (Leica SP8) at 37°C. Using a 63× oil immersion objective with 1.4× zoom, z-stacks of 35 μm were obtained every 2 min for 90 min. For a quantitative analysis of pore size and connectivity, hydrogels with rhodamine-labeled 4- PEG-VS (2.2% w/v) were casted into the microfluidic chip (with spatial constraints) and in the center of an Ibidi μ-slide with 8 wells (80821, without constraints). Gels were crosslinked for 90 min at 37°C. They were then washed twice with PBS and imaged using the same confocal microscope. Z-stacks of 30 μm were obtained and deconvoluted using Huygens Professional Software. These images were then processed with the MATLAB algorithm developed by Vandaele et al. (ibid.) that uses image segmentation and approximation of pores by spheres to quantify pore size, connectivity and other determinants of the porous architecture. An adaptive threshold sensitivity of 0.6 was chosen and pixel size was set to 72.22 nm. Pore size (external pore diameter in 3D) and pore connectivity (number of neighboring pores) were obtained for each group and 3D models of pore connectivity were created. Visualization of Fluid Flow on Chip: MMP-degradable acellular PEG gels (2.2% w/v 4-PEG- VS) were casted into microfluidic chips with attached luer connectors (AIM Biotech, LUC-1). After hydration for 24 h, flow imaging was performed on a wide field microscope (Olympus, IX83). Two different tracer solutions were prepared by diluting a 0.1% (w/v) stock solution of 70 kDa or 500 kDa FITC-dextran (both Sigma-Aldrich, FD70S-100MG and FD500S-100MG) 1:1000 in phenol red free DMEM. The chips were then mounted onto the microscope stage, the tracer solutions were filled into 20 ml syringes which were connected to the luer connectors on the chips via needles (0.80×22 mm, blunt, Braun), tubing (Semadeni Plastics, 4348) and male luer connectors (Cole Parmer, 45518-07) that were primed with liquid to prevent air from being trapped along the flow path. Using 2 syringe pumps (WPI, AL-1000), an interstitial flow was created by applying different flow rates (Q₁=334 μl min^-1 and Q₂=10 μl min^-1) to the medium channels. The perfusion of the PEG gel with tracer molecules was imaged in two different positions using a filter for FITC and a 20× air objective every 20 s for 8 min. The pumps were switched on between time-point 2 and 3. For quantification, the mean intensity in a rectangular area at positions P₁ and P₂ as measured in Fiji/ImageJ for each time-point and both tracers. This intensity was normalized by the mean intensity at time-point 0 for each tracer and position. Quantification of Permeability: In order to quantify the permeability of the macroporous PEG gels, the method described by Moreno-Arotzena et al. (Materials 2015, 8, 1636) was adapted. Acellular PEG gels and collagen type I hydrogels were used.2 mg ml^-1 collagen type I gel was prepared from an 8.91 mg ml^-1 stock solution (rat-tail, Corning, 354249) as described by Shin et al. (Nature protocols 2012, 7, 1247) and casted on-chip. To determine the permeability, all gels were first hydrated in PBS for 24 h after crosslinking. Medium reservoirs of an Ibidi μ-slide were filled with 60 ^O^SKHQRO^UHG^IUHH^'0(0^DQG^^^PO^V\ULQJH^EDUUHOV^ZLWKRXW^SOXQJHU^ZHUH^ then attached to the luer connectors.0.5 ml phenol red free DMEM were added to one of the barrels and 0.1 ml to the other one creating a height and pressure difference that caused interstitial flow through the gel. The heights h₁ and h₂ were measured to calculate the difference in height h_d. From this, the pressure difference ¨P was calculated using Equation 1, where g is the gravitational acceleration and ^ is the GHQVLW\^RI^WKH^PHGLXP^^^_DMEM=1000 kg m^-3). ǻ3 ^îJîKd (1) The change of height was measured every 15 min for the first hour and then every 30 min for ^^ K^ LQ^ WRWDO^^ $Q^ H[SRQHQWLDO^ IXQFWLRQ^ ZDV^ ILWWHG^ WR^ WKH^ GDWD^ DFFRUGLQJ^ WR^ ǻP(t)=P(0)×e^-ct to determine the exponent c. Since ¨P changes rapidly within the first hour, only data from time- points 0.75–4 h was considered. Permeability K and this constant c are related according to Darcy’s law in Equation 2, where ^ is the viscosity of DMEM (7.8×10^-4 Pa s), l is the length of the gel channel (1.70×10^-2 m), A_r is the cross section of one syringe barrel and A is the cross section of the gel channel in the direction of fluid flow. ^ = ^{^×ఓ×^×^^×ସ} ఘ_×^×^ (2) Histology and collagen quantification: Prior to cryosectioning, the fixed samples were cryoprotected overnight in 30% (w/v) sucrose (Sigma-Aldrich, S7903) in PBS at 4°C. The next day, the samples were soaked in 1:1 sucrose (30%) and optimal cutting temperature (OCT) compound (Tissue-Tek) solution for 4 h. Samples were then transferred into a cryomold (Tissue-Tek, 25×20×5 mm), covered in OCT and frozen in liquid nitrogen. Using a histology cryotome (Thermo Fisher, CryoStar NX70), sections of 10–^^^ ^P^ ZHUH^ FXt. For immunohistochemistry, non-specific antibody binding was blocked with 1% bovine serum albumin (w/v) (BSA, Sigma, 9048-46-8) and 5% serum (v/v) from the host of the secondary antibody for 1 h (Goat serum; Abcam, ab7481). Primary antibodies were diluted in PBS containing 1% BSA. Immunostaining of osteocalcin was performed using anti-osteocalcin (1:200, Abcam, ab93876) overnight at 4°C. Samples were washed 3×5 min with PBS before incubation with the secondary antibody (1:500 Goat anti-rabbit IgG H&L Alexa 555, Abcam, ab150082) for 1 h. Cells were then permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, 9002-93-1) in 0.1% BSA in PBS followed by three washes with PBS. Actin was stained with Phalloidin CruzFluor 647 Conjugate (1:200, Santa Cruz Biotechnology, sc-363797) and cell nuclei with Hoechst 33342 (1:1000, Sigma-Aldrich, B226) for 2 h protected from light. Immunohistochemistry staining was validated by a secondary antibody control without adding the primary antibody. Samples were mounted with Mowiol. Sections were imaged using confocal microscopy with a 63× oil immersion objective. Collagen was investigated using Picrosirius red staining (Sigma-Aldrich, 365548). Briefly, the sections were stained in picrosirius red (0.1% in saturated aqueous picric acid) for 1 h and washed in two changes of acidified water. Polarized light microscopy images (polarization filter angles at 0° and 90°) were taken in transmission mode with a Zeiss AxioImager.Z2 running ZEN Blue at 20× magnification. Collagen content was then quantified according to the method described by Rich and Whittaker^[23]. First, images were converted to 8-bit RGB. The color threshold in the Hue spectrum was next split into the different colors as: red 2̽9 and 230̽256, orange 10̽38, yellow 39̽51 and green 52̽128. The pixel area of each color was measured. The different hue ranges measured were expressed as a percentage of all pixels in the image. Fiber thickness and maturity of collagen fiber increases from green, yellow, orange to red. Statistics: Statistical analysis was performed in GraphPad Prism 8.2.0. Depending on the number of variables, ordinary one-way or two-way analysis of variance (ANOVA) were used followed by Tukey’s test for multiple comparisons if more than two groups were compared. For the comparison of only two groups, Welch’s t test was used. Further, p values are indicated by * (<0.05), ** (<0.01), *** (<0.001) and ****(<0.0001) in figures. Data is shown as mean ± SD and the number of samples or locations (n) is indicated in the figure caption. Whiskers of boxplots indicate minimum and maximum values. Synthesis of 4-PEG-VS: Synthesis was performed as previously disclosed in WO2017042301A1 and US2018264176A1. In a 2-neck flask, 4 ml of triethanolamine (TEOA, Sigma-Aldrich, 90279-100ML) buffer (200 m, pH 8.0) was purged with argon for 15 min to remove oxygen. Then, 1 g (50 μmol) of 20 kDa 4-PEG-thiol (Laysan Bio, SH-20K-1g) was added and stirred under argon protection. After dissolution, the solution was carefully transferred into a syringe and immediately added dropwise over vigorous stirring to 1.05 ml (10 mmol) of divinyl sulfone (DVS, 97%, Alfa Aesar, L12827-09) in 4 ml of the same buffer. The reaction was left to proceed for 2 h. The product was dialyzed (MW cut-off: 3.5 kDa) against ultrapure water with 6 water changes over at least 4 h each, sterile filtered, and lyophilized after removal of excessive water using a rotary evaporator (Heidolph, 55 mbar, 200 rpm, 40°C). The product was aliquoted and stored at -20°C until use. Synthesis of Rhodamine-labeled 4-PEG-VS: In a flask, 2 ml of ultrapure water was protected with argon IRU^^^^PLQ^^7KHQ^^^^^^PJ^^^^^^PRO^^RI^^^^N'D^^-PEG-thiol was dissolved in it. To WKLV^ IODVN^^ D^ VROXWLRQ^ FRQWDLQLQJ^ ^^^^^PJ^ ^^^^^ ^PRO^^ RI^ WHWUDPHWK\O^ UKRGDPLQH-5-maleimide (Sigma-Aldrich, 94506) in 2 ml PBS pH 7.4 was added dropwise over stirring. The conjugation happened within seconds but was left to proceed for 10 minutes. The resulting mixture was DGGHG^GURSZLVH^LQWR^^^^^^^^O^^^^^^PPRO^^RI^GLYLQ\O^VXOIRQH^LQ^^^PO^7(2$^EXIIHU^^^^^^P0^^S+^ 8.0), left to react for 60 min under stirring, dialyzed, sterile filtered, aliquoted, and lyophilized. All the handling was performed in the dark. This protocol substitutes 1/80th of the 4-arm-PEG ends in the upper limit case of 100% conjugation efficiency. Hydrogel Preparation: To prepare (labeled) PEG hydrogels, (rhodamine-labeled) 4-PEG-VS (20% w/v in HEPES), 40 kDa or 500 kDa dextran (Sigma-Aldrich, 31389-25G or 31392-10G, 8% w/v in PBS), RGD (China Peptides, 04010054463, N-C: GRCGRGDSPG SEQ ID NO 02, 3.2% w/v in PBS pH 7.4), HA (Sigma, 9067-32-7, 0.5%–1% w/v in Hanks’ Balanced Salt Solution (HBSS, Gibco, 14025-050) or phenol red free Dulbecco’s modified Eagle’s medium (DMEM, Gibco, 21063-029)) and crosslinker (PEG-2-SH (2 kDa, Laysan Bio, 4.92% w/v in PBS pH 7.4) or MMP-degradable (China Peptides, N-C: KCGPQGIWGQCK (SEQ ID NO 01), 3.21% w/v in PBS pH 7.4)) were thoroughly mixed in this order by pipetting up and down 15– 20 times to ensure PIPS, efficient crosslinking and even distribution of RGD. Rheology: For rheology, acellular gel precursor solutions as described above with varying concentrations of 4-PEG-VS (2.0–2.5% (w/v)) and HA (0.25–0.83% (w/v) in HBSS) as well as degradable and non-degradable crosslinker (in PBS pH 6) were prepared and analyzed on an Anton Paar rheometer MCR302 (82868246) using a PP20 plate and a glass bottom. Gels were crosslinked at 37°C for 60 min while a time-sweep oscillatory measurement was performed at 1 Hz, 5% strain and with a gap of 100 μm. For each sample, 40 μl gel solution were loaded into the center of the glass plate and after setting the PP20 plate to the desired gap position, mineral oil (Sigma-Aldrich, 330779) was placed around the gel to prevent dehydration. The viscosities of precursor solutions (referred to ‘‘Complex Viscosity’’, mPa.s) were defined by the average viscosity value at the first 30 seconds before the onset of gelation. HA is the deterministic component of the solution’s viscosity. Frequency sweep was performed to test time-dependent viscoelastic properties of the hydrogels (constant strain: 5%, 0.1-100 rad/s). For parameters, see the caption of Fig.2. CFD Simulation: Confocal images of rhodamine-labeled PEG hydrogel on chip were processed, and the pore geometry was reconstructed using Seg3D (University of Utah, UT, USA). To quantify the FSS within the scaffold that has highly irregular porous geometries, a multiscale and multiphase CFD model previously developed (Luo et al., Acta Biochim Biophys Sin 2011, 43, 210) was used. The model involves 2 scales, i.e. (i) the global scale that represents the whole microfluidic chip and (ii) the local scale that models the detailed micro- structures of subsections (dimension: 20×20×30 μm, n=4) from the whole scaffold (Fig.11a). In the global model, the scaffold region was modelled as porous media with a permeability of 8.67×10^-15 m², which was obtained from experimental measurement of a hydrogel with a composition matching the confocal microscopy data. To mimic the experimental condition for dynamic cell culture, two types of flow rates (i.e., 10 μl min^-1 and 100 μl min^-1 per port) were applied to the global model as inlet and outlet boundary conditions. Mass flux conservation was applied to the interface between porous media and free fluid. The global model was meshed with 450410 tetrahedral elements. The pressure gradient that was calculated from the global model was applied to the local CFD model for simulating the shear stress on PEG scaffold surfaces. In the local model, the fluid domain of each subsection was meshed by a XQLIRUP^ WHWUDKHGUDO^HOHPHQW^VL]H^RI^^^^^^P^^ZKLFK^JHQHUDWHG^^^^^^^^^^^^^^^^^^^^^^^^^^^^ and 1302269 elements, respectively, for subsections 1–4. In this study, the fluid was modelled as laminar flow with the dynamic viscosity of DMEM (7.8×10^-4 Pa s). The CFD models were solved by a finite volume method (FVM) using ANSYS CFX (ANSYS Inc., PA, USA) under the convergence criteria of root-mean-VTXDUH^UHVLGXDO^RI^WKH^PDVV^DQG^PRPHQWXPௗ^^^^-4. hMSC Culture: For 2D cell expansion, hMSCs were cultured in expansion medium containing DMEM with 10% fetal bovine serum (FBS, Gibco, 10270-106), 1% Antibiotic-Antimycotic (Anti- Anti, Gibco, 15240-062), 1% non-essential amino acids (NEAA, Gibco, 11140-035) and 1 ng ^ml-1 basic fibroblast growth factor (bFGF, Invitrogen, 13256-029) in T150 cell culture flasks (TPP, 90151) at 37°C with 5% CO₂ until reaching 80% confluency. Medium was exchanged 3 times per week. Cells were washed twice with PBS (37°C) before adding 0.25% trypsin-EDTA (Gibco, 25200-056) to detach them from the flask. Trypsin activity was blocked by adding control medium (DMEM + 10% FBS + 1% Anti-Anti). Cells were counted using a hemocytometer and resuspended in control medium at the desired concentration. hMSCs (p5- p8) were embedded inside macroporous hydrogels as described before and subsequently cultured in osteogenic differentiation medium ((phenol red free) control medium with 10 mM) ȕ-JO\FHURSKRVSKDWH^ ^ȕ-GP, Acros, 410991000), 50 μg ml^-1 L-ascorbic acid (Sigma-Aldrich, A92902-100G) and 100 nM dexamethasone (Sigma-Aldrich, D2915)). Medium was replaced 5 times a week in custom molds and static culture on chip (according to the manufacturer’s protocol). For dynamic cell culture, FSS was applied on-chip by connecting two syringe pumps to both inlets of one medium channel and applying a total flow rate of 20 μl min^-1 or 200 μl min- ¹ using phenol red free DMEM. Loading was performed 2×10 min daily similar starting on day 3 until day 7 and then again from day 10 until day 13 with 60 min in between each treatment. Three replicates were used per condition (static, low FSS, high FSS). Human Osteoblast Culture: Primary hOBs were obtained from a commercial supplier (PromoCell C-12720) from healthy donors. For 2D cell expansion, hOBs (passage 6) were cultured similarly as hMSCs until reaching 80% confluency. After embedding in PEG hydrogels, hOBs were cultured in the same osteogenic differentiation medium as hMSCs for 2 days before fixation. Ki67 Staining: To stain cells embedded in PEG hydrogels for the cell proliferation marker Ki67, fixed samples from day 0 and day 2 of osteogenic culture were used. For immunohistochemistry, cells were first permeabilized for 10 min with 0.2% Triton X-100 (Sigma-Aldrich, 9002-93-1), then non-specific antibody binding was blocked with 1% BSA (Sigma, 9048-46-8) and 5% goat serum (Abcam, ab7481) for 1 h. The primary antibody (Invitrogen, MA5-14520) was diluted in PBS containing 1% BSA (1:250). Immunostaining was performed overnight at 4°C. Samples were washed 3×5 min with PBS before incubation with the secondary antibody (1:500 Goat anti-rabbit IgG H&L Alexa 555, Abcam, ab150082) for 3 h. Cell nuclei were counterstained with Hoechst 33342 (Sigma-Aldrich, B226) followed by three washes with PBS. Ki67-positive cells were manually counted from confocal microscopy images and normalized by the number of nuclei in each image. Dendrite Analysis: To analyze the dendrite length of cells embedded within macroporous hydrogels, maximum intensity projections (100 μm) of confocal microscopy images of actin- stained samples were used. Images were processed using the ImageJ plugin NeuriteQuant to segment cells into cell bodies and dendrites and subsequently quantify the mean length of all dendrites of a single cell. Live/Dead Assay: In order to quantify cell viability, staining with Calcein Green AM (CaAM, Sigma-Aldrich, 56436-50UG) and Ethidium-homodimer-1 (EthD-1, Sigma-Aldrich, 460439) was performed. Staining solution (1:1000 EthD-1 and 1:500 CaAM in PBS) was applied after washing samples twice with PBS and then incubated for 15 min at 37°C protected from light before washing again with PBS. Samples were imaged using confocal microscopy with a 10× air objective. For analysis, maximum intensity projections (MIP) of z-stacks of 70–^^^^^P^HDFK^ were created in Fiji/ImageJ. Cells in green and red channel were either counted manually if discrimination between single cells was not possible or a custom-written macro was used Viability was then calculated as the percentage of live cells among all present cells in the MIP. Fixation and Actin-Nuclei Staining: At the end of the culture, cells were fixed by first washing them with PBS and then applying a solution of 4% paraformaldehyde (PFA, Sigma-Aldrich, 15- 812-7) for 15 min at room temperature. Samples were washed twice with PBS. Actin-nuclei staining was performed to further investigate cellular and subcellular morphology and cellular network formation. Gels were incubated in 1% BSA in PBS for 1.5 h at room temperature. Subsequently, cells were permeabilized in a solution of 0.2% Triton X-100 in 0.1% BSA in PBS for 10 min. Gels were washed 3 times with PBS. The staining solution containing dilutions of 1:1000 Hoechst 33342 and 1:200 Phalloidin CruzFluor 647 Conjugate or Phalloidin-TRITC (Sigma-Aldrich, P1951) in 0.1% BSA was prepared. On-chip samples were stained for 12–24 h at 4°C, samples in confocal dishes for 1.5 h at room temperature protected from light. Before image acquisition, on-chip samples were washed five times with 5 min between each wash and gels on confocal dishes were washed three times. Imaging was performed using confocal microscopy. For overview images, a 10× air objective was chosen, for investigation of cellular network formation, a 25× water objective was used and for imaging of subcellular morphology, 40× water objective was utilized. To quantitatively compare mean cell area between static and G\QDPLF^JURXSV^^^î^^WLOH^VFDQV^RI^^^^^^P^]-stacks were acquired with a 25× objective. Using the MIP of these images, mean cell area was calculated by determining the area of actin in Fiji/ImageJ and dividing it by the number of nuclei in the same image. Alizarin Red Staining: In order to assess matrix mineralization, calcium deposits were stained with Alizarin red. A staining solution (2 mg ml^-1 Alizarin red S (Sigma-Aldrich, A5533-25G) in distilled water, pH adjusted to 4.12) was applied to the samples on-chip after washing the gels twice with distilled water. The samples were stained for 30 min at room temperature and then washed 5 times until the water came out clear. Imaging at 5× magnification was performed on Leica DMi1 microscope. A color threshold was applied by selecting only the red channel in RGB color space in Fiji/ImageJ. The area of red color was then measured for each condition. Sequences SEQ ID NO 01 KCGPQGIWGQCK SEQ ID NO 02 GRCGRGDSPG SEQ ID NO 03 GPQGIWGQ SEQ ID NO 04 GCRD-GPQGIWGQ-DRCG SEQ ID NO 05 CGRGDSP Cited prior art documents: US2018264176A1 A. Akiva et al., Advanced Functional Materials 2021, 31, 2010524 H. Bahmaee et al., Frontiers in bioengineering and biotechnology 2020, 8, 1042 N. Broguiere et al., Biomaterials 2019, 200, 56 M. P. Lutolf et al., Nature biotechnology 2003, 21, 513. O. Moreno-Arotzena et al. Materials 2015, 8, 1636 G. Nasello et al., Frontiers in bioengineering and biotechnology 2020, 8, 336 J. Patterson and J.A. Hubbel, Biomaterials 31 (2010) 7836e7845 Q. Sun et al., Bone 2017, 105, 245 Y. Shin et al. (Nature protocols 2012, 7, 1247) J Vandaele et al., Soft Matter 2020, 16, 4210 L. Xu et al., Integrative Biology 2020, 12, 303

Claims

Claims 1. A method for generating a model of cell development, said method comprising a. providing a microfluidic chamber; b. providing a plurality of mammalian primary cells, particularly stem cells or osteoblasts, in a macroporous hydrogel, wherein the macroporous hydrogel comprises polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase, particularly wherein the extracellular endopeptidase is a matrix metalloproteinase; c. applying a flow of cell culture medium to said plurality of cells, thereby subjecting said cells to fluid shear stress. 2. The method according to claim 1, wherein said plurality of cells in a macroporous hydrogel is generated in a gelation step, by polymerization-induced phase separation process from a precursor solution comprising the cells. 3. The method according to claim 2, wherein said precursor solution comprises a. a first polymer susceptible to in situ crosslinking, and optionally, a crosslinking agent; and b. a second polymer not susceptible to crosslinking, (particularly not susceptible to crosslinking) wherein the first polymer and the second polymer are miscible when the first polymer is not crosslinked, and wherein the mixture comprised of the first polymer and the second polymer undergoes phase separation, separating said first and said second polymer into separate phases, if the first polymer is crosslinked. 4. The method according to claim 3, wherein the first polymer is susceptible to photo- crosslinking, and the precursor comprises a photoinitiator. 5. The method according to claim 3, wherein the first polymer is susceptible to thiol- Michael addition. 6. The method according to any one of claim 3 to 5, wherein the first polymer is a vinylsulfone-modified poly(ethyleneglycol) (PEG) or a norbornene-functionalized polyvinyl alcohol, and a thiol crosslinking agent is present in the composition. 7. The method according to claim 3 to 5, wherein the second polymer is selected from the group comprised of dextran sulfate, chondroitin sulfate, sulfated alginate, dextran, mannuronan, hyaluronan, alginate, particularly wherein the second polymer is selected from the group comprised of dextran, dextran sulfate, hyaluronan and chondroitin sulfate. 8. The method according to claim 2, wherein said precursor solution comprises a. a vinylsulfone-modified poly(ethyleneglycol) (PEG) and a crosslinker selected from the group comprising i. a crosslinker comprising an endopeptidase recognition oligopeptide and two thiol moieties, ii. a dithiol-PEG or tetrathiol-PEG; particularly wherein the vinylsulfone-modified PEG is a four-arm-PEG- vinylsulfone; b. as second polymer, dextran and/or hyaluronan (HA). 9. The method according to claim 8, wherein in the precursor solution, a. the vinylsulfone-modified poly(ethyleneglycol) (PEG) is present at 1.5% to 3.0% (w/v), particularly at 1.8% to 2.5%; b. HA is present at 0.15% to 1.0%, and dextran is present at 0.2-2.5%, preferably in 0.5-1%, particularly wherein HA is present at 0.25-0.5% and dextran is present at 0.5-1%. 10. The method according to any one of the preceding claims, wherein said flow of cell culture medium is adjusted to generate a fluid shear stress of 0.5-3.0 Pa, particularly of 1.2 to 2.50 Pa. 11. The method according to any one of the preceding claims, wherein the mammalian cells are primary cells, particularly wherein the mammalian cells are stem cells, more particularly wherein the mammalian stem cells are mesenchymal stem cells, yet even more particularly wherein the cells are human primary osteoblasts and /or osteocytes. 12. A method for imaging or assaying cell development, cell differentiation and / or collagen secretion, comprising generating a model of cell development by a method according to any one of the preceding claims, and visualizing collagen secreted by cells. 13. A microfluidic device, comprising a microfluidic chamber, said microfluidic chamber comprising a macroporous hydrogel comprising mammalian primary cells, the macroporous hydrogel comprising polymer chains crosslinked by linker molecules amenable to cleavage by an extracellular endopeptidase, wherein the microfluidic chamber comprises an inlet port and an outlet port, the inlet port being connectable to a cell culture medium influx, and the outlet port allow cell culture medium outflow to leave the chamber. 14. The microfluidic device according to claim 12, wherein the microfluidic chamber comprising the macroporous hydrogel is situated between a first stream of cell culture medium and a second stream of cell culture medium, the first stream being applied at a higher flow rate than the second stream, thereby generating a pressure gradient between the first and second stream. 15. The method according to any one of claims 1 to 11, or the microfluidic device according to any one of claims 12 to 13, wherein the linker molecules amenable to cleavage by an extracellular endopeptidase comprise a matrix metalloproteinase -sensitive peptide, particularly wherein the matrix metalloproteinase-sensitive peptide comprises SEQ ID NO 03 (GPQGIWGQ); more particularly wherein the matrix metalloproteinase-sensitive peptide is a di-cysteine peptide that is or comprises SEQ ID NO 01 (KCGPQGIWGQCK) or SEQ ID NO 04 (GCRD-GPQGIWGQ-DRCG). 16. The method according to any one of claims 1 to 11, or the microfluidic device according to any one of claims 12 to 14, wherein the polymer chains comprise a peptide molecule capable of promoting adherence of cells, particularly wherein the peptide molecule capable of promoting adherence of cells is a peptide comprising a fibronectin-derived arginyl-glycyl-aspartic acid motif, more particularly wherein the peptide is or comprises SEQ ID NO 05 (CGRGDSP) or SEQ ID NO 02 (GRCGRGDSPG).