Strategy to Replace Animal-Derived ECM by a Modular and Highly Defined Matrix

prob-lematic procedures currently used to produce cell culture medium supplements, extracellular matrix, and analytical tools such as animal-derived antibodies. Abstract Many extracellular matrices (ECM) used for modern cell culture are derived from animals. An alternative approach is the recombinant production of individual matrix protein components. A further development of this strategy uses a constant core protein polymer that is modifiable with functional domains of various ECM proteins. This way, a single, highly-defined ECM system could be used for a large variety of cell types. Self-assembling protein domains from human muscle sarcomeres, termed ZT material (ZT), have been shown to be suitable for this modular approach of generating ECMs. We explored in a proof-of-concept study whether ZT modified with the fibronectin 10 domain (ZT Fn10 ) is able to substitute bovine serum-derived fibronectin as a coating for neural crest cell (NCC)-based toxicity testing. Human NCCs were generated from pluripotent stem cells and used in the automated version of an NCC migration assay (cMINC). ZT Fn10 , but not the unmodified core material (ZT), allowed high migration activity. The classical cMINC setup, with bovine fibronectin coating, was used as positive control, and detailed analysis of NCC migration by time-lapse recording indicated that the novel ECM fully matched the bioactivity of the traditional ECM. A final set of

αvβ3, and α8β1 integrins (Perris et al., 1989;Testaz et al., 1999). Therefore, cell culture dishes for cMINC assays need to be coated with Fn or related ECM proteins. Interestingly, a variety of engineered biomaterials of defined composition have now become available that have proven to be viable substrate alternatives for in vitro cell culture (Cai and Heilshorn, 2014;Loo et al., 2015;Hellmund and Koksch, 2019). Biomaterials of homogenous composition and high purity optimally meet requirements for experimental standardization. In addition, biomaterials made of recombinant proteins offer a significant cost reduction for high-throughput studies, thanks to the economy of production and scalable yields of bacterial overexpression. Such biomaterials could permit developing a strategy for the replacement of animal-derived Fn in high-throughput assays like the cMINC assay.
In this respect, we developed a functionalized biopolymer, termed ZT, formed by the self-assembly of two human proteins: telethonin (Tel) and the two N-terminal immunoglobulin domains from titin (Z1Z2) (Bruning et al., 2010;Nesterenko et al., 2019). The complexation of Tel and Z1Z2 occurs naturally in the Z-disc of human muscle sarcomeres. In this complex, Tel acts as a molecular glue, becoming "sandwiched" between two antiparallel Z1Z2 and forming a robust intermolecular β-sheet that spans the three components Marino et al., 2006). The ZT polymer uses a Z1Z2-Z1Z2 fusion tandem (Z 1212 ) that leads to the spontaneous, propagative self-assembly of Z1Z2:Tel complexes into a polymer (Bruning et al., 2010). The ZT polymer can be readily functionalized by genetically encoding functional moieties into its protein building blocks prior to assembly. To date, various functional versions of the ZT polymer have been produced that display bioactive peptidic motifs (Bruning et al., 2010;Hill et al., 2019) as well as functional full-length proteins introduced by gene fusion Nesterenko et al., 2019). A ZT variant comprising the integrin-binding Fn 10 domain from human Fn (ZT Fn10 ) has been shown to successfully sustain the culturing of human embryonic stem cells (hESCs), murine mesenchymal stromal cells (mMSCs) , and human induced pluripotent stem cells (hiPSCs) (Nesterenko et al., 2019) over extended periods of time. Its robustness, the ease of its handling as a biocoating, and the economy of its production, make the fibronectin-mimetic ZT Fn10 a promising system to replace native Fn in high-throughput cell applications. Previous applications of the ZT material had focused on its cell adhesion and culturing properties, while its ability to support naturalistic cell migration in functional assays has not been tested so far.
In this study, we investigate the suitability of the ZT system to support NCC culturing and migration in the cMINC assay. In particular, we study whether cell migration on this recombinant substrate displays traits comparable to those on the native ECM-component Fn and whether the migration provides sufficient sensitivity to support its application to the screening of toxicants in the cMINC assay. media by providing information on commercially available serum-free media for several cell types.
Another group of animal-derived cell culture reagents is formed by extracellular matrix (ECM) components, such as the widely used Matrigel™ and plasma fibronectin. Matrigel™ is a heterogeneous glycoprotein mixture extracted from tumors grown in mice. Its complex composition suffers from high intrinsic variability, limiting experimental control and leading to data inconsistencies (Hughes et al., 2010). Plasma fibronectin (Fn) is a 250 kDa glycoprotein that assembles into dimers and consists of repetitive domains: 12 type I units, two type II units, and 15-17 type III units (Pankov and Yamada, 2002). The 10 th Fn type III repeat domain (Fn 10 ) is responsible for the integrin receptor binding of fibronectin and, thereby, its cell attachment (Chi-Rosso et al., 1997). Specifically, cell attachment to Fn 10 is mediated by the short peptide motif Arg-Gly-Asp (RGD). Due to its low recombinant yield and the complexity of its composition and assembly, most of the commercially available Fn for in vitro use is extracted from animal plasma (Oredsson et al., 2019). Consequently, Fn preparations are likely to contain other undefined plasma proteins as well.
Migration of neural crest cells (NCCs) is an important process in fetal development. During neurulation, NCCs delaminate from the dorsal part of the neural tube and migrate to target sites in the periphery where they differentiate into different cell types (LaBonne and Bronner-Fraser, 1999). Disturbed NCC migration can lead to severe malformations and disorders such as Hirschsprung's disease (Farlie et al., 2004). The migration of NCC is well studied in animals (Fuller et al., 2002;Bergeron et al., 2013;Usami et al., 2016), but few human in vitro data are available.
The migration inhibition of neural crest cell (MINC) assay was developed to evaluate developmental toxicants and signaling pathways (Zimmer et al., 2012). A high-throughput version of this in vitro assay, called circular migration inhibition of neural crest cell (cMINC) assay (Nyffeler et al., 2017b), is based on circular stoppers, which create a cell-free area around which cells are seeded; migration is initiated by removing the stoppers. The assay uses human NCCs generated from pluripotent stem cells to assess disturbances of their migration during fetal development (Nyffeler et al., 2017a). This migration requires extension of protrusions at the leading edge of the cell and disconnection from the ECM at the trailing edge (Conway and Jacquemet, 2019;Ridley et al., 2003). Thus, cell migration requires a constant switch between strong and weak attachment of the cell to the ECM as new adhesions are formed at the leading edge and old cell-ECM adhesions are disassembled (Shafqat-Abbasi et al., 2016).

Protein production
Expression clones of telethonin (Tel), Z 1212 , and Z 1212 Fn10 have been previously reported Bruning et al., 2010). In brief, all coding sequences were inserted into the pETM-11 vector (EMBL), which adds a His 6 -tag and a tobacco etch virus (TEV) protease cleavage site N-terminally to the inserted gene.
Proteins were expressed in BL21 E. coli cells (Invitrogen) grown at 37°C in Luria Bertani (LB) medium supplemented with 25 μg/mL kanamycin. At a culture OD 600 = 0.6, protein expression was induced with 1 mM IPTG, the incubation temperature was reduced to 18°C, and expression was carried out for 16-18 h overnight. Cells were harvested by centrifugation. Cell pellets with overexpressed Z 1212 , Z 1212 Fn10 and Tel proteins were resuspended in 50 mM Tris-HCl, 500 mM NaCl, pH 7.4 supplemented with an ethylenediaminetetraacetic acid (EDTA)-free protease inhibitor cocktail (Roche). Cells were lysed by sonication. Lysates were clarified by centrifugal ultrafiltration and filtered (0.2 μm pore filter; Sarstedt) prior to sample purification. Z 1212 and Z 1212 Fn10 were isolated from the soluble lysate fraction via Ni 2+ -IMAC (5 mL His-Trap HiPrep column; GE healthcare) in the buffer described above. The His 6 -tag was cleaved by incubation with TEV protease in dialysis overnight in 50 mM Tris-HCl pH 7.4, 500 mM NaCl. The sample was then further purified by subtractive Ni 2+ -IMAC, followed by dialysis (in 50 mM Tris-HCl, 100 mM NaCl pH 7.4) and gel filtration on a HiLoad 26/600 Superdex 200 column (GE Healthcare) equilibrated in dialysis buffer.
Tel was purified from the insoluble lysis fraction as described previously . Briefly, the insoluble fraction was resuspended in 25 mM Tris-HCl, 6 M urea, pH 7.4, lysed and centrifuged. The pellet was discarded and the supernatant was filtered and subsequently subjected to purification using Ni 2+ -IMAC. The Tel isolate was concentrated to 10 mg/mL in the buffer described above and stored at 6°C.
Production of the ZT Fn10 polymer by assembly of Z 1212 Fn10 and Tel Z 1212 , Z 1212 Fn10 and Tel samples were filtered using 0.22 μm Ultrafree-MC filters (Merck), and all subsequent handling was carried out in a clean bench under laminar flow. For polymerization of the proteins to form the ZT and ZT Fn10 substrates, a protocol adapted from Bruning et al. (2010) was used. Z 1212 or Z 1212 Fn10 samples were mixed with Tel at a 1:3 molar ratio in overnight dialysis in 50 mM Tris-HCl, 100 mM NaCl, pH 7.4 at room temperature and using a Slide-A-Lyzer™ MINI dialysis device 3.5K MWCO (Thermo Fisher Scientific) under sterile conditions. Samples were incubated for 24 h until self-assembly had led to a turbid sample pellet. The sample was then retrieved from the dialysis setting and stored at 6°C until further use. Self-assembly was confirmed by native polyacrylamide gel electrophoresis.
Coating of polystyrene plates with substrata Bovine plasma fibronectin (Sigma), ZT and ZT Fn10 assemblies were diluted in low-endotoxin Dulbecco's phosphate buffered saline (DPBS) without Ca 2+ and Mg 2+ (Merck Millipore) to 1 µg/ mL for bovine plasma fibronectin (Sigma) and to 1, 5, 10, 20 µg/ mL for ZT and ZT Fn10 . Coating of 96-well polystyrene plates (Corning) was performed by deposition, which ensued by adding 100 μL substrate solution to each well and incubating overnight at 37°C. After 24 h, solutions were removed by aspiration and wells dried by exposure to air for 15 min under a laminar flow.

Migration assay (cMINC)
The cMINC assay was performed as described earlier (Nyffeler et al., 2017b). Briefly, silicone stoppers (Platypus Technologies, Madison, WI, US) were placed centrally into each experimental well of a 96-well polystyrene plate (Corning) coat-cell. ii) Accumulated distance is the sum of incremental movements made by the cell during migration from the start to the end point. iii) Directness is the ratio between cell translocation and accumulated distance, where the closer the value is to 1, the more directional the migration is. iv) Cell speed is the ratio between the accumulated distance and the time length of the images acquired during cell migration.

Data handling and statistics
If not stated otherwise, values are expressed as mean ±SEM. If not indicated otherwise, experiments were performed at least three times (i.e., using three different cell preparations), with at least three technical replicates per condition.

Migration of neural crest cells on the ZT Fn10 polymer
In order to investigate whether the ZT Fn10 biomaterial may substitute the established bovine serum Fn coating in the cMINC assay setup, we compared the migration of NCCs on the novel ZT matrix to that on the classical coating. Two variants of the ZT polymer were compared in these experiments: the integrininteracting ZT Fn10 form and the non-functionalized ZT scaffold, known to be bioinert . The latter served as control to reveal any possible non-specific interactions between NCCs and the ZT core structure.
First, we evaluated the viability and retention of NCCs on the ZT substrates. For this, the polymers were passively adsorbed onto polystyrene 96-well cell culture plates, and NCCs were plated on the coated surfaces. The number of viable NCCs remaining on the plates was assessed 72 h after seeding. As positive control, the cell count on the classical Fn-coating (1 µg/mL) was evaluated. Non-coated dishes were also assessed. The results revealed that NCCs adhered about equally well to plates under all test conditions, including non-coated plates (Fig. 1A). This revealed that cell attachment was not dependent on a specific ECM component. This situation allowed us to start migration with similar numbers of cells on the different matrices.
In a second step, NCC migration capacity was monitored using the cMINC assay (Nyffeler et al., 2017b). Here, a clear dependence of migration on the type of coating was observed (Fig.  1b,c) (Fig. S1 3 , movies MS1-MS4 4 ). Cells cultured on the classical bovine Fn migrated towards the center of the well, while cells cultured on non-coated polystyrene migrated only to a small extent. The number of cells migrating on the ZT control material was comparable to the number of cells migrating on the non-coated polystyrene, confirming its bioinert character. In contrast, cells seeded on ZT Fn10 migrated quickly and with good directionality. ed with plasma fibronectin, ZT or ZT Fn10 . Cells were seeded into the stopper-containing wells at a density of 95,000 cells/ cm 2 . The following day, stoppers were removed to allow cells to migrate into the cell-free central area, and medium was refreshed. To test the effect of toxicants on NCC motility, 5x concentrated toxicant (cytochalasin D (CytoD), lithium chloride (LiCl), and polychlorinated biphenyl 180 (PCB180); (all from Sigma-Aldrich)) solution was added to the medium 24 h after stopper removal. After another 24 h, cell viability and migration endpoints were monitored. For this, cells were stained with Hoechst-33342 and calcein-AM (both from Sigma-Aldrich), and image acquisition was performed using a Cellomics ArrayScan VTI imaging microscope (Thermo Fisher). Hoechst-33342 and calcein-AM double-positive cells were determined by an automated algorithm described earlier (Stiegler et al., 2011;Krug et al., 2013) and defined as viable cells. For migration quantification, a free software tool 2 was used as described in Nyffeler et al. (2017b) to calculate the original stopper position and determine the number of viable cells within the migration area. Viability and migration were normalized to untreated or solvent control (0.1% DMSO). Concentration-response curves were generated with GraphPad Prism (GraphPad Software, La Jolla, CA, USA); each data point represents mean ±SEM of three biological replicates. For curve fitting, a four-parameter log-logistic curve fit with an upper asymptote forced to 100% was used.

Time-lapse experiment and cell tracking
The cMINC assay was performed using an Axio Observer Z1 microscope (Zeiss, Germany) equipped with an Axiocam MRM camera to track cell movement into the central "migration zone" for 24 h after stopper removal. Phase contrast images were acquired every 10 min with a 5x objective. To provide constant temperature and CO 2 supply during the recording time, a microscope equipped with an incubation system was used. Cell tracking was performed manually using the "Manual Tracking" plugin in ImageJ (Schneider et al., 2012) by selecting, in reverse time, viable cells that migrated into the central area and were visible throughout the full time-lapse recording. A total of 60 individual cells were tracked for each condition (namely, per media condition and concentration, three technical replicates were performed in two biological replicates, with 10 cells tracked per technical replicate). To determine cell distance and velocity, cell location coordinates on the viewed plate were obtained with ImageJ (Schneider et al., 2012) and transferred to the "Chemotaxis and Migration Tool V2.0" (ibidi, Germany). Cell translocation, accumulated distance, directness and cell speed parameters were calculated from the coordinates of the migrating cells. These parameters were defined as follows: i) Cell translocation refers to the straight line formed by connecting the start and end points of the migration path of the and availability are not limiting for ZT Fn10 , so that its implementation in high-throughput assays at this optimal concentration is realistic. At even higher concentrations we observed a trend to reduced migration, in good accordance with a bell-shaped migration behavior on ECM (Underwood et al., 1993).
In summary, ZT Fn10 allowed optimal migration that was fully comparable to Fn. The optimum concentration was 5 µg/mL, and at this concentration, the migration performance was even better than at standard assay conditions, i.e., 1 µg/mL Fn.

Examination of migration speed and directionality at the single cell level
To characterize the migration behavior of NCCs on ZT Fn10 in greater detail, the cMINC assay setup was used for single cell tracking. For this, cells were seeded onto coated plates as described above and migration was monitored by time-lapse video microscopy for 24 h at 10 min intervals. Different features of the recorded movement trajectories were quantified. These data fully confirmed the observations with the classical cMINC read-out: Cells seeded on optimal concentrations of ZT Fn10 migrated at higher speed, cov-During the development of the cMINC assay, the Fn-coating had been tested in detail. Maximal migration was observed at a concentration of approx. 5 µg/mL (Fig. S2A 3 ); however, 1 µg/mL was found to be the optimal balance between assay performance and costs (Nyffeler et al., 2017b), and this condition was then standardized by deposition of the assay protocol in public repositories (Krebs et al., 2019) and inclusion of the test in international screening programs (Bal-Price et al., 2018). Here, we retested a wide concentration range for comparison to the new substrate. Our data confirmed 1 µg/mL Fn as a suitable and robust standard condition (little variability, good assay window) (Fig. S2C 3 ). At low concentrations of Fn, migration was continuously improved by increasing the concentration of the coating material until a maximum was reached. At very high concentrations, migration was attenuated as cells tended to attach too strongly to move.
The current work shows that the optimal ZT Fn10 substrate concentration for NCC migration is also 5 µg/mL. This is in good agreement with data from a previous study that found that saturation of stem cell attachment was attained at a substrate concentration of around 5-10 µg/mL . Importantly, cost (B) Neural crest cell migration at 72 h after seeding on ZT Fn10 and on ZT. In both cases, data are shown as means from three cell preparations. ***, p < 0.001 as determined by two-way ANOVA (considering type of coating as factor) followed by Bonferroni's post-hoc test. (C) Exemplary fluorescence images of NCCs stained with Hoechst-33342 (nuclei, cyan) and calcein-AM (cytosol, magenta) at the end of migration experiments on diverse coatings (as indicated). The diameter of each of the circular areas displayed was 2 mm.

Comparative testing of test method positive controls on classical and novel coatings
As the ZT Fn10 substrate was shown to support NCC migration, we tested its suitability as a substitute of plasma Fn in the cMINC toxicity test. For this, we ran parallel tests on ZT Fn10 and Fn. As model toxicants, we used the well-established NCC migration inhibitors cytochalasin D (CytoD), lithium chloride (LiCl), and ered longer distances, and moved in a more directed fashion than those in the classical cMINC setup (Fig. 2). However, as ZT Fn10 is a fibronectin mimetic, the expectation was that when using both substrates at their optimal concentrations (5 µg/mL in both cases), the migration would be similar on both substrates. Effectively, this was confirmed experimentally (Fig. S2B 3 ), proving that NCC migration was equivalent on both substrates.

Fig. 2: Analysis of migration trajectories of neural crest cells
NCCs were plated around silicone stoppers and allowed to attach for 24 h. The stoppers were removed, and NCC migration was observed for 24 h using time-lapse imaging. The tracks were marked in the ImageJ program and evaluated manually. (A) Representative trajectories of neural crest cells upon migration into an initially empty circular area (delineated by green circles). (B) Speed (distance along the track per time), cell translocation (Euclidean distance from start to endpoint of track) and directness (cell translocation/track length, i.e., "low degree of zig-zagging"), each calculated from the trajectory of 20 cells from two cell preparations. ***, p < 0.001 as determined by ANOVA (considering type of coating as factor; performed separately for each endpoint) followed by Dunnett's post-hoc test.

Fig. 3: Migration inhibition of NCCs by model toxicants
NCCs were plated under otherwise similar conditions on dishes coated with fibronectin or ZT Fn10 . They were exposed to cytochalasin D (CytoD), lithium chloride (LiCl) and polychlorinated biphenyl 180 (PCB180) for 24 h. Initial values (100%) correspond to untreated cells (control cell medium containing 0.1% DMSO as "negative control"). Cell migration data are all given relative to negative control values. Each data point shown is the mean ±SEM of three biological replicates (= different cell preparations). Dashed lines at 75% visualize the threshold value for impaired cell migration.
polychlorinated biphenyl 180 (PCB180) (Fig. 3). The selected toxicants have largely different chemical structures and modes of action. They also act over a wide range of concentrations (100 nM-10 mM). Most importantly, they are all known to affect migration at lower concentrations than cell viability. This was fully confirmed here (see Fig. S3 3 for viability data). The results showed that concentration-response curves for impaired migration were similar for the novel ZT Fn10 coating material and for native Fn. We conclude that the ZT Fn10 substrate can be used instead of Fn for further toxicological applications of the cMINC assay.

Conclusions and outlook
This study was undertaken to examine the feasibility of exchanging animal-derived cell culture supplements for defined bioengineered recombinant material. The work exemplifies the type of bridging studies required to change established assay protocols. In some cases, assay conditions may need adaptations to new materials and assay performance might require re-examination. Here, the requirement for adaptations was minimal, with the new material being readily applicable to the toxicology application under consideration. Notably, the ECM-native Fn and the ZT Fn10 material proved to be applicable in similar concentration ranges. It is noteworthy that the concentration affected only a basic test system parameter (number of migrating cells under control conditions) but not the assay performance under toxicant challenge. These results suggest that ECM-substitute materials hold high promise in high-throughput toxicology applications and now encourage the exploration of these materials in other tests, e.g., chemotaxis assays for T-cells and neutrophils (Lin and Butcher, 2006;Cano et al., 2016). In this regard, it is worth emphasizing that the ZT Fn10 material has demonstrated excellent cell compatibility and that its sole protein composition, self-assembly and molecular domain granularity is reminiscent of that of typical biological matrices. Moreover, its modular construction readily allows the incorporation of alternative cell attachment and functional domains by genetic modification, as exemplified by other available ZT variants that carry fluorescent proteins and functional peptides (Nesterenko et al., 2019). In summary, our findings suggest that the replacement of animal-sourced cell substrates in in vitro toxicology screenings now appears feasible thanks to recent biomaterial advances.

Material availability
The ZT Fn10 substrate can be obtained from the authors (olga. mayans@uni-konstanz.de).