The microstructure of SBS gelatin mats
The mats produced by SBS had a flocculent, cotton tuft-like appearance (Fig. 1A), suggesting a 3D fiber topology in contrast to typical 2D electrospun fiber mats22,24. SEM analysis of dry mats demonstrated the placement of collagen fibers in wavy bundles that were orientated in all three spatial dimensions confirming the presence of an ECM-like 3D structure (Fig. 1B,C). As thermally crosslinked glucose-gelatin is hygroscopic, the material was anticipated to swell when hydrated and we asked whether the properties of the crosslinked gelatin mats in wet environment (such as cell culture) would differ from those in the dry state. Indeed, the SEM analysis of a water-soaked mats demonstrated that in hydrated state the material was considerably denser and the voids between the fibers were shrunk. These changes, however, did not alter the overall structure of crosslinked gelatin SBS mats and the fibrous 3D nature was well preserved (Fig. 1D,E). Comparing the SBS-produced fiber mats to a natural decellularized mouse liver ECM using SEM (Supplementary Fig. S1A-B) revealed that the microarchitecture and 3D fiber arrangement of the SBS mats showed considerable degree of similarity. However, the fibers of our SBS gelatin mats were significantly thicker than those in the natural liver ECM. When the fibers in SBS mats had a diameter of 1.52 ± 0.54 µm, the decellularized liver microstructure was denser and the matrix fibers were considerably thinner with a diameter of 19.6 ± 7 nm (Supplementary Fig. S1C).
The microstructure of gelatin-based SBS mats. (A) Photograph of thermally treated gelatin-based mat. A disc with 7 mm diameter that was used for 3D cell culturing is shown in the top part of the panel along with the corresponding punch. (B,C) SEM micrographs of dry gelatin-based SBS mat obtained after thermal cross-linking. (D,E) SEM micrographs of SBS mats after soaking in water and subsequent drying using supercritical CO2. Scale bars are 200 µm on left panels and 20 µm on right panels.
SBS mats are cytocompatible and allow 3D cell growth
To study the cytocompatibility of the mats we cultured immortalized hepatocyte cell line Hep G2 in cell culture medium containing fetal bovine serum on discs cut out from SBS mats that were fit into the wells of a 96-well tissue culture plate (Fig. 2). Cell morphology was evaluated by actin cytoskeleton staining. The cells showed normal morphology and were spread in between and around the matrix fibers. The cells formed a thick layer of pseudotissue in the 3D mat that consisted of tightly packed organoid-like structures that spanned 2–3 cell layers in depth (Fig. 2A,B). SEM of SBS mats cultured with Hep G2 cells showed fibrous matrix densely populated with cells (Fig. 2C,D). The viability of the cells on SBS mats was not reduced over time as no cytotoxicity was observed (Fig. 2E). On the contrary, the approximately two-times increase in the signal strength reflected the doubling of the cell numbers grown on SBS mats during the 96-h time window and demonstrated that the HepG2 cells were not just surviving but also proliferating on the SBS-mats (Fig. 2E).
Hep G2 cells cultured on SBS mats for 96 h. (A) 3D reconstruction showing the Hep G2 cells grown on a SBS mat. (B) Ortho-view of the Hep G2 cells grown on a SBS mat. Actin cytoskeleton of cells was visualized by phalloidin staining (yellow), cell nuclei were stained with DAPI (blue). (C,D) SEM micrographs at different magnifications of the surface of mats populated with Hep G2 cells. Scale bars are 100 µm (B) and 50 µm (C). (E) Cell viability assay of Hep G2 cells cultured on SBS mats for 96 h. RLU relative light units. Hep G2 cells were cultured in IMDM medium containing fetal bovine serum.
As natural ECM is a complex structure that contains several different structural and regulative proteins, we asked whether enrichment of the processed SBS gelatin mats with common active constituents of tissue ECM—laminin, vitronectin, or a mixture of basal membrane components (Matrigel, MG) would modulate cell growth. We cultured Hep G2 cells on either uncoated SBS mats or mats preincubated with MG, laminin-111 or vitronectin solution in tissue culture media (Supplementary Fig. S2). No significant changes in cell morphology or viability were detected suggesting that uncoated 3D gelatin mats provide adequate mechanical support and homing potential for the maintenance and growth of immortalized hepatocyte-like cells.
The mats are a suitable substrate for 3D growth of primary hepatocytes
To study whether the SBS mats would also support the growth of primary human liver cells we cultured human cryopreserved hepatocytes in serum-free hepatocyte growth medium on either SBS mats alone or mats enriched with MG, laminin-111 or vitronectin (Fig. 3A). As a comparison, the most commonly used primary hepatocyte culture method, collagen—MG bilayer culture, was used. The cells were cultured for 4 or 20 days and the SBS mats were studied using a confocal microscope. Similarly, to the immortalized Hep G2 cells, the primary human hepatocytes spread on and around the matrix fibers, spanning up to cell 3 layers in depth (Fig. 3B, Supplementary Fig. S3A). Cells grown in MG bilayer were flatter when compared to cells grown on SBS mats and grew as a monolayer (Fig. 3C, Supplementary Fig. S3B). To assess the cell shape differences in more detail we measured the cell length (x-axis) and cell depth (z-axis) from the z-series data from the confocal microscopy images. SBS mat-cultured hepatocytes appeared to have greater depth in the z-dimension and shorter length compared to hepatocytes in MG bilayers. Here the x : z ratio was 1.4 on average indicating a rounder and more sphere-like shape (Fig. 3D). Hepatocytes in MG bilayer cultures had the average x : z ratio 4.7 that is characteristic of a flatter, anisotropic cell shape (Fig. 3D). In long-term cultures (20 days), the hepatocytes that were grown on SBS mats retained their characteristic polygonal shape (Fig. 3A). Whereas the shape of the cells grown in MG bilayer for 20 days was significantly altered, the polygonal cell shape was replaced by an elongated irregular shape and the cells appeared to have reduced viability as exemplified by cell vacuolization (Supplementary Fig. S4). Addition of laminin-111 or vitronectin to the SBS mats did not significantly alter either the growth characteristics or the cells´ morphology during long term culture (Fig. 3A).
(A) Hepatocytes grown on SBS mats or in MG bilayer for 4 days (left panels) or 20 days (right panels). Scale bars 50 µm. 3D reconstruction of hepatocytes grown on mats (B) and MG bilayer (C) after 4 days of culture. Actin cytoskeleton of cells was visualized by phalloidin staining (yellow), cell nuclei were stained with DAPI (blue). The gelatin fibers appear as yellow or blue streaks due autofluorescence. (D) Quantification of cell shape using the cell length (x) to depth (z) ratio (x:z ratio). Summarized results from 3 independent hepatocyte culture replicates are presented, * indicates a statistically significant (P < 0.05) difference. Primary hepatocytes were cultured in serum-free hepatocyte defined growth medium.
SBS mats allow the maintenance of hepatocyte differentiation in long-term cultures
Hepatocyte differentiation status during long term culturing in serum-free hepatocyte growth medium was evaluated by studying the expression of hepatocyte-specific transcription factor HNF4α and albumin. After 24 days of culture the immunofluorescence analysis of hepatocytes showed strong expression of HNF4α (Fig. 4A, Supplementary Fig. S5A) and albumin (Fig. 4B) in cells grown on SBS mats. Cells grown in MG bilayer showed significantly lower levels of HNF4α and albumin proteins (Fig. 4C,D) indicating the reduction of hepatocyte differentiation status. Interestingly, there were no significant differences between the HNF4α mRNA levels in the hepatocytes grown on SBS mats or as MG bilayer (Supplementary Fig. S5C) suggesting the existence of a posttranslational step for regulation of the HNF4α protein levels. To further examine the differentiation status of the hepatocytes, we analysed the expression of hepatocyte dedifferentiation marker vimentin28. Vimentin protein and mRNA expression was moderately elevated in MG bilayer culture as demonstrated by the immunofluorescence analysis (Supplementary Fig. S4A-B) and RT-qPCR (Supplementary Fig. S5D) being in line with the reduction of the HNF4α protein.
The differentiation status of primary human hepatocytes grown in long-term cultures. Confocal microscopy images (A) of the expression of hepatocyte-specific transcription factor HNF4α and cleaved caspase 3 (CC-3) and quantification of the corrected cell fluorescence (C,E). Staining of albumin in hepatocytes (B) and quantification (D). Scale bars 20 µm. The dynamics of the synthesis of hepatocyte-specific metabolites albumin (F) and urea (G) was measured in the culture media of long-term hepatocyte cultures, n = 5. * indicates a statistically significant (P < 0.05) difference, ** P < 0.001 compared to MG bilayer. Primary hepatocytes were cultured in serum-free hepatocyte defined growth medium.
To evaluate the proportion of non-viable cells in the culture we studied the presence of cleaved caspase-3 (CC-3)—the marker of apoptosis29—in human primary hepatocyte long-term cultures (Fig. 4A). While the hepatocytes grown on SBS mats showed a low CC-3 signal, a large proportion of cells (in average 71%) grown in MG bilayer cultures was CC-3 positive (Fig. 4E).
To study the cell attachment and maintenance dynamics we evaluated the number of cells that remained unattached and were floating in the tissue culture media 3 h after plating and assumed that the rest of the cells were attached. Although the SBS mats had a lower cell attachment percentage (uncoated mat average 67.4%; vitronectin-coated mat 69.5%) the proportion of live matrix-attached cells remained essentially consistent from 4 to 24 days (Supplementary Fig. S6A). In comparison, the hepatocytes cultured in MG bilayer showed higher initial attachment to the 2D surface (in average 95%), however, this was followed by a fast and gradual drop in the survival rate as judged by the increase in the detached cell numbers (Supplementary Fig. S6A). To confirm the adequacy of this live cell number estimation we performed a luminescence-based cell viability assay at the end of the experiment at day 24 (Supplementary Fig. S6B-C). When we compared the calculated live cell numbers with the data obtained from the cell viability assay, we found that the numbers of cells grown on MG bilayers and 3D SBS-mats obtained by both methods did not differ significantly at the 24-day time point and concluded that our estimation of live cells in hepatocyte culture was reliable. The calculated viable cell numbers were in accordance with the measurements of two main hepatocyte metabolites—albumin and urea—from serum-free hepatocyte culture media that were consistent from 4 days onwards in cultures on SBS mats. However, the supernatants from MG bilayer cultures showed a consistent drop in the concentration of both metabolites (Fig. 4F,G). Media samples from the long term primary human hepatocyte cultures on SBS mats contained significantly more albumin (10–12 mg/dL) when compared to albumin content in the media collected from MG bilayer cultures (6–7 mg/dL) (Fig. 4F). When we normalised the albumin content to the estimated numbers of hepatocytes in culture, the differences in albumin production became even more pronounced showing approximately twofold higher albumin production per cell in hepatocytes cultured on SBS-mats compared to MG bilayer cultures at all time points (Supplementary Fig. S6D). There were no significant differences in urea concentrations at 24 days (Fig. 4G). Interestingly, we found that coating the SBS mats with vitronectin may somewhat improve the urea production per cell when compared to uncoated SBS mats (Supplementary Fig. S6E). Addition of laminin-111, vitronectin or MG-coating to the SBS mats did not significantly alter the production of albumin or urea in hepatocyte cultures on SBS mats (Supplementary Fig. S7).
SBS mats support the maintenance of hepatocyte polarity in culture
We examined the presence of two polarization markers, the junctional protein ZO1 and the transcytotic marker CD13 in human primary hepatocytes grown on uncoated or MG, laminin-111 or vitronectin–coated SBS mats; as a comparison we used MG bilayer culture. Immunofluorescence analysis showed that ZO1 is expressed at cell membranes at cell–cell contacts (Fig. 5A, Supplementary Fig. S8). ZO1 expression appeared consistent in short-term and long-term hepatocyte cultures on 3D mats irrespective of their coating (Fig. 5A, Supplementary Fig. S8, S9A). However, the hepatocytes grown as MG bilayer had significantly reduced ZO1 expression by day 20 in culture (Supplementary Fig. S9A). CD13 showed strong membrane localisation pattern in all hepatocytes grown on mats (Fig. 5B). Hepatocytes grown on MG bilayer showed diffuse and predominantly cytoplasmic expression of CD13 at 4 days of culture and the expression was significantly reduced by 20 days of culture (Fig. 5B, Supplementary Fig. S9B).
Expression of cell polarization markers in human primary hepatocytes grown on SBS mats or in MG bilayer. (A) Expression of tight junction component ZO1 (green) on hepatocytes cultured for 4 days (left panels) or 20 days (right panels). (B) Expression of transcytotic marker CD13 (orange) on hepatocytes cultured for 4 days (left panels) or 20 days (right panels). Cell nuclei were stained with DAPI (blue). Scale bars 20 µm. Primary hepatocytes were cultured in serum-free hepatocyte defined growth medium.
SBS mats facilitate the induction of hepatocyte-specific CYP gene expression
Next, we tested whether the human primary hepatocytes grown on SBS mats were able to upregulate the expression of CYP1A2, CYP2B6 and CYP3A4 mRNA in response to known inducers of CYP transcription (Fig. 6). RT-qPCR showed more than a 40-fold induction of CYP1A2 mRNA in response to omeprazole treatment in cells grown on vitronectin enriched SBS mats (in average 43-fold) and on MG bilayers (in average 45-fold); a 20-fold induction of CYP1A2 in average was detected in cells grown on un-enriched SBS mats. In laminin-111 and MG-enriched SBS mats the level of CYP1A2 induction was considerably lower, only 2–3- fold in average. Omeprazole induced the increase in CYP3A4 transcription equally well in un-enriched and vitronectin-enriched SBS mats with 22-fold and 24-fold changes, respectively. In MG bilayer and MG-enriched SBS mats the increase in CYP3A4 mRNA was 13–14-fold and the lowest induction was again seen in laminin-111-enriched SBS mats with an average of fourfold change. CYP3A4 induction in response to rifampicin treatment was the highest in un-enriched SBS mats and in MG bilayer cultures with averages of 85- and 98-fold change, respectively. The CYP3A4 mRNA increase was lowest in laminin-111- (20-fold) and MG-enriched SBS mats (16-fold) with an intermediate induction in cells grown in vitronectin-enriched SBS mats (35-fold). Rifampicin-induced increase in CYP2B6 mRNA was the highest in cells grown on un-enriched SBS mats (in average 45-fold), in rest of the samples the induction of CYP2B6 mRNA remained below 20-fold, in case of vitronectin-enriched SBS mats below tenfold.
Cytochrome P450 induction in hepatocytes cultured on SBS mats or in MG bilayer. CYP1A2 and CYP3A4 were induced by omeprazole, CYP2B6 and CYP3A4 were induced by rifampicin. The human primary hepatocytes were cultured for 20 days and subsequently treated with indicated compounds for 4 days, n = 5. CYP induction was measured by RT-qPCR, fold changes in mRNA expression were calculated relative to uninduced control cells. * indicates a statistically significant (P < 0.05) difference compared to MG bilayer; NS not significant. Primary hepatocytes were cultured in serum-free hepatocyte defined growth medium.
To further substantiate our findings, we performed functional analysis of CYP enzyme activity of hepatocytes grown for 24 days in SBS mats or in MG bilayer (Fig. 7). We found that the enzymatic activity of CYP1A2 and 2B6 was even higher in the hepatocytes grown on SBS mats at both basal level and when induced by omeprazole or rifampicin, respectively, when compared to the hepatocytes grown in MG bilayer. The activity of CYP3A4 in hepatocytes grown in SBS mats and MG bilayer system was similar at both basal level and when induced with omeprazole or rifampicin.
Quantification of CYP 1A2, 2B6 and 3A4 induction by enzymatic activity. CYP1A2 and CYP3A4 were induced by omeprazole, CYP2B6 and CYP3A4 were induced by rifampicin. The human primary hepatocytes were cultured for 20 days and subsequently treated with indicated compounds for 4 days, n = 5. CYP activity was quantified by luminescent method. * indicates a statistically significant (P < 0.05) difference, ** P < 0.01 compared to MG bilayer. RLU relative light units. Primary hepatocytes were cultured in serum-free hepatocyte defined growth medium.
