Liver donor age affects hepatocyte function through age-dependent changes in decellularized liver matrix

Abstract

The only treatment available for end stage liver diseases is orthotopic liver transplantation. Although there is a big donor scarcity, many donor livers are discarded as they do not qualify for transplantation. Alternatively, decellularization of discarded livers can potentially render them transplantable upon recellularization and functional testing. The success of this approach will heavily depend on the quality of decellularized scaffolds which might show variability due to factors including age. Here we assessed the age-dependent differences in liver extracellular matrix (ECM) using rat and human livers. We show that the liver matrix has higher collagen and glycosaminoglycan content and a lower growth factor content with age. Importantly, these changes lead to deterioration in primary hepatocyte function potentially due to ECM stiffening and integrin-dependent signal transduction. Overall, we show that ECM changes with age and these changes significantly affect cell function thus donor age should be considered as an important factor for bioengineering liver substitutes.

1. Introduction

World Health Organization estimates that 20 million individuals to have cirrhosis and/or liver cancer globally and 1–2 million die annually as a result of liver failure[1]. The gold standard to treat patients with end stage liver failure is orthotopic liver transplantation[2]. Although this approach leads to high survival rates, the number of donor livers fall short of the worldwide need. Different clinical approaches such as split liver transplantation and live donor transplantation have been developed to decrease the organ shortage[3]. However, the large gap between donor livers needed and those that qualify for transplantation remained stable throughout the years[4] showing the need for an immediate action to develop alternative strategies.

Although there is an increasing global demand for donor livers which results in 8 patients dying awaiting transplantation daily, about 250,000 livers are estimated to be discarded due to conditions such as advanced age and ischemic damage after cardiac death[5]. Use of such livers for expanding the donor pool while bypassing the drawbacks of impaired hepatocyte function is possible through decellularization which was shown in rat[6], porcine[7], and human livers[8]–[10] up to date. The advantage of using native liver matrix to develop a liver substitute over other liver bioengineering methods is that it presents an ideal structure for cell engraftment, function, and transplantation due to its preserved architecture and composition[11]–[13]. The liver extracellular matrix (ECM) has a crucial role in its development and functionality via facilitating cell attachment and migration, and controlling differentiation, repair and development[14]. As such, the otherwise non-transplantable pool of donor livers creates a tremendous and unique opportunity to obtain decellularized native liver scaffolds, where the crucial ECM component is preserved while the impaired hepatocyte function can be replaced through recellularization of healthy cells such as patient specific iPSC-derived functional hepatocytes.

In translating this approach to clinical settings, it is important to consider that the livers available for such use are the ones rejected for transplantation due to reasons including advanced age. A donor age dependent decline in liver regenerative capacity has been reported[15]–[18] and many liver diseases including alcohol-induced cirrhosis, viral hepatitis-induced cirrhosis, diabetic-associated chronic liver disease, hepatocellular carcinoma, and biliary cirrhosis were shown to be more prevalent in the elderly[19]–[21]. Even though the effects of aging on hepatocyte and overall liver function are well-known[22], the critical question of how the liver ECM changes with age and if these changes would affect the liver cell engraftment and the hepatocyte function, thus function of the liver substitute, has not yet been answered. In this study we report a new protocol for homogenously decellularizing whole human livers and show that older donor age is correlated with an opaquer scaffold appearance and affects ECM composition in both rats and humans (Figure 1). Further, our results demonstrate that these significant age-related differences in ECM lead to altered hepatocyte function, which is crucial to consider moving forward towards the development of functional decellularized liver substitutes, and important in understanding the fundamental changes in ECM biology in the context of aging.

Figure 1. The schematic representation of the study outline.

The differences observed between old versus young decellularized whole human liver scaffolds (1) were examined in the controlled rat liver model (2). The effects of age-based ECM changes on liver cell behavior was validated using human livers and cells (3).

2. Methods

2.1. Whole human liver decellularization

Donor human livers unsuitable for transplantation were provided by New England Donor Services. Six livers were grouped according to age as young (n=3; 18-year-old, 19-year-old, and 24-year-old) and old (n=3; 46-year-old, 48-year-old, and 52-year-old). None of the donors were listed to have history of liver disease and detailed demographics are provided in Supplementary Table 1. The livers were maintained frozen at −80°C until decellularization. The livers were cannulated at portal vein and hepatic artery and perfused at 80 mL/min from portal vein and at 40 mL/min from hepatic artery. The flow rate was increased to 360 mL/min for portal vein and to 180 mL/min for hepatic artery for 5 mins every hour for 12 hours each day. During these 5 minutes the livers were flipped and massaged to physically induce removal of cellular materials. The decellularization was started with deionized (DI) water perfusion for 16 hours. Next, the livers were perfused with 0.1%, 0.2%, and 0.5% sodium dodecyl sulfate (SDS) (Sigma Aldrich) for 24 hours each. Finally, the livers were washed with DI water for 2 hours, then with 1% Triton X-100 (Millipore Sigma) for 2 hours and with PBS for 3 hours. For analysis, biopsy samples were collected from each liver right before starting decellularization (T=0) at the tip of left lobe. At the end of decellularization biopsy samples from 8 different locations were collected: right lobe tip, right lobe center, left lobe tip, left lobe center, caudate lobe center, quadrate lobe center, upper back center of right lobe, and core of the liver for analysis.

2.2. Histological analysis

The native and decellularized human liver tissues were fixed with 10% formalin for 24 to 48 hours at Room Temperature (RT) and then maintained in 70% ethanol at 4°C. The tissues were then dehydrated and embedded in paraffin. The tissues were micro sectioned to 5 μm thick slices and stained with hematoxylin (Leica) and eosin (Leica) (H&E) to visualize the ECM and cell nuclei. In order to assess collagen fibers, Masson’s Trichrome staining was performed to 5 μm thick slices of paraffin embedded decellularized liver tissues using standard protocols. The stained sections were imaged using Nikon Eclipse E800.

2.3. Development of decellularized human and rat liver matrices for cell culture

Discarded young (19-year-old male (n=1)), and old (60-year-old female (n=1)) donor human livers were provided by New England Donor Services. None of the donors were listed to have history of liver disease (Supplementary Table 1). Rat livers were harvested from young (3-month-old) (n=3) and old (18-month-old) (n=3) rats in accordance with the Institutional Animal Care and Use Committee (IACUC) at Massachusetts General Hospital.

The whole rat livers and 4-cm³ sections of the human livers were frozen at −20°C for at least 1 hour and cryosectioned to 50-μm thick slices prior to start of decellularization, similar to our previously established protocol[9]. The decellularization was achieved through incubating the liver slices in DI water for 16 hours followed by 3 hours 0.01% and 1.5 hours 0.1% SDS under constant agitation. The decellularized tissues were then washed for 1.5 hours in 1% Triton X-100 to remove SDS followed by extensive PBS washes to remove any remaining detergent from the ECM. The yielding rat decellularized liver matrix (rDLM) and human decellularized liver matrix (hDLM) were then lyophilized and stored at 4°C until further use.

2.4. Assessment of DNA, total collagen, collagen isomerization, sulfated glycosaminoglycan, and growth factor content

The amount of DNA remaining in the whole human livers, rDLM and hDLM was assessed to ensure complete decellularization. DNA was isolated using Purelink genomic DNA kit (Invitrogen) following manufacturer’s instructions. The amount of DNA was measured using NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific) as defined by absorbance at 260-nm wavelength. Native rat and human livers were used as the control for the respective DNA amounts before decellularization. The tissue weight corrections for rDLM and hDLM were done against dry tissue weight for both native and decellularized tissues. The weight corrections for whole human livers were done against wet tissue weight for both native and decellularized tissues. DNA content for decellularized human livers is represented as the average of the values obtained from the biopsies of 8 different locations to better represent the homogeneity throughout the liver.

DNA fragment size analysis was performed using FlashGel electrophoresis kit (Lonza). The DNA isolated along with a DNA ladder (FlashGel DNA marker 100 bp-4000bp) was run in 1.2% FlashGel Cassette at 275 V for 5 minutes as suggested by the manufacturer. The gel was imaged using a digital camera.

Total collagen content of the rDLM and hDLM was assessed using total collagen kit (QuickZyme Biosciences) following manufacturer’s protocol and the total collagen amount was calculated as μg collagen per mg of dry tissue. The degree of isomerization of collagen type I was determined in young and old decellularized whole livers. Briefly, the liver samples were weighed and digested in 0.01 mg collagenase 1A (Sigma Aldrich) in PBS per mg tissue at 37°C for 16 hours with agitation. The α-CTX and β-CTX content was then determined using Alpha Crosslaps ELISA kit (BioVision) and Beta Crosslaps ELISA kit (MyBioSource), respectively and represented as pg α-CTX or β-CTX per mg tissue.

Total sulfated glycosaminoglycan (GAG) content was determined using a previously established protocol[9]. Briefly, the rDLM and hDLM were hydrolyzed in HCl (6 M) for 20 hours at 95°C. The hydrolyzed sample was then mixed with 50 μM of dimethylene blue solution and total amount of sulfated GAGs was determined with respect to a chondroitin sulfate standard at 525 nm.

The amount of growth factors in rDLM was determined using rat growth factor array (RayBiotech) following manufacturer’s instructions. The rDLM was homogenized prior to growth factor assessment. Concentration of the solubilized rDLM was determined through subtracting the dry weight of the insoluble portion from the total dry weight measured before homogenization.

2.5. Mass Spectrometry analysis

Rat (n=3) and human (n=2) liver matrices were homogenized in PBS prior to mass spectrometry analysis. The analysis was done at Taplin Mass Spectrometry Facility at Harvard Medical School. Briefly, the samples were dehydrated with acetonitrile for 10 minutes followed by removal of acetonitrile. Pieces were then completely dried in a speed-vacuum. Rehydration of the samples was with 50 mM ammonium bicarbonate solution containing 12.5 ng/μl modified sequencing-grade trypsin (Promega) at 4°C. After 45 min, the excess trypsin solution was removed and replaced with 50 mM ammonium bicarbonate solution to just cover the samples. Then, the samples were kept at 37°C overnight. Peptides were later extracted by removing the ammonium bicarbonate solution, followed by one-time wash with a solution containing 50% acetonitrile and 1% formic acid. The extracts were then dried in a speed-vacuum for approximately 1 hour. The samples were then stored at 4°C until analysis.

On the day of analysis, the samples were reconstituted in 5–10 μL of HPLC solvent A (2.5% acetonitrile, 0.1% formic acid). A nano-scale reverse-phase HPLC capillary column was created by packing 2.6 μm C18 spherical silica beads into a fused silica capillary (100 μm inner diameter x ^~30 cm length) with a flame-drawn tip[23]. After equilibrating the column, each sample was loaded via a Famos auto sampler (LC Packings) onto the column. A gradient was formed, and peptides were eluted with increasing concentrations of solvent B (97.5% acetonitrile, 0.1% formic acid).

As peptides eluted, they were subjected to electrospray ionization and then entered into an LTQ Orbitrap Velos Pro ion-trap mass spectrometer (Thermo Fisher Scientific). Peptides were detected, isolated, and fragmented to produce a tandem mass spectrum of specific fragment ions for each peptide. Peptide sequences (and hence protein identity) were determined by matching protein databases with the acquired fragmentation pattern by the software program, Sequest (Thermo Fisher Scientific)[24]. All databases included a reversed version of all the sequences and the data was filtered to between a one and two percent peptide false discovery rate.

2.6. Primary rat and human hepatocyte culture

The rDLM and hDLM were pepsin digested for solubilization prior to cell culture according to our previously established protocol[9]. Briefly, the freeze-dried rDLM and hDLM were incubated in pepsin (Sigma Aldrich) solution (1 mg/mL, pH=2) at a concentration of 10 mg rDLM per 1 mL pepsin solution for 24 hours at RT with constant agitation. Concentration of the solubilized rDLM and hDLM were assessed through determining the weight of undigested rDLM after centrifugation and freeze drying. The gels were then diluted using 0.01 M HCl to reach 1.25 mg/mL final concentration and maintained at 4°C until use.

Primary rat hepatocytes isolated from 3-month-old female Lewis rats were acquired from Cell Resource Core (Massachusetts General Hospital) at 90% viability and seeded immediately following isolation. Cryopreserved primary human hepatocytes isolated from a discarded donor liver of 10-year-old female donor with no liver disease history were acquired from Lonza (lot# 4227) at 87% viability. Both rat and human hepatocytes were maintained in sandwich culture of the respective liver matrix gels in 24 well plates where rat tail collagen was used as control. The gelation of collagen and the liver matrices was achieved through neutralization with 10x Dulbecco’s Modified Eagle Medium (DMEM) with 1/10 dilution in the respective matrix solution and incubation at 37°C for 1 hour. Rat hepatocytes were seeded on collagen, young rDLM, or old rDLM (1.25 mg/mL) gel covered wells at 250 × 10³ cells/cm² density. Human hepatocytes were seeded on collagen and young and old hDLM (1.25 mg/mL) gel covered wells at 250 × 10³ cells/cm² density.

The rat hepatocytes were allowed to attach for 30 min, and human hepatocytes were allowed to attach for 60 minutes then both were gently washed to remove dead and unattached cells. Following the wash, at least 3 wells for each group were imaged at 5 different spots. These images were used in determining initial attachment density through cell counting using ImageJ software for both rat and human hepatocytes. The rat hepatocytes were imaged 24 hours after seeding and the respective images were used to determine initial cell spreading through assessing the cell covered area percentage per well.

Twenty-four hours after seeding, both rat and human hepatocytes were covered with another layer of gel to form the complete sandwich culture of collagen or the respective rat or human liver matrix gel. The rat hepatocytes were maintained in hepatocyte growth media (DMEM supplemented with 10% fetal bovine serum, 0.5 U/mL insulin, 7 ng/mL glucagon, 20 ng/mL epidermal growth factor, 7.5 μg/mL hydrocortisone, 200 U/mL penicillin/streptomycin, and 50 μg/mL gentamycin) at 37°C supplemented with 10% CO₂ with daily media changes for the duration of the experiments. Human hepatocytes were maintained in hepatocyte growth medium (Lonza) at 37°C supplemented with 5% CO₂ with daily media changes. For all cellular assays, 24 hours after laying the second gel layer is considered as day 1.

2.7. Metabolic activity, albumin and urea assays

For measuring metabolic activity, PrestoBlue (Thermo Fisher Scientific) was added to culture media of rat hepatocytes on day 3 of culture at the concentration suggested by the manufacturer and incubated for 2 hours at 37°C, 10% CO₂. Following the incubation, media samples were used for the measurement.

For albumin and urea assays, the media collected from both rat and human hepatocytes at day 3 of culture was used. Level of albumin secreted in rat hepatocytes was determined following a direct, competitive enzyme-linked immunosorbent assay (ELISA) protocol as described before[25]. Albumin secreted by human hepatocytes was determined using Human Albumin ELISA kit (Abcam) following manufacturer’s instructions. The amount of urea secreted was determined using urea nitrogen direct kit (Stanbio) following the manufacturer’s instructions. The results were corrected to compensate for well-well differences in cell number using the cell counts.

2.8. Quantitative RT-PCR (qRT-PCR) and CYP activity assay

For qRT-PCR analysis, RNA was collected from hepatocytes using Purelink RNA kit (Invitrogen) and then converted to cDNA using iScript cDNA synthesis kit (Biorad) following manufacturer’s instructions. The respective cDNA was then used for qRT-PCR reaction using ViiA7 Real time PCR system (Thermo Fisher Scientific). The list of primers used is provided in Supplementary Table 2. All expression levels were normalized to GAPDH expression and results were represented relative to collagen sandwich cultures.

The activity of CYP1A1, CYP1A2, and CYP2B2 in rat hepatocytes was determined using ethoxyresorufin-O-deethylase (EROD) methoxyresorufin-O-deethylase (MROD), and pentoxyresorufin-O-deethylase (PROD) assay, respectively, as shown before[26] at day 6 of culture. The rat hepatocytes were induced for 48 hours using 2 μM 3-methylcholanthrene. For human hepatocytes CYP3A4 activity was measured using P450-Glo™ CYP3A4 Assay (Promega) following manufacturer’s instructions at day 7 of culture. Human hepatocytes were induced for 48 hours with 25 μM rifampicin before measurements. The results were corrected using cell counts to compensate for cell number differences in different wells.

2.9. Bile canaliculi assessment

Bile canaliculi formation was visualized on day 3 and 6 of culture for rat and day 5 and 7 for human hepatocytes. Briefly, after the cells were washed with PBS they were incubated in 400 nM 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (CDFDA) (Enzo Life Sciences) supplemented PBS. Cells were incubated for 5 min (37°C, 10% CO₂) then washed once with PBS and imaged immediately using EVOS fluorescence microscope (Thermo Fisher Scientific) to visualize the bile canaliculi and with phase/contrast to show cell integrity. In order to quantify bile canaliculi integrity in human hepatocytes, a total of 5 images per well from 3 wells for each group (collagen, young hDLM, and old hDLM) were analyzed for fluorescence intensity using ImageJ software.

2.10. Oxidative stress treatment and reactive oxygen species accumulation measurements

The oxidative stress treatment was induced by exposing the rat hepatocytes to 0.2 mM H₂O₂ containing media for 48 hours. Following this treatment, the media was changed to fresh media and cells were incubated for 24 hours in this media. The control group was incubated in media without H₂O₂ for 72 hours with a media change at 48 hours. At the end of the 72 hours, cells were labeled with live/dead assay (Life Technologies) following manufacturer’s instructions. At least 3 wells for each group were analyzed through taking 5 images per well using EVOS fluorescence microscope (Thermo Fisher Scientific). The cell survival was calculated through determining the number of live and dead cells, and the respective live cell percentage, using ImageJ software. The normalized survival percentage was determined by dividing the average live cell percentage of treated group to that of the corresponding control and multiplying by hundred. The oxidative stress treatment was administered either on day 3 or day 6 of culture.

To determine the mitochondrial reactive oxygen species (ROS) accumulation in response to oxidative stress, hepatocytes were exposed to 0.2 mM H₂O₂ for 48 hours followed by 24 hours normoxia, as described above. At the end of treatment, the mitochondrial ROS levels were measured using CellROX green reagent (Molecular Probes) following manufacturer’s instructions. The fluorescence intensity recorded is correlated with ROS amount thus the difference in fluorescence intensity between each respective group’s control and treatment was calculated and represented as increase in ROS.

2.11. Statistical Analysis

Microsoft Excel Office 365 (Version 16.39) and GraphPad Prism (Version 8.3.1) were used for statistical analysis. All experiments using rat hepatocytes were repeated using cells from 3 different isolations. The results are represented as average ± standard deviation of different experiments. All experiments using human hepatocytes were performed with cells isolated from a single donor with at least 3 different wells representing technical replicates. The results are represented as average ± standard deviation of different wells. All groups were compared to each other using Student’s t-test with Welch’s correction and the statistical significance is defined as p<0.05 for all experiments. Principal component analysis for showing the differences in young and old hDLM was performed using GraphPad Prism (9.0.0).

3. Results

3.1. Whole human decellularized liver matrix shows donor age-dependent structural differences

We aim to develop transplantable liver grafts using otherwise discarded human livers. Thus, complete removal of cellular materials from such grafts is a crucial first step. We have developed a protocol to successfully decellularize whole human livers by perfusion through portal vein and hepatic artery. Using flow rates ranging from 40 mL/min to 360 mL/min, we were able to decellularize whole human livers acquired from donors at different ages within as short as 4 days. We used an increasing SDS concentration followed by a series of washes with DI water, Triton X-100, and PBS (Figure 2A). The success of decellularization was monitored through the visible color change from brown to white throughout the decellularization process. We did not observe any further color change after 0.5% SDS perfusion. The color change showed that the decellularization protocol was effective regardless of donor age. Although all livers appeared white at the end decellularization protocol, old livers had an opaquer appearance compared to young livers (Figure 2B). We have confirmed the success of decellularization of whole livers through histological analysis at the start and end of the process. The H&E staining showed that the cellular materials were removed as indicated by the absence of positive staining for cell nuclei (Figure 2C) as opposed to before decellularization (T=0). It should be noted that at the time of biopsy collection for T=0 the livers have gone through freeze-thaw, thus the damaged appearance of hepatocytes is expected. Through histological analysis, we also assessed the structural integrity of the livers after decellularization. Both H&E and trichrome staining showed a denser collagen structure in old liver scaffolds compared to the young livers. In line with histological analysis, our DNA assessment showed that minimal amount of residual DNA remained in the livers following decellularization. We recorded less than 25 ng residual DNA per mg tissue on average in both old and young livers which is less than the minimum necessary 50 ng per milligram tissue for decellularization (Figure 2D). Altogether, our results indicated that whole human liver decellularization of discarded donor livers can be achieved within 4 days with no visible damage to the structural organization of the liver ECM. However, there are structural differences evidenced by the increased opaqueness and density of fibers which is dependent on donor age suggesting that there are potential compositional differences as well.

Figure 2. Whole human liver decellularization can be achieved regardless of donor age.

(A) The schematic representation of the decellularization method. (B) The images of livers from one 24-year-old (top) and one 52-year-old donor (bottom) through decellularization process. (C) The H&E staining of human livers sections right before (T=0) (left) and at the end of decellularization (middle), and trichrome staining of human liver sections showing collagen fibers at the end of decellularization (right) (H&E: cell nuclei, purple; ECM, pink. Trichrome: collagen, blue). (Scale bars=200 μm) (D) DNA content in young and old human livers before decellularization (native) and after whole liver decellularization (decellularized).

3.2. Aging leads to a decrease in growth factor and an increase in total collagen and sulfated GAG content in rat liver ECM.

Following up on our early observations with donor age-related changes in human liver ECM, we aimed to determine the compositional age-dependent differences in liver matrix. To do so, we used the rat liver as our model to minimize batch to batch variation. We extracted young and old rDLM respectively from 3 and 18-month-old rats using a protocol we adapted from our previous studies[27]. We aimed to decrease the SDS exposure of ECM to better capture the minute differences that manifest solely due to age. To achieve that we compared a milder protocol to the previously described protocol, using 3-month-old rat livers. For the milder protocol, we decreased the 0.01% SDS exposure to 3 hours from 4 hours, increased the exposure time of 0.1% SDS and 1% Triton X-100 from 1 hour to 1.5 hour each, while skipping the 0.2% and 0.5% SDS incubations completely (Supplementary Fig. 1A). We observed that DNA removal with the milder method was as successful as the previous method, while around 10% more GAG was preserved (Supplementary Fig. 1B and C) indicating better preservation of the ECM. Thus, for extracting rDLM from young and old rat livers, we used the milder decellularization protocol (Figure 3A). The success of decellularization was not affected by the rDLM age as less than 55 ng DNA was remaining per mg tissue in both young and old rDLM with more than 95% DNA removal (Figure 3B). In addition, DNA fragment size analysis showed that only positive staining was observed below the 200 bp marker for young rDLM and showed no staining for old rDLM, in support of the very low amounts of DNA detected (Supplementary Fig. 1D). Although the DNA removal was similar, total collagen and sulfated GAG content was significantly higher in old rDLM (p = 0.0005 and p = 0.004, respectively). In addition, content of important growth factors including bFGF (p = 0.004), HGF (p = 0.036), GM-CSF (p = 0.009), and VEGF-A (p = 0.005) were significantly lower in old rDLM, suggesting an age-dependent deterioration in growth factor deposition in liver ECM.

Figure 3. Rat liver ECM is significantly affected by age.

(A) The schematic representation of the decellularization method. The quantitative measurement of (B) DNA, (C) total collagen, (D) sulfated GAGs, and (E) growth factor (basic fibroblast growth factor (bFGF), platelet derived growth factor subunits A (PDGF-AA) and B (PDGF-BB), hepatocyte growth factor (HGF), granulocyte-macrophage colony-stimulating factor (GM-CSF), and vascular endothelial growth factor (VEGF-A)) content in decellularized young and old rDLM. (Statistical significance is shown by * p<0.05, and ** p<0.01) (Student’s t-test, n=3)

In order to examine the overall age-dependent changes in rat liver ECM, we performed mass spectrometry analysis. We have represented here the core matrisome proteins as defined by Naba et al.[28] and categorized them as collagens, proteoglycans and ECM glycoproteins. We have identified 8 ECM proteins to be unique to old rDLM (Figure 4A). In old rDLM laminin A4 (lama4), laminin B2 (lamb2), laminin C1 (lamc1), tubulointerstitial nephritis antigen-like 1 (tinagl1), agrin (agrn), and elastin (eln) were found to be the unique ECM glycoproteins (Supplementary Table 3). In addition to presence of unique factors, we have observed significant changes in the levels of numerous factors. Among collagens, col1a1 (p = 0.004), col6a1 (p = 0.039), col6a5 (p = 0.9×10⁻⁶), and col6a6 (p = 0.008) were found in significantly higher amounts in old rDLM while col18a1 (p = 0.046) was higher in young rDLM (Figure 4B). Similarly, majority of ECM glycoproteins were lower in young rDLM. The fibrinogen alpha, beta, and gamma chains (fga, fgb, and fgg) (p = 0.0004, 2.7 × 10⁻⁵, and 0.012, respectively), fibronectin (fn1) (p = 0.019), and TGFB1 (p = 0.023) content was higher in old rDLM while emilin and tenascin XB (tnxb) were lower compared to young rDLM (Figure 4C). Finally, among proteoglycans, the level of lumican (Lum) were found to be significantly higher (p = 0.001) in old rDLM (Figure 4D).

Figure 4. Rat liver ECM shows significant age-related changes in composition.

Mass spectrometry analysis showed that old rDLM has unique ECM components. (A) The distribution of unique core matrisome components between young and old rDLM. The abundance of different (B) collagens, (C) ECM glycoproteins, and (D) proteoglycans in young and old rDLM. (* represents statistical significance, p<0.05. Student’s t-test, n=3).

3.3. Primary hepatocyte metabolism is affected by liver ECM age.

After full characterization of the rat liver ECM at different age groups and finding out the many differences, we examined if these differences affected the cell behavior. Therefore, we cultured primary rat hepatocytes isolated from 3-month-old rats in young and old rDLM gel sandwich configuration while using collagen gel sandwich culture as control. We first determined if the initial cell attachment to different matrices was different. The number of cells attached per well within the first 30 minutes of seeding showed that there were no significant differences between attachment to young rDLM vs. collagen or old rDLM (Supplementary Fig. 2A, B). In addition, we determined the cell covered area percentage 24 hour after seeding to qualitatively determine the cell-ECM interactions with respect to rDLM age. We observed that the most coverage was achieved on young rDLM while there was no statistically significant difference between collagen and old rDLM (Supplementary Fig. 2C, Figure 5A). The difference in cell covered area despite the similar initial attachment suggests that the cell spreading on young rDLM is superior than that on old rDLM.

Figure 5. Primary hepatocyte metabolism is affected by the age dependent changes in liver ECM.

(A) The micrographs showing the hepatocyte morphology on collagen, young or old rDLM at day 3 of culture. (B) The overall hepatocyte metabolism represented by the relative fluorescence measured through presto blue assay. The amount of (C) albumin, and (D) urea secreted from hepatocytes cultured in collagen, young rDLM or old rDLM sandwich cultures for 3 days. (E) The qRT-PCR analysis results showing the mRNA expression of proliferation markers Ki67 and PCNA in primary hepatocytes grown on collagen, young rDLM or old rDLM at day 6 of culture. (Statistical significance is shown by * p<0.05, Student’s t-test, n =3) (Scale bars: 200 μm (left panel), 20 μm (right panel)).

To determine if the hepatocyte metabolism is affected by liver ECM age, we assessed the overall metabolic response of the cells using PrestoBlue assay. On day 3 of culture the metabolism of hepatocytes on young rDLM was significantly higher than those cultured on collagen (p = 0.024) and old rDLM (p = 0.011) while there was no significant difference between collagen and old rDLM (Figure 5B). Similarly, the albumin and urea secretion by hepatocytes was the highest with 192 ± 7 μg/mL albumin and 339 ± 9 μg/mL urea per 1×10⁶ cells in hepatocytes after 3 days of culture on young rDLM. When maintained on collagen and old rDLM albumin secreted was 90 ± 10 μg/mL and 122 ± 10 μg/mL per 1×10⁶ cells, respectively (Figure 5C). The urea secreted on collagen and old rDLM was 246 ± 5 μg/mL and 266 ± 12 μg/mL per 1×10⁶ cells, respectively (Figure 5D).

In addition, we determined if the proliferative capacity of the hepatocytes changed in response to different seeding substrates upon culture for 6 days. The expression of ki67 and proliferating cell nuclear antigen (PCNA) showed that the proliferative capacity of hepatocytes is higher on young rDLM compared to old rDLM (p = 0.0001 and 0.048, respectively) (Figure 5E).

3.4. The age of rDLM affects the liver specific functionality of primary rat hepatocytes.

A common method to investigate polarization of hepatocytes is to determine bile canaliculi formation. We assessed the bile canaliculi formation on different substrates on early and late stages of culture. A subset of hepatocytes seeded on all substrates successfully formed bile canaliculi by day 3 of culture as shown by the distinctive disposal of CDFDA, a fluorescent dye secreted into the canaliculi formed in between cells. However, by day 6 of culture the integrity of bile canaliculi was more disrupted in hepatocytes cultured on collagen and old rDLM compared to those young rDLM which displayed polarized phenotype (Figure 6A). Similarly, the expression of some major CYP450 enzymes including CYP2D3, CYP2C11, CYP2C6, CYP2E1, and CYP1A2 were significantly higher in hepatocytes cultured on young rDLM compared to both collagen (p = 4×10⁻⁵, 0.0003, 1×10⁻⁶, 0.0001, 0.036, respectively) and old rDLM (p = 0.006, 0.024, 0.007, 0.006, 0.041, respectively) (Figure 6B). The expression of these enzymes except for CYP1A2, in cells cultured on old rDLM was also higher than collagen (p = 0.001, 0.005, 0.001, 0.004, respectively) showing the advantage of native ECM over a single type protein coating. We have also determined the activity of CYP450 enzymes CYP1A1, CYP1A2, and CYP2B2. Activities of CYP1A1 and CYP2B2 were measured to be significantly higher when hepatocytes were cultured on young rDLM compared to both old rDLM (p = 0.043 and 0.02, respectively) and collagen (p = 0.025 and 0.049) (Figure 6C). However, even though the mRNA expression of CYP1A2 is higher on young rDLM compared to both collagen and old rDLM, the difference in its activity did not reach significance compared to collagen while it was significantly higher only compared to old rDLM (p = 0.028).

Figure 6. Hepatocyte-specific functionality shows deterioration when cultured on aged liver ECM.

(A) The fluorescent images showing the bile canaliculi formed by hepatocytes on collagen, young or old rDLM at day 3 (left) and day 6 (right) of culture. (B) The qRT-PCR analysis results showing the mRNA expression of important CYP450 enzymes CYP2D3, CYP2C11, CYP2C6, CYP2E1, and CYP1A2, and (C) the kinetic activity (pmol/min/10⁶ cells) of CYP1A1, CYP1A2, and CYP2B2 in hepatocytes in hepatocytes cultured on collagen, young or old rDLM for 6 days. (Statistical significance is shown by * p<0.05; no significance is shown by N.S. p>0.05) (Student’s t-test, n=3) (Scale bars=400 μm (left panel), 200 μm (right panel))

3.5. Advanced age in rat liver ECM leads to susceptibility to oxidative stress in hepatocytes.

The age-dependent deterioration of organ function is well-studied. The susceptibility to stresses such as oxidative stress is another well-documented outcome of aging. The contribution of cellular aging and functional deterioration is unquestionable, however, the contribution of external factors such as the age-dependent changes in liver ECM is not known. Here, we tested the effect of liver ECM age on hepatocyte survival and mitochondrial ROS accumulation following exposure to oxidative stress (Figure 7). We exposed the hepatocytes to H₂O₂ induced oxidative stress followed by fresh media incubation at day 3 and day 6 of culture and determined the cell survival (Figure 7, Supplementary Fig. 3). When the cells were exposed to stress starting on day 3 of culture, normalized cell survival was significantly higher on young rDLM compared to collagen (p = 0.007), however, there was no significant difference compared to old rDLM. Since we have observed some age-related deterioration on cell behavior at later stages of culture, for example bile canaliculi disruption only occurred on day 6 but not on day 3 of culture, we determined the oxidative stress response on different culture substrates at later culture stages by exposing the cells to oxidative stress starting day 6 of culture. When we applied the same stress to hepatocytes on day 6, the cell survival was higher on young rDLM compared to both collagen and old rDLM (p = 0.0014 and 0.0012, respectively) (Figure 7C). The survival under oxidative stress on collagen, however, was not different from survival on old rDLM. In addition to overall survival, mitochondrial ROS accumulation is an important indicator of cellular damage under oxidative stress. Thus, we determined the mitochondrial ROS accumulation on day 6 of culture following the stress treatment (Figure 7B, D). Interestingly, the mitochondrial ROS accumulation was indifferent between collagen and young rDLM, although it was significantly higher on old rDLM (p = 5.6×10⁻⁵ compared to young rDLM and 0.029 compared to collagen). As apoptosis is the outcome most associated with oxidative stress, we assessed the changes in apoptosis related gene expression following the stress treatment. The expression of BAX, caspase 3, and caspase 7 were determined following the oxidative stress treatment initiated on day 6 of culture. We determined that BAX expression spiked on young and old rDLM, the highest fold change being reached on young rDLM. The expression of caspase 3 and 7 followed a different trend where higher expression was recorded on old rDLM compared to young rDLM and collagen (young rDLM vs. old rDLM p =1.09×10⁻⁷, 0.0004, 0.0402, 0.0009, 0.0001 for BAX, caspase3, caspase7, p21 and p16, respectively) (Figure 7E). In addition, we assessed the expression of senescence-related markers p16 and p21 in hepatocytes cultured on different substrates (Figure 7F). We determined that both p21 and p16 expression was the lowest in cells maintained on young rDLM. There was a significant increase in expression of both markers on collagen and old rDLM substrates, indicating that cell-ECM interactions contribute to important guidance in cell fate.

Figure 7. Age-dependent ECM changes in liver affect the stress response of hepatocytes.

(A) The Live/Dead images showing the viability and (B) the fluorescence images showing mitochondrial ROS accumulation of hepatocytes following oxidative stress treatment that started on day 6 of culture. (C) The normalized cell survival (%), and (D) increase in ROS in hepatocytes cultured on different substrates in response to oxidative stress treatment. (E) The qRT-PCR analysis results showing the mRNA expression of apoptosis markers BAX, caspase3, caspase7, and (F) senescence markers p21 and p16 when hepatocytes on collagen, young or old rDLM are exposed to oxidative stress. (Statistical significance is shown by * p<0.05; no significance is shown by N.S. p>0.05) (Student’s t-test, n=3) (Scale bars=400 μm)

3.6. Human donor age affects matrix composition and hepatocyte functionality on decellularized liver matrix

Next, we assessed human DLM composition and its capacity to support hepatocyte function and regeneration as a function of donor age in an effort to translate out findings in rats. Following slice decellularization of human livers we obtained residual DNA content of 21.5±9 ng DNA/mg tissue for young and 42.7±7 ng DNA/mg tissue for old hDLM (Supplementary Fig. 4A). In support of the small quantity of residual DNA, fragment size analysis showed no positive DNA staining (Supplementary Fig. 4B). We determined the total collagen and sulfated GAG content in decellularized human livers. We observed that old hDLM had a higher collagen content with 98 ± 22 μg collagen per mg tissue compared to young hDLM which had 42 ± 8 μg collagen per mg tissue (p=0.013) (Supplementary Fig.4C). Similarly, GAG content was higher in old hDLM (7 ± 2 μg GAG per mg tissue) compared to young hDLM (3 ± 1 μg GAG per mg tissue) (p = 0.023) (Supplementary Fig.4D). In order to ensure that the slice decellularization did not result in excess removal of ECM components due to the surface area per unit volume of treatment solution compared to whole liver decellularization, we have compared the total collagen and sulfated GAG amounts between slice and whole liver decellularization (Supplementary Fig. 4B, C). Our results show that the age-matched livers had similar amounts of total collagen and sulfated glycosaminoglycans regardless of the decellularization method used.

We performed mass spectrometry analysis and determined the core matrisome proteins that are affected by donor age (Figure 8). Similar to the rat matrisome, we have grouped our findings to 3 groups: collagens, ECM glycoproteins, and proteoglycans. We have detected 4 unique components in young hDLM and 6 unique components in old hDLM (Figure 8A, Supplementary Table 4). Consistent with rat matrisome, we detected majority of collagens to be more abundant in old hDLM compared to young, including col1a1, col6a1, and col6a6 (Figure 8B). All ECM glycoproteins were detected to be higher in old hDLM (Figure 8C). In addition, heparan sulfate proteoglycan 2 (hspg2) and lum were found to be significantly more abundant in old hDLM compared to young hDLM (Figure 8D). We have also used the principal component analysis to compare the young and old hDLM. Our analysis showed that the old donors and young donors were separately clustered showing similarities with the age-matched samples while indicating the differences between young and old donor liver matrices (Figure 8E). We also determined the content of α- and β-CTX in young and old decellularized human livers (Supplementary Fig. 4E, F) and although there was no significant difference in α-CTX content, β-CTX was significantly higher in old liver matrix (p=0.003). Importantly, the ratio of α-CTX to β-CTX was 9-fold higher in young liver matrix (Supplementary Fig. 4G) similar to the difference observed between fetal bone compared to adult bone[29].

Figure 8. Donor age affects human liver ECM composition significantly.

Mass spectrometry analysis results showed that young and old hDLM have unique ECM components. (A) The distribution of unique core matrisome components between young and old hDLM. The abundance of different (B) collagens, (C) ECM glycoproteins, and (D) proteoglycans in young and old hDLM. (* represents statistical significance, p<0.05. Student’s t-test, n=2). (E) The principal component analysis of the mass spectrometry data showing the clustering of two young and two old donors separately, indicating the age-related differences in liver matrix.

In order to examine the effect of the age-related differences in human liver ECM on hepatocyte function, we cultured cryopreserved human hepatocytes on young (19-year-old) and old (60-year-old) hDLM substrates. We decellularized the human liver sections using the protocol we employed for rat livers and performed pepsin digestion for preparing the respective hydrogels. Human hepatocytes were then cultured in sandwich configuration where collagen gel sandwich culture was used as control. Hepatocytes showed no significant differences in morphology upon initial attachment to different aged ECM substrates (Figure 9A). Bile canaliculi formation was also observed regardless of age, comparable to collagen controls by day 5 of culture (Figure 9B). However, on day 7 the bile canaliculi were disintegrated almost completely on collagen and old hDLM while they remained intact in cells on young hDLMs (Figure 9C) even though the cells remained intact in all groups (Supplementary Fig. 5). The quantification of fluorescence intensity also showed that on day 7 of culture the bile canaliculi abundance and integrity was significantly higher on young hDLM compared to both old hLDM and collagen (young vs. old hDLM p=0.003 young hDLM vs. collagen p=0.013) (Figure 9D). The albumin and urea secretion on day 5 of culture was significantly higher in hepatocytes when cultured on young hDLM, compared to collagen controls and old hDLM (young vs. old hDLM, albumin secretion: p=0.008; urea secretion: p=0.043) (Figure 9E, F). Albumin secretion was significantly higher in both hDLM groups than collagen controls (p=0.019, 0.018 for young hDLM vs. collagen and old hDLM vs. collagen, respectively). In addition, we determined the mRNA expression of proliferation markers ki67 and PCNA and observed that culture on young hDLM resulted in significantly higher expression of both markers compared to old hDLM (young vs. old hDLM, ki67: p=2.6×10⁻⁸, and PCNA: p=0.035) (Figure 9G).

Figure 9. Human Hepatocytes show deteriorated functionality when cultured on older aged donor liver ECM.

(A) The phase contrast images on day 3, (B) and (C) the fluorescent images showing the bile canaliculi formed by hepatocytes on collagen, young, and old hDLM at days 5 and day 7 of culture, respectively. (D) The fluorescence intensity per field of view (fov) showing abundance of bile canaliculi in hepatocytes on day 7 of culture. (E) The albumin secretion, and (F) urea secretion of human hepatocytes cultured on different aged hDLMs and collagen controls on day 5 of culture. (G) The qRT-PCR analysis results showing the mRNA expression of proliferation markers ki67 and PCNA, and (H) of albumin and important P450 enzymes CYP1A2, CYP2B6, CYP2D6, CYP3A4, in human hepatocytes on day 7 of culture. (I) CYP3A4 activity of human hepatocytes cultured on different hDLMs and collagen controls on day 5 of culture. (* represents Statistically significant difference compared to collagen controls, and + represents statistically significant difference compared to old hDLM, p<0.05) (Student’s t-test, n=3) (Scale bars=200 μm)

The mRNA expression of important P450 enzymes showed a similar trend to albumin and urea secretion. Expression levels of albumin, CYP1A2, CYP2B6, CYP2D6, and CYP3A4 was significantly higher in hepatocytes cultured on young hDLM compared to collagen controls and old hDLM (young vs. old hDLM, p=0.008, 0.003, 0.015, 0.009, 0.012 for albumin, CYP1A2, CYP2D6, CYP2B6, and CYP3A4, respectively) (Figure 9H). In addition to mRNA expression, we determined the activity of CYP3A4 in hepatocytes (Figure 9I). Our results showed that CYP3A4 activity was significantly lower on collagen controls and old hDLM gels, (p=0.0014 for col vs. young hDLM, p=0.0094 for young vs. old hDLM) in line with the respective mRNA expression and with the other hepatocyte specific functionality we have assessed. We have observed a similar deterioration in function of human hepatocytes isolated from a 36-year-old donor when cultured on old hDLM compared to young hDLM (Supplementary Fig. 6). Overall, our data with human liver ECM and human hepatocytes showed strong correlation with our observations on rat liver ECM and rat hepatocyte interactions. All together, these results suggest an age-related alteration of the liver ECM which in turn takes role in deterioration of cellular function.

4. Discussion

Decellularization of whole organs is a promising approach to increase donor pool. One of the important challenges moving towards clinical applications is that the scaffold development will depend on the ECM of non-ideal livers most of which will be acquired from advanced age donors. Therefore, it is critical to study the changes introduced to ECM due to advanced donor age. To develop clinically relevant scaffolds, we have developed a 4-day long protocol to achieve complete and homogenous decellularization of whole human livers. This protocol allows for fast and efficient decellularization without damaging the structure of the ECM. Up to date, the reports of whole human liver decellularization show decellularization of 1 to 3 livers[8], [10], [30] and the effect of donor age dependent differences in decellularization efficiency is not explored. We have decellularized 6 whole human livers that were grouped into 2 categories, namely young and old. We compared the decellularization outcomes in terms of appearance, DNA removal and structural preservation of the ECM. We showed that our protocol is highly efficient in decellularization of discarded human livers regardless of donor age, although we observed a difference in appearance of the livers as old livers were opaquer. Both the histological analysis and DNA content assessment confirmed that the age-dependent difference observed in scaffold transparency is not caused by remaining cellular materials in the old liver scaffolds. Histological analysis further suggested that this difference might be caused by compositional and structural changes that the liver ECM undergoes with age as older liver scaffolds had a denser fiber organization. This is supported by the increased levels of fibrillar collagens and glycoproteins in old hDLM. Overall, although the decellularization efficiency is not affected by donor age, the compositional alterations in liver ECM might introduce new challenges in the utilization of these scaffolds in clinical applications.

The age-related decline in tissue and organ function is directly attributed to cellular deterioration, such as stem cell depletion and senescence, while the ECM has been considered to be dormant and provide structural support only. However, the microenvironmental changes and how these changes affect the cell function are newly being appreciated. Stearns-Reider et al. showed that the age-dependent changes in skeletal muscle ECM causes its stiffening and the related mechanotransduction signaling mechanism leads to the fibrogenic differentiation of muscle stem cells, thus result in stem cell depletion in skeletal muscle[31]. Similarly, in aged skin the changes in ECM lead to poor integrin-ECM binding, affecting cell fate[32], [33]. Similar effects of aging on ECM composition and structure and the related changes in cell fate and function have been reported for tendon[32] and heart[34], [35]. In liver, Delire et al. investigated the fibrosis development which was induced through CCl₄ injections in young (2-month-old) and old (15-month-old) mice[36]. They determined that the old mice developed more severe fibrosis with higher levels of inflammation and less fibrolysis was observed. As the growing evidence suggests, ECM is dynamic and crucial in determining tissue/organ functionality and regenerative capacity thus, it is very important to understand the age-related changes that take place in liver ECM to assess the feasibility of translating the decellularized liver scaffolds to clinical settings.

Up to date, the studies investigating the details of liver ECM has been focused on its remodeling leading to fibrosis and eventually to liver disease[37]–[40]. The contribution of age to liver fibrosis was investigated by Karsdal et al. by measuring the ECM components in serum of rats at different ages[41]. This study performed a systematic analysis of fibrogenesis, fibrolysis or ECM turnover markers with respect to age and CCl₄ induced liver fibrosis. They concluded that the overall ECM turnover rate was significantly different in older rats, stressing the importance of using animal models of relevant age for studying liver fibrosis and other diseases that are associated with aging. Similarly, our goal is to repurpose the discarded livers as decellularized scaffolds, which will provide a pool of advanced age livers thus, it is crucial to understand the age-related changes in liver ECM. Here we successfully decellularized young and old rat livers to characterize the changes in ECM composition solely due to age. We observed that the total collagen and GAG content increase with age in liver ECM. This is in line with liver fibrosis which is a condition more prevalent at older ages[41]. An important role of ECM is to bind to and secrete growth factors. Our results suggest that the age-related alterations lead to depletion in growth factor content. Specifically, bFGF, HGF, GM-CSF, and VEGF, all of which have important roles in liver function and regeneration[42], were significantly lower in old rDLM.

To gain a more in-depth understanding of the differences in ECM of young and old livers, we performed mass spectrometry analysis in both rat and human liver matrices. Overall, there were more types of ECM glycoproteins and collagens in old rDLM and hDLM. The higher collagen content detected by mass spectrometry supports our findings of increased total collagen with donor age in both rat and human livers. Similarly, Karsdal et al. showed an age-dependent difference in ECM turnover where type I and II collagen turnover was lower in old rats. In addition, the fibrillar type I collagen accumulation with fibrosis is also well-documented supporting the findings that show correlation between advanced age and fibrosis. In support of this, the presence of type III collagen only in old rDLM and higher levels of fibronectin and fibrinogen also coincide with early fibrotic events[43]. Further supporting the similarities between age-dependent changes in liver ECM and fibrosis development, we observed higher levels of collagen type VI in old rDLM and hDLM. The C-terminal fragment of the collagen type VI alpha 3 chain, which is also known as endotrophin, is shown to induce fibrosis and inflammation related genes in white adipose tissue and in tumor microenvironment[44], [45]. Importantly, our analysis showed a high number of ECM glycoproteins to be absent in young rDLM such as laminins. Similarly, in old human liver matrix ECM glycoproteins including laminins and fibrinogens were in significantly higher quantity. These factors all together form the basement membrane of the liver through self-assembling of collagen type IV and laminin into independent supramolecular networks which then link to nidogen and perlecan[46]. It should be noted that laminins are present in basement membrane of the healthy liver as well[46]. The reason for not detecting laminins in young rDLM can be attributed to minor ECM loss that is inevitable during decellularization. In contrast, we observed higher levels of tenascin XB in young rDLM and it was present only in young hDLM. This finding supports that tenascins are developmental components of the liver ECM and are not as abundant in adult liver[47]. As opposed to fibrillar collagens, non-fibrillar collagens like collagen types XIV and XVIII were more abundant or only present in young liver matrix. Collagen type XVIII was shown to be an important mediator of survival signals induced by α1ß1 integrin, making it a crucial factor in stress response and recovery[48]. In addition, endostatin, a signaling molecule released upon proteolytic cleavage of collagen type XVIII, is known to be mainly produced by hepatocytes and has anti-apoptotic and anti-fibrotic properties[49]. Similarly, collagen type XIV is an important component that binds to and regulates the assembly of collagen type I[50]. Interestingly, we detected elastin in old rDLM alone. Although in tissues like skin the decrease in elastin is associated with advanced age, in liver, elastin fibers aid in formation and stability of fibrotic septa, contributing to liver fibrosis[51]. Overall, the increase in fibrillar collagens and proteoglycans and glycoproteins, as well as a decrease in non-fibrillar collagens suggest that liver ECM stiffens with age.

The ECM of liver is important in regulating hepatocyte function. We determined that the differences we detected in young and old rDLM affected the hepatocyte metabolism and function. The fact that even though the similar number of hepatocytes attached to all substrates, they spread the most on young rDLM could be attributed to higher levels of collagen XVIII owing to its role in integrin binding. In addition, lower serum albumin levels in the elderly is well-documented. Our results indicate that the age-related changes in liver ECM might be at least partially involved in decreased serum albumin levels[52]. It is interesting to note that albumin and urea secretion as well as overall hepatocyte metabolism was similar in hepatocytes cultured on collagen and old rDLM substrates. This could be due to the increased collagen content in old rDLM, presenting a similar substrate for cell attachment and function. Contrarily, the CYP450 mRNA expression on old rDLM was significantly different than collagen. The bile canaliculi integrity also showed a similar trend to this indicating that the non-collagen components of the liver ECM take important role in regulation of CYP expression and bile canaliculi formation in hepatocytes. In addition, the lower cell survival and higher mitochondrial ROS accumulation in hepatocytes cultured on old rDLM is in line with the age-dependent decrease in growth factor and collagen type XVIII content. FGF is shown to have hepatoprotective roles[53] and HGF, among its many roles, was shown to ameliorate hepatic biliary fibrosis[54] and both of these factors were significantly reduced in old rDLM. Another effect of old rDLM was the decreased proliferative capacity observed compared to young rDLM. Although Ki67 expression was higher in young rDLM compared to both collagen controls and old rDLM, PCNA expression did not show a difference between collagen and young rDLM. The reason for this could be that collagen type I was shown to cause significantly higher induction of hepatocellular carcinoma cell proliferation compared to Matrigel through activation of ERK pathway[55]. Importantly, PCNA expression in young rDLM was significantly higher than old rDLM, in line with other functional deterioration observed. Finally, the lower expression of caspase 3 and 7 in hepatocytes cultured on young rDLM supports the higher survival and lower ROS levels detected following oxidative stress exposure. Interestingly, as opposed to caspase expression levels, we observed increased expression of BAX on young rDLM. It has been reported by Naderi et al. that human skin fibroblasts regulate BAX expression under oxidative stress differently[56]. The quiescent cells had low expression of BAX while the proliferative fibroblasts showed overexpression of BAX along with no increase in caspase 3 expression when exposed to 250 μM H₂O₂ for 1 hour. Given the higher expression of Ki67 and PCNA we detected on young rDLM, a similar mechanism might have resulted in the overexpression of BAX in hepatocytes cultured on young rDLM. Oxidative stress is also a well-known inducer of senescence. We examined the hepatocyte senescence induction under stress conditions through p16 and p21 expression levels. We found out that liver ECM alone plays an important role in senescence-inducing mechanisms under stress conditions suggesting that donor age could alone affect the regenerative capacity of the respective liver substitute.

In efforts to investigate the translatability of our results towards human DLM and hepatocytes, we have developed decellularized human liver matrices from young (19-year-old) and old (60-year-old) donors. Although the donor ages used as young and old are not exact matches for young and old rats used in this study, they represent a large enough age gap to determine the age-related differences. Our primary compositional characterization and mass spectrometry results as represented by individual matrisome components and principal component analysis demonstrated that age related differences observed in rats are similarly observed in humans as higher total collagen and sulfated GAG amounts as well as ECM glycoproteins and proteoglycans were recorded in old compared to young hDLM. This is also in line with higher rates of collagen deposition in fibrosis, which is positively correlated with age[16] and may be the reason behind the opaque appearance and denser structures observed in histological sections of decellularized whole livers from older donors. In addition, collagen type I is shown to undergo spontaneous β-isomerization with age in the CTX epitope leading to accumulation of β-CTX in bones while the non-isomerized form, α-CTX, remains more abundant in fetal bone. We also report a similar increase in β-isomerization of collagen with donor age in decellularized liver, showing another age-related change in liver matrisome.

In order to examine the effect of age-related differences in human liver ECM on cell function, we cultured human hepatocytes on hDLMs and collagen gels as controls. To allow for cell culture in sandwich configuration, we have developed the hDLM through slice decellularization. We made sure that the slice decellularization did not cause excess disruption of liver ECM compared to the whole liver decellularization by showing the similar collagen and GAG content in ECM isolated using both decellularization methods. Although surface area per unit volume of detergent is higher in slice decellularization, the detergent concentrations used, and the exposure time are shorter compared to whole liver decellularization which likely led to the similar ECM preservation despite the use of different approaches. The human hepatocytes were isolated from a 10-year-old donor which are age-matched with primary rat hepatocytes used providing better comparison to rat experiments. Similar to their rat counterparts, human hepatocytes were affected by the age of hDLM donor. Consistently, compared to collagen control gels, different hepatocyte-specific functions such as albumin and urea secretion, P450 enzyme expression and activity were higher on both young and old hDLM, stressing the importance of providing complex ECM components for better cell function. Among the hDLM, advanced donor age had a negative effect on all of the abovementioned functions. Altogether, our results strongly suggest a pattern of age-related deterioration in liver function induced by the advanced age of liver ECM. It should be noted that due to limited access to human livers at specific ages, the current conclusions are drawn from a single donor of the respective ages. Ideally, multiple donors of similar age should be used, and human hepatocytes isolated from various donors at same age should be cultured for making stronger conclusions.

In summary, here we showed that donor age has a significant effect on liver ECM composition where advanced age causes potential stiffening and decrease in growth factor deposition. Overall, we showed that these donor age-related differences are likely responsible for deteriorated hepatocyte function, and stress response. Our results also emphasize the importance of the unique and complex composition of native liver ECM on developing a fully functional liver substitute. Thus, with further investigating the human liver ECM from donors of different ages and disease backgrounds, crucial information can be collected and employed towards constructing viable liver substitutes with improved long-term functionality and stress response regardless of donor age. This way, functional human scale transplantable liver grafts can be engineered and made available, reducing liver failure caused deaths in the order of thousands of patients per year.

8. Data Availability

The data that support the findings of this study are available from the corresponding author, B.E.U., upon reasonable request.