Acta Biomaterialia 10 (2014) 875882

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Mechanostructure and composition of highly reproducible decellularized liver matrices

G. Mattei a,b,, V. Di Patria b, A. Tirella c, A. Alaimo b, G. Elia b, A. Corti d, A. Paolicchi d, A. Ahluwalia b
a

Department of Civil and Industrial Engineering, University of Pisa, Largo Lucio Lazzarino 1, 56122 Pisa, Italy Research Centre E. Piaggio, University of Pisa, Largo Lucio Lazzarino 1, 56122 Pisa, Italy c National Research Council, IFC, Via Moruzzi 1, 56124 Pisa, Italy d Department of Translational Research and New Technologies in Medicine and Surgery, University of Pisa, Via Roma 55, 56126 Pisa, Italy
b

a r t i c l e

i n f o

a b s t r a c t
Despite the increasing number of papers on decellularized scaffolds, there is little consensus on the optimum method of decellularizing biological tissue such that the micro-architecture and protein content of the matrix are conserved as far as possible. Focusing on the liver, the aim of this study was therefore to develop a method for the production of well-characterized and reproducible matrices that best preserves the structure and composition of the native extra cellular matrix (ECM). Given the importance of matrix stiffness in regulating cell response, the mechanical properties of the decellularized tissue were also considered. The testing and analysis framework is based on the characterization of decellularized and untreated samples in the same reproducible initial state (i.e., the equilibrium swollen state). Decellularized ECM (dECM) were characterized using biochemical, histological, mechanical and structural analyses to identify the best procedure to ensure complete cell removal while preserving most of the native ECM structure and composition. Using this method, sterile decellularized porcine ECM with highly conserved intra-lobular micro-structure and protein content were obtained in a consistent and reproducible manner using the equilibrium swollen state of tissue or matrix as a reference. A signicant reduction in the compressive elastic modulus was observed for liver dECM with respect to native tissue, suggesting a re-examination of design parameters for ECM-mimicking scaffolds for engineering tissues in vitro. 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Article history: Received 21 June 2013 Received in revised form 17 September 2013 Accepted 22 October 2013 Available online 30 October 2013 Keywords: Extracellular matrix Decellularization Biologic three-dimensional scaffold Biomaterial Tissue engineering

1. Introduction Maintaining hepatic functional characteristics in vitro has long been the holy grail of many researchers involved in the study of in vitro liver models. A number of scientists have suggested that the target has been elusive because hepatocytes in vitro are deprived of the multi-parametric, multi-stimuli in vivo milieu [1]. Reproducing the three-dimensional (3-D) features of the native liver is crucial, since the in vivo environment affects cell behaviour and function through a variety of soluble and insoluble signalling factors, including physical, chemical and mechanical cues (e.g., oxygen concentration, substrate stiffness) [2,3]. No single scaffold or scaffold material has been identied as optimal for hepatocytes, leaving this an open terrain for exploitation and development. The consensus is that a 3-D scaffold should provide an adhesive substrate that mimics extracellular matrix (ECM) ligands and matches both the stiffness and the porosity of the
Corresponding author at: Research Centre E. Piaggio, University of Pisa, Largo Lucio Lazzarino 1, 56126 Pisa, Italy. Tel.: +39 050 2217050; fax: +39 050 2217051. E-mail address: giorgio.mattei@centropiaggio.unipi.it (G. Mattei).

physiological matrix. Ideally, it should also be well characterized, reproducible and easy to fabricate, sterilize and store [4]. Biological scaffolds from decellularized tissues and organs have been used successfully for myriad applications in reconstructive surgery and regenerative medicine [5,6]. Recently, the success of whole organ perfusion, which involves sending detergents through the vasculature, as well as the deeper understanding of the role of matrix signals in guiding cell function, has reawakened an interest in matrix-derived materials and scaffolds [7]. Besides whole organ perfusion [810], common decellularization techniques include pressure gradients [1113], and immersion and agitation [14 16]. Most methods involve washing and saponication phases to remove blood and debris, respectively, and to hydrolyse cell membranes. Following an immersion and agitation approach, Lang et al. described a study to identify the most suitable combination of washing and saponication cocktail to decellularize thin slices of pig liver [14]. The authors optimized different decellularization/ oxidation procedures based on Triton X-100, ammonium hydroxide, phosphate buffered saline (PBS) and sodium chloride, coupled with a peracetic acid (PAA) treatment. They report 93% removal of cellular components from porcine liver tissue and preservation of

1742-7061/$ - see front matter 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actbio.2013.10.023

876

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

the key molecular components in the ECM. Others have used SDS and Triton X-100 perfused through the portal vein to remove cellular material from whole organs. Despite the recent explosion of interest, there is little consensus on the optimum method of decellularizing hepatic tissue such that the micro-architecture and protein content of the matrix are conserved as far as possible. Indeed, there is some concern that aggressive physical and chemical treatments may alter the protein structure, content and porous architecture of the matrices. In this context, the aim of this study was to generate well-characterized and reproducible off-the-shelf scaffolds of liver-derived ECM for studying hepatic tissue regeneration and for drug-related applications. Focusing on the technique of agitation and immersion, the present authors investigated the effects of different decellularization procedures on matrix composition and mechanostructural parameters, paying particular attention to the experimental design and initial testing conditions of samples in order to enable meaningful comparisons between different decellularization methods and controls. 2. Materials and methods

major families (i.e., detergent-free (DF), ionic (I) and non-ionic (NI)). First, a DF control family was provided to decouple the decellularizing effect due to mechanical agitation from that due to chemical agents (i.e., detergents). Triton X-100 and sodium dodecyl sulfate (SDS) were chosen as chemical decellularization detergents on the basis of their different modes of action and the reported effects on ECM [18]. Triton X-100 is a NI detergent that disrupts DNAprotein interactions as well as lipidlipid, lipidprotein and, to a lesser degree, proteinprotein interactions. SDS is an I detergent (i.e., anionic) that solubilizes cytoplasmic and nuclear cellular membranes, better at removing cell nuclei from dense tissues and organs than Triton X-100 is, but tends to denature proteins, disrupting the ultrastructure [19] and eliminating key growth factors [20]. Each protocol began with a rinsing day, to remove blood residues, and ended with a nal washing day, to remove detergent residues. Furthermore, the capability of NI Triton X-100 to form micelles with anionic SDS [21] was used in this work to enhance residual SDS removal from the I protocols. 2.3. Decellularization assessment

2.1. Hepatic tissue harvesting Four porcine livers were collected from 1-year-old healthy swine as a slaughter by-product. Pig liver consists of ve lobes (right lateral, right medial, left medial, left lateral and caudate lobe) wrapped in a tough brous capsule, i.e., Glissons capsule [17]. Individual lobes, except for the small caudate one, were sectioned from collected livers. After some fresh samples had been taken for mechanical tests (see Section 2.7), liver lobes were frozen at 20 C until use. Frozen livers were thawed at 4 C overnight, then punched with a tool to obtain 14-mm-diameter cylinders, which were subsequently cut in 3-mm-thick liver discs, avoiding macroscopic vasculature and Glissons capsule. Liver discs were frozen at 20 C until use. 2.2. Liver decellularization methods Several decellularization methods (Table 1), based on immersion and agitation as described in Supplementary material S1, were investigated, varying chemical detergent and treatment duration. The decellularization methods tested can be grouped into three Cell removal from liver discs obtained with each of the nine protocols was assessed by hematoxylin and eosin (H&E) staining and DNA quantication. Untreated frozen and thawed liver samples, henceforth termed fresh-frozen (FF), were used as controls. Decellularized liver ECM (dECM) discs were xed and parafn embedded. The dECM samples were then cut into 5-lm sections, stained with H&E and examined using an Olympus IX 81 microscope (Olympus Italia, Milan, IT). 2.4. Swelling tests Liver dECM samples obtained with all I and NI protocols were tested in triplicate (6 protocols 3 samples). Decellularized liver discs were freeze-dried at 50 C, 0.45 mbar for 48 h to determine their dry weight Wd. Then they were swollen in PBS 1X at room temperature and weighed every 12 h until a stable weight was obtained, i.e., the equilibrium swollen weight Weq. The equilibrium mass swelling ratio was calculated as Q eq W eq =W d . All measurements refer to the samples equilibrium swollen state, unless stated otherwise.

Table 1 Decellularization protocols used to obtain liver dECM; percentages refer to the weight/volume ratio (w/v) of detergent solutions in deionized H2O; only PBS 1 was used in the nal washing day for DF protocols, since the contribution of 0.1% w/v Triton X-100 to cell removal is very poor and not signicant compared with that of 1% w/v Triton X-100 or 0.1% w/v SDS used in the other protocols. Family Ionic Protocol I3 I4 I5 Non-ionic NI3 NI4 NI5 Detergent-free DF3 DF4 DF5 Not treated (control) FF FF-NS Day 1 Day 2 Day 3 Day 4 Day 5

PBS SDS 0.1% d Triton X-100 0.1% + d PBS 1 1 PBS SDS 0.1% SDS 0.1% 1 PBS SDS 0.1% SDS 0.1% 1 PBS Triton X-100 d Triton X-100 0.1% + d PBS 1 1% 1 PBS Triton X-100 Triton X-100 1% 1 1% PBS Triton X-100 Triton X-100 1% 1 1% PBS PBS 1 PBS 1 1 PBS PBS 1 PBS 1 1 PBS PBS 1 PBS 1 1 Fresh liver frozen at 20 C, then thawed at 4 C overnight Same as FF, but not equilibrium swollen prior to testing

d Triton X-100 0.1% + d PBS 1 SDS 0.1%

d Triton X-100 0.1% + d PBS 1

d Triton X-100 0.1% + d PBS 1 Triton X-100 1%

d Triton X-100 0.1% + d PBS 1

PBS 1 PBS 1 PBS 1

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

877

2.5. DNA content Total DNA content (micrograms per milligram equilibrium swollen sample) was determined for FF and all decellularized liver samples. DNA was extracted using the procedure outlined in Supplementary material S2. 2.6. Biochemical characterization Total protein content (TPC) was determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories Inc, Hercules, CA). FF and all decellularized liver samples were rst equilibrium swollen in PBS 1 to determine Weq, then solubilized in 1 M NaOH and subsequently diluted. To obtain an absolute and repeatable measurement of protein concentration, care was taken to obtain a reproducible initial matrix and a meaningful standard curve subject to the same solubilization process as the samples (see Supplementary material S3). Key liver ECM proteins (i.e., laminin, bronectin and collagen IV) were selectively analysed using Western blot (see Supplementary material S4 for experimental details). According to decellularization outcomes and TPC results (see Sections 3.1 and 3.4), Western blot analysis was performed in triplicate for protocols I3 and NI3 as representatives of I and NI decellularization families, respectively. The results were then compared with those obtained for protocol DF3 (from the DF family) and FF liver. Once again, to obtain meaningful comparisons between protocols, the same total protein quantity was loaded for each analysis. 2.7. Unconned compression experiments Samples were characterized mechanically in compression, using a twin column Zwick/Roell ProLine Z005 testing machine (Zwick/Roell, Ulm, Germany) equipped with a 10 N load cell. The testing setup and the method used to determine the elastic modulus are described in Supplementary material S5. The rst analysis was performed to assess whether the bulk compressive modulus of fresh liver depends on the sample harvesting site (i.e., intra-swine dependence) and whether it changes signicantly between animals (i.e., inter-swine dependence). Then, variations in the liver compressive modulus after a freezethaw cycle were investigated, collecting samples from FF liver lobes stored at 20 C and thawed at 4 C overnight. Finally, dECM from protocols I3 and NI3 were mechanically tested in compression as representative samples of I and NI decellularization families, respectively. The dECM samples obtained by decellularizing each of the four swine livers with the two aforementioned protocols were tested in triplicate (4 animals 2 decellularization protocols 3 samples). The resultant moduli were compared with that of untreated liver to assess whether and how different decellularization procedures affected liver bulk compressive properties. As the present results showed that protocol NI3 yielded the best liver dECM in terms of preserving ECM structure/ultrastructure and its constituents, micro-computed tomography (l-CT), histology and cytotoxicity tests were subsequently carried out on liver disc samples obtained using this NI protocol only. 2.8. l-CT and further histology

2.9. Sterilization of liver dECM and cytotoxicity tests Several sterilization procedures were investigated to obtain decellularized matrices usable as scaffolds for 3-D cell cultures or further processing to produce tissue derivatives. Protocol NI3 matrices were freeze dried and then treated by (i) exposing to chloroform vapour, (ii) H2O2 gas plasma, (iii) a combination of (i) and (ii), or (iv) PAA, before being placed in contact with hepatocytes in culture for cytotoxicity tests. Cell viability was evaluated using CellTiter Blue (Promega, Madison, WI). Details are given in Supplementary material S6. 2.10. Statistical analysis All experiments were carried out at least in triplicate. TPC, histology, mechanical tests and swelling measurements were performed on samples from at least two different pig livers. Comparisons between n groups of data (e.g., TPC, equilibrium mass swelling ratio) referring to one factor only (e.g., different decellularization protocols) were performed using one-way ANOVA followed by Tukeys Multiple Comparison Test. Two-way ANOVA followed by Tukeys Multiple Comparison Test was used to analyse the mechanical compressive results of fresh liver samples harvested from different lobes (1st factor of variability) of different pig livers (2nd factor). The same type of analysis was applied to the mechanical properties of liver dECM obtained by decellularizing samples from different pigs (1st factor) with two different protocols (2nd factor), and to cell viability assays after sterilization. Differences were considered statistically signicant at P < 0.05. 3. Results 3.1. Liver decellularized matrices Fig. 1 reports the histological analysis from the panel of protocols outlined in Table 1. The present authors did not detect any cells in either I or NI dECM at high magnication, but several cells were observed in all the DF protocols (Supplementary material S7). Notably, there is a gradual collapse in lobule structure with time, independent of the nature of the detergent. The use of SDS clearly promotes gradual removal of the intra-lobular matrix, and the removal occurs from the centre of the lobule outwards. 3.2. Swelling behaviour The duration of decellularization procedures (35 days) did not affect the equilibrium mass swelling ratios (Qeq), which were found to be dependent only on the decellularization family, as reported in Fig. 2. The I samples were characterized by a signicantly higher Qeq than the NI ones (respectively, 10.3 2.1 and 6.7 0.5, P = 0.018), reached in a shorter time, starting from the freeze-dried state (24 h compared with 36 h). This swelling behaviour can be explained by considering that NI liver dECM are characterized by a cell-free but ECM-rich intra-lobular network that hinders PBS absorption within the matrix, resulting in a lower Qeq, reached after a longer time interval with respect to I samples. The I matrices are characterized by a more open porous network composed principally of inter-lobular connective tissue, which allows for greater and more rapid penetration of water. However, the Qeq of FF and DF samples were similar (respectively, 1.9 0.4 and 2.3 0.4, P = 0.96), but signicantly lower than those of I (P = 7.19 105 for FF vs. I, and P = 1.03 104 for DF vs. I) and NI (P = 3.35 103 for FF vs. NI, and P = 5.70 103 for DF vs. NI) with equilibrium after 48 h. These values can be related to structural differences, as seen from histology, DNA and TPC analysis.

l-CT imaging was performed on freeze-dried dECM from protocol NI3 using a SkyScan 1174 system (Skyscan, Aartselaar, Belgium) with a resolution of 6.5 lm pixel1, 180 rotation. Histological staining was also performed on protocol NI3 dECM to identify connective components using silver, Mallorys trichrome and Alcian blue/PAS stains (10 lm slices were used, as thinner sections were lacking in contrast).

878

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

Fig. 1. H&E staining micrographs for all tested protocols as a function of time of immersion and agitation. There is a progressive shrinking of lobule dimensions with time and the intra-lobular matrix of the I family is clearly eroded. Scale bar 200 lm.

3.4. Evaluation of ECM constituents The FF and decellularized liver TPC are summarized in Fig. 3B. All TPC results were normalized to the equilibrium swollen weight of the sample, apart from FF-NS, i.e., FF not-swollen liver sample. FF-NS served to account for proteins lost during swelling. As expected, swelling samples to equilibrium results in a signicant decrease in TPC for FF liver, as a result of elimination of blood proteins and, in the present authors opinion, FF represents the control against which dECM should be compared. TPC was found to be 185 7 mg protein g1 wet sample for FF-NS, in agreement with published data [22]. One-way ANOVA analysis considering all the data showed that TPC did not change signicantly within I, NI and DF families, but there were signicant differences between groups, as reported in Fig. 3B. NI procedures were found to be less aggressive than I procedures in terms of protein retention within liver dECM. As conrmed by histology results, the DF family did not allow for complete cell removal, in fact the TPC values were found to be in-between those of the NI and FF liver. However, considering each family independently, the one-way ANOVA analysis showed a signicant decrease in TPC with time for both I and NI protocols, in agreement with the histological observations in Fig. 1. Since both I and NI protocols removed 97% of cells from day 3 onwards (Fig 3A), while clearly lobular shrinking with time (Fig. 1) was accompanied by protein erosion (Fig. 3B), only representative samples of NI, I and DF from day 3 were used for further tests. These three representative samples as well as FF untreated hepatic tissue were then analysed by Western blot in order to evaluate the presence of specic ECM proteins. Since the conservation of cell adhesion proteins and the basal lamina is crucial for the recellularization of 3-D scaffolds or materials derived from these matrices, the study focused on bronectin, laminin and collagen IV. As the TPC of control and fresh frozen livers is made up largely of cytoplasmic proteins, the ECM proteins are almost undetectable (Fig. 3C). However, both I and NI decellularization protocols

Fig. 2. Equilibrium swelling ratios (Qeq) for different sample families. The I samples are characterized by a signicantly higher Qeq than the NI ones, owing to the more porous network, which allows for a greater and more rapid penetration of water. However, FF and DF samples have a high cell density and a tighter network, leading to signicantly lower values of Qeq with respect to fully decellularized I and NI samples. Signicant differences in equilibrium swelling ratio between protocols are denoted by an asterisk (P < 0.05).

In particular, FF and DF samples have a high cell density and, therefore, fewer pores for water to penetrate with respect to the cellfree I and NI samples. 3.3. Total DNA analysis The total DNA in all samples is reported in Fig. 3A. Both the histology and swelling analyses are conrmed by these data. The DF protocols are clearly insufcient to remove cells, demonstrating the need to use chemical reagents along with forced mixing. Differences between the I and NI families were insignicant, suggesting that either reagent can be used for complete cell removal, which was reached after 3 days.

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

879

of protein degradation products) were observed, compared with the NI3 sample. Since all the DF protocols had traces of cells, further testing on DF samples was not carried out. 3.5. Compressive mechanical properties No signicant differences in the bulk compressive modulus (k) of fresh liver were found among liver lobes and different pigs. Moreover, k did not change signicantly between fresh and FF liver samples (1.62 0.13 kPa). Based on these results, liver disc samples harvested from all lateral and medial lobes were pooled together for all experiments. The resultant k of the decellularized samples was not dependent on pig, as expected. Moreover, no differences in k were observed between protocol I3 and NI3 liver dECM (respectively, 1.25 0.07 kPa and 1.31 0.09 kPa, P = 0.28). These values are signicantly different from that of untreated (i.e., fresh or FF) liver samples (P = 2.65 105), concluding that (i) cell removal was the main cause of the sharp decrease in k between untreated and decellularized liver samples, while (ii) k is independent of the decellularization protocol, indicating that it does not depend on the TPC of liver dECM. 3.6. Two-dimensional and 3-D architecture The 3-D architecture of dECM from protocol NI3 is depicted in the l-CT view in Fig. 4A (a video and slice are available in the Supplementary material les). NI3 has a dense matrix with an average (dry) pore size of 500 lm. Micrographs in Fig. 4BD show a reticular intra-lobular mesh rich with connective components, including mucosubstances. The brillar matrix in the micrograph in Fig. 4B has pores of 10 lm, similar to cellular dimensions (see Supplementary material S7). In comparison with the pink connective rich intra-lobular matrix in Fig. 4D, the cells in FF liver (Fig. 4E) stain bright red with Mallorys trichrome. 3.7. Liver dECM sterilization and cytotoxicity Fig. 5 shows the relative cytotoxic effects of the different sterilization methods investigated. Exposure to chloroform vapour and immersion in PAA were identied as the best methods of sterilizing liver dECM. The worst method was gas plasma treatment (with or without chloroform). The germicidal action of hydrogen peroxide plasma sterilization is based on the generation of free radicals, and it is likely that the low vitality observed was due to the action of free radicals on the matrix. 4. Discussion In order to achieve complete cell removal from small samples of hepatic tissue while preserving most of the ECM constituents and microscopic lobular architecture, it is necessary to establish quantiable parameters that enable different methods to be compared. It is also useful to identify the most crucial factors that can modify the outcome of the decellularization process. This study rst focused on establishing whether time plays a role in the decellularization process and which of the two most commonly used detergents is best for complete cell removal, through an assessment of histological features, swelling ratio and DNA and TPC. The histology and DNA results clearly show that a reagent for dislodging cells from the matrix is essential, as DF decellularization did not allow cell removal even after 5 days. Furthermore, there were signicant differences in TPC between families, while little change in DNA or TPC was observed over time for a given protocol family (Table 1 and Fig. 3A, B), indicating that time is not a crucial factor for cell removal. However, the decellularization processing

Fig. 3. Liver biochemical characterization before and after decellularization. (A) Total DNA content of FF and decellularized liver samples; data are expressed as lg DNA per mg of sample equilibrium swollen weight. (B) TPC of FF and decellularized liver samples; data are expressed as mg of protein per g of sample equilibrium swollen weight, except for the fresh tissue FF-NS, which was processed as it was. (C) Immunoblot analysis of ECM proteins; the same quantity of protein was loaded into each lane. Different letters indicate signicant differences between samples (P < 0.05), whereas the same letter means non-signicant differences.

increased the signal of all investigated ECM proteins. Although the observed signal increase appeared to be more pronounced in sample I3, more extensive banding patterns (suggesting the presence

880

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

Fig. 4. (AD) Three-dimensional structural and histological analysis of 3 day NI (NI3) liver dECM and (E) histology of control FF liver. (A) l-CT 3-D reconstruction of NI3 showing a dense matrix; (B) PAS/Alcian blue showing the mucin component of NI3; (C) silver stain highlighting the presence of a dense reticular component in NI3; and (D) Mallorys trichrome for NI3. The deep blue colour indicates connective tissue, while the pink stain shows elastic bres. (E) Mallorys trichrome of control FF liver: the deep red stains nuclei. Scale bar for histology images is 300 lm.

Fig. 5. Assessment of dECM sterilization methods through cell viability. Protocol NI3 dECM sterilized using four different methods (chloroform, chloroform + gas plasma, gas plasma, peracetic acid) and then placed in contact with HepG2 cells in culture for up to 48 h. Different letters indicate signicant differences in cell viability between sterilization methods (P < 0.05), whereas the same letter means non-signicant differences.

time should be kept to a minimum to avoid structural collapse, as demonstrated in Fig. 1. In most reports of tissue decellularization, ECM constitutents (e.g., proteins, glycosaminoglycans) are normalized to the dry weight of samples (e.g., Refs. [9,13,23,24]), and the evaluation of the aggressiveness of decellularization methods is somewhat arbitrary. In fact, matrix components expressed as milligrams per gram of dry tissue may be higher for decellularized tissue than for fresh tissue, making comparisons meaningless by privileging the more aggressive method, which removes more components unselectively. However, because the equilibrium swollen state is exceptionally stable and reproducible, highly reproducible quantitative data were obtained by normalizing all

measurements with respect to the equilibrium swollen state of samples, with modest standard deviations. Furthermore, the data show that the time to reach equilibrium and the equilibrium swelling ratio are simple and reliable indicators of the porosity and network density of dECM. After it had been established that the NI protocol is effective at cell removal and conserves more protein and matrix than the I protocol, and that 3 days are sufcient to ensure elimination of 97% DNA with both detergents, immunoblotting was performed to conrm the presence of adhesive and basal lamina proteins in selected samples. As the same quantity of protein was loaded into each lane, and as (i) no differences between the FF controls and DF3 and (ii) an increase in the band intensity for each of the investigated proteins (bronectin, laminin and collagen IV) were observed, the present results suggest that other proteins were arbitrarily removed by SDS. Moreover, the increase in band intensity was accompanied by an increase in band smearing. The latter is probably caused by the presence of degradation products, as suggested by Baptista et al. [9], hence the observed smear increasing from NI to I protocols may reect a higher degree of protein degradation as the treatment procedures become more aggressive. Matrix stiffness has been described as one of the most important biophysical parameters regulating cell function and differentiation. The current consensus is that the stiffness of a substrate should closely match that of the gross stiffness of biological tissue in order to elicit physiological cell responses. Attention was thus dedicated to the evaluation of the mechanical properties of liver dECM as a function of treatment protocol. As the mechanical properties of hepatic tissue are highly dependent on organ status and the test method employed [25], the present authors preferred to set-up an independent but self-consistent and reproducible platform for evaluating and comparing dECM and fresh tissue stiffness, starting from the equilibrium swollen state, as described in Supplementary material S5. To this end, a rst series of tests was performed to ascertain that the compressive modulus k is invariant for (i) different zones, (ii) different pigs and (iii) fresh and FF samples. The results are in agreement with those of Marchesseau et al. [25] and Tamura et al. [26]. The former highly cited paper lumps together data from 60 samples (5 animals 3 lobes 4 samples) to reduce the mechanical differences between lobes and animals: the present data conrm their a priori assumption. The latter performed a series of preconditioning compression tests on porcine liver specimens to determine the

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882

881

effects of freezing on the mechanical response of the tissue. They found that the hysteresis loops obtained after freezing samples matched the hysteresis loops of the same samples before freezing. From these results, they concluded that freezing did not signicantly change the mechanical properties of the tissue in compression. Interestingly, no difference in compressive modulus was observed between the I and NI protocols. However, in agreement with the trend recently reported by Evans et al. [27], there was a signicant lowering of k with respect to fresh and FF tissue, suggesting that the cells themselves also contribute to the stiffness of tissues. These studies relating the compressive properties of liver dECM to that of fresh tissue suggest that the design rules for ECM-mimicking scaffolds for recreating organs and tissues in vitro should be re-examined. The NI 3-day treatment protocol (NI3) was chosen as the optimum, and further characterization with l-CT and histological staining was performed to conrm the presence of additional matrix components and a well-preserved intra-lobular network. Assuming that the swelling ratio can be used to estimate the increase in pore size from the dry to equilibrium wet state, the dry pore dimensions observed in the l-CT images (500 lm) correspond to the average lobule dimensions in the H&E micrograph of NI3 in Fig. 3BD (1000 lm). The ner intra-lobular reticular network observed with silver staining (Figs. 3B and S7) was not resolved with l-CT. The sterilization of decellularized scaffolds or material derived thereof is a prerequisite for cell culture and eventual in vivo implantation. However, the sterilization method of choice should not degrade proteins nor alter the structure of the scaffold. Methods such as autoclaving, gamma irradiation, ethanol and ethylene oxide are known to induce changes in the protein chemistry and physical properties [28]. After assessing a variety of chemical and gas plasma-based sterilization techniques by measuring the viability of HepG2 cells over 48 h, it was concluded that chloroform vapour and PAA are not cytotoxic, with the method of choice falling on chloroform, being more benign to proteins [29,30]. In conclusion, a testing and analysis framework for identifying the best procedure for decellularizing hepatic tissue slices was established. The equilibrium swollen state and corresponding swelling ratio were shown to be useful indicators of the porosity of the scaffold and, more importantly, it was shown that using the equilibrium state as a reference enables reproducible quantitative comparisons between different decellularization protocols. Focusing on the conservation of mechanostructural parameters and protein content, the present authors singled out the NI protocol NI3 as the optimum. Finally, a signicant reduction in the stiffness of liver dECM was observed compared with intact hepatic tissue.

Appendix B. Figures with essential colour discrimination Certain gures in this article, particularly Figs. 1 and 4, are difcult to interpret in black and white. The full colour images can be found in the on-line version, at http://dx.doi.org/10.1016/ j.actbio.2013.10.023 References
[1] Mazzei D, Guzzardi MA, Giusti S, Ahluwalia A. A low shear stress modular bioreactor for connected cell culture under high ow rates. Biotechnol Bioeng 2010;106:12737. [2] Reilly GC, Engler AJ. Intrinsic extracellular matrix properties regulate stem cell differentiation. J Biomech 2010;43:5562. [3] Nelson CM, Bissell MJ. Modeling dynamic reciprocity: engineering threedimensional culture models of breast architecture, function, and neoplastic transformation. Semin Cancer Biol 2005;15:34252. [4] Serban MA, Prestwich GD. Modular extracellular matrices: solutions for the puzzle. Methods 2008;45:938. [5] Badylak SF, Taylor D, Uygun K. Whole-organ tissue engineering: decellularization and recellularization of three-dimensional matrix scaffolds. Annu Rev Biomed Eng 2011;13:2753. [6] Burch PT, Kaza AK, Lambert LM, Holubkov R, Shaddy RE, Hawkins JA. Clinical performance of decellularized cryopreserved valved allografts compared with standard allografts in the right ventricular outow tract. Ann Thorac Surg 2010;90:13015. [7] Wang Y, Cui C-B, Yamauchi M, Miguez P, Roach M, Malavarca R, et al. Lineage restriction of human hepatic stem cells to mature fates is made efcient by tissue-specic biomatrix scaffolds. Hepatology 2011;53:293305. [8] Uygun BE, Soto-Gutierrez A, Yagi H, Izamis M-L, Guzzardi MA, Shulman C, et al. Organ reengineering through development of a transplantable recellularized liver graft using decellularized liver matrix. Nat Med 2010;16:81420. [9] Baptista PM, Siddiqui MM, Lozier G, Rodriguez SR, Atala A, Soker S. The use of whole organ decellularization for the generation of a vascularized liver organoid. Hepatology 2011;53:60417. [10] Ross EA, Williams MJ, Hamazaki T, Terada N, Clapp WL, Adin C, et al. Embryonic stem cells proliferate and differentiate when seeded into kidney scaffolds. J Am Soc Nephrol 2009;20:233847. [11] Prasertsung I, Kanokpanont S, Bunaprasert T, Thanakit V, Damrongsakkul S. Development of acellular dermis from porcine skin using periodic pressurized technique. J Biomed Mater Res B 2008;85:2109. [12] Montoya CV, McFetridge PS. Preparation of ex vivo-based biomaterials using convective ow decellularization. Tissue Eng C 2009;15:191200. [13] Bolland F, Korossis S, Wilshaw S-P, Ingham E, Fisher J, Kearney JN, et al. Development and characterisation of a full-thickness acellular porcine bladder matrix for tissue engineering. Biomaterials 2007;28:106170. [14] Lang R, Stern MM, Smith L, Liu Y, Bharadwaj S, Liu G, et al. Three-dimensional culture of hepatocytes on porcine liver tissue-derived extracellular matrix. Biomaterials 2011:111. [15] Cebotari S, Tudorache I, Jaekel T, Hilker A, Dorfman S, Ternes W, et al. Detergent decellularization of heart valves for tissue engineering: toxicological effects of residual detergents on human endothelial cells. Artif. Organs 2010;34:20610. [16] Remlinger NT, Czajka CA, Juhas ME, Vorp DA, Stolz DB, Badylak SF, et al. Hydrated xenogeneic decellularized tracheal matrix as a scaffold for tracheal reconstruction. Biomaterials 2010;31:35206. [17] Kamimura K, Suda T, Xu W, Zhang G, Liu D. Image-guided, lobe-specic hydrodynamic gene delivery to swine liver. Mol Ther 2009;17:4919. [18] Gilbert TW, Sellaro TL, Badylak SF. Decellularization of tissues and organs. Biomaterials 2006;27:367583. [19] Ott HC, Clippinger B, Conrad C, Schuetz C, Pomerantseva I, Ikonomou L, et al. Regeneration and orthotopic transplantation of a bioarticial lung. Nat Med 2010;16:92733. [20] Reing JE, Brown BN, Daly KA, Freund JM, Gilbert TW, Hsiong SX, et al. The effects of processing methods upon mechanical and biologic properties of porcine dermal extracellular matrix scaffolds. Biomaterials 2010;31:862633. [21] Lee LT, Deas JE, Howe C. Removal of unbound sodium dodecyl sulfate (SDS) from proteins in solution by electrophoresis through triton X-100-agarose. J Immunol Methods 1978;19:6975. [22] Sohlenius-Sternbeck A-K. Determination of the hepatocellularity number for human, dog, rabbit, rat and mouse livers from protein concentration measurements. Toxicol In Vitro 2006;20:15826. [23] Vavken P, Joshi S, Murray MM. TRITON-X is most effective among three decellularization agents for ACL tissue engineering. J Orthop Res 2009;27: 16128. [24] Woods T, Gratzer PF. Effectiveness of three extraction techniques in the development of a decellularized bone-anterior cruciate ligament-bone graft. Biomaterials 2005;26:733949. [25] Marchesseau S, Heimann T, Chatelin S, Willinger R, Delingette H. Fast porous visco-hyperelastic soft tissue model for surgery simulation: application to liver surgery. Prog Biophys Mol Biol 2010;103:18596. [26] Tamura A, Omori K, Miki K, Lee JB, Yang KH, King AI. Mechanical characterization of porcine abdominal organs. Stapp Car Crash J 2002;46: 5569.

Acknowledgements The authors are grateful to Stefano Mazzoni and Andrea Pirone for the histology preparations and to the abattoir Desideri Luciano S.p.A. Via Abruzzi, 2 Pontedera (Pisa), Italy, for kindly supplying fresh hepatic tissue. The work leading to these results has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under grant agreement 304961 (ReLiver).

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.actbio.2013 .10.023.

882

G. Mattei et al. / Acta Biomaterialia 10 (2014) 875882 [29] Harrison MA, Rae IF. General techniques of cell culture. Cambridge: Cambridge University Press; 1997. [30] Schefer SU, Gonnermann J, Kamp J, Przybilla D, Pruss A. Remodeling of ACL allografts is inhibited by peracetic acid sterilization. Clin Orthop Relat Res 2008;466:18108.

[27] Evans DW, Moran EC, Baptista PM, Soker S, Sparks JL. Scale-dependent mechanical properties of native and decellularized liver tissue. Biomech Model Mechanobiol 2013;12:56980. [28] Wiegand C, Abel M, Ruth P, Wilhelms T, Schulze D, Norgauer J, et al. Effect of the sterilization method on the performance of collagen type I on chronic wound parameters in vitro. J Biomed Mater Res B 2009;90:7109.