Beyond Decellularization: Remnant Mitochondrial DNA Can Act as Hidden Damage-Associated Molecular Pattern

Abstract

Tissue decellularization aims to obtain bioscaffolds for regenerative applications by removing all cellular components while preserving the extracellular matrix (ECM) architecture. Although decellularization removes the majority of linear nuclear DNA (nDNA), residual amounts remain detectable. However, the fate of circular mitochondrial DNA (mtDNA) after decellularization has not yet been reported. Cell death or injury can cause the release of mtDNA, which is resistant to breakdown by exonucleases. Extracellular mtDNA acts as a damage-associated molecular pattern (DAMP) that can trigger immune responses. The aim of this study is to assess the presence of residual mtDNA in the liver, bile duct, and vascular scaffolds after decellularization and whether this causes inflammatory responses in macrophages. Decellularized tissues showed a marked reduction in total DNA content well below the threshold of 50 ng/mg tissue. However, in liver and vascular scaffolds, a relative increase in the mtDNA:nDNA ratio was detected in the remnant DNA fraction. Residual mtDNA in bioscaffolds acted as DAMPs causing macrophage activation, as shown by increased cell proliferation and cytokine production. Strategies to further reduce remnant mtDNA were tested. We found that treatment with the endonuclease enzyme HpaII was effective in degrading residual mtDNA. Importantly, mtDNA removal resulted in a significantly reduced macrophage activation. In conclusion, our study shows that mtDNA is relatively resistant to the decellularization procedure and can act as a DAMP in bioscaffolds. This underscores the importance of removing mtDNA from decellularized bioscaffolds to improve the immunocompatibility for biomedical applications.

1. Introduction

Tissue or whole organ decellularization intends to remove all cellular components, including DNA, while maintaining the structural and biochemical integrity of the extracellular matrix (ECM). However, the complete elimination of all cellular components remains challenging [1,2]. Generally, decellularization is considered successful when the ECM contains less than 50 ng of double-stranded DNA per milligram of dry tissue and lacks any visible nuclear material by microscopy [3,4,5]. However, so far, all published studies focus solely on total DNA content and do not distinguish between nuclear DNA (nDNA) and mitochondrial DNA (mtDNA). Each mitochondrion contains multiple copies of double-stranded, circular mitochondrial DNA (mtDNA) containing 37 genes. The liver is one of the most mitochondrially active organs, owing to its role in detoxification, gluconeogenesis, the urea cycle, and lipid metabolism, exhibiting higher mitochondrial copy numbers compared to other tissues [6]. Studies have shown that mtDNA copy numbers can vary by several orders of magnitude between cells and tissues. Hepatocytes, for instance, typically contain 800–2000 mitochondria per cell and have an mtDNA copy number of approximately 2100 per diploid nuclear genome, resulting in a relatively high mtDNA:nDNA ratio [6,7]. In contrast, endothelial cells lining veins and cholangiocytes lining bile ducts have lower energy demands and consumption, with fewer mitochondria and a lower mtDNA:nDNA ratio [8,9].

Unlike nDNA, mtDNA is circular, largely unmethylated, lacks histone packaging, and seems to have a shorter half-life in living cells [10]. Upon release from cells into serum or other biological fluids, circular DNAs, such as plasmids, extrachromosomal DNA, and mtDNA, show increased stability because exonucleases require free 3′ or 5′ ends for DNA degradation [11,12]. However, when mtDNA is linearized by the cleavage of targeted endonucleases, it is rapidly degraded from both 3′ and 5′ ends [13,14]. In biofluids, a substantial portion of cell-free mtDNA is packaged in extracellular vesicles or even in intact extracellular mitochondria, which physically shield the DNA from nuclease access and further increase its apparent stability in plasma and other fluids [15]. Whether decellularized scaffolds still contain intact extracellular mitochondria or mitochondrial elements is not clear. However, it has been hypothesized that matrix-bound vesicles, a subclass of extracellular vesicles that adhere to ECM proteins, remain present after decellularization [16]. These matrix-bound vesicles in scaffolds can either support regeneration or trigger an immune response [16,17,18,19,20,21,22]. Whether matrix-bound vesicles encapsulate mitochondria or mtDNA remains unknown.

The bacteria-like characteristics of mtDNA, being circular and small, lacking methyl groups and histone packaging, when released from cells evokes a pathogenic signal detected by Toll-like receptor 9 (TLR9) as a damage-associated molecular pattern (DAMP). The release of mtDNA has been shown to activate inflammatory responses [14,23,24,25,26,27]. Extracellular mtDNA can engage multiple pattern recognition receptors, triggering pro-inflammatory signaling in multiple cell lineages including macrophages [14]. Elevated mtDNA in circulation correlates with infectious and other diseases, and autoimmune disorders. Yang et al. showed that released mtDNA in mice with hepatocellular carcinoma promoted macrophage polarization and thereby decreased their sensitivity to treatment with sorafenib. This finding reveals that mtDNA functions not only as a marker of cellular damage but underlines its function as an active signaling molecule that modulates immune cell behavior [28].

The aim of this study is to quantify residual mtDNA in decellularized tissues and to identify strategies to further degrade and reduce mtDNA levels in decellularized scaffolds. In addition, we evaluated the macrophage inflammatory response triggered by remaining mtDNA. In the context of liver regenerative medicine, a functional vascular network and an intact biliary system are essential for nutrient delivery, waste removal, bile transport, and overall graft integration and survival. Therefore, the fraction of mtDNA was also determined in decellularized human bile ducts and human veins. Additionally, driven by the growing interest in using xenogenic materials in the field of tissue engineering, particularly porcine tissue due to its comparable size and architecture, residual mtDNA was also assessed in decellularized pig livers. This study underscores the importance of including residual mtDNA content as an important player in tissue engineering. The active removal of mtDNA should therefore be implied in decellularization procedures to ensure a safe scaffold for liver tissue engineering and regenerative medicine (TERM) [29].

2. Materials and Methods

2.1. Decellularization

Whole human livers (n = 6) and whole porcine livers (n = 4) were decellularized with an optimized protocol as described before [30]. Human livers with research consent were obtained after being declined for clinical liver transplantation. Porcine livers were obtained from the slaughterhouse (female German Landrace pigs, 90–110 kg). In short, livers were cannulated via the hepatic artery and portal vein, and connected to a peristaltic pump. The livers were pressure-controlled perfused with 20 L dH₂O, followed by continuous perfusion (10 L for porcine livers and 120 min for human livers) of 4% Triton X-100 (Carl Roth, Karlsruhe, Germany) + 1% NH₃ (Merck, Darmstadt, Germany) and cycles of reperfusion (five cycles for porcine livers and ten cycles for human livers) with 4% Triton X-100 + 1% NH₃ for 120 min per cycle. Next, the livers were continuously perfused with 50 L dH₂O to remove remaining detergent. The livers were stored in 10 L dH₂O for 7–14 days, with water refreshed every 2 days. Finally, the livers were perfused with 5 mg/L DNase type I (Merck, Darmstadt, Germany, 240 min for porcine livers and 480 min for human livers) to remove residual DNA and stored at −20 °C until further use.

Human bile ducts (n = 5) and human veins (n = 5) were decellularized as described previously by Willemse et al. [31] and Tejeda-Mora et al. [32], respectively. The cut ends of extrahepatic bile ducts were collected from donor livers during liver transplantation. Common iliac vein samples were obtained from deceased liver donors. In short, both tissues were rinsed with dH₂O and incubated in 4% Triton X-100 + 1% ammonia during constant shaking for ten cycles. The tissues were then rinsed with dH₂O and incubated in DNase type I for 4 h at 37 °C on an orbital shaker. The tissues were stored at −20 °C until further use.

The decellularization of all samples was confirmed by quantification of the DNA content (<50 ng/mg tissue) and the absence of cell nuclei after DAPI (4′,6-diamidino-2-phenylindole), and hematoxylin and eosin staining.

2.2. DNA Isolation and Quantitative PCR Analysis

DNA was isolated before and after decellularization using the QIAamp DNA Micro kit (Qiagen Benelux B.V., Venlo, the Netherlands) according to the manufacturer’s instructions, and the yield was measured in duplicate using a NanoDrop 2000 (ThermoFisher Scientific, Waltham, CA, USA). qPCR was performed in duplicate using the PowerTrack SYBR Masterix (ThermoFisher Scientific, Waltham, CA, USA), on a StepOnePlus system (Applied Biosystems, Carlsbad, CA, USA). Per sample, 25 ng DNA per reaction was used to determine the presence of nDNA and mtDNA. The details of the primer sets are listed in Table 1. SRY-box transcription factor 2 (SOX2) was selected as single-exon genomic reference gene, and ribosomal 45S Cluster 2 (RNR2) as a mitochondrial specific gene. The RNR2 gene (mt-16S rRNA) was selected because it serves as a specific marker for the methylation pattern of DAMP-related mitochondrial genes. Unlike protein-coding mtDNA regions, RNR2 is structurally, functionally, and biologically distinct, and is considered a dominant source of mitochondrial DAMPs through its activation of TLR9 and NLRP3. All primer sets were designed to be compatible with both human and porcine tissue. For each gene, two primer sets were included as technical replicate to ensure higher specificity. Primer efficiencies were assessed for all sets, ranging from 80.0% to 117.9% for SOX2 and 87.3% to 98.3% for RNR2, based on measurements in both human and porcine samples. The ratio of mtDNA and nDNA was determined by calculating average Ct values for all primers sets, and averaging the Ct values of both the mtDNA and nDNA sets. Next, the ratio was calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA). To calculate the absolute mtDNA copy number, we based our calculations on a single mtDNA molecule (16.6 kbp) with a mass of approximately 1.7 × 10⁻⁵ pg, corresponding to 1.7 × 10⁻⁸ ng per copy [33].

Table 1.

List of qPCR primers used to distinguish between nDNA and mtDNA.

Primer	Forward (5′ → 3′)	Reverse (5′ → 3′)	Amplicon (bp)
SOX2 set 1	CAGCCCATGCACCGCTACG	CCTGCTGCGAGTAGGACATGC	109
SOX2 set 2	TGAGCGCCCTGCAGTACAACT	GTGGGCGAGCCGTTCATGTAG	61
MT-RNR2 set 2	AGGGATAACAGCGCAATCCT	ACATCGAGGTCGTAAACCCT	63
MT-RNR2 set 3	CCTCGATGTTGGATCAGGACA	CGGTCTGAACTCAGATCACGTAG	95

The reduction in mtDNA signals after HpaII treatment was not due to the interference of the qPCR detection as none of the amplicon products contained a HpaII restriction site (Figure A3).

2.3. HpaII Incubation

HpaII is a restriction enzyme that recognizes the 5′-CCGG-3′ base pair sequence in DNA and cleaves leaving a 5′-CG overhang. HpaII only cuts unmethylated CCGG sites, which makes it useful for the digestion of residual unmethylated mtDNA. HpaII enzymatic treatment was performed on decellularized human liver fragments (n = 6). For this, HpaII (10 Units/μL, ThermoFisher Scientific, Waltham, CA, USA) was diluted in nuclease-free water and 10× Tango buffer containing BSA (ThermoFisher Scientific, Waltham, CA, USA) to an end concentration of 0.1 units per μg DNA, and was incubated overnight at 37 °C on a shaker. The samples were washed with PBS 1× (Lonza, Basel, Switzerland) containing 0.5 mM EDTA (ThermoFisher Scientific, Waltham, CA, USA) to inactivate the enzyme and washed three more times with PBS 1×. The untreated control samples were processed using the same buffer composition and washing steps as the HpaII-treated samples, with the enzyme replaced by distilled H₂O. This approach ensured that the ionic strength, BSA, and all other buffer components were identical between conditions, minimizing the potential confounding variables.

2.4. THP-1 Cell Culture

HpaII-treated (n = 6) and non-treated (n = 6) decellularized human liver fragments were transferred to a 48-wells plate. Per sample, 3 × 10⁵ human THP-1 monocytic leukemia-cells (ATCC, Virginia, USA) were added and cultured in RPMI 1640 (Biowest, Nuaillé, France) in a humidified CO₂ incubator at 37 °C. After 48 h, 30 ng/mL 12-O-Tetradecanoylphorbol-13-acetate (PMA, Merck, Darmstadt, Germany) was added to induce macrophage M0 differentiation. M0 macrophages cultured without decellularized human liver fragments served as the negative control, and wells containing cells treated with LPS (100 ng/mL, Merck, Darmstadt, Germany) served as the positive control.

2.5. THP-1 Adhesion Assay

To assess ECM-mediated activation, THP-1 cell adhesion was determined on non-treated liver ECM (n = 3) and HpaII-treated matched ECM samples (n = 3). Unbound THP-1 cells were removed by washing with PBS 1×. Adherent THP-1-cells were assessed by staining with DAPI (Vectashield antifade mounting medium with DAPI, Vector Laboratories Inc., Neward, CA, USA) and were visualized with confocal microscopy (Leica SP8 DLS Lightsheet microscope, Wetzlar, Germany) at 20× magnification. Per sample, nine representative regions of interest were imaged. The data was processed and the cell numbers were quantified by counting DAPI-stained nuclei using ImageJ version 1.54 (5 Dec 2024) (FIJI, U.S. National Institutes of Health). Images were first converted to binary data by applying a threshold. To separate adjacent nuclei, the Watershed function was applied. Nuclei were then identified and counted using the Analyze Particles tool, with circularity set between 0.50 and 1.00. The average number of nuclei per sample was calculated from the nine images and normalized to the number of cells per mm².

2.6. Immunofluorescence Staining

THP-1 cells, differentiated to the M0 state and subsequently cultured in the presence of HpaII-treated (n = 3) and non-treated (n = 3) ECM were fixed in 4% PFA for 15 min and stored in PBS 1× at 4 °C. The samples were permeabilized with 0.1% Triton X-100 in 1× PBS for 20 min and, next, non-specific epitopes were blocked using 5% BSA in 1× PBS for 60 min at RT. The primary antibody Ki-67 (Abcam Limited, Cambridge, UK, AB16667, diluted 1:100) was incubated overnight at 4 °C. The samples were washed five times in PBS 1× before incubation with the secondary antibody (Alexa Fluor 555, Thermo Fisher Scientific, Breda, The Netherlands, A32732, diluted 1:100) for 60 min at RT. Cytoskeletal staining with Phalloidin Alexa Fluor™ 488 (1:200, ThermoFisher Scientific, Waltham, CA, USA ) and nuclear staining with DAPI (Vectashield antifade mounting medium with DAPI, Vector laboratories Inc., Newark, CA, USA) were incubated for 30 min at RT. The stainings were imaged using an Invitrogen EVOS imaging system (ThermoFisher Scientific, Waltham, CA, USA) using 20× magnification, and the images were processed and the Ki-67⁺ cells were counted using ImageJ (version 1.54n) and the Analyze Particles tool.

2.7. Cytokine Production

The culture supernatant from M0 cells was collected and stored at −20 °C until their use in the enzyme-linked immunosorbent assay (ELISA) to measure tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and interleukin-1β (IL-1β) levels. ELISA was performed according to the manufacturer’s protocol (Thermo Fisher Scientific, Waltham, MA, USA), the samples were diluted if needed, and all the samples were measured in duplicate. The plates were measured on a Tecan Infinite plate reader (Tecan Group Ltd., MJannedorf, Switserland) at 450 nm. All the measured cytokine values were within the linear range of the assays.

2.8. Statistical Analysis

Data analysis was performed in GraphPad Prism 8.0. Paired samples were analyzed by a Wilcoxon paired t-test. The Mann–Whitney test was performed on other data sets.

3. Results

3.1. Persistence and Enrichment of mtDNA After Liver Decellularization

Six whole human livers were decellularized using pressure-controlled perfusion and mild detergents. Decellularization was confirmed by comparing the H&E staining of the native (Figure 1A) and decellularized tissue (Figure 1B). As shown in Figure 1C, the analysis of the DNA content revealed that the total DNA decreased from 1645 ± 244 ng/mg to 10 ± 4 ng/mg wet tissue weight during decellularization. Using qPCR, the relative level of nDNA- and mtDNA-specific sequences within the total DNA fraction was determined using two different primer sets for each. As shown in Figure 1D,E, a clear reduction in both nDNA and mtDNA was observed after decellularization. The ratio of mtDNA to nDNA was calculated using the average Ct values for both primer sets. A significant increase (p < 0.05) in the ratio of mitochondrial to nuclear DNA was observed (Figure 1F). The relative levels of nDNA and mtDNA were calculated per mg input tissue (Figure 1G). A nearly two-fold increase in the relative ratio of mtDNA to nDNA per mg tissue was determined (with ratios of ~800 and ~1488 before and after decellularization, respectively). This corresponds to ~9.14 × 10¹⁰ mtDNA copies per mg tissue at T = 0. For decellularized tissue, the mtDNA:nDNA ratio increased, but the total DNA content decreased to ~10 ng/mg tissue, resulting in ~5.52 × 10⁸ mtDNA copies per mg tissue.

Figure 1.

Persistence of mitochondrial DNA after decellularization of human and porcine liver tissue. Representative sample of H&E stained paraffin sections of human livers before (A) and after decellularization (B); scalebars: 400 µm. (C) The DNA content in ng per mg tissue before and after decellularization (n = 6). Relative level of nDNA (D) and mtDNA (E) before (T = 0) and after decellularization of human livers (n = 6), as quantified by qPCR using two independent primer sets for each. (F) Ratio of mtDNA:nDNA before (T = 0) and after decellularization for human livers (n = 6). Ratio between mtDNA and nDNA was calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA). (G) Calculated relative total DNA content per mg input tissue (T = 0) or scaffold (T = Decell) using 2^−Ct values found for nDNA and mtDNA. (H) Ratio of mtDNA:nDNA for pig livers before (T = 0) and after decellularization (n = 4). The ratio between mtDNA and nDNA was calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA) using both primer sets. (I) Relative total DNA content of pig livers per mg input tissue (T = 0) or scaffold (T = Decell), calculated using 2^−Ct values found for nDNA and mtDNA at T = 0 and T = Decell. Graphs present mean ± SD values, * p < 0.05, ** p < 0.01, and **** p < 0.0001.

As shown in Figure 1H,I, similar results were obtained for the decellularized porcine liver samples. For all four pig livers, the ratio increased after decellularization, supported with a significant decrease in decellularized tissues (Figure 1H, p < 0.05). Together these results indicate that, within the total residual DNA content, a significantly greater proportion of mtDNA remains present. The relative increase in mtDNA to nDNA suggests that during whole liver decellularization nDNA is degraded more easily and rapidly compared to mtDNA.

3.2. Differential Enrichment of mtDNA in Decellularized Vein and Duct Scaffolds

Both mtDNA and nDNA were assessed in human vein (n = 5) and bile duct (n = 5) samples, before and after decellularization (Figure 2A–D). As shown in Figure 2E,F, in bile ducts, no significant change in the relative levels of mtDNA and the mtDNA:nDNA ratio was detected after decellularization, and the relative levels of both nDNA and mtDNA remained stable (Figure A2D). This suggests that nDNA and mtDNA were removed to a similar extent during decellularization. In contrast, veins exhibited a significant increase in the mtDNA:nDNA ratio (p < 0.01, Figure 2B,C). Additionally, a significant rise in mtDNA relative levels was found following decellularization (p < 0.05, Figure A2E), indicating that relatively more mtDNA than nDNA remained in the decellularized veins. This indicates enrichment of mtDNA with respect to nDNA, showing the same trend as found in human and porcine liver samples. The ratio of mtDNA to nDNA in decellularized human liver, bile duct, vein, and porcine liver samples was plotted against the corresponding total DNA content ratio (calculated as DNA content at T = 0 divided by DNA content at T = Decell) (Figure 2G). The data show a distinct pattern for bile duct samples compared to liver and vein samples. While liver and vein samples cluster closely, showing relatively consistent mtDNA:nDNA ratios across varying DNA content ratios, bile duct samples exhibit a markedly different trend. This indicates that a lower DNA content ratio results in higher variability in the mtDNA:nDNA ratio, specifically valid for bile duct decellularization. While the exact reason for this phenomenon remains unclear, it might indicate that bile ducts undergo a different decellularization course, potentially reflecting tissue-specific structural characteristics.

Figure 2.

Persistent mtDNA in decellularized vein and bile ducts. The DNA content in ng per mg tissue before and after decellularization in veins (n = 5) (A) and bile ducts (n = 5) (D). Relative level of mtDNA before (T = 0) and after decellularization of human veins (n = 5) (B) and bile ducts (n = 5) (E). Ratio between mtDNA and nDNA was calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA) veins (C) and bile ducts (F). Graphs present mean ± SD values, * p < 0.05, and ** p < 0.01. (G) The ratio of mtDNA:nDNA at T = Decell expressed against the ratio of DNA content at T = 0 versus T = Decell for human livers (red, n = 6), bile ducts (green, n = 5), veins (blue, n = 5), and porcine livers (purple, n = 4).

3.3. Reduction in mtDNA After Linearization by Endonuclease Treatment

The hypothesis is that if mtDNA is linearized by the cleavage of targeted endonucleases, the linear double-stranded DNA becomes more susceptible to breakdown by exonucleases/DNA. Therefore, the treatment of decellularized liver grafts (n = 6) with the endonuclease enzyme HpaII was evaluated. The relative levels for nDNA (Figure 3A) remained unchanged after treatment with HpaII, whereas all mtDNA primer sets showed reduced (p < 0.05) relative levels after HpaII digestion (Figure 3B). The mtDNA:nDNA ratio, as calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA), decreased significantly (p < 0.05) after enzymatic treatment (Figure 3C). The significant reduction in the mtDNA:nDNA ratio after HpaII treatment indicate successful degradation and removal of mtDNA in decellularized human liver samples. The relative level of nDNA and mtDNA was calculated per mg input scaffold in untreated and HpaII-treated scaffolds (Figure 3D), showing a clear decrease in mtDNA levels following HpaII treatment, while nDNA levels remain stable.

Figure 3.

Treatment of decellularized human liver scaffolds with HpaII restriction enzyme. Relative quantity of nDNA (A) and mtDNA (B) of human decellularized liver scaffolds before (n = 6) and after targeted (HpaII) endonuclease treatment (n = 6). (C) Ratio between mtDNA and nDNA was calculated by 2^−Ct(mtDNA)/2^−Ct(nDNA). (D) Relative total DNA content of human scaffolds without (untreated) or with HpaII enzyme (HpaII-treated), calculated using 2^−Ct values found for nDNA and mtDNA. * p < 0.05 and ns = not significant.

3.4. Macrophages Are Activated by mtDNA in Decellularized Scaffolds

As a model for human tissue resident macrophage, the THP-1 cell line was used. The adhesion of macrophages to the human liver (n = 3) scaffold was determined before and after endonuclease treatment and quantified (Figure 4A,B). Compared to non-treated scaffolds, HpaII treatment resulted in a significant decrease (p < 0.05) in cell adherence (Figure 4C). Cell adherence showed the greatest decrease in human liver 3 sample after enzymatic treatment (from 106 ± 15 to 26 ± 7 cells/mm², respectively).

Figure 4.

HpaII treatment of human liver scaffolds reduced macrophage activation and proliferation. Representative images of DAPI-stained cell nuclei of THP-cells on untreated (A) and HpaII-treated (B) liver ECM, and number of adherent cells per mm². Scalebars: 100 µm. (C) Proliferation of M0 macrophages after 4 days of culture with decellularized scaffold, either untreated (n = 3) (D) or treated with HpaII (n = 3) (E). Representative images are shown. Cell nuclei (blue), Ki-67 (red), and F-actin (Phalloidin, green); scalebars: 200 µm. More ki-67 positive cells were detected in cultures with untreated scaffolds compared to HpaII-treated scaffolds. (F) Number of Ki-67⁺ cells per mm² plotted as min to max bars with line at mean. (G–I) Cytokine production by M0 macrophages after 4 days of culture with decellularized scaffolds. TNF-α, IL-1β, and IL-6 cytokine production was determined using ELISA. Untreated scaffolds (n = 6) induced production of all cytokines. The HpaII treatment of scaffolds (n = 6) significantly reduced TNF-α and IL-1β production. * p < 0.05 and ** p < 0.01.

THP-1 differentiated M0 macrophages were cultured with decellularized human liver scaffolds which were either untreated or treated with HpaII. Cell proliferation was determined via staining with Ki-67 after four days of culture. In cultures with untreated scaffolds, clear Ki-67 positive macrophages were observed, suggesting induced cell proliferation (Figure 4D). The number of Ki-67⁺ cells was significantly lower in macrophages cultured with HpaII-treated scaffolds (Figure 4E,F). To evaluate the pro-inflammatory response by M0 macrophages when cultured in the presence of decellularized liver grafts with (n = 6) and without (n = 6) enzymatic treatment, IL-6, IL-1β, and TNF-α production was determined by ELISA of the supernatant. A significant lower (p < 0.05) concentration of both IL-1β and TNF-α was found for conditions with untreated and HpaII-treated ECM (Figure 4G,H), indicating macrophage activation for the non-treated samples, and a lower reaction after enzymatic treatment. The same trend was visible in IL-6 concentrations (Figure 4I).

4. Discussion

mtDNA has emerged as an important mediator of innate immune activation. When released from stressed or damaged mitochondria, the cell-free mtDNA can act as a DAMP, activating innate immune receptors such as TLR9 and promoting local inflammatory responses [24,25,26,34,35,36,37,38]. The mechanical, oxidative, or enzymatic stress and subsequent cell death occurring during whole liver decellularization initiate the release of mtDNA. The concentration of residual mtDNA within a scaffold can potentially contribute to post-implantation inflammation or graft rejection, leading to complications for tissue engineering and transplantation studies. Therefore, mtDNA quantification should serve as an additional quality control, requiring the determination of a cut-off value to optimize decellularization protocols.

In the liver, hepatocytes contain a particularly high density of mitochondria due to the liver’s central role in energy metabolism and detoxification [24,25]. The remaining mtDNA in a decellularized liver bioscaffold represents a significant pathophysiological concern, promoting hepatic inflammation in clinical applications and potentially contributing to the progression of steatosis and fibrosis [25,26,39]. In our study, we observed that of the total DNA remaining after decellularization, a larger proportion is mtDNA compared to nDNA, suggesting that mtDNA is less effectively degraded during the decellularization process and retains better within the ECM. The DNase l step is included in decellularization protocols to remove residual nuclear material, but mtDNA appears to be partially protected. This protection is likely due to multiple factors, such as the retention of mtDNA within mitochondria remnants, which provide a physical barrier that limits the access of detergents and enzymes. The double membrane structure of mitochondria is more resistant to Triton X-100 compared to the nuclear membrane [40,41]. Additionally, the protection mtDNA could be associated with binding to nucleoid proteins such as TFAM, which has a role in mtDNA packaging [42,43]. It is also hypothesized that matrix-bound vesicles, a subclass of extracellular vesicles that adhere to ECM proteins, can encapsulate mitochondria and mtDNA, which may be released under stress and can either support regeneration or trigger an immune response [16,17,18,19,20,21].

The methylation-sensitive restriction enzyme HpaII recognizes the CCGG sequence and is efficient in cleaving unmethylated DNA [44]. Treatment with HpaII breaks mtDNA into smaller fragments that can be removed easier from the ECM and could lower the risk of immune responses. Since HpaII targets only specific DNA sequences and is too weak to degrade whole genomic DNA, it is therefore not a suitable substitution for the decellularization of whole livers. Treatment with HpaII is recommended as additional step for the complementary activity and optimization of current decellularization protocols. However, its clinical translation would require addressing several challenges, including enzyme penetration into large scaffolds, the scalability of the process, and complete removal of residual enzyme to meet biocompatibility and regulatory requirements [3,45]. These aspects will be critical for adapting this approach to clinical-grade ECM scaffolds.

mtDNA exhibits different immunoregulatory features from nDNA, which could account for the effects on adhesion, proliferation, and IL-6, IL-1β, and TNF-a production by (differentiated) THP-cells [27]. A significant reduction in IL-1β and TNF-α production following HpaII treatment of the ECM was observed. Our findings and IL-6 secretion data suggest the involvement of TLR9-related signaling; however, we did not directly assess pathway activation. IL-6 is a cytokine commonly associated with TLR9 activation, providing indirect evidence, but future studies incorporating pathway-specific readouts, such as NF-κB luciferase reporter assays or pharmacological inhibition, would be valuable to confirm pathway activation [46]. Nevertheless, overall, cytokine production remained relatively elevated with respect to the negative control (M0 cultured without ECM), suggesting that the ECM exerted stimulatory effects, potentially due to incomplete sterilization, residual endotoxins, or intrinsic biomaterial-related properties, underscoring the importance of creating a completely safe and clean bioscaffold [5]. A limitation of this study is the use of THP-1–derived macrophages, which, while widely employed as a model system, do not fully recapitulate the phenotype and signaling complexity of tissue-resident macrophages [47,48,49]. This is particularly relevant for ECM-mediated interactions, where primary cells may respond differently. Future studies should include validation with primary human macrophages and in vivo implantation models to confirm the translational relevance of our findings [47,50]. Interestingly, the degree of change in the mtDNA:nDNA ratio varied across the tissue types. Liver and vascular scaffolds exhibited significant mtDNA retention, whereas bile ducts showed no notable relative increase. These differences may reflect tissue-specific mitochondrial abundance and ECM composition. Endothelial cells lining the veins and cholangiocytes in the bile duct epithelium have lower energy demands and consumption in comparison to the liver, and thus have fewer mitochondria and a lower mtDNA to nDNA ratio [8,9]. In contrast, a significant difference in raw Ct value for mtDNA levels before and after the decellularization of veins was determined, suggesting that mtDNA is more easily removed from the vascular structures during decellularization in comparison to the bile duct. The vascular matrix is less tightly associated with structural components, making it more susceptible to washing out the remaining DNA. Importantly, substantial amounts of mtDNA remained detectable in both the decellularized vein and bile duct samples, and while the mtDNA:nDNA ratio remained unchanged in the bile ducts, these findings highlight that our decellularization procedure does not completely eliminate mitochondrial DNA. The remaining mtDNA in the decellularized vascular structures could potentially initiate inflammation in future in vitro and in vivo experiments [21,51,52]. Although this mechanism amplifies immune activation at the whole-organism level, the direct structural impact to individual tissues may be diminished compared to local high-concentration residues. The observed relative increase in the mtDNA:nDNA ratio in porcine liver scaffolds further emphasizes that this phenomenon is not species specific, raising concerns for xenogenic applications.

It should be foregrounded that the very low residual content of DNA in all decellularized bioscaffolds may allow its use in clinical setting, since the residual DNA content in bile ducts decreased 92.4% during decellularization (n = 5), and, for veins, the total DNA content decreased 98.7% (n = 5). To make a fair comparison between abundance of nDNA and mtDNA, 25 ng of total DNA was used for all qPCR experiments. Our study highlights that different tissues may vary in their retention of mitochondrial material during decellularization, which could have implications for assessing the effectiveness of decellularization protocols and the subsequent recellularization potential.

5. Conclusions

Residual mtDNA in decellularized scaffolds represents a hidden immunological risk that can compromise the success of tissue-engineered grafts. While conventional protocols effectively reduce total DNA content, mtDNA persists and acts as a potent DAMP, stimulating macrophage activation and cytokine release. Our study demonstrates that targeted enzymatic digestion using HpaII significantly reduces mtDNA levels and the associated immune responses, offering a practical approach to improve scaffold biocompatibility. These findings support the inclusion of mtDNA-specific assessment and removal strategies in decellularization protocols and regulatory frameworks. Addressing mtDNA persistence is essential to ensure the safety and efficacy of bioscaffolds for clinical applications in tissue engineering and regenerative medicine.

Abbreviations

ECM	Extracellular Matrix
EV	Extracellular vesicle
mtDNA	Mitochondrial DNA
nDNA	Nuclear DNA

Appendix A

Appendix A.1. mtDNA and nDNA in Decellularized Grafts

We found significant differences (p < 0.05) between the raw Ct values for nDNA and mtDNA before and after the decellularization in human liver samples (Figure A1). These differences indicate a lower level of nDNA with respect to mtDNA per sample, both before and after decellularization.

Figure A1.

Comparison of nDNA and mtDNA levels before and after decellularization of human livers. Raw Ct values for nDNA and mtDNA before (T = 0) and after (T = Decell) decellularization. * p < 0.05.

For pig livers, the total DNA content and relative level of nDNA and mtDNA before and after decellularization was determined (Figure A2A–C). A significant increase in the relative level of mtDNA in pig livers was found after decellularization. The relative level of nDNA before and after decellularization is shown for bile ducts (Figure A2D) and veins (Figure A2E). The relative levels found for nDNA in pig livers and bile ducts remained stable, while significantly decreased in veins.

Figure A2.

Total DNA content, relative level of nDNA and mtDNA for other decellularized grafts. The DNA content (ng DNA/mg tissue) for porcine livers (A) before and after decellularization. Relative level (2^−Ct × 10⁸) measured for nDNA (B) and mtDNA (C) before and after decellularization for porcine livers. Relative level (2^−Ct × 10⁸) measured for nDNA of human bile ducts (D) and veins (E). Graphs present mean ± SD, * p < 0.05, ** p < 0.01.

Appendix A.2. Mapping of HpaII Sites Across Human mtDNA

Figure A3 presents the revised Cambridge reference sequence of the human mitochondrial genome, deposited as NCBI Reference Sequence: NC_012920 [53,54]. Its ribosomal genes are marked in blue and the protein-coding genes in red. The arrows indicate the direction of transcription. HpaII’s restriction site positions are indicated in black.

Figure A3.

Mapping of HpaII sites across human mtDNA with genes and Hpall restriction sites.

Appendix A.3. THP-Cell Culture

For the blue-marked liver, macrophages formed cell clusters after 4 days of culture in the presence of the untreated ECM (Figure A4A). Additionally, multiple Ki-67⁺ cells were visualized. Cell clusters were also visible when cultured with the matched HpaII-treated ECM with several Ki-67⁺ cells (Figure A4B).

Differentiated macrophages were cultured with 100 ng/mL LPS to act as positive controls for the assays, and Ki-67 expression was assessed in these samples to determine cell proliferation (Figure A5).

Figure A4.

Staining of M0 macrophages after 4 days of culture with untreated treated ECM (A) and matched HpaII-treated ECM (B). Ki-67⁺ cells are stained in red, cell nuclei in blue and cell cytoskeleton in green. Scalebar: 200 µm.

Figure A5.

Staining of M0 macrophages positive control, after 4 days of culture with LPS. Ki-67⁺ (red) cells were visualized, cell nuclei are stained in blue and cell cytoskeleton in green. Scalebar: 200 µm.

Author Contributions

E.V.A.v.H., J.d.J., L.J.W.v.d.L. and M.M.A.V. designed the study. M.M.A.V., J.d.J. and L.J.W.v.d.L. obtained funding. J.W. procured and collected livers, ducts and veins. J.W. and E.V.A.v.H. performed decellularization of materials. H.P.R., E.V.A.v.H. and K.O.-V. isolated DNA and performed qPCR. E.V.A.v.H., K.O.-V., S.M.W.T. and K.L. performed and analyzed cell experiments. E.V.A.v.H. collected all data, performed analysis and drafted the figures. E.V.A.v.H. wrote the first draft of the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The use of human research livers was approved by the Erasmus MC Medical Ethics committee (MEC-2012-090, January, 2012). The use of human veins and human extrahepatic bile ducts was approved by the Erasmus MC Medical Ethics Committee (MEC-2014-060, April, 2014 ).

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author(s).

Conflicts of Interest

LJWvdL is scientific advisor for HealthBanks Biomedical Taiwan which is not related to the submitted work. All other authors have no conflicts of interest.

Funding Statement

This project was partly funded by TKI-LSH (Topconsortium Kennis en Innovatie-Life Sciences & Health) grant (RELOAD, EMC-LSH19002) and by the Convergence Health & Technology Flagship program “Organ Transplantation”, 2022.