The primary reagents and materials used in this study are listed as follows: acetonitrile (34851, Sigma-Aldrich); Alexa Fluor 488–labeled donkey anti-rabbit secondary antibody (A21206, ThermoFisher Scientific); Alexa Fluor 488–labeled donkey anti-mouse secondary antibody (A21202, ThermoFisher Scientific); ammonium bicarbonate (A643-500, Fisher Scientific); anti-collagen I primary antibody (1:100, 72026, CST); anti-collagen III primary antibody (1:100, ab7778, Abcam); anti-collagen VI primary antibody (1:200, ab182744, Abcam); anti-elastin primary antibody (1:50, MAB2503, Millipore); anti-GAPDH primary antibody (1:800, ab8245, Abcam); anti-LOXL1 primary antibody (1:50, sc-166632, Santa Cruz); anti-MAGP1 primary antibody (1:50, sc-166075, Santa Cruz); carbon tetrachloride (CCl₄, 10006418, Sinopharm Chemical Reagent Co., Ltd.); dithiothreitol (DTT, Prod 20290, ThermoFisher Scientific); donkey serum (SL050, Solarbio); the elastin van Gieson (EVG) staining kit (EVG-010, BASO); fast green dye (IF0020, Solarbio); formic acid (27001, Sigma-Aldrich); glutaraldehyde (P1126, Solarbio); HPLC-grade water (HP9021LT, Fisher Scientific); iodoacetamide (ICN10035105, Fisher Scientific); Lys-C (125-05061, Wako); NaCl (S9888-500G, Sigma-Aldrich); optimal cutting temperature (OCT) compound (SAKURA, Japan); PBS buffer (D8537, Sigma-Aldrich); the Pierce BCA protein assay kit (23227, ThermoFisher Scientific); PNGase F, peptide-N-glycosidase F (P0704L, BioLabs); protease inhibitor cocktail (20124ES03, Yeasen); the Residual SDS Assay Kit (C500055, YaJi Biological); SDS (BP166-500, Fisher Scientific); Sirius Red dye (S8060, Solarbio); the Tissue Genomic DNA Extraction Kit (DP304-02, TIANGEN); trifluoroacetic acid (A116-10X1AMP, Fisher Scientific); Tris-base (T8060, Solarbio); Triton X-100 (T8200, Solarbio); Trypsin (V511A, Promega); Tween 20 (T8220, Solarbio); and urea (U6504, Sigma-Aldrich). All solutions were prepared in HPLC-grade water or in dedicated buffers.

Wild-type C57BL/6J mice (10 weeks, male, purchased from Beijing HFK Bioscience Co., Ltd.) were euthanized by neck dislocation under anesthesia. The visceral adipose, brain, duodenum, heart, kidney, liver, lung, spleen, and stomach were surgically dissected and stored at −80°C until use. Small pieces of tissues were cut, weighted, placed into precooled tubes, and frozen in liquid nitrogen, and then stored at −80°C until use. The liver tissues from the control, liver fibrosis, and fibrosis resolution mouse models established in our previous study (Chen et al., 2019) were used for decellularization. Briefly, collagen α1(I)-GFP mice (male, 8 weeks old) were intraperitoneally injected with 12.5% carbon tetrachloride (CCl₄) in olive oil (1/7, v/v) at a dose of 0.01 mL/g body weight twice a week for 12 weeks (defined as “liver fibrosis”). Liver fibrosis mice subsequently underwent spontaneous recovery for another 12 weeks (defined as “fibrosis resolution”) after the cessation of CCl₄ intoxication. Control mice received an equal volume of olive oil for 12 weeks (defined as “control”). All mouse studies were approved by the Ethics Committee of Beijing Friendship Hospital, Capital Medical University, and carried out in accordance with the ARRIVE guidelines.

SDS decellularization workflow was modified based on Baiocchini et al. (2016) (Supplementary Material). We defined it as the enhanced SDS (E-SDS) method, which is summarized in Figure 1A and listed below.

• Needle puncture: small pieces of tissues (∼1 cm³) or whole organs were punctured (∼20 times/cm²) using the needle of 1 mL injector (0.5 mm × 16 mm, BD, United States). Note: a needle puncture should be executed on a precooled metal plate or culture dish placed on ice and completed as quickly as possible.

Proteolytic digestion was performed according to previous reports (Naba et al., 2015; Dussoyer et al., 2022). A brief workflow is shown in Figure 1B and described as follows.

• Reduction: decellularized dECMs (∼5–10 mg dry weight) were resuspended and reduced in an 8 M urea solution (50 μL) with 10 mM DTT upon continuous agitation (1,314 rcf) at 37°C for 2 h.

• Gross inspection: morphology and color of samples were recorded and photographed every 6 h during the decellularization process.

• 3D immunofluorescence: immunofluorescent staining of collagen I was performed in frozen 20-μm tissue sections. To specifically visualize collagen I, confocal z-stacks were captured at 0.3-μm increments between z-slices using a laser scanning confocal microscope (Olympus); z-slice contours were merged into a 3D contour surface by using the FV10-ASW 4.2 viewer software (Olympus) as previously described (Chen et al., 2019).

A nanoflow LC-MS/MS system was carried out using a quadrupole Orbitrap mass spectrometer (Orbitrap Eclipse, ThermoFisher Scientific) coupled online through an EASY-nLC 1,200 ultra-high pressure system (ThermoFisher Scientific) via a nano-electrospray ion source. Peptide samples (two replicates for each tissue) were loaded at 500 ng per injection on a 25 cm column (150-μm inner diameter, packed with ReproSil-Pur C18-AQ 1.9-µm silica beads; Beijing Qinglian Biotech Co., Ltd.), and were separated using a gradient from 6% to 12% solvent for 15 min, then 12%–30% solvent for 48 min, and stepped up to 40% solvent for 10 min, followed by a 10 min wash at 95% solvent at 300 nL/min (solvent: 80% acetonitrile and 0.1% formic acid). The total duration of the run was 85 min. An in-house-developed oven was used to keep the column temperature at 60°C. The MS instrument was operated in data-dependent acquisition mode, collecting MS spectra in the Orbitrap mass analyzer (resolution, 120,000; range, 350–2000 m/z) with an automatic gain control target of 4E⁵ and a maximum ion injection time of 50 ms. Following higher energy collisional dissociation with a normalized collision energy of 30%, MS/MS spectra were collected in the Orbitrap (15,000 resolution) with an AGC target of 5E⁴ and a maximum ion injection time of 22 ms.

The Proteome Discoverer suite (version 2.4, Thermo Fisher Scientific) was employed for raw data analysis. Tandem mass spectra were searched against the UniProtKB mouse proteome database, which contains both Swiss-Prot and TrEMBL mouse reference protein sequences (55,315 target sequences downloaded on 17 March 2022). The SEQUEST-HT search engine was used with the following parameters: fully tryptic specificity, maximum of two missed cleavages, minimum peptide length of six, fixed carbamidomethylation of cysteine residues (+57.02146 Da), variable modifications for oxidation of methionine residues (+15.99492 Da), precursor mass tolerance of 15 ppm and fragment mass tolerance of 0.02 Da for MS2 spectra collected in the Orbitrap. Peptide spectral matches and peptides with a false discovery rate (FDR) of less than 1% were filtered by using the percolator. After spectral assignment, the peptides were assembled into proteins and further filtered based on the combined probabilities of their constituent peptides to a final FDR of 1%. By default, the top matching protein or ‘master protein’ is the protein with the largest number of unique peptides and with the smallest value in percent peptide coverage. Only unique and razor (that is, parsimonious) peptides were considered for quantification. Protein expression matrices were deposited in the figshare database (DOI:10.6084/m9.figshare.22722343).

Proteins identified by LC-MS/MS were searched against mouse matrisome lists (http://matrisomeproject.mit.edu/) in silico defined (Shao et al., 2020) using gene IDs and followed by a final manual confirmation. The number of total proteins or matrisome proteins identified in each tissue by label-free LC-MS/MS was summarized. The matrisome number to total protein number ratio, matrisome abundance (sum of all the matrisome abundance) to total protein abundance (sum of all the protein abundance) ratio, number proportion of each matrisome category (ratio of the number of members in each matrisome category to the number of total matrisome members), and abundance proportion of each matrisome category (ratio of the total abundance of members in each matrisome category to total abundance of matrisome members) were calculated and compared. Venn analysis of matrisome proteins among nine mouse tissues was analyzed and visualized using the UpSetR (Conway et al., 2017) package. Internal interaction networks among common matrisome members in nine tissues were retrieved and visualized using the STRING database (https://cn.string-db.org/). The ClusterProfiler (Yu et al., 2012) package was used to identify the enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Heatmaps were drawn by using the pheatmap (https://CRAN.R-project.org/package=pheatmap) package. The MS/MS data from liver dECM scaffolds decellularized by E-SDS decellularization proposed by Baiocchini et al. (2016) were also compared.

Data are shown as mean ± standard deviation or percentage. The significance of the differences between two groups was determined by the unpaired Student’s t-test, the Chi-square trend test, or Fisher’s exact test. A p < 0.05 was considered statistically significant. The R 3.6.3, GraphPad Prism 7 (GraphPad Software Inc.), and SangerBox tool (http://sangerbox.com/) were used for statistical and bioinformatics analyses.

All mouse organs undergoing E-SDS decellularization became white (adipose, liver, lung, and stomach), off-white (heart, kidney, and spleen), or transparent (brain and duodenum) within a relatively short time period ranging from 6 to 48 h; with the exception of the brain, dECM scaffolds structurally kept the morphology and innate microscopic structure of native organs (Figure 2). Apart from gross inspection, Fast Green and Sirius Red staining were used to evaluate the extent of cell remnant elimination. As shown in Figure 2, compared to the native organs, all the decellularized organs lost the characteristic green color of the resident cells, and the remaining dECM scaffolds appeared pink, indicative of a successful decellularization with substantial ECM enrichment and rare cell debris reservation. Furthermore, elastin staining also confirmed an enrichment of elastic fibers after decellularization.

We next measured the nuclear materials in both native tissues and dECM scaffolds. In the first round of decellularization solution, the DNA content was highly detectable, whereas it was dramatically reduced in the second round solution in all the tissues, indicating decellularization mainly occurred in the first round of decellularization (24 h) (Figure 3A). A commercial tissue genomic DNA extraction kit was also applied to enrich DNA from native tissues and dECM scaffolds, but it revealed a significantly lower enrichment capacity of genomic DNA than when using the E-SDS method (Supplementary Figure S1). Taken together, the E-SDS pipeline has the potential to be used in mouse tissue or organ decellularization, achieving a relatively high efficiency in dECM scaffold purification.

Afterward, residual SDS content in the dECM scaffold was measured. As shown in Figure 3B, the dECM scaffolds before acetone precipitation contained higher levels of residual SDS, while SDS content was almost undetectable in dECM scaffolds after acetone wash, indicating that acetone has a strong capacity for precipitating SDS and that this step is indispensable to sufficiently reduce the cytotoxicity of E-SDS method–derived dECM scaffolds. Despite studies showing that the SDS detergent always elicited an altered microstructure and diminished ECM integrity (Xu et al., 2014), upon FG and SR and elastic staining (Figure 2), large empty spaces once occupied by cells could be clearly noticed (except in the brain dECM scaffold); collagenous and elastic fibers with different diameters were observed in all dECM scaffolds. Furthermore, a 3D reconstruction (Figure 3C) of specific immunostaining of collagen I clearly revealed collagenous fibers in the E-SDS-treated organs. We further performed SEM using the mouse liver dECM scaffold as an example. Ultrastructural analysis showed that the alignment and orientation of collagenous fibers were well preserved in the dECM scaffolds from control, fibrotic, and fibrosis-regressive mouse livers (Figure 3D). Given this, our E-SDS decellularization method is likely to have the potential to aid the accurate characterization of the structural information of collagen fibers in healthy or diseased tissue ECM, which may be of great interest to improve histological diagnosis.

Label-free proteomics analysis allowed us to unveil highly insoluble matrisome proteins in the dECM scaffolds from mouse tissues (figshare, DOI: 10.6084/m9.figshare.22722343). The matrisome number to total protein number ratios in nine dECM scaffolds ranged from 32.8% to 50.3% (Figure 4A). However, the abundance of matrisome proteins accounted for the most, as the matrisome abundance to total protein abundance ratios ranged from 78.5% in the brain to as high as 96.2% in the adipose (Figure 4A), suggesting that the E-SDS method can retain more extracellular insoluble proteins, with the potential to be used in MS profiling of insoluble matrisome proteins in wide-ranging tissues. Specifically, although the number of proteoglycans accounted for the most across all mouse tissues (Figure 4B), comparable with a previous report (Krasny et al., 2016), collagens appeared as the most abundant insoluble matrisome proteins in all dECM scaffolds (Figure 4B).

We next systematically compared the differences or similarities of insoluble matrisome proteins across nine mouse tissues. The common matrisome proteins in each tissue identified from two replicates were retained for subsequent analyses (Supplementary Figure S2). Most of the matrisome proteins were globally expressed across all the tissues, while a few of them exhibited a tissue specificity (Figure 5A). For instance, tubulointerstitial nephritis antigen (TINAG) and meprin A subunit alpha (MEP1A) were exclusively expressed in kidney dECM scaffold; laminin subunit gamma-2 (LAMC2), agrin (AGRN) and laminin, alpha 3 (LAMA3) were uniquely found in the lung dECM scaffold; tubulointerstitial nephritis antigen-like protein (TINAGL1) and annexin A7 (ANXA7) were only identified in the brain dECM scaffold. Except for tissue-specific matrisome members, a total of eight collagens, six ECM glycoproteins and one proteoglycan [heparan sulfate proteoglycan 2 (HSPG2)], were widely expressed across all tissues with variable abundance (Figure 5B). These common matrisome proteins highly interact with each other, probably constituting the core insoluble matrisome network in mouse tissues (Figure 5C). Functional enrichment analysis showed that these matrisome proteins were associated with fibrogenesis, immune, and proliferation–related KEGG pathways, covering the biological functions of the dECM scaffold (Figure 5D).

To assess the performance of the E-SDS method for precise insoluble matrisome coverage in tissue extracellular matrices, we compared dECM scaffolds using both E-SDS and Baiocchini’s SDS methods. Most of the identified matrisome proteins in the liver dECM scaffold using the E-SDS method were commonly identified by Baiocchini’s SDS method, and only nephronectin (NPNT) was exclusively detected by the E-SDS method (Figure 6A). The number proportion of matrisome members [42.8% (E-SDS) versus 5.5% (Baiocchini’s)] and abundance proportion of matrisome members [92.4% (E-SDS) versus 32.5% (Baiocchini’s)] were dramatically increased in the liver dECM scaffold using the E-SDS decellularization method (Figure 6B). Totally, the number or abundance of matrisome components from each category in the liver dECM scaffold using E-SDS and Baiocchini’s SDS methods was rather different (Figure 6C); in comparison, both the number and abundance of collagens and ECM glycoproteins accounted for most in E-SDS method–derived liver dECM scaffold (Figure 6C). Accordingly, it is likely that the E-SDS method could eliminate more ECM-affiliated proteins, ECM regulators, and secreted factors latent in the ECM scaffold than the Baiocchini’s SDS method. We further performed immunofluorescent staining to validate the results of LC-MS/MS proteomics. As shown in Figure 6D, it was nearly undetectable of intracellular glyceraldehyde-3-phosphate dehydrogenase (GAPDH, non-matrisome), indicating a rare cell debris reservation in liver dECM scaffold using both SDS methods; the secreted lysyl oxidase-like 1 (LOXL1, ECM regulators) and microfibrillar-associated protein 2 (MAGP1, ECM glycoproteins) were abundantly identified in the liver dECM scaffolds, derived from the Baiocchini’s SDS method but not the E-SDS method, which was comparable with the results of LC-MS/MS detection; the ECM structural proteins, such as collagen III, collagen VI, and elastin were highly enriched in liver dECM scaffolds using both SDS methods. Collectively, the E-SDS method coupled with proteomic profiling could identify and quantify the highly insoluble matrisome proteins in tissue extracellular matrices.