Comparative proteomic analysis of the composition of dECM and dECM-based inks compared to native tissues

Summary

Regenerative medicine and tissue engineering approaches based on decellularized extracellular matrix (dECM) present the advantage of a relatively biomolecule-rich matrix that directs cell function in a tissue-specific manner. To evaluate compositional changes during standard ink processing, six porcine tissues (artery, breast, dermis, epidermis, muscle, and nerve) were independently decellularized and formulated into biocompatible inks, tracking matrisome complexity via comparative liquid chromatography-tandem mass spectrometry (LC-MS/MS). Results revealed a core matrisome found overlapping in all decellularized tissues, alongside tissue-specific components correlating with predicted functional definitions. Although the proportion of collagens (mostly the α1 chains of collagen type I and III) increased in the final inks, a median of 55 matrisomal proteins was detected. This complexity is far superior to non-dECM-based inks in terms of mimicking native tissues. Our results support the use of dECM-based inks and biomaterials in mimicking native tissue ECM complexity, demonstrating tissue-specific composition, which can improve future therapeutic approximations.

Graphical abstract

Highlights

• •A conserved core matrisome overlaps across the analyzed native tissues
• •Processing selectively removes proteoglycans and secreted factors
• •Final bioinks are enriched in structural collagens
• •dECM inks preserve a complexity superior to non-dECM alternatives

Biological sciences; Proteomics; Biomaterials

Introduction

The extracellular matrix (ECM) is a dynamic network of macromolecules arranged according to cell and tissue specificity.¹ Tissue-resident cells secrete ECM in response to multiple inputs and create a microenvironment of crosslinked proteins and associated regulators, which is collectively known as the matrisome.^2,3 The matrisome provides an optimal spatial, biophysical, and biochemical microenvironment for cell function, which is dynamically regulated by tissue homeostasis.^4,5 Cells and the matrisome, thus, maintain a bidirectional crosstalk that modulates cell adhesion, migration, proliferation, and differentiation,^1,3,6 allowing the tissue to adapt to environmental needs.

The development of tissue-engineered constructs for regenerative medicine purposes is based on the use of biocompatible materials that aim to mimic the intricate architecture, mechanical, and biological properties of the original healthy tissue.^7,8,9 Further adding to the complexity from a materials engineering perspective, engineered constructs will sometimes have to adapt and resist chronically proinflammatory environments where engraftment is needed.¹⁰ However, due to their relatively low price and convenience of use, single-component scaffolds (such as gelatin, collagen, or alginate) that poorly mimic the complexity of native tissues are common alternatives to engineer grafts.³ Similarly, tumor-derived ECM such as Matrigel, which is a commercially available product, is often used as a base for hydrogel development. More recently, due to apparent issues with the specification of cells toward the desired phenotype(s), the focus is shifting to biomaterials based on tissue-specific decellularized extracellular matrix (dECM).^{11,12,13,14,15}

The dECM-based biomaterials are obtained by exposing tissues to chemical (e.g., detergents, hypo/hypertonic solutions), physical (e.g., temperature and mechanical force) and/or biological (e.g., enzymes) agents.¹⁶ The aim is to remove cells to prevent potential immunological reactions, while minimizing alterations to the biochemical composition, mechanical structure, and potential bioactivity of the tissues.¹⁷ Decellularized tissues are solubilized through protease and/or chemical treatments, yielding versatile materials suitable for creating foams, films, or suspensions. Of note, the digested soluble materials may be used as inks in 3D bioprinting applications.^18,19,20 However, this approach presents several shortcomings: (i) dECM-based inks usually present poor mechanical properties that impede their use in extrusion-based bioprinting,²¹ which may be circumvented by the use of alternative 3D printing strategies, such as volumetric bioprinting²²; (ii) the components of the ECM are susceptible to the effect of the decellularization agents they are exposed to, affecting their biological potential¹; (iii) relatively little is known on the compositional analysis of dECMs and dECM-based inks, which would be needed to develop a deeper understanding of ECM-derived product bioactivity^11,23; and (iv) the loss of specific ECM components that are required for proper bioactivity may imply the need of augmentation strategies to recapitulate bioactive, tissue-specific environments.²⁴ Occasionally, small organ dimensions may impede the extraction of significant ECM quantities, and thus alternative strategies must be sought.²⁵

Several studies have attempted to characterize ECM-derived biomaterial composition using proteomic approaches. For instance, a differential proteomic analysis of four porcine dECMs (liver, heart, skin, and cornea) revealed that each matrix presents a unique set of tissue-specific components and distinct compositional variation. This study further demonstrated that the signatures of both core matrisomes and matrisome-associated proteins differ significantly across decellularized tissue types.⁶ Similarly, researchers investigating the effects of demineralization and decellularization on bone ECM found that the matrisomal composition remains largely preserved following processing. However, they noted that the specific decellularization protocol employed can significantly influence the final protein profile, as different methods tend to enrich distinct ECM components.²⁶ More recently, the generation of a mass spectrometry-based atlas across 25 mouse organs revealed that while tissues share a common set of core matrisome proteins, their relative abundances can vary by up to 4,000-fold. This significant quantitative variation results in highly unique, tissue-specific matrix profiles that define the biological identity of each organ.²⁷

Despite these advances, several critical challenges remain. First, comprehensive comparative proteomic characterization across the entire processing workflow, from native tissue to dECM to ink, is lacking, limiting our understanding of how each processing step affects ECM composition.²⁸ Second, the high abundance of collagen in ECM samples hinders the detection of lower-abundance constituents that may be important regulators of cell function. Third, standardized protocols for dECM characterization do not exist, and different decellularization and digestion methods result in dramatically different protein retention profiles, making direct comparison between studies difficult.²⁹ Finally, the relationship between tissue-specific ECM composition and the functional properties of derived bioinks remains poorly understood, particularly regarding how processing-induced changes affect ink performance.^6,28

In this work, we conducted a comparative liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of the proteins present in the original tissue ECMs, dECMs, and digested dECM inks, aiming to evaluate the preservation of ECM components throughout the decellularization and solubilization processes. By systematically comparing these samples across six porcine tissue types, our study addresses the critical gap in understanding how each processing step affects matrisome composition. For that, the primary objective is to determine the extent to which essential ECM components are preserved during the native-to-ink pipeline. This involves identifying whether a universal core matrisome remains conserved across distinct porcine sources and assessing how specific processing steps affect the concentration of structural proteins. Furthermore, the research seeks to establish if the final inks retain enough tissue-specific protein signatures to accurately mimic the original organ’s microenvironment and effectively influence cellular behavior. Moreover, the study addresses for the first time the need to define a predictable protein profile that can serve as a standardized reference for future biomaterial development. Ultimately, we aim to uncover which key matrisome components are most critical for modulating bioactivity and interpreting cellular responses in engineered constructs.

Results

Decellularization of native porcine tissues and digestion to formulate inks for bioprinting

The following swine tissues were collected fresh, <2 h after sacrifice of 2-month-old Large White pigs, and cryopreserved: aortic artery, biceps femoris muscle, breast, sciatic nerve, and skin (the latter separated into dermal and epidermal layers). Based on tissue composition, distinct blends of biological and chemical decellularization agents were formulated for each tissue or tissue layer, and combined with physical actuators until satisfactory tissue decellularization standards³⁰ were achieved for each tissue (as detailed in STAR Methods; Figure S1; Table S1 and in previously published work).^31,32 Decellularized tissues were digested as specified in Table S2 to facilitate their use in ink formulation. Based on the inherent nature of each tissue, digestion protocols were chosen and designed to effectively remove cells depending on their composition and according to previous works. Biological triplicate samples of native tissue ECM, decellularized tissue ECM (pre-digestion), and decellularized tissue ECM (post-digestion) were then subjected to LC-MS/MS-based proteomic analysis (Figure 1).

Figure 1.

Workflow of the study

Six porcine tissues/tissue layers (artery, breast, dermis, epidermis, muscle, and nerve) were independently decellularized and digested, following tissue-optimized protocols. Triplicate samples of the resulting ECM at each step (ECM, dECM, and ink, respectively) were then analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) to obtain a proteomic atlas of the matrisomal proteins.

Decellularization and processing induce a distinct loss of ECM-component proteins

To investigate tissue-specific ECM signatures and how well they were conserved after tissue decellularization and sample processing for ink formulation, precursor and fragment tolerances were set at 20 ppm and 0.05 Da, and MaxQuant software searches were carried out against existing databases of pig proteins. Proteins identified in each sample with at least two peptides at false discovery rate (FDR) <1% were further analyzed (peak values are available at PRIDE, identifier number PXD059171). Intracellular and membrane proteins were discarded for further analysis, and ECM-specific proteins were classified by using the MatrisomeDB database,^33,34 which categorizes ECM proteins into two groups: the “core matrisome” consisting of collagens, proteoglycans, and glycoproteins, and the “matrisome-associated” components (affiliated proteins, regulators, and secreted factors), which include ECM-bound proteins and carbohydrates. Matrisomal proteins identified are listed in Zenodo (https://doi.org/10.5281/zenodo.14195914).

On average, matrisomal proteins constituted a mere 7.0% of the total proteins detected in native tissues (160.5 of 2,307.0; based on median values of all tissues (Figure 2A; Table S3). Analysis of dECMs allowed for the detection of 87.0% ± 24.2% of native proteins, which was perhaps unexpectedly high given the full decellularization obtained, as demonstrated by tissue histology and DNA content analyses (Figure S1). In contrast, the same analysis post-digestion (at the ink stage) only detected 23.8% ± 7.3% of the original tissue proteome (Table S3). Of note, the digestion step generates smaller peptides, which may hinder protein identification by LC-MS/MS.³⁵ The loss of component proteins with tissue decellularization was clearly tissue- and protocol-dependent (Figure 2B). For instance, while arteries, epidermis, and muscles retained the total number of identified proteins or even increased it (possibly due to the unmasking of proteins; see below), breast and nerve retained about 70.0% and the dermis retained only 43.6% of the total proteins (Figure 2B and Table S3). The stark difference found in dermal proteome preservation may be due to the use of trypsin at the decellularization step, which was required to eliminate fibroblasts due to the notably ECM-dense nature of the tissue.³⁶

Figure 2.

Representation of the contribution as well as conservation of the matrisome and its different categories for the different tissues and samples

(A) Quantification of the matrisome proportion within the whole proteome of artery (green), breast (blue), dermis (purple), epidermis (violet), muscle (brown), and nerve (orange).

(B) Categorization of the identified proteins by matrisome category, tissue, and step. Distribution of the proportions of MatrisomeDB categories is shown for the artery, breast, dermis, epidermis, muscle, and nerve. Top, middle, and bottom bars show the patterns for native tissue ECM (ECM), decellularized tissue ECM (dECM), and digested decellularized tissue ECM (Ink), respectively. MatrisomeDB categories are shown in decreasing color intensities.

(C) Venn diagrams of the identified proteins in each case, displaying the overlapping proteins and the proteins unmasked by tissue processing.

See also Figure S2 and Table S3.

As compared to total proteins, matrisomal proteins seemed to be significantly more resilient to the processing steps, retaining 96.4% ± 19.3% and 44.1% ± 15.0% of the native tissue matrisome at the dECM and ink stages, respectively. Decellularization of tissues did not significantly affect detection of matrisomal proteins, as would be expected due to their extracellular nature. In fact, detection of peptides was increased in several instances (Table S3). This could be explained by the fact that many abundant peptides, such as intracellular proteins in native tissue or collagen proteins, mask the intensity of other peptides within the spectrum. As the intensity of the highest peaks decreases, the smaller ones become more apparent.⁴ While every tissue was affected by the digestion step at the matrisome, the epidermis preserved the highest percentage of matrisomal proteins after digestion (72.2%), which might be due to the different digestion procedure used for this tissue as compared to the other five (see Table S2). Within detected matrisome proteins, the most abundant category was glycoproteins, as expected by the large glycoprotein number listed in the core matrisome, followed by the regulators and collagen families (Figure 2B).³⁷ As exceptions to this rule, the epidermis and breast tissue-derived ECMs, dECMs, and especially, inks showed a higher number of matrisomal regulators, followed by glycoproteins and affiliated proteins, which could possibly be explained by the use of tissue-specific protocols that better preserved these biomolecules.

In the light of these findings, Venn diagrams were used to illustrate matrisomal protein composition between ECM, dECM and ink samples for each tissue (Figure 2C). The vast majority of the proteins found in the dECM were also present in the ECM, with the exception of breast, dermis, and nerve, where 18, 46, and 24 proteins were found only in the native samples, respectively. Moreover, some dECMs also revealed uniquely identified proteins; specifically, 10 in epidermis, 6 in muscle, 5 in artery, 3 in breast, and 2 in dermis tissue. Likewise, most of the proteins present in the inks were shared among the three groups, although a few unique proteins were also identified.

Finally, it is important to note that dECM-based inks included a median of 53 matrisomal proteins (range 39–122; Table S3). This represents a staggering complexity, far superior in terms of mimicking the composition of native tissue to non-dECM-based inks.

Proteomic signatures of ECMs share a common core of matrisomal proteins

To further analyze the similarities and differences among matrisomal protein profiles detected in each tissue, the list of matrisome identifications was manually supervised to quantify the overlap of shared proteins among tissues (Figures 3 and S2). A Venn diagram shows the overlap of identified proteins found within the ECM of each tissue (Figure 3A). A total of 67 proteins were commonly shared among the samples, which in a Gene Ontology (GO) analysis appeared to be related to ECM organization, collagen fibril organization, extracellular structure organization, external encapsulating structure organization, and supramolecular fiber organization (Figure S3). These are functions commonly attributed to matrisomal core proteins. Notably, the arterial matrisome presented 32 unique proteins, while other tissues showed comparably lower numbers of unshared proteins. Overall, an overlap of 54.5% ± 11.2% of matrisome proteins was found among tissues, being lowest between artery and muscle (42.1%) and highest between breast and epidermis (87.1%) (Figure S2).

Figure 3.

Analysis of the similarities between native-ECM matrisomal protein profiles

(A) Venn diagram of the distribution of the shared proteins of artery (green), breast (blue), dermis (purple), epidermis (violet), muscle (brown), and nerve (orange).

(B) Partial Least-Squares Discriminant Analysis (PLS-DA) analysis of same tissues.

(C) PLS-DA-based variable influence/importance on projection (VIP)-score of the major contributors to the differences among tissues.

See also Figures S5 and S6.

The aforementioned results were based on the detection or absence of proteins, and they did not take into account the intensity of the peaks obtained by LC-MS/MS. To account for this variable, a multivariate dimensionality-reduction tool was used to convert multidimensional data to a lower-dimensional space. To this end, raw data were subjected to partial least-squares discriminant analysis (PLS-DA) (Figure 3B).^38,39 Together with this, in order to unveil the major proteins contributing to inter-tissue differences, a PLS-DA-based variable influence/importance on projection (VIP)-scoring analysis was conducted for every component explaining the covariance (Figure 3C).^40,41 The PLS-DA analysis demonstrated a clear discrimination of tissue-specific signatures, as observed in the distribution of groups based on the principal components. According to the axes defined by components 1 (PC1) and 2 (PC2), there was a clear distinction, on one hand, between epidermis and artery samples, and on the other hand, between these and the rest of the tissues. Based on the results of the VIP score analysis, it was shown that, for both PC1 and PC2, the proteins with the greatest contribution to the observed differences included filaggrin (FLG), anti-mullerian hormone (AMH), cathepsin H (CTSH), galectin 7 (LGALS7), secretory leukocyte peptidase inhibitor (SLPI) as well as numerous proteins of the S100 (S100A2, S100A6, S100A9, and S100A16), serpin (SERPINB2, SERPINB5, SERPINB7, SERPINB8, SERPINB10, and SERPINB13), and annexin families (ANXA1, ANXA2, ANXA8, and ANXA9). These proteins exhibited higher abundance in the epidermis group where they perform structural and functional roles in the epidermal barrier (FLG2, COL17A1, LAMC2, and LAMA3); differentiation and maintenance of keratinocytes (SERPINB13, ANXA8, and TGM5); stress response and skin repair (S100, LGALS7, ANXA2, SLPI, and ANXA1); as well as remodeling of the ECM (CTSH and CTSA). Conversely, fibromodulin (FMOD), prolargin (PRELP) and glypican-6 (GPC6) were overexpressed in the artery, contributing to the inter-tissue differences. In this case, both FMOD and PRELP have roles in ECM structuring. The first specifically regulates the organization of collagen type I and type III fibers, while the second promotes interaction between collagen and other ECM components, such as laminins and fibrillins. Additionally, GPC6 is a membrane proteoglycan that regulates cellular signaling by modulating pathways such as Wnt. The breast sample demonstrated an overexpression of different proteins grouped within the axis of component 3 (PC3), which enabled its discrimination from the other tissues. Among these proteins were filaggrin 2 (FLG2), S100 calcium binding protein A6 (S100A6), serpin family B member 13 (SERPINB13), and AMH. Finally, the axis defined by component 4 (PC4) primarily allowed for the distinction between dermis and muscle samples relative to nerve samples, due to an overexpression of proteins such as Von Willebrand Factor A domain containing 1 (VWA1), ficolin 2 (FCN2), nidogen 2 (NID2), S100 calcium binding protein B (S100B), α1 chain of collagen XXVIII (COL28A1), and ADAM metallopeptidase domain 10 (ADAM10) in the nerve. The proteins that contributed to these differences are involved in various functions such as ECM remodeling and cell adhesion (VWA1, NID2, COL28A1, FN1, MATN2, LAMA1, HMCN2, EMILIN3, and COL14A1), cell signaling and neural development (ADAM10, GPC4, PLXDC2, and WNT5B), damage response and neural protection (FCN2, S100B, SERPIND1, SERPINA3, KNG1, ITIH4, and CNTF), as well as the regulation of the blood-brain barrier and vascular integrity (LAMA1 and COL18A1).^42,43

Thus, PLS-DA analysis revealed that even if the proteins are shared among the ECMs of different tissues, the proportion and relevance of said proteins are different, allowing us to define separate groups, probably as a consequence of tissue-specific matrisomes. Overall, we found substantial overlap of core matrisomal proteins between native tissues, as well as the existence of a distinct proteomic signature for each tissue.

Composition of the core matrisomal proteomes of ECMs

With the intent of understanding the composition of the matrix in the native tissues, dECMs and inks, the profile of the matrisomal proteins (as defined by the different categories of MatrisomeDB) was plotted by averaging peak intensity (Figure 4). Overall, while all the native tissues showed a great diversity of proteins of the different matrisome categories, this diversity was reduced after the decellularization step, with proteoglycans and matrisome-associated proteins (affiliated, regulators, and factors) being particularly affected. This may be due to the fact that these types of proteins are hydrophilic and thus more susceptible to denaturation triggered by decellularizing agents and digestion procedures.^44,45 Unlike collagens, which usually compose the main structure of tissues, proteoglycans and matrisome-associated proteins play more dynamic roles, acting as intermediaries between cells and their environment.⁴⁶ In addition, the triple helix configuration of collagen and the carbohydrate chains present in the glycoproteins may have provided greater stability to these proteins, allowing a certain enzymatic resistance that contributed to their preservation.⁴⁷ Accordingly, the inks retained mainly collagens and glycoproteins, except for the epidermis.

Figure 4.

Proportions of the different matrisomal protein categories in each tissue at the ECM, dECM and ink processing stages

Distribution of the proportions of MatrisomeDB categories are shown in pie charts for the artery (green), breast (blue), dermis (purple), epidermis (violet), muscle (brown), and nerve (orange). Left, middle, and right columns show the patterns for native tissue ECM (ECM), decellularized tissue ECM (dECM), and digested decellularized tissue ECM (ink), respectively. MatrisomeDB categories are shown in decreasing color intensities as follows: Matrisome core: C (collagens), G (glycoproteins), P (proteoglycans) and matrisome-associated: A (affiliated), R (regulators), and F (factors).

As expected, there were some differences between tissues. Both arterial ECM and dECM showed a heterogeneous protein profile, with diverse presence of all matrisome categories. However, the digestion process induced a reduction of the associated proteins along with the proteoglycans, with a final composition based mainly on collagens (58.9%) and glycoproteins (32.7%). The breast tissue ECM showed an equilibrated matrisome, with high abundance of associated matrisome proteins (37.6%). The core was composed of 28.6% collagens, 22.7% glycoproteins, and 11.1% proteoglycans. The amount of core proteins augmented to 92.5% in the case of dECM, and 92.2% in the case of the ink. Interestingly, although most detected proteins were related to the core matrisome, 2.5% of regulators were present in the breast ink.

The main contribution to the dermal ECM was given by collagens (57.4%), followed by 13.8% of proteoglycans. Processing induced an increase in collagens that accounted for 95.8 % in dECM and 97.9% of the total intensity in the inks. Collagens were followed by the glycoprotein category as the second most abundant in both cases. Relatively small changes were observed in the digestion step (dECM to ink), possibly because the major proteome loss could be attributed to the trypsin treatment necessary to release the cells from the dermal ECM at the decellularization step. Due to the high abundance of collagen family proteins, a distribution of the different collagen types per tissue and sample type was calculated. Although collagen type I was predominant in the inks, 16 collagen isoforms were detected, with significant inter-tissue diversity (Figure S4).

An exception to collagen predominance was the epidermal ECM (11.2% collagens). A significant proportion of this ECM consisted of affiliated proteins (31.6%), regulators (19.8%), or secreted factors (30.7%), suggesting a particularly low presence of core matrisome proteins in this tissue. However, epidermal ink composition included 52.9% core proteins (49.8% of which corresponded to collagens) and 47.1% matrisome-associated proteins, respectively. Intriguingly, 36.0% of the epidermal ink was regulator proteins. Even if keratins play a significant role in the epidermal microenvironment, these were not considered for analysis in this study, as they are typically considered contaminants and removed in the pre-processing of the LC-MS/MS data of every tissue.

Native muscle tissue was mainly composed of collagens (51.0%), proteoglycans (14.5%), glycoproteins (13.4%), and regulators (12.5%). The heterogeneity and contribution of the different matrisome categories were maintained after decellularization, although the proportion of proteoglycans decreased slightly (8.6%). More notably, after the digestion process, only a small fraction of the original glycoproteins (2.3%) was retained, resulting in a composition mainly based on collagens, which accounted for 97.2% of the identified proteins.

Similar to breast tissue, the nerve sample revealed a balanced protein composition composed mainly of collagens (32.3%), regulators (24.2%), glycoproteins (18.8%) and proteoglycans (12.1%). The decellularization protocol had an impact on proteome diversity by reducing the presence of proteoglycans and matrisome-associated proteins, increasing collagens (57.5%) and glycoproteins (33.0%) instead. This phenomenon was more pronounced in nerve-derived inks, where proteoglycan groups, affiliated proteins, regulatory and secreted factors represented only 2.9% of the proteome composition.

Native tissue-specific proteomic signatures of ECMs partially align with GO terms

To further delineate tissue-specific proteomic signatures, a heatmap showing upregulated and downregulated clusters per tissue of the top 100 most abundant matrisomal proteins was construed for native tissue ECMs (Figure 5). Of interest, arteries, breasts, epidermis, muscle and nerves showed specifically upregulated protein clusters of 8–30 proteins that were consistently found in the biological triplicates. In contrast, the dermis showed a more generalized pattern of protein downregulation, indicating that its ECM composition was comparatively non-specific.

Figure 5.

Heatmap of matrisome proteins of each tissue ECM

The top 100 most abundant proteins are shown, clustered according to their contribution to the protein profiles of the different tissues and biological samples. In addition, dendrograms showing the clustering relationships both between samples and between proteins are plotted.

The genes encoding the proteins that were uniquely upregulated in each tissue were subjected to GO analysis. The proteins found clustering for the epidermal ECM signature corresponded to general ECM GO terms such as regulation of peptidase activity, but also included tissue-specific terms such as epidermis development. In the case of the arteries, in addition to common ECM-related GO terms, the identified proteins were related to tissue-specific processes such as regulation of plasminogen (PLG) activation, positive regulation of hemostasis and coagulation. The breast tissue ECM cluster was found to be related to more general terms such as negative regulation of blood coagulation and fibrinolysis. The proteins clustering for neural tissue were related to general ECM-related terms, as well as to the Wnt signaling pathway and notochord development. The ECM signature corresponding to the muscle was related to general ECM terms such as regulation of basement membrane organization or positive regulation of integrin-mediated signaling pathway, but terms related to muscle cell differentiation were also identified. Detailed results are presented in Table S4.

In summary, clusters with more complex compositions tended to be associated with more specific GO terms, while clusters characterized by general GO terms were often indicative of less well-defined clusters.

Proteomic signatures of ECMs at the differing processing steps scatter in PLS-DA analysis as a function of a limited number of up- and down-regulated proteins

To quantitatively capture the inter-group variation in protein composition among the processing steps, a cross-validated PLS-DA and VIP-scoring analysis were conducted using the normalized average label-free quantitation (LFQ) intensity proteomic data from the ECM, dECM, and ink processing stages of each tissue (Figure 6). The results of the cross-validation are represented in Figure S6.

Figure 6.

Partial least-squares discriminant analysis analysis of the normalized average label-free quantification intensity proteomic data from ECMs, dECMs, and inks

Distribution of samples are shown in decreasing color intensities for the ECMs (dark colors), dECMs (intermediate colors), and inks (light colors), for the following tissues: (A) artery (green) components 1 and 2, (B) artery components 3 and 4, (C) breast (blue), (D) dermis (purple) components 1 and 2, (E) dermis components 3 and 4, (F) epidermis (violet), (G) muscle (brown), and (H) nerve (orange).

See also Figure S6.

In all the studied tissues, the component 1 axis (PC1) distinguished between the ECM sample and the ink sample, while the component 2 axis (PC2) differentiated the decellularized matrix from the other two samples.

For the artery, the proteins contributing to the differences included the α1 and α3 chains of collagen VI (COL6A3 and COL6A1), the α1 chain of collagen XV (COL15A1), and transforming growth factor beta-induced (TGFBI), which were overexpressed in the native sample. In the dECM, there was an increase in the α6 chain of collagen IV (COL4A6), as well as proteins such as HtrA serine peptidase 3 (HTRA3) and procollagen-lysine 2-oxoglutarate 5-dioxygenase 2 (PLOD2). The final sample exhibited a higher abundance of the α1 and α2 chains of collagen I (COL1A1 and COL1A2) together with fibulin 5 (FBLN5). Validation revealed some dependency on replicates regarding components 3 and 4, making it hard to reach clear conclusions.

The VIP-scoring analysis demonstrated that in both breast and nerve tissues, all proteins encompassed in PC1 were overexpressed in the ECM, while those in PC2 were overexpressed in the dECM. In both cases, the ink sample exhibited downregulation of the identified proteins. In breast tissue, proteins overexpressed in the ECM included members of the annexin family, specifically A11, A6, A7, and A4 (ANXA11, ANXA6, ANXA7, and ANXA4), serpins F1 and G1 (SERPINF1 and SERPING1), and other proteins such as PLG and decorin (DCN). In the decellularized breast sample, there was an increased abundance of collagens, specifically the α1, α2, and α6 chains of collagen VI (COL6A1, COL6A2, and COL6A6), the α2 and α6 chains of collagen IV (COL4A2 and COL4A6), and the α1 chain of collagen XXI (COL21A1).

In the nerve ECM sample, overexpressed proteins included annexins A1 and A11 (ANXA1 and ANXA11), serpins G1, H1, and C1 (SERPING1, SERPINH1, and SERPINC1), and nidogens 1 and 2 (NID1 and NID2). In contrast, the dECM exhibited higher levels of fibrillins 2 and 3 (FBN2 and FBN3), the α3 chain of collagen VI (COL6A3), the α6 chain of collagen IV (COL4A6), and the α1 chains of collagens III, VIII, and XV (COL3A1, COL8A1, and COL15A1). Other overexpressed proteins in the dECM included laminin subunit α5 (LAMA5), versican (VCAN), and elastin (ELN).

Analysis of dermis samples revealed that the proteins contributing to variability along PC1 included the α1 and α2 chains of collagen I (COL1A1 and COL1A2), which were overexpressed in the ink sample. The remaining proteins within this component were more abundant in the native sample, including annexins A5 and A2 (ANXA5 and ANXA2), fibronectin 1 (FN1), FBLN5, the α1 chains of collagens VI and XIV (COL6A1 and COL14A1), the α6 chain of collagen VI (COL6A6), and several laminins (LAMC1, LAMB2, and LAMA4). Proteins in component 2 showed that the decellularized dermis sample had higher expression of dermatopontin (DPT) and the α1 chains of collagens (COL21A1 and COL5A1). Similar to the artery, components 3 and 4 depended on biological replicates, limiting conclusions.

In the epidermis, similar to breast and nerve samples, all PC1 proteins were overexpressed in the native sample. These included transglutaminase 2 (TGM2), inter-alpha-trypsin inhibitor heavy chain 2 (ITIH2), latent transforming growth factor β binding protein 2 (LTBP2), and families such as serpins (SERPINB5, SERPINF1, SERPINB10, and SERPINH1) and annexins (ANXA2 and ANXA8). Similarly, the dECM showed overexpression of most genes associated with PC2, including proteoglycan 3 (PRG3), ECM protein 1 (ECM1), thrombospondin 4 (THBS4), transglutaminase 1 (TGM1), and cystatin A (CSTA).

Lastly, in muscle tissue, the native sample showed overexpression of PC1 proteins, including the α1, α2, α3, and α6 chains of collagen VI (COL6A1, COL6A2, COL6A3 and COL6A6), as well as lumican (LUM), heparan sulfate proteoglycan 2 (HSPG2), transglutaminase 2 (TGM2), and osteoglycin (OGN). Conversely, the ink sample had higher levels of the α1 and α2 chains of collagen I (COL1A1 and COL1A2), the α2 chain of collagen IV (COL4A2), and the α1 chain of collagen III (COL3A1). The decellularized sample exhibited overexpression of most PC2 proteins, such as annexin 8 (ANXA8), the α1 chains of collagens IV and VII (COL4A1 and COL7A1), EGF-containing fibulin-like extracellular matrix protein 2 (EFEMP2), adipocyte enhancer-binding protein 1 (AEBP1), and microfibril associated protein 2 (MFAP2).

Some tissues are more heterogeneous than others with regard to the composition of abundant proteins

To delve further into the exact protein composition of ECMs, we depicted 98 of the proteins that displayed an abundance >1% in each tissue, aiming to ascertain if there were distinct patterns unique to each tissue type (Figure 7). The rest of the abundant proteins were grouped as “other”.

Figure 7.

Representation of the most abundant matrisomal proteins for each tissue and sample

Identified proteins with an abundance >1% for each tissue and sample are shown for (A) artery, (B) breast, (C) dermis, (D) epidermis, (E) muscle, and (F) nerve. Within each panel, the left, middle, and right pie charts show the patterns for native tissue ECM (ECM), decellularized tissue ECM (dECM) and digested decellularized tissue ECM (ink), respectively.

As expected, every tissue displayed a unique signature when plotting proteins by abundance. Arteries, breast, and nerve-derived ECMs showed a more heterogeneous composition. In contrast, the dermis, epidermis, and muscle ECMs were more restricted in abundant proteins. Native proportions were particularly well retained after the decellularization step of artery and epidermis. Remarkably, α1 chains of collagens type I (COL1A1) emerged as abundant in every tissue dECM, and became even more prominent in the ink state, along with α2 chains of collagens type 1 (COL1A2). As expected, elastin (ELN) appeared highly represented in the artery ink, corresponding to the elastic nature of the tissue. Aligned with previous results, the epidermal ink was most heterogeneous in composition. Together, tissue-specific signatures were partially retained after decellularization. Although relevant changes were observed in the abundance of ink-related matrisomal proteins as compared to native tissues, the final biomaterials represented complex mixtures with great potential impact on cell behavior.

Discussion

Due to their purported superiority in mimicking the biological and mechanical cues of the native tissues, dECM-based biomaterials are on the rise for the formulation of inks for 3D printing, as well as for direct use for regenerative medicine purposes.^11,12,13,48 However, tissue decellularization processes have the potential to significantly alter sample proteomic composition, necessitating a precise evaluation of the preservation of ECM components.⁴⁹ Notably, relatively little is known about how the different processing steps required for ink formulation affect the composition of dECM-derived inks and biomaterials.⁵⁰ In this work, by analyzing the proteomic signature of six decellularized tissues in the three main steps of the ink formulation process (ECM, dECM, and ink), we lay the foundations to better understand how ECM responds to decellularization- and digestion-associated processing, with regard to preservation of the original protein composition and ECM complexity.

The selected species was swine, for the following reasons: (i) porcine tissues are commonly available for most laboratories by sourcing them from research centers themselves and/or slaughterhouses, with limited ethical issues associated to their extraction due to their primary use for other purposes; (ii) porcine tissues share relevant anatomical and physiological similarities with human tissues, making them a relevant alternative for biomedical applications^51,52; (iii) these tissues offer the advantage of being a suitable source for the development of materials applicable to humans without eliciting adverse immunological responses.^53,54 Of interest, the apparent lack of immune response seems to occur despite the presence of abundant cellular proteins in the decellularized materials.⁵⁵ However, reports of immune response to porcine-derived dECM do exist, suggesting that elimination of some immunogenic epitopes such as α-galactosidase may suffice to avoid xenogenic rejection.⁵⁶ In any case and as a matter of fact, most biological implants currently in use in the clinic employing dECM-derived materials are porcine-based (such as AlloDerm, Permacol, and Strattice),^52,57,58 further demonstrating safety of human use. The selected tissues were based on easy access for tissue collection and their potential interest for purpose-built ink formulation, including major targets for tissue reconstruction efforts, such as skin, muscle, and breast; and inks of use for more general purposes such as vascular⁵⁹ and nerve⁵⁶ dECM-based inks.

The characterization of matrisome components in dECM products is challenging due to their complex protein composition profiles.⁶⁰ Traditional techniques such as immunohistochemistry, western blotting, or enzyme-linked immunosorbent assays have been applied to detect specific proteins or overall content description. However, those methods are limited to quantifying either targeted molecules or total amounts indiscriminately, and do not enable the comprehensive and high-throughput characterization of these complex materials.²³ In this regard, LC-MS/MS technology has proven effective in deciphering the molecular fingerprint of tissues as well as the composition of the matrisome, with numerous benefits, including high sensitivity, diverse sample analysis capabilities, and low detection limits.⁴

Regarding the ink composition, collagens and proteoglycans seemed to be the best-preserved proteins independently of the tissue, while the matrisome-associated groups were the most altered by ink production steps. In fact, it has been described that large macromolecules such as those comprising the core matrisome have increased processing resilience because of their relatively large size and the existence of intermolecular cross-links. Smaller constituents such as growth factors, chemokines, and other signaling molecules that are newly synthesized and not yet cross-linked are largely removed.^61,62

The PLS-DA and VIP score analyses evidenced the variation in the protein composition of the samples according to the processing stages. In accordance with the previous results, the ECM and dECM samples tended to share more similarities due to the better preservation of the protein profile, while the inks, as a result of the decrease in the number and abundance of proteins, tended to have a relatively distinct protein profile. In addition, the intrinsic variability of biological samples and the dynamic nature of ECMs may have increased this effect. Even so, there are cases in which the proteins are relatively more abundant in decellularized samples and in the derived inks, possibly due to a masking effect of the proteins. This effect has been described on numerous occasions, where the high abundance of collagen peptides hinders the detection of other less abundant but functionally important proteins.⁴

Additionally, fluctuations among the final materials may be shaped by factors such as the characteristics of the samples, inherent variations in the experimental procedures, and the LC-MS/MS method itself. In fact, sample digestion and extraction for MS have proven to be a challenging task due to the inherent complexity, heavy glycosylation and crosslinking, and general insolubility of ECM proteins, often leaving behind an insoluble pellet, which might result in incomplete characterization.^63,64 Moreover, most tissues are innervated and contain blood vessels, which are not totally removed in the decellularization process. Nevertheless, it is possible to mitigate this diversity by creating pools of dECM powder for individual tissues, all the while preserving the biological significance of the materials. Recently, Biehl et al. developed an automated, standardized decellularization protocol that was applied to produce ECM-based hydrogels from six independent porcine tissues.⁶⁵ Our approach fundamentally differs from theirs in that we pursued the best decellularization protocol for each tissue, thus optimizing ECM yield, but possibly at a price of increasing inter-tissue variability. Moreover, this study does not include the characterization of the final dECM-derived inks in terms of proteomic composition.

The development of biomimetic biomaterials entails an important step in basic and applied research, as well as for direct treatment applications based on tissue engineering. In this study, we presented six tissue-derived biomaterials with complex and heterogeneous biochemical compositions. Even if a proteomic loss occurred in the step from dECM to dECM-derived inks (which might be attributed to the digestion of the dECM powder), many proteins were still identified. Interestingly, tissue-specific proteins and signatures were highlighted for every tissue. Strikingly, the proportions of the matrisomal biomolecules were certainly unique. Indeed, the relation of the presence of biomolecules within such materials with their biological function, which is commonly ignored, was addressed in this study. In all the inks, at least 45 ECM-related proteins were identified, which are closely involved in biological processes such as cell migration, cell-ECM interactions, adhesion, and proliferation. This study represents a relevant resource for investigators interested in the development of novel tissue engineering-based solutions that better mimic the native tissue environment for future clinical applications.

Limitations of the study

While this study establishes a comprehensive proteomic baseline for dECM-based inks, several limitations should be acknowledged. First, the inherent insolubility of heavily crosslinked matrix proteins and the enzymatic digestion required for ink formulation caused a technical reduction in detectable proteins, particularly affecting hydrophilic matrisomal components like proteoglycans and regulatory factors. Second, tissue-specific protocol optimizations and the use of primary porcine sources introduced inherent biological and processing variability across batches. Finally, this work focuses strictly on compositional characterization; complementary long-term functional assays remain outside the scope of this study.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Ander Izeta (ander.izetapermisan@osakidetza.eus).

Materials availability

Materials generated in this study, including the tissue-specific dECM and dECM-derived inks, are available from the lead contact upon reasonable request. We may require a payment and/or a completed materials transfer agreement if there is potential for commercial application.

Data and code availability

• •The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD059171.
• •The datasets analyzed during the current study are available at Zenodo (https://doi.org/10.5281/zenodo.14195914).
• •All other data reported in the manuscript will be shared by the lead contact upon request.
• •The experimental workflow and metadata are available at protocols.io (https://doi.org/10.17504/protocols.io.8epv52zr5v1b/v1).
• •This study did not generate new or original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We thank Basatxerri for their support in sourcing fresh porcine tissue for the study. We express our gratitude to Prof. J. Ruiz Cabello and Dr. P.S. Valera for their valuable discussions regarding the analysis tools. This work was funded by Instituto de Salud Carlos III (ISCIII) and co-funded by the European Union (grant nos. PI25/00634, CERT22/00032, RD24/0014/0012, PT23/00142, and DTS24/00167), and Elkartek (bMG24; KK-2024/00041) and Hazitek grants (ITEAS; ZE-2022/00021) from the Department of Economic Development, Sustainability and Environment of the Basque Government. We also thank funding from the Department of Education of the Basque Government (IT1658-22). AIr was funded by a fellowship of the Predoctoral Training Program for Non-PhD Research Staff (PRE_2019_1_0031) of the Department of Education of the Basque Government and by Biogipuzkoa HRI. PV-A was supported by a fellowship paid by a donation made by Asociación Katxalin to Biogipuzkoa HRI in 2019, and an ERC Advanced Grant (ERC-2017-ADG-787510). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.

Author contributions

A. Irastorza and P.V.-A., conceptualization, methodology, software, validation, formal analysis, investigation, data curation, visualization, writing – original draft, writing – review and editing; L.Z.-O., data curation; P.G. and K.d.l.C., funding acquisition, resources, supervision, project administration, writing – review and editing; A. Izeta, conceptualization, funding acquisition, resources, supervision, project administration, writing – original draft, writing – review and editing. All authors read and approved the final manuscript.

Declaration of interests

All authors declare no financial or non-financial competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Porcine aortic artery	Biogipuzkoa HRI Animal Facility	N/A
Porcine breast	Biogipuzkoa HRI Animal Facility	N/A
Porcine skin (dermis and epidermis)	Biogipuzkoa HRI Animal Facility	N/A
Porcine biceps femoris	Biogipuzkoa HRI Animal Facility	N/A
Porcine sciatic nerve	Biogipuzkoa HRI Animal Facility	N/A
Decellularized aortic artery	This paper	N/A
Decellularized breast	González-Callejo et al.31	N/A
Decellularized dermis	Vázquez-Aristizabal et al.32	N/A
Decellularized epidermis	Vázquez-Aristizabal et al.32	N/A
Decellularized biceps femoris	This paper	N/A
Decellularized sciatic nerve	This paper	N/A
Artery dECM ink	This paper	N/A
Breast dECM ink	González-Callejo et al.31	N/A
Dermis dECM ink	Vázquez-Aristizabal et al.32	N/A
Epidermis dECM ink	Vázquez-Aristizabal et al.32	N/A
Muscle dECM ink	This paper	N/A
Nerve dECM ink	This paper	N/A
Chemicals, peptides, and recombinant proteins
Denarase	Kerry (formerly C-LEcta)	100 kU
Pepsin	Sigma-Aldrich	77160
Deposited data
UniProtKB database (includes Swiss-Prot and TrEMBL datasets for Sus scrofa)	Uniprot	https://www.uniprot.org
MatrisomeDB	Matrisome Project	http://matrisomedb.org
Mass spectrometry proteomics datasets	This paper; PRIDE/ProteomeXchange Consortium	PXD059171
Analyzed datasets and protein quantification matrices	This paper; Zenodo	https://doi.org/10.5281/zenodo.14195914
Experimental workflow and proteomic metadata	This paper; protocols.io	https://doi.org/10.17504/protocols.io.8epv52zr5v1b/v1
Software and algorithms
MaxQuant	Cox and Mann Lab	RRID:SCR_014485
BLAST	NCBI	RRID:SCR_004870
MetaboAnalyst 6.0	Xia Lab/McGill University	RRID:SCR_015519; https://www.metaboanalyst.ca
InteractiVenn	Heberle et al.44	https://www.interactivenn.net
Graphpad Prism 8.3.0	GraphPad Software	RRID:SCR_002798; https://www.graphpad.com

Experimental model and study participant details

Porcine tissue models and sample size

Porcine tissue collection was sourced from cadaveric animals according to Royal Decrees 118/2021 and 53/2013 and the three Rs principles, to ensure the protection of animals used in experiments and other scientific purposes. Tissue samples were collected postmortem from 2 months old Large White female pigs sacrificed either for surgical training of clinicians at the Biogipuzkoa Health Research Institute animal facility [after ethical review and approval by CEEA (23/11)] or for food consumption at the Basatxerri slaughterhouse. Use of postmortem samples in the latter setting does not need a specific ethical approval under Spanish law. All experiments conform to the relevant regulatory standards. Only female tissues were utilized in this study due to specific logistics and postmortem sample availability at the source facilities. While the exclusive use of female tissue represents a limitation regarding the generalization of the findings to male porcine backgrounds, sex-specific differences were not evaluated as the primary objective of this study. For baseline characterization of native tissues and final evaluation of dECM-based inks, a sample size of N = 3 biological replicates was utilized for each tissue type. For the intermediate dECM characterization, the experimental design included N = 3 independent biological replicates, with each biological sample analyzed in n = 3 technical replicates, yielding a total of 9 evaluations per tissue type.

Method details

Decellularization and formulation of dECM inks

Porcine samples were cleaned of surrounding tissue, and then sliced and stored at −80 °C until use. Porcine breast tissue was decellularized as described previously.³¹ Dermis and epidermis were isolated from porcine dorsal skin and decellularized, as detailed in earlier research.³²

For aortic artery, biceps femoris and sciatic nerve sample decellularization, tailored protocols were developed in accordance with the recommendations given by Crapo and collaborators,³⁰ as compiled in Table S1. Briefly, samples were freeze-thawed and exposed to a hypertonic solution based on sodium chloride, ethylenediaminetetraacetic acid (EDTA) and Tris overnight (O/N). Following, the three tissues were incubated in a detergent solution for 48 h. Specifically, artery and muscle samples were exposed to a sodium dodecyl sulfate (SDS) and sodium deoxycholate (SDC) solution, while nerve samples to a Tris/EDTA/SDS buffer. Then tissues were incubated in a nuclease (DNARASE, c-LEcta) treatment O/N and finally in a hypertonic solution based on Tris/NaCl for 4 h. A final step of sterilization by using a combination of ethanol (EtOH) and peracetic acid (PAA) for 2 h and a final 72-h washing step with phosphate buffered saline (PBS) were performed. Samples were then stored at −80 °C. Tissue decellularization was verified by haematoxylin-eosin staining, DAPI and DNA quantification as described in,³² results are shown in Figure S1.

For obtaining the dECM inks, samples were freeze-dried (Alpha 2–4 LSCplus, Christ) and pulverized in a Pulverisette 14 premium line mill (Fritsch). Each resulting powder was digested according to the tissue-specific optimized protocol as described in Table S2. The digests were then neutralized on ice to pH 7.4 and stored at 4 °C until use.

LC-MS/MS proteomic analysis

The proteomic analyses were performed by the Proteomics Platform of CIC bioGUNE (Derio, Spain). In total, three biological replicates of ECM, dECM and inks of each tissue were analyzed. Samples were incubated in a solution containing 7 M urea, 2 M thiourea, 4 % (w/v) 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) and 5 mM dithiothreitol (DTT) for 30 min at RT under agitation and digested following the filter-aided sample preparation (FASP) protocol,⁶⁶ with minor modifications. Trypsin was added to a trypsin:protein ratio of 1:50, and the mixture was incubated overnight at 37 ^°C, dried out in a RVC2 25 speedvac concentrator (Christ), and resuspended in 0.1 % formic acid (FA). Peptides were desalted and resuspended in 0.1 % FA using C18 stage tips (Millipore).

Samples were analyzed in a hybrid trapped ion mobility spectrometry – quadrupole time of flight mass spectrometer (timsTOF Pro with PASEF, Bruker Daltonics) coupled online to a nanoElute liquid chromatograph (Bruker). 200 ng of each sample were directly loaded in a 15 cm Bruker nanoelute FIFTEEN C18 analytical column (Bruker) and resolved at 400 nL/min with a 100 min gradient. Column was heated to 50 °C using an oven.

Protein identification and quantification was carried out using MaxQuant software using default settings. Searches were carried out against a database consisting of pig protein entries (Uniprot/Swissprot+TrEMBL/BLAST), with precursor and fragment tolerances of 20 ppm and 0.05 Da. Contaminants such as keratins and other proteins were discarded, and only proteins identified with ≥2 peptides at FDR < 1 % were considered for further analysis.

Quantification and statistical analysis

Data processing

The exact sample sizes and experimental replicates are detailed throughout the text. Briefly, a sample size of N = 3 independent biological replicates was utilized for each tissue type, with intermediate dECM characterizations analyzed in n = 3 technical replicates per biological sample.

Once the identifications were conducted, intracellular proteins were discarded, and only ECM-related proteins were plotted using GraphPad Prism 8.3.0 software and analyzed according to MatrisomeDB categories.³³ The mean and standard deviation (SD) for each sample (ECM, dECM and ink) of the Label-Free Quantification (LFQ) data were calculated and normalized to the total intensity value for each sample. Venn diagrams were done using Flaski 3.12.2⁶⁷ and InteractiVenn.⁴² Results related to gene ontologies (GO Biological Process 2023) were obtained with Enrichr.⁴³ For that, matrisome proteins of each sample were loaded and parameters were set in order to include the top 500 most relevant proteins encoding genes. The results were then sorted according to the adjusted p-value and the top 10 concepts were plotted in each case. PLS-DA, VIP scores and heatmap analysis were obtained with Metaboanalyst 6.0 software⁶⁸ by using one factor statistical analysis. Data was auto-scaled and analyzed without taking into account the order of groups. Cross validations were performed with the same software in order to estimate the predictive ability of the PLS-DA models, for that 5-fold CV method was used (Figure S6). In cross validation analysis, the accuracy parameter represents the proportion of correctly classified observations. R² indicates the explained variance, meaning the amount of variance in the data that the model can explain. Q² represents the predictive ability or power of the model, and is a measure of how well the model predicts the dependent variable during cross validation. All three parameters are represented in a range of values from 0 to 1, with values greater than 0.5 and close to 1 having the highest predictive power. VIP scores greater than 1 were used as a threshold to identify important variables. Biological interpretation of the results was conducted using GeneCards⁶⁹ and The Human Protein Atlas.⁷⁰ The workflow for the processing and analysis of the data is shown in Figure S7. No comparative inferential statistical hypothesis testing was performed in this study, as the primary objective focused on comprehensive proteomic profiling.