**2.2. Exploring the organization and the composition ECMs within tissues by histology**

Histology is an old, but still a powerful technique to image the organization of tissues at the scale of the cells. Organs or tissues have to be fixed to stabilize their organization after removing them from the living body and to be embedded in a hard material for being sectioned into thin slices to allow light to pass through the tissue and so be visible by the lenses of an optical microscope. Usually, tissues are fixed in paraformaldehyde and embedded in wax (paraffin). Tissues can also be frozen-fixed in liquid nitrogen. This approach may help to avoid some prejudicial effects of chemical fixations and do not require an embedding step to slice the tissue. However, frozen sections give pictures with less contrast and sharp edges, therefore formalinfixed paraffin embedded (FFPE) samples are preferred for routine examination. To be inter‐ pretable under light microscopic examination, the histological sections have to be stained to give some contrast to the different tissue and cells structures [15]. Numerous special stains have been developed to give a precise contrast to the organization and to the components of ECM, and this approach still keeps all its strength and efficiency to describe and accurately decorticate a tissue [16]. A significant further step in histology is the use of antibodies by the mean of immunohistochemistry (IHC) to specifically identify a target molecule, and at the same time, visualize its exact localization within the tissue [17]. However, IHC relies on the quality of the antibody, and it appears more difficult to obtain commercially reliable antibodies against ECM molecules than for intracellular molecules. A model of the accuracy and the potentials of histological techniques for investigation of ECMs are reported in a recent and thorough study on tooth root cementum using exclusively light microscopy techniques, with a combination of special stains, polarized light, and immunodetection [18].

#### *2.2.1. Special stains for histology of ECMs*

*Masson's and Gomori's trichrome*– they are ideal for connective tissues and in particular to stain fibrillar collagens. Trichrome staining indicates that three different dyes are used to discrim‐ inate the ECM materials from the cell cytoplasm and the nucleus. Selectivity of the dye for the different tissue structures is based mainly on size exclusion: a small dye will penetrate into low porosity elements while larger dye will penetrate preferably into the more porous structures. Gomori's trichrome stains collagen-based ECM in green, cell cytoplasm and noncollagenous tissue in pink to purple and cells nuclei in black to blue. Same discrimination of connective tissue comportments but with different colors is obtained with Masson's trichrome (collagen in blue, cytoplasm in red, and nucleus in black). This trichrome staining is longer and maybe more delicate to perform than the classical hematoxylin-eosin staining, however, this staining should be used each time when collagen-based samples (tissue or biomaterial) are investigated because of the high degree of contrast it enables compared to H&E [19].

*Von Kossa staining—* It is used to reveal mineral deposits within tissues. Mineralisation of ECM is observed in bones or teeth but can also appear pathologically in blood vessels, skin, or cartilage. Von Kossa staining detects calcium phosphate deposits by its substitution with silver nitrate to form black precipitates. This method is robust and useful to follow mineralization of bone substitute materials or progressive differentiation of naive cells into the osteoblastic lineage. It can also discriminate calcium–phosphate based materials from organic compounds.

play a role in ECM organization, but their important role is to be a reservoir for growth factors and to anchor signal molecules that are released through specific enzymes in particular after

**2.2. Exploring the organization and the composition ECMs within tissues by histology**

a combination of special stains, polarized light, and immunodetection [18].

*Masson's and Gomori's trichrome*– they are ideal for connective tissues and in particular to stain fibrillar collagens. Trichrome staining indicates that three different dyes are used to discrim‐ inate the ECM materials from the cell cytoplasm and the nucleus. Selectivity of the dye for the different tissue structures is based mainly on size exclusion: a small dye will penetrate into low porosity elements while larger dye will penetrate preferably into the more porous structures. Gomori's trichrome stains collagen-based ECM in green, cell cytoplasm and noncollagenous tissue in pink to purple and cells nuclei in black to blue. Same discrimination of connective tissue comportments but with different colors is obtained with Masson's trichrome (collagen in blue, cytoplasm in red, and nucleus in black). This trichrome staining is longer and maybe more delicate to perform than the classical hematoxylin-eosin staining, however, this staining should be used each time when collagen-based samples (tissue or biomaterial) are investigated because of the high degree of contrast it enables compared to

Histology is an old, but still a powerful technique to image the organization of tissues at the scale of the cells. Organs or tissues have to be fixed to stabilize their organization after removing them from the living body and to be embedded in a hard material for being sectioned into thin slices to allow light to pass through the tissue and so be visible by the lenses of an optical microscope. Usually, tissues are fixed in paraformaldehyde and embedded in wax (paraffin). Tissues can also be frozen-fixed in liquid nitrogen. This approach may help to avoid some prejudicial effects of chemical fixations and do not require an embedding step to slice the tissue. However, frozen sections give pictures with less contrast and sharp edges, therefore formalinfixed paraffin embedded (FFPE) samples are preferred for routine examination. To be inter‐ pretable under light microscopic examination, the histological sections have to be stained to give some contrast to the different tissue and cells structures [15]. Numerous special stains have been developed to give a precise contrast to the organization and to the components of ECM, and this approach still keeps all its strength and efficiency to describe and accurately decorticate a tissue [16]. A significant further step in histology is the use of antibodies by the mean of immunohistochemistry (IHC) to specifically identify a target molecule, and at the same time, visualize its exact localization within the tissue [17]. However, IHC relies on the quality of the antibody, and it appears more difficult to obtain commercially reliable antibodies against ECM molecules than for intracellular molecules. A model of the accuracy and the potentials of histological techniques for investigation of ECMs are reported in a recent and thorough study on tooth root cementum using exclusively light microscopy techniques, with

injuries and favor wound healing.

298 Composition and Function of the Extracellular Matrix in the Human Body

*2.2.1. Special stains for histology of ECMs*

H&E [19].

*Picro-sirius*—This staining method is specific of fibrillar collagens (type I–III), although it can slightly stain collagen type IV in BM. Sirius-red dye is an anionic molecule that arrange parallel to collagen molecules by interaction with basic (cationic) amino acids [20]. Collagenous tissue is stained in red while remaining tissue is pale yellow, and nuclei can be counterstained in black or deep blue if necessary. The parallel organization of picro-sirius is used to enhance the birefringence signal of collagen fibers observed under a polarized microscope. With the combination of polarized light and picro-sirius, it is possible to address the 3D orientation of the collagen fibers as the intensity of the birefringence depends on fibers orientation, with maximum birefringence intensity for fibers organized at 45°(+/–90°) toward the transmitted light axis, while no birefringence signal is observed with fibers organized parallel or orthog‐ onal to the transmission axis [21,22]. Picro-sirius is also used to make quantitative analysis of the collagen content in tissues, in particular to address the degrees of fibrosis.

*Verhoeff–Van Gieson*—This coloration method stains preferentially elastic fibers which have a higher affinity for the dye than the rest of the tissue due to hydrophobic interactions. The elastic fibers are stained black or dark brown while collagen fibers are red and the rest of the tissue appears pale yellow. This coloration is particularly appropriate to reveal atrophy of elastic tissues in case of vascular diseases and to discriminate arteries from veins due to differences in the elastic fibers organization and content [23].

*Alcian blue*—This dye is a cationic molecule and interacts specifically with negatively charged polysaccharides like sulfated and carboxylated GAGs. Because the selectivity of the dye is based on the charge of GAGs it is of importance to ascertain the pH of the alcian blue solution, as the carboxylic acid groups are in their acidic neutral form at pH < 2 and therefore will not interact with the dye. This specificity allows the discrimination between the carboxylated and the sulfated GAGs, as the latter are negatively charged at pH < 2. Polysaccharide's rich tissues are stained in light blue, counterstaining with nuclear fast red, making cells cytoplasm and nuclei colored in pink-red. Alcian blue is routinely used to stain cartilaginous ECM due to their large content of GAGs [24].

*Periodic Acid–Schiff* (PAS)—This staining is also sensitive to polysaccharides and is used to stain BM due to their high content in glycoproteins and proteoglycans. Unlike alcian blue, PAS staining reveals both charged and neutral polysaccharides because this coloration is based on the oxidation of monosaccharides with periodic acid and subsequent reaction of newly formed aldehyde groups with the Schiff reagent dye. BM are finely stained in pink-purple color, whereas the cytoplasm and the nuclei are colored in shades of blue when counterstained with Harris' hematoxylin. This staining method is also used to reveal other polysaccharides-rich compounds, such as mucins and glycogen [15].

#### *2.2.2. Immunohistochemistry of the extracellular matrix*

IHC enables the identification of a specific component of the ECM and to image its distribution within the tissue [25]. The target molecule is recognized by an antibody that reacts to a specific epitope and then is visualized by light microscopy through a chromogenic enzymatic reaction (alkaline phosphatase or horseradish peroxidase) or through a fluorescent dye with a fluores‐ cence microscope. The antibody is observed directly if the dye or the enzyme is linked to it, but most of the time it is detected indirectly by a labeled (by a fluorophore or an enzyme) secondary antibody which reacts to the first one through its Fc fragment. Frozen sections are more appropriate for Immunohistochemistry because they avoid the use of fixative that may alter the epitope, but frozen sections cannot be counterstained and so keeps the tissue organ‐ ization around the epitope not visible [26]. In contrast, FFPE samples are well preserved and can be counterstained with different dyes after antibody incubation and detection. However, if the fixative (generally 4% paraformaldehyde in neutral buffer) preserves the morphology of the tissue, it can severely compromise the antigenicity of the target molecule, and then make immunodetection inefficient or inoperative. Paraformaldehyde fixative triggers intra- and intermolecular cross-linking of proteins by the formation of methylene bridges between amino acids residues [25]. It may also alter the molecular structure of polysaccharides, lipids, and nucleic acids. The degree of cross-linking will depend on the concentration and the pH of the fixative solution, as well as on the time and the temperature at what the fixation is performed. The formation of intra- and intermolecular cross-linking modifies the secondary and tertiary structures of proteins that lower the detection by antibodies because of the modification of the target epitopes [27]. In the early 1990s, an antigen retrieval (AR) method was introduced to recover the antigenicity of FFPE tissue sections impaired by the fixation treatment [28]. The AR method originally refers to the high-temperature processing of FFPE sections, but with the development of other methods it is nowadays a generic term for any kind of treatment used to recover the original antigenicity of the FFPE sections [29]. The rationale of AR is the breaking of fixative-induced cross-links and methylene bridges that enable a renaturation of the proteins and a partial recovery of the epitopes. However, it has to be noticed that the true mechanism of AR is not yet understood, and it remains an empirical technique that requires several positive and negative controls to avoid true- or false-positive reactions [30]. AR is performed with the use of heat (called heat-induced antigen or epitope retrieval) or enzymes (referred-to PIER for proteolytic enzyme-induced epitope retrieval) to break fixative cross-links.

In heat-Induced epitope retrieval (HIER), three parameters appear essential in the outcome of the AR: temperature and pH of the solution and time of incubation of the sections [29,31]. Classically, sections are incubated for 10–20 minutes at 95°C in a water bath. Microwave and steam-cookers are also used to heat sections and have shown good AR properties, although the control of the temperature is more delicate. The pH of the solution is a critical factor because some epitopes will be revealed only in acidic or in alkaline buffer. The most common acidic buffer is citrate used in a pH range of 3–6. The most used alkaline buffer is Tris supplemented or not with EDTA at pH 8–10 [29,31]. All pH, temperature, and time have to be checked carefully because extreme conditions will damage the tissue sections.

Enzyme treatment is thought to break some of the fixative methylene bridges and to elicit the reconstitution of epitopes after a moderate digestion of proteins. It is generally performed with proteolytic enzymes such as pepsin, trypsin and proteinase K at a concentration of 0.05–1% for 10–30 minutes. Glycosidases, such as hyaluronidase, chondroitinase, and keratinase have shown valuable AR properties on polysaccharides-rich tissues and on glycosylated proteins [32]. The pH and temperature of the solution are adjusted to the optimal activity of the enzyme, and time of digestion and the concentration of the enzyme have to be carefully set to avoid overdigestion of the tissue sections which will lead to a loss of tissue structure and organization.

*2.2.2. Immunohistochemistry of the extracellular matrix*

300 Composition and Function of the Extracellular Matrix in the Human Body

IHC enables the identification of a specific component of the ECM and to image its distribution within the tissue [25]. The target molecule is recognized by an antibody that reacts to a specific epitope and then is visualized by light microscopy through a chromogenic enzymatic reaction (alkaline phosphatase or horseradish peroxidase) or through a fluorescent dye with a fluores‐ cence microscope. The antibody is observed directly if the dye or the enzyme is linked to it, but most of the time it is detected indirectly by a labeled (by a fluorophore or an enzyme) secondary antibody which reacts to the first one through its Fc fragment. Frozen sections are more appropriate for Immunohistochemistry because they avoid the use of fixative that may alter the epitope, but frozen sections cannot be counterstained and so keeps the tissue organ‐ ization around the epitope not visible [26]. In contrast, FFPE samples are well preserved and can be counterstained with different dyes after antibody incubation and detection. However, if the fixative (generally 4% paraformaldehyde in neutral buffer) preserves the morphology of the tissue, it can severely compromise the antigenicity of the target molecule, and then make immunodetection inefficient or inoperative. Paraformaldehyde fixative triggers intra- and intermolecular cross-linking of proteins by the formation of methylene bridges between amino acids residues [25]. It may also alter the molecular structure of polysaccharides, lipids, and nucleic acids. The degree of cross-linking will depend on the concentration and the pH of the fixative solution, as well as on the time and the temperature at what the fixation is performed. The formation of intra- and intermolecular cross-linking modifies the secondary and tertiary structures of proteins that lower the detection by antibodies because of the modification of the target epitopes [27]. In the early 1990s, an antigen retrieval (AR) method was introduced to recover the antigenicity of FFPE tissue sections impaired by the fixation treatment [28]. The AR method originally refers to the high-temperature processing of FFPE sections, but with the development of other methods it is nowadays a generic term for any kind of treatment used to recover the original antigenicity of the FFPE sections [29]. The rationale of AR is the breaking of fixative-induced cross-links and methylene bridges that enable a renaturation of the proteins and a partial recovery of the epitopes. However, it has to be noticed that the true mechanism of AR is not yet understood, and it remains an empirical technique that requires several positive and negative controls to avoid true- or false-positive reactions [30]. AR is performed with the use of heat (called heat-induced antigen or epitope retrieval) or enzymes (referred-to

PIER for proteolytic enzyme-induced epitope retrieval) to break fixative cross-links.

carefully because extreme conditions will damage the tissue sections.

In heat-Induced epitope retrieval (HIER), three parameters appear essential in the outcome of the AR: temperature and pH of the solution and time of incubation of the sections [29,31]. Classically, sections are incubated for 10–20 minutes at 95°C in a water bath. Microwave and steam-cookers are also used to heat sections and have shown good AR properties, although the control of the temperature is more delicate. The pH of the solution is a critical factor because some epitopes will be revealed only in acidic or in alkaline buffer. The most common acidic buffer is citrate used in a pH range of 3–6. The most used alkaline buffer is Tris supplemented or not with EDTA at pH 8–10 [29,31]. All pH, temperature, and time have to be checked Success of immunohistostaining mainly relies upon the quality of the antibody. Compared to soluble proteins, only few antibodies against ECM molecules are commercially available [32]. The ECM proteins are highly conserved in mammals making difficult the immunization of animals to generate an efficient antibody. Some antibodies are raised from synthetic peptides (5–20 amino acids) chosen from the primary amino-acid sequence of the target protein, but the epitope generated could be irrelevant to the secondary and tertiary structures of the native proteins [33]. Polysaccharides are either not or very slightly immunogenic, making very difficult to generate antibodies against the sugar part of proteoglycans. Moreover, ECM proteins are organized into dense fibers structures or meshwork or bear high polysaccharides chains that hinder the access of the antibody to the epitopes. In conclusion, IHC of the ECM is a delicate technique but remains the best option to obtain a picture of individual ECM components distribution within the different compartments of a tissue. As an example among several ones, by the means of monoclonal antibodies raised against laminin chains alpha-4 and -5, beta-1 and -2, and gamma-1, it has been possible to elucidate the particular composition and organization of the basement membrane surrounding islets of Langerhans in human pancreas [34]. The identification of a duplex BM surrounding intra-islets vessels with a specific laminin composition for each of the two BM has led to the proposition of a double-basement membrane model of human islets of Langerhans clearly distinct from the organization of basement membrane surrounding islets in mouse [35].

#### **2.3. Identification of ECM composition by proteomics with mass spectrometry**

The proteomic strategy is based on the isolation of a complex mixture of proteins from cells, tissues, or a whole organism and their identification by mass spectrometry and genomic database. Mass spectrometers commonly used for protein identification are MALDI–TOF (for Matrix–Assisted Lazer Desorption Ionization—Time of Flight) and ESI-Q-TOF (for Electro Spray Ionization—Quadripole—Time of Flight) that have their own characteristics and performances but do not change the general flow-chart of the sample preparation and identification [36,37]. After extraction and purification, proteins are separated by 1D or 2D sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE), respectively to their molecular weight (1D) or by both their isoelectric point (pH which net charge of protein is neutral) and molecular weight (2D). Mass spectrometers only detect charged molecules with an accuracy and sensibility that depends on the ratio of mass over charge (m/z), so the proteins samples have to be hydrolyzed into peptides before mass spectrometry analysis to obtain spectra at atomic resolution. Protein bands (1D) or spots (2D) are excised from the gel, hydrolysed into peptides by a proteolytic enzyme (frequently Trypsin), and loaded in the mass spectrometer to measure the exact mass of the peptides. Each protein from the original mixture is identified by matching the measured masses of their peptides with the expected masses of peptides calculated in silico from genomic database [37,38]. This technique allows a large-scale identification of components without the bias of predetermined molecular candidates as with antibody detection. It is thus possible in theory to have the exact protein composition of a tissue and follow its modification with time or diseases [39].

The total or relative amount of identified proteins can also be addressed. The SDS-PAGE migration pattern and intensity of protein band (1D) or spot (2D) give a "map" of the protein content of the target tissue or organ and can be used to identify particular band/spot that are modified in specific conditions, enabling discovery of new therapeutic targets [40]. Labeling methods exist to generate quantitative data with mass spectrometry. Samples of the control conditions are modified with nuclear isotopes 13C, 15N, or 18O, whereas the treated sample is left unmodified, and the relative abundance of both isotopic pics is compared [41]. A direct semiquantitative approach is also possible, with the mathematical integration of ion counts of the peptides identifying each protein to describe its relative abundance [42]. In the ECM analysis, relative quantification is a remarkable tool to identify the specific isoform of some multimeric ECM proteins, such as collagens and laminins, as the relative amount of each monomer will indicate under which form the ECM molecule is present in the tissue. To be more specific, collagen type V exists in the common heterotrimeric isoform [*α*1(V)]2*α*2(V) and a more scarce homotrimeric isoform [*α*1(V)]3. The relative amount of ion counts for the *α*1(V) chain over *α*2(V) chain will indicate if the *α*1(V) chain is associated only with *α*2(V) (*α*1 chain signal twofold of *α*2 chain signal) or if the investigated tissue contains both heterotrimeric [*α*1(V)]2*α*2(V) and homotrimeric [*α*1(V)]3 isoforms (*α*1 chain signal >> twofold of α2 chain signal). However, quantification by mass spectrometry can be restricted by the ionization properties of some proteins that will make them less detected and consequently underrepresented in the final analysis. Nevertheless, this highlights the potentials of proteomics and mass spectrometry in the study of ECM proteins, as such characterization of ECM proteins isoform will require several antibodies (i.e., one per protein chain) to identify one isoform by western blot or IHC [43].

The most critical steps of a proteomic analysis are the purification of the protein mixture and their identification from database. ECM proteins have a high molecular weight and are tightly associated with each other by covalent cross links that make them mostly insoluble. An important point in the analysis of ECM by mass spectrometry proteomic will be the proper solubilization of the ECM [44]. The tissue has to be first carefully decellularized to purify the ECM and eliminate the remaining intracellular proteins. This step requires the use of a detergent like SDS and will eliminate from the ECM part of the loosely bounded proteins like remodeling enzymes or growth factors [40]. The purified ECM can be solubilized by a combination of physical, chemical, and enzymatic methods. A physical method is the me‐ chanical breaking with a French press or grinding with mortar and pestle in liquid nitrogen. This step is important to homogenize correctly the purified ECM and make the following solubilization treatment effective. Ultrasound can also be used, but this process yields heat that can denaturate and break the proteins creating smears instead of protein bands or spots during SDS-PAGE separation if temperature is not carefully controlled. Homogenized ECM can be solubilized with a chaotropic agent like concentrated urea or guanidium chloride [45]. These molecules are efficient for solubilization, but a too high concentration is not compatible with SDS-PAGE separation and can impair the trypsin digestion. Highly cross-linked collagen fibrils or elastin microfibrils can remain insoluble after chaotropic extraction. Partial digestion with proteolytic enzymes such as pepsin is also used to favor ECM solubilization, but again, it has to be done carefully to not hydrolyze the ECM sample before SDS-PAGE separation. Deglycosylation with glucosidase such as PNGase or chondroitinase can unravel parts of the dense polysaccharide network of proteoglycan and unleash trapped ECM proteins [45]. Moreover, deglycosylation is also favorable for further trypsin digestion and peptide identi‐ fication from database. ECM proteins undergo several posttranslational modifications, such as hydroxylation, disulfide bridges, and glycosylation that can block digestive sites for trypsin, leading to inefficient peptide generation, or resulted in peptides of different masses than the expected masses from the genomic data base, leading to inappropriate identification of the protein. All these biochemical specificities of ECM proteins make proteomic discovery challenging and explain why only a few proteome of ECMs have been published so far. Nevertheless, this approach has a huge potential and consequently efficient solubilization and identification protocols are under development to make this technique more routinely usable in ECM and biomaterials characterization.

spectra at atomic resolution. Protein bands (1D) or spots (2D) are excised from the gel, hydrolysed into peptides by a proteolytic enzyme (frequently Trypsin), and loaded in the mass spectrometer to measure the exact mass of the peptides. Each protein from the original mixture is identified by matching the measured masses of their peptides with the expected masses of peptides calculated in silico from genomic database [37,38]. This technique allows a large-scale identification of components without the bias of predetermined molecular candidates as with antibody detection. It is thus possible in theory to have the exact protein composition of a tissue

The total or relative amount of identified proteins can also be addressed. The SDS-PAGE migration pattern and intensity of protein band (1D) or spot (2D) give a "map" of the protein content of the target tissue or organ and can be used to identify particular band/spot that are modified in specific conditions, enabling discovery of new therapeutic targets [40]. Labeling methods exist to generate quantitative data with mass spectrometry. Samples of the control conditions are modified with nuclear isotopes 13C, 15N, or 18O, whereas the treated sample is left unmodified, and the relative abundance of both isotopic pics is compared [41]. A direct semiquantitative approach is also possible, with the mathematical integration of ion counts of the peptides identifying each protein to describe its relative abundance [42]. In the ECM analysis, relative quantification is a remarkable tool to identify the specific isoform of some multimeric ECM proteins, such as collagens and laminins, as the relative amount of each monomer will indicate under which form the ECM molecule is present in the tissue. To be more specific, collagen type V exists in the common heterotrimeric isoform [*α*1(V)]2*α*2(V) and a more scarce homotrimeric isoform [*α*1(V)]3. The relative amount of ion counts for the *α*1(V) chain over *α*2(V) chain will indicate if the *α*1(V) chain is associated only with *α*2(V) (*α*1 chain signal twofold of *α*2 chain signal) or if the investigated tissue contains both heterotrimeric [*α*1(V)]2*α*2(V) and homotrimeric [*α*1(V)]3 isoforms (*α*1 chain signal >> twofold of α2 chain signal). However, quantification by mass spectrometry can be restricted by the ionization properties of some proteins that will make them less detected and consequently underrepresented in the final analysis. Nevertheless, this highlights the potentials of proteomics and mass spectrometry in the study of ECM proteins, as such characterization of ECM proteins isoform will require several antibodies (i.e., one per protein chain) to identify one isoform by

The most critical steps of a proteomic analysis are the purification of the protein mixture and their identification from database. ECM proteins have a high molecular weight and are tightly associated with each other by covalent cross links that make them mostly insoluble. An important point in the analysis of ECM by mass spectrometry proteomic will be the proper solubilization of the ECM [44]. The tissue has to be first carefully decellularized to purify the ECM and eliminate the remaining intracellular proteins. This step requires the use of a detergent like SDS and will eliminate from the ECM part of the loosely bounded proteins like remodeling enzymes or growth factors [40]. The purified ECM can be solubilized by a combination of physical, chemical, and enzymatic methods. A physical method is the me‐ chanical breaking with a French press or grinding with mortar and pestle in liquid nitrogen. This step is important to homogenize correctly the purified ECM and make the following

and follow its modification with time or diseases [39].

302 Composition and Function of the Extracellular Matrix in the Human Body

western blot or IHC [43].

#### **2.4. Three-dimensional organization of ECMs studied by electron microscopy**

Electron microscopy gives higher spatial resolution than light microscopy with the use a shorter wave-length from an electron beam. With resolution at the nanometres scale, and below with high resolution microscopes, it gives access to the molecular structure of ECM proteins and can image their supramolecular organization (i.e., fibrils and fibers assemblies) that are hardly distinguishable with optical microscopes. Transmission electronic microscopes (TEM) are built on the same scheme as optical/visible-light microscopes and so, equivalent techniques and processing of samples are required for both type of microscopy. In TEM, the electron beam pass through the samples to give rise to a projected image on an electron-sensitive surface like a phosphorescent screen, on a silver-film plate to record the image or nowadays on CCD cameras. The electron beam requires a vacuum pressure and cannot pass through thick samples of several micrometers which both represent a challenge for biological samples that are mainly wet, thick, and soft materials [46]. Biological tissues have to be fixed, dehydrated, and embedded in hard material (epoxy resin) and sliced with a diamond knife ultra-microtome into hundreds nanometres slices to be investigated by TEM. The electron beam interacts poorly with low atomic numbers atoms, such as carbon, oxygen, and nitrogen found in biological samples, so sections are stained with heavy metal solutions (commonly tungsten in phospho‐ tungstic acid, uranium in uranyl acetate, and lead in lead citrate) to give contrast [47]. Com‐ pared to histological staining, negative staining is more commonly used to prepare TEM sample to improve the contrast of organic materials: heavy metals dyes are absorbed by the background that creates contrast to the slightly stained specimen. The observation of ECM by TEM is nearly concomitant of its apparition in the late 1930s. Native collagen fibrils extracted from tissues and stained negatively with phosphotungtsic acid present a typical cross-striated pattern with a series of dark and light bands, spaced with a regular period of 67 nm. These observations have allowed the establishment of the assembly model of collagen molecules into collagen fibrils, known as the quarter-stagger model from Hodge and Petruska (1963). This model proposes a lateral stacking of collagen molecules, creating overlaps that exclude phosphotungstic dye and appear light, and a longitudinal collinear succession of collagen molecule spaced with a constant gap filled by Tungsten dye and appears dark under electron beam [48]. The cross-striated pattern is characteristic of fibrillary collagen, that are collagen type I, type II, and type III. On the other hand, network forming collagen type IV do not present any bands on TEM but is seen as a meshwork of hexagonal structures [49]. The resolution (roughly 1–5 nm) of TEM allows analyzing single macromolecules deposited on carbon film and stained by rotary-shadowing, creating a 3D electron sensitive replica of the specimen [50]. This method has revealed the semiflexible rod structure of collagen molecules terminated by a globular C-term pro-peptide and the cross-shaped triple chain structure of laminin mole‐ cules. TEM is particularly accurate to measure length of ECM molecules and diameters of fibrils and fibers assembly. These last parameters are important when analyzing a tissue because ECM fibrils diameters appear to be tissue-specific and modification of their size can be induced by pathologies such as diabetes, fibrosis, cancer, or aging and consequently impair tissue organization and function [51]. Compact bone tissue which supports most of the load of the body and muscles anchorage has to resist strong mechanical solicitations, but it is surprisingly light in weight structure if compared to human-engineered buildings. Bone tissue is made of an abundant organic ECM, strengthened with a mineral phase, and has highly hierarchical structure with length scales ranging from meters to nanometers that give its overall mechanical properties [52]. The shaft of long bones is organized in cylindrical osteons formed by successive concentric lamellae, themselves constituted by compact assemblies of collagen fibrils. An oblique transverse section of successive concentric lamellae made with ultra-microtome and observed with TEM revealed coexistence of two patterns of organization for collagen fibrils [53]. One is an alternation of parallel and orthogonal fibrils, with a regular 90° shift of fibrils orientation from one lamellae to another. The second is seen under TEM as arced structures, as if collagen fibrils were bent in between two series of longitudinal fibrils. The arced pattern is the consequence of the oblique sections into succession of collagen fibrils rotating with a tiny and constant angle from one lamellae to another, creating the illusion of bend structures [54]. From these TEM observations, a twisted plywood model of collagen fibril organization in bone has been proposed. This particular constant angle twist recall the organization observed in some liquid crystal phase, and it has been suggested that collagen molecules could have a liquid crystal behavior and autoassemble in higher-scales structures [21,55]. This finally underlines the potentials of transmission electron microscopy (TEM) to address ECM archi‐ tectures in tissues. As mentioned above, similar techniques and processing of samples used in optical microscopy are also applied with TEM. The different components of the ECM can be identified by immunolabeling with the same limitations for the necessity to retrieve antigens from the fixation and embedding processes. The antibodies are covalently linked to a gold particle to be seen by TEM and multiple labeling is possible with the use of a specific size of gold particle for each antibody [30].

TEM, however, needs a very thin specimen and cannot directly image a 3D structure. Unlike TEM, scanning electron microscope (SEM) uses electrons reflected from the surface of the sample as signals for image generation and provides information on surface topography, fibrillar organization, porosity, and also atomic composition of a bulk sample [56]. Samples have to be dehydrated to enter the low-vacuum chamber of the microscope and coated with an electron conducting layer (commonly gold) to ensure an adequate contrast and avoid charging phenomenon on the sample which are deleterious for the quality of the image. To keep their native 3D structure intact, biological samples are usually dehydrated by ethanol treatment and to a critical point drying. This procedure has enabled the evaluation of collagen fibrils diameters and spatial organization in reconstituted collagen hydrogels [57]. In biological tissues, the higher proportion of cells compared to ECM can minimize the access to the ECM fibrillar network. By a gentle decellularization method, the cellular counterpart of the tissue is removed and the native ECM frame remains [58]. This process mainly keeps in their original shape the reticular fibers of collagen and elastin but degrade most part of the laminins and GAGs network.
