Selective Visualization of Type II Collagen Using Sum‐Frequency Generation (SFG)
Salile Khandani, Yryx Y. Luna Palacios, Hannah Savage, Luis Chavez, Eric O. Potma
TL;DR
This paper introduces a new microscopy technique that can distinguish collagen Type II from Type I without labels, revealing its unique structural organization in cartilage.
Contribution
The study introduces a polarization-resolved SFG method to specifically visualize and differentiate collagen Type II from Type I.
Findings
Collagen Type II can be selectively identified using the XXY tensor element at 2860cm−1 in SFG measurements.
Collagen Type II in rat auricular cartilage forms pocket-like domains of aligned fibrils rather than a uniform network.
Polarization-resolved SFG microscopy provides label-free discrimination between collagen Types I and II.
Abstract
Collagen Types I and II share highly conserved triple‐helical backbones and similar C—H stretch vibrational spectra, which limits the ability of conventional spectroscopic or second‐harmonic generation methods to unambiguously distinguish between them in native and engineered matrices. By combining polarization‐resolved sum‐frequency generation (SFG) measurements with tensor‐based simulations of the C—H stretch response, this work identifies collagen's asymmetric CH2 mode measured via the XXY tensor element at 2860cm−1 as a robust optical marker that exhibits distinct spatial symmetries for collagen Type I and Type II. In rat auricular cartilage, analysis of the polarization‐resolved SFG signatures combined with vertex component analysis reveals pocket‐like domains of differently oriented collagen Type II fibrils rather than a uniformly aligned network. These findings establish…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
FIGURE 1
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6
FIGURE 7
FIGURE 8- —National Science Foundation10.13039/100000001
- —National Institutes of Health10.13039/100000002
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCollagen: Extraction and Characterization · Osteoarthritis Treatment and Mechanisms · Nasal Surgery and Airway Studies
Introduction
1
Collagen is the predominant structural protein in mammals and the principal determinant of the mechanical and biological integrity of connective tissues such as skin, bone, tendon, ligament, and cartilage. Among the fibrillar collagens, Types I and II play pivotal roles in maintaining tissue form and function: Type I provides tensile strength in fibrous tissues like tendon and bone, whereas Type II forms the fibrillar backbone of cartilage, supporting compressive resilience and joint articulation [1, 2]. Subtle changes in the relative abundance and architecture of these two collagen types accompany development, aging, degeneration, and repair, linking their structural organization directly to tissue health and regenerative capacity [3, 4]. Distinguishing collagen Types I and II within native and engineered matrices is therefore fundamental to elucidate disease mechanisms, evaluate biomaterial performance, and guide the design of therapeutic strategies.
Cartilage, the primary site of collagen Type II, is a specialized connective tissue composed of chondrocytes embedded within a highly organized extracellular matrix (ECM) that provides both structural and mechanical support [5, 6]. Elastic cartilage, such as that found in the external ear, combines a fibrillar network of Type II collagen with elastin fibers, together conferring exceptional flexibility and resilience [7]. Rat auricular cartilage serves as a widely used model for studying elastic cartilage due to its unique accessibility, flexibility, and defined microarchitecture, which make it ideal for analyzing ECM organization, remodeling, and tissue biomechanics [8, 9]. Within this tissue, Type II collagen forms a hierarchical network that maintains the shape, tensile strength, and structural integrity of cartilage [10], while elastin fibers interwoven through the collagenous scaffold allow reversible deformation, granting the tissue its characteristic mechanical adaptability [1]. Disorganization of this network—whether from injury, aging, or disease—alters local collagen orientation and signaling, marking early signs of matrix degeneration and impaired regeneration [4, 11, 12, 13, 14].
Traditional methods such as histology, electron microscopy, and immunohistochemistry have unveiled many aspects of cartilage architecture, yet they remain inherently destructive, requiring fixation, sectioning, and staining that may distort the native molecular arrangement [15, 16, 17]. These limitations have led to increasing interest in non‐destructive, label‐free imaging techniques capable of probing structural and molecular organization in intact tissues. Nonlinear optical (NLO) microscopy [18], in particular second‐harmonic generation (SHG) microscopy, has become a workhorse for visualizing fibrillar collagen, as the non‐centrosymmetric triple‐helical arrangement of peptide bonds gives rise to strong intrinsic SHG contrast without exogenous labels [19, 20]. SHG is sensitive to fibril orientation, packing, and organization, and has been widely applied to quantify collagen architecture in healthy and diseased tissues, particularly for collagen Type I [21, 22, 23, 24, 25, 26, 27]. However, SHG is fundamentally an intensity‐based, nonresonant process, and different fibrillar collagens with similar backbone symmetry can produce overlapping polarization responses. Indeed, previous work analyzing the feasibility of discriminating collagen Types I and II using polarization‐resolved SHG and extracted tensor‐element ratios (such as χ33/χ31) in tendon (Type I) and cartilage (Type II) showed only marginal differences [28]. Consequently, SHG microscopy has not emerged as a reliable method for unambiguously differentiating between collagen types.
Sum‐frequency generation (SFG) vibrational microscopy addresses this limitation by combining the structural selectivity of a second‐order nonlinear process with vibrational contrast [29, 30, 31, 32, 33]. In its collinear form, SFG microscopy can be integrated into a standard microscope frame [34], enabling rapid laser‐scanning microscopy in a similar fashion as in SHG microscopy, but offering additional chemical contrast [35, 36]. For instance, SFG microscopy has been used to discriminate collagen Type I in cancerous tissue from healthy samples, based on spectral signatures of selected tensor elements [37]. Polarization‐resolved SFG microscopy extends this capability by probing molecular orientation, vibrational symmetry, and supramolecular chirality [38, 39]. Because collagen's second‐order NLO response encodes its molecular arrangement in a bond‐specific manner, polarization‐resolved SFG enables the detection of subtle differences in fibrillar organization that cannot be distinguished with conventional SHG imaging [39].
In this study, we leverage the molecular sensitivity of polarization‐resolved SFG microscopy to confidently discriminate collagen Type II from Type I in intact rat ear cartilage. By analyzing polarization‐resolved SFG responses of CH‐stretch vibrational modes in collagen Types I and II—through both simulations and experiments—we observe a distinctive difference in the XXY polarization response of the asymmetric CH‐stretch vibrational mode between these two collagen types. This polarization‐dependent variation highlights the sensitivity of SFG to molecular and supramolecular structural differences between these important members of the collagen family. Our findings establish a framework for extracting orientation information from tissue‐resolved SFG signals, offering a label‐free route to structural diagnostics of cartilage integrity, early matrix degeneration, and the evaluation of regenerative constructs.
Methods
2
Sample Preparation
2.1
Ear specimens were harvested from adult rats and placed into cassettes for processing. The tissues were processed in an automated tissue processor (Epredia Thermo Scientific Excelsior AS, Thermo Fisher Scientific, Waltham, MA, USA) using a standard overnight program, which included sequential dehydration in graded ethanol, clearing in xylene, and paraffin infiltration. The samples were subsequently embedded in paraffin wax using an embedding workstation (Thermo Scientific HistoStar, Waltham, MA, USA). Paraffin blocks were kept on ice prior to sectioning to improve ribbon quality. The samples were cut to 7 μm thickness using a rotary microtome (Epredia HM355S, Germany). The sections were floated on a 40°C water bath to remove wrinkles and mounted onto glass coverslips.
Collagen Type I samples were obtained from fresh rat tail tendon. Excised segments were carefully flattened to produce tissue slices of approximately 100 μm thickness, then mounted in a hydrated state between two borosilicate No. 1 coverslips.
Histological Methods
2.2
Immunofluorescence staining for Type II collagen was performed on the paraffin sections of rat ear cartilage tissue. After deparaffinization and rehydration, antigen retrieval was carried out using a high‐pH antigen retrieval solution (eBioscience IHC Antigen Retrieval Solution, 10×, pH ≈9) in a heat‐induced protocol according to the manufacturer's instructions. The sections were allowed to cool to room temperature, rinsed in PBS, and then washed in PBS containing 0.1% Tween‐20. Non‐specific binding was blocked with 4% fish gelatin in PBS (with 0.1% Tween‐20) for 20–30 min at room temperature.
Tissues were then incubated overnight at 4°C with primary antibodies against Type II collagen, including rabbit anti‐collagen II (Proteintech, 28459‐1‐AP) and mouse anti‐collagen II (Abcam, ab34712), diluted in fish gelatin blocking buffer. After three washes in PBS with 0.1% Tween‐20, the sections were incubated for 1 h at room temperature with Alexa Fluor 555–conjugated donkey anti‐rabbit IgG secondary antibody and matched to the host species of the primary antibodies. The slides were washed again in PBS and mounted with VECTASHIELD Antifade Mounting Medium, and fluorescence images were acquired using a Keyence BZ‐X710 fluorescence microscope equipped with appropriate filter sets for Alexa Fluor 555.
Reflective SFG Microscope
2.3
SFG measurements were performed using a reflective SFG microscope setup, as detailed in our prior work [39]. In short, the system is driven by a picosecond Yb^3+^‐fiber laser (aeroPULSE, NKT) operating at 1031 nm and producing 6 ps pulses at 76 MHz. A portion of the near‐infrared (NIR) output pumps a mid‐IR (MIR) optical parametric oscillator (Levante IR, APE) to generate the tunable excitation beam (2700–3200 cm^−1^). The tunable MIR beam was used as the ω1 field, while the residual 1031 nm beam serves as the ω2 input for SFG as well as the pump beam for the two‐photon excited fluorescence (TPEF) signal. Incident NIR and MIR beam polarizations were independently controlled using linear polarizers and half‐wave plates prior to their collinear recombination on a dichroic mirror.
All‐reflective relay optics were employed to minimize chromatic dispersion, aberrations, and polarization distortion across wavelengths. The excitation beams were focused onto the sample using a reflective Schwarzschild–Cassegrain objective (Model 5007, Beck) with NA=0.65. For enhanced performance and a larger field‐of‐view, we also used a custom‐designed freeform reflective microscope objective (0.65 NA) [40]. The radiated SFG signal was collected in the forward direction by a refractive condenser, filtered by a 780–800 nm bandpass filter, analyzed with a linear polarizer (analyzer), and detected with a photon‐counting module (H16721‐50, Hamamatsu). A dichroic mirror in the detection path reflected signal wavelengths shorter than 650 nm to a second photo‐multiplier tube (H7422‐01, Hamamatsu), which was used to detect the TPEF signal in the 540–570 nm range. The beam focusing properties of the reflective objectives were previously validated to exhibit negligible depolarization effects [39].
For polarization sensitive measurements, we define the XY laboratory axes to be in the plane parallel to the sample stage, whereas the Z‐axis is along the main propagation direction of the incident optical fields. The main fibrillar direction of collagen Type I fibers was aligned along the Y‐axis of the laboratory frame, using the visible striations of the SFG image to determine fiber orientation.
Results and Discussion
3
Histological Assessment of Auricular Cartilage
3.1
In this work, we have used auricular cartilage of the rat ear as a model system for studying the SFG response of native collagen Type II. Our first goal was to verify the presence of collagen Type II in this model system by using a standard immunofluorescence staining procedure. Our focus was on the elastic cartilage that can be found throughout the scapha and toward the external meatus opening of the external ear.
The immunofluorescence analysis confirmed the presence and distribution of collagen Type II within rat auricular cartilage by widefield fluorescence microscopy. Sections stained with the Proteintech antibody demonstrated a clearly discernible collagen Type II signal outlining the elastic cartilage matrix and the pericellular regions surrounding lipochondrocytes, evidence of the abundant presence of collagen Type II within the tissue (Figure 1).
Wide‐field immunofluorescence validation of collagen Type II in rat ear cartilage. Antibody‐stained sections show collagen Type II distribution (green) throughout the elastic cartilage matrix.
We next performed NLO microscopic imaging of the ear tissue samples using TPEF contrast to visualize the fluorophore stained elastic cartilage and compare it to SHG collected from the same area. In Figure 2a, the TPEF signal of a lipochondrocyte‐rich region of the elastic cartilage is shown, highlighting the collagen‐rich regions surrounding the lipochondrocytes. The corresponding SHG signal is depicted in Figure 2b, showing the same pericellular regions observed in the TPEF image. These imaging results confirm that the SHG signal reports on the presence of collagen Type II in the auricular cartilage matrix. The sensitivity of SHG to collagen Type II is expected given the protein structure's noncentrosymmetry, as observed in previous studies [28].
Nonlinear optical imaging of collagen Type II in rat ear cartilage. (a) TPEF image from antibody‐stained elastic cartilage excited at 1031 nm. (b) SHG image of the same area, using the same 1031 nm excitation beam.
SFG Microscopy Visualizes Collagen II in Rat Ear Cartilage
3.2
The SHG imaging results suggest that collagen Type II can also be visualized with SFG microscopy. To verify this, we carried out imaging experiments by tuning the frequency of the MIR beam into C—H stretching region (2700–3200 cm^−1^). Figure 3a shows a region rich in collagen Type II, as evidenced by TPEF mapping of antibody‐stained elastic cartilage. Similar features are resolved in the SFG image of Figure 3b, recorded at the vibrational frequency of 2943 cm^−1^, thus confirming that SFG microscopy produces detectable signal from collagen Type II in cartilage tissue.
Nonlinear optical imaging of collagen Type II in rat ear cartilage. (a) TPEF image excited at 1031 nm of antibody‐stained elastic cartilage. (b) SFG image of the corresponding area recorded at 2943 cm−1.
The SFG spectra of collagen Type II from rat ear cartilage and collagen Type I (rat tail tendon), collected in the YYY polarization configuration, are shown in Figure 4. The collagen Type II spectrum in this frequency range reveals characteristic vibrational features of collagens, marked by the strong resonance near 2943 cm^−1^, which is attributed to the symmetric C—H stretching vibrations of proline and hydroxyproline rings [39, 41]. The overall spectral SFG profile closely resembles that of collagen Type I reported in previous studies [31, 32, 41, 42]. This spectral similarity arises from their conserved triple‐helical structures composed of repetitive Gly–X–Y sequences and high contents of proline and hydroxyproline, which create rather comparable carbon‐hydrogen vibrational environments.
SFG spectrum of collagen Type I from rat tail tendon and Type II from rat ear elastic cartilage, collected for the YYY polarization setting. Spectra were obtained by averaging the SFG signal over a region of interest of ∼100×100μm2.
Despite these similarities, several spectral differences between the collagen spectra can be identified as well. First, the collagen Type II spectrum features a pronounced shoulder on the low energy side of the main peak, in the 2860−2880cm−1 range. This spectral contribution is attributed to the asymmetric C—H stretching mode of glycine [39], which appears prominently in the YYY spectrum of collagen Type II but is absent in the corresponding collagen Type I spectrum. Second, whereas the main peak of collagen Type I displays a marked asymmetry, characterized by a shallow slope on the high energy side, the profile appears narrower and more symmetric in the collagen Type II spectrum. Although it is possible to leverage these differences for spectral discrimination between the two collagens, the overall spectral contrast remains subtle, complicating direct and unambiguous differentiation between collagen types. For this reason, we explored an alternative strategy for the selective visualization of collagen Type II, one that takes advantage of the polarization properties of the SFG technique.
Polarization‐Resolved SFG Selectively Identifies Collagen Type II
3.3
To examine the sensitivity of polarization‐resolved SFG for differentiating between collagen types, we employed a computational framework to simulate the expected SFG response arising from the methylene stretching modes in fibrillar collagen. Details on the computational model are given in reference [39]. We used the crystal structure of collagen Types I and II from the protein data bank (PDB) [43] to extract atomic coordinates and to determine the signal contributions from individual methylene bridge groups (referred to as the methylene group hereafter) within the fibrillar structure. In the simulations, the main fibrillar axis is aligned with the Y‐axis of the laboratory frame, whereas the transverse polarization direction of each of the incident fields is rotated by an angle Φ around the Z‐axis.
Figure 5a shows the calculated polarization dependence of the SFG signal for the χYYY2 tensor element of collagen Type I, for the case of the symmetric C—H stretching mode. With the fibril aligned along the Y‐axis, both polarization directions of the incident ω1 and ω2 fields, as well as the polarization orientation of the analyzer, are initially parallel to Y‐axis (Φ=0). Upon rotation of the field polarizations around Φ, the χYYY2 signal is modulated, producing the double‐lobed pattern seen in the polar plot of Figure 5a. A similar simulation for collagen Type II is provided in Figure 5b, which also reveals a double‐lobed pattern. Although there are small differences between the χYYY2 patterns for collagen Types I and II, they both display a high degree of dipolar symmetry, which complicates their unequivocal differentiation.
Simulated polarization dependence of selected SFG tensor elements for the CH2 stretching modes in collagen Types I and II. (a, b) Simulated responses for the symmetric stretch of the YYY configuration, and (c, d) the corresponding responses for the asymmetric stretch in the XXY configuration.
In general, we observed that most of the tensor elements exhibit polar plots that are similar for collagen Type I and Type II, both for the symmetric and asymmetric methylene stretching mode. This is not surprising, given the comparable overall fibrillar structure and methylene group arrangement in the two collagen types. However, we detected a notable difference in the XXY element of the asymmetric mode: while collagen Type I shows a clear butterfly‐shaped pattern, collagen Type II displays a double‐lobed pattern. We surmised that the distinct symmetries between these polar plots facilitate the discrimination between collagen Type I and Type II.
We next experimentally verified the dependence of the SFG XXY‐tensor element on the polarization orientation of the incident fields. In this setup, the polarization directions of the ω1 and ω2 incident fields were set orthogonal and jointly rotated by the angle Φ relative to the sample Y‐axis in the plane. The analyzer was simultaneously rotated by the same angle Φ. Figure 6a shows the SFG signal of rat tail tendon in the XXY configuration when Φ=0, obtained at 2860 cm^−1^ where the asymmetric stretch is prominent for both collagens in this configuration [39]. The polarization dependence of the χXXY2 signal is shown in Figure 6c, where the experimental data points are denoted by dots, while the thick solid lines represent the simulated asymmetric angular dependencies of the measured SFG signal. The resulting polar plot reveals the characteristic butterfly pattern for collagen Type I, similar to the simulation in Figure 5c.
SFG imaging and polarization‐resolved measurements of collagen Types I and II. (a) SFG image of collagen Type I from rat tail tendon. (b) Polarization‐resolved SFG measurements of the XXY tensor element of collagen Type I, tuned to the asymmetric methylene stretching mode (2860 cm−1). (c) SFG image of collagen Type II extracted from rat ear cartilage. (d) Polarization‐resolved SFG measurements of the XXY tensor element of collagen, tuned to the asymmetric methylene stretching mode (2860 cm−1).
Figure 6b depicts an SFG image of collagen Type II in rat ear cartilage for the XXY polarization configuration at 2860 cm^−1^. The corresponding polarization dependence of the collagen Type II signal of a selected area in the image is shown in Figure 6d, which reproduces the main double‐lobed features of the simulated pattern in Figure 5d. This striking divergence in both the experimental data points and simulated curves highlights the high sensitivity of polarization‐resolved SFG to subtle supramolecular structural differences between the two fibrillar collagens.
We used polarization‐resolved SFG microscopy to distinguish collagen Types I and II in rat ear tissue. Figure 7a shows a larger field‐of‐view SFG image (2860 cm^−1^) of tissue containing fibrous dermal structures (Zone I) and elastic cartilage (Zone II), which are expected to be rich in collagen Type I and Type II, respectively. While the total SFG intensity could not differentiate these types, polarization‐resolved analysis successfully did. The application of a two‐component vertex component analysis (VCA) [44] segmented the image by polarization property (Figure 7b). As shown in Figure 7c, the Zone I polar plot exhibits the butterfly profile characteristic of collagen I, while Zone II shows the double‐lobed pattern indicative of collagen II (Figure 7d). This demonstrates label‐free, type‐specific collagen identification in tissues.
Polarization‐resolved SFG distinguishes collagen types in rat ear tissue. (a) SFG image at 2860 cm−1 of rat ear tissue taken for XXY showing dermal collagen (Zone I) and elastic cartilage (Zone II) (b) Two‐component VCA segmentation based on polarization response. (c) Polar plots reveal a butterfly‐shaped averaged profile in Zone I (Collagen I), and (d) a bipolar averaged profile in Zone II (Collagen II).
SFG Imaging Detects Heterogeneity in Collagen Type II Orientation
3.4
Using the distinct polarization dependence of the XXY tensor element of collagen Type II at 2860cm−1, we studied the fibrillar orientation of collagen in elastic cartilage of the rat ear in more detail. Compared to collagen Type I in tendon tissue, which exhibits a high degree of fibrillar alignment on the >100μm scale, the corresponding alignment of collagen Type II fibrils in cartilage is expected to be significantly less prominent on these length scales. The SFG intensity image in Figure 8a appears to confirm this notion, showing limited signs of a unidirectional organization of collagen fibrils.
Polarization‐resolved SFG imaging of collagen Type II. (a) SFG image of collagen Type II in rat ear cartilage at 2860 cm−1. (b) Vertex component analysis (VCA) of the polarization‐resolved XXY SFG dataset at 2860 cm−1, showing two spatially segregated endmember domains in blue and red. (c) Polar plots of the polarization dependence for the blue and red VCA endmember regions, demonstrating distinct but bipolar XXY responses characteristic of collagen Type II.
Nonetheless, the intensity image reveals limited information about fibrillar orientation. Figure 8b depicts a VCA of an SFG data stack composed of images obtained at incrementally changing values of Φ. The VCA yields two distinct endmembers, here indicated by the blue and red domains, which correspond to distinct orientations of fibrillar collagen Type II. Each of these color‐coded regions exhibits the same SFG spectrum and a similar bipolar XXY polarization dependence, within experimental uncertainty, confirming that both domains arise from collagen Type II with identical local molecular structure. As is evident from Figure 8c, these domains differ only in their relative orientation of collagen fibrils, roughly separated by an angle of ΔΦ∼90°.
The emergence of two endmembers reflects differences in local fibrillar orientation and packing of collagen Type II, indicating that the cartilage matrix is organized into discrete pockets of differently oriented collagen Type II fibrils instead of a single uniformly aligned network. This heterogeneous, pocket‐like organization of collagen Type II contrasts with the highly aligned, cable‐like collagen Type I architecture, where dense bundles of parallel fibrils form a continuous load‐bearing network. In other words, while tendon collagen Type I forms long, uniformly aligned fibril bundles [45], we find that auricular cartilage collagen Type II is arranged into spatially separated domains of differently oriented fibrils on the ∼10μm length scale.
Conclusions
4
Polarization‐resolved SFG microscopy provides a specific, tensor‐based contrast that separates collagen Type II from collagen Type I in intact tissue. In rat ear cartilage, the C—H stretch SFG spectrum of collagen Type II shows similarity to that of collagen Type I, consistent with their conserved triple‐helical structure, so alternative strategies are desired to allow reliable type identification. By combining measurements with χ2 tensor simulations, the asymmetric CH_2_ mode at 2860 cm^−1^, measured in the XXY configuration, is identified as a discriminating channel. This difference in the XXY response provides an unambiguous, label‐free optical criterion to distinguish the two fibrillar collagens in native and engineered matrices.
The polarization‐resolved SFG data also describe how collagen Type II is organized in auricular cartilage. Pixel‐wise inspection of the XXY response using VCA shows regions that share an identical SFG spectrum and the same bipolar XXY polarization dependence, indicating that these areas all arise from collagen Type II. What varies between regions is the preferred fibrillar orientation and packing, revealing discrete orientation domains instead of a single uniformly aligned network.
Funding
This work was supported by the National Science Foundation (NSF), grant CMI‐2404006, and the National Institutes of Health (NIH), grant R21‐EB034084.
Conflicts of Interest
The authors declare no conflicts of interest.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1H. Watanabe , Y. Yamada , and K. Kimata , “Roles of Aggrecan, a Large Chondroitin Sulfate Proteoglycan, in Cartilage Structure and Function,” Journal of Biochemistry 124, no. 4 (1998): 687–693.9756610 10.1093/oxfordjournals.jbchem.a 022166 · doi ↗ · pubmed ↗
- 2V. C. Mow and R. Huiskes , Basic Orthopaedic Biomechanics & Mechano‐Biology (Lippincott Williams & Wilkins, 2005).
- 3A. Ripamonti , N. Roveri , D. Braga , D. J. S. Hulmes , A. Miller , and P. A. Timmins , “Effects of p H and Ionic Strength on the Structure of Collagen Fibrils,” Biopolymers 19, no. 5 (1980): 965.7378548 10.1002/bip.1980.360190503 · doi ↗ · pubmed ↗
- 4M. D. Grynpas , D. R. Eyre , and D. A. Kirschner , “Collagen Type II Differs From Type I in Native Molecular Packing,” Biochimica et Biophysica Acta (BBA)‐Protein Structure 626, no. 2 (1980): 346.10.1016/0005-2795(80)90129-47213652 · doi ↗ · pubmed ↗
- 5Y. Gao , S. Liu , J. Huang , et al., “The ECM‐Cell Interaction of Cartilage Extracellular Matrix on Chondrocytes,” Bio Med Research International 2014, no. 1 (2014): 648459.24959581 10.1155/2014/648459 PMC 4052144 · doi ↗ · pubmed ↗
- 6A. J. Sutherland , G. L. Converse , R. A. Hopkins , and M. S. Detamore , “The Bioactivity of Cartilage Extracellular Matrix in Articular Cartilage Regeneration,” Advanced Healthcare Materials 4, no. 1 (2015): 29–39.25044502 10.1002/adhm.201400165 PMC 4286437 · doi ↗ · pubmed ↗
- 7P. A. West , M. P. G. Bostrom , P. A. Torzilli , and N. P. Camacho , “Fourier Transform Infrared Spectral Analysis of Degenerative Cartilage: An Infrared Fiber Optic Probe and Imaging Study,” Applied Spectroscopy 58, no. 4 (2004): 376–381.15104805 10.1366/000370204773580194 · doi ↗ · pubmed ↗
- 8J. Chandra Rajan , “Separation of Type III Collagen From Type I Collagen and Pepsin by Differential Denaturation and Renaturation,” Biochemical and Biophysical Research Communications 83, no. 1 (1978): 180–186.358975 10.1016/0006-291x(78)90414-x · doi ↗ · pubmed ↗