Matrix-assisted laser desorption ionization–imaging mass spectrometry: A new methodology to study human osteoarthritic cartilage




Information about the distribution of proteins and the modulation that they undergo in the different phases of rheumatic pathologies is essential to understanding the development of these diseases. We undertook this study to demonstrate the utility of mass spectrometry (MS)–based molecular imaging for studying the spatial distribution of different components in human articular cartilage sections.


We compared the distribution of peptides and proteins in human control and osteoarthritic (OA) cartilage. Human control and OA cartilage slices were cut and deposited on conductive slides. After tryptic digestion, we performed matrix-assisted laser desorption ionization–imaging MS (MALDI-IMS) experiments in a MALDI–quadrupole time-of-flight mass spectrometer. Protein identification was undertaken with a combination of multivariate statistical methods and Mascot protein database queries. Hematoxylin and eosin staining and immunohistochemistry were performed to validate the results.


We created maps of peptide distributions at 150-μm raster size from control and OA human cartilage. Proteins such as biglycan, prolargin, decorin, and aggrecan core protein were identified and localized. Specific protein markers for cartilage oligomeric matrix protein and fibronectin were found exclusively in OA cartilage samples. Their distribution displayed a stronger intensity in the deep area than in the superficial area. New tentative OA markers were found in the deep area of the OA cartilage.


MALDI-IMS identifies and localizes disease-specific peptides and proteins in cartilage. All the OA-related peptides and proteins detected display a stronger intensity in the deep cartilage. MS-based molecular imaging is demonstrated to be an innovative method for studying OA pathology.

Osteoarthritis (OA) is the world's most common age-related joint disease and is characterized by degeneration of cartilage (1, 2). In addition, degradation processes that structurally change the joint are involved in OA, including synovial inflammation, osteophyte formation, and remodeling of subchondral bone (3). This pathology affects up to 70% of the population age >65 years. Cartilage degradation represents a key process during OA pathogenesis. In addition, treatments to manage OA are limited to controlling pain and improving function. Therapies to reduce cartilage degradation and joint destruction are a new challenge in OA treatment. Based on all these comments, methodologies that help us to study in detail the structure of cartilage could improve the knowledge of OA.

Mass spectrometry (MS) facilitates the molecular identification of lipids, peptides, and proteins through the determination of their molecular weight and their fragmentation behavior (4). Proteins from the cartilage and chondrocytes, such as prolargin (PRELP), fibromodulin (FM), aggrecan core protein, decorin, or biglycan, have been previously identified (5, 6). All of them interact with glycosaminoglycans (GAGs) and build the extracellular matrix (ECM) in which the chondrocytes are embedded (6). OA-related proteins, such as cartilage oligomeric matrix protein (COMP) and fibronectin (FN), have been detected by MS and are considered markers of inflammation and tissue remodeling (7). For the first time, the identification and localization (with high spatial resolution) of these 2 proteins have been achieved using the same sample and without any type of labeling.

Matrix-assisted laser desorption ionization–imaging MS (MALDI-IMS) can determine the distribution of hundreds of unknown compounds in a single molecular imaging measurement. MALDI-IMS has been used in the past several years to look for peptides, proteins, and lipids in specific areas of a tissue section with a spatial resolution <50 μm (4, 8, 9). After careful preparation, the tissue section is introduced into the mass spectrometer, and then the proteins, peptides, and lipids are desorbed from discrete pixels from the surface and in an ordered way. Each pixel is linked to the mass spectrum specific for that region. A plot of the intensity of a signal produces a map of the relative abundance of that compound over the imaged tissue (10).

This technology provides a powerful tool for the investigation of biologic processes, as it is not known in advance which segment of the molecular complexity will be revealed. IMS is not necessarily a targeted analytic method, and the technology has been applied to a variety of pathologies, such as neurodegenerative disorders and many cancer types (11, 12). In addition, specific protein patterns revealed by IMS have been shown to be predictive of diagnosis and prognosis (13–15). MALDI-IMS has also been used to detect and map pharmaceutical compounds in sections from dosed tissues (14, 15). Although this is a promising method, there are no previous studies that describe and compare the peptide distribution in human control and OA cartilage. IMS is a new technology that can help us localize and identify the key molecules in OA pathology.



We obtained α-cyano-4-hydroxycinnamic acid (HCCA), trifluoroacetic acid (TFA), trypsin, and chloroform from Sigma-Aldrich. Acetonitrile (ACN) and ethanol were purchased from Biosolve, and octyl glucoside was purchased from Sigma-Aldrich.

Cartilage procurement and processing.

Control human knee cartilage from adult donors with no history of joint disease was provided by the Tissue Bank and the Autopsy Service at Complexo Hospitalario Universitario A Coruña. The cartilage was macroscopically and histologically normal. OA cartilage was obtained from consenting donors who were undergoing joint replacement. Cartilage slices were removed from the condyles and frozen in liquid N2. Slices (10 μm or 4 μm thick) were cut in a cryostat, deposited on indium tin oxide high-conductive slides (Delta Technologies), and then frozen at –20°C. This study was approved by the local ethics committee in Galicia, Spain.

Tissue digestion and matrix deposition for MALDI-IMS.

Cartilage samples from 10 donors with OA (ages 51–84 years) and 10 control subjects (ages 51–91 years) were studied in duplicate. (Information on the workflow that was followed for the study is available online at The 10-μm–thick sections were washed for 30 seconds in 100% ethyl alcohol, 2 minutes in 70% ethyl alcohol, and 30 seconds in chloroform. A cocktail of 50 ng/μl trypsin and 0.01% octyl glucoside was applied in an automated manner (CHIP-1000; Shimadzu) (8, 16). The whole tissue section was microspotted with the enzyme in a 150-μm spacing raster scheme. Twenty nanoliters of trypsin per position was deposited in cycles of 250 pl per droplet to cover the entire tissue surface. The samples were incubated overnight at 37°C. Subsequently, the matrix solution, HCCA (10 mg/ml) diluted in 50% ACN and 50% H2O containing 0.1% TFA (1/1 volume/volume), was sprayed on top of the tissue section by a vibrational sprayer (ImagePrep; Bruker Daltonics).


Digital optical scans of all tissue sections were obtained prior to MALDI-IMS experiments using a 2,400 dots per inch desktop scanner. The resulting digital images were imported into the MALDI Imaging Pattern Creator software (Waters Corporation). Instrument calibration was performed using a standard calibration mixture of polyethylene glycol with an MW of 100–3,000 (Sigma-Aldrich). A MALDI-quadrupole time-of-flight SYNAPT HDMS system (Waters) operating with a 200-Hz Nd:YAG laser was configured to acquire data in the positive V-reflectron mode. Data were acquired at a raster size of 150 μm. Ion images were generated with Biomap software (Novartis Pharma).

Tissue digestion and matrix deposition for profiling experiments.

Profiling experiments for protein identification were performed on each donor tissue, applying 20 μl of 50 ng/μl trypsin and 0.01% octyl glucoside in water. After drying, 20 μl of matrix was spotted on the digested position. Data-dependent analysis of tryptic peptides was performed in a MALDI SYNAPT HDMS system. Every MS survey scan was followed by collisional fragmentation of the most intense ions and collection of MS/MS spectra. Direct MS/MS fragmentations were performed in those masses directly from the tissue on the peptides that differentiate the OA tissue from the control tissue. These target peptides were found after discriminant analysis of the entire data set. The obtained spectra were processed in MassLynx (Waters). The resulting data files were submitted to a Mascot (Matrix Science) search using the Swiss-Prot database.

Immunohistochemistry and morphologic studies.

Four-micrometer–thick tissue sections were fixed in acetone at 4°C for 10 minutes and washed twice in 10 mM phosphate buffered saline–Tween 20 (0.1%; Sigma-Aldrich). The slides were incubated in a peroxidase-blocking solution for 10 minutes (Dako) prior to FN immunohistochemistry studies. After rinsing, 100 μl of a 1:100 dilution of horseradish peroxidase–conjugated anti-FN antibody (IST-9, sc-59826; Santa Cruz Biotechnology) was applied and incubated in a humidified chamber for 1 hour at room temperature. The slides were washed in water, and 100 μl of a diaminobenzidine substrate (Dako) was applied for 10 minutes. The slides were rinsed with water, dehydrated, and mounted. Digital images were generated with a BX61 Olympus microscope connected to a DP71 Olympus camera. Quantitation was performed with AnalySIS software, version 5.0 (Olympus Biosystems). For nuclei and cytoplasm staining, the slides were immersed in Harris' hematoxylin solution (Sigma-Aldrich) for 8 minutes. After washing with water, they were rinsed in 95% ethyl alcohol and counterstained in an eosin solution for 30 seconds. Digital images were acquired with the Mirax system (Carl Zeiss) after dehydrating steps.

Multivariate analysis and data interpretation.

The intensity of all mass/charge (m/z) channels was normalized to the intensity of the m/z 190 matrix peak using Biomap software. Peak average intensities were calculated for each mass of interest. Principal components analysis (PCA) was used to investigate spectral similarities and differences between all samples studied. The PCA extracted linear combinations of variables, each of which is associated with the largest possible variance after removing the variance of prior principal components in an iterative manner. The final data analysis was performed with the principal components that explained the highest variance in the data set (80%). Discriminant analysis was performed to look for the peaks with the highest differences between control and disease groups. P values for statistical differences found in MALDI-IMS and immunohistochemistry experiments were calculated with the Mann-Whitney U test using GraphPad Prism software, version 5.00 for Macintosh. P values less than or equal to 0.05 were considered significant. The data are expressed as the mean ± SEM intensity. String software version 9.0 ( was used to create hypothetical maps of interactions among different proteins according to PubMed literature.


MALDI-IMS of control and OA human cartilage samples.

MALDI-IMS was used for the study of peptide distributions in control and OA cartilage. (A combined spectrum of a representative digested human control sample is available online at We detected peaks between 100 and 3,000 daltons including very abundant peptides, identified by MS/MS in the profiling experiments, such as m/z 1590.9 (PRELP), m/z 2270.2 (aggrecan core protein), m/z 2763.3 (decorin), and m/z 2846.3 and m/z 3081.6 (biglycan) (Table 1).

Table 1. Proteins identified directly from the tissue in control or in OA cartilage*
Cartilage, Swiss-Prot accession no.Protein nameObserved mass/charge ratioMascot algorithm scorePeptide sequence
  • *

    OA = osteoarthritic.

 P16112Aggrecan core protein2271.151K.VSLPNYPAIPSDATLEVQSLR.S
 O75339Cartilage intermediate-layer protein 11954.043R.TTDWTPAGSTGQVVHGSPR.E
 P49747Cartilage oligomeric matrix protein2256.1105R.FYEGPELVADSNVVLDTTMR.G
 P02458Collagen α1(II) chain2023.936K.QGDRGEAGAQGPMGPSGPAGAR.G
 P16112Aggrecan core protein819.430K.DEFPGVR.T
 Q8WYP5Protein ELYS1025.627K.KIEVPAQLK.E

The spectra of the human OA cartilage exhibited a different profile (further information is available online at Several peaks were common to both sample types, such as tryptic peptides of biglycan and aggrecan core protein. Peptides such as FN (m/z 1349.7, m/z 1593.9), COMP (m/z 2256.1), FM (m/z 1955.1), collagen α1(II) chain (m/z 2023.9), and protein ELYS (m/z 1025.6) were identified by MS/MS directly from the tissue surface (Table 1) and found to be distinctive for OA.

The images of the peaks were used to investigate the protein distribution differences in OA and control samples through the examination of the tryptic peptide intensities. Figure 1A shows a hematoxylin and eosin staining of an OA cartilage sample showing the superficial and the deep area. Figure 1B shows the representation of 2 homogeneously distributed peaks on a control tissue section, 1 of them corresponding to the protein PRELP (m/z 1590.9). Studying the intensities of some OA-specific peptides with the Biomap software, we observed a striking difference in the peak intensity distribution. Four peptides from FN were imaged in OA and control samples (Figure 2). We observed a difference in the normalized intensity of these peaks, using the same scale bar for all figures. First, we observed higher intensities of the FN-related peaks in the OA samples. In addition, all of them were predominantly observed in the deep area of the OA cartilage. The tryptic peptides from COMP, m/z 2256.1 and m/z 1613.8, exhibited the same pattern differences when control and OA samples were compared (Figure 3).

Figure 1.

Analyses of peptide distribution and hematoxylin and eosin (H&E) staining. A, H&E staining of an osteoarthritic cartilage sample. The deep area (close to the bone) and the superficial area (in contact with the synovial fluid) are clearly differentiated. B, Example of the distribution of the random peptide mass/charge (m/z) 2278.1 and the peptide m/z 1590.9 from the protein prolargin in a control cartilage sample after matrix-assisted laser desorption ionization–imaging mass spectrometry experiments. Scale bars show normalized intensities. Color figure can be viewed in the online issue, which is available at

Figure 2.

Distribution and intensities of 4 peptides from the fibronectin (FN) protein in control and osteoarthritic (OA) cartilage. Represented are the different intensities and distributions of FN peptides in control and OA cartilage samples after matrix-assisted laser desorption ionization–imaging mass spectrometry experiments and tandem mass spectrometry profiling identifications. Scale bars show normalized intensities. m/z = mass/charge.

Figure 3.

Comparison of the intensity and the distribution of 2 peptides from cartilage oligomeric matrix protein (COMP) in control and OA cartilage. Represented are the different intensities and distributions of COMP peptides in control and OA cartilage samples after matrix-assisted laser desorption ionization–imaging mass spectrometry experiments and tandem mass spectrometry profiling identifications. Scale bars show normalized intensities. See Figure 2 for other definitions.

PCA discriminant analysis and potential OA markers.

Discriminant analysis was used to generate a target list for MS/MS analysis. All the spectra generated in the IMS experiments from control and OA donors were combined and peak-picked prior to PCA. Peak picking was required to reduce the data set to a size that enables the computational methods needed. Discriminant analysis was performed on the principal components that described 80% of the total variance in the data set. The remaining variance was discarded as noise. The resulting discriminant functions classified the data into 2 groups: control (group 1) and OA (group 2) (Figure 4A). Figure 4B displays the factor spectra and reveals that the peaks of the positive part of the first discriminant function were specifically from OA samples, while the peaks of the negative part were more abundant in control samples.

Figure 4.

Results of principal components analysis (PCA) and discriminant analysis. A, The spectra of all the control (group 1) and osteoarthritic (OA) (group 2) cartilage samples after matrix-assisted laser desorption ionization–imaging mass spectrometry (MALDI-IMS) experiments were analyzed using PCA and discriminant analysis to classify peptides specific for each condition. The positive part of the first discriminant function (DF1) graph corresponds to OA cartilage samples and the negative part of the graph to control cartilage samples. B, Shown is the loading plot of the DF1 representing the peptides specific for the OA condition revealed by the MALDI-IMS approach. Color figure can be viewed in the online issue, which is available at

After this analysis, the peaks with the highest absolute loadings were selected to create a list of OA-specific peptides (further information is available online at These peaks constituted prime candidates for identification through MS/MS analysis. In several cases, the signal intensity was not sufficient to perform MS/MS analysis. Interestingly, the first 4 identified masses corresponded to tryptic peptides of FN and their isotopes (further information is available online at We also identified the peak at m/z 1954.0 as a unique peptide from cartilage intermediate-layer protein 1 (CILP-1). We subsequently examined the distribution of 2 of these OA-specific peaks (m/z 2574.4 and m/z 2376.3) in the OA samples using Biomap software. This analysis confirmed their predominant localization in the deep area (further information is available online at

Using our in-house–developed ChemomeTricks software, we plotted all these OA peak scores to study their localization. The reconstruction of the image showed that the intensities were higher in the deep area of the cartilage than in the superficial area (further information is available online at The abundance progressively changed from the deep to the superficial area in the OA sample (further information is available online at, and the presence of those peptides in the control sample exhibited a pronounced decrease in comparison to the OA sample. We plotted the intensity histograms of the OA-related peaks m/z 1349.7 and m/z 2377.3 (further information is available online at The maximum intensity observed in the control samples was much lower than in the OA samples. Moreover, in the OA samples we could distinguish 2 different distributions. The first one had a lower intensity (but higher than the intensities of the control distribution) and the second one had a higher intensity, showing the heterogeneous peptide distribution that we find in OA samples.

We quantified the differences found in the distribution of 2 FN peptides with the Biomap software. The highest intensity differences in the FN peptide at m/z 1401.7 between control and OA samples were found between the deep areas. At m/z 1401.7, we found a mean ± SEM intensity of 0.16 ± 0.02 in the deep area of the control samples compared with a mean ± SEM intensity of 0.47 ± 0.08 in the deep area of the OA tissue samples (P < 0.01) (n = 10 samples) (Figure 5A). Similar differences were found in the FN peptide m/z 1349.7 between control and OA samples. The mean ± SEM intensity at m/z 1349.7 was found to be 0.24 ± 0.05 in the deep area of the control cartilage compared with 1.00 ± 0.48 in the deep area of the OA cartilage (P < 0.05) (n = 10 samples). In addition to these differences between control and OA samples, we observed distinctive intensity differences between the superficial and deep areas in OA samples of the same peaks. At m/z 1401.7, we found a mean ± SEM intensity of 0.26 ± 0.05 in the superficial area compared with 0.47 ± 0.08 in the deep area (P < 0.01) (n = 10 samples). Similarly, at m/z 1349.7, the mean ± SEM intensity was 0.62 ± 0.29 in the superficial area compared with 1.00 ± 0.48 in the deep area (P < 0.05) (n = 10 samples) (Figure 5B).

Figure 5.

Validation of FN peptide distribution in control (C) and OA cartilage. A, Biomap software was used to quantify the differences in intensity of the FN peptide m/z 1401.7 between control and OA cartilage samples. Values are the mean ± SEM (n = 10 samples). & = P < 0.01 for deep OA cartilage versus deep control cartilage and for superficial (sup) OA cartilage versus deep OA cartilage. B, Biomap software was used to quantify the differences in intensity of the FN peptide m/z 1349.7 between control and OA cartilage samples. Values are the mean ± SEM (n = 10 samples). ∗ = P < 0.05 for deep OA cartilage versus deep control cartilage and for superficial OA cartilage versus deep OA cartilage. C, Shown are representative results of hematoxylin and eosin staining and FN immunohistochemistry in control and OA cartilage samples. Original magnification × 20. D, Shown are the results of quantitation of differences in FN positivity between deep areas of control and OA cartilage samples. Values are the mean ± SEM (n = 3 samples). ∗ = P < 0.05 versus control cartilage. See Figure 2 for other definitions. Color figure can be viewed in the online issue, which is available at

Immunohistochemistry validation of the peptide distribution.

The molecular imaging analysis described in the previous sections revealed a higher intensity of OA-related peaks in the deep area of the cartilage. Immunohistochemical staining against FN was performed to orthogonally validate the results found by IMS. As shown in Figure 5C, immunohistochemistry with hematoxylin counterstaining revealed an important increase in the deep area of the OA cartilage. The data in Figure 5D represent the average of the FN-positive pixels in the deep area of the OA and control samples, confirming the differences found with the MALDI-IMS approach (mean ± SEM 0.2 ± 0.00% in control samples versus 0.71 ± 0.32% in OA samples; P < 0.05) (n = 3 samples).


In this study, we developed a novel protocol by which to identify and localize peptides/proteins in human control and OA cartilage by MALDI-IMS (17). MALDI-IMS involves the visualization of the spatial distribution of proteins, peptides, or drug compounds within thin slices of samples. It is a promising tool for putative biomarker characterization and drug development. Many investigators have already used MALDI-IMS to localize proteins, peptides, and lipids in diseased tissues (13, 18). For example, MALDI-IMS has revealed novel prognostic markers in Barrett's adenocarcinoma and intestinal gastric cancer, and it has even been used for tumor classification (11, 13, 19–22). However, nothing has been described until now in relation to rheumatic pathologies.

For the first time, we studied the distribution of peptides in healthy and OA human cartilage by MALDI-IMS. Among the proteins in healthy cartilage, we identified unique peptides of aggrecan core protein (m/z 2271.1, m/z 1315.6, m/z 2270.2), decorin (m/z 2763.3), PRELP (m/z 1590.9, m/z 1352.7), and biglycan (m/z 2027.1, m/z 1312.7, m/z 3081.6, m/z 2846.3). We showed an example of the homogeneous distribution of m/z 1590.9 from PRELP and m/z 2278.1 in healthy samples. We saw a similar distribution of PRELP in OA cartilage (data not shown). All these identified proteins interact with GAGs and build the ECM in which the chondrocytes are embedded (23), and they modulate fibril formation of collagen, especially type I collagen, preventing the mineralization of the cartilage (24, 25).

In OA samples, we also identified peptides from aggrecan core protein (m/z 819.4), biglycan (m/z 2027.2, m/z 2846.4), PRELP (m/z 1044.5, m/z 1590.9, m/z 1070.6), collagen α1(II) chain (m/z 2023.9), and protein ELYS (m/z 1025.6), as well as FM (m/z 1955.1, m/z 1361.7, m/z 767.4, m/z 978.5). Under pathologic conditions, FM can activate classical pathways of direct complement binding and plays a role in inflammation of the joint (26).

Regarding the already known OA-related proteins described in the literature, we identified different peptides of COMP in OA cartilage samples. Our results indicate a higher presence of the peptides m/z 2256.1 and m/z 1613.8 in OA versus healthy cartilage. In addition, the distribution in OA cartilage was more abundant in the deep area than in the superficial area (27). Our findings are consistent with the observation that rheumatoid arthritis (RA) patients and OA patients who have high levels of COMP in serum also have bone erosion (28). Skiöldebrand et al demonstrated that galloping horses had high levels of COMP messenger RNA expression and protein in the deep areas of osteochondral fragments from the middle carpal joint (29). Using classic proteomics such as liquid chromatography, FN or COMP can be detected in synovial fluid from controls, but in a lower concentration than in samples from OA patients (30, 31). In our approach, we did not perform any protein separation, identifying the proteins directly from the cartilage tissue.

COMP is an integral structural component of the cartilage matrix that binds to types I, II, and IX collagen and types I and II procollagen. It has been found in high levels in the cartilage, synovial fluid, and plasma of OA and RA patients (32). Only a few studies have looked for the distribution of COMP in OA tissues. By microscopy, COMP has been shown to be present in higher levels in the intercellular matrix than inside the chondrocytes (33, 34), and in rabbits with anterior cruciate ligament transection, it stains positively in the area surrounding apoptotic chondrocytes (35). COMP has been related to endochondral ossification in the developing mouse joint, exhibiting a gradient reduction to the superficial zone (36). The 2 identified peptides are located in the C-terminal end of COMP, which binds types I, II, and IX collagen and regulates fibril formation. Mutations in the C-terminal end exert a dominant-negative effect on both intracellular and extracellular processes. This ultimately affects the morphology and proliferation of growth plate chondrocytes, eventually leading to chondrodysplasia and a reduction in long bone growth. The C-terminal end of COMP is also important because it can directly bind to FN (37). COMP also bound to other proteins associated with OA (FM, CILP, matrix metalloproteinase 3, β6 integrin, and β8 integrin) (further information is available online at Thus, using this technique, the identification of peptides in specific areas of a tissue can be directly linked to the presence of the protein (8, 38).

We also identified many peptides from other OA-related proteins like FN. Our results indicate a high presence of the peptides m/z 1401.7, m/z 1593.9, m/z 1431.8, and m/z 1323.7 in OA versus healthy tissues. In addition, all these masses were more abundant in the deep area of the OA cartilage than in the superficial area. By means of Biomap software, we quantified the intensity of the differences in abundance in healthy and OA samples and in superficial and deep areas. We also performed immunohistochemistry studies that validated these results. FN is a glycoprotein that is present at low levels in the ECM of control cartilage, and its increase in OA cartilage produces a change in chondrocyte phenotype and in the activity of metalloproteinases (39). Positive staining for FN in the superficial area of OA cartilage has been described (40); however, the authors pointed out the possibility that the observed FN staining could be due to the penetration of the protein from the synovial fluid into the cartilage.

Our results show high levels of FN in the deep area of OA cartilage, and other investigators have previously shown results in this direction. FN synthesis was measured by radioimmunoassay in bovine articular explants as well as in chondrocytes showing high levels in the deep area (41). Recently, FN was found in the deep area of equine cartilage, and a clear association has been shown between the presence of FN, caspase 3, and apoptosis of chondrocytes (42). It is known that FN fragments from synovial fluid, as well as from OA cartilage, contribute to tissue remodeling, inflammation, and cell death. One hypothesis is that some of the peptides found are related to these FN fragments. However, the antibody used for the immunohistochemistry studies has affinity for the heparin-binding extra domain A segment, and it is known to be specific for cytoplasmic FN and not for FN fragments.

A high abundance of COMP and FN proteins in OA cartilage could be expected. However, we would like to underscore the fact that, to our knowledge, this is the first time these proteins have been detected and identified by MALDI-IMS in human cartilage. In addition, we observed that all of the OA markers detected were increased in the deep area of the cartilage. These results confirm that the study of different parts of the cartilage can give us valuable information for understanding the biologic processes during OA development.

Performing discriminant analysis, we showed that we could differentiate the peptide profiles of OA and control cartilage. Interestingly, the results we obtained in a training data set with 3 healthy cartilage samples and 3 OA cartilage samples were confirmed when we increased the number of samples to 10 donors per group, showing the stability of the technique. Some of the most OA-abundant peaks corresponded to FN, validating the capabilities of the technique even further. One of the OA peptides classified by discriminant analysis corresponded to CILP-1 (m/z 1954.0) (further information is available online at Our in-house–developed software localized this mass in the deep area of the cartilage and with a different intensity in OA versus healthy tissue. CILP-1 has been described as an autoantigen in OA pathology and has been said to be regulated by growth factors such as transforming growth factor β (43, 44). In addition, CILP-1 has been found in higher concentrations in the middle and deep areas of aged cartilage as compared with young cartilage (45).

We used the peptide cutter tool from the Swiss-Prot database in peaks that we fragmented but were not identified by the Mascot algorithm. These peaks matched with FN (m/z 1355.7 and m/z 1357.7), α5 integrin (m/z 2376.3, m/z 2377.3, m/z 2378.3), and CILP-2 (m/z 2574.4, m/z 2575.4). Interestingly, different members of the integrin family can bind FN and COMP proteins, demonstrating the close relationship that exists between all these proteins (46, 47). Lorenzo et al found that CILP-2 (C1 isoform) could be used to differentiate OA from RA and nondisease conditions (48). All of the known and unknown OA-related peaks identified by discriminant analysis were also localized by our in-house–developed software in the deep area of the cartilage. Biomap software confirmed this distribution, as we showed for the peptides m/z 2574.4 and m/z 2376.3.

Graphic representation of the intensity versus the number of pixels of peptides with high scores in the first discriminant function revealed that their average intensity was higher in OA samples than in healthy samples. Two different distributions represented the heterogeneity of OA cartilage. This demonstrates the capability of MALDI-IMS to classify different types of tissues according to 1) the score after discriminant analysis and 2) the representation of the intensity versus the number of pixels for a specific mass.

Some of the proteins that were identified by MS/MS were subjected to a protein interaction analysis using String software version 9.0. This resulted in a theoretical OA protein interaction map. All of the included members (COMP, FN, CILP-1, and FM) are connected to different proteins of the integrin family (β5 integrin, β3 integrin, β7 integrin, α5 integrin, α2 integrin, α3 integrin, β8 integrin, β1 integrin, α4 integrin, α2β integrin) (further information is available online at Integrins are proteins that connect the cytoskeleton to the ECM. Cell signaling mediated through integrins regulates several chondrocyte functions, including differentiation, matrix remodeling, and cell survival (49). Other investigators have shown that during OA, abnormal integrin expression alters ECM/cell signaling and modifies chondrocyte synthesis, with a subsequent imbalance of destructive cytokines (50). Future experiments to validate the identity of these OA-specific masses could reveal these proteins to be prime species for diagnosis and potential drug targets. These results suggest the potential of MALDI-IMS for the localization and identification of known as well as unknown proteins in the cartilage.

In conclusion, we have simultaneously localized and identified for the first time peptides from human healthy and OA cartilage using MALDI-IMS. The abundance of all the detected OA peptides was higher in the deep area than in the superficial area of the OA cartilage. MALDI-IMS provides a new way to study and image molecules in different areas of the cartilage and to understand molecular signaling pathways and the physiology of OA cartilage in a fast and sensitive way.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Heeren had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Cillero-Pastor, Blanco, Heeren.

Acquisition of data. Cillero-Pastor, Kiss.

Analysis and interpretation of data. Cillero-Pastor, Eijkel, Blanco, Heeren.