Mapping glycation and glycoxidation sites in collagen I of human cortical bone

Highlights

• •Advanced glycation end products (AGEs) play a role in pathogenesis of diabetic complications.
• •The information about specific locations of AGEs within protein structures is lacking.
• •We identified and mapped major physiological AGEs in collagen I of diabetic human bone.
• •The map helps in studies of pathogenic mechanisms involving AGEs and collagen I.

Abstract

Accumulation of advanced glycation end products (AGEs), particularly in long-lived extracellular matrix proteins, has been implicated in pathogenesis of diabetic complications and in aging. Knowledge about specific locations of AGEs and their precursors within protein primary structure is critical for understanding their physiological and pathophysiological impact. However, the information on specific AGE sites is lacking. Here, we identified sequence positions of four major AGEs, carboxymethyllysine, carboxyethyllysine, 5-hydro-5-methyl imidazolone, and 5-hydro-imidazolone, and an AGE precursor fructosyllysine within the triple helical region of collagen I from cortical bone of human femurs. The presented map provides a basis for site-specific quantitation of AGEs and other non-enzymatic post-translational modifications and identification of those sites affected by aging, diabetes, and other diseases such as osteoporosis; it can also help in guiding future studies of AGE impact on structure and function of collagen I in bone.

Abbreviations

AGE

advanced glycation end product

CML

carboxymethyllysine

CEL

carboxyethyllysine

MG-H1

5-hydro-5-methyl imidazolone

G-H1

5-hydro-imidazolone

FL

fructosyllysine

CMHL

carboxymethyl hydroxylysine

GdnHCl

guanidinium hydrochloride;

LC-MSMS

liquid chromatography-tandem mass spectrometry.

1. Introduction

Post-translational modification (PTM) is one of the key steps in protein biosynthesis where the individual amino acid residues are structurally modified by specific enzymes to enhance protein stability and/or functionality. An example of these enzymatic PTMs is hydroxylation of lysine and proline residues and adduction of galactose and glucose moieties to hydroxylysine residues in collagen I, catalyzed by specific hydroxylases and transferases (reviewed in [1]).

Along with enzymatic modifications, there are also non-enzymatic (NE) alterations of protein structure mainly via oxidation, nitration, deamidation, halogenation and glycation of amino acid side chains resulting in NE-PTMs (reviewed in [2]). Although NE-PTMs are formed spontaneously, they occur at preferentially targeted sites within the protein structure [3], [4], [5], [6]. The contributing factors to this site specificity may be the nature of neighboring amino acid residues, local pH and ionic strength, and the site's surface exposure [7], [8], [9]. While low levels of NE-PTMs are present in normal physiology, their continued accumulation may affect protein functionality and lead to pathology.

Advanced glycation end products (AGEs) are NE-PTMs that derive from glycation and glycoxidation pathways where glucose and/or its degradation products react with lysine, arginine, and α-amino groups of N-terminal residues in proteins ([10], [11], [12], [13] and Fig. 1). Glycation of amino groups in lipids has also been reported [14]. Accumulation of AGEs has been implicated in pathogenesis of diabetic complications, including nephropathy, retinopathy, and neurodegenerative disorders [15], [16], [17], [18], osteoporosis [19], and in aging [20]. Knowledge about specific locations of AGEs and their precursors within protein primary structure is critical for understanding their pathophysiological impact because the functional consequences of AGEs likely depend on sequence position of the modification. Depending on their location, AGEs may affect local protein conformation, stability, and assembly; interfere with specific ligand binding sites; block accessibility of cleavage sites to proteases and cause other localized functional damage. This is especially true for long-lived extracellular matrix (ECM) proteins as they can accumulate higher levels of AGEs and are not protected by the robust cellular antioxidant defenses.

Fig. 1.

Structures of an AGE precursor fructosyllysine and major AGEs. a, Fructosyllysine (FL); b, Nε-carboxymethyllysine (CML); c, Nε-carboxyethyllysine (CEL); d, 5-hydro-5-methyl-imidazolone (MG-H1); e, 5-hydro-imidazolone (G-H1). The wavy lines indicate a cut-out from the collagen I polypeptide backbone.

Type I collagen is the most abundant ECM protein found in connective tissues including skin, tendon, and bone. However, the information on the AGE sites within collagen I triple-helical structure is lacking. There were few studies addressing specific locations of glycation and/or AGE sites in collagen I. In rodent tail tendons, sites of fructosyllysine (FL), an AGE precursor, have been reported within small collagen I sequence regions using either isotopic labeling or mass spectrometry [21,22]. A single MG-H1 site has been identified in collagen I from rat aorta [23]. Studies in human bone collagen I have been focused primarily on the total AGE levels [24,25], and the location of only one carboxymethyllysine (CML) site has been reported [25].

To address the existing scarcity of knowledge about the AGE locations in the bone matrix, we identified and mapped collagen I sequence positions of four major AGEs and an AGE precursor FL (Fig. 1). The map provides i) a knowledge base for future quantitation of AGE levels at the specific sites and identification of those sites affected by disease, aging, and other factors; ii) aids in studies investigating the potential impact of AGE modifications on functional sites within collagen I triple helix; and iii) serve as a benchmark in designing the in vitro protein glycation studies to ensure physiological relevancy.

2. Materials and methods

2.1. Bone samples

Femurs were harvested from cadavers of 30- to 90-year-old males who had diabetes for a period of 10 to 40 years (N = 12) and stored at −80 °C.  Using a low speed, water-irrigated, diamond-embedded, circular cutting saw (South Bay Technology Inc, Model 660, San Clemente, CA, USA) strips of cortical bone were extracted from the medial quadrant of each femur mid-shaft at room temperature. The periosteal and endosteal surfaces of the cortical strips were milled using a custom fixture to machine a uniform, rectangular beam specimen (nominal dimensions: L = 50.1 ± 2.13 mm x B = 4.04 ± 0.15 mm x W = 5.03 ± 0.23 mm). The distal end of the beam was cut, and 2 mm thick bone samples were obtained prior to homogenization for mass spectrometry.

2.2. Extraction of bone matrix proteins

Extraction of bone matrix proteins from femur samples was carried out as previously described [4], with minor changes. Briefly, flash frozen bone samples (∼40-mg) were pulverized under liquid nitrogen using a SPEX 6770 freezer/mill (SPEX SamplePrep LLC, Metuchen, NJ). The powdered samples (30 mg) were mixed with 0.5 mL of the 10 mM Tris buffer, pH 7.5, containing 6 M guanidinium hydrochloride (GdnHCl), 50 mM EDTA and Halt's protease inhibitor cocktail. Extraction was carried out for 72 h at 4 °C under continuous rotation. Supernatant was collected following centrifugation at 13,000 g for 15 min at 4 °C and subjected to three cycles of washing with 10 mM Tris buffer using 3 KDa Amicon Ultra-4 centrifugal filtration unit to achieve complete buffer exchange. The protein concentration in the samples was determined using Pierce™ BCA protein assay kit (ThermoFisher Scientific, Waltham, MA).

2.3. Mass spectrometry (LC-MS/MS)

The site-specific modifications in bone collagen I were analyzed as described previously [4], with some changes. Sample were diluted with 100 mM ammonium bicarbonate and reduced with dithiothreitol. Cysteine residues were carbamidomethylated with iodoacetamide, and three separate aliquots from each sample were digested with either sequencing-grade trypsin (Promega), ArgC protease (Genovis) or AspN protease (EMD Millipore) for 8 h at 37 °C. Following digestion, peptide preparations were desalted, reconstituted with 0.1% formic acid, and loaded onto a capillary column packed with 20 cm of C18 reverse phase material (Jupiter, 3 μm beads, 300 Å, Phenomenex). Peptides were loaded onto this capillary, reverse phase, analytical column (360 μm O.D. × 100 μm I.D.) using a Dionex Ultimate 3000 nano-LC and autosampler. The mobile phase solvents consisted of 0.1% formic acid, 99.9% water (solvent A) and 0.1% formic acid, 99.9% acetonitrile (solvent B). Peptides were gradient-eluted at a flow rate of 350 nL/min using a 90-minute gradient. The gradient consisted of the following steps: 2%−35% B from 1 to 73 min; 35%−95% B from 73 to 78 min; constant 95% B from 78 to 79 min; 95% B back to 2% B from 79 to 80 min; constant 2% B from 80 to 90 min (column re-equilibration).

A Q Exactive Plus mass spectrometer (Thermo Scientific), equipped with a nanoelectrospray ionization source, was used to mass analyze the eluting peptides using a data-dependent method. The instrument method consisted of MS1 using an MS AGC target value of 3 × 10⁶, followed by up to 20 MS/MS scans of the most abundant ions detected in the preceding MS scan. A maximum MS/MS ion time of 100 ms was used with a MS2 AGC target of 1 × 10⁵. Higher-energy collisional dissociation was set to 27 nce, and peptide match, isotope exclusion, and dynamic exclusion (10 s) were enabled. The data are publicly available [26,27].

Database searches were performed using SEQUEST as previously described [28], and results were assembled in Scaffold v 5.1.2 with minimum filtering criteria of 95% peptide probability and 99% protein probability. Searches were configured to use variable protein residue modifications: fructosylated lysine residue (FL, ΔM=162.0528); carboxymethylated lysine residue (CML, ΔM=58.0055), carboxymethylated hydroxylysine residue (CMHL, ΔM=74.0004); carboxyethylated lysine residue (CEL, ΔM=72.0211), 5-hydro-5-methyl imidazolone adducted arginine residue (MG-H1, ΔM=54.0106), and 5-hydro-imidazolone adducted arginine residue (G-H1, ΔM=39.9949). Searches additionally included enzymatic PTMs: hydroxylated lysine and proline residues (ΔM=15.9949), galactosyl (ΔM=178.0477) and glucosylgalactosyl (ΔM=340.1006) hydroxylysine residues, as well as carbamidomethylation of cysteine (ΔM=57.0215) and oxidation of methionine (ΔM=15.9949).

All sites of modification were confirmed by manual interpretation of the raw tandem mass spectra using Xcalibur 4.0 software (Thermo Scientific). The additional criteria included manual verification of the peptide peak in the elusion profile at the minimum base peak intensity levels of 1 × 10⁵, utilizing a window of 10 ppm around the theoretical monoisotopic m/z values of the observed precursor ions to generate extracted ion chromatograms, and confirmation that a mass of the observed leftmost peak in the peptide isotopic envelope is within the 4-ppm margin of the theoretical mass of the peptide. The MS² spectra were manually analyzed to verify the peak assignments and the presence of a mass shift that matches the modification mass at the corresponding residue.

3. Results and discussion

We identified sites of major AGEs i.e., carboxymethyllysine (CML), carboxyethyllysine (CEL), 5-hydro-5-methyl-imidazolone (MG-H1), 5-hydro-imidazolone (G-H1), and an AGE precursor fructosyllysine (FL) (Fig. 1) in collagen I triple helical region using mass-spectrometry. Because the levels of these modifications are elevated in diabetes and aging [29,30], the bone matrix was isolated from 12 diabetic patients with a wide range of ages and diabetes duration (Fig. S1).

Combining MS data generated from three different digestion protocols allowed for a triple helical region sequence coverage of 94.7% for α1(I) chain and 90.3% for α2(I) chain, thus suggesting that most of the modification sites within this collagen I region have been detected and identified. The identity of the sites and the nature of modifications have been verified by manual analysis of the MS spectra. The examples of the MS² spectra with peak assignments for different modifications are shown in Fig. 2 and Figs. S2-S5. In total, 35 FL and AGE sites have been identified within the triple helical domain of collagen I, with many of these sites containing several different modifications (Table 1 and Fig. 3). CML modifications can also form on hydroxylysine (Hyl) residues and are designated in Table 1 and Fig. 3 as carboxymethyl hydroxylysine (CMHL). Twenty of 35 AGE sites were found in α1 chain and 15 in α2 chain; 17 modifications were located to lysine residues and 18 modifications were located to arginine residues (Table 1 and Fig. 3). All the FL and AGE modifications were found in each of the twelve donors at the relative abundance levels (the base peak intensity) >1 × 10⁵, with the exception of two sites within the α2 chain, CML at K⁵¹⁰ and K⁶²¹, which were identified in only 4 donors out of 12 (blue font in Fig. 3). Interestingly, these 4 donors were also the oldest donors in the group.

Fig. 2.

Identification of carboxymethyllysine modification at K177 site of the α2 chain of collagen I from human bone. Shown is the MS2 spectrum of the precursor mass of 795.9203 (+2 ion) with the retention time of 53.89 min and the base peak intensity of 4.91 × 105. Following the ArgC digest, mass spectrometry data were collected and analyzed as described under Materials and Methods. Peak assignments were confirmed by manual analysis of the spectrum (see Table S1). The asterisks indicate hydroxylated proline residues; CML, carboxymethyllysine; #, internal ions.

Table 1.

Identification of AGE-modified peptides of collagen I from diabetic human bone.

Peptide sequence	Observed precursor ions, a m/z and charge	Accuracy b, ppm	Elution time c, min	Modification structure and site d
Collagen α1 chain
PGErGPPGPQGAR	454.5574 + 3	1.5	20.39 – 20.64	MG-H1 / R244
GFSGLDGAkGDAGPAGPK	553.9376 + 3	−0.5	37.84 – 37.93	CML / K277
GFSGLDGAkGDAGPAGPK	559.2681 + 3	−1.4	37.54 – 37.67	CMHL / K277
GFSGLDGAkGDAGPAGPK	558.6087 + 3	−0.9	38.07 – 38.22	CEL / K277
GrPGAPGPAGAR	575.2920 + 2	0.7	17.43 – 17.86	MG-H1 / R312
GEPGPVGVQGPPGPAGEEGkR	687.9998 + 3	−0.5	32.25 – 32.46	CML / K468
GGPGSrGFPGADGVAGPKGPAGER	560.0242 + 4	2.5	34.18 – 34.46	MG-H1 / R493
GGPGSrGFPGADGVAGPK	557.2691 + 3	0.0	33.65 – 33.90	MG-H1 / R493
GFPGADGVAGPkGPAGER	571.9452 + 3	1.2	36.30 – 36.57	CML / K505
GFPGADGVAGPkGPAGER	576.6167 + 3	0.5	36.53 – 36.71	CEL / K505
GPAGErGSPGPAGPK	468.9008 + 3	0.9	20.10 – 20.38	MG-H1 / R511
GSPGPAGPkGSPGEAGR	523.5817 + 3	0.4	19.41 – 19.66	CML / K520
GSPGPAGPkGSPGEAGR	528.2543 + 3	1.7	19.75 – 19.96	CEL / K520
GQAGVMGFPGPkGAAGEPGKAGER	605.7950 + 4	2.8	36.54 – 36.79	FL / K586
GQAGVMGFPGPkGAAGEPGK	634.9727 + 3	1.1	40.33 – 40.53	CML / K586
AGErGVPGPPGAVGPAGK	553.9543 + 3	2.0	29.07 – 29.25	MG-H1 / R598
GAAGLPGPkGDR	425.2176 + 3	0.0	23.30 - 23.37	FL / K742
GAAGLPGPkGDR	390.5356 + 3	−1.0	24.89 – 25.13	CML / K742
GAAGLPGPkGDR	395.8670 + 3	0.5	24.63 – 24.91	CMHL / K742
GDAGPkGADGSPGKDGVR	606.9535 + 3	−0.1	17.95 −18.22	FL / K751
GDAGPkGADGSPGK	644.2934 + 2	1.2	11.77 – 12.05	CML / K751
GDAGPkGADGSPGKDGVR	582.2745 + 3	−0.3	18.73 – 19.02	CEL / K751
GDAGPkGADGSPGKDGVR	576.9443 + 3	2.1	19.34 – 19.60	CEL / K751
GDAGPKGADGSPGkDGVR	572.2713 + 3	0.2	18.96 – 19.25	CML / K759
DkGESGPSGPAGPTGARGAPG	667.6459 + 3	−3.7	23.71 – 23.90	FL / K781
GLTGPIGPPGPAGAPGDkGESGPSGPAGPTGAR	976.4772 + 3	1.4	40.36 – 40.60	CML / K781
GArGSAGPPGATGFPGAAGR	599.9592 + 3	0.2	31.31 – 31.50	MG-H1 / R865
GVVGLPGQrGER	432.2342 + 3	1.4	31.28 – 31.50	MG-H1 / R967
GErGFPGLPGPSGEPGKQGPSGASGER	663.5681 + 4	2.3	38.70 – 39.06	MG-H1 / R970
GFPGLPGPSGEPGkQGPSGASGER	806.3788 + 3	0.1	39.75 – 40.17	FL / K984
GFPGLPGPSGEPGkQGPSGASGER	811.7097 + 3	−0.7	38.65 – 39.01	FL / K984
GFPGLPGPSGEPGkQGPSGASGER	771.6985 + 3	3.0	41.40 – 41.70	CML / K984
GFPGLPGPSGEPGkQGPSGASGER	777.0294 + 3	1.9	40.34 – 40.66	CML / K984
GPAGPQGPrGDKGETGEQGDR	530.7503 + 4	1.9	21.83 – 22.08	MG-H1 / R1093
GPAGPQGPrGDKGETGEQGDR	707.9874 + 3	−3.1	26.89 – 27.20	G-H1 / R1093
GPAGPQGPRGDkGETGEQGDR	557.7607 + 4	1.6	20.74 – 21.00	FL / K1096
GPAGPQGPRGDkGETGEQGDR	531.7494 + 4	2.6	21.19 – 21.40	CML / K1096
DkGETGEQG	489.7057 + 2	3.1	10.15 – 10.21	CML / K1096
Collagen α2 chain
PGErGVVGPQGAR	450.5688 + 3	−0.2	25.14 – 25.34	MG-H1 / R156
GFPGTPGLPGFkGIR	795.9194 + 2	2.5	53.74 – 53.93	CML / K177
GFPGTPGLPGFkGIR	803.9165 + 2	2.1	52.65 – 53.01	CML / K177
GFPGTPGLPGFkGIR	536.2801 + 3	2.1	51.24 – 51.51	CMHL / K177
GFPGTPGLPGFkGIR	541.6110 + 3	0.7	51.18 – 51.42	CMHL / K177
GrVGAPGPAGAR	568.3032 + 2	2.1	20.57 – 20.78	MG-H1 / R224
GSPGSrGLPGADGR	457.2197 + 3	−0.7	24.32 – 24.46	MG-H1 / R405
NGDkGHAGLAGAR	427.8813 + 3	0.0	19.44 – 19.70	CML / K510
GLHGEFGLPGPAGPrGER	469.2388 + 4	1.5	41.50– 41.70	MG-H1 / R588
GErGPPGESGAAGPTGPIGSR	659.6523 + 3	0.2	30.91 – 31.15	MG-H1 / R591
GPSGPPGPDGNkGEPGVVGAVGTAGPSGPSGLPGER	1085.8549 + 3	−0.4	43.11 – 43.39	CML / K621
GPSGPPGPDGNkGEPGVVGAVGTAGPSGPSGLPGER	1091.1902 + 3	3.0	41.05 – 41.37	CML / K621
GAAGIPGGKGEkGEPGLR	435.9776 + 4	0.7	27.72 – 28.01	CML / K657
GAAGIPGGKGEkGEPGLR	439.9766 + 4	1.1	27.50 – 27.77	CMHL / K657
DGArGAPGAVGAPGPAGATGDR	655.3124 + 3	0.5	26.42 – 26.66	MG-H1 / R675
GAPGAVGAPGPAGATGDrGEAGAAGPAGPAGPR	927.7826 + 3	1.2	32.65 – 32.94	MG-H1 / R693
GPrGDQGPVGR	383.8639 + 3	1.0	20.57 – 20.82	MG-H1 / R822
GYPGNIGPVGAAGAPGPHGPVGPAGkHGNR	562.0823 + 5	2.0	36.17 – 36.33	CML / K974
GPAGPSGPAGkDGR	462.5601 + 3	0.2	18.73 – 19.02	FL / K1064
GPAGPSGPAGkDGR	427.8771 + 3	−1.2	19.84 – 20.10	CML / K1064
GPAGPSGPAGkDGR	432.5509 + 3	3.2	20.30 – 20.50	CEL / K1064
GPAGPSGPAGKDGrTGHPGTVGPAGIR	499.6567 + 5	0.0	30.60 – 30.90	MG-H1 / R1067

^aConfirmed by manual analysis of the MS¹ spectra.

^bMass accuracy was calculated as ((observed m/z–theoretical m/z)/theoretical m/z)x10⁶.

^cA range of LC peak elution times for the corresponding peptide over the twelve individual bone samples.

^dAmino acid position is given for full collagen α1(I) or α2(I) chain sequences (UniProtKB – P02452, CO1A1_HUMAN and P08123, CO1A2_HUMAN, respectively). Hydroxylated proline (P) or lysine (K) residues are underlined; AGE-modified lysine (k) and arginine (r) residues are indicated by a corresponding lower-case letter in bold font. FL, fructosyllysine; CML, carboxymethyllysine; CMHL, carboxymethyl hydroxylysine; CEL, carboxyethyllysine; MG-H1, 5-hydro-5-methyl-imidazolone; G-H1, 5-hydro-imidazolone.

Fig. 3.

Location of the AGE precursor (FL) and AGE sites in human type I collagen fibril. Residue numbering for α1 (I) chain is shown above the sequence and numbering for α2 (I) chain is shown below the sequence. Non-triple helical N-terminal and C-terminal telopeptides are indicated by rectangular boxes. Amino acid position numbering is given for full length collagen α1(I) or α2(I) chain sequences (UniProtKB – P02452, CO1A1_HUMAN and P08123, CO1A2_HUMAN, respectively). The triple helical sequence regions that were not detected by LC-MSMS are highlighted in gray; the overall sequence coverage of collagen I triple-helical region was 94.7% and 90.3% for α1 and α2 chains, respectively. The guide sequence numbering is shown in green font above (α1(I) chains) or below (α2(I) chain) the sequences. The AGE-modified lysine or arginine residues at the specific sites are shown in red font with the names of modification shown in the red shaded boxes above (for α1 chain) or below (for α2 chain) the site. The sites not found in all the donors are shown in a blue font. The AGE modification names are shown as follows: FL, fructosyllysine; CML, carboxymethyllysine; CMHL, carboxymethyl hydroxylysine; CEL, carboxyethyllysine; MG-H1, 5-hydro-5-methyl-imidazolone; G-H1, 5-hydro-imidazolone. Modification names separated by a slash symbol (/) indicate that multiple modifications were found at the same lysine residue. See Table 1 for a list of the identified AGE-modified collagen I peptides.

Additionally, hydroxyproline (Hyp) modifications, formed enzymatically during normal collagen I processing, have been identified within the FL- and AGE-modified peptides (Table 1). Interestingly, we observed heterogeneity in proline hydroxylation, a phenomenon that has been previously reported in collagen I and other collagens [31,32]. In the α2 chain, two forms of the CML¹⁷⁷-modified peptide were found, with and without hydroxylation at P¹⁷⁴ (Fig. S6 and Fig. 2). The base peak intensity of the former peptide was about 3-fold lower than the base peak intensity of the latter peptide (Fig. S7). Both forms of the peptide are included in Table 1. Similar heterogeneity was observed for hydroxylation of P¹⁷⁴ in the CMHL¹⁷⁷-modified and P⁶⁴² in the CML⁶²¹-modified peptides (Table 1). We also detected heterogeneity in hydroxylation of P⁷⁵⁰ within the CEL⁷⁵¹-modified peptide of the α1 chain (Fig. S8, S9 and Table 1), with both forms of this peptide having similar base peak intensities (Fig. S10). Finally, hydroxylation of P¹⁰⁹² at the X-position within the G-H1¹⁰⁹³-modified peptide and the absence of hydroxylation of P⁹⁸² at the Y-position within the MG-H1⁹⁷⁰-modified peptide has been observed (Table 1).

With the minor exceptions (see Fig. 3), the same FL and AGE modifications were found at the same collagen I residues in the samples from all twelve donors, thus suggesting that specificity of modification sites is preserved across multiple individuals. Even though glycation and glycoxidation reactions are spontaneous and do not require enzymatic catalysis, the local microenvironment of a specific residue (i.e., neighboring amino acid residues, pH, ionic strength, and surface exposure) can play an important role in its susceptibly to modification [7], [8], [9]. Our data is consistent with and is supportive of the concept of the “hot spot” regions within biological macromolecules that are particularly prone to post-translational modifications [6].

The AGE map not only reveals the structure of post-translationally modified collagen I triple helix present in diabetic bone, but also allows for predictions of potential functional impact of these modifications. Comparison of our AGE map with the previously published map of collagen I functional sites [33], revealed that AGE modifications can potentially interfere with interactions between collagen I and its binding partners. Specifically, CML, CMHL, G-H1 and MG-H1 were found in collagen I triple helix within the proposed binding sites for chaperone HSP47, cartilage oligomeric matrix protein (COMP), decorin domain receptor 2 (DDR2), dermatan sulfate proteoglycan, and fibronectin ([33] and Table 2). Our analysis suggests that modification of K¹⁷⁷ residue of the α2(I) chain to CML and CMHL (Fig. 3 and Table 1) may also affect intermolecular collagen I crosslinking ([33] and Table 2), consistent with a previous report [22]. Interestingly, MG-H1 modifications at R⁵⁸⁸ and R⁵⁹¹ of the α2(I) chain (Fig. 3 and Table 1), are located next to a known α1β1 and α2β1 integrin binding site GFPGER within the triple helix ([33] and Fig. 3), however, no modification was found at the arginine residue within the binding site itself. It is possible that this residue was protected from modification in vivo due to the binding of the integrin receptor.

Table 2.

AGE modifications located within functional sites of collagen I.

Modification structure and site a	Collagen I functional site(s) b
Collagen α1 chain
MG-H1 / R312	Heat shock protein 47 (HSP47 chaperone) binding site
MG-H1 / R493	Heat shock protein 47 (HSP47 chaperone) binding site
MG-H1 / R511	Heat shock protein 47 (HSP47 chaperone) binding site
MG-H1 / R598	Heat shock protein 47 (HSP47 chaperone) binding site
CML / K781	Cartilage oligomeric matrix protein (COMP) and dermatan sulfate proteoglycan binding sites
MG-H1 / R967	Fibronectin, discoidin domain receptor 2 (DDR2) and heat shock protein 47 (HSP47 chaperone) binding sites
MG-H1 / R970	Discoidin domain receptor 2 (DDR2) and heat shock protein 47 (HSP47 chaperone) binding sites
MG-H1 / R1093	Heat shock protein 47 (HSP47 chaperone) binding site
G-H1 / R1093	Heat shock protein 47 (HSP47 chaperone) binding site
Collagen α2 chain
MG-H1 / R156	Heat shock protein 47 (HSP47 chaperone) binding site
CML / K177	Intermolecular cross-link site
CMHL / K177	Intermolecular cross-link site
MG-H1 / R224	Heat shock protein 47 (HSP47 chaperone) binding site
MG-H1 / R405	Heat shock protein 47 (HSP47 chaperone) binding site
MG-H1 / R675	Cartilage oligomeric matrix protein (COMP) binding site
MG-H1 / R693	Cartilage oligomeric matrix protein (COMP) binding site

^aAmino acid position is given for full collagen α1(I) or α2(I) chain sequences (UniProtKB – P02452, CO1A1_HUMAN and P08123, CO1A2_HUMAN, respectively).

^bSite location is from the map of human collagen I functional sites [33].

Because the atomic structures of the identified modifications are known, the map can also serve as a starting point for molecular dynamics simulation studies. These studies can point to disruptions in local triple-helical structure due to the presence of an AGE moiety at the specific residue, including changes in triple-helix hydration, and make predictions about the impact of multiple AGE sites on overall stability of the triple-helix [4].

The presented AGE map does not include non-enzymatic cross-links such as pentosidine and glucosepane because their unknown sequence positions preclude calculations of the precursor masses. The feasibility of determination of the in vivo AGE cross-link sequence positions is yet to be demonstrated. So far, only molecular modeling predictions of glucosepane location within collagen I have been attempted [34,35]. The results were inconsistent and depended on the employed computer models [35]. Given that AGE crosslinks can potentially affect the ability of collagen fibrils to dissipate energy by sliding relative to one another [36,37], identification of the pentosidine and glucosepane sites is an important goal of the future studies and would help to determine the causes of the elevated bone fracture risk in diabetes and aging. Identification of AGE crosslink sequence positions will also be an important step towards a complete AGE map of collagen I.

In summary, we present a map of glycation and AGE sites within the triple helical region of collagen I isolated from diabetic human bone. This is the only in vivo AGE map available for any human extracellular matrix protein. To the best of our knowledge, in vivo AGE mapping has been done for only few human proteins, most notably serum albumin [38], and the majority of site-specific AGE identifications have been performed in the in vitro modified proteins [39]. Our AGE map provides a basis for future site-specific quantitation of AGE levels and identification of those sites affected by diseases (e.g., diabetes and/or osteoporosis) and other physiologic conditions such as gender and aging. The map can also help in guiding studies investigating the structural and functional impact of AGEs within collagen I triple-helix of bone.

CRediT authorship contribution statement

Paul Voziyan: Conceptualization, Investigation, Formal analysis, Writing – original draft. Sasidhar Uppuganti: Investigation, Data curation. Micheal Leser: . Kristie L. Rose: Methodology, Investigation, Writing – review & editing. Jeffry S. Nyman: Conceptualization, Funding acquisition, Writing – review & editing.

Declaration of Competing Interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

• The data have been uploaded to the Mendeley Data public database and the links to the data are included in the References.