PROTEIN CROSSLINKING BY THE WITH

ASCORBIC ACID AND GLUCOSE

by

ZHENYU DAI

Submitted in partial fulfillment of the requirements for the Degree of Doctor of Philosophy

Dissertation Advisor: Vincent M. Monnier, M. D.

Department of Biochemistry

CASE WESTERN RESERVE UNIVERSITY

August 2007

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of

______

candidate for the Ph.D. degree *.

(signed)______(chair of the committee)

______

______

______

______

______

(date) ______

*We also certify that written approval has been obtained for any proprietary material contained therein.

Table of Contents

List of Tables……………………………………………………………………………2

List of Figures…………………………………………………………………………..3

Acknowledgements……………………………………………………………………..6

List of Abbreviations…………………………………………………………………....7

Abstract………………………………………………………………………………...10

Chapter 1. Introduction……………………………………………………………….. 12

Chapter 2. Histidino-threosidine: A novel crosslink from ascorbic acid degradation products

Introduction…………………………………………………………………..49

Experimental Procedures…………………………………………………….50

Results………………………………………………………………………..55

Discussion……………………………………………………………………72

Chapter 3. Glucosepane: Site-specific crosslinking of ribonuclease A

Introduction………………………………………………………………….76

Experimental Procedures……………………………………………………83

Results……………………………………………………………………….86

Discussion…………………………………………………………………..103

Conclusions and future studies…………………………………...………………….110

Bibliography…………………………………………………………………………113

1

List of Tables

Table 1.1 The pKa comparison of several ε- and one α- in

hemoglobin……………………………………………………………….31

Table 1.2 Distribution of glycated peptides in RNase A...... 34

Table 1.3 Distribution of glucose adducts among lysine residues on α1(I)CB3 and

α2CB3-5 fragments of prepared from rats of different ages…... 37

Table 2.1 Relative yield of agents for histidino-threosidine…………….. 69

Table 3.1 Comparative levels of selected Maillard Reaction adducts and cross-links

in normal human skin collagen in the eighth decade of life……………. 79

Table 3.2 Observed double peaks or triple peaks separated by Δm/z=6 in the tryptic

map of glycated ribonuclease A…………………………………………..88

Table 3.3 Peptide maps from other enzymatic digestion. a) Asp-N digestion. b)

chymotryptic digestion …………………………………………………..101

Table 3.4 The relative surface accessibility (RSA) values of lysine and

residues in ribonuclease A …………………………………..………. 106

2

List of Figures

Fig 1.1 Proposed mechanism of pentosidine formation from D-ribose, D-fructose,

and ascorbic acid via the common intermediate…………………………...17

Fig 1.2 Proposed mechanism of pentosidine formation by Lederer et al.…………18

Fig 1.3 Structure of formulas of important α-dicarbonyl intermediates…………... 23

Fig 1.4 Mechanism of glucosepane and crossline formation from glucose via N6-(2,3-

dihydroxy-5,6-dioxohexyl)-L-lysinate intermediate compound…………..24

Fig 1.5 Structure of DODIC, MODIC, and GODIC crosslinks………………..…..25

Fig 1.6 Specific α-helix peptides models with catalytic residues positioned at specific

Sites………………………………………………………………………...30

Fig 1.7 Pathway of glycation of protein and trapping methods to discriminate the

Schiff base and the Amadori products...... 33

Fig 1.8 Catalytic mechanism of phosphate bound in a basic microenvironment on the

Amadori rearrangement…………………………………………………….35

Fig 1.9 Diagrammatic representation of the aggregated forms of the collagen

superfamily of proteins…………………………………………………… 39

Fig 1.10 Location of lysyl oxidase derived crosslinks in different types of collagen..41

Fig 1.11 Location of the divalent immature and the trivalent mature lysyl oxidase

crosslinks derived from immature crosslinks……………………..……… 43

Fig 2.1 Reversed-phase HPLC UV and fluorescence profiles for the purification of

Z-histidino-threosidine………………………………………………………56

Fig 2.2 Absorption spectra of BOC-histidino-threosidine…………………………..58

3

Fig 2.3 Electron spray ionization-MS/MS spectra of histidino-threosidine and

histidino-threosidine analog………..…………………………………….….59

Fig 2.4 1H-NMR spectra of Z-histidino-threosidine and histidino-threosidine analog.61

Fig 2.5 1H – 1H correlation spectroscopy (COSY) of histidino-threosidine analog…..63

Fig 2.6 Proposed chemical structure of histidino-threosidine analog and

histidino-threosidine………………………………………………………….65

Fig 2.7 Heteronuclear multiple bond correlation (HMBC) spectroscopy of imidazole-

Threosidine…………………………………………………………………....66

Fig 2.8 13C NMR spectrum of histidino-threosidine analog……………….…………68

Fig 2.9 Effect of time, incubation ratios pH on Z-histidino-threosidine formation from

Z-lysine, Z-histidine and threose………………………………………...……70

Fig 2.10 Detection of histidino-threosidine by LC-ESI/MS/MS analysis in bovine lens

protein incubated with threose………………………………………………..71

Fig 2.11 Proposed mechanism of formation of histidino-threosidine and dideoxysone

formation with histidine and threose……………………………….………….75

Fig 3.1 Structures of crosslinks of the Maillard reaction categorized by their

precursors …...………………………………………………………………...77

Fig 3.2 Major glycation pathways expected in nonoxidative glucose-protein

incubation……………………………………………………………………….81

Fig 3.3 Sequencing of glycated ribonuclease A tryptic peptide 40CKPVNTFVHESL

ADVQAVCSQK61 with modification of K41 by Δm/z=160 (oxidized

Amadori). ………………………………………………………………………90

Fig 3.4 Sequencing of glycated ribonuclease A tryptic peptide 36DRCKPVNTFVHESL

4

ADVQAVCSQK61…………………………………………………………….92

Fig 3.5a Sequencing of glycated ribonuclease A tryptic peptide 36DRCKPVNTFVHES

LADVQAVCSQK61 with a possible intramolecular modification by DODIC

at residues K41 and R39……………………………………………………..94

Fig 3.5b the co-elution of peptide with m/z of 2896.4 and the second peptide with m/z

of 2914.4 suggest that the peptides with m/z value of 2914.4 may have

different origins. …………………………………………………………… .95

Fig 3.6 Both tryptic and chymotryptic peptides indicate presence of intra-molecular

glucosepane between K98 and R85……………………...………………….97

Fig 3.7a Sequencing assignment of the peptide with m/z value of 4064.0…………...99

Fig 3.7b In tryptic digestion of glycated ribonuclease A, two sets of peaks point to the

inter-crosslinking between K1 and R39…………………………………….100

Fig 3.8 Distance of all nearby lysine and arginine pairs in ribonuclease A…………105

Fig 3.9 Schematic representation of crosslink formation in RNase A…….………...112

5

Acknowledgements

My most sincere appreciation goes to:

My dissertation advisor and mentor, Dr. Vincent M. Monnier, for accepting me as a lab member, constant scientific guidance and pressure.

My wife, Xiaochen Hu and my family, for their support and love.

My advisory committee, Drs. Vernon E. Anderson, Lawrence M. Sayre, Pieter deHaseth, for their interest, guidance and intriguing discussion.

My defense committee member, Dr. Steven L. Sanders and Dr. Masaru Miyagi for advice on this dissertation.

My collaborator, Ms. Ina Nemet, Dr. Wei Shen, Dr. Benlian Wang, and Dr.

Gang Sun for scientific and technical assistance.

Members of Dr. Monnier’s laboratory, for all the help on science, technology.

6

List of Abbreviations

3-DG 3-deoxyglucosone

AG aminoguanidine

AGE(s) advanced glycation end product(s)

AGOEs advanced glycoxidation end products

BOC- t-butyl oxy carbonyl

CEL Nε-(1-carboxyethyl)lysine

CML Nε-carboxymethyl-lysine

CNBr or CB cyanogen bromide

COSY correlation spectroscopy

CTGF connective tissue growth factor

DEPT distortionless enhancement by polarization transfer

DHA dehydroascorbic acid

DODIC 3-deoxyglucosone-derived imidazolium cross-link

DTPA diethylenetriamine pentaacetic Acid

GBM glomerular basement membrane

GO glyoxal

GODIC glyoxal-derived imidazolium cross-link

GOLD glyoxal-lysine dimer

GSH glutathione

Hb Ao unmodified hemoglobin

Hb A1c glycated hemoglobin

7

HHLNL histidino-hydroxylysinonorleucine

HHMD histidino-hydroxymerodesmosine deH-LNL dehydro-lysinonorleucine

HLKNL hydroxylysinoketonorleucine

HL-pyr hydroxylysyl-pyridinoline

HMBC heteronuclear multiple bond coherence

HMQC heteronuclear multiple quantum coherence

K2P 1-(5-amino-5-carboxypentyl)-4-(5-amino-5-carboxypentyl-

amino)-3-hydroxy-2, 3-dihydropyridinium

LC-MS liquid chromatography-mass spectrometry

L-pyr lysyl-pyridinoline

MFP-1 Maillard fluorescent product-1

MGO methylglyoxal

MODIC methylglyoxal-derived imidazolium cross-link

MOLD methylglyoxal-lysine dimer

MALDI matrix-assisted laser desorption and ionization

MRM multiple reaction monitoring

NMR nuclear magnetic resonance

PM pyridoxamine

PTB N-phenacylthiazolium bromide

RAGE receptor for advanced glycation end products

RNase A ribonuclease A

ROS reactive oxygen species

8

RP-HPLC reversed phase-high performance liquid chromatography

SSAO semicarbazide-sensitive amine oxidase

TFA trifluoroacetic acid

Z or CBZ benzyloxycarbonyl

9

Protein Crosslinking by the Maillard Reaction with Ascorbic Acid and Glucose

Abstract

by

ZHENYU DAI

Nonenzymatic glycation has been implicated in diabetes, aging, and aging related disease such as Alzheimer’s disease. Nonenzymatic glycation has been implicated in the pathology of normal aging and diabetes. Glycation derived AGEs, especially AGE crosslinks has been hypothesized to be involved in the pathological process.

Lens and collagen have been suggested to be major tissues with glycation mediated damage. In lens, ascorbic acid has been believed to be the major glycation agent instead of glucose. No quantitatively important ascorbic acid derived AGE crosslink has been unequivocally elucidated. I therefore hypothesized that the major ascorbic acid derived crosslinks remain to be characterized, which would have potential implication in the pathological process of aging and diabetic lens.

On the other hand, glycation mediated crosslinks to collagen have been suggested to be linked with mechanical strength changes of extracellular matrix. I hypothesized that the interfibrillar and intrafibrillar AGE crosslinks will have different impact on mechanical strength of collagen and that detailed information on glycation

10

derived crosslinking sites will be needed to decipher this relationship. As a simplified

protein model, RNAse A was used to study the AGE crosslinking sites.

In the first part of this dissertation, I isolated a novel acid-labile yellow chromophore from the incubation of lysine, histidine and D-threose. The chemical

structure, a crosslink between lysine and histidine with addition of two threose molecules, was determined by one and two-dimensional NMR spectroscopy combined with LC- tandem mass spectrometry. The compound was found in lens protein incubated with threose at concentrations above 2 mM, but was not detected in the aging human lens.

Histidino-threosidine is to our knowledge the first Maillard reaction product known to involve histidine in a crosslink.

In the second part of this dissertation, glucosepane, a major AGE crosslink in vivo, was identified following incubation of ribonuclease A with glucose. An intra-molecular glucosepane between K41 and R39 found to be the major crosslink formed in non- oxidative physiological conditions. A minor intra-molecular glucosepane was also observed between K98 and R85. The only inter-crosslink observed between K1 and R39 was an intermolecular DODIC. To our knowledge, this is the first location of intra- molecular and inter-molecular Amadori derived crosslinks in proteins.

11

Chapter One

Introduction

1.1 . The Maillard Reaction In Vivo

Almost one hundred years ago, Louis Camille Maillard reported that non-

enzymatic reaction between reducing sugars and amino acids in protein will produce

protein-protein crosslinks and a brown pigment(1). He also hypothesized this so called

“Maillard Reaction” may have application not only for food industry, but also in diabetes

mellitus. However, the link between the Maillard Reaction, diabetes and aging was not

fully established until the 1980s. Since then a fast growing area of research has developed.

The Maillard Reaction in vivo begins with a reducing sugar attaching to amines,

primarily those present in proteins, lipids, or DNA(2). A Schiff base is formed as the

initial reaction which is highly reversible. The Schiff base can go on to form more stable

Amadori products which can undergo further complex rearrangements resulting in a mixture of compounds termed advanced glycation end products (AGEs)(3-5).

The quantitative relationship between blood glucose level, accumulation of AGEs

in tissues, and the extent of tissue pathology has been extensively studied in animals as

well as in humans. Before any Maillard reaction crosslinks were chemically characterized,

AGEs-like fluorescence (excitation at 370 nm, emission at 440 nm) has been shown in

increased amounts in collagen from aged and diabetic patients(6). In retinal vessels of

diabetic rats, AGEs specific fluorescence increased 2.6 fold after 26 weeks of diabetes(7).

12

Similar increases in AGEs specific fluorescence have been found in proteins from

diabetic lens(8) and renal cortex(9). AGE accumulation was accompanied with histological evidence of diabetic tissue damage. AGE specific antibodies showed that diabetic samples contain 10-45 times more AGEs than nondiabetic samples, suggesting

that AGEs level difference between diabetic patients and normal control is much higher

than difference of blood glucose level. The senescent extracellular matrix exhibits

characteristics including decreased solubility(10), decreased proteolytic digestibility(11;

12), and accumulation of yellow and fluorescent material(13), all of which are

commonly associated with diabetes. These age-related and diabetes-enhanced changes

are thought to result in part from AGE-derived crosslinks. Clinically, changes in

collagen-rich tissues, such as arteries, lungs, and joints, have been correlated with

hypertension, emphysema, and decreased joint mobility.

Once formed, AGEs cannot be degraded until the carrier proteins themselves are

degraded. Although accumulation rates of AGEs on proteins, lipids, and DNA will

depend on the extent of modification, the most abundant AGEs are found on long-lived

proteins that have slow turnover rates. Thus lens proteins, neuronal proteins, and collagen

are prominent targets of attack by AGEs (14).

Cellular effects of AGEs mediated by the receptor for AGEs (RAGE) are

another area of AGEs research. RAGE is a multiligand type I transmembrane

glycoprotein belonging to the immunoglobulin superfamily. RAGE ligands include AGEs,

amyloid-beta peptide, high mobility group box 1(a nuclear factor that enhances

transcription and a crucial cytokine that mediates the response to infection, injury and

inflammation), and several members of the S100 protein superfamily. RAGE activation

13

triggers an intracellular signaling pathway that leads to interleukin-1β, interleukin-6 and tumor necrosis factor-α release(15; 16). The ability of AGEs to decrease proteoglycan and collagen synthesis in articular cartilage chondrocytes could be mediated by

RAGE(17; 18). For recent reviews on RAGE see references (19-21).

Collagen provides the basic mechanical characteristics of most tissues so that

many of the deleterious effects of glycation in aging and the late complications of

diabetes are mediated mainly by glycation of collagen. For example, glycation results in

changes of collagen packing density (22), surface charge(23), and stiffness(24).

Glucose is the universal glycation agent in most tissues of human body, whereas

in lens ascorbic acid has been hypothesized to be one of the most important glycation

agent(25). At the time I started my research, no major AGE crosslink derived from

ascorbic acid in the human lens had been structural characterized. Crystallins in human

lens are highly polymerized proteins and ascorbic acid derived crosslinks could lead to

changes in crystallin structure. Thus the first objective of my thesis was to identify

ascorbic acid derived crosslinks in vitro for use in human lens research, with the specific

aim focusing on the elucidation of the chemical structure of a crosslink involving lysine,

histidine and threose, i.e. an ascorbic acid degradation product. The second objective of

my thesis was to map the sites of glucose-derived crosslinks in ribonuclease A as a model

for collagen research. In collagen, crosslinks derived from lysyl oxidase have been

unequivocally proven to be involved in determining mechanical characters of immature

collagen. In contrast, the glycation derived crosslinks have been hypothesized to be

responsible for the collagen stiffening in postmature collagen, during aging and in

diabetes. The type of crosslinks formed, i.e. inter- vs. intrafibrillar crosslinks, could have

14

a totally different mechanical impact on the tissue properties. Due to the highly repetitive structure of the helical domain of the Type I collagen molecule, and the anticipated difficulty in making unequivocal structural assignments, I opted to approach the problem of site specificity using ribonuclease A, i.e. a protein that has been widely used in glycation studies. Thus the specific aim of the second part research was to identify the glycation derived crosslinking sites in ribonuclease A as a protein model.

1.2 . Crosslink Formation, Mechanism and Precursors

With the increasing recognition of the complexity of the Maillard Reaction, it was realized that glycation-derived crosslinks, although present in relatively low amounts in vivo, may have damaging effects on protein structure and function, such as increasing the mechanical strength of collagen. To address the relationship between aging- related extracellular matrix stiffness and the Maillard Reaction derived crosslinks, chemical characterization of the structure of crosslinks derived from glycation became a necessity.

Below we discuss well characterized, in vivo existing glycation-derived crosslinks with focus on their formation mechanism and common precursors. The research in this field could lead to mechanism based drug design for reversal of the effects of glycation.

1.2.1 Pentosidine

The first glycation derived crosslink was characterized in our laboratory in

1989(26). The highly fluorescent character of this compound made it easy to detect but labor intensive to purify. The initial purification of pentosidine from 600 g of dura mater

15

from elderly donors yielded only 1 mg. The intriguing characteristic of pentosidine is its in vivo precursor and its formation pathway. It was originally believed that pentoses should be pentosidine precursors according to their in vitro incubation results. But the low level of pentose in vivo questioned this notion. Further investigations revealed the order of efficacy as pentosidine precursors: ribated lysine>ribose>glucated lysine>glucose ~ fructose>ascorbic acid and dehydroascorbic acid. Higher yields at alkaline pH, and with increasing concentrations of arginine, suggested that the dehydration of the ribose-derived Amadori compound proceeds via an α-dicarbonyl intermediate which can cyclize and, under basic conditions, condense with the guanido group of arginine (Fig1.1).

16

H O H H H O H H C H C N R1 H C N R1 C C O H C OH H C OH R1-NH2 C O R1-NH2 HO C H HO C H H C OH H C OH H C OH H C OH H C OH H C OH H C OH H C OH CH2OH CH2OH CH2OH CH2OH D-ribose D-ribated lysine Glycated lysine D-glucose

O

H2N H NH O C N R2 H HN H R1 N 1) OH- H N R2 2) [O] N N O R1 Pentosidine

N OH

R1

Pentosidine intermediate 1) [O]

2) -CO2

COOH H CHO CH2OH H2C N R1 H C O HC N R1 CH2 C O O O C O C O R-NH2 H C OH H C OH -CO H C OH 2 H C OH HO C H H C OH O O CH2OH Dehydroascorbate CH2OH CH2OH 2,3-Diketo gulonic acid

CH2OH CH2 CH2OH O O C O HO C H HO OH H C OH Ascorbate H C OH CH2OH D-fructose

R1 = Lysine residue R = Arginine residue 2

Fig 1.1 Originally proposed mechanism of pentosidine formation from D-glucose, D- ribose, D-fructose, and ascorbic acid via a common intermediate(27).

17

Interestingly, work from Baynes’ laboratory on the identification of

fluorescent products of the Maillard Reaction finally showed that Maillard Fluorescent

Product-1 (MFP-1) was actually pentosidine(28). Their work added more information

about pentosidine. They suggested that glucose or Amadori adducts of glucose to proteins

could be sources of pentosidine, potentially explaining the increase of pentosidine in

diabetic patients. More recently, Lederer and colleagues(29) fully clarified the mechanism of pentosidine formation from ribose and identified pentosinane as the immediate precursor of pentosidine (Fig 1.2).

H H O 2HC N R1 R1 -2H O C 2 HO C H N HO C H -OH- CH2 H C OH C O H C OH C HO O CH OH 2 H O 6-(3-hydroxy-5-oxo-2,3,4,5-tetrahydro- D-arabinose N6-(2-hydroxy-4,5-dioxopentyl)-lysine 1-pyridiniumyl)norleucine HN H -H O+ 3 + C N R2 H N 2

R1 R1 N N N H - N -H , -H2O H H N R2 N R2 HO N N H

Pentosidine Pentosinane

R = Lysine residue 1 R = Arginine residue 2

Fig 1.2 Proposed mechanism of pentosidine formation by Lederer et al.(29)

18

However, other precursors of pentosidine, such as ascorbic acid, fructose, and other lower sugars could not be rigorously excluded. Pentoses are also possible precursors due to their relatively high in vitro activity, although their much lower in vivo

concentrations make them less likely in vivo precursors. In vitro incubation results from our laboratory suggested that oxidation was required for the formation of pentosidine from glucose, ribose, ascorbic acid, and Amadori products. Structure stoichoimetry requires 2e oxidation if starting from pentose.

Baynes’ laboratory also addressed the significance of pentosidine in tissue proteins. The level of 7 pmol pentosidine/mg collagen was lower than found in oujr laboratory due to differences in calibration. Overall, the data clearly showed that the

levels of pentosidine formed in vivo were too low to affect protein structure, function, or

turnover of collagen. Even in vitro studies with lysozyme or RNase, which were extensively modified by glucose, revealed that pentosidine could account only for a small fraction of lysine modification and less than 1% of the crosslinks in protein dimers formed during glycation(28).

Of relevance to this thesis is that Nagaraj et al.(30) discovered pentosidine in

human lens and demonstrated that is could be formed from ascorbic acid.

1.2.2 Other AGE crosslinks

The aging of the eye lens is associated with yellow discoloration and non-

tryptophan fluorescence(31-33). Fluorescent AGEs derived from the Maillard Reaction

have been found to have spectral properties similar to those present in cataractous

lenses(34; 35). Furthermore, an increase in protein-bound non-tryptophan fluorescence

19

and a concomitant increase in high molecular weight proteins in diabetic(36; 37) and

galactosemic(38) cataracts indicate a role of reducing sugars in protein damage in

cataractogenesis.

Studies dedicated to the characterization of glycation-related fluorescent

molecules from our laboratory lead to the finding of LM-1 in the human lens(39).

There is no solid data on where LM-1 comes from in human lens. But ascorbic acid and

its degradation products could be a likely source.

The inability of glucose and methylglyoxal to act as LM-1 precursors made LM-1 a unique marker for metabolic pathways different from those of the existing in vivo

Maillard Reaction, i.e. the glycoxidation markers (pentosidine), methylglyoxal derived

markers (methylglyoxal-lysine dimer (MOLD) and argpyrimidine), and N ε-

(carboxymethyl)lysine (CML).

LM-1 was either non-detectable or present in very low amounts in other human

tissues. It thus acts as a novel Maillard Reaction marker of human lens aging that may

specifically reflect carbonyl stress by ascorbic acid in diabetic lens.

MOLD and glyoxal-lysine dimer (GOLD) are additional examples of crosslinks

that were first discovered in vitro and later found to also exist in vivo. MOLD and GOLD

were originally characterized from the reaction of methylglyoxal (MGO) or glyoxal (GO)

with hippuryllysine respectively by Baynes’ lab(40; 41). Nagaraj and colleagues detected

and measured MOLD in human serum proteins by reverse phase high performance liquid

chromatography (RP-HPLC) and found that MOLD is elevated in diabetes(42). Later,

20

Baynes and coworkers quantified MOLD and GOLD in human lens protein by liquid

chromatography/mass spectrometry (LC/MS)(43).

Glycation precursors include not only glucose, fructose, ribose, or ascorbic acid, but also some dicarbonyl compounds such as GO, MGO, and deoxyglucosone. These

intermediate compounds can originate from different sources in vivo. Nagaraj’s laboratory reported mean values of 313 pmol MOLD/ mg diabetic human serum protein, and 261 pmol MOLD/mg normal samples, although there was a considerable overlap.

Baynes and colleagues continued their work on the quantitation of

MOLD/GOLD adducts by LC/MS. MOLD and GOLD together account only for 0.1% chemical modifications of lysine in a senescent lens (around 0.8 mmol MOLD and 0.2 mmol GOLD/ mol of lysine at age 80). But MOLD and GOLD are present at much higher level in lens protein than pentosidine (0.004 mmol/mol lysine in lens protein) (28).

In collagen, MOLD was measured at 0.04 mol/mol collagen. In comparison to the level of enzymatic formed crosslinks (1-5 mol/mol collagen), it is unlikely that MOLD and

GOLD will have a major impact on collagen crosslinking in aging.

1.2.3 Glucosepane

Of particular relevance to this thesis are the studies on the discovery of glucosepane. In

2002, Biemel et al. presented novel data on the discovery of a major AGE named glucosepane(44). Before that, MOLD and GOLD represented quantitatively the most important in vivo crosslinks. In order to detect and quantify AGEs in vivo, most laboratories applied conventional acid hydrolysis to the protein samples in order to achieve total digestion of the protein. However, the AGEs found by this method needed

21

to be stable enough to survive 6 N HCl hydrolysis. Indeed Eble et al. suggested that the

major crosslinks are labile under conventional acid and alkaline hydrolysis conditions(45).

MGO and GO, the respective MOLD and GOLD precursors, are present in low

µM concentrations in the plasma of diabetics(46). At these levels of α-dicarbonyl

compound, model experiments showed that formation of MOLD and GOLD is a minor

process(47; 48). Thus physiological importance of MOLD and GOLD could be

overestimated.

Biemel et al. founds that glucosepane was present in human serum albumin (HSA)

at 13.1-19.8 pmol/mg HSA, and was elevated up to 42.3 pmol/mg HSA in diabetics. The

level is high considering the half-life of HSA (12-20 days). Not surprisingly, the authors

also detected higher levels of glucosepane (132.3-241.7 pmol/mg of protein) in the long

lived human lens proteins. Interestingly, although MOLD and GOLD have been found in

brunescent lenses, with respective concentrations of 180 and 18 pmol/mg of protein(47;

49), Biemel et al. could not detect MOLD and GOLD in either HSA or in human lens

protein.

The rationale for the high level of glucosepane in vivo has been suggested to be

related with the fact that the precursor of glucosepane N 6-(2,3-dihydroxy-5,6-

dioxohexyl)-L-lysinate (Fig 1.3) is irreversibly incorporated into a protein. Thus this

precursor is sterically protected from enzymatic detoxification(44). On the contrary, 3-

deoxyglucosone (3-DG), MGO, and GO exist in an equilibrium between free and protein-

bound form. Both NADPH-dependent aldose reductase and aldehyde reductase can convert 2-oxoaldehydes to monocarbonyl derivatives: 3-DG can be detoxified to 3- deoxy-D-fructose(50); MGO and GO also can be disposed by these enzymes and the

22

glyoxalase system((51). These enzymes are distributed throughout the body and readily

detoxify free dicarbonyl precursors of AGEs, but those protein-bound dicarbonyls can

hardly be accessed by these enzymes. Thus protein-bound N 6-(2,3-dihydroxy-5,6-

dioxohexyl)-L-lysinate represents a persistent glycation agent.

H 2HC N R1 H O C H C OH H O C O H O HO C H C CH2 C CH2 C O H C OH C O C H C OH CH H O C 3 H O CH2OH

N6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate 3-DG MGO GO

R1 = Lysine residue

Fig 1.3 Structural formulae of important α-dicarbonyl intermediates(44)

In this paper, the authors suggested that generation of glucosepane proceeds very likely through the intramolecular aldimine 6-(3,4-dihydroxy-6-oxo-3,4,5,6-tetrahydro-

2H-azepinium-1-yl)-norleucine ( Fig 1.4). Additionally, the authors believe that the precursor dideoxyosone could be a primary source of other Maillard products. As an example, the authors showed a possible formation pathway of crossline via the same precursor N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate. In this case, this precursor

forms a six- instead of the seven-member ring (Fig 1.4b).

23

a) H H O 2HC N R1 C H C OH H C OH HO R1 HO C H - N + R -NH -OH HO C H 1 2 CH2 H C OH -2H O 2 C O HO H C OH C O 6-(3,4-dihydroxy-6-oxo-3,4,5,6-tetrahydro- CH2OH H O 2H-azepinium-1-yl)-norleucine

6 HN D-glucose N -(2,3-dihydroxy-5,6-dioxohexyl)- H -H O+ L-lysinate 3 + C N R2 H2N

R1 R1 H N N N N H HO H HO N R N R2 2 N N HO H HO

Glucosepane

R1 = Lysine residue R2= Arginine residue

H b) 2HC N R H O 1 OH C H C OH OH H C OH - + R1-NH2 HO C H -OH HO C H CH O -2H O 2 H C OH 2 C N C O H C OH C H R1 CH2OH H O 6-(3-hydroxy-5-oxo-2,3,4,5-tetrahydro-1- N6-(2,3-dihydroxy-5,6- pyridiniumyl)norleucine D-glucose dioxohexyl)-L-lysinate OH OH

-H O OH 2 HO -OH- + O NH R1

OH OH OH OH

-H+ OH OH HO OH HO OH H H OH OH

N N R1 N R1 N R1 R1 Crossline Fig. 1.4. Mechanism of glucosepane and crossline formation from glucose via N 6-(2,3- dihydroxy-5,6-dioxohexyl)-L-lysinate intermediate compound according to Biemel and colleagues(44).

24

In the same paper, Biemel et al. also found and measured other dicarbonyl- derived AGEs crosslinks in vivo (Fig 1.5). 3-deoxyglucosone-derived imidazolium cross- link(DODIC), methylglyoxal-derived imidazolium cross-link(MODIC), and glyoxal- derived imidazolium cross-link(GODIC), which have been found in high yields from model reactions of proteins with 3-DG, MGO, and GO. DODIC is present in human lens protein only in minute quantity (1.3-8.0 pmol/mg protein), but it is the first 3-DG-derived crosslink detected in vivo. MODIC was measured at an intermediate level in human lens protein (40.7-97.2 pmol/mg protein). While DODIC and MODIC did not increase in brunescent lenses, GODIC showed enormous elevation in brunescent lens. The authors attribute this character to the fact that GODIC belongs to the subclass of AGEs, advanced glycoxidation end products (AGOEs), which require oxygen for their formation. Thus, the elevated level of GODIC supports the hypothesis that the brunescent lens is linked to enhanced oxidative stress.

H R1 N N H R N H H N R2 1 N R1 N N H N H H N R2 N R2 H2C H H N N H C OH H H3C H C OH

CH2OH

DOGDIC GODIC MODIC

R1 = Lysine residue R2= Arginine residue

Fig. 1.5. Structure of DODIC, MODIC, and GODIC

25

While Biemel et al. stated that the level of 250 pmol/mg protein identified glucosepane as a major AGEs crosslink in human lenses, similar conclusions were reached by Monnier’s laboratory in human collagen in which the glucosepane level increased with age up to 2 nmol/mg of collagen in non-diabetic individuals and further increased to 4.3 nmol/mg of collagen in some diabetic patients(52). This result raises important questions concerning the role of glucosepane in the human body as 4.3 nmol/mg represents 1-1.2% (mol/mol) of modification in arginine and lysine, i.e. about one glucosepane molecule for every two triple helical collagen molecules in diabetic individuals. For comparison, the estimated number of pentosidine crosslinks is ten times less in cartilage collagen(53), i.e. one pentosidine for every 20 collagen molecules. The significance of glucosepane level in collagen can be better understood when considering the level of the lysyl oxidase-derived physiological cross-links(1-5 mol/mol of collagen in skin)(54).

The higher levels of glucosepane in collagen compared to lens may be in part related to the lower glucose concentration in lens than the extracellular matrix. Another mechanism is that MGO in lens crystallins may be competing for the same modification sites as glucosepane. Several facts support these propositions: 1) Because MGO is generated primarily from intracellular metabolism(55), MGO would be expected to be lower in extracellular matrix (collagen), 2) the level of the MGO-derived crosslink MODIC has been found to be lower in human skin collagen versus lens, i.e. 0.38 versus 0.8 mmol/mol lysine at ages 85 and 70 years, respectively(43). At similar ages, the MGO-derived hydroimidazolone arginine adduct, MG-H1, has been detected at a level of 14 nmol/mg of protein(56). Another lysine-lysine crosslink named K2P reached 400 pmol/mg human

26

lens protein and up to 1400 pmol/ mg human lens protein in one brunescent lens(57).

Thus glucosepane could play a relatively minor role in human lens, but still be a major

crosslink in collagen. Nevertheless, since only 60-80% of old human collagen could be

enzymatically digested, glucosepane level could be even higher than detected.

1.3 Site specificity of glycation

It has been established that not all protein sites are glycated to the same extent.

This uneven modification of protein may have implications on the function of the protein.

Whether the glycation hot spots are located in the active center of enzymes or cellular

receptors is of particular interest to glycation researchers. Glycation derived crosslinks,

although present in relatively smaller amounts than simple AGE adducts, may have more

damaging effects on protein structure and function. In particular, crosslinking sites of

collagen by glycation may have effects on the mechanical characteristics of the

extracellular matrix, which are known to be affected in aging and diabetes.

Scanty research data on crosslinking sites by non-enzymatic glycation are

available. Because it is relevant to this thesis, I am discussing below studies on Amadori

product formation sites in different proteins as these may identify the determining factors on crosslinking specificity. I will emphasize the related studies on RNAse A since I will this protein as a model in our search for crosslinking sites.

1.3.1 Glycation Specificity of Hemoglobin A and human serum albumin

A naturally existing minor human hemoglobin, Hb AIC, has been discovered as

the product of condensation between glucose and the N-terminus of the β chain of the

27

major hemoglobin, Hb A0.(58-61). Hb AIC comprises about 4% of total hemoglobin in

normal red blood cells. Bunn and colleagues did pioneering work determining the

glycation sites of hemoglobin. They reported that an additional 8-10% of Hb A0 is

glycated at other amino groups, notably the ε-amino groups of lysines(62).

In 1980, Shapiro and Bunn determined the glycation sites(63) of hemoglobin in vivo as the order of β-Val-1, β-Lys-66, α-Lys-61, β-Lys-17, and α-Val-1. Interestingly, the prevalent glycation sites identified in vitro were not in the same order as those found in vivo finding: β-Val-1, α-Lys-16, β-Lys-66, β-Lys-17, α-Val-1, α-Lys-7, and β-Lys-120.

This paper demonstrated that glycation preferentially occurs at specific lysine residues at the N-terminus of proteins. Among twenty-two ε –amino groups of hemoglobin, only three groups are preferentially glycated in vivo, while six groups are glycated in vitro. The authors suggested that pKa of any individual amino group is not good predictors for reactivity of amino groups with glucose. The authors also believed that the relative surface accessibility of lysines cannot determine the glycation hot spots because most lysines have similar exposure to free glucose except for some conformational constraints. Instead they attributed the factors affecting the Amadori rearrangement as the glycation hot spots determining factors. Interestingly, all “hot” lysines residues for glycation found in this paper are located near carboxylate groups(64).

However, many unmodified lysines are also located near a carboxylate group. The authors explain this discrepancy by the orientation of these carboxylate groups in order to favor catalysis.

Acharya et al. further addressed the relative roles of two factors which may determine glycation specificity: the ease with which the amino groups can form the

28

aldimine adducts and the microenvironments of the aldimine which may or may not

facilitate the Amadori rearrangement(65). The reactivity of amino groups of hemoglobin

A toward reductive glycation (i.e., propensity for aldimine formation) is decreased in the

order as β-Val-1, α-Val-1, β-Lys-66, α-Lys-61, α-Lys-16. But the overall reactivity of hemoglobin A towards nonreductive glycation decreases in the order as α-Lys-16, β-Val-

1, β-Lys-66, β-Lys-82, α-Lys-61, α-Val-1. The authors suggested that the discrepancy of

the reactivity reflects the propensity of the microenvironment to facilitate the Amadori

rearrangement in nonenzymatic glycation.

The catalytic activity of the “microenvironment” may depend on the amino acid

sequence (nearest-neighbor linear effects) or the higher order structure

(tertiary/quaternary) of the protein (nearest-neighbor three-dimensional effects). To discriminate the relative importance of these two possibilities, Acharya et al. measured the Amadori rearrangement of α-Lys-16 respectively in the segment 1-30 of the isolated

α chain and the tetramer of hemoglobin A(66) and found that the higher order structure is important in catalyze the Amadori rearrangement. A more recent paper by Venkataraman et al. applied a novel but simplified idea to address the catalytic effects of proximal residues on the Amadori rearrangement(67). They designed specific α-helix peptides

(named KD) with catalytic residues positioned at specific sites (Fig 1.6). By comparing

KD2 with KD4, the authors concluded that the proximity of Asp10 to Lys6 might increase the pKa of the lysine’s ε-amino group so that initial Schiff base formation may be hindered for KD2. Conversely, Asp10 will catalyze the subsequent Amadori rearrangement in KD4 whereas no such effects would be present in KD2. Histidine has also been suggested as an acid-base catalyst near physiological pH(68-70). Thus, the

29

authors also determined its effect on the glycation of a peptide with histidine in the place of aspartic acid and showed that histidine can also catalyze the Amadori rearrangement, although less efficiently than aspartic acid. Within the peptide RKD4, the flanking by strongly basic arginine ameliorated the effect of aspartic acid on the pKa of Lys6, resulting in rapid accumulation of aldimine adducts (Schiff bases). The catalytic effect of aspartic acid on the Amadori rearrangement in RKD4 was similar to the effect of KD4.

Asp 10 Asp 10 Asp 10 Lys 8

Lys 6 Lys 6

Arg 3

KD4 KD2 RKD4

Fig 1.6 Spatial juxtaposition of Lys6 (the glycation site) and the catalytic aspartic acid in a peptide helix. Lys6 and Asp10 are proximate in KD4, while Lys6 and Asp8 are dramatically opposed in KD2. In RKD4, Arg3 and Asp10 lie on the same helix face as

Lys6(67). Resulsts showed that nearby basic residues will increase Schiff base accumulation on lysine residues whereas nearby acidic residues will favor the Amadori rearrangement.

Garlick and Mazer pioneered the investigation of glycation specificity on human serum albumin in vivo(71). They found that fifty percent of Amadori products in glycated

30

human serum albumin occur at lysine residue 525 and suggested nearby lysine residues

could be a factor favoring glycation. The same rule can be applied in human hemoglobin.

Again, the pKa of lysine was not a good predictor of glycation specificity based on comparison of the pKa of β-66, α-61, β-17 and the α-amino group at the N terminus of the α-chain in hemoglobin (Table 1.1).

Glycation hot spots Other glycation sites β-66 α-61 β-17 α-amino group of N- terminus (α-chain) pKa 11.0 10.8 11.5 7.2

Table 1.1 The pKa comparison of several ε-lysines and one α-lysine in hemoglobin(63;

72).

In vitro incubations of human and rat serum albumin with glucose showed some

indirect evidence of lys-199 as a glycation site(73; 74). The low pKa of 8.0-8.7 of lysine-

199(75; 76), which is significantly lower than in other ε-lysines, is believed to be a factor

which will favor the initial condensation of amino acids with glucose.

Another paper by Iberg et al. identified minor glycation sites in human serum

albumin besides confirmation of the major glycation site(77). Lys-439, Lys-199, Lys-281,

Lys-525 were unequivocally identified, whereas Lys-233, Lys-317, Lys-12, and Lys-534

were less reliably identified. The authors also considered catalysis of the Amadori

rearrangement as an explanation of the observed glycation specificity. The carboxylic

groups have been suggested as acid catalysts(71; 78), but the authors do not believe that

the fully ionized carboxyl group can donate a proton. Instead, they suggested that

appropriately located positively charged amino acids could afford local acid-base

catalysis. For evidence, the authors showed three glycation hot spots are located in a

31

sequence of basic amino acids: Lys-525 in a Lys-Lys sequence, Lys-439 in a Lys-His

sequence, and Lys-534 in a Lys-His-Lys sequence. Furthermore, Lys-199, Lys-281, Lys-

317, and Lys-439 are close to disulfide bridges, which place a remote positive amino acid close to these sites.

Lapolla et al. used mass spectrometry to detect glycated peptides of human serum albumin(79). They identified Lys-233, Lys-276, Lys-378, Lys-545, and Lys-525 as predominant glycation sites in vitro, which have been shown to conform to the lysine solvent accessible surface by computational modeling. Actually, these results are quite different from the previous research besides the major glycation site, Lys-525. The different sensitivity of individual glycated peptides to detection by mass spectrometry may explain the difference. The heavily in vitro glycated human serum albumin may also

exhibit different glycation specificity compared to the mild glycation environment in vivo.

1.3.2 Glycation Specificity on Ribonuclease A

Of particular relevance to this thesis is the work from Baynes et al. who pioneered research on ribonuclease A(RNase A) glycation specificity(78). They applied different protocols to discriminate between the Schiff base and Amadori product (Fig 1.7).

32

H O C HC N Protein

H C OH + NH2-Protein H C OH HO C H HO C H H C OH H C OH H C OH H C OH

CH2OH CH2OH

Glucose Schiff base (Aldimine)

H 2HC N Protein C O

HO C H NaBH3CN

H C OH

H C OH

CH2OH NaBH4 Amadori products (Ketoamine)

H H 2HC N Protein 2HC N Protein HO C H H C OH

HO C H + HO C H

H C OH H C OH

H C OH H C OH

CH OH CH2OH 2 Nε-mannitollysine Nε-glucitollysine

Fig 1.7 Pathway of glycation of protein and trapping methods(78). The Schiff base

adducts can be trapped by reduction with cyanoborohydride to N ε-glucitollysine, while the ketoamine is reduced by borohydride to a mixture of N ε-glucitollysine and N ε- mannitollysine.

The authors identified the glycation hot spots of RNase A by isolation of glycated peptides and subsequent amino acid analysis. Lys-1, 7, 37, and 41 were identified as preferential glycation sites and Lys-1 reacts at both its α- and ε-amino

33

groups. The pKa seems to affect the initial Schiff base formation because Nα-Lys-1, due

to its low pKa, generates more Schiff base than Nε-Lys-1. By comparing the initial Schiff

base distribution of Lys-1(α:ε=86:14) with its final Amadori adduct distribution

(α:ε=61:39), the Amadori rearrangement is clearly demonstrated as another factor that

influences the overall glycation rates. The authors provided the Schiff base and the

Amadori adducts distribution map of all four glycation sites and proved that both initial

Schiff base formation and the following Amadori rearrangement both contribute to

glycation sites preference (Table 1.2)(78).

Lysine modified Percentage of Schiff base Percentage of the

A B Amadori product

1 82 66 24

37 3 7 9 7 6 14 29

41 9 14 38

Column A and B represent protein labeled with 0.6 and 1.5 mol of Glc/mol of RNAse A.

Table 1.2. Distribution of glycated peptides in RNase A(78).

As evidence to prove the idea of Shapiro et al.(63), the authors noticed that both

Lys-1 and Lys-37 are adjacent to acidic amino acids, Glu-2 and Asp-38, which may be

involved in local catalysis of the Amadori rearrangement. In addition, the authors

indicated that Lys-7 and Lys-41 are located in the active site of RNase. Either or both of

these two residues can be involved in enzymatic activity(80-82). Combining the data on

hemoglobin and RNase A together, the authors found the common features in

environment of the most reactive site of hemoglobin and RNase A. Thus, Lys-41 in

RNase A and β-Val-1 are both located in high affinity binding sites for phosphate ion or

34

organic phosphates. The authors observed that the enhanced reactivity of Lys-41 was dependent on the concentration of phosphate ion in the incubation buffer as an evidence for their hypothesis. This hypothesis could also explain the higher reactivity of β-Val-1 than α-Val-1 in that only β-Val-1 is located adjacent to a phosphate binding site, a basic region with lysine, histidine and arginine residues. The glycation preference for lysines at the active site can have profound effects on protein allosteric sensitivity(83), ligand binding(84; 85), receptor recognition(86-88), and enzymatic activity.

Watkins et al. continued their research on the effect of phosphate on the specificity of protein glycation(89). Phosphate has been known to enhance the overall glycation rate(90; 91), increase the proportion of open chain reducing sugar in solution(92), and promote the enolization of sugars in solution(93). The authors provided a mechanism of catalysis of the Amadori rearrangement by phosphate (Fig 1.8). Lysine residues in Lys-Lys or other basic sequences have been identified as glycation hot spots in hemoglobin(63) and albumin(71; 77), whereby the basic sequence may affect glycation through binding of phosphate, bicarbonate, or other buffering ions.

Protein Protein Protein

Lys Lys Lys

NH NH NH

CH2 CH O O O CH O O O P C O P P H C OH C O H O O O O H O O HO C H

HO C H HO C H H C OH

H C OH H C OH H C OH

CH2OH H C OH H C OH

CH OH 2 CH2OH

Schiff base Eneaminol Ketoamine

Fig 1.8 Catalytic mechanism of phosphate bound in a basic microenvironment on the

Amadori rearrangement (89).

35

1.3.3 Glycation Specificity of Collagen

Although our long term aim is to detect the AGE crosslinking sites in collagen that are hypothesized to play a key role in connective tissue aging, research on collagen glycation specificity is very limited, in part due to the special structure of collagen. Reiser et al. pioneered the research on collagen glycation specificity(94). They found that the majority of Amadori products are detected only at specific lysine residues and that this specificity is well conserved during aging. They focused their study on two specific cyanogen bromide (CNBr, CB) digested peptides, CB3 and α2CB3-5. Of the 5 lysines in

CB3, Lys-434 is the favored glycation site. Of the 18 lysines and 1 hydroxylysine residue in α2CB3-5, Lys453, Lys-479, and Lys-924 represent more than 80% of the glycation adducts. With previous glycation specificity research in hemoglobin, albumin, RNase, among others, the authors also found acidic or basic residues close to these hot spots.

Lys-434 is the only lysine in CB3 which is immediately adjacent to an acidic residue

(Asp435). Lys-924 in α2CB3-5 is adjacent to an Asp residue (Asp923), and 3 Asp residues in position 926 of all three α chains. The microenvironment of Lys-453 has either basic or acidic residues: Arg-453 in the α1(I) chain, Glu-452 in both the α1(I) and

α2 chains, and Glu-455 in all three chains. The adjacent positive or negative residues near

Lys-479 could not be proven because the rat α2 chain had not been sequenced at that time.

After comparison of the glycation specificity at different ages of rat tendon collagen

(Table 1.3), the authors interestingly concluded that glycation specificity is well conserved during aging. Age-related changes included a loss of order in fibrillar array, shortening and thickening of the collagen bundle, increased irregularity, and increased

36

interfibrillar space(95). Such structural changes could influence the microenvironment of

selected lysine or hydroxylysine residues and consequently the glycation specificity. The

observation from Reiser et al. suggests that the primary sequence of collagen may be a

major determinant of the glycation sites, whereas the higher order of collagen is not. To

prove this hypothesis, the authors incubated the purified CB3 peptides with glucose with

the presumption that CB3 exists as a random coil in the glucose solution. They found that

even without high order structure, CB3 still conserved the glycation specificity observed

in intact collagen fibrils, thus proving the relative unimportance of higher order structure of collagen in determining the glycation sites.

A

Age (months) Lysine residue 531 434 408 479

6 18 71 1 10

18 17.4 68 4 10.5 36 23.5 71 3 2.5

B

Age Lysine residue (months) 453 479 974 924 Mixture

6 29 23.5 8.5 27.5 11.5 18 26.5 23 13.5 27.5 9.5

36 32.5 28.5 9.5 21.5 8

Table 1.3 Distribution of glucose adducts among lysine residues on (A) α1(I)CB3 and (B)

α2CB3-5 prepared from rats collagen of different ages(94). Data are expressed as

percentage of the total glucose adducts on the α1(I)CB3 and α2CB3-5 peptide.

37

Le Pape et al. also provided some indirect evidence suggesting the existence of

preferred glycation sites in collagen(96). These investigators analyzed the incorporation

rate of 14C glucose by normal and diabetic rat tail tendon collagen. The tendon collagen

from diabetic rats incorporated considerably less glucose than that of normal rats. The

authors suggested that the preferred glycation sites were blocked in diabetic rats.

1.4 Crosslinking and mechanical strength of collagen

During aging, the changes in mechanical properties of the collagen fibers are

reflected in the increased stiffness of skin, tendon, bone and joints. The fibers become

more rigid and brittle (97-99). These changes are deleterious to the locomotive function of the skeletal system, the elastic property of the vascular system, and the filtration property of the basement membrane. The increased stiffness of collagen was also reflected in its resistance to pepsin digestion(100).

Several factors affect the mechanical strength of collagen. The total collagen content of the tissue could be an important factor. The mechanical characteristics of the

various types of collagen are apparently different such that the composition ratio of the

different types of collagen will affect the overall mechanical characteristics in the

extracellular matrix(101) (Figure 1.9). The ratio of elastin to collagen content is believed

to be a determining factor of the intrinsic mechanical properties of the arterial wall.

However, it has been demonstrated that the absolute and relative concentrations of elastin

and collagen in the vessel wall are unaltered in both diabetic rats and diabetic

patients(100). Interestingly, but not surprisingly, the diameter of collagen fibers also

plays a significant role in determining mechanical strength of the collagen. The plastic

38

deformation of the tissue is directly related to small diameter fibers, whereas the ability to withstand high stress levels is related to large diameter fibers(102). The difference in diameter of collagen fibers (200 nm in tendon, 100 nm in skin, 50 nm in cartilage, 20 nm in cornea) can be related to their mechanical properties. The crosslinking effect on collagen’s mechanical properties has been proposed for a long time(103; 104). We now know that two distinct mechanisms of collagen fiber crosslinking are involved. One is a precise enzymatically controlled crosslinking of initially “immature” divalent crosslinks which undergo spontaneous transformation into mature multivalent crosslinks. The second mechanism is glycation-derived crosslinking.

(a) Fibrillar

Type I, II and III

(b) Non-fibrillar collagens

Type IV

7S domain

Fig 1.9 Schematic representation of the common fibrillar and non-fibrillar collagens. (a)

Types I, II and III possess uniform triple helical molecules, which form a quarter- staggered end-overlap macromolecular structure as a basis of the fibre structure. (b) The

39

type IV molecule is a longer more flexible triple helical molecule due to imperfections in

the helical structure (not shown) and aggregates through the N- and C-terminal non-

helical domains to produce a network structure as a basis for the basement membrane

structure(105).

1.4.1 Enzymatic Intermolecular Crosslinks

Enzymatic collagen crosslinking is initiated with oxidative deamination of a lysine or hydroxylysine (Hyl) residue from telopeptides (N- or C terminal of collagen,

which is more flexible compared with triple helical region in the middle of collagen) of collagen which will generate lysine derived aldehydes. Lysyl oxidase only acts on telopeptides when it binds to the highly conserved sequence (Hyl-Gly-His-Arg)(106).

Inhibition of this enzyme by copper deficiency and lathyrism has long been known to lead to extreme fragility of collagen containing tissues due to deficient crosslinking.

Several immature enzymatic crosslinks have been reported. If lysyl oxidase acts on lysine of collagen telopeptides as in skin and rat tail tendon, aldimine crosslinks predominate.

Aldimine crosslinks are effective crosslinks under physiological conditions, but they are readily cleaved at acid pH and elevated temperature. The second group of immature crosslinks is keto-imines when lysine oxidase acts on hydroxylated lysines. Keto-imines are stable to acid and heat which accounts for the insolubility of fetal bone and cartilage collagen. Another aldol-derived tetravalent crosslink, histidino-hydroxymerodesmosine

(HHMD), has been suggested to have a significant defect on the mechanical properties of collagen fibers. But its in vivo existence is still controversial(107; 108).

40

The locations of the divalent crosslinks have been identified for most of the crosslinks in fibrous collagen type I(109), type II(110), and type III(111). The crosslinking involve one deaminated lysine or hydroxylysine in the telopeptides (residues

9N and 16C) and one lysine or hydroxylysine in the helix (residues 87 and 930). The latter sites have the conserved sequence Hyl-Gly-His-Arg-Gly (Fig 1.10).

(a) -Hyl-Gly-His-Arg-

(b)

-Hyl-Gly-His-Arg-

Figure 1.10 Location of crosslinks in collagen a) Head to tail placement in fibrous

collagen. b) Head to tail antiparallel placement in type IV collagen of basement

membrane(112).

The existence of mature crosslinks derived from labile precursors is evidenced by

the fact that the reducible crosslinks (immature crosslinks) decline with age in skin

despite an increase in strength(113). Several mature crosslinks have been reported. The

41

immature crosslink, aldimine dehydro-hydroxylysino-norleucine (de-HLNL),

spontaneously reacts with histidine to form histidino-hydroxylysinonorleucine (HHL)

which has been found only in skin and cornea(114; 115). Hydroxylysyl-pyridinoline (HL-

Pyr) was identified as a common crosslink in a wide variety of tissues(116). It is formed

by two hydroxylysyl-aldehyde residues with a helical hydroxylysine. Similarly, lysyl-

pyridinoline (L-pyr) is formed by two hydroxylysyl-aldehyde residues with a helical

lysine which is found primarily in calcified tissues(117; 118). The pyridinium crosslinks are derived from the immature keto-imine crosslink, hydroxylysine-5-ketonorleucine

(HLKNL). The relatively low level of pyridinoline (1 per 5 collagen molecules) apparently could not explain higher insolubility and acid resistance in bones compared to cartilage which could be explained by the presence of additional crosslinks. Pyrrole crosslinks have been suggested to be a ubiquitous trivalent collagen crosslink(119).

Although this crosslink has not been chemically characterized due to the difficulty of its isolation, it could form from the reaction of a telopeptidyl lysine derived aldehyde and the immature crosslink HLKNL. Kleter et al. reported another modified pyrrole crosslink, pyrroleninone(120). The in vivo significance of these two pyrrole crosslinks still needs to be elucidated.

The location of mature crosslinks in collagen is critical for the mechanical properties of the tissue. More collagen molecules needed to be crosslinked into the original divalent immature crosslinks to add additional mechanical strength to collagen if the primary divalent crosslinks are present intra-fibril (Fig 1.11a). The Eyre and Henkel groups separately proposed that the crosslinks involve three molecules whereas other groups believed that crosslinking only two molecules was needed for the increase for the

42

mechanical strength(118; 121-123). The divalent immature crosslinks form at the 4D

stagger (see Fig 11) of the aligned collagen molecules, thus the additional lysine derived aldehyde or hydroxylysine derived aldehyde in mature crosslinks can only come from additional collagen molecules in parallel alignment. Bailey et al. proposed that another lysine derived aldehyde or hydroxylysine derived aldehyde involved in mature crosslinks

may come from collagen molecules on different microfibrils(124) (Fig 11). Such trans-

microfibre crosslinks obviously would increase the mechanical strength of collagen.

4D stagger (a)

(b)

Fig 1.11a) Location of the interfibrillar divalent immature crosslinks in immature collagen fibers aligned in the quarter-staggered arrangement. b) Location of the trivalent interfibrillar mature crosslinks derived from immature crosslinks(112).

1.4.2 Non-enzymatic Crosslinking

43

Glycation can affect the properties of collagen in many ways including altering its charge profile and hence its interaction with cells. But the most damaging effects of glycation are believed to result from the formation of glycation derived inter- molecular crosslinks.

Due to the special structure of collagen fibres, the site specificity of glycation- related inter-molecular crosslinks is likely to affect the mechanical strength of collagen.

The studies on the specificity are usually based on CB digestion and the results are controversial. Tanaka reported that ribose can attach to all the CB peptides whereas Le

Pape et al. showed that glucose will attach mainly to αICB6, and more specifically Reiser et al. suggested the preferential glycation occurs on particular lysine residues in α1CB3 and α2CB3,5(94; 125; 126). Interestingly, Wess et al. provided evidence from neutron scattering for preferential glycation sites existing in the gap region of the fibre(127). No precise glycation derived crosslinking sites have been suggested for collagen.

1.5 Blocking AGE Formation and Its in vivo Relevance.

Inhibitors of AGE formation (AGE inhibitors) and breakers of existing AGE

(AGE breakers) have been intensively studied in recent years due to their potential medical application. The application of AGE inhibitors and AGE breakers in animal models and preclinical studies provided indirect evidence supporting the relevance of

AGEs in complications of aging and diabetes.

1.5.1 Mechanism of AGE Inhibitors and AGE Breakers.

44

Aminoguanidine(AG) was the first proposed AGE inhibitor(128) and also the

first to enter clinical trials(129). A second AGE inhibitor, pyridoxamine (PM), has been

tested in clinical trials and showed significant activity in prevention of as well as

intervention against diabetic nephropathy(130; 131). In 1999, Vasan et al. reported the

first AGE-crosslink breaker, phenylthiazolium bromide (PTB)(132). Because of the

unstable nature of PTB in physiological buffers, a more stable analog of PTB, ALT-711

has been tested for the ability to reverse the AGE-derived crosslinks.

The inhibition of advanced glycation by AG is primarily related to its α, β- dicarbonyl scavenging activity. AG has other pharmacological activities, such as inhibition of nitric oxide synthase and semicarbazide-sensitive amine oxidase (SSAO), at in vivo concentrations that are required for anti-glycation activity(133; 134). AG also has antioxidant activity. Clinical trials of AG had to be terminated either because of toxicity or because the primary endpoint (creatinine doubling time) was not reached.

Three major mechanisms have been proposed to explain the effect of PM on glycation. First, PM inhibited the post-Amadori steps in the Maillard Reaction is involved(135). The second mechanism of PM inhibition of glycation is scavenging of toxic carbonyl compounds. Although carbonyl species are present in vivo only in low quantity, they could be one of the major sources of AGEs due to their reactivity. PM has been demonstrated to scavenge carbonyl species such as glyoxal(136), methylglyoxal(137), glycolaldehyde(136), and 1,4-dicarbonyls(138). In the Zucker rat model of diabetes and hyperlipidemia, PM treatment significantly decreased the plasma level of glyoxal and methylglyoxal and the AGEs in collagen(139; 140). The third major mechanism of PM protection of protein damage by glycation is by scavenging ROS from

45

autoxidation of glucose or oxidative degradation of Amadori intermediates. PM can

potentially reduce oxygen radicals via the hydrogen-donating activity of its phenolic

group(141-143).

The exact mechanism of crosslink breaking by PTB is not clearly understood.

The thiazolium structure could chemically break α-dicarbonyl compounds by cleaving

the carbon-carbon bond between the carbonyls as originally proposed by Vasan et

al.(132). This mechanism precludes the action of crosslink breakers on major currently

known AGE crosslink structures such as glucosepane, MOLD, GOLD, vesperlysines, etc

since none of these structures have functional groups that can be attacked by PTB. Nor

has it ever been demonstrated that chemical crosslinks formed in vivo are actually broken.

Indeed Baynes and colleagues reported that AGE breakers cleave model AGE crosslinks,

but fail to break AGEs crosslink in skin and tail collagen from diabetic rats as indicated

by pepsin and acid solubility(144). With the report describing the potent metal-chelating

activity of AGE breakers and their hydrolysis products(145), Baynes et al. concluded that

the beneficial effects of AGE breakers may not stem from the cleavage of preexisting

AGEs crosslinks, instead, their antioxidant, chelating, and possible enzyme-inhibitory

effects may have more direct effects on AGE formation through scavenging of MGO and

other dicarbonyl intermediates. Thus these purposed AGE breakers may function largely

as AGE inhibitors, whereby normal collagen turnover would explain the observed

decrease in collagen crosslinking with improvement in vascular elasticity and compliance.

Thus the relationship between supposed AGEs breaking functions and beneficial effects

of AGEs breaker needs to be further studied.

46

1.5.2 Effects of AGE Inhibitors and AGE Breakers on Mechanical Characteristics of

Collagen-Rich Tissues

Wolffenbuttel et al. reported the beneficial effects of AGE-breaker, ALT-711, on arterial elasticity(146). In this study, treatment with ALT-711 increased systemic arterial compliance by 25% and reduced characteristic input impedance of the aorta by

35%, both parameters indicated reduced vascular stiffness. This finding is comparable with the preventive effects of aminoguanidine in diabetes(147). Oral administration of

ALT-711 to aged dogs resulted in significant improvements in end-diastolic and stroke volume index, decreased left ventricular stiffness, and improvement in cardiac function(148). Administration of ALT-711 in aged monkeys lead to a decrease in pulse wave velocity and aortic stiffness(149). In human patients over the age of 50 with stiff cardiovasculature, ALT-711 treatment induced a significant reduction of arterial pulse pressure and an increase in large artery compliance(150). After application of ALT-711 on the skin of aged rats, the rats skin elasticity was significantly improved(151).

The effects of AG have been well established, but how these beneficial effects are correlated with AGE crosslink inhibition requires further evaluation. In diabetic rats,

AG attenuated the aortic accumulation of total collagen, especially collagen I, III, and IV and the increased expression of profibrotic cytokines TGF-β and Connective Tissue

Growth Factor (CTGF)(152-154). In STZ (Streptozotocin)-diabetic baboon and Otsuka

Long Evans Tokushima fatty rats, treatment of AG prevented glomerular basement membrane (GBM) thickening(155; 156). Oral administration of AG to Fisher 344 rats from 24-30 months of age prevented end-life increases in arterial stiffness and cardiac hypertrophy without affecting the total collagen and elastin content of the arterial

47

wall(157). This result suggests a reduction of AGEs-induced crosslinking in the extracellular matrix protein of the arterial wall. Sell and Monnier reported that AG administration to Fisher 344 rats from 6 months to 24 months of age moderately decreased tail tendon break time without affecting glycation or glycoxidation as reflected by pentosidine and CML levels(158).

Pyridoxamine prevented age-related increases in systolic blood pressure and mean arterial pressure in the Zucker rat, and inhibited the thickening of the aortic wall and that of smaller arteries in the heart and kidney(140). Pyridoxamine has a direct lipid lowering effect in vivo. Pyridoxamine administration lowered advanced lipoxidation products (ALEs) and AGEs such as CML and CEL without affecting Amadori products or pentosidine(130). Thus, pyridoxamine’s beneficial effects on the arterial wall may directly depend on its lipid lowering effects or inhibition of crosslinking via lipid peroxidation products. Indeed pyridoxamine adducts of lipoxidation intermediates, such as N -hexanoyl-pyridoxamine and N -nonanedioyl monoamide, have been detected in vitro and in pyridoxamine treated animals(159).

48

Chapter Two

Isolation, Purification and Characterization of Histidino-Threosidine, a Novel

Maillard Reaction Protein Crosslink from Threose, Lysine and Histidine

2.1 INTRODUCTION

Besides collagen, lens crystallins are important glycation targets. Post- synthetic modifications and crosslinks accumulate with age, decreasing solubility of lens crystallins. These processes are greatly accelerated during cataractogenesis, and can become extreme in brunescent cataracts. Unequivocal evidence now points to a major role of AGEs in the age-related pigmentation and crosslinking of human lens crystallins(160-162).

Ascorbic acid is involved in the formation and maintenance of collagen, possibly due its function as lysyl oxidase cofactor. Because collagen is the basis of connective tissues, ascorbic acid is needed for maintaining healthy blood vessels, skin, ligaments, cartilage, joint linings, and bones. Ascorbic acid also helps thyroid hormone production by acting as a cofactor of the critical enzymes in the hormone synthesis pathway. Ascorbic acid is also well known for its antioxidant character. However, ascorbic acid may act as an important source of glycation damage in human lens. The factor behind this controversy phenomenon is that ascorbic acid degradation products, such as threose, are responsible for the ascorbic acid damage(163). Fan X et al have proven that the elevated intralenticular ascorbic acid levels in mouse accelerated the aging process(164). Several facts support the importance of ascorbylation in lens: (1) ascorbic acid (ASA) concentration in the lens is higher than in most other tissues, i.e. 1 to

49

3 mM(165; 166); (2) ascorbic acid through its degradation products such as dehydroascorbic acid (DHA), , erythrose, and threose forms protein adducts and cross-links at a much higher rate (~70 fold) compared with glucose and fructose(167),

(3) peptide maps from ascorbylated lenses resemble closely those found in cataract

lens(160). Unfortunately, except for K2P and vesperlysine A, which cannot be formed

from glucose, protein modifications such as carboxymethyl-lysine, pentosidine, and

carboxyethyl-lysine are not specific for ascorbic acid, leaving a gap in our understanding

of the chemistry of ascorbic acid mediated crosslinking. I hypothesize that major ascorbic

acid derived AGE crosslinks of yet unknown structure could be important in glycation

derived damage in human lens. As described below the systematic search for ascorbic

acid-derived crosslinks in a model system lead to the discovery of a novel crosslink

involving lysine and histidine, given the common name histidino-threosidine.

2.2 EXPERIMENTAL PROCEDURES

2.2.1 Reagents

Reagents of the highest quality available were obtained from Sigma (St Louis,

MO, U.S.A.), unless indicated otherwise. Deionized water (18.2 MΩ cm-1) was used for all experiments. D-Threose was purchased from Omicron Biochemicals (South Bend,

Indiana, U.S.A.). Solvents for NMR experiments were purchased from Norell

(Landisville, NJ, U.S.A.). LC-18 reversed phase Superclean solid phase extraction (SPE)

tubes were purchased from Supelco (Sigma). CBZ-protected amino acids were from

Fluka/Sigma (St Louis, MO, U.S.A.).

50

2.2.2 Incubation of Z-Lysine, Z-Histidine with D-Threose

Z-Lysine (700 mg, 2.5 mmol), Z-histidine (722.5 mg, 2.5 mmol) and D-threose

(2.5 mmol) were dissolved in 100 mM sodium phosphate buffer (50 ml, pH 7.4) with 1 mM diethylenetriaminepentaacetic acid (DTPA). This mixture was filtered through a 0.22

o µm filter unit (Millipore), N2 bubbled for 30 seconds and incubated under N2 at 37 C for

3 weeks.

2.2.3 Preliminary Purification by SPE

Incubation samples (25 ml) were applied to LC-18 SPE tubes (SUPELCO, 60 g).

The SPE tubes were washed with 50 ml methanol (1 % TFA) and equilibrated with 50 ml water (1% TFA) before use. The incubation sample was passed through the tube and the

LC-18 SPE tube was washed with 100 ml water containing 1% TFA. The dark brown material was then eluted with 50 ml methanol containing 1% TFA.

2.2.4 High-performance Liquid Chromatography (HPLC)

The methanol eluate was injected into a preparative C-18 reversed-phase column

(Vydac 218TP1022, 22 x 250 mm, 10 µm; The Separations Group, Hesperia, CA). A

Waters HPLC (Waters Chromatography Div., Milford, MA) with Model 510 pumps, automatic injector (model 712 WISP), and a model 680 controller was used. The column was eluted at a flow rate of 5 ml/min with 12 % acetonitrile for 10 min, and developed with a gradient of acetonitrile (12 %-42 %) for 40 min. The column eluate was monitored with an online absorbance detector (Waters 484 tunable absorbance detector) and a fluorescence detector (Waters 470 scanning fluorescence detector) at 335 and 385

51

nm for the excitation and the emission, respectively. The chromatograms were recorded with chromatography software (Azur, France). Fractions which had m/z 720 (determined by LC/MS direct infusion) were collected by an online fraction-collector (FRAC-100,

Pharmacia Biotech) and freeze-dried. This fraction was redissolved in methanol and then further purified with two columns. The samples were first injected into a C-18 reverse- phase semi-preparative column (Vydac 218TP1010, 10 x 250 mm, 10 µm, The

Separations Group) using the same solvents and gradient as before, but with a lower flow rate: 3 ml/min. The fractions with m/z 720 were collected, freeze-dried, dissolved in methanol, and injected into a C-18 reverse-phase analytical column (Vydac 218 TP104,

4.6 x 250 mm, 10 µm; The Separations Group, Hesperia, CA) with the same gradient at a flow rate of 1 ml/min. After HPLC purification, the fractions were freeze-dried and stored at -80 oC.

2.2.5 In vitro Incubation of Z-Lysine and Z-Histidine with Sugars

Glucose, ribose, fructose, erythrulose, glyceraldehyde, methylglyoxal, DHA, threose, erythrose, glycolaldehyde were incubated with 10 mM of Z-lysine and Z- histidine in 50 ml 100 mM sodium phosphate buffer with 1 mM DTPA (pH 7.4). After 14 days at 37 oC, 100ul of each sample was injected into a C-18 analytical column with same gradient and solvents indicated above.

2.2.6 Spectroscopy

Absorption spectra were recorded with a Hewlett-Packard 8452A diode array spectrophotometer connected to an IBM PC/AT computer (Hewlett-Packard, Inc.,

Avondale, PA; IBM Corp., Boca Raton, FL) The sample of Z-histidino-threosidine for

52

proton NMR spectroscopy was exchanged three times with methanol-d4 under a nitrogen

atmosphere. The sample, dissolved in 400 µl of 100 % methanol-d4, was transferred to a

5-mm NMR tube and 1H-NMR spectrum acquired at 25 oC with a Varian Inova 600 MHz

NMR spectrometer. The samples of Z-histidino-threosidine for 13C-NMR, DEPT, COSY,

HMQC and HMBC were exchanged three times with dimethylsulfoxide-d6, transferred to

o a 5-mm NMR tube and scanned overnight at 25 C with the same NMR spectrometer,

with the assistance of Mr. Dale Ray of the Case Structural Biology Center. The sample of

histidino-threosidine for NMR spectroscopy was exchanged three times with D2O. The

NMR spectrum of histidino-threosidine was acquired with the same parameters as the

sample for Z-histidino-threosidine. LC-MS and LC-MS/MS were performed using a 2690

separation module with a Quattro Ultima triple quadrupole mass spectrometer (Waters-

Micromass, Manchester, U.K.). High resolution mass spectrometry was obtained on a

Kratos MS-25Ainstrument from Case Chemistry Department with the help of Mr. Jim

Faulk.

2.2.7 Determination of Histidino-threosidine in Bovine Lens Protein Incubated with

Threose

Bovine lenses (Pel-Freez Biologicals, Rogers, AR) were pooled, decapsulated,

homogenized in water (2.5 ml/lens) in a Con-Torgue glass homogenizer (Eberbach

Corporation, Ann Arbor, MI), and separated by centrifugation at 20,000×g. The pellet

was washed twice with water (2.5 ml/lens) followed by centrifugation. The three

supernatant fractions (water-soluble fractions) were pooled and extensively dialyzed

against water for 2 days and freeze-dried.

53

The water-soluble human lens protein was redissolved in 50 mg/ml 100 mM

sodium phosphate buffer with 1 mM DTPA (pH 7.4), with 0, 2, 10, 50 mM threose added

individually, and then incubated at 37oC for 2 weeks. After incubation, samples were dialyzed against water for 2 days and freeze-dried. The freeze-dried pellet was

sequentially digested at 37 oC for 24 h intervals by the addition of each of the following enzymes in PBS: (A) 0.12 U of peptidase (P7500, Sigma) /5 mg substrate; (B) 1.05 U of pronase E (165921, Roche) /5 mg substrate; and (D) 0.2 U of aminopeptidase M (102768,

Roche). Chloroform and toluene (0.15%, v:v) were added as antimicrobial agents. The resulting enzyme digested protein was analyzed by LC-MS/MS system. An Atlantis® dC-

18 column (2.1 x 50 mm, 3 µm; Waters, Milford, MA, U.S.A.) was used. The mobile

phase was 0.1 % TFA in water (Burdick & Jackson, Muskegon, MI, U.S.A.). The flow

rate was 0.2 ml/min (washing column with 90 % of acetonitrile (Burdick&Jakson,

Muskegon, MI, U.S.A.) and column re-equilibration was performed between every injection). The products were detected by positive electrospray ionization-mass spectrometric multiple reaction monitoring. The ionization source temperature was 130

°C and the desolvation gas temperature 400 °C. The cone gas and desolvation gas-flow rates were 850 and 150 l/h, respectively. The capillary voltage was 3.60 kV and the cone voltage 35 V. Argon gas was in the collision cell. The collision energies for m/z 84.00

and 279.20 fragment ions were 40 and 20 eV, respectively. Programmed molecular ion

and fragment ion m/z values were optimized to ± 0.1 Da for multiple-reaction-monitoring

detection of the analyte.

54

2.3. RESULTS

In preliminary studies, I carried out a number of systematic incubations

consisting of Z-lysine, Z-Arginine, and Z-histidine incubated with dehydroascobic acid in

100 mM Na/PO4. The mixture was analyzed by HPLC and revealed prominent peaks in the lys-his sytem leading us to choose this system for further analysis. I choose threose, a

proposed DHA degradation product, for further analysis as described below.

2.3.1 Isolation and Purification of of a Candidate Z-protected Crosslink with m/z

value of 720 from the Incubation of Z-lysine, Z-histidine and D,L-threose

After 20 days of incubation of 50 mM each of Z-lysine, Z-histidine, and threose in

0.1 M sodium phosphate buffer at 37 oC, the incubation mixture was applied to LC-18

reversed phase Superclean SPE tubes and eluted with methanol. This step removed

unreacted sugar and enriched the hydrophobic fraction of the incubation sample. The

methanol eluate was further fractionated by preparative reverse phase HPLC and

monitored with UV detection at 257 nm (specific for the Z-group) and fluorescence

detection at 335/385 nm (characteristic of the known fluorescent crosslink, pentosidine).

Fractions eluting after Z-lysine were collected and assayed by MS for m/z values

exceding 569 (i.e. the sum of the molecular weights of Z-lysine + Z-histidine) (Fig 2.1).

55

A Z-lysine-threosidine720

B

Minutes

Fig 2.1. Reversed-phase HPLC chromatograms showing the purification of Z- histidino-threosidine from incubation of Z-lysine, Z-histidine and D-threose by preparative column. (A) UV absorption profile at 257 nm (B) Fluorescence profile with excitation and emission maxima at 335 nm and 385 nm, respectively(168).

A peak eluting around 46 min exhibited m/z 720, which strongly suggested the presence of a potential crosslinked Z-amino acid. This peak was collected, pooled, freeze-dried and further purified on an analytical C-18 column. After purification, the Z- histidino-threosidine was subjected to LC-MS and MS/MS analyses (Data not shown).

56

MS spectra revealed a monoprotonated molecular ion peak with m/z 720, indicating a

molecular mass of 719. Based on the molecular mass of 719.2803 obtained by high-

resolution mass spectrometry, the empirical formula C36H41O11N5 was assigned to this

molecule.

2.3.2 Evidence for the Involvement of Histidine in the Crosslink

The imidazole group of histidine is known for its activity as a general acid/base

catalyst as well as its nucleophilicity(169; 170). Thus histidine might act as a catalyst without being a constituent in the final product. To determine whether the isolated

chromophore is a lysine-histidine crosslink or a lysine-lysine crosslink (with the

preliminary 1H NMR data, at least one lysine is involved in the crosslink), we incubated

Z-lysine, threose, and imidazole (chosen as a histidine analog) under identical conditions.

The incubation samples were analyzed by LC/MS. I found a compound with m/z 499, which corresponds to a Z-lysine-imidazole structure. If the imidazole group only acted as a catalyst and the compound was a lysine-lysine crosslink, the m/z should be again 720.

Since no molecule exhibits this m/z in the incubation mixtures with imidazole, the

imidazole group of histidine thus did not act as a catalyst in the formation of a putative

lysine-lysine cross-link, but was implicated as a component of the crosslinked molecule.

2.3.3 Structure Elucidation of Histidino-threosidine

For further characterization of the putitive crosslink, we prepared an analog from

BOC-lysine, BOC-histidine and D-threose. The BOC-compound was similarly purified by HPLC and the molecular weight was confirmed with mass spectrometry. Fig 2.2

57

shows the absorption spectrum of the BOC- protected histidino-threosidine which

exhibits a maximum absorption at 305 nm.

2

1.8

1.6

1.4

1.2

1

0.8 Aabsorbance

0.6

0.4

0.2

0 200 209 218 227 236 245 254 263 272 281 290 299 308 317 326 335 344 353 362 371 380 389 398 407 416 Wavelength (nm)

Fig 2.2 Absorption spectra of BOC-histidino-threosidine isolated from incubation of

BOC-lysine and BOC-histidine with D-threose(168).

The crosslink was deprotected by treatment with ice cold TFA for 2 hours. The

resulting compound was analyzed by MS and MS/MS. A major peak with m/z 452 was

obtained, indicating a molecular mass of 451 Da (not shown). The interpretation of the

MS/MS fragmentation pattern of this compound is shown in Fig 2.3A. The fragments

were assigned individually to the structure.

58

x10A 452.052 100

% 278.865 434.151 279.117 433.710 156.222 279.243 217.270 372.002 250.897 360.027 452.998 0 m/z 150 200 250 300 350 400 450 500

B x2 292.065 100 205.835 206.087

178.075

% 177.824 147.118 150.201 248.154 274.045 223.970 175.747 119.761 297.610 0 m/z 100 120 140 160 180 200 220 240 260 280 300

Fig 2.3 Electrospray ionization-MS/MS spectra of (A) histidino-threosidine and (B) analog of histidino-threosidine. For histidine-threosidine MS/MS was performed on the precursor m/z 452. Major fragments are 434 (loss of H2O), 416 (loss of 2H2O), 407 (loss

of COOH), 278.9 (loss of H2O and histidine), 250.9 (loss of COOH and histidine), 189

(loss of CHOHCH2OH, COOH and histidine), 162 (loss of CHO, CHOHCH2OH, COOH,

histidine), 156 (histidine). For histidino-threosidine analog, MS/MS was performed on

the precursor ion with m/z 292. Major fragments are 274 (loss of H2O), 263.8 (loss of

CHO), 248 (loss of CH3CH2CH2 ), 206 (loss of OH and imidazole ring), 195 (loss of

CHO and imidazole ring), 178 (loss of CHO, OH and imidazole ring), 150 (loss of –CH2- imidazole and –CHOHCH2OH)(168).

59

The determination of the structure of the crosslink using 1H NMR spectrum of Z-

histidino-threosidine presented a problem because the large proton signals of the Z group

(around 7.2 ppm) were potentially masking important aromatic protons with similar 1H

NMR chemical shifts. Therefore we used imidazole as a histidine analog and butylamine as a lysine analog to obtain NMR spectra. The incubation conditions and purification process were similar as before. The compound had the predicted m/z value of 292. The

MS/MS fragmentation pattern of this histidino-threosidine analog is shown in Fig 2.3B.

As shown in Fig 2-4, the proton spectrum of histidino-threosidine analog conserved all the important signals in Z- histidino-threosidine.

60

4',5', 5H from Z water methanol A

2H from Z

ε 5 9 α' 6 α 2 7

H-8 H-δ B HO HA HA OH Hβ Hα Hβ Hα

OH OH

H-5 H-6

H-7 H-α H-4'

H-5' H-2' H-β H-γ H-9

Fig 2.4 1H-NMR spectra of (A) Z-histidino-threosidine and (B) imidazo-threo-

butylamine. Spectrum A was acquired in methanol-d4, while spectrum B was acquired in

61

D2O. Inset shows resonance of H-8 and H-9α,β (please refer to Fig 2.6 about H

number)(168).

As shown in Fig 2.4B, the signal at 9.8 ppm suggests an aldehyde proton. Its

correlated carbon is around 179 ppm which further supports an aldehyde structure. The proton signals around 8.5, 7.25, and 7.3 ppm can be assigned to the three protons on the imidazole ring. The other 7.2 ppm proton indicates another ring in the cross-link. The shape of the proton resonance at 5.5 ppm suggests that it is derived from two mutually correlated protons on the same carbon. The HMBC spectrum shows this proton resonance to be correlated with 4 or 5 aromatic carbons, suggesting that it’s a –CH2 group between

two aromatic rings. The protons at 4.7 and 3.5 ppm are strongly correlated according to

the COSY spectrum (Fig 2.5). Their correlated carbons are at 66 and 65 ppm, consistent

with the presence of a –CHOH-CH2OH group in the structure. The splitting pattern of the

protons at 3.5 ppm is interesting (Fig 2.4B). Hα is first split by Hβ and then by HA. In the same way, Hβ is first split by Hα and then by HA. Thus in R and S conformation, 8 splits are formed. Small additional splitting may be due to long range coupling. Other proton

signals at 4.2, 1.6, 1.1, and 0.7 ppm can be assigned by the protons at alphabetic chain of

butylamine.

62

H-8

H-9

H-9

H-8

Fig 2.5 1H – 1H correlation spectroscopy (COSY) of imidazo-threo-butylamine. The cross signals between H-8 and H-9 are aligned in the panel(168).

63

Based on this information, a tentative structure for imidazo-threo-butylamine

emerged (Fig 2.6). The newly formed ring should have –CHO, -CH2-, -CHOH-CH2OH, and -H as its side chains. Assuming the new ring involves a 4 carbon backbone from threose, the pyrrolic structure shown in Fig 2.6 is proposed, except that the position of –

CHO and –CH2- can be exchanged. Exchangeability also exists between –H and -CHOH-

CH2OH position.

64

δ

A γ 1’ β N 5’ α 2’ H 5 N N 4’ 1 3’ 2 7 9 4 8 3 HO 6 OH O

H2N B α COOH

β γ

1’ β’ δ N COOH ε 2’ 5’ α’ H 3’ N 5 N 4’ H2N 1

2 7 9 4 8 3 HO 6 OH O

Fig 2.6 Possible chemical structures of imidazo-threo-butylamine (A) and histidino- threosidine (B)(168).

65

To further characterize the structure of the histidino-threosidine analog imidazo- threo-butylamine, the exact position of –CHO, -CH2-, -CHOH-CH2OH, and –H needed to be determined. The HMBC spectrum gave us much information (Fig 2.7)

C-2 C-5 C-3,C-4 C-5' C-2′ e C-4'd

H-9 H-ε H-8 H-7

H-5

H-4' H-5' H-2' H-6

Fig 2.7 Heteronuclear multiple bond correlation (HMBC) spectroscopy of imidazo-threo- butylamine (168).

With the information obtained from 1H and HMQC NMR spectra, the peak at

4.1 ppm was assigned to H-α as shown in Fig 2-6. In HMBC spectrum (Fig 2.7), this proton correlates with two carbons in the aromatic region (129 and 130 ppm) suggesting

66

these two carbons at the position 5 and 2 as shown in Fig 2.6. In addition, the carbon at

130 ppm showed residual one-bond C-H coupling, visible as satellite doublets centered

on the proton at 7.1 ppm. According to these data, the C-atom resonance at 129 ppm was

assigned to C-2, while the resonance at 130 ppm to C-5 and the 7.1 ppm resonance in the

1H NMR spectrum as H-5. In addition, the HMBC spectrum also showed correlation of

C-5 with H-α and H-8, which supports that the C-atom at position 4 links to the CHOH-

CH2OH side chain. Correlation of C-2 in the HMBC spectrum with H-α, H-7, H-5 and H-

6 is compatible with the proposed structure. Residual one bond correlation of C-atom at

135 ppm with the proton resonance at 8.5 ppm can be assigned to position 2' in the imidazole ring. This C-atom also showed correlation with H-7 confirming the presence of

13 a CH2 group between two rings in the structure. Similarly, C NMR chemical shifts at

122 and 120 ppm can be assigned to C-4' and C-5'. The signal at 124.5 ppm was assigned

to C-4 in the newly formed pyrrole ring and correlation of C-4 with H-8, H-9 and H-5 is

compatible with the proposed structure. Interestingly, the signal at 124.5 ppm also

correlated with H-6 which is four bonds away from position C-4. Finally, it was realized

that the signal at 124.5 ppm represents overlapping of C-4 and C-2 signals and according

to that, correlation between the C-atom at 124.5 ppm and proton at 9.6 ppm can be

explained as correlation between C-2 and H-6. With all of the information above, the

side chains at position 2 and 3 still can be exchangeable. Fig 2.8 shows the assignment of

the 13C NMR spectrum to imidazo-threo-butylamine, i.e. the proposed histidino-

threosidine analog structure.

67

3,4 4' 5 9 2 8 δ γ β 5' ε 6 2' 7

Fig 2.8 13C NMR spectrum of histidino-threosidine analog(168). The assignments

correspond to the numbering shown in Fig 2.6A.

To find which sugar or sugar degradation products resulted in the highest yield

of Z-histidino-threosidine upon incubation with Z-lysine and Z-histidine, systematic

incubations were performed and the peak areas which showed m/z 720 in HPLC profiles

were compared. Among sugars and sugar degradation products, threose and erythrose had

the highest activity, followed by erythrulose and DHA, while the yields from glucose,

ribose, fructose, glyceraldehyde, MGO, glycolaldehyde were very low (Table 2-1). The

fact that erythrulose, the main non-oxidative degradation product of ascorbic acid under

physiological conditions(163), was a precursor suggests that L-erythrulose can transform

into L-threose, L-erythrose and glycolaldehyde under conditions similar to physiological conditions(171). Generally, degradation products of ascorbic acid are all good precursors of histidino-threosidine. Somewhat surprising was the relatively low yield of histidino- threosidine from DHA in light of the data from Simpson & Ortwerth which imply that erythrulose is the single major degradation product of DHA under anaerobic conditions.

Assuming their observation is correct, this discrepancy is attributed to the fact that the

major DHA degradation product erythrulose is not a good precursor for histidino-

68

threosidine, and its ability to tautomerize to threose or erythrose is more restricted than

initially assumed.

PRECURSOR YIELD

D-Glucose 0

D-Ribose 0

D-Fructose trace

D,L-Glyceraldehyde trace

Methylglyoxal trace

Glycolaldehyde trace

D-Threose 34 (100)a

D-Erythrose 30 (88)

D-Erythrulose 3 (8.8)

Dehydroascorbate 0.8 (2.4)

a Yield is given as percentage in parentheses.

Table 2-1 Relative yield of glycation agents for histidino-threosidine (Peak areas were compared in HPLC chromatographs)(168).

69

Fig 2.9A shows the time dependent generation of Z-histidino-threosidine using threose as a precursor. With threose, Z- histidino-threosidine reaches a plateau in 4 days.

Fig 2.9B shows the effect of the ratio of Z-lysine, Z-histidine and threose on the generation of the compound. Increasing threose beyond the equimolar ratio totally eliminates the generation of Z-histidino-threosidine, suggesting large amounts of threose favors the generation of other products instead of Z- histidino-threosidine.

120

100 A 80 60 Y Data 40

20

0 02468101214

120 B 100 5:1 80 1:1 60 1:2 40 20 0 1:5

yield) yield) (relative (relative Z-histidino-threosidine 2 4 6 8 10121416

100 C 9.0 80

60 40 7.0

Y Data 20 0 5.0

2 4 6 8 10121416 Days Fig 2.9 Effect of (A) time, (B) incubation ratios (The numbers indicate the ratio of Z-

His/Z-Lys:threose, and (C) pH (on Z-histidino-threosidine formation from Z-lysine, Z- histidine and threose(168).

70

Finally, formation of Z- histidino-threosidine was favored in alkaline conditions, as shown in Fig 2.9C. When the pH value decreased to 5, its generation was totally inhibited.

Histidino-threosidine was detected in bovine lens protein incubated with 10 and 50 mM threose concentrations at levels of 560 and 2840 pmol/mg of protein, respectively (Fig

2.10).

8.62 HISTIDINO-THREOSIDINE a 5.00 %

2.26 12.6813.98 21.17 25.17 5 10.00 20.00

8.85 b HISTIDINO-THREOSIDINE

4.77

% 4.31 5.50 24.46 12.26 17.74 26.26 27.37 21 10.00 20.00

2.34 c 2.57 7.618.85 12.74 15.46

% 18.08 26.89 26.11 27.73

38 Time 10.00 20.00

Fig 2.10 Detection of histidino-threosidine (m/z 452 279) by LC-ESI/MS multiple reaction monitoring in bovine lens protein incubated with 50 (A), 10 (B) and 2 mM (C) threose(168).

71

2.4. DISCUSSION

The mechanism by which ascorbic acid and its degradation products crosslink proteins is by and large still unknown. While this work was in progress, candidate

crosslinks have been described by Reihl et al(172). These crosslinks involve arginine

and lysine in an imidazole ring. Our preliminary data suggested Z- lysine and Z-histidine

were strong candidates residues for the desired crosslinks based on the anticipated

hydrophobic behavior (e.g. long retention time) of the compounds. In doing so,

surprisingly few other candidate crosslinks emerged from which Z-histidino-threosidine

was isolated and characterized.

Histidino-threosidine has novel features in addition to features reminiscent of

previously described Maillard Reaction compounds. Neither glucose nor ribose were

precursors. Dehydroascorbate was a precursor, albeit in relatively low yields compared to

threose, erythrose and erythrulose. Incubations with fructose, glyceraldehyde,

methylglyoxal, and glycolaldehyde generated traces of histidino-threosidine. On the other

hand the pyrrole carbaldehyde structure has been observed in a number of studies.

Nagaraj and Monnier reported a structure named formyl threosyl pyrrole(173).

Surprisingly, except for pyrraline, which originates from 3-DG(174), there is currently no

evidence for the existence of pyrrole adducts or crosslinks in the lens. Histidino-

threosidine is the first structurally characterized Maillard crosslink found, although a histidine adduct with 4-hydroxynonenal has been previously reported(175). These authors confirmed the structure of an HNE-His imidazole Michael adduct. Similarly, a histidine

72

adduct derived from a fatty acid oxidation product, malondialdehyde, has been identified by Bailey’s group(176).

However, other studies have pointed to the beneficial effects and utilization of histidine as an anti-cataract agent. Carnosine is a dipeptide containing histidine.

Carnosine-like compounds can prevent AGE-induced crosslinking(177). Recently,

Seidler et al (178) suggested that histidine is the representative structure for an anti- crosslinking agent and the imidazolium group of histidine is essential for carnosine’s anti-crosslinking effects because methylation of the N-1 position of imidazole abolished anti-crosslinking activity of carnosine. Sayre et al. showed that carnosine also can inhibit

4-hydroxynonenal derived crosslinking by forming a 13-member cyclic adduct through initial Schiff base formation followed by conjugate addition of the imidazole group(179).

Here we provided experimental evidence that histidine can form a stable structure with glycated lysine.

A proposed mechanism of histidino-threosidine formation is shown on Fig 2.11A.

The reaction starts with nucleophilic attack of lysine nitrogen on a sugar aldehyde group which after loss of water and tautomerization gives an Amadori product. Although formation of some cross-links with two sugar moieties, such as triosidines, involves formation of bis-Amadori products on a lysine residue(180), this path cannot generate the potential histidino-threosidine structures A or B (Path A). However, nucleophilic attack of the lysine-Amadori compound on the keto-group of a histidine-Amadori (Path B) can generate lysine-histidine-bis-Amadori product, which after intramolecular enolization and aldol condensation, can generate histidino-threosidine B. Histidino-threosidine with an aldehyde group at position 2 probably follows the mechanism of crossline formation

73

proposed by Biemel et al(44), where dideoxysone vi (Path C), formed by enolization of

sugar part on histidine Amadori product (Fig 2.11B)(181), reacts with lysine Amadori

product and after ring closure generates histidino-threosidine A. Because Path C

involves reaction of Amadori product and 3,4-dideoxysone which is, according to Reihl

et al., not a major dideoxysone formed from Amadori product of C-4 sugars(181), this

makes Path C less likely than Path B and, accordingly, structure A less likely than

structure B. Loss of chirality on position 8 is the result of intramolecular tautomerization

of lysine Amadori product (equilibriums labeled with *).

Histidino-threosidine was detected in bovine lens proteins incubated with 10 and

20 mM but not 2 mM threose, leading to the conclusion that the formation of this

crosslink requires saturation with sugars, i.e. conditions that are unlikely found under

physiological conditions. One could speculate that the crosslink may be present in certain

forms of old and cataractous human lenses in which the oxidoreductases and other

detoxification enzymes are severely deficient. However, attempts to find the crosslink in

old and cataractous lenses using mass spectrometry failed (unpublished), and it is

difficult to conceive that any tissue would accumulate the high tetrose concentrations that

are needed to generate histidino-threosidine, even under pathological conditions. Thus, further research perhaps focusing on the crosslinks described by Reihl(172) will be needed to unequivocally address the role of ascorbic acid in crosslinking of proteins by ascorbic acid during aging.

74

R2 R O 1 N R1 N R2 N OH N O R N 1 H N N A HO or O HO HO HO N N HO O HO O R2 HO OH A B Path A HO OH R R1 1 O R NH NH 1 + * NH OH 2 HO HO HO O OH OH HO HO O R2 N O OH N R2 N vi OH Path B N O Path C i R1 O R R1 O R1 1 N N N N N N N * N * HO HO HO OH HO O O O OH O N R R2 N OH 2 OH HO HO HO HO N N H2O R R2 2

R1 R1 N R N N 1 N N N N O * HO HO O OH OH OH HO R R2 OH 2 OH HO HO N N HO R2 A R R1 1 N N N N N * N HO HO OH OH O OH R N O R2 O 2 R2 HO R HO H O 1 2 N N

HO O HO B

R R2 R2 2 R2 R2 R2 N N N N N N H O O N N 2 B R2 N N N N N OH HO HO + HO O HO N OH OH O OH OH H OH O OH OH OH OH O O

i ii iii iv v vi

Fig 2.11. Proposed mechanism of formation of histidino-threosidine with the most likely

candidate being structure B. (A) and proposed mechanism of dideoxysone formation with

histidine and threose (B)(168).

75

Chapter three

Identification of glucose-derived crosslinking sites in ribonuclease A

3.1 INTRODUCTION

In collagen, the major carbonyl compound of the extracellular matrix is glucose.

The major crosslinks identified so far from the Maillard Reaction with glucose and other carbonyl compounds are summarized in Fig 3.1.

76

A. From Glucose-Amadori B. From Pentoses OH HO H Lys HO HO OH N N N HO OH HO NH OH NH N N Arg N N N Arg N Lys Lys N Lys Lys Lys Glucosepane Crossline Pentosidine Vesperlysine A

C. From Tetroses D. From Trioses Lys OH HO NH HO HO N

OH HO N N N H NH N Lys Arg Lys HO N Lys Lys Threosidine Arg-Hydroxy-Triosidine Lysyl Hydroxy-Triosidine E. From Dioses and the oxoaldehydes glyoxal, methylglyoxal and 3-deoxyglucosone

Lys Lys Lys N Lys N H N NH N N H H N Arg N NH N N NH OH Arg HO N H3C O N OH HO Lys H3C Arg CH2OH MOLD MODIC DODIC DODIC-Ox

Lys Lys HN Lys N N CH2 NH O N NH N Lys NH Lys Arg GOLD GODIC GOLA F. “ODIC” crosslinks from ascorbic acid and its degradation products Unknown precursor Lys Lys Lys Lys N N N NH N NH NH HO NH N HO NH NH Arg N HO OH N Arg CH OH Arg CH2OH 2 HN Lys CH2OH

From L-xylosone From L-threosone From 3-deoxythreosone K2P

Fig 3.1 Structures of selected cross-links of the Maillard Reaction categorized by their precursors(182).

When a number of these crosslinks were determined in skin collagen from aging human beings, it became apparent that glucosepane emerged as the single major crosslink

(Table 3.1). In addition, its level is dramatically elevated in diabetes(182), a condition

77

that is associated with increased stiffness of all collagen-rich tissues. As shown in this table, it is clear that glucosepane is present in vivo at a level comparable to simple adducts such as CML. Biemel et al. provided convincing evidence showing that glucosepane is the dominant compound in both human serum albumin and lens protein(44). Our laboratory provided additional data supporting the premise that glucosepane is the single major glycation derived crosslink known to date in human skin and glomerular basement collagen that correlates with both aging and diabetes(52).

Glucosepane levels increased up to ~ 2 nmol/mg of collagen in old nondiabetic controls and further increased to ~ 4.3 nmol/mg in diabetic patients(52). Assuming a molecular mass of 100 kDa for a single strand of triple helical collagen, the level of glucosepane in collagen can be translated to one crosslink for every two and five collagen molecules in diabetic and nondiabetic aged controls, respectively. This level of modification might explain the accumulation of collagen matrix due to impaired proteolytic digestion in diabetes and aging, and perhaps even increase matrix stiffening, particularly if the crosslinks are intermolecular or interfibrillar.

78

Modification Mean value Mol% Relationship with

(pmol/mg protein) modification age

DODIC-Ox 4 <0.1 increase

GOLA, crossline unknown unknown unknown

GODIC 8 <0.1 increase

DODIC 9 <0.1 No increase

MODIC 12 <0.3 increase

GOLD 12 <0.3 increase

CEL 25 0.25 increase

Pentosidine 50 0.5 increase

Fructosyl- 90 0.9 increase

ornithine

MOLD 120? 1.2 increase

CML 600 6.0 increase

Fructosyl-lysine 750 7.5 shallow increase

Glucosepane 1000 10 increase

Ornithine 4000 40 increase

Total damage ~700 70 increase

Table 3.1 Comparative levels of selected Maillard Reaction adducts and cross-links in normal human skin collagen in the eighth decade of life(182)

79

Interestingly, Biemel et al. further demonstrated that a protein-linked dicarbonyl structure, N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate, is the precursor of glucosepane (Fig 1.4)(29). With the understanding of the higher reactivity of free dicarbonyl compounds as 3-DG, MGO, and GO, the discovery that protein-bound

Amadori products and dicarbonyl can be important sources of crosslinking is intriguing.

Biemel et al. suggested that N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate is irreversiblly linked to protein and this protein-bound character makes the precursor a persistent glycation agent that is resistant to in vivo detoxifying reductases.

In order to correlate tissue stiffening with glucosepane formation, an understanding as to potential sites of glucosepane formation is essential.

As an initial approach to the above question RNase A is used as a model protein for the determination of crosslinking sites by glucose. RNase A is a small protein of 13.7 kDa that has been widely studied in glycation chemistry. The strategy was to preincubate the protein under deareated conditions to prevent glycoxidative modifications, followed by removal of excess glucose by dialysis and reincubation with unmodified RNase A.

The final preparation was submitted to enzymatic digestion and MALDI TOF analysis, searching for peptides containing the full complement of six glucose-derived carbons.

Under the above conditions, the following pathways of protein modification are expected to dominate (Fig 3.2):

80

H H2C N R1 R1 H C OH A N HO C H N H N R CH HO 2 2 N C O HO H C H O Glucosepane

H H H O H C N R1 C HC N R1 * 2 H C OH H C OH C O R1-NH2 HO C H HO C H HO C H H C OH H C OH H C OH H C OH H C OH H C OH CH2OH CH2OH CH2OH Schiff base Amadori product

-H2O H O HC N R H 1 N R1 C C OH C C O C O R2-NH2 C H CH2 CH2 H C OH H C OH H C OH -H2O H C OH H C OH H C OH CH2OH CH2OH CH2OH 3-DG H2N H N R2 HN -H2O

R1 N H N H N R2 H N

H2C + NH3 H C OH

- H C OH R1= (CH2)4 C COO H CH2OH + NH3

- DOGDIC R2 (CH2)3 C COO H

81

H H H H2C N R1 H C N R1 H2C N R1 2 B C O C O C OH -H2O enolization HO C H HO C HO C

H C OH H C H C H C OH H C OH C OH CH2OH CH2OH CH2OH

1,5 H+ shift tautomerization

H H H N R N R H2C N R1 H2C 1 H2C 1 HC OH 1,3 H+ shift HC OH keto-enol HC OH tautomerization HO CH tautomerization HO C HO C H C H C H C HC OH C OH C OH H C O H C O CHOH

keto-enol tautomerization

H H2C N R1 HC OH

HO CH H CH

C O

C Fig 3.2A MajorH glycationO pathways expected in nonoxidative glucose-protein incubation.

Three stable glycation products are: Amadori product (m/z=162); glucosepane (m/z=108);

DODIC (m/z=126). The step marked as * is shown in detail in Fig 3.2B. These detailed steps are produced with help of V. Andeson.

82

Thus, according to Fig 3.2, the only currently known structures expected to

contain the full six glucose carbons, and to found in our experiments, are the Amadori

product, glucosepane and DODIC.

3.2 EXPERIMENTAL PROCEDURES

3.2.1 Reagents

RNase A from bovine pancreas (Type XII-A, >90% SDS-PAGE purity) was from Sigma

13 (St. Louis, MO). D-( C6)Glucose was also from Sigma (St. Louis, MO). Chelex® 100

Resin was from Bio-Rad (Hercules, CA). Deionized water (18.2 MΩ cm-1) was used for

all experiments. All other reagents were obtained in the highest quality available from

Sigma (St Louis, MO, U.S.A.), unless indicated otherwise.

3.2.2.Preparation of Glycated RNase A

Bovine RNase A at a concentration of 50 mg/ml was incubated with a mixture of 250

13 mM D-glucose and 250 mM D-( C6)Glucose in 100mM chelex-treated sodium phosphate buffer containing 1mM DTPA at 37oC for 2 weeks. Glucose was removed by dialyzing

the sample against 4 liters of the same buffer for two days with one buffer change.

Spectra/Por 7 dialysis tubing (Spectrumlabs, Rancho Dominguez, CA), 2 KDa MWCO)

was used for the dialysis. Upon determination of protein concentration, this “preglycated”

preparation was further incubated with an equal amount of freshly added native RNase A.

83

for 4 weeks at 37oC. Chloroform and toluene (0.15%, v:v) were added to prevent bacterial growth during the incubation.

3.2.3 Enzymatic Digestion for Peptide Mapping

Prior to enzymatic digestion, protein samples (20 ul) were diluted to 500 µl with water and concentrated back to 20 µl using an Ultrafree®-0.5 centrifugal filter units (Millipore,

Billerica, MA). Three cycles of concentration removed ~99% of the initial salt content.

The resulting protein sample was subjected to in-solution digestion by trypsin (Promega,

Madison, WI) in the presence of RapiGest SF (Waters, Milford, MA). The ratio of enzyme to protein substrates was around 1:100 and the digestion was carried out in a 50

o mM NH4HCO3 buffer (pH 8.0) at 37 C for 3 hours. The ultra-centrifuged protein was

also subjected to digestion by chymotrypsin and endoproteinase Asp-N (Roche Applied

Science, Indianapolis, IN). Chymotryptic digestion was carried out in100 mM Tris-HCl,

10 mM CaCl2, pH 7.8) whereas the Asp-N digestion buffer was 50mM Tris-HCl, pH 8.0.

In both digestions, the ratio of enzyme to protein substrates was around 1:100 and the digestion was carried out at 37oC (endoproteinase Asp-N) or at 25oC (chymotrypsin)

overnight. Before all these digestions, protein samples were reduced by 5 mM

Dithiothreitol (60oC, 30 mins) and alkylated by15 mM iodoacetamide (30 mins in the

dark).

3.2.4 Matrix-Assisted Laser Desorption and Ionization Mass Spectrometry

84

The digested peptides were analyzed by using a prOTOF™ 2000 MALDI O-TOF mass

spectrometer (PerkinElmer, Inc.). Samples were mixed with an equal volume of α-cyano-

4-hydroxycinnamic acid matrix solution (10 mg/ml in 50% acetonitrile, 0.1% TFA).

Typically, 1 μl of the mixture was applied onto the laser target probe and was air-dried before being introduced into the mass spectrometer. For identification of possible glycation modified peptides, a data base including all possible m/z values of Amadori, glucosepane, and DODIC modified peptides was generated, which was used to compare with the doublet or triplet peaks found in the MALDI experiment. Laser-desorbed positive ions were analyzed after acceleration by19 kV in the reflector mode for the peptide digest.

3.2.5 Liquid Chromatography-Tandem Mass Spectrometry Analysis

LC-MS/MS analyses of the proteolytic digests were performed using a quadrupole ion trap mass spectrometer (model LTQ) from Thermo-Finnigan (San Jose, CA) coupled with an Ettan MDLC system (GE Healthcare,Piscataway, NJ),chromatographed with a gradient of 0-60% acetonitrile - 0.1% formic acid for 30 min. The spectra were acquired by data-dependent methods, consisting of a full scan and then an MS/MS on 6 most abundant precursor ions at the collision energy of 35%. The m/z values of probably modified peptides estimated from MALDI measurement were put in the mass list for

MS/MS. The obtained data were submitted to BioWorks Rev. 3.3 for database search.

Further interpretation of the tandem mass spectra of the modified peptides was assisted by an Excel spreadsheet that generated predicted fragment ions from glycated peptides available from V. Anderson.

85

3.3 RESULTS

3.3.1 Sequence Coverage of Ribonuclease A from LC-MS/MS Analysis

The sequence coverage of RNase A following tryptic digestion from LC-MS/MS

experiments was at least 85%. The high sequence coverage of the enzymatic digestions

decreased the chance of missing some modified peptides whose corresponding native

peptides are not sensitive in mass spectrometry.

3.3.2 Amadori- and Glucosepane Modified Peptides Are Major forms of Glycation

in Ribonuclease A Incubated with Glucose.

It is well established that oxidative conditions during incubation of proteins with glucose favor the formation of certain AGEs over others, in particular the glycoxidation products

CML and pentosidine(183). A similar phenomenon is observed with ascorbic acid which

degrades preferentially into erythrulose under anaerobic conditions. Thus, in order to

enhance the formation of glucosepane and diminish glycoxidation, we have used metal-

depleted phosphate buffer and nonoxidative conditions. The results of the tryptic digest

of glycated RNase A confirmed the above hypothesis. Sixteen doublet peaks, Δm/z=6,

each corresponding to a native peptide with a modification by a single glucose-derived

product (or triplet peaks corresponding to native peptides with two glucose-derived

products) were observed in the tryptic map of RNase A. Most of these peaks could be

assigned to Amadori intermediates and/or glucosepane modified peptides (Table 3.2).

Five peptides were modified with Amadori intermediates only, except for one peptide

with m/z 2677.3 corresponding to 40C*KPVNTFVHESLADVQAVCSQK61 with residue

86

Lysine 41 modified with Δm/z=160, as unequivocally confirmed by tandem mass

spectrometry (Fig 3.3). The modification by Δm/z=160 instead of 162 (Amadori product)

is quite unexpected as it suggests oxidation of an Amadori product. The m/z 3251.6 peak

also could be assigned to the peptide 40C*KPVNTFVHESLADVQAVC*SQKNVAC*K66

with Amadori modification at K41. The m/z 1312.7 peak corresponds to peptides

1KETAAAKFER10 with Amadori modification at K1 or K7. The other glycation hot spots of Amadori modifications at K37 are comprised in the m/z 3514.7 and 3630.75

peaks, corresponding to 34NLTKDRC*KPVNTFVHESLADVQAVC*SQK61 and

38DRC*KPVNTFVHESLADVQAVC*SQKNVAC*K66. These peptides have in common

the sequence 38-61 and in addition to the Amadori product (Δm/z=162) an m/z increase in

108, consistent with the presence of an intramolecular glucosepane. Thus, these peptides

have a characteristic isotopic triplet peak ratio of 1:2:1. Besides the glycation hot spots at

K1, K7, K37 and K41, other lysine residues, such as K66, K104, and K98 were also

shown to be modified by Amadori products.

87

Detected Theoretical sequence Position Modified Modification m/z of residue type sequence 1312.69 KETAAAKFER (1312.76) 1-10 K1 or K7 Amadori 2677.30 * * 40-61 K41 Amadori-2 C KPVNTFVHESLADVQAVC SQK(2679.3) * * 2896.43 DRC KPVNTFVHESLADVQAVC SQK(2896.4) 38-61 R39 to intra- K41 molecular glucosepane * * 2914.43 DRC KPVNTFVHESLADVQAVC SQK(2914.4) 38-61 R39 to K41 2896+H2O? * * * 3020.33 NVAC KNGQTNC YQSYSTMSITDC R(3020.3) 62-85 K66 Amadori * 3029.48 TTQANKHIIVAC EGNPYVPVHFDASV(3029.55) 99-124 K104 Amadori

3233.58 3251.6-H2O * * 3251.61 C KPVNTFVHESLADVQAVCSQKNVAC K(3251.64) 40-66 K41 or Amadori K61 * * 3352.7 NLTKDRC KPVNTFVHESLADVQAVC SQK(3352.6) 34-61 K41 to Intra- R39 glucosepane * * 3514.75 NLTKDRC KPVNTFVHESLADVQAVC SQK(3514.85) 34-61 K37(Ama) Amadori and (1:2:1) K41 to intra- R39 molecular glucosepane * * * 3630.75 DRC KPVNTFVHESLADVQAVC SQKNVAC K(3630.85) 38-66 K61(Ama) Amadori and (1:2:1) K41 to intra- R39 molecular glucosepane 4064.01 KETAAAKFER 1-10 K1 or K7 inter- * * DRC KPVNTFVHESLADVQAVC SQK(4064.07) 38-61 R39 molecular DODIC

* * 4088.01 YPNC AYKTTQANKHIIVAC EGNPYVPVHFDASV 92-124 K98 and 2 Amadori (1:2:1) K104 (4088.08) 4226.11 KETAAAKFER 1-10 K1 or K7 Amadori and * * (1:2:1) DRC KPVNTFVHESLADVQAVC SQK(4226. 1) 38-61 R39 inter- molecular DODIC * * * 4523.05 NGQTNC YQSYSTMSITDC RETGSSKYPNC AYKTTQANK 67-104 K91 or Intra- K98 to glucosepane (4523.0) R85 * 4541.05 YPNC AYKTTQANK 92-104 K98 intra- * * 67-91 R85 molecular NGQTNC YQSYSTMSITDC RETGSSK(4541.01) glucosepane/ pseudo inter- molecular glucosepane

Table 3.2 Observed doublet or triplet peaks separated by Δm/z=6 in the MALDI spectrum of the tryptic digest of glycated RNase A. The data above show only the monoisotopic m/z value.

88

As shown in Table 3.2, the modified peak with m/z of 3233.58 cannot be readily assigned as an Amadori product or glucosepane modified peptide. But we can also assign it to the same peptide as Amadori modified

40C*KPVNTFVHESLADVQAVC*SQKNVAC*K66, with the loss of one molecule of water. This would correspond to a lysine bound 1,4-dideoxy-5,6-glucosone which has been suggested as the precursor of glucosepane(29). Thus detection of this dideoxy- glucosone modified peptide strengthens previous data suggesting that it is an important glucosepane precursor.

89

Da_ 041007 # 4778 RT: 25.02 AV: 1 NL: 1.15E4 y14 F: ITMS + c NSI d Full ms2 [email protected] [355.00-2000.00] 1571.7 100

95 b1 b2 b3 b4 b5 b6 b7 b8 b9 b 10b11b12b13b14b15b16b17b18b19b20b21b22 40C K* P V N T F V H E S L A D V Q A V C S Q K61 90 y22 y21 y20 y19y18y17y16y15y14y13y12y11y10 y9 y8 y7 y6 y5 y4 y3 y2 y1 85

80 y 8 y ++ 75 919.4 20 1115.0 b 70 12 1572.8 65

60 R el 55 ati y12 ve 50 1305.6 Ab b14 un 45 1758.8 da nc 40 y b8 10 1106.6 35 1105.8 y13 y15 1434.6 1670.7 30 b15 y4 25 y1857.9 522.2 16 y6 y b 1817.9 y17 7 y ++ y 13 20 692.4 820.4 18 9 1643.8 1016.9 1919.0 1034.5 y 15 y5 11 b16 621.3 1218.5 1985.8 10 b9 b2 b4 b5 b11 1248.5 1459.4 5 449.4 645.5 759.3

0 400 600 800 1000 1200 1400 1600 1800 2000 m/z

Fig 3.3 Tandem mass spectrum of glycated RNase A tryptic peptide with molecular

doubly charged ion at 1339. This peptide is assigned as peptide 40C*KPVNTFVHESLA

DVQAVC*SQK61 with modification of K41 by Δm/z=160 (oxidized Amadori). Every b

ion greater than b2 contained this increment while every y ion up to y20 was consistent

with the unmodified peptide sequence, strongly implicating the modification of K41 by

an oxidized Amadori product.

The glucose derived modifications are evidenced by the characteristic doublet

or triplet peaks (for examples see Fig 3.6 and 3.7b). These peaks can be partly explained

90

by Amadori modified peptides, while the non-Aamadori product peaks in Table 3.1 are

consistent with the presence of intra- or inter-molecular glucosepane modifications.

The most prominent candidate glucosepane containing peptide was peptide

38DRC*KPVNTFVHESLADVQAVC*SQK61, corresponding to a doublet peak with the

m/z 2896.4 and 2902.4, respectively. The LC/MS/MS fragmentation data shown in Fig

3.4 support the existence of an intramolecular glucosepane modification crosslinking K41 and R39. The appearance of a b4 ion that contains the mass increment localizes the

modification to the first four residues of the peptide. The very intense y23 ion containing

both R39 and K41, but not D38, further limits the location of the adduct. The presence of

y4-y17 ions consistent with the unmodified sequence confirms the identity of the peptide.

The absence of b2, b3, y21 and y22 ions is predicted for an intramolecular crosslink.

91

Da_ 041007 # 4635 RT: 24.27 AV: 1 NL: 2.15E4 F: ITMS + c NSI d Full ms2 [email protected] [385.00-2000.00] 100 4.50 2896. 2902.4 y23 95 b1 b2 b3 b4 b5 b6 b7 b8 b9 b10b11b12b13b14b15b16b17b18b19b20b21b22b23b24 1391.4 38 61 90 D R* C K* P V N T F V H E S L A D V Q A V C S Q K y y y y y y y y y y y y y y y y y y y y y y y y Intensity 85 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1

80

0 2895.0 m/z 2910.0 75 C:\Documents and Settings\zhenyu\Desktop\raw data\h6.txt (16:18 06/12/07) b22 70 1375.8

65

Rela60 tive Abu nda55 nce b16 50 1977.9

b21 45 1268.6

40 b8 1079.5 35 y8 b 919.4 19 30 1138.8 b10 b13 y11 1325.7 25 1218.8 1678.9 b18 20 1103.0 b4 b15 y14 668.4 y12 b12 1862.8 15 1571.6 b20 b7 1305.7 y9 1188.3 1591.7 b14 b6 978.5 10 1034.5 y10 1791.9 y4 864.6 b9 y5 1105.6 y6 1226.4 b y15 522.4 y7 11 1817.6 y17 5 621.4 692.3 1670.8 820.4 1462.7 y16 1918.6 0 400 600 800 1000 1200 1400 1600 1800 2000 m/z Fig 3.4 Tandem mass spectra of glycated RNase A tryptic peptide corresponding

36DRC*KPVNTFVHESLADVQAVC*SQK61 with precursor m/z 1449.4. The presence of ions b4 and y23, both containing the glucosepane mass increment, suggest the presence of intermolecular glucosepane crosslinking K41 and R39. b18-b22 and y23 ions are doubly charged.

Besides this peptide, several other modified peptides can be assigned to peptides with intramolecular glucosepane between R39 and K41, including

1) 34NLTKDRC*KPVNTFVHESLADVQAVC*SQK61,

2) 34NLTKDRC*KPVNTFVHESLADVQAVC*SQK61,

3) 38DRC*KPVNTFVHESLADVQAVC*SQKNVAC*K66. Peptide 1 and 2 actually are same peptides with different modifications: peptide 1 only has an intra-molecular

92

glucosepane modification, whereas peptide 2 has both an intra-molecular glucosepane and an Amadori modification, possibly at K37, resulting in the missed tryptic cleavage.

Of relevance in the above analyses and throughout these studies is that the peptide containing the glucosepane modification at K41 and R39 was a major signal, strongly suggesting that these sites are highly favored targets of glucosepane formation.

The m/z 2914.4 cannot be easily explained by either Amadori or glucosepane

modifications. Instead, it is consistent with the peptide 34DRC*KPVNTFVH

ESLADVQAVC*SQK61 with, however, a Δm/z of 126. This mass increment is 18 Da,

potentially H2O, larger than the intra-molecular glucosepane modification. Another

possible AGE crosslink, DODIC, also has a six carbon backbone that would yield M and

M+6 double peaks and would yield Δm/z=126. The LC/MS/MS fragmentation of this

peptide strengthened this tentative assignment. As shown in Fig 3.5a, the existence of b4

ion again limits the modification with Δm/z=126 to the fragment 34DRC*K37, suggesting

the modification could be DODIC. Interestingly, when we tried to find the triply charged

peptide with m/z 2914.4 in the LC/MS spectrum, we found that there are two major species with different elution times at 22.47 and 24.38 mins (Fig 3.5b). The second

2914.4 peak co-eluted with the peak with m/z value of 2896.4, leading us to speculate that the first 2914.4 peptide may be modified by e.g. DODIC whereas the second 2914.4 peptide might be modified by glucosepane (2896.4) plus a water molecule added in the mass spectrometer. Coelution may suggest in source fragmentation, i.e. loss of H2O.

93

Da_ 041007 #4199 RT: 22.45 AV: 1 NL: 1.30E5 F: ITMS + c NSI d Full ms2 [email protected] [255.00-2000.00] 100 y17-H2O 951.0 95 b1 b2 b3 b4 b5 b6 b7 b8 b9 b10b11b12b13b14b15b16b17b18b19b20b21b22b23b24 38 61 90 D R* C K* P V N T F V H E S L A D V Q A V C S Q K y24 y23 y22y21 y20y19y18y17y16y15y14y13y12y11y10 y9 y8 y7 y6 y5 y4 y3 y2 y1 85 b14 905.3 80 75 b b -H O 9 70 14 2 1244.7 896.5 65

Re60 b13 lati 848.9 ve 55 Ab b -H O b15 b21 un50 13 2 840.1 941.3 1277.1 da b18 y nc 45 8 1112.1 919.4 e 40 b10 35 b20 1343.6 b4 1197.0 30 686.5 y7 y10 b10 820.4 25 672.4 b 1105.6 12 y9 y12 805.4 1034.6 20 b19 1305.6 b23 y6 1147. 1385.0 15 692.4 y4 b17 y11 y13 522.3 1048.0 1218.6 10 y5 b11 b 1434.6 y 16 3 621.4 740.9 998.7 b11 y14 y15 5 362.1 1480. 1571.6 1670.7 0 400 600 800 1000 1200 1400 1600 1800 2000 m/z

Fig 3.5a Tandem mass spectrum of a glycated RNase A tryptic peptide, doubly protonated precursor m/z 972.3, corresponding to 36DRC*KPVNTFVHESLADVQAV

C*SQK61 , can be interpreted as an intramolecular modification by DODIC crosslinking residues K41 and R39. b11-b21, b23, and y17 ions are doubly charged.

94

RT: 0.00 - 65.00 22.47 NL: 100 6.60E6 m/z= 95 971.63- 972.63 90 MS Da_041007 85 80 22.64 Mr=2914.4 75

70

65 16.96 60 24.38

55

50

45 22.39

Relative Abundance 19.60 40

35 16.82

30

25 14.09 20 13.90 15 25.09

RT: 0.0010 - 65.00 25.65 24.52 NL: 100 26.11 5 13.02 9.64E6 10.28 27.89 29.62 24.28 39.02 40.16 46.51 47.16 54.72 m/z= 95 1.29 7.63 57.24 63.79 0 965.63- 0 5 10 15 20 25 30 35 40 45 50 55 60 966.63 90 Tim e (min) MS Da_041007 85

80

75 70 Mr=2896.4 65

60

55

50

45

Relative Abundance 40

35

30

25

20 23.94 15 21.69 25.18 21.15 10 16.42 25.41 15.10 5 26.49 10.33 14.02 29.92 2.32 8.97 36.46 39.12 45.48 47.44 50.02 54.62 56.63 64.63 0 0 5 10 15 20 25 30 35 40 45 50 55 60 Tim e (min)

Fig 3.5b Selected ion chromatograms demonstrating the co-elution of a peptide with Mr

2896.4 and the second peptide with Mr 2914.4 suggest that the peptides with Mr 2914.4 may have different origins.

95

A major question this work hoped to address is the extent to which glucose participates in intermolecular crosslinking. The peptide with m/z of 4541.0 matches the theoretical mass of crosslinked peptides 67NGQTNC*YQSYSTMSITDC*RETGSSK91 and 92YPNC*AYKTTQANK104, with a glucosepane crosslink. The glucosepane would necessarily crosslink R85 and K98, as K104 and K91 must be unmodified to allow for tryptic digestion. But with the existence of the nearby doublet peak with m/z of 4523.0, we need to reconsider the origin of the peak at 4541.0. These two hypothesized crosslinked peptides, 67NGQTNC*YQSYSTMSITD C*RETGSSK91 and

92YPNC*AYKTTQANK104, are actually linked together in the primary sequence. As indicated by the 3-D crystal structure of RNase A (Fig 3.8), residues K98 and R85 are very close to each other, making an intra-molecular glucosepane a more likely choice than an inter-molecular glucosepane. With all these factors considered, the peak at 4541.0 may actually originate from the peak with m/z of 4523.0. In glycated RNase A, with intra-molecular glucosepane exists between K98 and R85, the appearance of M+18 may be due to addition of water upon enzymatic hydrolysis of the internal K91-Y92 peptide.

Similarly, in the chymotryptic digest, intra-molecular glucosepane modified peptide

80SITDC*RETGSSKYPNC*AYKTTQANKHIIVAC*EGNPY115, which has an m/z value of 4259.8, will yield a 4277.9 peptide after digestion at Y97. In both cases, the ratio of original peptide and the second peptide derived from enzymatic cutting may depend on the efficiency of enzymatic digestion.

96

A 0.70 Tryptic digest 4547.0 Δm/z=18 4529.0

4541.0 4523.0

Intensity

0 4500 m/z 4580 C:\Documents and Settings\zhenyu\Desktop\raw data\h6.txt (12:33 05/23/07) B 0.4 5 Δm/z=18 Chymotryptic digest 4265.8

4259.8 4283.9 4277.9

Intensity

0 423 m/ 431 0 0 C:\Documents and Settings\zhenyu\Desktop\raw data\f2.txt (12:37 05/23/07) z

Fig 3.6 Both tryptic and chymotryptic peptides indicate the presence of intra-molecular glucosepane crosslink between K98 and R85. Incomplete enzymatic hydrolysis between the crosslinked residues would yield a pseudo-intermolecular-glucosepane, resulting in an additional m/z of 18 after digestion. Thus m/z 4523.0 corresponds to peptide

67NGQTNC*YQSYSTMSITDC*RETGSSKYPNC*AYKTTQANK104 with an intra- molecular glucosepane at K98 and R85. The m/z of 4259.8 corresponds to peptide

80SITDC*RETGSSKYPNC*AYKTTQANKHIIVAC*EGNPY115 with an intra-molecular glucosepane at R85 and K98/K104.

97

The search for inter-molecular crosslinking sites revealed unexpected results. The anticipated inter-molecular glucosepane crosslink has proven elusive.

However after searching for all glucosepane-based theoretical m/z values compatible with crosslinked peptides, we could not identify candidate glucosepane crosslinked tryptic peptides consistent either with m/z of 4064.01 (doublet peak) or 4226.1(1:2:1 triplet peak).

Interestingly, both of these m/z values matched theoretical values of inter-molecular crosslinks assuming the crosslink had a molecular mass 18 Da greater than glucosepane.

These data prompted a search for other inter-molecular crosslinks which contain all 6 carbons of a single glucose. The DODIC crosslink is consistent with these mass data.

Tandem mass spectrum of the 4064.01 peptide supported the presence of an inter- molecular crosslink with Δm/z of 126 between K1 and R39 (Fig 3.7a). The b ion, y ion, b’ ion and y’ ion series establish the presence of both peptides, 1KETAAAKFER10 and

38DRC*KPVNTFVHESLADVQAVC*S QK61. The appearance of b’4, and b’5 also suggests that the crosslinking site contains K1 instead of K7. In the same manner, the peptide with m/z value of 4226.1 also fits the crosslinking of peptide 1KETAAAKFER10 and 38DRC*KPVNTFVHESLADVQAVC*S QK61 by DODIC, with additional Amadori modification at K7 or K41 (Fig 3.7b).

98

Da 0410 #365 RT 20.1 AV:1 NL 1.90E F: ITMS + c NSI d Full ms2 [email protected] b14-H2O 100 981.5 b’1 b’2 b’3 b’4 b’5 b’6 b’7 b’8 b’9 b’10 95 1K* E T A A A K F E R10 y’10 y’9 y’8 y’7 y’6 y’5 y’4 y’3 y’2 y’1 90 b1 b2 b3 b4 b5 b6 b7 b8 b9 b10b11b12b13b14b15b16b17b18b19b20b21b22b23b24 85 38D R* C K P V N T F V H E S L A D V Q A V C S Q K61 b13 y24 y23 y22 y21y20y19y18y17y16y15y14y13y12y11y10y9 y8 y7 y6 y5 y4 y3 y2 y1 80 949.3 b b’ 20 75 4 1181.9 835.0

70 b12 920.66 b b17 65 y8 14 1082. 919.5 987.5 60 R b’5 b’8-H2O el 55 b11 1137.6 1247.4 y’7 at 792.4 877.8 iv 50 y10 e y22 45 1105. A 839.7 b19 y12 b 40 1148.8 y11 1305.6 y21 b16 35 786.5 1049.6 1218.

y’5 30 y’2 304.2 650.7 25 y4 y7 522.4 b18 20 820.4 1125.0 y’4 y6 15 290.0 692.4 y13 1434.6 y16 y5 y14 10 y9 1817.8 y3 1571.7 621.5 1034.4 y’4-NH3 5 362.1 562.1 0 40 60 80 100 120 140 160 180 200 m/

Fig 3.7a Tandem mass spectrum of the quadruply charged precursor ion, m/z 1016.8. y22 ion suggests the crosslinking at R39, while b’4, b’5, and y’4 ion suggests that K1 instead of K7 is the crosslinking site. b11-b14, b16-b20, b’4, b’5, y21 and y22 ions are triply charged, b’8 ion is quadruply charged.

99

0.40 4070.0 Peptide with one DODIC only

4064.0 Intensity

0 4040 m/z 4120 C:\Docum ents and Settings\zhenyu\Desktop\raw data\h6.txt (17:36 05/24/07) 4232.1

0.30 4238.1 4226.1 Peptide with one DODIC and one Aamadori Intensity

0 4200 m/z 4300 C:\Documents and Settings\zhenyu\Desktop\raw data\h6.txt (17:40 05/24/07)

Fig 3.7b MALDI mass spectrum of the tryptic digestion of glycated RNase A. Two sets of peaks point to the inter-molecular crosslinking between K1 and R39. The two peptides crosslinked are 1KETAAAKF ER10 and 38DRC*KPVNTFVHESLADVQAVC*SQK61.

In order to strengthen the above observations, digestions with the enzymes Asp-N and chymotrypsin were also carried out to verify the presence of the modification by Amadori, glucosepane and DODIC (Table 3.3a and 3.3b). These two

100

enzymes yield relatively low amounts of modified peptides and some of them cannot be assigned readily so that less information was obtained than with tryptic digestion.

4a Detected Theoretical sequence Position Modified Modification m/z of residue type sequence 1394.6 * 1898.9 DRC KPVNTFVHESLA (1898.9) 38-52 R39-K41 Intra-DODIC * 1480.7 DRC KPVNTFVH(1480.7) 38-48 R39-K41 Intra- glucosepane * 1880.9 DRC KPVNTFVHESLA (1880.9) 38-52 R39-K41 Intra- glucosepane 2444.1 Originated from 2336.0 Intra- glucosepane * 2830.3? DSSTSAASSSNYC NQMMKSRNLTK (2830.3) 14-37 K31 or Amadori K37 * * * 3604.6 DVQAVC SQKNVAC KNGQTNC YQSYSTMSIT (3604.6) 53-82 K61 or amadori K66 3734.7 Originated from 3572.5 Amadori * * 4142.8 ETGSSKYPNC AYKTTQANKHIIVAC EGNPYVPVHF 86-120 K91 or Amadori K98 or (4143.0) K104 * * * 4574.2 DC RETGSSKYPNC AYKTTQANKHIIVAC EGNPYVPVHF 83-120 K91 or Amadori K98 or (4574.2) K104 3.3b Detected Theoretical sequence Position Modified Modification m/z of residue type sequence 811.1 903.4 1076.0 1343.6 1381.6 * 1473.7 TKDRC KPVNTF(1473.7) 36-46 R39 to Intra- K41/K37 glucosepane 2124.8 * 2206.1 KTTQANKHIIVAC EGNPY(2206.1) 98-115 K98 or Amadori K104 * * 2271.0 SITDC RETGSSKYPNC AY (2271.0) 80-97 K91 amadori * * 4259.8 SITDC RETGSSKYPNC AY 80-97 R85 Inter- * 98-115 K98 or glucosepane KTTQANKHIIVAC EGNPY(4260.0) K104 Or intra- molecular glucosepane * * * 4277.8 SITDC RETGSSKYPNC AYKTTQANKHIIVAC EGNPY(4278.0) 80-115 R85 to intra- K98? molecular glucosepane

Table 3.3 Modified peptides mapped by enzymatic digestion. a) with Asp-N. b) with chymotrypsin.

101

In the Asp-N digestion map, peptides with m/z of 1480.7 and 1880.9 confirmed

the existence of a major intra-molecular glucosepane between K41 and R39. A relative

weak doublet peak (Δm/z=6), with m/z 1898.9 shows that an intra-molecular DODIC

crosslink could be present between K41 and R39. The doublet peak with m/z 2830.3 and

2836.3 corresponds to the peptide 14DSSTSAASSSNYC*NQMMKSRNLTK37, with an

Amadori modification at K31 or K37. Based on the data with trypsin and the known

glycation hot spots, modification at K37 is more likely than at K31. In the same manner,

the peptide with m/z 3604.6 is assigned to 53DVQAVC*SQKNVAC*KNGQTNC*YQSYS

TMSIT82, with Amadori modification at K61 or K66. Combined with the data from

tryptic digestion, K66 could be the more likely site. Both a weak doublet peak at 4142.8

and a strong double peak at 4574.2 suggest an Amadori modification at K91, K98 or

K104. The two sets of doublet peaks with m/z 2444.1 and 3734.7 did not correspond to modification of predicted Asp-N peptides. But it is obvious that peak 2444.1 comes from an intra-molecular glucosepane modified peptide whereas peak 3646.6 comes from an

Amadori modified peptide due to the appearance of their prominent corresponding native peptides (m/z=2336.0 and 3572.5, respectively).

Finally, in the chymotryptic digestion map, the peptide with m/z 1473.7 confirms the major intra-molecular glucosepane between K41 and R39. Furthermore, the peptide with m/z 2271.0 indicates that the peptide 80SITDC*RETGSSKYPNC*AY97 is modified by an Amadori product at K91, a site that was not identified by trypsin digestion. The peptides with m/z 4259.8 and 4277.8 have been discussed before and could represent the same intra-molecular glucosepane modified peptide before and after enzymatic hydrolysis between the crosslinked residues.

102

3.4 DISCUSSION

As Biemel et al. proposed in their paper (184), both glucosepane and DODIC

are derived from Amadori products. Consequently clarifying the glycation hot spots for

the Amadori product was a necessary step prior to investigating the sites of glucosepane

or DODIC formation. Two groups, Watkins NG et al. and Brock JW et al., agreed on the

glycation hot spots of RNase A with the order of K41, K7, K37 and K1(78; 185).

Watkins group showed that K1 is more glycated than K37 based on radioactivity measurement of [14C]glucose, which may be more reliable compared with the

semiquantitative methods of mass spectrometry. In their study, 38% of K41, 29% of K7,

24% of K1 and 9% of K37 are glycated by Amadori products. One can assume that

because the Amadori product is the precursor of glucosepane and DODIC, the sites that

are preferentially modified by Amadori product accumulation will have a great potential

for glucosepane formation.

These studies confirmed the four glycation hot spots listed above. Additional

glycation sites also were indicated by enzymatic peptide maps as K66, K91, K98, and

K104. This discrepancy with previous studies may be due to the different incubation

conditions and digestion procedures. As expected, the major crosslink is glucosepane

under nonoxidative conditions as evidenced by the dominant tryptic peptide with m/z

2896.4 (residues 38 to 61) in the digestion of glycated RNase. In this aspect, our in vitro

results conform well with the predominant level of glucosepane observed during analysis

of in vivo modifications(44; 52).

Most established protocols and routine methods used to identify and confirm

crosslinks in vivo utilize complete acid hydrolysis or exhaustive enzymatic digestion of

103

proteins to the constituent amino acids, and their acid-stable modifications. The crosslinks identified by these protocols will include both intra- and intermolecular crosslinks. Thus discrimination of the crosslinks as being primarily intra- or intermolecular, is important for understanding of the in vivo implication of crosslinking.

To form an intra-molecular glucosepane or to form an inter-molecular glucosepane should depend on the availability of nearby arginine residues at the major sites of Amadori product accumulation. Thus we calculated the distance between several pairs of nearby lysine and arginine residues from the crystal structure 1RBX deposited in the protein data bank (Dunbar,J., Yennawar,H.P., Banerjee,S. and

Farber,G.K.unpublished) using PyMOL (DeLano, W.L. The PyMOL Molecular Graphics

System (2002) on World Wide Web http://www.pymol.org). These distances are highlighted in Fig 3.8.

104

K37

R39

R10 K41 K98

R85

K7

Fig 3.8 Distance of all nearby lysine and arginine pairs in RNase A. Calculated lysine and arginine distance: K41 to R39----7.45 Ǻ; K37 to R39----13.67 Ǻ; K98 to R85----4.02 Ǻ,;

K7 to R10----8.11 Ǻ

105

From the values of the paired lysine/arginine distance, the most likely intra-

molecular crosslinks are K98-R85 and K41-R39. In the MALDI tryptic peptide map, the

intensity of the peptide with intra-molecular glucosepane between K41 and R39 (m/z

2896.4) is much higher than the peptide with intra-molecular glucosepane between K98 and R85 (m/z 4523.0). K41 has been unequivocally established as the most prevalent

Amadori accumulation site, whereas K98 is not a glycation hot spots in contradiction to our data showing the Amadori product on K98. Other factors may have minor effects on the level of intra-molecular glucosepane formation at different sites, such as the relative surface accessibility of lysine and arginine. As shown in Table 3.4, the relative surface accessibility (RSA) values of lysines and are predicted by the Structural

Analysis of Residue Interaction Graphs (SARIG) server provided by Weizmann Institute of Science (bioinfo2.weizmann.ac.il/~pietro/SARIG/). The more the arginine is exposed,

the greater propensity for intra-molecular glucosepane formation. The lowest

accessibility of K41 will not inhibit the intra-molecular glucosepane formation as long as

it has a local arginine “mate”. The highest accessibility of R39 makes it not only a

suitable local “mate” for K41, but also a good candidate for inter-molecular crosslinking.

site K1 K7 K31 K37 K41 K61 K66 K91 K98 K10 R10 R33 R39 R85 4 RSA 118.0 31.5 46.9 45.1 21.7 44.3 69.2 75.0 39.6 26.8 31.7 23.4 58.2 42.5

Table 3.4 the relative surface accessibility (RSA) values of lysines and arginines are

predicted by SARIG server.

The existence of both modifications with Δm/z 108 and 126 is very

interesting. These paired modifications suggest that both glucosepane and DODIC

106

modifications are occurring at the same sites. Initially, DODIC modifications were not

anticipated because of low levels observed in vivo as compared to glucosepane.

Glucosepane represents the dominant cross-link in collagen(52) and in lens protein

(132.3-241.7 pmol/mg of protein) whereas DODIC is present in much lower amounts

(1.3-8.0 pmol/mg of protein)(44). However the appearance of DODIC at the same sites as

intra-molecular glucosepane is quite reasonable because glucosepane and DODIC share

the same precursor-the aminoketose. An interesting part of this work is the relative peak

intensity of intra-molecular glucosepane and intra-molecular DODIC at different intra-

crosslinking sites. At the K41 and R39 crosslinking site, the intra-molecular glucosepane

containing peptide (m/z of 2896.4) is much more abundant than the intra-molecular

DODIC containing peptide (m/z of 1914.4). While at the K98 and R85 crosslinking site,

instead, the intra-molecular DODIC containing peptide (4541.0 is more abundant than the

intra-molecular glucosepane containing peptide (4523.0). I noticed that the peptide

67NGQTNC*YQSYSTMSITDC*RETGSSKYPNC*AYKTTQA NK104 has a missed

cleavage site at position K91. When this peptide is modified by intra-molecular

glucosepane and cut at position K91, its m/z value will exactly match with the same

peptide modified with intra-molecular DODIC. Thus the peptide with m/z of 4541.0 may

be derived from two sources, i.e. the peptide 67NGQTNC*YQSYS

TMSITDC*RETGSSKYPNC*AYKTTQANK104 with an intra-molecular DODIC modification, or the peptide 67NGQTNC*YQSYSTMSITDC*RETGSSKYPNC*AYKT

TQANK104 with intra-molecular glucosepane modification and an additional tryptic

cleavage at K91.

107

The most interesting part of this study was to identify the sites of glucose

derived intermolecular crosslinks. As noted above, the peak at m/z 4064.0 and 4226.1 fit the theoretical masses for intermolecular glucosepane containing peptides. Furthermore, the LC/MS/MS fragmentation data of the peptide with m/z of 4064.0, not only confirmed the presence of both crosslinked peptides, but also suggested that K1 and R39 were the

crosslinked residues. As shown in table 3.1, the m/z value of 4226.1 was consistent with

the same peptides being crosslinked with an additional Amadori modification at K7. As

shown in Table 3.5, the relative surface accessibilities of K1 and R39 are the highest

among lysines and arginines of RNase A, respectively. Although the accumulation of

Amadori product at K1 is not the dominant site in RNase A, K1 is one of the glycation

hot sites. With all these factors taken into consideration, the major inter-crosslink found between K1 and R39 is not so surprising.

The observation of DODIC as the only inter-crosslink observed is quite surprising. In in vitro incubations of bovine serum albumin with glucose(184), Biemel et

al. reported about 3 fold higher yields of glucosepane than DODIC. In both human serum

albumin and human lens protein samples, glucosepane is present at a higher level than

DODIC(44). The authors attribute the dominance of glucosepane in vivo to that of its

precursor, N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate, which is irreversibly linked to

the protein, whereas 3-DG, MGO, and GO exist in an equilibrium between a free and

loosely protein-associated form. This character of N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-

lysinate makes it inaccessible to the detoxifying reductases and thus becomes a persistent

glycation agent.

108

Although Lederer et al. proposed a mechanism of glucosepane formation through DODIC as a precursor(48), the absence of glucosepane in the incubation of BSA with 3-deoxyosone mixture excludes DODIC as a precursor of glucosepane(184). Using o-phenylenediamine as a trapping agent for protein bound dicarbonyl groups, Biemel et al. suggests that N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate is the only prominent lysine linked α-diketo compound formed from the reaction of glucose with lysozyme(186). Later they proposed a mechanism for the formation of glucosepane from

Amadori products via N 6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate and the seven member ring intermediate, azepanone.

Thus the formation of glucosepane and DODIC only shares the step from lysine to Amadori product. The formation of glucosepane through N 6-(2,3-dihydroxy-

5,6-dioxohexyl)-L-lysinate requires several enolization steps along the sugar backbone to affect the intramolecular oxidation-reduction reactions, whereas these steps are not required in DODIC formation. I observed that intra-molecular glucosepane is highly favored in the K41-R39 crosslink over DODIC. Interestingly, Biemel et al. observed that the three lysine residues K33, K96, and K116 of lysozyme, which all have partner arginines within 5 Å, have the highest rate of transformation from Amadori to N 6-(2,3- dihydroxy-5,6-dioxohexyl)-L-lysinate(186). This correlation suggested that the guanidino group of arginine has a catalytic effect on this transformation.

By analogy the lysine sites with nearby arginine residues in RNase A should favor the formation of glucosepane, whereas those lysine residues without nearby arginine(s) will be more likely to form DODIC. In our case, the most glycated lysine K41 is more prone to form intra-molecular glucosepane due to close approximation of R39.

109

Due to extremely low surface accessibility, K41 cannot form any inter-molecular crosslinks, including both inter-molecular glucosepane and inter-molecular DODIC.

Among all other glycation hot spots, K1 has very good surface accessibility and relatively high Amadori accumulation so that it’s a good site to form inter-crosslink. Without nearby arginine catalyzed enolization, DODIC formation is favored in this site. This process is presented in Fig 3.8.

Conclusions and future studies:

The work of my thesis can be summarized as follows: 1) I found a novel ascorbic acid derived crosslink which is, to my knowledge, the first

AGE crosslink involving histidine. It is a complex tetramolecular structure

which involves two moles of threose. While studies have shown that

erythrulose is an ascorbic acid degradation product, only small amount of

threose are expected to form by enolization from erythrulose. For all the

reasons above, it is thus not surprising that we were not able to detect the

crosslink in preliminary studies with human tissues. Futures studies should

thus focus on the recently described crosslinks described in 2004 by Reihl

and colleagues(172) and probe their presence in the human lens.

2) In the RNAse A model, I demonstrated with my colleagues that the

Amadori products, glucosepane, and DODIC are major glucose-derived

110

modifications under non-oxidative conditions. 3) I observed intramolecular glucosepane crosslink formation at sites R39-K41 and R85-K98. I also observed an intramolecular DODIC crosslink at R39-K41. 4) The only intermolecular crosslink was also observed at K1-R39, but surprisingly found to be a DODIC crosslink. These results are summarized in cartoon form in Fig. 3.9.

In future studies I plan to investigate glucosepane crosslink formation in collagen for which I designed several strategies: 1) According to my experience based on RNAse A, the lysine glycation hot spots with nearby arginine will have higher chances to form intramolecular glucosepane. Thus, Lys-434, Lys-453, Lys-479, Lys-924 are likely targets.

These, however will have to be confirmed experimentally. 2) In silico probing of the now in Pubmed available triple helical structure of type I collagen will greatly facilitate the search for candidate RK sites. 3) In vitro incubation of rat tail tendon collagen (i,e type I collagen) under non- oxidative conditions in presence of low excess of glucose concentration, and elimination of oxoaldehyde accumulation, will greatly favor glucosepane while minimizing DODICformation. Finally, 4) there is always the possibility of using 14C-labeled glucose and traditional peptide mapping by

HPLC. This approach, however, is obsolete in view of the powerful mass

111

spectrometric methods that are now available. Nevertheless, the isolation of highly crosslinked collagen regions by HPLC or gel electrophoresis may still be needed to unequivocally identify intermolecular crosslinks. In that regard the use of cyanogen bromide (CNBr, CB) digestion to generate well characterized fragments may be useful (94).

Amadori K98 K98 R85 R85

RNase A Amadori RNase A K1 K41 K1 R39 K41 Amadori R39

5,6-diketo K98 R85 K91-R85 Intra-molecular glucosepane Amadori RNase A K1 K41 5,6-diketo R39

K1-R39 Inter-molecular DOGDIC

K41-R39 Intra-molecular glucosepane

Fig 3.9 Schematic representation of crosslinking formation in RNase A.

112

Bibliography :

1. Maillard L GM: Action desacides amines sur les sucres: formation des melanoidines par voie methodique. C R Seances Acad Sci III 154:66-68, 1912

2. Mironova R, Niwa T, Handzhiyski Y, Sredovska A, Ivanov I: Evidence for non- enzymatic of Escherichia coli chromosomal DNA. Mol Microbiol 55:1801- 1811, 2005

3. Monnier VM, Kohn RR, Cerami A: Accelerated age-related browning of human collagen in diabetes mellitus. Proc Natl Acad Sci U S A 81:583-587, 1984

4. Brownlee M, Cerami A, Vlassara H: Advanced glycosylation end products in tissue and the biochemical basis of diabetic complications. N Engl J Med 318:1315-1321, 1988

5. Cerami A: Hypothesis. Glucose as a mediator of aging. J Am Geriatr Soc 33:626-634, 1985

6. Monnier VM, Vishwanath V, Frank KE, Elmets CA, Dauchot P, Kohn RR: Relation between complications of type I diabetes mellitus and collagen-linked fluorescence. N Engl J Med 314:403-408, 1986

7. Hammes HP, Martin S, Federlin K, Geisen K, Brownlee M: Aminoguanidine treatment inhibits the development of experimental diabetic retinopathy. Proc Natl Acad Sci U S A 88:11555-11558, 1991

8. Nakayama H, Mitsuhashi T, Kuwajima S, Aoki S, Kuroda Y, Itoh T, Nakagawa S: Immunochemical detection of advanced glycation end products in lens crystallins from streptozocin-induced diabetic rat. Diabetes 42:345-350, 1993

9. Mitsuhashi T, Nakayama H, Itoh T, Kuwajima S, Aoki S, Atsumi T, Koike T: Immunochemical detection of advanced glycation end products in renal cortex from STZ- induced diabetic rat. Diabetes 42:826-832, 1993

10. Schnider SL, Kohn RR: Effects of age and diabetes mellitus on the solubility and nonenzymatic glucosylation of human skin collagen. J Clin Invest 67:1630-1635, 1981

11. Hamlin CR, Luschin JH, Kohn RR: Partial characterization of the age-related stabilizing factor of post-mature human collagen--II. By the use of trypsin. Exp Gerontol 13:415-423, 1978

113

12. Hamlin CR, Luschin JH, Kohn RR: Partial characterization of the age-related stabilizing factor of post-mature human collagen--I. By the use of bacterial collagenase. Exp Gerontol 13:403-414, 1978

13. LeBella FS, and Paul, G. : J. Gerontol. 20:54-59, 1964

14. Verzijl N, DeGroot J, Thorpe SR, Bank RA, Shaw JN, Lyons TJ, Bijlsma JW, Lafeber FP, Baynes JW, TeKoppele JM: Effect of collagen turnover on the accumulation of advanced glycation end products. J Biol Chem 275:39027-39031, 2000

15. Taniguchi N, Kawahara K, Yone K, Hashiguchi T, Yamakuchi M, Goto M, Inoue K, Yamada S, Ijiri K, Matsunaga S, Nakajima T, Komiya S, Maruyama I: High mobility group box chromosomal protein 1 plays a role in the pathogenesis of rheumatoid arthritis as a novel cytokine. Arthritis Rheum 48:971-981, 2003

16. Hofmann MA, Drury S, Hudson BI, Gleason MR, Qu W, Lu Y, Lalla E, Chitnis S, Monteiro J, Stickland MH, Bucciarelli LG, Moser B, Moxley G, Itescu S, Grant PJ, Gregersen PK, Stern DM, Schmidt AM: RAGE and arthritis: the G82S polymorphism amplifies the inflammatory response. Genes Immun 3:123-135, 2002

17. DeGroot J, Verzijl N, Bank RA, Lafeber FP, Bijlsma JW, TeKoppele JM: Age- related decrease in proteoglycan synthesis of human articular chondrocytes: the role of nonenzymatic glycation. Arthritis Rheum 42:1003-1009, 1999

18. DeGroot J, Verzijl N, Budde M, Bijlsma JW, Lafeber FP, TeKoppele JM: Accumulation of advanced glycation end products decreases collagen turnover by bovine chondrocytes. Exp Cell Res 266:303-310, 2001

19. Kim W, Hudson BI, Moser B, Guo J, Rong LL, Lu Y, Qu W, Lalla E, Lerner S, Chen Y, Yan SS, D'Agati V, Naka Y, Ramasamy R, Herold K, Yan SF, Schmidt AM: Receptor for advanced glycation end products and its ligands: a journey from the complications of diabetes to its pathogenesis. Ann N Y Acad Sci 1043:553-561, 2005

20. Bierhaus A, Humpert PM, Stern DM, Arnold B, Nawroth PP: Advanced glycation end product receptor-mediated cellular dysfunction. Ann N Y Acad Sci 1043:676-680, 2005

21. Wada R, Yagihashi S: Role of advanced glycation end products and their receptors in development of diabetic neuropathy. Ann N Y Acad Sci 1043:598-604, 2005

22. Bai P, Phua K, Hardt T, Cernadas M, Brodsky B: Glycation alters collagen fibril organization. Connect Tissue Res 28:1-12, 1992

23. Haitoglou CS, Tsilibary EC, Brownlee M, Charonis AS: Altered cellular interactions between endothelial cells and nonenzymatically glucosylated laminin/type IV collagen. J Biol Chem 267:12404-12407, 1992

114

24. Silbiger S, Crowley S, Shan Z, Brownlee M, Satriano J, Schlondorff D: Nonenzymatic glycation of mesangial matrix and prolonged exposure of mesangial matrix to elevated glucose reduces collagen synthesis and proteoglycan charge. Kidney Int 43:853-864, 1993

25. Bensch KG, Fleming JE, Lohmann W: The role of ascorbic acid in senile cataract. Proc Natl Acad Sci U S A 82:7193-7196, 1985

26. Sell DR, Monnier VM: Structure elucidation of a senescence cross-link from human extracellular matrix. Implication of pentoses in the aging process. J Biol Chem 264:21597-21602, 1989

27. Grandhee SK, Monnier VM: Mechanism of formation of the Maillard protein cross- link pentosidine. Glucose, fructose, and ascorbate as pentosidine precursors. J Biol Chem 266:11649-11653, 1991

28. Dyer DG, Blackledge JA, Thorpe SR, Baynes JW: Formation of pentosidine during nonenzymatic browning of proteins by glucose. Identification of glucose and other as possible precursors of pentosidine in vivo. J Biol Chem 266:11654- 11660, 1991

29. Biemel KM, Conrad J, Lederer MO: Unexpected carbonyl mobility in aminoketoses: the key to major Maillard crosslinks. Angew Chem Int Ed Engl 41:801-804, 2002

30. Nagaraj RH, Sell DR, Prabhakaram M, Ortwerth BJ, Monnier VM: High correlation between pentosidine protein crosslinks and pigmentation implicates ascorbate oxidation in human lens senescence and cataractogenesis. Proc Natl Acad Sci U S A 88:10257- 10261, 1991

31. Satoh K: Age-related changes in the structural proteins of human lens. Exp Eye Res 14:53-57, 1972

32. Zigman S, Groff J, Yulo T, Griess G: Light extinction and protein in lens. Exp Eye Res 23:555-567, 1976

33. Truscott RJ, Augusteyn RC: The state of sulphydryl groups in normal and cataractous human lenses. Exp Eye Res 25:139-148, 1977

34. Monnier VM, Cerami A: Nonenzymatic browning in vivo: possible process for aging of long-lived proteins. Science 211:491-493, 1981

35. Monnier VM, Cerami A: Detection of nonenzymatic browning products in the human lens. Biochim Biophys Acta 760:97-103, 1983

115

36. Liang JN: Fluorescence study of the effects of aging and diabetes mellitus on human lens alpha-crystallin. Curr Eye Res 6:351-355, 1987

37. Larsen M, Kjer B, Bendtson I, Dalgaard P, Lund-Andersen H: Lens fluorescence in relation to metabolic control of insulin-dependent diabetes mellitus. Arch Ophthalmol 107:59-62, 1989

38. Nagaraj RH, Monnier VM: Non-tryptophan fluorescence and high molecular weight protein formation in lens crystallins of rats with chronic galactosemia: prevention by the aldose reductase inhibitor sorbinil. Exp Eye Res 51:411-418, 1990

39. Nagaraj RH, Monnier VM: Isolation and characterization of a blue fluorophore from human eye lens crystallins: in vitro formation from Maillard reaction with ascorbate and ribose. Biochim Biophys Acta 1116:34-42, 1992

40. Wells-Knecht Kj, Brinkman, E., Thorpe, S. R., and Baynes, J. W.: J. Org. Chem. 60:6246-6247, 1995

41. Brinkmann E, Wells-Knecht, K. J., Thorpe, S. R., and Baynes, J. W.: J. Chem. Soc. Perkin Trans. I 1:2817-2818, 1995

42. Nagaraj RH, Shipanova IN, Faust FM: Protein cross-linking by the Maillard reaction. Isolation, characterization, and in vivo detection of a lysine-lysine cross-link derived from methylglyoxal. J Biol Chem 271:19338-19345, 1996

43. Frye EB, Degenhardt TP, Thorpe SR, Baynes JW: Role of the Maillard reaction in aging of tissue proteins. Advanced glycation end product-dependent increase in imidazolium cross-links in human lens proteins. J Biol Chem 273:18714-18719, 1998

44. Biemel KM, Friedl DA, Lederer MO: Identification and quantification of major maillard cross-links in human serum albumin and lens protein. Evidence for glucosepane as the dominant compound. J Biol Chem 277:24907-24915, 2002

45. Eble AS, Thorpe SR, Baynes JW: Nonenzymatic glucosylation and glucose- dependent cross-linking of protein. J Biol Chem 258:9406-9412, 1983

46. Odani H, Shinzato T, Matsumoto Y, Usami J, Maeda K: Increase in three alpha,beta- dicarbonyl compound levels in human uremic plasma: specific in vivo determination of intermediates in advanced Maillard reaction. Biochem Biophys Res Commun 256:89-93, 1999

47. Glomb MA, Pfahler C: Amides are novel protein modifications formed by physiological sugars. J Biol Chem 276:41638-41647, 2001

116

48. Lederer MO, Klaiber RG: Cross-linking of proteins by Maillard processes: characterization and detection of lysine-arginine cross-links derived from glyoxal and methylglyoxal. Bioorg Med Chem 7:2499-2507, 1999

49. Chellan P, Nagaraj RH: Protein crosslinking by the Maillard reaction: dicarbonyl- derived imidazolium crosslinks in aging and diabetes. Arch Biochem Biophys 368:98-104, 1999

50. Feather MS, Flynn TG, Munro KA, Kubiseski TJ, Walton DJ: Catalysis of reduction of 2-oxoaldehydes (osones) by mammalian aldose reductase and aldehyde reductase. Biochim Biophys Acta 1244:10-16, 1995

51. Thornalley PJ: Amino Acids 6:15-23, 1994

52. Sell DR, Biemel KM, Reihl O, Lederer MO, Strauch CM, Monnier VM: Glucosepane is a major protein cross-link of the senescent human extracellular matrix. Relationship with diabetes. J Biol Chem 280:12310-12315, 2005

53. Verzijl N, DeGroot J, Oldehinkel E, Bank RA, Thorpe SR, Baynes JW, Bayliss MT, Bijlsma JW, Lafeber FP, Tekoppele JM: Age-related accumulation of Maillard reaction products in human articular cartilage collagen. Biochem J 350 Pt 2:381-387, 2000

54. Buckingham B, Reiser KM: Relationship between the content of lysyl oxidase- dependent cross-links in skin collagen, nonenzymatic glycosylation, and long-term complications in type I diabetes mellitus. J Clin Invest 86:1046-1054, 1990

55. Schalkwijk CG, Stehouwer CD, van Hinsbergh VW: Fructose-mediated non- enzymatic glycation: sweet coupling or bad modification. Diabetes Metab Res Rev 20:369-382, 2004

56. Ahmed N, Thornalley PJ, Dawczynski J, Franke S, Strobel J, Stein G, Haik GM: Methylglyoxal-derived hydroimidazolone advanced glycation end-products of human lens proteins. Invest Ophthalmol Vis Sci 44:5287-5292, 2003

57. Cheng R, Feng Q, Argirov OK, Ortwerth BJ: Structure elucidation of a novel yellow chromophore from human lens protein. J Biol Chem 279:45441-45449, 2004

58. Aallen DW, Schroeder, W. B. and Balog, J.: J. Am. Chem. Soc 80:1628-1634, 1958

59. Bookchin RMaG, P. M.: Biochem Biophys Res Commun:86-93, 1968

60. Bunn H, F., Haney, D. N., Gabbay, K.H. and Gallop, P. M.: Biochem Biophys Res Commun 67, 1975

61. Bunn HF, Haney, D. N., Kamin, S., Gabbay, K. H. and Gallop, P.M.: J. Clin. Invest. 57, 1976

117

62. Bunn HF, Shapiro R, McManus M, Garrick L, McDonald MJ, Gallop PM, Gabbay KH: Structural heterogeneity of human hemoglobin A due to nonenzymatic glycosylation. J Biol Chem 254:3892-3898, 1979

63. Shapiro R, McManus MJ, Zalut C, Bunn HF: Sites of nonenzymatic glycosylation of human hemoglobin A. J Biol Chem 255:3120-3127, 1980

64. Sack JS, Andrews, L. C., Magnus, K. A., Hauson, J.C., Rubin, J., and Love, W. E.: Hemoglobin 2:153-169, 1978

65. Acharya AS, Roy RP, Dorai B: Aldimine to ketoamine isomerization (Amadori rearrangement) potential at the individual nonenzymic glycation sites of hemoglobin A: preferential inhibition of glycation by nucleophiles at sites of low isomerization potential. J Protein Chem 10:345-358, 1991

66. Nacharaju P, Acharya AS: Amadori rearrangement potential of hemoglobin at its glycation sites is dependent on the three-dimensional structure of protein. Biochemistry 31:12673-12679, 1992

67. Venkatraman J, Aggarwal K, Balaram P: Helical peptide models for protein glycation: proximity effects in catalysis of the Amadori rearrangement. Chem Biol 8:611-625, 2001

68. Shilton BH, Walton DJ: Sites of glycation of human and horse liver alcohol dehydrogenase in vivo. J Biol Chem 266:5587-5592, 1991

69. Mori N, Bai Y, Ueno H, Manning JM: Sequence-dependent reactivity of model peptides with glyceraldehyde. Carbohydr Res 189:49-63, 1989

70. Mori N, Manning JM: Studies on the Amadori rearrangement in a model system: chromatographic isolation of intermediates and product. Anal Biochem 152:396-401, 1986

71. Garlick RL, Mazer JS: The principal site of nonenzymatic glycosylation of human serum albumin in vivo. J Biol Chem 258:6142-6146, 1983

72. Matthews JB, Hanonia, G. I. H., and Gurd, F. R. N.: Biochemistry 18:1919-1928, 1979

73. Day JF, Thornburg RW, Thorpe SR, Baynes JW: Nonenzymatic glucosylation of rat albumin. Studies in vitro and in vivo. J Biol Chem 254:9394-9400, 1979

74. Day JF, Thorpe SR, Baynes JW: Nonenzymatically glucosylated albumin. In vitro preparation and isolation from normal human serum. J Biol Chem 254:595-597, 1979 75. Means GE, Bender ML: Acetylation of human serum albumin by p-nitrophenyl acetate. Biochemistry 14:4989-4994, 1975

118

76. Gerig JT, Katz KE, Reinheimer JD: Reactions of 2,6-dinitro-4- trifluoromethylbenzenesulfonate with human serum albumin. Biochim Biophys Acta 534:196-209, 1978

77. Iberg N, Fluckiger R: Nonenzymatic glycosylation of albumin in vivo. Identification of multiple glycosylated sites. J Biol Chem 261:13542-13545, 1986

78. Watkins NG, Thorpe SR, Baynes JW: Glycation of amino groups in protein. Studies on the specificity of modification of RNase by glucose. J Biol Chem 260:10629-10636, 1985

79. Lapolla A, Fedele D, Reitano R, Arico NC, Seraglia R, Traldi P, Marotta E, Tonani R: Enzymatic digestion and mass spectrometry in the study of advanced glycation end products/peptides. J Am Soc Mass Spectrom 15:496-509, 2004

80. Richards FM, and Wyckoff, H. W.: The Enzymes (3rd Ed) 4:648-806, 1971

81. Blackburn S: Enzyme Structure and Function:327-376, 1976

82. Walter B, Wold F: The role of lysine in the action of bovine pancreatic ribonuclease A. Biochemistry 15:304-310, 1976

83. Bunn HF, Briehl RW: The interaction of 2,3-diphosphoglycerate with various human hemoglobins. J Clin Invest 49:1088-1095, 1970

84. McDonald MJ, Bleichman M, Bunn HF, Noble RW: Functional properties of the glycosylated minor components of human adult hemoglobin. J Biol Chem 254:702-707, 1979

85. Shaklai N, Garlick RL, Bunn HF: Nonenzymatic glycosylation of human serum albumin alters its conformation and function. J Biol Chem 259:3812-3817, 1984

86. Gonen B, Baenziger J, Schonfeld G, Jacobson D, Farrar P: Nonenzymatic glycosylation of low density lipoproteins in vitro. Effects on cell-interactive properties. Diabetes 30:875-878, 1981

87. Witztum JL, Mahoney EM, Branks MJ, Fisher M, Elam R, Steinberg D: Nonenzymatic glucosylation of low-density lipoprotein alters its biologic activity. Diabetes 31:283-291, 1982

88. Lorenzi M, Cagliero E, Markey B, Henriksen T, Witztum JL, Sampietro T: Interaction of human endothelial cells with elevated glucose concentrations and native and glycosylated low density lipoproteins. Diabetologia 26:218-222, 1984

119

89. Watkins NG, Neglia-Fisher CI, Dyer DG, Thorpe SR, Baynes JW: Effect of phosphate on the kinetics and specificity of glycation of protein. J Biol Chem 262:7207- 7212, 1987

90. Webb BH: J. Dairy Sci 2:81-96, 1935

91. Kato H: Bull. Agr. Cehm. Soc. Japan 20:273-278, 1956

92. Burton HS, and McWeeny, D.J.: Nature 197:266-268, 1963

93. Thornalley P, Wolff S, Crabbe J, Stern A: The autoxidation of glyceraldehyde and other simple monosaccharides under physiological conditions catalysed by buffer ions. Biochim Biophys Acta 797:276-287, 1984

94. Reiser KM, Amigable MA, Last JA: Nonenzymatic glycation of type I collagen. The effects of aging on preferential glycation sites. J Biol Chem 267:24207-24216, 1992

95. Smith L: Histopathologic characteristics and ultrastructure of aging skin. Cutis 43:414-424, 1989

96. le Pape A, Guitton JD, Muh JP: Modification of glomerular basement membrane cross-links in experimental diabetic rats. Biochem Biophys Res Commun 100:1214-1221, 1981

97. Torps S, Arridge, R.S.C., Armeniades, C.D., Baer, E.: Structure-property relationships i tendon as a function of age. Structure of Fibrous Biopolymers 26:197-221, 1975

98. Viidik A: Age-related changes in connective tissue. Lectures on Gerontology Biology of Ageing part A1:173-221, 1982

99. Uitto J: Connective tissue biochemistry of the aging dermis. Age-related alterations in collagen and elastin. Dermatol Clin 4:433-446, 1986

100. Reddi AS: Collagen metabolism in the myocardium of normal and diabetic rats. Exp Mol Pathol 48:236-243, 1988

101. Cox RH: Age-related changes in arterial wall mechanics and composition of NIA Fischer rats. Mech Ageing Dev 23:21-36, 1983

102. Parry DA: The molecular and fibrillar structure of collagen and its relationship to the mechanical properties of connective tissue. Biophys Chem 29:195-209, 1988

103. Verzar F: The ageing of connective tissue. Gerontologia 1:363-378, 1957

120

104. Bjorksten J: Aging: present status of our chemical knowledge. J Am Geriatr Soc 10:125-139, 1962

105. Bailey AJ: Molecular mechanisms of ageing in connective tissues. Mech Ageing Dev 122:735-755, 2001

106. Siegel RC: Lysyl oxidase. Int Rev Connect Tissue Res 8:73-118, 1979

107. Robins SP, Bailey AJ: The chemistry of the collagen cross-links. Characterization of the products of reduction of skin, tendon and bone with sodium cyanoborohydride. Biochem J 163:339-346, 1977

108. Bernstein PH, Mechanic GL: A natural histidine-based imminium cross-link in collagen and its location. J Biol Chem 255:10414-10422, 1980

109. Kang AH: Studies on the location of intermolecular cross-links in collagen. Isolation of a CNBr peptide containing -hydroxylysinonorleucine. Biochemistry 11:1828-1835, 1972

110. Miller EJ: Collagen cross-linking: identification of two cyanogen bromide peptides containing sites of intermolecular cross-link formation in cartilage collagen. Biochem Biophys Res Commun 45:444-451, 1971

111. Nicholls AC, Bailey AJ: Identification of cyanogen bromide peptides involved in intermolecular cross-linking of bovine type III collagen. Biochem J 185:195-201, 1980

112. Bailey AJ, Paul RG, Knott L: Mechanisms of maturation and ageing of collagen. Mech Ageing Dev 106:1-56, 1998

113. Bailey AJ, Shimokomaki MS: Age related changes in the reducible cross-links of collagen. FEBS Lett 16:86-88, 1971

114. Yamauchi M, Chandler GS, Tanzawa H, Katz EP: Cross-linking and the molecular packing of corneal collagen. Biochem Biophys Res Commun 219:311-315, 1996

115. Yamauchi M, London RE, Guenat C, Hashimoto F, Mechanic GL: Structure and formation of a stable histidine-based trifunctional cross-link in skin collagen. J Biol Chem 262:11428-11434, 1987

116. Fujimoto D, Moriguchi T, Ishida T, Hayashi H: The structure of pyridinoline, a collagen crosslink. Biochem Biophys Res Commun 84:52-57, 1978

117. Ogawa T, Ono T, Tsuda M, Kawanishi Y: A novel fluor in insoluble collagen: a crosslinking moiety in collagen molecule. Biochem Biophys Res Commun 107:1252-1257, 1982

121

118. Eyre DR: Crosslink maturation in bone collagen. The chemistry and Biology of Mineralised Connective Tissues:51-55, 1981

119. Kuypers R, Tyler M, Kurth LB, Jenkins ID, Horgan DJ: Identification of the loci of the collagen-associated Ehrlich chromogen in type I collagen confirms its role as a trivalent cross-link. Biochem J 283 ( Pt 1):129-136, 1992

120. Kleter GA, Damen JJ, Kettenes-van den Bosch JJ, Bank RA, te Koppele JM, Veraart JR, ten Cate JM: A novel pyrroleninone cross-link from bovine dentine. Biochim Biophys Acta 1381:179-190, 1998

121. Henkel W, Glanville RW, Greifendorf D: Characterisation of a type-I collagen trimeric cross-linked peptide from calf aorta and its cross-linked structure. Detection of pyridinoline by time-of-flight secondary ion-mass spectroscopy and evidence for a new cross-link. Eur J Biochem 165:427-436, 1987

122. Light N, Bailey AJ: Collagen cross-links: location of pyridinoline in type I collagen. FEBS Lett 182:503-508, 1985

123. Robins SP, Duncan A: Pyridinium crosslinks of bone collagen and their location in peptides isolated from rat femur. Biochim Biophys Acta 914:233-239, 1987

124. Bailey AJ, Light ND, Atkins ED: Chemical cross-linking restrictions on models for the molecular organization of the collagen fibre. Nature 288:408-410, 1980

125. Tanaka S, Avigad G, Eikenberry EF, Brodsky B: Isolation and partial characterization of collagen chains dimerized by sugar-derived cross-links. J Biol Chem 263:17650-17657, 1988

126. Le Pape A, Guitton JD, Muh JP: Distribution of non-enzymatically bound glucose in in vivo and in vitro glycosylated type I collagen molecules. FEBS Lett 170:23-27, 1984

127. Wess TJ, Wess L, Miller A, Lindsay RM, Baird JD: The in vivo glycation of diabetic tendon collagen studied by neutron diffraction. J Mol Biol 230:1297-1303, 1993

128. Brownlee M, Vlassara H, Kooney A, Ulrich P, Cerami A: Aminoguanidine prevents diabetes-induced arterial wall protein cross-linking. Science 232:1629-1632, 1986

129. Bolton WK, Cattran DC, Williams ME, Adler SG, Appel GB, Cartwright K, Foiles PG, Freedman BI, Raskin P, Ratner RE, Spinowitz BS, Whittier FC, Wuerth JP: Randomized trial of an inhibitor of formation of advanced glycation end products in diabetic nephropathy. Am J Nephrol 24:32-40, 2004

130. Degenhardt TP, Alderson NL, Arrington DD, Beattie RJ, Basgen JM, Steffes MW, Thorpe SR, Baynes JW: Pyridoxamine inhibits early renal disease and dyslipidemia in the streptozotocin-diabetic rat. Kidney Int 61:939-950, 2002

122

131. LECLERCQ B, F. ZHENG, M. BERHO, et al: Decreased mortality and albuminuria following pyridoxamine and enalapril therapy in an obese mouse model of type 2 diabetes mellitus with established nephropathy (Abstr.). J. Am. Soc. Nephrol 14:396A, 2003

132. Vasan S, Zhang X, Zhang X, Kapurniotu A, Bernhagen J, Teichberg S, Basgen J, Wagle D, Shih D, Terlecky I, Bucala R, Cerami A, Egan J, Ulrich P: An agent cleaving glucose-derived protein crosslinks in vitro and in vivo. Nature 382:275-278, 1996

133. Alderton WK, Cooper CE, Knowles RG: Nitric oxide synthases: structure, function and inhibition. Biochem J 357:593-615, 2001

134. Yu PH, Zuo DM: Aminoguanidine inhibits semicarbazide-sensitive amine oxidase activity: implications for advanced glycation and diabetic complications. Diabetologia 40:1243-1250, 1997

135. Booth AA, Khalifah RG, Todd P, Hudson BG: In vitro kinetic studies of formation of antigenic advanced glycation end products (AGEs). Novel inhibition of post-Amadori glycation pathways. J Biol Chem 272:5430-5437, 1997

136. Voziyan PA, Metz TO, Baynes JW, Hudson BG: A post-Amadori inhibitor pyridoxamine also inhibits chemical modification of proteins by scavenging carbonyl intermediates of carbohydrate and lipid degradation. J Biol Chem 277:3397-3403, 2002

137. Nagaraj RH, Sarkar P, Mally A, Biemel KM, Lederer MO, Padayatti PS: Effect of pyridoxamine on chemical modification of proteins by carbonyls in diabetic rats: characterization of a major product from the reaction of pyridoxamine and methylglyoxal. Arch Biochem Biophys 402:110-119, 2002

138. Amarnath V, Amarnath K, Amarnath K, Davies S, Roberts LJ, 2nd: Pyridoxamine: an extremely potent scavenger of 1,4-dicarbonyls. Chem Res Toxicol 17:410-415, 2004

139. IACOVELLA Gea: Pyridoxamine decreases levels of alpha-dicarbonyls in plasma of ZDF diabetic rats (Abstra.). Diabetes Metab Res Rev 52:A187, 2003

140. Alderson NL, Chachich ME, Youssef NN, Beattie RJ, Nachtigal M, Thorpe SR, Baynes JW: The AGE inhibitor pyridoxamine inhibits lipemia and development of renal and vascular disease in Zucker obese rats. Kidney Int 63:2123-2133, 2003

141. Miyata T, van Ypersele de Strihou C, Ueda Y, Ichimori K, Inagi R, Onogi H, Ishikawa N, Nangaku M, Kurokawa K: Angiotensin II receptor antagonists and angiotensin-converting enzyme inhibitors lower in vitro the formation of advanced glycation end products: biochemical mechanisms. J Am Soc Nephrol 13:2478-2487, 2002

123

142. Daumer KM, Khan AU, Steinbeck MJ: Chlorination of pyridinium compounds. Possible role of hypochlorite, N-chloramines, and chlorine in the oxidation of pyridinoline cross-links of articular cartilage collagen type II during acute inflammation. J Biol Chem 275:34681-34692, 2000

143. Jain SK, Lim G: Pyridoxine and pyridoxamine inhibits superoxide radicals and prevents lipid peroxidation, protein glycosylation, and (Na+ + K+)-ATPase activity reduction in high glucose-treated human erythrocytes. Free Radic Biol Med 30:232-237, 2001

144. Yang S, Litchfield JE, Baynes JW: AGE-breakers cleave model compounds, but do not break Maillard crosslinks in skin and tail collagen from diabetic rats. Arch Biochem Biophys 412:42-46, 2003

145. Price DL, Rhett PM, Thorpe SR, Baynes JW: Chelating activity of advanced glycation end-product inhibitors. J Biol Chem 276:48967-48972, 2001

146. Wolffenbuttel BH, Boulanger CM, Crijns FR, Huijberts MS, Poitevin P, Swennen GN, Vasan S, Egan JJ, Ulrich P, Cerami A, Levy BI: Breakers of advanced glycation end products restore large artery properties in experimental diabetes. Proc Natl Acad Sci U S A 95:4630-4634, 1998

147. Huijberts MS, Wolffenbuttel BH, Boudier HA, Crijns FR, Kruseman AC, Poitevin P, Levy BI: Aminoguanidine treatment increases elasticity and decreases fluid filtration of large arteries from diabetic rats. J Clin Invest 92:1407-1411, 1993

148. Asif M, Egan J, Vasan S, Jyothirmayi GN, Masurekar MR, Lopez S, Williams C, Torres RL, Wagle D, Ulrich P, Cerami A, Brines M, Regan TJ: An advanced glycation endproduct cross-link breaker can reverse age-related increases in myocardial stiffness. Proc Natl Acad Sci U S A 97:2809-2813, 2000

149. Vaitkevicius PV, Lane M, Spurgeon H, Ingram DK, Roth GS, Egan JJ, Vasan S, Wagle DR, Ulrich P, Brines M, Wuerth JP, Cerami A, Lakatta EG: A cross-link breaker has sustained effects on arterial and ventricular properties in older rhesus monkeys. Proc Natl Acad Sci U S A 98:1171-1175, 2001

150. Kass DA, Shapiro EP, Kawaguchi M, Capriotti AR, Scuteri A, deGroof RC, Lakatta EG: Improved arterial compliance by a novel advanced glycation end-product crosslink breaker. Circulation 104:1464-1470, 2001

151. Vasan S, Foiles P, Founds H: Therapeutic potential of breakers of advanced glycation end product-protein crosslinks. Arch Biochem Biophys 419:89-96, 2003

152. Basta G, Schmidt AM, De Caterina R: Advanced glycation end products and vascular inflammation: implications for accelerated atherosclerosis in diabetes. Cardiovasc Res 63:582-592, 2004

124

153. Oldfield MD, Bach LA, Forbes JM, Nikolic-Paterson D, McRobert A, Thallas V, Atkins RC, Osicka T, Jerums G, Cooper ME: Advanced glycation end products cause epithelial-myofibroblast transdifferentiation via the receptor for advanced glycation end products (RAGE). J Clin Invest 108:1853-1863, 2001

154. Twigg SM, Cao Z, SV MC, Burns WC, Brammar G, Forbes JM, Cooper ME: Renal connective tissue growth factor induction in experimental diabetes is prevented by aminoguanidine. Endocrinology 143:4907-4915, 2002

155. Birrell AM, Heffernan SJ, Kirwan P, McLennan S, Gillin AG, Yue DK: The effects of aminoguanidine on renal changes in a baboon model of Type 1 diabetes. J Diabetes Complications 16:301-309, 2002

156. Yamauchi A, Takei I, Makita Z, Nakamoto S, Ohashi N, Kiguchi H, Ishii T, Koike T, Saruta T: Effects of aminoguanidine on serum advanced glycation endproducts, urinary albilmin excretion, mesangial expansion, and glomerular basement membrane thickening in Otsuka Long-Evans Tokushima fatty rats. Diabetes Res Clin Pract 34:127- 133, 1997

157. Corman B, Duriez M, Poitevin P, Heudes D, Bruneval P, Tedgui A, Levy BI: Aminoguanidine prevents age-related arterial stiffening and cardiac hypertrophy. Proc Natl Acad Sci U S A 95:1301-1306, 1998

158. Sell DR, Nelson JF, Monnier VM: Effect of chronic aminoguanidine treatment on age-related glycation, glycoxidation, and collagen cross-linking in the Fischer 344 rat. J Gerontol A Biol Sci Med Sci 56:B405-411, 2001

159. Alderson NL, Metz, T.O., Chachich, M.E., Baynes, J.W., and Thrope, S.R. : Diabetes 50(Suppl. 2), 2001

160. Cheng R, Lin B, Lee KW, Ortwerth BJ: Similarity of the yellow chromophores isolated from human cataracts with those from ascorbic acid-modified calf lens proteins: evidence for ascorbic acid glycation during cataract formation. Biochim Biophys Acta 1537:14-26, 2001

161. Cheng R, Lin B, Ortwerth BJ: Separation of the yellow chromophores in individual brunescent cataracts. Exp Eye Res 77:313-325, 2003

162. Cheng R, Lin B, Ortwerth BJ: Rate of formation of AGEs during ascorbate glycation and during aging in human lens tissue. Biochim Biophys Acta 1587:65-74, 2002

163. Simpson GL, Ortwerth BJ: The non-oxidative degradation of ascorbic acid at physiological conditions. Biochim Biophys Acta 1501:12-24, 2000

125

164. Fan X, Reneker LW, Obrenovich ME, Strauch C, Cheng R, Jarvis SM, Ortwerth BJ, Monnier VM: Vitamin C mediates chemical aging of lens crystallins by the Maillard reaction in a humanized mouse model. Proc Natl Acad Sci U S A 103:16912-16917, 2006

165. Varma SD, Richards RD: Ascorbic acid and the eye lens. Ophthalmic Res 20:164- 173, 1988

166. Varma SD: Ascorbic acid and the eye with special reference to the lens. Ann N Y Acad Sci 498:280-306, 1987

167. Lee KW, Mossine V, Ortwerth BJ: The relative ability of glucose and ascorbate to glycate and crosslink lens proteins in vitro. off. Exp Eye Res 67:95-104, 1998

168. Dai Z, Nemet I, Shen W, Monnier VM: Isolation, purification and characterization of histidino-threosidine, a novel Maillard reaction protein crosslink from threose, lysine and histidine. Arch Biochem Biophys 463:78-88, 2007

169. Kipp BH, Faraj C, Li G, Njus D: Imidazole facilitates electron transfer from organic reductants. Bioelectrochemistry 64:7-13, 2004

170. Rothery EL, Mowat CG, Miles CS, Walkinshaw MD, Reid GA, Chapman SK: Histidine 61: an important heme ligand in the soluble fumarate reductase from Shewanella frigidimarina. Biochemistry 42:13160-13169, 2003

171. Argirov OK, Lin B, Olesen P, Ortwerth BJ: Isolation and characterization of a new advanced glycation endproduct of dehydroascorbic acid and lysine. Biochim Biophys Acta 1620:235-244, 2003

172. Reihl O, Lederer MO, Schwack W: Characterization and detection of lysine-arginine cross-links derived from dehydroascorbic acid. Carbohydr Res 339:483-491, 2004

173. Nagaraj RH, Monnier VM: Protein modification by the degradation products of ascorbate: formation of a novel pyrrole from the Maillard reaction of L-threose with proteins. Biochim Biophys Acta 1253:75-84, 1995

174. Nagaraj RH, Sady C: The presence of a glucose-derived Maillard reaction product in the human lens. FEBS Lett 382:234-238, 1996

175. Nadkarni DV, Sayre LM: Structural definition of early lysine and histidine adduction chemistry of 4-hydroxynonenal. Chem Res Toxicol 8:284-291, 1995

176. Slatter DA, Avery NC, Bailey AJ: Identification of a new cross-link and unique histidine adduct from bovine serum albumin incubated with malondialdehyde. J Biol Chem 279:61-69, 2004

126

177. Munch G, Mayer S, Michaelis J, Hipkiss AR, Riederer P, Muller R, Neumann A, Schinzel R, Cunningham AM: Influence of advanced glycation end-products and AGE- inhibitors on nucleation-dependent polymerization of beta-amyloid peptide. Biochim Biophys Acta 1360:17-29, 1997

178. Hobart LJ, Seibel I, Yeargans GS, Seidler NW: Anti-crosslinking properties of carnosine: significance of histidine. Life Sci 75:1379-1389, 2004

179. Liu Y, Xu G, Sayre LM: Carnosine inhibits (E)-4-hydroxy-2-nonenal-induced protein cross-linking: structural characterization of carnosine-HNE adducts. Chem Res Toxicol 16:1589-1597, 2003

180. Tessier FJ, Monnier VM, Sayre LM, Kornfield JA: Triosidines: novel Maillard reaction products and cross-links from the reaction of triose sugars with lysine and arginine residues. Biochem J 369:705-719, 2003

181. Reihl O, Rothenbacher TM, Lederer MO, Schwack W: Carbohydrate carbonyl mobility--the key process in the formation of alpha-dicarbonyl intermediates. Carbohydr Res 339:1609-1618, 2004

182. Monnier VM, Mustata GT, Biemel KL, Reihl O, Lederer MO, Zhenyu D, Sell DR: Cross-linking of the extracellular matrix by the maillard reaction in aging and diabetes: an update on "a puzzle nearing resolution". Ann N Y Acad Sci 1043:533-544, 2005

183. Miyata T, Wada Y, Cai Z, Iida Y, Horie K, Yasuda Y, Maeda K, Kurokawa K, van Ypersele de Strihou C: Implication of an increased oxidative stress in the formation of advanced glycation end products in patients with end-stage renal failure. Kidney Int 51:1170-1181, 1997

184. Biemel KM, Reihl O, Conrad J, Lederer MO: Formation pathways for lysine- arginine cross-links derived from hexoses and pentoses by Maillard processes: unraveling the structure of a pentosidine precursor. J Biol Chem 276:23405-23412, 2001

185. Brock JW, Hinton DJ, Cotham WE, Metz TO, Thorpe SR, Baynes JW, Ames JM: Proteomic analysis of the site specificity of glycation and carboxymethylation of ribonuclease. J Proteome Res 2:506-513, 2003

186. Biemel KM, Lederer MO: Site-specific quantitative evaluation of the protein glycation product N6-(2,3-dihydroxy-5,6-dioxohexyl)-L-lysinate by LC-(ESI)MS peptide mapping: evidence for its key role in AGE formation. Bioconjug Chem 14:619-628, 2003

127