Decrease in skin collagen glycation with improved glycemic control in patients with insulin dependent diabetes mellitus

Decrease in skin collagen glycation with

improved glycemic control in patients with

insulin-dependent diabetes mellitus.

T J Lyons, … , J A Dunn, J W Baynes

J Clin Invest. 1991;87(6):1910-1915. https://doi.org/10.1172/JCI115216.

Glycation, oxidation, and nonenzymatic browning of protein have all been implicated in the development of diabetic complications. The initial product of glycation of protein,

fructoselysine (FL), undergoes further reactions, yielding a complex mixture of browning products, including the fluorescent lysine-arginine cross-link, pentosidine. Alternatively, FL may be cleaved oxidatively to form N(epsilon)-(carboxymethyl)lysine (CML), while glycated hydroxylysine, an amino-acid unique to collagen, may yield

N(epsilon)-(carboxymethyl)hydroxylysine (CMhL). We have measured FL, pentosidine, fluorescence (excitation = 328 nm, emission = 378 nm), CML, and CMhL in insoluble skin collagen from 14 insulin-dependent diabetic patients before and after a 4-mo period of intensive therapy to improve glycemic control. Mean home blood glucose fell from 8.7 +/- 2.5 (mean +/- 1 SD) to 6.8 1.4 mM (P less than 0.005), and mean glycated hemoglobin (HbA1) from 11.6 +/-2.3% to 8.3 +/- 1.1% (P less than 0.001). These changes were accompanied by a significant decrease in glycation of skin collagen, from 13.2 +/- 4.3 to 10.6 +/- 2.3 mmol FL/mol lysine (P less than 0.002). However, levels of browning and oxidation products (pentosidine, CML, and CMhL) and fluorescence were unchanged. These results show that the glycation of long-lived proteins can be decreased by improved glycemic control, but suggest that once cumulative damage to collagen by browning and oxidation reactions has occurred, it may not be […]

Research Article

Decrease

in

Skin Collagen Glycation with Improved Glycemic

Control

in

Patients with Insulin-dependent Diabetes Mellitus

Timothy J. Lyons,** Karen E. Bailie,* Daniel G.Dyer,*John A. Dunn,'andJohn W.Baynes"1

*Department ofMedicine, Aitnagelvin Hospital, Londonderry,NorthernIreland, UnitedKingdom; tSir George E. Clark MetabolicUnit, Royal VictoriaHospital, Belfast, Northern Ireland, UnitedKingdom; andIDepartmentofChemistry, and

1School ofMedicine, UniversityofSouth Carolina, Columbia, South Carolina29208

Abstract

Glycation, oxidation, and nonenzymatic browning of protein have all been implicated in thedevelopment of diabetic compli-cations. The initial product of glycation of protein, fructosely-sine(FL),undergoes furtherreactions,yielding a complex mix-tureofbrowningproducts,includingthefluorescent

lysine-ar-ginine cross-link, pentosidine. Alternatively, FL may be cleavedoxidativelytoformNf-(carboxymethyl)lysine (CML), while glycated hydroxylysine, an amino-acid unique to colla-gen, mayyieldN4-(carboxymethyl)hydroxylysine (CMhL).We have measured FL,pentosidine,fluorescence(excitation=328 nm,emission =378 nm), CML, and CMhL in insoluble skin

collagen from14insulin-dependent diabetic patients beforeand

aftera 4-moperiod ofintensive therapy to improve glycemic control. Meanhome blood glucose fellfrom8.7±2.5 (mean±1 SD)to6.8±1.4mM(P<0.005),and mean glycated

hemoglo-bin

_(HbA,)

from 11.6±2.3%to83±1.1% (P<0.001). These changeswereaccompanied byasignificantdecrease in glyca-tionofskin collagen,from 13.2±4.3to10.6±2.3mmolFL/mol lysine (P<0.002). However, levelsof browningandoxidation

products (pentosidine, CML, and CMhL) and fluorescence

were unchanged. These results show that the glycationof long-livedproteinscan be decreased by improved glycemic control, but suggest that oncecumulative damagetocollagen by brown-ing and oxidationreactionshasoccurred, itmay not bereadily

reversed. Thus, in diabetic patients, institution and

mainte-nance ofgoodglycemic control atany time couldpotentially

limit the extent of subsequentlong-term damage toproteinsby

glycation and oxidation reactions. (J. Clin. Invest. 1991.

87:1910-1915.)Keywords:glycation*nonenzymatic browning

* Maillard reaction * diabetes * oxidation

Introduction

Themechanismsunderlyingthedevelopment ofthe

complica-tions of diabetes are notfullyunderstood. Eventhe relation-shipbetweenglycemiccontrol and theriskofdeveloping

com-plications remains unclear,althoughthereisnow a consensus thathyperglycemia does,initself,playanimportantroleinthe

AddressreprintrequeststoDr.TimothyJ.Lyons, Division ofEndocri-nology,Metabolism, andNutrition,MedicalUniversityofSouth Caro-lina, 171 AshleyAvenue,Charleston, SC29425,which isDr.Lyons's

currentaddress.

Receivedfor publication 30 October 1990 and in revisedform29

January1991.

development of retinopathy, nephropathy, and neuropathy (1). Theprocesses of glycation andnonenzymatic browningof

proteins provideanattractivehypothesistolink hyperglycemia

with the development of complications (2-4). The first step in

this reaction pathway, glycation, involvesthe nonenzymatic condensation ofglucosewithfree aminogroupsintheprotein, primarilythe e-aminogroups of lysine residues, formingthe

Amadori adduct, fructoselysine(FL)' (Fig. 1). FLmay react

furthertoinitiateacomplex series ofreactions, which leadto

theaccumulation ofcovalently attached brown and fluorescent

products,cross-links,and otherchemical modifications in

pro-teins. These reactionsare knowncollectively as Maillardor

nonenzymatic browning reactions (5, 6). This reaction is readily illustrated by the browning and cross-linking of

pro-teins on incubation with glucose in physiological buffers in vitro.

Anumberof chemicalandphysical changesoccurin hu-manskin collagen with age.Fluorescence, cross-linking, and

resistancetoenzymatic degradation increase with age,while solubilityandelasticitydecrease(reviewed in 7).Indiabetes, in

concert with increased glycation (8-10) and nonenzymatic

browning(1 1-13) of collagen, these age-related changes in the physicalandchemicalproperties ofcollagenappear tobe accel-erated(11-15). Thissuggests a rolefor glucoseand

nonenzy-maticbrowning reactionsin thedevelopment ofage-like chemi-cal andfunctional alterations of collagen in diabetes. Although

the extentofglycation ofskin collagendoes not appear to

corre-latedirectly withthe presenceofcomplicationsindiabetes (9, 10),thelong-term effect of increasedglycation, i.e.,the

brown-ingreaction,maybemore relevant.Thus, ingroupsof insulin-dependent diabetic patients, matched forage anddurationof

diabetes,thereisasignificantcorrelationbetweenskin collagen fluorescenceand theseverity ofretinopathy, nephropathy,

arte-rialstiffness, and joint stiffness (16).Increased free-radical-me-diated oxidative damagetobiomolecules,includingbothlipids

andproteins,has also beenproposedasamechanism contribut-ingtothedevelopmentof diabeticcomplications(reviewed in 17, 18).Indeed,freeradical reactionsmay inthemselves gener-atefluorescent products andcross-linksin proteins(19), and thus the changesincollagen withage andindiabetesmay be thecombinedresultofincreases inglycation, browning,and

oxidationreactions.

Severaldistinct chemicalproducts ofglycationand brown-ing reactions of proteinhave now beenmeasured in human proteins. These compounds (Fig. 1) includetheAmadori

ad-duct, FL, and three products of later stages of the Maillard

1. Abbreviations used in thispaper: CMhL,

N4carboxymethyl)hy-droxylysine; CML,

N'-(carboxymethyl)lysine;

FL,

Nf-(l-deoxy-fruc-tose-l-yl)L-lysine; HbA,, hemoglobin

A1;

MHBG,meanhome blood glucose.

1910 T.J. Lyons,K.E. Bailie,D.G.Dyer, J. A. Dunn,and J. W.Baynes J.Clin.Invest.

©TheAmerican Societyfor ClinicalInvestigation,Inc.

0021-9738/91/06/1910/06 $2.00

OR

hdH

CARSOXME LLYSINE

LYSNE

+w

"Z

400

Arginkw-GLUCOSE PENTOSIDOINE

Figure 1. Maillard reaction pathways for formation of fructoselysine,

pentosidine,andN-(carboxymethyl)lysine.FLis the Amadori com-pound, the first stable intermediate in the Maillard reaction. CML and the analogous compound CMhL (not shown), are formedby

ox-idativecleavage of Amadori adducts to lysine and hydroxylysine,

re-spectively.The mechanism of formation of the fluorescent crosslink, pentosidine, is unknown.

reaction: pentosidine2 (20-23), N-(carboxymethyl)lysine

(CML) (24-27), and NE-(carboxymethyl)hydroxylysine (CMhL)(27). Theconcentration oftheinitialproduct, FL, in

long-livedproteins, such as lens proteins (26, 28) and skin

col-lagen(27), increases in response to hyperglycemia in diabetes.

Amongthe laterproductsof the Maillardreaction,pentosidine

is a fluorescentcross-link formed betweenlysineand arginine

residuesduring the browning process (20-23), while CML and

CMhL, whicharecolorless, areformedbyoxidativecleavageof

carbohydrateadducts to lysine andhydroxylysine residues in protein,respectively (24-27).All threeof theselate-stage

prod-uctsof the Maillard reactionrequireoxygen fortheirformation

(20,24, 25), i.e., they are either direct products of oxygen

radi-cal reactions or are formed by further reaction of oxidation

products.In addition, they all accumulate gradually with age in

skin collagen (27), and at an accelerated rate in diabetes

(21, 23).

In this study, we have measured FL, pentosidine, CML,

and CMhL and Maillard-type fluorescence (excitation (Ex)

=328nm, emission (Em)=378 nm) in insoluble skin collagen

frompatients with insulin-dependent diabetes, both before and

afteraperiod of improved glycemic control. Our aim was to

discover ifimprovementsin control would lead to a decrease in

thelevelof any of these Maillard reaction products in diabetic

skincollagen,and thus, perhaps, to a reversal of the potentially

damagingeffects ofglycation, browning, and oxidation to

long-livedproteins in diabetes.

2. ThecompoundMaillard Fluorescent Product #1 (MFP-1)described inreferences22 and 23 was originally identified as a fluorescent cross-link formed during browning andcross-linkingof proteins by glucose.

MFP-I was isolated and characterized recently in our laboratory and shown by nuclear magnetic resonance spectroscopy and mass spectrom-etry to be identical to the compound, pentosidine, previously

charac-terizedby Sell and Monnier (20, 21). There is some uncertainty about whetherpentosidineis formed in vivo from ribose (20, 21), glucose (22, 23), or other sugars (23a).

Methods

Patientselection. Patients

_(eight

male, sixfemale)with

insulin-depen-dent diabetesmellitus, whowerein relativelypoorglycemic control,

but motivatedto_improve,wererecruited fromthediabetesclinics of

Altnagelvin

Hospital, Londonderry,and theRoyal VictoriaHospital,

Belfast, Northern Ireland. Theirmeanage was31.9±10.5 (19-51)yr

(mean±SD, range),andmean duration of diabetes 12.5±10.6 (0-38)

yr.Threepatientshadnewly

diagnosed

diabetes. Of the remainder,10 werereceiving twicedaily

injections

ofregularand NPH (Isophane)

insulin, andone a

single daily injection.

Meandailyinsulindosewas

74±11_(52-88)U.The_studywasapprovedbytheEthical andHuman

Subjects

Committees of the

_{participating}

institutions, and informed

consentwasobtainedfrom allvolunteers.

Study design.

Afteraninitial_assessment,eachpatientwastaughtto

perform

homeblood

glucose

monitoringusingaMemoryGlucometer

(AmesDiv.,MilesLaboratories_Inc.,Elkhart, IN)and wasgivena

Mem-ory Glucometeron loanfor the duration of the study. Patients

per-formed blood_glucosemeasurementsfour timesdaily (beforemealsand at_{bedtime) throughout}the_study. Fora2-wk "run-in" periodbefore

the first skin

_biopsy,

no effort was made to alter glycemiccontrol.

Duringthis_period, home blood_glucose monitoring results were

re-corded,and

hemoglobin

A, (HbA,)

wasmeasuredontwooccasions.

Attheendof the run-in_period,afull-thickness,elliptical(1.0

x

0.5

cm)skin

_biopsy

wasobtained,under local anesthesia, fromtheupper inner_aspectof the buttock. The_biopsysampleswerewashedin saline

and stored at -70'C. Patients then entered a program of intensive

managementto_improve

_glycemic

control,andwereseen on an

individ-ualbasis_byoneofus(T. LyonsorK.Bailie)atleastonce everytwo

weeks.Ateachvisit,the_patientsreceivedappropriateindividual

educa-tion, including adviceon theadjustment of insulin dosage. Between

visits,oneofus wasavailableatalltimestogiveadvicebytelephone.

Each_patientwasalso assessedinitially,andsubsequently reviewed,as

necessary,byadietitian. Throughout thestudy, HbAlmeasurements

were performedevery twoweeks. The results of home bloodglucose

monitoringwereanalyzed

using

theAmes"Glucofacts"program,and wereexpressedas"mean home bloodglucose" (MHBG)on aweekly basis.Attheconclusion of thestudy,asecond skin biopsywasobtained from the

corresponding

siteontheoppositebuttock. Themean

inter-val betweenfirst and second skinbiopsieswas 120±19(92-157)d.

Analyticalprocedures.

HbAIwasmeasuredbyagargel

electrophore-sis(29);thenormalrangeinourlaboratoryis 3.6-7.2%.Insoluble colla-genwasisolatedfrom skin

biopsy

samplesbymechanical scrapingand

solvent extractionas

previously

described(9).Thepreparationof

N-formyl-NE-fructoselysine

(the standard for the measurement ofFL),

CML,and CMhL have been describedpreviously(24, 25,27).A

radio-active_pentosidinestandardwaspreparedfromglucose,

N-acetylargin-ine,

and

_{N"-acetyl-[4,5-3H]ysine}

of knownspecificradioactivity, and

purified byreversed phaseHPLC_(23a). Levelsof FL, CML,CMhL,

and

lysine

in insoluble_collagen weremeasured bygas

chromatogra-phy/massspectroscopywith selected ion monitoring (SIM-GC/MS),as

previouslydescribed in detail_{(26, 27).}Briefly,formeasurement ofFL,

collagen sampleswerehydrolyzed in 7.8N

HCI

(24h,110C, under

N2),

yielding

40%conversion ofFLtotheanalyte, furosine(26, 27,30).

Because of_partial conversion ofFL to CML during the hydrolysis

reaction(andpossibleconversion ofglycatedhydroxylysinetoCMhL),

CML and CMhLweremeasuredseparatelyinNaBH4-reducedsamples

(27).Furosine_{(formed during}acid_hydrolysisof FL), CML,andCMhL weremeasuredastheirNo-trifluoroacetylmethylesterderivativesby

SIM-GC/MS.Theassayswerestandardizedusing standardcurves

pre-pared bystandardaddition,and theconcentrationswerenormalizedto

the

_lysine

_(forFLand_CML)orhydroxylysine(forCMhL)contentof

the_collagen.The_pentosidinecontentof skin collagenwasmeasured,

following NaBH4-reductionandhydrolysisin 6N

HCI

(24hat1

10°C),

byreversed phaseHPLC usingfluorometric detection (Ex= 328nm,

Em = 378nm) (22, 23a).

Quantitation

wasbasedonfluorescencearea

units,

using

astandardcurvepreparedwith knownamountsof

measurementof fluorescence (Ex= 328nm, Em=378nm), collagen was digested with pepsin (5%, wt/wt) in 0.5 M acetic acid (adjusted to pH2with 6 MHC1),for 24 h at 37°C. The samples were clarified by centrifugation for measurement of fluorescence. More than 98% ofthe hydroxyproline content of the skin sample, determined as previously described (9), was solubilized by this method. The fluorescence read-ings were normalized to the hydroxyproline content of the solution. For eachanalysis all samples were analyzed together in a single batch to avoid interassay variations. All laboratory work on the samples was also done"blind",i.e., with no knowledge of sample identity.

Statistics. Results are expressed throughout as mean±ISD.

Differ-encesbetween groupsbeforeandafter improved glycemic control were comparedusing a paired t test.

Results

Measurements ofboth MHBG and

_HbA,

(Fig. 2) show that

thereweresignificant improvements in glycemiccontrolinthe

patientgroupduring thecourseofthisstudy.Theoverall

differ-encesandstatistical analyses for theseandothermeasurements

discussed beloware summarized in Table I. Fig. 3showsthe

resultsofall the analysesfor individual patients.Itisapparent

that, inadditiontoMHBG (Figs.2 A and3A) and

_HbAl

(Figs.

2 Band 3 B), theextentof glycation of skincollagen also

de-creased significantly between the beginning and end ofthe

study(Fig. 3 C and Table I). Glycation of collagen correlated

significantly with MHBG (r = 0.65, P < 0.02) andHbA, (r

=0.89,P<0.0001)atthebeginning, butnot at theend,ofthe study. Similarly,MHBGduring the2-wkrun-inperiod corre-latedsignificantly with initialHbA, (r=0.56, P < 0.05), but, as

might beexpected,this correlationwaslostwith improved gly-cemiccontrol. Over the course of the study, the relative

de-crease inindividualpatient

_HbAl

([initial-final]/initial)

corre-mE 11

0,

0

0

0

02 m

E

0

c a

I

4.

0 2 4 6 8 1012 14

Time After FirstBiopsy (weeks)

Figure2. Sequentialmeanhomebloodglucose (A) and HbA, (B)

results(mean±I SD)for all 14patients duringthecourseof the study. The first skinbiopsieswereperformedattime 0.

TableI. Glycation and Oxidation Products inInsolubleSkin

Collagen beforeandafter Improved Glycemic Control

Measurement* Before After Ps

MHBG (mM) 8.7±2.5 6.8±1.4 <0.005

(4.0-12.8) (4.0-9.2)

HbAj(%)

11.6±2.3 8.3±1.1 <0.001 (76-16.6) (6.8-10.2)

mmolFL/mol Lys 13.2±4.3 10.6±2.3 <0.002 (7.2-22.7) (7.1-15.2)

mmolCML/mol Lys 0.78±0.46 0.78±0.45 NS

(0.20-1.81) (0.14-1.86)

mmol CMhL/mol Hyl 2.18±1.06 2.19±1.11 NS

(0.73-4.40) (0.49-4.48)

gmolpentosidine/mol Lys 9.6±6.6 9.7±6.4 NS (1.5-22.5) (1.0-23.7)

fluorescenceU/agHyp 1.25±0.62 1.22±0.51 NS (0.34-2.55) (0.40-2.33)

*_{Mean±SD;ranges shown in}parentheses.

$Significanceestimated bypairedttest;NS,P>0.1.

lated significantly with the relative decreases in MHBG (r

=0.68,P< 0.01) andFLin collagen (r= 0.72, P<0.005).

Thus, changes in mean blood glucose concentration were

mirrored by changes inglycation of both hemoglobin and col-lagen.This isapparentinFig. 3, A-C, which shows visually that the greatestreductions in collagen FLoccurred in those

pa-tients with highest initialFLvalues andwhoachievedthe great-estdecreasesin MHBG and

_HbA,.

Asshown inFig. 3, D-G and summarized in Table I, levels ofCML, CMhL,pentosidine,and totalfluorescencewere unaf-fected by the4-moperiodofimproved glycemiccontrol.All of these parameters showed wide variations amongindividuals,

reflecting differencesin age (Fig. 3), aswell as durationand

severityof diabetes. Therewere nosignificant correlations of

MHBG, HbA,,orcollagenFLwithCML, CMhL,pentosidine,

ortotalfluorescence.

One otherimportantobservation made in this study is

sum-marized inFig. 4andits legend. Indiabeticpatients, eitherat thebeginningorend ofthestudy, therewas astrongcorrelation betweenanytwoofthemeasuresoflong-termchemical modifi-cationof collagen, i.e.,nonenzymatic browning products (pen-tosidine),oxidation products(CMLandCMhL),andtotal

fluo-rescence. Thus, the four parameters measured inthis study

provideaconsistentassessmentof theextentof Maillard

reac-tion damagetocollagen. Discussion

Collagen, in itsvarious forms, isa ubiquitousproteinin the

body. Collagen abnormalitiesmaythereforehavewidespread

consequences, such asinterference withbasementmembrane function in small vessels andglomeruli,orwith theprocessesof growth andremodellingof tissues. Such problemsarewell

rec-ognized features oflong-term diabetes. Increasedglycationand

nonenzymatic browning of collagen and other structural

pro-teins by glucose is thoughttobe oneof perhaps several

mecha-nismscontributingtopathophysiologicalchanges

characteris-1912 T.J. Lyons, K. E. Bailie,D. G.Dyer, J.A.Dunn,and J. W Baynes

I

A

'1111

6 111111~~0

-0

1140.,--

4-

2-211

B

1

E

co

m

n 1

a:

Ir

r__ Before

12 A MAfter

4-6B

12-8

4

0-

20--a 15-E

E- tioi

fIl

Il

Increasing Age

(II

N

10 D o sod

m__L6L0

Eg

E 5

0.0-3_{0.8 crain Ae}

Figure 3. Individual changesin indicesof glycemic control (A, B), andglycation (C), non-enzymatic browning (D, E), andoxidation(F,

G) of collagen during the courseof the study. Two data points are missing from the pento-sidine (D) and total flu-orescence(E) results, becauseof insufficient sample.Asterisks in B denotepatientswith initial

_HbAj

>12%.

ticof aging (31)and thedevelopment ofthecomplications of diabetes (3, 5, 6, 16).The extentofglycationoflong-lived

pro-teins isdirectlyrelatedtoambient glucose concentration

(26-28) and does not change significantly withage inthe nondia-betic population (26-28, 32, 33). While a numberof earlier studies have shown thatglycation of collagen -is increased in

response to hyperglycemia in diabetes (8-10, 15) and have found a strong correlation between glycation ofhemoglobin

andcollagen (9, 10),ourworkdemonstrates thatglycation of

human skin collagen may besignificantly reducedwithin as short atimeas a4-moperiod ofimprovedglycemiccontrol. To our knowledge only one otherstudy (34) has addressed the

question of reversibility of glycation of collagen. Inthat case theinvestigators foundnodecrease inglycation oftail tendon

collagenindiabeticratstreatedwithinsulin foran8-wkperiod.

Thefailureto detectdifferencesin collagenglycationin these

animal experiments may beexplained by the shorter period studied.

Becauseoftheimpracticality ofmaintainingconstant gly-cemia andobtainingmultipleskinbiopsiesfrompatients,it is

difficulttoestimateanexacthalf-time for reversal ofglycation

ofhuman skincollagen.However,the results of thisstudy sug-gest that the half-time for response ofcollagen glycation to improvements in glycemiccontrol may beas shortas4mo. This conclusion is drawn fromanalysisofindividual percent decreases in

HbAl

andcollagen FL,normalizedtothepercent

decreaserequiredtoachieve themeansof the nondiabetic pop-ulation(5.5%HbAI,and4.6 mmolFL/mol Lysin skin

colla-gen)(27). Forexample,the response incollagen glycationwas

calculatedas 100 x

_(FLi

- FLf)/FLi -FLm), whereFLi,FLf referto the initialandfinal measurements of mmol FL/mol Lysincollagen,and FLm =4.6 mmolFL/molLysincollagen,

the mean value for the nondiabetic population (27). Bythis analysis there was an average 45% decrease in

HbA,

and25% decrease in glycation ofcollagen (r = 0.68, P < 0.01) toward the nondiabeticmeans.The greater response in

_HbA,

isconsistent

with theincreasedrateofturnoverofhemoglobin,compared to skin collagen. The 25% decrease in glycation ofcollagen

within a 4-mo period indicates that the upper limit for the half-time for reversal of collagen glycation is - 8 mo. The

actual half-time is undoubtedlyshorter since none of the pa-tients achieved normoglycemia instantaneously or maintained normoglycemiathroughout the course of the study. For the six patients in poorest control at the beginning of the study (initial

HbAI

> 12.0%, marked by an asterisk in Fig. 3 B), the average decreases toward the nondiabetic means during the course of thestudy were 66% for MHBG, 64% for

HbA,,

and 36% for

glycation ofcollagen (see Fig. 3,A-C). These results suggest that if glycemia were completely andinstantaneously normal-ized atthebeginningofthestudy (i.e.,100% decrease to normal inMHBG), it should have been possible to achieve a 50% de-creasetowards normal in glycation ofcollagen within the 4-mo

periodof ourstudy,suggestinganactualhalf-time closer to 4 mofor reversal of theexcessive glycationofcollagen. A more rigorous mathematicalanalysiscannot be justified because of the lack of information about the kinetics of collagenglycation

and the mechanisms by which collagen FL content was de-creased.

There are a number of possible explanations for the de-creaseinglycation of collagenduring improved glycemic con-trol. All assume a decrease in the rate ofglycationin response to the fall in ambientglucoseconcentration, plussome reac-tion in which theexisting glucoseadducts are consumed. As-sumingnegligible turnover of insoluble skincollagen(35-38), onepossibleexplanationfor the decrease inglycationis that the reaction ofglucose withlysine may be reversible, the hexose being released as glucose and mannose stereoisomers,

regener-To

0 ci vc 0s

9)

_#

5 10 15 20

ramolPentosidine/molLysine

25

Figure 4. Correlations between pentosidine and totalfluorescence

orCMLin skin colla-gen. These data are from analyses of the ini-3 tialbiopsysamples. All E sixof thepossible paired

B correlations between pentosidine,

fluores-E

E cence,CML,andCMhL hadr2 0.9 and P<

ating the unmodified lysine residue. As reviewed in reference

39, there isfirmevidence forreversal ofthe Amadori rearrange-mentwith modelAmadoricompounds (24), but little informa-tion on the rate or extent ofreversal of glycation of proteinsin

vivo.Itis alsopossiblethat theAmadoriadduct is consumed in aforwardreaction, resulting, forexample,inthe releaseofthe

carbohydrate inamodifiedform,such as 1- or 3-deoxygluco-sone(40),again regenerating the unmodified lysine residue in collagen. The precision ofouranalyses do notpermit us to

determine whether the loss ofFLin collagen is accompaniedby anincrease in thelysinecontentof the protein.Thus,it isalso

possible that the FL may beconsumedinfragmentation and oxidationreactions, leadingtoformation ofbrowningand oxi-dation products, including pentosidine, CML (CMhL), and

otherspecies.However, the furtherprogressofthesereactions should be limited by the decrease in theprecursor, FL.

In contrast toFL,levelsof CML,CMhL,pentosidine,and

totalfluorescence in collagendidnotrespond toimproved

gly-cemic controlwithin theperiodofthis study. This is consistent with the fact that, to ourknowledge, these compounds are

stable and irreversible chemicalmodifications of protein;they areknowntoaccumulatewithagein long-lived proteins,such aslenscrystallins(26) and the insoluble fraction of skin colla-gen(21, 27). Theirconstantconcentration in collagen during thelimitedcourseofthis study is also consistent with the

meta-bolic inertnessof insoluble human skin collagen (35-38), while the strong correlations between the concentrations of these compounds in collagen (Fig. 4) emphasize that they are all

sensingthe same chemical environment and stresses. While

ourresults show thatglycationof collagenmay be reducedby improved glycemic control, they also indicate that browning andoxidation products formedduringadvancedstagesof the Maillard reaction result inrelativelypermanent,perhaps irre-versible, modificationoftheprotein. These considerations add further weight to the argument that the establishment and maintenanceofgood glycemiccontrolmayinhibit the

develop-mentofdiabeticcomplications. Acknowledgments

The authors wishtothank Drs. F. A. O'Connor and W. E. Parkes

(Altnagelvin Hospital),and Drs. D. R.Hadden,J.A.Weaver, L. Ken-nedy, andA.B.Atkinson (Royal VictoriaHospital)forpermissionto studypatientsunder theircare.Dietetic consultationswereprovided by

Mrs. E.McColgan(Altnagelvin Hospital)and Mrs. J.Holmes (Royal VictoriaHospital). Weare gratefulfor the assistance of Mrs. G. M.

Flemingand Mr. T. R.McCormickofthe Departmentof Biochemis-try,Altnagelvin Hospital.

This workwassupported in part by Research Grant DK-19971 from the National Institute of Diabetes andDigestiveandKidney

Dis-eases.TheAmes Division, MilesLaboratories, Inc., generously pro-vided the memory glucometers (GlucometerM) and personal

com-puter used inthesestudies.

References

1.West,K.M.1982.Hyperglycaemiaasacauseoflongtermcomplications.

InComplicationsofDiabetes,2nd ed.H.Keen and R. J.Jarrett,editors. Edward

Arnold,London. 13-18.

2.Kennedy, L.,and J. W.Baynes. 1984.Non-enzymaticglycosylationand thechroniccomplicationsof diabetes:anoverview.Diabetologia.26:93-98.

3.Brownlee, M.,H.Vlassara,and A.Cerami.1984.

Non-enzymaticglycosyla-tionand thepathogenesisof diabeticcomplications.Ann.Intern.Med. 101:527-537.

4. Editorial. 1986. Browning and diabetic complications. Lancet. i: 1192-1193.

5. Baynes, J. W., and V. M. Monnier, editors. 1989. The MaillardReaction in Aging, Diabetes and Nutrition. Alan R. Liss, Inc., New York. 410 pp.

6. Finot, P. A., H. U. Aeschbacher, R. F. Hurrell, and R. Liardon, editors. 1990.The MaillardReaction inFood Processing, Human Nutrition and Physiol-ogy.BirkhauserVerlag, Basel. 516 pp.

7.Kohn, R. R. 1978. PrinciplesofMammalianAging. Prentice-Hall Inc., Englewood Cliffs, NJ. pp. 37-44.

8.Schnider, S. L., and R. R. Kohn. 1981. Effects of age and diabetes mellitus on thesolubility and nonenzymatic glucosylation ofhuman skin collagen. J. Clin. Invest. 67:1630-1635.

9. Lyons, T. J., and L. Kennedy. 1985. Nonenzymatic glycosylation of skin collagen in patients with limited joint mobility. Diabetologia. 28:2-5.

10.Vishwanath, V., K. E. Frank, C. A. Elmets, P. J. Dauchot, and V. M. Monnier. 1986. Glycation ofskin collagen in TypeIdiabetes mellitus: correlation with long-term complications. Diabetes. 35:916-921.

11.Monnier, V. M., R. R. Kohn, and A. Cerami. 1984. Accelerated age-re-latedbrowning of human collagen in diabetes mellitus. Proc. Natl.Acad.Sci. USA. 81:583-587.

12.Monnier, V. M., D. R. Sell, F. W. Abdul-Karim, and S. N. Emancipator. 1988. Collagen browning and cross-linking are increased in chronic experimental hyperglycemia: relevance to diabetes and aging. Diabetes. 37:868-872.

13.Dominiczak, M.H., J. Bell, N. H. Cox, D. C. McCruden, S. K. Jones,

A.Y.Finlay, I. W. Percy-Robb, and B. M. Frier. 1990. Increased collagen-linked fluorescence in skin of youngpatientswith TypeIdiabetes mellitus.Diabetes

Care. 13:468-472.

14.Hamlin, C. R., R. R. Kohn, and J. H.Luschin. 1975. Apparent acceler-atedaging of human collagen in diabetesmellitus. Diabetes.24:902-904.

15. Yue, D. K., S. McLennan, L.Delbridge,D. J.Handelsman, T. Reeve, and J. R.Turtle. 1983. The thermalstability ofcollagen in diabetic rats: correlation with severity ofdiabetes and non-enzymatic glycosylation. Diabetologia. 24:282-285.

16.Monnier, V. M., V.Vishwanath, K. E. Frank, C.A.Elmets, P. J. Dauchot, and R. R. Kohn. 1986.Relation betweencomplicationsof TypeIdiabetes

melli-tusandcollagen-linked fluorescence.N.Engl. J.Med.314:403-408.

17.Wolff,S. P. 1987. Thepotentialroleof oxidativestress in diabetes and its complications: novelimplicationsfor theory andpractice.InDiabetic

Complica-tions. M. J. C.Crabbe, editor.Churchill-LivingstoneInc., New York. 167-220. 18. Lyons, T. J., S. R.Thorpe, and J.W.Baynes. 1990.Glycationand

oxida-tion ofproteinsinaginganddiabetes.InGlucoseMetabolism,Diabetes and the Vascular Wall. N. Ruderman, M. Brownlee, and J. R. Williamson, editors. Ox-fordUniversityPress, New York.Inpress.

19. Jones, A. F., and J. Lunec.1987. Proteinfluorescenceanditsrelationship

tofree radicalactivity.Br.J.Cancer.55(Suppl.VIII):60-65.

20.Sell, D. R., and V. M.Monnier.1989.Structureelucidationofa

sene-scencecross-link from human extracellular matrix:implication ofpentoses in the

agingprocess. J.Biol. Chem.264:21597-21602.

21.Sell, D. R., andV.M.Monnier. 1990. End-stage renal disease and diabetes catalyze the formation ofapentose-derivedcrosslink fromaginghumancollagen.

J.Clin.Invest.85:380-384.

22. Baynes, J. W., J. A. Dunn, D. G. Dyer,K.J.Knecht,M. U.Ahmed, and S. R.Thorpe. 1990. Role ofglycationin development ofpathophysiologyin dia-betes andaging.InGlycatedProteinsinDiabetes Mellitus.R.G.Ryall, editor.

AdelaideUniversity Press, Adelaide, Australia. 219-236.

23. Dunn, J. W., D. G. Dyer, K. J. Knecht, S. R. Thorpe, D. R. McCance, K.

Bailie,G.Silvestri,T. J.Lyons, and J. W. Baynes. 1990. Accumulationof

Mail-lardreaction products in tissueproteins. In The MaillardReactionin Food

Pro-cessing,HumanNutrition andPhysiology.P.A.Finot,H.U.Aeschbacher,R. F. Hurrell, and R.Liardon, editors.Birkhlauser Verlag,Basel. 425-430.

23a.Dyer,D. G., J.A.Blackledge, S. R.Thorpe,and J.W.Baynes. 1991.

Formation ofpentosidine during non-enzymatic browningofproteinsby

glu-cose:identification of glucoseand othercarbohydratesaspossibleprecursors of

pentosidineinvivo.J. Biol.Chem.Inpress.

24.Ahmed, M. U., S. R. Thorpe, and J.W.Baynes. 1986. Identification of

N-4carboxymethyl)lysineas adegradationproduct offructoselysinein glycated

protein.J.Biol. Chem.261:4889-4894.

25.Ahmed,M.U.,J. A.Dunn,M. D.Walla,S.R.Thorpe, and J. W. Baynes. 1988.Oxidativedegradationofglucoseadductstoprotein: formationof

3-(N'-lysino)-lacticacidfrom modelcompoundsandglycated proteins.J.Biol. Chem. 263:8816-8821.

26.Dunn, J. A., J.S.Patrick,S. R.Thorpe, andJ.W.Baynes.1989. Oxidation ofglycated proteins: age-dependentaccumulation ofN'-(carboxymethyl)lysinein human lensproteins.Biochemistry.28:9464-9468.

27. Dunn, J. A., D. R.McCance, S. R. Thorpe, T. J. Lyons, and J. W. Baynes. 1991.Age-dependentaccumulation ofN-(carboxymethyl)lysineand

N-(carbox-ymethyl)hydroxylysinein human skincollagen. Biochemistry. 30:1205-1210.

28. Patrick,J.S., S. R.Thorpe,and J. W.Baynes. 1990.Non-enzymatic

glycosylation of proteindoesnotincrease withage in normal human lenses. J.

Gerontol.,Biol.Sci.45:B18-23.

29.Menard, L.,M.E.Dempsey, L.A.Blankstein,H.Aleyassine,M. Wacks, and J. S.Soeldner. 1980.Quantitativedetermination ofglycosylated hemoglobin Alby agargelelectrophoresis. Clin. Chem.26:1598-1602.

30.Erbersdobler, H. F. 1986. Twenty years of furosine: better knowledge aboutthebiologicalsignificanceofMaillard reaction products in food and nutri-tion. Dev. Food Sci. 13:481-491.

31.Monnier, V. M. 1990. Nonenzymatic glycosylation, the Maillard reaction and theaging process. J.Gerontol., Biol.Sci.45:B105-111.

32.Schleicher,E., and 0. H. Wieland. 1986. Kineticanalysisofglycationas a

tool forassessingthe half-life ofproteins.Biochim.Biophys.Acta. 884:199-205. 33.Garlick,R.L., H.F.Bunn, andR.G.Spiro.1988.Nonenzymatic

glyca-tion ofbasement membranesfromhumanglomeruliandbovinesources:effect of diabetes and age.Diabetes.37:1144-1150.

34. McLennan,S., D. K. Yue, M. Marsh, B.Swanson, L. Delbridge, T. Reeve, and J. R. Turtle.1986. Thepreventionandreversibility oftissuenon-enzymatic

glycosylation in diabetes.Diabetic Med.3:141-146.

35. Thompson, R. C., and J. E. Ballou. 1956. Studies of metabolic turnover

with tritium as a tracer. V. The predominantly non-dynamic state ofbody constit-uents in the rat. J. Biol. Chem. 223:795-809.

36.Prockop, D. J., and K. I. Kivirikko. 1968. Hydroxyprolineandthe metabo-lismofcollagen. In Treatise on Collagen, Vol. 2, Biology ofCollagen. B. S. Gould, editor. Academic Press Ltd., London. 215-246.

37. Molnar, J. A., N. Alpert, J. F. Burke, and V. R. Young. 1986. Synthesis anddegradation ratesof collagens in vivo in whole skin of rate, studied with 1802 labeling. Biochem. J. 240:431-435.

38. Molnar, J. A., N. M. Alpert, D. A. Wagner, S. Miyatani, J. F. Burke, and V. R.Young. 1988. Synthesis and degradation of collagens in skin of healthy and protein-malnourished rats in vivo, studied by"02labeling.Biochem. J. 250:71-76.

39.Baynes, J. W., N. G. Watkins, C. I. Fisher, C. J. Hull, J. S. Patrick, M. U. Ahmed,J. A. Dunn, and S. R. Thorpe. 1989. The Amadori product on protein: structureandreactions. In The Maillard Reaction in Aging, Diabetes and Nutri-tion. J. W. Baynes and V. M. Monnier, editors. Alan R. Liss, Inc., New York. 43-67.