Ann.N.Y.Acad.Sci.ISSN0077-8923 ANNALS OF THE NEW YORK ACADEMY OF SCIENCES Issue:TheYearinDiabetesandObesity (cid:2) Islet cell mass in diabetes and how it relates to function, birth, and death GordonC.WeirandSusanBonner-Weir SectiononIsletCellBiologyandRegenerativeMedicine,ResearchDivision,JoslinDiabetesCenter,DepartmentofMedicine, HarvardMedicalSchool,Boston,Massachusetts Addressforcorrespondence:GordonC.Weir,M.D.,SectiononIsletCellBiologyandRegenerativeMedicine,JoslinDiabetes Center,OneJoslinPlace,Boston,[email protected] Intype1diabetes(T1D)(cid:2) cellmassismarkedlyreducedbyautoimmunity.Type2diabetes(T2D)resultsfrom inadequate(cid:2)cellmassandfunctionthatcannolongercompensateforinsulinresistance.Thereductionof(cid:2)cell mass in T2D may result from increased cell death and/or inadequate birth through replication and neogenesis. Reductioninmassallowsglucoselevelstorise,whichplaces(cid:2)cellsinanunfamiliarhyperglycemicenvironment, leadingtomarkedchangesintheirphenotypeandadramaticlossofglucose-stimulatedinsulinsecretion(GSIS), whichworsensasglucoselevelsclimb.Toxiceffectsofglucoseon(cid:2)cells(glucotoxicity)appeartobetheculprit. Thisdysfunctionalinsulinsecretioncanbereversedwhenglucoselevelsareloweredbytreatment,afindingwith therapeuticsignificance.Restorationof(cid:2)cellmassinbothtypesofdiabetescouldbeaccomplishedbyeither(cid:2)cell regenerationortransplantation.Learningmoreabouttherelationshipsbetween(cid:2)cellmass,turnover,andfunction andfindingwaystorestore(cid:2)cellmassareamongthemosturgentprioritiesfordiabetesresearch. Keywords: betacell;islets;diabetes;neogenesis;insulinsecretion Introduction tionindiabetes,andthecomplicatedissuesof(cid:2)cell Whiletype1diabetes(T1D)clearlyresultsfroma birth,death,andreplenishment. loss of (cid:2) cells, there were decades of uncertainty (cid:2) cellmass about the contribution of (cid:2) cell failure to type 2 diabetes(T2D).Inthefirsthalfofthe20thcentury In a normal adult human pancreas, (cid:2) cell mass is itwasgenerallyassumedthat(cid:2)cellfailurewasim- approximately2%ofpancreaticweight.Aboutone portantforalldiabetes.Inthe1960s,however,the millionisletsarescatteredthroughoutthepancreas, development of the insulin radioimmunoassay led whichtranslatestoroughlyonebillion(cid:2)cells.Pan- to the finding that plasma insulin levels were of- creasestypicallyweighbetween60gand100g,so(cid:2) ten high in T2D, leading many to assume that the cellmassissomewherenear1–2g. pathogenesis could be explained by insulin resis- In1955MacleanandOgilvie1reported,inacare- tance alone. However, over the past 15 years there fulstudyonautopsypancreasesusinghistochemical hasbeengeneralagreementthat(cid:2)cellinadequacyis staining, that (cid:2) cell mass in older people with di- afundamentalpartofT2D.Thegrowingincidence abetes (presumably mostly T2D) was about 50% ofT2Disdrivenbyinsulinresistance,whichislikely of controls, although this finding was largely ig- theresultofatypicalWesternlifestylecharacterized nored. It was also found that (cid:3) cell mass was not byobesityandminimalexercise.Akeypointisthat altered.Similarestimatesof(cid:2)cellmasswerefound most people with insulin resistance will never de- in a few small studies, and then in 2003 a study velopT2D:diabetesonlyresultsiftheir(cid:2) cellsfail led by Butler and Butler reported similar find- toprovidesufficientinsulin.Thisreviewwillfocus ings using immunostaining in autopsy pancreases upontherelationshipbetween(cid:2)cellmassandfunc- from better-characterized subjects.2 There was doi:10.1111/nyas.12031 92 Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. Weir&Bonner-Weir Islet(cid:2)cellmassindiabetes yearsanumberofthemhavediedanddonatedtheir pancreases for study. At this writing, (cid:2) cells iden- tifiedbyinsulinimmunostaininghavebeenfound inall28subjects(S.Bonner-Weir,unpublished).It also appears that those few with more residual C- peptidehavemore(cid:2)cells.Thepresenceofthese(cid:2) cellsraisesimportantquestionsastowhether(cid:2)cells havespecialcharacteristicsthatresistdestructionor whethernew(cid:2)cellsarecontinuallyformedbyrepli- cationofexisting(cid:2)cellsand/orbyneogenesisand thendestroyedbypersistingautoimmunity. Figure 1. (cid:2)cellrelativevolumeinautopsypancreasesfrom Thequesttomeasure(cid:2) cellmasswith individualswhowerenondiabetic(ND)hadimpairedfasting glucose(IFTG)orhadtype2diabetesmellitus(TTDM).Figure imaging takenfromRef.2,withpermissionfromtheAmericanDiabetes Accuratemeasurementof(cid:2)cellmassinlivingsub- Association. jects would greatly enhance our understanding of the pathogenesis of both T1D and T2D, as well as considerable scatter in the data, but in both lean themonogenicformsofdiabetes,makingitama- and obese individuals the relative (cid:2) cell volume, a jorresearchpriority.Forbothtypesitwouldbeof surrogate for mass, was about 30–70% of nondi- greatvaluetofollowthedynamicsofthedeclinein abetic controls. Importantly, in subjects with im- (cid:2)cellmassbecauseitmightbepossibletolinkthese pairedfastingglucoselevels,the(cid:2)cellmasswasalso topathogeneticmechanisms.InT1Dthiscouldbe reduced,butlessso,toabout60%ofnormal(Fig.1). enormouslyhelpfulinevaluatingtheefficacyofvar- ThoroughstudiesfromKoreaandBelgiumreported iousimmuneinterventionsthatarebeingtestedin similarreductionsin(cid:2)cellmassinT2D.3,4Another subjectswithprediabetesandnew-onsetdiabetes.In finding important for understanding T2D is that bothtypesofdiabeteswewouldverymuchliketo obesenondiabeticsubjectsinthreestudieshadonly understandtheeffectsofmedicationandglycemic a20–40%increasein(cid:2)cellmasscomparedtolean controlon(cid:2)cellmass. controls.2,4,5Thisonlymodestincreasewassurpris- Unfortunately,ithasprovedverydifficulttode- ing,astheincreaseof(cid:2)cellmassininsulinresistant velop any methods that come close to being accu- miceisproportionallymuchhigher.6 rate. There was great interest in the type 2 vesic- InT1D,(cid:2) cellsaredecimatedbyautoimmunity ular monoamine transporter (VMAT2), which is and the process usually occurs over several years. expressed by (cid:2) cells, can be labeled with [11C]- Therearefewpathologicalstudiesand(cid:2)cellimag- dihydrotetrabenazine,andthenmeasuredwithPET inghasnotadvancedenoughtobehelpful,soouras- scanning.Unfortunately,inhumansubjectsthisap- sumptionsarelargelyextrapolatedfromC-peptide pearstobeinsufficientlyspecific.10NowtheGLP-1 secretion and insulin requirements. Because some receptor,whichmaybespecificenoughfor(cid:2)cells, individuals with new-onset T1D have remissions isreceivingmuchattention.11 Anotherapproachis that can last for several months, it is thought that totaginflammationinthepancreasofT1D,which (cid:2) cell mass at diagnosis might be as high as 50% correlates with the activity of autoimmunity; the ofnormalinsomeindividuals.7Weknowfromthe firstpublishedresultsinhumanslookpromising.12 DiabetesControlandComplicationsTrial(DCCT) Mostoftheapproachestoimaging(cid:2)cellsoverthe thatC-peptidelevelscanbesubstantialinsomesub- pastseveralyearshavebeenrecentlyreviewed.13One jectssoonafterdiagnosisbutthenfalltonegligible caveat is that to be useful a method will proba- levels in just a few years.8 A new remarkable find- blyneedtoaccuratelymeasuredifferencesin(cid:2)cell ing is that (cid:2) cells may never be completely wiped mass in the range of only 5%, because such small outinT1D.TheJoslinMedalistStudyincludessub- changesarelikelytomakeadifferenceinglycemic jectswhohavehadT1Dforover50years.9Inrecent control. Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. 93 Islet(cid:2)cellmassindiabetes Weir&Bonner-Weir Normal(cid:2) cellfunctionandmaintenanceof itcouldhaveanegativeinfluenceonhepaticinsulin mass action. (cid:2)cellshavearemarkableabilitytomakeandstore βcellhypertrophyandatrophy largequantitiesofinsulinandthensecretethisprod- Glucoseappearstobedominantamongthefactors uctwithexquisitetimingandprecision.Glucoseis influencingthemaintenanceof(cid:2)cellmass.Changes thedominantfactorcontrolling(cid:2)cellfunctionand in glucose levels seem to drive the major determi- survival,andauniqueglucoserecognitionmecha- nantsof(cid:2)cellmass,hyperplasia,hypertrophy,and nismallowscontrolbyextracellularconcentrations atrophy.Chronichypoglycemia,whichoccurswith of glucose that usually range between 4 mM and insulin-secretingtumors,leadstomarked(cid:2)cellat- 8 mM (70–140 mg/dL). This is accomplished by rophy,27 whereas chronic hyperglycemia can lead wide-openGLUT2glucosetransportersinrodents, to (cid:2) cell hyperplasia and hypertrophy.28,29 Using althoughprobablymorebyGLUT1inhumans.14,15 partialpancreatectomytocreateamodelofhyper- These transporters allow the extracellular and in- glycemiainrats,wefoundthatindividual(cid:2)cellsize tracellular glucose concentrations to be essentially increasedby85%.29Increasesinglucoselevelsstim- thesame.Thekeyglucosesignalingforinsulinse- ulatereplicationwithinaday,30 andovertimelead cretion is done by glucose metabolism, the rate of tohyperplasia.Glucosestimulationof(cid:2) cellrepli- whichiscontrolledbyglucosephosphorylationby cation has also been found in an in vivo model of glucokinase,withaK ofabout7mM.16,17Glucose glucoseinfusioninmice31 andaninvivomodelof m metabolismatthemitochondriallevelturnsonse- humanislettransplantation.32 cretionviathereasonablywell-understoodK+-ATP- Compensatoryβcellresponsetoinsulin dependent pathway and the less well-understood resistancewhenbloodglucoselevelsare K+-ATPindependentpathway.18The(cid:2)cellalsore- normal spondstosignalsfromthebrainviabothbranches There has been considerable debate about how (cid:2) of the autonomic nervous system19 and the in- cellsecretionandmasscanbeaugmentedininsulin cretin hormones from the intestine, gastrointesti- resistantstateswhenincreasesinglucoselevelscan- nalinsulinotropicpeptide(GIP),andglucagon-like notbedetermined.Wefavortheviewthatbecause peptide1(GLP-1).20 glucose is such a dominant determinant of (cid:2) cell The(cid:2)cellprovidesveryefficientcontrolofglu- functionandgrowth,thesechangesaremainlycon- coselevelswitheatingassecretedinsulinstartsac- trolledbyextremelyefficientglucosefeedbackon(cid:2) tivating the liver even before blood glucose levels cells.6,33,34 Theremaybesubtlechangesinglucose rise, due to vagal stimulation during the cephalic levels that make a difference and there is evidence phaseofinsulinsecretion17andsecretionofincretin ofincreasedactivityofglucokinase,35whichmeans hormones. As insulin secretion rises, glucagon se- thata(cid:2)cellcanbemoreresponsiveatlowerglucose cretion is suppressed, allowing this increased in- concentrations. There is much interest in the pos- sulin/glucagon ratio to enhance hepatic glucose sibility that some important signals are produced uptake at just the right time.21 Protection against by the liver because of the impressive (cid:2) cell com- hypoglycemiaisparticularlydependentuponrapid pensation found with knockout of hepatic insulin shutdownofinsulinsecretionbyfallingglucoselev- receptorsinmice.34Thesearchcontinues. els.22Thusglucoseandotherfactorscanturnonin- sulinsecretionwithinminutesand,whenremoved, Dysfunctionalinsulinsecretionas shutitoffasquickly. diabetesdevelops Oscillationsofinsulinsecretion When glucose levels chronically rise to levels only Insulin secretion exhibits oscillation with a peri- modestlyhigherthannormal,dramaticdysregula- odicity of about five minutes,23,24 which appears tion of insulin secretion appears. This was shown tobecoordinatedbyintrapanceaticneuralconnec- most impressively with a simple experiment pub- tions.25Becausetheseoscillationsaresoprominent lished over 35 years ago (Fig. 2).36 Adult humans intheportalvein,theylikelyhaveimportantinflu- withvariouslevelsoffastingglycemiareceivedrapid enceoninsulin’seffectsupontheliver.Disruption infusions of glucose intravenously to elicit acute oftheseoscillationshasbeenfoundinT2D,26 and glucose-simulated insulin secretion (GSIS). When 94 Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. Weir&Bonner-Weir Islet(cid:2)cellmassindiabetes T2D,partialimprovementinGSISwasfoundafter a 20-hour infusion of insulin40 and changes after gastric surgery were found weeks to months later. This is an important question because a thorough understandingofthetimingandmagnitudeofthe effectsofglucotoxitycouldhavetherapeuticvalue. Thecasefortheimportanceof glucotoxicityandlackofimportance oflipotoxicityandglucolipotoxicity Whileitisclearthatthediabeticmilieuisrespon- sible for (cid:2) cell dysfunction, there has been much discussionaboutthecontributionsofglucotoxicity, Figure2. IncrementsofacuteGSISinsubjectswithincreasing lipotoxicity,andglucolipotoxicity.41–43Thecasefor fastingplasmaglucoselevels.FiguretakenfromRef.36,with thedominanceofglucotoxicitycanbemadebecause permissionfromtheEndocrineSociety. oftheremarkablytightcorrelationbetweenglucose levelsandlossofacuteGSIS.Moreover,changesin the fasting glucose was normal at 4.5–5.6 mM insulinsecretionafterseveraldaysofglucoseinfu- (80–100mg/dL)alargespikeofinsulinsecretionap- sionsinratsresemblethoseseenindiabetics,44 yet pearedwithinjustafewminutes.However,themag- suchinfusionssuppressfreefattyacid(FFA)levels. nitudeofGSISwasmuchlowerwhenglucoselevels Also,similarlossofGSIScanbefoundwhenisolated rose above 5.6 mM and by the time they reached isletsareexposedtohigh-glucoseconcentrationsin 6.4 mM (115 mg/dL), a level in the range of im- culture.45,46 pairedfastingglucose(IFG),acuteGSIS,apredia- Thecaseforlipotoxicityisbasedalmostentirely beticstateequatedwithfirst-phaseinsulinsecretion, uponinvitroexposureofprimary(cid:2)cellsand(cid:2)cell wascompletelyobliterated.Nonetheless,the(cid:2)cells linestoFFAsinculture,especiallythesaturatedFFA functionedwellenoughtomaintaintheprediabetic palmitate.NotonlydoespalmitatedisruptGSISand statebecausetheycanrespondtomoreprolonged othertypesofinsulinsecretion,butalsoithasbe- glucosestimulationwithsecondphaserelease37and comeafavoritetoolinthestudyof(cid:2) celldeath.It to acute stimulation by incretin signals such as isnotatallclearthattheFFAconcentrationsused GLP-1,aswellasaminoacids.Thesefindingshave fortheseinvitrostudiesaresimilartothoseseenby nowbeenreproducedinmultiplehumanandani- (cid:2)cellsinvivo.Wehavelittleunderstandingofhow malstudies. FFAs are handled in the interstitial space and the Dysfunctionof (cid:2) cellsbecomesmoreseriousas environment of lipid membranes that surround (cid:2) thediabeticstateworsensandfunctionalmassde- cells.Ithasalsobeenpostulatedthatpalmitategiven teriorates.Agiven(cid:2)cellmassputsoutlessinsulin invitroto(cid:2)cellsmaybeconvertedtothetoxicprod- in response to stimuli. In another old study, sub- ucttripalmitin,whichisnotexpectedtobeformed jects with and without T2D received maximal (cid:2) in vivo.47 An apparent shortcoming in the use of cell stimulation from prolonged infusions of glu- FFAs or high glucose concentrations48 to study (cid:2) coseaugmentedwitharginine.38Itcanbeassumed celldeathisthatthefrequencyofdeathinsomany thatthe(cid:2)cellmassoftheseT2Dsubjectswasinthe invitrostudiesisfarhigherthanthefrequencyseen rangeof50%ofnormal,yettheirinsulinresponse invivo,whichraisesconcernsabouttherelevanceof tothismaximalstimuluswasonly15%ofnormal someofthemechanismsthathavebeenidentified. (Fig.3). There are a few in vivo studies using infusions of Importantly from a therapeutic perspective, the lipid,49buttheresultantFFAlevelsaresohighthat severedysfunctioninducedbythediabeticstatecan theirclinicalrelevancemustbequestioned. bereversedifglucoselevelsarebroughttonormal, Dataobtainedfrominvivostudiesonlipotoxic- as best shown by the full restoration of secretion ity in humans and animal models are also uncon- after bariatric surgery.39 It is surprising how little vincing. For example, FFA levels are high in obe- we know about the timing of this restoration. In sity, which is associated with exuberant GSIS. The Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. 95 Islet(cid:2)cellmassindiabetes Weir&Bonner-Weir humandiabetesoranimalmodelsispresentlyweak tononexistent Reductionofinsulincontentandabnormal proinsulinsecretion (cid:2) cells contain remarkable amounts of insulin, about10%ofthetotalproteincontentwithanaver- age(cid:2)cell,about20pgofinsulin.Insulinsynthesisis balancedbyinsulinsecretionandbytheautophagic destruction of secretory granules. Because insulin mRNA has a long half-life and stable concentra- tion,53thefine-tuningofreplenishmentaftermeals seemsmostlydrivenbyglucose-stimulatedtransla- tion.When(cid:2)cellsareexposedtothediabeticstate, theybecomedegranulatedsuchthatinsomeexper- imentalsystemsthecontentper(cid:2)cellmasscanbe reduced to lower than 5% of normal.54 Data from a variety of sources indicate that insulin depletion ismuchmoreseverewithhighratherthanmodest Figure 3. Subjects with noninsulin-dependent diabetes (NIDDM,T2D)andcontrolsubjectswhoseglucoselevelswere elevationsinbloodglucose.Whilesecretedinsulin increasedwithglucoseinfusionsfollowedbyacutestimulation accounts for some of this depletion, the reduction ofinsulinsecretionwithintravenousarginine.Figuretakenfrom ofinsulinsynthesismustbeacriticalcomponent.55 Ref.38,withpermissionfromtheEndocrineSociety. Of secreted insulin immunoreactivity from (cid:2) cells, about 2–4% is proinsulin, while circulating increase of glucose levels in prediabetes correlates proinsulininbloodcanbe10–40%ofthetotaldue verypreciselywithGSISdysfunction,36 butasimi- toitslongerhalf-life.Theproinsulintoinsulinratio larcorrelationwithFFAhasyettobedemonstrated. inbloodisincreasedinT2Dandtoalesserextent Ithasalsobeenpostulatedthatincreasedlipidstores IGT and has been used as a marker of (cid:2) cell dys- intheformofdropletscausetoxicity,butacorrela- function. This proportional change is thought to tionbetweenlipidstoresandsecretorydysfunction result from depletion of mature granules resulting in(cid:2)cellshasyettobeshown.50Moreover,(cid:2)cellse- inincreasedsecretionoftheincompletelyprocessed cretorydysfunctioncanbefoundindb/dbmicein- contentsofimmaturegranules.56 dependentofplasmalipidlevels.51TheobeseZucker Fivestagesofdiabetes;clinical diabeticfatty(ZDF)ratmodelofdiabeteshasbeen consequencesof(cid:2) celldysfunction proposedasanexampleoflipotoxicitybecausethe inducedbyprediabeticanddiabeticstates late stage islets contain large quantities of lipid as diabetesworsens.52Theproblemisthatithasnever To understand both T1D and T2D, it is helpful been shown that most of this lipid is located in (cid:2) to conceptualize five stages in their progression cellsratherthanthefibroblastsandadipocytesthat (Table1).57 Stage 1 is successfulcompensation for canbeseenintheseanatomicallydistortedislets. a modest fall in (cid:2) cell mass in T1D or for insulin Proponentsoflipotoxicityfoundthatinvitroex- resistanceinT2D.The(cid:2) cellssecretemoreinsulin posureof(cid:2)cellstoFFAsandhighglucosewaseven tomaintainanormoglycemicenvironment,which moredamaging,whichledtothewidelyusedterm in turn allows acute GSIS because (cid:2) cells are not glucolipotoxicity. The problem remains that FFAs exposedtoglucotoxicity.Withstage2,(cid:2) cellmass may be toxic only in an artificial in vitro environ- drops a little more and/or insulin resistance wors- ment. In conclusion, while lipotoxicity is a major ens,suchthatglucoselevelsrisetolevelsassociated issue for the pathogenesis of insulin resistance in withamarkedlossofacuteGSIS,whichwepostu- various insulin target tissues, evidence that lipo- lateiscausedbyglucotoxicity.Stage2isequivalent toxicityandthelipidpartofglucolipotoxicitycon- toimpairedglucosetolerance(IGT)orIFG,which tribute to (cid:2) cell secretory dysfunction or death in is stable because of maintenance of meaningful 96 Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. Weir&Bonner-Weir Islet(cid:2)cellmassindiabetes Table1.Fivestagesofdiabetes57 Stage1:Normalbloodglucoselevels:compensationforinsulinresistancewithincreased(cid:2)cellinsulinsecretionand mass.WithearlyautoimmunedestructionofT1D,increasedinsulinsecretioncompensatesfordecreased(cid:2)cell mass. Stage2:Stableadaptation:IGT,IFG,pre-T1D,islettransplantation.Inadequate(cid:2)cellmassallowsmodest hyperglycemia,whichleadstodysfunctionalGSISfromglucotoxicity. Stage3:Unstabledecompensation:T2DandearlyT1D;worseningglucotoxicityfor(cid:2)cellsandinsulintargettissues causesarapidriseinbloodglucoselevels. Stage4:Stabledecompensation:higherbloodglucoselevelsinT2DandearlyT1D,butenoughinsulinsecretionto controlketosis. Stage5:Decompensation:T1D;classicsymptomsofdiabetesandketosis;severereductionin(cid:2)cellmass. insulinsecretion,suchthatonlyabout5%ofindi- (cid:2) cellbirthanddeath vidualswithIGTprogresstoT2Dperyear.58 Stage To understand changes in (cid:2) cell mass one needs 3isanunstablephaseinwhichincreasingglucotox- to understand (cid:2) cell turnover, which is better un- icityproducesmore(cid:2)celldysfunctionandinsulin derstoodinrodentsthaninhumans.Birthcanre- resistance, leading to a rapid rise in glucose levels sultfromreplicationofpreexisting(cid:2) cellsorfrom tostage4,withmoreserioushyperglycemiainthe neogenesis, the formation of new islets thought rangeof9–17mM(160–300mg/dL).Thisrelatively to originate from cells in the pancreatic duct ep- rapid deterioration in control may not be caused ithelium.Evidenceisaccumulatingthatsomecells by a fall in (cid:2) cell mass but instead by increased withductidentificationmarkerscanserveasmulti- insulinresistancefromstressandovereating,forex- potent precursor cells.63,64 There has been debate ample.Evidencesupportingtheexistenceofstages and even controversy about the relative roles of 2–4inhumanswasrecentlyreportedbytheWhite- replicationversusneogenesis.Replicationhasbeen hall study group.59 Stage 5 has more severe (cid:2) cell shown to be particularly important in adult(cid:2) cell depletionasfoundinT1D. expansion65 and for regeneration after (cid:2) cell de- Theseconceptsaretherapeuticallysignificantfor pletion from diphtheria toxin in mice.66 In fetal T2D because aggressive treatment of subjects in life new islets are formed from cord- or duct-like stage 4 may be able to reduce glucose levels to the structures, and it is logical to think that postna- more stable stage 2 of IGT, or to an even better tal neogenesis is largely a recapitulation of this stage1.Therearealreadysuggestionsthataggressive process. In spite of discordant lineage tracing re- early treatment of newly diagnosed patients with sults in mice,63 it is now better appreciated that T2D with stage 4 glucose levels may lead to last- substantial neogenesis occurs during the neona- ingimprovementincontrol.60Thesestagesarealso tal period67,68 and in response to some forms of important for T1D because the (cid:2) cell appears to injury.63,69,70 gothroughthesamedevelopmentofdysfunctional In humans the majority of (cid:2) cell expansion oc- insulin secretion as autoimmunity reduces (cid:2) cell curs in early childhood with probably a greater mass.61 Forexample,subjectswithnew-onsetT1D contribution from replication than from neogen- maypresentwithhighglucoselevelsandthenhave esis.71 In adults, some studies find no evidence aremissionthatiseitherspontaneousorinducedby of any (cid:2) cell replication;72,73 however these find- immunosuppression.7DuringremissionC-peptide ings are not straightforward. For example, when levelsmightbeconsiderablyhigherthanwhenpa- adulthumanisletsaretransplantedintoimmuno- tientsinitiallypresentedwithhyperglycemia.62De- incompetent mice, they always have Ki67+ (cid:2) cells spite hopes that this could be explained by (cid:2) cell intherangeof0.2–0.7%.74 Perhapsthereissome- regeneration, relief from glucotoxicity leading to thing about the mouse in vivo environment that improved (cid:2) cell function seems like a better ex- turnsonquiescentcells.Itmaybesignificantthatal- planation. mostallstudieshaveusedpancreasesobtainedfrom Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. 97 Islet(cid:2)cellmassindiabetes Weir&Bonner-Weir Figure4. Sectionofpancreas(20×magnification)stainedforinsulin.Numerouspancreaticductsareshown,withinsulin-positive cellspresentinandnexttoductwall,suggestingnewisletformationfromexocrineducts(neogenesis).FiguretakenfromRef.2, withpermissionfromtheAmericanDiabetesAssociation. autopsiesorcadaverdonors,whichraisesquestions manscontainlipofuscin,whichtakesalongtimeto astowhetherthewarmandcoldischemiaofthese accumulate.78 conditions dampens the Ki67 expression. This is Thepossiblesignificanceofalowrateofturnover supportedbythefindingthatinpancreaticsamples shouldnotbeunderestimated.If0.5%of(cid:2)cellsare obtainedatsurgeryKi67positivitywasfoundtobe Ki67+ andKi67positivitylastsfor12hours,(cid:2) cell intherangeof0.5%.75 masscouldmorethandoubleinlessthanayear.The Thereappearstobealowrateof(cid:2)celldeath,as changesinmasswillofcoursealsobeinfluencedby judged by such techniques as TUNEL.2 Because (cid:2) thevariablesofdeath(fromapoptosisandnecrosis) cellmassisreasonablywellmaintainedinnondia- andneogenesis,whichareevenhardertoquantify. betics,itmakessensethattherewouldbesomelow Whatcausesthereductionof(cid:2) cellmass levelofturnover,withbirthbalancingdeath.Deter- inT2D? mining the contribution of neogenesis in humans is very difficult, but the presence of small clusters Manyconcludethatthereduced(cid:2)cellmassinT2D ofinsulin-containingcellswithin,protrudingfrom, reflectsanincreasedrateof(cid:2) celldeath,butwhile and near ducts (Fig. 4), and the discovery of cells this must often be true, there may also be a (cid:2) cell doublestainedforinsulinandtheductmarkercy- birthproblem.Forexample,T2Dmayresultfrom tokeratin, are consistent with some level of neo- an intrauterine growth disturbance or genetic and geneic activity.63 The severe hypoglycemia some- environmentalforcesthatslow(cid:2)cellreplicationor timesseenafterbariatricsurgeryisassociatedwith neogenesis. pancreaticpathologyconsistentwithneogenesis.76 Thereisalotofcircumstantialevidencethatthe Thereisspeculationthatthisisdriven bythevery prediabetic/diabeticstate,ormorespecificallyglu- highplasmalevelsofGLP-1seeninthesepatients.77 cotoxicity,leadstoanincreasedrateof(cid:2)celldeath Regardless of how (cid:2) cells are born, they are cer- from either apoptosis or necrosis. However, over- tainly long-lived; this is supported by the finding work in the absence of increases in glucose levels that a very high percentage of (cid:2) cells in adult hu- seems to be associated with reduced (cid:2) cell mass. 98 Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. Weir&Bonner-Weir Islet(cid:2)cellmassindiabetes TheobservationbytheButlergroupthatindividu- somehowcontributestocelldeathappearstohave alswithIFG(stage2)have,onaverage,a40%reduc- been answered. Examination of pancreases from tionin(cid:2)cellmass(Fig.1)2suggeststhatsomething subjects with T2D indicated that islets with large deleteriousoccursduringthecompensatorystage1. amyloid deposits often have fewer (cid:2) cells.85 When We know that glucose can stimulate replication,30 human IAPP, which can form amyloid unlike ro- andwecanarguethatitstillhasaneffectininsulin dent IAPP, is overexpressed in (cid:2) cells of mice or resistantstateswhenglucoselevelsappeartobenor- rats, the result is worsening of diabetes associated mal,asinstage1.Wehypothesizethatduringthis withmoreamyloiddeposition.86–88 Itnowappears stageofcompensationthereisincreasedreplication that the toxicity is exerted not by the extracellular of(cid:2)cells,andsuspectthatthereissomelimitation depositsbutbysmallfibrilsortoxicoligomersthat in the capacity for regeneration through glucose- candamagecellmembranes.89,90Thereisincreasing drivenreplicationleadingtoslowingofbirth.Thus, evidencethatthesecancausedamagewhileinside duringthistimeofoverworkalimitationinthebirth the(cid:2) cell;theymaywhenoutsideaswell.Itisstill ofnewcellscouldleadtoanolderpopulationof(cid:2) puzzlingthatamyloiddepositsareseeninisletsof cellsthathasahigherrateofcelldeath,whichcould T2Dandininsulinomasbutarefarlesscommonin contributetotheimpressivereductionin(cid:2)cellmass obesity, in spite of the increased secretion of both found in the earliest stages of diabetes of IGT and insulinandIAPP. IFG,andcouldhelpexplaintheincreasedincidence ofapoptosislaterinthedisease. Endoplasmicreticulumstress While investigators may advocate for the domi- The potential contribution of endoplasmic retic- nance of their favorite mechanism, it may be best ulum (ER) stress to (cid:2) cell apoptosis in T2D has totakethepositionthatseveralmechanismsmight received much attention in the past few years.91–93 contribute. ER stress is part of the unfolded protein response (UPR). This UPR is turned on when the cell has Oxidativestress problemswiththecomplexprocessoffoldingnewly Therehasbeengreatinterestinthepossibilitythat synthesizedproteins.Toprotectthecells,itrecruits glucotoxicity leads to oxidative stress, which then chaperoneproteinsthathelpwithfolding,acceler- causessecretorydysfunctionandanincreasedrate ates the degradation of unfolded proteins, and in- of(cid:2)celldeath.79Certainlyglucosestimulationleads hibitsthesynthesisofnewproteins.WhiletheUPR to an increase in reactive oxygen species (ROS), hasimportantbenefitsforfunctionandsurvival,a but this is not necessarily bad. There is interest- fully turned on stress response activates apoptotic ingresearchunderwaytosortoutthecontributions pathwaysthroughsuchmediatorsasC/EBPhomol- of ROS to important (cid:2) cell signaling mechanisms ogousprotein(CHOP)andc-JunN-terminalkinase that are part of normal (cid:2) cell function and to (cid:2) (JNK).Astudyofgeneexpressioninhuman(cid:2)cells cell dysfunction and death.80 This has led to the transplanted into mice showed that these cells, in hopethatantioxidanttreatmentmightslowthead- a transplant site exposed to mild hyperglycemia, verseprocesses.81Indeed,theadministrationofN- had an activated UPR but reduced expression of acetyl-cysteinehasbeenfoundto slow theadverse JNKandCHOP,94suggestingthatERstressismore effectsofROSontheprogressionofdiabetesinZDF helpfulthanharmfulinthatsituation.Inourstudy rats.82 ofpancreasesfromcadaverdonorswithT2D,gene Amyloid expressionwasperformedonlaser-captured(cid:2)cell- Thefirstdescriptionofamyloiddeposits,calledhya- enrichedtissuewithlittlechangeinERstressgenes, lineatthetime,intheisletsofpeoplewithdiabetes although there were a few changes suggesting that was reported in 1900.83 It took until 1987 to de- partsoftheUPRwereactivated.95 termine that this amyloid was formed from the (cid:2) Butler’s group stained for CHOP in pancreases cell secretory product islet associated polypeptide from subjects with T2D, and from lean and obese (IAPP),84whichinhumansandafewotherspecies subjectswithoutdiabetes.96 The(cid:2)cellsofboththe canformthe(cid:2)-pleatedsheetsthatmakeupamyloid. T2DandobeseindividualshadcytoplasmicCHOP The question as to whether amyloid deposits are staininginabout15–20%oftheir(cid:2)cells.However, just innocent bystanders or whether its formation nuclearstainingforCHOPwasseeninlessthan1% Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. 99 Islet(cid:2)cellmassindiabetes Weir&Bonner-Weir ofthe(cid:2)cellfromT2D,andnuclearstaininginthe Needtobettercharacterize(cid:2) cellsto obesesubjectswasevenlower.Thisisconsistentwith understandpathogenesis theviewthatonlysmallnumbersof(cid:2)cellsinT2D Most of our understanding comes from studying are dying at any one time, but that a larger num- largenumbersofisletcells,butmuchmorecouldbe berareshowingsignsofstressorfragility.Thus,the learnedaboutpathogenesisifwecouldlearnmore fewcellswithnuclearCHOPcouldbeontheirway abouttheheterogeneityof(cid:2) celldeath.Forexam- todeathwhilethosewithcytoplasmicCHOPmay ple,whyisitthatonlyoneinmanyhundredcellsis bemorevulnerablebutstillfunctional.Likewise,it chosentoenterthecellcycleandreplicate?Likewise, maybethatinT2Dmany(cid:2)cellshavesomeUPRac- whatisspecialabouttheonecellofmanythatdies? tivation,whichprotectsthem,butinamorefragile, We know from many studies there is considerable perhaps older, population of (cid:2) cells,the proapop- heterogeneity for the capacity to secrete insulin. A toticERstresspathwaysaremoreactive. human(cid:2)cellmayhavealifetimeofaseveralyears, so there must be differences among cells that are βcelldedifferentiation new, in the early stages of maturation, in midlife, Geneexpressionin(cid:2)cellsexposedtohyperglycemia in the stage of postmitotic senescence, when they canbestudiedaftersurgicalreductionof(cid:2)cellmass are old and fragile, and finally when they are dy- in rats with partial pancreatectomy.29,50,97–99 After ing. Also, what are the differences between a new variousperiodsoftime,the(cid:2)cellsinisolatedislets (cid:2) cell formed by neogenesis in a fetus versus one andintissuesectionsfromtheremnantpancreascan that formed during adulthood? And, what are the be studied. Impressive changes in gene expression differencesbetween(cid:2)cellsformedfromneogenesis werefoundin(cid:2)cellsexposedtoeitherverymodest comparedwiththoseproducedfromthereplication orseverehyperglycemia.Thesechangesincludedre- ofexistingcells?Howimportantistopicallocation, ducedexpressionofseveralkey(cid:2)celltranscription suchastheisletcenterornexttothemantleofnon- factors,changesrelatedtoglucosemetabolism,and (cid:2) cells? Could there be different fetal pathways of upregulation of genes related to stress. It has been developmenttheleadtothegenerationofdifferent temptingtocallthischangein(cid:2)cellphenotypeded- kindsof(cid:2)cells?Whatmightthedifferencesbebe- ifferentiation,butitisnotpossibletosaythatthese tween(cid:2)cellsinthedorsalversustheventrallobesof cells reverted to a phenotype that occurred earlier thepancreas?Whathappensto(cid:2)cellsinisletsthat indevelopment.Whileitmakessensethatsomeof arelessactiveandhavelessbloodflow?102Onemight thesechangescouldhaveadeleteriouseffectonse- guess that there is temporal cycling of (cid:2) cells that cretion,littlecanbesaidabouthowtheymightbe allowsthemtorestuntilneeded.Surelythediffer- relatedtoafallin(cid:2)cellmass. encesdescribedaboveareaccompaniedbychanges A new look at phenotypic change was taken by in gene and protein expression, so there must be the Accili group, which obtained data from mice ways to find markers for many of the different (cid:2) lacking FoxO1 and suggested the novel hypothesis celltypes.Flowcytometryandhistochemicalstain- that in T2D (cid:2) cells are dedifferentiated to a state ingmaybethemostpromisingwaystoisolatethese of immaturity whereby they express neurogenin3, different cell types for study. Even if only some of Oct4, Nanog, and L-Myc and no longer make in- thesepossibilitiesarerealized,itisexcitingtothink sulin.100 Someofthesecellsevenadoptedan(cid:3)cell of how much might be learned about how (cid:2) cells fate. The argument was made that restoration of function,areborn,anddie. differentiation in these cells may be a therapeutic pathtoincrease(cid:2)-cellmassinT2D.Thisworkwill Restorationof(cid:2) cellmassindiabetes no doubt stimulate new research, but at this early stagequestionsmustberaisedastowhetheraFoxO1 We know that restoration of (cid:2) cell mass can deficiencymodelisrelevanttoT2D.Certainlyinves- normalize blood glucose levels in both T1D and tigatorswilltakeacloserlookathumanpancreatic T2D. The proof-of-principle has been established T2Disletsforimmaturecells.Thepossibleconver- with islet transplantation in T1D103 and pancreas sionof(cid:2)cellsto(cid:3)cellsisinterestingbut(cid:3)cellmass transplantation for T2D.104 Unfortunately current inhumanT2Dhasnotbeenfoundtobeincreased approaches require cadaver pancreases and im- intwothoroughstudies.1,101 munosuppression,whichgreatlylimitsthenumber 100 Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. Weir&Bonner-Weir Islet(cid:2)cellmassindiabetes of diabetics that can be treated. To make (cid:2) cell conformalcoatingusingpolyethyleneglycol(PEG) replacement therapy more accessible, we need an orsimilarmaterials.121 abundant supply of well-characterized (cid:2) cells and Finally,thereisthedreamofstimulatingregener- must find safe ways to prevent autoimmune and ationof(cid:2) cellsinthepancreas.122 Drugsmightbe allogeneic killing of the newly provided (cid:2) cells. developedthatcouldstimulate(cid:2)cellreplicationor Possibilitiesforreplacementincludecellulartrans- neogenesis.Anotherpossibilityisthattheexocrine plantationandregenerationofpancreaticendocrine pancreas, or possibly even cells in the liver, could cells. bereprogrammedtobecome(cid:2) cells.123 Inastrik- Forasourceoftransplantablecells,isletcellsde- ing example of reprogramming, a single injection rivedfromhumanembryonicstemcellsorinduced ofadenovirusescarryinggenesforthreetranscrip- pluripotentstem(IPS)cellsareparticularlyattrac- tionfactorsimportantfor(cid:2)celldevelopment(Pdx1, tive because there have been impressive scientific neurogenin3,andMafA)intoamousepancreascan advances in recent years.105–107 Another hope is to producelargenumbersofcellswith(cid:2) cellcharac- findwaystoexpandhumanpancreaticcellsinvitro teristicsthatcanstoreandreleaseenoughinsulinto sotheycanbetransplanted.Thereareafewencour- reversethediabeticstate.124 aging studies suggesting that (cid:2) cells or pancreatic Summary ductcellscanbedirectedtoamesenchymalpheno- type,allowingexpansionandthenredifferentiation Reduction of (cid:2) cell mass is fundamental to the toisletcells.108,109 Thepossibility ofstimulating (cid:2) pathogenesis of both T1D and T2D and leads to cellreplicationhasalsobeenreceivingconsiderable severe dysfunction of insulin secretion. These two attention,withtheidentificationofnewcompounds abnormalitiesleadtotheconceptofthereduced(cid:2) that can stimulate division110,111 and the elucida- cellfunctionalmass.Weneedtoknowmoreabout tionofmechanismsthatmightbeexploitedinthe how reduced mass results from unbalanced (cid:2) cell future.112–116 Anotherapproachistocreatea(cid:2)cell turnover,andunderstandhowthisleadstosecretory linewithgeneticengineeringthatcanberedifferen- dysfunction.Diabetescanbepreventedorreversed tiatedbeforetransplantation,butsafetyconcernsare by inhibiting the fall of (cid:2) cell mass or by restor- likelytoremain.117 Finally,thepossibilityofxeno- ing mass through regeneration or transplantation. transplantation with porcine islets from adult or Theseissuesareamongthehighestprioritiesforthe perinatalpigscontinuestobeexplored.118,119 field. Findingwaystolimitautoimmunityandallore- Acknowledgments jection has been a daunting endeavor. With stem cells,itshouldbepossibletouselinesthatwillallow Our work has been supported by the National In- favorable human leukocyte antigen (HLA) match- stitutes of Health [R01 DK 66056 and DK 93909 ing. Transplantation of islet cells derived from iPS (S.B.-W.),P30DK36836 (JoslinDiabetes Research cells generated from people with T2D should face Center)], the Juvenile Diabetes Research Founda- noimmuneattack,butthosederivedfromsubjects tion, the American Diabetes Association, and the with T1D should face a strong autoimmune reac- Diabetes Research and Wellness Foundation. We tion. There are many approaches to control allo- thank the many collaborators, trainees, and lab- andautoreactivitythatarebeyondthescopeofthis oratory workers who made our work in this area review, but the possibility of protecting cells with possible. immunobarriermembranesisreceivingrenewedat- Conflictsofinterest tention.120 It seems likely that the first transplants ofisletcellsderivedfromstemcellswillemploythis Theauthorsdeclarenoconflictsofinterest. protectionmethod.Thegeneralapproachistouse References eithermacroencapsulation,inwhichmanyisletsare containedwithinadeviceoftencalledabioartificial 1. Maclean,N.&R.F.Ogilvie.1955.Quantitativeestimation pancreas,or microencapsulation,in whicha small ofthepancreaticislettissueindiabeticsubjects.Diabetes numberofisletsorcellaggregatesarecontainedin 4:367–376. smallhydrogelcapsulesthatcouldbe400–1500(cid:4)m 2. Butler,A.E.etal.2003.Beta-celldeficitandincreasedbeta- cellapoptosisinhumanswithtype2diabetes.Diabetes52: in diameter. Islet cells can also be protected with 102–110. Ann.N.Y.Acad.Sci.1281(2013)92–105(cid:2)c 2013NewYorkAcademyofSciences. 101