Anna Zampetaki1 Peter Willeit123 Simon Burr1 Xiaoke Yin1 Sarah R Langley1

Stefan Kiechl3 Ronald Klein4 Peter Rossing5 Nishi Chaturvedi6 and Manuel Mayr1

Angiogenic microRNAs Linked toIncidence and Progression of DiabeticRetinopathy in Type 1 DiabetesDiabetes 201665216ndash227 | DOI 102337db15-0389

Circulating microRNAs (miRNAs) have emerged as novelbiomarkers of diabetes The current study focuses on therole of circulating miRNAs in patients with type 1 diabetesand their association with diabetic retinopathy A total of29 miRNAs were quantified in serum samples (n = 300)using a nested case-control study design in two prospec-tive cohorts of the DIabetic REtinopathy CandesartanTrial (DIRECT) PROTECT-1 and PREVENT-1 ThePREVENT-1 trial included patients without retinopathyat baseline the PROTECT-1 trial included patients withnonproliferative retinopathy at baseline Two miRNAspreviously implicated in angiogenesis miR-27b andmiR-320a were associated with incidence and withprogression of retinopathy the odds ratio per SDhigher miR-27b was 057 (95 CI 040 082 P = 0002)in PREVENT-1 078 (057 107 P = 0124) in PROTECT-1and 067 (050 092 P = 0012) combined The respectiveodds ratios for higher miR-320a were 157 (107 231P = 0020) 143 (105 194 P = 0021) and 148 (117188 P = 0001) Proteomics analyses in endothelial cellsreturned the antiangiogenic protein thrombospondin-1as a common target of both miRNAs Our study iden-tifies two angiogenic miRNAs miR-320a and miR-27bas potential biomarkers for diabetic retinopathy

Recent studies have begun to unveil a powerful and un-expected role of microRNAs (miRNAs) in numerous forms ofdiseases providing a unique opportunity to translatethis knowledge into the clinical setting in the form of

miRNA-based therapeutics and diagnostics (12) miRNAsare small noncoding RNAs with cell-type specific expressionpatterns that orchestrate biological networks by modulatinggene expression miRNAs form base pairs with their targetmessenger RNAs and mediate gene silencing often regulat-ing multiple proteins within the same biological pathwaysGiven their importance in angiogenesis along with the tech-nical feasibility in manipulating their function in vivo vas-cular miRNAs have become central targets for therapeuticmanipulation including diabetes (3)

Additionally miRNAs circulate in blood We have pre-viously performed the first systematic analysis of circulatingmiRNAs in a population-based study (the Bruneck Study) andidentified distinct miRNA signatures associated with type2 diabetes (4) and risk of myocardial infarction (5) We havealso highlighted their platelet origin (6) and applied conceptsof network topology to explore their biomarker potential (7)Exciting opportunities exist to pursue miRNAs as novel bio-markers for risk estimation and patient stratification (8)

The current study addresses the role of circulatingmiRNAs in patients with type 1 diabetes (T1D) and theirassociation with microvascular complications in partic-ular diabetic retinopathy Our aims were threefold firstto evaluate whether miRNA profiles are independentlyassociated with retinopathy development and progres-sion in diabetes second to quantify the incrementaldiscriminatory power of miRNAs over and above tradi-tional risk markers and third to conduct experiments

1Kingrsquos British Heart Foundation Centre of Research Excellence Kingrsquos CollegeLondon London UK2Department of Public Health and Primary Care University of CambridgeCambridge UK3Department of Neurology Medical University of Innsbruck Innsbruck Austria4Department of Ophthalmology and Visual Sciences University of WisconsinndashMadison Madison WI5Steno Diabetes Centre University of Copenhagen Copenhagen Denmark6Institute of Cardiovascular Science University College London London UK

Corresponding author Nishi Chaturvedi nchaturvediuclacuk or ManuelMayr manuelmayrkclacuk

Received 21 March 2015 and accepted 16 September 2015

This article contains Supplementary Data online at httpdiabetesdiabetesjournalsorglookupsuppldoi102337db15-0389-DC1

AZ and PW share first authorship

NC and MM share senior authorship

copy 2016 by the American Diabetes Association Readers may use this article aslong as the work is properly cited the use is educational and not for profit andthe work is not altered

See accompanying article p 22

216 Diabetes Volume 65 January 2016

COMPLIC

ATIO

NS

using endothelial cells (ECs) and proteomics profiling toprovide mechanistic insight

RESEARCH DESIGN AND METHODS

Patient SampleThe DIabetic REtinopathy Candesartan Trials (DIRECT)trials have previously been described in detail (9) In brieftwo randomized double-blind parallel-design and placebo-controlled trials (PREVENT-1 and PROTECT-1) were con-ducted in 309 centers worldwide Both trials assignedpatients with normotensive normoalbuminuric T1D to eithercandesartan 16 mg once a day or placebo Exclusion criteriawere eye conditions that precluded capture of gradable ret-inal photographs (ie open-angle glaucoma dense cataractobscuring retinal view) stenotic valvular disease previoushistory of heart attack or stroke pregnancy lactation orrenal impairment (serum creatinine $110 mmolL forwomen $130 mmolL for men) The PREVENT-1 trial in-cluded patients without retinopathy at baseline (defined asEarly Treatment Diabetic Retinopathy Study [ETDRS] scalelevel 1010) the PROTECT-1 trial included patients withnonproliferative retinopathy at baseline (defined as ETDRSscale between 2010 and 4747) For the current studywe used a ldquonestedrdquo case-control approach Incident casesubjects were defined as patients who progressed threesteps or more on the ETDRS scale over the course of thestudy Case subjects were stratum matched with controlsubjects on sex HbA1c categories and categories of diabetesduration (plus an ETDRS summary score in PROTECT-1)These were randomly selected from participants whoseETDRS scale did not progress and who remained free ofintermittent microalbuminuria to the end of the study pe-riod Numbers selected for analysis were 70 case and 70control subjects in PREVENT-1 and 93 case and 93 controlsubjects in PROTECT-1 Analysis of miRNAs could not beperformed on all serum samples Thus the number of caseand control subjects in the present analysis was 62 and 64in PREVENT-1 and 81 and 93 in PROTECT-1 (n = 300)

miRNA MeasurementsTotal RNA from 200 mL of serum samples was preparedusing the miRNeasy kit (Qiagen) as previously described(4) A fixed volume of 3 mL of the 25 mL RNA eluate wasused as input for RT reactions using Megaplex Primer Pools(Human Pools A version 21 Life Technologies) RT reactionproducts were further amplified using the Megaplex PreAmpPrimers (Primers A version 21) TaqMan miRNA assayswere used to assess the expression of individual miRNAsDiluted preamplification product (05 mL) was combinedwith 025 mL TaqMan miRNA Assay (203) (Life Technol-ogies) and 25 mL TaqMan Universal PCR Master Mix NoAmpErase UNG (23) to a final volume of 5 mL Quanti-tative PCR was performed on a Life Technologies ViiA 7thermocycler at 95degC for 10 min followed by 40 cyclesof 95degC for 15 s and 60degC for 1 min All samples wererun in duplicates Relative quantification was performedusing the DataAssist version 301 (Life Technologies)

For normalization purposes the average of all miRNAtargets displaying Ct 32 cycles or an exogenous miRNA(cel-miR-39) spiked in during RNA extraction was ap-plied as previously described (5)

TransfectionHuman umbilical vein endothelial cells (HUVECs) werecultured as described previously (10) and seeded at 60ndash70confluency on the day before transfection Cells werewashed in serum-free DMEM and replaced with M199 con-taining ECs supplement without antibiotics or serum Hu-man retinal microvascular ECs (HRECs) (obtained fromCell Systems Corporation [CSC] Kirkland WA) wereseeded on CSC attachment factor (cat no 4Z0-210) andcultured in CSC complete media (cat no 4Z0-500) Priorto transfection HRECs were washed with CSC Serum FreeMedium without antibiotics (cat no SF-4Z0-500) miRNAmimics or nontargeting mimic control were transfected ata final concentration of 50 nmolL using LipofectamineRNAiMAX (Invitrogen) according to the manufacturerrsquosrecommendations (11)

Proteomic Profiling of the Endothelial SecretomeHUVECs were carefully washed in serum-free medium andthen incubated twice in fresh serum-free M199 for 30 minCells were stimulated in phorbol 12-myristate 13-acetate(PMA) (50 nmolL) containing M199 medium for 45 minand analyzed by mass spectrometry as previously described(10) The conditioned medium was collected centrifuged at1000 rpm for 5 min and concentrated using Amicon Ultra-15 Centrifugal Filter Unit with Ultracel-3 membrane (3 KDa)(Millipore) to 135 mL Then 05 molL Tris 1 SDS pH88 (15 mL) and 100 mmolL dithiothreitol reducing agent(78 mL) were added under agitation for 1 h at 55degC followedby incubation with 500 mmolL iodoacetamide (83 mL)for 1 h before precipitation with cold acetone (220degC) over-night at 220degC Samples were centrifuged (14000 rpmfor 10 min) the supernatant was discarded and the pro-tein pellet was left to air-dry Pellets were resuspended in01 molL pH 85 triethylammonium bicarbonate (100 mL)and incubated at 37degC under agitation for 30 min Pro-teins were digested in 04 mgmL trypsin (1 mL) overnightat 37degC The reaction was stopped by addition of 10formic acid (50 mL) Peptide samples were purified usingC18 spin columns (Thermo Scientific) The eluate was keptat 280degC before freeze drying (Christ Alpha 1-2 LD FreezeDryer) under vacuum at 255degC The samples were resus-pended in 2 acetonitrile and 01 formic acid (20 mL)and analyzed by reverse-phase nanoflow high-performanceliquid chromatography (3 mmolL 100 Aring 50 cm 3 75 mminner diameter column Acclaim PepMap C18 Thermo Sci-entific) interfaced with a Q Exactive Plus Orbitrap massspectrometer (Thermo Scientific) Each sample was sep-arated at a flow rate of 03 mLmin over 4 h as follows0ndash10 min2ndash10 of 80 acetonitrile and 01 formic acidin ddH2O (B) 10ndash200 min10ndash30B 200ndash210 min30ndash40B 210ndash220 min99B and 220ndash240 min2B Sam-ples were subject to a full mass spectrometry scan over a

diabetesdiabetesjournalsorg Zampetaki and Associates 217

range from 350 to 1600 mz whereby the 15 most abun-dant peaks were selected for tandem mass spectrometryand fragmented by higher-energy collisional dissociation

Database SearchRaw data were analyzed by Mascot and Sequest HT al-gorithms (Proteome Discoverer 14) to identify proteinsagainst Swiss-Prot protein database (version 201401) withtaxonomy set to all entries precursor mass toleranceto 10 ppm fragment mass tolerance to 20 mmu oxidationof methionine as a variable modification and carbamido-methylation of cysteine as a static modification and amaximum of two missed cleavage sites permitted Searchresults were loaded onto Scaffold software (version 43)where a protein probability 99 a peptide probability95 and a minimum number of two peptides per pro-tein were applied as filters to generate the list of identifiedproteins

Luciferase Reporter AssaysThe 39-untranslated region of the human neuropilin 1 (NRP1)and semaphorin 6A (SEMA6A) harboring a putative bind-ing site for miR-320a and miR-27b respectively wascloned into the XhoI and PmeI linkers of the dual-luciferasereporter vector psiCHECK-2 (Promega) The following primersets were used NRP F TTCAATGAGTATGGCCGACA NRPR GGATTTCGCTCAGTTTCC SEMA6A F CGCACAGAGGTGAACAGAAA and SEMA6A R GCCCAACATGGCATTTATCT

The reporter constructs (100 ng) were transfectedtogether with their respective miRNA mimics (50 nmolL)or mimic control in triplicate into HUVECs previouslyplated (post 12 h) in 6-well plates using LipofectamineRNAiMAX (Invitrogen) as described above After 48 h cellswere harvested in 200 mL Glo Lysis Buffer (Promega)and the activities of both Renilla and firefly were mea-sured Each lysate (30 mL) was analyzed using Dual-GloLuciferase reagents (Promega) Renilla luciferase activitywas normalized to constitutive firefly luciferase activityfor each well

ELISAsThrombospondin-1 (TSP-1) levels in the conditioned mediawere determined using the Human Thrombospondin-1Quantikine ELISA kit (Bio-Techne) according to themanufacturerrsquos instruction

Statistical AnalysismiRNA values were log transformed for analysis Pairwisecorrelations between miRNAs and other markers wereassessed using Pearson correlation coefficients in a pooleddata set of 145 control subjects and plotted with the Rpackage ldquocorrplotrdquo using hierarchical clustering and the singlelinkage method We analyzed the association of miRNAswith incident retinopathy in two steps First to identifythe miRNA subset with the highest prognostic ability forfuture retinopathy we used a L1-penalized logistic regres-sion technique This method helps prevent overfitting ofcollinear and high-dimensional data and implements the

ldquoleast absolute shrinkage and selection operatorrdquo (LASSO)algorithm shrinking regression coefficients toward zerorelative to the maximum likelihood estimates Theamount of shrinkage is determined by the tuning param-eter l1 which is progressively increased up to the valuethat shrinks all regression coefficients to zero We esti-mated the optimal tuning parameter l1 as the medianof 10000 fivefold likelihood cross-validations Plots ofb-coefficients (y-axis) versus l1 (x-axis) were generated usingthe R package ldquopenalizedrdquo Sensitivity analyses were con-ducted that 1) standardized miRNA concentrations to aglobal Ct average (instead of to cel-miR-39) and 2) ex-cluded participants from analysis with diabetic nephrop-athy or microalbuminuria Because the LASSO methodallows assessment of relevance and robustness of individ-ual explanatory variables but produces biased estimates ofb-coefficients we fitted as a second step ldquotraditionalrdquologistic regression models for identified miRNAs Analyseswere adjusted for age sex and diastolic blood pressureand conducted separately for the two trials A pooledestimate of association was calculated with random-effects meta-analysis Between-study heterogeneity wasassessed by the I2 statistic (12) Regarding the experimen-tal studies results are presented as the mean 6 SEM Forresults from transfected ECs paired or unpaired Studentt test one-way ANOVA with Dunnett or Tukey post hocanalysis and two-way ANOVAs were performed usingGraphPad Prism 5 software With allowance for loss dueto missing miRNA measurement our sample size allowedus to detect a difference in retinopathy incidence and inprogression in association with a standardized differenceof 04 and 05 respectively for any given miRNA signal with80 power and 5 significance Analysis was performedusing R 313 statistical software (13) and StataMP ver-sion 121 with two-sided tests and P 005

RESULTS

Associations of miRNAs With Incidence andProgression of Retinopathy in the DIRECT TrialsFrom the PREVENT-1 and PROTECT-1 trials of a total of3326 participants with T1D we selected 62 patients withincident diabetes retinopathy 93 patients with progressivediabetes retinopathy and 145 matched control subjectsSupplementary Table 1 compares clinical characteristics ofparticipants selected for the nested case-control analysisand those not selected A panel of 29 miRNAs of whichsome were previously shown to be associated with type 2diabetes (4) and cardiovascular disease (5) was measuredin serum and values were standardized to an exogenousspike-in normalization control cel-miR-39 Baseline char-acteristics of study participants are summarized in Table 1The ratio of males to females and mean diabetes durationwere similar in case and control subjects within each trialas anticipated owing to matching Mean HbA1c howeverwas a little higher in control subjects reflecting the factthat case and control subjects were stratum rather thanindividually matched by HbA1c Differences in age and

218 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

range from 350 to 1600 mz whereby the 15 most abun-dant peaks were selected for tandem mass spectrometryand fragmented by higher-energy collisional dissociation

Database SearchRaw data were analyzed by Mascot and Sequest HT al-gorithms (Proteome Discoverer 14) to identify proteinsagainst Swiss-Prot protein database (version 201401) withtaxonomy set to all entries precursor mass toleranceto 10 ppm fragment mass tolerance to 20 mmu oxidationof methionine as a variable modification and carbamido-methylation of cysteine as a static modification and amaximum of two missed cleavage sites permitted Searchresults were loaded onto Scaffold software (version 43)where a protein probability 99 a peptide probability95 and a minimum number of two peptides per pro-tein were applied as filters to generate the list of identifiedproteins

Luciferase Reporter AssaysThe 39-untranslated region of the human neuropilin 1 (NRP1)and semaphorin 6A (SEMA6A) harboring a putative bind-ing site for miR-320a and miR-27b respectively wascloned into the XhoI and PmeI linkers of the dual-luciferasereporter vector psiCHECK-2 (Promega) The following primersets were used NRP F TTCAATGAGTATGGCCGACA NRPR GGATTTCGCTCAGTTTCC SEMA6A F CGCACAGAGGTGAACAGAAA and SEMA6A R GCCCAACATGGCATTTATCT

The reporter constructs (100 ng) were transfectedtogether with their respective miRNA mimics (50 nmolL)or mimic control in triplicate into HUVECs previouslyplated (post 12 h) in 6-well plates using LipofectamineRNAiMAX (Invitrogen) as described above After 48 h cellswere harvested in 200 mL Glo Lysis Buffer (Promega)and the activities of both Renilla and firefly were mea-sured Each lysate (30 mL) was analyzed using Dual-GloLuciferase reagents (Promega) Renilla luciferase activitywas normalized to constitutive firefly luciferase activityfor each well

ELISAsThrombospondin-1 (TSP-1) levels in the conditioned mediawere determined using the Human Thrombospondin-1Quantikine ELISA kit (Bio-Techne) according to themanufacturerrsquos instruction

Statistical AnalysismiRNA values were log transformed for analysis Pairwisecorrelations between miRNAs and other markers wereassessed using Pearson correlation coefficients in a pooleddata set of 145 control subjects and plotted with the Rpackage ldquocorrplotrdquo using hierarchical clustering and the singlelinkage method We analyzed the association of miRNAswith incident retinopathy in two steps First to identifythe miRNA subset with the highest prognostic ability forfuture retinopathy we used a L1-penalized logistic regres-sion technique This method helps prevent overfitting ofcollinear and high-dimensional data and implements the

ldquoleast absolute shrinkage and selection operatorrdquo (LASSO)algorithm shrinking regression coefficients toward zerorelative to the maximum likelihood estimates Theamount of shrinkage is determined by the tuning param-eter l1 which is progressively increased up to the valuethat shrinks all regression coefficients to zero We esti-mated the optimal tuning parameter l1 as the medianof 10000 fivefold likelihood cross-validations Plots ofb-coefficients (y-axis) versus l1 (x-axis) were generated usingthe R package ldquopenalizedrdquo Sensitivity analyses were con-ducted that 1) standardized miRNA concentrations to aglobal Ct average (instead of to cel-miR-39) and 2) ex-cluded participants from analysis with diabetic nephrop-athy or microalbuminuria Because the LASSO methodallows assessment of relevance and robustness of individ-ual explanatory variables but produces biased estimates ofb-coefficients we fitted as a second step ldquotraditionalrdquologistic regression models for identified miRNAs Analyseswere adjusted for age sex and diastolic blood pressureand conducted separately for the two trials A pooledestimate of association was calculated with random-effects meta-analysis Between-study heterogeneity wasassessed by the I2 statistic (12) Regarding the experimen-tal studies results are presented as the mean 6 SEM Forresults from transfected ECs paired or unpaired Studentt test one-way ANOVA with Dunnett or Tukey post hocanalysis and two-way ANOVAs were performed usingGraphPad Prism 5 software With allowance for loss dueto missing miRNA measurement our sample size allowedus to detect a difference in retinopathy incidence and inprogression in association with a standardized differenceof 04 and 05 respectively for any given miRNA signal with80 power and 5 significance Analysis was performedusing R 313 statistical software (13) and StataMP ver-sion 121 with two-sided tests and P 005

RESULTS

Associations of miRNAs With Incidence andProgression of Retinopathy in the DIRECT TrialsFrom the PREVENT-1 and PROTECT-1 trials of a total of3326 participants with T1D we selected 62 patients withincident diabetes retinopathy 93 patients with progressivediabetes retinopathy and 145 matched control subjectsSupplementary Table 1 compares clinical characteristics ofparticipants selected for the nested case-control analysisand those not selected A panel of 29 miRNAs of whichsome were previously shown to be associated with type 2diabetes (4) and cardiovascular disease (5) was measuredin serum and values were standardized to an exogenousspike-in normalization control cel-miR-39 Baseline char-acteristics of study participants are summarized in Table 1The ratio of males to females and mean diabetes durationwere similar in case and control subjects within each trialas anticipated owing to matching Mean HbA1c howeverwas a little higher in control subjects reflecting the factthat case and control subjects were stratum rather thanindividually matched by HbA1c Differences in age and

218 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

blood pressure levels were also observed in the two studypopulations The complex correlation patterns of serummiRNAs and baseline characteristics are depicted inFig 1 miRNAs emerged to be highly correlated witheach other but largely independent from other character-istics such as age duration of diabetes blood pressureand HbA1c

We conducted penalized logistic regression analysesadjusted for age sex and diastolic blood pressure to identifyand evaluate the association of miRNAs with incidence andprogression of diabetic retinopathy (Fig 2 and Supplemen-tary Fig 1) The descriptive tables in Fig 2 show the di-rection of association and the percentage of cross-validatedmodels that included the respective miRNA The followingmiRNAs were identified in descending order of impor-tance miR-27b miR-320a miR-454 and miR-28-3p inPREVENT-1 and miR-320a miR-122 miR-221 and miR-27b in PROTECT-1 Generally these miRNAs were morecommonly selected for inclusion in cross-validated modelsfor incidence than for progression of diabetic retinopathy(90 vs10) Results for all 29 individual miRNAs areprovided in Supplementary Fig 2 but should be interpretedwith caution in light of multiple testing and type 1 error

Since combinations of miRNAs can be superior toindividual miRNAs (457) we then performed ldquotraditionalrdquologistic regression for the two miRNAs most consistentlyassociated with incidence and progression of diabetic reti-nopathy miR-27b and miR-320a Figure 3 shows estimatesof associations for miR-27b and miR-320a adjusted foreach other age sex and diastolic blood pressure Theodds ratio per SD higher miR-27b was 057 (95 CI 040082 P = 0002) in PREVENT-1 078 (057 107 P =0124) in PROTECT-1 and 067 (050 092 P = 0012)combined The respective odds ratios for higher miR-320a were 157 (107 231 P = 0020) 143 (105194 P = 0021) and 148 (117 188 P = 0001) Qual-itatively similar results were observed in analyses that 1)standardized miRNA levels to a Ct average of all miRNAs(instead of standardization to spiked-in cel-miR-39)(Supplementary Fig 3) 2) excluded cases with incidentmicroalbuminuria (Supplementary Fig 3) 3) used differ-ent levels of adjustment (unadjusted) or adjusted for agesex and systolic blood pressure or HbA1c) or 4) omittedoutliers of the miRNA distributions SupplementaryFig 4 summarizes raw values of miR-27b and miR-320afor case and control subjects Addition of information on

Table 1mdashBaseline characteristics of the study populations

PREVENT-1 PROTECT-1

Case subjects Control subjects PDagger Case subjects Control subjects PDagger

n 62 64 93 81

Age at randomization years 311 (81) 277 (81) 002 315 (91) 31 (85) 07

Male sex n () 33 (53) 37 (58) 07 58 (62) 42 (52) 02

Duration of diabetes years 73 (38) 73 (36) 10 109 (40) 106 (40) 07

Systolic blood pressure mmHg 116 (91) 119 (96) 004 117 (108) 118 (89) 07

Diastolic blood pressure mmHg 72 (77) 73 (65) 03 727 (70) 75 (64) 002

HbA1c 89 (15) 92 (18) 04 89 (12) 94 (15) 005

UAER mgmin 48 [33 65] 43 [35 60] 04 50 [30 65] 50 [35 95] 02

WHR all 081 (007) 083 (009) 008 085 (008) 082 (008) 01

WHR men 084 (007) 088 (009) 003 088 (007) 088 (007) 04

WHR women 077 (005) 077 (005) 09 078 (007) 077 (007) 05

Creatinine mmolL 897 (13) 900 (14) 09 881 (138) 867 (141) 06

GFR mLmin173 m2 88 [75 107] 90 [79 105] 08 92 [77 107] 94 [80 104] 08

GFR mLmin173 m2 83 [71 96] 83 [73 98] 08 86 [73 100] 88 [75 96] 06

HDL cholesterol mmolL 17 (03) 17 (04) 06 17 (04) 18 (04) 007

Total cholesterol mmolL 47 (11) 49 (09) 05 51 (11) 49 (08) 02

Insulin dose unitskgday 077 (021) 069 (024) 007 081 (030) 071 (020) 001

eGDR mgkgmin 92 (13) 90 (14) 04 86 (13) 91 (13) 002

Never smoker 34 (55) 47 (73) 50 (54) 54 (67)

Ex-smoker 7 (11) 2 (3) 5 (5) 8 (10)

Current smoker 21 (34) 15 (24) 002 38 (41) 19 (23) 008

Data are mean (SD) or median [interquartile range] unless otherwise indicated eGDR estimated glucose disposal rate GFR glomerularfiltration rate UAER urinary albumin excretion rate WHR waist-to-hip ratio DaggerP values were calculated using Fisher exact tests forcategorical variables two-group t tests for continuous normally distributed variables or Wilcoxon rank sum tests for continuous left-skewed variables Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) (44) 4-variable Modification of Diet in RenalDisease (MDRD) (45) Estimated glucose disposal rate as previously described (46)

diabetesdiabetesjournalsorg Zampetaki and Associates 219

miR-27b and miR-320a to a model of a panel of variablesassociated with disease risk (ie age sex duration ofdiabetes diastolic blood pressure and level of HbA1c)improved the area under the receiver operating curveby 0087 for PREVENT-1 (P = 0027) and by 0034 forPROTECT-1 (P = 0214) (Supplementary Fig 5)

Expression of miRNAs in ECsA PCR comparison was conducted to compare the endoge-nous expression of miR-27b and miR-320a to abundant

endothelial miRNAs miR-126 and miR-126 (or miR-126-3pand miR-126-5p respectively) There was no major differ-ence in the endogenous miRNA expression of miR-27bmiR-320a miR-126-3p and miR-126-5p between four differ-ent types of ECs human saphenous vein ECs human aorticECs HRECs and HUVECs (Supplementary Table 2) NotablymiR-320a was consistently more abundant in ECs than miR-27b Next mimics for miR-27b and miR-320a were cotrans-fected with luciferase reporter constructs containing putativebinding site of two confirmed target proteins for miR-27b

Figure 1mdashPairwise Pearson correlation coefficients of serum miRNA concentrations and other markers DBP diastolic blood pressureSBP systolic blood pressure T1DM type 1 diabetes mellitus UAER urinary albumin excretion rate

220 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

Figure 2mdashPenalized logistic regression analysis of the associations of 29 serum miRNA concentrations with diabetic retinopathy Asso-ciation with incident diabetic retinopathy in PREVENT-1 (A) Association with progression of diabetic retinopathy in PROTECT-1 (B) Modelsincluded age sex and diastolic blood pressure as unpenalized covariates The graph shows b-coefficients for different levels of penal-ization (l1 estimated in 100 steps) and is truncated at a l1 of 3 hence not all measured miRNAs are plotted The black dashed lineindicates the optimal tuning parameter l1 evaluated using 10000 fivefold likelihood cross-validations The descriptive table shows thedirection of association and the percentage of cross-validated models that included the respective miRNA

diabetesdiabetesjournalsorg Zampetaki and Associates 221

and miR-320a SEMA6A and NRP1 respectively SEMA6A isa negative regulator of mitogen-activated protein kinase sig-naling known to increase transcription of angiogenic effec-tor genes (14) NRP1 is a coreceptor for vascular endothelialgrowth factor (VEGF) and SEMA promoting EC prolifera-tion and microvessel density (15) Forty-eight hours aftertransfection luciferase activity was found to be decreased by

30 and 40 in SEMA6A and NRP1 respectively (Fig 4Aand B) Thereby the functionality of miRNA mimics wasconfirmed through silencing expression of known targets

Proteomics for miRNA Target IdentificationWe have demonstrated recently that direct and indirecttargets can be identified in a broader proteomics screen

Figure 3mdashStandard logistic regression analysis and meta-analysis of the association of miR-27b and miR-320a with diabetic retinopathy andTSP-1 as common target in ECs Association of miR-27b and miR-320a with diabetic retinopathy in PREVENT-1 PROTECT-1 and both studiescombined Associations of miR-27b and miR-320a with diabetic retinopathy were adjusted for each other plus age sex and diastolic bloodpressure SDs were defined in control subjects Study-specific results were combined using random-effects meta-analysis OR odds ratio

Figure 4mdashProteomics for miRNA target identification Luciferase reporter assays to confirm functional miRNA mimics Suppression ofSEMA6A (A) a known target of miR-27b and NRP1 (B) a target of miR-320a Spectral counts of TSP-1 as quantified by mass spec-trometry in the secretome of ECs transfected with miR-27b (C ) and miR-320a (D) compared with mimic controls Error bars representSEM P lt 005 in three independent experiments P lt 001

222 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

including targets that would not have been anticipatedbased on current bioinformatics prediction models (1116)To complement our clinical studies we used a proteomicsapproach to identify putative protein targets of miR-27band miR-320a in the secretome of ECs HUVECs werestimulated with PMA to induce exocytosis of Weibel-Paladebodies an endothelial cellndashspecific storage organelle allow-ing for protein analysis by mass spectrometry Among650 proteins identified (Supplementary Table 3) only

TSP-1 showed differential secretion from both miR-27bndashand miR-320andashtransfected ECs relative to control trans-fected cells (Fig 4C and D) The proteomics data wereindependently validated by ELISA (Fig 5A) Similar resultswere obtained in HRECs after miR-27b or miR-320a over-expression (Fig 5B) An alignment of the thrombospondin-1gene (THBS-1) mRNA region with the sequence of miR-320aand miR-27b is shown in Fig 6A The vertical bars andbold characters indicate sequences of the miR-320andash and

Figure 5mdashValidation of the proteomics findings by ELISA Reduced secretion of TSP-1 in the endothelial secretome upon transfection withmimics of miR-27b and miR-320a HUVECs after 45 min of PMA stimulation in serum-free medium (A) and HRECs after overnight serumdeprivation in the presence of endothelial supplements (B) P lt 005

diabetesdiabetesjournalsorg Zampetaki and Associates 223

and miR-320a SEMA6A and NRP1 respectively SEMA6A isa negative regulator of mitogen-activated protein kinase sig-naling known to increase transcription of angiogenic effec-tor genes (14) NRP1 is a coreceptor for vascular endothelialgrowth factor (VEGF) and SEMA promoting EC prolifera-tion and microvessel density (15) Forty-eight hours aftertransfection luciferase activity was found to be decreased by

30 and 40 in SEMA6A and NRP1 respectively (Fig 4Aand B) Thereby the functionality of miRNA mimics wasconfirmed through silencing expression of known targets

Proteomics for miRNA Target IdentificationWe have demonstrated recently that direct and indirecttargets can be identified in a broader proteomics screen

Figure 3mdashStandard logistic regression analysis and meta-analysis of the association of miR-27b and miR-320a with diabetic retinopathy andTSP-1 as common target in ECs Association of miR-27b and miR-320a with diabetic retinopathy in PREVENT-1 PROTECT-1 and both studiescombined Associations of miR-27b and miR-320a with diabetic retinopathy were adjusted for each other plus age sex and diastolic bloodpressure SDs were defined in control subjects Study-specific results were combined using random-effects meta-analysis OR odds ratio

Figure 4mdashProteomics for miRNA target identification Luciferase reporter assays to confirm functional miRNA mimics Suppression ofSEMA6A (A) a known target of miR-27b and NRP1 (B) a target of miR-320a Spectral counts of TSP-1 as quantified by mass spec-trometry in the secretome of ECs transfected with miR-27b (C ) and miR-320a (D) compared with mimic controls Error bars representSEM P lt 005 in three independent experiments P lt 001

222 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

including targets that would not have been anticipatedbased on current bioinformatics prediction models (1116)To complement our clinical studies we used a proteomicsapproach to identify putative protein targets of miR-27band miR-320a in the secretome of ECs HUVECs werestimulated with PMA to induce exocytosis of Weibel-Paladebodies an endothelial cellndashspecific storage organelle allow-ing for protein analysis by mass spectrometry Among650 proteins identified (Supplementary Table 3) only

TSP-1 showed differential secretion from both miR-27bndashand miR-320andashtransfected ECs relative to control trans-fected cells (Fig 4C and D) The proteomics data wereindependently validated by ELISA (Fig 5A) Similar resultswere obtained in HRECs after miR-27b or miR-320a over-expression (Fig 5B) An alignment of the thrombospondin-1gene (THBS-1) mRNA region with the sequence of miR-320aand miR-27b is shown in Fig 6A The vertical bars andbold characters indicate sequences of the miR-320andash and

Figure 5mdashValidation of the proteomics findings by ELISA Reduced secretion of TSP-1 in the endothelial secretome upon transfection withmimics of miR-27b and miR-320a HUVECs after 45 min of PMA stimulation in serum-free medium (A) and HRECs after overnight serumdeprivation in the presence of endothelial supplements (B) P lt 005

diabetesdiabetesjournalsorg Zampetaki and Associates 223

including targets that would not have been anticipatedbased on current bioinformatics prediction models (1116)To complement our clinical studies we used a proteomicsapproach to identify putative protein targets of miR-27band miR-320a in the secretome of ECs HUVECs werestimulated with PMA to induce exocytosis of Weibel-Paladebodies an endothelial cellndashspecific storage organelle allow-ing for protein analysis by mass spectrometry Among650 proteins identified (Supplementary Table 3) only

TSP-1 showed differential secretion from both miR-27bndashand miR-320andashtransfected ECs relative to control trans-fected cells (Fig 4C and D) The proteomics data wereindependently validated by ELISA (Fig 5A) Similar resultswere obtained in HRECs after miR-27b or miR-320a over-expression (Fig 5B) An alignment of the thrombospondin-1gene (THBS-1) mRNA region with the sequence of miR-320aand miR-27b is shown in Fig 6A The vertical bars andbold characters indicate sequences of the miR-320andash and

Figure 5mdashValidation of the proteomics findings by ELISA Reduced secretion of TSP-1 in the endothelial secretome upon transfection withmimics of miR-27b and miR-320a HUVECs after 45 min of PMA stimulation in serum-free medium (A) and HRECs after overnight serumdeprivation in the presence of endothelial supplements (B) P lt 005

diabetesdiabetesjournalsorg Zampetaki and Associates 223

miR-27bndashbinding site of the THBS-1 target gene Asexpected for direct miRNA targets there was comple-mentarity between the miR-27b seed-matching sequenceand the 39-untranslated region of the THBS-1 targetgene According to one algorithm (miRWalk) TSP-1was also predicted as a target for miR-320a The putativeseed-binding region however was within the coding re-gion of the THBS-1 gene To further confirm the func-tionality of these predicted interactions we fused theputative binding regions within the THBS-1 target geneto a luciferase reporter vector Coexpression of syntheticmiR-27b but not miR-320a decreased THBS-1 reporteractivity (data not shown) confirming TSP-1 as a directtarget of miR-27b (17) and an indirect target of miR-320a (Fig 6B)

DISCUSSION

Using data from two independent prospective cohorts weidentified miR-27b and miR-320a as potential biomarkersfor new-onset retinopathy or progression of retinopathyin patients with T1D Both miR-27 and miR-320ahave previously associated with metabolic syndrome and type2 diabetes (18) and have been implicated in angiogenesis

providing a mechanistic underpinning for the observedassociation with diabetic retinopathy in the DIRECTcohort (9)

Retinopathy in T1DRetinopathy remains the most frequent and most fearedcomplication of T1D (19) It is associated with elevatedrisks of other diabetes complications notably nephropa-thy and cardiovascular disease (20) Although acknowl-edged as a vascular phenomenon the only interventionproven to date to reduce onset or progression of retinop-athy is tight glycemic control This is difficult to achieveand is associated with elevated risks of hypoglycemicepisodes Significant numbers of individuals with T1Dcontinue to progress to severe possibly sight-threateningdisease even in the presence of good glycemic controlThere is therefore a need to discover novel biomarkersthat identify individuals at high risk of progressive reti-nopathy and targets for therapeutic intervention

miRNAs in T1DmiRNAs may offer distinct advantages over other bio-markers (8) Unlike messenger RNAs miRNAs are stablein blood As nucleic acids miRNAs can be measured by

Figure 6mdashAngiogenic miRNAs Alignment of THBS-1 mRNA region with miR-320a and miR-27b (A) Relative abundance of miR-320a andmiR-27b in ECs and suppression of TSP-1 secretion as a shared molecular pathway (B) Circulating miRNA changes in patients with T1Dand retinopathy however may represent markers of a milieu that is conducive to pathological angiogenesis rather than a local release fromthe retina

224 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

quantitative PCR methodology allowing the multiplexingof several miRNAs in a single experiment Thus far therole of miRNAs in diabetic retinopathy progression hasbeen assessed in small animal models (21ndash26) Moreoveran miRNA-dependent cross talk between HIF1a andVEGF was reported in the diabetic retina (27) In thisstudy we show that miR-320a and miR-27b are associ-ated with new-onset retinopathy or progression of reti-nopathy in patients with T1D These findings wereunaltered by excluding participants with incident persis-tent or intermittent microalbuminuria at any time duringthe follow-up

Mechanistic Links to AngiogenesisBesides its potential prognostic and diagnostic valuemiRNAs may participate in an unexplored mechanismcontributing to retinopathy in patients with T1D SeveralmiRNAs can target the same effector Diverse miRNAscan also act cooperatively or redundantly to regulate theeffectors of the same biological process miR-320a regu-lates glycolysis and represses angiogenic factors includingFlk1 VEGFc insulin-like growth factor 1 insulin-likegrowth factor 1 receptor and fibroblast growth factor(2829) miR-320a has also been implicated in tumorangiogenesis by silencing NRP1 (30) miR-27b is thoughtto promote angiogenesis by targeting antiangiogenic genes(31) including the transmembrane protein SEMA6A ThemiR-23-27-24 gene clusters are enriched in ECs andhighly vascularized tissues (32) miR-27b orchestrates en-dothelial tip cell formation (3334) By analyzing proteintargets of miR-27b and miR-320a in ECs we obtained amore comprehensive depiction of the interactions andregulatory feedback loops between angiogenic proteinsand our candidate miRNAs

Proteomics Approach for miRNA Target IdentificationThe ldquotargetomerdquo of most miRNAs remains an unexploredaspect of current biology Currently the available miRNAtarget prediction tools are based on an incomplete under-standing of miRNA target recognition and miRNA efficacyBioinformatic methodologies reveal numerous putative tar-gets but only one of five of the in silico predictions iscorrect and experimental confirmation is essential Pro-teomics methods can be useful for identifying miRNAtargets at the protein level in addition to the use of bio-informatics prediction algorithms We have pursued aproteomics approach to compare the effects of miR-27band miR-320a on the secretome of ECs (10) TSP-1 wasreturned as a common target of both miR-27b andmiR-320a (35) TSP-1 is an extracellular matrix proteinwhich inhibits EC proliferation migration and angiogen-esis (36) Its antiangiogenic effect is mediated through aninteraction with VEGF specifically via inhibition of VEGFreceptor-2 activation through engaging its receptor CD47(37) This has further been supported in vivo specificallyin the eye depletion of TSP-1 resulted in corneal neo-vascularization in mice (38) Of particular significanceto diabetic retinopathy a biphasic response of TSP-1

mediated through VEGF occurred in microvascular cells inthe ischemic retina (39) This response tightly regulatesVEGF and therefore indicates a potential negative feed-back mechanism of VEGF-induced angiogenesis throughTSP-1 (39) By now several studies have implicated TSP-1in pathological angiogenesis in the retina (40ndash42) TSP-1has been shown to inhibit neovascularization in diabeticmice (43) Moreover miR-27b rescued impaired angiogen-esis via TSP-1 suppression (17) Our findings confirm thatmiR-27b directly suppresses reporter activity for TSP-1In contrast miR-320a has an indirect effect on TSP-1secretion

Strengths and Limitations of the StudyOur study was prospective and hence measured miRNAlevels before occurrence of the disease outcome Using anested case-control approach we identified 155 partic-ipants with retinopathy incidenceprogression amongDIRECT trial participants (overall n = 3326) and comparedmiRNA profiles with those of matched control subjectsAssociations were independent of established risk factorsfor diabetic retinopathy including age sex HbA1c diabe-tes duration and blood pressure and were further under-pinned by our biologically plausible finding from theproteomics analysis that both miRNAs target TSP-1 Stillwhether miR-27b and miR-320a are causally involved indiabetes retinopathy or a marker of this disease remainsto be clarified The identified miRNAs are not retina spe-cific The changes in miR-27b and miR-320a may reflecta systemic predisposition for pathological angiogenesisThe cause for the differential regulation of circulatinglevels of miR-27b and miR-320a is currently unclear Formany biomarkers the cellular origin remains uncertain Forexample both ECs as well as platelets secrete TSP-1 Mea-suring its circulating levels does not reveal how much ofTSP-1 is endothelial or platelet derived The same limita-tion applies to miRNAs that are not tissue specific anddetected in the circulation Both miR-320a and miR-27bare also present in platelets (6) The opposing direction-ality of the association of miR-27b and miR-320a withdiabetic retinopathy could hint to a different cellular or-igin of these two circulating miRNAs ie miR-320a issecreted at much higher levels from ECs than miR-27bFurther studies are required to provide an in-depth un-derstanding of their cellular origin and to test the diag-nostic or therapeutic potential of these two circulatingmiRNAs in diabetic retinopathy We measured miRNAlevels only at baseline and hence could not assess or takeinto account within-person variability of the miRNAs overtime Finally model selection and parameter estimation wereperformed on the same data set without adjustment whichmay lead to an overestimation of the strength of associationobtained and underestimation of confidence limits

ConclusionsUnderstanding how circulating miRNAs could be har-nessed for assessing the risk of retinopathy in T1D is anessential area of research Our data in the DIRECT trials

diabetesdiabetesjournalsorg Zampetaki and Associates 225

show that miR-320a and miR-27b are associated with newonset of retinopathy and progression of retinopathyBesides identifying circulating miRNAs associated withretinopathy we also interrogated the targets of miR-27band miR-320a in ECs using a proteomics approach Takentogether the findings of our study identify miRNA bio-markers for retinopathy in two independent cohortsThese findings await confirmation in larger studies but ourtwo lead miRNAs may have clinical utility given theirestablished links to angiogenesis including the translationalcontrol of TSP-1 by miR-27b and its reduced endothelialsecretion by miR-320a

Acknowledgments The authors thank Raimund Pechlaner Departmentof Neurology Medical University of Innsbruck and Philipp Skroblin Cardiovas-cular Division Kingrsquos College London for help with preparing the manuscriptFunding AZ is an Intermediate Fellow of the British Heart Foundation(FS131830207) PW is an Erwin Schroumldinger Fellow in Epidemiologysponsored by the Austrian Science Fund MM is a Senior Fellow of theBritish Heart Foundation (FS130229892) This work was supported in partby JDRF (17-2011-658) Diabetes UK (120004530) the Fondation Leducq(MicroRNA-based Therapeutic Strategies in Vascular Disease [MIRVAD] [13CVD 02]) and the National Institute of Health Research Biomedical ResearchCenter based at Guyrsquos and St Thomasrsquo National Health Service FoundationTrust and Kingrsquos College London in partnership with Kingrsquos College HospitalDuality of Interest The DIRECT trial was jointly funded by AstraZeneca andTakeda Kingrsquos College London has filed patent applications on miRNA biomarkersNo other potential conflicts of interest relevant to this article were reportedAuthor Contributions AZ SB and XY performed the experimentsAZ PW NC and MM wrote the manuscript PW SRL SK and NCanalyzed data RK NC and MM conceived and designed the experimentsPR provided the serum samples MM is the guarantor of this work and assuch had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis

References1 Zampetaki A Mayr M MicroRNAs in vascular and metabolic disease CircRes 2012110508ndash5222 van Rooij E Olson EN MicroRNA therapeutics for cardiovascular diseaseopportunities and obstacles Nat Rev Drug Discov 201211860ndash8723 Kantharidis P Wang B Carew RM Lan HY Diabetes complications themicroRNA perspective Diabetes 2011601832ndash18374 Zampetaki A Kiechl S Drozdov I et al Plasma microRNA profiling revealsloss of endothelial miR-126 and other microRNAs in type 2 diabetes Circ Res2010107810ndash8175 Zampetaki A Willeit P Tilling L et al Prospective study on circulatingMicroRNAs and risk of myocardial infarction J Am Coll Cardiol 201260290ndash2996 Willeit P Zampetaki A Dudek K et al Circulating microRNAs as novelbiomarkers for platelet activation Circ Res 2013112595ndash6007 Zampetaki A Willeit P Drozdov I Kiechl S Mayr M Profiling of circulatingmicroRNAs from single biomarkers to re-wired networks Cardiovasc Res 201293555ndash5628 Mayr M Zampetaki A Willeit P Willeit J Kiechl S MicroRNAs within thecontinuum of postgenomics biomarker discovery Arterioscler Thromb Vasc Biol201333206ndash2149 Chaturvedi N Porta M Klein R et al DIRECT Programme Study GroupEffect of candesartan on prevention (DIRECT-Prevent 1) and progression (DIRECT-Protect 1) of retinopathy in type 1 diabetes randomised placebo-controlled trialsLancet 20083721394ndash1402

10 Yin X Bern M Xing Q Ho J Viner R Mayr M Glycoproteomic analysis of thesecretome of human endothelial cells Mol Cell Proteomics 201312956ndash97811 Abonnenc M Nabeebaccus AA Mayr U et al Extracellular matrix secretionby cardiac fibroblasts role of microRNA-29b and microRNA-30c Circ Res 20131131138ndash114712 Higgins JP Thompson SG Quantifying heterogeneity in a meta-analysisStat Med 2002211539ndash155813 Goeman JJ L1 penalized estimation in the Cox proportional hazards modelBiom J 20105270ndash8414 Dhanabal M Wu F Alvarez E et al Recombinant semaphorin 6A-1 ecto-domain inhibits in vivo growth factor and tumor cell line-induced angiogenesisCancer Biol Ther 20054659ndash66815 Miao HQ Lee P Lin H Soker S Klagsbrun M Neuropilin-1 expression bytumor cells promotes tumor angiogenesis and progression FASEB J 2000142532ndash253916 Zampetaki A Attia R Mayr U et al Role of miR-195 in aortic aneurysmaldisease Circ Res 2014115857ndash86617 Wang JM Tao J Chen DD et al MicroRNA miR-27b rescues bone marrow-derived angiogenic cell function and accelerates wound healing in type 2 di-abetes mellitus Arterioscler Thromb Vasc Biol 20143499ndash10918 Karolina DS Tavintharan S Armugam A et al Circulating miRNA profiles inpatients with metabolic syndrome J Clin Endocrinol Metab 201297E2271ndashE227619 Diabetes Control and Complications Trial (DCCT)Epidemiology of DiabetesInterventions and Complications (EDIC) Research Group Lachin JM White NHHainsworth DP Sun W Cleary PA Nathan DM Effect of intensive diabetes therapyon the progression of diabetic retinopathy in patients with type 1 diabetes 18 yearsof follow-up in the DCCTEDIC Diabetes 201564631ndash64220 van Hecke MV Dekker JM Stehouwer CD et al EURODIAB ProspectiveComplications Study Diabetic retinopathy is associated with mortality and car-diovascular disease incidence the EURODIAB Prospective Complications StudyDiabetes Care 2005281383ndash138921 Kovacs B Lumayag S Cowan C Xu S MicroRNAs in early diabetic reti-nopathy in streptozotocin-induced diabetic rats Invest Ophthalmol Vis Sci 2011524402ndash440922 Feng B Chen S McArthur K et al miR-146a-mediated extracellular matrixprotein production in chronic diabetes complications Diabetes 2011602975ndash298423 Silva VA Polesskaya A Sousa TA et al Expression and cellular localizationof microRNA-29b and RAX an activator of the RNA-dependent protein kinase(PKR) in the retina of streptozotocin-induced diabetic rats Mol Vis 2011172228ndash224024 McArthur K Feng B Wu Y Chen S Chakrabarti S MicroRNA-200b regu-lates vascular endothelial growth factor-mediated alterations in diabetic reti-nopathy Diabetes 2011601314ndash132325 Murray AR Chen Q Takahashi Y Zhou KK Park K Ma JX MicroRNA-200bdownregulates oxidation resistance 1 (Oxr1) expression in the retina of type 1diabetes model Invest Ophthalmol Vis Sci 2013541689ndash169726 Bai Y Bai X Wang Z Zhang X Ruan C Miao J MicroRNA-126 inhibitsischemia-induced retinal neovascularization via regulating angiogenic growthfactors Exp Mol Pathol 201191471ndash47727 Ling S Birnbaum Y Nanhwan MK Thomas B Bajaj M Ye Y MicroRNA-dependent cross-talk between VEGF and HIF1a in the diabetic retina Cell Signal2013252840ndash284728 Tang H Lee M Sharpe O et al Oxidative stress-responsive microRNA-320regulates glycolysis in diverse biological systems FASEB J 2012264710ndash472129 Wang X Huang W Liu G et al Cardiomyocytes mediate anti-angiogenesisin type 2 diabetic rats through the exosomal transfer of miR-320 into endothelialcells J Mol Cell Cardiol 201474139ndash15030 Wu YY Chen YL Jao YC Hsieh IS Chang KC Hong TM miR-320 regulatestumor angiogenesis driven by vascular endothelial cells in oral cancer by si-lencing neuropilin 1 Angiogenesis 201417247ndash26031 Kuehbacher A Urbich C Zeiher AM Dimmeler S Role of Dicer and Drosha forendothelial microRNA expression and angiogenesis Circ Res 200710159ndash68

226 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

quantitative PCR methodology allowing the multiplexingof several miRNAs in a single experiment Thus far therole of miRNAs in diabetic retinopathy progression hasbeen assessed in small animal models (21ndash26) Moreoveran miRNA-dependent cross talk between HIF1a andVEGF was reported in the diabetic retina (27) In thisstudy we show that miR-320a and miR-27b are associ-ated with new-onset retinopathy or progression of reti-nopathy in patients with T1D These findings wereunaltered by excluding participants with incident persis-tent or intermittent microalbuminuria at any time duringthe follow-up

Mechanistic Links to AngiogenesisBesides its potential prognostic and diagnostic valuemiRNAs may participate in an unexplored mechanismcontributing to retinopathy in patients with T1D SeveralmiRNAs can target the same effector Diverse miRNAscan also act cooperatively or redundantly to regulate theeffectors of the same biological process miR-320a regu-lates glycolysis and represses angiogenic factors includingFlk1 VEGFc insulin-like growth factor 1 insulin-likegrowth factor 1 receptor and fibroblast growth factor(2829) miR-320a has also been implicated in tumorangiogenesis by silencing NRP1 (30) miR-27b is thoughtto promote angiogenesis by targeting antiangiogenic genes(31) including the transmembrane protein SEMA6A ThemiR-23-27-24 gene clusters are enriched in ECs andhighly vascularized tissues (32) miR-27b orchestrates en-dothelial tip cell formation (3334) By analyzing proteintargets of miR-27b and miR-320a in ECs we obtained amore comprehensive depiction of the interactions andregulatory feedback loops between angiogenic proteinsand our candidate miRNAs

Proteomics Approach for miRNA Target IdentificationThe ldquotargetomerdquo of most miRNAs remains an unexploredaspect of current biology Currently the available miRNAtarget prediction tools are based on an incomplete under-standing of miRNA target recognition and miRNA efficacyBioinformatic methodologies reveal numerous putative tar-gets but only one of five of the in silico predictions iscorrect and experimental confirmation is essential Pro-teomics methods can be useful for identifying miRNAtargets at the protein level in addition to the use of bio-informatics prediction algorithms We have pursued aproteomics approach to compare the effects of miR-27band miR-320a on the secretome of ECs (10) TSP-1 wasreturned as a common target of both miR-27b andmiR-320a (35) TSP-1 is an extracellular matrix proteinwhich inhibits EC proliferation migration and angiogen-esis (36) Its antiangiogenic effect is mediated through aninteraction with VEGF specifically via inhibition of VEGFreceptor-2 activation through engaging its receptor CD47(37) This has further been supported in vivo specificallyin the eye depletion of TSP-1 resulted in corneal neo-vascularization in mice (38) Of particular significanceto diabetic retinopathy a biphasic response of TSP-1

mediated through VEGF occurred in microvascular cells inthe ischemic retina (39) This response tightly regulatesVEGF and therefore indicates a potential negative feed-back mechanism of VEGF-induced angiogenesis throughTSP-1 (39) By now several studies have implicated TSP-1in pathological angiogenesis in the retina (40ndash42) TSP-1has been shown to inhibit neovascularization in diabeticmice (43) Moreover miR-27b rescued impaired angiogen-esis via TSP-1 suppression (17) Our findings confirm thatmiR-27b directly suppresses reporter activity for TSP-1In contrast miR-320a has an indirect effect on TSP-1secretion

Strengths and Limitations of the StudyOur study was prospective and hence measured miRNAlevels before occurrence of the disease outcome Using anested case-control approach we identified 155 partic-ipants with retinopathy incidenceprogression amongDIRECT trial participants (overall n = 3326) and comparedmiRNA profiles with those of matched control subjectsAssociations were independent of established risk factorsfor diabetic retinopathy including age sex HbA1c diabe-tes duration and blood pressure and were further under-pinned by our biologically plausible finding from theproteomics analysis that both miRNAs target TSP-1 Stillwhether miR-27b and miR-320a are causally involved indiabetes retinopathy or a marker of this disease remainsto be clarified The identified miRNAs are not retina spe-cific The changes in miR-27b and miR-320a may reflecta systemic predisposition for pathological angiogenesisThe cause for the differential regulation of circulatinglevels of miR-27b and miR-320a is currently unclear Formany biomarkers the cellular origin remains uncertain Forexample both ECs as well as platelets secrete TSP-1 Mea-suring its circulating levels does not reveal how much ofTSP-1 is endothelial or platelet derived The same limita-tion applies to miRNAs that are not tissue specific anddetected in the circulation Both miR-320a and miR-27bare also present in platelets (6) The opposing direction-ality of the association of miR-27b and miR-320a withdiabetic retinopathy could hint to a different cellular or-igin of these two circulating miRNAs ie miR-320a issecreted at much higher levels from ECs than miR-27bFurther studies are required to provide an in-depth un-derstanding of their cellular origin and to test the diag-nostic or therapeutic potential of these two circulatingmiRNAs in diabetic retinopathy We measured miRNAlevels only at baseline and hence could not assess or takeinto account within-person variability of the miRNAs overtime Finally model selection and parameter estimation wereperformed on the same data set without adjustment whichmay lead to an overestimation of the strength of associationobtained and underestimation of confidence limits

ConclusionsUnderstanding how circulating miRNAs could be har-nessed for assessing the risk of retinopathy in T1D is anessential area of research Our data in the DIRECT trials

diabetesdiabetesjournalsorg Zampetaki and Associates 225

show that miR-320a and miR-27b are associated with newonset of retinopathy and progression of retinopathyBesides identifying circulating miRNAs associated withretinopathy we also interrogated the targets of miR-27band miR-320a in ECs using a proteomics approach Takentogether the findings of our study identify miRNA bio-markers for retinopathy in two independent cohortsThese findings await confirmation in larger studies but ourtwo lead miRNAs may have clinical utility given theirestablished links to angiogenesis including the translationalcontrol of TSP-1 by miR-27b and its reduced endothelialsecretion by miR-320a

Acknowledgments The authors thank Raimund Pechlaner Departmentof Neurology Medical University of Innsbruck and Philipp Skroblin Cardiovas-cular Division Kingrsquos College London for help with preparing the manuscriptFunding AZ is an Intermediate Fellow of the British Heart Foundation(FS131830207) PW is an Erwin Schroumldinger Fellow in Epidemiologysponsored by the Austrian Science Fund MM is a Senior Fellow of theBritish Heart Foundation (FS130229892) This work was supported in partby JDRF (17-2011-658) Diabetes UK (120004530) the Fondation Leducq(MicroRNA-based Therapeutic Strategies in Vascular Disease [MIRVAD] [13CVD 02]) and the National Institute of Health Research Biomedical ResearchCenter based at Guyrsquos and St Thomasrsquo National Health Service FoundationTrust and Kingrsquos College London in partnership with Kingrsquos College HospitalDuality of Interest The DIRECT trial was jointly funded by AstraZeneca andTakeda Kingrsquos College London has filed patent applications on miRNA biomarkersNo other potential conflicts of interest relevant to this article were reportedAuthor Contributions AZ SB and XY performed the experimentsAZ PW NC and MM wrote the manuscript PW SRL SK and NCanalyzed data RK NC and MM conceived and designed the experimentsPR provided the serum samples MM is the guarantor of this work and assuch had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis

References1 Zampetaki A Mayr M MicroRNAs in vascular and metabolic disease CircRes 2012110508ndash5222 van Rooij E Olson EN MicroRNA therapeutics for cardiovascular diseaseopportunities and obstacles Nat Rev Drug Discov 201211860ndash8723 Kantharidis P Wang B Carew RM Lan HY Diabetes complications themicroRNA perspective Diabetes 2011601832ndash18374 Zampetaki A Kiechl S Drozdov I et al Plasma microRNA profiling revealsloss of endothelial miR-126 and other microRNAs in type 2 diabetes Circ Res2010107810ndash8175 Zampetaki A Willeit P Tilling L et al Prospective study on circulatingMicroRNAs and risk of myocardial infarction J Am Coll Cardiol 201260290ndash2996 Willeit P Zampetaki A Dudek K et al Circulating microRNAs as novelbiomarkers for platelet activation Circ Res 2013112595ndash6007 Zampetaki A Willeit P Drozdov I Kiechl S Mayr M Profiling of circulatingmicroRNAs from single biomarkers to re-wired networks Cardiovasc Res 201293555ndash5628 Mayr M Zampetaki A Willeit P Willeit J Kiechl S MicroRNAs within thecontinuum of postgenomics biomarker discovery Arterioscler Thromb Vasc Biol201333206ndash2149 Chaturvedi N Porta M Klein R et al DIRECT Programme Study GroupEffect of candesartan on prevention (DIRECT-Prevent 1) and progression (DIRECT-Protect 1) of retinopathy in type 1 diabetes randomised placebo-controlled trialsLancet 20083721394ndash1402

10 Yin X Bern M Xing Q Ho J Viner R Mayr M Glycoproteomic analysis of thesecretome of human endothelial cells Mol Cell Proteomics 201312956ndash97811 Abonnenc M Nabeebaccus AA Mayr U et al Extracellular matrix secretionby cardiac fibroblasts role of microRNA-29b and microRNA-30c Circ Res 20131131138ndash114712 Higgins JP Thompson SG Quantifying heterogeneity in a meta-analysisStat Med 2002211539ndash155813 Goeman JJ L1 penalized estimation in the Cox proportional hazards modelBiom J 20105270ndash8414 Dhanabal M Wu F Alvarez E et al Recombinant semaphorin 6A-1 ecto-domain inhibits in vivo growth factor and tumor cell line-induced angiogenesisCancer Biol Ther 20054659ndash66815 Miao HQ Lee P Lin H Soker S Klagsbrun M Neuropilin-1 expression bytumor cells promotes tumor angiogenesis and progression FASEB J 2000142532ndash253916 Zampetaki A Attia R Mayr U et al Role of miR-195 in aortic aneurysmaldisease Circ Res 2014115857ndash86617 Wang JM Tao J Chen DD et al MicroRNA miR-27b rescues bone marrow-derived angiogenic cell function and accelerates wound healing in type 2 di-abetes mellitus Arterioscler Thromb Vasc Biol 20143499ndash10918 Karolina DS Tavintharan S Armugam A et al Circulating miRNA profiles inpatients with metabolic syndrome J Clin Endocrinol Metab 201297E2271ndashE227619 Diabetes Control and Complications Trial (DCCT)Epidemiology of DiabetesInterventions and Complications (EDIC) Research Group Lachin JM White NHHainsworth DP Sun W Cleary PA Nathan DM Effect of intensive diabetes therapyon the progression of diabetic retinopathy in patients with type 1 diabetes 18 yearsof follow-up in the DCCTEDIC Diabetes 201564631ndash64220 van Hecke MV Dekker JM Stehouwer CD et al EURODIAB ProspectiveComplications Study Diabetic retinopathy is associated with mortality and car-diovascular disease incidence the EURODIAB Prospective Complications StudyDiabetes Care 2005281383ndash138921 Kovacs B Lumayag S Cowan C Xu S MicroRNAs in early diabetic reti-nopathy in streptozotocin-induced diabetic rats Invest Ophthalmol Vis Sci 2011524402ndash440922 Feng B Chen S McArthur K et al miR-146a-mediated extracellular matrixprotein production in chronic diabetes complications Diabetes 2011602975ndash298423 Silva VA Polesskaya A Sousa TA et al Expression and cellular localizationof microRNA-29b and RAX an activator of the RNA-dependent protein kinase(PKR) in the retina of streptozotocin-induced diabetic rats Mol Vis 2011172228ndash224024 McArthur K Feng B Wu Y Chen S Chakrabarti S MicroRNA-200b regu-lates vascular endothelial growth factor-mediated alterations in diabetic reti-nopathy Diabetes 2011601314ndash132325 Murray AR Chen Q Takahashi Y Zhou KK Park K Ma JX MicroRNA-200bdownregulates oxidation resistance 1 (Oxr1) expression in the retina of type 1diabetes model Invest Ophthalmol Vis Sci 2013541689ndash169726 Bai Y Bai X Wang Z Zhang X Ruan C Miao J MicroRNA-126 inhibitsischemia-induced retinal neovascularization via regulating angiogenic growthfactors Exp Mol Pathol 201191471ndash47727 Ling S Birnbaum Y Nanhwan MK Thomas B Bajaj M Ye Y MicroRNA-dependent cross-talk between VEGF and HIF1a in the diabetic retina Cell Signal2013252840ndash284728 Tang H Lee M Sharpe O et al Oxidative stress-responsive microRNA-320regulates glycolysis in diverse biological systems FASEB J 2012264710ndash472129 Wang X Huang W Liu G et al Cardiomyocytes mediate anti-angiogenesisin type 2 diabetic rats through the exosomal transfer of miR-320 into endothelialcells J Mol Cell Cardiol 201474139ndash15030 Wu YY Chen YL Jao YC Hsieh IS Chang KC Hong TM miR-320 regulatestumor angiogenesis driven by vascular endothelial cells in oral cancer by si-lencing neuropilin 1 Angiogenesis 201417247ndash26031 Kuehbacher A Urbich C Zeiher AM Dimmeler S Role of Dicer and Drosha forendothelial microRNA expression and angiogenesis Circ Res 200710159ndash68

226 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

show that miR-320a and miR-27b are associated with newonset of retinopathy and progression of retinopathyBesides identifying circulating miRNAs associated withretinopathy we also interrogated the targets of miR-27band miR-320a in ECs using a proteomics approach Takentogether the findings of our study identify miRNA bio-markers for retinopathy in two independent cohortsThese findings await confirmation in larger studies but ourtwo lead miRNAs may have clinical utility given theirestablished links to angiogenesis including the translationalcontrol of TSP-1 by miR-27b and its reduced endothelialsecretion by miR-320a

Acknowledgments The authors thank Raimund Pechlaner Departmentof Neurology Medical University of Innsbruck and Philipp Skroblin Cardiovas-cular Division Kingrsquos College London for help with preparing the manuscriptFunding AZ is an Intermediate Fellow of the British Heart Foundation(FS131830207) PW is an Erwin Schroumldinger Fellow in Epidemiologysponsored by the Austrian Science Fund MM is a Senior Fellow of theBritish Heart Foundation (FS130229892) This work was supported in partby JDRF (17-2011-658) Diabetes UK (120004530) the Fondation Leducq(MicroRNA-based Therapeutic Strategies in Vascular Disease [MIRVAD] [13CVD 02]) and the National Institute of Health Research Biomedical ResearchCenter based at Guyrsquos and St Thomasrsquo National Health Service FoundationTrust and Kingrsquos College London in partnership with Kingrsquos College HospitalDuality of Interest The DIRECT trial was jointly funded by AstraZeneca andTakeda Kingrsquos College London has filed patent applications on miRNA biomarkersNo other potential conflicts of interest relevant to this article were reportedAuthor Contributions AZ SB and XY performed the experimentsAZ PW NC and MM wrote the manuscript PW SRL SK and NCanalyzed data RK NC and MM conceived and designed the experimentsPR provided the serum samples MM is the guarantor of this work and assuch had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis

References1 Zampetaki A Mayr M MicroRNAs in vascular and metabolic disease CircRes 2012110508ndash5222 van Rooij E Olson EN MicroRNA therapeutics for cardiovascular diseaseopportunities and obstacles Nat Rev Drug Discov 201211860ndash8723 Kantharidis P Wang B Carew RM Lan HY Diabetes complications themicroRNA perspective Diabetes 2011601832ndash18374 Zampetaki A Kiechl S Drozdov I et al Plasma microRNA profiling revealsloss of endothelial miR-126 and other microRNAs in type 2 diabetes Circ Res2010107810ndash8175 Zampetaki A Willeit P Tilling L et al Prospective study on circulatingMicroRNAs and risk of myocardial infarction J Am Coll Cardiol 201260290ndash2996 Willeit P Zampetaki A Dudek K et al Circulating microRNAs as novelbiomarkers for platelet activation Circ Res 2013112595ndash6007 Zampetaki A Willeit P Drozdov I Kiechl S Mayr M Profiling of circulatingmicroRNAs from single biomarkers to re-wired networks Cardiovasc Res 201293555ndash5628 Mayr M Zampetaki A Willeit P Willeit J Kiechl S MicroRNAs within thecontinuum of postgenomics biomarker discovery Arterioscler Thromb Vasc Biol201333206ndash2149 Chaturvedi N Porta M Klein R et al DIRECT Programme Study GroupEffect of candesartan on prevention (DIRECT-Prevent 1) and progression (DIRECT-Protect 1) of retinopathy in type 1 diabetes randomised placebo-controlled trialsLancet 20083721394ndash1402

10 Yin X Bern M Xing Q Ho J Viner R Mayr M Glycoproteomic analysis of thesecretome of human endothelial cells Mol Cell Proteomics 201312956ndash97811 Abonnenc M Nabeebaccus AA Mayr U et al Extracellular matrix secretionby cardiac fibroblasts role of microRNA-29b and microRNA-30c Circ Res 20131131138ndash114712 Higgins JP Thompson SG Quantifying heterogeneity in a meta-analysisStat Med 2002211539ndash155813 Goeman JJ L1 penalized estimation in the Cox proportional hazards modelBiom J 20105270ndash8414 Dhanabal M Wu F Alvarez E et al Recombinant semaphorin 6A-1 ecto-domain inhibits in vivo growth factor and tumor cell line-induced angiogenesisCancer Biol Ther 20054659ndash66815 Miao HQ Lee P Lin H Soker S Klagsbrun M Neuropilin-1 expression bytumor cells promotes tumor angiogenesis and progression FASEB J 2000142532ndash253916 Zampetaki A Attia R Mayr U et al Role of miR-195 in aortic aneurysmaldisease Circ Res 2014115857ndash86617 Wang JM Tao J Chen DD et al MicroRNA miR-27b rescues bone marrow-derived angiogenic cell function and accelerates wound healing in type 2 di-abetes mellitus Arterioscler Thromb Vasc Biol 20143499ndash10918 Karolina DS Tavintharan S Armugam A et al Circulating miRNA profiles inpatients with metabolic syndrome J Clin Endocrinol Metab 201297E2271ndashE227619 Diabetes Control and Complications Trial (DCCT)Epidemiology of DiabetesInterventions and Complications (EDIC) Research Group Lachin JM White NHHainsworth DP Sun W Cleary PA Nathan DM Effect of intensive diabetes therapyon the progression of diabetic retinopathy in patients with type 1 diabetes 18 yearsof follow-up in the DCCTEDIC Diabetes 201564631ndash64220 van Hecke MV Dekker JM Stehouwer CD et al EURODIAB ProspectiveComplications Study Diabetic retinopathy is associated with mortality and car-diovascular disease incidence the EURODIAB Prospective Complications StudyDiabetes Care 2005281383ndash138921 Kovacs B Lumayag S Cowan C Xu S MicroRNAs in early diabetic reti-nopathy in streptozotocin-induced diabetic rats Invest Ophthalmol Vis Sci 2011524402ndash440922 Feng B Chen S McArthur K et al miR-146a-mediated extracellular matrixprotein production in chronic diabetes complications Diabetes 2011602975ndash298423 Silva VA Polesskaya A Sousa TA et al Expression and cellular localizationof microRNA-29b and RAX an activator of the RNA-dependent protein kinase(PKR) in the retina of streptozotocin-induced diabetic rats Mol Vis 2011172228ndash224024 McArthur K Feng B Wu Y Chen S Chakrabarti S MicroRNA-200b regu-lates vascular endothelial growth factor-mediated alterations in diabetic reti-nopathy Diabetes 2011601314ndash132325 Murray AR Chen Q Takahashi Y Zhou KK Park K Ma JX MicroRNA-200bdownregulates oxidation resistance 1 (Oxr1) expression in the retina of type 1diabetes model Invest Ophthalmol Vis Sci 2013541689ndash169726 Bai Y Bai X Wang Z Zhang X Ruan C Miao J MicroRNA-126 inhibitsischemia-induced retinal neovascularization via regulating angiogenic growthfactors Exp Mol Pathol 201191471ndash47727 Ling S Birnbaum Y Nanhwan MK Thomas B Bajaj M Ye Y MicroRNA-dependent cross-talk between VEGF and HIF1a in the diabetic retina Cell Signal2013252840ndash284728 Tang H Lee M Sharpe O et al Oxidative stress-responsive microRNA-320regulates glycolysis in diverse biological systems FASEB J 2012264710ndash472129 Wang X Huang W Liu G et al Cardiomyocytes mediate anti-angiogenesisin type 2 diabetic rats through the exosomal transfer of miR-320 into endothelialcells J Mol Cell Cardiol 201474139ndash15030 Wu YY Chen YL Jao YC Hsieh IS Chang KC Hong TM miR-320 regulatestumor angiogenesis driven by vascular endothelial cells in oral cancer by si-lencing neuropilin 1 Angiogenesis 201417247ndash26031 Kuehbacher A Urbich C Zeiher AM Dimmeler S Role of Dicer and Drosha forendothelial microRNA expression and angiogenesis Circ Res 200710159ndash68

226 Angiogenic microRNAs and Diabetic Retinopathy Diabetes Volume 65 January 2016

32 Zhou Q Gallagher R Ufret-Vincenty R Li X Olson EN Wang SRegulation of angiogenesis and choroidal neovascularization by membersof microRNA-232724 clusters Proc Natl Acad Sci U S A 20111088287ndash829233 Urbich C Kaluza D Froumlmel T et al MicroRNA-27ab controls endothelialcell repulsion and angiogenesis by targeting semaphorin 6A Blood 20121191607ndash161634 Biyashev D Veliceasa D Topczewski J et al miR-27b controls venousspecification and tip cell fate Blood 20121192679ndash268735 Stenina-Adognravi O Thrombospondins old players new games Curr OpinLipidol 201324401ndash40936 Tolsma SS Stack MS Bouck N Lumen formation and other angiogenicactivities of cultured capillary endothelial cells are inhibited by thrombospondin-1Microvasc Res 19975413ndash2637 Kaur S Martin-Manso G Pendrak ML Garfield SH Isenberg JS Roberts DDThrombospondin-1 inhibits VEGF receptor-2 signaling by disrupting its associa-tion with CD47 J Biol Chem 201028538923ndash3893238 Cursiefen C Masli S Ng TF et al Roles of thrombospondin-1 and -2 inregulating corneal and iris angiogenesis Invest Ophthalmol Vis Sci 2004451117ndash112439 Suzuma K Takagi H Otani A Oh H Honda Y Expression of thrombospondin-1in ischemia-induced retinal neovascularization Am J Pathol 1999154343ndash354

40 Kermorvant-Duchemin E Sennlaub F Sirinyan M et al Trans-arachidonicacids generated during nitrative stress induce a thrombospondin-1-dependentmicrovascular degeneration Nat Med 2005111339ndash134541 Choi J Lin A Shrier E Lau LF Grant MB Chaqour B Degradome productsof the matricellular protein CCN1 as modulators of pathological angiogenesis inthe retina J Biol Chem 201328823075ndash2308942 Sorenson CM Wang S Gendron R Paradis H Sheibani N Thrombospondin-1deficiency exacerbates the pathogenesis of diabetic retinopathy J Diabet MetabolS12005 doi 1041722155-6156S12-00543 Ii M Takenaka H Asai J et al Endothelial progenitor thrombospondin-1mediates diabetes-induced delay in reendothelialization following arterial injuryCirc Res 200698697ndash70444 Levey AS Stevens LA Schmid CH et al CKD-EPI (Chronic Kidney DiseaseEpidemiology Collaboration) A new equation to estimate glomerular filtrationrate Ann Intern Med 2009150604ndash61245 Levey AS Coresh J Greene T et al Chronic Kidney Disease EpidemiologyCollaboration Expressing the Modification of Diet in Renal Disease Studyequation for estimating glomerular filtration rate with standardized serum cre-atinine values Clin Chem 200753766ndash77246 Williams KV Erbey JR Becker D Arslanian S Orchard TJ Can clinicalfactors estimate insulin resistance in type 1 diabetes Diabetes 200049626ndash632

diabetesdiabetesjournalsorg Zampetaki and Associates 227