MicroRNA-15a and -16-1 act via MYB to elevate fetalhemoglobin expression in human trisomy 13Vijay G. Sankarana,b,c,d, Tobias F. Mennee,f, Danilo Š�cepanovi�cg, Jo-Anne Vergilioh, Peng Jia, Jinkuk Kima,g,i,Prathapan Thirua, Stuart H. Orkind,e,f,j, Eric S. Landera,b,k,l, and Harvey F. Lodisha,b,l,1

aWhitehead Institute for Biomedical Research, Cambridge, MA 02142; bBroad Institute of Massachusetts Institute of Technology and Harvard University,Cambridge, MA 02142; Departments of cMedicine and hPathology and eDivision of Hematology/Oncology, Children’s Hospital Boston, Boston, MA 02115;Departments of dPediatrics and kSystems Biology, Harvard Medical School, Boston, MA 02115; lDepartment of Biology and iHoward Hughes Medical Institute,Massachusetts Institute of Technology, Cambridge, MA 02142; gHarvard–Massachusetts Institute of Technology Division of Health Sciences and Technology,Cambridge, MA 02142; fDepartment of Pediatric Oncology, Dana-Farber Cancer Institute, Boston, MA 02115; and jThe Howard Hughes Medical Institute,Boston, MA 02115

Contributed by Harvey F. Lodish, December 13, 2010 (sent for review November 8, 2010)

Many human aneuploidy syndromes have unique phenotypicconsequences, but in most instances it is unclear whether thesephenotypes are attributable to alterations in the dosage of specificgenes. In human trisomy 13, there is delayed switching andpersistence of fetal hemoglobin (HbF) and elevation of embryonichemoglobin in newborns. Using partial trisomy cases, we mappedthis trait to chromosomal band 13q14; by examining the genes inthis region, two microRNAs, miR-15a and -16-1, appear as topcandidates for the elevated HbF levels. Indeed, increased expres-sion of these microRNAs in primary human erythroid progenitorcells results in elevated fetal and embryonic hemoglobin geneexpression. Moreover, we show that a direct target of thesemicroRNAs, MYB, plays an important role in silencing the fetal andembryonic hemoglobin genes. Thus we demonstrate how the de-velopmental regulation of a clinically important human trait canbe better understood through the genetic and functional study ofaneuploidy syndromes and suggest that miR-15a, -16-1, and MYBmay be important therapeutic targets to increase HbF levels inpatients with sickle cell disease and β-thalassemia.

erythropoiesis | globin gene regulation

Human syndromes that are attributable to chromosomalimbalances or aneuploidy provide a unique opportunity to

understand the phenotypic consequences of altered gene dosage(1, 2). Such observations also provide the prospect of gaininginsight into the mechanisms mediating normal human de-velopment and physiology. However, in the vast majority ofinstances there is a limited understanding of how alterations inspecific genetic loci contribute to the consequent phenotypicfeatures seen in aneuploidy syndromes. Trisomy of chromosome13 is one of the few viable human aneuploidies and is associatedwith a number of unique features (1), including a delayed switchfrom fetal to adult hemoglobin and persistently elevated levels offetal hemoglobin (HbF) (1, 3–5). This trait is of considerableinterest given that it is one of the few quantitative and objectivebiochemical phenotypes described in such syndromes. Addi-tionally, the regulation of HbF is of great interest given the well-characterized role of elevated HbF in ameliorating clinical se-verity in sickle cell disease and β-thalassemia (6, 7).During human development a series of switches occurs in-

volving the transcription of the globin genes residing within theβ-globin locus on human chromosome 11. A transient lineage ofred blood cells, the primitive erythroid lineage, is produced inthe first few weeks of human gestation (8). These cells producea unique embryonic β-like globin chain, ε-globin (9). Additionallysmall amounts of other β-like globin genes are expressed in thislineage (8, 9). Subsequently, definitive erythroid cells are pro-duced from long-term self-renewing hematopoietic stem cells.Initially these cells predominantly express the β-like fetal he-moglobin gene, γ-globin, and are produced in the fetal liver (10).Around the time of birth, when production of erythroid and

other hematopoietic cells shifts to the bone marrow, the pre-dominant postnatal site for hematopoiesis, another switchoccurs, resulting in down-regulation of γ-globin and concomitantup-regulation of the adult β-globin gene (6, 8, 10). There isa limited understanding of the molecular control of these globingene switches that occur in human ontogeny, particularly withregard to the fetal-to-adult hemoglobin switch within the de-finitive erythroid lineage. Recent insight into these mechanismshas come from the field of human genetics (6) and resulted in theidentification of the transcription factor BCL11A as a majorregulator of this process (7). However, it is apparent that thisfactor cannot solely be responsible for this switch in ontogeny.

ResultsMapping of partial trisomy cases provides an opportunity todeduce genotype–phenotype relationships (1, 11). Given thedramatic reduction in births with trisomy 13 following theavailability of prenatal diagnosis (12), mapping of such traitsmust rely on cases that have previously been cytogeneticallymapped. Analysis of partial trisomy 13 cases has suggested thatspecific regions on the proximal part of chromosome 13 may beassociated with elevations in HbF (1). Using eight well-anno-tated cases with detailed cytogenetic mapping data available(11), chromosomal band 13q14 appears to be unambiguouslyassociated with elevated HbF levels (Fig. 1A). By accounting forall 57 partial trisomy cases that have been reported with HbFmeasurements (SI Appendix, Fig. S1) (13, 14), with varyingdegrees of detail reported for cytogenetic mapping, a clear as-sociation with 13q14 is again deduced (Fig. 1B). This finding isstrongly supported by Bayesian chromosomal region associationmodels we developed (SI Appendix).We then used an integrative genomic approach to identify

candidates within the region implicated from the partial trisomycases, by analyzing a gene expression compendium to search forgenes with preferential expression in erythroid precursors(CD71+) relative to other cell types (SI Appendix) (15). Of the 76genes in the region, 14 (18%) passed this test (SI Appendix, Fig.S2 and Table S1). We could further filter potential candidatesby examining whether the histone 3 lysine 4 trimethylation(H3K4me3) modification, a well-characterized marker of active

Author contributions: V.G.S. designed research; V.G.S., T.F.M., J.-A.V., and P.J. performedresearch; V.G.S., T.F.M., D.S., P.T., S.H.O., E.S.L., and H.F.L. contributed new reagents/analytic tools; V.G.S., D.S., J.-A.V., J.K., P.T., S.H.O., E.S.L., and H.F.L. analyzed data; andV.G.S., E.S.L., and H.F.L. wrote the paper.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE25678).1To whom correspondence should be addressed. E-mail: lodish@wi.mit.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1018384108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1018384108 PNAS | January 25, 2011 | vol. 108 | no. 4 | 1519–1524

GEN

ETICS

transcription, was present in the proximal promoter of the genesfrom chromosomal band 13q14 in erythroid cell lines (defined ashaving a peak of H3K4me3 within 1.5 kb upstream from thetranscription start site). Using data derived from the ENCODEproject (16), we were able to find a number of unique peaks (SIAppendix, Table S2) and of the gene list established from theexpression data, we could focus on 9 candidate genes (SI Ap-pendix, Table S3). Of these, we noted that a top candidate in thisregion was a precursor RNA (DLEU2) for two microRNAs, 15aand 16-1, which have identical seed targeting sequences (Fig.1C). This was of particular interest, given the role that micro-

RNAs play in modulating various aspects of hematopoiesis anderythropoiesis specifically (17, 18). MiR-15a and -16 are expres-sed throughout human erythroid maturation in adult bone mar-row cells and increase modestly with terminal differentiation(Fig. 1 D and E), consistent with prior observations in cord bloodand erythroid cell lines (19, 20).These observations suggest that increased expression of miR-

15a/16-1 in trisomy 13 could potentially result in elevated levelsof HbF expression. To directly test this hypothesis, we useda lentiviral vector to increase expression of these microRNAs inadult bone marrow-derived hematopoietic progenitors that sub-

1312

11.211.1

1112

13

14

21

22

31

32

33

34

q

pElevated HbF

Normal HbF

Chr.

13

Partial Trisomy Cases

pter−q11 q12 q13 q14 q21 q22 q31 q32−qter0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Elevated HbFNormal HbF

Chr. 13 Band

Po

rtio

n w

ith

Tri

som

y

Chromosome BandChromosome Bands Localized by FISH Mapping Clones

RefSeq Genes13q14.11 13q14.12 13q14.2 13q14.3

LOC646982LOC646982LOC646982

FOXO1MRPS31

SLC25A15SUGT1L1

MIR621ELF1ELF1WBP4

KBTBD6KBTBD7

MTRF1NARG1LNARG1LNARG1LOR7E37POR7E37PC13orf15KIAA0564

KIAA0564DGKHDGKHAKAP11

TNFSF11TNFSF11

C13orf30

EPSTI1

EPSTI1DNAJC15

ENOX1ENOX1

CCDC122C13orf31C13orf31

LOC121838

SERP2

TSC22D1TSC22D1

NUFIP1KIAA1704

GTF2F2

KCTD4

TPT1SNORA31

LOC100190939SLC25A30

COG3FAM194B

SPERT

SIAH3ZC3H13

CPB2

CPB2LCP1

C13orf18

LRCH1LRCH1LRCH1

ESDHTR2AHTR2A

SUCLA2NUDT15

MED4ITM2B

RB1LPAR6LPAR6LPAR6

RCBTB2CYSLTR2

FNDC3AFNDC3A

MLNRCDADC1

CAB39LCAB39L

SETDB2SETDB2

PHF11PHF11

RCBTB1ARL11

EBPLKPNA3

LOC220429C13orf1C13orf1C13orf1

DLEU2TRIM13TRIM13TRIM13TRIM13

KCNRGKCNRG

MIR16-1MIR15A

DLEU1

FAM10A4

DLEU7RNASEH2BRNASEH2B

GUCY1B2

FAM124ASERPINE3

INTS6

INTS6INTS6WDFY2

DHRS12

DHRS12FLJ37307FLJ37307

CCDC70

ATP7BATP7BALG11

UTP14C

NEK5NEK3NEK3

NEK3NEK3

THSD1PTHSD1THSD1VPS36CKAP2CKAP2

LOC220115HNRNPA1L2HNRNPA1L2

SUGT1

SUGT1LECT1LECT1

1 3 5 7 90.0

0.2

0.4

0.6

0.8

1.0

Days of Differentation

Rel

ativ

e Ex

pres

sion

1 3 5 7 90

20

40

60

80

Days of Differentation

Rel

ativ

e E x

pres

sion

A B

C

D E

miR-15a miR-16

Fig. 1. Identification of miR-15a/16-1 as candidates for causing the elevated HbF levels in trisomy 13. (A) A set of eight well-annotated cases, along with thefull trisomy of chromosome 13, demonstrates that chromosomal band 13q14 is umambiguously associated with the trait of elevated HbF levels (11). Cases withand without elevated HbF levels are labeled in purple and blue, respectively. Cases are considered to have elevated HbF levels when the measured level isgreater than two SDs above the mean level for age (3, 21). (B) A compilation of all 57 partial trisomy cases reported shows that chromosomal band 13q14 ismost frequently associated with elevated HbF (13, 14). The proportion of cases with elevated or normal HbF levels in patients with trisomy of each chro-mosomal band is shown in this diagram. The association with chromosomal band 13q14 finding is supported using Bayesian probabilistic models (SI Ap-pendix). (C) A diagram showing all of the genes on chromosomal band 13q14. Of these, using an integrative genomic approach, miR-15a and -16-1 appear astop candidates (highlighted with a red box). (D) Pattern of miR-15a expression during human adult erythroid progenitor differentiation (shown relative toRNU19 expression; n = 4 per time point). The range of stages covers the cells from early CFU-E progenitors (day 1 of differentiation) to polychromatophilicerythroblasts (day 9 of differentiation), as described previously (7). (E) Pattern of miR-16 expression during human adult erythroid progenitor differentiation(shown relative to RNU19 expression; n = 4 per time point).

1520 | www.pnas.org/cgi/doi/10.1073/pnas.1018384108 Sankaran et al.

sequently underwent synchronous differentiation toward the ery-throid lineage (7). By increasing expression of miR-15a and -16by an average of 1.5-fold (SI Appendix, Fig. S3), similar to whatwould be expected in the context of a trisomy, we found that thelevels of γ-globin gene expression were robustly increased by anaverage of 2.4-fold (Fig. 2A). Because trisomy 13 can result inelevated γ-globin synthesis even in newborns with elevated HbFlevels at baseline (21), we increased expression of these micro-RNAs by 2- to 3-fold in human erythroleukemia K562 cells (SIAppendix, Fig. S4), which endogenously express high levels ofγ-globin, and found that the γ-globin levels could be further in-creased by 50% (Fig. 2B). Newborns with trisomy 13 also showelevated expression of the embryonically expressed ε-globin, asillustrated by the persistence of low level hemoglobin Gower 2(α2ε2) expression (4). This increase was recapitulated in the pri-mary bone marrow-derived cells with increased miR-15a/16-1expression (Fig. 2C). Since alterations in γ-globin expression mayaccompany altered differentiation of cells, we examined the len-tivirally transduced primary adult erythroid progenitors and foundnomajor differences in the morphology or phenotype of cells withincreased miR-15a/16-1 expression compared with controls (Fig.2D and SI Appendix, Fig. S5). To gain further insight into themechanism by which these microRNAs may be acting to elevateγ-globin expression, we assessed cell cycle progression in the syn-chronously differentiating primary erythroid cells (7). Interest-ingly, we noted that cell cycle progression was slowed by miR-15a/16-1 overexpression at the early stages of differentiation thatrepresent colony-forming unit erythroid cells (CFU-Es) or pro-erythroblasts (G1- and S-phase difference, P < 0.001), but was

equivalent to controls at the later stages that represent more ma-ture (basophilic) erythroblasts (Fig. 2 E and F) (7). Such findingsare reminiscent of the observations made using S-phase inhibitorsfor HbF induction in primate models, where the most responsivestages appeared to be the CFU-Es and proerythroblasts (22, 23).To gain further insight into the molecular etiology for these

observations, we examined whether specific miR-15a/16-1 targetsmay be potential mediators of the increased HbF expression.Because evolutionary conservation of seed-matched targets showsgreat utility in identifying bona fide microRNA targets (24), weused a metric of target context and conservation (aggregate PCT,which is a Bayesian estimate of the probability that a site is con-served due to selective maintenance of miRNA targeting ratherthan by chance) and compared this with relative expression ofmRNAs in the erythroid target tissues of interest relative to a largenumber of other tissues (Fig. 3A and SI Appendix, Fig. S6). Usingthis approach,MYBwas onemiR-15a/16-1 target that appeared tobe of great interest, given that it was highly expressed in erythroidprogenitors and had two conserved 8mer miR-15a/16-1 targetingsites (Fig. 3B). This was particularly notable, because commongenetic variants from genomewide association studies have sug-gested that polymorphisms in the MYB locus are important me-diators of variation in HbF levels in humans (25), which is fur-ther supported by the finding that overexpression of MYB in celllines causes a decrease in γ-globin expression (26). We found thatMYB protein levels were reduced with even a modest (two- tothreefold) increase in miR-15a/16-1 expression in erythroid celllines (Fig. 3C) and a luciferase reporter assay confirmed that

A B C

Control

miR-15

a-16-1

0

10

20

30

40

Perc

enta

ge o

f-G

lobi

n

Control

miR-15

a-16-1

0.0

0.5

1.0

1.5

2.0

Rel

ativ

e-G

lobi

n Ex

pres

sion

Control

miR-15

a-16-1

0.00000

0.00005

0.00010

0.00015

Perc

enta

ge o

f-G

lobi

n

**

D

Control

miR

15a-16

E F

Control

miR-15

a-16

0

40

80

120

Perc

enta

ge o

f Cel

ls

Control

miR-15

a-16

0

40

80

120

G1SG2/M

Perc

enta

ge o

f Cel

lsεγ

γ

Fig. 2. Increased expression of miR-15a/16-1 in human erythroid cells results in elevated HbF and embryonic globin gene expression. (A) Percentage ofγ-globin gene expression (as a percentage of all human β-like globin genes) in cells transduced with pLVX-puro control or pLVX-miR-15a/16-1 lentivirus (n = 3per group; ***P < 0.001). Measurement was at the basophilic erythroblast stage of differentiation (days 6–7 of differentiation). (B) Relative amount ofγ-globin gene expression in K562 cells transduced with pSMPUW control or miR-15a/16-1 containing lentivirus, following selection (n = 3 per group; *P < 0.02).(C) Percentage of ε-globin gene expression (as a percentage of all human β-like globin genes) in primary bone marrow CD34-derived cells transduced withpLVX-puro control or pLVX/miR-15a/16-1 lentivirus (n = 3 per group; **P < 0.01). (D) Representative cytospin images of primary bone marrow CD34-derivedcells transduced with pLVX-puro control or pLVX/miR-15a/16-1 lentivirus (taken with a 63× objective lens). All cells show similar size and morphologicaldistribution on days 5–6 of differentiation. At other stages of differentiation the control and miR-15a/16-1 transduced cells also had the same morphology. (Eand F) Cell cycle analysis of primary bone marrow CD34+-derived cells transduced with pLVX-puro control or pLVX/miR-15a/16-1 lentivirus on day 4 (E) and day7 (F) of differentiation (n = 3–4 per group). All data are shown as the mean ± the SD.

Sankaran et al. PNAS | January 25, 2011 | vol. 108 | no. 4 | 1521

GEN

ETICS

theMYB 3′-UTR is a direct target of these microRNAs (Fig. 3D),consistent with previous studies (20, 27, 28).To test whether MYB may be a critical mediator of γ-globin

expression, we reduced MYB expression with two shRNAs insynchronously differentiating adult erythroid progenitors (Fig.3E). This knockdown robustly increased γ-globin expression, as

occurs to a lesser extent with miR-15a/16-1 elevation (Figs. 2Aand 3F). Concomitantly, we found that expression of the em-bryonic globin chain, ε-globin, was also dramatically increased byMYB knockdown (Fig. 3G). Our examination of expression datafrom a recent study of MYB siRNA treatment of umbilical cordblood erythroid progenitors (29) demonstrated robust elevations

A

1.1

-5

0

5

10

Aggregate PCT

Log 2

Expr

essi

onR

atio

MYB 3’UTR 654-661 5’...AUGAAAAACGUUUUUUGCUGCUA...3’MYB 3’UTR 681-688 5’...CUUAGCCUGUAGACAUGCUGCUA...3’ ||||||| hsa-miR-15a GUGUUUGGUAAUACACGACGAU-5’hsa-miR-16-1 GCGGUGAUAAAUGCACGACGAU-5’

B

MYB

C

D

Control

miR-15

a-16-1

0.0

0.4

0.8

1.2

Rel

ativ

e Lu

cife

rase

Act

ivity

MYB

GAPDH

Control

miR

-15a-1

6

E F G

***

Control

shMYB 1

shMYB 2

0.0

0.5

1.0

Rel

ativ

eMYB

Exp

r ess

ion

Control

shMYB 1

shMYB 2

0

20

40

60

80Pe

rcen

tage

of

-Glo

bin

Control

shMYB 1

shMYB 2

0.0000

0.0002

0.0004

0.0006

0.0008

Perc

enta

ge o

f-G

lobi

n

*** ***

*** *** ******

H

0.0

0.2

0.4

0.6

0.8

Enri

chm

ent

Sco

re

Up in shMYB Down in shMYB

I

Propidium Iodide

% o

f Max

imu

m

20

40

60

80

100

shMYB 1 - 39.94 %shMYB 2 - 36.43 %

Control - 68.65% Cells in S-phase

γ ε

Fig. 3. MYB is a target of miR-15a/16-1 and regulates HbF expression. (A) By comparing the aggregate PCT of a variety of miR-15 or -16 seed targets(24) with the log2 normalized relative expression in early (CD34+) hematopoietic/erythroid progenitors (relative to a panel of 78 other human cells andtissues), MYB appears to be a standout candidate target (highlighted in red with arrow). The x-axis plots aggregate PCT (24) on a linear scale, whereasthe y-axis shows relative expression in the erythroid progenitors as a log2 ratio. (B) Two 8mer target sites for miR-15a and -16-1 are located in the 3′-UTR of MYB. (C ) Ectopic expression of miR-15a/16-1 in K562 cells at a level two- to threefold of normal results in reduction in MYB protein levels;GADPH was used as a loading control for this Western blot. (D) Cotransfection of 293T cells with control or miR-15a/16-1 expression vector (in pLVX)and a MYB 3′-UTR construct luciferase reporter (in psiCHECK-2 vector) show reduced luciferase activity with elevated microRNA expression. (E )Relative MYB expression is shown on day 5 of differentiation in primary human CD34+-derived cells transduced with pLKO.1 control or pLKO.1 withshRNAs targeting MYB (n = 3 per group; ***P < 0.001). (F and G) Percentage of γ-globin (F ) and ε-globin (G) gene expression as a percentage of allhuman β-like globin genes on day 7 of differentiation in cells transduced with pLKO.1 control or pLKO.1 vector expressing two different shRNAstargeting MYB (n = 3 per group; ***P < 0.001). (H) Gene set enrichment analysis (30) of a monocyte gene expression signature, composed of 371 genes(31), evaluated using the expression array data of shMYB cells versus controls (n = 4 per group), demonstrates global up-regulation of the monocytegene signature in the shMYB cells. Enrichment plot is shown above heat map (below) using a green line to show the running enrichment score. (I)Representative flow cytometry profiles of propidium iodide staining in K562 cells transduced with empty or shMYB containing pLKO.1 lentiviruses.Data are shown as an average of three experiments for each group (P < 0.001 for both shMYB experiments relative to controls). All data are shown asthe mean ± SD.

1522 | www.pnas.org/cgi/doi/10.1073/pnas.1018384108 Sankaran et al.

of both γ-globin and ε-globin gene expression, even with higherbaseline levels of these globin genes at this stage of human on-togeny (SI Appendix, Fig. S7).Examination of cells with a knockdown of MYB demonstrated

an increased presence of myeloid (primarily monocyte) cells andsigns of precocious erythroid differentiation by morphologicalexamination (SI Appendix, Fig. S8). To more precisely define themolecular basis of these changes, expression analysis was per-formed with cells from one of the MYB knockdowns along witha set of controls. We did not observe any significant difference inmRNA expression levels of known regulators of erythropoiesisand globin gene expression, including BCL11A, GATA1, KLF1(EKLF), ZFPM1 (FOG-1), and SOX6, that we and others havepreviously described (7, 10) (SI Appendix, Fig. S9). Gene setenrichment analysis (30) was used with gene sets derived fromlineage-specific and differentiation stage-specific expressiondatasets (SI Appendix). A marked increase in the expression ofthe monocyte gene set was notable in the shRNA-treated cellscompared with controls (Fig. 3H) (31). In permissive conditions,a similar type of knockdown can also result in increased pro-duction of megakaryocytes (19). Moreover, when gene sets werecreated from the MYB knockdown expression sets comparedwith controls, the up-regulated genes were found to be signifi-cantly enriched in the later stages of erythroid differentiation,supporting the notion that precocious erythroid differentiationwas occurring in this context (SI Appendix, Fig. S10) (32). To-gether, these findings suggest that MYB is necessary for thenormal differentiation kinetics of adult erythroid cells and re-duction in the level of this gene results in altered erythroid dif-ferentiation kinetics and the increased presence of cells fromother lineages. Consistent with the findings with moderate MYBknockdown by miR-15a/16-1, these shRNAs were able to resultin a marked slowing of cell cycle progression (Fig. 3I) andmarked up-regulation of γ-globin expression in erythroid celllines (SI Appendix, Fig. S11). In support of the slowing of cellcycle progression, we noted that the expression of several cell cy-cle regulators were altered in our expression data from the MYBknockdown cells compared with controls (SI Appendix, Fig. S12).The altered cell cycle and differentiation kinetics of cells withMYB knockdown are consistent with findings in erythroid cellsusing hypomorphic myb alleles in mice (33). Our findings suggestthat MYB plays a critical role coordinating globin gene expres-sion, cell cycle regulation, and erythroid differentiation.Alterations in γ-globin expression can occur in the context of

stress erythropoiesis or other states where erythroid differenti-ation is perturbed. Patients with trisomy 13 and elevated HbFlevels do not have anemia (4, 21), but the evaluation of in vivoerythropoiesis in these patients has not previously been reported.To properly ascertain this, we examined autopsy specimens frompatients with trisomy 13 from the archives ranging over fourdecades at a single institution (SI Appendix). Initially all availabletrisomy 13 cases were selected for evaluation, of which 17 wereused for our final analysis on the basis of confirmation of thediagnosis of trisomy 13 and presence of appropriate histologicalsamples (SI Appendix, Table S4). In all of the samples examined,erythropoiesis appeared appropriate for age, and full maturationof the erythroid lineage could be appreciated (Fig. 4 A, C, andD). This suggests that the elevated HbF levels in trisomy 13 canoccur without grossly perturbed erythropoiesis. In examiningother aspects of hematopoiesis, we found a dramatic increase inmegakaryocyte numbers in over 70% of the cases examined (Fig.4 B and C and SI Appendix, Table S4). In addition, the majorityof the cases (64%) showed abnormal nuclear morphology of themegakaryocytes, suggestive of decreased ploidy in these cells(Fig. 4 B–D and SI Appendix, Fig. S13). These findings werespecific to patients with trisomy 13, as autopsy specimens from thesame institution, in the same time period, and in patients of similarages with other diagnoses lack these findings (SI Appendix). In-

terestingly, MYB knockdown in human cells and hypomorphicalleles in mice show similar phenotypes in megakaryocytes (19,34–36), suggesting that miR-15a/16-1 overexpression and theconsequent reduction in MYB expression may alter other aspectsof hematopoiesis in trisomy 13 patients.

DiscussionWe have demonstrated, using a combination of genetic andfunctional approaches in human cells that the overexpression ofmiR-15a/16-1 results in elevations in HbF gene expression andthis likely explains why patients with a trisomy of chromosome 13have a delayed fetal-to-adult hemoglobin switch and persistenceof fetal hemoglobin. This effect is mediated, at least in part,through down-modulation of the MYB transcription factor,which we have shown is a potent negative regulator of HbF ex-pression and is a direct target of miR-15a and -16-1 (Fig. 5). It ispossible that other genes on chromosome 13 may also contributeto this phenotype, but miR-15a and -16-1 appear to have a major

A

B

C

D

Fig. 4. Pathology from trisomy 13 autopsy cases reveals normal erythro-poiesis and perturbed megakaryopoiesis. (A) Bone marrow (BM) sectionfrom a trisomy 13 case reveals normal erythroid maturation with a normalimmature:mature progenitor ratio. (B) Abnormal and increased numbers ofmegakaryocytes that have hypolobulated nuclei with a “staghorn” ap-pearance are seen in a BM section. (C) Normal erythropoiesis and increasedmegakaryocytes with hypolobulated nuclei seen in a BM section. (D) Normalerythropoietic maturation and a megakaryocyte with a staghorn appear-ance are seen on this BM section. Examples of some dysplastic mega-karyocytes with staghorn and hypolobulated nuclei are highlighted in theimages with cyan arrows. All images are shown at 400× magnification andslides were stained with hematoxylin and eosin.

Midgestation Birth

Fetal

Hemoglobin

MYB

MYB

Euploid Erythroid

Progenitor

Trisomy 13

Progenitor

MicroRNAs

15a and 16-1

Fig. 5. A model demonstrating how elevations of microRNAs 15a and 16-1in trisomy 13 can result in elevated fetal hemoglobin expression. Normally,the basal level of these microRNAs can moderate expression of targets suchas MYB during erythropoiesis. In the case of trisomy 13, elevated levels ofthese microRNAs results in additional down-regulation of MYB expression,which in turn results in a delayed switch from fetal-to-adult hemoglobin andpersistent expression of fetal hemoglobin.

Sankaran et al. PNAS | January 25, 2011 | vol. 108 | no. 4 | 1523

GEN

ETICS

impact. Our study demonstrates a unique and previously un-appreciated pathway that regulates this intensively studied de-velopmental process. The exact relevance of our findings tonormal physiology is not clear, but it appears to be likely thatthese pathways may play an important role in the normal fetal-to-adult hemoglobin switch (Fig. 5). Consistent with this notion,common genetic variants close to the MYB gene appear to beimportant regulators of HbF levels in adult humans (25). Furtherwork will be needed to gain insight into the exact role of thesefactors in the mechanisms mediating human hemoglobinswitching. Nonetheless, it is apparent from our work that miR-15a, -16-1, and MYB could be important therapeutic targets toelevate HbF expression to ameliorate the severity of sickle celldisease and β-thalassemia.Our findings demonstrate the power of using human genetic

approaches to understand developmental processes, where theuse of model organisms can be limited (37). Importantly, ourfindings suggest that alterations of gene dosage at specific ge-netic loci likely underlie the numerous phenotypes observeduniquely with specific aneuploidy syndromes (1). In relativelyfew cases has this been strongly supported by functional evi-dence. Similar types of approaches may help to uncover othergenotype–phenotype correlations in these fascinating syndromes

that can teach us a great deal about normal human developmentand physiology.

Materials and MethodsDetails of the cell culture approaches, constructs used, lentiviral preparation,RNA analysis, flow cytometry, protein methods, and analysis can be found inSI Appendix. CD34+ cells were cultured and differentiated using a two-phaseculture method in serum-free conditions, with appropriate cytokines addedat the various stages of the culture (7). RNA extraction, quantitative RT-PCR,and microarray expression analysis were performed in a manner similar towhat has previously been described (7, 38). Lentivirus production and in-fection was carried out using a modified spin-infection method for theerythroid progenitors that were grown in suspension (39). Flow cytometrywas performed using standard approaches (38) with data analysis occurringin the FlowJo 7.5.5 software suite.

ACKNOWLEDGMENTS. We thank D. Nathan and S. Lux for their inspirationand guidance; C. Epstein, D. Bartel, C. Walkley, C. Sieff, J. Menne, J.Hirschhorn, J. Flygare, M. Bousquet, D. Nguyen, B. Wong, Y. Jeong, H. B.Larman, and G. Bell for valuable advice; T. DiCesare for valuable assistance inproducing the model illustration; and C. Kitidis-Mitrokostas and N. Cohen fortechnical assistance. T.F.M. was supported by the Kay Kendall LeukaemiaFund. D.S. was supported by a Department of Energy Computational ScienceGraduate Fellowship (Grant DE-FG02-97ER25308). This work was supportedby National Institutes of Health Grants R01 DK068348 and P01 HL32262(to H.F.L.).

1. Epstein CJ (1986) The Consequences of Chromosome Imbalance: Principles,Mechanisms, and Models (Cambridge Univ Press, Cambridge, New York).

2. Korbel JO, et al. (2009) The genetic architecture of Down syndrome phenotypesrevealed by high-resolution analysis of human segmental trisomies. Proc Natl Acad SciUSA 106:12031–12036.

3. Pinkerton PH, Cohen MM (1967) Persistence of hemoglobin F in D/D translocationwith trisomy 13-15 (D1). JAMA 200:647–649.

4. Huehns ER, Hecht F, Keil JV, Motulsky AG (1964) Developmental hemoglobinanomalies in a chromosomal triplication: D1 trisomy syndrome. Proc Natl Acad Sci USA51:89–97.

5. Powars D, Rohde R, Graves D (1964) Foetal haemoglobin and neutrophil anomaly inthe D1-trisomy syndrome. Lancet 1:1363–1364.

6. Orkin SH, Higgs DR (2010) Medicine. Sickle cell disease at 100 years. Science 329:291–292.

7. Sankaran VG, et al. (2008) Human fetal hemoglobin expression is regulated by thedevelopmental stage-specific repressor BCL11A. Science 322:1839–1842.

8. McGrath K, Palis J (2008) Ontogeny of erythropoiesis in the mammalian embryo. CurrTop Dev Biol 82:1–22.

9. Peschle C, et al. (1985) Haemoglobin switching in human embryos: Asynchrony ofzeta—alpha and epsilon—gamma-globin switches in primitive and definiteerythropoietic lineage. Nature 313:235–238.

10. Sankaran VG, Xu J, Orkin SH (2010) Advances in the understanding of haemoglobinswitching. Br J Haematol 149:181–194.

11. Yunis JJ (1977) New Chromosomal Syndromes (Academic, New York).12. Parker MJ, Budd JL, Draper ES, Young ID (2003) Trisomy 13 and trisomy 18 in

a defined population: Epidemiological, genetic and prenatal observations. PrenatDiagn 23:856–860.

13. Tharapel SA, Lewandowski RC, Tharapel AT, Wilroy RS, Jr. (1986) Phenotype-karyotype correlation in patients trisomic for various segments of chromosome 13.J Med Genet 23:310–315.

14. Rogers JF (1984) Clinical delineation of proximal and distal partial 13q trisomy. ClinGenet 25:221–229.

15. Su AI, et al. (2004) A gene atlas of the mouse and human protein-encodingtranscriptomes. Proc Natl Acad Sci USA 101:6062–6067.

16. Rosenbloom KR, et al. (2010) ENCODE whole-genome data in the UCSC GenomeBrowser. Nucleic Acids Res 38(Database issue):D620–D625.

17. Chen CZ, Li L, Lodish HF, Bartel DP (2004) MicroRNAs modulate hematopoietic lineagedifferentiation. Science 303:83–86.

18. Zhao G, Yu D, Weiss MJ (2010) MicroRNAs in erythropoiesis. Curr Opin Hematol 17:155–162.

19. Lu J, et al. (2008) MicroRNA-mediated control of cell fate in megakaryocyte-erythrocyte progenitors. Dev Cell 14:843–853.

20. Zhao H, Kalota A, Jin S, Gewirtz AM (2009) The c-myb proto-oncogene and microRNA-15a comprise an active autoregulatory feedback loop in human hematopoietic cells.Blood 113:505–516.

21. Bard H (1972) Postnatal fetal and adult hemoglobin synthesis in D1 trisomy syndrome.

Blood 40:523–527.22. Papayannopoulou T, Torrealba de Ron A, Veith R, Knitter G, Stamatoyannopoulos G

(1984) Arabinosylcytosine induces fetal hemoglobin in baboons by perturbing

erythroid cell differentiation kinetics. Science 224:617–619.23. Letvin NL, et al. (1985) Influence of cell cycle phase-specific agents on simian fetal

hemoglobin synthesis. J Clin Invest 75:1999–2005.24. Friedman RC, Farh KK, Burge CB, Bartel DP (2009) Most mammalian mRNAs are

conserved targets of microRNAs. Genome Res 19:92–105.25. Thein SL, Menzel S, Lathrop M, Garner C (2009) Control of fetal hemoglobin: New

insights emerging from genomics and clinical implications. Hum Mol Genet 18(R2):

R216–R223.26. Jiang J, et al. (2006) cMYB is involved in the regulation of fetal hemoglobin

production in adults. Blood 108:1077–1083.27. Klein U, et al. (2010) The DLEU2/miR-15a/16-1 cluster controls B cell proliferation and

its deletion leads to chronic lymphocytic leukemia. Cancer Cell 17:28–40.28. Chung EY, et al. (2008) c-Myb oncoprotein is an essential target of the dleu2 tumor

suppressor microRNA cluster. Cancer Biol Ther 7:1758–1764.29. Bianchi E, et al. (2010) c-myb supports erythropoiesis through the transactivation of

KLF1 and LMO2 expression. Blood 116:e99–e110.30. Subramanian A, et al. (2005) Gene set enrichment analysis: A knowledge-based

approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA

102:15545–15550.31. Watkins NA, et al. (2009) A HaemAtlas: Characterizing gene expression in

differentiated human blood cells. Blood 113:e1–e9.32. Welch JJ, et al. (2004) Global regulation of erythroid gene expression by transcription

factor GATA-1. Blood 104:3136–3147.33. Vegiopoulos A, García P, Emambokus N, Frampton J (2006) Coordination of

erythropoiesis by the transcription factor c-Myb. Blood 107:4703–4710.34. Kasper LH, et al. (2002) A transcription-factor-binding surface of coactivator p300 is

required for haematopoiesis. Nature 419:738–743.35. Emambokus N, et al. (2003) Progression through key stages of haemopoiesis is

dependent on distinct threshold levels of c-Myb. EMBO J 22:4478–4488.36. Carpinelli MR, et al. (2004) Suppressor screen in Mpl-/- mice: c-Myb mutation causes

supraphysiological production of platelets in the absence of thrombopoietin

signaling. Proc Natl Acad Sci USA 101:6553–6558.37. Sankaran VG, et al. (2009) Developmental and species-divergent globin switching are

driven by BCL11A. Nature 460:1093–1097.38. Sankaran VG, Orkin SH, Walkley CR (2008) Rb intrinsically promotes erythropoiesis by

coupling cell cycle exit with mitochondrial biogenesis. Genes Dev 22:463–475.39. Moffat J, et al. (2006) A lentiviral RNAi library for human and mouse genes applied to

an arrayed viral high-content screen. Cell 124:1283–1298.

1524 | www.pnas.org/cgi/doi/10.1073/pnas.1018384108 Sankaran et al.