Human leukemia mutations corrupt but do not abrogate GATA-2 … · Human leukemia mutations corrupt but do not abrogate GATA-2 function Koichi R. Katsumuraa,b, Charu Mehtaa,b, Kyle

Page 1

Human leukemia mutations corrupt but do notabrogate GATA-2 functionKoichi R. Katsumuraa,b, Charu Mehtaa,b, Kyle J. Hewitta,b, Alexandra A. Soukupa,b, Isabela Fraga de Andradea,b,Erik A. Ranheimc, Kirby D. Johnsona,b, and Emery H. Bresnicka,b,1

aUniversity of Wisconsin–Madison Blood Research Program, Department of Cell and Regenerative Biology, Wisconsin Institutes for Medical Research,University of Wisconsin School of Medicine and Public Health, Madison, WI 53705; bUniversity of Wisconsin Carbone Cancer Center, University of WisconsinSchool of Medicine and Public Health, Madison, WI 53705; and cDepartment of Pathology and Laboratory Medicine, University of Wisconsin School ofMedicine and Public Health, Madison, WI 53705

Edited by Stuart H. Orkin, Children’s Hospital and the Dana Farber Cancer Institute, Howard Hughes Medical Institute and Harvard Medical School, Boston,MA, and approved September 4, 2018 (received for review July 31, 2018)

By inducing the generation and function of hematopoietic stemand progenitor cells, the master regulator of hematopoiesis GATA-2 controls the production of all blood cell types. HeterozygousGATA2 mutations cause immunodeficiency, myelodysplastic syn-drome, and acute myeloid leukemia. GATA2 disease mutationscommonly disrupt amino acid residues that mediate DNA bindingor cis-elements within a vital GATA2 intronic enhancer, suggestinga haploinsufficiency mechanism of pathogenesis. Mutations alsooccur in GATA2 coding regions distinct from the DNA-bindingcarboxyl-terminal zinc finger (C-finger), including the amino-terminal zinc finger (N-finger), and N-finger function is not estab-lished. Whether distinct mutations differentially impact GATA-2mechanisms is unknown. Here, we demonstrate that N-finger mu-tations decreased GATA-2 chromatin occupancy and attenuated tar-get gene regulation. We developed a genetic complementationassay to quantify GATA-2 function in myeloid progenitor cells fromGata2 −77 enhancer-mutant mice. GATA-2 complementation in-creased erythroid and myeloid differentiation. While GATA-2disease mutants were not competent to induce erythroid differ-entiation of Lin−Kit+ myeloid progenitors, unexpectedly, theypromoted myeloid differentiation and proliferation. As themyelopoiesis-promoting activity of GATA-2 mutants exceededthat of GATA-2, GATA2 disease mutations are not strictly inhib-itory. Thus, we propose that the haploinsufficiency paradigmdoes not fully explain GATA-2–linked pathogenesis, and an amal-gamation of qualitative and quantitative defects instigated byGATA2 mutations underlies the complex phenotypes of GATA-2–dependent pathologies.

GATA-2 | hematopoiesis | MDS | AML | leukemia

Mechanisms underlying the heterogeneous malignancy acutemyeloid leukemia (AML) are incompletely understood,

and there is a vital need to develop efficacious therapies (1).Although major progress has been made in developing molecu-larly targeted and transplant therapies, the 5-y survival of geri-atric and pediatric AML patients remains at 10–20% and 60–70%, respectively (2). Elucidating how myeloid cell geneticnetworks are corrupted may unveil opportunities for AML bio-marker and therapeutics development. Rigorous studies havedefined AML genetic and epigenetic landscapes and the vexingclonal evolution during disease progression (3–9). Germlinemutations that predispose to myelodysplastic syndrome (MDS)and AML, such as those disrupting GATA-2 expression andfunction (10–12), have the potential to reveal clues regardingmechanisms governing disease initiation and progression.GATA-2 is essential for multilineage hematopoiesis (13),

triggers hemogenic endothelium to produce hematopoietic stemcells (HSCs) (14, 15), regulates HSC activity (16–18), and stim-ulates myelo-erythroid progenitor cell differentiation, pro-liferation, and survival (19–21). Gata2-null mice exhibit impairedmultilineage hematopoiesis and die at ∼embryonic day (E) 10.5(13). Additional instructive mouse models for analyzing GATA-

2 function include the embryonic-lethal Gata2 intronic (+9.5)enhancer mutant with defective HSC genesis (18) and erythroidprecursor function (19) and distal (−77) (20) enhancer mutantwith defective myelo-erythroid progenitor differentiation. Theresults with these models, and the finding that GATA-2 overexpressionin bone marrow suppresses hematopoiesis (22), indicate thatGATA-2 levels/activity must be constrained within a physiologicalwindow.In accord with critical GATA-2 functions discovered in mice,

heterozygous human GATA2 mutations are pathogenic andcause immunodeficiency that often progresses to MDS and AML(23, 24). GATA2 mutations also cause other AML-linked fa-milial diseases, and GATA2 is mutated frequently in high-riskMDS (25).GATA2mutations often occur in the DNA binding C-finger and inhibit DNA binding (26). GATA2 +9.5 enhancermutations decrease GATA-2 expression (18, 27). In 3q21q26AML, the −77 enhancer is repositioned next to MECOM encodingthe EVI1 oncogene (28, 29). Decreased GATA2, concomitant withelevated EVI1, underlies this malignancy. In addition, GATA-2 overexpression in AML can predict poor prognosis (30). Inaggregate, mouse and human data emphasize the need to avertdeclines and increases in GATA-2, both being pathogenic.GATA-2 establishes and maintains cell-type–specific genetic

networks, and heterozygous GATA2 mutations that reduce GATA-2 levels/activity may differentially affect network integrity in distinct

Significance

GATA-2 functions in stem and progenitor cells to control bloodcell development, and its mutations cause blood diseases (im-munodeficiency, myelodysplasia, and myeloid leukemia). HowGATA-2 mutations cause these diseases is unclear. We innovateda genetic complementation assay to analyze functional ramifi-cations of GATA-2 disease mutations. The activities of GATA-2 and mutants were quantified in blood progenitor cells frommice engineered to express a low level of GATA-2 due to de-letion of an essential Gata2 enhancer. Unexpectedly, the mu-tants were not only competent to induce myeloid cells, but theiractivities exceeded that of GATA-2. These results transform thecurrent paradigm that disease mutations are solely inhibitory,and ectopically low GATA-2 levels/activity constitute the diseasemechanism.

Author contributions: K.R.K. and E.H.B. designed research; K.R.K., C.M., K.J.H., A.A.S., I.F.d.A.,and K.D.J. performed research; K.D.J. contributed new reagents/analytic tools; K.R.K., E.A.R.,and E.H.B. analyzed data; and K.R.K. and E.H.B. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1813015115/-/DCSupplemental.

Published online October 9, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1813015115 PNAS | vol. 115 | no. 43 | E10109–E10118

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 2

contexts (26). Inadequate or excessive target gene activity wouldboth corrupt networks. Oncogenic Ras-dependent, multisite GATA-2 phosphorylation, coupled with GATA-2–dependent positive auto-regulation of GATA2 transcription, can elevate GATA-2 levels/activity and therefore disrupt physiological GATA-2 function (31,32). Because GATA-2 stimulates AML cell proliferation and sur-vival in vitro (32), elevating or reducing GATA-2 may instigate orcontribute to leukemogenesis.GATA factor C-fingers mediate DNA binding (33, 34), and C-

finger mutations impair GATA-2 function (23, 24). Although N-finger function remains enigmatic, N-finger mutations occur inpatients with erythroleukemia (35) and AML with biallelic muta-tion of CEBPA (36, 37). The N-finger was reported to bind DNAwith sequence-specificity in vitro (38, 39). The GATA-1 N-fingerbinds the critical coregulator FOG-1 (40). This interaction is me-diated by GATA-1 V205 (40), and V205 mutation disrupts ery-throid maturation in mice and generates familial dyserythropoieticanemia in humans (41). Although the GATA-1 and GATA-2 N-fingers are well conserved, GATA-2–expressing hematopoieticstem and progenitor cells (HSPCs) do not express FOG-1.R216 mutations in patients with X-linked gray platelet syndrome(42) attenuate GATA-1 function without influencing FOG-1 binding (43). This mutation reduces binding to sites contain-ing single or palindromic GATA motifs. Analogous GATA-2(R307W) and GATA-3 (R276) residues can be mutated in leu-kemia patients. Because the GATA-2 N-finger can be mutated inleukemia, and unlike GATA-1, FOG-1 is not expressed in theGATA-2–expressing cells, dissecting molecular consequences ofthese mutations has the potential to inform GATA factor mech-anisms and pathologies. Herein, we analyzed the mechanisticramifications of N-finger disease mutations in diverse systems,including a genetic complementation assay to quantify GATA-2function in Gata2 −77 enhancer mutant primary myelo-erythroidprogenitor cells. Our discovery that GATA-2 disease mutationsunexpectedly enhance select GATA-2 functions in primary cellsdemands a reconsideration of the paradigm that inhibitory diseasemutations strictly decrease GATA-2 levels/activity. These resultsprovide a perspective into the haploinsufficiency model of GATA-2-linked pathologies.

ResultsGATA-2 N-Finger Increases GATA-2 Endogenous Target Gene ChromatinOccupancy and Activation. Because human disease mutations canunveil unique mechanistic insights, we analyzed the functionalconsequences of GATA-2 N-finger mutations (R293Q, P304H,R307W, A318T, and R330Q) (Fig. 1A) from AML patients in amouse aortic endothelial (MAE) cell assay (31, 32) in which ex-ogenous GATA-2 expression activates endogenous target genes.Increasing levels of murine GATA-2 (98% sequence identity tohuman) or mutants were expressed in MAE cells, ensuring thatmutants were analyzed at levels resembling that of GATA-2 (Fig.1B). While GATA-2 activated the target genes Hdc, Gfi1, andGrb10 (14, 19, 20, 31), N-finger leukemia mutations attenuatedthe transcriptional response (Fig. 1C).To elucidate the mechanism underlying the compromised ac-

tivity of N-finger mutants, we tested whether the subcellular lo-calization of the mutants resembles or differs from that ofGATA-2. Immunofluorescence analysis indicated that GATA-2 and the N-finger mutants were exclusively nuclear-localized inMAE cells (Fig. 1D). Using a quantitative chromatin immuno-precipitation (ChIP) assay with anti-HA antibody, we comparedGATA-2 and R307WN-finger mutant activities to occupy chromatin.R307W was analyzed, because this mutation strongly inhibited targetgene induction. The R307W mutation decreased, but did not abro-gate, GATA-2 occupancy at the Gata2 +9.5 enhancer (18, 44), Hdc+3.7 (31), and Lyl1 promoter (45) loci (Fig. 1E). GATA-2 did notoccupy Gata2 −18.3, Hdc +11.2, Lyl1 −1.3, and Necdin promotersites, providing evidence for specificity of chromatin occupancy.

These results indicate that GATA-2 N-finger integrity promoteschromatin occupancy at target genes without grossly changingnuclear localization.

Structure-Based Design of GATA-2 Mutants to Dissect GATA-2Leukemia Mutant Mechanisms. GATA factor N-fingers are highlyconserved (Fig. 2A), implying important functions. To assesswhether the inhibitory consequences of N-finger leukemia mu-tants are recapitulated by a mutation that disrupts zinc fingerstructure, C295A was engineered into GATA-2 (Fig. 2A). Underconditions in which GATA-2, C295A, and R307W were expressedat comparable levels (Fig. 2B), both C295A and R307W signifi-cantly decreased induction of GATA-2 target gene expression inMAE cells (Fig. 2C). Similar to the N-finger disease mutants,C295A was exclusively nuclear-localized (Fig. 2D).Previously, we demonstrated that Ras signaling induces multi-

site GATA-2 phosphorylation and increases GATA-2–dependenttranscriptional activation (31, 32). We tested whether N-fingerleukemia mutants are competent to respond to Ras signaling.Expression of Ras(G12V) shifted GATA-2 migration to a slow-mobility species (2.2-fold increase), which we demonstrated pre-viously to be a hyperphosphorylated isoform (Fig. 2E). Ras(G12V) also induced a mobility shift with R307W (2.4-fold in-crease) and increased its capacity to activate Hdc expressionsevenfold, comparable to the fold-induction achieved with Ras(G12V)/GATA-2 (Fig. 2 E and F). Thus, although R307W-mediated transcriptional activation was strongly compromised, itremained competent to respond to Ras(G12V).Because the C-finger mediates DNA binding (33), and N-

finger function in vivo is unresolved, it was instructive to com-pare the consequences of cysteine mutations in the N- and C-fingers. We engineered C349A to disrupt C-finger structure (Fig.2A) and analyzed its activity in MAE cells. The predominantC349A isoform exhibited a slow mobility, even without Ras(G12V) expression, and the intensity of this isoform was 2.3-foldgreater than that of GATA-2 (Fig. 2G). This is consistent withour report that the T354M disease mutant linked to MDS/AML(11, 12) exhibits reduced chromatin binding and is predomi-nantly hyperphosphorylated without Ras(G12V) (31); however,multisite phosphorylated T354M does not increase target geneexpression, based on impaired chromatin binding. Ras(G12V)also induced the slow mobility C295A isoform 1.7-fold (Fig. 2G).Whereas Ras(G12V) increased GATA-2– and C295A-dependentHdc induction fivefold, C349A was inactive in the absence orpresence of Ras(G12V) (Fig. 2H), consistent with the disrupted C-finger that would not be competent for chromatin binding.GATA-1 V205 binds FOG-1, which mediates transcriptional

activation and repression of many GATA-1 target genes (40, 46,47). FOG-1 is not expressed in GATA-2–expressing cells (e.g.,MAE cells) and does not mediate GATA-2 biological activity inHSPCs. In principle, the GATA-2 residue equivalent to GATA-1V205 might mediate transcriptional regulation through a relatedor novel mechanism. We tested this model by engineering mu-tations at GATA-2 V296, equivalent to GATA-1 V205, andanalyzing its activity in MAE cells at expression levels resemblingGATA-2 (Fig. 2I). V296G and V296M retained the capacity toinduce Hdc (Fig. 2J). Thus, GATA-2 N-finger–dependent tran-scriptional activation is impaired by leukemia mutations, requiresN-finger structure, and has distinct molecular determinants fromthe GATA-1 N-finger.

GATA-2 Genetic Complementation Assay with Gata2 −77 EnhancerMutant Progenitor Cells: GATA-2 Leukemia Mutants Promote Myelopoiesis.To elucidate the functional consequences of GATA-2 mutations,we compared the activity of expressed wild-type and mutantGATA-2 in primary mouse bone marrow and fetal liver HSPCs.GATA-2 or R307W expression in −77+/+ bone marrow cellsdecreased CFU-GM (SI Appendix, Fig. S1), consistent with the

E10110 | www.pnas.org/cgi/doi/10.1073/pnas.1813015115 Katsumura et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 3

report that GATA-2 overexpression suppresses bone marrowHSPCs (22). We also expressed wild-type or mutant GATA-2 inmouse fetal liver lineage-negative (Lin−) hematopoietic precur-sors (Fig. 3 A and B). Although we predicted that R307W andthe T354M C-finger leukemia mutant would be inactive or haveattenuated activity, unexpectedly, they increased CFU-GMin −77+/+ Lin− fetal liver cells (Fig. 3B, Right).Because human GATA-2 disease mutations are heterozygous

(23, 24), expressing GATA-2 mutants in wild-type (−77+/+) he-matopoietic progenitors is not optimal for elucidating physio-logical or pathological mechanisms. We therefore devised agenetic complementation assay using −77−/− fetal liver myelo-erythroid progenitor cells (20), in which endogenous GATA-2 expression is reduced. It was not possible to use −77−/− bonemarrow, because this homozygous mutation is embryonic-lethal(20). Genetic complementation analysis in mutant cells is a pow-erful strategy to dissect mechanisms, as exemplified by studies withGATA-1 (48–51). While the MAE system enables quantification ofGATA-2 activity to regulate endogenous target genes (31), noGATA-2 genetic complementation systems have been described.Using Lin− myelo-erythroid progenitor cells, we compared GATA-2and mutant activities to induce myeloid and erythroid differentiation

of −77−/− cells. The −77 enhancer deletion decreased Gata2 ex-pression by 69% (Fig. 3C).As described previously (20), −77−/− Lin− myelo-erythroid

progenitor cells have little to no capacity to generate erythroidcolonies (CFU-E and BFU-E) (Fig. 3D). Wild-type progenitorsformed very few myeloid colonies (CFU-GM) (Fig. 3D), whichreflected the 1-d culture with erythroid factors postinfection anddiffered from our report in which wild-type progenitors wereanalyzed without culturing generate large numbers of CFU-GM(20). HA–GATA-2 expression rescued CFU-E and BFU-E, andincreased CFU-GM (Fig. 3D). Comparison of GATA-2, R307W,and T354M activities revealed that R307W, but not T354M,resembled GATA-2 in rescuing BFU-E at similar expressionlevels (Fig. 3D). Unexpectedly, this analysis revealed thatR307W and T354M increased CFU-GM 7- and 2.5-fold greaterthan GATA-2 (Fig. 3D). To test whether rescue involved GATA-2 expression at levels resembling endogenous GATA-2 or over-expression, we titrated the GATA-2–expressing retrovirus andanalyzed HA–GATA-2 and endogenous GATA-2 expression si-multaneously with anti–GATA-2 antibody. HA–GATA-2 levelsthat rescued CFU-E and BFU-E and elevated CFU-GM werecomparable to that of endogenous GATA-2 (Fig. 3E). Furthermore,

** ****

** ****

******

** ********

****

***

50

GATA-2 R293Q P304H R307W A318T R330Q

0

100

150Hdc

Cm

RN

A L

evel

s (R

elat

ive

Un

its)

**

**

* *

0

3

6

9

Grb10

0

5

10

15

Gfi1

A

B

***** ** ***

**

** *

GATA-2R307W

Tubulin

GATA-250

50

Mr

x 10

-3

HA

-GA

TA

-2 O

ccu

pan

cy(R

elat

ive

Un

its)

0

0.015

0.030

0.045

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

PI anti-HAE

Gata2 (+9.5) Hdc (+3.7) Lyl1 Pro Necdin ProGata2 (-18.3) Hdc (+11.2) Lyl1 (-1.3)

* *

* **

*

GATA-2 R307W A318T R330Q

50

50 Tubulin

GATA-2

Mr

x 10

-3

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

D

HA

DAPI

Merge

Empty GATA-2 R293Q P304H R307W A318T R330Q

GATA-2 R293Q P304H

Mr

x 10

-3

50

50

GATA-2

Tubulin

R293

V296

P304

R307

A318

R330

Em

pty

N-ZnF1 480

C-ZnF

Em

pty

Em

pty

Empty

Fig. 1. GATA-2 N-finger leukemia mutations attenuate chromatin occupancy and target gene activation. (A) GATA-2 structural model based on humanGATA-3 crystal structure. (B) Representative Western blot analysis with anti-HA antibody of wild-type and mutant proteins transiently expressed in MAE cells.(C) qRT-PCR analysis of mRNA levels of GATA-2 target genes in MAE cells transiently expressing HA–GATA-2 and mutant proteins (n = 6). (D) Immunoflu-orescence analysis with anti-HA antibody in MAE cells expressing HA–GATA-2 and mutant proteins. (Scale bars, 10 μm.) (E) Quantitative ChIP analysis of HA–GATA-2 in MAE cells transiently expressing HA–GATA-2 or HA-R307W (n = 5). *P < 0.05, **P < 0.01, ***P < 0.001.

Katsumura et al. PNAS | vol. 115 | no. 43 | E10111

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 4

the level of R307W that strongly induced CFU-GMwas comparableto that of endogenous GATA-2 (Fig. 3F). These results highlightthe unique utility of the genetic complementation assay for quali-tatively and quantitatively analyzing GATA-2–dependent pro-genitor activity. Importantly, the assay was conducted underconditions in which ectopically introduced wild-type or mutantGATA-2 was expressed at physiological GATA-2 levels.Because R307W and T354M increased the percentage of

CFU-GM among the colonies from 37% to 86% (R307W, P =0.0014) or 92% (T354M, P = 0.0009), we analyzed colony cel-lularity. Granulocytes were more abundant in colonies derivedfrom R307W- or T354M-expressing progenitors, in comparisonwith colonies derived from GATA-2–expressing progenitors

(Fig. 3 G and H). Flow cytometric analysis revealed that R307Wor T354M increased Mac1+Gr1+ myeloid cells (Fig. 3I).By eliminating the culture step with Epo and stem cell factor,

which favors erythroid precursors at the expense of myelo-erythroid progenitors, we improved the genetic complementa-tion assay for analyzing both erythroid and myeloid differentia-tion. We also utilized a more refined myelo-erythroid progenitorpopulation, FACS-purified Lin−Kit+ myelo-erythroid progeni-tors, rather than the bead-sorted Lin− population. CFU activitywas assayed immediately after Lin−Kit+ cell isolation to mini-mize the probability of cellularity transitions ex vivo (Fig. 4A).Gata2 expression was threefold lower in −77−/− progenitors incomparison with wild-type progenitors (Fig. 4B). Resembling

C295A

GATA-2

Tubulin

R307W

36

24

12

0

50

50

Hdc

mR

NA

Lev

els

(Rel

ativ

e U

nit

s)

Gfi115

10

5

0

* * * *

** ** ***

*** ******

BA

D

100

200

300

0

Ras(G12V)

GATA-2C295AC349A

G R ECV

N

GATATP

LW R R

DGTGHY

LC C

NACG L Y

Zn++

AG T C

CA

N

QTTTTT

LW R R

NANGDP

VC C

NACG L Y

Zn++

A

C295A C349A

Tubulin50

Ras(G12V)

GATA-2C295AC349A

Mr

x 10

-3G

50

50

GATA-2

Tubulin

GATA-2 V296G V296M C295A

70

140

210

0GATA-2 V296G V296M C295A

I

HA DAPI Merge

Empty

GATA-2

C295A

LPLPPCEARECVNCGATATPLWRRDRTGHYLCNACGLYHKMNGKARSCSEGRECVNCGATATPLWRRDGTGHYLCNACGLYHKMNG

V205

V296

GATA-2GATA-1N-Finger

15Ras(G12V)

C

H

J

Em

pty

Em

pty

R307WGATA-2

C295AGATA-2

* *

E

Ras(G12V)

GATA-2R307W

Tubulin

50

50

0.5

0

1.0

1.5

Hd

c m

RN

A L

evel

s(R

elat

ive

Un

its)

Ras(G12V)

GATA-2R307W

15 Ras(G12V)

Mr

x 10

-3

F **

**

****

***

**

Empty

Empty

Mr

x 10

-3

GATA-2

50GATA-2 (dark)

50

Hyper

Hyper

Phos

PhosHypo

Hypo

Hd

c m

RN

A L

evel

s(R

elat

ive

Un

its)

Hd

c m

RN

A L

evel

s(R

elat

ive

Un

its)

Mr

x 10

-3

GATA-2HyperPhosHypo

Fig. 2. Structural determinants of GATA-2 function. (A, Upper) Schematic representation of C295A and C349A mutants. (Lower) Murine GATA-1 and GATA-2N-finger amino acid sequences. (B) Representative Western blot analysis with anti-HA antibody of MAE cells transiently expressing HA–GATA-2 ormutant proteins. (C) qRT-PCR analysis of mRNA levels of GATA-2 target genes in MAE cells transiently expressing HA–GATA-2 or mutant proteins (n = 4).(D) Immunofluorescence analysis with anti-HA antibody in MAE cells expressing HA–GATA-2 or the C295A mutant. (Scale bars, 10 μm.) (E ) RepresentativeWestern blot analysis with anti-HA antibody of HA–GATA-2 and HA–R307W transiently expressed in MAE cells with or without Ras(G12V). (F) qRT-PCRanalysis of mRNA levels of Hdc genes in MAE cells transiently expressing HA–GATA-2 and R307W with or without Ras(G12V) (n = 5). (G) RepresentativeWestern blot analysis with anti-HA antibody of HA–GATA-2 or mutant proteins transiently expressed in MAE cells with or without Ras(G12V). This Western blotutilized a large SDS/PAGE gel to achieve maximal isoform separation. (H) qRT-PCR analysis of Hdc mRNA levels in MAE cells transiently expressing HA–GATA-2 ormutant proteins with or without Ras(G12V) (n = 3). (I) Representative Western blot analysis with anti-HA antibody of MAE cells expressing HA–GATA-2 or mutant proteins. (J) qRT-PCR analysis of Hdc mRNA levels in MAE cells transiently expressing HA–GATA-2 or mutant proteins (n = 5). *P < 0.05,**P < 0.01, ***P < 0.001.

E10112 | www.pnas.org/cgi/doi/10.1073/pnas.1813015115 Katsumura et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 5

Lin− cells, −77−/− Lin−Kit+ cells produced almost no colonies(Fig. 4C). GATA-2 expression elevated BFU-E and CFU-GM6.4- and 2.6-fold, respectively (Fig. 4D). While R307W did notrescue BFU-E, it increased CFU-GM 3.9-fold relative to theGATA-2 control (Fig. 4D). T354M also increased CFU-GM butnot BFU-E (Fig. 4E). Morphological analysis revealed only rare

erythroid cells in colonies from R307W-expressing −77−/− mye-loid progenitors in comparison with GATA-2–expressing pro-genitors (Fig. 4 F and G). Macrophages were abundant in −77−/−cells infected with empty vector, as described (20). Thus, R307Wpreferentially induces granulocytes. The distinct morphology ofthe cells derived from these −77−/− Lin−Kit+ progenitors,

Colonyassay

Fetal liver(E14.5)

-77+/+ -77-/-Lin- cellisolation

Infection

BFU-E CFU-GM

50

50

HA-GATA-2

Tubulin

EmptyGATA-2

R307WT354M

-77-/-

A

Empty

GATA-2

R307W

T354M

Empty

Empty

GATA-2

R307W

T354M

Empty

0

4

12

16

0

50

75

100

D

****

8

****

25

E

B

Gat

a2 m

RN

A L

evel

s(R

elat

ive

Un

its)

0

0.5

1.0

1.5

-77+/+ -77-/-

-77-/--77+/+

**

-77+/+-77-/-

GATA-2

Tubulin

GATA-2

HA-GATA-2

0

200

600

800

400

0

5

15

20

10

0

2.5

7.5

10

5.0

BFU-E CFU-GMCFU-E

GATA-2 GATA-2 GATA-2

Culture withEpo and SCF

50

50

Mr

x 10

-3

GATA-2

Tubulin

EmptyGATA-2

R307WT354M

-77+/+ C Lin- Cells

Co

lon

ies

per

1000

Lin

- C

ells

Co

lon

ies

per

1000

Lin

- C

ells

Empty GATA-2 R307W T354M-77-/--77+/+

EmptyG

Enucleated Macrophage Granulocyte

Per

cen

t o

f C

ells

25

50

75H

2.61

8.97

87.5

4.81

25.8

65.1

26.1

48.0

15.3

20.3

38.0

35.6

I

-77-/-

Empty GATA-2 R307W T354M

Gr1

Mac1

Empty

GATA-2

R307WT354M

Empty0

Empty

GATA-2

R307WT354M

EmptyEmpty

GATA-2

R307WT354M

EmptyEmpty

GATA-2

R307WT354M

EmptyEmpty

GATA-2

R307WT354M

EmptyEmpty

GATA-2

R307WT354M

Empty

**

**

*********

Mac

1+ Gr1

+ Cel

ls (

%)

15

30

45

0

Empty

GATA-2

R307WT354M

****** **

Ery

Ery

EryEry

Neu

Neu

Neu NeuNeuNeu

ProNeu

Proerythroblast/Promyeloblast

Basophilicerythroblast

Orthochromaticerythroblast

0

600

900

300

CFU-E

Empty

GATA-2

R307W

T354M

Empty

***

*****

***

*

* *

**

*

*

***

***

** **

-77-/--77+/+

F

GATA-2 R307W

BFU-E CFU-GM

GATA-2 R307W

0

4

8

10

6

2

0

40

60

20

-77+/+ -77-/-

GATA-2 R307W

GATA-2

Tubulin

HA-GATA-2

Co

lon

ies

per

1000

Lin

- C

ells

***

***

***

***

**

** **

****

**

**

Co

lon

ies

per

1000

Lin

- C

ells

BFU-E CFU-GM

EmptyGATA-2

R307WT354M

0

6

12

18

0

7

14

21

EmptyGATA-2

R307WT354M

**

**

Empty

Empty

0103

104

105

0 103 104 105

0103

104

105

0 103 104 105

0103

104

105

0 103 104 105 0 103 104 105

0103

104

105

Mr

x 10

-3

-77-/--77+/+50

50Mr

x 10

-350

50Mr

x 10

-3

-77-/--77+/+

Empty

Empty

Empty

Empty

Empty

Empty

Empty

Fig. 3. GATA-2 leukemia mutants increase myeloid progenitor cell activity in a primary cell genetic complementation assay. (A) Schematic representation ofgenetic complementation assay. (B, Left) Representative Western blot analysis of −77+/+ fetal liver cells expressing HA–GATA-2 with anti-HA antibody. (Right)Quantitative analysis of CFU activity of −77+/+ fetal liver cells (n = 3). (C) qRT-PCR analysis of Gata2 mRNA levels in Lin− cells from wild-type or −77−/− fetallivers (n = 5). (D, Upper) Representative Western blot analysis with anti-HA antibody of −77−/− fetal liver cells expressing HA–GATA-2 or mutants. (Lower)Quantitative analysis of CFU activity of −77−/− fetal liver cells (n = 7). −77+/+ fetal liver cells infected with control vector were used as control. (E, Upper)Representative Western blot analysis with anti–GATA-2 antibody (recognizes both HA–GATA-2 and endogenous GATA-2) of −77−/− fetal liver cells expressingHA–GATA-2. (Lower) Quantitative analysis of CFU activity of −77−/− fetal liver cells (n = 3). −77+/+ fetal liver cells infected with control vector were used ascontrol. (F, Left) Representative Western blot analysis with anti-GATA-2 antibody of −77−/− fetal liver cells expressing HA–GATA-2 or R307W. (Right)Quantitative analysis of CFU activity of −77−/− fetal liver cells (n = 4). −77+/+ fetal liver cells infected with control vector were used as control. Significancerelative to −77−/− cells infected with empty vector was evaluated. (G) Representative image of Giemsa-stained cells from colonies. (Scale bars, 20 μm.) Ery,erythroblasts; Mac, macrophage; Neu, neutrophil; Pro, proerythroblast/promyeloblast. (H) Quantification of Giemsa stain (n = 3). (I) Flow cytometric analysisof cells isolated from colonies and stained with Gr1 and Mac1 (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001.

Katsumura et al. PNAS | vol. 115 | no. 43 | E10113

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 6

in comparison with the −77−/− Lin− cells (Fig. 3) reflected theassay modification in which the progenitors analyzed in Fig. 4were immediately subjected to colony assay, rather than cul-tured for 1-d postinfection. Under both conditions, however,R307W induced granulocytes.Because R307W and T354M chromatin binding is reduced,

based on ChIP analysis at select target genes, and they increasedCFU-GM, we asked whether GATA-2 DNA binding activity is

inconsequential for increasing CFU-GM. We tested C349A, inwhich the cysteine mutation destroys the structural integrity ofthe DNA-binding C-finger. C349A had little to no activity in thegenetic complementation assay, using CFU as a read-out (SIAppendix, Fig. S2 A and B), suggesting that C-finger structureand DNA binding competence is required to induce CFU-GM.To assess whether additional GATA-2 leukemia mutants in-duce CFU-GM, we analyzed GATA-2 A318T, which has been

G

H

F -77+/+

Empty Empty GATA-2 R307W-77-/-

Ery

Ery ProNeu

Mac

Mac

MacEry

Neu

Neu Neu

NeuNeu

Per

cen

t o

f C

ells

0

60

90

30

Empty

GATA-2

R307WEmpty

Proerythroblast/Promyeloblast

Basophilicerythroblast

Orthochromaticerythroblast

Enucleated Macrophage Granulocyte

*

*

** *** *

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WEmpty

-77-/--77+/+

mR

NA

Lev

els

(Rel

ativ

e U

nit

s)

0

1.0

1.5

0.5

Gata2

Empty

GATA-2

R307W

Empty

0

3.0

4.5

1.5

Mpo

0

1.4

2.1

0.7

Alas2** **

*****

0

1.6

2.4

0.8

Slc4a1

** **

0

2

3

1

Epb4.9

* *

Empty

GATA-2

R307W

Empty

Empty

GATA-2

R307W

Empty

Empty

GATA-2

R307W

Empty

0

1.6

2.4

0.8

Ctsg

* *

0

1.0

1.5

0.5*

*Myb

Tal1

0

3.0

4.5

1.5*

0

2

3

1

Lyl1

0

1.0

1.5

0.5

**

**Elane

0

1.2

1.8

0.6

Gata2(HA)

Empty

GATA-2

R307W

Empty

0

5.0

7.5

2.5

Kit

** *

Empty

GATA-2

R307W

Empty

-77-/--77+/+

ED

Gat

a2 m

RN

A L

evel

s(R

elat

ive

Un

its)

0

0.8

1.2

0.4

A

-77-/-

**

-77-/- Lin-Kit+ Cells

-77+/+

B

6

4

2

0

75

50

25

0

BFU-E CFU-GM

*** ******

****

**

Co

lon

ies

per

1000

Lin

- Kit

+ C

ells

C

Colony assay

Fetal liver(E14.5)

Lin-Kit+ cellisolation

Infection

Lin-Kit+ Cells

-77-/-

0

8

12

4

0

60

90

30

BFU-E CFU-GM

***

*** ***

Empty GATA-2 R307W

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WEmpty

Empty

GATA-2

R307WT354M

-77-/--77+/+

Co

lon

ies

per

1000

Lin

- Kit

+ C

ells

Empty

GATA-2

R307WT354M

Fig. 4. GATA-2 leukemia mutants stimulate myelopoiesis ex vivo. (A) Schematic representation of rescue assay with Lin−Kit+ cells. (B) qRT-PCR analysis ofGata2mRNA levels in Lin−Kit+ cells from wild-type or −77−/− fetal livers (n = 3). (C) Representative image of the dishes subjected to CFU assay. (D) Quantitativeanalysis of CFU activity of −77−/− myeloid progenitor cells. −77+/+ fetal liver cells infected with control vector were used as control (n = 8). (E) Quantitativeanalysis of CFU activity of −77−/− Lin−Kit+ cells (n = 3). (F) Representative image of Giemsa-stained cells from colonies derived from Lin−Kit+ cells. (Scale bars,20 μm.) Ery, erythroblasts; Mac, macrophage; Neu, neutrophil; Pro, proerythroblast/promyeloblast. (G) Quantification of Giemsa stain (n = 4). (H) qRT-PCRanalysis of mRNA levels in cells isolated from colonies (n = 8). *P < 0.05, **P < 0.01, ***P < 0.001.

E10114 | www.pnas.org/cgi/doi/10.1073/pnas.1813015115 Katsumura et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 7

detected more frequently than R307W. A318T induced CFU-GM in Lin− cells and Lin−Kit+ cells (SI Appendix, Fig. S2C–E), and granulocytes were more abundant in colonies from−77 Lin− cells and −77 Lin−Kit+ cells (SI Appendix, Fig. S2 F andG). Thus, all GATA-2 leukemia mutants tested gain a functionto increase CFU-GM, differing from GATA-2 and the inactiveC349A mutant.The gain-of-function myelopoiesis-stimulating activity of

GATA-2 leukemia mutants was surprising, given the paradigmthat mutations create insufficient GATA-2 activities/levels tocontrol stem and progenitor cell transitions and function. Incertain cases, cis-element mutations in MDS/AML decreaseGATA-2 expression, consistent with haploinsufficiency. How-ever, because elevated GATA-2 would deregulate target genes,which include multiple disease-linked genes (14, 19, 20, 32), andcan correlate with poor prognosis of AML (30), ectopically lowor high GATA-2 levels/activity will disrupt the integrity of ge-netic networks that ensure normal hematopoiesis.Gene-expression analysis in R307W-expressing cells isolated

from colonies revealed elevated myeloid gene (Mpo, Ctsg, andElane) and reduced erythroid gene (Slc4a1, Epb4.9, and Alas2)expression (Fig. 4H). Genes expressed in HSPCs and erythroidcells (e.g., Myb, Kit, and Tal1) were reduced in −77−/− cells, andGATA-2 rescued expression. While R307W did not affect Tal1expression, it was more effective than GATA-2 in elevating Mybexpression (Fig. 4H). These results reinforce the conclusion from

CFU analysis that the leukemia mutants are more effective thanGATA-2 in inducing CFU-GM.

GATA-2 Leukemia Mutant Stimulates Myelo-Erythroid Progenitor CellCycle Progression. Consistent with the increased colony number,GATA-2 or R307W expression significantly increased the num-ber of cells within colonies. R307W increased cell numberstwofold greater than GATA-2 (Fig. 5A). Because GATA-2 (21,52) and select target genes regulate cell cycle progression, andR307W has strong activity to induce granulocytes, we comparedGATA-2 and R307W activities to impact cell cycle progressionin the −77−/− Lin−Kit+ cell genetic complementation assay.GATA-2 or R307W expression significantly increased S and G2/Mphase cells in the Mac1+Gr1+ myeloid cell population (Fig. 5B and C). A greater percentage of R307W-expressing cells residedin S phase in comparison with GATA-2–expressing cells. R307Wwas not more effective than GATA-2 in stimulating cell cycleprogression of Mac1+Gr1− cells or Mac1−Gr1− cells (Fig. 5C).The Mac1+Gr1+ myeloid cell population in S phase decreasedsignificantly in −77−/− vs. −77+/+ cells (Fig. 5 D and E). Cellsurvival was analyzed in the −77−/− Lin−Kit+ cell genetic com-plementation assay. Although altered survival was not detected inMac1+Gr1+ cells and Mac1−Gr1− cells, GATA-2 or R307W ex-pression significantly increased live cells in Mac1+Gr1− cells (Fig.5F). Analysis of the erythroid cell population revealed the loss ofthis population in −77−/− cells. However, there was no obvious

A Empty GATA-2 R307W

****

****

G1 S G2/M

****

-77-

/- M

ac1+

Gr1

+ C

ells

Per

cen

t o

f C

ells

B

0

3

6

9

Cel

l Nu

mb

er (

x105 )

EmptyGATA-2

R307W

*

**

**

0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

200

-77-/- Mac1+Gr1+

0

75

100

50

25

-77-/- Mac1+Gr1-

0

75

100

50

25

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

-77-/- Mac1-Gr1-

**** *

***

* *

***

**

***

G1 S G2/M G1 S G2/M

0

75

100

50

25

Empty

GATA-2

R307W

Empty

GATA-2

R307W

Empty

GATA-2

R307W

CDAPI

400

0

200

400

0

100

200

0

0

G1

0

50

75

100

25

S

0

8

12

16

4

G2/M

0

5.0

7.5

10.0

2.5

-77-/- Mac1+Gr1+

-77-/--77+/+40 80 120 40 80 120

100

200

300

200

400-77+/+ -77-/-D E

*

DAPI

0

20

40

60

80

100

-77-/--77+/+

Live Early Late Dead

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

-77-/- Mac1-Gr1-

Live Early Late Dead

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

-77-/- Mac1+Gr1-

0

20

40

60

80

100

***

***

**

-77-/- Mac1+Gr1+

Per

cen

t o

f C

ellsF

Live Early Late Dead

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

Em

pty

Em

pty

GA

TA-2

R30

7WE

mp

tyE

mp

tyG

ATA

-2R

307W

20

40

60

80

100

0 00 0 P

erce

nt

of

Cel

ls

Mac

1+G

r1+

Cel

ls

* **

**

Fig. 5. GATA-2 leukemia mutant increases cell cycle progression in a genetic complementation assay. (A) Quantification of cell number in colonies derivedfrom −77−/− Lin−Kit+ cells infected with empty vector, GATA-2 expression vector, or R307W expression vector. (B) Flow cytometric analysis of cell cycle in−77−/− Mac-1+Gr-1+ cells isolated from colonies. (C) Quantification of cell cycle status (n = 5). (D) Flow cytometric analysis of cell cycle in −77+/+ Mac-1+Gr-1+ cellsand −77−/− Mac-1+Gr-1+ cells. (E) Quantification of cell cycle status (n = 4). (F) Quantification of early apoptotic (Annexin V+DRAQ7−), late apoptotic (AnnexinV+DRAQ7+), and dead cells (Annexin V−DRAQ7+) by flow cytometry (n = 4). *P < 0.05, **P < 0.01, ***P < 0.001.

Katsumura et al. PNAS | vol. 115 | no. 43 | E10115

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 8

change in survival of R1 population erythroid precursor cells (SIAppendix, Fig. S3). This analysis indicates that GATA-2 stimulatescell cycle progression in −77−/− myelo-erythroid progenitor cells,and while the leukemia mutant not only retains this activity, itsactivity can exceed that of GATA-2.

DiscussionThe paradigm for how GATA2 mutations instigate MDS/AMLinvolves haploinsufficiency (27): inadequate GATA-2 productionto fulfill its requirement to establish/maintain genetic networksthat govern HSPC transitions. Heterozygous GATA2 coding re-gion mutations inhibit DNA binding or corrupt the ORF, while+9.5 enhancer mutations reduce GATA2 expression. Both classesof mutations yield subphysiological GATA-2 levels. High-levelGATA-2 expression in humans has been correlated with poorprognosis of AML (30), and GATA-2 overexpression can suppressbone marrow hematopoiesis in mice (22).We devised a genetic complementation assay that enables

quantification of GATA-2–dependent myelo-erythroid pro-genitor differentiation, endogenous target gene regulation, andcellular functions. We discovered that GATA-2 mutants pre-dicted to be inactive or to have attenuated activity retain activityin primary cells. Of particular interest were their activities toinduce granulocytes and stimulate cell cycle progression, whichexceeded that of GATA-2. Whereas GATA-2 induced −77−/−myeloid progenitor cells to produce both erythroid and myeloidcells, R307W selectively increased granulocytes (Fig. 6). Thus,leukemia mutations corrupt, without abrogating, GATA-2 function.It is therefore instructive to consider the relationship betweenour findings and AML pathogenesis involving GATA2 muta-tions. We propose that insufficient or elevated GATA-2 levels/activity corrupt GATA-2–dependent genetic network integrity,and both GATA-2 loss-of-function and gain-of-function mayconstitute pathogenic mechanisms.Although the functional consequences of GATA-2 N-finger

mutations detected in a subset of AML patients were unclear,prior studies implicated the GATA-1 N-finger in DNA bindingin certain contexts (38). However, whether the N-finger is anessential determinant of chromatin occupancy in physiologicalcontexts is uncertain. Herein, we demonstrated that N-fingerleukemia mutations resembled C-finger mutations in attenuat-ing GATA-2 chromatin binding and target gene activation.However, analysis of C295A and C349A mutations that disruptN- and C-finger structure, respectively, indicated that N-fingermutants were more effective than C-finger mutants in activatingtarget genes. N-finger, but not C-finger, mutants were responsiveto Ras(G12V)-mediated GATA-2 activation. These resultshighlight functional differences in N- and C-finger mutants.C-finger mutants have been described as germline mutations

in familial MDS/AML (10–12). Somatic N-finger mutations werereported in patients with acute erythroid leukemia (35) andAML with biallelic mutation of CEBPA (53). Analyses of familialMDS/AML have identified the co-occurrence of T354M withacquired ASXL1 mutations (54, 55). Other mutations reportedto co-occur with GATA2 mutations include NRAS, WT1, andDDX41, among others (56, 57). In addition, CDC25C mutationsin familial platelet disorder, which predisposes to AML, caninvolve subsequent GATA2 somatic mutations (58). Furtherstudies with large patient cohorts are required to rigorously an-alyze genotype–phenotype relationships.The GATA-1 N-finger is an essential determinant of GATA-

1 function (40). V205 mediates FOG-1 binding and FOG-1–dependent transcriptional activation and repression (40).While R216 does not impact FOG-1 binding, it contributes tothe regulation of select target genes (43). Our study revealedthat GATA-2 R307, the structural equivalent of GATA-1R216, is critical for GATA-2 function, while GATA-2 V296,which is equivalent to GATA-1 V205, is dispensable for

GATA-2–mediated Hdc induction. GATA-1 R216 is importantfor GATA-1 recognition of palindromic GATA-motifs (59),and in vivo analysis demonstrated that R216 mutation de-creased occupancy at sites containing single GATA-motifs(60). Our studies revealed that GATA-2 R307W strongly in-creased CFU-GM. Mutations of this conserved arginine ofGATA-1, GATA-2, and GATA-3 were described in hemato-logic malignancy and anemia patients.How do GATA-2 leukemia mutants induce myeloid cell pro-

liferation despite their crippled transcriptional activity? It is in-structive to consider the consequences of mutating GATA-1V205, which strongly reduces GATA-1–mediated activation andrepression of target genes that require FOG-1, the majority ofGATA-1 target genes. V205 mutations inhibit chromatin occu-pancy at select target genes and induce ectopic chromatin oc-cupancy at sites not normally occupied by GATA-1 (47). Thischromatin redistribution mechanism may have broader appli-cability to transcription factor mutations that influence DNAsequence-specificity or coregulator–transcription factor interac-tions and therefore indirectly impact chromatin occupancy. Be-cause various factors are implicated in binding GATA factors, inprinciple, mutations that impact chromatin occupancy mightinhibit or enhance such interactions, thereby altering networksestablished/maintained by interactors. In addition, DNA binding-impaired GATA factor mutants might retain the capacity to berecruited into chromatin complexes via protein–protein interac-tions. Regardless of these potential mechanisms, it will be in-structive to consider the relationship between the unexpectedmyelopoiesis-inducing activity resulting from human GATA-2disease mutations and leukemogenesis.

Macrophage-biaseddifferentiation

Myelo-erythroiddifferentiation

Excessive granulocytedifferentiation/proliferation

erythroid granulocyte macrophage

Myeloid Progenitors(-77-/- Lin-Kit+)

Control

GA

TA-2

Mutant

B

GATA-2GeneRegulation

ChromatinOccupancy

R307W T354M A318T V296MC295A C349A

HighLow

A

ColonyFormation

ErythroidProliferation

ND

ND

ND

ND

ND

ND

NDND

MyeloidProliferation ND ND

Fig. 6. GATA-2 leukemia mutations: gain-of-function and loss-of-functionconsequences. (A) Molecular activities of wild-type and mutant GATA-2.DNA binding capacity of T354M was described previously (31). ND, not de-termined. (B) The −77−/− myeloid progenitor cells differentiate preferentiallyinto macrophages ex vivo. While GATA-2 expression induces erythroid cellsand granulocytes, GATA-2 leukemia mutants stimulate granulocyte differ-entiation and proliferation, and this activity can exceed that of GATA-2.Thus, GATA-2 leukemia mutants exhibit a gain-of-function activity tostimulate myelopoiesis, with a concomitant loss of activity to stimulateerythropoiesis.

E10116 | www.pnas.org/cgi/doi/10.1073/pnas.1813015115 Katsumura et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 9

Materials and MethodsCell Culture. MAE cells (31) were maintained in medium 200 supplementedlow-serum growth supplement (Thermo Fisher Scientific) and 1% penicillin/streptomycin (Thermo Fisher Scientific). Cells were transfected withNucleofector II (Lonza). Fetal liver hematopoietic precursor cells were cul-tured in StemPro-34 (Thermo Fisher Scientific) with 1× nutrient supplementwith 2 mM glutamax (Thermo Fisher Scientific), 1% penicillin/streptomycin(Thermo Fisher Scientific), 100 μM monothioglycerol (Sigma), 1 μM dexa-methasone (Sigma), 0.5 U/mL of erythropoietin, and 1% conditioned mediumfrom a Kit ligand-producing Chinese hamster ovary (CHO) cells. Cells werecultured in a humidified incubator at 37 °C and 5% carbon dioxide (61).

Plasmids.Murine GATA-2 cDNAwas cloned into pcDNA4TOFHA vector (kindlyprovided by Danny Reinberg, New York University, New York) and pMSCVvector (kindly provided byMitchellWeiss, St. Jude Children’s ResearchHospital,Memphis, TN). XZ-201Ras(G12V) was kindly provided by Jing Zhang, Univer-sity of Wisconsin–Madison, Madison, WI.

Quantitative Real-Time RT-PCR. Total RNA was purified with TRIzol (ThermoFisher Scientific). cDNAwas prepared by annealing 1 μg of RNAwith 250 ng ofa 1:5 mixture of random hexamer and oligo (dT) primers heated at 68 °C for10 min. This was followed by incubation with Murine Moloney LeukemiaVirus Reverse Transcriptase (Thermo Fisher Scientific) with 10 mMDTT, RNAsin(Promega), and 0.5 mM dNTPs at 42 °C for 1 h. This mixture was heat-inactivated at 95 °C for 5 min and diluted to a final volume of 100 μL.

Quantitative ChIP. ChIP analysis in MAE cells was conducted as describedpreviously (62). Samples containing 3 × 106 cells were cross-linked with 1%formaldehyde for 10 min. Lysates were immunoprecipitated with rabbitpolyclonal anti-HA antibody using rabbit preimmune serum (Covance) as acontrol. DNA was quantitated by real-time PCR (Applied Biosystems Viia7 instrument) with SYBR green fluorescence, and the amount of product wasdetermined relative to a standard curve created from serial dilution ofinput chromatin.

Protein Analysis. Protein samples were isolated by centrifugation of 1 × 106

cells from each condition, washing with cold PBS, and lysing in SDS samplebuffer (25 mM Tris, pH 6.8, 2% β-mercaptoethanol, 3% SDS, 0.005% bro-mophenol blue, 5% glycerol). Samples were boiled for 10 min and storedat −80 °C. Samples were resolved by SDS/PAGE, and proteins were detectedby semiquantitative Western blotting with ECL Plus (Pierce). For primaryfetal liver cells, FEMTO supersignal (Pierce) was used.

Immunofluorescence. Cells were cytospun and fixed with 3.7% paraformal-dehyde in PBS for 10 min at room temperature. Slides were washed with PBS,and cells were permeabilized with 0.2% Triton X-100 for 10 min at room

temperature. After washing, slides were blocked with 10% BlokHen (AvesLabs) in 0.1% Tween 20 in PBS for 1 h at 37 °C and then incubated withprimary antibody (anti-HA, Covance HA11) in 2% BlokHen at 4 °C overnight.After washing, slides were incubated with secondary antibody for 1 h at37 °C. Slides were washed and mounted using Vectashield mounting me-dium with DAPI (Vector Laboratories).

Colony Forming Unit Assay. For CFU assays, dissociated Lin− cells or Lin−Kit+

cells from E14.5 fetal livers were plated in duplicate in Methocult M3434complete media (StemCell Technologies) at 1 × 103 cells per 35-mm plate.Plates were incubated for 8 d, and colonies were identified and enumerated.For subsequent analysis, cells were isolated from the plates using PBS con-taining 50% calf bovine serum and centrifuged for 10 min to removemethylcellulose.

Flow Cytometry. For Lin−Kit+ cell sorting, E14.5 fetal liver cells were dissoci-ated and resuspended in PBS with 10% FBS and passed through 25-μm cellstrainers to obtain single-cell suspensions before antibody staining. Aftercells were stained with FITC-conjugated CD5 (11-0051-85), CD8 (11-0081-85),CD19 (11-0193), IgM (11-5890), Il7Ra (11-1271), AA4.1 (11-5892; ThermoFisher Scientific), B220 (103206), CD3 (100306), CD4 (100406), TER-119(116206), and PE Cy7-conjugated c-Kit (105814; Biolegend) antibodies,Lin−Kit+ cells were collected on a FACSAria II cell sorter (BD Biosciences). Foranalysis of myeloid cells, cells isolated from colonies were fixed in 2%paraformaldehyde for 10 min at 37 °C. After permeabilization overnightat −20 °C in 95% methanol, cells were incubated for 1 h in HBSS/4% FBS at4 °C. After incubation with Fc block on ice for 15 min, cells were stained withanti-mouse Mac1-APCe780 (47-0112-82; Thermo Fisher Scientific) and anti-mouse Gr1-PE-Cy7 (108416; Biolegend) at room temperature for 30 min.DAPI was added at this stage for cell cycle analysis. Cells were washed twicein PBS before analysis and analyzed on a LSR II flow cytometer (BD Biosci-ences). The data were analyzed using FlowJo v10.1 software (TreeStar) andModFit LT software (Verity Software House).

Apoptosis Assay. To quantify apoptosis after Mac1/Gr1 staining, cells werewashed in Annexin V buffer (10mMHepes, 140mMNaCl, 2.5mMCaCl2, pH 7.4)and stained with Annexin V-Pacific blue (A35122; TermoFisher) and DRAQ7(ab109292; Abcam) for 15 min in the dark at room temperature.

Statistical Analysis. Statistical significance was determined by Student’s t-testusing web-based GraphPad (https://www.graphpad.com).

ACKNOWLEDGMENTS. This work was supported by National Institutes ofHealth Grants DK68634 and DK50107 (to E.H.B.) and K01DK113117 (toK.J.H.), and by the Carbone Cancer Center P30CA014520.

1. Döhner H, et al. (2017) Diagnosis and management of AML in adults: 2017 ELN rec-ommendations from an international expert panel. Blood 129:424–447.

2. Khwaja A, et al. (2016) Acute myeloid leukaemia. Nat Rev Dis Primers 2:16010.3. Figueroa ME, et al. (2010) DNA methylation signatures identify biologically distinct

subtypes in acute myeloid leukemia. Cancer Cell 17:13–27.4. Shih AH, et al. (2015) Mutational cooperativity linked to combinatorial epigenetic

gain of function in acute myeloid leukemia. Cancer Cell 27:502–515.5. Ribeiro AF, et al. (2012) Mutant DNMT3A: A marker of poor prognosis in acute my-

eloid leukemia. Blood 119:5824–5831.6. Welch JS, et al. (2012) The origin and evolution of mutations in acute myeloid leu-

kemia. Cell 150:264–278.7. Jan M, et al. (2012) Clonal evolution of preleukemic hematopoietic stem cells pre-

cedes human acute myeloid leukemia. Sci Transl Med 4:149ra118.8. Klco JM, et al. (2014) Functional heterogeneity of genetically defined subclones in

acute myeloid leukemia. Cancer Cell 25:379–392.9. Uy GL, et al. (2017) Dynamic changes in the clonal structure of MDS and AML in re-

sponse to epigenetic therapy. Leukemia 31:872–881.10. Dickinson RE, et al. (2011) Exome sequencing identifies GATA-2 mutation as the cause

of dendritic cell, monocyte, B and NK lymphoid deficiency. Blood 118:2656–2658.11. Hsu AP, et al. (2011) Mutations in GATA2 are associated with the autosomal dominant

and sporadic monocytopenia and mycobacterial infection (MonoMAC) syndrome.Blood 118:2653–2655.

12. Hahn CN, et al. (2011) Heritable GATA2 mutations associated with familial myelo-dysplastic syndrome and acute myeloid leukemia. Nat Genet 43:1012–1017.

13. Tsai FY, et al. (1994) An early haematopoietic defect in mice lacking the transcriptionfactor GATA-2. Nature 371:221–226.

14. Gao X, et al. (2013) Gata2 cis-element is required for hematopoietic stem cell gen-eration in the mammalian embryo. J Exp Med 210:2833–2842.

15. de Pater E, et al. (2013) Gata2 is required for HSC generation and survival. J Exp Med210:2843–2850.

16. Rodrigues NP, et al. (2005) Haploinsufficiency of GATA-2 perturbs adult hematopoi-etic stem-cell homeostasis. Blood 106:477–484.

17. Ling KW, et al. (2004) GATA-2 plays two functionally distinct roles during the on-togeny of hematopoietic stem cells. J Exp Med 200:871–882.

18. Johnson KD, et al. (2012) Cis-element mutated in GATA2-dependent immunodefi-ciency governs hematopoiesis and vascular integrity. J Clin Invest 122:3692–3704.

19. Mehta C, et al. (2017) Integrating enhancer mechanisms to establish a hierarchicalblood development program. Cell Rep 20:2966–2979.

20. Johnson KD, et al. (2015) Cis-regulatory mechanisms governing stem and progenitorcell transitions. Sci Adv 1:e1500503.

21. Rodrigues NP, et al. (2008) GATA-2 regulates granulocyte-macrophage progenitor cellfunction. Blood 112:4862–4873.

22. Persons DA, et al. (1999) Enforced expression of the GATA-2 transcription factorblocks normal hematopoiesis. Blood 93:488–499.

23. Spinner MA, et al. (2014) GATA2 deficiency: A protean disorder of hematopoiesis,lymphatics, and immunity. Blood 123:809–821.

24. Dickinson RE, et al. (2014) The evolution of cellular deficiency in GATA2 mutation.Blood 123:863–874.

25. Wlodarski MW, et al.; EWOG-MDS (2016) Prevalence, clinical characteristics, andprognosis of GATA2-related myelodysplastic syndromes in children and adolescents.Blood 127:1387–1397.

26. Katsumura KR, Bresnick EH, Group GFM; GATA Factor Mechanisms Group (2017) TheGATA factor revolution in hematology. Blood 129:2092–2102.

27. Hsu AP, et al. (2013) GATA2 haploinsufficiency caused by mutations in a con-served intronic element leads to MonoMAC syndrome. Blood 121:3830–3837, S3831–S3837.

28. Yamazaki H, et al. (2014) A remote GATA2 hematopoietic enhancer drives leukemogenesisin inv(3)(q21;q26) by activating EVI1 expression. Cancer Cell 25:415–427.

29. Gröschel S, et al. (2014) A single oncogenic enhancer rearrangement causes con-comitant EVI1 and GATA2 deregulation in leukemia. Cell 157:369–381.

Katsumura et al. PNAS | vol. 115 | no. 43 | E10117

GEN

ETICS

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0

Page 10

30. Vicente C, et al. (2012) Overexpression of GATA2 predicts an adverse prognosis for

patients with acute myeloid leukemia and it is associated with distinct molecular

abnormalities. Leukemia 26:550–554.31. Katsumura KR, Yang C, Boyer ME, Li L, Bresnick EH (2014) Molecular basis of crosstalk

between oncogenic Ras and the master regulator of hematopoiesis GATA-2. EMBO

Rep 15:938–947.32. Katsumura KR, Ong IM, DeVilbiss AW, Sanalkumar R, Bresnick EH (2016) GATA factor-

dependent positive-feedback circuit in acute myeloid leukemia cells. Cell Rep 16:

2428–2441.33. Martin DI, Orkin SH (1990) Transcriptional activation and DNA binding by the

erythroid factor GF-1/NF-E1/Eryf 1. Genes Dev 4:1886–1898.34. Omichinski JG, et al. (1993) A small single-“finger” peptide from the erythroid

transcription factor GATA-1 binds specifically to DNA as a zinc or iron complex. Proc

Natl Acad Sci USA 90:1676–1680.35. Ping N, et al. (2017) Exome sequencing identifies highly recurrent somatic GATA2 and

CEBPA mutations in acute erythroid leukemia. Leukemia 31:195–202.36. Fasan A, et al. (2013) GATA2 mutations are frequent in intermediate-risk karyotype

AML with biallelic CEBPA mutations and are associated with favorable prognosis.

Leukemia 27:482–485.37. Greif PA, et al. (2012) GATA2 zinc finger 1 mutations associated with biallelic CEBPA

mutations define a unique genetic entity of acute myeloid leukemia. Blood 120:

395–403.38. Ghirlando R, Trainor CD (2003) Determinants of GATA-1 binding to DNA: The role of

non-finger residues. J Biol Chem 278:45620–45628.39. Pedone PV, et al. (1997) The N-terminal fingers of chicken GATA-2 and GATA-3 are

independent sequence-specific DNA binding domains. EMBO J 16:2874–2882.40. Crispino JD, Lodish MB, MacKay JP, Orkin SH (1999) Use of altered specificity mutants

to probe a specific protein-protein interaction in differentiation: The GATA-1:FOG

complex. Mol Cell 3:219–228.41. Nichols KE, et al. (2000) Familial dyserythropoietic anaemia and thrombocytopenia

due to an inherited mutation in GATA1. Nat Genet 24:266–270.42. Tubman VN, et al. (2007) X-linked gray platelet syndrome due to a GATA1 Arg216Gln

mutation. Blood 109:3297–3299.43. Campbell AE, Wilkinson-White L, Mackay JP, Matthews JM, Blobel GA (2013) Analysis

of disease-causing GATA1 mutations in murine gene complementation systems.

Blood 121:5218–5227.44. Grass JA, et al. (2006) Distinct functions of dispersed GATA factor complexes at an

endogenous gene locus. Mol Cell Biol 26:7056–7067.45. Johnson KD, et al. (2007) Friend of GATA-1-independent transcriptional repression: A

novel mode of GATA-1 function. Blood 109:5230–5233.

46. Tsang AP, et al. (1997) FOG, a multitype zinc finger protein, acts as a cofactor fortranscription factor GATA-1 in erythroid and megakaryocytic differentiation. Cell 90:109–119.

47. Chlon TM, Doré LC, Crispino JD (2012) Cofactor-mediated restriction of GATA-1 chromatin occupancy coordinates lineage-specific gene expression. Mol Cell 47:608–621.

48. Gregory T, et al. (1999) GATA-1 and erythropoietin cooperate to promote erythroidcell survival by regulating bcl-xL expression. Blood 94:87–96.

49. Weiss MJ, Yu C, Orkin SH (1997) Erythroid-cell-specific properties of transcriptionfactor GATA-1 revealed by phenotypic rescue of a gene-targeted cell line. Mol CellBiol 17:1642–1651.

50. Bresnick EH, et al. (2018) Mechanisms of erythrocyte development and regeneration:Implications for regenerative medicine and beyond. Development 145:dev151423.

51. Grass JA, et al. (2003) GATA-1-dependent transcriptional repression of GATA-2 viadisruption of positive autoregulation and domain-wide chromatin remodeling. ProcNatl Acad Sci USA 100:8811–8816.

52. Tipping AJ, et al. (2009) High GATA-2 expression inhibits human hematopoietic stemand progenitor cell function by effects on cell cycle. Blood 113:2661–2672.

53. Theis F, et al. (2016) Clinical impact of GATA2 mutations in acute myeloid leukemiapatients harboring CEBPA mutations: A study of the AML study group. Leukemia 30:2248–2250.

54. Bödör C, et al. (2012) Germ-line GATA2 p.THR354MET mutation in familial myelo-dysplastic syndrome with acquired monosomy 7 and ASXL1 mutation demonstratingrapid onset and poor survival. Haematologica 97:890–894.

55. Churpek JE, et al. (2015) Genomic analysis of germ line and somatic variants in familialmyelodysplasia/acute myeloid leukemia. Blood 126:2484–2490.

56. Drazer MW, et al. (2018) Prognostic tumor sequencing panels frequently identifygerm line variants associated with hereditary hematopoietic malignancies. Blood Adv2:146–150.

57. Makishima H, et al. (2017) Dynamics of clonal evolution in myelodysplastic syndromes.Nat Genet 49:204–212.

58. Yoshimi A, et al. (2014) Recurrent CDC25C mutations drive malignant transformationin FPD/AML. Nat Commun 5:4770.

59. Yu C, et al. (2002) X-linked thrombocytopenia with thalassemia from a mutation inthe amino finger of GATA-1 affecting DNA binding rather than FOG-1 interaction.Blood 100:2040–2045.

60. Hasegawa A, et al. (2016) GATA1 binding kinetics on conformation-specific bindingsites elicit differential transcriptional regulation. Mol Cell Biol 36:2151–2167.

61. McIver SC, et al. (2018) Dissecting regulatory mechanisms using mouse fetal liver-derived erythroid cells. Methods Mol Biol 1698:67–89.

62. Im H, et al. (2004) Measurement of protein-DNA interactions in vivo by chromatinimmunoprecipitation. Methods Mol Biol 284:129–146.

E10118 | www.pnas.org/cgi/doi/10.1073/pnas.1813015115 Katsumura et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 6,

202

0