CRISPR Gene Editing in Stem Cells

with the remaining 15 derived from normal cells), whereas the

del(7q) was present in over 70% of the starting cells (Table

S1). To further investigate this and to determine whether the

in vitro culture that is necessary to initiate reprogramming or

the reprogramming process per se accounts for this skewing,

we compared the clonal composition of cells from patient 2

before (day 0) and after in vitro culture (day 3) (Figures S1B

and S1C). This showed that in vitro culture did not select for

normal cells but rather resulted in preferential growth of the

MDS clone over the normal cells, since the variant allele fre-

quency (VAF) of both

SRSF2

P95L and

PHF6

C280Y clonal so-

matic mutations increased over time in culture. Similarly, cells

from patients 7 and 8 analyzed for the del(5q) abnormality after

culture and immediately before the initiation of reprogramming

were found to consist mostly of clonal MDS cells, as the copy

number of chr5q was almost 1 in both samples and 8 out

of 10 metaphases of patient 7 cells harbored the del5q by

karyotyping (Figures S1D and S1E). Finally, we compared

the outcome of two different reprogramming protocols using

in vitro expansion and reprogramming of either hematopoietic

progenitors or erythroblasts, performed in parallel with the

same aliquot of starting cells divided in two (Table S2). Eryth-

roblast reprogramming, similarly to hematopoietic progenitor

reprogramming, preferentially gave rise to normal iPSCs. These

results show that the relative reprogramming disadvantage of

Figure 2. Disease-Stage-Specific iPSCs Capture

Phenotypes of Graded Severity

(A) Scheme of hematopoietic differentiation protocol.

(B) Fraction of CD34

cells generated by all the different

lines tested on day 8 of differentiation (extended data

are shown in Figure S3C). Note color coding (key is

shown in the top right panel of this figure). Mean and

SEM of different lines are shown. For lines differentiated

more than once (Figure S3C), the average value is

shown.

cells that have lost

CD34 expression by day 14 of differentiation generated

by the different iPSC lines (extended data are shown in

Figure S3F). Mean and SEM of different lines are shown.

For lines differentiated more than once, and the average

value is shown.

(D) Fraction of CD34

cells maintaining CD90 expression

at the indicated days of differentiation (extended data

are shown in Figure S4B, top). For lines differentiated

more than once, the average value is shown.

(E) Fraction of CD41a

/CD45 cells, corresponding to

megakaryocyte progenitors, at the indicated days of

differentiation (extended data are shown in Figure S4B,

bottom). For lines differentiated more than once, the

average value is shown.

(F) Colony assays for megakaryocyte progenitors (CFU-

Mk) in unsorted, CD41a

, and CD41a sorted cells from

N-2.12 iPSCs at day 8 of differentiation. The average of

two independent experiments is shown.

(G) Representative images of a medium (top) and large

(bottom) CFU-Mk colony. Scale bars, 50

See also Figures S3 and S4 and Table S3.

MDS cells from some patients is not due to

in vitro culture and suggest that some MDS-

associated genetic lesions, but not others,

exert a negative effect on or even completely abolish reprog-

ramming potential.

Disease-Stage-Specific iPSCs Capture Cellular

Phenotypes of Graded Severity or Disease Specificity

We selected a panel of iPSC lines representative of the different

disease stages for phenotypic characterization following he-

matopoietic differentiation using a protocol that enables the

derivation and study of hematopoiesis with definitive features

(Figures 2A, S3A, and S3B). All lines gave rise to comparable per-

centages of CD34

cells early in differentiation (day 8), providing

evidence against early developmental defects in mesoderm for-

mation or hematopoietic lineage specification that could com-

pound the identification of disease-relevant phenotypes (Figures

2B and S3C). In contrast, striking differences were observed in

the timing and emergence of CD45

hematopoietic progenitor

cells (HPCs) starting at the low-risk MDS stage (Figures 2C

and S3D–S3F). Preleukemic cells, like normal cells, generated

CD45

HPCs that comprised 90% of the cells by day 14 of dif-

ferentiation, with approximately half having lost CD34 expres-

sion, as a sign of further maturation beyond the progenitor stage

(Figures 2C, S3E, and S3F). Low-risk MDS CD45

HPCs ap-

peared later and matured later than normal HPCs, as evidenced

by loss of CD34 (Figures 2C, S3E, and S3F). High-risk MDS

iPSCs produced CD45

HPCs with a delay, as well as markedly

318

Cell Stem Cell

, 315–328, March 2, 2017