Science - USA (2022-04-08)

andFGFR2in lung cancer (P=9×10−^4 and 1 ×
10 −^2 , respectively),ARRDC3in kidney cancer
(P=4×10−^2 ),PIK3C2Bin bladder cancer (P=
8×10−^3 ),BCL6in leukemia (P= 1 × 10−^2 ), and
XBP1in breast cancer (P=8×10−^4 ) (fig. S44B).
These analyses provide additional support for
the plausibility of some of the mutation events
in this category, in addition to their location in
regulatory regions and enrichment for canon-
ical cancer genes (Fig. 2C).

Experimental evaluation of regulatory regions
and noncoding mutations around XBP1

Although many events in the regulatory cat-
egory fell into the promoter regions of known
cancer genes (Figs. 2 and 3A), some events
occurred outside of canonical regulatory re-

gions. For example,XBP1mutations, which

were present in ~6% of the breast cancer

patients in our WGS cohort, did not primarily

target theXPP1promoter but rather clustered

in a narrow, noncoding region downstream of

XBP1(Fig. 3A and fig. S45A), a pattern unlikely

to occur by random chance (fig. S45B).

Previous studies have connectedXBP1to

breast cancer ( 38 , 39 ) and estrogen receptor

signaling ( 40 , 41 ). Concordantly, Gene Set En-

richment Analysis showed estrogen receptor–

dependent signaling to be the most differentially

expressed pathway (FDR = 7 × 10−^4 ) between

breast cancer samples with high versus low

XBP1expression (Fig. 5 and fig. S46). Further-

more,XBP1was only expressed in prediction

analysis of microarray 50 [PAM50 ( 42 )] expres-

sion types related to hormone receptor signal-

ing (luminal A/B,HER2-enriched types) but

not in other breast tumors (basal-like type)

(fig. S47). In addition, the average ATAC-seq

signal aroundXBP1was 1.83-fold higher in

receptor-positive versus receptor-negative breast

tumors (P< 0.001, basal-like versus non–basal-

like PAM50 subtype, Mann-WhitneyUtest)

(fig. S46, D and E), suggesting that regulatory

regions aroundXBP1exhibited primary activ-

ity in the hormone receptor–related subtype.

We confirmed somatic mutations aroundXBP1

using Sanger sequencing in breast tumors

from our WGS cohort (fig. S48).

We used two experimental assays to further

assess mutations nearXBP1and to provide

proof-of-principle support for the possible

Dietleinet al.,Science 376 , eabg5601 (2022) 8 April 2022 8 of 12

0

50

100

ATAC-seq(% of max)

CCDC117 XBP1

3D promoter

interactions

ATAC-seq

in XBP1-pos

breast cancer

0

50

100

mutation density

(% of max)

29.16 29.18 29.2 29.22 29.24

Position on Chr 22

mutation density

in breast cancer

1.5

2.0

2.5

3.0

wt mut wt mut

mut near XBP1

log expression

XBP1levels mut vs. wt tumors

PCAWG CCLE

*** **

EARLY

LATE

HALLMARKESTROGEN

RESPONSE

0.0

1.0

2.0

3.0

-0.25 0.0 0.25 0.5 0.75

max. GSEA score

-log FDR

GSEAXBP1-pos vs. neg tumors

EARLY

ESTROGEN

RESPONSE

0.0

0.2

0.4

0.6

0 5 10 15 20

LATE

ESTROGEN

RESPONSE

0.0

0.2

0.4

0.6

0 5 10 15 20

gene rank (x1000) gene rank (x1000)

GSEA score GSEA score

-30

-30

-15

-15

0

0

15

15

30

30

45

45

effect score replicate 1

effect score replicate 2

effective sgRNA no effect

-30 -15 0 15 30 45

0.4

4

40

400

effect score replicate 1

ATAC-seq signal

effective sgRNA no effect

0.0

0.5

1.0

1.5

2.0

x-fold

expression

XBP1

#429F#493R#588F#920F#966F#977F#1401R#1403R#1414R#1683R#1894R#1949F#1954F#2093F#2138F#2307F#2310R#2371F#2420F#2481R

Position

on Chr 22

29.16 29.18 XBP1 29.2 29.22 29.24

0

12

25

37

50

mutated regions

ATAC-seq peaks

% of sgRNAs

with effect on

XBP1 expression

local fraction ofeffective sgRNAs

A

B

CD

E

G

4) sort cells in equisized bins

low middle high

5) score sgRNA effects

cell count

6) map effects to target regions

1) 2,923 CRISPR sgRNAs 2) transfection of 10 cells^7 3) expression in flow cytometry

effect

cells with low

expression

efficient

sgRNA

no

effect

count

lowmiddlehigh lowmiddlehigh

F H

IJ

region targeted by sgRNAs

Fig. 5. Noncoding somatic mutations occur in regulatory regions aroundXBP1.
(A) CRISPRi screening of regions aroundXBP1using a library of 2923 sgRNAs
in breast cancer cells (CAMA1). Regulatory regions were localized based on
sgRNAs, for which KRAB-mediated silencing of their target region led to
decreasedXBP1expression in flow cytometry (orange). (B) Fractions of effective
sgRNAs (y-axis) plotted against their position aroundXBP1(x-axis). Positions
of ATAC-seq peaks (teal, bottom), noncoding mutations (purple, bottom),
and target regions of the sgRNAs (top) are annotated. (CandD) Efficacies of
sgRNAs (sliding window of 10 adjacent sgRNAs) compared between experimental
replicates [x-axis versusy-axis (C)] and the ATAC-seq signal of their target
regions in breast cancer [y-axis (D)]. (E) Bar graphs displaying theXBP1
expression ratio before and after CRISPRi in regulatory regions (orange) and
nonregulatory regions (gray) for individual sgRNAs. Error bars reflect the

SD across cells. (F) Mutation densities (purple), ATAC-seq signals (teal), and three-

dimensional interactions in the breast cancer genome of MCF7 (ChIA-PET, black)

plotted against their genomic position aroundXBP1(x-axis). (G)XBP1expression

compared between breast tumors with [purple, mutated (mut)] and without [gray,

wild-type (wt)] mutations aroundXBP1in PCAWG (left) and CCLE (right). Boxes

indicate the 25/75% interquartile range, vertical lines extend to 10/90% percentiles,

and horizontal lines reflect distribution medians ofXBP1expression. Significant

differences (Mann-WhitneyUtest) are annotated with asterisks: *P< 0.05, **P< 0.01,

***P< 0.001. (H) Gene Set Enrichment Analysis analyzing expression differences

in tumors with high versus lowXBP1expression by computing an enrichment score

(x-axis) and a significance value (y-axis) for each hallmark signature. (IandJ) Gene

ranks (x-axis) are plotted against enrichment scores (y-axis) for early (I) and late

(J) estrogen response signatures (black).

RESEARCH | RESEARCH ARTICLE