RESEARCH ARTICLE

Blocking Genomic Instability Prevents

Acquired Resistance to MAPK Inhibitor
Therapy in Melanoma
Prashanthi Dharanipragada1, Xiao Zhang1, Sixue Liu1, Shirley H. Lomeli1, Aayoung Hong1, Yan Wang1,2,
Zhentao Yang1, Kara Z. Lo3, Agustin Vega-Crespo3, Antoni Ribas2,3,4,5, Stergios J. Moschos6,7,
Gatien Moriceau1, and Roger S. Lo1,2,5

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

 ABSTRACT Blocking cancer genomic instability may prevent tumor diversification and escape
from therapies. We show that, after MAPK inhibitor (MAPKi) therapy in patients and
mice bearing patient-derived xenografts (PDX), acquired resistant genomes of metastatic cutaneous
melanoma specifically amplify resistance-driver, nonhomologous end-joining (NHEJ), and homologous
recombination repair (HRR) genes via complex genomic rearrangements (CGR) and extrachromosomal
DNAs (ecDNA). Almost all sensitive and acquired-resistant genomes harbor pervasive chromothriptic
regions with disproportionately high mutational burdens and significant overlaps with ecDNA and CGR
spans. Recurrently, somatic mutations within ecDNA and CGR amplicons enrich for HRR signatures,
particularly within acquired resistant tumors. Regardless of sensitivity or resistance, breakpoint–
junctional sequence analysis suggests NHEJ as critical to double-stranded DNA break repair underly-
ing CGR and ecDNA formation. In human melanoma cell lines and PDXs, NHEJ targeting by a DNA-PKCS
inhibitor prevents/delays acquired MAPKi resistance by reducing the size of ecDNAs and CGRs early on
combination treatment. Thus, targeting the causes of genomic instability prevents acquired resistance.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

SIGNIFICANCE: Acquired resistance often results in heterogeneous, redundant survival mechanisms,
which challenge strategies aimed at reversing resistance. Acquired-resistant melanomas recurrently
evolve resistance-driving and resistance-specific amplicons via ecDNAs and CGRs, thereby nominating
chromothripsis–ecDNA–CGR biogenesis as a resistance-preventive target. Specifically, targeting DNA-
PKCS/NHEJ prevents resistance by suppressing ecDNA/CGR rearrangements in MAPKi-treated melanomas.

INTRODUCTION in patients treated with pathway-targeted therapies. Variable lev-

els of focal gene amplifications can confer phenotypic plasticity,
Acquired resistance to targeting of oncogenic pathways is the
fine-tuning tumor cell fitness in response to the selective pressure
rule rather than the exception. Genetic mechanisms of acquired
of therapies.
resistance across malignancies often occur as focal amplifica-
BRAF inhibitor (BRAFi) therapy of BRAFV600MUT metastatic
tions of resistance-driver genes. Scientific efforts to reverse the
cutaneous melanoma targets the MAPK pathway and leads
consequences of resistance evolution have focused largely on
quickly to acquired resistance (1). Genetic mechanisms of
targeting acquired vulnerabilities. In contrast, relatively little is
acquired BRAFi resistance most often result in reactivation
known regarding the mechanisms—for example, genomic insta-
of the MAPK pathway, which can be suppressed by adding
bility pathways—that enable rapid resistance evolution of cancer
a MEK inhibitor (MEKi; refs. 2–5). However, with com-
bined BRAFi and MEKi therapy of patients with BRAFV600MUT
melanoma, less than 20% of patients survive past 5 years (6).
1
Division of Dermatology, Department of Medicine, David Geffen School
of Medicine, University of California, Los Angeles, Los Angeles, California. Acquired resistance to BRAFi + MEKi therapy in patients
2
Department of Molecular and Medical Pharmacology, David Geffen School occurs through high-amplitude gene amplifications that,
of Medicine, University of California, Los Angeles, Los Angeles, California. individually or in combination, can reactivate the MAPK
3
Division of Hematology/Oncology, Department of Medicine, David Geffen pathway (7–9). Hence, there is a need to understand the
School of Medicine, University of California, Los Angeles, Los Angeles,
California. 4Division of Surgical Oncology, Department of Surgery, David
origins of such genomic instability. For patients with mela-
Geffen School of Medicine, University of California, Los Angeles, Los Angeles, noma driven by NRAS mutations, MAPK inhibitor (MAPKi)
California. 5Jonsson Comprehensive Cancer Center, David Geffen School therapy is currently not available, as MEKi monotherapy is of
of Medicine, University of California, Los Angeles, Los Angeles, California. limited clinical activity (10). However, agents added to MEKi
6
Division of Medical Oncology, Department of Medicine, The University
to suppress acquired MEKi resistance may deliver clinically
of North Carolina at Chapel Hill, Chapel Hill, North Carolina. 7Lineberger
Comprehensive Cancer Center, The University of North Carolina at Chapel meaningful efficacy (11). Recent analysis of patient-derived
Hill, Chapel Hill, North Carolina. xenografts (PDX) of NRASMUT melanoma with acquired
Note: P. Dharanipragada and X. Zhang contributed equally to this article. MEKi resistance points to similar genomic events (i.e., focal
G. Moriceau and R.S. Lo share senior authorship of this article. amplifications of BRAFWT, CRAFWT, and NRASMUT genes) as
Corresponding Author: Roger S. Lo, University of California, Los Angeles,
drivers of acquired resistance (11).
10833 Le Conte Avenue, 52-121 CHS Department of Medicine, Division Here, we test the hypothesis that chromothripsis and deriv-
of Dermatology, Los Angeles, CA 90095-1750. Phone: 310-825-5420; ative amplicons of resistance-driver genes confer melanoma
E-mail: [email protected] with genetic variants necessary to resist MAPKi therapy. Cuta-
Cancer Discov 2023;13:880–909 neous melanoma is a cancer in which the chromothrip­sis
doi: 10.1158/2159-8290.CD-22-0787 burden is already high without prior targeted therapy, and
This open access article is distributed under the Creative Commons Attribution- chromothripsis appears to be a key evolutionary mechanism
NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license. by which cancer rapidly generates and accumulates highly
©2023 The Authors; Published by the American Association for Cancer Research dynamic structural variants (SV; ref. 12). SV-related amplicons

APRIL 2023 CANCER DISCOVERY | 881

RESEARCH ARTICLE Dharanipragada et al.

can be identified as intrachromosomal complex genomic rear- (NSG) mice at doses sufficient to elicit tumor regression
rangements (CGR) and extrachromosomal DNAs (ecDNA, and then generated acquired MAPKi-resistant tumors (n = 6
aka double minutes), which may be temporally related struc- vehicle-treated tumors; n = 12 acquired-resistant tumors;
tures that confer a range of genomic–signaling plasticity (13). n = 6 normal tissues; Supplementary Fig. S1A; refs. 11,
Earlier studies have established the prognostic importance of 22). Sequencing depths across MAPKi-sensitive/naive tumors
ecDNAs in clinical tumor tissues and the potential relevance of (mean, 38×; median, 32×; range, 18–94×), acquired-resistant
ecDNAs to therapeutic resistance in cancer cell lines (14, 15). tumors (mean, 33×; median, 27×; range, 11–85×), and patient-
The non-Mendelian inheritance of ecDNAs, their accessible matched normal tissues (mean, 38×; median, 33×; range,
chromatin, as well as enhancer hijacking and transcriptional 12–98×) were comparable. The median numbers of genetic
hub congregation by ecDNAs are all mechanisms that facili- alterations [somatic single-nucleotide variants (SNV), indels
tate rapid adaptations to extreme stress such as oncogene- (ID), and copy-number variations (CNV)] for sensitive and
targeted therapy (16–19). Finally, iterative cycles of CGR and resistant tumors were, respectively: (i) 81 SNVs/megabase
ecDNA biogenesis drive the amplification of a model resist- (Mb), 5 IDs/Mb, and 261 CNVs/genome and (ii) 97 SNVs/
ance gene (DHFR) to the chemotherapy methotrexate (20). Mb, 7 IDs/Mb, and 253 CNVs/genome.
To test the hypothesis that chromothripsis, CGRs, and
ecDNAs play key roles in the evolution of MAPKi resist- Recurrence, Quantity, Chromosomal Origin, and
ance in clinical melanoma, we analyzed by whole-genome Complexity of ecDNAs and CGRs in Acquired-
Resistant versus MAPKi-Naive Melanoma

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

sequencing (WGS) three tumor cohorts (along with patient-
matched normal tissues): (i) patient-matched pre-MAPKi Using WGS data, we reconstructed ecDNA- and CGR-
and acquired MAPKi-resistant tumors from patients with derived, high copy-number (CN) amplicons in 58 BRAFV600MUT
BRAFV600MUT melanoma, (ii) acquired MAPKi-resistant mela- and NRASMUT MAPKi-sensitive/naive and acquired-resistant
noma metastatic to multiple organ sites (including the brain) tumors (n = 17 sensitive tumors; n = 41 acquired-resistant
from rapid autopsies of deceased patients with BRAFV600MUT tumors; n = 19 patients). We detected ecDNAs and CGRs at a
melanoma, and (iii) acquired MEKi-resistant melanoma and highly recurrent rate: ∼70% of sensitive and ∼77% of acquired-
untreated melanoma from BRAFV600MUT and, importantly, resistant genomes (Supplementary Table S3). With all three
NRASMUT melanoma PDXs. We evaluated (i) resistance- cohorts combined, we observed a greater number of ecDNAs
specific recurrence of amplicons harboring known and puta- and CGRs in acquired-resistant (versus sensitive) genomes
tive resistance-driver genes in CGRs and ecDNAs, (ii) overlaps (acquired-resistant: n = 941 total, n = 31 average per tumor;
in the genomic coordinates between CGR–ecDNAs and chro- sensitive: n = 241 total, n = 16 average per tumor; unpaired
mothripsis, and (iii) inferred double-stranded DNA break Student t test, P = 0.09; Fig. 1). In a patient-matched analysis
(DSB) repair pathway(s) based on analysis of breakpoint– of the clinical cohort, 12 of 15 pairwise comparisons showed
junctional sequences of CGR and ecDNA amplicons. Using a higher number of ecDNA and CGR amplicons in acquired-
human melanoma cell lines and PDXs, we tested the efficacy resistant genomes (Wilcoxon test, P = 0.033; Supplementary
and dissected the mechanisms of blocking genomic instabil- Fig. S1B). Overall, chr7p (26 of 53 genomes) and chr7q (23 of
ity via targeting of NHEJ-mediated DSB repair to prevent the 53 genomes) emerged as hotspots (fragile sites) for CGR and
evolution of acquired MAPKi resistance in both BRAFV600MUT ecDNA amplicon formation (Fig. 1). In addition, we observed
and NRASMUT melanoma. recurrent resistance-associated CGRs in chr9p (8 of 41 resist-
ant tumors across all three clinical–autopsy–PDX cohorts)
and lower-frequency resistance-associated CGRs (in 2q, 4q,
RESULTS
6p, 8q, 9q, 10q, 18p, 22p, Xp) and ecDNAs (in 6q, 10p). In the
Advanced Cutaneous Melanoma Cohorts and WGS total cohort, regardless of whether the tumor was sensitive or
Data Characteristics resistant to MAPKi, the number of rearranged segments was
To date, mutational profiles of acquired MAPKi resist- higher within ecDNAs (versus CGRs; P = 3.5e-08, unpaired
ance (in patients with BRAFV600MUT melanoma and treated Student t test; Supplementary Fig. S1C), suggesting ecDNAs
with BRAFi or BRAFi + MEKi therapy) have been limited as highly dynamic foci of genomic instability. Furthermore,
to whole-exome analysis (5, 7–9, 21). Thus, we assembled the mean number of rearranged segments (ecDNAs + CGRs)
three cohorts of tissues (Supplementary Table S1) for WGS- was highest in the RAM (∼10 segments per tumor) fol-
based analysis of SVs (Supplementary Table S2). The first lowed by the clinical (∼5 segments per tumor) and PDX
cohort consisted of patient-matched normal tissues as well as (∼4 segments per tumor) cohorts (Supplementary Fig. S1D;
BRAFV600MUT melanoma tumors before MAPKi therapy and, Kruskal–Wallis rank test, P = 0.01). This may reflect the
after initial responses, at disease progression (n = 10 normal highly advanced evolutionary state of terminal metastatic
tissues; n = 10 pretreatment tumors; n = 17 acquired-resistant melanoma in the RAM cohort. Polyploidy has been associ-
tumors; n = 10 patients). The second cohort consisted of ated with the onset of chromothripsis (12). Accordingly, we
rapid autopsy melanoma (RAM) tissues (n = 3 normal tis- observed higher numbers of CGRs and ecDNAs in genomes
sues; n = 1 sensitive BRAFV600MUT tumor; n = 12 acquired with higher ploidy (Supplementary Fig. S1E).
resistant tumors; n = 3 deceased subjects who were treated
with MAPKi; n = 6 metastatic organ sites). The third cohort CGR and ecDNA Amplicons Drive MAPKi
consisted of BRAFV600MUT or NRASMUT PDX tumors. We sub- Resistance
jected PDXs (n = 6 models; 1 BRAFMUT and 5 NRASMUT Analysis of amplicons due to intrachromosomal CGRs
models) to MAPKi therapy in NOD-scid IL2R gamma null and ecDNAs uncovered a significant (unpaired Student

882 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

 DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

Cohorts
Clinical
1q
Rapid autopsy
1p
2q PDX
2p
3q Genotype
3p
BRAF MUT
4q
5q NRAS MUT
Chromosome locations

5p
6q
Sensitive
6p
7q Acquired resistant
7p
8q
8p Genomic alterations
9q CGR
9p ecDNA
10q

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

10p
11p Number of segments
12q
13q 20
14q 40
14p 60
15q
15p
16q
17q
18p
19p
20q
21p
22q
22p
Xq
Xp
BRAF CGR
NRAS ecDNA
EGFR
HRAS
MYC
RAC1
M l_PDX .V1
6. mp 1. D. P

M _P X R2
Pt P 1.D ne

PT Pt5.Dine

R 12 AM t9. t9. line

M l_Ple.DD.D P

M _P X V1
Pt P sel 3

Pt4.D ase P1
Pt .D .D line
Pt DD DP A

9. t7. ine

M l_P X2 .R2

M _P X V1

M l_PDX .V1
M l_PDX .R4
M _P X R5

X6 1
6. 5. P1

Ba D e

R 140 3003.B in.DD.D 10

2. t1. P1

Pt t4.B t3.Dline

M PDX27.R1

M l_P DX .R3

el D 2. 2

M l_PDX .R2
AM M 0 Br n. .D ve
5. .D 2B

Pt P sel 2

M l_PDX 7.VP
Pt P sel 2
P1

A 4 0 . in D P

M l_PDX .R1

M _P X R4

D 6.V5

3
R RA M1 006 ren g.DD.D 9
R M1 .01 12. t9. D. P2

R M1 14 006Braal.D D.DP
R 1 .01 rai .S D.DP1

R M1 6.A .Lu in. .D 3

M l_PDX D.DP
el D 27 1
R RA 1301. rai DD siti 2

M usc 2.D D.DP

AM 40 . .L er D P
3. t2 ine

P sel C

14 06 Lymung .DD.DP
P1
P

7. t6. in

M l_PDX .R
P

_P X R

.R
A 4 d n D P
A 0 3 ra D P

00 .Ly ph .D .D
AM 2. .B n. en P
4. D. P1

M h D D
R M1 006 06 Liv .D .D
3
i

1 0 a D P
Ba D

el D 1.

el D 2.
Ba D
Ba .D

P D D
AM 2 .B 01 D D

el D 3.
Pt sel

el D 4.
el D 4.
Ba D
Ba P

4
3
P P e

3
P P e

4
1
s
s

2
Ba

t
4 D
1.
Pt

e
e
e

e
e

e
e
e
e
e
A M 4 .
AM R
A
R

Figure 1. Landscape of ecDNAs and CGRs in patient-matched MAPKi-sensitive and acquired resistant melanoma. Chromosomal distribution of ecDNA
and CGR amplicons in three cohorts: BRAFV600MUT clinical cohort (baseline tumors, n = 8; resistant tumors, n = 11; patients, n = 8), RAM tissues (sensitive
tumor, n = 1; resistant tumors, n = 12; deceased patients, n = 3), and PDXs (vehicle-treated or sensitive, n = 6; resistant tumors, n = 12; patients of origin,
n = 6). Os and Xs, ecDNAs and CGRs, respectively; sizes, numbers of CN variant segments (minimal CN of 4.5; reference sizes shown). Pretreatment and
post–acquired resistance tumors from Pt10 and Pt11 are not shown given no detection of ecDNAs and CGRs.

t test, P = 0.0002) association between acquired-resistant by RAC1 (23). Moreover, we validated a recurrent ecDNA by
tumors and CGRs/ecDNAs harboring bona fide MAPKi direct isolation and high-depth sequencing [NRAS amplifica-
resistance–driver genes. BRAF (CN range, 4.5–27), NRAS tion, CN 13, Pt9-double drug-disease progression (DD-DP)1;
(CN, 5–13), HRAS (CN, 13–16), MYC (CN, 12–15), and EGFR Fig. 2A] using a new approach referred to as CRISPR-CATCH
(CN, 4.6–5; Figs. 1 and 2A–D) when amplified are known to (24). This alternative technique confirmed the circularized
drive acquired MAPKi resistance and MAPK pathway reacti- junctions of an 890-kb driver ecDNA within this acquired-
vation (5, 7–9, 11). RAC1 (CN, 6–7) in an ecDNA amplicon resistant clinical tumor sample. In addition, we used DNA-
was specifically observed in acquired resistance in NRASMUT FISH to validate the resistance-specific presence and the CNs
melanoma PDXs, which suggests regulation of MLK3–CRAF of BRAF and NRAS amplicons (Fig. 2E). In total, among the 17

APRIL 2023 CANCER DISCOVERY | 883

 RESEARCH ARTICLE Dharanipragada et al.

A BRAF MUT—Clinical (Pt9-DD-DP1)

E BRAF /CEN7/DAPI NRAS/CEN1/DAPI NRAS/CEN1/DAPI
175 35
150 30
Coverage

125 25

Mel_PDX27-V1

Mel_PDX1-V1

Mel_PDX2-V1
100 20

CN
75 15
50 10
25 5
0 0
TRIM33 NRAS
chr1
42518432
42539638
114441294

116371336

Mel_PDX27-R2
B

Mel_PDX1-R1

Mel_PDX2-R2
BRAF MUT—rapid autopsy (RAM12.01-Brain-DDP10)
1,200 80
1,000 60
Coverage

800
40

CN
600
400 20
200
0 0
CDK6 BRAF
Mel_PDX27 Mel_PDX1 Mel_PDX2
chr7 60 *** 15 *** 15 ***
92757501

93431000
140031001

141584000

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

40 10

NRAS CN

NRAS CN
10

BRAF CN
C 175
BRAF MUT
—PDX (Mel_PDX27-R2)
20 5 5
150 35
30
Coverage

125 25
100 20 0 0
CN

75 0
15 V1 R2 V1 R1 V1 R2
50 10
25 5
0 0
BRAF
chr7
139668001

141933200

D NRAS MUT—PDX (Mel_PDX2-R2) F BRAF MUT—clinical (Pt2-DP1)

350 70 3,500 200
300 60 3,000 175
Coverage
Coverage

250 50 2,500 150

200 40 2,000 125
100
CN

1,500

CN
150 30 75
100 20 1,000 50
50 10 500 25
0 0 0 0
TRIM33 ATP1A1TENT5C NOTCH2 CREB2L2 TRIM24 BRAF EZH2 XRCC2
RBM15 NRAS
chr1 7 17 chr7 11
136414501

141637000
148317501

148882000
152566501

153202000
74487202
74497776
42051582
42072132
48185908
48196458
107638001

121714000
58086567
58097117
62266856
62277406
62374042
62384592
26734580
26755131

Figure 2. Genes amplified by ecDNAs and CGRs in acquired MAPKi resistance and enriched pathways. A–D, SV view of reconstructed amplicons in rep-
resentative acquired MAPKi-resistant melanoma tumors in the clinical (A), RAM (B), BRAFV600MUT PDX (C), and NRASMUT PDX (D) cohorts. Horizontal black
and red lines indicate, respectively, genomic segments with similar CNs and genes. Short vertical lines and arcs indicate discordant read pairs linking two
amplicons via an SV junction. Long vertical lines indicate break ends that map from amplicon into low-complexity regions that cannot be traced further.
Each line/arc representing discordant reads is colored based on differences from expected distance or orientation. Yellow arrows point out resistance-
driver genes within ecDNAs. DD-DP, double drug (BRAFi + MEKi)-disease progression. E, Representative images of DNA-FISH showing indicated patient-
matched tumors (n = 3 pairs) and CNs of BRAF or NRAS in either vehicle-treated/MAPKi-sensitive (top) or acquired MAPKi-resistant (middle) melanoma.
CEN, centromere; DAPI, nuclear stain. Scale bars, 25 μm. Quantification of DNA-FISH (bottom) from 40 nuclei per tumor; results shown as mean ± SD.
P values (unpaired two-tailed Student t test): ***, P < 0.001. F, SV view of reconstructed amplicons in clinically acquired resistance showing XRCC2 as a
putative resistance-driving gene. (continued on following page)

of 19 patients included in this study whose tumors harbored of MAPK-reactivation genes were highly positively correlated
ecDNAs/CGRs, MAPK-reactivation genes were amplified via with mRNA level changes (Spearman correlation rs = 0.64,
ecDNAs/CGRs specifically in acquired-resistant tumors (but P = 0.001; Supplementary Table S5 and Supplementary
not in any MAPKi-sensitive/naive tumor) at a high frequency Fig. S1F). In a minority of patients (Pt5, ecDNA+CGR+;
(60% or 23 of 38 resistant tumors; Supplementary Table S3). Pt11, ecDNA−CGR−; Supplementary Table S6), we observed
Expectedly, MAPKi resistance–driver gene amplification lower-level but linear amplifications of BRAF (CN, 3–4)
via ecDNAs/CGRs often co-occurred with other resistance- specifically in acquired-resistant tumors. The BRAFV600MUT
specific CN alterations and somatic mutations reported melanoma tumors from both patients were treated with
earlier (Supplementary Table S4; refs. 7–9). Using tumor- only BRAFi (not BRAFi + MEKi), suggesting that low-level
matched RNA-sequencing (RNA-seq) data available from a linear amplifications suffice in driving acquired resistance to
subset of tumors, CN alterations (resistant versus sensitive) BRAFi monotherapy.

884 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

G H
BRAF CGR genes in
ecDNA genes in EGFR
Sensitivity only
Sensitivity only 1,000 Resistance only
400 Resistance only NRAS Sensitivity & resistance
Sensitivity & resistance 800 Resistance & ecDNA
Resistance & CGR genes in sensitivity
Number of genes

Number of genes
300 genes in sensitivity 600

400
200
NRAS 300
BRAF

100 200

BRAF RAC1 100 HRAS

NRAS BRAF
0 BRAF NRAS
0

1
2
3
4
5
6
7
AM Pt9
_P 01
el 27

M PD 1
_ 2
_ 3
el X4
X6
1
2
3
4
5
6
7

M M1 9
_P 01
el 27

M PD 1
_ 2
_ 3
el X4
X6

Pt
Pt
Pt
Pt
Pt
Pt
Pt

el X
el X
el X
Pt
Pt
Pt
Pt
Pt
Pt
Pt

A Pt

el X
el X
el X

el 12
M DX
M PD

M PD

M PD

D
el 2
M DX
M PD

M PD

M PD

_P
_P

_
_
_
_

R
R

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

I Signaling by BRAF and RAF1 fusions
Signaling by BRAF and RAF1 fusions

Mel_PDX27
Oncogenic MAPK signaling
Oncogenic MAPK signaling
Glycoside metabolic process
Acrosome assembly Pt3
Glycosyl compound metabolic process
Cellular component assembly involved in morphogenesis
Polyketide metabolic process
Secretory granular organization
0 2 4 6 8 0 2 4 6
–Log10 (P) –Log10 (P)

Keratinization NRF2 pathway

Mel_PDX1
Matrisome Potassium ion channel
Pt4

Peptide cross-linking Glutathione conjugation

Neutrophil degranulation Detoxification of nitrogen compound
GPCR ligand binding Chitin metabolic pathways
0 10 20 30 40 0 2 4 6 8
–Log10 (P) –Log10 (P )
Protein containing complex organization Response to interferon alpha
Reversible hydration of carbon dioxide

Mel_PDX4
Calcium activation phospholipid scrambling
Pt9

Organelle fission EPHB-mediated forward signaling

Nitrogen metabolism Class I PI3K signaling events
Triglyceride catabolism Establishment or maintenance of cell polarity
0.0 2.5 5.0 7.5 0 1 2 3 4 5
–Log10 (P) –Log10 (P)
T-cell receptor signaling pathway Peptide cross-linking
PI3K–AKT–mTOR-signaling pathway
RAM12.01

Mel_PDX6
Formation of the cornified envelope
Terminal pathway of complement Keratinization
Negative regulation of biosynthetic process GPCR signaling
Jak–STAT signaling pathway Cholesterol biosynthesis
0 2 4 6 0 5 10 15
–Log10 (P) –Log10 (P)

Figure 2. (Continued) G, Number of ecDNA-amplified genes (i) shared between pre/sensitive and post/acquired resistant tumors, (ii) specifically in
pre/sensitive tumors, (iii) specifically in post/acquired resistant tumors, and (iv) in post/acquired resistant tumors shared with CGR-amplified genes in
pre/sensitive tumors. Cohorts: clinical (n = 8 patients), RAM (n = 1 patient), and PDX models (n = 6 patients). Patients without sensitivity (n = 2) and with-
out ecDNAs and CGRs detected (n = 2) were excluded. H, As in G, except showing the number of CGR-amplified genes in the same (i) to (iii). (iv) Number of
CGR-amplified genes in post/acquired resistant tumors shared with ecDNA-amplified genes in pre/sensitive tumors. I, Pathway enrichment analysis of
genes amplified by ecDNAs and/or CGRs exclusively in acquired resistant tumors (n = 8). Only tumors with sufficient genes for enrichment analysis were
analyzed. GPCR, G protein-coupled receptor.

APRIL 2023 CANCER DISCOVERY | 885

RESEARCH ARTICLE Dharanipragada et al.

Nonhomologous end-joining (NHEJ) and homologous NRAS amplifications, respectively, harbored coamplification
recombination repair (HRR) of DSBs are thought to be of superenhancers (Supplementary Table S8), which suggests
critical for CGR and ecDNA generation after chromothripsis that genomic plasticity may facilitate epigenomic plasticity
(20). Intriguingly, we observed that XRCC2 (RAD51-like; CN, and alter gene expression on a large scale in MAPKi-resistant
49), a key HRR gene, and XRCC6 (KU70; CN, 14–15), a key tumors. In further support of ecDNAs playing a role in epige­
NHEJ gene, were amplified specifically in acquired MAPKi- nomic plasticity, we observed that H3K27 acetylated genomic
resistant melanomas as ecDNAs (Fig. 2F; Supplementary regions (extracted from ENCODE data) were rearranged to the
Table S3). Moreover, acquired MAPKi resistance was associ- proximity of and coamplified with ecDNA genes (e.g., c-MYC
ated with ecDNA or CGR amplicons spanning other DSB and BRAF, respectively) in acquired resistance (Supplementary
repair genes in NHEJ (TRIM33 CN, 7–11; PAXIP1 CN, 3) and Fig. S1H). Enhancer docking sites within ecDNA amplicons
HRR (SSBP1 CN, 5; BRCA2 CN, 6; RFC3 CN, 6; TRIM33 CN, may also influence the differentiation state(s) of MAPKi-
7–11; SYCP1 CN, 7–9, TRIM24 CN, 5–11) pathways (Sup- resistant tumors by acting in trans on intrachromosomal genes
plementary Table S3). Thus, acquired MAPKi resistance is such as HOXA9, HOXA11, HOXA13, and LIFR (RAM12.01-
specifically associated with CGR and ecDNA amplicons in Brain-DD-DP10, RAM12.01-Brain-DD-DP3, and RAM12.01-
MAPK-reactivation and DSB repair genes, which suggests Brain-DD-DP9) and acting as mobile regulatory elements
functional interplay as codrivers of resistance. for genes such as SMO (Pt3-DP1; Supplementary Table S8)
Other resistance-specific genes amplified by ecDNAs and/ and ATP1A1 (Mel_PDX2-R2; Supplementary Fig. S1G). Thus,

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

or CGRs may also contribute functionally to the resistant CGR and ecDNA amplicons coamplify MAPKi resistance–
phenotype. Therefore, we identified the genes and their CNs specific and –driver genes with their enhancers and may
in ecDNA (Fig. 2G) or CGR (Fig. 2H) amplicons specifically harbor additional enhancer or superenhancer activities in cis
associated with sensitivity, resistance, or both. Importantly, or trans that concomitantly reprogram the transcriptome of
genes amplified by either ecDNAs or CGRs specifically in resist- acquired resistance.
ance highly outnumbered those specific to sensitive tumors,
which indicates that gene amplification by ecDNAs and CGRs Pervasive Chromothriptic Genomic Spans in
contributes to disease progression on MAPKi therapy. More­ Melanoma Overlap with Genomic Coordinates
over, genes amplified by either ecDNAs or CGRs in both sensi- of ecDNAs and CGRs
tive plus resistant tumors constituted a very small fraction, Chromothripsis, defined as a mutational phenomenon
which is also consistent with the notion that gene amplifica- leading to extensive genomic rearrangements and extensive
tion by ecDNAs and CGRs contributes to disease progres- CN oscillations, drives cancer initiation and progression (12).
sion. We rarely detected overlapping genes in patient-matched Prior studies have colocalized chromothriptic regions with
sensitivity-associated CGRs and resistance-associated ecDNAs ecDNAs (20, 28–30), and a recent study suggested chro-
(and vice versa; Fig. 2G and H). To explore the functional mothripsis as a pathway for ecDNA formation in general
contributions of ecDNA- and CGR-amplified genes to the (30) and specifically in ecDNA-driven methotrexate resist-
acquired MAPKi-resistant phenotype, we performed pathway ance (20). To evaluate the specific contributions of chro-
enrichment analysis using genes amplified by ecDNAs/CGRs mothripsis before and after evolution of MAPKi resistance,
specifically detected in resistance from eight evaluable cases we analyzed WGS data (Supplementary Table S2) from 58
(see Methods; Fig. 2I). This analysis nominated alterations patient-matched, BRAFV600MUT or NRASMUT MAPKi-sensitive/
in MAPK (in Pt3, PDX27), PI3K–AKT (RAM12.01, PDX4), naive and acquired-resistant tumors [along with 19 patient-
immune (Pt4, RAM12.01, PDX4), G protein-coupled receptor matched normal genomic DNAs (gDNA)]. We observed a high
(GPCR; Pt4, PDX6), cellular differentiation/morphogenesis prevalence of chromothripsis (high-confidence calls based on
(Pt3, Pt4, Pt9, PDX4, PDX6), and metabolism (Pt9, RAM12.01, recently published criteria; ref. 12) in MAPKi-sensitive (16
PDX27, PDX1, PDX6) pathways as resistance phenotypes of 17; mean total span per genome, 464 Mb) and acquired-
driven by ecDNAs/CGRs (Supplementary Table S7). resistant (40 of 41; mean total span per genome, 476 Mb)
Recent studies have suggested that enhancers within melanomas. The chromosome distributions of chromoth-
ecDNAs influence oncogene expression by either coamplifi- ripsis overlapped but were distinct between sensitive and
cation with target oncogenes within the ecDNAs (i.e., cis inter- acquired-resistant genomes (Supplementary Fig. S2A and
action) or regulation of intrachromosomal genes (i.e., trans S2B). Chromothriptic genomic spans in MAPKi-sensitive/
interaction; refs. 16, 25, 26). Thus, we annotated CGR and naive and acquired MAPKi-resistant melanomas overlap 76%
ecDNA amplicons in both sensitive and acquired resistant and 36%, respectively, with those reported earlier for a subset
genomes with enhancers listed in GeneHancer (27). Notably, of The Cancer Genome Atlas skin cutaneous melanoma (12),
MAPKi-resistance genes (e.g., BRAF, NRAS, HRAS, and EGFR) indicating again that MAPKi selects for chromothripsis in
and known oncogenes (e.g., EZH2, CREB3L2, CARD11, and distinct genomic regions. Interestingly, the single MAPKi-
EP300) were coamplified with their corresponding enhanc- sensitive melanoma without chromothripsis (Mel_PDX27-V1)
ers within CGRs and ecDNAs of MAPKi-resistant genomes still gave rise to two (of three) acquired-resistant tumors with
(Supplementary Fig. S1G and Supplementary Table S8). We chromothripsis. The recurrence and span sizes of chromo-
also observed enhancers and associated DSB repair genes, thriptic events in our melanoma cohorts are higher than those
such as XRCC2, XRCC6, and RAD21, within close proximity reported for cutaneous melanoma in the Pan-Cancer Analysis
(−0.5 kb to +1.1 kb; Supplementary Fig. S1G and Supplemen- of Whole Genomes study (mean total span per genome, 120 Mb;
tary Table S8). Furthermore, 10 of 11 and 4 of 5 acquired- ref. 12). These differences may be due to advancing disease and
resistant genomes with CGR/ecDNA-associated BRAF and therapy resistance.

886 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

If chromothripsis were a precursor step for ecDNA and acquired-resistant versus MAPKi-naive/sensitive tumors (Wil-
CGR generation during melanoma disease progression, we coxon rank sum test, P = 0.005; Supplementary Fig. S2C), sug-
expected a nonrandom overlap of affected genomic spans. gesting that MAPKi-selected chromothriptic regions in some
Consistent with expectation, we observed that the genomic individuals feature a relatively stronger mutator phenotype.
spans of ecDNAs and CGRs overlapped significantly or non- We have reported that MAPKi selection of clinical melano-
randomly with chromothriptic regions in ∼29% (5 of 17) mas alters the mutational spectra (9). To characterize how a
of MAPKi-sensitive/naive genomes and in ∼54% (22 of 41) MAPKi-associated mutator phenotype affects chromothrip-
of acquired MAPKi-resistant genomes (Fig. 3A). For every tic genomic spans selected by MAPKi, we culled chromothrip-
genome, we computed the sizes of ecDNAs + CGRs (a − x), sis-associated somatic mutations unique to MAPKi-sensitive/
chromothriptic regions (b − x), their overlap (x), and genomic naive (mean, 64 SNVs/Mb) versus MAPKi-resistant genomes
regions devoid of these genomic alterations (g − a − b). These (mean, 58 SNVs/Mb) and analyzed signatures of single-
four quantities were represented in 2 × 2 contingency tables base substitutions (SBS; Fig. 3B), double-base substitutions
to carry out the Fisher exact test. Using hypergeometric (DBS), and IDs (Supplementary Fig. S2D; 32, 33). In both
distribution, we then estimated the probability of observ- MAPKi-sensitive/naive and acquired-resistant tumors, we
ing random overlaps between ecDNA + CGR spans and frequently detected signatures of tobacco smoking (SBS4)
chromothriptic spans. In each tumor genome with over- and ultraviolet radiation (SBS7a and SBS38). Among chro-
laps between ecDNA + CGR and chromothriptic regions, we mothripsis-associated somatic mutations unique to either

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

observed a significant nonrandom convergence (Fisher exact MAPKi-sensitive/naive or acquired-resistant tumors, we
test, P < 0.00001), suggesting chromothripsis as an origin of also detected a signature of defective HRR (SBS3), and this
ecDNAs and CGRs (Fig. 3A). trended higher in resistant (12 of 31) versus sensitive (3 of
16) tumors, although this difference was not significant.
Tumor Mutational Burdens and Single-Base Unique but infrequent to acquired MAPKi resistance was the
Substitution Signatures of Chromothripsis, detection of mutational signatures of polymerase eta somatic
ecDNAs, and CGRs hypermutation (SBS9) and defective POLD1 proofreading
Chromothripsis generation is hypothesized to involve activity (SBS10d, SBS20). Unique and recurrent in acquired
micronuclei, which expose entrapped chromosomes to muta- MAPKi resistance was the detection of mutational signatures
gens and predispose them to replication, transcription, as of (i) defective DNA mismatch repair (MMR; SBS6, SBS20,
well as DNA repair defects (31). Therefore, we identified SBS26, and SBS44; 14 of 31 resistant tumors; 10 of 16
and characterized the somatic SNVs within the chromo- patients) and (ii) defective base excision repair (BER) as well
thriptic regions across the melanoma genomes in our three as damage due to reactive oxygen species (ROS) or mutations
cohorts. The numbers of chromothripsis-associated SNVs in NTHL1 and MUTYH (SBS18, SBS30, and SBS36). As excep-
ranged from 31 to 2,526 per Mb (mean 154 mutations/Mb, tions, SBS26 and SBS36 signatures were also detected at a low
median 96 mutations/Mb; Supplementary Table S2). We next percentage in two sensitive tumors (one of which had been
tested the hypothesis that chromothripsis may dispropor- exposed to MAPKi). BER/ROS signatures were observed more
tionally contribute to the tumor mutational burdens (TMB), frequently in the BRAFV600MUT (10 of 22 resistant tumors, 6 of
especially in acquired-resistant genomes, by calculating the 11 patients) than in the NRASMUT (1 of 9 resistant tumors, 1
numbers of SNVs within chromothriptic regions relative to of 5 patients) subset. This differential frequency may be due
the numbers of SNVs within nonchromothriptic regions. to the longer MAPKi exposure in the clinical (versus PDX
Acquired-resistant tumors harbored an average of 195 SNVs/ or experimental) setting. Consistently, the BER mutational
Mb in chromothriptic regions versus an average of 93 SNVs/ signature (SBS18) was detected in Mel_PDX3-R5, which was
Mb in nonchromothriptic regions. Because the ratio of the among the acquired-resistant PDX tumors with longer dura-
chromothriptic region (∼476 Mb) to nonchromothriptic tions of MAPKi treatment (Supplementary Fig. S1A). Overall,
region (∼2,619 Mb) is 0.18 and given the TMB of 170 SNVs/ resistance-specific (versus sensitivity-specific) chromothriptic
Mb in acquired resistance, we expected only 31 SNVs/Mb somatic SBSs enriched for signatures of defects in BER (Wil-
(0.18 × 170) in chromothriptic regions of acquired-resistant coxon rank sum test, P = 0.04) and MMR (Wilcoxon rank sum
tumors if SNVs were evenly distributed across the genome. test, P = 0.005). Moreover, resistance-specific (versus sensitiv-
Moreover, MAPKi-sensitive or pretreatment tumors harbored ity-specific) nonchromothriptic SBSs also enriched for signa-
an average of 121 SNVs/Mb in chromothriptic regions versus tures of defects in BER (Wilcoxon rank sum test, P = 0.05),
an average of 84 SNVs/Mb in nonchromothriptic regions. MMR (Wilcoxon rank sum test, P = 0.008), as well as HRR
Again, an observed average of 121 SNVs/Mb in chromo- (Wilcoxon rank sum test, P = 0.04; Supplementary Table S9).
thriptic regions was higher than the expected 20 SNVs/Mb We did not identify any resistance-specific or -enriched DBS
in the chromothripsis regions, which was calculated based and ID signature (Supplementary Fig. S2D).
on a 0.17 ratio of the chromothriptic region (∼464 Mb) to We also identified SBS signatures within ecDNA and CGR
nonchromothriptic region (∼2631 Mb) and the observed sequences and determined whether certain SBS signatures
average TMB of 115 SNVs/Mb in MAPKi-sensitive tumors. are enriched in acquired MAPKi-resistant (versus MAPKi-
Hence, in both acquired-resistant and MAPKi-sensitive mela- sensitive/naive) tumors or ecDNA/CGR amplicons (versus
noma genomes, chromothriptic regions were enriched for noninvolved genomic regions regardless of tumor sensitivity
SNVs compared with nonchromothriptic regions (Fisher or resistance; Fig. 3C). We extracted somatic SNVs unique
exact test, P < 0.00001). Furthermore, the ratio of SNVs in to MAPKi-sensitive/naive (n = 12) and acquired-resistant
chromothriptic to nonchromothriptic genomes was higher in (n = 28) tumors (in a patient-matched fashion) within

APRIL 2023 CANCER DISCOVERY | 887

RESEARCH ARTICLE Dharanipragada et al.

A
Mel_PDX27-V1 Chromothripsis
Mel_PDX2-V1
Pt3-Baseline ecDNAs + CGRs
Pt7-Baseline overlapping
Pt10-Baseline chromothripsis
Mel_PDX1-V1 *** Nonoverlapping
Pt2-Baseline
ecDNAs + CGRs
Sensitive

Mel_PDX3-V1
Pt11-Baseline
Pt5-Baseline
Mel_PDX6-V1 ***
Pt6-Baseline
Pt4-Baseline ***
Pt1-Baseline
Pt9-Baseline ***
RAM12.01-Sensitive
Mel_PDX4-V1 ***
Mel_PDX27-R1

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

Pt2-DP1
Mel_PDX27-R3
Pt1-DP3 ***
Mel_PDX27-R2 ***
Pt1-DP1
RAM12.01-Brain-DD-DP10 ***
Mel_PDX1-R2 ***
Pt3-DP1 ***
RAM14006-Lung-DD-DP
Pt6-DP2 ***
Mel_PDX3-R5
Mel_PDX4-R2 ***
Pt10-DP1
Pt5-DP1
Mel_PDX1-R1 ***
Pt4-DD-DP-3C ***
RAM12.01-Brain-DD-DP3 ***
RAM14006-Brain-DD-DP
Resistant

Mel_PDX6-R3
RAM14006-Lymph1-DD-DP
RAM14006-Liver-DD-DP
Pt5-DP2
Pt11-DP2
Pt4-DD-DP-2B ***
RAM13003-Brain-DD-DP ***
Pt9-DD-DP2 ***
RAM14006-Adrenal-DD-DP
Pt11-DP1
RAM13003-Lung-DD-DP ***
Pt9-DP2 ***
Pt7-DP1
Mel_PDX4-R5 ***
Mel_PDX2-R2 ***
RAM14006-Lymph2-DD-DP
Mel_PDX4-R4 ***
Mel_PDX3-R4 ***
Pt4-DD-DP-1A ***
RAM12.01-Brain-DD-DP9 ***
RAM14006-Muscle-DD-DP
Pt9-DD-DP1 ***
0 250 500 750
Size of region (Mb)

Figure 3. Associations of chromothripsis, ecDNAs, and CGRs with mutagenic and double-stranded DNA repair pathways. A, Pattern of overlap or
nonoverlap between ecDNA + CGR and chromothriptic genomic spans within individual sensitive (n = 17; top) and acquired resistant (n = 41; bottom) mel-
anoma tumors. Sizes (Mb) of chromothriptic regions are in cyan, ecDNAs + CGRs overlapping with chromothripsis in red, and ecDNAs + CGRs nonoverlap-
ping with chromothripsis in purple. P value (Fisher exact test) of nonrandom overlap between chromothriptic and ecDNA + CGR regions: ***, P < 0.00001.
(continued on following page)

888 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

B Cohorts
Clinical
Rapid autopsy
1.00 PDX
Proportion of mutational signatures

Genotype
BRAF MUT
NRAS MUT
0.75
Sensitive
Acquired
resistant
0.50 Signatures
SBS3
SBS4
SBS5
0.25 SBS6
SBS7a
SBS9
SBS10d
0.00 SBS18
SBS20
Pt1-Baseline
Pt2-Baseline
Pt3-Baseline
Pt4-Baseline
Pt5-Baseline
Pt6-Baseline
Pt7-Baseline
Pt9-Baseline
Pt10-Baseline
Pt11-Baseline
RAM12.01-Sensitive
Mel_PDX1-V1
Mel_PDX2-V1
Mel_PDX3-V1
Mel_PDX4-V1
Mel_PDX6-V1

Pt1-DP1
Pt1-DP3
Pt2-DP1
Pt3-DP1
Pt4-DD-DP-1A
Pt4-DD-DP-2B
Pt4-DD-DP-3C
Pt5-DP1
Pt5-DP2
Pt6-DP2
Pt7-DP1
Pt9-DP2
Pt9-DD-DP1
Pt9-DD-DP2
Pt10-DP1
Pt11-DP1
Pt11-DP2
RAM12.01-DD-DP10
RAM12.01-DD-DP3
RAM12.01-DD-DP9
Mel_PDX27-R2
Mel_PDX27-R3
Mel_PDX1-R1
Mel_PDX1-R2
Mel_PDX2-R2
Mel_PDX3-R4
Mel_PDX3-R5
Mel_PDX4-R2
Mel_PDX4-R4
Mel_PDX4-R5
Mel_PDX6-R3
SBS25
SBS26
SBS30
SBS32

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

SBS36
SBS38
SBS44

C
Signatures Cohorts
SBS2 Rapid autopsy
SBS3 Clinical
1.00 SBS4
SBS5 PDX
Proportion of mutational signatures

SBS6
SBS7a Genotype
0.75 SBS7b BRAF MUT
SBS9
SBS10b NRAS MUT
SBS10c
0.50 SBS25 Sensitive
SBS26 Acquired
SBS30 resistant
SBS38
0.25 SBS44
SBS86
SBS87
SBS92
0.00
SBS2 Signature score
SBS3 2.0
SBS4
SBS5 1.5
SBS6 1.0
SBS7a
Mutational signatures

SBS7b 0.5
SBS9
SBS10b
SBS10c
SBS25
SBS26
SBS30
SBS38
SBS44
SBS86
SBS87
SBS92
Pt2-Baseline
Pt3-Baseline
Pt4-Baseline
Pt5-Baseline
Pt6-Baseline
Pt7-Baseline
Pt9-Baseline
Mel_PDX27-V1
Mel_PDX1-V1
Mel_PDX2-V1
Mel_PDX4-V1
Mel_PDX6-V1
Pt1-DP1
Pt1-DP3
Pt2-DP1
Pt3-DP1
Pt4-DD-DP1A
Pt4-DD-DP2B
Pt4-DD-DP3C
Pt5-DP1
Pt5-DP2
Pt6-DP2
Pt7-DP1
Pt9-DD-DP1
Pt9-DD-DP2
Pt9-DP2
RAM1201-DD-DP3
RAM1201-DD-DP9
RAM1201-DD-DP10
Mel_PDX27-R1
Mel_PDX27-R2
Mel_PDX27-R3
Mel_PDX1-R1
Mel_PDX1-R2
Mel_PDX2-R2
Mel_PDX3-R4
Mel_PDX4-R2
Mel_PDX4-R4
Mel_PDX4-R5
Mel_PDX6-R3

Figure 3. (Continued) B, SBS signatures of chromothriptic genomes of MAPKi-sensitive/naive (left, n = 16) versus MAPKi-resistant (right, n = 31)
melanoma tumors. Excluded from analysis, acquired resistant tumors without patient-matched sensitive tumors and tumors without chromothripsis.
C, Top, SBS signatures of ecDNAs + CGRs of MAPKi-sensitive/naive (left, n = 12) versus MAPKi-resistant (right, n = 28) melanoma tumors. Bottom, SBS
signature enrichment scores are shown by both heat scale and dot size. Increasing dot sizes indicate increasing enrichment of indicated signatures
within ecDNAs + CGRs compared with regions devoid of these events (score > 1). Enrichment score = 1 is considered the cutoff; only scores >1 and <1
are shown. Excluded from analysis are acquired-resistant tumors without patient-matched sensitive tumors and tumors without ecDNAs and CGRs.
(continued on next page)

APRIL 2023 CANCER DISCOVERY | 889

RESEARCH ARTICLE Dharanipragada et al.

D E 100 100
80 80

% ecDNA + CGR breakpoints

60

HRAS
BRAF
100 60
% ecDNA + CGR breakpoints

Sensitive 40
Acquired 40

75 resistance 20 20
0 0
*** 100 100
50 80
80

NRAS

EGFR
60 60
25
40 40
20 20
0
0 0
NHEJ Alt-NHEJ HRR NHEJ Alt-NHEJ HRR NHEJ Alt-NHEJ HRR

Figure 3. (Continued) D, Breakpoint–junctional sequence analysis of ecDNAs + CGRs inferring indicated DNA DSB repair processes: NHEJ, alternative
NHEJ (alt-NHEJ), and HRR. Amplicons from all three cohorts of tumors combined for analysis (sensitive, n = 17; resistant, n = 41). Homologous sequences

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

of size 0–1 bp (NHEJ), 2–8 bp (alt-NHEJ), and >8 bp and large insertions (HRR). ***, P value (Kruskal–Wallis test) = 6.35e−16. E, As in D, except only for
resistant tumor–derived ecDNA and CGR amplicons that harbor BRAF (n = 11), NRAS (n = 5), HRAS (n = 2), and EGFR (n = 4) genes.

ecDNA/CGR amplicons. CGR + ecDNA-associated TMB kataegis in ecDNAs (33). We also observed positive enrich-
trended higher in acquired-resistant (mean, 130 SNVs/Mb) ment scores of SBS3 (defective HRR) in 47% (9 of 19) of
compared with MAPKi-sensitive/naive (mean, 80 SNVs/Mb) ecDNA+ acquired-resistant tumors compared with 17% (4 of
tumors (Wilcoxon rank sum test, P = 0.06). Enrichment of 24) of CGR+ acquired-resistant tumors (Wilcoxon rank sum
SBS3 (defective HRR) was more recurrent in the ecDNAs/ test, P = 0.07; Supplementary Fig. S2F). In short, ecDNA
CGRs of resistant tumors (11 of 28) compared with patient- and CGR amplicons, whether in MAPKi-sensitive/naive or
matched sensitive tumors (1 of 12), although this difference acquired-resistant tumors, harbor SBS patterns sugges-
did not reach statistical significance (Wilcoxon rank sum tive of specific defects or deficiencies in HRR. In acquired
test, P = 0.08; Fig. 3C). Moreover, we derived mutational MAPKi resistance, APOBEC activity likely contributes to
signature enrichment scores by calculating the normalized ecDNA mutagenesis and hence shapes the mutanome of
ratios of signature proportions within ecDNAs/CGRs and ecDNA amplicons. Further analysis might lead to insights
within regions devoid of these events (scores >1 defined as into ecDNA biogenesis and suppressive strategies.
positive enrichment; see Methods). We found that 25% (7 of
28) of acquired-resistant tumors displayed positive enrich- NHEJ Underlies the Formation of ecDNAs
ment scores for defective HRR signatures (SBS3) within and CGRs
ecDNAs and CGRs, whereas only 8% (1 of 12) of sensitive We analyzed the breakpoint–junctional sequences of CGRs
tumors displayed positive enrichment (Wilcoxon rank sum and ecDNAs to infer DSB repair processes underlying resistance-
test, P = 0.3; Fig. 3C). In either sensitive or resistant tumors, associated amplicons (Fig. 3D). Alternative end-joining refers
we observed positive enrichment scores for SBS signatures to mechanisms of DSB repair that may compensate for
reflective of APOBEC cytidine deaminase activity (SBS2; HRR- and NHEJ-based repairs and comprises single-strand
2 of 12 in sensitivity; 3 of 28 in resistance) and polymer- annealing (SSA), microhomology-mediated end-joining
ase epsilon exonuclease domain mutations (SBS10b; 3 of (MMEJ), and other end-joining pathways (34). SSA is indi-
12 in sensitivity; 3 of 28 in resistance). Positive enrich- cated in the breakpoint junctions by complementary repeat
ment scores were also noted with lower recurrence for sequences >25 nucleotides and MMEJ by shorter tracks of
SBS signatures of defective POLD1 proofreading (SBS10c), sequence homology (2–20 nucleotides). Moreover, replicative
defective MMR (SBS44), chemotherapy treatment (SBS86, processes, such as fork stalling and template switching as well
SBS87), and tobacco smoking (SBS82). Overall, 50% (6 of as MMEJ, can contribute to the generation of CGRs. Break-
12) of sensitive/naive and 32% (9 of 28) of acquired-resistant point junctions derived from replicative processes (e.g., repli-
tumors displayed enrichment of unique SBS signatures cative fork collapsing when it encounters a nick) are expected
within ecDNA + CGR sequences compared with uninvolved to have microhomologies, insertions, and relatively long tem-
genomic regions. Lastly, we addressed whether the enrich- plated insertions (35, 36). Indeed, signatures of replication
ment of unique SBS signatures differed within ecDNAs ver- processes and templated insertion, as well as that of NHEJ,
sus CGRs in acquired resistance (Supplementary Fig. S2E). were detected pan-cancer (12). Analysis of breakpoint–junc-
We observed positive enrichment scores of SBS2 (APOBEC tional sequences of all resistance- and sensitivity-associated
cytidine deaminase activity) in 26% (5 of 19) of ecDNA+ CGRs and ecDNAs inferred NHEJ as the main mechanism
acquired-resistant tumors compared with 4% (1 of 24) of of double-stranded DNA fragment joining or rearrange-
CGR+ acquired-resistant tumor (Wilcoxon rank sum test, ment (Fig. 3D; Supplementary Fig. S3A). A lower number of
P = 0.05; Supplementary Fig. S2F). This finding is consistent breakpoint–junctional sequences displayed short and long
with the recent discovery of the co-occurrence of APOBEC3 homologous sequences as well as insertions (>10 bp), which

890 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

suggested DSB repair by alternative NHEJ (alt-NHEJ) and we used next-generation DNA-PKi (VX984 and AZD7648)
HRR, respectively (Fig. 3D). Analysis of breakpoint–junctional and PARPi (olaparib) and corroborated their combinatorial
sequences of specifically ecDNAs and CGRs harboring MAPK- efficacy with MAPKi in parental cell lines (Supplementary
reactivation or MAPKi resistance–driver genes also revealed a Fig. S4C–S4E and Supplementary Fig. S5A–S5F). In M245,
similar pattern, with NHEJ dominating the landscape of DSB we further corroborated the pharmacologic findings with
repair mechanisms (Fig. 3E). Of unknown significance or genetic studies. Using independent short hairpin RNAs
etiology, a higher incidence of double-stranded DNA ligation (shRNA) against two critical NHEJ genes (PRKDC, which
by HRR was detected in NRASMUT melanoma, either sensitive encodes DNA-PKCS, and DNA ligase IV, or LIG4), we showed
or acquired-resistant to MAPKi, compared with BRAFV600MUT that their protein knockdown, while not having a significant
melanoma (Supplementary Fig. S3B). In acquired-resistant effect on growth without MAPKi, strongly suppressed the
genomes, we also detected a higher incidence of double- frequency of DTPP emergence (Fig. 4P and Q).
stranded DNA ligation by NHEJ for CGRs compared with The acquired-resistant sublines used in the prior experi-
ecDNAs (Supplementary Fig. S3C). ments have been adapted chronically (i.e., months to years)
to MAPKi. Hence, we also tested whether a shorter delay (i.e.,
Inhibitors of DNA-PKCS and/or PARP1/2 Prevent weeks) in the cotreatment of parental cell lines with DNA-PKi
Acquired MAPKi Resistance in Melanoma Cell (and/or PARPi) would hinder the efficacy in preventing DTPP
Lines formation. In both cell lines (M229 BRAFV600MUT and M245

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

DNA-dependent protein kinase catalytic subunit (DNA- NRASMUT), we found that a 3- to 3.5-week delay in cotreat-
PKCS) and PARP1/2 are involved in multiple DSB repair ment with DNA-PKi and/or PARPi reduced the suppression
pathways, particularly NHEJ (DNA-PKCS), HRR (DNA- of resistance (Fig. 4R and S compared with Fig. 4A and J,
PKCS, PARP1/2), and MMEJ (PARP1/2; refs. 37, 38). Thus, respectively). In the same parental cell lines, we then com-
we tested the activity of the specific DNA-PKCS inhibi- pared cotreatment with DNA-PKi (and/or PARPi) during the
tor (DNA-PKi) NU7026 and PARP1/2 inhibitor (PARPi) first versus second half of the overall MAPKi treatment course
ABT888 (veliparib), individually and combinatorially, in (Fig. 4T and U). Notably, we observed that cotreatment with
preventing the clonal emergence of drug-tolerant proliferat- DNA-PKi, PARPi, or DNA-PKi + PARPi during the first half
ing persisters (DTPP; refs. 39, 40) from human BRAFV600MUT of the MAPKi treatment course was remittive and during
(n = 3; M229, M249, M395) or NRASQ61MUT (n = 3; M202, the second half inferior (compared with continuous cotreat-
M207, M245; Fig. 4) melanoma cell lines chronically treated ment). Thus, the upfront and initial phase of MAPKi therapy
with BRAFi + MEKi or MEKi, respectively. We hypothesized appears to be the key window of opportunity for cotreatment
that DNA-PKi interferes with MAPKi-elicited, de novo rear- with DNA-PKi (and/or PARPi) to suppress acquired MAPKi
rangement of specific SVs including ecDNAs and CGRs. resistance, suggesting a preventive mechanism of action.
We further hypothesized that DNA-PKi is more effective To support further a causal link between acquired MAPKi
at preventing rather than reversing resistance once fully resistance and ecDNAs or homogeneously staining regions
established by chronic MAPKi treatment. Hence, we tested (HSR) in human melanoma cell lines, we first sought to
treatments with NU7026 and/or ABT888 in combination determine the modes of amplification of bona fide resistance-
with MAPKi in acquired MAPKi-resistant sublines (n = 9) driver genes. In three parental cell lines (BRAFV600MUT, M249
that are isogenic to the parental human BRAFV600MUT and and M395; NRASMUT, M245) and five isogenic acquired
NRASQ61MUT cell lines (Fig. 4A–O). BRAFV600MUT sublines MAPKi-resistant sublines, we performed DNA-FISH on cells
with acquired-resistance to BRAFi + MEKi are annotated in metaphase and probed against BRAF (amplification of
as double-drug resistant (DDR), whereas NRASMUT sub- which drives acquired MAPKi resistance in BRAFV600MUT mel-
lines with acquired-resistance to MEKi are annotated sim- anoma; refs. 5, 7–9), RAF1, and NRAS (amplification of which
ply by their clone (C) numbers. We began by testing the drives acquired MAPKi resistance in NRASMUT melanoma;
single-agent anticlonogenic growth activity of DNA-PKi refs. 11, 22), as well as centromeres of chr7, chr3, and chr1,
and PARPi on parental melanoma (and nonmelanoma, see respectively (Fig. 5A–C). We observed in acquired resistance
below) cell lines without MAPKi treatment in order to select that BRAF is amplified either as HSRs (M249 DDR4 and
noninhibitory concentrations to test in combination with DDR5) or as ecDNAs (M395 DDR), RAF1 as a mixture of
MAPKi (Supplementary Fig. S4A–S4G). In combination HSRs and ecDNAs (M245 C3), and NRAS as HSRs (M245
with MAPKi, NU7026 synergistically and dose dependently C5). Expectedly, these driver ecDNAs and HSRs were not
prevented DTPP formation in all parental cell lines tested detected in the isogenic parental cell lines (Fig. 5A–C). We
(Fig. 4A, D, G, I, J, and N), whereas ABT888 displayed then tested whether MAPKi withdrawal would select against
activity in 3 of 6 parental cell lines (Fig. 4A, D, and J). In resistant clones with ecDNAs and/or HSRs that amplify the
cases in which ABT888 individually was active in prevent- aforementioned resistance-driver genes and, consequently,
ing DTPP formation, NU7026 plus ABT888 led to even restore at least partial MAPKi sensitivity. Importantly, we
greater suppression of acquired MAPKi resistance (Fig. 4A, observed that MAPKi withdrawal from acquired-resistant
D, and J). Consistently in all acquired MAPKi-resistant sub- sublines significantly reduced metaphase cells with driver
lines, NU7026 and ABT888 at the lower range of the con- ecDNAs and HSRs (Fig. 5A–C) and significantly enhanced
centrations tested displayed no or reduced anticlonogenic their MAPKi sensitivity (Fig. 5D–F). Using the M249 paren-
activity compared with their activities observed in isogenic tal cell line, which is highly sensitive to resistance suppres-
parental lines (Fig. 4A–O). More recent generations of DNA- sion by either DNA-PKi or PARPi (Fig. 4D), we tracked the
PKi display improved selectivity and potency (37, 41). Hence, emergence of BRAF HSR early on treatment (day 31) with

APRIL 2023 CANCER DISCOVERY | 891

RESEARCH ARTICLE Dharanipragada et al.

A M229 (BRAF V600MUT) B M229 DDR1 (BRAF V600MUT)

BRAFi + MEKi BRAFi + MEKi
NU7026 (μmol/L) 0 2 4 6 8 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 2 4 8 10 12

NU7026 (μmol/L) 0 2 4 8 6 10 NU7026 (μmol/L) 0 2 4 8 6 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 2 4 8 10 12

1.5 1.5
Relative % area

Relative % area
**
1.0 1.0
**
***

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

*
0.5 *** *** *** ***
0.5 ***
*** *** *** *** *** *** *** ***
*** ***
0.0 *** *** *** 0.0
(μmol/L) 0 2 4 6 8 10 0 2 4 8 10 12 0 + 2 + 4 + 8 10 12 (μmol/L) 0 2 4 6 8 10 0 2 4 8 10 12 0 + 2 + 4 + 8 10 12
2 4 8 6+0+ 2 4 8 6+0+
1 1
NU7026 ABT888 NU7026 + ABT888 NU7026 ABT888 NU7026 + ABT888
BRAFi + MEKi BRAFi + MEKi

C M229 DDR4 (BRAF V600MUT) D M249 (BRAF V600MUT)

BRAFi + MEKi BRAFi + MEKi
NU7026 (μmol/L) 0 2 4 6 8 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 0.5 1 2 3 4

NU7026 (μmol/L) 0 2 4 8 6 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 0.5 1 2 3 4

1.5
Relative % area

Relative % area

1.0
1.0 **
** ***
*** *** 0.5
0.5 ***
***
*** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** ***
0.0 0.0
(μmol/L) 0 2 4 6 8 10 0 2 4 8 10 12 0 2 4 8 10 12 (μmol/L) 0 2 4 6 8 10 0 .5 1 2 3 4 0 .5 1 2 3 4
0 0 + + + +
2 + 4 + 8 +6 + 0 + 2 + 4 6 8 10
1
NU7026 ABT888 NU7026 + ABT888 NU7026 ABT888 NU7026 + ABT888
BRAFi + MEKi BRAFi + MEKi

Figure 4. DNA-PKi and/or PARPi cotreatment prevents acquired MAPKi resistance in human melanoma cell lines. A–O, Long-term clonogenic growth
of isogenic parental and acquired MAPKi-resistant BRAFV600MUT (A–H) or NRASQ61MUT (I–O) human melanoma cell lines showing acquired resistant colonies
to BRAFi + MEKi (PLX4032 at 0.5 μmol/L, AZD6244 at 0.5 μmol/L in A and G; PLX4032 at 0.25 μmol/L, AZD6244 at 0.25 μmol/L in D; PLX4032 at
1 μmol/L, AZD6244 at 1 μmol/L in B, C, E, F, and H) or MEKi (trametinib at 0.005 μmol/L in I and 0.01 μmol/L in J and N; trametinib at 0.1 μmol/L in K, L, M,
and O) and their suppression by indicated cotreatments with DNA-PKi (NU7026) and/or PARPi (ABT888) at indicated concentrations. Top, representative
cultures; bottom, quantifications over n = 4 fields (mean ± SD). Data representative of 2 to 4 independent repeats. Seeding densities (cells/well in 6-well
dishes) and culture durations (days): 5,000; 31 (A), 1,000; 17 (B), 5,000; 18 (C), and 40,000; 40 (D). (continued on following page)

BRAFi + MEKi or combinations with DNA-PKi, PARPi, or treatments with DNA-PKi and/or PARPi. During the earliest
DNA-PKi + PARPi (Fig. 5G). Consistent with BRAF HSRs as a window on MAPKi treatment, expansion of ecDNA- and
driver of acquired MAPKi resistance, we observed their induc- CGR-involved genomic spans may be critical for specific
tion by BRAFi + MEKi but suppression by combinatorial resistance-driving ecDNAs and CGRs to emerge later. We

892 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

E M249 DDR4 (BRAF V600MUT) F M249 DDR5 (BRAF V600MUT)

BRAFi + MEKi BRAFi + MEKi
NU7026 (μmol/L) 0 2 4 6 8 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 0.5 1 2 3 4 ABT888 (μmol/L) 0 0.5 1 2 3 4

NU7026 (μmol/L) 0 2 4 6 8 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 0.5 1 2 3 4 ABT888 (μmol/L) 0 0.5 1 2 3 4
Relative % area

Relative % area
1.0 * 1.0 *
*** ***

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

0.5 *** 0.5 *** ***
*** *** ***
*** ***
0.0 0.0
(μmol/L) 0 2 4 6 8 10 0 0.5 1 2 3 4 0 0.5 + 1 + 2 + 3 + 4 (μmol/L) 0 2 4 6 8 10 0 0.5 1 2 3 4 0 0.5 + 1 + 2 + 3 + 4
2 + 4 6 8 10 2 + 4 6 8 10
NU7026 ABT888 NU7026 + ABT888 NU7026 ABT888 NU7026 + ABT888
BRAFi + MEKi BRAFi + MEKi

G M395 (BRAF V600MUT) H M395 DDR (BRAF V600MUT)

BRAFi + MEKi BRAFi + MEKi
NU7026 (μmol/L) 0 2 4 6 8 10 NU7026 (μmol/L) 0 2 4 6 8 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 2 4 8 10 12

NU7026 (μmol/L) 0 2 4 8 6 10 NU7026 (μmol/L) 0 2 4 8 6 10

ABT888 (μmol/L) 0 2 4 8 10 12 ABT888 (μmol/L) 0 2 4 8 10 12
Relative % area

Relative % area

1.0
1.0 ** ***
** * ** *** *** *** *** *** ***
*** *** ***
0.5 0.5 ***
*** ***
*** *** *** *** *** *** ***
***
0.0 0.0
(μmol/L) 0 2 4 6 8 10 0 2 4 8 10 12 0 2 4 8 10 12 (μmol/L) 0 2 4 6 8 10 0 2 4 8 10 12 0 +2 4 8 10 12
2 + 4 + 8 +6 + 0 + 2 4 + 8 +6 + 0 +
1 1
NU7026 ABT888 NU7026 + ABT888 NU7026 ABT888 NU7026 + ABT888
BRAFi + MEKi BRAFi + MEKi

Figure 4. (Continued) Seeding densities (cells/well in 6-well dishes) and culture durations (days): 5,000; 14 (E), 5,000; 12 (F), 20,000; 30 (G), and
20,000; 21 (H). (continued on next page)

reasoned that DNA-PKi, which is generally more effective ecDNA and CGR spans detected in vehicle-treated genomes;
than PARPi in preventing acquired resistance (Fig. 4), should Supplementary Table S10). Consistent with our hypothesis,
suppress MAPKi-elicited ecDNA + CGR genomic spans. To DNA-PKi cotreatment with MAPKi blunted the expansion
test this hypothesis, we performed WGS on two parental of ecDNA + CGR genomic spans in both cell lines (Fig. 5H).
cell lines (M229 and M245) treated with vehicle, MAPKi, Compared with treatment with MEKi alone, M245 cells
or MAPKi + DNA-PKi for a short duration (10–11 days); treated with MEKi + DNA-PKi gained de novo ecDNAs (0.06
identified ecDNAs and CGRs; and calculated treatment- Mb in chr3) and lost CGRs (0.4 Mb in chr1 and chr2;
specific genomic spans of ecDNAs + CGRs (by filtering out Fig. 5H).

APRIL 2023 CANCER DISCOVERY | 893

RESEARCH ARTICLE Dharanipragada et al.

I M202 (NRAS Q61MUT) J M245 (NRAS Q61MUT) K M245 C3 (NRAS Q61MUT)

MEKi MEKi MEKi
NU7026 0 2 5 10 NU7026 0 5 10 15 NU7026 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)

ABT888 0 1 2 5 ABT888 0 1 2 5 ABT888 0 1 2 5

(μmol/L) (μmol/L) (μmol/L)

NU7026 NU7026 NU7026

(μmol/L) 0 2 5 10 (μmol/L) 0 5 10 15 (μmol/L) 0 5 10 15
ABT888 0 1 2 5 ABT888 0 1 2 5 ABT888 0 1 2 5
(μmol/L) (μmol/L) (μmol/L)
Relative % area

Relative % area

Relative % area
1.0 1.0 1.0
*** ** ** * ***
*** ***
0.5 *** ***
0.5 0.5

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

*** *** *** ***
*** *** ***
0.0 0.0 *** *** *** *** *** 0.0
(μmol/L) 0 2 5 10 0 1 2 5 0 1 2 5 (μmol/L) 0 5 10 15 0 1 2 5 0 1 2 5 (μmol/L) 0 5 10 15 0 1 2 5 0 + 1+ 2 + 5
2 + 5 +10 + 5 +10 +15 + 5 10 15
NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888 ABT888
MEKi MEKi MEKi

L M245 C4 (NRAS Q61MUT) M M245 C5 (NRAS Q61MUT) N M207 (NRAS Q61MUT)

MEKi MEKi MEKi
NU7026 0 5 10 15 NU7026 0 5 10 15 NU7026 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)

ABT888 0 1 2 5 ABT888 0 1 2 5 ABT888 0 5 10 15

(μmol/L) (μmol/L) (μmol/L)

NU7026 NU7026 NU7026

(μmol/L) 0 5 10 15 (μmol/L) 0 5 10 15 (μmol/L) 0 5 10 15
ABT888 0 1 2 5 ABT888 0 1 2 5 ABT888 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)
Relative % area

Relative % area

Relative % area

1.0 1.0 1.0

*** *** *** ***
***
0.5 *** 0.5 *** 0.5
*** *** *** *** *** *** ***
*** *** *** *** ***
*** ***
0.0 0.0 0.0
(μmol/L) 0 5 10 15 0 1 2 5 0 1 2 5 (μmol/L) 0 5 10 15 0 1 2 5 0 1 2 5 (μmol/L) 0 5 10 15 0 5 10 15 0 5 10 15
+ +
5 10 15 + 5 +10 +15 + 5 +0 + 5 +
1 1
NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888 ABT888
MEKi MEKi MEKi

Figure 4. (Continued) Seeding densities (cells/well in 6-well dishes) and culture durations (days): 40,000; 30 (I), 50,000; 31 (J), 5,000; 17 (K), 5,000; 22
(L), 5,000; 16 (M), and 20,000; 16 (N). (continued on following page)

DNA-PKi Suppresses Acquired MAPKi Resistance the MAPKi regimen, we used MEKi (trametinib), KRASG12C
in KRASG12C Pancreatic Ductal Adenocarcinoma inhibitor (KRASG12Ci; AMG510 or MRTX849) + MEKi, or
and Non–Small Cell Lung Carcinoma Cell Lines type II RAF inhibitor (RAFi; BGB-283) + MEKi. As with mela-
As DNA-PKi plus MAPKi may constitute an effective noma cell lines, we first identified concentrations of DNA-
combination for BRAFV600MUT and NRASMUT melanoma, we PKi (NU7026) or PARPi (ABT888) that did not affect the
explored this combinatorial efficacy (± PARPi) in human pan- clonogenic growth of KRASG12C PDAC and NSCLC cell lines
creatic ductal adenocarcinoma (PDAC) and non–small cell without MAPKi (Supplementary Fig. S4F and S4G). We then
lung carcinoma (NSCLC) cell lines driven by KRASG12C. For assayed for acquired MAPKi-resistant growth in these cell

894 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

O M207 C1 (NRAS Q61MUT)

P M245 (NRAS Q61MUT)
MEKi PRKDC
shVector sh1 sh2 sh3 sh4 2.0
NU7026 (μmol/L) 0 5 10 15 shVector

Cell viability
1.5 PRKDC sh1
PRKDC
1.0 PRKDC sh2
0.5 PRKDC sh3
TUBULIN PRKDC sh4
0.0
ABT888 (μmol/L) 0 5 10 15 0 1 2 3 4 5 6
Days
PRKDC
shVector sh1 sh2 sh3 sh4

14 days
DMSO
NU7026 (μmol/L) 0 5 10 15
ABT888 (μmol/L) 0 5 10 15

29 days
MEKi
Relative % area

1.0 *** ***

Relative % area

Relative % area
*** 1.5
*** *** 1.0

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

1.0 ***
0.5
*** *** 0.5 0.5 ***
*** *** ***
0.0 0.0 0.0
(μmol/L) 0 5 10 15 0 5 10 15 0 5 10 15

r
1
2
3
4

r
1
2
3
4
5 +0 + 5 +
to

to
sh
sh
sh
sh

sh
sh
sh
sh
c

c
Ve

Ve
1 1 PRKDC PRKDC
sh

sh
NU7026 ABT888 NU7026 +
DMSO MEKi
ABT888
MEKi

Q M245 (NRAS Q61MUT) R

LIG4 M229 (BRAF V600MUT)
1.5
shVector sh1 sh2

Relative % area
Cell viability

Days: 1 22 36 Combo agent(s) 1.0

LIG4 1.0
shVector BRAFi + MEKi ***
*** ***
0.5 LIG4 sh1 NU7026 + 0.5
TUBULIN LIG4 sh2 DMSO NU7026 ABT888 ABT888
0.0
0.0
0 1 2 3 4 5 6
Days

N SO

T8 6
+ NU 88
T8 026
AB 02
LIG4

88
M
7

AB 7
U
D
shVector sh1 sh2
BRAFi + MEKi
14 days
DMSO

S M245 (NRAS Q61MUT

)
Days: 1 25 40 Relative % area
Combo agent(s) 1.0
29 days
MEKi

MEKi ***
NU7026 + 0.5
DMSO NU7026 ABT888 ABT888
Relative % area

Relative % area

0.0
1.0 *** 1.0
*** ***
N SO

T8 6
+ NU 88
T8 026
AB 02

0.5 0.5 88
M
7

AB 7
U
D

0.0 0.0
MEKi
r
1
2

r
1
2
to

to
sh
sh

sh
sh
c

c
Ve

Ve

LIG4 LIG4
sh

sh

DMSO DMSO

Figure 4. (Continued) Seeding density (cells/well in 6-well dishes) and culture duration (days): 5,000; 16 (O). Data representative of 2 to 4 independ-
ent repeats. P and Q, Western blot, MTT assay (top right), and clonogenic growth assay (bottom) of M245 cells transduced by lentivirus harboring shVec-
tor control or shRNAs of PRKDC (P) or LIG4 (Q). TUBULIN, loading control. Clonogenic growth with vehicle (14 days, 5,000 cells/well in 6-well dishes) or
MEKi (29 days, 50,000 cells/well in 6-well dishes) treatments. MEKi, trametinib at 0.01 μmol/L. Representative results of 2 independent experiments;
quantification of n = 4 fields (mean ± SD). R and S, As in A and J, respectively, except DNA-PKi and/or PARPi cotreatments were performed in M229 (R)
and M245 (S) at a later stage of MAPKi treatment (see time-point schema). Inhibitor concentrations for M229: BRAFi + MEKi (PLX4032 at 0.5 μmol/L,
AZD6244 at 0.5 μmol/L), NU7026 (8 μmol/L), ABT888 (4 μmol/L), and NU7026 + ABT888 (4 μmol/L + 4 μmol/L). Inhibitor concentrations for M245: MEKi
(trametinib at 0.01 μmol/L), NU7026 (5 μmol/L), ABT888 (1 μmol/L), and NU7026 + ABT888 (5 μmol/L + 1 μmol/L). Left, representative cultures; right,
quantifications over n = 4 fields (mean ± SD). Data representative of 2 independent repeats. (continued on next page)

APRIL 2023 CANCER DISCOVERY | 895

RESEARCH ARTICLE Dharanipragada et al.

T M229 (BRAF V600MUT) U M245 (NRAS Q61MUT)

BRAFI + MEKi MEKi
NU7026 + NU7026 +
DMSO NU7026 ABT888 ABT888 DMSO NU7026 ABT888 ABT888

Day 15–30 Day 1–15 Day 1–30

Day 14–28 Day 1–14 Day 1–28

DMSO DMSO
NU7026
NU7026
1.5 ABT888 1.5
ABT888
Relative % area

Relative % area
NU7026 + ABT888
NU7026 + ABT888
1.0 1.0

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

** ** *** ***
0.5 *** 0.5 ***
*** ***
*** *** *** ***
0.0 *** 0.0 *** *** ***
Day 1–28 Day 1–14 Day 14–28 Day 1–30 Day 1–15 Day 15–30
BRAFi + MEKi MEKi

Figure 4. (Continued) T and U, As in A and J, respectively, except DNA-PKi and/or PARPi cotreatments were performed in M229 (T) and M245 (U)
cells at the same concentrations in R and S for the entire duration (day 1–28 in T; day 1–30 in U), the first half (day 1–14 in T; day 1–15 in U) or the second
half (day 14–28 in T; day 15–30 in U) of the total treatment course. Top, representative cultures; bottom, quantifications over n = 4 fields (mean ± SD).
Data representative of 2 independent repeats. P values (one-way ANOVA followed by the Tukey multiple comparisons test) comparing indicated cultures
versus MAPKi-only cultures (A–O and R–U) or shVector culture (P and Q): *, P < 0.05; **, P < 0.01; ***, P < 0.001.

lines, without or with DNA-PKi and/or PARPi (Fig. 6). Inter- resistance, without incurring overt signs of toxicities or
estingly, we observed that DNA-PKi effectively suppressed significant reductions in body weight. In NRASMUT mela-
acquired MAPKi resistance in two of two KRASG12C PDAC noma (Mel_PDX1; Fig. 7C), DNA-PKi treatment (10 mg/
cell lines (Fig. 6A–F) and one of two KRASG12C NSCLC cell kg/d) did not elicit tumor regression, and MEKi (trametinib,
lines (Fig. 6G–J). Moreover, PARPi was generally ineffective 3 mg/kg/d) elicited only transient tumor regression. In
as a combinatorial agent with MAPKi. Future studies should contrast, the combination of MEKi + DNA-PKi led to
explore the contributions of chromothripsis, ecDNAs, and minimally palpable tumors, without incurring toxicity. In
CGRs to acquired MAPKi resistance in human KRASG12C two additional NRASMUT melanoma PDXs (Mel_PDX2 and
PDAC and NSCLC. Mel_PDX4), we consistently observed superior efficacy of
MEKi + DNA-PKi over MEKi alone (Fig. 7D and E). In
DNA-PKi Forestalls Resistance In Vivo and Mel_PDX4, we initiated dosing with both DNA-PKi at 8
Reduces ecDNA and CGR Size mg/kg/day and MEKi at 3 mg/kg/day but stopped dosing
Given the broader resistance-preventive activity of DNA- DNA-PKi (in combination with MEKi) on day 67 (Fig. 7E).
PKi, we tested NU7026’s in vivo efficacy in five cutaneous With longer follow-up (Supplementary Fig. S6B), discon-
melanoma PDXs. In Mel_PDX16 (BRAFV600MUT) melanoma, tinuation of DNA-PKi dosing in this group of mice treated
using well-established tumors (∼500–700 mm3; Fig. 7A), initially with MEKi + DNA-PKi did not lead to tumor
BRAFi + MEKi (vemurafenib at 90 mg/kg/d, trametinib at relapse. Based on cell line findings (Fig. 4), we expected that
0.7 mg/kg/d) elicited only tumor growth inhibition, and delayed DNA-PKi cotreatment in vivo would also dimin-
DNA-PKi treatment (8 mg/kg/d) had no discernible effect ish its preventive mechanism of action. In Mel_PDX27
on tumor growth compared with vehicle-treated tumors. In (BRAFV600MUT) melanoma (Fig. 7B; Supplementary Fig. S6A),
contrast, the triplet of BRAFi + MEKi + DNA-PKi elicited we repeated BRAFi + MEKi (vemurafenib at 90 mg/kg/day,
tumor regression transiently for ∼14 days until the tumors trametinib at 0.7 mg/kg/day) treatment, and, at the time of
acquired resistance. There were no overt signs of toxici- disease progression (when the mean tumor volume returned
ties or significant reductions in body weight in any exper- to the mean pre-MAPKi volume of ∼500 mm3), we divided
imental group. In Mel_PDX27 (BRAFV600MUT) melanoma the tumors/mice into two groups. The first group continued
(Fig. 7B; Supplementary Fig. S6A), BRAFi + MEKi (vemu- on BRAFi + MEKi treatment at the same dosages, whereas
rafenib at 90 mg/kg/day, trametinib at 0.7 mg/kg/day) the second group received DNA-PKi treatment (8 mg/kg/d)
elicited transient tumor regression, whereas DNA-PKi treat- added on top of BRAFi + MEKi. Consistent with the cell
ment (8 mg/kg/day) did not elicit tumor regression com- line results (Fig. 4A–O), DNA-PKi cotreatment did not elicit
pared with vehicle-treated tumors. In contrast, the triplet of discernible tumor regression on tumors that had already
BRAFi + MEKi + DNA-PKi significantly forestalled acquired acquired MAPKi resistance in vivo (Fig. 7F).

896 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

A M249 (BRAF V600MUT)

DDR4 – BRAFi – MEKi DDR4 – BRAFi – MEKi
Parental DDR4 + BRAFi + MEKi 8 days 26 days
30 *** ***

BRAF HSR area/

HSR HSR HSR
BRAF/CEN7/DAPI

20
**

cell
*
10

0
Withdrawal 0 8 26 0 8 26
(days) DDR4 DDR5
DDR5 – BRAFi – MEKi DDR5 – BRAFi – MEKi
Parental DDR5 + BRAFi + MEKi 8 days 26 days
HSR HSR HSR
BRAF/CEN7/DAPI

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

B M395 (BRAF V600MUT)
DDR – BRAFi – MEKi DDR – BRAFi – MEKi 400

BRAF ecDNA CN/cell

Parental DDR + BRAFi + MEKi 7 days 25 days ***
ecDNA ecDNA ecDNA 300
BRAF/CEN7/DAPI

200

100

0
Withdrawal 0 7 25
(days) DDR

C M245 (NRAS Q61MUT) **

100
RAF1 ecDNA CN/cell
Parental C3 + MEKi C3 – MEKi 14 days C3 – MEKi 26 days 80
ecDNA ecDNA ecDNA 60
RAF1/CEN3/DAPI

40
20
0
Withdrawal 0 14 26
(days) C3

100 ***
** *** 120 HSR
HSR HSR HSR
RAF1 HSR area/

80 ecDNA
RAF1/CEN3/DAPI

100
Frequency (%)

60 80
cell

40 60
40
20 20
0 0
Withdrawal 0 14 26 Withdrawal 0 14 26
(days) C3 (days) C3
Parental C5 + MEKi C5 – MEKi 9 days
HSR HSR 15 * HSR
120
NRAS HSR area/
NRAS/CEN1/DAPI

Non-HSR
100
Frequency (%)

10
80
cell

60
5 40
20
0 0
Withdrawal 0 9 Withdrawal 0 9
(days) C5 (days) C5

Figure 5. Resistance-driver ecDNAs and HSRs dynamically track with resistance, and DNA-PKi prevents the size expansion of ecDNAs and CGRs early
on MAPKi treatment. A–C, Metaphase DNA-FISH of paired parental and acquired resistant BRAFV600MUT (A and B) or NRASQ61MUT (C) cell lines without or
with drug withdrawal showing ecDNA or HSR amplicons harboring BRAF, RAF1, or NRAS. Left, representative images; right, quantifications per cell. Scale
bars, 15 μm. P values (unpaired two-tailed Student t test): *, P < 0.05; **, P < 0.01; ***, P < 0.001. (continued on next page)

APRIL 2023 CANCER DISCOVERY | 897

 RESEARCH ARTICLE Dharanipragada et al.

D M249 (BRAF V600MUT) E M395 (BRAF V600MUT)

Parental Parental Parental
DDR4 + BRAFi + MEKi DDR5 + BRAFi + MEKi DDR + BRAFi + MEKi
DDR4 – BRAFi – MEKi 26 days DDR5 – BRAFi – MEKi 30 days DDR – BRAFi – MEKi 25 days
150
Relative units (3 days)

Relative units (3 days)

Relative units (3 days)

100
150
100 80
100 60
50 40
50
20

***

***
***

***

***
0 0 0
2 0
5 2
5
1
2
5

2 0
5 2
5
1
2
5

2 0
5 2
5
1
2
5
0. 0.
0.

0. 0.
0.

0. 0.
0.
+
+
+

+
+
+

+
+
+
+
+

+
+
1
2
5

1
2
5

+
+
1
2
5
0.

0.

0.
BRAFi + MEKi (μmol/L) BRAFi + MEKi (μmol/L) BRAFi + MEKi (μmol/L)
Relative units (6 days)
Relative units (6 days)

200

Relative units (6 days)

250 150
200 150
150 100
100
100

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

50
50
***

50
***

***
***
***
***
0 0 0
2 0
5 2
5
1
2
5

2 0
5 2
5
1
2
5

2 0
5 2
5
1
2
5
0. 0.
0.

0. 0.
0.

0. 0.
0.
+
+
+

+
+
+

+
+
+
+
+
1
2
5

+
+
1
2
5

+
+
1
2
5
0.

0.

0.
BRAFi + MEKi (μmol/L) BRAFi + MEKi (μmol/L) BRAFi + MEKi (μmol/L)

F M245 (NRAS Q61MUT) H M229-d11

ecDNA
Parental Parental CGR
C3 + MEKi C5 + MEKi 1.5
C3 – MEKi 30 days C5 – MEKi 17 days
120 120
Relative units (3 days)

Relative units (3 days)

1.0
Average total span of ecDNAs + CGRs (Mb)

100 100
80 80
60 60 0.5
*** ***
*** ***

40 40
20 20 0.0
0 0
MAPKi MAPKi + DNA-PKi
0
0.02
0.05
0.1
0.2
0.5
1
2
5

0
0.02
0.05
0.1
0.2
0.5
1
2
5

M245-d10
MEKi (μmol/L) MEKi (μmol/L) 0.4 ecDNA
CGR
Relative units (6 days)

Relative units (6 days)

120 150
0.3
100
80 100
0.2
60
40 50
0.1
***

20
***
***

0
0
0
0.02
0.05
0.1
0.2
0.5
1
2
5

0
0.02
0.05
0.1
0.2
0.5
1
2
5

0.0
MEKi (μmol/L) MEKi (μmol/L) MEKi MEKi + DNA-PKi

G M249 (BRAF V600MUT)

BRAFi + MEKi 31 days *
15
BRAF HSR area/

DMSO DMSO NU7026 ABT888 NU7026 + ABT888

HSR HSR 10
BRAF/CEN7/DAPI

cell

0
D O
N SO
AB 26

+ NU 88
T8 026
S

70
T8

88
M
M

AB 7
U
D

BRAFi + MEKi

898 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

We then explored early on-treatment pharmacodynamic insights into the underlying genomic instability processes
markers that are associated with the superiority of that generate preexisting and therapy-elicited clonal diver-
MAPKi + DNA-PKi over MAPKi alone in forestalling or sification in advanced cutaneous melanoma. We identified
preventing acquired resistance. We posited that the com- chromothripsis as well as ecDNAs and CGRs as highly
bination of DNA-PKi with MAPKi (versus MAPKi alone) recurrent and pervasive genomic SVs in MAPKi-naive/sen-
would suppress the total size of ecDNAs and CGRs spe- sitive and acquired MAPKi-resistant melanoma in both the
cifically generated due to MAPKi treatment, agnostic of clinical setting (BRAFV600MUT melanoma, in which MAPKi
content genes, regulatory elements, and known or putative therapy is a standard-of-care therapy) and the experi-
roles in driving resistance. Using all five PDX models, we mental setting (NRASMUT melanoma, in which there is a
collected early on-treatment tumors and vehicle-treated lack of targeted therapy options). CGRs can derive from
tumors (Supplementary Fig. S6C and S6D). Expectedly, reintegration of ecDNAs or breakage–fusion–bridge (BFB)
p-ERK was suppressed strongly in both groups of tumors cycles (20, 43). We did not detect any BFB event in any of
treated with MAPKi or MAPKi + DNA-PKi (Supplementary our tumors, MAPKi-sensitive/naive or acquired-resistant,
Fig. S6E). Binding of the DNA-PK holoenzyme (DNA- favoring ecDNA reintegration as the main route of CGR
PKCS + Ku70 + Ku80) to DSBs elicits autophosphorylation generation in the setting of metastatic cutaneous mela-
of DNA-PKCS on serine 2056. We found that DNA-PKi noma. Consistently, the acute stress of MAPKi therapy
treatment, with or without MAPKi, suppressed nuclear favors ecDNAs, whereas stable or chronic stress favors

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

p-DNA-PKCS foci (Supplementary Fig. S6F), which is con- reintegration of ecDNAs into chromosomes as HSRs (13).
sistent with NU7026 being able to suppress the kinase Moreover, following chromothripsis, chimeric circulariza-
activity and autophosphorylation of the activated DNA-PK tion of DNA and reintegration of DNA circles into chro-
holoenzyme. We found that short-term MEKi treatment in mosomes constitute a major source of SVs and linear
Mel_PDX1 induced γH2AX nuclear foci in >50% of tumor genome mutagenesis (44, 45). Importantly, the selective
cells (Supplementary Fig. S6G), consistent with a recent pressure of MAPKi therapy is evidenced by the amplifica-
report that found MAPKi treatment resulting in γH2AX tion of bona fide resistance-driver genes via ecDNAs and
nuclear foci or DSBs in melanoma cell lines (42). Expect- CGRs. The report here of resistance-specific ecDNAs and
edly, DNA-PKi cotreatment with MEKi further induced CGRs amplifying a wide array of coding and noncod-
DSBs marked by γH2AX foci. We then generated WGS ing sequences warrants future investigations into their
data to enable an analysis of CGR and ecDNA amplicons resistance-causative mechanisms.
(Supplementary Table S11). CGR and ecDNA amplicons Even though chromothripsis is regarded as a potential
identified in vehicle-treated PDX tumors were considered precursor of ecDNAs, it creates oscillating CNs of genomic
as background and were removed from those identified in segments but does not cause high-level amplifications. We
MAPKi- or MAPKi + DNA-PKi–treated tumors. Consistent found that, within each tumor, ecDNAs (mean total size per
with our hypothesis, MAPKi + DNA-PKi (versus MAPKi) genome, 7 Mb; mean size per ecDNA, 343 kb) and their rein-
treatment was associated with reductions in the average tegrated CGR counterparts (mean total size per genome, 6
total genomic spans of CGRs and ecDNAs in five of five Mb; mean size per CGR, 598 kb) almost always span genomic
PDX models analyzed (Fig. 7G). We also analyzed the break- regions bounded by larger chromothriptic regions (mean
point–junctional sequences of MAPKi treatment–specific total size per genome, 474 Mb; mean size per chromothriptic
ecDNAs and CGRs to infer the relative contributions of region, 120 Mb), which is consistent with a chromothriptic
DSB repair pathways (Supplementary Fig. S6H). Consist- origin of ecDNA and CGR amplicons. Chromothripsis can
ent with the functional role of DNA-PKi, its combina- occur as a result of micronuclei formation around lagging
tion with MAPKi suppressed the contribution of NHEJ, chromosomes or chromosome bridge formation due to tel-
with potentially compensatory increases in alt-NHEJ or omere crisis (46–48). Both aberrant processes are associated
HRR processes. with a loss of primary or micronuclear membrane integ-
rity and subsequent mutagenesis. This is consistent with
our finding of enhanced mutational density within chromo-
DISCUSSION thriptic genomic regions, especially within acquired-resistant
Preexisting tumor heterogeneity and de novo diversifica- genomes, as well as a resistance-specific mutator phenotype
tion in response to targeted therapy are thought to fuel enriched for signatures of excessive single-stranded DNA
the evolution of acquired resistance. Here, we provide damage and/or deficient repair (BER and MMR). Intriguingly,

Figure 5. (Continued) D–F, Three-day (top) and six-day (bottom) MTT assay of indicated cell lines treated with graded concentrations of BRAFi + MEKi
(D and E) or MEKi (F). Cell viability was normalized to respective DMSO/vehicle groups. Inner brackets, comparisons between acquired resistant sublines
without and with drug withdrawal. Outer brackets, comparisons between acquired resistant sublines and their isogenic parental cell lines. P values (two-
way ANOVA test): *, P < 0.05; ***, P < 0.001. G, Representative images (left) and quantification (right) of metaphase DNA-FISH showing HSR harboring
BRAF in M249 cells treated with vehicle or BRAFi + MEKi with or without NU7026 and/or ABT888 for 31 days. Inhibitor concentrations: BRAFi + MEKi
(PLX4032 at 0.25 μmol/L, AZD6244 at 0.25 μmol/L), NU7026 (4 μmol/L), ABT888 (2 μmol/L), and NU7026 + ABT888 (4 μmol/L + 2 μmol/L). CEN,
centromere; DAPI, nuclear stain. Scale bars, 15 μm. P values (unpaired two-tailed Student t test): *, P < 0.05. H, Average total genomic spans of treatment-
specific ecDNAs + CGRs in M229 and M245 cell lines (background ecDNA + CGR spans detected in vehicle-treated cells were filtered). Inhibitor concen-
trations: BRAFi + MEKi (PLX4032 at 1 μmol/L, AZD6244 at 1 μmol/L for M229) or MEKi (trametinib at 0.02 μmol/L for M245), DNA-PKi (NU7026 at
8 μmol/L for both cell lines).

APRIL 2023 CANCER DISCOVERY | 899

RESEARCH ARTICLE Dharanipragada et al.

A MIAPaCa-2 (KRAS G12C) B MIAPaCa-2 (KRAS G12C) C MIAPaCa-2 (KRAS G12C)

RAFi + MEKi (BGB-283 + trametinib) KRASi + MEKi (AMG510 + trametinib) KRASi + MEKi (MRTX849 + trametinib)
NU7026 0 1 2 5 NU7026 0 1 2 5 NU7026 0 1 2 5
(μmol/L) (μmol/L) (μmol/L)

ABT888 0 5 10 15 ABT888 0 5 10 15 ABT888 0 5 10 15

(μmol/L) (μmol/L) (μmol/L)

NU7026 NU7026 NU7026

(μmol/L) 0 1 2 5 (μmol/L) 0 1 2 5 (μmol/L) 0 1 2 5
ABT888 0 5 10 15 ABT888 0 5 10 15 ABT888 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)
Relative % area

Relative % area
Relative % area
1.0 1.0 ** ** 1.0
*** *** *** ***
*** ***
0.5 *** *** 0.5 *** *** *** 0.5 *** ***

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

*** *** ***
*** *** *** ***
*** *** *** *** ***
0.0 0.0 0.0
(μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15 (μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15 (μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15
+
1 2 +5 + +
1 2 +5 + 1 +2 + 5 +
NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888 ABT888
RAFi + MEKi KRASi + MEKi KRASi + MEKi

D XWR200 (KRAS G12C) E XWR200 (KRAS G12C) F XWR200 (KRAS G12C)

RAFi + MEKi (BGB-283 + trametinib) KRASi + MEKi (AMG510 + trametinib) KRASi + MEKi (MRTX849 + trametinib)
NU7026 0 5 10 15 NU7026 0 5 10 15 NU7026 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)

ABT888 0 2 5 10 ABT888 0 2 5 10 ABT888 0 2 5 10

(μmol/L) (μmol/L) (μmol/L)

NU7026 NU7026 NU7026

0 5 10 15 0 5 10 15 0 5 10 15
(μmol/L) (μmol/L) (μmol/L)
ABT888 0 2 5 10 ABT888 0 2 5 10 ABT888 0 2 5 10
(μmol/L) (μmol/L) (μmol/L)
Relative % area
Relative % area

Relative % area

1.0 1.0 1.0

*** *** *** *** *** ***
0.5 *** 0.5 0.5
*** ***
*** *** *** *** *** ***
*** *** *** *** ***
0.0 *** 0.0 *** *** ***
*** *** *** 0.0
(μmol/L) 0 5 10 15 0 2 5 10 0 2 5 10 (μmol/L) 0 5 10 15 0 2 5 10 0 2 5 10 (μmol/L) 0 5 10 15 0 2 5 10 0 2 5 10
+ +
5 10 5 + +
5 10 +5 + + +
5 10 5 +
1 1 1
NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888 ABT888
RAFi + MEKi KRASi + MEKi KRASi + MEKi

Figure 6. DNA-PKi cotreatment prevents acquired MAPKi resistance in KRASG12 human PDAC and NSCLC cell lines. A–J, Long-term clonogenic growth
of PDAC (MIAPaCa-2, XWR200; A–F) or NSCLC (H358, H2122; G–J) cell lines showing acquired resistant colonies to MEKi alone, type II RAFi + MEKi, or
KRASG12Ci + MEKi and their suppression by indicated cotreatments with DNA-PKi (NU7026) and/or PARPi (ABT888) at indicated concentrations. Top, rep-
resentative cultures; bottom, quantifications over n = 4 fields (mean ± SD). Data representative of 2 independent repeats. KRASi, KRAS inhibitor. Seeding
densities (cells/well in 6-well dishes) and culture durations: MIAPaCa-2 (10,000; 30 days for A–C) and XWR200 (150,000; 23 days for D, 30 days for E, 26
days for F). (continued on following page)

a recent study proposed that a specific BER defect may predis- in promoting the total segment sizes of ecDNAs and CGRs
pose micronuclei-associated or cytoplasmic chromosomes to generated early on MAPKi therapy. This finding is consist-
breakage, a key step toward chromothripsis (49). ent with prior literature supporting NHEJ as key to ecDNA
Translationally, we produced in vivo evidence supportive formation (28, 50). Selection for numeric expansion of spe-
of the well-studied role of DNA-PKCS in NHEJ as critical cific ecDNAs and CGRs is considered a direct mechanism

900 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

G H358 (KRAS G12C) H H358 (KRAS G12C)

Figure 6. (Continued) Seeding densities
RAFi + MEKi (BGB-283 + trametinib) KRASi + MEKi (AMG510 + trametinib) (cells/well in 6-well dishes) and culture dura-
NU7026 0 1 2 5 NU7026 0 1 2 5 tions: H358 (40,000; 14 days for G, 29 days for
(μmol/L) (μmol/L) H, 26 days for I) and H2122 (20,000, 22 days
for J). Concentrations of type II RAFi (BGB-
283) + MEKi (trametinib) or KRASG12Ci (AMG510
or MRTX849) + MEKi (trametinib) or MEKi
ABT888 0 5 10 15 ABT888 0 5 10 15 (trametinib): 0.5 μmol/L + 0.01 μmol/L (A),
(μmol/L) (μmol/L) 0.02 μmol/L + 0.01 μmol/L (B), 0.01 μmol/L + 0.01
μmol/L (C), 0.1 μmol/L + 0.001 μmol/L (D),
0.005 μmol/L + 0.001 μmol/L (E), 0.001
NU7026 NU7026 μmol/L + 0.001 μmol/L (F), 0.2 μmol/L + 0.001
( mol/L) 0 1 2 5 0 1 2 5
(μmol/L) μmol/L (G), 0.005 μmol/L + 0.001 μmol/L (H),
ABT888 0 5 10 15 ABT888 0 5 10 15 0.002 μmol/L + 0.001 μmol/L (I), and 0.02 μmol/L
(μmol/L) (μmol/L) (J). Data representative of 2 to 4 independent
repeats. P values (one-way ANOVA followed by
Tukey multiple comparisons test) comparing
indicated cultures versus MAPKi treatment-only

Relative % area
Relative % area

1.0 cultures: **, P < 0.01; ***, P < 0.001.

1.0
*** *** *** ***
*** *** *** ***
0.5 *** 0.5 ***

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

*** *** ***
0.0 0.0 ***
(μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15 (μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15
1 +2 + 5 + 1 +2 + 5 +
NU7026 ABT888 NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888
RAFi + MEKi KRASi + MEKi

I H358 (KRAS G12C) J H2122 (KRAS G12C)

KRASi + MEKi (MRTX849 + trametinib) MEKi (trametinib)
NU7026 0 1 2 5 NU7026 0 2 5 10
(μmol/L) (μmol/L)

ABT888 0 5 10 15 ABT888 0 1 2 5
(μmol/L) (μmol/L)

NU7026 NU7026
(μmol/L) 0 1 2 5 (μmol/L) 0 2 5 10
ABT888 0 5 10 15 ABT888 0 1 2 5
(μmol/L) (μmol/L)
Relative % area
Relative % area

1.0 1.0
*** ***
*** *** *** ***
*** ***
0.5 0.5
***
***
0.0 *** *** *** *** ***
0.0
(μmol/L) 0 1 2 5 0 5 10 15 0 5 10 15 (μmol/L) 0 2 5 10 0 1 2 15 0 1 2 15
1 +2 + 5 + +
2 5 +0 +
1
NU7026 ABT888
NU7026 + NU7026 ABT888 NU7026 +
ABT888 ABT888
KRASi + MEKi MEKi

promoting tumor fitness in response to a given stressor. chemosensitization (37). The rationale is based on cata-
Therefore, reduction of the total ecDNA/CGR genomic strophic DSBs that would result in excessive DNA damage
spans early on DNA-PKi cotreatment should serve to reduce repair stress and hence the synergistic induction of death
the reservoir of ecDNAs/CGRs available for natural selec- in cancer cells, especially those overexpressing DNA-PKCS.
tion by MAPKi therapy. Here, we rationalize the combination of DNA-PKi and
DNA-PKCS, the target of DNA-PKi, subserves other less- MAPKi based on dual concepts. First, the rapid induc-
characterized cancer survival pathways (37). It is possible tion of genomic instability mechanisms, in particular
that another beneficial mechanism of action of DNA-PKi the generation of ecDNA and CGR amplicons, is criti-
coexists. DNA-PKi has been proposed in combination with cal for genomic diversification and perhaps epigenomic
agents that directly induce DNA damage, such as radi- reprogramming necessary for melanomas to adapt quickly
otherapy or chemotherapy, with the intent of radio- or to MAPKi therapy. In this rationale, DNA-PKi suppresses

APRIL 2023 CANCER DISCOVERY | 901

RESEARCH ARTICLE Dharanipragada et al.

A 3,000 Mel_PDX16 (BRAF MUT) Vehicle

DNA-PKi
BRAFi + MEKi
2,500 BRAFi + MEKi + DNA-PKi
Mel_PDX16 (BRAF MUT)
25

Tumor volume (mm3)

2,000
20
1,500 7.2E-3 Vehicle

Weight (g)
15
DNA-PKi
1,000 BRAFi + MEKi
10
BRAFi + MEKi + DNA-PKi
500 5

0 0
0 16 33 36 39 42 44 47 49 51 54 56 58 61 63 65 68 70 0 16 33 39 47 54 61 68
Days Days

B Mel_PDX27 (BRAF MUT)

4,500 Vehicle
4,000 DNA-PKi
3,500 BRAFi + MEKi Mel_PDX27 (BRAF MUT)
35
Tumor volume (mm3)

BRAFi + MEKi + DNA-PKi

3,000 30
2,500 25

Weight (g)
2,000 20 Vehicle
1,500 15 DNA-PKi
1,000 10 BRAFi + MEKi

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

1.2E-5 BRAFi + MEKi + DNA-PKi
500 5
0 0
0 24 48 53 55 57 60 62 64 67 69 71 74 76 78 81 83 85 88 0 24 48 55 62 69 76 83
Days Days

C Mel_PDX1 (NRAS MUT)

6,000
Vehicle
DNA-PKi Mel_PDX1 (NRAS MUT)
5,000 25
MEKi
Tumor volume (mm3)

4,000 MEKi + DNA-PKi 20

Vehicle
Weight (g)

3,000 15
DNA-PKi
MEKi
2,000 10 MEKi + DNA-PKi

1,000 5
3.9E-6
0 0
0 17 34 37 41 43 45 48 50 52 55 57 59 62 64 66 0 17 24 41 48 55 62
Days Days

D 3,500 Mel_PDX2 (NRAS MUT

)
Vehicle
Vehicle
DNA-PKi
DNA-PKi
3,000
MEKi
MEKi
MEKi + DNA-PKi Mel_PDX2 (NRAS MUT)
Tumor volume (mm3)

2,500 MEKi + DNA-PKi 30

2,000 25

20
1,500
Weight (g)

15 Vehicle
1,000 DNA-PKi
10 MEKi
500 MEKi + DNA-PKi
3.2E-7 5
3.2E-7
0 0
0 21 43 48 50 52 55 57 59 62 64 66 69 71 73 76 78 80 0 21 43 50 57 64 71 78
Days Days

E Mel_PDX4 (NRAS MUT)

3,500 Vehicle
Mel_PDX4 (NRAS MUT)
DNA-PKi 35
3,000
MEKi
30
Tumor volume (mm3)

2,500 MEKi + DNA-PKi

25
2,000
Weight (g)

20
Vehicle
1,500
15 DNA-PKi
1,000 MEKi
10
P = 8.91E-15 MEKi + DNA-PKi
500 5
0 0
0 21 42 46 48 50 53 55 57 60 62 64 67 69 71 74 76 78 0 21 42 49 56 63 70 77
Days Days

Figure 7. DNA-PKi cotreatment with MAPKi reduces the size of ecDNAs and CGRs and prevents acquired resistance in vivo. A–E, Measurements of
tumor volumes (left) and body weights of mice (right) in two BRAFV600MUT (A and B) and three NRASMUT (C–E) cutaneous melanoma PDX models. Vehicle or
indicated treatments initiated on well-established tumors on days 33 (A), 48 (B), 34 (C), 43 (D), and 42 (E) after tumor fragment implantation, as marked
by upside-down red triangles. Dosage of inhibitors DNA-PKi (NU7026), BRAFi (vemurafenib), and MEKi (trametinib; mg/kg/day): DNA-PKi, 8; BRAFi, 90;
MEKi, 0.7 (A and B); DNA-PKi, 10; MEKi, 3 (C); DNA-PKi, 6; MEKi, 5 (D); and DNA-PKi, 8 (stopped on day 67 when in combination with MEKi, as marked by
an upside-down orange triangle); MEKi, 3 (E). One tumor per mouse; number of mice per experimental group (from top to bottom): 5, 5, 7, 7 (A); 4, 5, 8,
8 (B); 5, 5, 8, 5 (C); 3, 4, 8, 8 (D); and 5, 5, 6, 9 (E). Results are shown as mean ± SEM. P values, Student t test. Body weights are shown as average values.
(continued on following page)

902 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

F Mel_PDX27 (BRAF MUT) BRAFi + MEKi

1,600
BRAFi + MEKi + DNA-PKi
Mel_PDX27 (BRAF MUT)
1,400 35

1,200 30
Tumor volume (mm3)

1,000 25

Weight (g)
800 20
BRAFi + MEKi
600 15 BRAFi + MEKi + DNA-PKi
400 10

200 5

0 0
0 33 66 70 73 75 77 80 82 84 87 89 91 94 96 98 101 0 33 66 73 80 87 94 101
Days Days

G Mel_PDX16 Mel_PDX27 Mel_PDX1

0.3
0.25 ecDNA
ecDNA ecDNA
CGR 0.3 CGR CGR

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

0.20
ecDNAs + CGRs (Mb)
Average total span of

0.2
0.2 0.15

0.10
0.1
0.1
0.05

0.0 0.0 0.00

BRAFi + MEKi BRAFi + MEKi BRAFi + MEKi BRAFi + MEKi BRAFi + MEKi MEKi MEKi + DNA-PKi
+ DNA-PKi + DNA-PKi (8) + DNA-PKi (10)

Mel_PDX2 Mel_PDX4
ecDNA ecDNA
CGR CGR
0.09
ecDNAs + CGRs (Mb)
Average total span of

0.06

2
0.03

0.00 0
MEKi MEKi + DNA-PKi MEKi MEKi + DNA-PKi

Figure 7. (Continued) F, As in Fig. 5B (BRAFi + MEKi treatment group), except at the earliest time of acquired resistance (day 84, marked by an
upside-down orange triangle), tumors or mice were divided into two groups (n = 5 each). Group 1 continued on BRAFi + MEKi treatments. Group 2
received DNA-PKi (8 mg/kg/day) in combination with BRAFi + MEKi treatments. G, Average total spans of ecDNAs + CGRs specific to MEKi-treated
versus MEKi + DNA-PKi–treated PDXs (background ecDNAs + CGRs detected in vehicle-treated tumors were filtered) in Mel_PDX16 (BRAFi + MEKi–
treated, n = 3; BRAFi + MEKi + DNA-PKi–treated, n = 3), Mel_PDX27 (BRAFi + MEKi–treated, n = 3; BRAFi + MEKi + DNA-PKi (8 mg/kg/day)–treated, n = 3;
BRAFi + MEKi + DNA-PKi (10 mg/kg/day)–treated, n = 2), Mel_PDX1 (MEKi-treated, n = 2; MEKi + DNA-PKi–treated, n = 2), Mel_PDX2 (MEKi-treated,
n = 2; MEKi + DNA-PKi–treated, n = 2), and Mel_PDX4 (MEKi-treated, n = 2; MEKi + DNA-PKi–treated, n = 3). Tumors were collected for analysis on days 6
(Mel_PDX16), 7 (Mel_PDX27), 11 (Mel_PDX1), 9 (Mel_PDX2), and 8 (Mel_PDX4) after initiating treatments.

NHEJ, which is necessary for the efficient formation of (52). Excessive DNA SSBs can be converted into DNA DSBs,
ecDNAs and CGRs. Second, MAPKi is potentially an inducer engendering DNA damage repair stress and/or chromosome
of DNA damage and/or DNA damage repair deficiency. In breakage. The latter may be a pathway to chromothripsis
this context, MAPKi has been shown to induce DNA dam- and ecDNA/CGR generation (49).
age in early drug-tolerant persister subpopulations (42). A Our demonstration of ecDNAs and CGRs driving acquired
therapy-induced oxidative metabolic adaptation (51) has MAPKi resistance advances the concept that multiple resist-
been proposed to cause ROS-induced mutagenesis, which ance mechanisms, genetic and epigenetic, as well as direct
can be repaired by DNA single-stranded break (SSB) repair (drug–target or MAPK pathway reactivation) and indirect
processes such as BER and MMR. DNA-PKCS, in addition to (nondrug–target pathway activation), are simultaneously
DSB repair, can also bind to and is activated by DNA SSBs causal of clinically acquired resistance. Future work needs to

APRIL 2023 CANCER DISCOVERY | 903

RESEARCH ARTICLE Dharanipragada et al.

dissect this hybrid genomic–epigenomic model with ecDNAs enzymatic fragmentation of gDNA, the libraries were constructed
and CGRs at the center of therapeutic targeting efforts. Find- by end repairing and A-tailing the fragmented DNAs, ligation
ings here also advance the concepts that preventing, instead of adapters, and PCR amplification. After library construction,
of reversing, acquired resistant phenotypes may be more indexed libraries were quantified for equal molar pooling and
paired-end sequenced with a read length of 2 × 150 bp on the Illu-
impactful clinically and that targeting DNA-PKCS and NHEJ,
mina NovaSeq 6000 S4 platform.
and potentially MMEJ and HRR, lies at the center of this Whole-genome libraries were prepared for patient-matched nor-
approach in stabilizing cancer genomes during oncogene- mal tissues and tumors as follows: (i) 10 BRAFV600MUT clinical
targeted therapies. Finally, prevention of chromothripsis, patients (baseline tumors, n = 10; resistant tumors, n = 17), (ii) three
such as by reducing chromosome fusion and missegregation BRAFMUT RAM patients (sensitive tumor, n = 1; resistant tumors,
events and minimizing primary and micronuclei membrane n = 12), and (iii) six PDX models. PDX models included one BRAFMUT
ruptures, may expand our molecular armamentarium against PDX (vehicle-treated tumor, n = 1; resistant tumors n = 3) and five
targeted therapy resistance. NRASMUT PDXs (vehicle-treated tumors, n = 5; resistant tumors,
n = 9). For early on-treatment samples, we prepared whole-genome
libraries from (i) one human BRAFV600MUT (M229; DMSO, n = 1;
METHODS BRAFi + MEKi, n = 1; BRAFi + MEKi + DNA-PKi, n = 1) and one
Human Subjects NRASMUT (M245; DMSO, n = 1; MEKi, n = 1; MEKi + DNA-PKi,
n = 1) cutaneous melanoma cell line and (ii) two BRAFV600MUT (vehi-
Patient characteristics related to clinical tissues are presented cle, n = 6; BRAFi + MEKi, n = 6; BRAFi + MEKi + DNA-PKi, n = 9)

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

in Supplementary Table S1. Patient-derived tissues were obtained and three NRASMUT (vehicle, n = 6; MEKi, n = 6; MEKi + DNA-PKi,
with written informed consent and approval by local institutional n = 7) cutaneous melanoma PDX models. Also, whole-genome
review boards. libraries were prepared for vehicle-treated tumors and early on-
treatment tumors from two NRASMUT PDX models. In total, paired-
Mice end sequencing was performed on 123 (104 tumors or cell lines, 19
NSG (NOD scid gamma) mice were obtained from the Radiation patient-matched normal tissues) genomes using Illumina NovaSeq
Oncology breeding colony at University of California, Los Angeles S4 with a read length of 2 × 150 bp and at a sequencing depth of 11
(UCLA). Male or female mice were used at 4 to 6 weeks of age. All to 98× (median, 26×).
animal experiments were conducted according to the guidelines
approved by the UCLA Animal Research Committee. WGS Data Analysis
WGS reads were mapped to the GRCh38/hg38 human reference
PDX Models and In Vivo Treatments genome using BWA-MEM (53). Alignments were sorted, and PCR
To develop PDX models, tumor fragments derived from meta- duplicates were removed using Samtools (54). CN variations were
static melanoma, with approval by the local institutional review called using two depth-of-coverage–based methods CNVkit (55)
boards, were transplanted subcutaneously in sex-matched NSG mice and ReadDepth (56). Default parameters were used for CNVkit.
(4–6 weeks old). One tumor fragment was implanted in each mouse. We used an FDR of 0.05 and overdispersion of 1 for ReadDepth
Tumors were measured with a caliper every 2 days, and tumor analysis. Structural variations reported by at least two SV detection
volumes were calculated using the formula (length × width2)/2. methods: SvABA (57), TIDDIT (58), and DELLY (59) were consid-
Tumors with volumes around 500 mm3 were randomly assigned ered. SVs in both DELLY and TIDDIT are determined by combining
into experimental groups. Special mouse diets were generated to discordant read pairs and split-reads, whereas TIDDIT additionally
reduce stress to animals by incorporating trametinib (LC Labo- uses depth-of-coverage signatures. SvABA utilizes discordant reads
ratories) into chow to achieve daily trametinib dosing at 3 or and genome-wide local assembly strategies for predicting SVs from
5 mg/kg/day or combined vemurafenib (LC Laboratories) and the genome. For high-coverage data (>15×), default parameters
trametinib dosing at 90 mg/kg/day and 0.7 mg/kg/day, respec- were considered for SvABA, TIDDIT, and DELLY. Parameter mini-
tively (Test Diet). DNA-PKi (NU7026, Selleckchem) was dissolved mum number of points (-l) was set to 5 in TIDDIT for low-coverage
in saline (UCLA Division of Laboratory Animal Medicine phar- data.
macy) and administered intraperitoneally at 6, 8, or 10 mg/kg/day.
We derived model- or patient-matched vehicle-treated tumors and Analysis of ecDNAs and CGRs
acquired trametinib-resistant tumors as plotted in Supplementary We carried out reconstruction of focal gene amplifications and elu-
Fig. S1A. We also derived model-matched vehicle-treated tumors cidation of ecDNA and CGRs using AmpliconArchitect (60). Briefly,
and early on-treatment tumors as plotted in Supplementary Fig. AmpliconArchitect determines the list of potential intervals for each
S6C and S6D. amplicon to be reconstructed, and within each amplicon, CNs and
SVs are estimated using read depth and discordant read signatures.
WGS It then constructs breakpoint graphs detailing sequence edges and
gDNA and total RNA were extracted from frozen tumor tissue breakpoint edges, and predicts copy counts of all edges. Simple cycles
preserved in RNALater or snap-frozen tumor tissue using the are then decomposed from the breakpoint graphs. Finally, Ampli-
Qiagen AllPrep DNA/RNA Mini Kit and the Ambion mirVana conClassifier classifies the amplicons into ecDNAs, CGRs, and linear
miRNA Isolation Kit. Normal gDNA from peripheral blood mono- amplicons. Amplicons are classified as ecDNAs if the segment(s)
nuclear cells (PBMC) were extracted from fresh or frozen PBMCs forms a head-to-tail structure, with size >10 kb, and CN >4.5; as
using the Qiagen FlexiGene DNA Kit. All gDNA were quantified complex genomic rearrangements for noncircular amplicons con-
using a NanoDrop (Thermo Fisher Scientific) and Qubit fluo- taining DNA segments from different chromosomes or regions that
rometer using the dsDNA BR Assay (Life Technologies), and then are far apart (>1 Mb) on chromosomes; or as linear amplicons for
gDNA size and quality were tested using TapeStation (Agilent) to linear amplifications. In our study, the initial set of CNV seed regions
ensure gDNA libraries were prepared using equal gDNA input and was inferred by ReadDepth and CNVKit. SV view of amplicons and
presence of a high-molecular-weight band. Whole-genome libraries circle plots of simple cycles were generated using functions available
were prepared using the Roche KAPA HyperPrep Kit. Briefly, after in AmpliconArchitect.

904 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

We carried out pathway enrichment analysis of genes within centromeric region of chr1; BRAF/CEN7 amplification probe (BRAF-
ecDNAs and/or CGRs using the Molecular Signatures Database CHR07-20-ORGR) targeting BRAF or centromeric region of chr7;
with pathways listed in Gene Ontology, Kyoto Encyclopedia of and RAF1/CEN3 amplification probe (RAF1-CHR03-20-ORGR) tar-
Genes and Genomes, Reactome, and Pathway Interaction Data- geting RAF1 or centromeric region of chr3, all from Empire Genom-
bases. For each tumor sample, we identified somatic SNVs using ics. The probes were mixed with the provided hybridization buffer
Strelka2 (61) with default parameters. Next, we estimated SBS sig- in a 1:4 ratio and applied onto the tissues or cells. FFPE samples
natures in regions within and without ecDNAs and CGRs using the were then denatured at 75°C in a slide moat for 7 minutes, whereas
nonnegative matrix factorization–based tool MutationalPatterns metaphase samples were denatured at 73°C for 2 minutes. We then
(62) with COSMIC SBS signatures V3.3 as reference. We carried performed hybridization overnight at 37°C in a humidified chamber.
out mutational signature analysis using SNVs in ecDNAs and/or The samples were then washed at 73°C with 0.3% Igepal/0.4× SSC for
CGR regions of (i) acquired MAPKi-resistant (mutations in patient- 2 minutes, followed by another 2-minute wash with 0.1% Igepal/2×
matched sensitive tumors subtracted from mutations in resistant SSC at room temperature. Finally, the tissue samples were stained
tumors) and (ii) MAPKi-sensitive genomes. To characterize whether with ProLong Diamond Antifade Mountant with DAPI (Invitrogen,
SBS mutational signatures are preferentially detected within (ver- P36966) and covered by coverslips. Images were acquired on a Leica
sus outside of) ecDNAs and/or CGRs amplicons, we calculated Confocal SP8-STED/FLIM/FCS microscope. For ecDNA quantifica-
the ratios of the proportions of signatures within ecDNA and/or tion, we counted the number of specific gene foci in each nucleus by
CGR sequences and the proportion of signatures outside ecDNAs ImageJ (version, 1.52a). For HSR quantification, we counted the area
and/or CGR sequences and defined these ratios as SBS mutational of gene foci larger than 2 μm2 to exclude ecDNA and non-HSR gene
signature enrichment score. We considered an enrichment score foci in each nucleus by ImageJ (version, 1.52a).

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

of 1 as the cutoff threshold. Scores >1 indicate SBS signatures
enriched within ecDNAs and/or CGRs, whereas scores <1 indicate RNA-seq Analysis
SBS signatures enriched in the background (non-ecDNA– and/or
Total RNAs were extracted from frozen tumor tissue preserved in
non-CGR–involved genomic regions). For clarity, only scores >1
RNALater of snap-frozen tumor tissues using the Qiagen AllPrep
and <1 were plotted.
DNA/RNA Mini Kit and the Ambion mirVana miRNA Isolation
We extracted enhancer elements and their connected genes from
Kit. Total RNAs were quantified by the Qubit RNA High Sensitivity
GeneHancer (version4-4, GeneCards). Genehancer is an integrated
kit (Thermo Fisher Scientific) and/or using a NanoDrop (Thermo
database of human enhancers and their inferred target genes, mined
Fisher Scientific). RNA size and quality were measured using an
from four different genome-wide databases: ENCODE, FANTOM,
Agilent 2100 Bioanalyzer (Agilent Technologies). RNA libraries were
the VISTA enhancer browser, and the Ensembl regulatory build.
constructed using the NuGen Universal Plus mRNA-Seq Kit. Briefly,
The enhancer–target genes associations were obtained from eQTLs,
after fragmentation of total RNA and double-stranded cDNA gen-
CHi-C, eRNA coexpression, transcription factor coexpression, and
eration using a mixture of random and oligo(dT) primers, the RNA
gene–enhancer distance methods. Annotations of superenhancers
libraries were constructed by end repairing, adapter ligation, strand
were obtained from dbSUPER (63). Enhancers were first assessed
selection, and PCR amplification. Libraries were quantified for equal
for their presence on CGR and ecDNA amplicons, and, if so, their
molar pooling and paired-end sequenced with a read length of 2 × 50
connected oncogenes were searched for their presence within CGR/
bp on the Illumina NovaSeq 6000 S4 platform.
ecDNA amplicons (cis interaction) or elsewhere in the chromosome
We mapped paired-end reads of patient-matched tumor samples
(trans interaction). Furthermore, we obtained H3K27 acetylation
from 10 patients and 6 PDX models to the GRChr38 human ref-
peaks from seven cell lines (GM12878, H1-hESC, HSMM, HUVEC,
erence genome using STAR aligner (64) with default parameters.
K562, NHEK, and NHLF) listed in ENCODE. The peaks aligned to
Log2 fold change values were calculated for each acquired resist-
ecDNA regions in Pt9-DD-DP2 (MYC) and Mel_PDX27-R2 (BRAF)
ant tumor compared with patient- or model-matched pretreatment
were obtained from the UCSC Genome Browser.
tumor or vehicle-treated tumor, respectively. Genes with an abso-
lute fold change >2 were considered significant. Gene counts were
Metaphase Chromosome Spread estimated using FeatureCounts (65) with GENCODE (version 38)
Cells in metaphase were prepared by colcemid (Sigma) treatment annotations. Feature counts were then normalized using trimmed
at 10 μg/mL for several hours (depending on the growth rate). The mean of M values (TMM) approach and log2 transformed (66). Fold
single-cell suspensions were collected, washed with PBS, and then changes for each gene were estimated by computing the difference
treated with 0.075 M KCl (Gibco) for 15 minutes. We then fixed and between the normalized values of acquired resistant tumors and
washed cells with 3:1 methanol:acetic acid. The final cell pellet was MAPKi-sensitive/naive transcriptomes.
resuspended with the fixative and dropped onto a humidified slide.
Analysis of Chromothripsis
DNA-FISH To infer chromothriptic events in the genomes, we applied Shat-
Formalin-fixed, paraffin-embedded (FFPE) tissue samples were terSeek (12), an in silico scoring algorithm based on the clusters of
baked at 90°C for 25 minutes in an oven and immersed in 100% xylene interleaved SVs, CN oscillations, and interchromosomal SVs. CN
and then 100% ethanol, each for 10 minutes to deparaffinize tissues. and SVs described above were considered as input for ShatterSeek.
Air-dried tumor tissues were pretreated in 90°C to 95°C 10 mmol/L High-confidence chromothriptic events were selected based on the
in citric acid buffer (pH 6.8, Thermo Fisher Scientific, 327162500) for statistical criteria recommended by the authors. Briefly, a high-
30 minutes and washed in 2× SSC buffer (Invitrogen, 15557044) for confidence chromothriptic event is characterized with (i) a cluster of
5 minutes. Then, FFPE slides were digested in 37°C pepsin solution SVs (>6 DUP/DEL/h2hINV/t2tINV), (ii) oscillating CN between two
(Thermo Fisher Scientific, J6167906) for 20 to 30 minutes, washed in states (>7 CN events), (iii) chromosomal enrichment and distribution
2× SSC buffer for 5 minutes, and dehydrated in ascending ethanol of DNA breakpoints (P < 0.05), (iv) randomness of fragment joins
series (70%, 85%, and 100%), each for 2 minutes. For metaphase DNA- (P > 0.05), and/or (v) interchromosomal rearrangements between
FISH, fixed cells in interphase or metaphase on slides were dehy- multiple chromosomes. Somatic SNVs spanning chromothripsis
drated in ascending ethanol series (70%, 85%, and 100%), each for regions were identified using Strelka2 (61) with default parameters
2 minutes. We used the following DNA-FISH probes: NRAS/CEN1 and defined as chromothripsis-associated SNVs. Regions with and
amplification probe (NRAS-CHR01-20-ORGR) targeting NRAS or without chromothripsis were extracted and TMBs were computed

APRIL 2023 CANCER DISCOVERY | 905

RESEARCH ARTICLE Dharanipragada et al.

within these regions for each tumor sample. Ratios between TMBs Human PRKDC sh4: GGTCATGGATTCAAGAAAT
within and outside of chromothriptic regions were calculated. Human LIG4 sh1: TTCGACGCCACACCGTTTATT
Mutational signature analysis was carried out using SNVs in chro- Human LIG4 sh2: TATGTCAGTGGACTAATGGAT
mothriptic regions of (i) acquired MAPKi-resistant (mutations in
patient-matched sensitive tumors subtracted from mutations in
Cell Growth Assays
resistant tumors) and (ii) MAPKI-sensitive genomes. Mutational
(SBS, DBS, and ID) signatures for each sample were predicted using For clonogenic assay, cells were plated at indicated cell densities
nonnegative matrix factorization–based tool MutationalPatterns in 6-well plates and treated with inhibitor(s) the next day. Inhibi-
(62), and the extracted signatures were compared with the COSMIC tors and media were replenished every 2 days for the number
SBS, DBS, and ID signature database. of days indicated. Colonies were fixed in 4% paraformaldehyde,
followed by staining with 0.1% crystal violet. For the MTT assay,
Junctional Sequence Analysis of Amplicon Breakpoints cells were plated at 2,000 cells per well in 96-well plates (Fig. 4P
and Q); parental and acquired resistant sublines were seeded at
Sequences of CGRs and ecDNAs identified in all the genomes were 4,000 cells per well and treated with graded concentrations of
analyzed. Breakpoints of amplicons harboring key MAPKi-resistance MAPKi the next day (Fig. 5D–F); and media were replenished every
genes, BRAF, NRAS, HRAS, and EGFR, were extracted, and sequences 2 to 3 days. Methylthiazolyldiphenyl-tetrazolium bromide (MTT;
spanning these junctions were further analyzed. Homologous 100 μL) solution (0.5 mg/mL, Sigma, M5655) was added to each well
sequences and insertions for the breakpoint junctions were obtained and incubated at 37°C for 2 hours for MTT formazan formation.
from SvABA, and a possible mechanism for DNA DSB repair was Formazan was then dissolved in 100 μL DMSO, and the absorbance
inferred. First, HRR was identified with large homologous sequences

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

values were measured using an ELISA reader (SpectraMax Plus 384)
and insertions (≥8 bp). Second, MMEJ or alt-NHEJ was identified at the wavelength of 570 nm, blanked with DMSO solution. Experi-
with breakpoints comprising homologous sequences (2–8 bp). Third, ments were performed with 4 replicates.
NHEJ was identified when neither of the sequence types described
above was identified in the vicinity of the breakpoints.
Western Blots
Cell Lines and Inhibitor Treatments Cells were lysed in RIPA buffer (Thermo Fisher Scientific) with pro-
All cell lines were routinely tested for Mycoplasma and profiled tease inhibitor cocktail (Thermo Fisher Scientific) and phosphatase
and identified by RNA-seq and the GenePrint 10 system (Pro- inhibitor cocktails (Thermo Fisher Scientific) for Western blotting.
mega) at periodic intervals during the course of this study. All M BCA protein assay (Thermo Fisher Scientific, PI23227) was used to
series cell lines were established from patient-derived tumors at determine the protein concentration. Antibodies used in Western
the UCLA with institutional review board approval. All M cell lines blot were as follows: TUBULIN (Cell Signaling Technology, 2144S),
with acquired MAPKi resistance were derived in the Lo Laboratory PRKDC (Cell Signaling Technology, 38168S), and LIG4 (Cell Signal-
and published previously (3, 5, 8, 9, 11, 22, 40, 67). We maintained ing Technology, 14649S).
H358 (ATCC) in RPMI 1640 (Gibco) with 10% heat-inactivated FBS
(Omega Scientific, FB-02) and 2 mmol/L glutamine in a humidified, Immunofluorescence Analysis
5% CO2 incubator at 37°C; all other cell lines were maintained in Tumor tissues were fixed in formalin followed by paraffin embed-
high-glucose DMEM (Omega Scientific, DM-22) with 10% heat- ding. After deparaffinization and rehydration, tissue sections were
inactivated FBS (Omega Scientific, FB-02) and 2 mmol/L glutamine antigen retrieved by heat. Permeabilization and blocking were fol-
in a humidified, 5% CO2 incubator at 37°C. We obtained inhibitors lowed by overnight incubation with primary antibodies [p-ERK1/2
from the following sources: PLX4032 (Plexxikon), AZD6244 (Sell- (Cell Signaling Technology, 4370), p-DNA-PKcs (Abcam, ab124918),
eck Chemicals), trametinib (LC Laboratories), NU7026 (Abcam, p-H2AX (Cell Signaling Technology, 9718)]. Immunofluorescence
ab120970), ABT888 (Enzo, ALX-270-444-M005), VX984 (MCE, HY- was performed with Alexa Fluor–conjugated secondary antibodies
19939S), AZD7648 (TargetMol, T7122), olaparib (LC Laboratories, (Life Technologies, A-21429). Nuclei were counterstained by DAPI.
763113-22–0), MRTX849 (Selleckchem, S8884), AMG510 (Selleck- Signals were captured with a Zeiss microscope (AXIO Imager A1)
chem, S8830), and BGB-283 (BeiGene, via a Material Transfer Agree- mounted with a charge-coupled device camera (Retiga EXi QImag-
ment with UCLA). All inhibitors were dissolved in DMSO and stored ing), and images were captured by Image-Pro plus 6.0.
at −20°C.

Statistical Methods
Lentivirus Production, Transduction, and shRNA Sequences
Statistical analysis for genomic data was conducted in R.4.02,
shRNAs for PRKDC and vector control (pGIPZ) were obtained
Python 3.8.0, and Python 2.7.17. Statistical analyses, described for
from Robert Damoiseaux, PhD (Molecular Screening Shared
specific experiments, for cell line– or PDX tumor–based assays were
Resource, UCLA); shRNAs for LIG4 were from Sigma. All shRNAs
performed using GraphPad Prism.
were packaged into lentiviral particles for infection. The lentiviral
viruses were generated by transfection of the constructs together
with pMD2.G, pRSV-Rev, and pMDLg/pRRE into HEK-293T cells Data Availability
using calcium phosphate. Fourteen hours after transfection, media The BAM files of WGS data are deposited in the European
were replaced with preheated fresh media. Virus particles were Genome-phenome Archive (https://www.ebi.ac.uk/ega/) with the
harvested 24 and 48 hours later and filtered by a 0.45 μm filter unit accession number EGAS00001006874. This study did not generate
(Millipore). Cells were transduced with recombinant lentivirus with custom codes. Any additional information regarding data reported in
10 μg/mL polybrene (Millipore, TR-1003-G) for 48 hours and then this article is available from the corresponding author upon request.
selected by puromycin (Sigma, P8833) for 1 week. shRNA targeting There are no restrictions on data availability.
sequences were as follows:

Human PRKDC sh1: CAGTGAAAGTCTGAATCAT Authors’ Disclosures

Human PRKDC sh2: GTCATGGATTCAAGAAATA A. Ribas reports personal fees (honoraria) from Amgen, Bristol
Human PRKDC sh3: GAGCTTACATGCTAATGTA Myers Squib, Chugai, Genentech, Merck, Novartis, Roche, Sanofi,

906 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

and Vedanta, personal fees (scientific advisory board member, hono- fact, this article is hereby marked “advertisement” in accordance with
raria, and stock) from 4C Biomed, Appia, Apricity, Arcus, High- 18 USC section 1734.
light, Compugen, ImaginAb, Kalthera/ImmPACT Bio, MapKure,
Merus, Rgenix, Lutris, PACT Pharma, Synthekine, Tango, Advaxis, Note
CytomX, Five Prime, RAPT, Isoplexis, and Kite/Gilead, and grants
Supplementary data for this article are available at Cancer Discovery
from Agilent and Bristol Myers Squibb outside the submitted work.
Online (http://cancerdiscovery.aacrjournals.org/).
S.J. Moschos reports grants from Merck during the conduct of the
study, as well as grants from Merck, Amgen, and Syndax Pharma and Received July 12, 2022; revised November 27, 2022; accepted
other support from Castle Biosciences outside the submitted work. January 23, 2023; published first January 26, 2023.
R.S. Lo reports grants from Merck, OncoSec, Pfizer, and Bristol
Myers Squibb outside the submitted work, as well as a patent for
U.S. Patent Application Serial No.: 63/268,028 issued. No disclo-
REFERENCES
sures were reported by the other authors.
1. Flaherty KT, Puzanov I, Kim KB, Ribas A, McArthur GA, Sosman JA,
et al. Inhibition of mutated, activated BRAF in metastatic melanoma.
Authors’ Contributions N Engl J Med 2010;363:809–19.
P. Dharanipragada: Data curation, formal analysis, investigation, 2. Flaherty KT, Infante JR, Daud A, Gonzalez R, Kefford RF, Sosman J,
visualization, methodology, writing–original draft, writing–review et al. Combined BRAF and MEK inhibition in melanoma with BRAF
and editing. X. Zhang: Formal analysis, investigation, visualiza- V600 mutations. N Engl J Med 2012;367:1694–703.
tion, methodology, writing–review and editing. S. Liu: Data cura- 3. Nazarian R, Shi H, Wang Q, Kong X, Koya RC, Lee H, et al. Melano-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

tion, formal analysis, methodology, writing–review and editing. mas acquire resistance to B-RAF(V600E) inhibition by RTK or N-RAS
S.H. Lomeli: Resources, methodology, writing–review and edit- upregulation. Nature 2010;468:973–7.
ing. A. Hong: Formal analysis, visualization, methodology, writ- 4. Shi H, Moriceau G, Kong X, Koya RC, Nazarian R, Pupo GM, et al.
ing–review and editing. Y. Wang: Methodology, writing–review and Preexisting MEK1 exon 3 mutations in V600E/KBRAF melanomas
editing. Z. Yang: Resources, methodology, writing–review and edit- do not confer resistance to BRAF inhibitors. Cancer Discov 2012;2:
ing. K.Z. Lo: Resources, methodology. A. Vega-Crespo: Resources. 414–24.
5. Shi H, Moriceau G, Kong X, Lee MK, Lee H, Koya RC, et al. Melanoma
A. Ribas: Resources. S.J. Moschos: Resources, methodology, writ-
whole-exome sequencing identifies (V600E)B-RAF amplification-
ing–review and editing. G. Moriceau: Resources, formal analysis,
mediated acquired B-RAF inhibitor resistance. Nat Commun 2012;
visualization, methodology, writing–review and editing. R.S. Lo:
3:724.
Conceptualization, resources, data curation, formal analysis, super-
6. Long GV, Eroglu Z, Infante J, Patel S, Daud A, Johnson DB, et al.
vision, funding acquisition, validation, investigation, visualiza- Long-term outcomes in patients with BRAF V600-mutant metastatic
tion, methodology, writing–original draft, project administration, melanoma who received dabrafenib combined with trametinib. J Clin
writing–review and editing. Oncol 2018;36:667–73.
7. Hugo W, Shi H, Sun L, Piva M, Song C, Kong X, et al. Non-genomic
Acknowledgments and immune evolution of melanoma acquiring MAPKi resistance.
Cell 2015;162:1271–85.
We thank all members of the Lo Laboratory for their critical 8. Moriceau G, Hugo W, Hong A, Shi H, Kong X, Yu CC, et al. Tunable-
comments. This research was supported by grants (to R.S. Lo) from combinatorial mechanisms of acquired resistance limit the efficacy
the NIH (1R01CA176111A1, 1R21CA215910-01, R21CA255837- of BRAF/MEK cotargeting but result in melanoma drug addiction.
01, and 1P01CA168585), the Melanoma Research Alliance (MRA; Cancer Cell 2015;27:240–56.
Team Science Awards 2020 and 2022), and the V Foundation for 9. Shi H, Hugo W, Kong X, Hong A, Koya RC, Moriceau G, et al.
Cancer Research (Translational Award). Additional funding was Acquired resistance and clonal evolution in melanoma during BRAF
provided by NIH 1P01CA168585 (to G. Moriceau), an MRA Der- inhibitor therapy. Cancer Discov 2014;4:80–93.
matology Fellows Award (P. Dharanipragada, S. Liu, and Z. Yang), 10. Dummer R, Schadendorf D, Ascierto PA, Arance A, Dutriaux C,
an MRA Young Investigator Award (G. Moriceau), a Jonsson Com- Di Giacomo AM, et al. Binimetinib versus dacarbazine in patients
prehensive Cancer Center (JCCC) Postdoctoral Seed Grant (to Z. with advanced NRAS-mutant melanoma (NEMO): a multicen-
Yang), JCCC Postdoctoral Fellowships (to P. Dharanipragada, X. tre, open-label, randomised, phase 3 trial. Lancet Oncol 2017;18:
Zhang, S. Liu, and Z. Yang), and the NIH T32CA009120 Tumor 435–45.
Immunology Postdoctoral Fellowship (to A. Hong). S. Liu is the 11. Hong A, Piva M, Liu S, Hugo W, Lomeli SH, Zoete V, et al. Dura-
ble suppression of acquired MEK inhibitor resistance in cancer by
recipient of a Career Development Award from the Melanoma
sequestering MEK from ERK and promoting antitumor T-cell immu-
Research Foundation. Additional support came from Mary Tanner
nity. Cancer Discov 2021;11:714–35.
and Maurizio Grimaldi (R.S. Lo), the Ressler Family Foundation
12. Cortes-Ciriano I, Lee JJ, Xi R, Jain D, Jung YL, Yang L, et al. Compre-
(R.S. Lo), and the University Cancer Research Fund [Lineberger
hensive analysis of chromothripsis in 2,658 human cancers using
Comprehensive Cancer Center, The University of North Carolina at whole-genome sequencing. Nat Genet 2020;52:331–41.
Chapel Hill (UNC-CH)]. We thank Mr. Vincent J. Moylan Jr, M.S., 13. Song K, Minami JK, Huang A, Dehkordi SR, Lomeli SH, Luebeck J,
P.A. (Assistant Professor of Pathology and Laboratory Medicine, et al. Plasticity of extrachromosomal and intrachromosomal BRAF
UNC-CH), Medical Examiner for Orange County, NC, and Dr. amplifications in overcoming targeted therapy dosage challenges.
Leigh B. Thorne (Professor of Pathology and Laboratory Medi- Cancer Discov 2022;12:1046–69.
cine, UNC-CH) for performing warm autopsies and quality-control 14. Nathanson DA, Gini B, Mottahedeh J, Visnyei K, Koga T,
analysis of tissues used. We thank Xinmin Li, PhD (Director), and Gomez G, et al. Targeted therapy resistance mediated by dynamic reg-
the Technology Center for Genomics and Bioinformatics at UCLA ulation of extrachromosomal mutant EGFR DNA. Science 2014;343:
for excellent technical support. We also thank Robert Damoiseaux, 72–6.
PhD (Director), and the Molecular Screening Shared Resource for 15. Turner KM, Deshpande V, Beyter D, Koga T, Rusert J, Lee C, et al.
sharing resources. Extrachromosomal oncogene amplification drives tumour evolution
and genetic heterogeneity. Nature 2017;543:122–5.
The publication costs of this article were defrayed in part by the 16. Helmsauer K, Valieva ME, Ali S, Chamorro Gonzalez R, Schopflin R,
payment of publication fees. Therefore, and solely to indicate this Roefzaad C, et al. Enhancer hijacking determines extrachromosomal

APRIL 2023 CANCER DISCOVERY | 907

RESEARCH ARTICLE Dharanipragada et al.

circular MYCN amplicon architecture in neuroblastoma. Nat Com- 40. Song C, Piva M, Sun L, Hong A, Moriceau G, Kong X, et al. Recurrent
mun 2020;11:5823. tumor cell-intrinsic and -extrinsic alterations during MAPKi-induced mel-
17. Hung KL, Yost KE, Xie L, Shi Q, Helmsauer K, Luebeck J, et al. ecDNA anoma regression and early adaptation. Cancer Discov 2017;7:1248–65.
hubs drive cooperative intermolecular oncogene expression. Nature 41. Harnor SJ, Brennan A, Cano C. Targeting DNA-dependent protein
2021;600:731–6. kinase for cancer therapy. ChemMedChem 2017;12:895–900.
18. Wu S, Bafna V, Mischel PS. Extrachromosomal DNA (ecDNA) in 42. Yang C, Tian C, Hoffman TE, Jacobsen NK, Spencer SL. Melanoma
cancer pathogenesis. Curr Opin Genet Dev 2021;66:78–82. subpopulations that rapidly escape MAPK pathway inhibition incur
19. Wu S, Turner KM, Nguyen N, Raviram R, Erb M, Santini J, et al. DNA damage and rely on stress signalling. Nat Commun 2021;12:1747.
Circular ecDNA promotes accessible chromatin and high oncogene 43. Cowell JK, Miller OJ. Occurrence and evolution of homogeneously
expression. Nature 2019;575:699–703. staining regions may be due to breakage-fusion-bridge cycles follow-
20. Shoshani O, Brunner SF, Yaeger R, Ly P, Nechemia-Arbely Y, Kim DH, ing telomere loss. Chromosoma 1983;88:216–21.
et al. Chromothripsis drives the evolution of gene amplification in 44. Koche RP, Rodriguez-Fos E, Helmsauer K, Burkert M, MacArthur IC,
cancer. Nature 2021;591:137–41. Maag J, et al. Extrachromosomal circular DNA drives oncogenic
21. Van Allen EM, Wagle N, Sucker A, Treacy DJ, Johannessen CM, genome remodeling in neuroblastoma. Nat Genet 2020;52:29–34.
Goetz EM, et al. The genetic landscape of clinical resistance to RAF 45. Rosswog C, Bartenhagen C, Welte A, Kahlert Y, Hemstedt N, Lorenz W,
inhibition in metastatic melanoma. Cancer Discov 2014;4:94–109. et al. Chromothripsis followed by circular recombination drives onco-
22. Hong A, Moriceau G, Sun L, Lomeli S, Piva M, Damoiseaux R, et al. gene amplification in human cancer. Nat Genet 2021;53:1673–85.
Exploiting drug addiction mechanisms to select against MAPKi- 46. Ly P, Cleveland DW. Rebuilding chromosomes after catastrophe:
resistant melanoma. Cancer Discov 2018;8:74–93. emerging mechanisms of chromothripsis. Trends Cell Biol 2017;27:
23. Chadee DN, Xu D, Hung G, Andalibi A, Lim DJ, Luo Z, et al. Mixed- 917–30.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

lineage kinase 3 regulates B-Raf through maintenance of the B-Raf/ 47. Ly P, Teitz LS, Kim DH, Shoshani O, Skaletsky H, Fachinetti D, et al.
Raf-1 complex and inhibition by the NF2 tumor suppressor protein. Selective Y centromere inactivation triggers chromosome shattering
Proc Natl Acad Sci U S A 2006;103:4463–8. in micronuclei and repair by non-homologous end joining. Nat Cell
24. Hung KL, Luebeck J, Dehkordi SR, Colon CI, Li R, Wong IT, et al. Biol 2017;19:68–75.
Targeted profiling of human extrachromosomal DNA by CRISPR- 48. Maciejowski J, Li Y, Bosco N, Campbell PJ, de Lange T. Chromothrip-
CATCH. Nat Genet 2022;54:1746–54. sis and kataegis induced by telomere crisis. Cell 2015;163:1641–54.
25. Morton AR, Dogan-Artun N, Faber ZJ, MacLeod G, Bartels CF, 49. Tang S, Stokasimov E, Cui Y, Pellman D. Breakage of cytoplasmic
Piazza MS, et al. Functional enhancers shape extrachromosomal chromosomes by pathological DNA base excision repair. Nature
oncogene amplifications. Cell 2019;179:1330–41. 2022;606:930–6.
26. Zhu Y, Gujar AD, Wong CH, Tjong H, Ngan CY, Gong L, et al. 50. Kim H, Nguyen NP, Turner K, Wu S, Gujar AD, Luebeck J, et al.
Oncogenic extrachromosomal DNA functions as mobile enhancers Extrachromosomal DNA is associated with oncogene amplification
to globally amplify chromosomal transcription. Cancer Cell 2021;39: and poor outcome across multiple cancers. Nat Genet 2020;52:891–7.
694–707. 51. Labrie M, Brugge JS, Mills GB, Zervantonakis IK. Therapy resistance:
27. Fishilevich S, Nudel R, Rappaport N, Hadar R, Plaschkes I, Iny Stein T, opportunities created by adaptive responses to targeted therapies in
et al. GeneHancer: genome-wide integration of enhancers and target cancer. Nat Rev Cancer 2022;22:323–39.
genes in GeneCards. Database (Oxford) 2017;2017:bax028. 52. Hammarsten O, DeFazio LG, Chu G. Activation of DNA-dependent
28. Rausch T, Jones DT, Zapatka M, Stutz AM, Zichner T, Weischenfeldt J, protein kinase by single-stranded DNA ends. J Biol Chem 2000;275:
et al. Genome sequencing of pediatric medulloblastoma links cata- 1541–50.
strophic DNA rearrangements with TP53 mutations. Cell 2012;148: 53. Li H, Durbin R. Fast and accurate short read alignment with Burrows-
59–71. Wheeler transform. Bioinformatics 2009;25:1754–60.
29. Stephens PJ, Greenman CD, Fu B, Yang F, Bignell GR, Mudie LJ, et al. 54. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al.
Massive genomic rearrangement acquired in a single catastrophic The sequence alignment/map format and SAMtools. Bioinformatics
event during cancer development. Cell 2011;144:27–40. 2009;25:2078–9.
30. Zhang CZ, Spektor A, Cornils H, Francis JM, Jackson EK, Liu S, 55. Talevich E, Shain AH, Botton T, BBC CNVkit.: Genome-wide copy
et al. Chromothripsis from DNA damage in micronuclei. Nature number detection and visualization from targeted DNA sequencing.
2015;522:179–84. PLoS Comput Biol 2016;12:e1004873.
31. Krupina K, Goginashvili A, Cleveland DW. Causes and consequences 56. Miller CA, Hampton O, Coarfa C, Milosavljevic A. ReadDepth: a
of micronuclei. Curr Opin Cell Biol 2021;70:91–9. parallel R package for detecting copy number alterations from short
32. Alexandrov LB, Kim J, Haradhvala NJ, Huang MN, Tian Ng AW, sequencing reads. PLoS One 2011;6:e16327.
Wu Y, et al. The repertoire of mutational signatures in human cancer. 57. Wala JA, Bandopadhayay P, Greenwald NF, O’Rourke R, Sharpe T,
Nature 2020;578:94–101. Stewart C, et al. SvABA: genome-wide detection of structural variants
33. Bergstrom EN, Luebeck J, Petljak M, Khandekar A, Barnes M, Zhang T, and indels by local assembly. Genome Res 2018;28:581–91.
et al. Mapping clustered mutations in cancer reveals APOBEC3 58. Eisfeldt J, Vezzi F, Olason P, Nilsson D, Lindstrand A. TIDDIT, an
mutagenesis of ecDNA. Nature 2022;602:510–7. efficient and comprehensive structural variant caller for massive par-
34. Chang HHY, Pannunzio NR, Adachi N, Lieber MR. Non-homologous allel sequencing data. F1000Res 2017;6:664.
DNA end joining and alternative pathways to double-strand break 59. Rausch T, Zichner T, Schlattl A, Stutz AM, Benes V, Korbel JO.
repair. Nat Rev Mol Cell Biol 2017;18:495–506. DELLY: structural variant discovery by integrated paired-end and
35. Korbel JO, Campbell PJ. Criteria for inference of chromothripsis in split-read analysis. Bioinformatics 2012;28:i333–i9.
cancer genomes. Cell 2013;152:1226–36. 60. Deshpande V, Luebeck J, Nguyen ND, Bakhtiari M, Turner KM,
36. Maher CA, Wilson RK. Chromothripsis and human disease: piecing Schwab R, et al. Exploring the landscape of focal amplifications in
together the shattering process. Cell 2012;148:29–32. cancer using AmpliconArchitect. Nat Commun 2019;10:392.
37. Dylgjeri E, Knudsen KE. DNA-PKcs: a targetable protumorigenic 61. Kim S, Scheffler K, Halpern AL, Bekritsky MA, Noh E, Kallberg M,
protein kinase. Cancer Res 2022;82:523–33. et al. Strelka2: fast and accurate calling of germline and somatic vari-
38. Patterson-Fortin J, D’Andrea AD. Exploiting the microhomology- ants. Nat Methods 2018;15:591–4.
mediated end-joining pathway in cancer therapy. Cancer Res 2020; 62. Blokzijl F, Janssen R, van Boxtel R, Cuppen E. MutationalPatterns: com-
80:4593–600. prehensive genome-wide analysis of mutational processes. Genome
39. Sharma SV, Lee DY, Li B, Quinlan MP, Takahashi F, Maheswaran S, Med 2018;10:33.
et al. A chromatin-mediated reversible drug-tolerant state in cancer 63. Khan A, Zhang X. dbSUPER: a database of super-enhancers in mouse
cell subpopulations. Cell 2010;141:69–80. and human genome. Nucleic Acids Res 2016;44:D164–71.

908 | CANCER DISCOVERY APRIL 2023 AACRJournals.org

DNA-PKCS/NHEJ Underlie ecDNAs/CGRs Driving MAPKi Resistance RESEARCH ARTICLE

64. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. 66. Robinson MD, Oshlack A. A scaling normalization method for differ-
STAR: ultrafast universal RNA-seq aligner. Bioinformatics 2013;29:15–21. ential expression analysis of RNA-seq data. Genome Biol 2010;11:R25.
65. Liao Y, Smyth GK, Shi W. featureCounts: an efficient general purpose 67. Shi H, Hong A, Kong X, Koya RC, Song C, Moriceau G, et al. A Novel
program for assigning sequence reads to genomic features. Bioinfor- AKT1 mutant amplifies an adaptive melanoma response to BRAF
matics 2014;30:923–30. inhibition. Cancer Discov 2014;4:69–79.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/13/4/880/3319047/880.pdf by guest on 29 June 2024

APRIL 2023 CANCER DISCOVERY | 909