Article

Pancreatic cancer intrinsic PI3Ka activity

accelerates metastasis and rewires
macrophage component
Benoit Thibault1,2,† , Fernanda Ramos-Delgado1,2,†, Elvire Pons-Tostivint1,2,†, Nicole Therville1,2,
Celia Cintas1,2, Silvia Arcucci1,2, Stephanie Cassant-Sourdy1,2, Gabriela Reyes-Castellanos3,
Marie Tosolini1 , Amelie V Villard1,2, Coralie Cayron1,2, Romain Baer1,2, Justine Bertrand-Michel4,
Delphine Pagan1, Dina Ferreira Da Mota1,5, Hongkai Yan6,7, Chiara Falcomat
a6,7, Fabrice Muscari1,5,
Barbara Bournet1,5, Jean-Pierre Delord1,5, Ezra Aksoy8, Alice Carrier3 , Pierre Cordelier1,
Dieter Saur6,7 , Celine Basset1,2,5 & Julie Guillermet-Guibert1,2,*

Abstract Keywords pancreatic cancer; phosphoinositide; PI3K isoforms; targeted

therapy; tumour-stroma dialog
Pancreatic ductal adenocarcinoma (PDAC) patients frequently suf- Subject Categories Cancer; Signal Transduction
fer from undetected micro-metastatic disease. This clinical situa- DOI 10.15252/emmm.202013502 | Received 23 September 2020 | Revised 17
tion would greatly benefit from additional investigation. April 2021 | Accepted 23 April 2021 | Published online 25 May 2021
Therefore, we set out to identify key signalling events that drive EMBO Mol Med (2021) 13: e13502
metastatic evolution from the pancreas. We searched for a gene See also: GL Raja Arul & ME Fernandez‐Zapico (July 2021)
signature that discriminate localised PDAC from confirmed meta-
static PDAC and devised a preclinical protocol using circulating
cell-free DNA (cfDNA) as an early biomarker of micro-metastatic Introduction
disease to validate the identification of key signalling events. An
unbiased approach identified, amongst actionable markers of In humans, PI3Ks are composed of 8 isoforms distributed into 3
disease progression, the PI3K pathway and a distinctive PI3Ka classes. Each class I PI3K dimers, namely called PI3Ka, PI3Kb,
activation signature as predictive of PDAC aggressiveness and PI3Kc and PI3Kd, are composed of a catalytic subunit (p110a,
prognosis. Pharmacological or tumour-restricted genetic PI3Ka- p110b, p110c and p110d) and a regulatory subunit (p85 for a, b and
selective inhibition prevented macro-metastatic evolution by c, and p101/p87 for c). PI3Ka and PI3Kb are ubiquitously
hindering tumoural cell migratory behaviour independently of expressed. PI3Kc and PI3Kd are restricted to the cardiovascular
genetic alterations. We found that PI3Ka inhibition altered the system and leucocytes in normal tissues, but can be overexpressed
quantity and the species composition of the produced lipid second in solid tumours (Vanhaesebroeck et al, 2010). PI3Ks are lipid
messenger PIP3, with a selective decrease of C36:2 PI-3,4,5-P3. kinases that phosphorylate phosphatidylinositol 4,5–biphosphate
Tumoural PI3Ka inactivation prevented the accumulation of pro- (PIP2) into PIP3 which acts as a second messenger and regulates
tumoural CD206-positive macrophages in the tumour-adjacent various functions in normal and tumour cells via the PI3K/Akt/
tissue. Tumour cell-intrinsic PI3Ka promotes pro-metastatic mTOR pathway. The PI3K/Akt axis is frequently hyper-activated in
features that could be pharmacologically targeted to delay macro- cancers and has been tested as a clinical target in recent years
metastatic evolution. (Pons-Tostivint et al, 2017; Goncalves et al, 2018). PI3K inhibitors
are currently described as cytostatic agents, as PI3K activity is

1 Centre de Recherches en Cancerologie de Toulouse, Inserm, CNRS, Universite de Toulouse, Toulouse, France
2 LABEX TouCAN, Toulouse, France
3 Aix Marseille Univ, CNRS, INSERM, Institut Paoli-Calmettes, CRCM, Marseille, France
4 Lipidomics, I2MC, Inserm, Toulouse, France
5 Institut Universitaire du Cancer de Toulouse – Oncopole (IUCT-O), Hopitaux de Toulouse, Institut Claudius Regaud Toulouse, France
6 Division of Translational Cancer Research, German Cancer Research Center (DKFZ) and German Cancer Consortium (DKTK), Heidelberg, Germany
7 Chair of Translational Cancer Research and Institute of Experimental Cancer Therapy, Klinikum rechts der Isar, School of Medicine, Technische Universit€at Mu
€nchen,
Munich, Germany
8 Centre for Biochemical Pharmacology, William Harvey Research Institute, Queen Mary University of London, London, UK
*Corresponding author. Tel: +33 5 82 74 16 52; E-mail: [Link]@[Link]
†
These authors contributed equally to this work

ª 2021 The Authors. Published under the terms of the CC BY 4.0 license EMBO Molecular Medicine 13: e13502 | 2021 1 of 20
 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

critically driving oncogenesis in a cell-autonomous manner. pathway was the most differentially expressed with the lowest P-
However, the clinical importance of other cell functions regulated value (Fig 1A). Differential enrichment analysis demonstrates that
by this pathway in a non-cancer cell-autonomous manner, particu- the PI3K/Akt/mTOR pathway is constantly (equally) enriched in
larly on macrophages, has been underestimated. localised and metastatic PDAC compared to normal parenchyma.
Pancreatic ductal adenocarcinoma (PDAC) is a lethal cancer This trend was confirmed following Reactome pathway analysis;
(Neoptolemos et al, 2018) where activation of class I PI3K is high however, we found that PI3K cascade to FGFR2 was significantly
and linked to poor prognosis (Schlieman et al, 2003). Localised, increased in PDACmet as opposed to PDACloc, suggesting differen-
locally advanced and metastatic PDAC are characterised by early tial activation of receptor tyrosine kinase (RTK)-coupled PI3K in
surgical relapse and failure of long-term disease control with these samples. Considering that PI3Ka is key for insulin signalling
chemotherapies. Molecular characterisation of large cohorts of (Vanhaesebroeck et al, 2005), angiogenesis (Graupera et al, 2008)
PDAC patients demonstrates that oncogenic KRAS mutations on G12 and PDAC initiation (Baer et al, 2014), we then designed a PI3Ka
position are found in more than 80% of all patients. There are multi- activation gene signature, based on expression levels of PI3Ka-
ple altered signalling pathways downstream of oncogenic KRAS, regulated curated genes (Appendix Fig S1A). PI3Ka activation scor-
including PI3K/Akt pathway (Witkiewicz et al, 2015; Conway et al, ing allowed us to cluster 8/9 PDACmet patients (Fig 1C). Conver-
2019). Fewer than 5% of patients present PIK3CA oncogenic muta- sely, only 2/9 PDACloc clustered with PDACmet. We further
tion, but this mutation mimics the KRAS oncogenic pathway (Eser validated the PI3Ka activation signature by confirming that cancer
et al, 2013). Our research group and others have demonstrated that cells isolated from peritoneal metastasis (two primary culture of
the lipid kinase PI3Ka drives the initiation of pancreatic cancer patient-derived cells) presented a significantly higher expression of
downstream of oncogenic KRAS (Baer et al, 2014; Wu et al, 2014). FOXA1, RRM2, BIRC5, FGFR4, PHGDH, TYMS, as well as MYBL2,
However, little is known of the importance of this PI3K isoform in PTTG1, KIF2C, CDC20, CCNB1 albeit in a lower extent compared to
the progression of localised tumours towards a metastatic disease. non-tumoural ductal cells (Appendix Fig S1B). Those cells also
Cell-free DNA (cfDNA) and, more precisely, circulating tumour presented increased levels of pS473Akt/Akt levels compared to non-
DNA (ctDNA) appears in clinical oncology as an attractive tumoural ductal cells (Appendix Fig S1C) analysed by Western blot.
biomarker for early cancer detection, diagnosis and prognosis (Diehl This gene subset of the PI3Ka activation signature was increased in
et al, 2008; Dawson et al, 2013; Alix-Panabi
eres & Pantel, 2016; all metastatic patients (Fig 1C). We extended our findings to two
Abdallah et al, 2020). In cancer patients, ctDNA represents a vari- larger independent cohorts of PDAC patients (Dataset EV2). High
able fraction of cfDNA (Dawson et al, 2013) and is distinguished by scoring of PI3Ka activation was significantly increased in patients
the presence of specific cancer-associated mutations. The release of with the poorest prognosis, regardless of their stage (Fig 1D). In
cfDNA can be due to apoptosis and necrosis of cancer cells (or both cohorts, PI3Ka activation signature was mostly associated with
healthy cells), and it can be secreted directly by the tumour or pure basal-like RNA subtype as described in Appendix Fig S2B. The
micro-environment cells such as immune and inflammatory cells PI3Ka activation signature discriminates between localised patients
(Schwarzenbach et al, 2011). Because cfDNA has been studied as an and those with an early risk of relapse and death (Fig 1E). Even
exploratory biomarker of micro-metastatic disease in PDAC though oncogenic mutations of PI3Ka are rare in PDAC (Eser et al,
(Pietrasz et al, 2017; Lee et al, 2019), we proposed that detection of 2013), as confirmed in the PAAD database (Dataset EV3,
cfDNA, as a sign of early metastatic disease, could predict therapeu- Appendix Fig S2), high scoring of non-mutated PI3Ka activation
tic effectiveness towards metastatic evolution. was a worse prognostic factor, irrespective of the stage of the
Taking two unbiased approaches to analyse patient data, we disease. In conclusion, activation of non-mutated PI3Ka appears as
demonstrated that gene expression signatures of PI3K activation a strong prognostic factor of aggressive disease.
were a novel way to molecularly identify the most aggressive
primary tumours. We then identified a novel pharmacological Full annihilation of PI3Ka prevents pancreatic cancer
target, PI3Ka, that drives pro-inflammatory features towards macro- cell migration
metastatic evolution. PI3Ka inhibitors could be included in the
immunomodulatory and anti-cancer therapeutic arsenal in PDAC. We then investigated whether pancreatic cancer cells depended on
the activity of a specific PI3K isoform. We used a panel of pancreatic
tumour cell lines generated by KRAS mutation combined with other
Results genetic alterations including PIK3CA mutation and PTEN deletion
(Dataset EV4A). We compared two different pharmacological strate-
PI3K and PI3Ka-specific transcriptomic signature predicts gies of PI3K inhibition using compounds with either isoform-selective
aggressive pancreatic cancer or pan-PI3K pharmacological profiles as shown by inhibitory concen-
tration 50 (IC50) on recombinant protein in vitro (Fig 2A, Dataset
We sought to determine in an unbiased manner which signalling EV4B). We analysed protein expression levels on four cell lines and
pathway are associated with aggressive features in PDAC. we observed detectable effects on pS473Akt levels (used as read-out
We analysed publicly available data set to distinguish a normal of PI3K activity) for all inhibitors at 1 and 10 µM, thus allowing
pancreas from chronic pancreatitis (CP) and primary tumours from comparison of the differential downstream actions of PI3K isoforms.
localised PDAC (PDACloc) or metastatic PDAC (PDACmet) (Fig 1A– The most potent inhibitors, BKM120 and GDC0941, almost annihi-
C, Dataset EV1). PDACmet and CP patients share the same enrich- lated the pS473Akt levels at 10µM in the four cell lines and at 1 µM in
ment of mRNA expression-based hallmarks of biological pathways only one cell line (R211, data shown in main figure) (Fig 2B–D,
compared to normal, except for 3 hallmarks. The PI3K/Akt/mTOR Appendix Fig S3) suggesting that effective concentrations (EC50) are

2 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

A Hallmarks D PDAC patients – all UICC stages

Corr p value Corr p value Corr p value Corr p value
PDACmet PDACloc vs CP vs PDACmet PAAD (TCGA) database
Pathways vs normal normal normal vs PDACloc
Complement 0.0002 0.0000 0.0066 0.2773
Apical junction 0.0016 0.0000 0.0000 0.1524
Coagulation 0.0020 0.0000 0.0000 0.1487
Glycolysis 0.0021 0.0403 0.0482 0.8633
IL-2 STAT5 signaling 0.0022 0.0001 0.0014 0.0524 PI3Kα activation: High+Medium Low
Reactive oxygen species pathway 0.0038 0.0075 0.0087 0.9235
Interferon alpha response 0.0042 0.0002 0.0008 0.3250
Heme metabolism 0.0043 0.0002 0.0006 0.1213
Interferon gamma response 0.0046 0.0001 0.0034 0.1668
IL-6 JAK STAT3 signaling 0.0049 0.0010 0.0230 0.1679
PI3K Akt mTOR Signaling 0.0077 0.0032 0.5960 0.3675
Hedgehog signaling 0.0081 0.9943 0.0002 0.0657
Early estrogen response 0.0098 0.0003 0.0000 0.0355
Myogenesis 0.0103 0.1399 0.0031 0.0048
TNFa signaling via NFKB 0.0121 0.0003 0.0038 0.0167
Inflammatory response 0.0155 0.0140 0.0082 0.0568
Oxidative phosphorylation 0.0205 0.0002 0.0002 0.0234
Mitotic spindle 0.0218 0.0030 0.0061 0.6924
Epithelial mesenchymal transition 0.0219 0.0000 0.0004 0.0031
Xenobiotic metabolism 0.0226 0.0829 0.2647 0.0163
Allograft rejection 0.0230 0.0335 0.0141 0.7627 PACA-AU database
Wnt beta catenin signaling 0.0301 0.0000 0.0157 0.3125
UV response up 0.0445 0.6722 0.0604 0.1190
B PI3K-related Reactome
Pathways Corr p value PDACmet vs PDACloc
PI3K Akt activation 0.0603
PI3Kα activation: High+Medium Low
PI3K Akt signaling in cancer 0.0741
PI3K cascade FGFR2 0.0162

Normal
C CP
PDACloc
PDACmet
PI3Kα activation signature

E
Localized PDAC (in TCGA)
1.0

0.8
Overall survival

0.6 p = 0.0413

0.4

0.2 PI3Kα activation:

Med n=9
Lo n=12
0.0
0 500 1000 1500 2000 2500
Time (days)
Figure 1.

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 3 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

◀ Figure 1. A PI3Ka-specific transcriptomic signature predicts pancreatic cancer aggressiveness.

A List of 19 HALLMARK pathways (out of 50) significantly altered (ranked with P-value – top: low, bottom: high) in a cohort of 9 metastatic pancreatic adenocarcinoma
patients (PDACmet), 9 localised PDAC (PDACloc), 8 chronic pancreatitis (CP) compared to 9 normal pancreas samples (normal); corrected P-values correspond to the
comparison indicated. ANOVA test corrected using the Benjamini and Hochberg method (BH) on SES (sample enrichment score). Significantly altered corrected P-
values are highlighted in colour. The signature in red corresponds to the signature with the lowest P-value for discriminating PDAC from CP.
B List of PI3K-related pathway gene signatures (Reactome) altered in the same cohort; corrected P-values are shown. ANOVA test corrected using the Benjamini and
Hochberg method (BH) on SES (sample enrichment score). Significantly altered corrected P-values are highlighted in colour.
C Transcriptomic signature indicative of PI3Ka activation in the same cohort. Genes whose expression is increased only in metastatic patients are underlined. The
unsupervised hierarchy of each sample is shown in the top part of the figure; the unsupervised hierarchy of genes is shown on the left; the list of genes is detailed on
the right. Blue = low expression, Red = high expression.
D Scoring of the PI3Ka activation transcriptomic signature was used to cluster patients with high, medium (both groups were combined, high+medium: red) and low
(black) scoring levels in the primary tumours of confirmed PDAC patients from PAAD (TCGA) or PACA-AU (BH corrected P-values). The survival curves of each cluster
were then plotted, and the statistical significance was calculated using the log rank test.
E PI3Ka activation scores from localised PDAC in the TCGA database according to their UICC staging were retrieved, clustered into two groups (low and med) (BH
corrected P-values) and their overall survival was plotted, with log rank testing.

higher in pancreatic cancer cells compared to in vitro IC50. a- targeting each class I PI3K catalytic subunit, with the latter condition
selective inhibitors and pan-PI3K inhibitors with low in vitro IC50 on mimicking a pan-PI3K inhibitor. Pools of p110a-targeting siRNAs or
PI3Ka significantly decreased pS473Akt levels in the four cell lines at pools of p110a/b/c/d siRNAs lead to decreased expression of PI3K
1 µM, suggesting that inhibitor selectivity profile is conserved. catalytic subunits (Fig 2I) and induced similar migration inhibition
When a-selective inhibitors, A66 and BYL-719, were used at low on R211 and PDAC8661 cells compared to pools of scramble siRNAs
concentrations of 0.1 µM, significant effects were observed for (Fig 2J). We noticed that, in PDAC8661, decreased expression of
assays related to migratory phenotype as shown in Appendix Figs S4 p110a mRNA increased p110b and p110d mRNA expression, without
and S5 and in Fig 2E. In all human and murine cell lines tested, A66 leading to a compensatory increased migration. We also created
and BYL-719 presented a concentration-dependent capacity to inhibit pools of PANC-1 cells stably transfected with two hairpins targeting
pancreatic cancer cell migratory hallmarks, cell motility PIK3CA or two hairpins targeting PIK3CB (genes encoding for PI3Ka
(Appendix Fig S4) and directed cell migration (Fig 2E, Appendix Fig and PI3Kb, respectively), as well as one scramble hairpin (Fig 2K).
S4). The EC30 for migration was reached in 7/9 and 9/9 cell lines for Only shRNA targeting PI3Ka significantly decreased cell migration
A66 and BYL-719, respectively (Appendix Fig S6B). On the contrary, (Fig 2L). Genetic approaches confirm the selective action of the
these parameters were not attained with the c-specific inhibitor PI3Ka on the migratory phenotype of pancreatic cancer cells.
AS252424 or the b and b/d-specific inhibitors except for AZD8186,
which reached the EC30 for migration in two cell lines (Appendix Fig PI3Ka inhibition regulates selective PI-3,4,5-P3 species
S6B). The R6065 cell line (harbouring PTEN deletion) was an excep-
tion, in which two b-selective inhibitors had an effect on migration To explain the downstream differences observed with PI3Ka and
assays at 0.1 µM (Appendix Fig S4B). This demonstrated that motil- pan-PI3K inhibitors, we researched a concentration where both inhi-
ity and migration cell activities are sensitive to PI3Ka inhibition in bitors induced the same effect on Akt phosphorylation (Fig 3A and
pancreatic cancer cell. Considering that PI3K cascade to FGFR2 was B). Interestingly, at this concentration, the a-selective inhibitor
significantly increased in PDACmet as opposed to PDACloc, FGFR presented a moderate effect on the number of living cells whereas it
signal activation was prevented in R211 cell line by treatment with drastically affected cell migration; the pan-PI3K inhibitor, BKM120,
2 µM of AZD4547. AZD4547 significantly decreased R211 cell migra- while being as potent on inhibiting Akt phosphorylation, did not
tion (Fig 2F). The concomitant inhibition of FGFR and PI3Ka by reduce these parameters significantly (Fig 3C and D). This compel-
simultaneous BYL-719 and AZD4547 treatment did not inhibit cell ling result could be explained by the differential selective decrease of
migration more than individual treatment which suggested that at PI-3,4,5-P3 or of PIP2 (comprising both the substrate PI-4,5-P2 and the
least a part of PI3Ka pro-migratory signal was due to FGFR activa- product PI-3,4-P2) total levels by each inhibitor, respectively (Fig 3E).
tion. Effects on cell survival and proliferation were significant at A66 as opposed to BKM120 selectively reduced the proportion of
higher concentrations (10 µM) (Fig 2G, Appendix Fig S5). We C36:2 PIP3 (Fig 3F). BKM120 led to the modification of the percent-
assessed apoptotic R211 cells (with IncuCyte Annexin V assay) after ages of other PIP3 species (Fig 3F). The distributions of PIP and PIP2
2 days of treatment with a-specific or pan-PI3K inhibitors and found subspecies were not modified (Appendix Fig S8). Altogether, these
a selective increase in cell death induced by all PI3Ka inhibitors and data suggest that, in PDAC, PI3Ka-specific actions include production
by three out of four pan-PI3K inhibitors (Fig 2H). The growth inhibi- of PIP3 with distinctive acylation pattern, that could explain the selec-
tory GI30 for cytotoxicity was reached in a lower number of cell lines tive promotion of cell motility and migration by PI3Ka.
for both A66 and BYL-719 (Fig 2G, Appendix Figs S5 and S6A). Cell
migration and cell survival of the non-tumoural ductal pancreatic PI3Ka inhibition targets cell migration and cell survival
cell line, HPNE hTERT, were also sensitive to PI3Ka-selective inhibi- regardless of the genetic landscape of pancreatic
tion, but to a lesser extent for the migration assay (Appendix Fig S7). adenocarcinoma cells
To validate the pharmacological approach, we used a genetic strat-
egy and treated R211 and PDAC8661 murine pancreatic tumour cells PI3Ka oncogenic action is commonly described as directly coupled
with pools of scramble siRNA, siRNA targeting p110a or mixed siRNA only to oncogenic KRAS or tyrosine kinase receptors, and not to

4 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine

5 of 20
EMBO Molecular Medicine 13: e13502 | 2021
D

L
F
B

H
Benoit Thibault et al

ª 2021 The Authors

G
A

K
I

Figure 2.
 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

◀ Figure 2. Full annihilation of PI3Ka activity is required to prevent pancreatic cancer cell migration.
A In vitro IC50 of PI3K inhibitors, obtained on recombinant proteins and based on the literature. The Greek letter on the x axis indicates the most potently targeted
isoform.
B–D (B) Murine pancreatic tumour cells R211 were treated for 15 min with the vehicle, a, b, b/d, c-specific or pan-PI3K inhibitors at 1 or 10 µM in the presence of 10%
FBS and the protein levels of pAkt (Ser473 and Thr308), total Akt and b Actin were observed by Western blot. PAkt on (C) Ser473 and (D) Thr308 were quantified and
normalised with b Actin. n = 2 in each group.
E Murine pancreatic tumour cells R211 were treated with the vehicle, a, b, b/d, c-specific or pan-PI3K inhibitors at 0.1 or 1 µM and simultaneously subjected to a
Boyden chamber migration assay. Migrating cells were quantified after 24 h. Representative image of filter after Crystal violet staining is shown. Scale = 200 µm.
n = 3 in each group.
F Murine pancreatic tumour cells R211 were treated with vehicle, BYL-719 1 µM, AZD4547 2 µM or both treatments and subjected to a Boyden chamber migration
assay. Migrating cells were quantified after 24 h. n = 4 in each group.
G Murine pancreatic tumour cells R211 were treated with the vehicle, a, b, b/d, c-specific or pan-PI3K inhibitors, and living cells were quantified after 3 days with a
MTT colorimetric assay. Metabolically active cells are considered as living cells. n ≥ 3 in each group.
H Murine pancreatic tumour cells R211 were treated with the vehicle, a or pan-PI3K inhibitors and apoptotic (IncuCyte Annexin V) cells were quantified after 2 days.
n = 3 in each group.
I Relative p110a, b, c or d mRNA expression (compared to scramble siRNA) after inhibition of expression by siRNA targeting p110a or a combination of siRNA
targeting each class I PI3K isoform. n ≥ 3 in each group.
J R211 and PDAC8661 cells were treated with siRNA scramble, siRNA targeting p110a or a combination of siRNA targeting each class I PI3K isoform (pools) and
subjected to a Boyden chamber migration assay. Migrating cells were quantified after 24 h. n = 3 in each group.
K Relative p110a or b mRNA expression (compared to scramble shRNA stably transduced cells) in human pancreatic tumour cells PANC-1 stably transduced with
shRNA targeting p110a or b. n = 4 in each group.
L PANC-1-transduced cells were subjected to a Boyden chamber migration assay. Migrating cells were quantified after 24 h. n = 5 in each group.
Data information: Mean
SEM (*P < 0.05, **P < 0.01, ***P < 0.001, n ≥ 3 independent experiments except for WB experiment, Student’s t-test. When not precised,
comparisons are performed with vehicle.

PTEN alterations (Thorpe et al, 2015). To challenge this concept, Pharmacological PI3Ka inhibition prevents the rapid progression
we analysed the impact of the genetic alterations (on KRAS, of cfDNA-positive PDAC
PIK3CA, PTEN) and of the organ of origin in determining the role
of each class I PI3K in tumour cell migration and cytotoxic sensitiv- When we compared resected pT3 tumours with and without patho-
ity in response to PI3K inhibitors. With the data shown in Fig 2, logical nodal involvement (pT3N0M0, two tumours versus
Appendix Figs S4E and F, and S5D and E, we calculated the corre- pT3N1M0, four tumours), we observed a stronger pAkt Substrate
lation between the in vitro IC50 of PI3K inhibitors for all class I IHC staining, indicative of PI3K/Akt activity, in the primary site that
PI3Ks (Fig 2A, Dataset EV4B), and their capacity to inhibit migra- are associated with nodal involvement. While PDAC patients with
tion in nine cell lines (correlation test values are shown in Fig 4A– increased levels of ctDNA present a worse prognosis (Pietrasz et al,
C, individual values for each cell lines in Appendix Fig S9). The 2017; Lee et al, 2019), cfDNA and ctDNA could also be indicative of
ability of all PI3K inhibitors (at 1 µM) to regulate cell migration underlying micro-metastatic disease. Increased pAkt Substrate IHC
mainly depended on their capacity to target PI3Ka, and in a less staining was increased in pT3N1M0 patients with higher level of
frequent way on PI3Kb, d or c (Fig 4A, Appendix Fig S9). We cfDNA with detected KRAS mutation, which could suggest a correla-
reported the P-value of the effect versus IC50 correlation test for tion with dependency to PI3K activity (Fig 5A and B).
each class I isoform and found that PI3Ka was significantly associ- We quantified cfDNA in the KPC model to evaluate the effects of
ated with pancreatic cell migration (Fig 4B, Appendix Fig S9C), cell PI3Ka-selective pharmacological inhibition on established tumours
motility (Appendix Fig S9A and E), cytotoxicity (Fig 4C, presenting high levels of cfDNA. In the KPC model, aggressive
Appendix Fig S9D) and pSer473Akt phosphorylation (Appendix Fig pancreatic tumours spontaneously develop under KRAS and p53
S9B and F). As a positive control, we confirmed the demonstrated oncogenic mutations (Hingorani et al, 2005); however, detectable
isoform dependency that was published by others in other solid levels of cfDNA have not been described. Tumours were diagnosed
and liquid cancer cell lines (Park et al, 2008; Torbett et al, 2008; through high-resolution US imaging as well as quantification of
Lynch et al, 2018). Most cancer cell lines of pancreatic origin cfDNA (for the setting of threshold limit, see Materials and Methods
depended on PI3Ka activity, and sometimes on PI3Kd and PI3Kc to below). We quantified the cfDNA in blood plasma samples by longi-
regulate their migration and cell viability despite the genetic tudinally measuring the relative concentration of two expressed
context. This result suggests that PI3Ka would be a target of choice genes, TP53 and GAPDH, and then correlated those findings with
for pancreatic cancer patients who, in spite of their genetic hetero- the anatomo-pathological results from the pancreas and metastatic
geneity associated with class I PI3K activation (mutant KRAS, dele- site organs (Dataset EV5). The longitudinal average levels of cfDNA
tion of PTEN, mutation of PIK3CA), depend on intrinsic basal correlated with disease progression and mouse lethality (Fig 5C).
PI3Ka activity, thus corroborating the prognostic value of the PI3Ka We analysed two independent cohorts of KPC mice and were able to
activity signature to detect pro-metastatic features (Fig 1). Epithe- discriminate between localised primary tumour and metastasis, by
lial or mesenchymal features of the murine cell line panel were cfDNA levels at the time of sacrifice (Fig 5C and D, Appendix Fig
assessed, and cells were classified as either epithelial or mesenchy- S10A–E, Dataset EV5). A small fragment of cfDNA was shown by
mal phenotype (Appendix Fig S9G). The importance of PI3Ka activ- others to be specific to tumour cells (Thierry et al, 2010). Analysis
ity for cell migration and cytotoxic response was similar in both of cfDNA integrity revealed a distinct 160–210 bp fragment selec-
groups (Appendix Fig S9H and I). tively increased in mice that developed metastatic PDAC compared

6 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

A R211 B R211
R211

Actin
Ser473/ Actin
100

P-Akt Ser473/
80

vehicle)
ofvehicle)
p 0.8433
60 NS

Normalized P-Akt
40

(%of
(%
Normalized
20

0
A66 BKM120
1 μM 1 μM

C D E
R211 R211 R211 R211 R211
120 NS p 0.0846 120 150
Living cells % (vs vehicle)

p=0.085 150 150

Migration % (vs vehicle)

100

PIP3 % (vs vehicle)

PIP2 % (vs vehicle)
100 p 0.016
* PIP % (vs vehicle)
*** p 0.0001 ** p 0.0013
80 80 100 100 100
60 60

40 40 50 50 50
20 20

0 0 0 0 0
A66 BKM120 A66 BKM120
A66 1μM

BKM120 1μM
Vehicle

A66 1μM

BKM120 1μM
Vehicle
A66 1μM

BKM120 1μM
Vehicle
1 μM 1 μM 1 μM 1 μM

F
Vehicle A66 1μM BKM120 1μM

C40:0, C40:2,
C40:0, C40:2, C40:0, C40:2,
C40:3, C40:6 C38:1, C38:2, C38:3, C40:3, C40:6
C40:3, C40:6 11% C38:4, C38:5 8%
11.5%
C38:1, C38:2, C38:3, 6%
C38:1, C38:2, C38:3,
C38:4, C38:5
C38:4, C38:5 C36:3, C36:4
9%
12% 3%

C36:3, C36:4 C36:1

C36:3, C36:4 C36:2
3% C32:0 2.5%
C32:0 3% 6% R211
41% C32:0
C36:2 42% C36:0
46.5%
C36:1 6% C36:2 0.5% PIP3
C34:1
2.5% 3% C34:2
C36:1 7%
3% 1% C34:0
C36:0 C34:1 C32:2
1% C34:1 C36:0 7% 8.5% C32:1
5.5% C32:1 5.5%
C34:2 C32:2 8.5% 1.5%
C34:2 C34:0 C32:1 C32:2 8%
2.5% 4% 8% 3%
6% 1%
C34:0
2.5%

Figure 3. PI3Ka inhibition regulates selective PI-3,4,5-P3 species.

A, B (A) Murine pancreatic tumour cells R211 were treated for 15 min with the vehicle, a-specific A66 or pan-PI3K inhibitor BKM120 at 0.01, 0.05, 0.1, 0.5, 1 or 5 µM in
the presence of 10% FBS and protein levels of PAkt (Ser473), total Akt and b Actin were observed by Western blot. (B) PAkt on Ser473 was quantified and
normalised with b Actin. n = 3 in each group.
C, D R211 cells were treated with the vehicle, a-specific inhibitor A66 or pan-PI3K inhibitor BKM120 at 1 µM and (C) living cells were quantified after 3 days with a MTT
colorimetric assay or (D) migrating cells were quantified after 24 h in a Boyden chamber assay and compared to vehicle (DMSO). n = 3 in each group.
E Murine pancreatic tumour cells R211 were treated for 15 min with the vehicle (0.01% DMSO) or a-specific A66 or pan-PI3K inhibitor BKM120 at 1 µM.
Phospholipids were extracted, total PIP, PIP2 and PIP3 were quantified and compared to the vehicle (DMSO). n = 3 in each group.
F The proportion of each PIP3 subtype was represented. Subtypes whose proportion was modified compared to DMSO were underlined. n = 3 in each group.
Data information: Mean
SEM (*P < 0.05, **P < 0.01, ***P < 0.001, n ≥ 3 independent experiments, Student’s t-test).

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 7 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

B C

Figure 4. PI3Ka inhibition is effective on pro-tumoural features regardless of the genetic landscape of pancreatic adenocarcinoma.
A, B (A) Migration inhibition capacities of PI3K inhibitors at 1 µM were tested on pancreatic tumour cells (R211, PDAC8661, PANC-1, 10593, 10158, R6344, R6430, R6065,
R6141). The mean of these values was plotted against the in vitro IC50 for each class I PI3K and each inhibitor. IC50 is determined on recombinant proteins.
Pearson correlation tests were performed and its P-values were presented separately for each class I PI3K isoform. (B) Individual P-values of these correlation tests
were represented for each of the 9 pancreatic (R211, PDAC8661, PANC-1, 10593, 10158, R6344, R6430, R6065, R6141,) breast (MDA-MB-231, MDA-MB-468) and
prostate (PC3) cancer cell lines; Pearson correlation P-values for each PI3K isoform are plotted separately. n = 3 in each group.
C Cytotoxic capacities of PI3K inhibitors at 10 µM were tested on the 10 pancreatic (R211, PDAC8661, DT4994, PANC-1, 10593, 10158, R6344, R6430, R6065, R6141),
breast (MDA-MB-231, MDA-MB-468), prostate (PC3) and acute myeloid cells (NOMO-1, HL-60), and the correlation with the in vitro IC50 for each class I PI3K of
each inhibitor was determined. Pearson’s correlation tests were performed, and P-values represented for each isoform and cell line. n ≥ 3 in each group.
Data information: The dotted red line corresponds to a threshold P-value of 0.05 obtained with a Pearson correlation test. Mean
SEM; n ≥ 3 independent experiments.

to mice with high-grade PanINs and localised PDAC (Appendix Fig spleen (Fig 6E), delayed ascites development (Fig 6F) and mice
S10B, Dataset EV5). survival (Fig 6G). Although most vehicle-treated mice had to be
We then treated KPC mice featuring aggressive carcinoma (i. e. sacrificed during the treatment, only one BYL-719-treated mouse
featuring a US-detected tumour and a high level of cfDNA, presented signs of tumour evolution that required immediate
Appendix Fig S11), with the PI3Ka-selective inhibitor, BYL-719, or euthanasia. A correlation was established between the proliferative
with vehicle (Fig 6A) (Andre et al, 2019). pS473-Akt levels were index of cancer cells (assessed by ki67 index in primary tumour and
reduced by BYL-719 treatment in all tested tissues, while pERK- metastatic sites) and levels of cfDNA, which reflected the global
T202 Y204 levels decreased only in the pancreas following BYL-719 tumour burden (Appendix Fig S13A). BYL-719 significantly reduced
treatment (Fig 6B). We interrupted the cohort treatment when cell proliferation assessed by Ki67 index in both primary and meta-
vehicle-treated mice presented signs of macroscopic metastatic static sites (Fig 6H–J, Appendix Fig S13B, for validation of cohort
dissemination, ascites detection, or ethical limit points. The BYL- size, see the value of the Power test in Dataset EV5), associated with
719 treatment line significantly slowed tumour volume progression decreased tumour grading and changes in tumour cell/stroma ratios
(Fig 6C, Appendix Fig S12) and reduced the number of macro- (Appendix Fig S14). The anticipated secondary effects of BYL-719
metastatic foci (Fig 6D), distant metastatic area in lungs, liver and treatment on insulin secretion were observed (Appendix Fig S15A

8 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

pAKT Substrate
A
Patient #0 – T3,N0,M0 Patient #1 – T3,N1,M0 Patient #2 – T3,N1,M0

[pancreas section]

50μm 25μm 50μm 25μm 50μm 25μm

Low cfDNA (ng/mL) 7.48 Low cfDNA (ng/mL) 13.93 High cfDNA (ng/mL) 84.84
Mutated Kras (%) 0.19 Mutated Kras (%) 0.14 Mutated Kras (%) 0.35
%pAKT Substrate per FOV

40
B C D
n=2
KPC mice
*** p<0.0001
p<0.0317
30 100 0.8 * * p<0.0105
Percent survival

0.7

cfDNA (A.U.)
n=2 0.6
20 0.5
*** 0.4
n=2 50
10 p<0.0001 0.20
0.15 ** p<0.0012
Low cfDNA (n=25)
0.10 ***p<0.0001
High cfDNA (n=39) 0.05
0
Low Low High 0
0 10 20 30 40 50 Mice with:
Normal High PDACloc PDACmet
cfDNA cfDNA cfDNA (n=19) grade (n=12) (n=12)
N0 N1 N1
Weeks
PanIN
(n=12)

Figure 5. cfDNA-positive PDAC present increased lethality and metastatic dissemination.

A Representative images of PI3K activation (as shown by positive pAkt substrate staining) on the primary tumours of PDAC patients, with values of quantification of
cfDNA and mutated KRAS allele frequency.
B Quantification of pAkt substrate staining in T3 patients (total number of patients = 6, n = 2 in each group). Mean
SEM.
C Survival curve of KPC mice presenting different levels of cfDNA. n in each group is indicated. Log rank test, ***P < 0.001.
D Quantification of cfDNA in KPC mice at different stages of the disease. n in each group is indicated. PanIn = Pancreatic Intraepithelial Neoplasia. Non-parametric
Mann–Whitney, *P < 0.05, **P < 0.01, ***P < 0.001.

and B). We did not see any significant impact on levels of cleaved (KC;p110a+/lox, A260, see also Appendix Fig S17, Appendix Supple-
caspase 3-positive cells (apoptotic marker: Appendix Fig S15C) or mentary Methods) (Fig 6N). Taken together, these data demonstrate
on c-H2AX-positive foci (DNA damage marker: Appendix Fig S15D). that in vivo inhibition of PI3Ka prevents the evolution of micro-
Furthermore, in a context of early pancreatic cancer lesions metastatic foci into macro-metastatic foci and the rapid progression
induced by oncogenic KRAS in an inflammatory condition (caeru- of cfDNA-positive PDAC.
lein injections), the highly selective pharmacological inactivation of
PI3Ka with another compound, namely GDC-0326, completely Tumour-intrinsic PI3Ka alters tumour cell chemokine secretion
prevented the maintenance of pre-cancer lesions (Appendix Fig and promotes the acquisition of tumour-associated
S16A–D), and features linked to stromal remodelling (Appendix Fig inflammatory macrophage characteristics in the
S16B and E). The high level of cleaved caspase 3 detected in epithe- peritumoural tissue
lial lesions could be prevented by PI3Ka inactivation (Appendix Fig
S16B and F). Transition from micro- to macro-metastasis is now well described as
Tail vein injection of murine pancreatic cancer cells, engineered promoted by tumour-extrinsic factors including immune cells
to express secreted luciferase (R211-Luc cells) in Nude mice, was (Celi

a-Terrassa & Kang, 2016). Hence, we first analysed systemic
carried out in order to confirm the action of BYL-719 treatment on alterations on circulating blood cells during PDAC progression in
the evolution of micro-metastatic foci. Administration of BYL-719 our KPC model. We performed a full blood count to assess immune
treatment for 21 days significantly prevented tumour cell growth in cell populations (Dataset EV5), as macrophages promote PDAC
lungs (Fig 6K and L, Dataset EV5). Interestingly, vehicle-treated progression (Padoan et al, 2019). We noticed that metastatic KPC
mice presented an increased area of pancreatic epithelial marker mice presented significantly increased counts of white blood cells
CK19 (Fig 6M). Similar results were observed in C57/B6 mice when (Fig 7A), monocyte (Fig 7B) and granulocyte (Fig 7C) counts
comparing the area of metastatic foci after tail vein injection of compared to KPC with localised tumours. The lymphocyte counts
syngeneic Kras mutant pancreatic cells (KC;p110a+/+, A338) remained the same (Fig 7D).
compared to syngeneic Kras mutant pancreatic cells partly lacking To test whether tumour-intrinsic PI3Ka could be involved in this
PI3Ka activity through a genetic inactivation of one allele of Pik3ca tumour/stroma interaction, we used derived cell lines from

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 9 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

SPLEEN LUNG PANCREAS

A B C

BYL-719

BYL-719

BYL-719
Vehicle
in situ KPC model

Vehicle

Vehicle
p 0.0216
Aggressive BYL-719
KPC mice * KPC mice
10 Vehicle (n=6)
Carcinoma 70 kDa

Tumour volume
pAKT

(Fold change)
(tumour + high 50 kDa BYL-719 (n=6)
Ser473
cfDNA) 70 kDa
AKT 50 kDa
Days 0 11 50 kDa 5
Sacrifice pERK
40 kDa
T202 Y204 35 kDa
50 kDa
ERK 40 kDa
35 kDa 0

D KPC mice E F G
40 p 0.0411
Number of metastatic foci

Vehicle (n=6)
Metastasis Relative Area

p 0.0281 0.15
BYL-719 (n=6) 0.0281 * KPC mice
6 100

Ascites development
*

Percent Survival
Vehicle (n=6) Vehicle (n=6)

(Number of Mice)
30
BYL-719 (n=6) BYL-719 (n=6)
0.2273 0.10 4 p 0.0422
*
20
p 0.2273 50
ns
2 Vehicle (n=6)
0.05 BYL-719 (n=6)
10
0 5 10 15 0 5 10 15
0 0.00 Days of Treatment Days of treatment
<0.5mm >0.5mm

H KPC mice
VEHICLE BYL-719 I ** p 0.0022
H&E ki67 H&E ki67 600
Vehicle (n=6)
[pancreas section]
Primary tumour

BYL-719 (n=6)

ki67-positive
cells/ FOV
400

200

50μm 50μm 50μm 50μm 0 KPC primary tumour

J * p 0.0341
[liver section]
Metastasis

1000

ki67-positive
Vehicle (n=6)

cells/ FOV
BYL-719 (n=6)
500
50μm 50μm 50μm 50μm

0
KPC metastasis

R211-luc
K BYL-719 L VEHICLE BYL-719

Nude mice 21
0
2 weeks

Days Sacrifice

8×106 p 0.019
Vehicle (n=6)
*
(photon count)
Luminescence

6×106 BYL-719 (n=8)

4×106
3 weeks

2×106

0
2 3
Weeks after injection
M N KC;p110α+/+ or KC;p110α+/lox

p 0.0173

0.15 * C57/B6 mice

0 21
Days Sacrifice
0.20
Metastasis relative area
CK19 per FOV

p 0.0094
0.10 0.15
0.0094

**

0.10
0.05
0.05

0.00 0.00
+

Vehicle BYL-719
lo
+/

+/
α

α
0

0
11

11
;p

;p
C

C
K

Figure 6.

10 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

◀ Figure 6. Pharmacological PI3Ka inhibition prevents the rapid progression of cfDNA-positive PDAC.
A KPC mice diagnosed with an aggressive carcinoma were given daily oral doses of BYL-719 (50 mg/kg) or of vehicle (0.5% methyl cellulose with 0.2% Tween-80).
n = 6 in each group
B The protein levels of pAkt (Ser473), total Akt, pERK (Thr202 and Tyr204) and total ERK were observed by Western blot in spleen, lung and pancreas tissue lysates
after treatment with oral doses of vehicle (0.5% methyl cellulose with 0.2% Tween-80) or of BYL-719 (50 mg/kg).
C–E (C) Quantification of tumour volume (expressed in fold change) and (D, E) quantification of micro- and macro-metastatic foci and the relative metastasis area
(comprising metastases in the spleen, liver and lung) of KPC mice treated with oral doses of vehicle (0.5% methyl cellulose with 0.2% Tween-80) or of BYL-719
(50 mg/kg).
F Development of ascites in KPC mice after treatment with oral doses of vehicle (0.5% methyl cellulose with 0.2% Tween-80) or of BYL-719 (50 mg/kg).
G Survival curve of KPC mice treated with the vehicle or BYL-719. Log rank test, *P < 0.05.
H–J (H) Representative images and (I) quantification of Ki67-positive cells in pancreatic tumours and in (J) metastasised liver sections after treatment with oral doses of
vehicle (0.5% methyl cellulose with 0.2% Tween-80) or of BYL-719 (50 mg/kg).
K, L (K) Treatment regimen of the tail vein injection experiment in nude mice treated with the PI3Ka inhibitor or vehicle (n is indicated), quantification at two time
points and (L) representative images of luminescence via IVIS® Spectrum in vivo imaging system.
M Quantification and representative CK19 IHC of R211-luc lung micro-metastasis from vehicle or BYL-719-treated mice. n = 6 in vehicle; n = 5 in BYL-719 group.
N Tail vein injection experiment in C57/B6 mice of murine syngeneic pancreatic cancer cells (Kras mutated A338, Kras mutated and half PI3Ka inactive, A260; n > 8 in
each group), quantification of micro- and macro-metastasis at final time point.
Data information: Mean
SEM (*P < 0.05, **P < 0.005) C–E, I–J, M, N: non-parametric Mann–Whitney; G: log rank test.

tumoural lesions induced either by Kras mutant or Kras mutant in with BYL-719 or with genetic PI3Ka inactivation were used to assess
pancreatic cells partly lacking PI3Ka activity, presenting differential cytokine production of IC21 macrophage cell line (Fig 7P). From the
metastatic potential (Fig 6N). We found that the genetic inactivation tested cytokines, only TNFa secretion was significantly decreased
of PI3Ka in pancreatic cancer cells led to an altered cytokine secre- by both pharmacological and genetic inactivation of PI3Ka (Fig 7Q
tion pattern in vitro (a panel of 16 cytokines were tested), with and R). IL-3 blocking antibody was added to the conditioned
decreased levels of IL-3 in two mutant Kras pancreatic cancer cell medium and decreased TNFa production by IC21 macrophage cells
lines partly lacking PI3Ka activity (genetic inactivation) compared compared to control antibody-treated cells (Fig 7R). Exogenous
to three mutant Kras pancreatic cancer cell lines, including R211 TNFa (20ng/mL) promoted R211 pancreatic cancer cell migration
cells (Fig 7E, Appendix Fig S17). Pharmacological inactivation of and PI3Ka inhibition by BYL-719 inhibited cell migration induced by
PI3Ka in three different mutant Kras pancreatic cancer cell lines TNF-a (Fig 7S).
significantly and reproducibly decreased IL3 levels (Fig 7F). PDAC Our data show that the rapid progression of aggressive cfDNA-
patients with a high PI3Ka activity signature presented a significant positive PDAC driven by PI3Ka involves changes in the inflamma-
increase on the gene signature for the selective immune population tory cytokine context in conjunction with inflammatory pro-
of cd T lymphocytes (LTcd) and not on generic signatures of other tumoural macrophage characteristics in peritumoural tissue.
immune cell populations (Fig 7G, Dataset EV2C). The cd T lympho- Tumour-induced increased TNFa secretion by macrophages could
cyte signature is known to be associated with poor prognosis and further promote PI3Ka-driven tumour cell migration.
highly inflammatory conditions, promoting differential macrophage
differentiation (Daley et al, 2016).
We therefore analysed immune cell composition in the cohort of Discussion
KPC mice treated with BYL-719 or vehicle. BYL-719 did not modify
the overall F4/80+ macrophage count (Fig 7H and I), but signifi- In the path of the recent success of targeted therapies in pancreatic
cantly prevented their differentiation into pro-tumourigenic CD206+ metastatic patients (Cintas et al, 2018; Golan et al, 2019), further
macrophages (tumour-associated inflammatory macrophages) in attempts should be made to prevent their rapid progression. We
tumour-adjacent tissues (Fig 6H and J), with no significant dif- have demonstrated that such strategy could be to target the PI3Ka-
ference in CD4+ and CD8+ infiltration (Appendix Fig S18). To test driven signal that critically sustains several aspects of pancreatic
whether oncogenic PI3Ka also mimicked oncogenic Kras on tumour/ oncogenicity, including those at the origin of tumour-induced envi-
stroma interaction, we assessed F4/80+ macrophage recruitment and ronment rewiring and metastasis evolution.
infiltrating CD206+-macrophages around tumours and found them at In the clinical setting, single agents targeting PI3K had a limited
a similar rate in peritumoural tissue from hyper-activated oncogenic impact (Shapiro et al, 2014), mainly due to the lack of selection of
PI3Ka (p110aH1047R) and KRAS mutant tumours (Fig 7K and L). patients with advanced disease (Le Tourneau et al, 2015). In meta-
Oncogenic PI3Ka did not trigger inflammation in normal pancreas of static breast cancer (MBC) patients, when PIK3CA mutation was
young mice (1–3 months old) as assessed by quantifying infiltrating detected in circulating tumour DNA, progression-free survival (PFS)
F4/80+ macrophages (Appendix Fig S18). Inactivation of PI3Ka improvement became largely significant (Delaloge & DeForceville,
exclusively in pancreatic epithelial cells was sufficient to completely 2017), and BYL-719 (alpelisib) was recently granted marketing
prevent CD206+ macrophage infiltration around high-grade lesions authorisation for the treatment of breast cancer in combination with
(Fig 7M and N, Appendix Fig S18C and D). Increased peritumoural endocrine therapy (Andre et al, 2019). Our study suggests that
CD206+ staining is associated with further development of metastatic contrary to MBC, even without PIK3CA oncogenic mutation, the
foci and ascites (Fig 7O, Appendix Fig S18E). PI3K pathway could be a driver of pancreatic metastatic evolution,
To assess macrophage function when tumoral PI3Ka is inhibited, therefore, a druggable target. Patients with PI3Ka activation signa-
conditioned medium from pancreatic tumour cells treated or not ture are enriched in pure basal-like phenotype. This RNA-based

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 11 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

A 20
B 5
C *** p<0.0001 D
*** pp0.0005 * p 0.0330
5
*** p 0.0002 15

Granulocyte count
White Blood Cells

*** p 0.0005

Monocyte count

Lymphocyte count
15 ** 0.0060 4 4

(103/mm3)

(103/mm3)
(103/mm3)

(103/mm3)
3 10
10 * p 0.0132
2 2 p 0.0048
5 ** 5
1 1
0 0 0
Normal PDACloc PDACmet Normal PDACloc PDACmet Normal PDACloc PDACmet 0
Norm PDACl PDACme
Mice with: (n=18) (n=35) (n=29) Mice with: (n=18) (n=35) (n=29) Mice with: (n=18) (n=35) (n=29) Mice with: al oc t
(n=18) (n=35) (n=29)
Patients: PACA-AU
E F G 8 pvalue=0.0058**

SES score- LTγδ

15 2.0

IL-3 secretion (vs vehicle)

Vehicle
BYL-719 6
1.5
IL-3 (pg/mL)

10
* p 0.028 4
1.0
2
5
R211
0.5
0
Genotype of 0 Genotype of 0.0 PI3Kα
independent cell lines: KC;p110α+/+ KC;p110α+/lox independent cell lines: KC activation: high med low

H KPC mice
VEHICLE BYL-719 I NS p 0.4848
Peritumoural Intratumoural Peritumoural Intratumoural Vehicle (n=6)
F4/80 positive cells

BYL-719 (n=6)
[pancreas section]

150

F4/80 positive cells

per FOV
100

50
50μm 50μm 50μm 50μm

0
CD206 positive cells

KPC mice
[pancreas section]

J
100
** p 0.0087

CD206 positive cells

Vehicle (n=6)
BYL-719 (n=6)

per FOV
50
50μm 50μm 50μm 50μm

mice mice mice mice

0
KC;p110α+/+(n=8) KC;p110α+/+ (n=8) KPC;p110α+/+ (n=4) KPC;p110α+/+ (n=5) KPC mice
K p110αH1047R (n=7)
L p110αH1047R (n=7)
M KPC;p110α+/lox (n=4) N KPC;p110α+/lox (n=5) O
KPC;p110αlox/lox (n=4)
F4/80 positive cells per

250 100 60 80 p 0.0115

NS p 0.1526
* p 0.0357
CD206 positive cells

*
CD206 positive cells

NS p 0.3357 50
CD206 positive cells

CD206 positive cells

200 80 60
40
per FOV

per FOV

p 0.0167
40
FOV

150
per FOV

60 *
per FOV

30 40
100 40 20
20 20
50 20 10
0 0 Normal High PDACloc PDACmet
0 0 0 High grade lesions Mice with: (n=3) grade (n=5) (n=5)
Low grade PanIN
PanIN
(n=7)
P Q R S
+/+
KC;p110α+/+ KC;p110α KC;p110α+/lox CM KC;p110α+/+ Vehicle CM KC;p110α+/+ TNFα
+vehicle +BYL719 +vehicle CM KC;p110α+/+ BYL-719 CM KC;p110α+/lox
400 1000
Condionned medium ** p 0.0025 R211
TNF secretion (pg/mL)

(CM) 800
p 0.009

300 R211
TNF (pg/mL)

200 Vehicle
600 *
*
Migration % (vs vehicle)

TNF-α 20 ng/mL
p 0.0014

200
400 150
p 0.0004

BYL-719 1 μM
+/ blocking IL-3 Ab
100 * TNF-α + BYL-719
*
IC21 macrophages 200 100
*
*
*
0 0 50
secreted from: IC21 secreted from: IC21 IC21
Cytokine dosage + ctrl Ab + blocking IL-3 Ab
0

Figure 7.

12 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

◀ Figure 7. Tumour-intrinsic PI3Ka alters tumour cell chemokine secretion and promotes the acquisition of tumour-associated inflammatory macrophage
characteristics in the peritumoural tissue.
A–D Blood count of (A) white blood cells, (B) monocytes, (C) granulocytes and (D) lymphocytes in KPC mice, with cohort size in each group.
E Basal level of IL-3 was quantified in KC;p110a+/+ (R211, A338 and A338L) and KC;p110a+/lox (A260 and A94L) cell lines isolated from mice primary tumours or lung
metastases (L). Each point is a mean of n = 3 independent values.
F IL-3 relative level (compared to vehicle) was determined in 3 KC (R211, A338 and A338L) cell lines isolated from mice primary tumours under treatment with
vehicle (DMSO) or 1µM of BYL-719. n = 3 independent values.
G Violin plot demonstrating the link between PI3Ka activation and activation of a selective population of cdLT in the PACA-AU cohort.
H–J (H) Representative images and quantification of (I) F4/80-positive cells and (J) CD206-positive cells after the pharmacological inhibition of PI3Ka. Black arrowheads
show positive immune cells.
K, L Quantification of (K) F4/80 and (L) CD206-positive cells in KC and oncogenic PI3Ka mice.
M, N Quantification of CD206-positive cells in (M) low-grade and (N) high-grade PanIN lesions after the genetic inactivation of PI3Ka in epithelial pancreatic cells.
O Quantification of CD206 positive in KPC mice at different stages of PDAC progression.
P Treatment of macrophage cell line with tumour cell-conditioned medium.
Q, R Quantification of secreted TNFa in indicated conditions. n = 3 independent values for each cell line and treatment.
S Murine pancreatic tumour cells R211 were treated with vehicle, TNF-a 20 ng/ml, BYL-719 1 µM or both treatments and subjected to a Boyden chamber migration
assay. Migrating cells were quantified after 24 h. n = 4 independent values.
Data information: n in each group is indicated in the figure. A-D, I-O, Mann–Whitney; F, Q-S, Student t-test; G, ANOVA test corrected using the Benjamini and Hochberg
method (BH) on SES (sample enrichment score). Mean
SEM (*P < 0.05, **P < 0.005, ***P < 0.0001).

PDAC subtype is also known as the most aggressive subtype of PI3Ka inhibitor is efficient regardless of the mesenchymal/epithelial
PDAC patients (Collisson et al, 2011; Puleo et al, 2018). Mueller et al status of the tested cell lines. In PDAC, an epithelial-to-
(2018) showed that this subtype of murine cell lines showed a strong mesenchymal transition-independent metastasis programme is iden-
gene set enrichment for Ras downstream signalling pathways, tified that promotes macro-metastatic foci (Chen et al, 2018).
including PI3K/Akt signalling, further corroborating our finding. Macro-metastatic foci are enriched in E-cadherin expressing tumour
In PDAC, high levels of ctDNA as a marker of micro-metastatic cells in KPC model (Aiello et al, 2016). The efficiency of PI3Ka
disease were correlated with poor prognosis (Pietrasz et al, 2017). targeting is not impacted by the heterogeneous genetic landscape of
Experimental data from other investigators also argue that dissemi- pancreatic cancer patients. A minimal PI3Ka activity is important
nating cells are detected very early in the disease history (Rhim for migratory ability of cells and contributes to their metastatic
et al, 2012). Our longitudinal analysis of cfDNA levels demonstrates potential regardless of the genetic landscape of the tumour. In the
an early increase of these parameters in an experimental pancreatic model described by Thorpe et al (2015), PI3K isoform efficiency is
cancer model. Quantification of the concentration of the 160–210 bp for some patients predicted with the driving genetic alterations lead-
fragment increases the prediction rate of survival, as evidenced in ing to PI3K hyperactivation: oncogenic Kras mutation promotes
patients in other studies using circulating mutation rates (Mouliere increased dependence to PI3Ka; loss of PTEN, increased dependence
et al, 2018). to PI3Kb; oncogenic PI3Ka mutation, increased dependence to
In terms of toxicity, the main concern for using potent PI3Ka PI3Ka. We find that even Pten-depleted PDAC cell lines require a
inhibitors remains the induction of insulin feedback which could minimal PI3Ka activity to migrate. We support this result by demon-
feed the tumours (Goncalves et al, 2018). These concerns could be strating that a basal level of PI3Ka activation is sufficient to drive
resolved by the clinical management of glycaemia during treatment these pro-metastatic features (Fig 4). This result fits with the rare
(Rodon et al, 2013; Goncalves et al, 2018). It has to be noted that detection of oncogenic mutations in PI3Ka encoding gene (Waddell
this insulin feedback occurs to be reduced by age and that PDAC is et al, 2015) despite the crucial role of PI3Ka activity in this cancer
mostly detected in patients > 40 years (Foukas et al, 2013). initiation (Baer et al, 2014; Wu et al, 2014). PI3Ka importance in
Our data also demonstrate that oncogenic KRAS-PI3Ka coupling controlling PDAC cell migration can occur independently from Kras
leads to specific functions of clinical relevance in the pancreatic mutation and is also promoted by FGFR signalling. There is a vast
context. In mutant KRAS-driven lung cancers, inactivation of PI3Ka literature showing that FGFR2 is promoting PDAC cell migration
yielded only a partial response (Castellano et al, 2013), MAPK path- in vitro, in vivo and in patients (Nomura et al, 2008). We could
way activity being key in this context. From our results, there are speculate that remodelling the actin cytoskeleton appears to be key
four non-exclusive explanations of the importance of KRAS-PI3Ka in PDAC progression so that tumour cells can extrude themselves
coupling in this organ setting. from the strong desmoplastic and low vascularised pancreatic
Firstly, in the pancreas, PI3Ka selective inactivation decreased primary tumours. This could also be the case for the evolution of
the phosphorylation of Erk1/2. Our data are not in line with data micro-metastasis: others found that, in metastatic sites, macrophage
from other studies in this regard (Alagesan et al, 2015; Junttila et al, VCAM binding critically activates PI3K-Ezrin in breast metastatic
2015); other authors used pan-PI3K inhibitors. MEK inhibition has a cells and promotes transition to macro-metastatic foci (Chen et al,
minimal anti-tumoural action due to the induction of strong feed- 2011). Ezrin plays a critical role in maintaining cell shape and lamel-
back activation on Akt (Kong et al, 2016). Selectivity towards PI3Ka lipodial extensions (Lamb et al, 1997). Early on in the discovery of
could prevent the induction of compensatory signals towards the oncogenic KRAS properties, it was demonstrated that the PI3K
MAPK pathway. signal towards actin remodelling was the first major signalling event
Secondly, the action of PI3Ka on actin cytoskeleton remodelling leading to cell transformation (Rodriguez-Viciana et al, 1997).
in pancreatic cancer cells is sensitive to pharmacological inhibition. Subsequently, actin remodelling was also linked to PI3Ka-driven

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 13 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

glucose metabolism regulation (Hu et al, 2016). Actin cytoskeleton favour the prominent role of PI3Ka downstream KRASG12D in
remodelling and associated cellular functions such as migration and PDAC. However, the multiple PI3K isoform engagement could also
motility are sensitive to unique and isoform-selective PI3Ka inhibi- explain why some tumours appear to escape from BYL-719 treat-
tion. Our data describing production of PIP3 with distinctive acyla- ment in vivo. Interestingly, while all BYL-719-treated tumours
tion pattern points to a novel mechanism to explain PI3K isoform presented a significant decrease in CD206 staining, the effects on
selectivity. Currently, published data are increasingly showing that CD4 and CD8 immune cell population were found heterogeneous;
the acylation state of phospholipids could modulate their localisa- this is associated with the fact that some BYL-719-treated mice
tion and function (Choy et al, 2017). In this context, we can provide presented (albeit in a reduced size and number) macro-metastatic
additional information to show that pan-PI3K inhibitors or isoform- foci. Anti-tumoural action of PI3Ka partial genetic inactivation was
specific inhibitors could target different PIP3 subspecies and conver- also found heterogeneous in C57/B6 background. However, in the
sion to PI-3,4-P2 ultimately leading to clear-cut effects on migration nude mice model, that is devoid of lymphocytes but presents macro-
or cytotoxicity and explaining isoform selectivity. phages, three mice out of eight displayed lower effects of BYL-719
Thirdly, pro-tumoural macrophages in the organ around the treatment, suggesting that other parameters than immune cell
tumour could favour metastasis evolution. Others found that activa- rewiring could be involved in resistance to treatment. In vitro, some
tion of macrophages associated with increased apoptotic tumour cell lines were also sensitive to other PI3K isoform inhibitors (e.g.
cells accelerated growth and not implantation of prostate metastatic PI3Kb). Both PI3Kc and PI3Ka are important for pancreatic cancer
nodules (Roca et al, 2018). Similarly, systemic secretions of the (Torres et al, 2019); the possible crosstalk between these two
tumour-associated macrophages that are found around the primary isoforms should be investigated. In a future study, we aim to dissect
pancreatic tumours could favour growth of metastatic micro- the mechanisms of resistance to PI3Ka inhibition, in an aim to
metastatic foci. Tumour-intrinsic PI3Ka indirectly rewires immune increase the efficiency of this therapeutic agent on PDAC evolution.
cell composition in the tumour site; this increase of CD206-positive In conclusion, our data demonstrate that PI3K-targeting agents
cell counts in the tissue corresponding to tumour-associated could be effective in the management of micro-metastatic disease
inflammatory macrophages could act as a distant event and promote (assessed by cfDNA), in PDAC patients, preventing macro-
metastatic evolution. IL-3 regulates varied inflammatory responses metastatic evolution. PI3K inhibitors also trigger indirect
that promote the rapid clearance of pathogens but also contribute to immunomodulatory actions. Our work provides sufficient ground to
pathology in chronic inflammation. Therapeutic interventions support the emergence of an extended translational study in PDAC
manipulating this cytokine are so far only developed in AML patients to corroborate these findings.
(Dougan et al, 2019), but increased understanding of its action in
solid tumours is needed before their intervention being included to
the arsenal of immunotherapies. Suppression of macrophages in the Materials and Methods
KPC mice and their primary tumours by clodronate treatment did
not delay lethality, but decreased incidence of macro-metastasis Transcriptomics and bioinformatics analysis
(Griesmann et al, 2017). The tumour-intrinsic action of PI3Ka on
tumour inflammation that we described here could be added to its We selected transcriptional profiling data sets of normal pancreas,
direct immunomodulatory action in pancreatic cancer found by chronic pancreatitis and pancreatic primary tumoural tissues from
others (Sivaram et al, 2019). Hence, PI3Ka pro-metastatic action localised (PDACloc) or metastatic patients (PDACmet). Published
acts both via a direct actin cytoskeleton remodelling and FGFR- data on human samples were retrieved from public databases
induced tumour cell migration as well as via tumour-extrinsic E_EMBL_6 (Abdollahi et al, 2007) from compatible platforms,
promotion of TNFa secretion by macrophages and subsequent normalised using the RMA method (R 3.2.3, bioconductor version
further activation of migratory phenotype. These data are in line 3.2), collapsed (collapse microarray), filtered (SD > 0.25) and statis-
with our early work (Guillermet et al, 2003; Bousquet et al, 2006) tically tested using an ANOVA test corrected using the Benjamini
that showed that class IA PI3K activity in pancreatic cancer cells is and Hochberg method (BH). For each sample, individual scoring for
critical to activate NF-jB activity, preventing TNFa-induced cell hallmarks or Reactome (actualised list of genes downloaded from
death and promoting cell survival and migration. Finally, Fig 1A MSigDB version [[Link]. org/gsea/msigdb] and
showed that Hallmark TNFa signalling via NFKB is one of the 6 hall- Reactome [[Link]]) was performed using Autocom-
marks significantly changed in metastatic PDAC patients, further pare_SES software (available at [Link]
validating our finding. TNFa was previously found to promote tran- softwares/products/autocompare_ses) using the “greater” (indicat-
sition from micro- to macro-metastasis as BM-derived EPCs are ing an enriched gene set) Wilcoxon tests with frequency-corrected
known to be critical regulators of the angiogenic switch in progres- null hypotheses (Tosolini et al, 2016), followed by values in each
sion of micro-metastasis to lethal macro-metastasis (Gao et al, group of patients compared using an ANOVA test. Hierarchical
2008), and tumour-derived TNF signalling had been linked in vivo patient clustering was performed using the PI3Ka activation signa-
to differentiation of myeloid progenitor cells to “byphenotypic” ture. The PI3Ka gene signature was designed as the intersection of
myeloid/ECs (Li et al, 2009). genes up-regulated and down-regulated in 20 breast tumours after
Fourthly, the repartition of each KRAS mutation is different in BYL-719 treatment (Bosch et al, 2015) and LINCS shRNA CMAP sig
each solid cancer, with G12D mutation being the most common in gene list (Zhang et al, 2017) (Appendix Fig S1). This list was
PDAC. Recent evidence shows that each KRAS mutation drives dif- narrowed down to 20 genes, which expression was found compati-
ferent signalling and engages different pathways (Cayron & ble with the quality criteria (filter) detailed above. Unsupervised
Guillermet-Guibert, 2020; Hobbs et al, 2020). The published data hierarchical clustering of the E_EMBL_6 data set was performed

14 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

focusing on the expression of these 20 genes regulated by PI3Ka in Digital PCR (ddPCR) System (Bio-Rad), according to the manufac-
cancer. turer’s recommendation. Samples with high-molecular DNA were
Confirmed PDAC samples from public databases were selected excluded from the analysis. Allelic frequency percentages for KRAS
for further bioinformatics analysis. In detail, mRNA expression data mutation were obtained using QuantoSoft software (Bio-Rad).
and clinical data from confirmed PDAC patients of PAAD (TCGA)
(175 patients) and PACA-AU (267 patients) cohorts were retrieved. Inhibitors and ligands
Amongst the 175 well-annotated TGCA patients, 21 patients were
considered localised according to their UICC staging (T = 0, 1 or 2, For in vitro use, all PI3K inhibitors (Knight et al, 2004; Jackson et al,
N = 0, M = 0). For each patient, a PI3Ka activation signature score 2005; Pomel et al, 2006; Ali et al, 2008; Folkes et al, 2008; Raynaud
or immune cell infiltration score (to quantify LTcd, NK, LT CD8, et al, 2009; Burger et al, 2011; Jamieson et al, 2011; Fritsch et al,
Monocyte-Macrophage-DC, B cells, granulocytes, LT CD4) was 2014; Barlaam et al, 2015) (Dataset EV4) and the FGFR inhibitor
given using SES auto compare software, and patients were hierar- AZD4547 were purchased (CliniSciences) and dissolved in dimethyl
chically clustered in three groups corresponding to high, medium or sulfoxide (DMSO) to obtain a stock concentration of 10 mM, subse-
low scoring. Scores in each group were statistically tested using an quently diluted as indicated and compared to the diluted DMSO
ANOVA test corrected according to the Benjamini and Hochberg vehicle (vehicle). In vivo, BYL-719 (ApexBio) was dissolved in 0.5%
method (BH). For PI3Ka activation signature, high and medium methyl cellulose with 0.2% Tween-80 and administered by oral
groups were then pooled. The overall survival of patients in each gavage at 50 mg/kg daily. TNF-alpha was purchased (Sigma-
cluster was plotted and statistical differences were calculated using Aldrich) and dissolve in PBS to obtain a stock concentration of
the log rank test. The prognosticator value of PI3Ka activation 1 mg/ml.
signature and IUCC staging were tested independently and
compared to the value of clinical T,N,M staging using the multivari- Cell lines
ate Cox test and the PAAD database. We verified that within each
patient cluster, there was no enrichment in terms of genetic changes All the cell lines described in Dataset EV4 were obtained from the
associated with the PI3K/Akt pathway and, in particular, that onco- American Type Culture Collection (ATCC, Manassas, VA) or from
genic mutations of PI3Ka and PTEN were infrequent and equally genetically engineered mouse models or from CRB, Toulouse, IUCT-
distributed in each group of patients (mutational pattern available O. Mutant KRAS, with or without partially deficient PI3Ka activity
in PAAD cohort only: Dataset EV3, Appendix Fig S2). SES scores for pancreatic cancer cell lines were derived from the pancreas and lung
PI3Ka activation were also calculated in PDAC RNA subtypes using of KC or KC;p110a+/lox animals (aged 10–13 months). Peritoneal
classification method from Puleo et al (2018). metastatic cells pASC1 and pASC3 are primary cells isolated from
metastatic patients with PDAC in the IUCT-O. R211-Luc cells are
Human pancreatic samples R211 cells modified to express the luciferase. General methods were
used and are detailed in the Appendix and Dataset EV4. IC21 cells
Patient samples from BACAP collection were collected and stored were treated for 1 day with the conditioned medium, pre-incubated
with the “CRB Cancer des H^ opitaux de Toulouse”. Informed consent with control Antibody (0.5 µg/ml, MAb rat IgG1 isotype control) or
was obtained from all subjects and the experiments conformed to anti-IL3 antibody (0.5 µg/ml, Clone MP2-8F8, Catalog number: BX-
the principles set out in the WMA Declaration of Helsinki and the BE0282-1MG InVivoMab anti-mouse IL-3) and TNFa measured from
Department of Health and Human Services Belmont Report. The supernatant. Lipids were extracted and derivatised using TMS-
BACAP collection has been declared to the Ministry of Higher diazo-methane as previously described (Clark et al, 2011).
Education and Research (DC-2008-463), and transfer agreements
(AC-2008-820) (AC-2013-1955) have been obtained following siRNA, stably transfected cell lines
approval by ethical Committees. Clinical and biological annotations
of the samples have been declared to the CNIL (Comite National Cells were transfected with Lipofectamine 2000 (ThermoFisher
Informatique et Libertes—French national Data Protection Agency). Scientific) with SMARTpool ON-TARGETplus mouse siRNA (Dhar-
All patient records and information were anonymised and encrypted macon) targeting: Pik3ca, Pik3cb, Pik3cg, Pik3cd according to the
prior to analysis. IHC is detailed in Appendix, antibody used in manufacturer’s protocols as used in (Kingham & Welham, 2009;
Dataset EV4. Höland et al, 2014; Huang et al, 2015). ON-TARGETplus Non-
Blood was collected from patients diagnosed with pancreatic targeting control siRNAs (Dharmacon) were used as controls.
adenocarcinoma and analysed by digital droplet PCR (ddPCR) for Twenty-four hours after transfection, the cells were used for migra-
KRAS mutation. Briefly, 10 ml of blood was centrifuged twice in tion or RT–qPCR experiments. Panc-1 cells were stably transfected
PAXgene blood ccfDNA tubes (Qiagen) at 1,200 g for 10 min at 4°C by lentiviral transduction with pLVTHM-p110alpha1 (forward:
and 16,000 g for 10 min at 4°C. Cell-free plasma was collected and 50 CGCGTCCCCGCGAAATTCTCACACTATTATTTCAAGAGAATAAT
total DNA was extracted from 3ml of plasma using the QIAamp AGTGTGAGAATTTCGCTTTTTGGAAAT; reverse: 50 CGATTTCCAAA
Circulating Nucleic Acid Kit (Qiagen) according to the manufac- AAGCGAAATTCTCACACTATTATTCTCTTGAAATAATAGTGTGAG
turer’s recommendation. Circulating cell-free DNA (cfDNA) ranging AATTTCGCGGGGA); pLVTHM-p110alpha2 (forward: 50 CGCGTCC
from 110-210 base pairs was qualified and quantified using the CCGCACAATCCATGAACAGCATTTTCAAGAGAAATGCTGTTCATG
DNF-474 high sensitivity ngs fragment analysis kit (Agilent). For GATTGTGCTTTTTGGAAAT; reverse: 50 CGATTTCCAAAAAGCACA
ddPCR, 5 ng of cfDNA were analysed using the ddPCRTM KRAS G12/ ATCCATGAACAGCATTTCTCTTGAAAATGCTGTTCATGGATTGTGC
G13 Screening Kit #1863506 (Bio-Rad) and the QX200 Droplet GGGGA), pLVTHM-p110beta1 (forward: 50 CGCGTCCCCCACATT

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 15 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

GACTTTGGACATATCGCGTCCCATATGTCCAAAGTCAATGTGGTTTT mice (Charles River). Three-week post-injection, lungs were inflated

TGGAAAT; reverse: 50 CGATTTCCAAAAACCACATTGACTTTGGAC then fixed and embedded in paraffin.
ATATTCTCTTGAAATATGTCCAAAGTCAATGTGGGGGGA), pLVTHM-
p110beta2 (forward: 50 CGCGTCCCCCGACAAGACTGCCGAGAGATT Pancreatic inflammation
TTCAAGAGAAATCTCTCGGCAGTCTTGTCGTTTTTGGAAAT; reverse:
50 CGATTTCCAAAAACGACAAGACTGCCGAGAGATTTCTCTTGAAA Pancreatic injury was induced on young KC mice (8–12 weeks old)
ATCTCTCGGCAGTCTTGTCGGGGGA), pLVTHM-scr (forward: 50 CG by a series of six hourly intra-peritoneal injections of caerulein
CGTCCCCTTCTAGAGATAGTCTGTACGTTTCAAGAGAACGTACAGA (75 µg/kg of body weight, Bachem) repeated after 48 h. Eight days
CTATCTCTAGAATTTTTGGAAAT; reverse: 50 CGATTTCCAAAAAT later, a treatment/intervention protocol was developed to allow
TCTAGAGATAGTCTGTACGTTCTCTTGAAACGTACAGACTATCTCT formation of pre-cancer lesions: mice were treated with the vehicle
AGAAGGGGA) and selected positively by cell sorted flow cytometry (0.5% methyl cellulose with 0.2% Tween-80) or with GDC0326
for GFP expression. (10 mg/kg) by gavage (Appendix Fig S16A).

Animal models KPC cohort follow-up (ultrasound imaging, blood collection and
cytokine profiling)
All animal procedures were conducted in compliance with the
Ethics Committee pursuant to European legislation translated into Ultrasound imaging was performed using the VisualSonics Vevo2100
French Law as Decret 2013-118 dated 1st of February 2013 (APAFIS High-Resolution System equipped with an ultrasound transducer in
3601-2015121622062840). the 25–55 MHz range. Animal preparation and imaging procedures
The LSL-KRASG12D (K) and LSL-p53R172H (P) knock-in from D. were performed as described in Sastra and Olive (2013). KPC mice
Tuveson, Mouse Models of Human Cancers Consortium Repository, were monitored once a week from 12 weeks old onwards; when a
Frederick National Cancer Institute, Pdx1-Cre (C) from D.A. Melton, tumour was detected, ultrasound scans were performed every other
Harvard University, Cambridge, MA, Pdx-1Cre (C) from D. Tuveson day. The tumour area was measured by delimiting the tumour
and p110alox/lox from B. Vanhaesebroeck, University College border and determining the major axis; at least five replicates were
London strains were interbred against a mixed background (CD1/ performed per mice per ultrasound. Tumour volume was calculated
SV129/C57Bl6) or in a C57/B6 background. KPC mice are of mixed using the formula V = (4/3) × p × (Length/2)2 × (Depth/2). The
gender. KPC mice and compound mice with p110alox line were bred tumour volume fold change corresponds to the increased fold change
in three animal houses (mixed background, Melton’s Pdx1-Cre: in the tumour volume after treatment onset.
CRCT, Anexplo, Toulouse, France, 2 sites) (Therville et al, 2019) Blood samples were collected every 2 weeks via retro-orbital
and (C57B6 background, Tuveson’s Pdx1-cre: CRCM, Marseille, collection from the age of 12 weeks onwards. Blood counts were
France). The Ptf1aCre/+-LSL-KRASG12D/+ (KC; p110a+/+) and Ptf1aCre/+- performed using Yumizen H500 haematology analyser (HORIBA),
LSL-PIK3CAH1047R/+ (p110aH1047R) strains were bred at the Technische calibrated for murine blood. A maximum volume of 100–150 ll was
Universit€
at M€ unchen (Technical University, Munich). All mice were collected on each occasion. The blood was collected using Pasteur
housed and bred under specific pathogen-free conditions maintained glass pipettes and then transferred to Eppendorf tubes containing
in the accredited animal facility. Mice were housed with a 12-h day– 20 ll 0.5 M EDTA. Plasma was separated by centrifuging blood at
night cycle with lights on at 7:00 AM in a temperature (22
1°C) 1,500 g for 20 min at 4°C within 3 h of blood collection. The plasma
and humidity (55
5%)-controlled room. All mice were allowed free was stored at 80°C until required for further use.
access to water and food. All cages contained wood shavings, bedding A MILLIPLEXMAP (Merc Millipore # MCYTMAG70PMX32BK)
and a cardboard, an igloo or a sizzle nest tube for environmental assay was performed using 25 ll of non-diluted murine blood
enrichment. KPC mice were treated 5 days a week with vehicle or plasma. We specifically tested for a panel of 32 murine chemokines
BYL-719 administered by oral gavage at 50mg/kg daily starting the and cytokines: Eotaxin, G-CSF, GM-CSF, IFNc, IL-1a, IL-1b, IL-2, IL-
day after the tumour detection date. 3, IL-4, IL-5, IL-6, IL-7, IL-9, IL-10, IL-12 (p40), IL-12 (p70), IL-13,
IL-15, IL-17, IP-10, KC, LIF, LIX, MCP-1, M-CSF, MIG, MIP-1a, MIP-
Tail vein injection 1b, MIP-2, RANTES, TNFa and VEGF. IL3 and TNFa could not be
detected on a comparative scale.
All obtained mice were acclimatised for at least 1 week. Vendor
health reports indicated that the mice were free of known viral, cfDNA extraction from murine blood plasma
bacterial and parasitic pathogens. 5 × 104 R211-Luc cells were
injected into the tail vein of female 8-week-old nude mice (Charles For cfDNA extraction, blood plasma was re-centrifuged at 18,000 g
River). The mice were treated 5 days a week with vehicle or BYL- for 10 min at room temperature to reduce debris contamination.
719 administered by oral gavage at 50 mg/kg daily for 3 weeks start- cfDNA was extracted from blood plasma using the QIAmp DNA
ing on the injection date. Two- and three-week post-injection, mice Mini Kit (QIAGEN) protocol except for eluting the cfDNA in 50 ll of
were injected i.p. with 150 mg/kg of RediJect D-Luciferin (Perkin elution buffer. cfDNA samples were stored at 20°C until required
Elmer) and monitored for Luciferase expression after 6–8 min in the for further use. CfDNA quantification was performed by qPCR as
IVIS Spectrum in vivo imaging system (Perkin Elmer). Luminescence described below.
(photo count) was measured for each mouse. Lungs were inflated For quantifying the relative cfDNA in blood plasma, we initially
then fixed and embedded in paraffin. 1 × 105 A338 or 1 × 105 A260 plotted a standard curve using extracted DNA from a murine cell line
cells were injected into the tail vein of female 8-week-old C57/B6 R211 (without LSL cassette, expressing mutated KRAS and TP53)

16 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

and from pancreatic extracts from mice expressing the LSL cassette
The paper explained
(not recombined). The qPCR was performed using 1 lL of DNA and
SsoFast Eva Green Supermix (Bio-Rad). For the standard curve, we Problem
did serial dilutions up to 1/1,000 of the two DNA extracts and a qPCR Pancreatic cancer is one of the most lethal solid cancers and is char-
of two different genes, p53 and GAPDH. We designed and selected acterised by rapid progression after primary tumour detection. The
the most specific and efficient primers (Sigma-Aldrich) using key signalling events driving this fast evolution into macro-metastatic
disease are still unknown.
Primer-Blast (NCBI). The obtained standard curves allowed subse-
quent quantification of cfDNA in mouse plasma samples.
Results
The following primers were used: p53 gene (Forward: 50 -CC Two unbiased approaches led to the identification of a high PI3Ka
AGCTCAGCCTTTGTAGTGAA-30 ; reverse: 50 -GTGCAGCCCTAAGCA activation signature in pancreatic primary tumours with bad progno-
TCTAGC-30 ; chromosome 11), GAPDH gene (Forward: 50 -AGCCC sis. Our in vitro data showed that PI3Ka is a major positive regulator
CAGGCTATCTGATGT-30 ; 50 -ATAGCTGATGGCTGCAGGT-30 ; chro- of cancer cell escape from the primary tumour through actin
cytoskeleton remodelling. PI3Ka was inhibited in two preclinical
mosome 6).
models of pancreatic cancer. First, in a model of micro-metastatic
cfDNA thresholds were calculated by compiling all cfDNA disease (extra-pancreatic dissemination that goes undetected by ultra-
measurements from mice with normal pancreas, high-grade PanINs, sound (US) imaging), mice presenting US-detected primary pancreatic
localised and metastatic PDAC (Dataset EV5). For the survival curve tumours and increased circulating cell-free DNA (cfDNA) were treated.
(Kaplan–Meier), mice were categorised as having a normal, low or In a second model, tumour cell implantation and early proliferation in
metastatic organs after intravascular injection were analysed. A clini-
high level of cfDNA according to their mean cfDNA values. As for
cally relevant PI3Ka-selective inhibitor (BYL-719/Alpelisib), currently
thresholds, the low range corresponds to mice presenting 0 – 0.003 tested in pancreatic cancer patients without patient stratification,
AU of cfDNA and the high range to mice presenting cfDNA above was used in both models. Inhibition of PI3Ka delayed primary tumour
the mentioned level. In order to establish these ranges, we analysed and micro-metastasis evolution, showing that PI3Ka activity drove the
the histology of each mouse at end point. cfDNA was quantified evolution of micro-metastatic disease towards the macro-metastatic
stage in these models. Mechanistically, tumour-intrinsic PI3Ka activity
using blood collected at sacrifice/death. All raw data and threshold
increased pro-tumoural characteristics in peritumoural immune cells
calculations are shown in Dataset EV5. via increased IL-3 cytokine production. In return, inflammatory macro-
The Fragment AnalyzerTM was used to determine the size of phages increased TNFa production, facilitating tumour cell migration.
DNA fragments in blood plasma. The DNF-474 High Sensitivity NGS
Fragment Kit was used to characterise cfDNA. Three different Impact
cfDNA fragmentation profiles were obtained: for normal and healthy In pancreatic cancer patients, PI3Ka-targeting agents could be effec-
mice, the electropherogram did not present any fragment; for mice tive in the management of micro-metastatic disease assessed by
cfDNA, preventing macro-metastatic evolution. PI3Ka inhibitors also
with high-grade PanINs and localised PDAC, a 160–210 bp fragment trigger indirect immunomodulatory actions and could be added in the
was always found; for mice with metastatic PDAC, the electrophero- arsenal of immunomodulatory agents.
gram presented the aforementioned 160–210 bp fragment, in addi-
tion to larger fragments but at lower concentrations.

Statistical analysis Acknowledgements

We are grateful to SigDYN members, past and present, for their technical
Experimental data provided at least three experimental replicates support, sample banks, common tools, scientific and protocol discussions, CRB
and three to five measurement replicates, except for WB experi- and BACAP consortium for patient samples, UMS006/CREFRE, Anexplo Plat-
ments. Statistical analyses were performed with GraphPad Prism form, Toulouse (mouse breeding and experimental zone; ENI core platform),
using student t-test for in vitro data; non-parametric Mann–Whitney the CRCT core technology platform in particular Laetitia Ligat, Carine Valle
for all in vivo data; Log rank test was used for all survival curves: and Emeline Sarot, ImagIN platform (FX Fresnois), Dr. Barbara Garmy-Susini
*P < 0.05, **P < 0.01, ***P < 0.001. Correlation analysis was for access to her laboratory while moving our own facilities, the staff of Histol-
performed using Pearson r test. Statistical relevance of the murine ogy platform (I2MC, Inserm 1048, Toulouse, France), MetaToul-Lipidomique
KPC cohort size was determined using a Power Calculation test Core Facility (I2MC, Inserm 1048, Toulouse, France), MetaboHUB-ANR-11-
([Link]/.../power_calculator_14_may_2014.xls), INBS-0010 for lipidomic analysis, advices and technical assistance, and
those data are shown in Dataset EV5. For patient data analysis, Genentech for GDC0326. JGG is a member of COST action EU Pancreas
ANOVA test corrected using the Benjamini and Hochberg method BM1204. JGG’s laboratory belongs to Toucan, Laboratoire d’Excellence, ANR, an
(BH) on SES (sample enrichment score). integrated research programme on Signal-targeted Drug Resistance. JGG’s
laboratory for this topic was/is funded by Europe EU-ERG FP7 (270696 PaCa/
PI3K), ARC (PJA20171206596; salary for RB), Toucan ANR Laboratory of Excel-
Data availability lence, MSCA-ITN/ETN PhD-PI3K (Project ID: 675392, salary for FR-D and SA),
Fondation de France (salary for BT), GSO, Ligue Nationale Contre le Cancer
This study includes no data deposited in external repositories. (salary for CC and CC), Fondation Toulouse Cancer Sante (revision experi-
Normalised raw data of analysed public data are available in Dataset ments). Both FR-D and SA disseminated their research to high school students,
EV1 enabling a further analysis on the built cohorts (normal, PDAC as part of their commitment in MSCA-ITN funding. DS laboratory for this
loc, PDAC met, CP). research was funded by Deutsche Forschungsgemeinschaft (DFG, German
Research Foundation)—Project ID 329628492—SFB 1321 (DS) and the Euro-
Expanded View for this article is available online. pean Research Council (ERC CoG No. 648521 (D.S.).

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 17 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

Author contributions inhibitor of PI3Kb and PI3Kd for the treatment of PTEN-deficient cancers. J
BT, FR-D, EP-T, NT, CC, SA, SC-S, GR-G, MT, AVV, CC, RB, JB-M, DP, DFM, AC, JG- Med Chem 58: 943 – 962
G: experiments; BT, FR-D, EPT, SA, GRC, MT, AVV, CC, AC, PC, CB, JG-G: formal Bosch A, Li Z, Bergamaschi A, Ellis H, Toska E, Prat A, Tao JJ, Spratt DE, Viola-
analysis of data; BT, FR-D, EP-T, NT, CC, EA, PC, DS, CB, JG-G: methodology; BT, Villegas NT, Castel P et al (2015) PI3K inhibition results in enhanced
FR-D, EP-T, MT, JG-G: visualisation of data; BT, FR-D, JG-G: writing—original estrogen receptor function and dependence in hormone receptor-positive
draft; BT, FR-D, EP-T, NT, MT, JB-M, PC, JG-G: materials and methods writing; breast cancer. Sci Transl Med 7: 283ra51
BT, FR-D, EP-T, JG-G: review of writing; all authors: editing MS; GRC, HY, CF, Bousquet C, Guillermet-Guibert J, Saint-Laurent N, Archer-Lahlou E, Lopez
FM, BB, AC, DS, JG-G: provided samples; J-PD, EA, AC, PC, DS, CB, JG-G: supervi- F, Fanjul M, Ferrand A, Fourmy D, Pichereaux C, Monsarrat B et al
sion; BT, FR-D, EP-T, JB-M, DS, JG-G: funding acquisition; JG-G: conceptualisa- (2006) Direct binding of p85 to sst2 somatostatin receptor
tion, project administration, project supervision, validation. reveals a novel mechanism for inhibiting PI3K pathway. EMBO J 25:
3943 – 3954
Conflict of interest Burger MT, Pecchi S, Wagman A, Ni Z-J, Knapp M, Hendrickson T, Atallah G,
EPT was funded by Cellgene. WO2021001431-A1 is a filed patents pertaining Pfister K, Zhang Y, Bartulis S et al (2011) Identification of NVP-BKM120 as
to the results presented in the paper. a potent, selective, orally bioavailable class I PI3 kinase inhibitor for
treating cancer. ACS Med Chem Lett 2: 774 – 779
For more information Castellano E, Sheridan C, Thin M, Nye E, Spencer-Dene B, Diefenbacher M,
i [Link] Moore C, Kumar M, Murillo M, Grönroos E et al (2013) Requirement for
ii [Link] interaction of PI3-kinase p110a with RAS in lung tumor maintenance.
Cancer Cell 24: 617 – 630
Cayron C, Guillermet-Guibert J (2020) The type of KRAS mutation drives
References PI3Ka/c signalling dependency: implication for the choice of targeted
therapy in pancreatic adenocarcinoma patients. Clin Res Hepatol
Abdallah R, Taly V, Zhao S, Pietrasz D, Bachet J-B, Basile D, Mas L, Zaanan A, Gastroenterol 45: 101473
Laurent-Puig P, Taieb J (2020) Plasma circulating tumor DNA in pancreatic Celi

a-Terrassa T, Kang Y (2016) Distinctive properties of metastasis-initiating
adenocarcinoma for screening, diagnosis, prognosis, treatment and follow- cells. Genes Dev 30: 892 – 908
up: a systematic review. Cancer Treat Rev 87: 102028 Chen Q, Zhang XH-F, Massague J (2011) Macrophage binding to receptor
Abdollahi A, Schwager C, Kleeff J, Esposito I, Domhan S, Peschke P, Hauser K, VCAM-1 transmits survival signals in breast cancer cells that invade the
Hahnfeldt P, Hlatky L, Debus J et al (2007) Transcriptional network lungs. Cancer Cell 20: 538 – 549
governing the angiogenic switch in human pancreatic cancer. Proc Natl Chen Y, LeBleu VS, Carstens JL, Sugimoto H, Zheng X, Malasi S, Saur D, Kalluri
Acad Sci USA 104: 12890 – 12895 R (2018) Dual reporter genetic mouse models of pancreatic cancer identify
Aiello NM, Bajor DL, Norgard RJ, Sahmoud A, Bhagwat N, Pham MN, Cornish an epithelial-to-mesenchymal transition-independent metastasis program.
TC, Iacobuzio-Donahue CA, Vonderheide RH, Stanger BZ (2016) Metastatic EMBO Mol Med 10: e9085
progression is associated with dynamic changes in the local Choy CH, Han B-K, Botelho RJ (2017) Phosphoinositide diversity, distribution,
microenvironment. Nat Commun 7: 12819 and effector function: stepping out of the box. BioEssays 39: 1700121
Alagesan B, Contino G, Guimaraes AR, Corcoran RB, Deshpande V, Cintas C, Douche T, Therville N, Arcucci S, Ramos-Delgado F, Basset C,
Wojtkiewicz GR, Hezel AF, Wong K-K, Loda M, Weissleder R et al (2015) Thibault B, Guillermet-Guibert J (2018) Signal-targeted therapies and
Combined MEK and PI3K inhibition in a mouse model of pancreatic resistance mechanisms in pancreatic cancer: future developments reside
cancer. Clin Cancer Res 21: 396 – 404 in proteomics. Cancers 10: 174
Ali K, Camps M, Pearce WP, Ji H, Ru €ckle T, Kuehn N, Pasquali C, Chabert C, Clark J, Anderson KE, Juvin V, Smith TS, Karpe F, Wakelam MJO, Stephens LR,
Rommel C, Vanhaesebroeck B (2008) Isoform-specific functions of Hawkins PT (2011) Quantification of PtdInsP3 molecular species in cells
phosphoinositide 3-kinases: p110 delta but not p110 gamma promotes and tissues by mass spectrometry. Nat Methods 8: 267 – 272
optimal allergic responses in vivo. J Immunol 180: 2538 – 2544 Collisson EA, Sadanandam A, Olson P, Gibb WJ, Truitt M, Gu S, Cooc J,
Alix-Panabi
eres C, Pantel K (2016) Clinical applications of circulating tumor Weinkle J, Kim GE, Jakkula L et al (2011) Subtypes of pancreatic ductal
cells and circulating tumor DNA as liquid biopsy. Cancer Discov 6: adenocarcinoma and their differing responses to therapy. Nat Med 17:
479 – 491 500 – 503
Andre F, Ciruelos E, Rubovszky G, Campone M, Loibl S, Rugo HS, Iwata H, Conway JR, Herrmann D, Evans TJ, Morton JP, Timpson P (2019) Combating
Conte P, Mayer IA, Kaufman B et al (2019) Alpelisib for PIK3CA -mutated, pancreatic cancer with PI3K pathway inhibitors in the era of personalised
hormone receptor-positive advanced breast cancer. N Engl J Med 380: medicine. Gut 68: 742 – 758
1929 – 1940 Daley D, Zambirinis CP, Seifert L, Akkad N, Mohan N, Werba G, Barilla R,
Baer R, Cintas C, Dufresne M, Cassant-Sourdy S, Schönhuber N, Planque L, Torres-Hernandez A, Hundeyin M, Mani VRK et al (2016) cd T cells support
Lulka H, Couderc B, Bousquet C, Garmy-Susini B et al (2014) Pancreatic pancreatic oncogenesis by restraining ab T cell activation. Cell 166:
cell plasticity and cancer initiation induced by oncogenic Kras is 1485 – 1499.e15
completely dependent on wild-type PI 3-kinase p110a. Genes Dev 28: Dawson S-J, Tsui DWY, Murtaza M, Biggs H, Rueda OM, Chin S-F, Dunning
2621 – 2635 MJ, Gale D, Forshew T, Mahler-Araujo B et al (2013) Analysis of circulating
Barlaam B, Cosulich S, Degorce S, Fitzek M, Green S, Hancox U, Lambert-van tumor DNA to monitor metastatic breast cancer. N Engl J Med 368:
der Brempt C, Lohmann J-J, Maudet M, Morgentin R et al (2015) Discovery 1199 – 1209
of (R)-8-(1-(3,5-difluorophenylamino)ethyl)-N, N-dimethyl-2-morpholino-4- Delaloge S, DeForceville L (2017) Targeting PI3K/AKT pathway in triple-
oxo-4H-chromene-6-carboxamide (AZD8186): a potent and selective negative breast cancer. Lancet Oncol 18: 1293 – 1294

18 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benoit Thibault et al EMBO Molecular Medicine

Diehl F, Schmidt K, Choti MA, Romans K, Goodman S, Li M, Thornton K, through mobilization of aldolase from the actin cytoskeleton. Cell 164:
Agrawal N, Sokoll L, Szabo SA et al (2008) Circulating mutant DNA to 433 – 446
assess tumor dynamics. Nat Med 14: 985 – 990 Huang L, Sherchan P, Wang Y, Reis C, Applegate RL, Tang J, Zhang JH (2015)
Dougan M, Dranoff G, Dougan SK (2019) GM-CSF, IL-3, and IL-5 family of Phosphoinositide 3-kinase gamma contributes to neuroinflammation in a
cytokines: regulators of inflammation. Immunity 50: 796 – 811 rat model of surgical brain injury. J Neurosci 35: 10390 – 10401
Eser S, Reiff N, Messer M, Seidler B, Gottschalk K, Dobler M, Hieber M, Jackson SP, Schoenwaelder SM, Goncalves I, Nesbitt WS, Yap CL, Wright CE,
Arbeiter A, Klein S, Kong Bo et al (2013) Selective requirement of PI3K/ Kenche V, Anderson KE, Dopheide SM, Yuan Y et al (2005) PI 3-kinase
PDK1 signaling for Kras oncogene-driven pancreatic cell plasticity and p110beta: a new target for antithrombotic therapy. Nat Med 11: 507 – 514
cancer. Cancer Cell 23: 406 – 420 Jamieson S, Flanagan J, Kolekar S, Buchanan C, Kendall J, Lee W-J, Rewcastle
Folkes AJ, Ahmadi K, Alderton WK, Alix S, Baker SJ, Box G, Chuckowree IS, G, Denny W, Singh R, Dickson J et al (2011) A drug targeting only p110a
Clarke PA, Depledge P, Eccles SA et al (2008) The identification of 2-(1H- can block phosphoinositide 3-kinase signalling and tumour growth in
indazol-4-yl)-6-(4-methanesulfonyl-piperazin-1-ylmethyl)-4-morpholin-4- certain cell types. Biochem J 438: 53 – 62
yl-thieno[3,2-d]pyrimidine (GDC-0941) as a potent, selective, orally Junttila MR, Devasthali V, Cheng JH, Castillo J, Metcalfe C, Clermont AC, Otter
bioavailable inhibitor of class I PI3 kinase for the treatment of cancer. DD, Chan E, Bou-Reslan H, Cao T et al (2015) Modeling targeted inhibition
J Med Chem 51: 5522 – 5532 of MEK and PI3 kinase in human pancreatic cancer. Mol Cancer Ther 14:
Foukas LC, Bilanges B, Bettedi L, Pearce W, Ali K, Sancho S, Withers DJ, 40 – 47
Vanhaesebroeck B (2013) Long-term p110a PI3K inactivation exerts a Kingham E, Welham M (2009) Distinct roles for isoforms of the catalytic
beneficial effect on metabolism. EMBO Mol Med 5: 563 – 571 subunit of class-IA PI3K in the regulation of behaviour of murine
Fritsch C, Huang A, Chatenay-Rivauday C, Schnell C, Reddy A, Liu M, embryonic stem cells. J Cell Sci 122: 2311 – 2321
Kauffmann A, Guthy D, Erdmann D, De Pover A et al (2014) Knight ZA, Chiang GG, Alaimo PJ, Kenski DM, Ho CB, Coan K, Abraham RT,
Characterization of the novel and specific PI3Ka inhibitor NVP-BYL719 and Shokat KM (2004) Isoform-specific phosphoinositide 3-kinase inhibitors
development of the patient stratification strategy for clinical trials. Mol from an arylmorpholine scaffold. Bioorg Med Chem 12: 4749 – 4759
Cancer Ther 13: 1117 – 1129 €ger C, Regel
Kong Bo, Wu W, Cheng T, Schlitter AM, Qian C, Bruns P, Jian Z, Ja
Gao D, Nolan DJ, Mellick AS, Bambino K, McDonnell K, Mittal V (2008) I, Raulefs S et al (2016) A subset of metastatic pancreatic ductal
Endothelial progenitor cells control the angiogenic switch in mouse lung adenocarcinomas depends quantitatively on oncogenic Kras/Mek/Erk-
metastasis. Science 319: 195 – 198 induced hyperactive mTOR signalling. Gut 65: 647 – 657
Golan T, Hammel P, Reni M, Van Cutsem E, Macarulla T, Hall MJ, Park J-O, Lamb RF, Ozanne BW, Roy C, McGarry L, Stipp C, Mangeat P, Jay DG (1997)
Hochhauser D, Arnold D, Oh D-Y et al (2019) Maintenance olaparib for Essential functions of ezrin in maintenance of cell shape and lamellipodial
germline BRCA-mutated metastatic pancreatic cancer. N Engl J Med 381: extension in normal and transformed fibroblasts. Curr Biol 7: 682 – 688
317 – 327 Le Tourneau C, Delord J-P, Gonçalves A, Gavoille C, Dubot C, Isambert N,
Goncalves MD, Hopkins BD, Cantley LC (2018) Phosphatidylinositol 3-kinase, Campone M, Tredan O, Massiani M-A, Mauborgne C et al (2015)
growth disorders, and cancer. N Engl J Med 379: 2052 – 2062 Molecularly targeted therapy based on tumour molecular profiling versus
Graupera M, Guillermet-Guibert J, Foukas LC, Phng L-K, Cain RJ, Salpekar A, conventional therapy for advanced cancer (SHIVA): a multicentre, open-
Pearce W, Meek S, Millan J, Cutillas PR et al (2008) Angiogenesis label, proof-of-concept, randomised, controlled phase 2 trial. Lancet Oncol
selectively requires the p110alpha isoform of PI3K to control endothelial 16: 1324 – 1334
cell migration. Nature 453: 662 – 666 Lee B, Lipton L, Cohen J, Tie J, Javed Aa, Li L, Goldstein D, Burge M, Cooray P,
Griesmann H, Drexel C, Milosevic N, Sipos B, Rosendahl J, Gress TM, Nagrial A et al (2019) Circulating tumor DNA as a potential marker of
Michl P (2017) Pharmacological macrophage inhibition decreases adjuvant chemotherapy benefit following surgery for localized pancreatic
metastasis formation in a genetic model of pancreatic cancer. Gut 66: cancer. Ann Oncol 30: 1472 – 1478
1278 – 1285 Li B, Vincent A, Cates J, Brantley-Sieders DM, Polk DB, Young PP (2009) Low
Guillermet J, Saint-Laurent N, Rochaix P, Cuvillier O, Levade T, Schally AV, levels of tumor necrosis factor alpha increase tumor growth by inducing
Pradayrol L, Buscail L, Susini C, Bousquet C (2003) Somatostatin receptor an endothelial phenotype of monocytes recruited to the tumor site.
subtype 2 sensitizes human pancreatic cancer cells to death ligand- Cancer Res 69: 338 – 348
induced apoptosis. Proc Natl Acad Sci USA 100: 155 – 160 Lynch JT, Polanska UM, Hancox U, Delpuech O, Maynard J, Trigwell C,
Hingorani SR, Wang L, Multani AS, Combs C, Deramaudt TB, Hruban RH, Eberlein C, Lenaghan C, Polanski R, Avivar-Valderas A et al (2018)
Rustgi AK, Chang S, Tuveson DA (2005) Trp53R172H and KrasG12D Combined inhibition of PI3Kb and mTOR inhibits growth of PTEN-null
cooperate to promote chromosomal instability and widely metastatic tumors. Mol Cancer Ther 17: 2309 – 2319
pancreatic ductal adenocarcinoma in mice. Cancer Cell 7: 469 – 483 Mouliere F, Chandrananda D, Piskorz AM, Moore EK, Morris J, Ahlborn LB,
Hobbs GA, Baker NM, Miermont AM, Thurman RD, Pierobon M, Tran TH, Mair R, Goranova T, Marass F, Heider K et al (2018) Enhanced detection of
Anderson AO, Waters AM, Diehl JN, Papke B et al (2020) Atypical circulating tumor DNA by fragment size analysis. Sci Transl Med 10:
KRASG12R mutant is impaired in PI3K signaling and macropinocytosis in eaat4921
pancreatic cancer. Cancer Discov 10: 104 – 123 Mueller S, Engleitner T, Maresch R, Zukowska M, Lange S, Kaltenbacher T,

Höland K, Boller D, Hagel C, Dolski S, Treszl A, Pardo OE, Cwiek P, Salm F, Konukiewitz B, Öllinger R, Zwiebel M, Strong A et al (2018) Evolutionary
Leni Z, Shepherd PR et al (2014) Targeting class IA PI3K isoforms routes and KRAS dosage define pancreatic cancer phenotypes. Nature 554:
selectively impairs cell growth, survival, and migration in glioblastoma. 62 – 68
PLoS One 9: e94132 Neoptolemos JP, Kleeff J, Michl P, Costello E, Greenhalf W, Palmer DH (2018)
Hu H, Juvekar A, Lyssiotis C, Lien E, Albeck J, Oh D, Varma G, Hung Y, Ullas S, Therapeutic developments in pancreatic cancer: current and future
Lauring J et al (2016) Phosphoinositide 3-kinase regulates glycolysis perspectives. Nat Rev Gastroenterol Hepatol 15: 333 – 348

ª 2021 The Authors EMBO Molecular Medicine 13: e13502 | 2021 19 of 20

 17574684, 2021, 7, Downloaded from [Link] by Algeria Hinari NPL, Wiley Online Library on [11/12/2023]. See the Terms and Conditions ([Link] on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EMBO Molecular Medicine Benoit Thibault et al

Nomura S, Yoshitomi H, Takano S, Shida T, Kobayashi S, Ohtsuka M, Kimura oral pan-class I PI3K inhibitor, in patients with advanced solid tumors.
F, Shimizu H, Yoshidome H, Kato A et al (2008) FGF10/FGFR2 signal Clin Cancer Res 20: 233 – 245
induces cell migration and invasion in pancreatic cancer. Br J Cancer 99: Sivaram N, McLaughlin PA, Han HV, Petrenko O, Jiang Y-P, Ballou LM, Pham
305 – 313 K, Liu C, van der Velden AW, Lin RZ (2019) Tumor-intrinsic PIK3CA
Padoan A, Plebani M, Basso D (2019) Inflammation and pancreatic cancer: represses tumor immunogenecity in a model of pancreatic cancer. J Clin
focus on metabolism, cytokines, and immunity. Int J Mol Sci 20: 676 Invest 129: 3264 – 3276
Park S, Chapuis N, Bardet V, Tamburini J, Gallay N, Willems L, Knight Za, Therville N, Arcucci S, Vertut A, Ramos-Delgado F, Da Mota DF, Dufresne M,
Shokat Km, Azar N, Viguie F et al (2008) PI-103, a dual inhibitor of class IA Basset C, Guillermet-Guibert J (2019) Experimental pancreatic cancer
phosphatidylinositide 3-kinase and mTOR, has antileukemic activity in develops in soft pancreas: novel leads for an individualized diagnosis by
AML. Leukemia 22: 1698 – 1706 ultrafast elasticity imaging. Theranostics 9: 6369 – 6379
Pietrasz D, Pecuchet N, Garlan F, Didelot A, Dubreuil O, Doat S, Imbert- Thierry AR, Mouliere F, Gongora C, Ollier J, Robert B, Ychou M, Del Rio M,
Bismut F, Karoui M, Vaillant J-C, Taly V et al (2017) Plasma circulating Molina F (2010) Origin and quantification of circulating DNA in mice with
tumor DNA in pancreatic cancer patients is a prognostic marker. Clin human colorectal cancer xenografts. Nucleic Acids Res 38: 6159 – 6175
Cancer Res 23: 116 – 123 Thorpe LM, Yuzugullu H, Zhao JJ (2015) PI3K in cancer: divergent roles of
Pomel V, Klicic J, Covini D, Church DD, Shaw JP, Roulin K, Burgat- isoforms, modes of activation and therapeutic targeting. Nat Rev Cancer
Charvillon F, Valognes D, Camps M, Chabert C et al (2006) Furan- 15: 7 – 24
2-ylmethylene thiazolidinediones as novel, potent, and selective Torbett NE, Luna-Moran A, Knight ZA, Houk A, Moasser M, Weiss W, Shokat
inhibitors of phosphoinositide 3-kinase gamma. J Med Chem 49: KM, Stokoe D (2008) A chemical screen in diverse breast cancer cell lines
3857 – 3871 reveals genetic enhancers and suppressors of sensitivity to PI3K isoform-
Pons-Tostivint E, Thibault B, Guillermet-Guibert J (2017) Targeting PI3K selective inhibition. Biochem J 415: 97 – 110
signaling in combination cancer therapy. Trends Cancer 3: 454 – 469 Torres C, Mancinelli G, Cordoba-Chacon J, Viswakarma N, Castellanos K,
Puleo F, Nicolle R, Blum Y, Cros J, Marisa L, Demetter P, Quertinmont E, Grimaldo S, Kumar S, Principe D, Dorman MJ, McKinney R et al (2019) p110c
Svrcek M, Elarouci N, Iovanna J et al (2018) Stratification of pancreatic deficiency protects against pancreatic carcinogenesis yet predisposes to
ductal adenocarcinomas based on tumor and microenvironment features. diet-induced hepatotoxicity. Proc Natl Acad Sci USA 116: 14724 – 14733
Gastroenterology 155: 1999 – 2013.e3 Tosolini M, Algans C, Pont F, Ycart B, Fournie J-J (2016) Large-scale
Raynaud FI, Eccles SA, Patel S, Alix S, Box G, Chuckowree I, Folkes A, Gowan S, microarray profiling reveals four stages of immune escape in non-Hodgkin
De Haven Brandon A, Di Stefano F et al (2009) Biological properties of lymphomas. Oncoimmunology 5: e1188246
potent inhibitors of class I phosphatidylinositide 3-kinases: from PI-103 Vanhaesebroeck B, Ali K, Bilancio A, Geering B, Foukas LC (2005) Signalling by
through PI-540, PI-620 to the oral agent GDC-0941. Mol Cancer Ther 8: PI3K isoforms: insights from gene-targeted mice. Trends Biochem Sci 30:
1725 – 1738 194 – 204
Rhim A, Mirek E, Aiello N, Maitra A, Bailey J, McAllister F, Reichert M, Beatty Vanhaesebroeck B, Guillermet-Guibert J, Graupera M, Bilanges B (2010) The
G, Rustgi A, Vonderheide R et al (2012) EMT and dissemination precede emerging mechanisms of isoform-specific PI3K signalling. Nat Rev Mol Cell
pancreatic tumor formation. Cell 148: 349 – 361 Biol 11: 329 – 341
Roca H, Jones JD, Purica MC, Weidner S, Koh AJ, Kuo R, Wilkinson JE, Wang Y, Waddell N, Pajic M, Patch A-M, Chang DK, Kassahn KS, Bailey P, Johns AL,
Daignault-Newton S, Pienta KJ et al (2018) Apoptosis-induced CXCL5 Miller D, Nones K, Quek K et al (2015) Whole genomes redefine the
accelerates inflammation and growth of prostate tumor metastases in mutational landscape of pancreatic cancer. Nature 518: 495 – 501
bone. J Clin Invest 128: 248 – 266 Witkiewicz AK, McMillan EA, Balaji U, Baek GuemHee, Lin W-C, Mansour J,
Rodon J, Dienstmann R, Serra V, Tabernero J (2013) Development of PI3K Mollaee M, Wagner K-U, Koduru P, Yopp A et al (2015) Whole-exome
inhibitors: lessons learned from early clinical trials. Nat Rev Clin Oncol 10: sequencing of pancreatic cancer defines genetic diversity and therapeutic
143 – 153 targets. Nat Commun 6: 6744
Rodriguez-Viciana P, Warne PH, Khwaja A, Marte BM, Pappin D, Das P, Wu C-Y, Carpenter ES, Takeuchi KK, Halbrook CJ, Peverley LV, Bien H, Hall JC,
Waterfield MD, Ridley A, Downward J (1997) Role of phosphoinositide 3- DelGiorno KE, Pal D, Song Y et al (2014) PI3K regulation of RAC1 is
OH kinase in cell transformation and control of the actin cytoskeleton by required for KRAS-induced pancreatic tumorigenesis in mice.
Ras. Cell 89: 457 – 467 Gastroenterology 147: 1405 – 1416.e7
Sastra SA, Olive KP (2013) Quantification of murine pancreatic tumors by Zhang Y, Kwok-Shing Ng P, Kucherlapati M, Chen F, Liu Y, Tsang YH, de
high-resolution ultrasound. Methods Mol Biol 980: 249 – 266 Velasco G, Jeong KJ, Akbani R, Hadjipanayis A et al (2017) A pan-cancer
Schlieman MG, Fahy BN, Ramsamooj R, Beckett L, Bold RJ (2003) Incidence, proteogenomic atlas of PI3K/AKT/mTOR pathway alterations. Cancer Cell
mechanism and prognostic value of activated AKT in pancreas cancer. Br J 31: 820 – 832.e3
Cancer 89: 2110 – 2115
Schwarzenbach H, Hoon DSB, Pantel K (2011) Cell-free nucleic acids as License: This is an open access article under the
biomarkers in cancer patients. Nat Rev Cancer 11: 426 – 437 terms of the Creative Commons Attribution License,
~a I, Pandya SS,
Shapiro GI, Rodon J, Bedell C, Kwak EL, Baselga J, Bran which permits use, distribution and reproduction in
Scheffold C, Laird AD, Nguyen LT et al (2014) Phase I safety, any medium, provided the original work is properly
pharmacokinetic, and pharmacodynamic study of SAR245408 (XL147), an cited.

20 of 20 EMBO Molecular Medicine 13: e13502 | 2021 ª 2021 The Authors