Research Article

Oncogenic KRAS-Dependent Stromal

Interleukin-33 Directs the Pancreatic
Microenvironment to Promote Tumor Growth
Katelyn L. Donahue1, Hannah R. Watkoske2,3, Padma Kadiyala4, Wenting Du2, Kristee Brown2,
Michael K. Scales5, Ahmed M. Elhossiny6, Carlos E. Espinoza2, Emily L. Lasse Opsahl1, Brian D. Griffith2,
Yukang Wen2, Lei Sun2, Ashley Velez-Delgado5, Nur M. Renollet3, Jacqueline Morales5, Nicholas M. Nedzesky3,
Rachael K. Baliira1, Rosa E. Menjivar2,7, Paola I. Medina-Cabrera1, Arvind Rao6,8,9,10,11, Benjamin Allen5,8,

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

Jiaqi Shi8,12, Timothy L. Frankel2,8, Eileen S. Carpenter8,13, Filip Bednar1,2,8, Yaqing Zhang2,8, and
Marina Pasca di Magliano2,5,8
 ABSTRACT Pancreatic cancer is characterized by an extensive fibroinflammatory micro-
environment. During carcinogenesis, normal stromal cells are converted to
cytokine-high cancer-associated fibroblasts (CAF). The mechanisms underlying this conversion,
including the regulation and function of fibroblast-derived cytokines, are poorly understood. Thus,
efforts to therapeutically target CAFs have so far failed. Herein, we show that signals from epithelial
cells expressing oncogenic KRAS—a hallmark pancreatic cancer mutation—activate fibroblast auto-
crine signaling, which drives the expression of the cytokine IL33. Stromal IL33 expression remains high
and dependent on epithelial KRAS throughout carcinogenesis; in turn, environmental stress induces
interleukin-33 (IL33) secretion. Using compartment-specific IL33 knockout mice, we observed that
lack of stromal IL33 leads to profound reprogramming of multiple components of the pancreatic tumor
microenvironment, including CAFs, myeloid cells, and lymphocytes. Notably, loss of stromal IL33

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

leads to an increase in CD8+ T-cell infiltration and activation and, ultimately, reduced tumor growth.

Significance: This study provides new insights into the mechanisms underlying the programming
of CAFs and shows that during this process, expression of the cytokine IL33 is induced. CAF-derived
IL33 has pleiotropic effects on the tumor microenvironment, supporting its potential as a thera-
peutic target.

Introduction pancreas-specific expression of KRASG12D drives the transfor-

mation of healthy tissue into premalignant precursor lesions
The 5-year survival rate for pancreatic ductal adenocar-
known as pancreatic intraepithelial neoplasia (PanIN), which
cinoma (PDA) is currently 13%, making it one of the most
progress over time into invasive PDA (5, 6). Furthermore,
fatal human malignancies; to date, PDA is the third leading
proof of principle studies in mouse models have shown that
cause of cancer-related death in the United States (1). On-
inactivation of oncogenic KRAS induces tumor regression at
cogenic mutations in the small GTPase KRAS, most frequently
all stages of carcinogenesis (7, 8).
KRASG12D, occur in more than 95% of PDA cases (2–4). In mice,
Formation of PanIN and progression to PDA is paralleled
by the accumulation of a fibroinflammatory stroma that in-
1Cancer
cludes cellular components, such as fibroblasts and immune
Biology Program, University of Michigan, Ann Arbor, Michigan.
2Department of Surgery, University of Michigan, Ann Arbor, Michigan. cells (the latter mostly immunosuppressive in nature), and
3College of Literature, Science, and the Arts, University of Michigan, Ann abundant extracellular matrix (9, 10). We and others have
Arbor, Michigan. 4Immunology Graduate Program, University of Michigan, shown that oncogenic KRASG12D activity drives accumula-
Ann Arbor, Michigan. 5Department of Cell and Developmental Biology, tion of the stroma; in turn, inactivation of oncogenic KRAS
University of Michigan, Ann Arbor, Michigan. 6Department of Computa-
tional Medicine and Bioinformatics, University of Michigan, Ann Arbor,
reverses some, but not all, stromal changes (8, 11–16). Indeed,
Michigan. 7Cellular and Molecular Biology Program, University of Michigan, it is telling that an altered microenvironment has been linked
Ann Arbor, Michigan. 8Rogel Cancer Center, University of Michigan, Ann to resistance to oncogenic KRAS inhibition in mouse models
Arbor, Michigan. 9Cancer Data Science Resource, University of Michigan, and human patients (14, 17).
Ann Arbor, Michigan. 10Department of Radiation Oncology, University of
Fibroblasts are a prevalent component of the precursor le-
Michigan, Ann Arbor, Michigan. 11Department of Biostatistics, University
of Michigan, Ann Arbor, Michigan. 12Department of Pathology and Clinical sion microenvironment (PME) and the tumor microenviron-
Labs, University of Michigan, Ann Arbor, Michigan. 13Division of Gastroen- ment (TME), defined in these contexts as PanIN-associated
terology and Hepatology, Department of Internal Medicine, University of fibroblasts (PAF) and cancer-associated fibroblasts (CAF), re-
Michigan, Ann Arbor, Michigan. spectively. The healthy pancreas includes heterogenous pop-
Current address for W. Du: BeiGene Co., Ltd., Shanghai, China; current ulations of resident fibroblasts and fibroblast-like pancreatic
address for M.K. Scales, Department of Dermatology, Duke University
stellate cells (PSC); lineage tracing experiments completed
School of Medicine, Durham, North Carolina; and current address for R.E.
Menjivar, Department of Anatomy, University of California San Francisco, by our group and others have demonstrated that resident
San Francisco, California. fibroblasts and PSCs contribute to PAFs and CAFs (18, 19).
Corresponding Author: Marina Pasca di Magliano, Rogel Cancer Center,
*
Other sources of CAFs have been described, including meso-
University of Michigan, 1500 E. Medical Center Drive, Room 6306, Ann thelial cells, pericytes, endothelial cells, and adipocytes (20).
Arbor, MI 48109. Email: marinapa@umich.edu The overarching role of PAFs and CAFs remains poorly under-
Cancer Discov 2024;14:1964–89 stood, with evidence supporting pro- and anti-tumorigenic
doi: 10.1158/2159-8290.CD-24-0100 functions (for review see refs. 16, 21). Likewise, the mech-
This open access article is distributed under the Creative Commons Attribution- anism through which normal fibroblasts and other cells of
NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license. the healthy pancreas are converted to PAFs and CAFs also
©2024 The Authors; Published by the American Association for Cancer Research remains unclear.

October 2024 CANCER DISCOVERY | 1965

RESEARCH ARTICLE Donahue et al.

Furthermore, pancreatic CAFs are heterogenous—although Beyond this characterization of epithelial IL33 (34–39), the
their classification continues to evolve, a general framework role of stromal IL33 has been largely ignored, despite previ-
for CAF heterogeneity describes three groups: “myofibro- ous descriptions of IL33 in this compartment in mouse and
blastic CAFs” (myCAF), “inflammatory CAFs” (iCAF), and human PDA and in other pancreas diseases, such as pancre-
“antigen-presenting CAFs” (apCAF; refs. 22, 23). myCAFs atitis (12, 40, 41). Furthermore, the mechanisms underlying
exist in close proximity to tumor cells and are marked by high the activation of stromal IL33 during carcinogenesis remain
expression of α-smooth muscle actin (α-SMA); these cells are unknown.
major sources of ECM and ECM-remodeling proteins (22). Here, we show that pancreatic PAFs/CAFs are a dominant
Conversely, iCAFs express little-to-no α-SMA and instead source of IL33 in PanIN and PDA. We show that deletion of
produce and secrete a wide variety of signaling molecules, compartment-specific stromal IL33 in an orthotopic model of
including many immunoregulatory chemokines and growth PDA alleviates immunosuppression and suppresses tumor
factors (22). Of note, these classifications are largely based on growth. Additionally, we investigate the mechanisms of IL33
single-cell RNA sequencing (scRNA-seq) and are not fully re- induction and show that its expression by PAFs/CAFs is depen-
flected at the protein level (e.g., most pancreatic CAFs have dent on tumor and fibroblast-derived signaling factors. In sum-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

some level of α-SMA protein expression). Epithelial cells ex- mary, this work explores the role of stromal-derived IL33 in the
pressing oncogenic KRAS activate a secretory phenotype in PME and TME and provides new insight into the mechanisms
CAFs (12, 22); as a result, CAF-derived cytokines promote through which KRAS-mutant cells reprogram their surround-
immunosuppression and tumor progression (24, 25). The ings; at the same time, it highlights a new node of intercellular
mechanisms underlying CAF subtype polarization seem in- signaling between tumor cells, fibroblasts, and immune cells.
trinsic and extrinsic. For instance, a subset of CAFs expresses
the surface marker CD105—this population is stable and re- Results
tains its status in vitro and in vivo (26). Additionally, apCAFs,
a specialized population of CAFs expressing MHC-II com- IL33+ Stromal Cells are Abundant in Human and
ponents (23), distinctly originate from mesothelial cells (27), Mouse PDA
yet in the case of the myCAF and iCAF phenotypes, CAF polar- We first assessed the abundance and compartmentalization
ization is driven by extracellular signals, with TGFβ signaling of IL33 within the PDA microenvironment by performing
promoting the myCAF phenotype and NF-κB-JAK1/2-STAT3 IL33 IHC on surgically resected human tumors and matched
activation promoting the iCAF phenotype (22). Although adjacent normal regions (Fig. 1A). We observed robust IL33
the myCAF/iCAF distinction is commonly used in the field, staining in tumor cells and disease-associated stromal re-
ex vivo polarization studies highlight the plasticity of these gions, in contrast with low protein levels in the adjacent
two groups, suggesting that these classifications may be non- normal tissues. To interrogate which stromal cells expressed
binary and instead exist on a spectrum, similar to macro- IL33, we analyzed a human scRNA-seq dataset comprising
phage polarization (22, 28). 16 PDA patient samples and three adjacent normal samples
CAFs are the main source of tumor-promoting, immuno- previously created by our group (Fig. 1B; Supplementary
suppressive cytokines, such as IL6 and SAA3 (25, 29, 30); we Fig. S1A; ref. 42). Notably, overall IL33 transcript abundance
have recently shown that these cytokines are activated in PAFs was higher in the PDA dataset than in the adjacent normal
during the earliest stages of carcinogenesis through signals (Fig. 1C). Furthermore, we found IL33 expression in fibro-
from KRAS-mutant epithelial cells (12). Another cytokine ac- blasts (cluster 2), pericytes (cluster 3), and endothelial cells
tivated in PAFs at this juncture is IL33, an IL1 family member. (cluster 4) within the TME, in addition to expression by the
IL33 is an unusual cytokine—under homeostatic conditions, CK19+ (Cytokeratin-19, KRT19) ECAD+ (E-Cadherin, CDH1)
it is sequestered in the nucleus of cells expressing it, such as ductal/ductal-like malignant cells (cluster 1).
endothelial cells, barrier epithelial cells, and fibroblast-like Next, we examined the compartmentalization of IL33 ex-
cells. Upon tissue damage, IL33 is released into the extracel- pression across commonly used genetically engineered mouse
lular space (31, 32). Released IL33 binds to its receptor, ST2, models (GEMM) of PDA progression. We curated a murine
which may be expressed by a number of cell types, including scRNA-seq dataset of healthy pancreas, PanIN, and PDA tis-
mast cells, group 2 innate lymphoid cells (ILC2), CD4+ T sues to query Il33 gene expression (Fig. 1D; Supplementary
helper 2 cells (Th2), regulatory T cells (Tregs), NK cells, and Fig. S1B and S1C; refs. 23, 43–45). As in the human samples,
macrophages (31). IL33 has been linked to tumor promo- Il33 expression was present in CK19+ ECAD+ cells and stromal
tion and restriction across various solid tumor types (33). In cells, but it was most highly expressed in fibroblasts across
PDA, epithelial IL33 has been described in multiple studies, each tissue type, followed by mesothelial and endothelial cells
whereas stromal IL33 has remained unexplored. Epithelial (Fig. 1E). We then performed coimmunofluorescence (co-IF)
IL33 is activated downstream of oncogenic KRAS in prema- staining of healthy murine pancreata (WT), spontaneous
lignant (34) and tumor cells (35). Inhibition of epithelial IL33 PanIN (Ptf1a-Cre;LSL-KrasG12D, hereafter KC), and sponta-
slows PanIN progression (36), whereas administration of re- neous pancreatic tumors (Ptf1a-Cre;Trp53R172H/+;LSL-KrasG12D,
combinant IL33 promotes PanIN formation (34). In advanced hereafter KPC) with IL33, the epithelial marker ECAD, and
disease, the effect of IL33 has been studied with controversial an antibody that recognizes the C-terminal intracellular do-
results: administration of recombinant IL33 into orthotopic mains of PDGFRα and PDGFRβ. PDGFRα and PDGFRβ are
tumor-bearing mice has revealed an antitumor role for this expressed by fibroblasts (including PAFs/CAFs) and mesothe-
cytokine (37, 38); however, orthotopic injection of IL33 de- lial cells (Supplementary Fig. S1B), as well as by PSCs (41, 46),
ficient tumor cells also results in less tumor growth (35). whereas PDGFRβ is expressed by pericytes (47)—importantly,

1966 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

A B hAdj. Normal hPDA

n = 3; cells = 9,109 n = 16; cells = 49,018

10 1. CK19+ ECAD+
Patient #1 Patient #2 2. Fibroblast
3. Pericyte
4. Endothelial

UMAP_2
5. Myeloid
Adj. Normal

0
6. T Lymphocyte
7. NK Cell
8. B Lymphocyte
9. Acinar
−10 10. Endocrine
11. Platelet
12. RBC
IL33 50 µm IL33 13. Proliferating
−10 0 10 −10 0 10
UMAP_1
C hAdj. Normal hPDA
IL33

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

T S S
PDA

3 10 10

T 2

UMAP_2

UMAP_2
0 0
1
IL33 IL33
0 3 3
−10 −10
4 1 4 1
2 2

−10 0 10 −10 0 10
D mHealthy mPanIN mPDA
UMAP_1 UMAP_1
n = 3; cells = 6,859 n = 3; cells = 14,207 n = 3; cells = 18,834

10
E mPDA
5 Lymphocyte mPanIN
UMAP_2

Percent
mWT
0 Expressed
0 mPDA
−5 20 Myeloid mPanIN
40
−10 60 mWT
80 mPDA
−15
−10 0 10 −10 0 10 −10 0 10 Fibroblast mPanIN
Average
UMAP_1
Expression
1. CK19 ECAD
+ +
3. Tuft Cell 5. Mesothelial 7. Myeloid 9. RBC mWT
2. Acinar 4. Endothelial 6. Fibroblast 8. Lymphocyte 10. Proliferating 2
mPDA
1
F Mesothelial mPanIN
mHealthy (WT) mPanIN (KC) mPDA (KPC) 0
mWT
IL-33/PDGFRD/E/E-Cadherin

mPDA

Endothelial mPanIN
<0.0001
mWT
<0.0001 mPDA
IL-33 CTCF per PDGFRD/E+ ROI

Tuft Cell
<0.0001 mPanIN
60,000
50 µm A mPDA

40,000 Acinar mPanIN

mWT
20,000
mPDA

0 CK19+ ECAD+ mPanIN

3)
= )
KP n = )
7
55

mWT
18
38
KC = 2

(n
(n

IL-33 IL-33 IL-33 Il33

C
T
W

Figure 1. IL33+ stromal cells are abundant in human and mouse PDA. A, Human IHC staining of IL33 in matched adjacent normal (“Adj. Normal”) and
PDA regions. S, stromal area; T, tumor area. B, UMAP visualization of human scRNA-seq dataset split into adjacent normal and PDA groups. n = number of
patients in each dataset. C, Feature plot of IL33 transcription levels in human scRNA-seq. D, UMAP visualization of murine scRNA-seq dataset split into
healthy, PanIN, and PDA groups. E, Dot plot representation of Il33 transcription levels across cell types in the murine scRNA-seq dataset. F, Co-IF staining
of murine tissues [healthy (wildtype) aka WT, PanIN aka KC (Ptf1a-Cre; LSL-KrasG12D), and PDA aka KPC (Ptf1a-Cre; Trp53R172H/+;LSL-KrasG12D)]. IL33
(green), PDGFRα/β (red), E-Cadherin (white), DAPI (blue). IL33 CTCF was quantified per individual ROI; each ROI encompasses one PDGFRα/β+ cell. N = 3
mice were quantified per group. N in the figure represents the number of ROIs measured per group. P values represent one-way ANOVA testing between
groups. Line = mean CTCF.

October 2024 CANCER DISCOVERY | 1967

RESEARCH ARTICLE Donahue et al.

these cell types may all give rise to PAFs/CAFs in the pancre- orthotopic treatment schedule whereby we gavaged mice
atic disease context. We observed an abundance of IL33 in once a day for 5 days prior to surgery, but then let mice rest
ECAD− PDGFRα/β+ cells in the PanIN and PDA microenvi- for 2 days before tumor implantation, withholding tamoxifen
ronments, as well as occasional expression in the healthy pan- chow for the entirety of the experiment (Fig. 2G). We took this
creas (Fig. 1F). Taken together, our findings show that IL33 is approach to prevent systemic toxicity of prolonged tamoxifen
robustly expressed in transformed pancreatic tissues and that treatment and to assess the relative contribution of PDGFRA+
the stroma is a major source of IL33 in the TME. healthy cells in total CAF IL33 expression. The tumors from
this treatment model exhibited a comparable trend in tumor
Stromal IL33 Promotes PDA Growth size reduction between groups, as seen in our tamoxifen chow
Our observations revealed that the PAF/CAF compartment model (Supplementary Fig. S2C). Two experimental and two
is the highest expressor of stromal IL33 in pancreatic disease. control tumors (one male and one female per group) were
Therefore, to decipher the impact of stromal IL33 in PDA, we pooled and submitted for scRNA-seq. The resulting data re-
utilized the Pdgfra-CreERT2/+ mouse, which targets fibroblasts, vealed a complex TME that included tumor cells as well as
PSCs, and mesothelial cells (the cell types that ultimately fibroblasts, immune cells, and other stromal compartments

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

give rise to the majority of the PAF/CAF compartment; refs. (Fig. 2H; Supplementary Fig. S2D and S2E). Despite with-
18–20, 27). We crossed these mice with Il33f/f-eGFP mice to cre- holding tamoxifen chow, we saw a drastic reduction in Il33
ate Pdgfra-CreERT2/+;Il33f/f-eGFP GEMMs (CreER;Il33f/f) whereby transcription in the fibroblast cluster of the experimental
exons 5 to 7 of Il33 are excised upon administration of tamox- group, implying that these CAFs are mostly derived from
ifen (Fig. 2A). We induced recombination in these mice via cells expressing Pdgfra during homeostatic conditions (Sup-
tamoxifen by oral gavage once a day for 5 days prior to or- plementary Fig. S2F).
thotopic injection of syngeneic PDA cells (cell line 7940b; In addition, we observed a population of EMT-like cells ex-
ref. 48) and maintained mice on tamoxifen chow through- pressing the mesenchymal markers Pdgfra, Pdgfrb, Pdpn (Podo-
out the experiment (Fig. 2B). Tumors were harvested 3 weeks planin), and Col1a1 (Collagen Type 1 Alpha 1), relatively high
after implantation. Using this treatment strategy, we induced Krt19 compared with other stromal groups, and high EMT
targeted deletion of Il33 from PDGFRA+ cells prior to the es- markers such as Ero1l (endoplasmic reticulum oxidoreduc-
tablishment of disease and continued to induce recombination tase 1 alpha), Vegfa (vascular endothelial growth factor A),
throughout tumor growth, possibly targeting nascent CAFs Twist1 (twist family BHLH transcription factor 1), and Snai1
derived originally from PDGFRA− sources. PDGFRα+ cells (snail family transcriptional repressor 1), that simultane-
were sorted from the resulting tumors and evaluated for re- ously lacked expression of the mesothelial markers Lrrn4
combination efficiency via western blot; we found drastically (Leucine Rich Repeat Neuronal 4) and Msln (Mesothelin) and
reduced IL33 levels in the experimental groups using this the pericyte marker Cspg4 (Chondroitin Sulfate Proteoglycan 4;
method (Fig. 2C). Overall, tumors in CreER;Il33f/f mice were Supplementary Fig. S2D). Because our implanted tumor cells
∼40% smaller than controls, demonstrating that CAF IL33 are derived from a male mouse, yet we used male and female
has a protumorigenic role in PDA (Fig. 2D). host mice in our orthotopic transplantation study, we inves-
To characterize changes that might explain tumor size sup- tigated whether this EMT-like cell population was of host or
pression, we performed immunostaining for markers of cell tumor cell origin by querying expression of the X-inactivation
proliferation and cell death (Fig. 2E and F; Supplementary lncRNA Xist, which is highly expressed in the cells of females
Fig. S2A and S2B). For the former, tumor cells (E-Cadherin+), but not of males (49, 50). We saw that our tumor cell popu-
CAFs (PDGFRα/β+ intratumoral cells), and the proliferation lation had no meaningful Xist expression, but our EMT-like
marker Ki67 were measured by co-IF. We saw a clear reduction cells exhibited robust Xist expression, similar to that of other
in overall proliferation in the CreER;Il33f/f model (Supplemen- host populations such as ductal cells, mesothelial cells, fibro-
tary Fig. S2A), including a strong reduction in the fraction blasts, and pericytes (Supplementary Fig. S2G). This provided
of proliferating tumor cells and a trending decrease in pro- evidence that the EMT-like cells are a unique host-derived
liferating CAFs (Fig. 2E). We also observed increased levels Krt19high mesenchymal population, possibly arising in re-
of the cell death marker Cleaved Caspase 3 (CC3) in experi- sponse to tissue disruption in areas adjacent to invasive
mental tissues (Fig. 2F). Interestingly, although we observed disease. When we examined the expression of Il33 across all
an increase in overall CC3 in CreER;Il33f/f tumors, the relative cell populations in our scRNA-seq dataset, we detected Il33
fraction of CC3 colocalizing with E-Cadherin and PDGFRα/β in tumor cells and, to a greater extent, in the Pdgfra-expressing
were slightly reduced in the experimental context (69.5% vs. stromal cell types, that is, fibroblasts, mesothelial cells, and
60.1% and 5.2% vs. 4.5% of total CC3 staining, respectively; EMT-like cells; these stromal cells all displayed reduced Il33
Supplementary Fig. S2B). Consequently, these data show levels in the CreER;Il33f/f mice (further supporting the idea
an increase in CC3 in an E-Cadherin− PDGFRα/β− popula- that the EMT-like population originates from the host, as
tion from 25.3% of total CC3 staining in control tumors to the transplanted cancer cells carry neither the CreER nor the
35.4% in CreER;Il33f/f tumors, possibly indicating cell death Il33 floxed allele; Supplementary Fig. S2F). In contrast, we
in E-Cadherin− tumor cells or a PDGFRα/β− stromal com- observed no change in Il33 expression within the tumor cells
partment. Together these data link the decreased tumor size (Supplementary Fig. S2F).
to loss of proliferation and reduced cell survival. We next dissected changes in the tumor cell population in
To delineate the impact of CAF IL33 loss on the TME, we the CreER;Il33f/f mice. We first compared the tumor cell tran-
repeated our orthotopic experiment and harvested tissues scriptome to the established “classical” and “basal” gene sig-
for scRNA-seq. In this experiment, we adapted our initial natures; these molecular subtypes have been associated with

1968 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

A B OT
C CreER CreER;Il33f/f
#1 #2 #1 #2
Injection
Pdgfra +Tamoxifen +Tamoxifen IL-33 (light)
Cre ER
Il33 Il33 IL-33 (dark)
Exons 5-7 EGFP EGFP +Tamoxifen Chow Harvest
LoxP LoxP LoxP −7 days
−2 days 0 day PdgfrD
CreER 21 days
or
CreER;Il33f/f Vinculin
D E F
Ki67/PDGFRD/E/E-Cadherin CC3
Tumor Weight/Body Weight

0.0443
1.5

%Ki67+ of E-Cadherin+
60 0.0257

1.0
40

CreER
CreER
0.5
20 1.5

%CC3+ Area per FOV

0.0292
0.0
CreER CreER;Il33f/f 0 1.0
Ki67 CreER CreER;Il33f/f

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

0.0564 0.5

%Ki67+ of PDGFRD/E+
400 40 0.0790
Tumor Weight (mg)

CreER;Il33f/f
CreER;Il33f/f

300 30 0.0
CreER CreER;Il33f/f
200 20

100 10

0 50 µm 0 50 µm
Ki67 CreER CreER;Il33f/f
CreER CreER;Il33f/f
OT
G Injection H
+Tamoxifen CreER CreER;Il33f/f
n = 2; cells = 5,371 n = 2; cells = 4,816

Single Cell
−7 days Sequencing 19 19
CreER −2 days 0 days
21 days
or
CreER;Il33f/f 10 9 9

I CreER vs. CreER;Il33f/f − Tumor Cells

13 11 17
6 13 11 17
6
14 14
UMAP_2

PANCREAS_BETA_CELLS 2 2
MITOTIC_SPINDLE 0 4 4
16 10 16 10
EPITHELIAL_MESENCHYMAL_TRANSITION
15 15
APICAL_JUNCTION
ESTROGEN_RESPONSE_EARLY 12 1 7 12 1 7
MTORC1_SIGNALING
P53_PATHWAY
3 3
−10 18 18
APOPTOSIS
INTERFERON_GAMMA_RESPONSE 8 8
UV_RESPONSE_UP 21 21
HYPOXIA 5 5
CHOLESTEROL_HOMEOSTASIS
padj −10 −5 0 5 10 15 −10 −5 0 5 10 15
ADIPOGENESIS
XENOBIOTIC_METABOLISM 0.020
UMAP_1
FATTY_ACID_METABOLISM 1. Tumor 5. Endothelial 9. Macrophage 13. CD8+ T Cell 17. Mast Cell
0.015
ESTROGEN_RESPONSE_LATE
0.010 2. EMT-like 6. Mesothelial 10. Dendritic Cell 14. NK Cell 18. B Cell
INTERFERON_ALPHA_RESPONSE
TNFA_SIGNALING_VIA_NFKB 0.005 3. Acinar 7. Fibroblast 11. Monocytic Mac. 15. CD4+ T Cell 19. Plasma Cell
OXIDATIVE_PHOSPHORYLATION 4. Ductal 8. Pericyte 12. Granulocyte 16. ILC2 20. Proliferating
−2 −1 0 1 21. RBC
Normalized Enrichment Score

J Il1rl1 (ST2) K L
Left: Cre Right: Cre;Il33f/f OT
Tumor Weight/Body Weight

800 2.0
Expression Level

Injection 0.3575
0.0469
Tumor Weight (mg)

3 600 1.5
2
1 Harvest 400 1.0
0 Il1rl1+⁄+ 0 day 21days 200 0.5
or
Naive CD4

Exhausted CD4

Th1

Th17

Treg

ILC2

Mast Cell

Il1rl1−⁄−
0 0.0
Il1rl1+⁄+ Il1rl1−⁄− Il1rl1+⁄+ Il1rl1−⁄−

Figure 2. Stromal IL33 promotes PDA growth. A, Genetic scheme of Pdgfra-CreERT2/+;Il33f/f murine model. Tamoxifen induces activation of the Cre-
ERT2 fusion protein, allowing recombination to occur. B, Experimental design for the Pdgfra-CreERT2/+;Il33f/f orthotopic tumor model. OT, orthotopic,
CreER = Pdgfra-CreERT2/+, CreER;Il33f/f = Pdgfra-CreERT2/+;Il33f/f. C, Western blot of PDGFRα+ cells sorted from CreER and CreER;Il33f/f orthotopic
tumors. Two tumors/mice were pooled in each lane. D, Relative and absolute tumor sizes from CreER and CreER;Il33f/f orthotopic tumors. (E + F) Immu-
nostainings of CreER and CreER;Il33f/f tumors: E, = Co-IF staining of Ki67 (green), PDGFRα/β (red), E-Cadherin (white), and DAPI (blue), F, = IHC staining
of Cleaved Caspase-3 (CC3). In quantification, each dot represents one animal. G, Treatment schedule for the Pdgfra-CreERT2/+;Il33f/f orthotopic tumor
model adapted for scRNA-seq. H, UMAP visualization of orthotopic scRNA-seq dataset split into CreER and CreER;Il33f/f groups. I, Waterfall plot depict-
ing differential pathway enrichment in tumor cells based on the Hallmark collection of annotations. Positive normalized enrichment scores are enriched
in the control group. Pathways of interest are bolded. padj = Bonferroni-corrected P value. J, Violin plot depicting expression of Il1rl1 (ST2) in select
leukocytes from scRNA-seq. K, Experimental design for Il1rl1+/+ and Il1rl1−/− orthotopic tumor experiment. L, Relative and absolute tumor sizes from
Il1rl1+/+ and Il1rl1−/− orthotopic tumors. Tumor weight/body weight ratios are relative to the control group. Histogram data are mean ± standard deviation.
Experiments with two conditions were compared using a two-tailed Student t test.

October 2024 CANCER DISCOVERY | 1969

RESEARCH ARTICLE Donahue et al.

improved and worse survival in PDA, respectively (51). We Fig. S2D). To parse out which CD4+ T cells expressed ST2,
found that the tumor cells of CreER;Il33f/f mice had an enrich- we subclustered them and discovered that Tregs, but not
ment for the “classical” gene signature, whereas the basal sig- other helper T-cell populations present in the tumors, were
nature score was unchanged across the two groups (Supple- ST2-positive (Fig. 2J; Supplementary Fig. S3D). Interestingly,
mentary Fig. S2H). We then took an unbiased approach and IL33 activation of mast cells, ILC2s, and Tregs has been linked
assessed differential pathway activation between tumor cells to disease progression in solid tumors, including IL33-ILC2/
in the control and experimental animals using the Hallmark Treg activation in pancreatic cancer (35, 36, 39, 60). Given
collection of molecular signatures (Fig. 2I). In accordance with the presence of ST2+ immune cell populations in the tumors,
our staining result (Fig. 2E and F; Supplementary Fig. S2B), we performed an additional orthotopic PDA implantation ex-
we found enrichment for the “MITOTIC_SPINDLE” signature periment using Il1rl1+/+ and Il1rl1−/− mice (Fig. 2K). Tumors in
in the control group and “P53_PATHWAY” and “APOPTOSIS” Il1rl1−/− mice were ∼30% smaller than controls, further impli-
in the experimental group. We additionally detected enrich- cating IL33-ST2 signaling as protumorigenic in PDA (Fig. 2L).
ment for the “EPITHELIAL_MESENCHYMAL_TRANSITION” Overall, these data suggest that CAF IL33 supports PDA.
signature in the control tumor cells, indicating that tumor The loss of IL33 from PDGFRA+ stromal cells correlates with

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

cells in the CreER;Il33f/f TME have diminished EMT capabil- a decrease in tumor cell proliferation and an increase in tissue
ities. Notably, the experimental tumor cells also displayed apoptosis. Furthermore, the tumor-promoting effect of IL33
enrichment for the cellular stress signatures “HYPOXIA” and is at least in part mediated by secreted IL33 activating signal-
“UV_RESPONSE_UP.” Furthermore, CreER;Il33f/f tumor cells ing in ST2-expressing cells.
were enriched for multiple indicators of a proinflammatory
TME, including “TNFA_SIGNALING_VIA_NFKB,” “IN- Loss of Stromal IL33 Alters the ST2+ Immune
TERFERON_ALPHA_RESPONSE,” and “INTERFERON_ Cell Secretome, Resulting in a Shift in CAF
GAMMA_RESPONSE,” suggesting that the CreER;Il33f/f TME Differentiation
is more immune-permissive. We next investigated gene expression changes in ST2+ cell
We next explored the mechanism by which the loss of stro- populations (ILC2s, mast cells, and Tregs) to assess the ef-
mal IL33 induces a tumor-restrictive phenotype. Although fect of stromal IL33 loss (Fig. 3A). Mast cells in CreER;IL33f/f
canonically IL33 is thought to be constitutively maintained tumors expressed less Ccl3 (C–C motif chemokine ligand 3)
in the nucleus (31), it can also be secreted from living cells and Il6, whereas ILC2s showed a decrease in Csf2 (colony-
through stress-induced mechanisms (35, 41, 52–54). Of par- stimulating factor 2, GM-CSF); this reduction was expected,
ticular importance to this work, a recent study in pancreati- as these factors are linked to IL33 signaling in gastric cancer
tis demonstrated that PSCs can secrete IL33 in response to (Fig. 3A; ref. 60). IL33-ST2 signaling also induces ILC2s to
oxidative stress (41). To assess whether this was also true in produce IL4, IL5, and IL13 in PDA, as well as LIF (leukemia
our cancer model, we grew normal murine pancreatic fibro- inhibitory factor) in pancreatitis (35, 41). Although the expres-
blasts in vitro (cell line CD1WT; ref. 25) and either cultured sions of Il4, Il13, and Lif were decreased in mast cells and ILC2s
them in normal DMEM or polarized them to CAFs by pro- in CreER;Il33f/f tumors, the level of ILC2 Il5 was unexpectedly
viding murine PDA (cell line 9805; ref. 55) tumor conditioned increased (Fig. 3A). Furthermore, we detected upregulation
media (CM) for 24 hours. We then spiked in hydrogen per- of the EGFR ligand Areg (Amphiregulin) by ILC2s and Tregs
oxide (H2O2) to induce oxidative stress for an additional in CreER;Il33f/f tumors (Fig. 3A). This result was surprising as
24 hours, collected the resulting CAF CM, and profiled the IL33 positively regulates ILC2 Areg expression in pancreatitis
level of extracellular IL33 via ELISA (Supplementary Fig. S3A). and other instances of intestinal inflammation (41, 61), and
Fibroblasts grown in DMEM did not secrete IL33 (Supple- Il1rl1 expression positively correlates with an immunosup-
mentary Fig. S3B), fitting with the observation in vivo that nor- pressive AREG+ KLRG1+ Treg phenotype in a murine model
mal pancreatic fibroblasts express minimal IL33 at baseline of KRASG12D-driven lung cancer (62). Thus, these data possi-
(Fig. 1E and F). CAFs treated with PDA CM also showed lit- bly highlight tissue-specific functions for IL33. The increase
tle-to-no IL33 secretion (Supplementary Fig. S3B). However, in Areg was particularly interesting as AREG directly stimu-
CAFs treated with PDA CM + H2O2 did secrete IL33 into the lates myofibroblast activation and fibrosis in pancreatitis (41)
extracellular space (Supplementary Fig. S3B). Oxidative stress and has recently been shown to contribute to heterogeneity
is a well-characterized hallmark of the PDA TME (56–59); within the myCAF population (63). In our tumors, we found
it is therefore likely that this secretion mechanism also exists that ILC2s and Tregs were the highest expressors of Areg across
in vivo. In fact, when we compared enrichment of the Hall- all cell types in CreER;Il33f/f mice, whereas CAFs expressed the
mark reactive oxygen species pathway across cell types in our AREG receptor Egfr (Supplementary Fig. S4A). Given this up-
tumorigenesis scRNA-seq dataset, we detected increasing en- regulation of Areg in CreER;Il33f/f tumors, we next looked for
richment for oxidative stress during disease progression in changes in CAF activation and polarization. We performed
all cell populations but found that fibroblasts were unique immunostainings to check for differences in total CAFs
in displaying a strong enrichment for reactive oxygen species (intratumoral PDGFRα/β+ cells), activated myofibroblasts
activation even in healthy tissues (Supplementary Fig. S3C). (α-SMA and PDGFRα/β colocalization), and total collagen
The notion that IL33 is secreted from CAFs into the TME deposition (Gomori trichrome); we detected no substantial
prompted us to check for changes in the activation profiles change between the two groups based on these parameters
of cells expressing the IL33 receptor, ST2, in our CreER;Il33f/f (Supplementary Fig. S4B). We then assessed transcriptomic
tumors. Expression of ST2 (Il1rl1) was restricted to mast cells, changes within the fibroblast population using scRNA-seq
ILC2s, and a small subset of CD4+ T cells (Supplementary data (note this and subsequent downstream scRNA-seq

1970 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

A C Enriched Signaling in CreER Enriched Signaling in CreER;Il33f/f

Pericyte Endothelial

Try4/GM10334

b1
Macrophage Acinar

Lama4
Cd a s 3

Itg
G bs
Gdf10

fb
Cxcl1 Pos xb
CD8+ T

22 6

Thb tga9/
Th

P dg
Igf1

6
Treg CreER;Il33 b NK
f/f

d gf 9 Mast

2
P gals

I
CreER

s1
L

Tn
tn
Fibroblast
Mast Cell CreER;Il33
f/f

Ptn

p1
Sp
CreER Fibroblast 1
am
Nc
ILC2 CreER;Il33
f/f

CreER Tncfb1
Tg b
Il1 rres2
Ra a5a

Klrg1
Areg
Csf2
Ccl3
Il13
80 Il4
Il5

Il6
Lif
Sem
Entpd1
Percent
20
40
60

Fibroblast Thbs1
0

Expressed
receiving

Col1
Il1a/b
Average 2 1 0 Tnf

a1/a
Expression Macrophage

Th
B

bs
CreER vs. CreER;Il33f/f

1
Endothelial

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

Il1a/b
Fibroblasts

Il1a/b
Sell

EGF
Col4a1/a2
Inhba
ILC2 Monocytic Mac.

S p eg
gf

Areg
p1
APOPTOSIS

Ar
N 1r Tumor

Agrn
MYC_TARGETS_V1 1 Granulocyte
Ductal F Treg
GLYCOLYSIS EMT
MYOGENESIS
HYPOXIA D F CreERT2/+
TGF_BETA_SIGNALING 1 CreERT2/+;Il33f/f
CHOLESTEROL_HOMEOSTASIS 2 1. iCAF 4
HEDGEHOG_SIGNALING 2 2. iCAFhi, myCAFlo
MTORC1_SIGNALING 5 6 4
UMAP_2

P53_PATHWAY Il33
3. myCAF
TNFA_SIGNALING_VIA_NFKB
0
−1.5 −1 −0.5 0.0 4. myCAF , iCAFhi lo

Normalized Enrichment Score

5. apCAF

lo

lo

lo
−2
0.04
0.03
0.02
0.01

AF

AF

AF
AF

AF

AF
3

iC

yC

C
yC

pC
6. myCAFhi, apCAFlo

ap
,i
padj

m
hi
,m

,a
AF

hi
−8 −4 0 4

hi

AF
AF

yC
UMAP_1

yC
iC

m
E

m
3 4 5 7
Pdpn Clec3b Acta2 Saa3
Pan-Fibroblast

3 4 4 4
myCAF

apCAF
iCAF

Pdgfra Ly6c1 Col8a1 H2−Ab1

3 4 5 5
Pdgfrb Col14a1 Ccn2 Slpi
, i F lo
AF my F lo

ap F lo

F lo

AF my F lo

ap F lo

, i F lo
AF my F lo

ap F lo

, i F lo
AF y F lo

ap F lo
F

pC F

AF

pC F

AF

pC F

AF

yC m AF

pC F

AF
m h,i iCA

, a CA

m h,i iCA

, a CA

m h,i iCA

, a CA

, a CA
hi A

hi A

hi A

hi A

A
C

iC

C
AF yC
C

AF yC
C

AF yC
C

AF yC
C
i

m
yC m

yC m

yC m
,
hi

hi

hi

hi
hi
m ,
AF

AF

AF

AF
yC

yC

yC

yC
iC

iC

iC

iC
m

G H Enriched Signaling in CreER Enriched Signaling in CreER;Il33f/f

Pericyte Pericyte
100% Mesothelial Macrophage
23.5 13.2 Endothelial Fibroblast Macrophage Mesothelial Fibroblast
Acinar Endothelial Monocytic Mac.
iCAFhi Ductal Ductal Dendritic
13.5 Monocytic Mac.
26.7% EMT EMT
iCAFhi Dendritic
75% Granulocyte
% of Total Fibroblasts

35.9
59.4% 22.1 Tumor Granulocyte Tumor CD8+ T
CD8+ T NK
NK CD4+ T
Kng2 CD4+ T Treg
Gdf10 Treg Il1b Mast
50% 9.6 B
Clec2d ILC2 Cx3cl1 Plasma
myCAF hi
Vegfd Mast Tgfb1
31.4 Igf1 B Vegfa
20.0 63.1% 2 Plasma
Cxcl5/1stn Cxcl1
/2
myCAF hi Po nxb m1
25% T Nca
c
35.7%
4.2 Ptn Tn Th
bs
11.5 1
10.2 C3
apCAF
p1

0% apCAF 4.9
Sp
Co

Fibroblast Fibroblast
l1a

Mif
1/

CreER CreER;Il33f/f sending sending

a2

Figure 3. Loss of stromal IL33 alters the ST2+ immune cell secretome, resulting in a shift in CAF differentiation. A, Gene expression of activation
markers split by CreER and CreER;Il33f/f from scRNA-seq. B, Waterfall plot depicting differential pathway enrichment in fibroblasts based on the Hallmark
collection of annotations. Negative normalized enrichment scores are enriched in the experimental group. Pathways of interest are bolded. No genesets
were enriched in the control group with Bonferroni-corrected P value (padj) of < 0.05. C, Chord diagram visualizing differentially enriched (Bonferroni-
corrected P value < 0.05 and fold-change ≥0.25) predicted to interact with fibroblasts. Edge widths are proportional to predicted interaction strength.
D, UMAP visualization of fibroblasts from the CreER and CreER;Il33f/f scRNA-seq datasets. E, Gene expression of markers representing CAF subtypes.
F, Il33 expression in each CAF population split by experimental group. G, Histogram depicting the frequency of each CAF population across the CreER
and CreER;Il33f/f scRNA-seq datasets. H, Chord diagram visualizing differentially enriched (Bonferroni-corrected P value < 0.05 and fold-change ≥0.25)
fibroblast-derived ligands and their predicted interaction partners. Edge widths are proportional to predicted interaction strength.

October 2024 CANCER DISCOVERY | 1971

RESEARCH ARTICLE Donahue et al.

analyses (Fig. 3B–H) include only the “fibroblast” population changes in NF-κB signaling or TGFβ signaling between the
and not the “mesothelial” or “EMT-like” cells). Interestingly, the IL33 WT and KO cells, either at baseline or upon treatment
top enriched pathway in CreER;Il33f/f fibroblasts was “TNFA_ with PDA CM (Supplementary Fig. S4E). As these pathways
SIGNALING_VIA_NFKB,” which is downstream of AREG define myCAF/iCAF/apCAF differentiation in vivo, it seems un-
activation (Fig. 3B). Other pathways differentially activated likely that the nuclear activity of IL33 plays a role in regulating
in the IL33-deficient fibroblasts included “HEDGEHOG_ CAF polarization. A more likely explanation is that a change
SIGNALING” and “TGF_BETA_SIGNALING” (both linked in factors secreted by ST2-expressing immune cells affects
to myCAF differentiation; refs. 11, 22), as well as “P53_ fibroblast differentiation status. Importantly, we also mea-
PATHWAY,” “HYPOXIA,” and “APOPTOSIS” (the latter of sured proliferation in these cell lines and found no differ-
which complements our CC3 IHC staining; Fig. 2F). To un- ences between the IL33 WT or KO groups, under DMEM or
derstand the drivers of these changes in fibroblasts, we used PDA CM treatment (Supplementary Fig. S4F). These data
our scRNA-seq data to plot an unbiased differential predicted suggest that intrinsic IL33 is dispensable to CAF survival
interaction analysis between cells across the TME and fibro- and cell growth.
blasts in the control and CreER;Il33f/f context (Fig. 3C). Our As an important function of PDA CAFs is to activate protu-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

analysis highlighted increased signaling between ILC2- and morigenic intercellular signaling across the TME (12, 21, 22,
Treg-derived AREG and fibroblasts in CreER;Il33f/f tumors; at 24, 29), we next performed the scRNA-seq interaction tech-
the same time, it revealed strong interaction potential between nique to predict how fibroblast-derived signaling was altered
fibroblasts and myeloid-derived Il1a, Il1b, and Tnf (Tumor ne- in the context of IL33 loss (Fig. 3H). In control fibroblasts, a
crosis factor), as well as increases in autocrine fibroblast Il1b variety of ECM-remodeling proteins including Collagen Type
and Tgfb1 (transforming growth factor beta 1) signaling, pos- 1 (Col1a1 and Col1a2), the glycoprotein Tenascin XB (Tnxb),
sibly contributing to the NF-κB and TGFβ pathway activa- and the heparin-binding protein Pleiotrophin (Ptn) were dif-
tion seen in the pathway analysis (Fig. 3B). ferentially enriched compared with the experimental group.
Given that the NF-κB and TGFβ pathways work in opposi- Several fibroblast-immune interactions were also higher in
tion to regulate the iCAF/myCAF/apCAF phenotypes in PDA, control fibroblasts, with the strongest predicted interaction
we hypothesized that CreER;Il33f/f CAFs may exhibit a mod- stemming from Complement C3 (C3) signaling to macro-
ified phenotypic landscape. To investigate this, we subclus- phages, which has been shown to activate an immunosup-
tered fibroblasts and identified six unique groups, all present pressive, protumorigenic myeloid phenotype in cancer (68).
in control and experimental samples (Fig. 3D). Although We also saw expression of Cxcl12 (C–X–C motif chemokine
some of these clusters aligned similarly to the iCAF, myCAF, ligand 12), a cytokine linked to PDA cell immune evasion, in
and apCAF molecular signatures (Fig. 3E; refs. 22, 23) others control CAFs, which was lost in the CreER;Il33f/f model (24).
displayed hybrid expression of multiple markers (“iCAFhi, Conversely, upon fibroblast IL33 loss, we observed an increase
myCAFlo,” “myCAFhi, iCAFlo,” and “myCAFhi, apCAFlo”), con- in regulators of cell adhesion, plasticity, and vascularization,
sistent with the plastic nature of fibroblast populations. Nota- including Thbs1 (thrombospondin 1; ref. 69), Tgfb1 (70), Spp1
bly, each CAF cluster displayed some level of Il33 expression in (osteopontin; ref. 71), Tnc (tenascin-C; ref. 72), and Vegfa
control tumors, with higher expression in iCAF- and apCAF- (Fig. 3H; ref. 73). Furthermore, we observed an increase in
high cells than in myCAF-high cells; in CreER;Il33f/f fibroblasts, genes encoding for immune-regulatory proteins acting on
expression was diminished, as expected, across populations myeloid cells. These include the chemoattractants Cxcl1 and
(Fig. 3F). Looking broadly at these subtypes, a shift in CAF Cxcl2 (potentially recruiting granulocytes), as well as Cx3cl1
phenotype between control and experimental tumors emerged. (potentially recruiting macrophages). We also saw enrichment
Although fibroblasts in the control tumors mainly exhibited an for signaling between fibroblast Mif (macrophage migration
immunosuppressive secretory phenotype (iCAF/iCAFhi; 59.4% inhibitory factor) and macrophages and granulocytes; this
of total fibroblasts), CreER;Il33f/f tumors exhibited an opposite is notable, as Mif activates a proinflammatory phenotype in
profile, with 63.1% of fibroblasts displaying a myofibroblastic myeloid cells (74). To understand the causes of this shift in
phenotype (myCAF/myCAFhi; Fig. 3G). the fibroblast secretome, we compared the expression levels of
IL33 is a nuclear cytokine with a DNA binding motif; as the differentially expressed ligands across our CAF compart-
such, it can operate intracellularly to affect the transcrip- ments (Supplementary Fig. S4G). We found that the different
tome (33, 64, 65). This intrinsic function of IL33 is highly secretome signatures between control and CreER;Il33f/f fibro-
context-dependent (31, 66, 67), and whether it factors in blasts could be attributed to the shift from an immunosuppres-
pancreatic CAFs is unknown. To query the cell-autonomous sive secretory (iCAF/iCAFhi) phenotype to a myofibroblast-like
effect of IL33 as it may relate to CAF phenotype regulation, (myCAF/myCAFhi) phenotype, although we also observed tran-
we harvested pancreatic fibroblasts from a healthy Il33f/f-eGFP scriptional upregulation of Mif, Cxcl1, Cxcl2, and Vegfa within
mouse and clonally generated two IL33 KO cell lines (“KO #14” the apCAF-like population upon Il33 loss.
and “KO #16”) and two IL33 WT cell lines (“WT #1” and Altogether, our data show that loss of IL33 from PDGFRA+
“WT #20”; Supplementary Fig. S4C). We treated these fibro- stromal cells profoundly reprograms the TME, including
blasts with murine PDA CM to induce CAF polarization for changes in cytokine expression in ST2+ immune cells. In turn,
24 hours before harvesting for protein or RNA. As expected, we observed a shift in CAF populations, resulting in differences
CM treatment induced IL33 accumulation in wildtype pan- in their secretome. The result of this phenomenon is the loss of
creatic fibroblasts but not in IL33 KO cell lines, confirming expression of tumor-promoting and immunosuppressive cyto-
efficient gene inactivation (Supplementary Fig. S4D). We then kines such as Cxcl12, and an increase in cytokines that poten-
performed bulk RNA-seq. We observed no considerable tially support a proinflammatory immune composition.

1972 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

Inactivation of Stromal IL33 Enables Cytotoxic changes in CD8+ T-cell function (Fig. 4C). We found that the
T-cell Activity strongest predicted interactions in control tumors were be-
Tumor-infiltrating myeloid cells play a central role in the ini- tween CD8+ T cells and collagens originating from fibroblasts
tiation, progression, and maintenance of PDA through their and EMT-like cells; in KRASG12D-driven murine models of lung
secretion of tumor-promoting factors and their suppression cancer, collagen interactions directly induce CD8+ T-cell ex-
of CD8+ T cells (28). Given the changes in multiple immune- haustion and correlate with low CD8+ T-cell infiltration (77).
regulatory factors in CreER;Il33f/f tumors, we sought to de- Conversely, the strongest signal received by CD8+ T cells in
termine whether loss of PDGFRA+ stromal cell IL33 caused the CreER;Il33f/f model was MHC-I presentation from tumor
changes in myeloid populations within the TME. We first cells (Fig. 4C), suggesting a potential for increased tumor
stained tumors for macrophages (F4/80) and granulocytes cell killing and correlating with our increase in CC3 staining
(myeloperoxidase) to measure cell abundance (Fig. 4A; Sup- (Fig. 2F). We also observed multiple CD8+ T-cell recruiting fac-
plementary Fig. S5A). Although macrophage levels were tors enriched in CreER;Il33f/f tumors, including Cxcl9, Ccl7, and
Ccl2 from EMT-like cells, endothelial cells, mesothelial cells,
unchanged, granulocyte infiltration exhibited a dramatic in-
and macrophages (Fig. 4C). Beyond the interaction analysis, the
crease; the latter was predicted by the upregulation of fibro-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

overall gene expression of these chemokines also trended up-
blast Cxcl1 and Cxcl2 upon Il33 loss (Fig. 3H; Supplemen-
ward in other cell types in CreER;Il33f/f tumors, including Cxcl9
tary Fig. S4G). Notably, increased neutrophil infiltration via
in tumor cells, fibroblasts, monocytic macrophages, and den-
CXCL1-CXCR2 signaling is tied to CD8+ T-cell exclusion
dritic cells, Ccl7 and Ccl2 in acinar cells, and Ccl2 in fibroblasts,
and pancreatic tumor growth (75).
dendritic cells, granulocytes, and mast cells (Supplementary
We next looked for functional changes within myeloid pop-
Fig. S5D). Furthermore, CD4+ T cells also expressed the corre-
ulations using scRNA-seq data. In our dataset, we identified
sponding receptors for these chemokines, Cxcr3, Ccr2, Ccr5, and
traditional macrophages (Adgre1hi, Mrc1hi, C1qchi, Cd14lo, and
Ccr4, in the CreER;Il33f/f model (Supplementary Fig. S5D).
Ccr2lo) as well as a distinct population of monocytic macro-
To determine whether these transcriptional changes in-
phages (Adgre1lo, Mrc1lo, C1qclo, Cd14hi, and Ccr2hi; Fig. 2H;
deed affect T-cell function, we costained tumors for CD8 and
Supplementary Fig. S2D). We examined the macrophage pop-
the cytotoxicity marker Granzyme-B (Fig. 4D). We measured
ulations by comparing their transcriptomes to a curated list
an increase in the number of CD8+ and Granzyme-B+ CD8+
of proinflammatory or immunosuppressive markers (Fig. 4B).
cells, indicating a heightened cytotoxic T-cell response in tu-
Both cell types showed upregulation of multiple proinflam-
mors lacking stromal IL33. We then costained CD8 and Ki67
matory markers when stromal IL33 was deleted, including but
(to measure CD8+ T-cell proliferation) and CD44 (which is en-
not limited to Il1a, Tnf, Il6, Cxcl9, and MHC2 complex genes
riched on memory T cells; ref. 78). We detected a modest trend
H2-Ab1 and H2-Eb1. We also detected downregulation of im-
in CD8+ T-cell proliferation increasing in CreER;Il33f/f tumors
munosuppressive markers, including Chil3 (Chitinase-like 3,
as well as an increase in the expression of CD44 on CD8+
aka Ym1) and Mrc1 (CD206). Interestingly, Tgfb1 and Vegfa,
T cells, suggesting a potentially heightened memory pheno-
two immunosuppressive markers, were upregulated in the
type, although further markers are needed to evaluate the
CreER;Il33f/f group but only by monocytic macrophages. We
memory status of this cell population (Supplementary Fig.
then evaluated changes in granulocytes by performing differen-
S5E). We also stained for CD4 and the Treg marker Foxp3
tial expression analysis (Supplementary Fig. S5B). Although the
and found an increase in overall CD4+ T cells in CreER;Il33f/f
CreER;Il33f/f tumors had many more granulocytes than the
tumors (Fig. 4E), as predicted by our sequencing data (Sup-
control tumors, the gene expression between the two groups
plementary Fig. S5D). This included increases in Foxp3+ and
was mostly unchanged. However, we noted upregulation of
Foxp3− CD4+ T cells. Notably, the ratio of Foxp3+ to Foxp3−
Il1b in the CreER;Il33f/f-infiltrating granulocytes, whereas
cells remained unchanged between the CreER and CreER;Il33f/f
the control granulocytes expressed more Apoe (apolipopro-
groups, suggesting that the observed increase in Tregs was
tein E); both genes are linked to NF-κB activation and tumor
driven by an overall increase in CD4+ T-cell recruitment and
promotion in pancreatic cancer (44, 76). Furthermore, dif-
not by changes in Treg polarization.
ferential pathway analysis showed enrichment for indicators
In summary, we observed that removing stromal-derived
of activation in CreER;Il33f/f granulocytes, including “TNFA_
IL33 shifted the abundance and immunosuppressive poten-
SIGNALING_VIA_NFKB,” “ALLOGRAFT_REJECTION,”
tial of myeloid cells and that signaling changes across the
“INTERFERON_ALPHA_RESPONSE,” and “INTERFERON_
CreER;Il33f/f TME ultimately resulted in an increase in the
GAMMA_RESPONSE” (Supplementary Fig. S5C).
recruitment and activation of CD8+ T cells, along with an
Altogether, our data indicate multiple avenues of potential
increase in helper T cells.
direct and indirect CD8+ T-cell modulation. These include not
only supportive phenotypes such as increased proinflamma-
tory polarization in macrophages (Fig. 4B) and a reduced im- Expression of Fibroblast IL33 is Extrinsically
munosuppressive profile in CAFs (Fig. 3G), but also increased Induced by Epithelial KRASG12D and Requires
granulocyte recruitment and activation (Supplementary Fig. JAK1/2-STAT3 Activation Throughout
S5A–S5C), which is potentially CD8+ T-cell suppressive. Given Tumorigenesis
this multifaceted phenotype, we performed an unbiased dif- Given the pleiotropic effect of stromal IL33 in the TME, we
ferential enrichment analysis of all signals from cells across sought to better understand its mechanism of regulation in or-
the TME received by CD8+ T cells in control and CreER;Il33f/f der to identify potential therapeutic vulnerabilities. We thus uti-
tumors to gain a comprehensive understanding of possible lized the iKRASG12D model, whereby pancreatic epithelial cells

October 2024 CANCER DISCOVERY | 1973

RESEARCH ARTICLE Donahue et al.

A CreER CreER;Il33f/f
B
CreER;Il33f/f
Monocytic Mac.
CreER

CreER;Il33f/f
Macrophage
F4/80

CreER

Il1a
Il1b
Cxcl10
Cd80
Cd86

Tlr2
Il6
Cxcl9
H2−Ab1
H2−Eb1
Ccl5
Cxcl16
Tlr4
Tgfb1
Vegfa
Ccl24
Chil3
Arg1
Mrc1
Cd163
Tnf
Percent Average
50 µm Expressed Expression

−0.5
−1.0
1.0
0.5
0.0
25
50
75
Pro-Inflammatory Immunosuppressive
%F4/80+ Area per FOV

15
0.2166 C
10
Enriched Signaling in CreER Enriched Signaling in CreER;Il33
5 CD8+ T receiving CD8+ T receiving

0
CreER CreER;Il33f/f

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

Mast L Treg
ga
Lgalsls9 Pdcd1lg2
NK Fn1
9 EMT
MHC−I (H2-T23) Collagen (Col4a1/a2) Cd80
Nectin 2 MHC−I (H2-Q4/T23 Granulocyte
)
Pericyte Thbs32 F11r )
l1 2-K1
Cxc Ptn Lg
Pv als9 −I (H Tumor
MH MHC
C− r Ductal
I (H
2-T Monocytic Mac. bs1
22
Th

M
/T2
Fibroblast

HC
Endothelial 3) f
Tn 9

−I
Collagen (Col1a1 Macrophage Cxcl

(H
/a2)

En hbs l2

2-
tp 1
T Cc

D1
Sp d1
Th p1

/K
1

Cc 4/Gm1
Ccl7

1)
Pericyte

bs

Try
Lama4
Icam2
Cxcl9

MHC−
EMT

l2
Fibroblast

I (H2-Q
Mesothelial

033
Endothelial Acinar

4
6)
D E
CD8/Granzyme-B/E-Cadherin CD4/Foxp3
CreER

CreER

50 µm 25 µm 50 µm 25 µm
CreER;Il33f/f

CreER;Il33f/f

80 150 80
Total CD4+ Cells per FOV

%Foxp3+ of total CD4

4 0.0451
%CD8+ Area per FOV

0.3765
%GzmB+ of CD8+ Area

0.0475 0.0007
3 60 60
100
2 40 40
50
1 20 20
0 0 0 0
CreER CreER;Il33f/f CreER CreER;Il33f/f CreER CreER;Il33f/f CreER CreER;Il33f/f

Figure 4. Inactivation of stromal IL33 enables cytotoxic T-cell activity. A, IHC staining of F4/80 in CreER and CreER; Il33f/f tumors. B, scRNA-seq gene
expression of curated proinflammatory and immunosuppressive markers, grouped by cell type and split by experimental group. C, Chord diagram visualiz-
ing ligands differentially enriched (Bonferroni-corrected P value < 0.05 and fold-change ≥0.25) in CreER and CreER;Il33f/f tumors that interact with CD8+
T cells. Edge widths are proportional to predicted interaction strength. Chemokines are bolded. D and E, Co-IF staining of CreER and CreER;Il33f/f tumors:
(D) = CD8 (green), Granzyme-B (red), E-Cadherin (white) and DAPI (blue), (E) = CD4 (yellow), Foxp3 (magenta), and DAPI (cyan). For staining quantification,
each dot represents one animal, and values were compared using a two-tailed Student t test. Histogram data are mean ± standard deviation.

express KrasG12D in a doxycycline-dependent, inducible and re- although the mechanism by which this occurs remains un-
versible manner (referred to as “KRASG12D ON” and “KRASG12D known. As such, we first endeavored to understand whether
OFF,” respectively; Fig. 5A). Using the iKRASG12D model, we stromal IL33 expression remains dependent on oncogenic KRAS
have shown that expression of PAF Il33 is dependent on epithe- throughout tumorigenesis. We generated an “atlas” of pre-
lial oncogenic KRAS at the onset of tumor initiation (12), viously published and newly generated iKRASG12D scRNA-seq

1974 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

A rtTa
B
p48
Cre –Dox “KRASG12D OFF”
+cae
KRASG12D “3 or 5 weeks
Rosa26 TetO OFF

Early PanIN
STOP rtTa EGFP rtTa EGFP ON
LoxP LoxP LoxP +dox KRASG12D ON or +3 days OFF”
Dox “KRAS G12D
ON”
72 hours 48 3 or 5 weeks
+Dox rtTa iKRASG12D hours ON “3 or
KRASG12D 5 weeks ON”
TetO 3 days
C
iKrasG12D ON iKrasG12D OFF “14 weeks ON”

Late PanIN
10
“15 weeks ON
10 +dox KRASG12D ON ON OFF +1 week OFF”
iKRASG12D; 14 weeks
5 1 week 1 week
2 Trp53R172H/+
9
UMAP_2

6
0 OT injection
iKRASG12D;Trp53R172H/+
“OT PDA ON
4
OFF +3 days OFF”

PDA
−10 8 1 +dox KRAS ON
G12D or
7 11 3
6 FVBN 2 weeks
ON “OT PDA ON”

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

UMAP_1 −10 −5 0 5 10 −10 −5 0 5 10 −10 −5 0 5 3 days
1. CK19+ ECAD+ 7. NK Cell 3 week 5 week 14+ week OT PDA
2. Acinar 8. T Lymphocyte
3. Endothelial 9. ILC2
4. Mesothelial
5. Fibroblast
10. B Lymphocyte
11. RBC
E Fibroblast Il33
6. Myeloid 12. Proliferating 1.311659e–18 7.327054e–64 1.408223e–166 2.244752e–15

Expression Level
4
D 3
iKrasG12D ON iKrasG12D OFF 2
1
10 10 Il33 0
5 5
4 KrasG12D: ON OFF ON OFF ON OFF ON OFF
UMAP_2

UMAP_2

3
0 0 3 week 5 week 14+ week OT PDA
2
1
−10
4
−10
4
0
G +cae +/– JAK1/2i

−15 −10 −5 0 5 10 −15 −10 −5 0 5 10 +dox KRASG12D ON ON Harvest

72 48 3 weeks 3 days
UMAP_1 UMAP_1 iKRASG12D
hours hours

F Fibroblast ON vs OFF
H
HALLMARK “IL6_JAK_STAT3_SIGNALING” IL-33/PDGFRD/E/E-Cadherin

IL-33 CTCF per PDGFRD/E+ ROI

40,000 0.0002
3 week iKRASG12D 5 week iKRASG12D
FDR = 9.96e–6 FDR = 3.14e–2 30,000
0.6
+JAK1/2i
enrichment score

NES = 1.97 0.50

Control

NES = 1.58
enrichment score

0.4
20,000
0.25
0.2
0.00 10,000
0.0
50 µm
0 5,000 10,000 15,000 20,000
rank 0 5,000 10,000 15,000 20,000 0
rank

Control
(n = 305)

JAK1/2i
(n = 270)
14+ week iKRASG12D;Trp53R172H/+ OT PDA iKRASG12D;Trp53R172H/+ I
FDR = 2.92e–2 FDR = 6.51e–4 Pdgfra
Cre ER
+Tamoxifen
K Il33
enrichment score

0.50 NES = 1.56 0.50 NES = 1.75

enrichment score

Stat3 Stat3 2.0

0.0466
Relative mRNA

0.25 0.25 Exons 18-20

LoxP LoxP LoxP 1.5
0.00
0.00 1.0
0 5,000 10,000 15,000 20,000
rank
0 5,000 10,000 15,000 20,000
rank
J OT
Injection 0.5

+Tamoxifen
0.0
CAF
CreER;Stat3f/f
CreER

Sorting
CreER +
−7 days 0 days
or 21 days RNA
Extraction
CreER;Stat3f/f

Figure 5. Expression of fibroblast IL33 is extrinsically induced by epithelial KRASG12D and requires JAK1/2-STAT3 activation throughout tumorigenesis.
A, Genetic scheme of the iKRASG12D mouse. Doxycycline induces reversible expression of KRASG12D in pancreatic epithelial cells. B, Diagram representing
the various iKRASG12D treatment models and collection points across tumorigenesis. Cae, caerulein; OT, orthotopic. C, UMAP visualization of iKRASG12D
scRNA-seq dataset. Projection on the left is colored by cell type (all datasets merged). Projections on the right are split by iKRASG12D “ON” and “OFF” status
and are colored by timepoint. D, Feature plot representation of Il33 expression levels split by iKRASG12D “ON” and “OFF” status (all timepoints merged). E, Violin
plots depicting fibroblast Il33 expression level per timepoint and split by iKRASG12D “ON” and “OFF” status. Wilcoxon rank sum tests were performed between
iKRASG12D “ON” and “OFF” pairings per each timepoint, and Bonferroni adjusted P values are displayed above violins. F, GSEA enrichment plots of the Hallmark
“IL6_JAK_STAT3_SIGNALING” pathway based on fibroblast iKRASG12D “ON” and “OFF” differential gene expression analysis within each timepoint. G, Treatment
scheme for iKRASG12D “ON” model + JAK1/2 inhibitor. H, Co-IF staining of IL33 (green), PDGFRα/β (red), E-Cadherin (white), DAPI (blue). IL33 CTCF was
quantified per individual ROI; each ROI encompasses one PDGFRα/β+ cell. N = 3 mice were quantified per group. N in the figure represents the number of ROIs
measured per group. P values represent a two-tailed Student t test. Line = Mean CTCF. I, Genetic scheme of Pdgfra-CreERT2/+;Stat3f/f (CreER;Stat3f/f) murine
model. Tamoxifen induces activation of the Cre-ERT2 fusion protein, allowing recombination to occur. J, Diagram representing the treatment schedule for the
CreER;Stat3f/f orthotopic tumor model. K, Expression levels of Il33 in CAFs from J as measured by RT-qPCR. Values are normalized to Ppia (Cyclophilin A) and
relative to the CreER group. Two-tailed Student t test was performed to compare groups; data are mean ± standard deviation.

October 2024 CANCER DISCOVERY | 1975

RESEARCH ARTICLE Donahue et al.

data representing distinct stages of PDA development. This even in the presence of oncogenic KRAS. Next, to condition-
included two “Early PanIN” timepoints (3 and 5 weeks post- ally disrupt stromal JAK1/2-STAT3 signaling, we bred Pdgfra-
pancreatitis; refs. 7, 12, 79), a “Late PanIN” timepoint wherein CreERT2/+;Stat3f/f (CreER;Stat3f/f) mice, which lose exons 18
mice also have a full-body Trp53R172H/+ knock-in mutation to to 20 of Stat3 upon activation of recombination by tamoxi-
accelerate tumorigenesis (55), and a “PDA” model of synge- fen (Fig. 5I). We gavaged mice with tamoxifen once a day for
neic orthotopically injected iKRASG12D;Trp53R172H/+ tumor cells 5 days to suppress Stat3 in Pdgfra+ cells and then implanted
(Fig. 5B; refs. 55, 79). We also included matched “KRASG12D syngeneic tumor cells orthotopically to model mature PDA
OFF” groups for comparison at each timepoint (7, 12, 55). (Fig. 5J; Supplementary Fig. S6G). The resulting tumors
These datasets were batch-corrected and analyzed collectively. also displayed a notable decrease in growth (Supplementary
The resulting Uniform Manifold Approximation and Pro- Fig. S6H), in accordance with our CreER;Il33f/f and Il1rl1−/−
jection for Dimension Reduction (UMAP) visualization re- orthotopic models. We sorted PDGFRα+ cells from these
vealed a diverse cellular landscape including CK19+ ECAD+ tumors, extracted RNA, and assessed via RT-qPCR to reveal
cells (encompassing ductal/malignant-ductal cells), stromal a reduction in Il33 mRNA levels in the CreER;Stat3f/f model
cells including fibroblasts and mesothelial cells, and immune (Fig. 5K). Collectively, these findings suggest that KRASG12D-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

cells at each timepoint (Fig. 5C; Supplementary Fig. S6A dependent activation of JAK1/2-STAT3 in fibroblasts is re-
and S6B). Notably, the fibroblast compartment (cluster 5) re- quired to induce and maintain fibroblast IL33 expression
mained the highest expressor of Il33 across cell types in the throughout tumorigenesis.
“KRASG12D ON” and “KRASG12D OFF” contexts, followed by the
mesothelial cluster (cluster 4; Fig. 5D). When broken down Tumor Cell-Initiated Autocrine Signaling Drives
by timepoint, fibroblast Il33 expression was substantially re- IL33 Upregulation in Pancreatic Fibroblasts
duced in each model when KRASG12D was turned “OFF” Because KRASG12D expression in vivo inevitably alters multi-
(Fig. 5E). This pattern was not true for all genes enriched in ple compartments throughout the microenvironment, we
PME/TME fibroblasts: for example, Igf1 (insulin-like growth previously developed an in vitro system to determine whether
factor 1) and Timp2 (TIMP metallopeptidase inhibitor 2) were tumor cell signals directly activate fibroblast JAK1/2-STAT3. To
initially KRASG12D sensitive but became KRASG12D independent accomplish this, we utilized a previously established iKRASG12D;
as disease progressed (Supplementary Fig. S6C). Interestingly, Trp53R172H/+ tumor cell line (9805; ref. 55). Low-passage 9805
the expression of Il33 in mesothelial cells (cluster 4) did not cells (≤p17) maintain doxycycline dependence in vitro and con-
display meaningful dependence on KRASG12D (Fig. 5D; Sup- tinue to modulate “KRASG12D ON” and “KRASG12D OFF.” We
plementary Fig. S6D). Unfortunately, we did not detect an collected CM from “KRASG12D ON” and “KRASG12D OFF”
EMT-like cell population in this dataset comparable with 9805 cultures and provided it to normal pancreatic fibroblasts
the one found in our orthotopic CreER and CreER;Il33f/f tu- (CD1WT; Fig. 6A; ref. 25). We have demonstrated already that
mors, and as such, we were unable to draw any conclusions 24-hour treatment of CD1WT with “KRASG12D ON” CM, but
about the regulation of Il33 in this compartment. Presently, not “KRASG12D OFF” CM instigates Il33 transcriptional up-
our findings build off existing knowledge to show that fibro- regulation via a heat-liable signaling factor(s) and that this
blast Il33 upregulation remains dependent on the activity upregulation is blocked by concurrent treatment of CD1WT
of epithelial KRASG12D throughout PanIN development and with a JAK1/2i (12). We repeated this assay and also found that
in PDA, whereas mesothelial cells, despite upregulating Il33 24 hours of “KRASG12D ON” CM exposure caused accumula-
during pancreatic disease (Fig. 1E), may regulate Il33 through tion of IL33 protein and Il33 mRNA upregulation, whereas
a KRASG12D-independent mechanism. the “KRASG12D OFF” CM did not (Fig. 6B; Supplementary
We next investigated the mechanism by which the expres- Fig. S7A). We further validated this phenomenon using a phar-
sion of fibroblast IL33 is modulated in PanIN and PDA. Our macologic approach by treating 9805 cells with the KRASG12D
group has previously shown that Il33 upregulation in pancre- inhibitor MRTX1133 or the MEK1/2 inhibitor Trametinib,
atic fibroblasts in vitro requires JAK1/2-STAT3 signaling (12). generating CM, and providing it to CD1WT fibroblasts (Supple-
We therefore sought to assess the in vivo relationship between mentary Fig. S7B). As expected, MRTX1133- and Trametinib-
stromal IL33 and JAK1/2-STAT3 activation. First, we per- treated tumor CM resulted in reduced Il33 expression in
formed differential gene expression profiling between fibro- fibroblasts compared with fibroblasts treated with CM alone.
blasts in each KRASG12D ON/OFF pairing and found that the As a negative control, we treated tumor cells with the KRASG12C
Hallmark “IL6_JAK_STAT3_SIGNALING” signature was en- inhibitor Sotorasib, which had no effect. We also saw that
riched in the KRASG12D ON group at each timepoint (Fig. 5F). 24 hours of “KRASG12D ON” CM exposure induced pSTAT3
We then performed co-IF staining of our 3-week PanIN tis- activation, and that upregulation of IL33 and activation of
sues and saw decreases in pSTAT3 and IL33 protein levels pSTAT3 were blocked by concurrent treatment with a JAK1/2i
within PDGFRα/β+ cells at the 3-day “OFF” timepoint (Sup- in a dose-dependent manner (Fig. 6B; Supplementary Fig. S7A).
plementary Fig. S6E). Thus, high JAK1/2-STAT3 signaling The “KRASG12D OFF” CM failed to induce pSTAT3, indicat-
positively correlates with stromal IL33 mRNA and protein ex- ing that the JAK1/2-activating signaling factor(s) of interest
pression. Subsequently, we treated our 3-week PanIN-bearing coming from cancer cells is KRASG12D-dependent. Notably,
iKRASG12D mice with the JAK1/2 inhibitor ruxolitinib for 3 days “KRASG12D ON” CM also upregulated the iCAF marker Il6 and
prior to harvesting tissue (Fig. 5G). PDGFRα/β+ cells from downregulated the myCAF marker Acta2 (α-SMA), suggesting
the treated mice had decreased pSTAT3 and IL33 expression a phenotypic switch to a more iCAF-like polarization at this
(Fig. 5H; Supplementary Fig. S6F), consistent with the notion 24-hour interval (Supplementary Fig. S7A). Although the
that JAK1/2-STAT3 activity is required for IL33 expression expression of Il6 was also JAK1/2-dependent, Acta2 was not.

1976 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

A B

M
EM

C
+ ON CM

FF
iKRASG12D ON

M
“ON CM”

+O
+dox JAK1/2i (µmol/L) – – – 0.03 0.3 3
Harvest IL-33
CAFs
iKRASG12D
iKRASG12D; IL-33/D-Tubulin 1 1.5 9.2 6.4 2.2 1.1
Tumor Cells
Trp53R172H/+ iKRASG12D OFF “OFF CM” p-STAT3 (Y705)
–dox pSTAT3/STAT3 1 1.1 2.5 2.4 1.0 0.1
Harvest
STAT3
Fibroblasts CAFs
Healthy
D-Tubulin

C Il33 Il6 D
0.0006 0.0124 Acta2 rIL-6 (ng/mL)
+ON +ON rLIF (ng/mL)
<0.0001 0.0003
0.0002 DMEM CM 0.03 0.3 3 30 DMEM CM 0.06 0.6 6
5 0.0017 5 0.0024 2.0 0.0032 <0.0001
IL-33
Relative mRNA

Relative mRNA

Relative mRNA
4 4
1.5 IL-33/D-Tubulin 1 2.8 1.0 1.2 2.5 2.6 1 3.6 1.6 1.5 3.2
3 3
1.0 p-STAT3 (Y705)
2 0.9430 2 0.3300 pSTAT3/STAT3 1 1.6 1.0 1.3 1.7 3.2 1 2.4 1.6 1.8 3.0
0.6028

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

1 1 0.5
STAT3
0 0 0.0
+ON CM (hours) – – 24 20+4 +ON CM (hours) – – 24 20+4 +ON CM (hours) – – 24 20+4 D-Tubulin
+JAK1/2i (hours) – 4 – 4 +JAK1/2i (hours) – 4 – 4 +JAK1/2i (hours) – 4 – 4

Il6
E F Il33 <0.0001
<0.0001 Acta2
<0.0001 <0.0001
Log10(Relative mRNA)

+OFF CM (hours) – – – – – 24 <0.0001

Log10(Relative mRNA)

Log10(Relative mRNA)
<0.0001 <0.0001
100 10 10 <0.0001
+ON CM (hours) – 2 4 8 24 – 0.0005
<0.0001
IL-33
10 1
p-STAT3 (Y705) 1
STAT3 1 0.1
D-Tubulin
0.1 0.1 0.01
Hours: 0 1 2 4 8 16 24 48 8 48 Hours: 0 1 2 4 8 16 24 48 8 48 Hours: 0 1 2 4 8 16 24 48 8 48
+ON CM +OFF CM
pSTAT3 Relative Band Intensity

+ON CM +OFF CM +ON CM +OFF CM

IL-33 Relative Band Intensity

0.9496 0.8223

15 <0.0001
6 0.0113 G DMEM Group 1: Group 2:
0.1106
Collect Add Back Add Fresh Treatment
4 Fibro. CM Matching Fibro. CM + GolgiStop
10 0.0038
0.3296
+“ON CM” Harvest
0.0709
RNA
5 2 0.9860 +
0.9624
18 hours 6 hours Protein
+rLIF (6 ng/mL) Wash plates,
0 0 & Re-Treat:
Hours: 0 2 4 8 24 24 Hours: 0 2 4 8 24 24
+ON CM +OFF +ON CM +OFF
CM CM

H Lif I
0.0768
Il6
0.4021 Acta2
Il33 0.3782

+ON CM
DMEM
0.0072 0.0417 0.4945 +ON CM
<0.0001

+rLIF
0.8757
DMEM

0.0136 0.0333 0.0004 0.2773 <0.0001 0.3387 +rLIF

25 20 8 0.8430 0.7733 4 + GolgiStop
0.0002 0.0016 <0.0001 0.4804
IL-33
Relative mRNA

Relative mRNA
Relative mRNA
Relative mRNA

20 6 3
15
IL-33/D-Tubulin 1 9.3 11.8 0.4 1.6 1.7
15
10 4 2 p-STAT3 (Y705)
10 pSTAT3/STAT3 1 2.0 1.9 0.5 0.9 2.2
5 2 1
5 STAT3
0 0 0 0 D-Tubulin
+GolgiStop +GolgiStop +GolgiStop +GolgiStop
DMEM

+ON CM

+rLIF

DMEM
+ON CM
+rLIF
DMEM
+ON CM
+rLIF
DMEM

+rLIF
+ON CM

DMEM
+ON CM
+rLIF

DMEM

+ON CM

+rLIF
DMEM

+ON CM

+rLIF

DMEM

+ON CM

+rLIF

Figure 6. Tumor cell-initiated autocrine signaling drives IL33 upregulation in pancreatic fibroblasts. A, Ex vivo culture scheme for iKRASG12D; Trp53R172H/+
(cell line 9805) CM generation and healthy pancreatic fibroblasts (cell line CD1WT). B, Western blot of CD1WT whole cell lysates after 24 hours of treatment
with DMEM, iKRASG12D “OFF” CM, iKRASG12D “ON” CM, or concurrent iKRASG12D “ON” CM and JAK1/2i. C, RT-qPCR of CD1WT after treatment with DMEM,
JAK1/2i (4 hours, 0.3 μmol/L), iKRASG12D “ON” CM (24 hours), or pretreatment of iKRASG12D “ON” CM for 20 hours followed by spike-in of JAK1/2i (0.3 μmol/L)
for an additional 4 hours (24 hours total iKRASG12D “ON” CM treatment). Groups were compared with ordinary one-way ANOVA. D, Western blot of CD1WT
whole cell lysates after 24 hours of treatment with DMEM, iKRASG12D “ON” CM, rIL6 (left) or rLIF (right). E, Representative western blot of CD1WT whole
cell lysates after treatment with DMEM, iKRASG12D “OFF” CM, or iKRASG12D “ON” CM for increasing intervals of time. Densitometry quantification for IL33
normalized to loading control (α-tubulin) and pSTAT3 normalized to total STAT3 are shown. Quantification is relative to the 0-hour timepoint. Ordinary
one-way ANOVA was performed to compare each timepoint to the control. F, RT-qPCR of CD1WT after treatment with DMEM, iKRASG12D “OFF” CM, or
iKRASG12D “ON” CM for increasing intervals of time. Values are log10 transformed to better visualize large changes in gene expression level. Ordinary
one-way ANOVA was performed to compare each timepoint to the 0-hour timepoint. Only comparisons with P value < 0.05 are shown. G, Experimental
scheme to block autocrine signaling in CD1WT. CD1WT were treated with DMEM, iKRASG12D “ON” CM, or rLIF for 18 hours, and then, the resulting CM was
set aside. Cells were washed with PBS and then given back their original 18-hour CM or given GolgiStop (1.3 μL/2 mL) + fresh DMEM, iKRASG12D “ON” CM,
or rLIF media. Cells were incubated for an additional 6 hours before harvesting CD1WT RNA and protein. H, RT-qPCR of CD1WT after autocrine blocking
experiment. Two-tailed Student t test was performed to compare groups of interest (all tested comparisons shown). I, Western blot of CD1WT whole cell
lysates after autocrine blocking experiment. In all experiments with iKRASG12D CM, doxycycline is used as a vehicle control. In all experiments with JAK1/2i
(ruxolitinib), DMSO was used as a vehicle control. All replicates represent complete, independent experiments. RT-qPCR values are normalized to Ppia
(Cyclophilin A) and relative to the untreated DMEM group. Histogram data are mean ± standard deviation.

October 2024 CANCER DISCOVERY | 1977

RESEARCH ARTICLE Donahue et al.

We then queried whether JAK1/2 was also required to main- Interestingly, we found that TGFβ induced heightened expres-
tain IL33 expression after its initial induction and subsequent sion of the myCAF marker Acta2 (as predicted given the rela-
CAF polarization. We tested this by pretreating CD1WT with tionship between TGFβ and the myCAF phenotype; ref. 22) and
“KRASG12D ON” CM for 20 hours and then supplementing moderate upregulation of Il6 but suppressed Il33 expression.
cultures with JAK1/2i for an additional 4 hours (Fig. 6C). By Lif was also unaffected by rTGFβ at this timepoint. We further
RT-qPCR, we found that this acute spike-in of the JAK1/2i investigated the relationship between TGFβ and IL33 by com-
suppressed Il33 and Il6 expression, corroborating our in vivo bining the treatment of CD1WT with “KRASG12D ON” CM and
findings that JAK1/2 activation is continuously required to a TGFβ receptor inhibitor (Supplementary Fig. S7G). The ad-
maintain IL33 upregulation in PDA CAFs (Fig. 5H–K). Next, dition of the inhibitor did not impact the ability of “KRASG12D
we investigated the relationship between CAF polarization, ON” CM to upregulate IL33, despite dose-dependent loss of
JAK1/2 activation, and IL33 expression in a human model. We pSMAD2/3 in this assay. Therefore, our data indicate that
generated CM from primary patient-derived PDA organoids TGFβ is not involved in the upregulation of fibroblast IL33.
and provided it to primary fibroblasts collected from either In fact, Il33 expression may be antagonized by high TGFβ ac-
adjacent normal (normal-like fibroblasts) or tumor regions tivity, similar to the generation of the iCAF phenotype.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

(CAFs). PDA organoid CM induced upregulation of IL33, as Overall, our findings presented a conundrum: whereas
well as IL6 and CXCL1, in all three fibroblast lines (Supple- rIL1α, rTNFα, and rTGFβ all induce expression of Lif and/
mentary Fig. S7C). Furthermore, upregulation of IL33 and IL6 or Il6, but not Il33, LIF and IL6 induce Il33 expression. We
was suppressed in the presence of JAK1/2i. Thus, a cancer cell- considered that the timing of activation might explain this
derived factor(s) induces gene expression of IL33 and IL6 in a disconnect. We thus investigated the kinetics of STAT3 phos-
JAK1/2-dependent manner in human and mouse fibroblasts. phorylation and IL33 protein accumulation following the
To identify cancer cell-secreted proteins that are KRAS- “KRASG12D ON” CM treatment. Interestingly, we found that ac-
dependent, we performed a multiplex ELISA on “KRASG12D ON” tivation of pSTAT3 did not occur until between 2 and 4 hours
and “KRASG12D OFF” CM. We detected low amounts of IL1α, following CM treatment and continued to increase until
TNFα, and IL6, as well as abundant LIF, in the “KRASG12D ON” 24 hours (Fig. 6E). IL33 protein accumulation followed a com-
CM, all of which were reduced or undetectable in “KRASG12D parable pattern. We also assessed our fibroblasts at the RNA
OFF” CM (IL1β was also probed for, but not detected, in ei- level and found that Il33 mRNA gradually increased from
ther condition; Supplementary Fig. S7D). IL1α and TNFα ac- 2 hours to peak at ∼16 hours post-“KRASG12D ON” CM expo-
tivate NF-κB: in pancreatic fibroblasts, it has been shown that sure (Fig. 6F). Of note, we also observed sustained increases
NF-κB activation induces the production and secretion of IL6 in Il6 and decreases in Acta2 at the 16 hours timepoint,
and LIF, which in turn autocrine activate JAK1/2 signaling suggesting a major phenotypic CAF switch at this juncture
and induce an iCAF phenotype (22). Therefore, we sought to (Fig. 6F). As treatment with “KRASG12D ON” CM did not
determine whether any of the detected KRASG12D-dependent immediately initiate STAT3 phosphorylation—despite CM
cancer cell-derived factors were sufficient to induce IL33 ex- containing some IL6 and LIF, which are JAK1/2 activating
pression. For this purpose, we treated CD1WT fibroblasts ligands—our data support a model by which tumor cells may
with recombinant IL6 (rIL6) and rLIF (activating JAK1/2) secrete an insufficient concentration of ligands or lack secre-
or rIL1α and rTNFα (activating NF-κB). Within 24 hours of tion of other factors required to activate JAK1/2 signaling
treatment with rIL6 and rLIF, both cytokines were able to acti- in fibroblasts. We then tested whether a secondary message
vate STAT3 phosphorylation and IL33 expression to levels that by fibroblasts was necessary to induce JAK1/2 signaling via
mirrored the “KRASG12D ON” CM treatment group (Fig. 6D). a feedforward loop. We designed an in vitro assay in which
We then treated CD1WTs with rIL1α or rTNFα for 1 and we treated CD1WT with DMEM alone, “KRASG12D ON” CM,
24 hours and collected RNA. However, neither cytokine acti- or rLIF for 18 hours to initiate IL33 upregulation, pSTAT3
vated Il33 expression; rather, expression of Il33 was reduced activation, and CAF polarization. We collected the resulting
below baseline with each (Supplementary Fig. S7E). Interest- CD1WT CM, washed cells with PBS, and treated them for
ingly, 1 hour of either rIL1α or rTNFα upregulated Il6 and an additional 6 hours with either (i) their matching CD1WT
Lif in CD1WTs, but only rIL1α-induced Il6 remained high at CM (i.e., giving the CM back to the originating cells) or
the 24-hour timepoint, whereas Lif expression was either lost (ii) a Golgi blocking agent (GolgiStop) plus fresh treatment
or suppressed below baseline expression at that interval (Sup- (DMEM, “KRASG12D ON” CM, or rLIF) matched to the orig-
plementary Fig. S7E). rIL1α and rTNFα also suppressed Acta2 inal treatment condition (Fig. 6G). This design resulted in
at 24 hours (Supplementary Fig. S7E), suggesting that each fibroblasts receiving either (i) up to 24 hours of potential
factor was individually sufficient to push the fibroblasts to a autocrine signaling factors or (ii) withdrawal of any auto-
“less myCAF-like” state—similar to treatment with full tumor crine signaling factors for the last 6 hours of treatment. We
cell CM—despite not upregulating Il33. then collected RNA and protein from these fibroblasts. Our
It has also been reported that TGFβ signaling can induce results showed that “KRASG12D ON” CM-dependent Il33 up-
Il6 expression in fibroblasts (80) and PSCs (81) and Lif expres- regulation was lost when autocrine signaling was blocked
sion in other cell types (82, 83); furthermore, this signaling (Fig. 6H and I). Importantly, the Golgi blocking agent also
directly induces upregulation of IL33 in tumor-initiating cells prevented the “KRASG12D ON” CM from activating pSTAT3,
(TIC) in squamous cell carcinoma (SCC; ref. 54). To investigate indicating that tumor-derived factors alone are indeed not
whether TGFβ could also lead to Il33 upregulation in pancre- sufficient to activate this pathway (Fig. 6I). Furthermore, rLIF
atic fibroblasts, we treated CD1WT for 24 hours with rTGFβ did not rescue IL33 upregulation when autocrine signaling was
and harvested the resulting RNA (Supplementary Fig. S7F). blocked, despite successfully activating pSTAT3 (Fig. 6H and I).

1978 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

Interestingly, upregulation of Il6 was not affected by treat- Tumor Cell

ment with the Golgi blocker with or without CM; in con-
trast, the Golgi blocker reversed the downregulation of Lif
previously seen after 24 hours of “KRASG12D ON” CM (Fig. 6H; KrasG12D
Supplementary Fig. S7E). The Golgi blocker also prevented TNF-D
the downregulation of Acta2 in response to rLIF treatment IL-1D
(Fig. 6H). 3
pSTAT3
F
Taken together, our data demonstrate that IL33 upregu- Fibroblast 6/LI
lation in PAFs/CAFs is dependent on at least two autocrine IL- IL-33
signaling events. In the first loop, fibroblast pSTAT3 activa-
Fibroblast IL-33 Fibroblast IL-33
tion is dependent on tumor cell-initiated fibroblast autocrine
signaling (Fig. 6I). In the second loop, pSTAT3 activation
alone (without tumor CM) is sufficient to drive IL33 upreg- Treg
Treg ILC2
ulation in healthy pancreatic fibroblasts (Fig. 6D) but only in

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

the context of active fibroblast autocrine signaling (Fig. 6I). ILC2

Additionally, it is possible that this second loop is dependent

on pSTAT3, despite being agnostic of tumor CM treatment CAF
(Fig. 6I). Overall, our observations shed light on a complex
CAF
signaling network involving cancer cell-derived signaling fac- Myeloid
Myeloid
tors and autocrine activation that evolves over time as healthy
fibroblasts are reprogrammed to CAFs. These data corrobo- Granzyme-B
rate the preexisting model whereby pancreatic tumor cell CM CD8 T+

promotes the iCAF phenotype and suppresses the myCAF CD8+ T

phenotype in culture (22); at the same time, they demonstrate MHC-1
that IL33 expression is uniquely regulated compared with the
Tumor Tumor
PDA-promoting iCAF cytokines IL6 and LIF (25, 84).

Figure 7. Tumor cell KRASG12D initiates upregulation of fibroblast IL33,

Discussion promoting immunosuppression in PDA. Working model. During PanIN and
Pancreatic cancer is characterized by a complex network of PDA, KRASG12D-dependent tumor cell-derived signaling factors initiate
fibroblast autocrine signaling, including the JAK1/2-STAT3 pathway.
cancer, stromal, and immune cell crosstalk (85) that requires This fibroblast reprogramming results in the upregulation of IL33.
further unraveling to improve therapeutic interventions. CAFs Furthermore, at least one additional autocrine loop is required for IL33
are an abundant population within the stroma, but their upregulation (possibly also dependent on pSTAT3). CAF IL33 is secreted
in response to oxidative stress, in which it signals to ST2+ immune cells
role during cancer progression, whether pro- or anti-tumor, ILC2s and Tregs, promoting an immunosuppressive TME and tumor
remains controversial (86). This lack of understanding likely growth. When stromal IL33 is removed, ILC2s and Tregs exhibit an altered
stems from the complex biologic functions of CAFs: beyond secretory gene signature, and a shift in CAF and myeloid cell polarization
providing structural support, they secrete extracellular ma- is seen. This ultimately results in the recruitment and activation of CD8+ T
cells, and the suppression of tumor growth.
trix and serve as a signaling hub, secreting multiple cytokines
and growth factors (12, 15, 21). The function of specific CAF-
derived cytokines and mechanisms through which epithelial have been identified in multiple human and murine tumors
neoplastic cells drive cytokine expression programs in CAFs including breast cancer; these cells activate an antitumoral in-
remain unclear. One of the cytokines activated in pancreatic fi- flammatory response upon IL33 stimulation (bioRxiv 2023.
broblasts in response to epithelial oncogenic KRAS is IL33 (12). 02.14.528486). In pancreatic cancer, the role of IL33 may vary
Although IL33 expression in cancer-associated stromal cells has depending on the stage of the disease. Injection of recombi-
been previously noted (40), its function and regulation have re- nant IL33 into mice bearing KRAS–mutant pancreatic epithe-
mained unexplored. lia accelerates PanIN formation, and inactivation of IL33 in
Here, we demonstrate that stromal IL33 is activated early KRASG12D-expressing epithelial cells hinders it, despite robust
during carcinogenesis and that PAF/CAF IL33 remains depen- expression of IL33 in other stromal cell populations (12, 34, 36).
dent on epithelial KRAS at all stages of cancer progression, In invasive diseases, the relationship between IL33 and ST2+
from low-grade PanIN to malignant disease. Furthermore, we ILC2s has been increasingly explored; the downstream ef-
show that stromal IL33 supports tumor progression (Fig. 7). fect is controversial, with studies supporting pro- and anti-
Our study complements previous literature addressing the tumoral effects (35, 37). Concurrently, IL33 to ST2+ Treg
role of epithelial IL33 in cancer. IL33 expression is elevated in signaling promotes CD8+ T-cell exclusion and increases tu-
many solid tumor types; its regulation is cell type-dependent, mor growth (39).
whereas its functional role is largely determined by the nature Importantly, in addition to differences in the function of
of the cells within the tumor expressing the IL33 receptor, ST2 IL33 during PanIN and PDA, there are likely species-dependent
(33, 87). In a murine model of SCC, TICs upregulate IL33 in differences in the regulation and role of epithelial and stro-
response to TGFβ signaling; in turn, secreted IL33 drives a mal IL33 that should be emphasized. Outside of the disease
TIC-macrophage positive feedback loop that promotes cancer context, homeostatic IL33 compartmentalization can differ
invasion and drug resistance (54). Conversely, ST2+ NK cells between humans and mice (e.g., although nearly all human

October 2024 CANCER DISCOVERY | 1979

RESEARCH ARTICLE Donahue et al.

endothelial cells express IL33, only a subset of murine endo- found a potential immune cell-fibroblast paracrine signaling
thelial cells show IL33 expression; ref. 31). This observation axis. Beyond its effect on fibroblasts, AREG might act on the
may also be reflected in our scRNA-seq data: although murine pancreatic epithelium to reinforce the KRAS-MAPK signaling
endothelial cells have little-to-no Il33, our human endothelial axis, which is important in the onset of pancreatic carcino-
cells have substantial IL33 expression in adjacent normal and genesis even in the presence of oncogenic KRAS (88, 89). CAFs
PDA tissues, suggesting that IL33 expression in this compart- in CreER;Il33f/f tumors also displayed increased activation of
ment may be disease agnostic (it should be noted, however, Hedgehog and TGFβ signaling, both of which are linked to
that some level of dysplasia is expected in adjacent normal myCAF differentiation (11, 22). In addition to an increase
tissues due to their close localization with the tumor, making in Areg, stromal IL33 deletion results in increased NF-κB li-
them a limited control for such a comparison). Although we gands, namely Il1a, Il1b, and Tnf from myeloid cells and fibro-
found that the fibroblast compartment was the highest ex- blasts, as well as increased fibroblast Tgfb1, all of which may
pressor of IL33 across multiple GEMMs of PanIN and PDA, act on the CAF compartment. The overarching result of these
with minimal expression of epithelial IL33 by RNA and pro- pleiotropic changes was a shift in fibroblasts toward a myofi-
tein staining, our human PDA data show robust expression broblastic phenotype and a reduction in immunosuppressive

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

of IL33 in tumor and stromal cells, suggesting that tumor cell cytokine production; these findings support previous litera-
IL33 may play a larger role in human disease than in our mu- ture showing that the myofibroblast-activating TGFβ path-
rine models. way overrides NF-κB activation and the subsequent iCAF
Indeed, it is further possible that the type of cell expressing phenotype when the pathways are in competition (22). It
IL33 plays an important role in its impact on the TME and should be emphasized that despite this transcriptomic analy-
tumor growth. Nuclear IL33 may have a unique cell-intrinsic sis, we did not detect a change in the abundance of the myo-
transcriptomic effect that promotes or hinders tumor pro- fibroblast marker α-SMA by immunostaining in our tissues.
gression. Another consideration is the regulation of IL33 se- Indeed, our data depict a nuanced molecular landscape in
cretion; regarding the relative contribution of each cell type’s which there is much overlap between secretory and myofibro-
IL33 in the extracellular space, it is possible that different blastic CAFs in vivo [to this point, of the six CAF subtypes we
extracellular cues are necessary to stimulate active IL33 secre- detected, five display moderate-to-high Acta2 (α-SMA) gene
tion from each population, and that some cues are more com- expression]. Thus, our data highlight the need to continually
mon than others. Here, we show that tumor CM-induced IL33 address the concept of CAF heterogeneity in the context of
upregulation does not automatically instigate IL33 secretion cell plasticity.
from pancreatic fibroblasts; instead, our work corroborates a Interestingly, nuclear IL33 has also been implicated in the
mechanism by which oxidative stress causes fibroblast IL33 regulation of NF-κB and TGFβ signaling (64, 65), suggesting
secretion (41)—a relationship that has also been shown in that there may be cell-intrinsic effects of IL33 loss driving
TICs (54). Other stress stimuli that have been shown to cause CAF repolarization, in addition to cues from the microenvi-
IL33 release in epithelial cells include fungus (35, 52) and extra- ronment. However, our in vitro assessment of IL33 WT and
cellular ATP (53). Overall, the cell-intrinsic and extracellular IL33 KO fibroblasts with and without tumor CM treatment
impacts of IL33 warrant further investigation about cells of showed no change in NF-κB or TGFβ signaling, fibroblast pro-
origin, with emphasis on the interspecies differences between liferation, or fibroblast survival, thus providing no evidence of
human and murine biology, in order to fully understand the a cell-intrinsic effect.
scope of IL33 function in PanIN and PDA and apply these Ablation of stromal IL33 caused loss of cancer cell pro-
findings therapeutically. liferation, increased apoptosis, prompted a transcriptional
To focus on the role of stromal IL33 in PDA, we deleted IL33 shift toward the more differentiated “Classical” molecular
from PDGFRα+ cells and found a suppression of tumor subtyping, and resulted in an overall reduction in tumor size
growth, despite intact IL33 expression in cancer cells. We while profoundly altering the microenvironment. We observed
detected ST2 expression in ILC2s, mast cells, and Tregs, and a similar reduction in tumor size when syngeneic orthotopic
found fewer Il4+ and Il13+ ILC2s and mast cells in CreER;Il33f/f tumors were implanted in mice lacking the IL33 receptor, ST2.
tumors, supporting previous observations whereby IL33 in- This indicates that the effects of IL33 are largely dependent on
duces expression of these cytokines (37, 41). We observed up- its secreted form, rather than the DNA binding, intracellular
regulation of AREG—an EGFR ligand—by Tregs and ILC2s functionality of IL33. Notably, the effects we observed in the
upon inactivation of IL33. This was unexpected, given that CreER;Il33f/f model occur in the presence of epithelial IL33,
IL33 stimulation elicits AREG upregulation by ILC2s in the supporting the notion that the stroma is the main source of
gut and during pancreatitis (41, 61), and again, indicates tis- secreted IL33 in this system. We observed direct and indirect
sue-dependent mechanisms of action. effects of IL33 on the TME; ST2+ cells such as ILC2s, Tregs and
Thus, the loss of stromal IL33 induced elevated expres- mast cells had altered cytokine production upon fibroblast
sion of AREG, which was in turn predicted to act on CAFs, IL33 loss. At the same time, in macrophages, we observed a
which express a high degree of EGFR. Consistent with this loss of immunosuppressive markers such as Arg1 (Arginase-1)
notion, CAFs displayed elevated NF-κB pathway enrichment, and Mrc1 (CD206) in addition to an increase in proinflam-
a canonical downstream mediator of AREG-EGFR signaling. matory genes including Cd80, Cd86, Tnf, Il1a, and Il1b, and
A direct link between AREG and CAF differentiation has MHC2 presentation machinery (H2-Ab1 and H2-Eb1). Tumors
been recently described (63), with fibroblast-derived AREG lacking stromal IL33 also had increased recruitment and acti-
shifting myCAFs to a tumor-promoting subtype. However, vation of granulocytes which, although generally thought to
we did not detect AREG expression in fibroblasts, rather, we be immunosuppressive (90), also expressed the T–cell activating

1980 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

component Cd80 in the CreER;Il33f/f context, adding to the adhesion kinase activation has been implicated in IL33 ex-
proinflammatory milieux of the TME. Given that we did not pression (92, 93), and in PSCs, PDGF-BB and IFNγ can both
detect ST2 expression in these myeloid cell populations by cause IL33 upregulation (40).
scRNA-seq, these changes are likely due to indirect effects of Together, our findings shed light on the mechanisms
IL33 loss, potentially stemming from changes in the produc- through which cancer cells reprogram the stroma, particularly
tion of other cytokines from ST2+ cells. CAFs—mechanisms that could be targeted therapeutically to
In the context of stromal IL33 loss, tumor cell/TME to “normalize” fibroblasts in the setting of pancreatic cancer,
T-cell interactions were altered in such a way that the overall as proposed (94). Our data demonstrate the rapid nature by
result was an increase in CD8+ T-cell recruitment and activa- which CAF polarization can shift due to the addition or loss of
tion, with likely cytotoxic activity, as evidenced by increased extracellular stimuli, further complicating potential efforts to
CC3 and Granzyme-B. To this point, we also found increased establish sustained CAF reprograming as a therapeutic ap-
infiltration by helper T cells, which are necessary for driving proach. Importantly, our use of the pharmacologic KRASG12D
an antitumoral cytotoxic response (91). We were surprised to inhibitor, MRTX1133, showed a reduction of CAF IL33 expres-
find no change in the relative abundance of Th2s or ILC2s sion that mirrored the genetic ablation of oncogenic KRAS,

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

by scRNA-seq, as the recruitment of these cell types has been highlighting a potential synergistic effect whereby the TME may
previously linked to extracellular IL33 activity; however, this be reprogrammed in response to targeting cancer cells—a highly
was likely due to the compounded changes in recruitment relevant observation, as KRASG12D inhibitors are currently en-
factors seen across cell populations when stromal IL33 was re- tering the clinic (95, 96). Ultimately, targeting IL33 might pro-
moved. Overall, our data indicate that extracellular stromal vide a new avenue for combination therapy in pancreatic cancer.
IL33 has tumor-promoting activity through modulation of
the immune microenvironment.
Given the functional importance of stromal IL33, we then
Methods
sought to address the mechanism through which oncogenic Human Samples
KRAS-expressing cells drive IL33 expression in pancreatic Surgical specimens were obtained from patients referred for Whip-
fibroblasts. We found that PAF/CAF IL33 expression in mu- ple or distal pancreatectomy according to IRB HUM000025339.
rine models of PanIN and PDA is dependent on fibroblast Written informed consent forms were obtained from the patients.
pSTAT3 and epithelial oncogenic KRAS activity, consistent Studies were approved by the Institutional Review Boards of the
with our previous observations in vitro (12). Similarly, secreted University of Michigan Medical School and were conducted in ac-
factors derived from human PDA organoids induce IL33 up- cordance with recognized ethical guidelines.
regulation in human pancreatic fibroblasts and CAFs with
dependence on fibroblast JAK1/2 activation. Using a multi- Mice
cytokine ELISA platform, we showed that pancreatic cancer Mice were housed in the specific pathogen-free animal facility at
cells secrete the NF-κB-activating ligands IL1α and TNFα as the Rogel Cancer Center at the University of Michigan. Animals were
well as the JAK1/2-STAT3-activating ligands IL6 and LIF in overseen by the University of Michigan Unit for Laboratory Animal
Medicine. Male and female age-matched mice were utilized for exper-
an oncogenic KRAS-dependent manner. Treatment of healthy
iments, and littermates were used whenever possible. Mouse studies
pancreatic fibroblasts with these cytokines revealed a complex were approved by the Institutional Animal Care and Use Committee
mechanism whereby gene expressions of many cytokines (in- at the University of Michigan (protocol #PRO00011612).
cluding Il33) are upregulated or downregulated with differ- iKRASG12D [Ptf1a (aka p48)-Cre;TetO-KrasG12D;Rosa26rtTa/+] mice were
ent temporal dynamics. Il33 and Il6, but not Lif, are gradually generated as previously described (7, 12). KRASG12D expression in the
induced by recombinant LIF over the course of 16 hours, which pancreatic epithelium was induced in adult mice (8–14 weeks of age)
in turn suppresses the expression of Acta2, encoding for by supplementing 5% sucrose water with doxycycline (0.2 g/L, #D9891,
α-SMA. Conversely, Il6 and Lif, but not Il33, are induced by re- Sigma-Aldrich). After maintaining mice on doxycycline water for 3 to
combinant IL1α and TNFα, although these activation relation- 4 days, acute pancreatitis was induced via intraperitoneal injection of
ships are sustained differently over time. Furthermore, despite caerulein (75 μg/kg, #C9026, Sigma-Aldrich) 8 times a day for 2 days.
Mice were then aged for 3 or 5 weeks after the last day of caerulein, at
being a direct activator of IL33 in SCC TICs, TGFβ induces Il6
which point a cohort of mice was randomly selected to switch from
but not Il33 upregulation in pancreatic fibroblasts (54). It is doxycycline water to regular water, thus terminating the expression
also notable that we detected the highest levels of IL33 expres- of KRASG12D. After 3 additional days of doxycycline (“3- or 5-week
sion in vivo in iCAFs, which are canonically polarized by acti- KRASG12D ON”) or regular water (“3- or 5-week KRASG12D ON + 3-day
vation of NF-κB and JAK1/2-STAT3 and suppressed by TGFβ KRASG12D OFF”), animals were sacrificed and pancreata were collected.
signaling (22). Interestingly, tumor CM-induced upregulation iKRASG12D;Trp53R172H/+ (Ptf1a-Cre;TetO-KrasG12D;Rosa26rtTa/+;Trp53R172H/+)
of Il33 is almost completely ablated when combined with mice were generated as previously described (55). Expression of
GolgiStop, which prevents fibroblast autocrine signaling; KRASG12D in this model was induced by providing doxycycline chow
thus, a paracrine and an autocrine component are required (1.0 gm/kg, #F3949, Bio-Serv). Mice were aged to 14 weeks and then
for Il33 expression. We also found that tumor-instigated fibro- harvested (“14-week KRASG12D ON”) or remained on doxycycline
chow to 15 weeks before being switched back to a regular diet for an
blast STAT3 phosphorylation was dependent on fibroblast
additional week before harvesting pancreata (“15-week KRASG12D ON
autocrine signaling, in accordance with existing reports in + 1-week KRASG12D OFF”).
the literature (22). Importantly, direct STAT3 activation in Pdgfra-CreERT2/+ (#032770, Jackson Laboratory, mixed background
fibroblasts treated with GolgiStop did not rescue IL33 accu- at time of utilization) mice were crossed with either Il33f/f-eGFP mice
mulation, indicating that additional autocrine pathways are (#030619, Jackson Laboratory, C57BL/6J) or Stat3flox mice (#016923,
required to activate IL33 upregulation. In other cell types, focal Jackson Laboratory, C57BL/6J) to generate Pdgfra-CreERT2/+;Il33f/f-eGFP

October 2024 CANCER DISCOVERY | 1981

RESEARCH ARTICLE Donahue et al.

mice and Pdgfra-CreERT2/+;Stat3f/f mice, respectively. These mice were Histology and Immunohistochemistry
aged 8 to 12 weeks and then treated with tamoxifen (4 mg/day, Murine tissues were incubated in 10 mL of 10% neutral-buffered
#T5648, Sigma-Aldrich) dissolved in corn oil via oral gavage once a formalin at room temperature (RT) for 24 hours before being sub-
day for 5 days. Two days after their last gavage, mice underwent orthot- mitted to the University of Michigan Tissue & Molecular Pathology
opic tumor implantation surgery. Experimental and control animals Shared Resource core facility for paraffin embedding and sectioning.
used in Pdgfra-CreERT2/+;Il33f/f-eGFP experiments were additionally Gomori trichrome staining was performed following the manufac-
switched to tamoxifen chow (400 mg/kg, #TD.130860, Teklad Cus- turer’s guidelines (#87021, Epredia).
tom Diets) immediately following the last administration of tamox- For IHC staining, slides were deparaffinized and rehydrated
ifen gavage and remained on tamoxifen chow until experiment end by serial incubation in Xylene (5 minutes, two times), 100% eth-
(except in the case of animals intended for scRNA-seq, which received anol (5 minutes, two times), 95% ethanol (1 minute, two times), and
standard chow diet). diH2O (5 minutes). Slides were then subjected to antigen retrieval
Il1rl1−/− mice were originally generated by Dr. Shizuo Akira’s lab with Antigen Retrieval Citra Solution (#HK086-9K, BioGenex) and
and backcrossed to C57BL/6 by Dr. Stefan Writz’s group (97, 98). Mice microwaving as per the manufacturer’s instructions. Upon cooling,
were aged to 8 weeks old before undergoing orthotopic tumor im- slides were quenched in 0.3% H2O2 in methanol for 15 minutes
plantation surgery. Il1rl1+/+ C57BL/6 mice were used as controls. before blocking in 1% BSA (#BP1600-100, Fisher Scientific) in PBS

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

for 30 minutes at RT. After blocking, slides were incubated in prima-
Murine Cell Lines ry antibody diluted in PBS in a humidified chamber overnight at
All cell lines were maintained in DMEM (#11965, Gibco) + 10% 4°C. Primary antibodies were used at the following concentrations:
heat-inactivated FBS (#A3840202, Gibco) + 1% penicillin–streptomycin human IL33 (1:50, #AF3625, R&D Systems), Cleaved Caspase 3
(PS; #15070063, Gibco) at 37°C and 5% CO2. Cells were routinely (1:100, #9664L, Cell Signaling), F4/80 (1:200, #70076, Cell Signaling),
tested for mycoplasma using the MycoAlert PLUS Mycoplasma De- and myeloperoxidase (1:200, #AF3667, R&D Systems). The next day,
tection Kit (#LT07-710, Lonza), and cell lines used in orthotopic sur- slides were incubated in a biotinylated secondary antibody diluted
geries tested negatively prior to injection. 1:300 in PBS for 45 minutes at RT. Vectastain Elite ABC Kit, Peroxi-
The iKRASG12D;Trp53R172H/+ cell lines A9993 (aka “iKRASG12DF1,” dase (PK6100, Vector Laboratories) complex was then added follow-
FVB/N, female) and 9805 (mixed background, male) were previously ing the manufacturer’s protocol, and slides were developed using the
derived from spontaneous pancreatic tumors (7, 55). Both cell lines DAB (3,3′-diaminobenzidine) Substrate Kit, horseradish peroxidase
were maintained in media supplemented with doxycycline (1 μg/mL, without nickel (#SK4100, Vector Laboratories). Next, slides were coun-
#D9891, Sigma-Aldrich). Only passage 17 and below were utilized terstained in hematoxylin (#26381-02, Electron Microscopy Services)
to ensure doxycycline dependence in these cell lines. The male KPC for 2 minutes and then dehydrated by serial incubation in 95% eth-
(Pdx-Cre;LSL-KrasG12D;LSL-Trp53R172H) cell line 7940b (C57BL/6J) was anol (10 seconds, two times), 100% ethanol (2 minutes, two times),
provided by Dr. Gregory Beatty (48). The fibroblast cell line CD1WT and Xylene (2 minutes, two times). The resulting slides were mounted
was previously derived from a healthy murine pancreas (25). with Cytoseal (#23-244256, Epredia).
To generate IL33 wildtype (“IL33 WT”) and knockout (“IL33 KO”)
cell lines, primary fibroblasts were isolated under sterile conditions Immunofluorescence
from the pancreas of a healthy male Il33f/f-eGFP mouse using the out- For immunofluorescent (IF) stains, slides were deparaffinized
growth method (99). Fibroblasts were established and infected with and rehydrated as for IHC. Antigen retrieval was done by microwaving
adenovirus (Ad5 CMV-eGFP for IL33 WT lines or Ad5 CMV-eGFP and Antigen Retrieval Citra Solution as previously described, or EDTA
and Ad5 CMV-Cre for IL33 KO lines) as previously described (100). Unmasking Solution (#14747, Cell Signaling, following manufac-
Successfully infected cells were selected by flow cytometry and assessed turer’s recommendations) when staining for pSTAT3 and/or IL33
at the genomic and protein levels for recombination efficiency. primary antibodies. Upon cooling, slides were blocked in 1% BSA
in PBS for 30 minutes (or 10% goat serum if IL33 primary anti-
Orthotopic Surgeries body was used) and then incubated in primary antibody overnight
To establish orthotopic KPC tumors, 7940b cells (passage 16 or in a humidified chamber at 4°C. Primary antibodies were used at
lower) were injected into the pancreata of adult mice (7.5 × 104 cells/ the following concentrations: IL33 (1:50, #AF3625, R&D Systems),
animal for the Pdgfra-CreERT2/+;Il33f/f-eGFP with tamoxifen chow PDGFRα/β (1:100, #AB32570, Abcam), E-Cadherin (1:50, #14472S,
model and 5 × 104 cells/animal for all other orthotopic experiments). Cell Signaling), pSTAT3 (Y705; 1:100, #9145S, Cell Signaling), Ki67
Equivalent injections were given to control animals; controls used (1:100, #AB15580, Abcam), α-SMA (1:1,000, #A2547, Sigma-Aldrich),
were C57BL/6 mice for the Il1rl1−/− experiment and Pdgfra-CreERT2/+ CD8 (1:400, #98941, Cell Signaling), Granzyme-B (1:500, #17215,
mice for the Pdgfra-CreERT2/+;Il33f/f-eGFP and Pdgfra-CreERT2/+;Stat3f/f Cell Signaling), CD4 (1:100, #25229, Cell Signaling), CD44 (1:100,
experiments. All Pdgfra-CreERT2/+ controls were also given tamoxifen #MA1-10225, Invitrogen), and Foxp3 (1:100, #12653, Cell Signaling).
oral gavage/chow alongside experimental animals in accordance with The following day, nuclei were stained by incubating slides in DAPI
the experimental design. 7940b cells were resuspended in a 1:1 ratio (4′,6-diamidino-2-phenylindole, dihydrochloride; 1:30,000 in PBS,
of RPMI 1640 medium (#11875, Gibco) supplemented with 10% FBS #D1306, Invitrogen) for 10 minutes in the dark at RT. Secondary anti­
and Matrigel matrix basement membrane (#354234, Corning) at a con- body incubation was subsequently done at 1:300 in PBS for 45 minutes
centration of 1,875 cells/μL (for 7.5 × 104 cells/animal; 40 μL/injection) at RT in the dark. Slides were mounted with Prolong Gold Antifade
or 1,000 cells/μL (for 5 × 104 cells/animal; 50 μL/injection; for full sur- Mountant (#P36934, Invitrogen). The Tyramide SuperBoost Kit
gery procedure, see ref. 101). Tumors were allowed to grow for 3 weeks (#B40922, #B40912, or #B40923, Invitrogen) was used when staining
prior to harvest. with multiple primary antibodies raised in the same host species.
To establish the iKRASG12D orthotopic model, 5 × 104 A9993 cells
were implanted into syngeneic FVB/N mice (1,000 cells/μL in a 1:1 Imaging and Quantification
ratio of RPMI 1640 medium supplemented with 10% FBS and Matri- IHC and IF slides were imaged on an Olympus BX53F micro-
gel matrix basement membrane; ref. 55). Mice were provided doxycy- scope outfitted with an Olympus DP80 digital camera and CellSens
cline chow for 2 weeks to maintain KRASG12D expression, and then Standard software (Olympus). Select IF stains (those that included
randomly selected to return to regular diet for an additional 3 days IL33 or pSTAT3 primary antibody) were imaged on the Leica Stel-
(KRASG12D “OFF”) before tumor harvest. laris 8 Falcon Confocal Microscopy System using LAS X software

1982 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

(Leica Microsystems). Investigators were blind to the experimental and a scale factor of 10,000. Variable genes were identified using the
condition of the slide when taking images, and at least three mice per FindVariableFeatures function. When scRNA-seq datasets from dif-
group were assessed. fering run dates were merged, batch correction was performed using
IHC and IF images were quantified using Fiji (102). To quantify the IntegrateData workflow in Seurat (105). All genes were scaled and
% positive staining per field of view (FOV), three to eight images per centered using ScaleData. Principal component analysis (PCA) was
slide (depending on the size of the tissue) were taken on the Olym- performed using RunPCA. Cell clusters were identified using Find-
pus system at 20× magnification with care to capture only tissue Neighbors and FindClusters with dimensions that captured ∼90%
area within the FOV. The fraction of stained area for each image was variance as defined by the PCA result. UMAP clustering was complet-
averaged within each mouse, and the resulting % positive scores are ed using RunUMAP. To compare gene expression profiles between
reported in the text (each datapoint is a single mouse). To quanti- groups, we used FindMarkers and Wilcoxon rank sum test with
fy the % positive cells, E-Cadherin+ or PDGFRα/β+ cells (∼100–150 Bonferroni correction. In addition to Seurat, the R package scCus-
cells each per FOV) or CD4 cells (all cells per FOV) were manually tomize version 1.1.1 (RRID:SCR_024675) was used for visualization.
categorized as Ki67/Foxp3 positive or negative; at least three images Pathway analysis was performed with the R package fgsea version
per mouse were quantified, and total % positive scores were averaged 1.26.0 (bioRxiv 060012) and the MSigDB mouse-ortholog hallmark
within each mouse and represented in the figure. To quantify the gene sets were downloaded using the R package msigdbr version 7.5.1
fluorescence intensity of IL33 and/or pSTAT3 within fibroblasts, (RRID:SCR_022870). Pathway enrichment scores were also calculated

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

three to five images were taken of the tissues on the Leica system using Seurat’s AddModuleScore functionality.
using the 40× objective. Then, ≥30 regions of interest (ROI) com- To infer ligand–receptor interactions, we used the R package Cell-
posed of individual PDGFRα/β+ cells were hand-selected for each Chat version 1.6.1 (106) following their standard workflow. Differ-
image (for images with fewer than 30 PDGFRα/β+ cells, all available ential gene expression via Wilcoxon rank sum test with Bonferroni
PDGFRα/β+ cells were selected). Using the single channel image correction was performed using identifyOverExpressedGenes with
for the protein of interest, the integrated density was recorded for thresh.pc = 0.1 (minimum 10% of cells expressing the given gene in
each ROI. The mean gray value of three background ROIs was also the cluster), tresh.fc = 0.1 (log fold-change minimum), and thresh.
recorded. Corrected total cell fluorescence (CTCF) was then calcu- P = 0.05 (P value threshold). For final visualization, ligands with less
lated for each ROI as CTCF = Integrated Density − (Area of ROI × than 0.25% fold-change were removed. Interactions between the
Average Mean Gray Value of Background ROIs). The CTCF for each ligand-expressing cluster and the receptor-expressing cell popula-
ROI is represented in the results section to profile the range and tions were visualized using netVisual_chord_gene with slot.name =
distribution of CTCF scores in each group. Colocalization of CD8/ “netP” and then manually annotated to include the relevant gene
Granzyme-B and PDGFRα/β/α-SMA were measured using Just An- names found in slot.name = “net.”
other Colocalization Plugin in ImageJ (103). The complete R script encompassing the analysis and visualization
for each dataset presented in the manuscript is publicly available on
Generation of Single Cell RNA Sequencing Datasets GitHub (https://github.com/PascaDiMagliano-Lab/).
We analyzed a number of scRNA-seq datasets previously published
by our group and our collaborators, which are detailed in Supplemen-
Tumor Conditioned Media Generation
tary Table S1. Here, we present a number of new murine scRNA-seq The 9805 cell line was plated at 1 × 106 cells per 10-cm dish in
datasets including three KC PanIN datasets, seven iKRASG12D data- DMEM + 10% FBS + 1% PS + doxycycline (1 μg/mL). The cells were al-
sets, and the Pdgfra-CreERT2/+;Il33f/f orthotopic datasets [NCBI Gene lowed to adhere overnight, and then, the plates were washed with PBS
Expression Omnibus (GEO) collection GSE269888]. The detailed and replenished with DMEM + 10% FBS + 1% PS with or without dox-
GEO accession numbers for these datasets can also be found in Sup- ycycline (1 μg/mL; establishing “KRASG12D ON” and “KRASG12D OFF”
plementary Table S1. conditions, respectively). The cells were again left overnight to transition
To prepare tissues for sequencing, samples were disrupted to sin- between “KRASG12D ON” and “KRASG12D OFF” status. The following
gle-cell suspension by mechanically mincing the tissue, followed by day, the cells were washed with PBS and replenished with low-serum
enzymatic digestion with Collagenase V (1 mg/mL in RPMI, #C9263, DMEM + 1% FBS + 1% PS with or without doxycycline. After 48 hours,
Sigma-Aldrich) at 37°C for 30 minutes under constant shaking. the resulting CM was centrifuged at 1,000 g for 10 minutes at 4°C to
After digestion, suspensions underwent serial filtration through pellet debris. The supernatant was decanted, and then, the CM was
500-, 100-, and 40-μm mesh cell strainers. We then enriched live cells stored at either 4°C (for up to 1 week) or −80°C (for up to 3 months)
using the MACS Dead Cell Removal Kit (#130-090-101, Miltenyi before use. Samples of “KRASG12D ON” and “KRASG12D OFF” CM were
Biotec). The resulting single-cell suspensions were brought to the submitted to the Michigan Diabetes Research Center Clinical Core to
University of Michigan Advanced Genomics Core for cDNA library measure extracellular LIF, IL6, IL1α, TNFα, and IL1β using MILLIP-
preparation and sequencing. Samples were run using 50-cycle paired- LEX (MilliporeSigma) and the Luminex 200 Instrument System.
end reads on the HiSeq 4000 (Illumina) or NovaSeq 6000 (Illumina)
to a depth of 100,000 reads. The University of Michigan Advanced iKRASG12D JAK1/2i In vivo
Genomics Core also performed processing and alignment of the raw For JAK1/2i treatment, iKRASG12D mice 8 to 14 weeks of age were
data using Cell Ranger (versions 2.2.0–6.1.2 depending on the date of given doxycycline chow for 4 days and then received caerulein-induced
sequencing) with default settings and an initial expected cell count acute pancreatitis as previously described. Three weeks following the
of 10,000. Details on the sequencing and alignment method for each last dose of caerulein, mice were randomized and received either
sample can be found on NCBI GEO. vehicle or the JAK1/2i ruxolitinib (#INCB18424, MedChemExpress)
via oral gavage once a day for 4 days (180 mg/kg in 100 μL of 10%
Single cell RNA Sequencing Analysis DMSO in corn oil). Mice were harvested the same day following the
Downstream analysis of the scRNA-seq data was done using R ver- last dose of ruxolitinib.
sion 4.3.0 “Already Tomorrow,” R Studio version 2023.03.1 + 446, and
R package Seurat version 4.3.0.1 (104). We excluded genes that ap- Pharmacologic Inhibition of Pancreatic Fibroblasts
peared in <3 cells, as well as cells with <100 genes, <1,000 or >100,000 CD1WT were plated at 3 × 105 cells per well in six-well dishes with
transcripts, or >15% mitochondrial gene composition. Genes were DMEM + 10% FBS + 1% PS. The next day, cells were washed with
normalized using NormalizeData with the “LogNormalize” method PBS and treated with 2 mL of fresh, low-serum media (DMEM + 1%

October 2024 CANCER DISCOVERY | 1983

RESEARCH ARTICLE Donahue et al.

FBS + 1% PS), 1 mL of “KRASG12D ON” or “KRASG12D OFF” CM + IL33 Secretion

1 mL of low-serum media, or a combination of 1 mL of “KRASG12D CD1WT were plated 2 × 105 cells per well into four wells of a six-
ON” CM + 1 mL of low-serum media + either JAK1/2i (ruxolitinib) or well plate in triplicate and left overnight. The following day, two wells
TGFβRi (Galunisertib, #HY13226, MedChemExpress) at 0.03, 0.3, or were provided 2 mL of fresh, low-serum DMEM + doxycycline and
3 μmol/L. DMSO and doxycycline were given as vehicle controls for two wells received 1 mL of “KRASG12D ON” CM + 1 mL of low-serum
JAK1/2i/TGFβRi and “KRASG12D ON” CM, respectively. Cells were DMEM. After 24 hours of incubation, we administered 100-μmol/L
harvested 24 hours after treatment. H2O2 to one control and one “KRASG12D ON” CM well by pipetting
In the shortened adaptation of the JAK1/2i experiment, CD1WT directly onto the plate. The cells were incubated for an additional
were plated as described and then treated with 2 mL of low-serum 24 hours, and then, media was collected. Media from three wells were
DMEM + doxycycline vehicle or 1 mL of “KRASG12D ON” CM + 1 mL pooled (∼6 mL at the time of collection), centrifuged at 1,000 g for
of low-serum media. 20 hours following treatment, JAK1/2i or an 10 minutes at 4°C to pellet any debris, and then concentrated using
equivalent volume of DMSO was added directly to the well to reach Amicon Ultra 4 Centrifugal Filters (#UFC801024, MilliporeSigma)
a total concentration of 0.3 μmol/L. CD1WT were harvested 4 hours as per the manufacturer’s directions. The resulting concentrated
following JAK1/2i treatment. CD1WT CM was submitted for IL33 ELISA at the Rogel Cancer Cen-
ter Immunology Core, alongside samples of plain “KRASG12D ON”
rIL6, rLIF, rIL1α, rTNFα, and rTGF-β Treatment

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

CM, and low-serum DMEM to serve as controls.
CD1WT were plated at 3 × 105 cells per well in six-well dishes with
DMEM + 10% FBS + 1% PS. The following day, cells were washed Isolation and Culture of Primary Human PDA Organoids and
with PBS and treated with either 2 mL of low-serum DMEM + dox- Fibroblasts
ycycline vehicle, 1 mL of “KRASG12D ON” CM + 1 mL of low-serum
Adjacent normal fibroblasts, tumor fibroblasts, and tumor or-
DMEM, or 2 mL of low-serum DMEM + doxycycline vehicle + rIL6
ganoids were generated from resected patient tumors. Tissue was di-
(0.03–30 ng/mL, #406-ML, R&D Systems), rLIF (0.06–6 ng/mL,
gested as described previously (107). To culture organoids, cells were
#8878-LF, R&D Systems), or rTGFβ (2 or 20 ng/mL, #7666-MB,
plated in growth factor-reduced Matrigel matrix basement mem-
R&D Systems). Cells were harvested 24 hours after treatment. For
brane and grown in human complete feeding media [50% L-WRN CM
rIL1α (20 ng/mL, #400-ML, R&D Systems) and rTNFα (20 ng/mL,
(108) and 50% DMEM/F12 (#12634010, ThermoFisher) supplement-
#410-MT, R&D Systems) experiments, three additional wells were giv-
ed with B27 (#17504044, ThermoFisher) to 1× final concentration,
en 2 mL of low-serum DMEM + doxycycline vehicle in parallel with
10-mmol/L nicotinamide (#N0636, Sigma-Aldrich), 1.25-mmol/L
the untreated control group to serve as the 1-hour treatment wells.
N-acetyl-cysteine (#A9165, Sigma-Aldrich), 50-ng/mL hEGF (#236-EG,
One hour before harvest, these wells were given 1 mL of “KRASG12D
R&D Systems), 100-ng/mL hFGF (#345-FG, R&D Systems), 10-nmol/L
ON” CM + 1 mL of low-serum media or 2 mL of rIL1α (20 ng/mL) or
hGastrin1 (#3006, Tocris Bioscience), 10.5-μmol/L Y27632 (#1254,
rTNFα (20 ng/mL) + doxycycline vehicle and collected simultane-
Tocris Bioscience), and 2× PS (107, 109)]. To culture fibroblasts, cells
ously with the 2-hour treatment group.
were grown in fibroblast media (DMEM-F12 + 10% FBS + 1× PS). All
primary cell cultures were characterized by whole exome sequencing
Tumor Conditioned Media Time Course as described previously (107).
CD1WT were plated at 1.5 × 105 cells per well in six-well dishes
with DMEM + 10% FBS + 1% PS. The following day, cells were washed Human Fibroblast Treatment with Organoid CM and JAKi
with PBS, then one well was switched to 1 mL of “KRASG12D ON”
Tumor organoids (cell line 61620400) were seeded at 50% conflu-
CM + 1 mL of low-serum DMEM and another was switched to 1 mL
ence and cultured in human complete feeding media; organoid CM
of “KRASG12D OFF” CM + 1 mL of low-serum DMEM (this was the
was collected after 72 hours. Adjacent normal (cell line 19809) and tu-
24-hour timepoint for western blot collection series and the 48-hour
mor fibroblasts (cell lines 1249 and 21498) were cultured with 100%
timepoint for the RNA collection series). All other wells were given
fibroblast media or 50% organoid CM and 50% fibroblast media in the
2 mL of low-serum DMEM + doxycycline. Cells were then treated in
presence or absence of 0.5-μmol/L JAK1/2i (ruxolitinib) for 24 hours
descending order over time to allow all cells to be harvested simulta-
before harvesting total RNA. DMSO was used as a vehicle control
neously. The 0-hour timepoint never received CM.
when JAK1/2i was not provided.

Autocrine Blocking of Pancreatic Fibroblasts

Cell Proliferation Assay
CD1WT were plated at 3 × 105 cells per well in six-well dishes with
“IL33 WT” and “IL33 KO” cell lines were seeded at 3 × 103 cells
DMEM + 10% FBS + 1% PS. The next day, cells were washed with
per well in DMEM + 10% FBS + 1% PS in four opaque-walled
PBS and treated with either 2 mL of fresh, low-serum DMEM, 1 mL
96-well plates and left to adhere overnight. The following day,
of “KRASG12D ON” CM + 1 mL of low-serum DMEM, or 2 mL of low-
cell density was measured on 1 plate (day 0); the remaining plates
serum DMEM supplemented with 6-ng/mL rLIF (doxycycline was
were aspirated and washed with PBS, and cells were treated with
given as a vehicle control to all wells not receiving “KRASG12D ON”
low-serum DMEM or a 1:1 ratio of low-serum DMEM and 9805
CM). Two identical sets of CD1WT were treated under these condi-
“KRASG12D ON” tumor CM. Conditions without tumor CM were
tions in parallel (groups 1 and 2). Cells were incubated for 18 hours,
given doxycycline as a control. Plates were collected and read every
at which point the resulting CD1WT CM was removed by pipetting
24 hours for 3 days. Proliferation was measured using CellTiter-Glo
and set aside, and plates were washed with PBS. Immediately follow-
Luminescent Cell Viability Assay (#G7570, Promega) following the
ing the wash, the fibroblast CM was added back to the well it originated
manufacturer’s instructions. Two wells per condition were averaged
from for the wells in group 1. For the wells in group 2, we instead
for each measurement.
added fresh aliquots of low-serum DMEM + doxycycline (2 mL),
1 mL of “KRASG12D ON” CM + 1 mL of low-serum DMEM, or 2 mL of
low-serum DMEM + doxycycline supplemented with 6-ng/mL rLIF. Pharmacologic Inhibition of KRAS in Murine Tumor Cells
Group 2 cells also received the protein transport inhibitor GolgiStop The 9805 tumor cell line was grown as described above for the
(1.3 μL/2 mL, #554724, BD Biosciences). All cells were incubated for generation of tumor CM; briefly, cells were plated in doxycycline-
an additional 6 hours and then harvested for RNA or protein. containing DMEM + 10% FBS + 1% PS in 10 cm dishes (day 1; d1)

1984 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

and then, the following day, washed with PBS and switched to RNA Extraction and RT-qPCR
doxycycline-deficient low-serum DMEM (for “KRASG12D OFF” CM) RNA was extracted using the RNeasy Plus Mini Kit (#74134,
or doxycycline-proficient low-serum DMEM (for “KRASG12D ON” QIAGEN) following the manufacturer’s protocol. RNA levels and
CM; d2). On d3, cells were again washed and replenished with low- quality were assessed via nanodrop, and cDNA was generated using
serum DMEM (for “KRASG12D OFF” CM), low-serum DMEM + doxy- the High-capacity cDNA Reverse Transcription Kit (#4368814, Ap-
cycline (for “KRASG12D ON” CM), or low-serum DMEM + doxycycline + plied Biosystems) supplemented with RNAse Inhibitor (#N8080119,
small molecule inhibitor: MRTX1133 (0.5 μmol/L, #HY134813, Applied Biosystems). Samples were prepared for qPCR using Fast
MedChemExpress), Sotorasib (0.5 μmol/L, #HY114277, MedChe- SYBR Green Master Mix (#4385612, Applied Biosystems). Primers are
mExpress), or Trametinib (50 nmol/L, #S2673, Selleck Chemicals). listed in the Supplementary Table S2. Reactions were run on a Quant-
On d4, inhibitors were again spiked into the media at equivalent con- Studio 6 Pro (Applied Biosystems). Cyclophilin A (Ppia) and GAPDH
centrations to ensure continued inhibition of KRAS. Tumor CM was were used as the housekeeping genes in all mouse and human
collected as described above on d5. DMSO was used as a vehicle con- RT-qPCR experiments, respectively.
trol for the untreated conditions.
CD1WT cells were plated at 3 × 105 cells per well in six-well dishes
Bulk RNA Sequencing
with DMEM + 10% FBS + 1% PS. The next day, cells were washed with
PBS and treated with 2 mL of fresh, low-serum media (DMEM + 1% Extracted RNA was submitted to the University of Michigan

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

FBS + 1% PS) or 1 mL of low-serum media + 1 mL of the various tumor Advanced Genomics Core for bulk RNA-seq and preprocessing.
CMs. Doxycycline was given as vehicle control for “KRASG12D ON” CM. Samples were subjected to 151-bp paired-end sequencing accord-
Cells were harvested for RNA 24 hours after treatment. ing to the manufacturer’s protocol (Illumina NovaSeq 6000). BCL
Convert Conversion Software version 3.9.3 (Illumina) was used to
Western Blotting generate demultiplexed Fastq files. Quality control was performed
using FastQC version 0.11.8 (https://www.bioinformatics.babra-
Cells were lysed in their culture dish in RIPA buffer (#R0278, Sig- ham.ac.uk/projects/fastqc/). Reads were mapped to the reference
ma-Aldrich) supplemented with protease and phosphatase inhibitor genome GRCm38 (ENSEMBL) using STAR version 2.7.8a (110) and
cocktail (#87785, Sigma-Aldrich). Protein concentration was quanti- assigned count estimated to genes with RSEM version 1.3.3 (111).
fied using the Pierce BCA Protein Assay Kit (#23225, ThermoScien- Alignment options followed ENCODE standards for RNA-seq. Down-
tific) following the manufacturer’s protocol. Equal concentrations stream analysis was performed using DESeq2 version 1.40.2 (112).
of protein were separated via SDS-PAGE and transferred to polyvi- Differentially expressed genes were defined by an FDR of 0.05 and
nylidene difluoride membranes using a wet tank system (Bio-Rad). fold changes of 0.5 or more. The MSigDB mouse-ortholog hallmark
Membranes were rinsed in water and then allowed to dry completely gene sets were downloaded using the R package msigdbr version
before being reactivated in methanol and stained using Ponceau-S 7.5.1 (RRID:SCR_022870). Pathway analysis was performed using
(#P7170, Sigma-Aldrich). Membranes were then blocked at RT with the R package fgsea version 1.26.0 (bioRxiv 060012). The raw data
gentle shaking for 1 hour in 5% milk in TBST [1× TBS (#1706435, from this experiment are available on NCBI GEO accession number
Bio-Rad) + 0.1% Tween-20 (#P7949, Sigma-Aldrich)]. After blocking, GSE269889.
membranes were incubated in primary antibody diluted either in 5%
milk in TBST (for total protein targets) or 5% BSA in TBST (for phos-
phorylation site targets) at 4°C with gentle rocking. Primary anti- Quantification and Statistical Analysis
bodies were used at the following concentrations: mouse IL33 (1:100, GraphPad Prism 10 software was used for all statistical analyses
#AF3625, R&D Systems), PDGFRα (1:500, #3164S, Cell Signaling), outside of scRNA-seq experiments. Histograms are presented as
pSTAT3 (Y705; 1:1,000, #9145S, Cell Signaling), STAT3 (1:1,000, mean ± standard deviation. Two-tailed Student t test and one-way
#9139S, Cell Signaling), p-SMAD2 (S465/467)/SMAD3 (S423/425; ANOVA with Tukey test were performed as described in the figure
1:500, #8828S, Cell Signaling), SMAD2/3 (1:1,000, #8685S, Cell Sig- legend. All in vitro assays were performed at least three times with
naling), α-tubulin (1:2,000, #3873S, Cell Signaling), vinculin (1:2,000, cells of different passage numbers to ensure biologic variability.
#13901S, Cell Signaling). Membranes were incubated in primary an- Outliers were detected using Grubbs’ test by GraphPad Prism and
tibody for 24 to 48 hours, then washed with TBST, and incubated excluded where appropriate. To compare gene expression profiles
in HRP-conjugated secondary antibody diluted in 5% milk in TBST between groups using scRNA-seq data, we used a Wilcoxon rank
for 2 hours at RT. Bands were visualized using Clarity Western ECL sum test with Bonferroni correction. Exact P values for each compar-
Substrate (#1705060, Bio-Rad) and the ChemiDoc Imaging System ison can be found within the figures or figure legends, with P < 0.05
(Bio-Rad). Densitometry was performed in Fiji. considered significant.

Selection of PDGFRα+ Cells Data availability

To harvest PDGFRα+ cells from Pdgfra-CreERT2/+;Il33f/f orthotopic The scRNA-seq datasets generated in this study are publicly
tumors for protein, single-cell suspensions were created as described available at NCBI GEO accession number GSE269888. Preexist-
above, and cells were labeled and subjected to CD140a (PDGFRα) ing scRNA-seq data analyzed in this study are available at the
magnetic selection as per the manufacturer’s instructions (#130- accession numbers listed in Supplementary Table S1. Bulk RNA-
101-502, Miltenyi Biotec). The resulting PDGFRα+ cells were pelleted seq data generated in this study are also available at NCBI GEO
and lysed in RIPA + protease and phosphatase inhibitor cocktail for GSE269889.
western blot.
To harvest PDGFRα+ cells from Pdgfra-CreERT2/+;Stat3f/f orthotopic
tumors for RNA, tumors were processed into single-cell suspensions Authors’ Disclosures
and cells were stained for 30 minutes with antimouse PE-CD140a E.L. Lasse Opsahl reports grants from NIH/NCI during the con-
(PDGFRa; #135905, BioLegend), FITC-CD31 (#102405, BioLegend), duct of the study. A. Rao reports grants from NIH, ACS, DoD, UM
FITC-CD326 (#1182010, BioLegend), and FITC-CD45 (#157214, Bi- MIDAS, and UM MICDE; grants and other support from Agilent;
oLegend). Fibroblasts were identified as PDGFRa+, CD31−, CD326−, and other support from ASI during the conduct of the study, as well
and CD45−; sorted using Sony SH800 Cell Sorter; and processed for as other support from Telperian, Inc., Tempus Inc., Voxel Analyt-
RNA isolation. ics, LLC, and TCS Ltd. outside the submitted work. B. Allen reports

October 2024 CANCER DISCOVERY | 1985

RESEARCH ARTICLE Donahue et al.

grants from NIH during the conduct of the study. J. Shi reports Received January 30, 2024; revised May 18, 2024; accepted July 1,
grants from NIH during the conduct of the study. E.S. Carpenter re- 2024; published first July 3, 2024.
ports grants from Cornerstone Pharmaceutical and nonfinancial sup-
port from Beigene outside the submitted work. No disclosures were
reported by the other authors. References
1. Siegel RL, Giaquinto AN, Jemal A. Cancer statistics, 2024. CA Cancer
Acknowledgments J Clin 2024;74:12–49.
2. Sarantis P, Koustas E, Papadimitropoulou A, Papavassiliou AG,
The authors would like to thank Dr. Gregory Beatty (University
Karamouzis MV. Pancreatic ductal adenocarcinoma: treatment hur-
of Pennsylvania) for generously sharing the murine pancreatic cancer dles, tumor microenvironment and immunotherapy. World J Gas-
cell line 7940b. We would like to thank Dr. Nicholas Lukacs (Uni- trointest Oncol 2020;12:173–81.
versity of Michigan) for providing the Il1rl1−/− mouse model. The 3. Witkiewicz AK, McMillan EA, Balaji U, Baek G, Lin WC, Mansour J,
Ptf1a-Cre mouse was generously gifted by Dr. Chris Wright (Vander- et al. Whole-exome sequencing of pancreatic cancer defines genetic
bilt University). We thank Lee Olsen for her editing services as well diversity and therapeutic targets. Nat Commun 2015;6:6744.
as Michael Mattea and Christopher Strayhorn for providing his- 4. Buscail L, Bournet B, Cordelier P. Role of oncogenic KRAS in the
topathology support. We thank Matthew Felten for his assistance diagnosis, prognosis and treatment of pancreatic cancer. Nat Rev

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

with the submission of sequencing data to NCBI GEO. Moreover, Gastroenterol Hepatol 2020;17:153–68.
we would like to acknowledge technical support from the University 5. Aguirre AJ, Bardeesy N, Sinha M, Lopez L, Tuveson DA, Horner J,
of Michigan Advanced Genomics Core, Michigan Diabetes Research et al. Activated Kras and Ink4a/Arf deficiency cooperate to pro-
Center Clinical Core, Rogel Cancer Center Immunology Core, Rogel duce metastatic pancreatic ductal adenocarcinoma. Genes Dev
Cancer Center Tissue and Molecular Pathology Shared Resource, and 2003;17:3112–26.
University of Michigan Microscopy Core. Funding: K.L. Donahue 6. Hingorani SR, Petricoin EF, Maitra A, Rajapakse V, King C,
was supported by NIH/NCI F31-CA265085. K.L. Donahue and E.L. Jacobetz MA., et al. Preinvasive and invasive ductal pancreatic cancer
Lasse Opsahl were funded by the NIH/NCI Cancer Biology Training and its early detection in the mouse. Cancer Cell 2003;4:437–50.
7. Collins MA, Bednar F, Zhang Y, Brisset JC, Galban S, Galban CJ, et al.
Program T32-CA009676. Work in the M. Pasca di Magliano labora-
Oncogenic Kras is required for both the initiation and maintenance
tory was supported by NIH/NCI grant U54CA274371. M. Pasca di
of pancreatic cancer in mice. J Clin Invest 2012;122:639–53.
Magliano and F. Bednar were supported by NIH/NCI R01-CA271510.
8. Ying H, Kimmelman AC, Lyssiotis CA, Hua S, Chu GC, Fletcher-
Dr. M. Pasca di Magliano and Dr. B. Allen were also supported by
Sananikone E, et al. Oncogenic Kras maintains pancreatic tumors
NIH/NCI R01-CA275182. M. Pasca di Magliano and T.L. Frankel through regulation of anabolic glucose metabolism. Cell 2012;149:
were supported by NIH/NCI R01-CA268426. T.L. Frankel was also 656–70.
supported by VA BLR&D Merit Award 5I01BX005777, NIH/National 9. Sherman MH, Beatty GL. Tumor microenvironment in pancreatic
Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) cancer pathogenesis and therapeutic resistance. Annu Rev Pathol
5R01DK128102, and NIH/NCI U01CA274154. Moreover, F. Bednar 2023;18:123–48.
was supported by the Association of Academic Surgery, Association 10. Ying H, Dey P, Yao W, Kimmelman AC, Draetta GF, Maitra A., et al.
of VA Surgeons, and American Surgical Association Foundation. Genetics and biology of pancreatic ductal adenocarcinoma. Genes
P. Kadiyala was supported by the NIH/National Institute of Allergy Dev 2016;30:355–85.
and Infectious Diseases Research Training Grant in Experimental 11. Steele NG, Biffi G, Kemp SB, Zhang Y, Drouillard D, Syu L, et al.
Immunology T32-AI007413 and NIH/National Institute of General Inhibition of Hedgehog signaling alters fibroblast composition in
Medical Sciences (NIGMS) Training Program in Translational Re- pancreatic cancer. Clin Cancer Res 2021;27:2023–37.
search T32-GM113900. W. Du, M.K. Scales, and R.E. Menjivar were 12. Velez-Delgado A, Donahue KL, Brown KL, Du W, Irizarry-Negron V,
supported by the NIH/Eunice Kennedy Shriver National Institute of Menjivar RE, et al. Extrinsic KRAS signaling shapes the pancreatic
Child Health & Human Development Training Program in Organo- microenvironment through fibroblast reprogramming. Cell Mol
genesis T32-HD007505. Furthermore, M.K. Scales was supported by Gastroenterol Hepatol 2022;13:1673–99.
13. Zhang Y, Yan W, Mathew E, Kane KT, Brannon A III, Adoumie M,
NIH/NCI F31-CA232655 and the Bradley Merrill Patten Memorial
et al. Epithelial-myeloid cell crosstalk regulates acinar cell plasticity
Scholarship. R.E. Menjivar was supported by the NIH/NIGMS Cellular
and pancreatic remodeling in mice. Elife 2017;6:e27388.
and Molecular Biology Training Grant T32-GM007315 and the NIH/
14. Hou P, Kapoor A, Zhang Q, Li J, Wu CJ, Li J, et al. Tumor microenvi-
NCI F31-CA257533. A. Velez-Delgado was supported by the NIH/
ronment remodeling enables bypass of oncogenic KRAS dependency
NIGMS Cellular Biotechnology Training Program T32-GM008353 in pancreatic cancer. Cancer Discov 2020;10:1058–77.
and the NIH/NCI F31-CA247037. A.M. Elhossiny was supported 15. Dey P, Li J, Zhang J, Chaurasiya S, Strom A, Wang H, et al. Oncogenic
by the Rackham International Student Fellowship. P.I. Medina- KRAS-driven metabolic reprogramming in pancreatic cancer cells
Cabrera, A. Velez-Delgado, and R.E. Menjivar were supported by the utilizes cytokines from the tumor microenvironment. Cancer Discov
University of Michigan Rackham Merit Fellowship. P.I. Medina- 2020;10:608–25.
Cabrera was supported by the Cancer Biology Graduate Program at 16. Biffi G, Tuveson DA. Diversity and biology of cancer-associated fi-
the University of Michigan. B.D. Griffith was supported by the NIH/ broblasts. Physiol Rev 2021;101:147–76.
NCI Surgical Oncology Research Training Program T32-CA009672 17. Ning W, Marti TM, Dorn P, Peng RW. Non-genetic adaptive resis-
and Society of University Surgeons Resident Research Award. A. Rao tance to KRASG12C inhibition: EMT is not the only culprit. Front On-
was supported by the NIH/NCI R37-CA214955-01A1 and CCSG Bio- col 2022;12:1004669.
informatics Shared Resource 5 P30 CA046592. J. Shi was supported 18. Garcia PE, Adoumie M, Kim EC, Zhang Y, Scales MK, El-Tawil YS,
by NIH/NCI R37CA262209. Finally, Y. Zhang was funded by NIH/ et al. Differential contribution of pancreatic fibroblast subsets to the
NCI R50CA232985. E.S. Carpenter was supported by the VA BLR&D pancreatic cancer stroma. Cell Mol Gastroenterol Hepatol 2020;10:
CDA IK2BX005875, American College of Gastroenterology CDA, and 581–99.
NIH/NIDDK T32-DK094775. 19. Helms EJ, Berry MW, Chaw RC, DuFort CC, Sun D, Onate MK,
et al. Mesenchymal lineage heterogeneity underlies nonredundant
functions of pancreatic cancer-associated fibroblasts. Cancer Discov
Note 2022;12:484–501.
Supplementary data for this article are available at Cancer Discovery 20. Zhang T, Ren Y, Yang P, Wang J, Zhou H. Cancer-associated fibroblasts
Online (http://cancerdiscovery.aacrjournals.org/). in pancreatic ductal adenocarcinoma. Cell Death Dis 2022;13:897.

1986 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

21. Sherman MH, Magliano MPD. Cancer-associated fibroblasts: les- 42. Steele NG, Carpenter ES, Kemp SB, Sirihorachai VR, The S,
sons from pancreatic cancer. Annu Rev Cancer Biol 2023;7:43–55. Delrosario L, et al. Multimodal mapping of the tumor and periph-
22. Biffi G, Oni TE, Spielman B, Hao Y, Elyada E, Park Y, et al. IL1- eral blood immune landscape in human pancreatic cancer. Nat
Induced JAK/STAT signaling is antagonized by TGFβ to shape CAF Cancer 2020;1:1097–112.
heterogeneity in pancreatic ductal adenocarcinoma. Cancer Discov 43. Hosein AN, Huang H, Wang Z, Parmar K, Du W, Huang J, et al. Cel-
2019;9:282–301. lular heterogeneity during mouse pancreatic ductal adenocarcinoma
23. Elyada E, Bolisetty M, Laise P, Flynn WF, Courtois ET, Burkhart RA, progression at single-cell resolution. JCI Insight 2019;5:e129212.
et al. Cross-species single-cell analysis of pancreatic ductal adeno- 44. K emp SB, Carpenter ES, Steele NG, Donahue KL, Nwosu ZC,
carcinoma reveals antigen-presenting cancer-associated fibroblasts. Pacheco A, et al. Apolipoprotein E promotes immune suppression in
Cancer Discov 2019;9:1102–23. pancreatic cancer through NF-κB-Mediated production of CXCL1.
24. Feig C, Jones JO, Kraman M, Wells RJ, Deonarine A, Chan DS, Cancer Res 2021;81:4305–18.
et al. Targeting CXCL12 from FAP-expressing carcinoma-associated 45. Du W, Menjivar RE, Donahue KL, Kadiyala P, Velez-Delgado A,
fibroblasts synergizes with anti-PD-L1 immunotherapy in pancreatic Brown KL, et al. WNT signaling in the tumor microenvironment
cancer. Proc Natl Acad Sci U S A 2013;110:20212–7. promotes immunosuppression in murine pancreatic cancer. J Exp
25. Zhang Y, Yan W, Collins MA, Bednar F, Rakshit S, Zetter BR, et al. Med 2023;220:e20220503.
Interleukin-6 is required for pancreatic cancer progression by pro- 46. Jin G, Hong W, Guo Y, Bai Y, Chen B. Molecular mechanism of pan-

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

moting MAPK signaling activation and oxidative stress resistance. creatic stellate cells activation in chronic pancreatitis and pancreatic
Cancer Res 2013;73:6359–74. cancer. J Cancer 2020;11:1505–15.
26. Hutton C, Heider F, Blanco-Gomez A, Banyard A, Kononov A, 47. Winkler EA, Bell RD, Zlokovic BV. Pericyte-specific expression of
Zhang X, et al. Single-cell analysis defines a pancreatic fibroblast PDGF beta receptor in mouse models with normal and deficient
lineage that supports anti-tumor immunity. Cancer Cell 2021;39: PDGF beta receptor signaling. Mol Neurodegener 2010;5:32.
1227–44.e20. 48. Long KB, Gladney WL, Tooker GM, Graham K, Fraietta JA, Beatty GL.
27. Huang H, Wang Z, Zhang Y, Pradhan R, Ganguly D, Chandra R, IFNγ and CCL2 cooperate to redirect tumor-infiltrating monocytes
et al. Mesothelial cell-derived antigen-presenting cancer-associated to degrade fibrosis and enhance chemotherapy efficacy in pancreatic
fibroblasts induce expansion of regulatory T cells in pancreatic can- carcinoma. Cancer Discov 2016;6:400–13.
cer. Cancer Cell 2022;40:656–73.e7. 49. Loda A, Heard E. Xist RNA in action: past, present, and future. PLoS
28. DeNardo DG, Ruffell B. Macrophages as regulators of tumour im- Genet 2019;15:e1008333.
munity and immunotherapy. Nat Rev Immunol 2019;19:369–82. 50. Brown CJ, Ballabio A, Rupert JL, Lafreniere RG, Grompe M,
29. Djurec M, Graña O, Lee A, Troulé K, Espinet E, Cabras L, et al. Tonlorenzi R, et al. A gene from the region of the human X inactiva-
Saa3 is a key mediator of the protumorigenic properties of cancer- tion centre is expressed exclusively from the inactive X chromosome.
associated fibroblasts in pancreatic tumors. Proc Natl Acad Sci U S A Nature 1991;349:38–44.
2018;115:E1147–56. 51. Moffitt RA, Marayati R, Flate EL, Volmar KE, Loeza SGH,
30. Fukuda A, Wang SC, Morris JPT, Folias AE, Liou A, Kim GE, et al. Hoadley KA, et al. Virtual microdissection identifies distinct tu-
Stat3 and MMP7 contribute to pancreatic ductal adenocarcinoma mor- and stroma-specific subtypes of pancreatic ductal adenocar-
initiation and progression. Cancer Cell 2011;19:441–55. cinoma. Nat Genet 2015;47:1168–78.
31. Cayrol C, Girard J-P. Interleukin-33 (IL-33): a critical review of its 52. Katz-Kiriakos E, Steinberg DF, Kluender CE, Osorio OA, Newsom-
biology and the mechanisms involved in its release as a potent extra- Stewart C, Baronia A, et al. Epithelial IL-33 appropriates exosome
cellular cytokine. Cytokine 2022;156:155891. trafficking for secretion in chronic airway disease. JCI Insight
32. Kakkar R, Lee RT. The IL-33/ST2 pathway: therapeutic target and 2021;6:e136166.
novel biomarker. Nat Rev Drug Discov 2008;7:827–40. 53. Kouzaki H, Iijima K, Kobayashi T, O’Grady SM, Kita H. The dan-
33. Yeoh WJ, Vu VP, Krebs P. IL-33 biology in cancer: an update and fu- ger signal, extracellular ATP, is a sensor for an airborne allergen and
ture perspectives. Cytokine 2022;157:155961. triggers IL-33 release and innate Th2-type responses. J Immunol
34. Alonso-Curbelo D, Ho Y-J, Burdziak C, Maag JLV, Morris JPT IV, 2011;186:4375–87.
Chandwani R, et al. A gene-environment-induced epigenetic pro- 54. Taniguchi S, Elhance A, Van Duzer A, Kumar S, Leitenberger JJ,
gram initiates tumorigenesis. Nature 2021;590:642–8. Oshimori N. Tumor-initiating cells establish an IL-33-TGF-β niche
35. Alam A, Levanduski E, Denz P, Villavicencio HS, Bhatta M, Alhorebi L, signaling loop to promote cancer progression. Science 2020;369:
et al. Fungal mycobiome drives IL-33 secretion and type 2 immunity eaay1813.
in pancreatic cancer. Cancer Cell 2022;40:153–67.e11. 55. Collins MA, Brisset J-C, Zhang Y, Bednar F, Pierre J, Heist KA, et al.
36. Burdziak C, Alonso-Curbelo D, Walle T, Reyes J, Barriga FM, Metastatic pancreatic cancer is dependent on oncogenic Kras in
Haviv D, et al. Epigenetic plasticity cooperates with cell-cell in- mice. PLoS One 2012;7:e49707.
teractions to direct pancreatic tumorigenesis. Science 2023;380: 56. Mello AM, Ngodup T, Lee Y, Donahue KL, Li J, Rao A, et al. Hypoxia
eadd5327. promotes an inflammatory phenotype of fibroblasts in pancreatic
37. Moral JA, Leung J, Rojas LA, Ruan J, Zhao J, Sethna Z, et al. ILC2s cancer. Oncogenesis 2022;11:56.
amplify PD-1 blockade by activating tissue-specific cancer immu- 57. Martinez-Useros J, Li W, Cabeza-Morales M, Garcia-Foncillas J.
nity. Nature 2020;579:130–5. Oxidative stress: a new target for pancreatic cancer prognosis and
38. Dixit A, Sarver A, Zettervall J, Huang H, Zheng K, Brekken RA, et al. treatment. J Clin Med 2017;6:29.
Targeting TNF-α-producing macrophages activates antitumor im- 58. Hao X, Ren Y, Feng M, Wang Q, Wang Y. Metabolic reprogramming
munity in pancreatic cancer via IL-33 signaling. JCI Insight 2022;7: due to hypoxia in pancreatic cancer: implications for tumor forma-
e153242. tion, immunity, and more. Biomed Pharmacother 2021;141:111798.
39. Zeng Q, Saghafinia S, Chryplewicz A, Fournier N, Christe L, Xie YQ, 59. Abou Khouzam R, Lehn JM, Mayr H, Clavien PA, Wallace MB,
et al. Aberrant hyperexpression of the RNA binding protein FMRP in Ducreux M, et al. Hypoxia, a targetable culprit to counter pancreatic
tumors mediates immune evasion. Science 2022;378:eabl7207. cancer resistance to therapy. Cancers (Basel) 2023;15:1235.
40. Masamune A, Watanabe T, Kikuta K, Satoh K, Kanno A, 60. Eissmann MF, Dijkstra C, Jarnicki A, Phesse T, Brunnberg J, Poh AR,
Shimosegawa T. Nuclear expression of interleukin-33 in pancre- et al. IL-33-mediated mast cell activation promotes gastric cancer
atic stellate cells. Am J Physiol Gastrointest Liver Physiol 2010;299: through macrophage mobilization. Nat Commun 2019;10:2735.
G821–32. 61. Monticelli LA, Osborne LC, Noti M, Tran SV, Zaiss DM, Artis D.
41. Yang X, Chen J, Wang J, Ma S, Feng W, Wu Z, et al. Very-low-density IL-33 promotes an innate immune pathway of intestinal tissue pro-
lipoprotein receptor-enhanced lipid metabolism in pancreatic stellate tection dependent on amphiregulin-EGFR interactions. Proc Natl
cells promotes pancreatic fibrosis. Immunity 2022;55:1185–99.e8. Acad Sci U S A 2015;112:10762–7.

October 2024 CANCER DISCOVERY | 1987

RESEARCH ARTICLE Donahue et al.

62. Li A, Herbst RH, Canner D, Schenkel JM, Smith OC, Kim JY, et al. 82. Peñuelas S, Anido J, Prieto-Sánchez RM, Folch G, Barba I, Cuartas I,
IL-33 signaling alters regulatory T cell diversity in support of tumor et al. TGF-beta increases glioma-initiating cell self-renewal through
development. Cell Rep 2019;29:2998–3008.e8. the induction of LIF in human glioblastoma. Cancer Cell 2009;
63. Mucciolo G, Araos Henríquez J, Jihad M, Pinto Teles S, Manansala JS, 15:315–27.
Li W, et al. EGFR-activated myofibroblasts promote metastasis of 83. Zhang F, Yan Y, Cao X, Guo C, Wang K, Lv S. TGF-β-driven LIF ex-
pancreatic cancer. Cancer Cell 2024;42:101–18.e11. pression influences neutrophil extracellular traps (NETs) and con-
64. Ali S, Mohs A, Thomas M, Klare J, Ross R, Schmitz ML, et al. The tributes to peritoneal metastasis in gastric cancer. Cell Death Dis
dual function cytokine IL-33 interacts with the transcription factor 2024;15:218.
NF-κB to dampen NF-κB–Stimulated gene transcription. J Immunol 84. Shi Y, Gao W, Lytle NK, Huang P, Yuan X, Dann AM, et al. Target-
2011;187:1609–16. ing LIF-mediated paracrine interaction for pancreatic cancer therapy
65. Park JH, Ameri AH, Dempsey KE, Conrad DN, Kem M, Mino- and monitoring. Nature 2019;569:131–5.
Kenudson M, et al. Nuclear IL-33/SMAD signaling axis promotes 85. Halbrook CJ, Lyssiotis CA, Pasca di Magliano M, Maitra A. Pancre-
cancer development in chronic inflammation. EMBO J 2021;40: atic cancer: advances and challenges. Cell 2023;186:1729–54.
e106151. 86. Helms E, Onate MK, Sherman MH. Fibroblast heterogeneity in
66. Gautier V, Cayrol C, Farache D, Roga S, Monsarrat B, Burlet-Schiltz O, the pancreatic tumor microenvironment. Cancer Discov 2020;10:
et al. Extracellular IL-33 cytokine, but not endogenous nuclear IL-33, 648–56.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

regulates protein expression in endothelial cells. Sci Rep 2016;6: 87. Fournié JJ, Poupot M. The pro-tumorigenic IL-33 involved in anti-
34255. tumor immunity: a yin and yang cytokine. Front Immunol 2018;
67. Travers J, Rochman M, Miracle CE, Habel JE, Brusilovsky M, 9:2506.
Caldwell JM, et al. Chromatin regulates IL-33 release and extracellu- 88. Ardito CM, Grüner BM, Takeuchi KK, Lubeseder-Martellato C,
lar cytokine activity. Nat Commun 2018;9:3244. Teichmann N, Mazur PK, et al. EGF receptor is required for KRAS-
68. Zha H, Wang X, Zhu Y, Chen D, Han X, Yang F, et al. Intracellular induced pancreatic tumorigenesis. Cancer Cell 2012;22:304–17.
activation of complement C3 leads to PD-L1 antibody treatment 89. Blasco MT, Navas C, Martín-Serrano G, Graña-Castro O, Lechuga CG,
resistance by modulating tumor-associated macrophages. Cancer Martín-Díaz L, et al. Complete regression of advanced pancreatic
Immunol Res 2019;7:193–207. ductal adenocarcinomas upon combined inhibition of EGFR and
69. Daubon T, Léon C, Clarke K, Andrique L, Salabert L, Darbo E, et al. C-raf. Cancer Cell 2019;35:573–87.e6.
Deciphering the complex role of thrombospondin-1 in glioblastoma 90. Jiang W, Li X, Xiang C, Zhou W. Neutrophils in pancreatic cancer:
development. Nat Commun 2019;10:1146. potential therapeutic targets. Front Oncol 2022;12:1025805.
70. Derynck R, Turley SJ, Akhurst RJ. TGFβ biology in cancer progres- 91. Tay RE, Richardson EK, Toh HC. Revisiting the role of CD4+ T cells
sion and immunotherapy. Nat Rev Clin Oncol 2021;18:9–34. in cancer immunotherapy–new insights into old paradigms. Cancer
71. Zhao H, Chen Q, Alam A, Cui J, Suen KC, Soo AP, et al. The role of Gene Ther 2021;28:5–17.
osteopontin in the progression of solid organ tumour. Cell Death 92. Griffith BGC, Upstill-Goddard R, Brunton H, Grimes GR, Biankin AV,
Dis 2018;9:356. Serrels B, et al. FAK regulates IL-33 expression by controlling chro-
72. Sun Z, Schwenzer A, Rupp T, Murdamoothoo D, Vegliante R, matin accessibility at c-Jun motifs. Sci Rep 2021;11:229.
Lefebvre O, et al. Tenascin-C promotes tumor cell migration and me- 93. Serrels B, McGivern N, Canel M, Byron A, Johnson SC, McSorley HJ,
tastasis through integrin α9β1-mediated YAP inhibition. Cancer Res et al. IL-33 and ST2 mediate FAK-dependent antitumor immune
2018;78:950–61. evasion through transcriptional networks. Sci Signal 2017;10:
73. Yang Y, Cao Y. The impact of VEGF on cancer metastasis and sys- eaan8355.
temic disease. Semin Cancer Biol 2022;86:251–61. 94. Sherman MH, Yu RT, Engle DD, Ding N, Atkins AR, Tiriac H, et al.
74. Calandra T, Roger T. Macrophage migration inhibitory factor: a reg- Vitamin D receptor-mediated stromal reprogramming suppresses
ulator of innate immunity. Nat Rev Immunol 2003;3:791–800. pancreatitis and enhances pancreatic cancer therapy. Cell 2014;159:
75. Bianchi A, De Castro Silva I, Deshpande NU, Singh S, Mehra S, 80–93.
Garrido VT, et al. Cell-autonomous Cxcl1 sustains tolerogenic cir- 95. Kemp SB, Cheng N, Markosyan N, Sor R, Kim IK, Hallin J, et al.
cuitries and stromal inflammation via neutrophil-derived TNF in Efficacy of a small-molecule inhibitor of KRASG12D in immunocom-
pancreatic cancer. Cancer Discov 2023;13:1428–53. petent models of pancreatic cancer. Cancer Discov 2023;13:298–311.
76. Das S, Shapiro B, Vucic EA, Vogt S, Bar-Sagi D. Tumor cell-derived 96. Wang X, Allen S, Blake JF, Bowcut V, Briere DM, Calinisan A, et al.
IL1β promotes desmoplasia and immune suppression in pancreatic Identification of MRTX1133, a noncovalent, potent, and selective
cancer. Cancer Res 2020;80:1088–101. KRASG12D inhibitor. J Med Chem 2022;65:3123–33.
77. Peng DH, Rodriguez BL, Diao L, Chen L, Wang J, Byers LA, et al. 97. Hoshino K, Kashiwamura S, Kuribayashi K, Kodama T, Tsujimura T,
Collagen promotes anti-PD-1/PD-L1 resistance in cancer through Nakanishi K, et al. The absence of interleukin 1 receptor-related T1/
LAIR1-dependent CD8+ T cell exhaustion. Nat Commun 2020;11: ST2 does not affect T helper cell type 2 development and its effector
4520. function. J Exp Med 1999;190:1541–8.
78. Tietze JK, Wilkins DE, Sckisel GD, Bouchlaka MN, Alderson KL, 98. McHedlidze T, Waldner M, Zopf S, Walker J, Rankin AL,
Weiss JM, et al. Delineation of antigen-specific and antigen- Schuchmann M, et al. Interleukin-33-dependent innate lymphoid
nonspecific CD8+ memory T-cell responses after cytokine-based can- cells mediate hepatic fibrosis. Immunity 2013;39:357–71.
cer immunotherapy. Blood 2012;119:3073–83. 99. Todaro GJ, Green H. Quantitative studies of the growth of mouse
79. Zhang Y, Lazarus J, Steele NG, Yan W, Lee HJ, Nwosu ZC, et al. embryo cells in culture and their development into established lines.
Regulatory T-cell depletion alters the tumor microenvironment J Cell Biol 1963;17:299–313.
and accelerates pancreatic carcinogenesis. Cancer Discov 2020;10: 100. Scales MK, Velez-Delgado A, Steele NG, Schrader HE, Stabnick AM,
422–39. Yan W, et al. Combinatorial Gli activity directs immune infiltra-
80. Eickelberg O, Pansky A, Mussmann R, Bihl M, Tamm M, Hildebrand P, tion and tumor growth in pancreatic cancer. PLoS Genet 2022;18:
et al. Transforming growth factor-beta1 induces interleukin-6 e1010315.
expression via activating protein-1 consisting of JunD homodi- 101. Aiello NM, Rhim AD, Stanger BZ, Orthotopic injection of pancreatic
mers in primary human lung fibroblasts. J Biol Chem 1999;274: cancer cells. Cold Spring Harb Protoc 2016;2016:pdb.prot078360.
12933–8. 102. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M,
81. Aoki H, Ohnishi H, Hama K, Shinozaki S, Kita H, Yamamoto H, et al. Pietzsch T, et al. Fiji: an open-source platform for biological-image
Existence of autocrine loop between interleukin-6 and transforming analysis. Nat Methods 2012;9:676–82.
growth factor-beta1 in activated rat pancreatic stellate cells. J Cell 103. Bolte S, Cordelières FP. A guided tour into subcellular colocalization
Biochem 2006;99:221–8. analysis in light microscopy. J Microsc 2006;224:213–32.

1988 | CANCER DISCOVERY October 2024 AACRJournals.org

Stromal IL33 Promotes Pancreatic Cancer Growth RESEARCH ARTICLE

104. Hao Y, Hao S, Andersen-Nissen E, Mauck WM III, Zheng S, Butler A, 108. Miyoshi H, Stappenbeck TS. In vitro expansion and genetic modifi-
et al. Integrated analysis of multimodal single-cell data. Cell 2021; cation of gastrointestinal stem cells in spheroid culture. Nat Protoc
184:3573–87.e29. 2013;8:2471–82.
105. Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck 109. Boj SF, Hwang CI, Baker LA, Chio II, Engle DD, Corbo V, et al.
WM III, et al. Comprehensive integration of single-cell data. Cell Organoid models of human and mouse ductal pancreatic cancer.
2019;177:1888–902.e21. Cell 2015;160:324–38.
106. Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan C-H, 110. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S,
et al. Inference and analysis of cell-cell communication using Cell- et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics
Chat. Nat Commun 2021;12:1088. 2013;29:15–21.
107. Carpenter ES, Kadiyala P, Elhossiny AM, Kemp SB, Li J, Steele NG, 111. Li B, Dewey CN. RSEM: accurate transcript quantification from
et al. KRT17high/CXCL8+ tumor cells display both classical RNA-Seq data with or without a reference genome. BMC Bioinfor-
and basal features and regulate myeloid infiltration in the pan- matics 2011;12:323.
creatic cancer microenvironment. Clin Cancer Res 2024;30:2497– 112. Love MI, Huber W, Anders S. Moderated estimation of fold change and
513. dispersion for RNA-seq data with DESeq2. Genome Biol 2014;15:550.

Downloaded from http://aacrjournals.org/cancerdiscovery/article-pdf/14/10/1964/3500394/cd-24-0100.pdf by guest on 24 November 2024

October 2024 CANCER DISCOVERY | 1989