development of aggressive pancreatic ductal ...development of aggressive pancreatic ductal...

CIR-16-0311; 28/9/2017; 22:56:45

Research Article

Development of Aggressive Pancreatic DuctalAdenocarcinomas Depends on GranulocyteColony Stimulating Factor Secretion inCarcinoma CellsMichael W. Pickup1, Philip Owens2, Agnieszka E. Gorska2, Anna Chytil2, Fei Ye3,Chanjuan Shi4, Valerie M.Weaver1, Raghu Kalluri5, Harold L. Moses2, andSergey V. Novitskiy2

Abstract

The survival rate for pancreatic ductal adenocarcinoma (PDAC)remains low. More therapeutic options to treat this disease areneeded, for the current standard of care is ineffective. Using ananimal model of aggressive PDAC (Kras/p48TGFbRIIKO), we dis-covered an effect of TGFb signaling in regulation of G-CSFsecretion in pancreatic epithelium. Elevated concentrations ofG-CSF in PDAC promoted differentiation of Ly6Gþ cells fromprogenitors, stimulated IL10 secretion from myeloid cells, anddecreased T-cell proliferation via upregulation of Arg, iNOS,VEGF, IL6, and IL1b fromCD11bþ cells. Deletion of csf3 in PDACcells or use of a G-CSF–blocking antibody decreased tumorgrowth. Anti–G-CSF treatment in combination with the DNA

synthesis inhibitor gemcitabine reduced tumor size, increasedthe number of infiltrating T cells, and decreased the number ofLy6Gþ cells more effectively than gemcitabine alone. Humananalysis of human datasets from The Cancer Genome Atlas andtissue microarrays correlated with observations from our mousemodel experiments, especially in patients with grade 1, stage IIdisease. We propose that in aggressive PDAC, elevated G-CSFcontributes to tumor progression through promoting increases ininfiltration of neutrophil-like cells with high immunosuppressiveactivity. Such amechanism provides an avenue for a neoadjuvanttherapeutic approach for this devastating disease. Cancer ImmunolRes; 5(9); 718–29. �2017 AACR.

IntroductionPancreatic ductal adenocarcinoma (PDAC), one of the most

lethal types of cancers, is characterized by stromal expansion withmarked fibrosis. We generated pancreatic epithelium–specific(p48 ¼ Ptf1a) TGFb receptor type II (TbRII) knockout mice inthe context of activated Kras (Kras/p48TbRII-KO) and found thataggressive PDAC develops and recapitulates the histologic fea-tures of the human disease (1). These mice have shorter tumorlatency and survival rates comparedwith Kras/p48mice andothermodels, such as Pdx/Smad4 or Pdx/p53 (KPC; refs. 2, 3). Abro-gation of TGFb signaling in Kras/p48 mice elevates epithelial

STAT3 activity and correlates with decreased survival andadvanced tumor stage in patients with PDAC. Treatment of micewith a STAT3 inhibitor and gemcitabine enhances drug deliveryand therapeutic response through stroma remodeling withoutdepletion of tumor stromal content (4, 5). With STAT3 being acomponent of inflammatory signaling, the influence of theimmune system on the development of these tumors could playa role in the progression of the disease. As such, these PDACtissues show expression of CXCL 1, 2, and 5 chemokines, whichare involved in the chemotaxis of myeloid cells (6). Treatment ofthese mice with a CXCR2 inhibitor decreases tumor progression,angiogenesis, and increases overall survival (6). In KPC mice, thecombination of a CXCR2 inhibitor with a PD1-blocking antibodyimproved T-cell infiltration and extended survival (7). From this,we can conclude that immune cell infiltration plays a rolein tumor progression and also impedes antitumor immunity.Treatment based on stromal depletion in Kras/p48TbRII-KO miceresulted in invasive, undifferentiated tumors with enhanced hyp-oxia, EMT, and decreased animal survival. However, althoughthese mice are not responsive to gemcitabine, anti-CTLA4 immu-notherapy was effective at reducing tumor growth (8). Given theinefficacy of current immune modulating therapies, an in-depthunderstanding of immune interactions as well as the stromalcontext in these models could uncover therapeutic options.

Thepresence and functionofmyeloid cells in tumor tissueplaysa role in the regulationof an antitumorigenic immune response aswell as drug resistance. Tumors developing in the Kras/p48TbRII-KO

mice have an increased number of tumor-associatedmacrophages

1Center for Bioengineering and Tissue Regeneration, Department of Surgery,University of California, San Francisco (UCSF), San Francisco, California.2Department of Cancer Biology and the Vanderbilt-Ingram ComprehensiveCancer Center, Vanderbilt University, Nashville, Tennessee. 3Division of CancerBiostatistics, Department of Biostatistics, Vanderbilt University, Nashville, Ten-nessee. 4Department of Pathology, Vanderbilt University School of Medicine,Nashville, Tennessee. 5Department of Cancer Biology, Metastasis ResearchCenter, University of Texas MD Anderson Cancer Center, Houston, Texas.

Note: Supplementary data for this article are available at Cancer ImmunologyResearch Online (http://cancerimmunolres.aacrjournals.org/).

Corresponding Author: Sergey V. Novitskiy, Vanderbilt University, 2220Pierce Ave, Nashville, TN, 37232. Phone: 615-875-9482; Fax: 615-343-7448; E-mail: [email protected]

doi: 10.1158/2326-6066.CIR-16-0311

�2017 American Association for Cancer Research.

CancerImmunologyResearch

Cancer Immunol Res; 5(9) September 2017718

Research. on September 14, 2020. © 2017 American Association for Cancercancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst August 3, 2017; DOI: 10.1158/2326-6066.CIR-16-0311

http://crossmark.crossref.org/dialog/?doi=10.1158/2326-6066.CIR-16-0311&domain=pdf&date_stamp=2017-9-28

http://cancerimmunolres.aacrjournals.org/

CIR-16-0311; 28/9/2017; 22:56:45

(TAMs) in the tumor tissue and have increased expression ofCXCL1, 2, 5, and 16 chemokines, which affects the functions ofimmune cells (1, 6). Accumulation of TAMs in pancreatic tumortissue induces chemoresistance by reducing gemcitabine-inducedapoptosis via upregulation of cytidine deaminase (9). TargetingTAMs and inflammatorymonocytes by inhibiting either CSF1R orCCR2 decreases the number of tumor-initiating cells andimproves chemotherapeutic response (10). Similarly, blockingmyeloid growth factors, such as M-CSF (orthotopic model) andgranulocytemacrophage colony stimulating factor (GM-CSF;KPCmice), during pancreatic cancer development decreases tumorgrowth and improves responses to T-cell checkpoint immuno-therapy (3, 11). Such evidence supports the idea that TAMsrepresent an impediment toward effective antitumoral immuneresponses. Alternatively, targeted depletion of G-MDSCs by 1A8Ab in KPCmice can increase intratumoral CD8þ T cells, apoptosisof tumor cells, and remodeling of the tumor stroma (12). Inpatients, pancreatic cancer converts CD14þ blood monocytes tomonocyte-like myeloid-derived suppressor cells (M-MDSC) byactivation of the STAT3 pathway. These cells enhance stemnessandmesenchymal properties of tumor tissue (13, 14). Also, moregranulocyte-like MDSCs (G-MDSC), with high expression ofarginase 1, but not M-MDSCs were found in human pancreatictumor tissue versus circulating peripheral blood (15). As thesecells affect T-cell activity, they could provide an additional layer ofT-cell inhibition to be overcome for effective antitumorigenicimmunity to occur. Given the ineffectiveness of checkpoint inhi-bitors in PDAC clinical trials, finding therapeutics to overcomethese additional hurdles is imperative.

In the current study, we investigated the immune responseduring aggressive PDAC development. In Kras/p48 mice withabrogated TGFb signaling (TbRII deletion), we determine thatdifferent subpopulations of immune cells influence tumor pro-gression. We analyzed mechanisms involved in increasing thenumber of Ly6Gþ cells and mechanisms regulating their immu-nosuppressive functions. By using csf3 shRNA and blockingantibody to G-CSF (anti–G-CSF), we demonstrate a positivetherapeutic effect of blocking G-CSF in combination with gemci-tabine. Analysis of human TCGA datasets and tissue microarrays(TMA) helped us to identify a group of patients with similarchanges as we found in the mouse model.

Materials and MethodsCell lines and mice

All studies were performed onKras/p48 andKras/p48TGFbRII-KO

mice, which were established and maintained as describedpreviously (1, 6). p48-Cremicewere used todelete TGFb receptorII in the pancreatic epithelium and crossed with the KrasG12D

mice to promote tumorigenesis. These mice have pure BALB/cbackgrounds and have spontaneous tumor formation in thepancreas by 3 to 4 weeks old. For additional experiments, 8- to10-week-old female C57BL/6 and BALB/c mice were purchasedfrom Harlan Inc.

Kras/p48 and Kras/p48TGFbRII-KO pancreatic tumor cell lineswere derived from primary tumors of Kras/p48 and Kras/p48TGFbRII-KO mice, respectively, established, and cultured inDMEM/10% FBS as described previously (1, 6). These tumorcells were implanted to the pancreas (2.5� 105/25 mL of collagengel) or subcutaneously injected (5 � 105/100 mL of PBS). Forexperiments using MDSCs (CD11bþGr1þ), 4T1 tumor cells (5 �

105/100 mL of PBS) were subcutaneously injected to na€�ve BALB/cmice. The spleen of these mice was isolated on the fourth weekafter tumor injection for future MDSC isolation by magneticmicrobeads (Miltenyi Biotec; ref. 16). 4T1 breast cancer cell line(CRL-2539) was obtained from ATCC and maintained followingstandard protocol. To examine the effect of neutralizing G-CSF ontumor growth, mice were intraperitoneally injected with anti–G-CSF (R&D Systems, Inc.) twice a week, 10 to 50 mg/mouse asdescribed previously (17). Gemcitabine (EMD Millipore) wasinjected intraperitoneally twice a week in 50 mg/kg (6). For CD4and CD8 depletion experiment, CD4 Ab (clone GK1.5), CD8 Ab(clone 2.43), and a rat IgG2b isotype control (clone LTF-2) werepurchased from Bio X Cell. Mice were injected intraperitoneallyonce a week with 100 mg of Ab. The studies were approved byIACUC at Vanderbilt University Medical Center (Nashville, TN).

Flow cytometry analysisTo prepare single-cell suspensions, extracted tumors were

chopped into small pieces, incubated in DMEM with 1 mg/mLcollagenase P (Roche) for 30 minutes at 37�C, and then passedthrough a 70-mm cell strainer. After treatment with FcR BlockingReagent (Miltenyi Biotec Inc.), tumor single-cell suspensions(106 cells/mL) were labeled with specific antibody for 20minuteson ice. All antibodies were obtained from BioLegend. The cellswere analyzed on LSRII flow cytometry (Becton Dickinson), andthe datawere analyzedwith FlowJo software.Nonviable cellswereexcluded by using DAPI. Cell sorting was performed onFACSAriaIII.

Magnetic cell separationTumor-infiltrating myeloid cells (CD11bþ) and MDSCs

(CD11bþGr1þ) were magnetically separated from tumor tissueof pancreas or spleen of tumor-bearing mouse using CD11b orGr1 magnetic microbeads following application protocols of themanufacturer (Miltenyi Biotec Inc.).

Bone marrow isolation and differentiationBone marrow cells were harvested from the femurs of wild-

type mice. Hematopoietic progenitor cells (Lin�) were isolatedusing Lineage Cell Depletion Kit and LS columns from MiltenyiBiotec Inc. according to the manufacturer's instructions. Hemato-poietic progenitor cells were cultured on 24-well plates at 5� 104

cells/mL concentration in RPMI medium containing 10% FBS,20mmol/LHEPES, 50 mmol/L 2-mercaptoethanol, 1� antibiotic-antimycotic solution (Sigma) and supplemented with 25%conditionedmedia frompancreas carcinoma cells. For generationof dendritic cells (DC; for MLR and IL10 studies), whole bonemarrow cells were cultured on 24-well plates at 5� 105 cells/wellconcentration in RPMImedium containing 10% FBS, 20mmol/LHepes, 50 mmol/L 2-mercaptoethanol, 1� antibiotic-antimycoticsolution (Sigma) and supplemented with GM-CSF (20 ng/mL)and IL4 (10 ng/mL; both from R&D Systems, Inc.) for 5 to 6 daysunder humidified atmosphere of air/CO2 (19:1) at 37�C asdescribed previously (18).

Functional T-cell assaysDCs were differentiated as described above. T cells from na€�ve

BALB/c mice were isolated using R&D Systems Mouse T-cellEnrichment columns, according to the manufacturer's protocol.T cells were plated at 105 cells per well in 96-well plate. Each well

G-CSF Regulation of Aggressive PDAC

www.aacrjournals.org Cancer Immunol Res; 5(9) September 2017 719




CIR-16-0311; 28/9/2017; 22:56:45

contained 100,000 T cells, 10,000 DCs, and various numbers ofCD11b cells. Cells were incubated for 72 hours. 3[H]-thymidinewas then added at 1 mCi per 200 mL of cells per well for anadditional 18 hours, followed by cell harvesting and radioactivitycount using a liquid scintillation counter.

For nonspecific T-cell stimulation, T cells from na€�ve BALB/cmice were isolated using R&D Systems Mouse T-cell Enrichmentcolumns, according to the manufacturer's protocol, and mixedwith CD3/CD28 Dynabeads (Invitrogen) according to the man-ufacturer's protocol and different number of CD11bþ cells. Cellswere incubated for 72 hours. 3[H]-thymidine was then added at1 mCi per 200 mL of cells per well for an additional 18 hours,followed by cell harvesting and radioactivity count using a liquidscintillation counter.

Quantitative RT-PCRTotal RNA was extracted from sorted or cultured cells using an

RNeasy Mini Kit (QIAGEN Inc.). cDNA was synthesized usingInvitrogen Superscript First-strand Synthesis System for RT-PCR(Invitrogen Inc.). Primers specific for IL4, IL10, IL6, IL2, IFNg ,TNFa, FoxP3, IL17a, Arg, iNOS, VEGF,MMP2,MMP9, andG-CSFwere used, and relative gene expression was determined using ABIPRISM 7900HT Sequence Detection System (PE Applied Biosys-tems). The comparative threshold cycle method was used tocalculate gene expression normalized to b-actin as a gene refer-ence. Primer sequences are available upon request.

Tissue microarray of pancreatic tissuesA TMA was purchased from US Biomax and contained 10

normal pancreases, 11 chronic pancreatitis, 42 PDAC, and 20adjacent normal pancreas tissues. TMA slides were concurrentlyevaluated by Leica Biosystems software and by the pathologist(C. Shi; ref. 4). Nuclear and cytoplasmic staining were scored asfollows: the staining index was considered as the sum of theintensity score (0, no staining; 1þ, weak; 2þ, moderate; 3þ,strong) and the distribution score (0, no staining; 1þ, stainingof <33% of cells; 2þ, between 33% and 66% of cells; and 3þ,staining of >66% of cells). Staining indices were classified asfollows: 3þ or higher, strong staining; 1þ to 2þ, weak staining;and 0, negative staining. CD15 and G-CSF were scored as positiveif any detectable cytoplasmic staining was present.

ELISA and protein arrayCytokine levels in conditioned media of tumor cells, DCs, and

shRNA transformed tumor cells were measured using the MouseG-CSF and IL10 ELISA Kits (R&D Systems) following the manu-facturer's protocol. Proteome Profiler Mouse XL Cytokine Arraywith 111different cytokine antibodies (R&DSystems)was used tomeasureproteins in conditionedmedia isolatedCD11b cells frompancreas tumor tissue following the manufacturer's protocol.

Histology and IHCTissues were embedded directly in an optimal cutting temper-

ature compound without fixation or placement in 10% formalinovernight and then embedded in paraffin and sectioned at 5 mm.Sections were dewaxed in xylene and rehydrated in successiveethanol baths. For IHC, the MOM Kit was used (Vector). Ki67,CD3, CD31, F4/80, and CD3 staining were performed in Trans-lational Pathology Shared Resources (Vanderbilt University,Nashville, TN). IHC sections were photographed using the AXIO

Scope A1 ZEISSmicroscope and AxioVision 4.8 software. HumanTMA slides were stained according manufacture protocol forCD15 (Abcam) and G-CSF (Abcam) and then were scannedby using Leica SCN400 Slide Scanner with 40� objective.Quantification of IHC was performed using NIH ImageJ(http://rsbweb.nih.gov/ij/docs/examples/stained-sections/index.html) as described previously (19).

ShRNA knockdownPrepackaged lentiviral transduction particles containing veri-

fied MISSION shRNA constructs targeting the coding domainsequence of G-CSF and MISSION Non-Target shRNA ControlTransduction Particles in pLKO.1-puro plasmid vectors werepurchased from Sigma-Aldrich. Cells were seeded at a density of1 � 103 per well in a sterile 96-well plate and incubated for 24hours before addition of twomultiplicity of infection of lentiviralparticles. Cells were maintained in 2 mg/mL puromycin media.Knockdown levels were confirmed by ELISA.

Statistical analysisData were presented as mean � SEM. Multiple comparisons

between treatment groups and control untreated group wereperformed using one-way ANOVA, followed by Dunnett proce-dure for multiplicity adjustment. Two-group comparison wasperformed using two-sample t tests. Correlation analyses for thegene expressionbetween the signatures of interestwere performedusing data representing 185 patients from TCGA datasets. Thegene expression was normalized across all arrays. Gene symbolswere assigned using the manufacturer-provided annotation, butthe analysis was performed at probe level. The associationbetween RFS and Smad4 was analyzed using the univariate Coxproportional hazards regression model. When an interaction wasidentified, a Kaplan–Meier survival curvewas created for the "highversus low" of the Smad4 gene expression by dichotomizing allthe samples among the subtype (Fig. 4). All tests were two-tailed.

ResultsTumor development correlates with excess G-CSF

Fibroblast function and mechanical properties influence Kras/p48TbRII-KO tumor progression (1, 6, 8). We sought to examinechanges in immune infiltration and function during PDAC devel-opment; thus, we performed flow cytometry analysis of pancreastumor tissue from Kras/p48TbRII-KO animals in comparison withtissue from Kras/p48 mice. We observed an increase in thenumber of immune cells (CD45þ) in pancreatic tissue of Kras/p48TbRII-KO versus Kras/p48 and versus normal tissue (Fig. 1A).After analyzing different subpopulations of immune cells, wefound more CD11bþGr1þ cells in pancreatic tissue, spleen, andbone marrow (Fig. 1B and C). There is a significant increasenumber of CD11bþGr1þ cells in spleen of Kras/p48TbRII-KO miceversus Kras/p48 and versus control mice (P < 0.05) and also inKras/p48mice versus controlmice (P<0.05). In thebonemarrow,we found more of these cells only in Kras/p48TbRII-KO mice (P <0.05). As Gr1 Ab recognizes both Ly6C and Ly6G proteins on thecell surface, we carried out additional flow cytometry analysis ofthe Gr1þ population of cells. We discovered an increased numberof CD11bþGr1high cells that represent Ly6Gþ cells (neutrophils,G-MDSCs). However, there was not an increase in Ly6Cþ cells(Fig. 1D). Pancreatic carcinoma cell lines with deleted TbRIIisolated from mice of mixed genetic background (C57BL/6,

Pickup et al.

Cancer Immunol Res; 5(9) September 2017 Cancer Immunology Research720



http://rsbweb.nih.gov/ij/docs/examples/stained-sections/index.html

http://rsbweb.nih.gov/ij/docs/examples/stained-sections/index.html


CIR-16-0311; 28/9/2017; 22:56:46

0

1

2

3

4

0

2

4

6

8

10

Kras/p48TGFβRII-KO

0

5

10

15

20

25

C

Normal pancreas Kras/p48

A

Kras/p48TGFβRII-KO

B

Gr1

CD

11b

Normal pancreas Kras/p48

Kras/p48TGFβRII-KO

E

F

Normal pancreas Kras/p48 Kras/p48TGFβRII-KO

6.21%13.5%

20.0%

3.0% 3.34% 8.12%

% o

f CD

45

CD11b+Gr1high

**

0

10

20

30

%

CD45

*

*

ng/m

L/m

g of

Pro

tein

G-CSF, ELISA

0

200

400

600

800

1×106 2.5×105

Number of cells/well

ng/m

L

G-CSF, ELISA

Kras/p48

CD11b

Kras/p48 Kras/p48TGFβRII-KO

33.2% 41.6%

0

10

20

30

40

50CD11b+

* Ly6G49.1

Ly6C10.1

Ly6C11.7

Ly6G60.4

Ly6C

Ly6G

0

20

40

60

80

%

% o

f CD

11b

*

Ly6G+Ly6Clow

*

**

**

*

***

2.34% 3.12%

CD11b+Gr1low

D

Tum

orS

plee

nB

one

mar

row

Figure 1.

Expansion of Ly6Gþ cells in aggressive PDAC. A, Flow cytometry analysis of CD45þ cells in normal, Kras/p48, and KRas/p48TGFbRII-KO tissue and percent toDAPI� live cells quantification. Plots are gated as DAPI�. Ten mice per group were analyzed. � , P < 0.05. B, IHC for Gr1 (RB6-8C5) of normal pancreas,pancreas isolated from Kras/p48 mice, and Kras/p48TGFbRII-KO mice at 5 to 6 weeks of age. C, Representative flow cytometry plots of CD11bþGr1þ cells in pancreatictissue, spleen, and bone marrow from normal, Kras/p48, and KRas/p48TGFbRII-KO mice. Plots are gated as CD45þDAPI�. Ten mice per group were analyzed.D, Quantitative data of Gr1-high (solid line) and Gr1-low (dotted line) cells in tumor tissue showed on C (top). � , P < 0.05; �� , P < 0.01. Ten mice per group wereanalyzed. E, ELISA data of G-CSF levels in pancreatic tissue (left) and basal secretion in cultured cell lines (right). Data correspond to the mean � SEM of3 individualmice from two experiments. Pancreatic cell lines fromKras/p48mice andKras/p48TGFbRII-KOmicewere cultured in 6-well plate in 3mL ofDMEM/10%FBSfor 24 hours and level of G-CSF was measured in conditioned media. n ¼ 3. � , P < 0.05; �� , P < 0.01. F, Flow cytometry analysis of CD11bþ cells and Ly6Gversus Ly6C cells (gated as CD11bþ) after differentiation of bonemarrow progenitors (Lin� cells) for 5 days in the presence of 25% conditionedmedia from Kras/p48and Kras/p48TGFbRII-KO pancreatic carcinoma cell lines. Data correspond to the mean � SEM of 3 individual mice from two experiments. � , P < 0.05.






CIR-16-0311; 28/9/2017; 22:56:50

129) have excess CXCL1/2/5/16, VCAM, and G-CSF (6). Wevalidated these findings in our model using ELISA to measurelevels of G-CSF in tissue and established carcinoma cell linesisolated from BALB/c animals. We observed increased G-CSFprotein in tissue of Kras/p48TbRII-KOmice aswell as in conditionedmedia from cultured pancreas carcinoma cell lines (Fig. 1E). Todetermine the functionality of this altered cytokine profile inpromoting the differentiation of Ly6Gþ cells, we supplementedprogenitors (Lin� cells) derived from bone marrow with tumorcell conditionedmedia during their differentiation. After 5 days oftreatment, we found more myeloid (CD11bþ) and Ly6Gþ cells(Fig. 1F) if the differentiation was supplemented with Kras/p48TbRII-KO tumor cell conditioned media than if the differenti-ation was supplemented with Kras/p48 cell conditioned media.Thus, TGFb may regulate G-CSF in pancreas epithelium, leadingto accumulation of Ly6Gþ cells during tumor progression.

G-CSF upregulates immunosuppressive functions of myeloidcells

Ly6Ghigh cells isolated from tumors can be neutrophils orT-cell–suppressive G-MDSCs (20). To analyze the functions ofthese infiltrating immune cells in PDAC tissue, we FACS sortedvarious immune cell populations from tumor tissue to determinecellular function (Fig. 2A). qPCR analysis of sorted T cells (CD3þ)showed shifts in gene expression toward a Th2 response throughan increase in IL4, IL10, and IL6 expression (Fig. 2B). Markers of aTh1 response (IFNg), Tregs (Foxp3), and Th17 (IL17) were notchanged, but IL2 was downregulated (Fig. 2B). Analysis of sortedLy6Gþ cells showed increases in genes corresponding to M2-like/protumorigenic responses, increased gene expression of iNOS,Arg, IL6, IL10, VEGF, and MMP9; however, expression of MMP2was decreased (Fig. 2C). Gene expression changes in Ly6Cþ

population of cells and Ly6C�Ly6G� (MF, DC) were similar toLy6Gþ cells.

G-CSF was increased in pancreas tumor tissue and culturedpancreas carcinoma cell lines (Fig. 1). Increased tissue stiffness inKras/p48TbRII-KO mice can also elevate amounts of G-CSF (5). Wesought to address the functional role for G-CSF in activity ofLy6Gþ cells. First, we used a model for enrichment and isolationof MDSCs (CD11bþGr1þ); we implanted the 4T1 tumor cell lineinto BALB/c mice and isolated Gr1þ cells from spleen of thesetumor-bearing mice. Isolated MDSCs were incubated with con-ditionedmedia collected fromeither Kras/p48or Kras/p48TbRII-KO

pancreas carcinoma cells. Secreted factors Arg, iNOS, IL6, andVEGFwere upregulated from the tumor cells when incubatedwiththe G-CSF–rich carcinoma cell conditioned media (Fig. 2C).However, no differences in IL10, MMP2, and MMP9 expressionwere found. Second, using the same source of MDSCs, we incu-bated them with various concentration of G-CSF and foundsimilar changes in Arg, iNOS, IL10, MMP2, MMP9, and VEGFexpression (Fig. 2C). These data highlighted the potential influ-ence of IL10, as G-CSF regulates Th1/Th2 cells and influences thealternative polarization of macrophages through upregulation ofIL10 (21).Weperformed analysis ofmyeloid cell secretion of IL10in response to conditioned media from carcinoma cells. To dothis, we differentiated DCs from bone marrow progenitors andincubated them with conditioned media collected from pancre-atic carcinoma cell lines. We found that carcinoma cells bythemselves do not produce IL10 (Supplementary Fig. S1A). How-ever, conditioned medium from Kras/p48TbRII-KO cell line stim-ulated secretion of IL10 frommyeloid cells. This effect is primarily

caused by G-CSF, as adding anti–G-CSF to the medium blockedupregulation of IL10 (Supplementary Fig. S1A). No effect onstimulation of IL10 secretion was observed using conditionedmedia from Kras/p48 carcinoma cell lines.

To analyze the ability of myeloid cells from PDAC tissue tomodulate T-cell proliferation, we stimulated T-cell proliferationwith CD3/CD28 beads and mixed them with CD11bþ cellsisolated from Kras/p48TbRII-KO and Kras/p48 tumors with andwithout blocking G-CSF Ab. In addition, we stimulated T-cellproliferation with DCs from a different mouse background (allo-geneic MLR) and mixed in CD11bþ cells from the same tumors(Supplementary Fig. S1B). We discovered that myeloid cells(CD11bþ) from Kras/p48TbRII-KO PDAC tissue were better T-cellproliferation suppressors than CD11bþ cells isolated from Kras/p48 mice (Fig. 2D) that were dependent on G-CSF. We concludethat increased immunosuppressive function of myeloid cells inour PDAC in model was caused by G-CSF stimulation.

Inhibition of G-CSF suppresses pancreatic tumor growthAnti–G-CSF treatments decrease tumor growth of orthotopic

lymphoma and lung cancer models (17). Anti–GM-CSF limitstumor growth in KPC mice (3), a model of spontaneous PDA inwhich expression of oncogenic KrasG12D and mutant p53R172H istargeted to the pancreas by Cre recombinase under the control ofthe pancreas-specific promoter Pdx-1. Moreover, inhibition ofmacrophage accumulation in tumor tissue by inhibition ofCSF1Rand CCR2 decreases tumor-initiating cells and improves chemo-therapeutic responses in PAN02, KCM, and Kras-INK pancreastumor models (10). We subcutaneously implanted Kras/p48 andKras/p48TbRII-KO pancreas carcinoma cell lines and treated micewith anti–G-CSF (Fig. 3A). After 4weeks of anti–G-CSF treatment,the average Kras/p48TbRII-KO tumor weight from mice receivingblocking Ab was decreased compared with mice receiving controlIgG Ab injections (Fig. 3B). Tumor size of Kras/p48 versus Kras/p48TbRII-KO was lower, which correlates with our data from Kras/p48TbRII-KO mice with spontaneous formation of pancreatictumors (1, 6). This model showed no effect of G-CSF blockingAb on tumor weight of Kras/p48 tumors. IHC analysis of tumortissue showednodifferences inCD3 staining but less proliferation(Ki-67) and angiogenesis (CD31) in Kras/p48TbRII-KO mice withanti–G-CSF treatment compared to control group (Fig. 3C).Corresponding with the tumor weight data, no differences innumber of CD11bþLy6Gþ cells were observed in Kras/p48 ani-mals (Fig. D). The number of CD11bþLy6Gþ cells in tumor tissuewas decreased in Kras/p48TbRII-KO mice with no changes in num-ber of macrophages (Fig. 3D, E). Because of the magnitude ofchanges in tumor growth observed in the Kras/p48TbRII-KO com-pared with growth in the Kras/p48 mice and no differences inanti–G-CSF Ab treatment, we decided to perform the next experi-ments using only the Kras/p48TbRII-KO cell line. We found nodifference in blocking effect of anti–G-CSF in the subcutaneousmodel or the orthotopic transplantation model (SupplementaryFig. S2A and S2B). Because of the stress placed on themice duringthe orthotopic transplants as well as the immune responseobserved from the tissue manipulation, our subsequent experi-ments used only the subcutaneous tumor model.

G-CSF can be secreted bymany types of cells, includingmyeloidcells, fibroblasts, and endothelial cells. To determine how impor-tant epithelialG-CSF secretion is,weused shRNA to targetG-CSF inKras/p48TGFbRIIKO cell lines (SupplementaryFig. S2C).Threeweeksafter tumor cells were subcutaneously injected, we found that the

Pickup et al.





CIR-16-0311; 28/9/2017; 22:56:50

1:1 2:1 4:1 8:10

10,000

20,000

30,000

40,000

*

Kras/p48TGFβRII-KOKras/p48

Kras/p48 RII-KO TGFβ + anti–G-CSF AbKras/p48 + anti–G-CSF Ab

*

A

0.00

0.02

0.04

0.06

0.00

0.01

0.02

0.03

0.04

0.05

0.00

0.01

0.02

0.03

0.04

0.05

0.0000

0.0002

0.0004

0.0006

0.0008

0.000

0.001

0.002

0.003

0.004

0.00

0.01

0.02

0.03

0.04

0.000

0.005

0.010

0.015

0.000

0.001

0.002

0.003

0.004

B

ΔΔC t

, mRN

AΔΔ

C t, m

RNA

IL2IL6IL10IL4

IFNγ TNFα Foxp3 IL17a

Kras/p48TGFβRII-KOKras/p48

* **

C

*

D

Sor

ted

cells

from

tum

or ti

ssue

Con

ditio

ned

med

iaG

-CS

F

0.000

0.001

0.002

0.003

0.00

0.01

0.02

0.03

0.04

0.05

0.000

0.001

0.002

0.003

0.004

0.005

0.000

0.005

0.010

0.015

0.020

0.000

0.002

0.004

0.006

0.008

0.000

0.002

0.004

0.006

0.008

0.0

0.1

0.2

0.3

0.4

0.00

0.02

0.04

0.06

0.08

0.000

0.002

0.004

0.006

0.008

Kras/p48 RII-KOTGFβ

Kras/p48

IL10IL6ArgiNOS VEGF TNFα MMP9MMP2

* * ***

*

*

*

0.000

0.005

0.010

0.015

0.020

0.025

0 50100 5010 0 5010

ND

**

G-CSF, ng/mLG-CSF, ng/mL G-CSF, ng/mL

*

** * * *

ΔΔC t

, mRN

AΔΔ

C t, m

RNA

ΔΔC t

, mRN

A

0.00

0.05

0.10

0.15

0.000

0.002

0.004

0.006

0.008

0.000

0.005

0.010

0.015

0.020

0.025

0.00

0.02

0.04

0.06

0.08

0.00

0.02

0.04

0.06

0.08

0.000

0.002

0.004

0.006

0.00

0.05

0.10

0.15

0.20


Kras/p48 Kras/p48 RII-KOTGFβ






Kras/p48

0.000

0.001

0.002

0.003

0.00

0.02

0.04

0.06

0.08**








Kras/p48

0.00000

0.00002

0.00004

0.00006

0.00008

0 5010

G-CSF, ng/mL

0.0

0.1

0.2

0.3

0 5010

G-CSF, ng/mL

*

*

0.00

0.01

0.02

0.03

0.04

0.05

0 10 50

G-CSF, ng/mL

*

0.00

0.02

0.04

0.06

0 5010

G-CSF, ng/mL

*

*

0.000

0.005

0.010

0.015

0.020

0 5010

G-CSF, ng/mL

Single-cell suspension(CD45+DAPI-)

Proliferation, T cells

CP

M. H

[3]

T cells : CD11b cells

Figure 2.

Functions of myeloid cells. A, Cell sorting strategy. B, T-cell (CD3þ) gene expression profile. Cells were sorted as shown in A. Top row, gene expressionfor Th2 cytokines presented as DDCt normalized to Gapdh; bottom row, gene expression for Th1 cytokines presented as DDCt normalized to Gapdh. Datacorrespond to the mean� SEM of 3 individual mice from two experiments. � , P < 0.01. C,Gene expression analysis of protumorigenic cytokines previously identifiedas cytokine mediators of MDSC immune suppression. Top, gene expression measured by qPCR on sorted Ly6G; middle, gene expression analysis of Gr1þ

cells, from the spleen of 4T1 tumor-bearingmice, treatedwith conditionedmedia fromKras/p48 (openbars) or fromKras/p48TGFbRII-KO (closedbars) carcinoma cells;bottom row, gene expression analysis of Gr1þ cells, from the spleen of 4T1 tumor-bearing mice, treated with G-CSF, 20 ng/mL for 6 hours. Data correspondto the mean � SEM of 3 individual mice from two experiments. �� , P < 0.05; � , P < 0.01. D, T cells (CD3þ) were isolated by negative selection from spleen ofna€�ve BALB/c mice, mixed with CD11b cells isolated from pancreatic tissue by using CD11b magnetic microbeads (Miltenyi Biotec), anti–G-CSF was used inconcentration 40 ng/mL. Data correspond to the mean � SEM of three individual mice. � , P < 0.05.






CIR-16-0311; 28/9/2017; 22:56:51

A

0 1 32Weeks

4

s.c. tumorinjection

Micesacrificei.p. anti–G-CSF Ab

shR

NA

cont

rol

shR

NA

G-C

SF

0

50

100

150

200

Tumor weight

mg

shRNAG-CSF

shRNAcontrol

P = 0.006

F

B

0.0

0.5

1.0

1.5

Tumor weight

P = 0.036

Anti–G-CSFIgG control

Kras/p48TGFβRII-KO Kras/p48

g

E

0

2

4

6

8

10

Cel

ls, x

106

Macrophages G-MDSC

P = 0.021

Cells in tumor tissue

Ly6G

CD

11b

IgG Anti–G-CSF IgG Anti–G-CSF

Kras/p48TGFβRII-KO Kras/p48D

Ly6G

shRNA G-CSF

6.6 %shRNA control

Ly6C

36.0 %G

CCL5/RANTES

CCL6/C10

CCL12/M

CP-5

CCL22

CXCL1/KC

CXCL2/MIP-2

CXCL13

CXCL16IL-1r

aLIX

MMP-9

Myelopero

xidas

e

Osteopontin

(OPN)

0.1

1

10

Fol

d ch

ange

s

Protein arrayH

0

200

400

600

800

shRNA-control shRNA-GCSF

CD4 CD8 depletionIgG control

P = 0.34

P = 0.011

P = 0.009

P = 0.023

Tumor weight

mg

J

0 1 32

Weekss.c. tumor Mice

sacrifice

GEM (50 mg/kg)Anti–G-CSF (50 μg/mouse) U

ntre

ated

255 mg

GE

MG

EM

+an

ti–G

-CS

F

168 mg

121 mg

Averageweight

P = 0.23

Tumor weight

P = 0.049

P = 0.055

P = 0.008

0

100

200

300

K

0

5

10

15

20

0

5

10

15%

Are

a po

sitiv

e fo

r K

i67

IgG Anti–G-CSF

Ki67 Ki67

CD3 CD3

CD31 CD31

0

1

2

3

4

5

% A

rea

posi

tive

for

CD

3%

Are

a po

sitiv

e fo

r C

D31

P = 0.031

P = 0.25

P = 0.012

C

shRNA control shRNA G-CSF

F4/80 F4/80

CD3 CD3

I

Pickup et al.





CIR-16-0311; 28/9/2017; 22:56:59

tumor weight was decreased in mice with G-CSF knockdowntumors (Fig. 3F). Using flow cytometry analysis of G-CSF–KOtumor tissue, we found Ly6GþLy6Clow cell numbers were reducedby approximately 84% compared with control tumors (Fig. 3G).To determine whether G-CSF knockdown also altered myeloidcell functions,we isolatedCD11bþ cells fromthe implanted tumortissue andmeasured concentrations of secreted proteins.We foundthat myeloid cells isolated from tumor tissue with knockdownof G-CSF produced less protumorigenic proteins, such as CXCL1(KC), CXCL5 (LIX), MMP9, MPO, and CCL5 (RANTES; Fig. 3H).IHCanalysis showedmoreF4/80þ cells andmoreT cells (CD3þ) inG-CSF knockdown tumor tissue (Fig. 3I). On the basis of in vitroimmunosuppressive function ofG-MDSC cells isolated fromKras/p48TbRII-KO mice indicated above (Fig. 2), we sought to establishthe role of T cells during Kras/p48TbRII-KO tumor progression. Wesubcutaneously implanted Kras/p48TbRII-KO shRNA control andshRNA G-SCF pancreatic carcinoma cell lines into mice withdepleted CD4 and CD8 cells (Fig. 3J; Supplementary Fig. S2D).We discovered that depletion of T cells reverted the growth inhi-bition effect in shRNA G-CSF cell lines, demonstrating the immu-nosuppressive role of G-CSF in establishing a protumorigenicmicroenvironment, which promotes tumor progression.

Typical treatment of pancreatic cancer involves gemcitabine(22), even though this and other chemotherapies are not veryeffective and myeloid cells can mediate gemcitabine resistance(9). We sought to determine whether neoadjuvant myeloiddepletion using anti–G-CSF would enhance the efficacy of gem-citabine. Thus, we subcutaneously implanted Kras/p48TGFbRIIKO

cell lines and let the tumors grow for one week (Fig. 3K). A weekafter implantation, we began treatment with anti–G-CSF Ab and/or gemcitabine twice a week for the next 2 weeks. Mice weresacrificed 3 weeks after implantation. Tumor growth with gemci-tabine alone or gemcitabine in combinationwith anti–G-CSFwasdecreased when compared with untreated mice. Tumor weightdecreased more in mice receiving gemcitabine and anti–G-CSFthan in mice receiving chemotherapy alone. Treatment with onlyanti–G-CSF did not affect tumor growth or tumor weight. Fromthis, we conclude that G-CSF inhibition during pancreatic tumorgrowth not only decreases the number of granulocytes and con-verts TAMs to an M1-like (antitumorigenic) type, but alsoincreases effectiveness of chemotherapy.

G-CSF in human pancreatic cancersTo determine the relevance of these findings to human pan-

creatic cancer, we analyzed publicly available TCGA gene expres-sion profiles of pancreatic cancer tissues with clinical data relatedtometastasis, tumor stage, and tumor grade over a 10-year period.As our model of PDAC showed that TGFb signaling abrogation in

pancreatic epithelial cells resulted in immune modulation, wesought to correlate expression of Smad4, a key gene of TGFbsignaling, with patient survival. We found that survival isdecreased with low expression of Smad4 (Fig. 4A). Although nodifferences were found in patients with low versus high metasta-sis, expression of Smad4 lowered as tumor stage and gradeincreased (Fig. 4B). Next, we used a TMA of human pancreatictissue with 208 tissue cores (US Biomax, Inc) to determinedifferences in neutrophil infiltration. TMAs were stained forCD15, a marker of neutrophils, and for G-CSF (Fig. 4C). TMAIHC slides were validated by Leica software in parallel with apathologist. We found that tumor tissue has a high number ofCD15þ cells when compared with normal tissue, chronic pan-creatitis, or adjacent normal tissue (ANT). Segregating tumors bystage and grade, we found that tumors with grade 2 and 3 or stageII and III have the highest number of CD15þ cells (Fig. 4D). In theanalysis of G-CSF, we found a lower level of this protein in tumortissue compared with normal, CP, or ANT likely due to morefibrosis in tumor tissue and less epithelial cells that secrete G-CSF.However, epithelial scoring showed more G-CSF in tumor epi-thelium than in normal tissue (Fig. 4E). Analysis of differentstages and grades showedus that tumorswith grade3, stage II havemore G-CSF (Fig. 4F). Correlation analysis between CD15 andG-CSFwas only in grade 1 tumors (r¼ 0.75, P¼ 0.0002,a¼ 0.05)and stage II tumors (r¼ 0.36, P¼ 0.049,a¼ 0.05; SupplementaryFig. S3). These findings indicate that in human pancreatic tumors,as well as in mouse models, decreased TGFb signaling wasassociated with increases in the number of neutrophils (CD15þ

cells) and level of G-CSF in malignant pancreatic epithelium.Our data suggest that downregulation of TGFb signaling

in human pancreatic cancer (usually through inactivatingmutations in Smad4) is associated with decreased survival ofpatients and with increased G-CSF levels with accompanyingincreased number of CD15 cells, which, based on our mousemodel, have tumor-promoting properties. Collectively, ourfindings indicate a strategy for targeting pancreatic tumorgrowth by blocking G-CSF protein, especially in patients withgrade 1/stage II tumors.

DiscussionTGFb signaling promotes PDAC progression, as indicated by

the fact that Smad4,which encodes a signalmediator downstreamfromTGFb, is deleted ormutated in 55%of patients (23), and lowexpression predicts poor patient survival (Fig. 4A). We generatedan animalmodel of aggressive ductal carcinoma by deleting TGFbreceptor II (TbRII) in the pancreatic epithelium (p48) alongwith aKrasG12D mutation (1). These mice developed well-differentiated

Figure 3.A, B,Schematic of treatmentwith either control (openbars) or anti–G-CSF (10mg/mouse, MAB414; closed bars) andKRas/p48TGFbRII-KO or Kras/p48, tumorweight 4weeks after s.c. cell injection (5 � 105), n ¼ 5. C, IHC for Ki-67, CD3, and CD31 of tumor tissue isolated from mice treated IgG (control group, left column)andmice treated with anti–G-CSF (right column) 4weeks after tumor injection.D, Representative flow cytometry plots of tumor infiltrated immune cells in pancreastissue. Cells are gated as CD45þ.E,Quantitative data for the presence ofmacrophages andG-MDSC cells in pancreas tissue of Kras/p48TGFbRII-KOmicewith IgG (openbars) and anti–G-CSF treatment (closed bars), F, representative photograph of tumor tissue isolated from mice 3 weeks after s.c. injection of shRNA control(clone 1) and shRNA G-CSF (clone 1) carcinoma cells (left) and tumor weight (right), n ¼ 5. G, Representative flow cytometry plots of Ly6G/Ly6C analysis ofCD45þCD11bþ cells from tumor tissue shown in F. H, Cytokine array data analysis. CD11bþ cells were isolated from tumor tissue showed on F for collection ofconditioned media. Data shows fold changes in shRNA G-CSF tumors vs. shRNA control. Graph shows only proteins with statistically significant changes fromtotal 111 proteins (see details in Material and Methods section), n¼ 3. I, IHC for F4/80 and CD3 of tumors shown in F. J,Weight of tumor tissue isolated from mice 4weeks after s.c. injection of shRNA control (clone 2) and shRNA G-CSF (clone 5) carcinoma cells treated in parallel with IgG control (open bars) or depletiondue to use of CD4 and CD8 antibody (closed bars). K, Scheme of mice treatment with gemcitabine (GEM) and anti–G-CSF. Tumor tissue isolated frommice 3 weeksafter s.c. injection of Kras/p48TGFbRII-KO carcinoma cell line and treatment during last 2 weeks and tumor weight, n ¼ 5. Scale bars, 100 mm.






CIR-16-0311; 28/9/2017; 22:56:59

PDAC with 100% penetrance and a median survival of 59 days.The clinical and histopathologic manifestations of the combinedKrasG12D and deletion of TbRII in mice recapitulated humanPDAC. Using the same mouse model in comparison with KPCmice, it was found that the loss of TGFb signaling engages apositive feedback loop whereby epithelial STAT3 signaling mod-ulates pancreatic cancer malignancy by increasing matricellularfibrosis and tissue tension (8). In PDAC patient biopsies, highermatricellular protein deposition and increased activated Stat3 isassociated with SMAD4 mutation and shorter survival (5). Dele-tion of TbRII in pancreatic epithelium is associated with increasedexpression of CXCL1/5/16 chemokines, which in turn induceCTGF expression from CXCR2þ pancreatic stromal fibroblasts.Using a CXCR2 inhibitor reduces tumor progression in the Kras/p48TGFbRIIKO model (6). With such a robust change in inflamma-tory signaling and the efficacy of immune modulating CXCR2treatments, using thismodel, we could dissect themechanisms bywhich tumors circumvent antitumoral immunity and developstrategies to combat them. Pancreatic cancer is associated withdesmoplasia,which is anegative prognostic factor for patients (8).In Kras/p48TbRII-KOmice, depletion of fibroblasts in the context ofpancreatic cancer led to invasive, undifferentiated tumors withdiminished survival (6). Taken together, these data support theconcept that alterations in TGFB signaling may be used as adeterminant for differential therapeutic options. Given the stro-mal response observed in these models, targeting the alteredcytokine profile resulting from attenuation of the signaling path-way could be efficacious.

STAT3 is themain transcription factor regulating the expansionof MDSCs (CD11bþGr1þ) in tumor tissue (24). Ablation ofSTAT3 expression in conditional STAT3 knockout mice or selec-tive STAT3 inhibitors reduces the expansion of MDSCs andincreases T-cell responses in tumor-bearing mice (24). Tumor-induced STAT3 activation in MDSCs enhances stemness andmesenchymal properties in human pancreatic cancer (13).MDSCs consist of two subsets of CD11bþLy6GþLy6Clow granu-locytic and CD11bþLy6G�Ly6Chigh monocytic cells. Almost alltumors demonstrate expansion of the granulocytic subset ofMDSCs (16). In our study, we also found an increased numberof granulocytic (Ly6Gþ) MDSCs in pancreatic tissue (Fig. 1) inparallel with increased secretion of G-CSF from the pancreaticepithelium with deleted TbRII when compared with cells withintact TGFb signaling. In vitro studies demonstrated the ability ofconditioned media from Kras/p48TGFbRIIKO carcinoma cells toincrease differentiation of CD11bþ cells, especially the Ly6Gþ

subset. These cells in pancreatic tumors have increased expressionof immunosuppressive genes, and they suppress T-cell prolifer-ation in vitro and elicit a polarized response toward protumori-genic Th2 cells (Fig. 2). We found that secreted G-CSF frompancreatic epithelial cells upregulates secretion of IL10 frommyeloid cells. This protein is an immunosuppressive cytokinewith the ability to switch responsive myeloid cells fromM1 toM2and Th cells from Th1 to Th2 (25, 26). Thus, the data presentedhere and by others suggest that inhibition of STAT3 signalingshould be revisited, with patients identified as those who haveabrogated TGFb signaling and high levels of infiltrating myeloidcells. New treatments targeting MDSC differentiation throughG-CSF could be utilized.

In the KPC mouse tumor model (3), excess GM-CSF drivesdevelopment of CD11bþGr1þ cells with T cell–suppressiveproperties, but it is unknown what subset of MDSCs is respon-

sible for these effects, how blocking of GM-CSF might affectstandard chemotherapy, or how GM-CSF correlates with stageand grade of human tumors. Inhibition of GM-CSF decreasestumor growth and the number of CD11bþGr1þ cells andincreases the number of cytotoxic CD8þ T cells (3, 27). Targeteddeletion of granulocytic MDSCs in PDAC increased intratu-moral accumulation of activated CD8þ T cells and apoptosis oftumor epithelial cells and also remodeling of the tumor stroma(12). However, suppression not only of Gr1þ cells but alsomonocytes and macrophages have positive antitumorigeniceffects on pancreatic cancer. Targeting TAMs and monocytesby inhibiting CSF1R (M-CSF receptor) or CCR2 (CCL2/MCP-1receptor) decreased a number of tumor-infiltrating cells inpancreatic tumors (14). These treatments also improve chemo-therapeutic efficacy. CSF1R blockade also upregulates T-cellcheckpoint proteins PDL1 and CTLA4 and treatment togetherwith PDL1/CTLA4 antagonist reduces tumor growth (11).Blocking CSF1R also increases sensitivity of tumor cells togemcitabine through elimination of cytidine deaminase thatis secreted by TAMs, which causes resistance of tumor cells tochemotherapy (9). In our studies, we found that knockdown ofG-CSF in pancreatic carcinoma cells reduced tumor growth inparallel with decreased number of Ly6Gþ cells and increasednumber of T cells and antitumorigenic M1-like macrophages(F4/80þ) based on protein array data. Combination treatmentof gemcitabine with blocking G-CSF was more efficient indecreasing tumor growth compared with gemcitabine alone(Fig. 4). From this, we conclude that anti–G-CSF provides analternative means of neoadjuvantly increasing the efficacy ofconventional chemotherapeutic interventions.

Patients with PDAC exhibit elevated level of MDSCs(CD33þHLA-DR�) upon progression of disease comparedwith healthy donors or patients with stable disease (28).MDSCs can be a marker of chemotherapy failure in pancreaticcancer patients (29). Circulating and tumor-infiltrating granu-locytic (Lin�HLA-DR�CD33þCD15þ), but not monocytic(Lin�HLA-DR�CD33þCD14þ) MDSCs were detected in pan-creatic cancer patients when compared with healthy donors andpatients with chronic pancreatitis (15), which correlates withour mouse tumor model. Chemotherapy-derived inflammationaccelerates the accumulation of MDSCs in human pancreaticcancer and blocking GM-CSF may benefit PDAC patients bydownregulation of the immunosuppressive properties ofMDSCs (20). PDAC is the fourth most common cause of cancerin North America and activating KRAS mutation and Smad4loss occur in approximately 90% and 55% of PDAC, respec-tively. The combination of KRAS oncogene and TGFb signalingalteration (via loss of Smad4) promotes malignant transfor-mation of human pancreatic epithelial cells and metastasis(30). Our analysis of TCGA datasets showed no differences inSmad4 expression in patients with or without metastasis, butSmad4 is downregulated during tumor progression Smad4mutation or loss associates with an increased number ofCD15þ cells and G-CSF levels through attenuating TGFb sig-naling as we discovered in our mouse model, especially inpatients with grade 1/stage II tumor based on TMA analysis.Our data identify a cohort of patients most likely to benefitfrom anti–G-CSF treatment.

In summary, our results demonstrated that aggressive pan-creatic adenocarcinoma tumors with deletion of TbRII in pan-creatic epithelium had elevated expression of G-CSF. This

Pickup et al.





CIR-16-0311; 28/9/2017; 22:56:59

Normal Grade 1 Grade 2 Grade 30

1

2

3

4

P = 0.03

P = 0.004

P = 0.006

% o

f Pos

are

a

A

Time (weeks)

Sur

viva

l rat

eTumor grade vs.

Smad4Tumor stage vs.

Smad4

Log

(Sm

ad4)

Log

(Sm

ad4)

Normal pancreas TumorC

DTMA, CD15

FTMA, G-CSF TMA, G-CSF

Survival curves by Smad4 expression

Smad4=LowSmad4=High

Smad4=LowSmad4=High

CD15 CD15

G-CSF G-CSF

G3G2G1

8.0

8.5

9.0

9.5

10.0

10.5

11.0

8.0

8.5

9.0

9.5

10.0

10.5

11.0

IVIIIIIBI IIAIBIA

0

200

400

600

G-CSF TMA Epithelial scoring

Ave

rage

H-s

core

Normalpancreas

Tumortissue

B

E

Normal Grade 1 Grade 2 Grade 30

200

400

600

800

P = 0.04

P = 0.0009

P = 0.01

Area

× in

tens

ity

Normal Stage I Stage II Stage III Stage IV0

200

400

600

800

P = 0.02

P = 0.75

P = 0.21

P = 0.03

Area

× in

tens

ity

Normal Stage I Stage II Stage III Stage IV0

1

2

3

4

5 P < 0001

P = 0.0006

P < 017

P = 0.12

% o

f Pos

are

a

Figure 4.

Increased number of CD15þ cells and expression of G-CSF correlates with grade 1/stage II in human pancreatic cancer. A and B, TCGA dataset analysis ofpancreatic cancer patient survival for Smad4 high and low expressing patients and analysis of Smad4 expression segregated by metastatic incidence, tumor grade,and tumor stage. C, Representative tissue section from TMA stained for CD15 and G-CSF by IHC. D and F, CD15 (D) and G-CSF analysis (F) of TMA staininggrouped by grade and tumor stage. Normal, n¼ 17; tumor, n ¼ 76; grade 1, n ¼ 17; grade 2, n ¼ 28; grade 3, n¼ 31; stage I, n¼ 43; stage II, n¼ 30; stage III, n¼ 12;stage IV, n ¼ 4. E, Epithelial scoring of G-CSF in TMA used for C, D, and F. Scale bar, 100 mm.






CIR-16-0311; 28/9/2017; 22:57:4

resulted in an increase in the number of infiltratingCD11bþLy6ClowLy6Ghigh cells into the tumor tissue. Also, G-CSF upregulated immunosuppressive genes, such as iNOS, Arg,

IL6, IL10, MMP9, and VEGF. These myeloid cells had anincreased ability to suppress T-cell proliferation and as a resultdecreased the antitumor immune response (Fig. 5). Our studiesprovide insights into the development of aggressive ductaladenocarcinoma. Mutations in genes encoding TGFb signalingmolecules during PDAC development are very common, butmanipulation of the TGFb pathway during tumor progressionremains controversial due to the dual roles of TGFb on tumorprogression. As G-CSF has protumorigenic effects on pancreaticcancer, the targeting of this cytokine could be considered whenit is associated with mutations in TGFb signaling componentsin patients with grade 1/stage II pancreatic cancer.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: M.W. Pickup, R. Kalluri, H.L. Moses, S.V. NovitskiyDevelopment of methodology: M.W. Pickup, A.E. Gorska, S.V. NovitskiyAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.):M.W. Pickup, P. Owens, A. Chytil, C. Shi, V.M.WeaverAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): M.W. Pickup, P. Owens, F. Ye, S.V. NovitskiyWriting, review, and/or revision of the manuscript: M.W. Pickup, F. Ye,H.L. Moses, V.M. Weaver, S.V. NovitskiyAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): A. ChytilStudy supervision: S.V. Novitskiy

Grant SupportThis work was supported by NIH grants CA200681 (to S.V. Novitskiy) and

CA151925 (to H. Moses), and the Robert J. and Helen C. Kleberg Foundationand the T.J.Martell Foundation toH.L.Moses andCA068485 for core laboratorysupport.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received November 4, 2016; revised May 18, 2017; accepted July 25, 2017;published OnlineFirst August 3, 2017.

References1. Ijichi H, Chytil A, Gorska AE, Aakre ME, Fujitani Y, Fujitani S, et al.

Aggressive pancreatic ductal adenocarcinoma in mice caused by pan-creas-specific blockade of transforming growth factor-beta signalingin cooperation with active Kras expression. Genes Dev 2006;20:3147–60.

2. Herreros-Villanueva M, Hijona E, Cosme A, Bujanda L. Mouse models ofpancreatic cancer. World J Gastroenterol 2012;18:1286–94.

3. Bayne LJ, Beatty GL, Jhala N, Clark CE, Rhim AD, Stanger BZ, et al. Tumor-derived granulocyte-macrophage colony-stimulating factor regulates mye-loid inflammation and T cell immunity in pancreatic cancer. Cancer Cell2012;21:822–35.

4. Nagathihalli NS, Castellanos JA, Shi C, Beesetty Y, Reyzer ML, Caprioli R,et al. Signal transducer and activator of transcription 3, mediated remodel-ing of the tumor microenvironment results in enhanced tumor drugdelivery in a mouse model of pancreatic cancer. Gastroenterology2015;149:1932–43.

5. Laklai H, Miroshnikova YA, Pickup MW, Collisson EA, KimGE, Barrett AS,et al. Genotype tunes pancreatic ductal adenocarcinoma tissue tension toinduce matricellular fibrosis and tumor progression. Nat Med 2016;22:497–505.

6. Ijichi H, Chytil A, Gorska AE, Aakre ME, Bierie B, Tada M, et al. InhibitingCxcr2 disrupts tumor-stromal interactions and improves survival in amouse model of pancreatic ductal adenocarcinoma. J Clin Invest 2011;121:4106–17.

7. Steele CW, Karim SA, Leach JD, Bailey P, Upstill-Goddard R, Rishi L, et al.CXCR2 inhibition profoundly suppressesmetastases and augments immu-notherapy in pancreatic ductal adenocarcinoma. Cancer Cell 2016;29:832–45.

8. Ozdemir BC, Pentcheva-Hoang T, Carstens JL, Zheng X, Wu CC, SimpsonTR, et al. Depletion of carcinoma-associated fibroblasts and fibrosisinduces immunosuppression and accelerates pancreas cancer with reducedsurvival. Cancer Cell 2014;25:719–34.

9. Weizman N, Krelin Y, Shabtay-Orbach A, Amit M, Binenbaum Y, Wong RJ,et al. Macrophages mediate gemcitabine resistance of pancreatic adeno-carcinoma by upregulating cytidine deaminase. Oncogene 2014;33:3812–9.

10. Mitchem JB, Brennan DJ, Knolhoff BL, Belt BA, Zhu Y, Sanford DE, et al.Targeting tumor-infiltrating macrophages decreases tumor-initiating cells,relieves immunosuppression, and improves chemotherapeutic responses.Cancer Res 2013;73:1128–41.

Aggressive PDACTGF receptor II deletion

(p48/KrasG12D)[Genes & Dev, 2006]

CXCL1CXCL5[JCI, 2011]

G-CSF

Myeloid cells

Rec

ruitm

ent

Mo

bili

zati

on

Dif

fere

nti

atio

nA

ctiv

atio

nA

rg, i

NO

S, V

EG

F

IL10

Immune system suppression(mainly T cell)

Tumor progression

1

2

3

4

5

6

Figure 5.

Schematic summarizing data and working hypotheses. The following numbersrefer to the number in the schematic. 1, Tgfbr2KO pancreatic epithelial cellssecrete abundant CXCL1/CXCL5 due to lack of TGFb suppression ofchemokine expression (6). 2, The CXCL chemokines recruit to the tumormicroenvironment myeloid cells (CD11bþ). 3, In parallel with chemokines,carcinoma cells secrete increased level of G-CSF. 4, G-CSF promotesdifferentiation of myeloid cells to Ly6Gþ cells, upregulates immunosuppressivegenes, such asArg, iNOS, andVEGF, andupregulate secretion of IL10,making (5)myeloid cells that are immunosuppressive with protumorigenic properties.6, The net result is enhanced tumor growth and decrease survival.


Pickup et al.




CIR-16-0311; 28/9/2017; 22:57:5

11. Zhu Y, Knolhoff BL, Meyer MA, Nywening TM, West BL, Luo J, et al. CSF1/CSF1Rblockade reprograms tumor-infiltratingmacrophages and improvesresponse to T-cell checkpoint immunotherapy inpancreatic cancermodels.Cancer Res 2014;74:5057–69.

12. Stromnes IM, Brockenbrough JS, Izeradjene K, Carlson MA, Cuevas C,Simmons RM, et al. Targeted depletion of an MDSC subset unmaskspancreatic ductal adenocarcinoma to adaptive immunity. Gut 2014;63:1769–81.

13. Panni RZ, Sanford DE, Belt BA, Mitchem JB, Worley LA, Goetz BD, et al.Tumor-induced STAT3 activation in monocytic myeloid-derived suppres-sor cells enhances stemness and mesenchymal properties in humanpancreatic cancer. Cancer Immunol Immunother 2014;63:513–28.

14. Mace TA, Ameen Z, Collins A,Wojcik S,MairM, YoungGS, et al. Pancreaticcancer-associated stellate cells promote differentiation of myeloid-derivedsuppressor cells in a STAT3-dependent manner. Cancer Res 2013;73:3007–18.

15. Khaled YS, Ammori BJ, Elkord E. Increased levels of granulocytic myeloid-derived suppressor cells in peripheral blood and tumour tissue of pancre-atic cancer patients. J Immunol Res 2014;2014:879897.

16. Youn JI, Nagaraj S, Collazo M, Gabrilovich DI. Subsets of myeloid-derived suppressor cells in tumor-bearing mice. J Immunol 2008;181:5791–802.

17. Shojaei F, Wu X, Qu X, Kowanetz M, Yu L, Tan M, et al. G-CSF-initiatedmyeloid cell mobilization and angiogenesis mediate tumor refractorinessto anti-VEGF therapy in mouse models. Proc Natl Acad Sci USA2009;106:6742–7.

18. Herber DL, Cao W, Nefedova Y, Novitskiy SV, Nagaraj S, Tyurin VA, et al.Lipid accumulation and dendritic cell dysfunction in cancer. Nat Med2010;16:880–6.

19. Pickup MW, Hover LD, Guo Y, Gorska AE, Chytil A, Novitskiy SV, et al.Deletionof the BMP receptor BMPR1a impairsmammary tumor formationand metastasis. Oncotarget 2015;6:22890–904.

20. Youn JI, Collazo M, Shalova IN, Biswas SK, Gabrilovich DI. Characteriza-tion of the nature of granulocytic myeloid-derived suppressor cells intumor-bearing mice. J Leukoc Biol 2012;91:167–81.

21. Rutella S, Zavala F, Danese S, Kared H, Leone G. Granulocyte colony-stimulating factor: a novel mediator of T cell tolerance. J Immunol2005;175:7085–91.

22. Cancer.Net. Pancreatic cancer: treatment options. Alexandria, VA: Amer-ican Society for Clinical Oncology. Available from: http://www.cancer.net/cancer-types/pancreatic-cancer/treatment-options.

23. Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, Angenendt P, et al. Coresignaling pathways in human pancreatic cancers revealed by global geno-mic analyses. Science 2008;321:1801–6.

24. GabrilovichDI,Nagaraj S.Myeloid-derived suppressor cells as regulators ofthe immune system. Nat Rev Immunol 2009;9:162–74.

25. Martinez FO, Gordon S. The M1 and M2 paradigm of macrophageactivation: time for reassessment. F1000Prime Rep 2014;6:13.

26. Mosmann TR, Moore KW. The role of IL-10 in crossregulation of TH1 andTH2 responses. Immunol Today 1991;12:A49–53.

27. Pylayeva-Gupta Y, LeeKE,HajduCH,MillerG, Bar-SagiD.Oncogenic Kras-induced GM-CSF production promotes the development of pancreaticneoplasia. Cancer Cell 2012;21:836–47.

28. Wachsmann MB, Pop LM, Vitetta ES. Pancreatic ductal adeno-carcinoma: a review of immunologic aspects. J Investig Med 2012;60:643–63.

29. Markowitz J, Brooks TR, Duggan MC, Paul BK, Pan X, Wei L, et al. Patientswith pancreatic adenocarcinoma exhibit elevated levels of myeloid-derived suppressor cells upon progression of disease. Cancer ImmunolImmunother 2015;64:149–59.

30. Leung L, Radulovich N, Zhu CQ, Wang D, To C, Ibrahimov E, et al. Loss ofcanonical Smad4 signaling promotes KRAS driven malignant transforma-tion of human pancreatic duct epithelial cells and metastasis. PLoS One2013;8:e84366.





http://www.cancer.net/cancer-types/pancreatic-cancer/treatment-options

http://www.cancer.net/cancer-types/pancreatic-cancer/treatment-options


2017;5:718-729. Published OnlineFirst August 3, 2017.Cancer Immunol Res Michael W. Pickup, Philip Owens, Agnieszka E. Gorska, et al. Stimulating Factor Secretion in Carcinoma CellsAdenocarcinomas Depends on Granulocyte Colony Development of Aggressive Pancreatic Ductal

Updated version

10.1158/2326-6066.CIR-16-0311doi:

Access the most recent version of this article at:

Material

Supplementary

DC1

http://cancerimmunolres.aacrjournals.org/content/suppl/2017/08/03/2326-6066.CIR-16-0311.Access the most recent supplemental material at:

Cited articles

http://cancerimmunolres.aacrjournals.org/content/5/9/718.full#ref-list-1

This article cites 29 articles, 10 of which you can access for free at:

Citing articles

http://cancerimmunolres.aacrjournals.org/content/5/9/718.full#related-urls

This article has been cited by 4 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

SubscriptionsReprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://cancerimmunolres.aacrjournals.org/content/5/9/718To request permission to re-use all or part of this article, use this link



http://cancerimmunolres.aacrjournals.org/lookup/doi/10.1158/2326-6066.CIR-16-0311

http://cancerimmunolres.aacrjournals.org/content/suppl/2017/08/03/2326-6066.CIR-16-0311.DC1

http://cancerimmunolres.aacrjournals.org/content/suppl/2017/08/03/2326-6066.CIR-16-0311.DC1

http://cancerimmunolres.aacrjournals.org/content/5/9/718.full#ref-list-1

http://cancerimmunolres.aacrjournals.org/content/5/9/718.full#related-urls

http://cancerimmunolres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerimmunolres.aacrjournals.org/content/5/9/718


development of aggressive pancreatic ductal ...development of aggressive pancreatic ductal...

Documents