Loss of CXCR4 in Myeloid Cells Enhances Antitumor Immunity ... · Research Article Loss of CXCR4 in Myeloid Cells Enhances Antitumor Immunity and Reduces Melanoma Growth through NK

Research Article

Loss of CXCR4 in Myeloid Cells EnhancesAntitumor Immunity and Reduces MelanomaGrowth through NK Cell and FASL MechanismsJinming Yang1,2, Amrendra Kumar1,3, Anna E. Vilgelm2, Sheau-Chiann Chen4,Gregory D. Ayers4,5, Sergey V. Novitskiy6, Sebastian Joyce1,3, and Ann Richmond1,2

Abstract

The chemokine receptor, CXCR4, is involved in cancergrowth, invasion, and metastasis. Several promising CXCR4antagonists have been shown to halt tumor metastasis inpreclinical studies, and clinical trials evaluating the effective-ness of these agents in patients with cancer are ongoing.However, the impact of targeting CXCR4 specifically onimmune cells is not clear. Here, we demonstrate that geneticdeletion of CXCR4 in myeloid cells (CXCR4MyeD/D) enhancesthe antitumor immune response, resulting in significantlyreduced melanoma tumor growth. Moreover, CXCR4MyeD/D

mice exhibited slowed tumor progression compared withCXCR4WT mice in an inducible melanocyte BrafV600E/Pten�/�

mouse model. The percentage of Fas ligand (FasL)–expressingmyeloid cells was reduced in CXCR4MyeD/D mice as comparedwith myeloid cells from CXCR4WT mice. In contrast, there wasan increased percentage of natural killer (NK) cells expressing

FasL in tumors growing in CXCR4MyeD/D mice. NK cells fromCXCR4MyeD/D mice also exhibited increased tumor cell killingcapacity in vivo, based on clearance of NK-sensitive Yac-1 cells.NK cell–mediated killing of Yac-1 cells occurred in a FasL-dependent manner, which was partially dependent upon thepresence of CXCR4MyeD/D neutrophils. Furthermore,enhanced NK cell activity in CXCR4MyeD/D mice was alsoassociated with increased production of IL18 by specificleukocyte subpopulations. These data suggest that CXCR4-mediated signals from myeloid cells suppress NK cell–mediated tumor surveillance and thereby enhance tumorgrowth. Systemic delivery of a peptide antagonist of CXCR4to tumor-bearing CXCR4WT mice resulted in enhancedNK-cell activation and reduced tumor growth, supportingpotential clinical implications for CXCR4 antagonism insome cancers. Cancer Immunol Res; 6(10); 1186–98. �2018 AACR.

IntroductionChemokine receptor 4 (CXCR4) is a 7-transmembrane G

protein–coupled receptor that interacts with its endogenousligand CXCL12, also known as stromal cell–derived factor-1(SDF-1), regulates many key physiologic processes (1). However,CXCR4 is also highly expressed in more than 23 human cancers,where it has been reported to be expressed by tumor cells aswell asstromal cells, enabling it to promote tumorigenesis, progression,metastasis and influence relapse, and prognosis (2). CXCR4antagonism has been shown to disrupt tumor–stromal interac-tions, reduce tumor growth and metastatic burden, and even

overcome cancer cell resistance to cytotoxic drugs (3). CXCL12-based peptides and CXCR4-based small-molecule antagonists(4, 5) are in phase I/II clinical trials in patients with advancedsolid tumors. The CXCR4/CXCL12 axis not only is a therapeutictarget on tumor cells, but also is involved in inflammation andimmunity in the tumor microenvironment (6). However, theimpact of systemic targeting of CXCR4 on the immune cells hasnot been clearly elucidated.

In this study, we used genetic knockout of CXCR4 in myeloidcells to demonstrate that disruption of CXCR4/CXCL12 signalingin these cells inhibits the outgrowth of circulating B16melanomacells in the lung and inhibits tumor growth in an inducibleBrafV600E/Pten null melanoma mouse model. We illustrate thatloss of expression of CXCR4 in myeloid cells results in enhancedexpression of cytokine IL18 that activates natural killer (NK) cellsand enhances antitumor immunity. The CXCR4 peptide antago-nist, LY2510924, also enhances antitumor activity in part byactivating NK cells. Together, our data provide new insightinto the mechanism by which CXCR4 antagonism inhibitstumor growth.

Materials and MethodsCell lines and establishment of tumor models

PyMT breast cancer cells were provided by the Hal Moseslaboratory. This cell line was established from a spontaneousPyMT tumor growing in C57BL/6 mice. The cells have beenpreviously characterized (7–9). The YAC1 cells were obtainedfrom ATCC. The B16F0 cells were also obtained from ATCC,

1Tennessee Valley Healthcare System, Department of Veterans Affairs, VanderbiltUniversity Medical Center, Nashville, Tennessee. 2Department of Pharmacology,Vanderbilt University Medical Center, Nashville, Tennessee. 3Department ofPathology, Microbiology and Immunology, Vanderbilt University Medical Center,Nashville, Tennessee. 4Department of Biostatistics, Vanderbilt University MedicalCenter, Nashville, Tennessee. 5Division of Cancer Biostatistics, Departmentof Biostatistics, Vanderbilt University, Nashville, Tennessee. 6Department ofMedicine, Vanderbilt University, Nashville, Tennessee.

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

Corresponding Author: Ann Richmond, Vanderbilt University, PRB432, 2220Pierce Avenue, Nashville, TN 37232. Phone: 615–343–7777; Fax: 615–936–2911;E-mail: [email protected]

doi: 10.1158/2326-6066.CIR-18-0045

�2018 American Association for Cancer Research.

CancerImmunologyResearch

Cancer Immunol Res; 6(10) October 20181186

on August 19, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst August 14, 2018; DOI: 10.1158/2326-6066.CIR-18-0045

http://crossmark.crossref.org/dialog/?doi=10.1158/2326-6066.CIR-18-0045&domain=pdf&date_stamp=2018-9-17

http://cancerimmunolres.aacrjournals.org/

expanded, aliquoted, and frozen in liquid nitrogen. Aliquots wereused in the experiments here. The PYMT cells were obtained fromHarold LMoses at passage 3, expanded, frozen back and passage 4to 5 cells were used for experiments here. All cell cultures weremycoplasma free. Cultures are testedmonthly formycoplasma usinga sensitive PCR technique (e-MycoTM Plus, LiliF Diagnostics).Any mycoplasma positive cultures are discarded. We did notgenetically reauthenticate the cell lines, but we verified thecytological and histologic authenticity of the cells in culture andin mouse models.

PyMT breast cancer cells (1� 106) derived from C57BL/6 micewere intravenously injected into C57BL/6 CXCR4myeD/D or litter-mate control CXCR4WT mice. B16F0 melanoma cells wereobtained from ATCC. B16F0 melanoma cells (1 � 106) wereintravenously injected into CXCR4myeD/D mice (11 mice/group)or littermate CXCR4WT mice (9 mice/group).

An inducible melanoma mouse model was established bybreeding mice such that alleles of brafCA, Tyr::CreER andptenlox4-5/Lox4-5 are all present in each mouse. SubsequentlyCre-mediated conversion of brafV600 to brafV600E and thedeletion of exons 4 and 5 of pten were induced with topicaladministration of 4-hydroxytamoxifen (4-HT; ref. 10). Thesemice received a transplant of bone marrow cells (1 � 106)from CXCR4myeD/D or CXCR4WT mice and were subsequentlytreated with 5 mmol/L 4-HT to induce melanoma formation,followed by bone marrow reconstitution.

Myeloid CXCR4 knockout modelsAll animal experiments were approved by the Vanderbilt

University Institutional Animal Care and Use Committee. Todelete Cxcr4 in myeloid cells in C57BL/6 mice, the geneencoding Cre-recombinase under the control of the murineLyzM gene regulatory region (11) was introduced into CXCR4f/f

mice obtained from The Jackson Laboratories. The resultingmice were then bred to mice harboring the loxP-flankedtdTomato (mT) following the EGFP (mG) cassette, which wasinserted into the Gt(ROSA)26Sor locus (The Jackson Labora-tories). These mT/mG mice served as a Cre-reporter strain andafter Cre-mediated recombination myeloid cells that areCXCR4-null are green. Mice with CXCR4-null myeloidcells are designated as CXCR4myeD/D mice. LittermatesLysMCre::mT/mG mice without CXCR4f/f alleles were used asCXCR4WT controls.

Bone marrow transplant and inducible/spontaneousmelanoma models

Recipient C57BL/6 mice carrying BrafV600E/Ptenf/f/Tyr-Crealleles (obtained from The Jackson Laboratories) were given100 mg/L neomycin,10 mg/L polymyxin B in pH2 water 1 weekbefore transplant and then continuously for 6 weeks after trans-plantation. Mice received one dose of 10-Gy irradiation (CesiumGamma irradiator). Four hours later, mice were injected via tailvein with bonemarrow cells (1� 106) fromC57BL/6 donormice(CXCR4myeD/D mice or myeloid CXCR4WT mice). The reconstitu-tion of bone marrow in recipient mice was examined by moni-toring peripheral fluorescent green-positive myeloid and tomato-positive lymphocytes using flow cytometry. Themelanomas wereinduced with topical administration of 4-hydroxytamoxifen at5 mmol/L daily for 3 days. One and a half months after bonemarrow transplantation, tumor growth was evaluated and statis-tically analyzed by two-way ANOVA.

Establishment of NK-sensitive target Yac-1–reporter cell lineYac-1 cells were obtained from ATCC and maintained in

RPMI 1640 medium supplemented with 5% fetal bovine serum,2 mmol/L glutamine, 10 mmol/L HEPES (pH 7.4), and antibio-tics (100 units of penicillin/mL and 100 mg of streptomycin/mL).A lentiviral vector, pRBow-FLuc-BFP, contains dual expressioncassette with a single MCV promoter driving expression of botha luminescent and a blue fluorescent protein (BFP). Yac-1 cellswere infected with viral particles in complete RPMI mediumsupplemented with 8 mg/mL of polybrene (Sigma). Twenty-fourhours after infection, the cells expressing the construct wereselected with blasticidin (12 mg/mL) for 10 days. Cells expressingFluc were identified by flow cytometry and chemiluminescence,respectively. The selected cells were used for ex vivo and in vivoimaging NK cell cytotoxicity assay.

In vivo imaging of NK cell cytotoxicityTo determine NK cell cytotoxicity, YAC-1-luciferase (Luc) cells

(8 � 106 cells in 1 mL HBSS) were injected into mice via the tailvein. After being anesthetized with isoflurane, mice were theninjected intraperitoneally with 3mg D-luciferin in 200 mL PBS foreach imaging session. Whole body images were taken 10minutesafterD-luciferin injection. Themicewere repeatedly imaged in thesame position for 2minutes at 1, 4, and 24 hours after injection ofYAC-1–Luc cells using an IVIS-200 imaging system (XenogenImaging Technologies) to acquire the photons of light emittedfrom the mice. Regional luciferase signals were quantified usingLiving Image 4.1 software (Xenogen).

NK cytotoxicity assay ex vivoNKeffector cellswere derived frommurine spleen vianegatively

selection approach using EasySep mouse NK Cell IsolationKit (Stemcell Technologies) and the NK purity was over 80%.YAC-1–Luc target cells were washed once in RPMI media andplated at a concentration of 10,000 cells per 0.1 mL in 96-wellmicroplates. Target cells were cocultured with increasing numbersof purified NK cells, in a final volume of 0.2mL at 5%CO2, 37�C.After a 4-hour incubation, cocultured cells were washed 3 timeswith PBS. The cells were lysed using the Luciferase Assay Systemwith Reporter Lysis Buffer (Promega), and the intracellular lucif-erase activity was determined by reading 10 seconds of chemilu-minescence per the manufacturer's protocol (Promega). Platingtarget cells in media without effector cells served as a control.

Depletion of immune cells in vivoFor functional abrogation of macrophages, mice were treated

daily via oral gavage with 200 mg of colony-stimulating factor-1receptor inhibitor (CSF-1R; Novartis Pharmaceuticals Corpora-tion) or 20% captisol as vehicle control for 10 days (12). Fordepletion of neutrophils, mice were peritoneally injected with250 mg of Ly6G monoclonal antibody (mAb; Clone 1A8, Bio XCell) or the same amount of IgG2a isotype control mAb (Clone2A3, Bio X Cell) daily for 3 days andmaintained at 100 mg/mouseevery other day (13). For depletion of CD8þ T cells, mice wereperitoneally injected with 200 mg of CD8a mAb (Clone YTS169.34, Bio X Cell) or the same amount of IgG2a isotype controlmAb (Clone 2A3, Bio XCell) daily for 3 days andmaintainedwith100 mg/mouse every other day (14). For specifically depletingNK cells, mice were intravenously injected with 300 mg rabbitantibody to asialo-GM1 (WakoChemicals), or a volume of rabbitserum containing an equivalent amount of control IgG, 2 days

CXCR4 Regulates Tumor Immunity

www.aacrjournals.org Cancer Immunol Res; 6(10) October 2018 1187




before tumor implantation and maintained with an injection of300 mg anti-asialo-GM1 or IgG control twice a week (15).

Flow cytometry analysis and antibodiesFor flow cytometry analyses, tissues were minced on a pro-

grammable dissociator and digested with an enzyme solution ofcollagenase 1 (1,500 CDU), Dispase II (1 mg/mL) and DNase 1(0.1 mg/mL). The detail of antibodies used (listed in supplemen-tary data), staining, and flow cytometry analyses protocols isaccording to our previous published methodology (16). Forintravenous antibody staining and flow cytometry, animals wereinjected intravenously with 2 mg allophycocyanin-conjugatedanti-mouse CD45 mAb. Mouse lungs were cut into 1- to 2-mmslices and digested in buffer containing 2 mg/mL collagenase and0.1 mg/mL DNase I (8). Digested lungs were passed through a70-mm strainer to obtain a single-cell suspension. Mouse spleenswere pressed through 40 mm strainer using syringe plunger toobtain a single-cell suspension. Cells were incubated with GhostDye TM Violet 510, an amine reactive viability dye used todiscriminate live/dead cells, and washed with FACS buffer (PBScontaining 2% v/v FBS). After blocking Fc receptors with anti-mouse CD16/CD32 mAb in FACS buffer for 15 minutes, cellswere incubatedwithmAbs tomouse CD45R (B220)-FITC, CD8a-FITC, NK1.1-PE, CD3e-BV421 in addition to BV421-conjugatedCD1d tetramers loaded with GalCer (CD1d-tet) for additional 1hour on ice as described (14, 17). Cells werewashed twice in FACSbuffer and data acquired with FACSCanto II (Becton Dickenson).FACSdata, comparing leukocytes fromCXCR4myeD/D toCXCR4WT

(.fcs files) were analyzed using FlowJo software (Version 10.1).For cell counting, absolute numbers of lymphocyteswere countedusing AccuCheck counting beads (Thermo Fisher). Frequency ofNK cells, type I NKT cells, and type II NKT cells within totallymphocyte gate was obtained and used to calculate absolutenumbers of each of these populations.

Cytokine quantitationIL18 concentrations in the lung tumor tissue and serum were

determined by ELISA (cat. 7625; R&D Systems) 16 days afterB16F0 melanoma cells were intravenously implanted intoCXCR4WT and CXCR4MyeD/D mice.

Statistical analysisData are summarized in figures using the mean � SD. Treat-

ment effects in standard two-group experiments were comparedusing the Wilcoxon rank-sum test. The experimental differencesbetween 2 groups by variables, such as difference betweenCXCR4WT and CXCR4myeD/D mice across effector ratios, wereassessed in the context of a two-way ANOVA. Pairwise differencesbetween two groups were compared using model-based meancomparisons. The Benjamini and Hochberg (BH) correctionwas used to adjust P value for multiple comparison as noted(denoted by asterisks) in text and figure legends to control thewithin experiment false discovery rate to less than 5%. Also, �, P <0.05; ��, P < 0.01; ��, P < 0.001.

ResultsDeletion of CXCR4 in myeloid cells results in an antitumorphenotype

Murine C57BL/6 mice with deletion of CXCR4 in myeloidcells were generated by tissue-specific LysM-Cre-mediated

recombination. The resulting CXCR4myeD/D mice exhibitedbarely detectable expression of both Cxcr4 transcript andCXCR4 protein (Supplementary Fig. S1A–S1C). To determinehow CXCR4 expression in myeloid cells affects immunity in anexperimental melanoma metastasis model, syngeneic B16F0melanoma cells were intravenously injected into CXCR4myeD/D

mice littermate CXCR4WT mice. Two weeks later, lung weight inCXCR4myeD/D tumor-bearing mice was 52% lower than in theCXCR4WT tumor-bearing mice (398 � 192 mg vs. 762� 70 mg,P < 0.001; Fig. 1A and B). In a separate experiment, 2 weeksafter PyMT breast cancer cells were intravenously injected intoCXCR4myeD/D or littermate control CXCR4WT mice, the lungtumor weight in CXCR4WT mice was over 70% greater than inCXCR4myeD/D mice compared with CXCR4WT mice (P < 0.01,n ¼ 6; Supplementary Fig S1D–S1E) Thus, disruption of theCXCR4/CXCL12 signaling axis in myeloid cells altered thetumor microenvironment and effectively suppressed outgrowthof both melanoma or breast cancer cells in the lung.

To further examine whether disruption of CXCR4/CXCL12signaling inmyeloid cells influences tumorigenesis ofmelanoma,a 4-hydroxytamoxifen (4-HT)-inducible braf/pten mouse mela-noma model with bone marrow transplant from CXCR4MyeD/D

and CXCR4WT donor mice was used. One and one-half monthsafter bone marrow transplantation, 100% melanoma incidenceoccurred in both groups receiving the bone marrow cells fromCXCR4MyeD/D and CXCR4WT donor mice. However, the tumorweight in mice receiving myeloid CXCR4MyeD/D bone marrowwas 67% lower than that of recipients of CXCR4WT bone marrow(P < 0.05; Fig. 1C and D). Data shown here demonstrate thatdisruption of CXCR4/CXCL12 signaling in myeloid cells sup-presses B-RafV600E/Pten�/�-driven melanocyte transformationand melanoma formation.

To examine whether disruption of CXCR4/CXCL12 signalingaltered the immune cell population in blood, immune cellsfrom peripheral blood were stained with specific antibodiesand analyzed by flow cytometry. Cells with the tissue-specificLysM-Cre-mediated deletion of myeloid CXCR4 were identifiedbasedonGFP expression. This analysis showed that theperipheralmyeloid cell population was increased by 0.85-fold inCXCR4MyeD/D mice in comparison with that of CXCR4WT mice(Fig. 1E, P < 0.01, n ¼ 6). Neutrophils, but not macrophages,were significantly increased by 1.33-fold in the CXCR4MyeD/Dmice(P<0.001). In addition, B cells were decreased by 35%(P<0.001)in mice with CXCR4 deletion in myeloid cells. These datasuggest a functional link between the immune cell alterationsand antitumor phenotype in the CXCR4myeD/D mouse. However,we failed to observe a difference in the total CD3þ T-cell infiltra-tion into the tumor in the inducible melanoma tumors grown inmice that were transplanted with bone marrow cells fromCXCR4myeD/DorCXCR4WTmice (P¼ 0.74, n¼10; SupplementaryFig. S1I–S1J).

A cell-cycle analysis was performed in neutrophils derived frombone marrow and peripheral blood of non–tumor-bearingCXCR4MyeD/D or CXCR4WT mice. An increased percentage of cellsin S phase was observed in the bone marrow neutrophils fromCXCR4MyeD/D mice by 0.54-fold in comparison with CXCR4WT

mice (P < 0.01, n ¼ 6; Supplementary Fig. S1Fa). In addition,thepercent peripheral neutrophils in sub-G0phase of the cell cyclewas significantly reduced by 97% in CXCR4MyeD/D mice in com-parison with that from CXCR4WT mice (P < 0.01, n ¼ 6; Supple-mentary Fig. S1Fb). Thus, blockage of CXCR4 signaling in

Yang et al.

Cancer Immunol Res; 6(10) October 2018 Cancer Immunology Research1188




neutrophils is associated with a 1.33-fold increase in numbersof neutrophils in the peripheral blood of CXCR4MyeD/D miceand a 0.54-fold increase in the bone marrow neutrophils in Sphase. To further investigate the role of CXCL12 on bonemarrow neutrophils, these cells isolated from wild-typeC57BL/6 mice were exposed to increasing concentrations ofCXCL12 (1–100 ng/mL) in vitro over 3 days. Results showed aninduction of proliferation on day 1 in response to �6.2 ng/mL(Supplementary Fig. S1G), followed by a slight decline in

Ki67þ cells and a 2-fold increase in annexin V staining on day2 that peaked with 12.5 ng/mL CXCR4 (SupplementaryFig. S1H), indicating that CXCL12 may play a bifunctionalrole in vitro. However, CXCL12 in the bone marrow niche hasbeen reported to hold neutrophils in the marrow and whenCXCR4 is antagonized there is a release of neutrophils into theperipheral blood (31). Thus, it is highly possible that prema-ture release from the bone marrow is the major mechanism ofincreased neutrophils in peripheral blood.

A a

c

bB

CXCR4wt CXCR4D/D

C D

E

0

20

40

60

Myeloid Mac Neu CD4 CD8 B cell NK DC

Perip

hera

l im

mun

e pr

ofile

CXCR4WT

P < 0.01

P < 0.01 P < 0.01 P < 0.01

CXCR4WT

CXCR4D/D

CXCR4D/D

CXCR4D/D

CXCR4WT

P < 0.05

Tum

or v

olum

e (m

m3 )

Lung

tum

or w

eigh

t (m

g)

1,000

800

600

400

200

0

1,000

800

600

400

200

0

Figure 1.

Myeloid CXCR4-null mice exhibit an antitumor phenotype. A, Experimental lung metastasis. Syngeneic melanoma cells (B16F0, 1 � 106) were intravenouslyinjected into CXCR4MyeD/D mice (n ¼ 11) or littermate myeloid CXCR4WT mice (n ¼ 9). Two weeks after injection, the lungs were weighed. The lung tumorburden was calculated as the weight of tumor-bearing lung minus that of tumor-free lung and analyzed by Wilcoxon rank-rum test. B, Representative photoof the tumor-bearing lung was from CXCR4WT mouse (a), from CXCR4MyeD/D mouse (b), or from tumor-free mouse (c). C, Inducible spontaneous melanoma.The BrafCA::Ptenf/f::Tyr-Cre C57BL/6 mice were irradiated and then transplanted with bone marrow cells from CXCR4MyeD/D mice or CXCR4WT mice. Recipientmice were topically treated with 4-hydroxytamoxifen to induce tumor formation and 1.5 months after bone marrow transplantation, tumor volume wasdetermined and analyzed by Wilcoxon rank-sum test. D, Representative photo for the skin of tumor-bearing mouse with bone marrow transplants from CXCR4WT

or CXCR4MyeD/D mouse. E, Profile of immune cell populations from peripheral blood. Peripheral blood was collected from CXCR4MyeD/D mice (n ¼ 6) or CXCR4WT

mice (n ¼ 6). CD45þ leukocytes were prepared and stained with fluorochrome-conjugated antibodies specific for immune cell-surface markers as indicated:myeloid cells (CD11bþ), macrophages (CD11bþ/F4/80þGr1�), and neutrophils (CD11bþ/Gr1þ). The stained cells were subjected to flow cytometry analysis.Percentage of subtypes of immune cells in the total CD45þ cell population was graphed and statistically analyzed by the two-way ANOVA with model-basedmean comparisons and BH P value correction.






NK cells, but not NKT cells, facilitate antitumor effect of CXCR4loss in myeloid cells

To identify whether the innate or acquired immune responseplays a role in the antitumor phenotype, CXCR4MyeD/D mice weretreated with antibodies against asialo-GM1 or CD8. This treat-ment efficiently depleted NK cells by 93% (SupplementaryFig. S2A) and CD8þ T cells by 98% (Supplementary Fig. S2B) inthe respective mice, when compared with mice treated with an

isotype-matched nonspecific antibody. Syngeneic B16F0 mela-noma cells were intravenously implanted into these mice. After 2weeks of tumor growth in the lung, we observed that NK celldepletion (Fig. 2A), but not CD8þ T-cell depletion (Supplemen-tary Fig. S2C), reversed the suppression of tumor growth inCXCR4myeD/D mice. Hence, NK cells appear to be required forthe enhanced antitumor immunity in the CXCR4MyeD/D tumor-bearing mice.

A

C D

0.00.20.40.60.81.0

2468

Isotype IgG Anti-Asialo GM1

# N

K C

ell

MV (CD45+)

IST(CD45-)

MV (CD45+)

IST(CD45-)

MV (CD45+)

IST(CD45-)

0.0

0.2

0.4

0.6

0.8

# N

KT-

II C

ells

Isotype IgG

Anti-Asialo GM1

0.0

0.4

0.8

1.2

# N

KT-

I Cel

ls

Isotype IgG

Anti-Asialo GM1

E F

P < 0.01

CXCR4WT CXCR4D/D

CXCR4WT CXCR4D/D

P < 0.01

800

600

400

200

0

Lung

tum

or w

eigh

t (m

g)

% N

K C

ells

/CD

45+ c

ells

% N

K C

ell-e

xpre

ssin

g IF

Ng

Sham

Asialo G

M1

P < 0.01

25

20

15

10

5

0

B15

10

5

0

Figure 2.

CXCR4MyeD/Dmice acquire antitumor immunity dependent uponNKcells.A,DepletionofNKcells attenuated antitumor immunityofCXCR4MyeD/Dmice. CXCR4MyeD/D

mice (10/group) were treated for 2 days with asialo-GM1 antibody or normal rabbit serum containing an equivalent amount of IgG. Over 93% NK cells weredepleted in asialo-GM1 antibody-treated mice. Mice were intravenously injected with 1.2 � 105 B16F0 melanoma cells. Two weeks after tumor cell implantation,the lung weight was determined and compared with that of mice not injected with B16F0 cells (Wilcoxon rank-sum test). B, Tumor NK cell infiltration. 2� 105 B16F0melanoma cells were intravenously injected into CXCR4MyeD/D mice and CXCR4WT mice (8/group). Sixteen days after injection of tumor cells, the lungtumor-infiltrating NK cells were analyzed by flow cytometry (Wilcoxon rank-sum test). C, Intratumoral quantitation of IFNg-expressing NK cells was determinedby FACS (Wilcoxon rank-sum test). D, Depletion of interstitial and vascular NK cells. C57BL/6mice (5/group) were treated with asialo-GM1 antibody or isotype IgG.Two days after treatment, micewere injected intravenously with B16F0melanoma cells (1.2� 105). Fiveminutes before euthanasia, mice were intravenously injectedwith 2 mg allophycocyanin-conjugated anti-mouse CD45 mAb. The interstitial (IST, CD45�) and marginated vascular (MV, CD45þ) populations in lung tumors wereanalyzed by flow cytometry. E, Simultaneously, the interstitial and marginated vascular populations type II NKT (E) or type I NKT (F) in lung tumors weredetermined by flow cytometry. For A–C, group comparisons were made using Wilcoxon rank-sum test.

Yang et al.





To examine theNK cell response in the absence of CD8þ T cells,NK cell surface markers of activation (CD69, CD107a, andNKG2D), inactivation (NKG2A), and intramembrane Fas ligand(FasL; CD178) were analyzed on NK cells isolated from tumorsgrowing in mice depleted of CD8þ T cells by anti-CD8 IgG, ascompared with treatment with isotype control IgG. There was nosignificant alteration in NK cell surface markers in CXCR4myeD/D

mice after CD8þ T lymphocyte depletion (SupplementaryFig. S2D). Therefore, we conclude that CD8þ T cells are notrequired for the enhanced activation of NK cells observed inCXCR4myeDD mice.

NK cells that express IFNg were reported to have greater anti-tumor activity (18). To further study the mechanisms of NK cellsantitumor immunity mediated by myeloid CXCR4, we examinedthe expression of IFNg in NK cells infiltrating the melanomatissues in the lung of CXCR4MyeD/D mice and CXCR4WT mice afteri.v. injection of melanoma tumor cells. The results indicate that2-fold more NK cells infiltrated into the metastatic tumor site ofCXCR4MyeD/D mice in comparison with that of CXCR4WT miceafter tail-vein injection with B16F0 melanoma cells (Fig. 2B,P < 0.01, n ¼ 8). The percentage of IFNg-expressing NK cells inthe lung was 2.1-fold higher in the tumor-bearing CXCR4MyeD/D

mice than in CXCR4WT mice (Fig. 2C, P < 0.01, n ¼ 8).To evaluate whether NKT cells might contribute to the total NK

antitumor immunity in the CXCR4mye-/- mice, NKT cells weremonitored in vascular and tumor tissues of B16F0 melanoma–bearing mice after treatment with anti–asialo-GM1 to deplete NKcells (Fig. 2D). Type II NKT with properties of immune-suppres-sion, but not type I NKT cells with properties of enhancedantitumor immunity, were depleted in lung tumors after treat-ment with asialo-GM1 antibody (Fig. 2E and F). Because anincrease in tumor growth was observed when asialo-GM1 anti-body was used to deplete both type II NKT cells and NK cells,leaving the antitumor activity of the type I NKT cells intact, ourdata suggest that NK cells are primarily responsible for theantitumor response modulated by loss of CXCR4 in myeloidcells (Supplementary Fig. S2A).

Myeloid CXCR4 promotes innate immunity through theFas/FasL signaling pathway

To ascertain whether the myeloid CXCR4 signaling pathway isrequired for mediation of tumor-specific NK cell cytotoxicity,Yac-1 cells were genetically engineered to express firefly luciferase(Fluc) as a reporter permitting the imaging and quantification ofFluc expression in vivoover a time course of 1, 4, and 24hours afterinjection of luciferin. Subsequently, NK-sensitive Fluc-expressingtarget Yac-1 cells (8 � 106) were injected via tail vein intoCXCR4MyeD/D or littermate CXCR4WT mice (8 mice/group).We observed that target Yac-1-Fluc cells remaining in theCXCR4MyeD/D mice were reduced by 60% 4 hours after injectionof luciferin (P¼ 0.02) and 97%24 hours after injection (P < 0.01)compared with that of CXCR4WT mice (Fig. 3A and B), but not atthe 1-hour postinjection time point (P ¼ 0.67). To confirm thisfinding, NK cells were isolated by negative selection from thespleens ofCXCR4MyeD/Dor littermate CXCR4WTmice. The purifiedNK cells were cocultured with Yac-1-Fluc cells at different ratios asindicated in vitro for 4 hours, and subsequent NK cytotoxicity wasevaluated. In comparison with the NK cells from CXCR4WT mice,the cytotoxicity ofNK cells fromCXCR4MyeD/Dmice demonstratedenhanced killing over a range of effector to target (E/T) ratios(Fig. 3C).

To examine the cytokine profiles in the mice, serum wascollected from CXCR4MyeD/D or CXCR4WT mice and subjected tocytokine array analysis detecting 62 cytokines (SupplementaryTable S1). Results in Fig. 3A show that FasL was 7.9 � 0.5-foldlower in CXCR4MyeD/D mice compared with CXCR4WT mice,whereas other cytokineswereunchangedor alteredonly by2-fold.Membrane-bound FasL, but not soluble FasL, is essential forFas-mediated cytotoxic activity (19). This promptedus to examinethe Fas/FasL expression on immune cells. Transmembrane FasL(CD178) expression by peripheral leukocytes was determined byflow cytometry in CXCR4WT mice and CXCR4myeD/D mice. Weobserved a 7.1-fold lower FasL-expressing macrophages inCXCR4MyeD/D mice (P < 0.001) in CXCR4myeD/D comparedwith CXCR4WT mice. There was also 9.2-fold reduction inFasL-expressing neutrophils (P < 0.001) but a 1.6-fold increasein FasL-expressing B cells (P < 0.001) and DCs (P < 0.01) inCXCR4MyeD/D mice. FasL expression in NK cells was significantlyincreased by 2-fold (P < 0.05), whereas FasL expression remainedunchanged in CD4þ and CD8þ T lymphocytes in CXCR4MyeD/D

mice (Fig. 3D). Nevertheless, there were no significant differencesin the expression of Fas receptor on immune cells fromCXCR4MyeD/D mice as compared with CXCR4WT mice, except forF4/80þ macrophages, B220þ B cells, and CD11cþ dendritic cells,where expression of the FAS receptor (CD95) was alteredby �10% in CXCR4MyeD/D mice as compared with CXCR4WT

(Supplementary Fig. S3B).To examine whether FasL-mediated signals affect NK cell cyto-

toxicity, NK cells were prepared from spleens of CXCR4MyeD/D

mice. Antibody blocking FasL (10 mg/mL) or isotype-matchedcontrol antibody (10 mg/mL)was added into coculture ofNK cellsand Yac-1-Fluc cells to neutralize soluble FasL in themedium andtransmembrane FasL on NK cell surface. FasL neutralization onNK cells inhibited cytotoxicity against target Yac-1 cells (P < 0.01,n¼5; Fig. 3E). Thisfinding suggests that FasLdirectlymediatesNKcell cytotoxicity in the YAC-1 tumor model.

NK cells also kill target cells by releasing antitumor cytotoxicmolecules such as perforin and granzymes. Surface expression ofCD107a is a functional marker of NK-cell degranulation. Toinvestigate whether NK-cell degranulation was also a mechanismfor the enhanced FasL-mediated NK-cell antitumor cytotoxicityin vivo in CXCR4myeDD mice, Yac-1 cells (8 � 106) were intrave-nously injected into CXCR4WT or CXCR4MyeD/D mice (5 mice/group). Peripheral leukocytes were sampled over a time course of0, 1, 4, and 24 hours after Yac-1 cell injection. The inducedexpression of CD107a on the NK cells was analyzed by flowcytometry. We found that NK cells exhibited this degranulationmarker especially at the 4-hour time point, but the percentage ofCD107þNK cells did not vary betweenCXCR4WT or CXCR4MyeD/D

mice (Supplementary Fig. S3C). Moreover, our finding that NKcell number in peripheral blood was altered over the short timecourse after injection of YAC1 cells in vivo (Supplementary Fig.S3D) suggests that the circulating NK cells in CXCR4MyeD/D miceare depleted as they move into the lung to attack the YAC1 cells.Because a greater percentage of NK cells from CXCR4MyeD/D miceexpress FASL (Fig. 3D), it is presumed that these cells are availablefor YAC1 killing through both degranulation and the Fas/FasLdeath pathway.

Neutrophils are essential for enhanced NK-cell activity in vivoTo identify which myeloid cells (neutrophil or macrophage)

initiated NK-cell cytotoxicity in vivo, over 95% neutrophils were






CXCR4WT CXCR4Δ/ΔCtrl

4 h

24 h

1 h

ImagingtimeA B

C

D

E

0

10

20

30

40

50

F480Mac

Ly6GNeu

B220B cells

CD4CD4

CD8aCD8

CD11cDC

NK1.1NK

% C

ells

exp

ress

ing

FasL

CXCR4WT

CXCR4

2e+05

3e+05

4e+05

5e+05

6e+05

7e+05

20 10 5 1E/T Ratio

Yac1

-fluc

repo

rter

CXCR4WT

CXCR4

P < 0.01

P < 0.05P < 0.01

P < 0.01 P < 0.01 P < 0.01

P < 0.01P < 0.01

P < 0.01

1 4 24Imaging time (h)

P < 0.01

P < 0.01

ns CXCR4WT

CXCR4 Δ/Δ

Δ/Δ

Δ/Δ

Luminescence

2,000

1,500

1,000

500

Counts

Yac1

alone

NK+lgG

NK+Fas

L Ab

8.0 ¥ 106

Yac

1-L

uci

rep

ort

er6.0 ¥ 106

4.0 ¥ 106

2.0 ¥ 106

0

Yac

1-F

luc

rep

ort

er

250,000200,000150,000100,000

40,000

20,000

0

Figure 3.

CXCR4/CXCL12 negatively regulates NK cell cytotoxicity. A, NK-cell cytotoxicity in vivo. 1 � 108 Yac1-Fluc cells were intravenously injected into CXCR4MyeD/D orCXCR4WT mice. Images were taken at 1, 4, or 24 hours after Yac1 cells injection. Animals without Yac1 cell injection served as negative imaging backgroundcontrol (Ctrl). Datawere expressed as logmean� SD (n¼ 8,Wilcoxon rank-sum testwithBHP value correction).B,Representative photos of imagedmice.C,NK-cellcytotoxicity in vitro. Murine NK cells were negatively isolated from the spleen of the CXCR4MyeD/D or CXCR4WT mice. The NK cells were cocultured with Yac1-Flucreporter cells for 4 hours. Luciferase activity was determined for the un-lysed Yac1 cells to reflect the reverse NK-cell killing activity. Data are expressed asmean � SD (n ¼ 3) and statistically analyzed by two-way ANOVA with least-squares means using (BH) P value adjustment for multiple comparisons. D,mFasL (CD178) expression on immune cells. Peripheral immune cells from CXCR4MyeD/D mice (n ¼ 5) or CXCR4WT mice (n ¼ 5) were stained with CD178-APCand F4/80-BV421, Ly6G-APC/Cy7, CD8-Alexa Flour700, CD4-pacific blue, B220-Alexa Flour700, Nk1.1-APC/Cy7, and CD11c-AF700. The cells were sorted andanalyzed by flow cytometry for CD178 expression on the various cell populations. Two-wayANOVAwithmodel-basedmean comparisons and BHP value correction.E, Neutralization of FasL affects NK-cell cytotoxicity ex vivo. NK cells were negatively selected from the spleens of CXCR4MyeD/D mice and cocultured withYac1-luci reporter cells at a 20:1 ratio for 5 hours. Anti-FasL antibody (10 mg/mL) or isotype IgG was applied. The luciferase activity was determined in thesurviving target cells (n ¼ 5). P values were determined by Wilcoxon rank-sum test to evaluate differences between NKþIgG and NKþFasL mAb.

Yang et al.





eliminated from mice by injection of specific Ly6G mAb intoCXCR4MyeD/D mice (Supplementary Fig. S4A). This resulted in a46.2% decline in NK cell cytotoxicity toward target Yac-1 cells byat 4 hours (P¼ 0.02, n¼ 6), and 83% at 24 hours (P < 0.01, n¼ 6)compared withmice treated with IgG2a isotypemAb (Fig. 4A andB). Next, we sought to investigate the molecular determinants ofNK-cell activation by neutrophils. To accomplish this, murineneutrophils were depleted with 250 mg of Ly6G mAb for 3 daysfollowedby intravenous injectionwithYac-1 cells for the effector–target cell interaction in lungs. After a 24-hour interaction, lungleukocytes were isolated, and NK cells were identified byflow-cytometric analysis. CD3�NK1.1þ cells expressing CD69,NKG2A, NKG2D, CD27, and intracellular IFNg and Ki67 werequantified. We observed that in comparison with the neutrophil-sufficient CXCR4MyeD/D mice, neutrophil-deficient CXCR4MyeD/D

mice showed significant reduction of NK cells expressing CD69,

IFNg , and Ki67 (P < 0.05, n ¼ 6, Fig. 4C). However, there was noreduction inNK cells expressing the differentiationmarkers CD27or NKG2A. Thus, neutrophils are required for NK-cell prolifera-tion and activation, but not for NK-cell maturation, based onCD27 and NKG2A expression.

We next sought to determine whether macrophages contribut-ed to the activationofNK cells. The polarization/differentiation ofmacrophages was blocked by daily oral gavage of 200 mg ofBLZ945, a CSF-1R inhibitor, for 10 days (20). Results demon-strated thatNK-cell cytotoxicity, asmonitored by YAC-1 luciferaseactivity in vivo, was not influenced by treatment of mice withCSF-1R inhibitor (P > 0.05, n ¼ 4; Supplementary Fig. S4B). Inaddition, when mouse macrophages were physically depleted bydelivery of clodronate (1 mg), similar results were achieved aswhen macrophages were functionally inactivated with BLZ945(Supplementary Fig. S4C).

IgG2a Ly6GCtrl

4 h

24 h

1 h

Imagingtime

A B

C

0

20

40

60

CD69 IFNg CD27 NKG2A NKG2D Ki67

% N

K C

ells

exp

ress

ing

Intact neutrophilsDepleted neutrophils

CXCR4wt CXCR4 Δ/Δ

ns

P < 0.01

P < 0.01 P < 0.05

1 4 24Imaging time (h)

P < 0.01

P < 0.01

nsLy6GIgG2a

Luminescence2,000

1,500

1,000

500

Counts

Yac

1-F

luc

rep

ort

er

1,500,000

1,000,000

500,000

80,00060,00040,00020,000

0

% M

DS

C in

lun

g t

um

or

D

40

30

20

10

0

Figure 4.

Neutrophils are responsible for NK cytotoxicity in vivo. A, NK cell cytotoxicity toward YAC1 cells in neutrophil depleted mice. The neutrophils in CXCR4MyeD/D micewere depleted with 250 mg/Ly6G mAb as described in Materials and Methods. Subsequently 1 � 108 Yac1-Fluc cells were intravenously injected into the mice.Imaging was performed at indicated time points after Yac1 cells injection. Mice without Yac1 cell injection served as the negative imaging background control (Ctrl).Data analysis was performed with Wilcoxon rank-sum test with BH P value correction. B, Representative photos of imaging of mice. C, Depletion ofneutrophils abrogatedNK activation. Micewere injected in the peritoneumwith 250 mg of Ly6GmAbor IgG isotype control daily for 3 days, and then 1� 108 Yac1 cellswere intravenously injected. Twenty-four hours after cell injection, the lung leukocytes were isolated, and NK cells were sorted using Percy/cy5.5 conjugatedCD3 and APC-conjugated NK1.1þ cells by FACS. Subsequent cell-surface expression of CD69-APC, NKG2A-APC, NKG2D-APC, CD27-pacific blue, andintracellular IFNg-Alex Fluor700 and Ki67-pacific blue was analyzed. Data were analyzed by two-way ANOVAwith model-basedmean comparisons and BH P valuecorrection.D,MDSCpopulation inmetastatic tumor. 2� 105B16F0melanoma cellswere intravenously injected intoCXCR4MyeD/Dmice andCXCR4WTmice (8/group).After 16 days of injection, the lung tumor infiltrated MDSCs were analyzed by flow cytometry. The Wilcoxon rank-sum test was used for data analysis.






CXCR4 is expressed on myeloid-derived suppressor cells(MDSC), and its inhibition reduces MDSC recruitment to tumors(21). Thus, loss ofCXCR4onmyeloid cells could result in reducedrecruitment of MDSCs to tumors, which could account for thereduction in tumor growth in CXCR4myeD/D mice. To investigatethe role of MDSCs in metastatic lung melanoma of CXCR4MyeD/D

mice, the percentages of tumor-infiltrating CD11bþ/Ly6Gþ/Ly6Clowmyeloid cells were analyzed by flow cytometry.We failedto observe a difference in the size of the tumor MDSC populationbetweenCXCR4MyeD/Dmice andCXCR4WTmice (Fig. 4DP¼0.17,n ¼ 8).

NK cell activation requires cytokine participationIL18 can activate NK cells (22) and invariant NK T cells (iNKT)

cells (23). Neutrophils are the principal source of IL18 (22).Expression of IL18 was quantified in neutrophils and macro-phages from CXCR4WT and CXCR4MyeD/D mice. Twenty-fourhours after mice were intravenously injected with 8 � 106 Yac-1cells, intracellular IL18 of lung myeloid cells was determined.Neutrophils deficient in CXCR4 showed enhanced IL18 expres-sion, in comparison with CXCR4WT neutrophils (Fig. 5A and B,84.2% � 5.4% vs. 56.6% � 8.0%, P < 0.01, n ¼ 6). To confirmwhether neutrophil-derived IL18 may provide a potential source

Figure 5.

Neutrophils activate NK cells through cytokines. A, Yac1 cells (1 � 108) were intravenously injected into myeloid CXCR4D/D or CXCR4WT mice. Twenty-fourhours after injection, intracellular IL18 expression in the lung myeloid cells was determined by FACS. Data were analyzed by the Wilcoxon rank-sum testwith BH P value correction (n ¼ 6). B, A representative graph is shown. C, IL18 expression. B16F0 melanoma cells (2 � 105) were intravenously injected intoCXCR4MyeD/D or CXCR4WT mice (n ¼ 8). Sixteen days after injection, IL18 in lung tumors was determined with mouse IL18 ELISA. D, Simultaneously, IL18 levelin serum was measured. For C and D, data were analyzed by Wilcoxon rank-sum test.

Yang et al.





for NK activation in metastatic lung tumor, IL18 concentrationsin the lung tumor tissue and serum were determined 16 daysafter B16F0 melanoma cells were intravenously implantedinto CXCR4WT and CXCR4MyeD/D mice. IL18 concentration wassignificantly increased in either of tumor tissue or serum ofCXCR4MyeD/D mice compared with CXCR4WT mice (Fig. 5C andD, tumor tissue 1.69-fold increase; serum 1.5-fold increase, P <0.01, n¼ 8). Exogenous delivery of CXCL12 significantly reducedthe production of IL18 by cultured bonemarrow cells over 2 daysby 1.7-fold (Supplementary Fig. S5C, P < 0.01, n ¼ 8). Given theobserved increase in annexin V staining after 2 days of CXCL12treatment (Supplementary Fig. S1H), we cannot rule out thepossibility that some of the reduction is based upon reducedviability of CXCL12-treated cells. Thus, neutrophils with CXCR4deficiency promote NK-cell activation via enhanced IL18production.

CXCR4 antagonist enhances NK cell antitumor activityindependent of CD8þ T cells

A potent and highly selective CXCR4 antagonist, the smallcyclic peptide LY2510924, is currently in clinical trials for met-astatic renal cancer, relapsed or refractory acutemyeloid leukemia(AML), and other advanced refractory solid tumors (24). Thistargeted therapy is based on the mechanism that CXCR4 is over-expressed in a variety of human cancers where it directly stimu-lates tumor cell proliferation and metastasis, leading to pooroverall survival. To investigate the potential effect of LY2510924on metastatic tumor progression in vivo, immunocompetentC57BL/6 mice were intravenously injected with syngeneic malig-nant B16F0 melanoma cells and then treated with 3 mg/kgsubcutaneous injection of LY2510924 twice a day (25). After 2weeks of treatment, the tumor burden in the lungs was reduced61% compared with PBS vehicle control treated animals (P <0.001, n ¼ 9; Fig. 6A). Drug treatment did not affect mouse bodyweight, indicating reduced tumor burden was not due to toxicity.Mice treated with the CXCR4 antagonist exhibited alterations inthe peripheral blood immune cell profile (Fig. 6B), which weresimilar to changes induced byCXCR4 loss (Fig. 1E). BothmyeloidCXCR4 deletion and CXCR4 antagonism resulted in elevation ofboth neutrophil and NK cell populations (P < 0.01, n ¼ 6). Toexamine antitumor immunity in the tumor microenvironment,tumor tissue was digested, and tumor-infiltrating leukocytes werestained with fluorescently labeled antibodies and analyzed byflow cytometry. We observed that the ratio of NK cells to totalCD45þ cells within tumor was 1.6-fold higher in mice treatedwith LY2510924 compared with PBS-treated controls (P < 0.001,n ¼ 8, Fig. 6C). The activation marker CD69 on NK cells was1.9-fold higher in NK cells from LY2510924-treated mice com-pared with PBS-treated control mice (P < 0.001, n ¼ 8; Fig. 6D).However, the expression of cell-surface markers, i.e., CD25,CD62L, andCD69, or intracellularmarkers, Foxp3, IL4, and IFNg ,on tumor-infiltrating CD3þ/CD4þ T cells was not altered byLY2510924 treatment of tumor-bearing mice (SupplementaryFig. S6A and S6B). In contrast, the tumor-infiltrating CD103þ

CD8þ T cells were significantly decreased after LY2510924 treat-ment (P < 0.001, n ¼ 8), whereas alterations in cell-surfacemarkers CD69, CD107b, and PD-1 were not observed on theCD8þ T cells in these tumors (Supplementary Fig. S6C).

We also examined the effects of LY2510924 on the growth oforthotopic implants of B16F0melanoma tumors (SupplementaryFig. S6D–S6F) and achieved similar results as observed with

melanoma tumors in the lung (Fig. 6A–C). Because IL18 acts asan NK cell activator (26), intracellular IL18 production in theinfiltrating cells was investigated. We observed that LY2510924induced significant IL18production in tumor-infiltratingmyeloidcells as compared with PBS control, with its major source comingfrom neutrophils and a minor source from macrophages andCD4þ T cells (Fig. 6E). Thus, the CXCR4 antagonist LY2510924confers antitumor activity, at least in part through induction ofIL18 that can activate NK cells.

To determine whether a subset of melanoma patients maybenefit from CXCR4 inhibition, we examined the survival ofmelanoma patients with high expression of the myeloid markerITGAM, for which themRNA expression z-score thresholdwasþ1(Supplementary Fig. S7). When survival data were analyzed inrelation to a high or low CXCR4 expression using a TCGA data setof 469 skin cutaneous melanoma samples, analysis revealedthat melanoma patients with high levels of CXCR4 exhibited asignificantly reduced progression-free survival (PFS) comparedwith patients with a low expression of CXCR4. In contrast, whenPFS in all the melanoma tumors (without selecting the highITGAM tumors) was examined, CXCR4 expression did not affectthePFS (Supplementary Fig. S7A andS7B). These data suggest thatmelanoma clinical trials with a CXCR4 antagonist might considerselecting patientswith tumorswith a high frequency of infiltratingmyeloid cells.

DiscussionThe CXCR4/CXCL12 signaling pathway is involved inmultiple

stages of tumorigenesis. CXCL12 is highly expressed in the tumormetastatic sites such as the liver, lung, lymph nodes, and bonemarrow and is also secreted by fibroblasts in vitro and in vivo (27).CXCR4 is overexpressed by tumor cells, and this expression isconsistent with increased recurrence and poor overall survival inmultiple cancers, including breast, lung, kidney, colon, ovarian,brain cancers, lymphoma, and leukemia (27–29). The criticalroles of CXCR4 in cancer have triggered the development ofspecific antagonists for clinical application (30, 31).

Cytokines modulate innate immunity responses to cancers(22, 23, 26, 32). The importance of IL18 has been recognizedin NK-cell activation and in direct induction of IFNg productionby NK cells (33). Thus, the crucial effector cytokine that controlsNK-cell activation is IFNg , the bulk of which is derived from NKcells (33). Jaeger and colleagues suggested a role for neutrophils asnonredundant regulatory cells ensuring the terminal maturationof NK cells; neutrophil-induced NK-cell maturation may occurnot only in the bone marrow but also at the peripherywhere neutrophils are able to interact with NK cells (34). Neu-trophils provide a major source of IL18, which induces NK-cellactivation. However, CXCR4/CXCL12 signaling is also essentialfor development of NK cells in MX-Cre-generated CXCR4-deficient mice (35), underscoring discriminatory power ofCXCR4/CXCL12 deficiency in liver and lymphocytes. Our resultsshow that elevations in IL18 in various leukocytes may trigger theenhanced NK cell–mediated tumor-cell killing observed inCXCR4myeD/D mice.

NK cells acquire effector function through a licensing processand exert an antitumor effect, which is also consistentwith reportsdescribing neutrophils as pivotal activators of NK cells and NKTcells (23, 36). This study sought to test the hypothesis thatCXCR4null neutrophils enhance NK cell antitumor immunity






through a Fas/FasL signal pathway. FasL is an apoptosis-inducingmember of the TNF cytokine family, and its receptor Fas plays anessential role in the shutdown of chronic immune responses (37).Tumor-associated fibroblasts secrete CXCL12 (31), which poten-

tially contributes to the induction of apoptosis in CD4þ andCD8þ T lymphocytes via a CXCR4 receptor by upregulation ofthe Fas/FasL signal pathway (38, 39). The effector mechanism ofNK-cell cytotoxicity may also be through the perforin-mediated

B

E

0

20

40

60

CD11b Mac Neu CD4 CD8 B cell NK DC

% P

erip

hera

lCD

45+

cells PBS

LY2510924

0

20

40

60

80

CD11c F480 Ly6G CD4 CD8 CD19

% IL

18 e

xpre

ssin

g ce

lls PBSLY25109024

P < 0.01

P < 0.01

P < 0.01

P < 0.01

P < 0.01 P < 0.01P < 0.01

P < 0.01

Lung

tum

or w

eigh

t (m

g)

C D

% C

D69

+ N

K c

ells

% N

K1.

1+ C

D45

cel

ls

A

Figure 6.

The CXCR4 antagonist LY2510924 inhibits tumor growth and induces NK activation. A, LY2510924 antitumor activity. C57BL/6 mice (9 mice/group) wereintravenously injected with B16F0melanoma cells (1� 105; ref. 14) and treated subcutaneously with 3mg/kg of LY2510924, or PBS vehicle twice a day. After 2 weeksof treatment, mice were sacrificed and the weight of tumor in tumor-bearing lungs was determined by subtracting the weight of the tumor-free lung.Data were plotted and statistically analyzed by the Wilcoxon rank-sum test. B, Peripheral immune cell profile. Blood was collected 2 weeks after PBS orLY2510924 treatment. Erythrocytes were excluded with lysis buffer and leukocytes were stained with specific antibodies for flow cytometry analysis using two-wayANOVA with model-based mean comparisons and BH P value correction. C and D, NK-cell distribution and activation. Leukocytes were isolated from thetumor-bearing lungs, stained with CD45-APC/Cy7, NK1.1-PE, and CD69-APC and analyzed by flow cytometry (Wilcoxon rank-sum test). E, Intracellular IL18expression in leukocytes that infiltrated into lung tumors was analyzed by flow cytometry. Data were analyzed by two-way ANOVA with model-based meancomparisons and BH P value correction.

Yang et al.





granule exocytosis pathway (40). However, because membrane-bound FasL was highly expressed on NK cells of CXCR4myeD/D

mice, it is likely that the increased NK-cell activity toward theYAC-1 tumor cells in the CXCR4myeD/D mice is, at least in part,FASL dependent (41). Our data suggest that the process ofneutrophils licensing NK-cell antitumor cytotoxicity is controlledin part by CXCR4/CXCL12-Fas/FasL signal cascades.

Systemic blockade of CXCR4 with the CXCR4 antagonistLY2510924 resulted in enhanced production of IL18 byneutrophils as well as enhanced antitumor immunity, provid-ing insight into how CXCR4 antagonists may be useful in theclinic for treatment of cancer. In agreement with our results,Alterio and colleagues reported that interference of stromalCXCR4 signaling by using heterozygote CXCR4þ/� lineagemice or the CXCR4 antagonist AMD3100 inhibits lung metas-tasis (42). We conclude that the CXCR4/CXCL12 axismodulates innate immune responses through inhibition ofneutrophil licensing of NK cell cytotoxicity, and loss of thisCXCR4-mediated role results in enhanced NK-cell activation,which is neutrophil-dependent and correlates with FasL andIL18 expression (Fig. 7).

NK cells contribute to murine antitumor immunity and areassociated with clinical prognosis in human cancers (43, 44).Consistent with our observation that CXCR4 antagonismenhancedNK cell antitumor activity through amechanism involv-ing neutrophils, we showed that melanoma patients with highCXCR4 expression combined with infiltration of tumor-associat-ed myeloid cells had poorer PFS than those with low CXCR4expression. CXCR4 antagonists are entering clinical trials formetastatic melanoma and breast cancer (45, 46), and phase Iclinical trials have shown that combining CXCR4 antagonist withpembrolizumab is safe (45). Based upon the data reported here,we anticipate that patients with tumors that have high myeloidinfiltration and high CXCR4 expression will benefit from

these antagonists, especiallywhen combinedwith immune check-point inhibitors.

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

Authors' ContributionsConception and design A. RichmondDevelopment of methodology: J. Yang, A. Kumar, S.V. NovitskiyAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): J. Yang, A. Kumar, A. RichmondAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): J. Yang, A. Kumar, A.E. Vilgelm, S.-C. Chen,G.D. Ayers, A. RichmondWriting, review, and/or revision of the manuscript: J. Yang, A. Kumar,A.E. Vilgelm, S.-C. Chen, G.D. Ayers, S. Joyce, A. RichmondStudy supervision: A. RichmondOther (supervised the experimental plan and analyses of NK and NKT cellsafter in vivo depletion of cells with anti-asialo-GM1 treatment): S. Joyce

AcknowledgmentsThis work was supported by awards from theDepartment of Veterans Affairs,

TVHS through an SRCS Award and a MERIT Award to A. Richmond(101BX002301), by grants from the NIH CA34590 (to A. Richmond),CA200681 (to S.V. Novitskiy), and CA P30-068485. It is also supported by aCDAAward toA.E. Vilgelm from theHarry LloydCharitable Trust forMelanomaResearch and by a Breast Cancer Research Foundation Award to A.E. Vilgelm(IIDRP-16-001). We are thankful to Andrew C. Johnson for outstandingtechnical support and to Mark Boothby for scientific discussions and advice.

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 February 7, 2018; revised June 7, 2018; accepted August 9, 2018;published first August 14, 2018.

IL18

A B

mFasL Promote Inhibit

CXCR4wt Fas

IFNγ

IL18

CXCR4 Fas

IFNγ

D/D

Figure 7.

Model summarizing the proposed signaling events in the myeloid CXCR4 deletion mice. A, myeloid cells in CXCR4WT mice promote NK cell apoptosisthrough the Fas signal pathway and inactivation of NK cells through restricted IL18 production, resulting in reduced NK cell antitumor immunity. B, Knockoutof CXCR4 in myeloid cells enhances neutrophil release of IL18, which boosts the percentage of IFNg-expressing NK cells. These activated NK cells exhibitenhanced NK cell–mediated tumor cell killing. NK cells from mice with knockout of CXCR4 in myeloid cells exhibit increased Fas ligand–mediated killing ofFas-expressing tumor cells, conferring enhanced antitumor immunity.






References1. Ganju RK, Brubaker SA, Meyer J, Dutt P, Yang Y, Qin S, et al. The alpha-

chemokine, stromal cell-derived factor-1alpha, binds to the transmem-brane G-protein-coupled CXCR-4 receptor and activates multiple signaltransduction pathways. J Biol Chem 1998;273:23169–75.

2. Chatterjee S, Behnam Azad B, Nimmagadda S. The intricate role of CXCR4in cancer. Adv Cancer Res 2014;124:31–82.

3. Wong D, Korz W. Translating an Antagonist of Chemokine ReceptorCXCR4: from bench to bedside. Clin Cancer Res 2008;14:7975–80.

4. Kryczek I, Wei S, Szeliga W, Vatan L, ZouW. Endogenous IL-17 contributesto reduced tumor growth and metastasis. Blood 2009;114:357–9.

5. McDermott DH, Liu Q, Velez D, Lopez L, Anaya-O'Brien S, Ulrick J, et al. Aphase 1 clinical trial of long-term, low-dose treatment of WHIM syndromewith the CXCR4 antagonist plerixafor. Blood 2014;123:2308–16.

6. ViolaA, LusterAD.Chemokines and their receptors: drug targets in immunityand inflammation. Ann Rev Pharmacol Toxicol 2008;48:171–97.

7. BierieB,ChungCH,Parker JS, StoverDG,ChengN,ChytilA, et al. Abrogationof TGF-beta signaling enhances chemokine production and correlates withprognosis in human breast cancer. J Clin Invest 2009;119:1571–82.

8. Sai J, Owens P, Novitskiy SV, Hawkins OE, Vilgelm AE, Yang J, et al. PI3Kinhibition reduces mammary tumor growth and facilitates antitumorimmunity and Anti-PD1 responses. Clin Cancer Res 2017;23:3371–84.

9. Pickup MW, Hover LD, Polikowsky ER, Chytil A, Gorska AE, Novitskiy SV,et al. BMPR2 loss in fibroblasts promotes mammary carcinoma metastasisvia increased inflammation. Mol Oncol 2015;9:179–91.

10. DankortD,CurleyDP,Cartlidge RA,NelsonB, Karnezis AN,DamskyWE Jr,et al. Braf(V600E) cooperates with Pten loss to induce metastatic mela-noma. Nat Genet 2009;41:544–52.

11. Gordon S. Alternative activation of macrophages. Nat Rev Immunol2003;3:23–35.

12. StrachanDC, Ruffell B,Oei Y, BissellMJ, Coussens LM, PryerN, et al. CSF1Rinhibition delays cervical and mammary tumor growth in murine modelsby attenuating the turnover of tumor-associatedmacrophages and enhanc-ing infiltration by CD8(þ) T cells. Oncoimmunol 2013;2:e26968.

13. Saxon JA, Sherrill TP, Polosukhin VV, Sai J, Zaynagetdinov R, McLoed AG,et al. Epithelial NF-kappaB signaling promotes EGFR-driven lung carcino-genesis via macrophage recruitment. Oncoimmunol 2016;5:e1168549.

14. Yang J, Hawkins OE, Barham W, Gilchuk P, Boothby M, Ayers GD, et al.Myeloid IKKbeta promotes antitumor immunity by modulating CCL11and the innate immune response. Cancer Res 2014;74:7274–84.

15. Crowe NY, Smyth MJ, Godfrey DI. A critical role for natural killer T cells inimmunosurveillance ofmethylcholanthrene-induced sarcomas. J ExpMed2002;196:119–27.

16. Yang J, Splittgerber R, Yull FE, Kantrow S, Ayers GD, Karin M, et al.Conditional ablation of Ikkb inhibits melanoma tumor development inmice. J Clin Invest 2010;120:2563–74.

17. Kumar A, Bezbradica JS, Stanic AK, Joyce S. Characterization and functionalanalysis of mouse semi-invariant natural T cells. Curr Protoc Immunol2017;117:14.13.1–55.

18. PietraG,ManziniC,RivaraS,VitaleM,CantoniC,PetrettoA, et al.Melanomacells inhibit natural killer cell function by modulating the expression ofactivating receptors and cytolytic activity. Cancer Res 2012;72:1407–15.

19. O'Reilly LA, Tai L, Lee L, Kruse EA, Grabow S, Fairlie WD, et al. Membrane-bound Fas ligand only is essential for Fas-induced apoptosis. Nature2009;461:659–63.

20. Pyonteck SM, Akkari L, Schuhmacher AJ, Bowman RL, Sevenich L, QuailDF, et al. CSF-1R inhibition alters macrophage polarization and blocksglioma progression. Nat Med 2013;19:1264–72.

21. Yang L, Huang J, Ren X, Gorska AE, Chytil A, Aakre M, et al. Abrogation ofTGF beta signaling in mammary carcinomas recruits Gr-1þCD11bþmyeloid cells that promote metastasis. Cancer Cell 2008;13:23–35.

22. Sporri R, JollerN,HilbiH,Oxenius A. Anovel role for neutrophils as criticalactivators of NK cells. J Immunol 2008;181:7121–30.

23. Hagglof T, Sedimbi SK, Yates JL, Parsa R, Salas BH, Harris RA, et al.Neutrophils license iNKT cells to regulate self-reactive mouse B cellresponses. Nat Immunol 2016;17:1407–14.

24. GalskyMD,VogelzangNJ,Conkling P, RaddadE, Polzer J, Roberson S, et al.A phase I trial of LY2510924, a CXCR4 peptide antagonist, in patients withadvanced cancer. Clin Cancer Res 2014;20:3581–8.

25. Peng SB, Zhang X, PaulD, Kays LM,GoughW, Stewart J, et al. Identificationof LY2510924, a novel cyclic peptide CXCR4 antagonist that exhibitsantitumor activities in solid tumor and breast cancer metastatic models.Mol Cancer Ther 2015;14:480–90.

26. Trinchieri G. Interleukin-12 and the regulation of innate resistance andadaptive immunity. Nat Rev Immunol 2003;3:133–46.

27. Muller A,HomeyB, SotoH,GeN,CatronD, BuchananME, et al. Involvementof chemokine receptors in breast cancer metastasis. Nature 2001;410:50–6.

28. Balkwill F. Cancer and the chemokine network. Nat Rev Cancer 2004;4:540–50.

29. Orimo A, Gupta PB, Sgroi DC, Arenzana-Seisdedos F, Delaunay T, NaeemR, et al. Stromal fibroblasts present in invasive human breast carcinomaspromote tumor growth and angiogenesis through elevated SDF-1/CXCL12secretion. Cell 2005;121:335–48.

30. Peng SB, Peek V, Zhai Y, Paul DC, LouQ, Xia X, et al. Akt activation, but notextracellular signal-regulated kinase activation, is required for SDF-1alpha/CXCR4-mediated migration of epitheloid carcinoma cells. Mol Cancer Res2005;3:227–36.

31. ZhouY, LarsenPH,HaoC, YongVW.CXCR4 is amajor chemokine receptoron glioma cells and mediates their survival. J Biol Chem 2002;277:49481–7.

32. Leung BP, Culshaw S, Gracie JA, Hunter D, Canetti CA, Campbell C, et al. Arole for IL-18 in neutrophil activation. J Immunol 2001;167:2879–86.

33. TakedaK, TsutsuiH, Yoshimoto T, AdachiO, YoshidaN, Kishimoto T, et al.Defective NK cell activity and Th1 response in IL-18-deficient mice.Immunity 1998;8:383–90.

34. Jaeger BN, Donadieu J, Cognet C, Bernat C, Ordonez-Rueda D, Barlogis V,et al. Neutrophil depletion impairs natural killer cell maturation, function,and homeostasis. J Exp Med 2012;209:565–80.

35. Noda M, Omatsu Y, Sugiyama T, Oishi S, Fujii N, Nagasawa T. CXCL12-CXCR4 chemokine signaling is essential for NK-cell development in adultmice. Blood 2011;117:451–8.

36. Ueda R, Narumi K, Hashimoto H, Miyakawa R, Okusaka T, Aoki K.Interaction of natural killer cells with neutrophils exerts a significantantitumor immunity in hematopoietic stem cell transplantation recipients.Cancer Med 2016;5:49–60.

37. Hughes PD, Belz GT, Fortner KA, Budd RC, Strasser A, Bouillet P. Apoptosisregulators Fas and Bim cooperate in shutdown of chronic immuneresponses and prevention of autoimmunity. Immunity 2008;28:197–205.

38. Colamussi ML, Secchiero P, Gonelli A, Marchisio M, Zauli G, Capitani S.Stromal derived factor-1 alpha (SDF-1 alpha) induces CD4þ T cell apo-ptosis via the functional up-regulation of the Fas (CD95)/Fas ligand(CD95L) pathway. J Leukoc Biol 2001;69:263–70.

39. HerbeinG,MahlknechtU, Batliwalla F,GregersenP, Pappas T, Butler J, et al.Apoptosis ofCD8þT cells ismediatedbymacrophages through interactionof HIV gp120 with chemokine receptor CXCR4. Nature 1998;395:189–94.

40. Trapani JA, Smyth MJ. Functional significance of the perforin/granzymecell death pathway. Nat Rev Immunol 2002;2:735–47.

41. Screpanti V, Wallin RP, Ljunggren HG, Grandien A. A central role for deathreceptor-mediated apoptosis in the rejection of tumors by NK cells.J Immunol 2001;167:2068–73.

42. D'Alterio C, Barbieri A, Portella L, Palma G, Polimeno M, Riccio A, et al.Inhibition of stromal CXCR4 impairs development of lung metastases.Cancer Immunol Immunother 2012;61:1713–20.

43. Palucka AK, Coussens LM. The Basis of Oncoimmunology. Cell 2016;164:1233–47.

44. Vesely MD, Kershaw MH, Schreiber RD, Smyth MJ. Natural innate andadaptive immunity to cancer. Annu Rev Immunol 2011;29:235–71.

45. Pernas S, Martin M, Kaufman PA, Gil-Martin M, Gomez Pardo P, Popez-Tarruella S, et al. Balixafortide plus eribulin in HER2-negative metastaticbreast cancer: a phase 1, single-arm, dose-escalation trial. Lancet Oncol2018;19:812–24.

46. Andtbacka RH, Pierce RH, Campbell JS, YusakM,MilhemM, Ross M, et al.X4P-001, an orally bioavailable CXCR4 antagonist, enhances immune cellinfiltration and activation in the tumor microenvironment of melanoma[abstract]. In: Proceedings of theAmericanAssociation forCancer Research;2018 Apr 15–16; Chicago, IL. Philadelphia (PA): AACR; 2018. p 155.Abstract nr 613.


Yang et al.




2018;6:1186-1198. Published OnlineFirst August 14, 2018.Cancer Immunol Res Jinming Yang, Amrendra Kumar, Anna E. Vilgelm, et al. MechanismsReduces Melanoma Growth through NK Cell and FASL Loss of CXCR4 in Myeloid Cells Enhances Antitumor Immunity and

Updated version

10.1158/2326-6066.CIR-18-0045doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerimmunolres.aacrjournals.org/content/suppl/2018/08/14/2326-6066.CIR-18-0045.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerimmunolres.aacrjournals.org/content/6/10/1186.full#ref-list-1

This article cites 45 articles, 17 of which you can access for free at:

Citing articles

http://cancerimmunolres.aacrjournals.org/content/6/10/1186.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected]

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

Permissions

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

.http://cancerimmunolres.aacrjournals.org/content/6/10/1186To 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-18-0045

http://cancerimmunolres.aacrjournals.org/content/suppl/2018/08/14/2326-6066.CIR-18-0045.DC1

http://cancerimmunolres.aacrjournals.org/content/6/10/1186.full#ref-list-1

http://cancerimmunolres.aacrjournals.org/content/6/10/1186.full#related-urls

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

mailto:[email protected]

http://cancerimmunolres.aacrjournals.org/content/6/10/1186


Loss of CXCR4 in Myeloid Cells Enhances Antitumor Immunity ... · Research Article Loss of CXCR4 in Myeloid Cells Enhances Antitumor Immunity and Reduces Melanoma Growth through NK

Documents