MK2 contributes to tumor progression by promotingM2 macrophage polarization and tumor angiogenesisLucia Suarez-Lopeza,b,c,d, Ganapathy Srirama,b,c,1, Yi Wen Konga,b,c,1, Sandra Morandella,b,c, Karl A. Merricka,b,c,Yuliana Hernandeza,b,c, Kevin M. Haigisd,e, and Michael B. Yaffea,b,c,f,2

aDepartment of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02142; bDepartment of Biology, Massachusetts Institute ofTechnology, Cambridge, MA 02142; cCenter for Personalized Cancer Medicine, David H. Koch Institute for Integrative Cancer Research, MassachusettsInstitute of Technology, Cambridge, MA 02142; dCancer Research Institute and Department of Medicine, Beth Israel Deaconess Medical Center, HarvardMedical School, Boston, MA 02215; eHarvard Digestive Disease Center, Harvard Medical School, Boston, MA 02215; and fDivisions of Acute Care Surgery,Trauma, and Surgical Critical Care and Surgical Oncology, Department of Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston,MA 02215

Edited by Melanie H. Cobb, University of Texas Southwestern Medical Center, Dallas, TX, and approved March 21, 2018 (received for review December20, 2017)

Chronic inflammation is a major risk factor for colorectal cancer.The p38/MAPKAP Kinase 2 (MK2) kinase axis controls the synthesisof proinflammatory cytokines that mediate both chronic inflammationand tumor progression. Blockade of this pathway has been previouslyreported to suppress inflammation and to prevent colorectaltumorigenesis in a mouse model of inflammation-driven colorectalcancer, by mechanisms that are still unclear. Here, using whole-animal and tissue-specific MK2 KO mice, we show that MK2activity in the myeloid compartment promotes tumor progressionby supporting tumor neoangiogenesis in vivo. Mechanistically, wedemonstrate that MK2 promotes polarization of tumor-associatedmacrophages into protumorigenic, proangiogenic M2-like macro-phages. We further confirmed our results in human cell lines, whereMK2 chemical inhibition in macrophages impairs M2 polarization andM2 macrophage-induced angiogenesis. Together, this study providesa molecular and cellular mechanism for the protumorigenic functionof MK2.

MK2 | macrophage polarization | tumor angiogenesis | inflammation |colon cancer

Colorectal cancer (CRC), including hereditary, sporadic, andcolitis-associated CRC (CAC), is the third most commonly

diagnosed cancer among both men and women in the UnitedStates. There are projected to be 135,430 individuals newly di-agnosed with CRC and 50,260 deaths from the disease in 2017(1). Strong epidemiologic and experimental evidence implicateschronic inflammation as a risk factor for developing CRC (2, 3).Indeed, patients with ulcerative colitis, a form of inflammatorybowel disease (IBD), are up to 30-fold more likely to develop aCRC (4) with the risk of cancer directly related to the severity,extent, and duration of disease (5).The precise mechanisms responsible for cellular transforma-

tion and tumor progression during chronic inflammation are notfully understood. Multiple lines of evidence demonstrate that aninflammatory context favors the development of oncogenic muta-tions and epigenetic events in epithelial cells, accompanied byfurther progression of precancerous lesions to malignant tumors(6). The inflammatory tumor microenvironment, which is composedof surrounding stroma (fibroblasts, endothelial cells, pericytes,and mesenchymal cells) and innate and adaptive immune cells(7) is a determinant for tumor progression. Within the immunecompartment, myeloid cells are major players in cancer-relatedinflammation. Myeloid cells have the potential to develop anantitumoral immune response, but they are often instructed by thetumor toward an immunosuppressive phenotype. These myeloidsuppressor cells favor tumor growth by stimulating tumor cell pro-liferation, angiogenesis, malignant progression, metastasis, andresistance to therapy (8). Myeloid cells present within tumorsinclude several different cell types: inflammatory monocytes,neutrophils (tumor-associated neutrophils, TAN), dendritic cells,

myeloid-derived suppressor cells (MDSC), and macrophages (calledtumor-associated macrophages, TAMs). During tumor evolution,TAMs generally show an alternative form of activation—referred toas M2 macrophages—which influence multiple fundamental aspectsof tumor biology. Contrary to proinflammatory, classically activatedM1 macrophages, M2 macrophages are immunosuppressive andpromote both tissue remodeling and neoangiogenesis (9).Diverse cell populations in the tumor and the tumor micro-

environment communicate with each other through cytokine andchemokine production, and act in autocrine and paracrine mannersto control and shape tumor growth, both intrinsically and ex-trinsically. Intrinsically, cytokines like IL-6, TNF-α, and IL-1β ac-tivate tumor cell proliferation, expansion, and survival through theactivation of intracellular signaling molecules, such as STAT3 andNF-κB (10). Extrinsically, cytokines are regulators of recruitment,activation state, and tumor promoting/suppressing activities ofstromal and immune cells involved in tissue remodeling, tumorangiogenesis, and metastasis (6).Mitogen-activated protein kinase (MAPK) cascades are central

components of intracellular signaling networks, as they translate

Significance

Colon cancer is the third most common cancer type and thesecond leading cause of cancer-related death. Chronic inflam-mation is a major contributor to colon cancer development andprogression, but the molecular details of how inflammationcontributes to colon cancer remain unclear. The p38MAPK/MAPKAP Kinase 2 (MK2) pathway is a central mediator of cellstress and inflammation. Here, we examine the importance ofthis pathway in an inflammation-driven model of colon can-cer using whole-body and tissue-specific MK2 knockouts, incombination with measurements of macrophage polarization,endothelial proliferation, and morphogenesis. We demon-strate that the MK2 pathway promotes colon tumor devel-opment by regulating the polarization of macrophages intoan M2 tumor-promoting state that modulates the tumor mi-croenvironment and enhances tumor angiogenesis.

Author contributions: L.S.-L. and M.B.Y. designed research; L.S.-L., G.S., Y.W.K., S.M.,K.A.M., Y.H., K.M.H., and M.B.Y. performed research; S.M. and Y.H. contributed newreagents/analytic tools; L.S.-L., G.S., Y.W.K., and K.M.H. analyzed data; and L.S.-L., G.S.,and M.B.Y. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1G.S. and Y.W.K. contributed equally to this work.2To whom correspondence should be addressed. Email: myaffe@mit.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1722020115/-/DCSupplemental.

Published online April 16, 2018.

E4236–E4244 | PNAS | vol. 115 | no. 18 www.pnas.org/cgi/doi/10.1073/pnas.1722020115

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

multiple external stimuli into specific cellular responses. Thep38αMAPK/MAPKAP Kinase 2 (MK2) axis is a master regu-lator of cellular stress, responding to a plethora of external stresssignals such as osmotic shock, heat, DNA damaging agents, in-flammatory cytokines, and pathogen-associated molecular patterns(PAMPs) to drive specific cellular responses. Following inflam-matory stimuli (i.e., Toll-like receptors, TNF-α, or IL-1 receptoractivation), activation of the p38α-MK2 pathway posttranscrip-tionally up-regulates proinflammatory cytokines and mediators in-cluding IL-1β, IL-4, IL-6, IL-8, GM-CSF, IFN-γ, TNF-α, and COX2(11). Consistent with its putative role in regulation of proinflam-matory cytokine expression, genetic ablation of MK2 was recentlyreported to abrogate tumor formation in a mouse model of colitis-associated cancer, although the underlying mechanisms remainunclear (12).Here, we show that MK2 function within myeloid cells is the

most relevant activity necessary for tumor promotion under con-ditions of chronic inflammation. We observe that MK2 is crucialfor M2 activation of tumor-associated macrophages and tumorneoangiogenesis. These findings support myeloid-specific blockadeof MK2 as a chemopreventive strategy for IBD patients to preventCRC progression.

ResultsMK2 Promotes Tumorigenesis in a Murine Model of Colitis-AssociatedCancer. To evaluate the role of MK2 in colitis-associated tumordevelopment, we crossed our previously described mouse modelfor targeted and reversible MK2 inhibition (MK2 Cre-versible orCV) (13) with a β-actin–Cre driver mouse line to generate whole-body MK2 knockout (KO) mice (Fig. S1 A and B). Loss of MK2protein was confirmed both by Western blot (WB) and immu-nohistochemistry on diverse murine tissues (Fig. S1 C and D).Inflammation-driven colon tumorigenesis was modeled in these

MK2 KO mice using the azoxymethane (AOM)/dextran sodiumsulfate (DSS) protocol. This widely utilized procedure includes theadministration of a DNA damaging agent, AOM, followed by cy-cling doses of DSS to promote colitis (14) (Fig. 1A). This treatmentgenerated visible tumors in the distal colon after 100 d. WT andMK2 KO mice were subjected to this treatment protocol, colonswere harvested, longitudinally opened, and the number of tumorswere counted and tumor areas measured under the dissecting scope.In contrast to Ray et al. (12) who reported a complete absenceof tumors in MK2-deficient mice after AOM/DSS treatment,we observed tumors in both theWT andMK2 whole-body knockoutanimals (Fig. 1B). However, MK2 depletion led to significantlysmaller tumors and reduced total tumor burden in the MK2 KOmice compared with MK2-proficient mice (Fig. 1C), consistent withthe proinflammatory role of MK2. There was also a trend towardfewer tumors in MK2 KO mice, although this failed to reachstatistical significance. Tumors were processed for histology andblindly evaluated by an expert pathologist. Neoplasms were iden-tified as low-grade adenomas in both groups (Fig. 1D).

Myeloid-Specific Depletion of MK2 Phenocopies Constitutional MK2Loss in a Murine Model of Colitis-Associated Cancer. MK2 is ubiq-uitously expressed, and its activity has been demonstrated to becrucial for both epithelial and immune cell responses to multipletypes of cellular stress, such as DNA damage, osmotic stress, heat,and inflammation. To gain additional insight into the mechanismthrough which MK2 promotes colorectal tumorigenesis, we nextexamined the contribution of MK2 signaling in specific cellularlineages by genetically engineering conventionally floxed MK2conditional KO mice (Fig. S2 A–C). We generated tissue-specificknockouts by crossing this line to tissue-specific Cre driver mousestrains. We used Villin-Cre and LysM-Cre mice to specifically depleteMK2 in the intestinal epithelia (IEC-KO) and myeloid lineages(LysM-KO), respectively.

MK2 loss in myeloid cells was confirmed by WB on magnet-ically isolated CD11b+ cells from the bone marrow of LysM-KOmice (Fig. S2D). CD45+CD11b+ cells from blood, bone marrow,and spleen were additionally FACS sorted and processed forRNA isolation to measure MK2 expression by RT-PCR. Asexpected, we observed 80% and 90% loss of expression in bloodand bone marrow-derived cells, respectively, but only a 60%inhibition in myeloid cells isolated from spleen (Fig. S2E). Thispattern of MK2 inhibition is consistent with previous work onLysM-Cre mice, where depletion efficiency of 83–98% was ob-served in mature macrophages and near 100% in granulocytes(15). Partial depletion (16%) was detected in CD11c+ splenic den-dritic cells, which are closely related to the monocyte/macrophagelineage (15). MK2 expression in intestinal mucosa was also mea-sured by immunohistochemistry (IHC) on LysM-KO mice, wherethere was a clear down-regulation ofMK2-positive infiltrated immunecells, with MK2 expression retained in intestinal epithelia andsome subsets of lymphocytes (Fig. S2G).MK2 loss in intestinal epithelial cells (IEC) was confirmed by

Western blotting of isolated intestinal crypts from both smallintestine and colon (Fig. S2F), as previously reported in theVillin-Cre mouse model (16). MK2-specific loss in gut epitheliumwas further confirmed by IHC on colon swiss rolls (Fig. S2G).Remarkably, the colonic tumor phenotypes of these tissue-

specific MK2 KOs differed substantially. MK2 loss in intestinalepithelia resulted in enhanced tumor incidence compared with con-trols, suggesting that MK2 activity in the epithelial compartmentopposes tumor formation (Fig. 2 A and B). We have previouslydemonstrated that the MK2 pathway plays an important role inmaintenance of the G1/S and G2/M checkpoints after DNAdamage, extending the time available for DNA repair (17, 18).When IEC MK2 KO animals were treated with AOM, and theircolons were examined 24 h later, there was a notable increase in

A MK2 WT MK2 KOB

**Tumor size

AREA

(mm

2) MK2 WTMK2 KO

0

10

20

30

40

50

0

5

10

15

20Tumor number

Tum

or n

umbe

r nsTumor burden

0

50

100

150

200 *C

MK2 WT MK2 KOD

DSS DSSDSSDSSDSS

0 5 26 29 47 52 68 73 89 94 100

AOM, single dose

Tumorevaluation

AREA

(mm

2)

Fig. 1. MK2 promotes tumor growth in a murine model of colitis-associatedcancer. (A) Summary of treatment protocol. Chronic inflammation was in-duced by 2.5% DSS administration in the drinking water for 5 d, every 21 dfor a total of five cycles. AOM (10 mg/kg) was i.p. administered at the start ofthe treatment to generate colon tumors. (B) Representative pictures fromdistal colons from WT (Left) and MK2 KO (Right) mice at the time of tumorevaluation. Note the larger tumor size in MK2 WT mice. (C) Tumors werecounted under the dissecting scope in a genotype-blinded manner, and areaswere quantified using ImageJ. *P < 0.05; **P < 0.01; ns, not significant, Stu-dent’s t test. (D) Representative tumor histology following hematoxylin/eosinstaining from WT and MK2 KO animals. (Scale bar, 300 μm.) Mouse imagescourtesy of Clker/Barretto.

Suarez-Lopez et al. PNAS | vol. 115 | no. 18 | E4237

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

the number of cells within the lower two-thirds of the intestinalcrypts (where intestinal stem cells and transient amplifying cellpopulations are located) that stained positively for γH2AX, amarker of persistent DNA damage and double-strand breaks,compared with floxed controls (Fig. S7A). The number of apoptoticcells in these crypts, however, was similar between the KO andfloxed controls (Fig. S7B), suggesting that increased amounts ofDNA damage after AOM treatment in the amplifying cell pop-ulation in the intestinal crypts of the IEC KO may contribute to theincreased number of tumors that were observed in these animals.In contrast to the intestinal epithelial cell-specific KO mice,

myeloid-specific MK2 KO (LysM-Cre driven) mice developedsignificantly smaller tumors than littermate controls, recapitulatingthe global MK2 KO phenotype (Fig. 2 C and D). These findingssuggest that MK2 activity in the myeloid cell lineages plays a dominantrole in promoting tumor progression, and this myeloid-specificrole was subsequently explored in more detail.

MK2 Deficiency Impairs M2 Polarization in Tumor-Associated Macrophages.To better explore the cellular basis for the tumor-promoting role ofMK2 in the myeloid cells, we investigated the recruitment of specificmyeloid cell subsets to tumor stroma in MK2 KO mice. IHC wasused to measure myeloid cell infiltration into tumors by staining formyeloid cell type-specific markers: pan-myeloid (CD11b), neutro-phils (MPO), neutrophils/monocytes (Gr1), and macrophages (F4/80). No significant differences in broad categories of myeloid cellrecruitment were observed between WT and MK2 global KO mice(Fig. S3). Furthermore, no significant differences in myeloid cellrecruitment to the tumor stroma were observed in the myeloid-specific MK2 KO and intestinal epithelial-specific MK2 KO tumors(Fig. S4). However, we noted striking differences when we measuredspecific markers of classical (M1) or alternative (M2) activation ofmacrophages. AOM/DSS tumors recruited high numbers of argi-nase-1–expressing cells, which is associated with M2-like macro-

phages. Arginase-1–positive cells, however, were almost completelyabsent in MK2 KO mice tumors (Fig. 3A). Conversely, iNOS-positive cells (typical of M1 macrophages) were significantly higherin MK2 KO tumors compared with WT (Fig. 3B).These observations suggest that MK2 is potentially involved in

macrophage polarization. To directly explore whether MK2 isrequired for M1/M2 polarization in a cell-autonomous manner,we generated bone marrow-derived macrophage cultures fromboth MK2 WT and KO animals. Bone marrow-derived cellsdifferentiated into macrophages with comparable efficiency inboth the MK2 WT and KO, with ∼70% CD11b+F4/80+ cells(Fig. S4). This finding agrees well with the observation thatsimilar numbers of F4/80+ cells were recruited to tumors in MK2WT and KO mice (Fig. S3D).Macrophage polarization to the M2 phenotype was induced by

IL-4 stimulation and assessed by flow cytometry. Expression ofCD206, Arginase-1, and Resistin-like alpha (Retnla), which arehallmarks of murine M2 macrophages (19, 20), were significantlylower in MK2 KO macrophages compared with WT (Fig. 3 Cand D), suggesting a defect in M2 polarization in the absence ofMK2. We also measured IL-10 expression, a hallmark M2 cyto-kine, and observed a trend toward lower levels of IL-10 in MK2KO macrophages.We next used LPS/IFN-γ stimulation to drive M1 polarization.

This was assessed by CD86 surface expression and nitric oxide(NO) release, which are hallmarks of the M1 phenotype (21)(Fig. 3 E and F). MK2 depletion did not affect M1 polarizationefficiency when assessed by CD86 and NO release. However, themRNA levels of the canonical M1 effectors iNOS, TNF-α, andIL1-β were significantly lower in the MK2 KO compared withWT macrophages (Fig. 3G). These observations suggest thatMK2 is not required for M1 polarization but regulates M1 cy-tokine expression, which is in good agreement with previousreports showing that MK2 promotes the synthesis of TNF-αdownstream of TLR4 activation (22) and regulates the expres-sion of IL-1β (23).Curiously, we also observed IL-10 expression after M1 stimuli

in both in WT and MK2 KO macrophages, even more than afterM2 stimuli. This could be due to prolonged exposure to LPS, asreported by Chang et al. (24), although in both M1 and M2,polarized cells we found a trend toward reduced IL-10 expres-sion with loss of MK2 (Fig. 3D).Interestingly, myeloid-specific depletion of MK2 in vivo re-

capitulated these in vitro observations from macrophage cul-tures. Both the number of Arginase-1 and iNOS-positive cellswere reduced in colonic tumors from LysM-KO mice comparedwith WT littermate controls (Fig. 3 H and I). Thus, combiningthese in vivo and in vitro findings, we conclude that MK2 broadlyregulates macrophage biology. MK2 is required for efficientM2 polarization in a cell autonomous manner, but is also impor-tant for optimal M1 macrophage function, since MK2 regulates theextent of production of several M1 cytokines.

Myeloid MK2 Depletion Halts Tumor Angiogenesis.We observed thatglobal or myeloid-specific loss of MK2 resulted in a reduction intumor size and a defect in macrophage M2 polarization. We nextexplored if these two phenomena were functionally connected.M2 macrophages are well-known tumor promoting agents (25).Among their tumor-promoting functions, they are major con-tributors to tumor angiogenesis because they secrete factors thathelp tumors to neovascularize and promote tissue remodelingthat facilitates cell invasion and metastasis (9). We thereforeexamined the role of MK2 in tumor vascularization in vivo byimmunostaining for the endothelial marker CD31. Interestingly,CD31-positive staining was significantly reduced in both whole-body and myeloid-specific KO mice tumors compared with theircorresponding controls, suggesting a myeloid-driven role for MK2in promoting tumor angiogenesis (Fig. 4A).

FL/FL IEC-KO LysM-KOFL/FLC

0

5

10

15

20

AREA

(mm

2)

02468

10

0

5

10

15 *ns

FL/FLIEC-KO

FL/FLLysM-KO

ns

A

DB Tumor size Tumor numberTumor size Tumor number

0

5

10

15

Tum

ornu

mbe

r *

0

50

100

150ns

Tumor burden Tumor burden

01020304050

ns

AREA

(mm

2)

T um

ornu

mbe

r

A REA

(mm

2)

AREA

(mm

2)

Fig. 2. Myeloid-specific depletion of MK2 leads to smaller tumors afterchronic inflammation. (A) Representative pictures of colon tumors inducedin epithelia-specific MK2 KO mice. Note the increase in tumor number in theepithelial-specific MK2 KO. (B) Colons were harvested at the end of experi-ment, and tumors were counted and measured under a dissecting scopeusing ImageJ in a genotype-blinded fashion. (C) Representative pictures ofcolon tumors induced in myeloid-specific MK2 KO mice. Tumors are partic-ularly evident at the ends of the colon and are smaller in the myeloid-specificKO. (D) Colons were harvested at the end of experiment and tumors werecounted and measured as in B. *P < 0.05; ns, not significant, Student’s t test.Mouse images courtesy of Clker/Barretto.

E4238 | www.pnas.org/cgi/doi/10.1073/pnas.1722020115 Suarez-Lopez et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

To further demonstrate a direct link between MK2-dependentmyeloid secreted factors and angiogenesis, we established an invitro protocol to evaluate endothelial responses to MK2-dependentmacrophage secreted factors. In these experiments, the humanmonocytic cell line U-937 was differentiated into macrophages usingphorbol 12-myristate 13-acetate (PMA). Once cells were differenti-ated into M0 macrophages, M1 and M2 polarization was induced bystimulation with specific cytokine combinations, in the presence orabsence of the MK2 inhibitor PF 3644022. Similar to what we ob-served in murine MK2 KO macrophages, M1 polarization efficiencyof U-937 cells in response to LPS and IFN-γ was not significantlyaffected by the presence of the MK2 inhibitor, as assessed by the up-

regulation of human M1 hallmark surface markers HLA-DR andCD86 (26) (Fig. S5A). The levels of M1 cytokine production,however, were reduced by the PF3644022 drug, exactly as ob-served with the murine MK2 KO macrophages (Fig. S5B). WhenM2 polarization was induced by IL4 and IL-13, M2 polarizationefficiency was markedly reduced in the presence of MK2 inhibitor,as judging by both surface expression of the M2 marker CD163(Fig. S5C) as well as by the pattern of M2 gene expression forTGM-2, IL-10, IL1RN, and PPARG, all of which are character-istic of M2 polarization in human macrophages (Fig. S5D) (21).In a separate series of experiments, U937-derived macro-

phages were stimulated with the cytokine polarization mixtures

A

C

E

H I

D

F

G

B

Fig. 3. MK2 deficiency impairs M2-like polarization of TAMs. (A) Representative pictures of Arginase-1 immunodetection in WT (Left) and whole-animal MK2KO (Right) colon tumors. Stained slides were scanned and quantified in an automated manner using the positive nuclei algorithm in ImageScope andnormalized by area analyzed in square millimeters. (B) Representative pictures of Nos2 immunodetection in WT (Left) and whole-animal KO (Right) colontumors. Stained slides were scanned and quantified as in A. (C) Bone marrow-derived macrophages from WT and KO mice were polarized to M1 andM2 phenotypes. Twenty-four hours after polarization, surface marker expression was quantified by FACS. Percentages of CD206-positive cells were calculatedout of total macrophages (CD11b+F4/80+ cells). (D) Expression of hallmark M2 genes was measured by Q-PCR, actin was used for normalization, and foldchange versus the corresponding unstimulated control was calculated. (E) Percentages of CD86-positive cells were calculated out of total macrophages(CD11b+F4/80+ cells) by FACS. (F) Nitrite concentrations in the supernatants of MK2 WT and KO macrophages polarized to M1 or M2 as determined by theGriess reaction. (G) Expression of hallmark M1 genes was measured by Q-PCR. Actin was used for normalization and fold change versus the correspondingunstimulated control was calculated. (H) Representative pictures of Arginase 1 immunodetection in myeloid-specific KO tumors (Right) and littermate controls(Left). Stained slides were analyzed as in A. (I) Representative pictures of Nos2 immunodetection in myeloid-specific KO tumors (Right) and littermate controls(Left). Stained slides were analyzed as in B. (Scale bars: 200 μm.) *P < 0.05; **P < 0.01, ****P < 0.0001; Student’s t test (A, B, H, and I), two way-ANOVA. Data inC–G is from at least three independent experiments. Mouse images courtesy of Clker/Barretto.

Suarez-Lopez et al. PNAS | vol. 115 | no. 18 | E4239

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

A

B C

D E

Fig. 4. MK2 loss in the myeloid compartment halts tumor angiogenesis. (A) Representative pictures of blood vessels detection by CD31 staining inMK2 whole-animal KO (Upper) and myeloid-specific KO (Lower) and their corresponding controls. Inset in Upper Left shows typical blood vessel morphology whenviewed under higher magnification. Stained slides were scanned and automated quantified using the positive pixel algorithm in ImageScope. Strong positive pixelscounts were normalized by area of tumor analyzed. *P < 0.05, Student’s t test. (B) ECIS cell impedance (resistance) of confluent HUVEC cultures over time. Woundswere induced by electroablation as indicated, and confluency recovery was monitored over time in the presence or absence of macrophage-derived conditionedmedia. Data represent three independent experiments. (C) Rate of recovery for the first 25% of confluence was calculated per experimental group. *P < 0.05; **P <0.01; ****P < 0.0001. Student’s t test. (D) HUVEC tubule formation morphogenesis assays were performed in the presence or absence of various types of macrophage-derived conditioned media. (E) Number of nodes, junctions, segments, and mesh area were measured in an automated manner using the angiogenesis analyzer plug-in in ImageJ. Data were normalized to values from untreated control (no conditioned media), which represents 100%. *P < 0.05; **P < 0.01; ***P < 0.001, one-wayANOVA. (Scale bars: A, 200 μm; D, 400 μm.) Mouse images courtesy of Clker/Barretto.

E4240 | www.pnas.org/cgi/doi/10.1073/pnas.1722020115 Suarez-Lopez et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

described above, with or without the MK2 inhibitor, for 24 h,then washed and recultured in cytokine/inhibitor-free media foran additional 24 h to generate macrophage-conditioned media.This media was then tested for its effects on endothelial cellproliferation, migration, and blood vessel formation. Endothelialcell proliferation and migration was assessed in HUVEC woundhealing assays using an electric cell-substrate impedance system(ECIS) device, following standard published procedures (27–29).This system enables accurate dynamic tracking of transepithelialelectrical resistance as a readout of cellular confluency andpermeability of tight junctions. In brief, HUVEC cells weregrown to confluency, and a radial clear zone of defined size(354 μm diameter) was created by electroablation. The closure ofthe wound was followed over time (Fig. 4B). Media from U-937–derived M2 macrophages significantly enhanced the rate ofclosure relative to media from the parental nonpolarized cells(purple versus black line), in agreement with a proangiogenicrole for M2 macrophages. In contrast, media from M1-polarizedU-937 macrophages markedly impaired the ability of HUVECcells to proliferate, migrate, and heal the wound (orange versusblack line). The presence of the MK2 inhibitor during the first24 h of polarization resulted in a small but reproducible statis-

tically significant decreased rate of recovery, suggesting thatMK2 signaling regulates secreted factors that promote endo-thelial cell proliferation. Inhibition of MK2 signaling in M1 cellsresulted in an even further impairment of their conditionedmedia to drive endothelial wound healing relative to controls(Fig. 4B, pink and yellow lines). The rate of wound healingduring the first 25% of closure was quantified. As shown in Fig.4C, the wound closure velocity was significantly higher when HUVECcells were in the presence of M2-derived media, whereas M1media slowed down recovery. MK2 inhibition decreased therate of monolayer recovery in both cases.We also studied the ability of these conditioned media to di-

rectly promote blood vessel formation using tubule formationassays, as described by Kubota et al. (30). In these experiments,HUVEC cells were seeded onto matrigel, which induces theirdifferentiation into capillary-like structures (Fig. 4D) (30). Inagreement with a proangiogenic role of M2 macrophages (31,32), M2-derived conditioned media significantly increased thenumber of junctions, nodes, segments, and total mesh area in thetubule networks that were generated relative to media fromnonpolarized controls. Interestingly, M1-conditioned media alsopromoted the morphogenic capacity of HUVEC cells, albeit to alesser extent (Fig. 4E), in contrast to the inhibitory effect seenpreviously on cell proliferation (Fig. 4 B and C). Interestingly,conditioned media from MK2-inhibited M2 macrophages, butnot from MK2-inhibited M1 macrophages, showed significantlyless morphogenic potential than their corresponding uninhibitedcontrols, as evidenced by lower numbers of junctions, nodes,segments, and total mesh area following MK2 inhibition. Takentogether, these results demonstrate an important role of MK2 inmacrophage M2 polarization, and its subsequent effects on endo-thelial cell proliferation, migration, and blood vessel morphogenesis.

DiscussionIn this study, we have identified MK2 as a regulator of M1/M2 polarization in tumor-associated macrophages. MK2 regulatesseveral aspects of macrophage biology, including both regulatingM2 polarization and M1 cytokine production. In agreement with aproangiogenic role of M2-like macrophages, tumors developed inMK2 KO mice showed markedly less vascularization and reducedtumor size in a colitis-associated cancer model.Our results are in agreement with previous studies that link

either p38 or MK2 to inflammation-driven tumorigenesis (12, 33,34). Here, however, we have established the underlying molec-ular and cellular mechanisms by selective ablation of MK2 ac-tivity in myeloid versus epithelial cellular linages and identifiedthe myeloid compartment as most relevant for tumor promoteractivities of MK2 in the colon.The AOM/DSS model is based on the DNA alkylating func-

tion of the mutagenic molecule AOM, and the tumor promotingeffects of chronic inflammation as a result of cyclic colonicdamage induced by DSS. Because of its high reproducibility andpotency, as well as the simple and affordable mode of applica-tion, the AOM/DSS has become a well-accepted model forstudying colon carcinogenesis and a powerful platform for che-mopreventive intervention studies (35). Both MK2 whole-bodyand myeloid-specific KO mice developed significantly smallertumors, which indicates an epithelial-independent promotingrole of MK2 in colonic tumorigenesis. Conversely, specific MK2loss in epithelial cells resulted in an increased number of ade-nomas. We have previously shown that MK2 is involved in DNAdamage repair and maintenance of cell cycle checkpoints incancer cells (13, 36, 37). It is plausible then, that after AOM-induced DNA damage, MK2 is required for DNA damage repair,cell cycle arrest, and/or apoptosis in intestinal cells, influencing thenumber of mutated cells that will progress into adenomas. Inagreement with this, the frequency of DNA double-strand breaksafter AOM insult was significantly higher, with no corresponding

Chronic Inflammation

M1 Macrophages

Normal ephitelialcells

Precancerous lession

Angiogenic switch

M2 Macrophages

Defective Angiogenicswitch

Cancer cellsproliferation

Tumor angiogenesis Poor Tumor angiogenesis

TUMOR PROGRESSION TUMOR PROGRESSION

MK2 proficient

MK2 defective

M1 Macrophages

Fig. 5. A proposed model for MK2-dependent macrophage function ininflammation-driven colonic tumorigenesis. Early inflammation is dominated byclassically activated M1 macrophages. Persistent chronic inflammation allows riseof oncogenic mutations, which can be eliminated byM1macrophages. However,when the lesions progress, macrophages in the tumor microenvironment acquirean M2-like phenotype. M2-like macrophages promote tumor neovascularizationand angiogenesis. This macrophage shift from the M1 to M2 state is also knownas “the angiogenesis switch.”MK2, as a key regulator of macrophage biology, isrequired for that angiogenesis switch in TAMs; therefore, MK2 inhibition leadsto poor tumor vascularization and delayed tumor progression. Mouse imagescourtesy of Clker/Barretto. Tumor angiogenesis, macrophages, and carcinomaimages adapted from Servier Medical Art.

Suarez-Lopez et al. PNAS | vol. 115 | no. 18 | E4241

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

increase of apoptotic cells, in IEC-KO mice than in controls(Fig. S7). This imbalance of damaged cells and apoptosis couldplausibly lead to the accumulation of mutated cells that havethe potential of becoming an adenoma after chronic inflamma-tion. We focused here on the mechanism by which myeloidMK2 inhibition delays tumor progression as a potential thera-peutic strategy for cancer prevention in IBD patients.In vivo MK2 loss led to decreased infiltration of arginase-

positive cells in tumors, which are normally identified as M2macrophages. However, myeloid suppressor cells also expressarginase (38). We therefore cannot exclude the possibility that atleast some of the Arginase-1–positive cells are MDSCs. Our invitro experiments however, in both murine and human systems,demonstrated a clear role of MK2 in broad macrophage biology,including polarization into M2 macrophages and macrophageproduction of proangiogenic factors, consistent with the pro-posed mechanism for the observed mouse phenotypes. A func-tional role for MK2 in other myeloid cell types, includingneutrophils and dendritic cells (39–41), could also contribute tothe observed phenotype.CRC is associated with a marked increase in the number of

macrophages, particularly M2, which has previously been shownto correlate with metastasis and poor prognosis of the disease(42). It has been proposed that a gradual switching of TAMpolarization takes place, from M1 in early tumorigenesis thatmay contribute to elimination of cancer cells, to M2-like as thetumor progresses and becomes a “wound that does not heal”(43). TAMs influence tumor progression at different levels.TAMs influence the intrinsic properties of tumor cells, as well asthose of the tumor microenvironment, for example, by producinggrowth factors, such as EGF, which stimulates cancer cell pro-liferation. TAMs also produce proteolytic enzymes that digestthe extracellular matrix to promote tumor cell disseminationand metastasis. Furthermore, TAMs promote angiogenesis andlymphangiogenesis, as well as tissue remodeling by, for example,stimulating the deposition of fibrous tissue. All of these pro-cesses can support tumor development and progression (32).AOM/DSS tumors also accumulate M2 macrophages as theyprogress (44), and clodronate-mediated depletion of macro-phages significantly decreases tumor burden in this model (45).Consistent with the reduced number of M2 macrophages, we

observed a decrease in CD31 staining in tumors, which labelsboth blood and lymph vessels, both in systemic and myeloid-specific MK2 KO. Therefore, we hypothesized that MK2 de-pletion leads to inefficient macrophage-driven angiogenesis anda consequent reduction in tumor size. We tested our hypothesison isolated in vitro systems, where we could interrogate theMK2 macrophage-specific contribution to endothelial cells pro-liferation, migration and morphogenesis. We demonstrated thatmacrophages secrete proangiogenic factors that are MK2 de-pendent. Indeed, MK2 regulates the expression of IL-8, IL-6,VEGF, COX-2, IL-1α, and IL-1β (23, 46–48), all cytokines withangiogenic capacities. The exact identity and specific mechanismof regulation of proangiogenic factors by MK2 in our system arenot yet known.We also looked at the adaptive immune components of the

tumor microenvironment in both the myeloid-specific and in-testinal epithelial-specific MK2 KO animals. Interestingly, wesaw opposite effects of MK2 inhibition in these two KO geno-types. In the LysM-KO versus floxed controls, there were nosignificant differences in T-cell recruitment into the tumorstroma, but there was decreased B-cell recruitment (Fig. S4 D–H).In contrast, the IEC-KO tumors showed an increased number ofCD3-positive T cells and reduced numbers of CD8 and FoxP3-positive T cells compared with the floxed littermate controls,while B cell numbers were not affected (Fig. S4 D–H). Theseobservations further indicate that MK2 has opposing roles in theepithelial and myeloid compartments, particularly with respect

to modulation of the adaptive arm of the immune system, inaddition to its role in myeloid-driven tumor neoangiogenesis.Blockade of the p38/MK2 axis has been proposed as a

mechanism to treat diverse chronic inflammatory disorders, in-cluding IBD (49). Given the unacceptable side effects of systemicp38 inhibition, MK2 inhibitors are currently under active de-velopment, encouraged by the full viability and increased re-sistance to endotoxic shock and collagen-induced arthritis seenin MK2 KO mice (48). Our whole-body and tissue-specific KOresults are in agreement with recent studies testing the efficacy ofMK2 inhibition on regression of colon tumors in vivo (50). Ourstudy, together with others, encourages the potential use ofMK2 inhibition not only to prevent chronic inflammation butalso for cancer prevention and treatment in high cancer riskIBD patients.

MethodsMice. To generate MK2 null mice, the MK2CV/CV mice previously generated inour laboratory (which are fully MK2 proficient in the absence of Crerecombinase activity; ref. 13), were crossed with β-actin promoter-driven Cremice (51), in which Cre is ubiquitously expressed in all tissues. Mice with agerm-line inversion of exon 2 were obtained, and the Cre driver gene sub-sequently bred out of the colony, resulting in a mouse line with a stableexon 2 inversion, and a complete loss of MK2 expression in all tissue types, asverified by PCR, WB, and IHC. Examples of selected tissues from these whole-body MK2 KO animals are shown in Fig. S1 C and D. Animals were genotypedusing the PCR strategy depicted in Fig. S1B (P1: 5′-GAGCTCTCCACCATCGA-GAC-3′; P2: 5′-GCAGACAGCCCACTATGGAT-3′; P3: 5′-GCTGCCTCTCCTTCTGT-AGAC-3′). Both mice strains were backcrossed 10 generations to C57/BL6N(Taconic) strain.

To generate mice carrying conditional alleles of MK2, we constructed atargeting vector based on the pBluescript variant pKS-DTA (generouslyprovided by A. Ventura, David H. Koch Institute for Integrative Cancer Re-search, Massachusetts Institute of Technology, Cambridge, MA). The Bacclone RP24-125B12 (Bacpac Resource Center) spanning the entire genomiclocus encoding MK2 served as the template for PCR-directed cloning. Thefinal targeting construct consists of 2.7 kb of the 5′ flanking sequence beforeexon 2, a LoxP site, exon 2, followed by a second LoxP site, a FRT-Pgk::Neo-FRT cassette to select for integration of the targeting construct into ES-cells,and exons 3–10. The targeting construct was linearized using AscI, whichcuts at a unique site in the pBluescript parent backbone of the targetingconstruct, and electroporated in 129/B6 hybrid ES cells. Transfected ES cellclones were selected by growth in media containing neomycin for 1 wk.Proper integration was monitored using Southern blot analysis from SpeIgenomic DNA digests to monitor the 5′ integration event using a 300-bpfragment from the genomic sequence in intron 1–2 that lies outside thetargeting construct as probe A (Fig. S2 A and B). The probe was generated byPCR amplification followed by radiolabeling with 32P-α-dCTP using randompriming. In the WT genomic MK2 sequence, the SpeI sites in intron 1 andintron 10 gives rise to a 16.3-kb band, while correct 5′ integration of thetargeted allele gives rise to a 11.0-kb band due to the presence of a SpeI sitewithin the Pgk::Neo cassette. The 3′ integration event was monitored bySouthern blotting using a 300-bp probe B corresponding to the 3′ flankingregion outside the targeting construct. The targeted allele gives rise to a7.2-kb band (Fig. S2 A and B).

One ES cell clone containing the integrated MK2 floxed (FL) allele wasinjected into host albino FEBN blastocysts and implanted into pseudopreg-nant mice. Chimeric mice were crossed to C57-FLPe mice to eliminate the FRT-Pgk::Neo-FRT cassette. The offspring was genotyped by PCR using primersflanking exon 2 and spanning the second LoxP site (P1 and P2) (Fig. S2C).

MK2 FL/FL mice were then crossed with Villin-Cre (16) and Lyz2 tm1(cre)Ifo

(15) strains to generate the specific depletion of MK2 on intestinal epithelia(IEC-KO) or myeloid linages (LysM-KO), respectively. IEC-KO mice and controllittermates were maintained in a mixed C57/BL/6N × 129SvJ background,whereas LysM-KO and control littermates were backcrossed five generationsto C57/BL/6N. Animals containing tissue specific knockouts were alwayscompared with their matched littermate controls.

Murine Colitis-Associated Cancer. Eight- to 12-wk-old male mice were ad-ministered 2.5% DSS (MP Biochemicals) in the drinking water for 5 d every21 d, for a total of five cycles to induce chronic inflammation. Colon tumorswere generated by i.p. administration of 10 mg/Kg AOM (Sigma) beforechronic DSS administration (14). Colons were harvested and tumors were

E4242 | www.pnas.org/cgi/doi/10.1073/pnas.1722020115 Suarez-Lopez et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

examined 100 d after AOM administration under a dissecting scope. Tumorimages were taken and tumor size was measured using ImageJ (52).

All mouse studies described here were approved by the MassachusettsInstitute of Technology Institutional Committee for Animal Care and con-ducted in compliance with the Animal Welfare Act regulations and otherfederal statutes relating to animals and experiments involving animals andadhere to the principles set forth in the Guide for the Care and Use ofLaboratory Animals, National Research Council, 1996 (Institutional AnimalWelfare Assurance No. A-3125–01).

Intestinal Crypts Isolation. Eight- to 10-wk-old male mice were CO2 eutha-nized, and both small intestines and colons were collected and washed incold PBS before incubation in 4 mM EDTA, 1 mM DTT PBS solution withrocking for 45 min at 4 °C. After incubation, tissue was transferred to freshPBS and shaken manually for 2 min to release intestinal crypts. The musclelayer was removed and the released crypts in PBS were filtered through a70-μm filter to remove villi. Intestinal crypts were lysed in RIPPA buffercontaining a commercial protease inhibitor mixture (Roche) for proteinextraction studies.

Immunohistochemistry. Colons were flushed with PBS and swiss-rolled beforefixation in 10% neutral buffered formalin and paraffin embedding. Four-micrometer sections were dewaxed and rehydrated before heat-mediatedantigen retrieval in citrate buffer (pH 6). Anti-MK2 (CST D4E11, 1:200), antiF4/80 (Thermo Scientific MF48000, 1:100), anti-MPO (Thermo Ab-1 no. RB-373-A1, 1:200), anti-GR1 (eBioscience 14-5931-82, 1:200), anti-CD11b (eBioscience14-0112 1:200), anti-Arg1 (CST 93668S, 1:100), anti-iNOS (CST 13120S, 1:400),anti-CD31 (Abcam, ab28364, 1:200), anti-CD3 (Abcam ab5690, 1:200), anti-FoxP3(eBioscience 14-0808-30, 1:200), anti-CD4 (eBioscience 14-9766-80, 1:200), anti-CD8 (14-5773-80, 1:200), anti-B220 (eBioscience 16-0452-85), anti-γH2AX (CST9718, 1:400), anti-CC3 (CST 9661, 1:400) antibodies were used for immunohis-tochemistry. Impact DAB (Vector) secondary antibodies were used, and sampleswere hematoxylin counterstained before mounting. Slides were scanned in aLeica slide scanner before analysis in Aperio ImageScope software.

CD11b+ Cells Isolation.Mice were CO2 euthanized and femurs and tibias werecollected, cleared of adherent muscle, and flushed with a 21 G syringe tocollect bone marrow in cold PBS containing 2 mM EDTA and 2% FBS. Redblood cells (RBCs) in the sample were lysed in RBC lysis buffer (eBiosciences)for 5 min at room temperature. Biotin-labeled CD11b antibody (MiltenyiBiotec) was used to isolate myeloid cells using CELLection Biotin Binder kit(Thermo Scientific) following the manufacturer’s instructions. Both CD11b−

and CD11b+ cells were then processed for Western blotting to measureMK2 expression.

Blood was collected by cardiac puncture, and leukocytes were recoveredafter RBC lysis. Spleens were collected and mechanically dissociated betweenfrosted glass slides. Splenocytes were recovered after RBC lysis and processedfor FACS sorting. CD45+CD11b+ cells were isolated in a MoFlo sorter andprocessed for RNA extraction to measure MK2 expression.

Q-PCR. RNA was extracted using TRIzol reagent (Ambion) according to themanufacturer’s instructions, and 1–2 μg of total RNA was used for reversetranscription using the SuperScript III First-Strand Synthesis Kit (Invitrogen)and oligo-dT priming as per the manufacturer’s instructions. For qPCR, cDNAwas amplified using SYBR green PCR mastermix (Applied Biosystems)according to the manufacturer’s cycling conditions for 40 cycles on a Bio-RadC1000 Thermal Cycler. Data were analyzed using the delta-delta Ct methodas described previously (53) and plotted as fold change versus control. Primersused are listed in Table S1.

Western Blotting. For Western blotting, 15 μg of total protein was resolved by10% SDS/PAGE before immunoblotting. Primaries antibodies were used asfollows: anti-MK2 (CST D4E11, 1:2,000), anti-GAPDH (Sigma G8796, 1:2,000),anti–γ-tubulin (Sigma T7816, 1:2,000). Fluorescently labeled secondary antibodieswere used for protein detection using an Odyssey fluorescence imaging system.

Murine Bone Marrow-Derived Macrophage Differentiation and Polarization.Bone marrow cells were isolated from the femurs and tibias of MK2 WTand MK2 KO mice by flushing with IMDM (10% FBS with antibiotics) or bybriefly centrifuging the bones at 15,000 × g for 15 s at 4 °C in a 1.5-mLEppendorf tube. After RBC lysis in RBC lysis buffer (eBioscience), 2 × 106 cellsper mL were seeded on 10-cm bacterial plates in IMDM supplemented with10 ng/mL murine M-CSF. An equal volume of M-CSF–supplemented mediawas added on top of the existing media 72 h after seeding. On day 7 after

seeding, cells were detached with 5 mM EDTA (in PBS without Ca and Mg)and replated on 24-well tissue culture plates at 200,000 cells per well and/oron 6-cm tissue culture plates at 4 million cells per plate in IMDM (10% FBSwith antibiotics). Cells were allowed to attach for 12 h. To induce polariza-tion into M0, M1, or M2, cells were either unstimulated (M0), stimulatedwith 100 ng/mL LPS (Sigma) and 50 ng/mL murine IFN-γ (Gibco) (M1), or with10 ng/mL murine IL-4 (Gibco) (M2) and harvested 24 h after stimulation.

Measurement of NO Production from Macrophages. NO production wasquantified by measuring the concentration of nitrite, which is spontaneouslyformed by oxidation of NO in culture supernatants. The Griess Reagent Kit fornitrite quantitation was used (Thermo Scientific). Briefly, 150 μL of culturesupernatant was mixed with 130 μL of water and 20 μL of Griess Reagent[made fresh by mixing equal volumes of 0.1% N-(1-naphthyl)ethylenedi-amine dihydrochloride and 1% sulfanilic acid]. After 1-h incubation at roomtemperature in the dark, the reaction was read at 540 nm using a Tecanplate reader spectrophotometer. Standard curve for nitrite concentrations(1–100 μM) was derived using a sodium nitrite solution.

Human Macrophages Differentiation and Polarization. The monocytic cell lineU-937 was purchased from ATCC and grown in RPMI 1640 medium 10% FBS,37 °C, 5% CO2. For macrophage differentiation, 2 × 106 cells were seeded insix-well plates and stimulated with PMA for 48 h. After macrophage dif-ferentiation, cells were stimulated with 100 ng/mL LPS (Sigma) and 50 ng/mLhuman IFN-γ (Gibco) (M1) or with 10 ng/mL human IL-4 (Gibco) and 10 ng/mLhuman IL-13 (M2) and harvested 48 h after stimulation. Where indicated,10 μM of the MK2 chemical inhibitor PF3644022 (Tocris) was added.

FACS Analysis of Macrophage Populations. Cells were detached by incubationin PBS containing 5 mM EDTA for 5–10 min. Single-cell suspensions wereincubated for 10 min at room temperature with Fc blocking (CD16/32) an-tibody (eBioscience) before staining with fluorochrome-conjugated anti-bodies against the following antigens: CD11b-APC-Cy7, F4/80-FITC, CD86-Pe(eBioscience). Cells were then fixed and permeabilized (eBioscience) beforeCD206-APC staining. Aqua Live/Dead dye (Invitrogen) was used to excludedead cells. Human U937 cells were processed similarly but staining panelconsisted of the following antibodies: CD11b-APC-Cy7, HLA DR-FITC, CD86-Pe, CD163-PeCy7. DAPI was used to exclude dead cells. Multiparameteranalysis was performed on a LSR Fortessa (BD), and bidimensional dot plotswere generated using FlowJo software.

ECIS Wound Healing Assay. ECIS plates (96W1E+; Applied Biophysics) werecleaned and stabilized by incubating with 200 μL per well of 10 mM L-cys-teine at room temperature for 15 min. Wells were washed twice with salinebefore coating with 200 μL per well warm 1% gelatin for 30 min. Wells werewashed two times with saline. HUVEC cells were purchased from ATCC andcultured in EGM-2 media supplemented with endothelia specific growthfactors (EGM-2, Bulletkit; Lonza), 37 °C, 5% CO2. HUVEC cells were seeded in300 μL per well on ECIS plates and grown to confluence. Impedance levelswere used to verify that confluence was achieved and maintained. Once cellsreached confluence, 150 μL of HUVEC media was removed from each welland replaced with 150 μL of conditioned media collected from polarized U-937 cells. The cells were allowed to acclimatize to the fresh media for 30 minbefore undergoing two wounding cycles (6,000 Hz, 20 s, 400 uA) (27–29).ECIS resistance readings were collected and plotted against time afterwounding. Linear regression was used to fit the data points, and the slopewas used to determine the time required for each group to recover to 25%of prewounding cell confluency. The rate of recovery was then calculatedfrom the time taken for the cells to cover 23.71 μm.

Tubule Formation Assay. Twenty-four thousand HUVEC cells were seeded ontop off 120 μL of matrigel (Corning) in complete cell culture media (EGM-2Bulletkit; Lonza) and 15% of conditioned or control media. Tubules wereallowed to form overnight and analyzed by phase contrast microscopy. Im-ages were automatically analyzed in ImageJ with the angiogenesis analyzerplug-in. Plates were also fixed and stained with Diff Quick stain kit (Thermo)for bright-field pictures acquisition, following manufacturer’s instructions.

Statistical Methods. Unless stated otherwise, all data were plotted and an-alyzed in GraphPad Prism software, using Student’s t test or ANOVA followedby the Bonferroni post hoc analysis. Data represent the mean ± SEM.

Suarez-Lopez et al. PNAS | vol. 115 | no. 18 | E4243

IMMUNOLO

GYAND

INFLAMMATION

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0

Illustrations. Mouse illustrations were adapted from freely distributed clipartobtained from www.clker.com. Images used in Fig. 5 were adapted fromSMART servier medical art under a creative commons license.

ACKNOWLEDGMENTS. We thank Phaedra Ghazi and Katherine Baldwinfrom the K.M.H. laboratory and Pau Creixell and other members of theM.B.Y. laboratory for helpful advice and discussion; veterinary pathologistDr. R. T. Bronson (Tufts University) for assistance with tumor histopathology

analysis; and the Koch Institute Swanson Biotechnology Center, especiallythe ES/Transgenics Facility, the Hope Babette Tang (1983) Histology Facility,and the Flow Cytometry Core Facility and the Koch Institute support (core)Grant P30-CA14051 from the National Cancer Institute. L.S.-L. is a recipientof Research Fellowship Award 346496 from Crohn’s and Colitis Foundation ofAmerica. This project was funded by NIH Grants R01-GM104047 and R35-ES028374, Starr Cancer Consortium Award I9-A9-07, the Charles and MarjorieHolloway Foundation, and the MIT Center for Precision Cancer Medicine.

1. Siegel RL, et al. (2017) Colorectal cancer statistics, 2017. CA Cancer J Clin 67:177–193.2. Söderlund S, et al. (2009) Decreasing time-trends of colorectal cancer in a large cohort

of patients with inflammatory bowel disease. Gastroenterology 136:1561–1567, quiz1818–1819.

3. Ekbom A, Helmick C, Zack M, Adami H-O (1990) Ulcerative colitis and colorectalcancer. A population-based study. N Engl J Med 323:1228–1233.

4. Viscido A, et al.; GISC: Italian Group for the Study of the Colon and Rectum (2001)Survival and causes of death in Italian patients with ulcerative colitis. A GISC na-tionwide study. Dig Liver Dis 33:686–692.

5. Rutter M, et al. (2004) Severity of inflammation is a risk factor for colorectal neoplasiain ulcerative colitis. Gastroenterology 126:451–459.

6. Grivennikov SI, Greten FR, Karin M (2010) Immunity, inflammation, and cancer. Cell140:883–899.

7. de Visser KE, Eichten A, Coussens LM (2006) Paradoxical roles of the immune systemduring cancer development. Nat Rev Cancer 6:24–37.

8. Engblom C, Pfirschke C, Pittet MJ (2016) The role of myeloid cells in cancer therapies.Nat Rev Cancer 16:447–462.

9. Solinas G, Germano G, Mantovani A, Allavena P (2009) Tumor-associated macro-phages (TAM) as major players of the cancer-related inflammation. J Leukoc Biol 86:1065–1073.

10. Neurath MF (2014) Cytokines in inflammatory bowel disease. Nat Rev Immunol 14:329–342.

11. Duraisamy S, et al. (2008) MK2: A novel molecular target for anti-inflammatorytherapy. Expert Opin Ther Targets 12:921–936.

12. Ray AL, et al. (2016) Blockade of MK2 is protective in inflammation-associated co-lorectal cancer development. Int J Cancer 138:770–775.

13. Morandell S, et al. (2013) A reversible gene-targeting strategy identifies syntheticlethal interactions between MK2 and p53 in the DNA damage response in vivo. CellRep 5:868–877.

14. Neufert C, Becker C, Neurath MF (2007) An inducible mouse model of colon carci-nogenesis for the analysis of sporadic and inflammation-driven tumor progression.Nat Protoc 2:1998–2004.

15. Clausen BE, Burkhardt C, Reith W, Renkawitz R (1999) Conditional gene targeting inmacrophages and granulocytes using LysMcre mice. Transgenic Res 8:265–277.

16. el Marjou F, et al. (2004) Tissue-specific and inducible Cre-mediated recombination inthe gut epithelium. Genesis 39:186–193.

17. Reinhardt HC, et al. (2010) DNA damage activates a spatially distinct late cytoplasmiccell-cycle checkpoint network controlled by MK2-mediated RNA stabilization. MolCell 40:34–49.

18. Cannell IG, et al. (2015) A pleiotropic RNA-binding protein controls distinct cell cyclecheckpoints to drive resistance of p53-defective tumors to chemotherapy. Cancer Cell28:623–637.

19. Stein M, Keshav S, Harris N, Gordon S (1992) Interleukin 4 potently enhances murinemacrophage mannose receptor activity: A marker of alternative immunologic mac-rophage activation. J Exp Med 176:287–292.

20. Raes G, et al. (2002) Differential expression of FIZZ1 and Ym1 in alternatively versusclassically activated macrophages. J Leukoc Biol 71:597–602.

21. Martinez FO, Gordon S (2014) The M1 and M2 paradigm of macrophage activation:Time for reassessment. F1000Prime Rep 6:13.

22. Ronkina N, et al. (2007) The mitogen-activated protein kinase (MAPK)-activatedprotein kinases MK2 and MK3 cooperate in stimulation of tumor necrosis factorbiosynthesis and stabilization of p38 MAPK. Mol Cell Biol 27:170–181.

23. Frevel MAE, et al. (2003) p38Mitogen-activated protein kinase-dependent and -independentsignaling of mRNA stability of AU-rich element-containing transcripts. Mol Cell Biol 23:425–436.

24. Chang EY, Guo B, Doyle SE, Cheng G (2007) Cutting edge: Involvement of the type IIFN production and signaling pathway in lipopolysaccharide-induced IL-10 pro-duction. J Immunol 178:6705–6709.

25. Mantovani A, Allavena P, Sica A, Balkwill F (2008) Cancer-related inflammation.Nature 454:436–444.

26. Gordon S (2003) Alternative activation of macrophages. Nat Rev Immunol 3:23–35.27. van Gils JM, et al. (2009) HPA-1a alloantibodies reduce endothelial cell spreading and

monolayer integrity. Mol Immunol 46:406–415.28. Konya V, et al. (2013) Endothelial E-type prostanoid 4 receptors promote barrier

function and inhibit neutrophil trafficking. J Allergy Clin Immunol 131:532–40.e1-2.

29. Spiering D, et al. (2009) MEK5/ERK5 signaling modulates endothelial cell migrationand focal contact turnover. J Biol Chem 284:24972–24980.

30. Kubota Y, Kleinman HK, Martin GR, Lawley TJ (1988) Role of laminin and basementmembrane in the morphological differentiation of human endothelial cells intocapillary-like structures. J Cell Biol 107:1589–1598.

31. Zajac E, et al. (2013) Angiogenic capacity of M1- and M2-polarized macrophages isdetermined by the levels of TIMP-1 complexed with their secreted proMMP-9. Blood122:4054–4067.

32. Mantovani A, Marchesi F, Malesci A, Laghi L, Allavena P (2017) Tumour-associatedmacrophages as treatment targets in oncology. Nat Rev Clin Oncol 14:399–416.

33. Johansen C, et al. (2009) MK2 regulates the early stages of skin tumor promotion.Carcinogenesis 30:2100–2108.

34. Gupta J, et al. (2014) Dual function of p38 a MAPK in colon cancer : Suppression ofcolitis-associated tumor initiation but requirement for cancer cell survival. Cancer Cell25:484–500.

35. De Robertis M, et al. (2011) The AOM/DSS murine model for the study of colon car-cinogenesis: From pathways to diagnosis and therapy studies. J Carcinog 10:9.

36. Manke IA, et al. (2005) MAPKAP kinase-2 is a cell cycle checkpoint kinase that reg-ulates the G2/M transition and S phase progression in response to UV irradiation. MolCell 17:37–48.

37. Reinhardt HC, Aslanian AS, Lees JA, Yaffe MB (2007) p53-deficient cells rely on ATM-and ATR-mediated checkpoint signaling through the p38MAPK/MK2 pathway forsurvival after DNA damage. Cancer Cell 11:175–189.

38. Gallina G, et al. (2006) Tumors induce a subset of inflammatory monocytes with im-munosuppressive activity on CD8+ T cells. J Clin Invest 116:2777–2790.

39. Soukup K, et al. (2015) The MAPK-activated kinase MK2 attenuates dendritic cell-mediated Th1 differentiation and autoimmune encephalomyelitis. J Immunol 195:541–552.

40. Hannigan MO, et al. (2001) Abnormal migration phenotype of mitogen-activatedprotein kinase-activated protein kinase 2-/- neutrophils in Zigmond chambers con-taining formyl-methionyl-leucyl-phenylalanine gradients. J Immunol 167:3953–3961.

41. Coxon PY, et al. (2003) MAPK-activated protein kinase-2 participates in p38 MAPK-dependent and ERK-dependent functions in human neutrophils. Cell Signal 15:993–1001.

42. Forssell J, et al. (2007) High macrophage infiltration along the tumor front correlateswith improved survival in colon cancer. Clin Cancer Res 13:1472–1479.

43. Mantovani A, Sica A (2010) Macrophages, innate immunity and cancer: Balance,tolerance, and diversity. Curr Opin Immunol 22:231–237.

44. Wang WEI, et al. (2015) Dynamic changes and functions of macrophages and M1/M2 subpopulations during ulcerative colitis-associated carcinogenesis in an AOM/DSSmouse model. Mol Med Rep 11:2397–2406.

45. Bader JE, et al. (2018) Macrophage depletion using clodronate liposomes decreasestumorigenesis and alters gut microbiota in the AOM/DSS mouse model of coloncancer. Am J Physiol Gastrointest Liver Physiol 314:G22–G31.

46. Dean JL, Brook M, Clark AR, Saklatvala J (1999) p38 mitogen-activated protein kinaseregulates cyclooxygenase-2 mRNA stability and transcription in lipopolysaccharide-treated human monocytes. J Biol Chem 274:264–269.

47. Neininger A, et al. (2002) MK2 targets AU-rich elements and regulates biosynthesisof tumor necrosis factor and interleukin-6 independently at different post-transcriptionallevels. J Biol Chem 277:3065–3068.

48. Gaestel M, Kotlyarov A, Kracht M (2009) Targeting innate immunity protein kinasesignalling in inflammation. Nat Rev Drug Discov 8:480–499.

49. Gaestel M (2006) MAPKAP kinases–MKs–Two’s company, three’s a crowd. Nat RevMol Cell Biol 7:120–130.

50. Ray AL, Berggren KL, Restrepo Cruz S, Gan GN, Beswick EJ (2017) Inhibition ofMK2 suppresses IL-1β, IL-6, and TNF-α-dependent colorectal cancer growth. Int JCancer 142:1702–1711.

51. Lewandoski M, Meyers EN, Martin GR (1997) Analysis of Fgf8 gene function in ver-tebrate development. Cold Spring Harb Symp Quant Biol 62:159–168.

52. Schneider CA, Rasband WS, Eliceiri KW (2012) NIH image to ImageJ: 25 years of imageanalysis. Nat Methods 9:671–675.

53. Dávalos V, et al. (2012) Human SMC2 protein, a core subunit of human condensincomplex, is a novel transcriptional target of the WNT signaling pathway and a newtherapeutic target. J Biol Chem 287:43472–43481.

E4244 | www.pnas.org/cgi/doi/10.1073/pnas.1722020115 Suarez-Lopez et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

, 202

0