Supporting Information

O’Connell et al. 10.1073/pnas.1109493108 SI Materials and Methods software. For quantification of microvessel density, regions of Cells and Transgenic Mice. 4T1 mammary carcinoma cells and CT26 highest vessel density were first identified at low magnification colorectal cancer cells were grown in DMEM supplemented with (magnification of 10×). Then, under high magnification (mag- 10% (vol/vol) FBS and 100 U/mL penicillin/streptomycin. The nification of 40×), microvessel density was determined by + S100A4-GFP (1), S100A4-tk (2), -C KO (3), S100A4-Cre counting CD31 blood vessels in 10 or more visual fields. All flox/flox (4), VEGF-A (5), and Rosa-lox-Stop-lox-YFP (6) transgenic patient samples were processed according to institutional mice have been described elsewhere. guidelines and an institutional review board-approved protocol.

+ Immunostaining. Harvested tumors and lungs were fixed in 10% Microarray Analysis. S100A4 stromal cells were isolated by (vol/vol) neutral buffered formalin, dehydrated, and embedded fluorescence-activated cell sorting (FACS) from normal lung in paraffin. Antigen retrieval was performed on deparaffinized tissue and metastatic lung tissue of S100A4-GFP mice. RNA was sections with 10 mM citrate buffer (pH 6.0) before im- isolated using the Qiagen RNeasy Plus Micro Kit following munostaining. Alternatively, harvested tumors and lungs were the manufacturer’s instructions and was then amplified with the embedded in O.C.T. medium (TissueTek) and immunostaining NuGen WT-Ovation Pico Kit. Microarray hybridization to the was performed on frozen sections. Anti-tk antibody was obtained Illumina Mouse Ref-8 Expression BeadChip was performed by from William C. Summers (Yale University, New Haven, CT). the Molecular Genetics Core Facility at Children’s Hospital in Anti-S100A4 antibody was a gift from Eric G. Neilson (Van- Boston. Microarray data have been deposited in National Center derbilt University, Nashville, TN). Anti-CD31 antibody was ac- for Biotechnology Information’s Expression Omnibus quired from Santa Cruz Biotechnology, anti-BrdU antibody was (GEO) and are accessible through GEO Series accession no. acquired from Roche, anti-Ki67 antibody was acquired from GSE31711. Microarray targets were sorted for ECM molecules, AbCam, anti-CK8 antibody was acquired from the National In- chemokines, and growth factors using the public stitutes of Health (NIH) Hybridoma Bank, anti–Tenascin-C identification numbers from the Gene Ontology Consortium antibody was acquired from AbCam, and anti–VEGF-A anti- [GO:0005604 (), GO:0008009 (chemokine body was acquired from NeoMarkers. TUNEL staining was activity), and GO:0008083 (growth factor activity)]. performed using an In Situ Cell Death Detection Kit (Roche). Cytospins were also prepared from the bone marrow of trans- FACS. FACS was performed on single-cell suspensions prepared planted mice at death and stained for Y using from lung tissue. To prepare a single-cell suspension, lung tissue StarFISH reagents (Cambio, Ltd.) per the manufacturer’s in- was cut into 1 mm3 pieces and incubated in 1 mg/mL collagenase/ structions to determine the percentage of bone marrow re- dispase solution (Roche) for 1 h in a 37 °C shaker. The resulting constitution by donor bone marrow (Fig. S9). For quantification suspension was filtered through a 40-μm cell strainer, treated of immunostaining, the number of positively stained cells was with ammonium chloride-potassium (ACK) lysis buffer for re- counted in 10 or more visual fields at a magnification of 40×, moval of red blood cells, and resuspended in a 2% (vol/vol) FBS whereas relative expression was quantified using NIH ImageJ in PBS plus propidium iodide solution before FACS analysis.

1. Xue C, Plieth D, Venkov C, Xu C, Neilson EG (2003) The gatekeeper effect of epithelial- 4. Bhowmick NA, et al. (2004) TGF-beta signaling in fibroblasts modulates the oncogenic mesenchymal transition regulates the frequency of breast cancer metastasis. Cancer potential of adjacent epithelia. Science 303:848e851. Res 63:3386e3394. 5. Dong J, et al. (2004) VEGF-null cells require PDGFR alpha signaling-mediated stromal 2. Iwano M, et al. (2001) Conditional abatement of tissue fibrosis using nucleoside fibroblast recruitment for tumorigenesis. EMBO J 23:2800e2810. analogs to selectively corrupt DNA replication in transgenic fibroblasts. Mol Ther 3: 6. Srinivas S, et al. (2001) Cre reporter strains produced by targeted insertion of EYFP and 149e159. ECFP into the ROSA26 locus. BMC Dev Biol 1:4. 3. Talts JF, Wirl G, Dictor M, Muller WJ, Fässler R (1999) Tenascin-C modulates tumor stroma and monocyte/macrophage recruitment but not tumor growth or metastasis in a mouse strain with spontaneous mammary cancer. J Cell Sci 112:1855e1864.

O’Connell et al. www.pnas.org/cgi/content/short/1109493108 1of6 ABCMet S

Met S

D Met E F S Met

S S Met S Met Met S Met

Fig. S1. S100A4+ stromal cells are recruited to metastatic lung tissue in human cancers. Sections of lung tissue from patients with cancer were stained with an antibody specific for S100A4 (brown or red). Metastatic areas and stromal regions are separated by dotted lines (Met, metastasis; S, stromal region). (Scale bars: 20 μm.) No specific staining for S100A4 is observed in normal lung tissue from an uninvolved area (A), whereas S100A4+ cells are observed in the stromal regions associated with metastases (B–E). Patient samples represent a variety of cancer types, including renal cell carcinoma (B), colorectal cancer (C, D, and F), and breast cancer (E).

A S100A4-GFP Ki67 DAPI Merged 30 cells

+ **** 20

10

0 per x400 visual field Normal Lung

# S100A4-GFP Normal Metastatic Lung Lung

+ 100

75

50 **** cells /S100A4-GFP

+ 25

0 Normal Metastatic Metastatic Lung % Ki67 Lung Lung

B tk tk 2.5 2.0 cells + 1.5 1.0 * 0.5 0 Control S100A4-tk GCV # S100A4-tk per x400 visual field Control S100A4-tk Tumor Tumor GCV

tk tk 12.5 10.0 cells + 7.5 5.0 2.5 Control S100A4-tk GCV 0 ** # S100A4-tk

per x400 visual field Control S100A4-tk Metastatic Lung Metastatic Lung GCV

Fig. S2. Proliferating S100A4+ stromal cells in the metastatic microenvironment can be targeted for selective ablation on GCV treatment of S100A4-tk mice. (A) Frozen sections of normal and metastatic lung tissue from S100A4-GFP mice were immunostained for Ki67 (red) with DAPI nuclear counterstain (blue) to assess proliferating S100A4+ stromal cells (green). White arrows point to S100A4-GFP+ single-positive cells; yellow arrowheads point to Ki67+/S100A4-GFP+ double-positive cells. (Scale bars: 20 μm.) The total number of S100A4-GFP+ cells and proliferating Ki67+/S100A4-GFP+ cells as a percentage of total S100A4-GFP+ cells was quantified on an average of 10 high-powered visual fields (magnification of 400×). (B) Representative photomicrographs of immunostaining for tk (red) with DAPI nuclear counterstain (blue) in 4T1 mammary tumors and metastatic lung lesions of control and GCV-treated S100A4-tk mice (n = 6 in each group). Arrows point to S100A4-tk+ cells. (Scale bars: 20 μm.) (A and B) Bar graphs present a quantitative assessment of S100A4-tk+ stromal cells in the tumor and metastatic microenvironments. Mean ± SEM. *P < 0.05; **P < 0.01; ****P < 0.0001.

O’Connell et al. www.pnas.org/cgi/content/short/1109493108 2of6 eattcae ntelvrwsquanti was liver the in area metastatic isev.mtsai ie isefrom tissue liver metastatic vs. tissue muotiig(rw)wt eaoyi ula onesan(upe najcn isescin fmtsai ugtsu rmptet ihbr with patients from tissue lung metastatic of sections tissue adjacent in (purple) shown. counterstain are nuclear hematoxylin with (brown) immunostaining S5. Fig. S4. Fig. Mean implantation. cell PBS; S3. Fig. O oto ( control ’ onl tal. et Connell n )adGVtetdW ie(TGCV; (WT mice WT GCV-treated and 7) = n baino S100A4 of Ablation C ramn ftmrbaigW ie T acrclswr rhtpclyipatdit h amr a aso B-rae Tmc (WT mice WT PBS-treated of pads fat mammary the into implanted orthotopically were cells cancer 4T1 mice. WT tumor-bearing of treatment GCV 104adC4 xrsini eattcln iseo ainswt ratcne.Rpeettv htmcorpso 104adCD45 and S100A4 of photomicrographs Representative cancer. breast with patients of tissue lung metastatic in expression CD45 and S100A4 )adGCV-treated and 8) = www.pnas.org/cgi/content/short/1109493108 ± SEM. + toa el teutslvrmtsai fC2 ooetlcne el.( cells. cancer colorectal CT26 of metastasis liver attenuates cells stromal B A S100A4-tk S100A4-GFP Normal Liver Control S100A4-GFP fi d eattclsosaeecrldb otdlns Mean lines. dotted by encircled are lesions Metastatic ed. AB

( 3 n Tumor volume (mm ) 2500 3000 1000 1500 2000 n )mc.Rpeettv htgah fHEsandlvrscinaeson Saebr:40 bars: (Scale shown. are section liver H&E-stained of photographs Representative mice. 4) = 500 ie( visual (8 mice ) rmr uo oue( volume tumor Primary 7). = 0 S100A4 PBS WT fi Metastatic Liver S100A4-GFP lsprgop.(cl as 50 bars: (Scale group). per elds S100A4-tk GCV GCV WT A n ecn eattcae nln ise( tissue lung in area metastatic percent and ) CD45 % Metastatic area 0.5 1.5 0 1 2 PBS WT

μ +

. ( m.) # S100A4-GFP cells % Metastatic area per x400 visual field ± 10 20 30 40 10 15 20 25 B 0 5 0 A T6cne el eeas netditapeial into intrasplenically injected also were cells cancer CT26 ) E.** SEM. Quanti ) GCV WT Control omlMet Normal P fi < aino S100A4 of cation .1 *** 0.01; S100A4-tk GCV *** ** P B < eeasse 4datrcancer after d 24 assessed were ) 0.001. + toa el nnra liver normal in cells stromal μ . Percent m.) atcancer east 3of6 A B Selective Expression Profiling of Fibroblast-Derived Top 20 Basement Membrane / ECM Proteins Rank Gene Fold-Change Normal Control S100A4-tk GCV Lung Metastatic Lung Metastatic Lung 1 Timp1 1.6 2 Lama2 1.3 I + +++ +++ 2 Col18a1 1.3 Collagen III + +++ +++ 2 Col7a1 1.3 Collagen XVIII + +++ +++ 2 Acan 1.3 Entactin +++ ++ ++ 6 Fras1 1.2 Fibronectin ED-A + +++ + 6 Smoc1 1.2 -1 (α1) + + + 6 Col4a6 1.2 Laminin-10 (α5) +++ ++ ++ 6 Lamc3 1.2 Perlecan +++ ++ ++ SPARC + ++ + 6 Lamc2 1.2 Tenascin-C - +++ + 6 Col4a5 1.2 TSP1 + +++ +++ 6 Col4a4 1.2 + ++ + 6 Lama4 1.2 6 Tnc 1.2 6 Col17a1 1.2 6 Fbln1 1.2 C TN-C TN-C 17 Col28a1 1.1 17 Col8a2 1.1 17 Lad1 1.1 17 Lamc1 1.1

Wildtype TN-C KO

D TN-C/CD31 H&E

Wildtype Wildtype

Fig. S6. Contribution of Tenascin-C to the metastatic microenvironment by S100A4+ stromal cells. (A) Selective expression profiling of candidate ECM proteins by immunostaining in normal lung tissue and metastatic lung tissue of control mice, as well as in metastatic lung tissue of GCV-treated S100A4-tk mice. Results are displayed as no staining (−), low expression (+), moderate expression (++), and high expression (+++). (B) Top 20 ECM molecules up-regulated by S100A4+ metastasis-associated stromal cells from a gene expression profile comparing S100A4+ stromal cells from normal lung (n = 3) and metastatic lung (n = 4). (C) Representative photomicrographs of metastatic lung sections from WT (Wildtype) and TN-C KO mice labeled with an antibody to Tenascin-C (green). Specific staining was absent in metastatic nodules of TN-C KO mice. (D) Representative image of Tenascin-C (green) and CD31 (red) double staining in the metastatic microenvironment of WT mice. No colocalization between Tenascin-C and CD31 was observed. H&E-stained panel displays the corresponding region in an adjacent section. Metastatic lesions are encircled by dotted lines. (Scale bars: 40 μm.)

O’Connell et al. www.pnas.org/cgi/content/short/1109493108 4of6 ABTop 20 Chemokines S100A4-Cre;Rosa-LSL-YFP Metastatic Lung and Growth Factors Gene Fold-Change Vegfa 2.9 Il6 2.7 Hbegf 2.3 Cxcl2 2.1 Csf2 2.1 Areg 2.0 Il5 2.0 Lif 1.9 S100A4 Armet 1.7 YFP Il1b 1.6 Il1a 1.6 Mdk 1.6 Cxcl10 1.5 Gdf9 1.5 Gmfg 1.4 Hdgf 1.4 Artn 1.4 Egf 1.3 Gdf5 1.3 Gdf15 1.3 DAPI Merged

C Wildtype S100A4-Cre;VEGF-Af/f

VEGF-A VEGF-A

Fig. S7. Contribution of VEGF-A to the metastatic microenvironment by S100A4+ stromal cells. (A) Top 20 chemokines and growth factors up-regulated by S100A4+ metastasis-associated stromal cells from a gene expression profile comparing S100A4+ stromal cells from normal lung (n = 3) and metastatic lung (n = 4). (B) Frozen sections of metastatic lung tissue from S100A4-Cre;Rosa-lox-Stop-lox-YFP mice were labeled with an antibody to S100A4 (red) and DAPI nuclear counterstain (blue). (Scale bars: 50 μm.) Arrows point to S100A4+/YFP+ cells (green), 85% of which also labeled positively with S100A4 antibody (red). (C) VEGF- flox/flox f/f A expression in the metastatic microenvironment of S100A4-Cre;VEGF-A (S100A4-Cre;VEGF-A ) mice and WT (Wildtype) littermates. Frozen sections of metastatic lung tissue from WT and S100A4-Cre;VEGF-Aflox/flox mice were labeled with an antibody to VEGF-A (green). (Scale bars: 20 μm.) Specific staining was absent in metastatic nodules of S100A4-Cre;VEGF-Aflox/flox mice.

O’Connell et al. www.pnas.org/cgi/content/short/1109493108 5of6 WT BM > WT Cre BM > Cre

Cre BM > WT WT BM > Cre

25 * WT BM > WT 20 15 Cre BM > Cre 10 Cre BM > WT 5

% Metastatic area WT BM > Cre 0

Fig. S8. VEGF-A from bone marrow-derived S100A4+ stromal cells does not functionally contribute to metastatic colonization. Quantification of metastatic flox/flox flox/flox burden in WT mice transplanted with WT bone marrow (WT BM > WT, n =5),S100A4-Cre;VEGF-A mice transplanted with S100A4-Cre;VEGF-A flox/flox flox/flox bone marrow (Cre BM > Cre, n = 3), WT mice transplanted with S100A4-Cre;VEGF-A bone marrow (Cre BM > WT, n = 3), and S100A4-Cre;VEGF-A mice transplanted with WT bone marrow (WT BM > Cre, n = 4) 24 d after orthotopic injection of 4T1 cancer cells. Representative H&E-stained lung sections are displayed with dotted lines encircling metastatic lesions. (Scale bars: 50 μm.) Mean ± SEM. *P < 0.05.

ABC

Female BM Male BM Male BM > Female

Fig. S9. Assessment of percentage of bone marrow (BM) reconstitution by Y chromosome staining. Cytospin preparations were generated from female bone marrow as a negative control (A), male bone marrow as a positive control (B), and bone marrow from experimental females with male donor bone marrow from our various bone marrow transplantation studies (C). Fluorescent in situ hybridization for the Y chromosome was performed on the cytospin prepa- rations. More than 90% of the bone marrow cells from experimental females in our bone marrow transplantation studies were positive for the Y chromosome, demonstrating >90% reconstitution of recipient bone marrow by donor bone marrow.

O’Connell et al. www.pnas.org/cgi/content/short/1109493108 6of6