immunology.sciencemag.org/cgi/content/full/4/39/eaau8943/DC1

Supplementary Materials for

RIG-I activation is critical for responsiveness to checkpoint blockade

Simon Heidegger*, Alexander Wintges, Florian Stritzke, Sarah Bek, Katja Steiger, Paul-Albert Koenig, Sascha Göttert,

Thomas Engleitner, Rupert Öllinger, Tatiana Nedelko, Julius C. Fischer, Vladimir Makarov, Christof Winter, Roland Rad, Marcel R.M. van den Brink, Jürgen Ruland, Florian Bassermann, Timothy A. Chan, Tobias Haas, Hendrik Poeck*

*Corresponding author. Email: simon.heidegger@tum.de (S.H.); hendrik.poeck@tum.de (H.P.)

Published 13 September 2019, Sci. Immunol. 4, eaau8943 (2019)

DOI: 10.1126/sciimmunol.aau8943

The PDF file includes:

Materials and Methods Fig. S1. Characterization of RIG-I–deficient B16.OVA clones. Fig. S2. Tumor cell–intrinsic RIG-I signaling promotes localized cross-presentation of tumor-associated antigen by CD103+ DCs in TdLNs. Fig. S3. Tumor cell–intrinsic RIG-I deficiency is associated with reduced TIL frequencies and decreased expression of proteins involved in T cell lytic function. Fig. S4. Anti–CTLA-4–mediated antitumor immunity does not rely on tumor cell–derived IFN-I. Fig. S5. Melanoma cell–intrinsic STING signaling induces IFN-I production but not programmed cell death. Fig. S6. Local RIG-I activation renders poorly immunogenic tumors susceptible to checkpoint inhibition. Fig. S7. Antitumor synergy between CTLA-4 blockade and local RIG-I activation is not restricted to melanoma. Fig. S8. Proposed model: Tumor-intrinsic RIG-I signaling promotes checkpoint inhibitor–mediated anticancer immunity. Legend for table S1 Table S2. Low RIG-I–encoding DDX58 expression in melanoma biopsies is an independent risk factor for death. References (53–57)

Other Supplementary Material for this manuscript includes the following: (available at immunology.sciencemag.org/cgi/content/full/4/39/eaau8943/DC1)

Table S1. Single-gene data murine tumor RNA-seq (Excel). Table S3. Raw data file (Excel).

MATERIALS AND METHODS

Tumor cell lines. The 4T1 cell line was cultured in complete RPMI-1640 medium.

Media and reagents. RPMI-1640 medium (Invitrogen) was supplemented with 10% (vol/vol)

FBS (Hyclone), 3 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin and 0.1%

β-mercaptoethanol (all from Sigma-Aldrich). DMEM medium (Invitrogen) was supplemented

with 10% FBS, 100 U/ml penicillin and 100 μg/ml streptomycin. Double-stranded in vitro-

transcribed 3pRNA (sense, 5′- UCA AAC AGU CCU CGC AUG CCU AUA GUG AGU CG -3′)

was generated as described (18). ISD single strand oligonucleotides were purchased from

Sigma-Aldrich (Munich, Germany) and were annealed by heating to 75 °C for 30 minutes and

re-cooling to room temperature.

CRISPR-Cas-mediated genome editing.

Single guide RNA target sequences:

Ddx58 (RIG-I) #1 5′-GCTGCTCCGCTGTCATTCTGTGG-3′

Ddx58 (RIG-I) #2 5′-GCCCTTGCTTCTGTTTGCAGAGG-3′

Ddx58 (RIG-I) #3 5′-GGCTGATGAGGATGATGGAGCGG-3′

Tmem173 (STING) 5′-CAGTAGTCCAAGTTCGTGCG-3′

Irf3 5′-GCATGGAAACCCCGAAACCG-3′

Irf7 5′-CTACGACCGAAATGCTTCCA-3′

Casp3 (Caspase-3) 5′-CGGGGTACGGAGCTGGACTG-3′

Mlkl 5′-GATGCAGTTGCAAATTAGCG-3′

If not stated otherwise, the Ddx58-deficient cell clones used for tumor experiments were

generated with sgRNA #2 (Panc02 cell line) or sgRNA #3 (B16.OVA and CT26 cell lines).

Preparation of cell suspensions from tumors and draining lymph nodes. Mice were

sacrificed and tumors as well as tumor-draining (inguinal) lymph nodes were removed using

forceps and surgical scissors. Tumors and lymph nodes were minced and homogenized by

filtering through a 100 µm nylon strainer (BD Bioscience). The tumor cell suspensions were

washed in complete RPMI before subsequent analysis. Tumor-infiltrating leukocytes were

enriched by 40/80 Percoll (Biochrom, Berlin, Germany) density gradient centrifugation

Histology and immunohistochemistry. Tumor samples were fixed in 10% neutral-buffered

formalin solution and embedded in paraffin. 2 µm-thin sections were analyzed both

histologically (Hematoxylin-Eosin, H.E.) and immunohistochemically (CD8a - clone GHH8,

Dianova). Antibody binding was visualized with diaminobenzidine (DAB) after a biotinylated

secondary rabbit anti-rat antibody (Vector Laboratories) and Streptavidin Peroxidase

(Medac). Stained slides were digitalized (Leica Biosystems, Wetzlar, Germany, AT-2) and

evaluated using the Aperio Imagescope software (version 12.3, Leica Biosystems, Wetzlar,

Germany). Immunohistochemical stainings were analyzed by documenting the percentage of

tumor tissue covered by CD8+ cells.

Cytokine concentrations in tumor tissue. Mice were injected sc with 2.4 x 105 B16.OVA

cells in each flank. When tumors were readily visible, 25 µg 3pRNA complexed in 3.5 µl

in vivo-jetPEI (vehicle) or just vehicle were injected into the right-sided tumors. 24 h later,

bilateral tumor tissue was removed and was frozen in liquid nitrogen. Tumors were

mechanically disrupted using mortar and pestle and were then suspended in Bio-Plex™ Cell

Lysis Buffer (Biorad, Munich, Germany). Homogenized tissue lysates were spun down at

20.000 G for 30 min, and supernatants were collected for further analysis. Protein

concentration was measured using the Pierce™ BCA Protein Assay Kit (Thermo Fisher

Scientific). IFN-I and cytokine levels were determined by ELISA (PBL Assay Science or

eBioscience) according to manufacturer´s protocol.

Immunoblotting. Tumor cell proteins were extracted using the radioimmunoprecipitation

assay (RIPA) buffer (Invitrogen), including complete protease inhibitor cocktail (Roche).

Protein yield was measured by BCA assay (Thermo Fischer Scientific) according to the

manufacturer’s protocol. Protein separation by SDS-PAGE was done on 10-12%

polyacrylamide gels for 90 min at 80 V. Proteins were blotted to a nitrocellulose blotting

membrane (GE Healthcare) for 90 min at 0.3 A and membranes were subsequently blocked

for 90 min in 5 % bovine serum albumine (BSA) or 5 % milk in 1x TBST (tris-buffered saline

and Tween 20). After incubation with primary antibodies (see table below) in blocking buffer

over night at 4°C and three additional washing steps with 1x TBST (each lasting for at least

10 min), secondary antibodies coupled with horse-radish-peroxidase (HRP) were incubated

for 1-2 h at room temperature. After three additional washing steps, signals were visualized

using Pierce ECL western blotting substrate according to the manufacturers’ protocol on the

INTAS science imaging system.

Western blot primary antibodies:

Target (murine) Isotype Clone Distributor

RIG-I Rat IgG2a SS1A Enzo

Phospho-MLKL (Ser345) Rabbit IgG Polyclonal Cell Signaling

β-Actin Rabbit IgG 13E5 Cell Signaling

Western blot HRP-linked secondary antibodies:

Target Isotype Clone Distributor

Rat IgG Goat IgG Polyclonal Cell Signaling

Rabbit IgG Goat IgG Polyclonal Cell Signaling

Flow cytometry.

Fluorochrome-coupled antibodies:

Gene expression analysis by quantitative real-time PCR. Total RNA was isolated from

cells lysed in TRIzol (Ambion) and was reversely transcribed using standard methods and

kits according to manufacturer´s protocols (RNeasy Mini Kit, Qiagen; SuperScript III Reverse

Transcriptase, Invitrogen). The specific primer pair sequences are listed below. Transcript

amplification was visualized with the qPCR Core kit for SYBR Green I (Eurogentec) using a

LightCycler 480 II (Roche) real-Time PCR system. The relative transcript level of each gene

was calculated according to the 2−ΔΔCt method with normalization to β-Actin.

Target (murine) Isotype Clone Distributor

CD3 Rat IgG2b, κ 17A2 Biolegend

CD4 Rat / IgG2b, κ GK1.5 eBioscience

CD11b Rat IgG2b, κ M1/70 Biolegend

CD11c Armenian Hamster IgG N418 eBioscience

CD8a Rat / IgG2a, κ 53-6.7 Biolegend

CD45 Rat IgG2b, κ 30-F11 Biolegend

CD64 Mouse IgG1, κ X54-5/7.1 Biolegend

CD80 (B7-1) Armenian Hamster IgG 16-10A1 Biolegend

CD86 (B7-2) Rat / IgG2a, κ GL1 eBioscience

CD103 LOU/M IgG2a, κ M290 BD Pharmingen

FOXp3 Rat IgG2b, κ MF-14 Biolegend

Granzyme B Rat / IgG2a, κ NGZB eBioscience

IFNγ Rat / IgG1a, κ XMG1.2 Biolegend

Ly6C Rat IgG2c, κ HK1.4 Biolegend

H-2Kb-bound SIINFEKL Mouse IgG1, κ 25-D1.16 Biolegend

MHC II (I-A/I-E) Rat IgG2b, κ M5/114.15.2 Biolegend

NK1.1 Mouse IgG2a, κ PK136 Biolegend

PD-L1 Rat IgG2b, κ 10F.9G2 Biolegend

TIL mRNA primer sequence:

mIFNG1 fwd 5′-ATCTGGAGGAACTGGCAAAA-3′

mIFNG1 rev 5′-TTCAAGACTTCAAAGAGTCTGAGG-3′

mGZMB fwd 5′-TGCTGCTCACTGTGAAGGAA-3′

mGZMB rev 5′-TTACCATAGGGATGACTTGCTG-3′

mCCL5 rev 5′-TGCCCACGTCAAGGAGTATTT-3′

mCCL5 fwd 5′-TCGAGTGACAAACACGATGC-3′

mPerforin fwd 5′-CAAGGTAGCCAATTTTGCAGC-3′

mPerforin rev 5′-GGCGAAAACTGTACATGCGAC-3′

mICOS fwd 5′-GCACTGGAGGAGAAGACTGC-3′

mICOS rev 5′-GAAGACAAAGACACGGCAGAA-3′

mLAG3 fwd 5′-CCGGGCCATAGAGGAGAT-3′

mLAG3 rev 5′-AGCAGTCCCAGAAGCAAAAA-3′

mHAVCR2 fwd 5′-TTTTCAGGTCTTACCCTCAACTG-3′

mHAVCR2 rev 5′-CATAAGCATTTTCCAATGACCTT-3′

mPDCD1 fwd 5′-TGCAGTTGAGCTGGCAAT-3′

mPDCD1 rev 5′-GGCTGGGTAGAAGGTGAGG-3′

B16 melanoma mRNA primer sequence:

mActin fwd 5′-CACACCCGCCACCAGTTCG-3′

mActin rev 5′-CACCATCACACCCTGGTGC-3′

mIFNb1 fwd 5′-ATAAGCAGCTCCAGCTCCAA-3′

mINFb1 rev 5′-GCAACCACCACTCATTCTGA-3′

Tumor challenge and treatment. For the bilateral tumor model, mice were injected sc with

Panc02 cells (1 x 106 right flank, 0.5 x 106 left flank) or 4T1 (2.4 x 105 right flank, 0.8 x 105 left

flank) cells on day 0. When tumors were readily visible (volume approximately 100 mm3),

25 µg 3pRNA complexed in 3.5 µl in vivo-jetPEI (Polyplus) were repeatedly injected into the

right-sided tumors, typically on day 6, 9 and 12. Anti-CTLA-4 (clone 9H10) or appropriate

isotype controls (200 µg ip, all from BioXCell, West Lebanon, NH) were administered

intraperitoneally at the same time points.

Tumor cell debris phagocytosis assay. Tumor cells were fluorescently labeled with CFSE

(2.5 µM, Invitrogen) and were subsequently transfected with 3pRNA (3 µg/ml). 48 h later,

tumor cells were cocultured with bone marrow-derived dendritic cells. After 6h, the uptake of

tumor cell debris by CD11c+ dendritic cells was determined by flow cytometry.

Murine melanoma RNA-seq library preparation and sequencing. Total cellular RNA was

extracted from bulk murine tumor samples using the RNeasyMini Kit (Qiagen). Library

preparation for bulk 3’-sequencing of poly(A)-RNA was done as described previously (53).

Briefly, barcoded cDNA of each sample was generated with a Maxima RT polymerase

(Thermo Fisher) using oligo-dT primer containing barcodes, unique molecular identifiers

(UMIs) and an adapter. 5’ ends of the cDNAs were extended by a template switch oligo

(TSO) and after pooling of all samples full-length cDNA was amplified with primers binding to

the TSO-site and the adapter. cDNA was tagmented with the Nextera XT kit (Illumina) and 3’-

end-fragments finally amplified using primers with Illumina P5 and P7 overhangs. The P5

and P7 sites were exchanged to allow sequencing of the cDNA in read1 and barcodes and

UMIs in read2 to achieve a better cluster recognition. The library was sequenced on a

NextSeq 500 (Illumina) with 75 cycles for the cDNA in read1 and 16 cycles for the barcodes

and UMIs in read2.

Murine melanoma RNA-seq analysis. Gencode gene annotations version M18 and the

mouse reference genome GRCm38.p6 were derived from the Gencode homepage

(https://www.gencodegenes.org/). Dropseq tools v1.12 was used for mapping the raw

sequencing data to the reference genome (54). The resulting UMI filtered countmatrix was

imported into R v3.4.4. Prior to differential expression analysis with DESeq2 1.18.1 (55), a

dummy variable was introduced with distinct levels for each combination of original

covariates describing the dataset (Genotype: Wild-type or RIG-I-/- ; Treatment: Ctrl or 3pRNA

or Isotype or anti-CTLA-4; Timepoint: day7 or day13). Dispersion of the data was estimated

with a parametric fit including the dummy variable as explanatory variable in the model. The

Wald test was used for determining differentially regulated genes between experimental

groups and shrunken log2 fold changes were calculated afterwards, with setting the type

argument of the lfcShrink function to ‘normal’. A gene was determined to be differentially

regulated if the adjusted p-value was below 0.05. Results from the differential expression

analysis were ranked according to the shrinked log2 foldchange and used as input for the

preranked GSEA (56). Pathway analysis was conducted within the Hallmark geneset

collection from the MSigDB v6.2 (57). RNA sequencing data for this study have been

deposited in the European Nucleotide Archive (accession #: PRJEB32241).

SUPPLEMENTARY FIGURES

Fig. S1. Characterization of RIG-I–deficient B16.OVA clones.

(A) Representative western blot of different B16.OVA cell clones after attempted

CRISPR/Cas9-mediated Ddx58 (RIG-I) deletion with wild-tpye (WT) control. The RIG-I-

deficient (RIG-I-/-) clone 3.6 was used for most subsequent experiments. (B-D) Selected

RIG-I-/- B16.OVA clones were transfected with 3pRNA in vitro using Lipofectamin 2000.

Results are from one representative experiment (n = 3 independent experiments). (B) Cell

death induction was assessed by Annexin-V and 7-AAD staining. Data give the frequency of

apoptotic cells (combined Annexin-V+ 7-AAD- and Annexin-V+ 7-AAD+ cells). (C-D) Release

of IFN-I was determined by ELISA. Data give fold change IFN-I production in comparison to

the wild-type ‘Vehicle’ control group. (E) In vitro growth dynamics of the RIG-I-/- B16.OVA

clone 3.6 and a selected IRF3/7-/- B16.OVA clone. (F) Mice were implanted with either WT,

RIG-I-/- B16.OVA cells clone 3.6 or clone 3.7 in the right flank. Tumor-bearing animals were

treated with anti-CTLA-4 antibody as described in Figure 1A. Data give overall survival of n =

5-10 individual mice per group pooled from two independent experiments. (G-H) As

described in Figure 1, mice were implanted with either WT or RIG-I-/- B16.OVA cells in each

flank. The specific RIG-I ligand 3pRNA was repeatedly injected into the right-sided tumor;

anti-CTLA-4 antibody was administered ip. (G) Individual growth curves of local, 3pRNA-

injected and distant tumors in n = 10 mice. (H) Long-term surviving animals were

rechallenged iv with WT B16.OVA cells and formation of pulmonal pseudometastases was

analyzed 14 days later. Naive mice were used as control for successful tumor induction. All

data are representative of at least two independent experiments. (I-J) Mice were

simultaneously implanted with a WT B16.OVA tumor on one side and a RIG-I-/- B16.OVA

tumor on the contralateral side. The specific RIG-I ligand 3pRNA was repeatedly injected

either in the WT (blue group) or the RIG-I-/- tumor (red group); anti-CTLA-4 antibody was

administered ip. (I) Growth of distant, non-RNA-injected tumors in both groups (blue = distant

RIG-I-/- tumor, red = distant WT tumor) is shown. Data give mean + S.E.M. of n = 10

individual mice per group pooled from two independent experiments. (J) Overall survival of

mice depicted in (I). Ctrl, control. WT, wild-type.

Fig. S2. Tumor cell–intrinsic RIG-I signaling promotes localized cross-presentation of

tumor-associated antigen by CD103+ DCs in TdLNs. (A) Wild-type (WT) or RIG-I-/-, CFSE-

labeled B16.OVA cells were transfected with 3pRNA for 24 h, washed and co-cultured with

bone-marrow derived dendritic cells (BMDCs). Uptake of fluorescently-labeled tumor cell

debris by BMDCs was determined by flow cytometry after 6 h. Graphs show percentage of

CFSE+ CD11c+ DCs ± S.E.M of n = 4-5 samples that were pooled from two independent

experiments. Representative histograms are gated on CD11c+ DCs and show CFSE

fluorescence intensity. (B-C) As described for Figure 2, mice were implanted with either WT

or RIG-I-/- B16.OVA cells in each flank. The RIG-I ligand 3pRNA was injected into the right-

sided tumor; anti-CTLA-4 antibody was administered ip. SIINFEKL cross-presentation by

CD103+ DCs was analyzed on day 15 after tumor induction. (B) The gating strategy is shown

in the left panel. Representative histograms are gated on Singlet CellDeathMarker- CD11chigh

MHC-IIhigh CD64- Ly6C- CD11b- CD103+ DCs and show MHC-I-SIINFEKL fluorescence

intensity on DCs from local (3pRNA-injected), tumor-draining lymph nodes (TdLN).

(C) Graphs show pooled data of DCs in TdLN of contralateral, non-injected tumors and give

mean of n = 10 individual mice normalized to the WT ‚isotype‘ control group. All data are

representative of at least two independent experiments.

Fig. S3. Tumor cell–intrinsic RIG-I deficiency is associated with reduced TIL

frequencies and decreased expression of proteins involved in T cell lytic function.

As described in Figure 2, mice were implanted with either wild-type (WT) or RIG-I-/- B16.OVA

cells in each flank. The RIG-I ligand 3pRNA was injected into the right-sided tumor; anti-

CTLA-4 antibody was administered ip. Tumor analyses were performed on day 15 after

tumor induction. (A) The frequency of tumor-infiltrating CD8+ leukocytes was determined by

immunohistochemistry. Bars give mean of n = 10-13 individual mice that were pooled from at

least two independent experiments. (B) Gene expression analysis within lysates of distant,

non-3pRNA-injected tumors was performed with real-time PCR. Data are normalized to gene

expression in isotype-treated mice bearing WT B16.OVA tumors (control) and give mean ±

S.E.M. of n = 10 individual mice that were pooled from two independent experiments. (C) WT

mice were inoculated with either WT or RIG-I-/- B16.OVA cells in the right flank and were

repeatedly treated with either anti-CTLA-4 or intratumoral 3pRNA as described in Figure 1D.

On day 7 and 13, next generation sequencing of bulk tumor RNA was performed and

summarized Gene Set Enrichment Analysis (GSEA) results are shown. Colors indicate the

direction of the pathway association with the respective treatment. Blue colors denote

negative association to the respective pathway, with less pathway activity in the treatment

group, whereas red colors denote enhanced pathway activity in the respective treatment

group. Dot sizes indicate the association strength to the respective pathway in terms of false

discovery rate (FDR) value.

Fig. S4. Anti–CTLA-4–mediated antitumor immunity does not rely on tumor cell–

derived IFN-I.

(A) Wild-type (WT) or IFNaR1-deficient (Ifnar1-/-) recipient mice were bilaterally inoculated

with WT B16.OVA cells. Tumor-bearing mice were injected with 3pRNA +/- anti-CTLA-4 as

described in Figure 3A. Data give mean tumor growth + S.E.M. of n = 4-8 individual mice per

group that were pooled from two independent experiments. (B) Mice bearing either WT or

RIG-I-/- tumors were treated with vehicle (jetPEI) or 3pRNA as described in Figure 3C.

Cytokine concentrations in local (3pRNA-injected) and distant (untreated) tumor tissue were

analyzed by ELISA. Data show values of n = 5 (vehicle) and n = 15 (3pRNA) individual mice

with mean. Data were pooled from three independent experiments. (C-D) WT and IRF3/7-

deficient (IRF3/7-/-) B16.OVA clones were transfected with 3pRNA in vitro. Release of IFN-I

was determined by ELISA. Error bars give the mean ± S.E.M. of quadruplicate samples. An

asterisk without brackets indicates comparison to the appropriate ‚vehicle‘ control group. (E)

WT mice were bilaterally inoculated with either WT or IRF3/7-/- B16.OVA cells and were

repeatedly treated with anti-CTLA-4 +/- 3pRNA as described in Figure 3E. Mean tumor

growth + S.E.M. of 3pRNA-injected and distant tumors in n = 10-15 individual mice per group

that were pooled from two independent experiments. (F) Wild-type and IRF3/7-/- B16.OVA

clones were transfected with 3pRNA in vitro as described in Figure S4C. Cell death induction

was assessed by Annexin-V and 7-AAD staining. Results are from one representative

experiment (n = 3 independent experiments).

Fig. S5. Melanoma cell–intrinsic STING signaling induces IFN-I production but not

programmed cell death.

(A) Wild-type (WT) or STING-deficient (STING-/-) B16.OVA clones were transfected with

3pRNA or ISD in vitro. After the indicated time points (6h for 3pRNA), Ifnb1 mRNA

expression was determined by real-time PCR. Data give the mean ± S.E.M. of triplicate

samples and are representative of two independent experiments. (B-C) WT cells were

stimulated as described above. At the indicated time points (48h for 3pRNA), cell death

induction was assessed by Live/Dead reagent and (B) Annexin-V or (C) cleaved caspase-3

staining and was analyzed by flow cytometry. Results are from one representative out of

three independent experiments in total. (D-F) WT recipient mice were implanted with

B16.OVA WT or STING-/- cells in each flank and were repeatedly treated with intratumoral

ISD injections +/- anti-CTLA-4 in equivalence to Figure 1D. Data are derived from the same

experiments that are depicted in Figures 6D-E, additionally including ISD-treated groups. (D-

E) Mean + S.E.M. tumor growth of ISD-injected and distant B16.OVA tumors. (F) Overall

survival of treated B16.OVA tumor-bearing mice. All data show values of n = 8-10 individual

mice per group that were pooled from two independent experiments. ND, not determined.

Fig. S6. Local RIG-I activation renders poorly immunogenic tumors susceptible to

checkpoint inhibition.

Mice were implanted with poorly immunogenic B16-F10 melanoma cells in each flank and

were repeatedly treated with anti-CTLA-4 +/- intratumoral 3pRNA as described in Figure 1D.

(A) Mean tumor growth + S.E.M. of 3pRNA-injected and (B) distant B16-F10 tumors and

(C) overall survival of n = 10 treated WT recipient mice per group. (D) Frequency of TRP2

Tetramer+ CD8+ T cells in peripheral blood was determined on day 15 after tumor induction.

(E-F) Some mice were additionally injected with anti-CD8α (cytotoxic T cells) or anti-CD4

(helper T cells) depleting antibodies, beginning two days prior to tumor induction. Mean

tumor growth + S.E.M. of (E) 3pRNA-injected and (F) distant B16-F10 tumors in WT mice

receiving additional depleting antibodies was analyzed. Figures give data of n = 10-20

individual mice per group that were pooled from at least two independent experiments. WT,

wild-type.

Fig. S7. Antitumor synergy between CTLA-4 blockade and local RIG-I activation is not

restricted to melanoma.

Using different tumor cell lines, mice were implanted with tumor cells in each flank and were

repeatedly treated with anti-CTLA-4 +/- 3pRNA as described in Figure 1D. (A-B) Tumor

growth of 3pRNA-injected and distant Panc02 pancreatic carcinomas. (C) Overall survival in

treated animals bearing bilateral Panc02 tumors. (D-E) Tumor growth of 3pRNA-injected and

distant 4T1 mammary carcinomas. All data give mean tumor growth + S.E.M. of n = 5-9

individual mice per group and are representative of two independent experiments. (F-H)

Tumor cell lines (B16.OVA melanoma, Panc02 pancreatic carcinoma, C26 colon carcinoma

and 4T1 mammary carcinoma) were transfected with 3pRNA in vitro using Lipofectamin

2000. (F-G) Release of IFN-I was determined by ELISA. All data are representative of two

independent experiments. (H) Cell death induction was assessed by Annexin-V and 7-AAD

staining. Results are from one representative experiment (n = 3 independent experiments).

Fig. S8. Proposed model: Tumor-intrinsic RIG-I signaling promotes checkpoint

inhibitor–mediated anticancer immunity.

SUPPLEMENTARY TABLES

Table S1. Single-gene data murine tumor RNA-seq (Excel).

The single gene level data underlying the gene set enrichment analysis (GSEA) of our

murine tumor RNA-Seq data presented in Figure S3C can be found in a separate Excel

sheet in the supplementary data.

Table S2. Low RIG-I–encoding DDX58 expression in melanoma biopsies is an

independent risk factor for death.

Uni- and multi-variable Cox regression analysis for disease-free and overall survival in

patients with malignant melanoma. HR, hazard ratio with 95% confidence interval. T4, locally

advanced disease. N pos, presence of lymph node metastasis. M1, presence of distant

organ metastasis.

Univariable Cox regression

Overall survival (OS)

Variable Values n Events HR p

DDX58 low 457 218 1.95 (1.49–2.56) < 0.001

TMEM173 low 457 218 1.19 (0.91–1.55) 0.2

T = T4 386 183 2.25 (1.64–3.07) < 0.001

N pos 403 195 1.71 (1.28–2.3) < 0.001

M M1 431 207 1.57 (0.8–3.07) 0.2

Age.in.decades numeric 457 218 1.28 (1.17–1.4) < 0.001

Gender Male 457 218 1.14 (0.86–1.51) 0.4

Disease free survival (DFS)

Variable Values n Events HR p

DDX58 low 399 256 1.36 (1.06–1.74) 0.01

TMEM173 low 399 256 1.23 (0.96–1.58) 0.09

T = T4 338 213 1.79 (1.33–2.4) < 0.001

N pos 355 232 2.19 (1.68–2.87) < 0.001

M M1 375 243 2.11 (1.25–3.57) 0.005

Age.in.decades numeric 399 256 1.23 (1.13–1.34) < 0.001

Gender Male 399 256 1.19 (0.91–1.54) 0.2

Multivariable Cox regression

Overall survival (OS)

Variable Values n Events HR p

DDX58 low 352 171 1.85 (1.35–2.51) < 0.001

T = T4 352 171 2.22 (1.59–3.11) < 0.001

N pos 352 171 1.89 (1.38–2.61) < 0.001

M M1 352 171 1.78 (0.72–4.41) 0.2

Age.in.decades 352 171 1.19 (1.07–1.32) 0.001

Gender Male 352 171 1.07 (0.77–1.48) 0.7

Disease free survival (DFS)

Variable Values n Events HR p

DDX58 low 311 201 1.34 (1.01–1.79) 0.04

T = T4 311 201 1.71 (1.25–2.34) < 0.001

N pos 311 201 2.25 (1.67–3.02) < 0.001

M M1 311 201 2.5 (1.25–4.97) 0.009

Age.in.decades 311 201 1.19 (1.08–1.31) < 0.001

Gender Male 311 201 1.1 (0.82–1.49) 0.5