immunology.sciencemag.org/cgi/content/full/5/49/eabc2728/DC1

Supplementary Materials for

Resistance to PD1 blockade in the absence of metalloprotease-

mediated LAG3 shedding

Lawrence P. Andrews, Ashwin Somasundaram, Jessica M. Moskovitz, Andrea L. Szymczak-Workman, Chang Liu, Anthony R. Cillo, Huang Lin, Daniel P. Normolle, Kelly D. Moynihan, Ichiro Taniuchi, Darrell J. Irvine, John M. Kirkwood,

Evan J. Lipson, Robert L. Ferris, Tullia C. Bruno, Creg J. Workman, Dario A. A. Vignali*,

*Corresponding author. Email: dvignali@pitt.edu

Published 17 July 2020, Sci. Immunol. 5, eabc2728 (2020)

DOI: 10.1126/sciimmunol.abc2728

The PDF file includes:

Fig. S1. Validation of the LAG3NC conditional knock-in mouse. Fig. S2. Tumor growth and analysis of TILs isolated from Lag3NC.L/L and Lag3NC.L/L CD4Cre mice. Fig. S3. Restriction of LAG3NC to T cells affects anti-PD1–mediated tumor regression. Fig. S4. Single-cell RNA-seq analysis of MC38 tumor-infiltrating T cells isolated from Lag3NC.L/L and Lag3NC.L/L CD4Cre mice. Fig. S5. Transcriptomic analysis of LAG3NC conventional CD4+ T cells in the context of PD1 blockade. Fig. S6. Transcriptomic analysis of LAG3NC regulatory CD4+ T cells in the context of PD1 blockade. Fig. S7. Transcriptomic analysis of LAG3NC CD8+ T cells in the context of PD1 blockade. Fig. S8. Tumor growth of Lag3NC.L/L ThPOKCreERT2 mice and analysis of tumor-infiltrating conventional CD4+ T cells isolated from Lag3NC.L/L and Lag3NC.L/L CD4Cre mice in the context of PD1 blockade. Fig. S9. Tumor growth of Lag3NC.L/L Foxp3CreERT2 mice and analysis of tumor-infiltrating regulatory CD4+ T cells isolated from Lag3NC.L/L and Lag3NC.L/L CD4Cre mice in the context of PD1 blockade. Fig. S10. Tumor growth of Lag3NC.L/L E8ICre.GFP mice and analysis of tumor-infiltrating CD8+ T cells isolated from Lag3NC.L/L, Lag3NC.L/L CD4Cre, and Lag3NC.L/L E8ICre.GFP mice in the context of PD1 blockade. Fig. S11. Adoptive transfer of CD8+ pmel into B16-gp100 tumor-bearing Lag3NC.L/L and Lag3NC.L/L CD4Cre mice. Fig. S12. Inhibition of ADAM10-mediated LAG3 shedding in vitro and in vivo. Fig. S13. LAG3 and ADAM10 expression on PBL and TIL isolated from patients with metastatic melanoma (cohort A).

Fig. S14. LAG3 and ADAM10 expression on PBL isolated from patients with advanced skin cancer (cohort B). Fig. S15. LAG3 and ADAM10 expression on PBL and TIL isolated from patients with HNSCC (cohort C). Fig. S16. LAG3 and ADAM10 expression on PBL isolated from treatment-naïve patients with HNSCC (cohort D).

Other Supplementary Material for this manuscript includes the following: (available at immunology.sciencemag.org/cgi/content/full/5/49/eabc2728/DC1)

Table S1 (Microsoft Excel format). Fold change and corrected P value for BD between random and sample conditions. Table S2 (Microsoft Excel format). Cohort A clinical information of patient with metastatic melanoma. Table S3 (Microsoft Excel format). Patient cohort B clinical information. Table S4 (Microsoft Excel format). Cohort C clinical information of patient with HNSCC patient. Table S5 (Microsoft Excel format). Cohort D clinical information of treatment-naïve patient with HNSCC. Table S6 (Microsoft Excel format). Raw data file.

Fig. S1

Fig. S1. Validation of the LAG3NC

conditional knock-in mouse. (A) LAG3 D4 domain, connecting peptide and transmembrane domain showing 12 amino acid residues of the connecting peptide removed (HSARRISGDLKG in red) in exon 7 to result in a non-cleavable form of LAG3. (B and C) Splenocytes were isolated from Lag3

NC.L/L,

Lag3NC.L/L

CD4Cre

or Lag3NC.L/L

E8ICre.GFP

mice and stimulated for 48 hours with 1 µg/ml anti-CD3. (D) Splenocytes were isolated from Lag3

NC.L/L ThPOK

CreERT2

mice that have received five consecutive intraperitoneal injections of tamoxifen (1 mg in 5% EtOH/sunflower oil) or vehicle and stimulated for 48 hours with 1 µg/ml anti-CD3. (E) Splenocytes were isolated from Lag3

NC.L/L Foxp3

CreERT2.GFP Rosa26LSLtdTomato

mice that have received tamoxifen or vehicle and stimulated as in (D). (F) CD4

+ or CD8

+ T cells were sorted from splenocytes isolated from Lag3

NC.L/L,

Lag3NC.L/L

CD4Cre

or Lag3NC.L/L

E8ICre.GFP

mice and stimulated with plate-bound anti-CD3 and soluble anti-CD28 (5 µg/ml) for 96 hours. Soluble LAG3 (sLAG3) was detected by ELISA in the supernatant. (G) sLAG3 was assessed in serum of Lag3

NC.L/L, Lag3

NC.L/L CD4

Cre, Lag3

NC.L/L E8I

Cre.GFP or LAG3.KO mice by ELISA. Results represent the mean of two independent experiments. **P < 0.01, ***P < 0.001 and ****P < 0.0001 by (F and G) unpaired t test.

Fig. S2

Fig. S2. Tumor growth and analysis of TILs isolated from Lag3NC.L/L and Lag3

NC.L/L CD4

Cre mice. Tumor

growth of Lag3NC.L/L and Lag3

NC.L/L CD4

Cre mice receiving (A) 5X10

5 MC38 adenocarcinoma cells subcutaneously

or (B) 1.25X105 B16-F10 melanoma cells intradermally. TIL was harvested at day 14 from Lag3

NC.L/L and Lag3

NC.L/L CD4

Cre mice injected with 5X10

5 MC38 adenocarcinoma cells subcutaneously and stained for (C)

LAG3 (MFI), (D) LAG3 (%), (E) TIGIT, (F) PD1, (G) 2B4, (H) TIM3, (I) Bcl2, (J) cleaved Caspase-3 (cCasp3), (K)

Ki67, (L) phospho-AKT (pAKT) and (M) phospho-S6 (pS6) on CD4+Foxp3

+, CD4

+Foxp3

– and CD8

+ T cells.

Results represent the mean of two (B, L and M) or three (A and C to K) independent experiments. *P < 0.05 and **P < 0.01. n.s., not significant by (A and B) two-way ANOVA and (C to M) unpaired t test. Error bars represent the means + SEM.

Fig. S3

Fig. S3. Restriction of LAG3NC

to T cells affects anti-PD1–mediated tumor regression. (A) Mean tumor growth curves of Lag3

NC.L/L and Lag3NC.L/L

CD4Cre

mice receiving 5X105 MC38 adenocarcinoma cells

subcutaneously and anti-PD1 or IgG (200 g) on days 6, 9 and 12 by intraperitoneal injection. (B) Individual tumor growth curves following secondary MC38 adenocarcinoma cell injection (2.5X10

5

cells

subcutaneous; day 42) of Lag3

NC.L/L and Lag3NC.L/L

CD4Cre

mice

following tumor resection (day 12) of primary MC38 tumor injection (5X10

5

cells) subcutaneously or sham control animals. Tumor growth of C57BL/6 mice receiving (C) B16-F10 (1.25X10

5

cells)

or (D) B16-gp100 (1.25X105

cells)

intradermally and immunized with Amph-gp100 or Amph-E7

vaccine subcutaneously on days 4 and 11 (20 g), with anti-PD1 or IgG as in (A). Results represent the mean of three (A) and two (B to D) independent experiments. *P < 0.05 and ****P < 0.0001. n.s., not significant by (A, C and D) two-way ANOVA. Error bars represent the means + SEM.

Fig. S4

Fig. S4. Single-cell RNA-seq analysis of MC38 tumor-infiltrating T cells isolated from Lag3NC.L/L and

Lag3NC.L/L

CD4Cre

mice. Lag3NC.L/L and Lag3

NC.L/L CD4

Cre mice were injected with 5X10

5 MC38 adenocarcinoma

cells subcutaneously and received anti-PD1 or IgG (200 g) by intraperitoneal injection on days 6, 9 and 12. (A) Flow cytometric gating strategy for sorting of CD45.2

+ CD4

+/CD8α

+/CD8β

+ cells from MC38 TIL. (B) Flow

cytometric analysis of MC38 TIL isolated from Lag3NC.L/L and Lag3

NC.L/L CD4

Cre mice receiving anti-PD1 or IgG at

d14 as in (A) showing relative proportions of CD8+, CD4

+Foxp3

– and CD4

+Foxp3

+ T cells.

Fig. S5

Fig. S5. Transcriptomic analysis of LAG3NC

conventional CD4+ T cells in the context of PD1 blockade.

Single-cell RNA-seq analysis of MC38 tumor-infiltrating CD4+Foxp3

– Tconvs isolated from Lag3

NC.L/L and Lag3NC.L/L

CD4Cre

mice at day 14 following injection with 5X105 MC38 adenocarcinoma cells subcutaneously and receiving

anti-PD1 or IgG (200 g) on days 6, 9 and 12 by intraperitoneal injection. (A) FltSNE projection showing overall

clustering of CD4+Foxp3

– Tconvs by sample type. (B) Number of cells per cluster identified in (A).

(C) Heatmap of

the top 10 differentially expressed genes in each cluster identified in (A). (D) Genes within select gene sets upregulated in cluster 2 relative to other clusters.

Fig. S6

Fig. S6. Transcriptomic analysis of LAG3NC

regulatory CD4+ T cells in the context of PD1 blockade. Single-

cell RNA-seq analysis of MC38 tumor-infiltrating CD4+Foxp3

+ Treg cells isolated from Lag3

NC.L/L and Lag3NC.L/L

CD4

Cre mice at day 14 following injection with 5X10

5 MC38 adenocarcinoma cells subcutaneously and receiving

anti-PD1 or IgG (200 g) on days 6, 9 and 12 by intraperitoneal injection. (A) Clustering of CD4+ Foxp3

+ Tregs by

DRAGON revealed a total of nine clusters across all samples. (B) Number of cells per cluster identified in (A). (C) Scaled sample enrichment in clusters identified in (A). (D) Gene set enrichment analysis showing enriched pathways in each cluster identified in (A).

(E) FltSNE projection showing overall clustering of CD4

+ Foxp3

+ Treg

cells by sample type. (F) Heatmap of the top 10 differentially expressed genes in each cluster identified in (A).

Fig. S7

Fig. S7. Transcriptomic analysis of LAG3NC

CD8+ T cells in the context of PD1 blockade. Single-cell RNA-

seq analysis of MC38 tumor-infiltrating CD8+ T cells isolated from Lag3

NC.L/L and Lag3

NC.L/L CD4

Cre mice at day 14

following injection with 5X105 MC38 adenocarcinoma cells subcutaneously and receiving anti-PD1 or IgG (200 g)

on days 6, 9 and 12 by intraperitoneal injection. (A) Clustering of CD8+ T cells by DRAGON revealed a total of six

clusters across all samples. (B) Number of cells per cluster identified in (A). (C) Scaled sample enrichment in clusters identified in (A). (D) Gene set enrichment analysis revealed enriched pathways in each cluster identified in (A). (E) FltSNE projection showing overall clustering of CD8

+ T cells by sample type. (F) Heatmap of the top 10

differentially expressed genes in each cluster identified in (A).

Fig. S8

Fig. S8. Tumor growth of Lag3NC.L/L

ThPOKCreERT2

mice and analysis of tumor-infiltrating conventional CD4+

T cells isolated from Lag3NC.L/L

and Lag3NC.L/L

CD4Cre

mice in the context of PD1 blockade. (A) Mean tumor growth curves of Lag3

NC.L/L and Lag3

NC.L/L ThPOK

CreERT2 mice receiving 5X10

5 MC38 adenocarcinoma cells

subcutaneously and anti-PD1 or IgG (200 g) on days 6, 9 and 12 by intraperitoneal injection, as well as five consecutive intraperitoneal injections of tamoxifen (1 mg in 5% EtOH/sunflower oil) from days 0 to 4. (B) TIL was harvested at day 14 from Lag3

NC.L/L or Lag3

NC.L/L CD4

Cre mice injected with 5X10

5 MC38 adenocarcinoma cells

subcutaneously receiving anti-PD1 or IgG (200g) on days 6, 9 and 12 by intraperitoneal injection. Mice that received anti-PD1 were stratified into non-responders (N) and responders (R) to treatment. (C) IFN-γ and (D) TNF-α from CD4

+Foxp3

– TIL was assessed in mice described in (B), following re-stimulation with phorbol myristate

acetate (PMA) and ionomycin for 4 hours in the presence of brefeldin A. (E) KLRG1 staining was assessed in mice described in (B). (F) Representative flow plots of KLRG1 and TNF-α from CD4

+Foxp3

– TIL described in (B). (G)

Cleaved caspase-3 and (H) Bcl2 staining of CD4+Foxp3

– TIL was assessed in mice described in (B). Results

represent the mean of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. n.s., not significant by (A) two-way ANOVA and (B to E, G and H) Mann-Whitney Test.

Fig. S9

Fig. S9. Tumor growth of Lag3NC.L/L

Foxp3CreERT2.GFP

mice and analysis of tumor-infiltrating regulatory CD4+

T cells isolated from Lag3NC.L/L

and Lag3NC.L/L

CD4Cre

mice in the context of PD1 blockade. (A) Survival, (B) mean tumor growth curves and (C) individual tumor growth curves of Foxp3

CreERT2.GFP, Lag3

NC.L/L and Lag3

NC.L/L

Foxp3CreERT2.GFP

mice receiving 5X105 MC38 adenocarcinoma cells subcutaneously and anti-PD1 or IgG (200 g)

on days 6, 9 and 12 by intraperitoneal injection, as well as five consecutive intraperitoneal injections of tamoxifen (1 mg in 5% EtOH/sunflower oil) from days 0 to 4. (D) TIL was harvested at day 14 from Lag3

NC.L/L or Lag3

NC.L/L

CD4Cre

mice injected with 5X105 MC38 adenocarcinoma cells subcutaneously receiving anti-PD1 or IgG (200 g)

on days 6, 9 and 12 by intraperitoneal injection and frequency of Foxp3+

Tregs (%CD4+ T cells) was assessed. (E)

BrdU+ Ki67

+ in CD4

+Foxp3

+ TIL was assessed in mice that received an intraperitoneal injection of BrdU 12 hours

before harvest. (F) Cleaved caspase-3 and (G) Bcl2 staining in CD4+Foxp3

+ TIL was assessed in mice, described

in (D). Results represent the mean of three independent experiments. *P < 0.05, **P < 0.01 and ****P < 0.0001. n.s., not significant by (A) log-rank (Mantel-Cox), (B) two-way ANOVA and (D to G) Mann-Whitney Test.

Fig. S10

Fig. S10. Tumor growth of Lag3NC.L/L

E8ICre.GFP

mice and analysis of tumor-infiltrating CD8+ T cells isolated

from Lag3NC.L/L

, Lag3NC.L/L

CD4Cre

, and Lag3

NC.L/L E8I

Cre.GFP mice in the context of PD1 blockade. (A) Mean

tumor growth curves of Lag3NC.L/L

and Lag3NC.L/L

E8ICre.GFP

mice receiving 5X10

5 MC38 adenocarcinoma cells

subcutaneously and anti-PD1 or IgG on days 6, 9 and 12 (200 g) by intraperitoneal injection. (B) TIL was harvested at day 14 from Lag3

NC.L/L, Lag3

NC.L/L E8I

Cre.GFP or Lag3

NC.L/L CD4

Cre mice injected with 5X10

5 MC38

adenocarcinoma cells subcutaneously receiving anti-PD1 or IgG (200 g) on days 6, 9 and 12 by intraperitoneal injection. Mice that received anti-PD1 were stratified into non-responders (N) and responders (R) to treatment. (C) IFN-γ and (D) TNF-α from CD8

+ TIL was assessed in mice described in (B) following restimulation with phorbol

myristate acetate (PMA) and ionomycin for 4 hours in the presence of brefeldin A. (E) BrdU+ Ki67

+ in CD8

+ TIL was

assessed in mice that received an intraperitoneal injection of BrdU 12 hours before harvest. (F) Cleaved caspase-3 and (G) Bcl2 staining of CD8

+ TIL was assessed in mice described in (B). Results represent the mean of three

independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. n.s., not significant by (A) two-way ANOVA and (B to G) Mann-Whitney Test.

Fig. S11

Fig. S11. Adoptive transfer of CD8+ pmel into B16-gp100 tumor-bearing Lag3

NC.L/L and Lag3

NC.L/L CD4

Cre

mice. (A) Lag3NC.L/L

or Lag3NC.L/L

CD4Cre

mice received an adoptive transfer of 1x105

pmel (Thy1.1+) cells the day

before inoculation with 1.25X105

B16-gp100 melanoma cells intradermally. Mice received anti-PD1 (200 g) intraperitoneally on days 6, 9 and 12. (B) TIL, draining (DLN) and non-draining (NDLN) lymph nodes were harvested at day 15 and pmel (Thy1.1

+ CD8

+) infiltration was assessed. IFNγ was measured on both Thy1.1

+ and

Thy1.2+ CD8

+ T cells following re-stimulation with phorbol myristateacetate (PMA) and ionomycin for 4 hours in the

presence of brefeldin A. IFN-γ was correlated with (C) tumor size or (D) Thy1.1+CD8

+ T

cell frequency. (E)

Thy1.1+CD8

+ T

cell frequency was correlated with tumor size. Results represent the mean of three independent

experiments. n.s., not significant by (B) Mann-Whitney Test.

Fig. S12

Fig. S12. Inhibition of ADAM10-mediated LAG3 shedding in vitro and in vivo. (A) Foxp3Cre-YFP

mice were injected with 1.25X10

5 B16-F10 melanoma cells. At day 14 post tumor inoculation, CD8

+, CD4

+Foxp3

– and

CD4+Foxp3

+ TIL and peripheral populations were sorted. RNA was extracted using QIAGEN RNEasy MicroKit,

cDNA synthesized and Adam10 mRNA was assessed. (B) Purified CD4+Foxp3

– (YFP

–) or (C) CD8

+ T cell

splenocytes isolated from Foxp3Cre-YFP

mice and stimulated with anti-CD3 and cultured with GI254023X inhibitor (GI) or DMSO for 72 hours. Inhibition of soluble LAG3 release detected by ELISA. (D) Alzet osmotic pumps releasing 20mg/kg/d GI254023X inhibitor or DMSO were implanted in C57BL/6 mice and injected with 5X10

5 MC38

adenocarcinoma cells subcutaneously and anti-PD1 on days 6, 9 and 12 (200 g) by intraperitoneal injection. Soluble LAG3 was detected in the serum of the mice by ELISA. (E) TIL, draining (DLN) and non-draining (NDLN) lymph nodes were isolated from C57BL/6 or Lag3

NC.L/L CD4

Cre mice, as in (D) at day 15 post inoculation. IFN-γ from

CD8+ TIL was assessed following restimulation with phorbol myristate acetate (PMA) and ionomycin for 4 hours in

the presence of brefeldin A. (F) Tumor size was measured at day 15. Results represent the mean of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. n.s., not significant by (A to F) unpaired t test. Error bars represent the means + SEM.

Fig. S13

Fig. S13. LAG3 and ADAM10 expression on PBL and TIL isolated from patients with metastatic melanoma (cohort A). Melanoma patient cohort A (Table S2). (A) Lymphocytes were isolated from peripheral blood (PBL) and tumor (TIL) from melanoma patients (n = 14), or healthy donors (n = 10) and stained fresh for (A to F) LAG3 or (G to L) ADAM10 on CD4

+Foxp3

– T cells, CD4

+Foxp3

+ T cells and CD8

+ T cells. Paired analysis of LAG3 and

ADAM10 expression was assessed on (M and N) CD4+Foxp3

– T cells, (O and P) CD4

+Foxp3

+ T cells and (Q and

R) CD8+ T cells. *P < 0.05, ** P <0.01, ***P < 0.001 and ****P < 0.0001. n.s., not significant by (A to L) unpaired t

test.

Fig. S14

Fig. S14. LAG3 and ADAM10 expression on PBL isolated from patients with advanced skin cancer (cohort B). Advanced metastatic melanoma patient cohort B (Table S3). Patients were stratified as responders or progressors following SOC anti-PD1, or anti-PD1 and anti-CTLA4 therapy. Lymphocytes were isolated from peripheral blood pre- and post-treatment and expression of (A to C) LAG3 expression on CD4

+Foxp3

– T cells,

CD4+Foxp3

+ T cells and CD8

+ T cells was assessed. (D and E) The change of LAG3 and ADAM10 expression

was assessed for CD4+ Foxp3

+ T cells and CD8

+ T cells and patients were stratified by responsiveness to

treatment. (F to H) ADAM10 expression on CD4+Foxp3

– T cells, CD4

+ Foxp3

+ T cells and CD8

+ T cells was

assessed. Paired analysis of LAG3 and ADAM10 expression was assessed on (I) CD4+Foxp3

+ T cells and (J)

CD8+ T cells. (K and L) LAG3:ADAM10 ratio was assessed for CD4

+Foxp3

+ T cells and CD8

+ T cells. *P < 0.05.

n.s., not significant by (A to C, F to H, K and L) Wilcoxon test and (D and E) unpaired t test.

Fig. S15

Fig. S15. LAG3 and ADAM10 expression on PBL and TIL isolated from patients with HNSCC (cohort C). Head and neck squamous cell carcinoma (HNSCC) patient cohort C (Table S4). (A) Lymphocytes were isolated from peripheral blood (PBL) and tumor (TIL) from HNSCC patients (n = 29), or healthy donors (HD; n = 5) and stained for (A to C) LAG3, (D to F) ADAM10 and (G to I) ADAM17 on CD4

+Foxp3

– T cells, CD4

+Foxp3

+ T cells

and CD8+ T cells. Paired analysis of LAG3 and ADAM10 expression was assessed on (J) CD4

+Foxp3

– T cells, (K)

CD4+Foxp3

+ T cells and (L) CD8

+ T cells. LAG3 expression on PBL or TIL was stratified by patient HPV status as

HPV positive (+), HPV negative (–) or non-determined (ND), on (M) CD4+Foxp3

– T cells, (N) CD4

+Foxp3

+ T cells

and (O) CD8+ T cells. *P < 0.05, **P < 0.01 and ***P < 0.001. n.s., not significant by (A to I and M to O) unpaired t

test.

Fig. S16

Fig. S16. LAG3 and ADAM10 expression on PBL isolated from treatment-naïve patients with HNSCC (cohort D). Head and neck squamous cell carcinoma (HNSCC) patient cohort D (Table S5). (A) Soluble LAG3 (sLAG3) was detected by ELISA in patient serum. Patients were stratified by stage of disease. (B) Lymphocytes were isolated from patient peripheral blood and ADAM10 expression was assessed on CD4

+Foxp3

– T cells,

CD4+Foxp3

+ T cells and CD8

+ T cells. Paired analysis of LAG3 and ADAM10 expression in patient PBL on (C)

CD4+Foxp3

+ T cells and (D) CD8

+ T cells. (E) Survival curve of advanced HNSCC patients (n = 25) with high

LAG3 (>23.5%) or low LAG3 (<23.5%) expression on CD4+Foxp3

– T cells. **P < 0.01 and ****P < 0.0001. n.s., not

significant by (A and B) unpaired t test and (E) log-rank (Mantel-Cox).