CDK4/6 inhibition triggers ICAM1-driven immune response and sensitizes LKB1 mutant lung cancer to immunotherapy

Abstract

Liver kinase B1 (LKB1) mutation is prevalent and a driver of resistance to immune checkpoint blockade (ICB) therapy for lung adenocarcinoma. Here leveraging single cell RNA sequencing data, we demonstrate that trafficking and adhesion process of activated T cells are defected in genetically engineered Kras-driven mouse model with Lkb1 conditional knockout. LKB1 mutant cancer cells result in marked suppression of intercellular adhesion molecule-1 (ICAM1). Ectopic expression of Icam1 in Lkb1-deficient tumor increases homing and activation of adoptively transferred SIINFEKL-specific CD8⁺ T cells, reactivates tumor-effector cell interactions and re-sensitises tumors to ICB. Further discovery proves that CDK4/6 inhibitors upregulate ICAM1 transcription by inhibiting phosphorylation of retinoblastoma protein RB in LKB1 deficient cancer cells. Finally, a tailored combination strategy using CDK4/6 inhibitors and anti-PD-1 antibodies promotes ICAM1-triggered immune response in multiple Lkb1-deficient murine models. Our findings renovate that ICAM1 on tumor cells orchestrates anti-tumor immune response, especially for adaptive immunity.

Fig. 1. Mutant Kras/Lkb1 driven lung cancer cell atlas was established.

a Schematic illustration for generation of the genetically engineered Kras-driven mouse model to establish Kras^G12D/+ mice with conditional knockout of Lkb1. b Micro-CT scan of the mouse lung. Left: Kras^G12D/+ sgTomato (K) mouse. Right: Kras^G12D/+ sgLkb1 (KL) mouse. c IHC staining of LKB1 in mouse lung tumor sections in K and KL mouse. Scale bar, 50 μm (40×). n = 1 experiment with n = 5 mice. d UMAP plot of 14,260 cells from lung tumor samples of K (n = 1) and KL (n = 1) mouse, colored by their 11 major cell types. e Heatmap of canonical cell-type markers of 11 major cell types. f Immune cell-type composition of each sample. g Expression level of Tcf7 in activated T cells and that of Ctla4 in exhausted T cells from lung tumor samples of K and KL mouse. ***p = 0.001, ****p < 0.0001. h Enrichment analysis on differentially expressed genes (DEGs) in KL versus K along the activated T cells using clusterProfiler for gene sets in GO terms. (Data were presented as violin plot (median, 25–75%, range). Statistical significance was tested with a two-tailed Mann–Whitney U test. ns, not significant; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).

Fig. 2. LKB1 regulates ICAM1 expression in non-small cell lung cancer.

a Enrichment analysis on DEGs in the cancer cells between KL (n = 701 cells) versus K (n = 662 cells) using clusterProfiler for gene sets in GO terms. b Venn diagram showing the overlap of DEGs between LKB1 wild-type and mutated groups both in tissues (GSE72094, n = 374 for LKB1-WT, n = 68 for LKB1-MUT) and cells (A549 cells, n = 3 for LKB1-WT, n = 3 for LKB1-MUT). Overrepresentation analysis was performed using these overlap genes, and the top 20 immune-specific signatures were revealed. Ranking of the genes among these enriched signatures according to their frequency. c Plot displays ICAM1 expression level in lung adenocarcinoma patient samples with (n = 73) and without (n = 437) LKB1 mutation. d Representative images and corresponding quantification of IHC staining of ICAM1 in human LUAD tumor sections in LKB1 wild-type and mutated groups. Scale bar, 50 μm (40×). n = 11 (WT) or n = 7 (MUT) biologically independent samples examined over 1 independent experiment. **p = 0.0128. e Relative RPKM values of ICAM1 in LKB1 mutant versus LKB1 wild-type lung cancer cells from CCLE, with or without KRAS co-mutations. Mutant KRAS: n = 7 with mutant LKB1, n = 14 with WT LKB1. WT KRAS: n = 10 with mutant LKB1, n = 38 with WT LKB1. Middle line: median; box edges: 25th and 75th percentiles; whiskers: the upper and lower ends of the whiskers signify the maxima and minima, respectively. Most extreme points that do not exceed ± interquartile range × 1.5; further outliers are marked individually. **p = 0.0066, **p = 0.0011. f Immunoblot (IB) of the indicated proteins in lung cancer cells with LKB1-loss or LKB1-intact. n = 3 experiments. g, h A549 and H460 lung cancer cells were transfected with lentivirus expressing the indicated genes (Ctrl, LKB1-WT, and LKB1-Mut). ICAM1 expression was analyzed by immunoblot (g, n = 3 experiments.) and flow cytometry (h; left, representative images; right, quantification of results. n = 3 biologically independent samples examined over 1 independent experiment. ****p < 0.0001, ****p < 0.0001.) (Results are presented as mean ± SEM. One-way ANOVA followed by Tukey’s multiple comparisons test, two-tailed Student’s t-test or two-tailed Fisher’s exact test was used to analyze the data. **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).

Fig. 3. ICAM1 is important for the homing and activation of tumor-specific T cells in LKB1 deficient lung cancer.

a LLC1-OVA tumor cells (LLC1-control-Icam1^WT-OVA; LLC1-shLkb1-Icam1^WT-OVA; LLC1-shLkb1-Icam1^OE-OVA) were implanted. Congenitally-marked (APC⁺) and stimulated CD8⁺ T cells expressing the OT-I were transferred on day 7 and analyzed on day 12. b Infiltration of OT-I CD8⁺ T cells (left) and expression of CD69 on OT-I cells (right) were analyzed by flow cytometry. For CD8⁺ T cells, n = 12, n = 12 and n = 13 in control-Icam1^WT, shLkb1-Icam1^WT and shLkb1-Icam1^OE groups, respectively. ****p < 0.0001, ****p < 0.0001. For CD69, n = 12, n = 11 and n = 12 in control-Icam1^WT, shLkb1-Icam1^WT and shLkb1-Icam1^OE groups, respectively. ***p = 0.0003. c Flow cytometry analysis of CD69 and PD-1 expression on activated T cells with or without co-culturing with LLC1 tumor cells (shLkb1-Icam1^WT, shLkb1-Icam1^OE) for 24 h. The E:T ratio (effector to target) was 1:1, 5:1, and 10:1 respectively. n = 3 biologically independent samples examined over 1 independent experiment. CD69, *p = 0.038, **p = 0.0012, ***p = 0.0009; PD-1, **p = 0.0036, ***p = 0.0007, ***p = 0.0009. d Flow cytometry analysis of CD69 expression on activated T cells co-culturing with LLC1 tumor cells (control-Icam1^WT, n = 9; shLkb1-Icam1^WT, n = 9; control-Icam1^OE, n = 6; shLkb1-Icam1^OE, n = 9) for 24 h. The E:T ratio (effector to target) was 10:1. ****p < 0.0001, ****p < 0.0001. e Flow cytometry analysis of CD69 and CD44 expression on activated T cells with co-culturing of LLC1 tumor cells (LLC1-shLkb1-Icam1^OE), with or without LFA-1 antibody blockade therapy for 24 h. The E:T ratio (effector to target) was 10:1. n = 6 biologically independent samples examined over 1 independent experiment. ****p < 0.0001, ***p = 0.0006. f Correlation analysis of ICAM1 expression and cytolytic activity (CYT) patient samples (TCGA) with or without LKB1 mutations. LKB1 wild-type, n = 374 samples, p < 0.0001; LKB1 mutation, n = 68 samples, p = 0.257. g Correlation analysis of IHC staining with anti-CD8a and anti-ICAM1 antibodies on a tissue microarray (TMA) of lung adenocarcinoma patients (n = 152 TMA elements). p < 0.0001. h IHC of ICAM1 and CD8 expression in LKB1-WT and LKB1-mutant lung cancer patients. Scale bar, 20 μm (40×), 50 μm (20×). n = 11 (LKB1-WT) or n = 7 (LKB1-mutant) biologically independent samples examined over 1 independent experiment. (Results were presented as mean ± SEM. One-way ANOVA followed by Tukey’s multiple comparisons test or two-tailed Student’s t-test was used to analyze the data. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).

Fig. 4. ICAM1 overexpression reactivated the TME and sensitized anti-PD-1 immunotherapy in Lkb1 deficient lung tumors.

a Total CD8⁺ T cells relative to CD45⁺ cells in tumor tissue, relative mean fluorescence intensity (MFI) of CD69 and CD44 on CD8⁺ T cells, percentage of CD62L⁺CD44⁺ (central memory T cells) and CD62L⁻CD44⁺ (effector memory T cells) in CD8⁺ T cells, and total NK cells relative to CD45⁺ cells in tumor tissue and relative MFI of CD69 on NK cells from mice bearing LLC1 tumors (LLC1-shControl-Icam1^WT, LLC1-shLkb1-Icam1^WT, LLC1-shLkb1-Icam1^OE) were analyzed by flow cytometry. n = 1 independent experiment. CD8⁺ T-cell, n = 13 in shControl-Icam1^WT, n = 8 in shLkb1-Icam1^WT, n = 16 in shLkb1-Icam1^OE; *p = 0.0193. CD69, n = 12 in shControl-Icam1^WT, n = 6 in shLkb1-Icam1^WT, n = 9 in shLkb1-Icam1^OE; *p = 0.0436. CD44, n = 5 samples in each group. Effector memory, n = 5 samples in each group. Central memory, n = 5 in shControl-Icam1^WT, n = 7 in shLkb1-Icam1^WT, n = 6 in shLkb1-Icam1^OE. NK cells, n = 7 in shControl-Icam1^WT, n = 6 in shLkb1-Icam1^WT, n = 5 in shLkb1-Icam1^OE; *p = 0.0415. CD69, n = 10 in shControl-Icam1^WT, n = 5 in shLkb1-Icam1^WT, n = 7 in shLkb1-Icam1^OE; *p = 0.0333. b LLC1-shLkb1-Icam1^WT or LLC1-shLkb1-Icam1^OE luc cells were injected into the left chest of mice and tumor formation was detected using a bioluminescence imager every week. Bioluminescent images in mice bearing lung tumors treated with isotype and anti-PD-1. c Quantification of bioluminescence results. n = 3 in Icam1^WT receiving IgG, n = 5 in Icam1^WT receiving anti-PD-1 Ab, n = 4 in Icam1^OE receiving IgG, n = 5 in Icam1^OE receiving anti-PD-1 Ab. *p = 0.0477. d C57BL/6 mice were subcutaneously inoculated with Lewis lung cancer cells with Lkb1 knockdown and stably Icam1 overexpression (Icam1^OE) or its parental (Icam1^WT) cell line. They were administered with different treatments (control immunoglobulin G (IgG), or anti-PD-1 Ab). Tumor size (left, n = 6 in Icam1^WT receiving IgG, n = 5 in Icam1^WT receiving anti-PD-1 Ab, n = 9 in Icam1^OE receiving IgG, and n = 8 in Icam1^OE receiving anti-PD-1 Ab. ***p = 0.0006.) and survival (right, n = 8 in Icam1^WT receiving IgG, n = 6 in Icam1^WT receiving anti-PD-1 Ab, n = 8 in Icam1^OE receiving IgG, and n = 8 in Icam1^OE receiving anti-PD-1 Ab. p = 0.0079) in different treatment arms were monitored. Kaplan–Meier curves of progression-free survival according to the expression level of ICAM1 in lung adenocarcinoma patients (e GSE126044 and GSE135222, n = 56 in total.) and melanoma patients (f GSE93157 and Liu et al., n = 46 in total.) following immune checkpoint inhibitor therapy. (Results are presented as mean ± SEM. Mixed-effects model followed by Tukey’s multiple comparison test or log-rank test was used to analyze the data. ns, not significant; *p < 0.05; ***p < 0.001. Source data are provided as a Source Data file.).

Fig. 5. CDK4/6 inhibitors were carefully selected for LKB1 mutant lung cancer based on the mechanistic link between LKB1-RB-CDK4 interaction.

a Combined drug screening using drugs from both the Genomics of Drug Sensitivity in Cancer (GDSC) database based on sensitivity to LKB1 mutation and an FDA-approved immunology compound library. b Quantitative RT-qPCR of relative ICAM1 expression in A549 and H460 cells treated with 5 selected drugs (cyclophosphamide monohydrate; vinorelbine tartrate; vorinostat; rapamycin; palbociclib). n = 4 biologically independent samples examined over 1 independent experiment. ****p < 0.0001, ****p < 0.0001. A549 cells transfected with lentivirus expressing the indicated genes (LV-Ctrl, LV-LKB1-WT, LV-LKB1-Mut), treated ± 500 nM palbociclib were harvested for quantitative RT-qPCR (c) and flow cytometry analysis (d). n = 3 biologically independent samples examined over 1 independent experiment. ***p = 0.0002, ****p < 0.0001, ***p = 0.007 (c). n = 4 biologically independent samples examined over 1 independent experiment. *p = 0.023, *p = 0.0378 (d). e, f Impact of CDK4/6i on LKB1 mutant lung cell line profiles. e Expression of immune resistance program genes (columns) that were most differentially expressed in palbociclib-treated (green) versus control (pink) cell lines (rows). Expression is normalized in each cell line. f Immune resistance scores in cell lines (A549, H460, and H2030) treated with palbociclib (“Pal”) or with DMSO vehicle (“con”) (GSE110397). Middle line: median; box edges: 25th and 75th percentiles. n = 2 biologically independent samples. g A549 cells transfected with lentivirus expressing the indicated genes (LV-Ctrl, LV-LKB1-WT, LV-LKB1-Mut), treated ± 500 nM palbociclib, transfected ± siRB were harvested for immunoblot. n = 3 independent experiments. h Immunoprecipitation analysis of the interaction among LKB1, RB, and CDK4 performed in H1299 cells expressing intact LKB1. n = 3 independent experiments. i Immunoblot analysis of the indicated proteins performed in A549 or H460 cells expressing the indicated genes in plasmids. n = 3 independent experiments. j ChIP assay was performed with cell lysates from A549/vector, A549/LKB1-OE, A549/CDK4-OE and A549/vector-palbociclib cells. A pair of primers flanking the p65 binding site within the ICAM1 promoter were used in PCR. Real-time PCR was employed to the ChIP assay. n = 4 biologically independent samples examined over 1 independent experiment. **p = 0.0015, *p = 0.0249, ****p < 0.0001. (Results are presented as mean ± SEM. One-way ANOVA followed by Tukey’s multiple comparisons test or two-tailed Student’s t-test was used to analyze the data. *p < 0.05; **p < 0.01; ***p < 0.001. Source data are provided as a Source Data file.).

Fig. 6. CDK4/6i sensitizes LKB1-deficient tumors to anti-PD-1 therapy.

a, b C57BL/6 mice were subcutaneously inoculated with Lewis lung cancer cells with Lkb1 knockdown (LLC1-shLkb1) and administered different treatments (control immunoglobulin G (Vehicle), palbociclib or anti-PD-1 Ab, or co-treatment with palbociclib and anti-PD-1 Ab). Tumor size (a, n = 7 for vehicle, n = 7 for palbociclib, n = 6 for anti-PD-1 Ab, n = 6 for combination. Two-way ANOVA followed by Tukey’s multiple comparisons test was performed to compare the tumor growth curves in different treatment groups, **p = 0.0022, ***p = 0.0003, ****p < 0.0001) and survival (b, n = 9 for vehicle, n = 12 for palbociclib, n = 9 for anti-PD-1 Ab, n = 8 for combination. Log-rank test was used to analyze the survival data, ****p < 0.0001, ****p < 0.0001, ****p < 0.0001) in different treatment arms were monitored. c Tumor volumes were measured beginning on day 7 and continuing every two days until day 21. d, e LLC1-shLkb1-luc cell line was constructed and injected into the left chest of mice. Tumor formation was detected. Mice harboring lung tumors were administered different treatments (control immunoglobulin G (Vehicle), palbociclib or anti-PD-1 Ab, or co-treatment with palbociclib and anti-PD-1 Ab). Representative bioluminescent images (d) and quantification of results (e). n = 4 biologically independent mice examined over 1 independent experiment. One-way ANOVA followed by Tukey’s multiple comparisons test was performed. **p = 0.0097. f, g Genetically engineered Kras-driven mouse model with conditional deletion of Lkb1 (KL GEMM) Representative MRI images (f) of KL GEMM lung tumors prior to treatment and after two weeks of treatment. n = 3, n = 3, n = 3 and n = 4 in vehicle, anti-PD-1 Ab, palbociclib and co-treatment groups, respectively. The contours of lung tumors were sketched. Waterfall plot (g) shows tumor volume response to the treatment. Each column represents one mouse. One-way ANOVA followed by Tukey’s multiple comparisons test was performed. *p = 0.0312. (Results are presented as mean ± SEM. A mixed-effects model followed by Tukey’s multiple comparisons was performed to compare the tumor growth curves in different treatment groups. Log-rank test was used to analyze the survival data. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).

Fig. 7. Using CDK4/6 inhibitor and anti-PD-1 antibody triggers an active tumor immune microenvironment.

a–c RNA-sequencing of murine tumor specimens after receiving different treatments was performed and analyzed. a Heatmap of stimulatory Immunomodulators (IMs) and inhibitory IM in different treatment arms (n = 3 samples in each groups). b Murine Microenvironment Cell Population counter (mMCPcounter) analysis of immune cell infiltration level in different treatment arms (n = 3 samples in each groups). c Cell–cell interactions by calculating the scores of communication signatures in different cell types across different treatment arms (n = 2 samples in each groups). d Total CD8⁺ T cells relative to cells in tumor tissue (left), MFI of IFN-γ in CD8⁺ T cells (middle), and MFI of PD-1 in CD8⁺ T cells (right) from mice bearing LLC1-shLkb1 tumors receiving the indicated treatments. For CD8⁺ T cells, n = 8 samples in each group; **p = 0.0016, **p = 0.0025, ***p = 0.0003. For IFN-γ, n = 6 (vehicle), n = 8 (palbociclib), n = 7 (anti-PD-1 Ab), and n = 7 (combination), respectively; *p = 0.0165, **p = 0.0027, **p = 0.0099. For PD-1, n = 7 samples in each group; *p = 0.0422, **p = 0.0081. e Total NK cells relative to cells in tumor tissue (left, n = 7 samples in each group) and MFI of CD107a in NK cells (right, n = 7 in vehicle, n = 9 in palbociclib, n = 6 in anti-PD-1 Ab and n = 8 in combination group) from mice bearing LLC1-shLkb1 tumors receiving the indicated treatments. NK cells, *p = 0.019, **p = 0.0052. CD107a, *p = 0.0351. f MFI of ICAM1 in tumor cells (middle, n = 10) and MFI of PD-L1 in tumor cells (right, n = 10) from mice bearing LLC1/shLkb1 tumors receiving the indicated treatments. ICAM1, *p = 0.0119, **p = 0.0039. PD-L1, *p = 0.0399. (Results are presented as mean ± SEM. One-way ANOVA followed by Tukey’s multiple comparisons test was used to analyze the data. ns, not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).

Fig. 8. In situ analyses and blocking treatment unveiled mechanism of synergistic effect of combined CDK4/6-PD-1 targeting therapy.

a IHC staining using anti-pRB, anti-ICAM1, and anti-CD8a antibodies on resected mouse samples following different arms of treatment in subcutaneous tumor models. n = 5 biologically independent mice examined over 1 independent experiment. b IHC staining using anti-ICAM1, and anti-CD8a antibodies on resected mouse samples following vehicle and combination treatment in GEMM KL models. Scale bar, 200 μm. n = 5 biologically independent mice examined over 1 independent experiment. c Immunofluorescence (IF) analysis of ICAM1⁺cells (green), CD8⁺ T cells (red) in resected KL GEMM samples (upper) and orthotopic mouse models (underneath) following different arms of treatment. Scale bar, 100 μm. n = 5 biologically independent mice examined over 1 independent experiment. d Lysates of mouse lung tumor tissues having received different treatments (vehicle, palbociclib, anti-PD-1 Ab, or combination of palbociclib with anti-PD-1 Ab) were mixed with a cocktail of biotinylated detection antibodies, and then incubated with the Mouse Cytokine Array. Array images were detected using X-ray films. n = 1 experiment with n = 3 mice. e, f C57BL/6 mice were inoculated with LLC1-shLkb1 cells and received the indicated depletion treatments (CD8-depleted, NK1.1-depleted, or ICAM1-depleted treatment). Tumor size (e n = 9 for vehicle, n = 10 for combination, n = 8 for ICAM1-depleted, n = 6 for CD8-depleted, and n = 7 for NK1.1-depleted. ****p < 0.0001, ***p = 0.0009, **p = 0.0069, *p = 0.0113.) and survival (f n = 8 for vehicle, n = 7 for combination, n = 10 for ICAM1-depleted, n = 6 for CD8-depleted, and n = 6 for NK1.1-depleted. ***p = 0.0003, **p = 0.0018, **p = 0.0059, *p = 0.0108.) were monitored. g C57BL/6 mice were inoculated with LLC1/shLkb1 with or without Icam1 knockout cells and received the combination treatments versus vehicle. Tumor size were monitored. n = 5, n = 7, n = 7, and n = 6, respectively. **p = 0.0053. h Total CD8⁺ T cells relative to CD45⁺ cells in tumor tissue, relative MFI of CD44 on CD8⁺ T cells from mice bearing LLC1-shLkb1 tumors with or without Icam1 knockout and receiving the indicated treatments were analyzed by flow cytometry. CD8⁺ T cells, n = 15 for sgCon with CDKi treatment, n = 9 for sgIcam1 with CDKi treatment, n = 18 for sgCon with combination treatment, and n = 12 for sgIcam1 with combination treatment; **p = 0.0095. CD44, n = 9 for sgCon with CDKi treatment, n = 12 for sgIcam1 with CDKi treatment, n = 9 for sgCon with combination treatment, n = 9 for sgIcam1 with combination treatment; **p = 0.0015. (Results are presented as mean ± SEM. A mixed-effects model followed by Tukey’s multiple comparisons was performed to compare the tumor growth curves in different treatment groups. Log-rank test was used to analyze the survival data. One-way ANOVA followed by Tukey’s multiple comparisons test was used to analyze the flow cytometry data. n.s., not significant; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Source data are provided as a Source Data file.).