Cell autonomous single agent eCNTFR-Fc treatment results in suppression of the RAS/MAPK signaling axis.
To assess the effect of eCNTFR-Fc on tumor development, we used an autochthonous, highly aggressive GEM model of lung adenocarcinoma12,13. Intratracheal instillation of adenovirus (Ad5-CMV-Cre) in KrasG12D−LSL/Trp53f/f (KP) mice led to consistent development of adenocarcinoma. Tumors were induced in the lungs of KP mice and treated with vehicle or eCNTFR-Fc. Mice were sacrificed at one- or three-weeks after treatment initiation and lung cells harvested for analysis (Fig. 1a). To understand the effects of eCNTFR-Fc on tumorigenesis, treated tumors were analyzed using single-cell RNAseq and spatial transcriptomics. scRNAseq analysis was performed on 24,783 cells across all treatments. Cell types were identified based on reference cell annotations and manual curation (Fig. 1b and Extended Data Fig. 2a). Single cell transcriptome analysis of the tumor compartment identified effects of eCNTFR-Fc on genes related to immune modulation at 3 weeks (Fig. 1c). For example, eCNTFR-Fc upregulated Gst01, a gene coding for Glutathione transferase Omega 1, a protein previously shown to play a pro-inflammatory role14. eCNTFR-Fc also downregulated SelenBP1 expression which is associated with immune infiltration and is negatively correlated with the presence of NK cells, T helper cells, central and effector memory T cells and CD8 + T cells15. Pathway analysis showed enrichment of pathways related to immune function including antigen processing and presentation, viral infection and T cell receptor signaling (Fig. 1d). In contrast, the MAPK and PD-L1/PD1 immune checkpoint pathways were downregulated in tumor cells as shown by GSEA enrichment analysis (Fig. 1e-f and Extended Data Fig. 2b-c) at both 1- and 3-weeks post treatment. Downregulation of MAPK is consistent with prior data indicating that eCNTFR-Fc inhibits KRAS signaling3. PD-L1 signaling is possibly reduced due to decreased expression of Jun, Hif-1 a, and Nfkb in eCNTFR-Fc treated cells, all known regulators of PD-L1 expression16,17. Overall, eCNTFR-Fc treatment resulted in consistent dysregulation of 108 genes in the tumor cell compartment in the KrasG12D/Trp53f/f model.
To further assess the consequences of eCNTFR-Fc on tumor cells, spatial analysis of changes in protein expression was performed on the Nanostring DSP Platform. Lung sections were stained for CD45 (yellow), PanCK (green), and DNA (blue) to establish overall tissue morphology. Regions of interest (ROIs) of 300 µm in diameter were selected for molecular profiling with a 43-plex oligonucleotide–antibody cocktail designed to query the MAPK pathway and key immune subtypes (see methods). Examples of the lung morphologies are shown in Extended Data Fig. 3. Protein expression analysis in situ between vehicle and eCNTFR-Fc treated tumor cells also identified decreases in p42/44 MAPK and PD-L1 (Fig. 1g-h). Differential protein expression analysis demonstrated decreases in total RAS, MEK, and ERK expression (Fig. 1i), which aligns with the decreases in gene expression seen using single cell transcriptome analysis.
RAS/MAPK pathway alteration in response to inhibition of the CLCF1/CNTFR signaling axis was also seen in human lung cancer cell lines (Extended Data Fig. 4). Human LUAD cell lines with stably integrated Cas9 were transfected with sgRNA guides to Control or CNTFR (Synthego) to generate Control (sgNT) or CNTFR knockout cell lines. Three KRAS mutant lines, (A549 (G12S), H23 (G12C), H358 (G12C)), and KRAS wildtype lines (H1437 and H1975) were evaluated. As previously reported, cell lines with KRAS mutations were generally more sensitive to KRAS knockout/knockdown18 and KRAS mutant cell lines were more susceptible to eCNTFR treatment3. An exception was the KRAS G12C mutant cell line, H23, which did not show sensitivity to KRAS knockout (Extended Data Fig. 4a) consistent with previous reports18–21. The two most KRAS-dependent cell lines, determined by either knockout or with RNA interference, A549 and H358, were the most affected by CNTFR knockout (Extended Data Fig. 4b). Individual growth curves for A549, H358 and H23 with control or CNTFR knockouts conditions are shown in Extended Data Fig. 4c-e. Cells with a knockout of CNTFR had a lower capacity to form colonies compared to the non-targeting guides in A549 and H358, but not H23 (Extended Data Fig. 4f-h), consistent with the growth assay results.
To further define the tumor-intrinsic effects of CLCF1-CNTFR signaling, kinase signaling was analyzed under serum-starved and serum-stimulated conditions in sgNT and sgCNTFR A549 clones. CNTFR knockout resulted in abrogation of the JAK/STAT pathway and partial inhibition of the MEK/ERK and AKT pathways (Extended Data Fig. 5a). Western blotting also confirmed changes to the kinetics and intensity of ERK phosphorylation (Extended Data Fig. 5b-e). The effects on ERK signaling were not observed in H23 cells, which, as noted above, are refractory to KRAS inhibition (Extended Data Fig. 5f-g). In KRAS mutant (G12S) LUAD cell line A549, CNTFR knockout led to complete abrogation of the JAK/STAT and partial inhibition of the MEK/ERK and AKT pathways (Extended Data Fig. 5b-c).
While H23 cells were not sensitive to either KRAS inhibition or CNTFR knockout alone, knockout of CNTFR increased sensitivity to KRAS G12C inhibition in H23 and H1792 knockout cell lines (Extended Data Fig. 6a-b). The same effect was observed when WT H23 cells were treated with eCNTFR-Fc in combination with KRAS G12C inhibition (Extended Data Fig. 6c). When KRAS G12C inhibition was combined with CNTFR knockout, there was an almost complete abrogation of p-ERK, indicating a synergistic effect on inhibition of MAPK signaling (Extended Data Fig. 6d-e). Overall, these results indicate that the CLCF1-CNTFR axis supports proliferation in the presence of KRAS inhibition and that abrogation of this axis using eCNTFR-Fc is a potential therapeutic strategy relevant to human cancers that carry KRAS mutations.
Blockade of CLCF1-CNTFR signaling remodels the tumor microenvironment.
Single cell transcriptome analysis identified significant changes to immune cell populations in response to eCNTFR-Fc treatment. After one week of treatment with eCNTFR-Fc, macrophages were significantly depleted, whereas both B and T cells expanded (Fig. 2a). After three weeks of treatment, these changes were more apparent, with a strong enrichment of M1-like (Cd11c+Cd206loCd86hi) macrophages and additional depletion of M2-like (Cd11c+Cd206hi) macrophages. At both treatment timepoints, there were fewer CD8 + T cells with exhaustion markers (Lag3, Ctla4, Pd-1, Tigit, and Havcr2) and an increase in cytotoxic CD8 + T cells, NKT cells, and NK cells.
The effect of eCNTFR-Fc treatment on T cells was further evaluated using differential gene expression analysis of both naïve and CD8 + subsets (Fig. 2b-c). After one week of eCNTFR-Fc treatment, there was a significant upregulation of ribosomal protein-coding genes (Fig. 2b), which is known to occur prior to T cell expansion22, consistent with the observed enrichment of mature effector T cell populations at the three-week timepoint. Differential gene expression analysis of eCNTFR-Fc treated CD8 + T cells and NKT cells at the three-week timepoint identified downregulation of these ribosomal proteins and upregulation of expression of Id2 and Gzma (Granzyme A)22, suggesting these cell populations have passed the expansion phase and have reached a mature cytotoxic phenotype (Fig. 2c). In parallel, MAPK signaling increased in T cells after eCNTFR-Fc treatment, in contrast to the observed effect in the tumor cell population and suggesting that this is an indirect effect.
Spatial analysis was used to analyze proteins associated with CLCF1-dependent changes in the immune microenvironment. After three weeks of treatment with eCNTFR-Fc, the T cell activation markers CD28 and MHC II were upregulated, whereas the T cell exhaustion markers CTLA4, PDL1, and PD-1 were downregulated23,24 (Fig. 2e-f). CD28 is required for PD-1 blockade to efficiently kill cancer cells25. Ly6C/G expression was also decreased in CD45 + cells at both eCNTFR-Fc treatment timepoints (Fig. 2g). The Ly6 antibody has dual specificity, binding to both Ly6C and Ly6G, which marks two separate populations. Ly6C marks a subset of monocytes/macrophages, and Ly6G marks neutrophils26–28, thus these two populations cannot be distinguished using this marker. The single cell analysis corroborates losses in both populations, with greater losses seen in the neutrophil populations when comparing vehicle to eCNTFR-Fc (Fig. 2a). Neutrophils have been shown to contribute to the immunosuppressive environment and facilitate immune evasion, reducing the effectiveness of immune checkpoint inhibition29. Taken together, eCNTFR-Fc treatment leads to an immune phenotype consistent with decreased immunosuppression and enrichment of activated effector immune cells.
Blockade of CLCF1 signaling potentiates the effect of checkpoint blockade.
The changes to the immune compartment after eCNTFR-Fc treatment suggest that the blockade of CLCF1 signaling could potentiate the effect of checkpoint inhibitor therapy in lung cancer. To test this, we tested the efficacy of combined therapy with eCNTFR-Fc and αPD-1 in the KP GEM model. At 8-weeks post tumor initiation, mice were treated with vehicle, eCNTFR-Fc alone, αPD-1 alone, or a combination of eCNTFR-Fc and αPD-1 for 28 days, and tumors were collected at 16 weeks (Fig. 3a). A representative image of lungs under each of the different treatments is shown in Fig. 3b. Mice treated with single agent eCNTFR-Fc demonstrated decreased tumor burden compared to vehicle-treated controls, while single agent αPD-1 had little effect (Fig. 3c). The effect of eCNTFR alone was less significant than in prior work3, most likely due to the higher tumor burden in mice treated in this study (see methods). Strikingly, combination of eCNTFR-Fc and αPD-1 led to a greater than additive effect and significantly decreased tumor burden when compared to vehicle, suggesting that eCNTFR-Fc is sufficient to overcome the poor T cell responses, which usually renders immunotherapy ineffective in the KP lung adenocarcinoma model30–35.
To further dissect this drug interaction, we used a syngeneic allograft model where mouse LUAD cells, harvested from a GEM model mouse, were injected and grew subcutaneously in the mouse flank. Mice were implanted and treated with vehicle, eCNTFR-Fc alone, αPD-1 alone, or a combination of eCNTFR-Fc and αPD-1 for up to 28 days, and tumors were collected once mice reached endpoint (Fig. 3d). While eCNTFR-Fc or αPD-1 alone did have an effect on tumor growth in this model, with 7 of 18 tumors having decreased tumor volume and 2 achieving a complete response to eCNTFR-Fc alone and 4 of 19 tumors out of having decreased tumor volume and 3 achieving a complete response to αPD-1 alone. Mice that received a combination of eCNTFR-Fc and αPD-1 had significantly decreased tumor growth (Fig. 3e-f). 11 of 22 mice had tumor regression, with 7 mice achieving a complete response, and the remaining mice exhibiting more intermediate responses (Fig. 3e-f). Combination treatment also significantly increased the survival of mice in comparison to the vehicle and single agent treated mice (Fig. 3g).
Composition of the immune system is altered after combination therapy with eCNTFR and αPD-1.
To evaluate the mechanistic basis for the enhanced response to αPD-1 when combined with eCNTFR-Fc, we performed single cell sequencing of lung tumors treated with single or combination therapy (Fig. 4a). Treatment with αPD-1 alone resulted in the depletion of Cd206hi M2-like macrophages and an increase of Cd86hiCd206lo M1-like macrophages, similar to eCNTFR-Fc, but without the reciprocal enrichment of T cell effector populations. Cytotoxic NK, NKT, Cd8 + T cells and naïve T cells all were decreased after treatment with αPD-1. Notably, the only enrichment in T-cell populations after αPD-1 was an increase in exhausted CD8 + T cells, which can no longer mount an effective anti-tumor response. In contrast, when eCNTFR-Fc was combined with αPD-1, there was an overall decrease in the relative number of macrophages. As with single agent eCNTFR-Fc treatment, there was enrichment in cytotoxic CD8 + T cells, NKT cells, and NK cells, but additionally, there were increases in CD4 + T cells and T Regs (marked by Cd4 and Foxp3, respectively). Interestingly, T Regs have classically been shown to have decreased anti-tumor responses36; however, some newer studies suggest careful control of CD4 expressing cells, including T Regs, are required for an optimal immune environment to promote anti-tumor efficacy of immunotherapy37.
Subsetting of the macrophage population allowed for further characterization of these cells and increased resolution, allowing for the identification of 3 macrophage subpopulations (Fig. 4b): one M1-like (immunostimulatory) and two M2-like (immunosuppressive) populations. Treatment with eCNTFR-Fc resulted in the depletion of CD206 + M2-like macrophages, similar to single agent αPD-1 treatment (Fig. 4c). We observed expression of secreted cytokines in one subset of the M2-like macrophages. Expression of a set of cytokines has been correlated to low response to immunotherapy in NSCLC38,39. Using this “refractory to PD-1” gene signature, we then calculated a score for the expression of the gene signature on a per-cell basis and identified an M2-like subset with a high PD-1 treatment refractory signature score (Fig. 4d), suggesting that these cells may partially mediate response to αPD-1. Additionally, it may suggest that eCNTFR-Fc contributes to the sensitization to αPD-1 by preventing the differentiation of naïve monocytes into this subset of macrophages, preventing subsequent secretion of refractory cytokines.
Although αPD-1 treatment alone is sufficient to show the depletion of M2-like macrophages, this does not translate to the sensitivity to αPD-1 in the GEM model. This is likely due to the lack of reciprocal activation of effector cell populations and the possibility of secretion of refractory cytokines by multiple cell types (Extended Data Fig. 7), including other myeloid populations, monocytes, DCs and neutrophils, stromal cells, and tumor cells themselves. When looking at the response of tumor cells to treatment, tumor cells treated with αPD-1 alone showed increased expression of the genes associated with PD-1 refractory signature responses (Fig. 4e). However, treatment with eCNTFR-Fc alone or in combination with αPD-1 led to significant loss of the refractory signature, suggesting that suppression of the CLCF1/CNTFR signaling axis is preventing the release of these cytokines. This is not overly surprising as several genes in the signature are members of the IL6 subfamily or are targets of STAT3, which are canonically activated downstream components of CLCF1/CNTFR signaling40,41. To further confirm whether tumor cells are contributing to the cytokine expression changes in response to CLCF1/CNTFR signaling, we evaluated cytokine expression under serum-starved and serum-stimulated conditions in A549 control and CNTFR knockout cell lines using a cytokine array. CNTFR knockout led to decreased expression of IL-6 and IL-11 under both serum-starved and stimulated conditions (Extended Data Fig. 8). VEGF42 had decreased expression in CNTFR knockout cells and VEGF expression did not respond as well to serum stimulation, consistent with the observed decrease in STAT3 activation. VEGF has also been shown to directly influence T cell response by increasing the expression of immunosuppressive checkpoints43,44. Similarly, we saw decreased expression of Angiopoietin-2, previously described as a biomarker of poor response to immunotherapy45 and linked to poor prognosis for patients with NSCLC46.
To evaluate whether this response to combination treatment requires T cell activation, tumors were treated with a CD8 antibody to deplete CD8 + T cells in conjunction with the eCNTFR-Fc and αPD-1 combination therapy. CD8 + T cell depletion completely abolished the effect of the eCNTFR-Fc and αPD-1 combination (Fig. 4f-g), indicating that this effect is T-cell dependent. In summary, the combination of eCNTFR-Fc and αPD-1 treatment alters the cellular composition of the TME, leading to a phenotype of decreased immunosuppression and increased enrichment of effector immune cells and T cell mediated anti-tumor activity.