SETD2 is the main transferase for the trimethylation of histone H3 at lysine 36 (H3K36me3)
in mammals. SETD2 plays important role in repairing DNA double strand breaks and maintaining
chromatin integrity.
1
In renal carcinoma, SETD2 deficiency caused DNA replication fork instability and DNA
damage.
2
The absence of SETD2 was also reported to have strong tumor-promoting effects in lung
adenocarcinomas (LUAD).
3
However, the other roles of SETD2 remain poorly understood in LUAD. In this study,
we utilized comprehensive omics-data analysis to determine that SETD2 was associated
with DNA damage and immune-related signals in LUAD. SETD2 knockdown induced DNA damage
and cGAS activation in LUAD cells, and reduced the number of cells at the G1 phase.
Moreover, SETD2 deficiency was conducive to mutation burden, immune cell infiltration,
and immunotherapy responses. High SETD2 expression was associated with high radiocurability.
These findings suggest that SETD2 may be a promising biomarker of therapeutic responses
for LUAD patients, and offer novel insights into immunotherapy and radiotherapy.
In the combined analysis of RNA sequencing (RNA-seq) from non-small cell lung cancer
in The Cancer Genome Atlas (TCGA NSCLC), assay for transposase-accessible chromatin
(ATAC) sequencing and H3K36me3 chromatin immunoprecipitation sequencing from GSE110318
and Roadmap confirmed that SETD2 was associated with expression and promoter opening
of DNA damage checkpoints (Fig. 1A), and that their promoters were covered by H3K36me3
(Fig. 1B). The cGAS/STING pathway senses double-stranded DNA (ds-DNA) and activates
the type I interferon signaling and innate immunity.
4
We next analyzed the 3 RNA-seq datasets (GSE89214, GSE75234, and GSE150809) of shSETD2
or H3.3K36 mutation vs. controls. As expected, shSETD2 or H3.3K36 mutation promoted
the type I interferon signaling and innate immunity (Fig. 1C–E).
Figure 1
SETD2 deficiency induced DNA damage, promoted the cGAS/STING signaling, enhanced genomic
mutations and immune cell infiltration, and was related to immunotherapy and radiotherapy.
(A) Correlation between SETD2 expressions and ATAC signals of DNA damage checkpoints.
(B) H3K36me3 signals of DNA damage checkpoints. (C) GSEA of shSETD2 vs. control in
GSE89214. (D) GSEA of control vs. H3.3K36 mutations in GSE75234. (E) GSEA of shSETD2
vs. control in GSE150809. (F) Representative immunofluorescence (200 ×) indicated
increased ds-DNA in siSETD2 H1975 cells. (G) The quantitative results of ds-DNA immunofluorescence.
(H) Representative immunofluorescence (200 ×) indicated increased pH2AX in siSETD2
H1975 cells. (I) The quantitative results of pH2AX immunofluorescence. (J) qRT-PCR
of cGAS/STING pathway components in H1975 cells with or without siSETD2. (K) The frequent
genomic mutations in patients with or without SETD2 mutations in TCGA cohorts. (L)
Proportion of lymphocyte infiltration of patients with or without SETD2 mutations
in the TCGA LUAD cohort. (M–O) Kaplan–Meier analysis of SETD2 mutation vs. WT group
in the ICI cohorts. (P) Correlation between SETD2 expression and survival fraction
at 2 Gy in GSE20298. (Q) Kaplan–Meier analysis of SETD2 expression in the 3 cohorts
with radiotherapy. (R) SETD2 expression and GMscore of patient clusters in the 3 cohorts
with radiotherapy.
Figure 1
H1975 cells had higher levels of SETD2 than the other NSCLC cells (Fig. S1). Immunofluorescence
indicated that SETD2 knockdown significantly increased ds-DNA and pH2AX in H1975 cells,
suggesting more DNA damage (Fig. 1F–I). As demonstrated in Figure 1J, siSETD2 had
high efficiency, and SETD2 deficiency induced the cGAS/STING pathway. Moreover, SETD2
was associated with cell cycle checkpoints in TCGA NSCLC cohort (Fig. S2A, B). SETD2
had the lowest expression in the G2/M cluster from single cell RNA-seq data GSE131907
(Fig. S2C, D). Less siSETD2 cells were arrested at the G1 phase compared with the
control group (Fig. S2E, F).
The negative regulation of DNA damage by SETD2 may be potentially beneficial to genomic
stability. We next explored the effects of SETD2 mutations on genomic mutation burden,
which was a sign of genomic instability. In TCGA pan-cancer datasets, 513 SETD2 mutation
sites were found, of which 188 sites with annotations were possibly oncogenic (Fig. S3),
suggesting that SETD2 mutations might change its tumor suppressor function. Moreover,
we found that driver mutations, but not variant of uncertain significance, were significantly
associated with lower expression levels of SETD2 in TCGA cohorts (Fig. S4). The whole
gene mutation rates were increased in patients with SETD2 mutations in the TCGA LUAD,
LUSC, and pan-cancer cohorts (Fig. 1K). In MSK-IMPACT cohort (pan-cancer, n = 10,945;
LUAD, n = 1357; LUSC, n = 170), patients with SETD2 non-synonymous mutations had high
genomic mutations (Fig. S5A). Similar results were found in the other 3 independent
data sets (Broad, LUAD, n = 183; MSS Mixed Solid Tumors, pan-cancer, n = 249; Cancer
Cell Line Encyclopedia, pan-cancer, n = 1739). In colon cancer of the CPTAC-2 cohort,
SETD2-mutated patients had higher percentages of microsatellite instability (Fig. S5B).
Next, we fairly checked the relations between each cut-off point of SETD2 expressions
and genome altered fractions. In Cancer Cell Line Encyclopedia pan-cancer data (n = 1739)
and NSCLC data (n = 135, Fig. S6), low expression of SETD2 was associated with increased
genome altered fraction in the majority of cut-off points and all significant points.
Similar results were also found in TCGA NSCLC (Fig. S7).
The cGAS/STING pathway activation and unstable genome may favor immune infiltration.
A total of 100 pathological slides were obtained from 10 randomly selected patients
with or without SETD2 mutations in TCGA LUAD cohort. We selected 10 images of 910 × 910
pixels in the region of tumor cells for each patient. The representative images were
shown in Figure S8. Patients with SETD2 mutations had higher proportions of lymphocyte
infiltration (Fig. 1L) and fewer tumor cells, but they were not related to macrophages
and stromal cells (Fig. S9).
Clinically, Kaplan–Meier analysis indicated that the upregulated SETD2 was associated
with longer survival in TCGA LUAD, GSE50081 (NSCLC, n = 172), and Raponi (LUSC, n = 130)
cohorts (Fig. S10). Using multivariate Cox regression to correct covariates (pathological
stage, age, and gender) in TCGA LUAD, SETD2 expression (P = 0.16), age (P = 0.13)
and pathological stage were related to the prognosis (Table S1). High immune infiltration
and mutation burden may be beneficial to immune checkpoint inhibitor (ICI) responses.
As expected, the SETD2 mutation group showed the favorable prognosis in the ICI cohorts
(MSK-IMPACT with anti-PD-1/PD-L1/CTLA-4, pan-cancer, n = 1661; CheckMate-012 with
anti-PD1 plus anti-CTLA-4, NSCLC, n = 75; DFCI GSE4573 with anti-CTLA-4, melanoma,
n = 110, Fig. 1M−O). Patients with low SETD2 expression had the favorable prognosis
in the DFCI cohort with immunotherapy (Fig. S11). High expression of SETD2 attenuated
the immunotherapy responses in all the significant cut-off points of SETD2 in the
GSE35640 cohort (melanoma with MAGE-A3, n = 65, Fig. S12).
Since DNA damage is related to radiation, we speculated that SETD2 was associated
with radiotherapy responses. SETD2 expression showed a strong negative correlation
with the survival fraction at 2 Gy in colorectal cancer cell data GSE20298 (P = 0.0004,
Fig. 1P). The high levels of SETD2 were correlated with superior prognosis in the
3 cohorts with radiotherapy (Fig. 1Q). In terms of multivariate Cox regression, SETD2
expression and pathological stages were the predictors of radiocurability (Table S2).
Previous studies suggested that an unstable genome contributed to radiotherapy resistance.
5
To expand the use scenario of genomic mutation, we developed a transcriptome-based
genomic mutation score (GMscore). The design process was outlined in Figure S13A.
Using only TCGA LUAD for training, our GMscore significantly discriminated between
the high- and low-frequent mutation groups in TCGA LUAD, LUSC, BRCA, and SKCM cohorts
(Fig. S13B). In the 3 cohorts with radiotherapy, patients were divided into 3 groups
(Fig. S13C). The high survival group was associated with the highest SETD2 expression
and lowest GMscore in comparison with the middle and low survival groups (Fig. 1R;
Fig. S13D).
Generally, our work demonstrated that defective SETD2 induced DNA damage, activated
the cGAS/STING pathway, and reduced G1 phase cells in LUAD. SETD2 downregulation favored
immune infiltration and immunotherapy responses. Moreover, deficient SETD2 was positively
correlated with genomic mutations, which was associated with unfavorable radiotherapy
prognosis.
Conflict of interests
The authors declare that they have no competing interests.
Funding
This work was supported by National Natural Science Foundation, China (No. 81972852,
81800429); Key Research & Development Project of Hubei Province, China (No. 2020BCA069);
Nature Science Foundation of Hubei Province, China (No. 2020CFB612); Young and Middle-Aged
Medical Backbone Talents of Wuhan, China (No. WHQG201902); Application Foundation
Frontier Project of Wuhan, China (No. 2020020601012221); Translational Medicine and
Interdisciplinary Research Joint Fund of Zhongnan Hospital of Wuhan University, China
(No. ZNJC201922, ZNJC202007).