Taylor & Francis Taylor & Francis Group

Neurological Research

Neurological Research A Journal of Progress in Neurosurgery, Neurology and Neurosciences

ISSN: 0161-6412 (Print) 1743-1328 (Online) Journal homepage: www.tandfonline.com/journals/yner20

APOBEC3C is a novel target for the immune treatment of lower-grade gliomas

Shufa Zhao, Yuntao Li, Jie Xu & Liang Shen

To cite this article: Shufa Zhao, Yuntao Li, Jie Xu & Liang Shen (2024) APOBEC3C is a novel target for the immune treatment of lower-grade gliomas, Neurological Research, 46:3, 227-242, DOI: 10.1080/01616412.2023.2287340

To link to this article: https://doi.org/10.1080/01616412.2023.2287340

+

View supplementary material

Published online: 26 Nov 2023.

Submit your article to this journal ☒

Article views: 2635

View related articles

View Crossmark data

CrossMark

Citing articles: 3 View citing articles ☒

Taylor & Francis Taylor & Francis Group

L Check for updates

APOBEC3C is a novel target for the immune treatment of lower-grade gliomas

Shufa Zhaoª*, Yuntao Lia*, Jie Xua and Liang Shen (D b

ªDepartment of Neurosurgery, Huzhou Cent Hospital, Affiliated Cent Hospital Huzhou University, Huzhou, Zhejiang, China; bDepartment of Neurosurgery, The affiliated Changzhou Second People’s Hospital of Nanjing Medical University, Changzhou, Jiangsu, China

ABSTRACT

Background: Apolipoprotein B mRNA editing catalytic polypeptide-like (APOBEC) type 3C (A3C) has been identified as a cancer molecular biomarker in the past decade. However, the practical role of A3C in lower-grade gliomas (LGGs) in improving the clinical outcome remains unclear. This study aims to discuss the function of A3C in immunotherapy in LGGs.

Methods: The RNA-Sequencing (RNA-seq) and corresponding clinical data were extracted from UCSC Xena and the results were verified in the Chinese Glioma Genome Atlas (CGGA). Weighted gene co-expression network analysis (WGCNA) was used for screening A3C-related genes. Comprehensive bioinformation analyses were performed and multiple levels of expres- sion, survival rate, and biological functions were assessed to explore the functions of A3C. Results: A3C expression was significantly higher in LGGs than in normal tissues but lower than in glioblastoma (GBM), indicating its role as an independent prognosis predictor for LGGs. Twenty-eight A3C-related genes were found with WGCNA for unsupervised clustering analysis and three modification patterns with different outcomes and immune cell infiltration were identified. A3C and the A3C score were also correlated with immune cell infiltration and the expression of immune checkpoints. In addition, the A3C score was correlated with increased sensitivity to chemotherapy. Single-cell RNA (scRNA) analysis indicated that A3C most probably expresses on immune cells, such as T cells, B cells and macrophage.

Conclusions: A3C is an immune-related prognostic biomarker in LGGs. Developing drugs to block A3C could enhance the efficiency of immunotherapy and improve disease survival.

Abbreviation: A3C: Apolipoprotein B mRNA editing catalytic polypeptide-like (APOBEC) type 3C; LGGs: lower-grade gliomas; CGGA: Chinese Glioma Genome Atlas; WGCNA: Weighted gene co- expression network analysis; scRNA: Single-cell RNA; HGG: higher-grade glioma; OS: overall survival; TME: tumor microenvironment; KM: Kaplan-Meier; PFI: progression-free interval; IDH: isocitrate dehydrogenase; ROC: receiver operating characteristic; GS: gene significance; MM: module member- ship; TIMER: Tumor IMmune Estimation Resource; GSVA: gene set variation analysis; ssGSEA: single- sample gene-set enrichment analysis; PCA: principal component analysis; AUC: area under ROC curve; HAVCR2: hepatitis A virus cellular receptor 2; PDCD1: programmed cell death 1; PDCD1LG2: PDCD1 ligand 2; PTPRC: protein tyrosine phosphatase receptor type C; ACC: Adrenocortical carci- noma; BLCA: Bladder Urothelial Carcinoma;BRCA: Breast invasive carcinoma; CESC: Cervical squa- mous cell carcinoma and endocervical adenocarcinoma; CHOLCholangiocarcinoma; COADColon adenocarcinoma; DLBC: Lymphoid Neoplasm Diffuse Large B-cell Lymphoma; ESCA: Esophageal carcinoma; GBM: Glioblastoma multiforme; HNSC: Head and Neck squamous cell carcinoma; KICH: Kidney Chromophobe; KIRC: Kidney renal clear cell carcinoma; KIRP: Kidney renal papillary cell carcinoma; LAML: Acute Myeloid Leukemia; LGG: Brain Lower Grade Glioma; LIHC: Liver hepatocel- lular carcinoma; LUAD: Lung adenocarcinoma; LUSC: Lung squamous cell carcinoma; MESO: Mesothelioma; OV: Ovarian serous cystadenocarcinoma; PAAD: Pancreatic adenocarcinoma; PCPG: Pheochromocytoma and Paraganglioma; PRAD: Prostate adenocarcinoma; READ: Rectum adenocarcinoma; SARC: Sarcoma; SKCM: Skin Cutaneous Melanoma; STAD: Stomach adenocarci- noma; TGCT: Testicular Germ Cell Tumors; THCA: Thyroid carcinoma; THYM: Thymoma; UCEC: Uterine Corpus Endometrial Carcinoma; UCS: Uterine Carcinosarcoma; UVM: Uveal Melanoma

ARTICLE HISTORY Received 27 July 2023 Accepted 21 November 2023

KEYWORDS

Glioma; APOBEC3C; immune treatment; weighted gene co-expression network analysis (WGCNA); single-cell RNAS

1 Background

Glioma is a malignant tumor that seriously affects the lives of patients. The prognosis of patients with lower- grade gliomas (LGGs), also known as diffuse low- grade and intermediated-grade gliomas (WHO grade II and III), is slightly better than that of higher-grade gliomas (HGG) [1]. Glioblastoma, one type of HGG,

shows more aggressive biological characteristics, with a median survival time of fewer than 15 months [2]. Additionally, the overall survival (OS) of LGG patients ranges between 5 to 10 years, and most LGGs will progress to HGG within 10 years [1].

Despite the rising costs in glioma treatment and research, limited improvement in clinical prognosis has

CONTACT Liang Shen soochowneuro@163.com

Department of Neurosurgery, The affiliated Changzhou Second People’s Hospital of Nanjing Medical University, Changzhou, Jiangsu, China

*These authors have contributed equally to this work and share first authorship.

+ Supplemental data for this article can be accessed online at https://doi.org/10.1080/01616412.2023.2287340

been achieved. As a result, multiple novel approaches to glioma treatment have been proposed over the past sev- eral decades. For example, Chinese patent medicine ther- apy, electromagnetic field therapy, immunotherapy and other treatments have been studied and could potentially be supplemented to traditional treatment methods (including surgery, radiotherapy and chemotherapy) [3-6]. Immunotherapy is an available treatment method for some solid tumors, but diverse clinical studies have demonstrated that its therapeutic efficacy in glioma is limited [7,8]. The genesis of a poor response to immu- notherapy probably lurks in disorganized blood vessels in the glioma microenvironment. Therefore, immunother- apy in glioma should be modified to achieve reasonable therapeutic efficacy. Consequently, exploring new bio- markers with huge potential in affecting the prognosis of glioma for immunotherapy has sparked a heated controversy.

The apolipoprotein B mRNA editing catalytic poly- peptide-like (APOBEC) family of proteins contains 11 members that can be classified into four families, and play a significant role in innate antiviral immunity [9]. APOBEC3 (A3) is one of the four families and contains 7 types (A3A, A3B, A3C, A3DE, A3F, A3G and A3H). Its encoding genes are located on chromosome 22 [10,11]. A3C is a pervasive research target of HIV-1 and HBV according to the published studies [12-15]. In addition to its regulatory role in viral reverse transcription, A3C has also been found to be altered in malignant cancers [16,17]. A few studies have investigated the function of A3C in the development of glioma cells [18,19], but the function of A3C in glioma immunotherapy is rarely mentioned. In order to further explore the function of A3C, we first performed immunofluorescence in glioma tissue, and found that A3C located where CD31 was expressed. CD31 was identified as a cellmarker of T cell [20] or endothelial cell [21], which indicated that A3C may be involved in the immune response of glioma. This study attempts to uncover the mechanisms of A3C in affecting the prognosis of LGGs. We hypothesize that gene expression reprogramming following A3C expres- sion might modulate glioma cell growth.

2 Methods

2.1 Immunofluorescence

The LGG samples [22] were donated by the participants who underwent surgical tumor removal at the depart- ment of neurosurgery, the affiliated Changzhou Second People’s Hospital of Nanjing Medical University. Paraffin sections of glioma were dewaxed and tissue antigen repair was performed with citrate buffer (0.01 M citrate, 0.05% Tween 20, PH6.0) at 100℃ for 20 min. Three percent H2 O2 was used to eliminate endogenous peroxidase activity at room temperature for 10 min. The sections were blocked with blocking buffer (5%BSA,0.2% Triton

X-100 in PBS) for 1 hour at room temperature and incubated for 24 hours at 4℃ in a primary antibody diluted in blocking buffer. After that, sections were incu- bated in fluorescent secondary antibody diluted in block- ing buffer. Last, sections were mounted onto glass slides with an antifading mounting medium and DAPI (S2110, Solarbio, China). Antibodies used: Anti-A3C (ab221874, Abcam, U.S.A.), Anti-CD31 (ab9498, Abcam, U.S.A.), Goat Anti-Rabbit IgG (Alexa Fluor® 488) (ab150077, Abcam, U.S.A.), Goat Anti-Mouse IgG (Alexa Fluor® 647) (ab150115, Abcam, U.S.A.).

2.2 Prognostic effect of A3C in gliomas

For a preliminary understanding of the relationship between A3C and malignant cancers, the RNA-seq data of 33 types of cancers and the corresponding clinical information were downloaded from the UCSC Xena (https://xena.ucsc.edu/). A3C expres- sion in cancer tissues was primarily assessed by comparing it with normal tissues if the data was available. Then, A3C in gliomas was further ana- lyzed in The Cancer Genome Atlas (TCGA), which was identified as the internal group, while the data from CGGA was chosen as the external group. Only five normal brain samples were included in TCGA. For an accurate result, the 206 normal cortex tissue samples from Genotype-Tissue Expression (https://commonfund.nih.gov/GTEx/) were added to the differential gene expression ana- lysis. Finally, 211 normal samples, 529 LGGs sam- ples and 168 GBM samples were included in the boxplot to compare the differential expressions. Kaplan-Meier (KM) analysis was employed to eval- uate the relationship between A3C expression, overall survival (OS) and progression-free interval (PFI) in TCGA. Next, the predictive power was verified according to the prognosis prediction in glioma subtypes, the independence from prognostic factors (age, gender, WHO grade, histology, isoci- trate dehydrogenase (IDH) mutation status, and 1p/19 codeletion status) and the 1-, 3-, and 5-year time-dependent receiver operating characteristic (ROC) values.

A3C-related genes were identified by weighted gene co-expression network analysis (WGCNA), a powerful method to screen key modules and genes [23]. The main steps of WGCNA were described in the previous study [24]. The gene set most associated with A3C was regarded as the A3C module eigengene and was included for further analysis. The genes with gene significance (GS) >0.5 and module membership (MM) >0.8 in the eigengene module were identified as A3C- related genes.

2.4 Unsupervised clustering analysis and validation

Twenty-eight A3C-related genes were adopted for unsu- pervised clustering analysis based on the gene expression. A consensus clustering algorithm was used to confirm the number of clusters and determine the stability of the samples [25]. KM survival curves were plotted to com- pare the prognosis between different clusters. Moreover, the distribution of three kinds of data was described in clusters. For further verification, a heat map was plotted to describe the distribution of the three data types in different clusters. These three types of data include the distribution of immune cell (dendritic cells, macrophage cells, neutrophils, CD8 T cells, CD4 T cells and B cells) infiltration from the Tumor IMmune Estimation Resource (TIMER) database (https://cistrome.shi nyapps.io/timer/), the key molecular prognostic markers and clinical information of LGGs, and the differential expression of A3C-related genes.

2.5 Gene set variation analysis (GSVA) and single-sample gene-set enrichment analysis (ssGSEA)

GSVA was performed using the ‘GSVA’ package to explore the difference in biological processes between the clusters [26]. A curated gene set of ‘c2.cp.kegg.v7.2.symbols’ acquired from the GSEA database (https://www.gsea-msigdb.org/gsea/index. jsp) was used for GSVA analysis. Kyoto encyclo- pedia of genes and genomes (KEGG) was consid- ered significant if the adjusted P value was less than 0.05. The top 20 KEGG pathways from the cluster with the best prognosis and the cluster with the worst prognosis were plotted on the heat map. Twenty-three immune cells computed by reference gene sets were provided by the pre- vious study [27]. Immune cell infiltration, includ- ing activated B cells, activated CD4 T cells, activated CD8 T cells, activated dendritic cells and so on, was quantified by the relative abun- dance in glioma tumor microenvironment (TME) using the ssGSEA algorithm.

2.6 Validation of prognosis in clusters

Although the samples were assigned to different clus- ters, the difference in prognosis between clusters was still unexplored. Firstly, principal component analysis (PCA) according to the A3C-related genes expression was performed to assess the sample distribution. PCA 1 and PCA 2 were included to calculate a score that we defined as the A3C score [25]. Secondly, KM survival plot curves were used to explore the relationship between the A3C score and OS in TCGA and CGGA. In addition, the prognostic prediction inde- pendence of the A3C score was assessed using uni- variate and multivariate Cox proportional hazard

regression analyses. Thirdly, the relationship between A3C expression and A3C score was also investigated.

2.7 Immune cell infiltration and chemotherapeutic drug sensitivity

The difference in immune cell infiltration among the clusters based on the A3C-related gene expression was explored. The immune score, stromal score and com- bined score provided by Yoshihara et al [28] and Ceccarelli et al. [29] were visualized in clusters to portray the difference. Correlations between A3C expression and immune cell infiltration were analyzed by Pearson corre- lation. The same was performed to analyze the correla- tion between A3C expression and 20 immune checkpoints expression (CD274, hepatitis A virus cellular receptor 2 (HAVCR2), programmed cell death 1 (PDCD1), and so on). Chemotherapy is an essential anti- tumor treatment. However, drug resistance leads to treat- ment failure in malignant tumors. Therefore, the func- tion of the expression of A3C-related genes, expression of A3C, and A3C score in chemotherapeutic drug sensitivity were assessed with CellMiner (https://discover.nci.nih. gov/cellminer/home.do) [30,31].

2.8 Acquisition and processing of scRNA-seq of LGGs

For a further exploration of the function of A3C in LGGs, scRNA-seq data was analyzed using Seurat (version 4.1.1). The scRNA-seq data was extracted from Gene Expression Omnibus (GSE1387941, GSE117891). Harmony was the algorithms of scRNA data integration [32]. After cell filtering and quality control 22,871 cells with 27,568 genes were included. The RunPCA function was used to project the cells into principle components in Seurat, and then RunUMAP function was included for a 2-dimensional visualization [33]. The most variable 2000 genes were selected for achieving 15 clusters with a resolution of 0.4. Cells were assigned into 11 major cell clusters according to the cell type-specific markers: Oligodendrocyte (‘ERMN’, ‘GRIA2’) [34,35], Astrocyte (‘AQP4’, ‘CHI3LI’) [36,37], Macrophage (‘TREM1’, ‘C1QC’) [38,39], Neuro (‘TOP2A’) [40], Neural progenitor cell (‘PBK’) [41], Cancer cell (‘RSPH1’) [39], Conventional dendritic cell 2b (‘S100A9’) [42], T cell (‘IL32’, ‘CD2’) [43,44], B cell (‘MZB1’) [45], Endothelial cell (‘CLDN5’) [46], and one cluster failed to group in any cell type.

2.9 Statistical analysis

The R (version 4.1.2) was used for statistical ana- lyses. Survival rates were calculated using KM curve with a log-rank test. Differentially expressed genes (DEGs) were screened by Wilcox test.

Figure 1. Arrows show double staining for A3C (green) and CD31 (red) in glioma tissue.

DAPI

A3C

CD31

merge

Figure 2. The flowchart of data processing of comprehensive biological analysis.

The expression of A3C in cancers and corresponding tissues

Prognostic role of A3C in glioma and subtypes, verified in CGGA

A3C is prognostic marker in LGG

Screening A3C related genes using WGCNA

28 A3C related genes, GS>0.5, |MM|>0.8

Unsupervised clustering analysis and validation

Three clusters

Assess the difference between clusters

Prognostic role: Survival analysis

Gene function enrichment analysis: DEGs and GSVA

Immune assessment: Immune cells infiltration, immune score, stromal score, combined score

Assess the distribution of samples: PCA

Calculate A3C score according PCA 1 and PCA 2

Prognostic role: Survival analysis, independence assessment

Correlation assessment, Chemotherapeutic drug sensitivity

scRNA analysis

Prognostic predictors were identified by univariate and multivariate Cox regression analysis. Then, (receiver operating characteristic) ROC curves

were adopted for assessing the predictive accuracy. P value < 0.05 was considered statistically significant.

Figure 3. (a) the RNA-seq transcription data of A3C in cancers from TCGA. (b) Scatter plots of A3C in LGG, GBM and corresponding normal tissues. (c-j) survival curves of A3C expression in glioma from TCGA and CGGA. OS in (c) glioma, (d) GBM, PFI in (e) GBM, (f) OS and (g) PFI in LGG from TCGA. OS in (h) glioma, OS in (i) GBM and (j) LGG from CGGA.

a

Type E Normal Tumor

**


*












**

8

6

O

A3C expression

4

2

0

1

+

ACC

BLCA

BRCA

CESC

CHOL

COAD

DLBC

ESCA

GBM

HNSC

KICH

KIRC

LAML

LGG

LIHC

LUAD

LUSC

MESO

OV

PAAD

PCPG

PRAD

READ

SARC

SKCM

STAD

TGCT

THCA

THYM

UCEC

UCS

UVM

b

Type Normal @LGG @GBM

C

high

low

p<0.001

d

high

low

p=0.055

1.00

1.001



6

0.75

0.75

A3C expression

0.50

0.50

4.

8

3 Ő

0.25

0.25

2

0.00

0.00

0

2

4

6

8

Time(years)

10

12

14

16

18

20

0

2

Time (years)

4

6

8

high

345 101 35

19

10

5

2

1

0

0

0

high- 83

10

1

0

0

0

low

346 178 63

38

17

14

7

4

1

0

0

low.

84

15

2

1

0

Normal

LGG

GBM

0

2

4

6

Time (years)

8

10

12

14

16

18

20

0

2

Time(years)

4

6

8

e

high

low

p=0.012

f

high

low

p<0.001

g

high

- low

p<0.001

1.00

1.00-

1.001

0.75

0.75

0.75

旺 0.50

0.50-

PF

0.50

8

0.25

0.25

0.25-

p<0.001

0.00

0.00

0.00

0

1

Time (years)

2

3

4

0

2

4

6

8

Time (years)

10

12

14

16

18

20

0

2

4

6

Time (years)

8

10

12

14

high- 83

14

5

2

high-

262 116 47

30

17

10

4

3

1

0

0

low

0

83

25

7

2

1

low-

262

138

48

26

10

9

5

2

0

0

0

high.

low

261

262

79

112

28

29

16

8

Unico

NOT

0

14

1

0

1

0

1

Time (years)

2

3

4

0

2

4

6

8

10

12

14

16

18

0

2

4

Time (years)

8

Time (years)

20

6

10

12

14

h

1.00-

high

low

p<0.001

i

1.00-

+

high

= low

p=0.111

j

1.00-

high

low

p<0.001

0.75

0.75

0.75

8 0.50

00

0.50

8 0.50

0.25.

0.25

0.25.

0.00

0.00

0.00

0

2

4

Time (years)

6

8

10

12

0

2

4

Time (years)

6

8

10

0

2

4

Time (years)

6

8

10

12

high-462

179

87

45

14

3

low

0

462

306

179

111

22

0

high

173

32

10

4

1

0

high-

285

187

104

50

18

5

66

low

174

45

15

9

2

1

low

0

288

220

136

92

59

19

0

0

2

4

Time(years)

6

8

10

12

0

2

4

Time (years)

6

8

10

0

2

4

Time (years)

6

8

10

12

3 Results

3.1 The expression of A3C in cancers and prognostic role in glioma

The immunofluorescence was double stained for A3C and CD31, showing extensive colocalization (Figure 1). Figure 2 shows the comprehensive biological analysis flowchart. In TCGA, 23 kinds of cancers corresponded

to normal tissue RNA-seq data, among which 13 can- cers had a higher expression of A3C than normal tis- sues, and only two cancers had a lower expression in tumor tissues (Figure 3(a)). The expression of A3C in LGG samples is higher than in normal samples, but it is lower than in GBM (Figure 3(b)). The Survival curve indicated that A3C is a poor biomarker for glioma (Figure 3(c)), and A3C has little effect on OS and

Figure 4. (a) univariate and multivariate Cox proportional hazard regression analyses in univariate and (b) multivariate Cox proportional hazard regression analyses in different subtypes. In (c) univariate and (d) multivariate Cox analyses from TCGA, A3C was identified as an independent OS predictor. (e) the time-dependent ROC curves for the 1-, 3- and 5-year survival rates in TCGA. In CGGA, (f) univariate and (g) multivariate Cox analyses verified that A3C is an independent prognosis predictor and the corresponding (h) time-dependent ROC curves for the 1-, 3- and 5-year survival rates.

a

b

pvalue

Hazard ratio

pvalue

Hazard ratio

Glioma

<0.001

2.430(2.175-2.716)

Glioma

<0.001

1.239(1.159-1.326)

GBM

0.059

1.226(0.992-1.514)

GBM

0.047

1.098(1.001-1.203)

LGG

<0.001

2.299(1.907-2.770)

LGG

<0.001

1.306(1.181-1.445)

IDH Mutation type

<0.001

1.807(1.357-2.406)

IDH Mutation type

0.113

1.092(0.979-1.217)

IDH Wild type

<0.001

1.454(1.252-1.690)

IDH Wild type

0.001

1.160(1.059-1.272)

1p/19q

Codeletion

0.002

2.265(1.350-3.802)

1p/19q

Codeletion

0.171

0.845(0.663-1.076)

1p/19q

Non-codeletion

<0.001 2.286(2.019-2.589)

1p/19q

Non-codeletion

0.004

1.124(1.037-1.218)

1.0

2.0

4.0

0.71

1.0

1.41

Hazard ratio

Hazard ratio

c

d

Hazard ratio

e

pvalue

pvalue

Hazard ratio

1.0

Age

<0.001 1.057(1.040-1.073)

<0.001

1.045(1.026-1.064)

GenderM

0.700 1.079(0.732-1.591)

0.315 1.232(0.820-1.852)

True positive rate

0.8

Grade !!

<0.001 3.173(2.066-4.875)

0.001

2.150(1.351-3.422)

0.6

Histology

0.005 1.378(1.101-1.725)

+

0.175 0.819(0.614-1.093)

0.4

IDH Wild type

<0.001 6.647(4.435-9.962)

<0.001

2.691(1.565-4.629)

1p/19q Non-codeletion

<0.001 2.625(1.595-4.319)

0.044

1.933(1.019-3.666)

0.2

5- year (AUC=0.702)

0.0

3- year (AUC=0.762)

A3C

<0.001 2.345(1.908-2.882)

-

<0.001

1.787(1.366-2.338)

1- year (AUC=0.826)

0

2

4

6

8

0

1

2

3

4

0.0

0.2

0.4

0.6

0.8

1.0

Hazard ratio

Hazard ratio

False positive rate

f

pvalue

Hazard ratio

g

pvalue

Hazard ratio

h

1.0

Age

0.005 1.018(1.005-1.032)

0.011

1.015(1.003-1.028)

Gender M

0.916 0.986(0.765-1.271)

0.980

1.003(0.773-1.302)

True positive rate

0.8

Grade |

<0.0013.143(2.387-4.140)

<0.001

2.921(2.204-3.872)

0.6

Histology

<0.0011.838(1.564-2.161)

0.099

1.285(0.953-1.733)

0.4

IDH Wild type

<0.0012.183(1.666-2.860)

0.078

1.309(0.970-1.767)

1p/19q Non-codeletion

<0.0013.620(2.573-5.093)

0.041

1.939(1.026-3.662)

0.2

5- year (AUC=0.676)

<0.0011.604(1.408-1.826)

0.0

3- year (AUC=0.661)

A3C

+

0.003

1.245(1.075-1.443)

1- year (AUC=0.629)

0

1

2

3

4

5

0

1

2

3

0.0

0.2

0.4

0.6

0.8

1.0

Hazard ratio

Hazard ratio

False positive rate

progression-free interval in GBM (Figure 3(d,e)). However, A3C plays a prognostic role in LGGs (Figure 3(f,g)). Likewise, a similar function was vali- dated in CGGA (Figure 3(h-j)). Moreover, univariate and multivariate Cox proportional hazard regression analyses (Figure 4(a,b)) indicated that the prognostic role of A3C is stable in LGGs.

3.2 Independent prognostic role of the expression of A3C

According to the training cohort, A3C is identified as an independent prognostic gene from age, gen- der, WHO grade, histology, IDH mutation and 1p/ 19q codeletion status (Figure 4(c,d)). The area

under ROC curve value of A3C in predicting the 1-, 3-, and 5-year OS were 0.826, 0.762, and 0.702, respectively, which implied a mild prognostic accu- racy in the training cohort (Figure 4(e)). In the testing cohorts, A3C was also identified as an inde- pendent predictor and had valuable area under ROC curve (AUC) values (Figure 4(f-h)).

After excluding the samples with incomplete clinical information, 424 LGG samples from TCGA were left. Subsequently, 3047 genes were identified as variant genes with variance was greater than the quartiles and were included in the co-expression network. All

Figure 5. (a) cluster Dendrogram analysis, and (b) eigengene adjacency heatmap showing the correlation between A3C and the other modules. (c) the MM vs. GS plot for the green-yellow module showed that MM and GS are highly correlated (Cor = 0.86, p < 0.001). (d) Consensus matrix heatmap for k = 3. (e) survival curves of three clusters.

a

Cluster Dendrogram

b

Eigengene adjacency heatmap

c

MM vs. GS

1.0

1

0.8

Cor=0.86, p<0.001

0.8

Height

0.8

0.6

0.6

A3C

0.6

GS for OS

y.

0

0.4

0.4

2

0

5

0

0.2

Dynamic Tree Cut

0.0

0

Merged

dynamic

1

0.2

0.4

0.6

0.8

A3C

MM in greenyellow module

d

consensus matrix k=3

e

1.00-

p<0.001

cluster

1

WN 3

A

0.75

B

Survival probability

C

0.50

0.25

0.00

I

0

2

4

6

8

10

12

14

16

18

20

Time (years)

Number at risk

cluster

A

175

60

28

15

6

4

2

0

0

0

0

B

155

62

26

17

8

6

4

1

0

0

0

C

94

25

14

7

3

2

1

1

1

0

0

0

2

4

6

8

10

12

14

16

18

20

Time (years)

the samples were clustered by Pearson’s correlation coefficient. Finally, 402 LGG samples were selected in a targeted cluster by removing the outliers. The soft threshold value to construct a scale-free network was determined to be four (R2 = 0.94, slope= - 1.4). Dynamic tree clipping was used to screen out 10 modules, and the green-yellow module, having a closer relationship with A3C than other modules, was considered biologically significant (Figure 5(a,b)). The correlation coefficient between MM and GS was 0.86 (Figure 5(c)). Twenty-eight genes that had a | MM| > 0.8 and a p <0.001 were deemed as A3C- related genes, which were prognostic genes based on the mortality analyses conducted using Cox propor- tional hazard ratios and KM analyses.

3.4 Sample cluster and survival analysis

424 samples were classified into three modification patterns based on the expression of 28 A3C-related genes by unsupervised clustering (Figure 5(d)). Three patterns were also termed clusters A-C, including 175

samples in cluster A, 155 samples in cluster B, and 94 samples in cluster C. Survival analyses revealed sig- nificant differences between the clusters. Cluster B had the best prognosis with a median survival time of about 12 years, while cluster C demonstrated the worst prognosis with a median survival time of close to 4 years (Figure 5(e)).

A3C is known for its immunomodulatory function in HIV-1. However, little is known about whether A3C has an equal immunomodulatory effect in malignant tumors. The infiltration of six types of immune cells, including dendritic, macrophage, neutrophil, CD8 T cell, CD4 T cell and B cell, presents a distinct character in the three clusters, especially in cluster B and cluster C. Likewise, several prognosis-related clinical features, such as 1p19q codeletion, IDH mutation and WHO grade, varied across the three clusters, especially in

234

Q

Dendritic

Dendritic

1

Macrophage

b

Neutrophil

CD8 T cell

0

S. ZHAO ET AL.

4

CD4 T cell

-1

0

1

B cell

Macrophage

-2

Cluster

KEGG VIRAL MYOCARDITIS

APOBEC3C

1

0.4

1p19q Codeletioon

IDH Mutation

C

B

-0.6

WHO Grade

0

Age

Cluster

Neutrophil

1

KEGG VIRAL MYOCARDITIS

ALOX5AP

ARHGDIB

0

ARPC1B

CD8 T cell

1

NATURAL_KILLER_CELL_MEDIATED_CYTOTOXICITY

LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION JAK_STAT_SIGNALING_PATHWAY

B2M

0

C1QA

CD4 T cell

TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY

APOPTOSIS

Cluster

C1QC

1

C3

B_CELL_RECEPTOR_SIGNALING_PATHWAY

0

CD53

B cell

1

COMPLEMENT_AND_COAGULATION_CASCADES

CYTOSOLIC_DNA_SENSING_PATHWAY

CD74

CLIC1

0

CTSA

APOBEC3C

CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION

1

CYBA

HEMATOPOIETIC_CELL_LINEAGE

FCER1G

0

GMFG

1p19q Codeletion

1

TYPE ___ DIABETES_MELLITUS

HAVCR2

HCLS1

0

GRAFT_VERSUS_HOST_DISEASE

HLA-DRA

IDH Mutation

1

AUTOIMMUNE_THYROID_DISEASE

ALLOGRAFT_REJECTION

HLA-E

4

LAPTM5

0

WHO Grade

2

INTESTINAL_IMMUNE_NETWORK_FOR_IGA_PRODUCTION

NPC2

2

LEISHMANIA_INFECTION

OLFML3

SYSTEMIC_LUPUS_ERYTHEMATOSUS

S100A11

1

Age

SASH3

0

1

ANTIGEN_PROCESSING_AND_PRESENTATION

ASTHMA

SERPINA1

0

-2

PRIMARY_IMMUNODEFICIENCY

SLC39A1

Cluster

SPI1

ImU A

TMEM109

C

-4

TMSB4X

GSVA items were displayed according to clusters B and C.

T cell, B cell) infiltrations, and distribution of 1p/19q codeletion, IDH mutation, WHO grade and age in three clusters. (b) the top 20

Figure 6. (a) the expression of A3C and 28 A3C-related genes, immune cells (dendritic, macrophage, neutrophil, CD8 T cell, CD4

cluster B and cluster C. The 28 A3C-related genes had a higher expression in cluster C, indicating a poorer prognosis than the other clusters (Figure 6(a)). Notably, cluster B was associated with the best prognosis, while cluster C indicated the worst prognosis. DEGs between clusters B and C were further analyzed by GSVA for functional biological annotation. The top 20 GSVA items were visualized, and the result implies that many immune-related pathways, including natural killer cell- mediated cytotoxicity, B cell receptor signaling pathway, primary immunodeficiency and so on, occurred due to the gene expression difference between the two clusters (Figure 6(b)). According to the GSEA results put forward

by Charoentong et al. [27], the infiltration of 23 subpo- pulations of immune cells related to adaptive immunity and innate immunity was assessed in three clusters. The results implied that 22 immune cell subpopulations have a different infiltration in different clusters, except for type 2 T-helper cells (Figure 7(a)). PCA in Figure 7(b) revealed three principal components based on the dis- tribution of the samples in three clusters. The three groups were combined in pairs, and then the DEGs were screened. Finally, a total of 167 genes with inter- secting differential expressions were obtained (Figure 7(c)). Patients with higher A3C-related gene scores demonstrated a markedly prolonged survival

Figure 7. (a) immune cell infiltration in three clusters. (b) the PCAs of samples from three clusters. (c) Venn diagram describing the numbers of inserting genes. The survival curves of A3C score in LGG from (d) TCGA and (e) CGGA. Multivariate Cox analyses showed that the A3C score is an independent prognosis predictor in (f) TCGA and (g) CGGA.

a

Cluster

A

B

C

1.00

ns

Immune infiltration

0.75

0.50

0.25

0.00

Activated B cell

Activated CD4 T cell

Activated CD8 T cell

Activated dendritic cell’

CD56bright natural killer cell’

CD56dim natural killer cell

Eosinophil

Gamma delta T cell’

Immature B cell’

Immature dendritic cell

MDSC

Macrophage

Mast cell’

Monocyte

Natural killer T cell’

Natural killer cell

Neutrophil

Plasmacytoid dendritic cell’

Regulatory T cell

T follicular helper cell’

Type 1 T helper cell ’

Type 17 T helper cell’

Type 2 T helper cell

b

c

d

p<0.001

APOBEC3C score

1.004

Low

4

Cluster

A

Survival probability

High

B

C

B-A

1022

C-A

0.75-

2

0.50-

PC2

17

250

167

336

OD

13

0.25-

0

APOBEC3C score

0.00

0

2

4

6

8

10

12

14

16

18

20

-2

Number at risk

Time (years)

C-B

Low-

249

84

40

23

9

6

3

1

1

0

0

-10

-5

0

5

10

High-

175

63

28

16

8

6

4

1

0

0

0

PC1

0

2

4

6

8

10

12

Time (years)

14

16

18

20

e

f

g

1.00-

p<0.001

APOBEC3C score

Survival probability

Low

pvalue

Hazard ratio

pvalue

Hazard ratio

0.75

High

Age

<0.001

1.089(1.063-1.115)

0.009

1.016(1.004-1.028)

0.50

Gender M

0.564

0.863(0.524-1.422)

0.705

1.051(0.812-1.361)

0.25

Grade !!!

0.002 2.453(1.383-4.352)

<0.001 2.942(2.224-3.893)

IDH Wild type

0.054

1.910(0.988-3.693)

0.017

1.418(1.065-1.888)

APOBEC3C Score

0.00

0

2

4

6

8

10

12

14

1p/19q

Non-codeletion

0.302

1.460(0.712-2.996)

<0.001 2.306(1.550-3.432)

Number at risk

Time (years)

Low{ 247

143

87

44

21

9

2

0

APOBEC3C score Low <0.001 3.317(1.634-6.734)

.

<0.001 2.034(1.499-2.759)

High

264

229

171

127

79

47

10

0

0

2

4

6

Time (years)

8

10

12

14

0

1

1234

5

6

0

1

2

3

Hazard ratio

Hazard ratio

time compared to those with lower scores (Figure 7(d,e)). Moreover, the A3C score was identified as an indepen- dent factor of clinical characteristics and molecular fea- tures (Figure 7(f,g)). High A3C expression samples were more likely to have a higher percent weight of low A3C score than that of low A3C expression samples (83% vs 35%) (Figure 8(a)). Samples in cluster C accompanied with the worst outcome were almost assigned to the high A3C group and got an A3C score (Figure 8(b,c). In Figure 8(d-f), the A3C expression, immune score, stro- mal score and combined score in cluster A, cluster B and cluster C were significantly different. The correlation

between A3C score and immune cell infiltration was equivalent to the correlation between A3C and immune cell infiltration in Figure 9. The A3C score revealed a strong co-expression relationship with nature killer T cell, natural killer cell, myeloid-derived suppressor cell (MDSC), immature B cell, activated dendritic cell, gamma delta T cell (cor >0.7). As shown in Figure 10, the expression of A3C had a significant positive correlation with the A3C score (Cor = 0.77). There was a positive co- expression relationship between A3C and immune checkpoints, and the correlation values with CD86, HAVCR2, PDCD1 ligand 2 (PDCD1LG2), and protein

Figure 8. (a) relationship between A3C and the A3C score. (b) Sankey distributional diagram showed the relationship in clusters, A3C, A3C score, and survival status. (c) immune score, (d) stromal score and (e) combined scores in three clusters.

a

b

C

100

APOBEC3C Score

Cluster A BEC

17%

High

60

p < 0.001

Low

High

High

Alive

p < 0.001

75

I

Percent weight

APOBEC3C Expression

p < 0.001

65%

40

50

83%

3

20

25

35%

Low

Low

Dead

0

0

0

high

low

A

B

C

APOBEC3C

Cluster

APOBEC3C

APOBEC3C score

Survival status

Cluster

d

e

f

Cluster BABBEC

Cluster BABBEC

Cluster 中A 中B 中C

p < 0.001

p < 0.001

p < 0.001

p < 0.001

5000.

p < 0.001

10000

p < 0.001

4000

p < 0.001

4000-

p < 0.001

p < 0.001

Immune Score

Stromal Score

Combined Score

7500

3000.

5000

2000

2000-

2500

1000-

0

0

0

A

B

C

A

B

C

A

B

C

Cluster

Cluster

Cluster

cellCD4 T cellCD8 T celldendritic cellkiller cellcellcellcelldendritic cellT cellcelldendritic cellcellcellcellcellcell
naturalnatural killer
Thelperhelper
A3C scoreA3CActivated BActivatedActivatedActivatedCD56 brightCD56 dimEosinophilGamma deltaImmature BImmatureMDSCMacrophageMast cellMonocyteNatural killerNatural killerNeutrophilPlasmacytoidRegulatory TT follicularType 1 T helperhelper
Type 17 TType 2 T
A3C score1
A3C-0.77
Activated B cell-0.220.310.8
Activated CD4 T cell-0.370.37-0.03
Activated CD8 T cell-0.610.560.230.27
Activated dendritic cell-0.730.740.350.330.550.6
CD56 bright natural killer cell-0.390.140.090.110.430.08
CD56 dim natural killer cell0.33-0.18-0.1-0.21-0.24-0.11-0.260.4
Eosinophil-0.270.190.30.320.20.330.16-0.16
Gamma delta T cell-0.740.69-0.020.510.550.630.18-0.070.090.2
Immature B cell-0.770.780.540.330.620.760.24-0.290.410.56
Immature dendritic cell-0.580.590.460.150.420.670.23-0.090.430.340.68
MDSC-0.780.740.450.420.640.820.3-0.20.480.60.840.70
Macrophage-0.590.630.560.270.590.750.21-0.160.430.420.770.680.86
Mast cell-0.620.590.390.420.50.750.13-0.170.630.430.70.650.820.74-0.2
Monocyte0.200.18-0.150.080.02-0.180.18-0.1-0.1-0.02-0.010.020.20.04
Natural killer T cell-0.880.770.270.40.50.770.22-0.180.290.740.760.620.770.60.67-0.12
Natural killer cell-0.820.790.480.320.520.80.19-0.210.40.580.850.680.850.740.780.010.86-0.4
Neutrophil-0.260.220.170.060.310.190.25-0.070.240.210.250.230.390.290.280.040.20.28
Plasmacytoid dendritic cell-0.550.490.110.060.320.570-0.040.030.450.460.350.430.30.44-0.030.570.550.23-0.6
Regulatory T cell-0.610.660.40.420.590.70.21-0.210.470.50.740.620.830.780.690.060.590.70.40.33
T follicular helper cell-0.60.650.550.320.490.750.18-0.170.50.380.790.70.840.830.780.120.660.810.270.380.79
Type 1 T helper cell-0.660.650.380.230.390.80.04-0.010.430.510.690.710.730.660.74-0.060.770.80.210.510.570.71-0.8
Type 17 T helper cell-0.450.490.470.060.350.60.07-0.060.270.30.620.550.590.60.520.120.540.630.30.470.490.60.59
Type 2 T helper cell-0.110.17-0.10.54-0.060-0.09-0.23-0.030.290.07-0.11-0.02-0.13-0.05-0.150.160.09-0.12-0.050.07-0.02-0.01-0.09-1

Figure 9. A3C, as well as A3C scores, have close correlation with the infiltration of immune cells, such as T cells, B cells, and macrophages.

Figure 10. The correlation between A3C, A3C scores and clarified immune checkpoints.
A3C scoreA3CB2MCD274CD40CD48CD86CD8ACTLA4HAVCR2IL23ALDHALDHBLGALS9PDCD1PDCD1LG2PTPRCTNFRSF18TNFRSF4TNFSF9VTCN1YTHDF11
A3c score
A3C-0.77
B2M-0.700.680.8
CD274-0.540.620.53
CD40-0.630.630.580.480.6
CD48-0.530.600.680.470.57
CD86-0.710.710.640.510.650.530.4
CD8A-0.230.360.330.480.410.510.28
CTLA4-0.400.430.470.370.430.500.420.450.2
HAVCR2-0.740.710.640.500.680.500.950.250.35
IL23A-0.330.260.350.190.280.320.190.150.280.220
LDHA-0.380.390.260.430.420.210.300.470.190.330.10
LDHB0.28-0.34-0.22-0.42-0.28-0.21-0.17-0.45-0.27-0.23-0.20-0.38
LGALS9-0.720.690.670.450.690.570.850.220.330.910.330.26-0.21-0.2
PDCD1-0.450.450.490.430.460.530.490.500.440.460.330.30-0.330.43
PDCD1LG2-0.760.780.660.670.610.550.730.420.500.710.240.42-0.360.640.42-0.4
PTPRC-0.680.710.630.580.640.560.900.380.490.860.150.35-0.300.720.440.78
TNFRSF180.000.140.120.360.230.190.020.550.390.000.190.38-0.360.020.270.160.06-0.6
TNFRSF4-0.300.280.340.210.440.310.090.260.330.100.310.25-0.160.210.350.170.070.44
TNFSF90.31-0.20-0.17-0.02-0.02-0.1-0.160.320.04-0.17-0.010.21-0.20-0.180.12-0.20-0.180.430.08-0.8
VTCN10.02-0.09-0.12-0.18-0.14-0.10-0.06-0.23-0.06-0.060.04-0.150.10-0.07-0.12-0.14-0.08-0.10-0.05-0.19
YTHDF1-0.270.230.250.08-0.020.130.01-0.120.120.050.21-0.02-0.150.070.160.120.06-0.040.23-0.300.22-1

tyrosine phosphatase receptor type C (PTPRC) were all greater than 0.7.

3.6 Chemotherapeutics sensitivity

There were 28 A3C-related genes, 20 of which were correlated with OS. The 20 A3C-related genes demon- strated a close relationship between their expression levels and corresponding drug sensitivity. The gene- drug relationship was described using correlation sig- nificance. The 16 top-ranking results are presented in Figure 11(a). The majority of the relationships could be found in S3. Moreover, the high A3C score samples were most probably sensitive to half maximal inhibi- tory concentration (IC50) of pyrimethamine, NVP. BEZ235 and rapamycin (Figure 11(b-d)).

3.7 ScRNA analysis and Immunofluorescence

After quality control and normalization 22,871 cells containing 27,568 genes were screened for analysis. Finally, the cells were projected onto two dimensions

and clustered into 15 clusters by uniform manifold approximation and projection (UMAP). The 15 clus- ters were further divided into 11 groups according to marker genes, representing Neuro, Endothelial cell, B cell, T cell, Conventional dendritic cell 2b, Cancer cell, Neural progenitor cell, Macrophage, Astrocyte, Oligodendrocyte and Unknown cluster (UMAP, Figure 12(a)). A3C mainly expressed on B cells, T cells and macrophages, and its average expression on T cells is the highest (Figure 12(b)).

4 Discussion

The range of OS of LGGs is much broader than HGG. In general, LGGs eventually progress to HGGs and prolonging the glioma at the lower-grade state may be an effective strategy. Exploring the genetic heteroge- neity of LGGs provides an important practical insight. However, the various treatments currently in use for LGGs are unsatisfactory, although some achievements have been made. In the last decade, the immunother- apy has become a novel approach to cancer treatment

Figure 11. (a) the top 16 gene-drug relationships. A3C score was most probably sensitive to IC50 of (b) pyrimethamine, (c) NVP. BEZ235 and (d) rapamycin.

a

SASH3, Nelarabine

AIF1, Chelerythrine

Drug sensitivity score

Cor=0.923, p<0.001

Drug sensitivity score

CHI3L2, Nelarabine Cor=0.885, p<0.001

SIGLEC10, Nelarabine

Drug sensitivity score

Cor=0.872, p<0.001

Drug sensitivity score

Cor=0.811, p<0.001

·

.

.

6

6

6

2.

4.

4.

4

.

2

2

2

0

·

0

.

0

·

0

2.

.

.

0

2

4

6

0

2

4

Gene Expression

Gene Expression

6

0.0

Gene Expression

0.5

1.0

1.5

2.0

0

2

4

6

8

Gene Expression

PTGS1, Epothilone B

AIF1, Nelarabine

LST1, Chelerythrine

Drug sensitivity score

Drug sensitivity score

Drug sensitivity score

Drug sensitivity score

CD300A, Fluphenazine Cor=0.783, p<0.001

Cor =- 0.800, p<0.001

Cor=0.791, p<0.001

Cor=0.788, p<0.001

0

·

4

6-

·

6

·

2.

4

2

·

4.

4

2

0

2.

6

.

0

2

.

·

0-

-

0

1

Gene Expression

2

3

0

2

4

6

8

0

2

4

0

1

Gene Expression

Gene Expression

2

Gene Expression

3

SASH3, Chelerythrine

SPI1, Imexon

Drug sensitivity score

Drug sensitivity score

HAVCR2, Fluphenazine Cor=0.779, p<0.001

Drug sensitivity score

Drug sensitivity score

NCKAP1L, Imexon Cor=0.764, p<0.001

Cor=0.779, p<0.001

Cor=0.768, p<0.001

4

·

6

4

4.

·

2

·

4

3

3.

2

2

0

:

.

2

1

1

2

+

0

:

O

.

0

J

·

·

1

A

0

2

4

6

0

1

2

3

0

2

4

0

1

2

3

Gene Expression

Gene Expression

Gene Expression

Gene Expression

HAVCR2, Isotretinoin

HCST, Chelerythrine Cor=0.757, p<0.001

CD86, Isotretinoin

Drug sensitivity score

Drug sensitivity score

Drug sensitivity score

Drug sensitivity score

CD4, Chelerythrine Cor=0.739, p<0.001

Cor=0.763, p<0.001

Cor=0.752, p<0.001

.

.

·

·

4

·

2

4

2.

2

·

0

2

0

0-

.

2

0

{

2

0

Gene Expression

1

2

3

0

Gene Expression

2

4

0.0

0.5

Gene Expression

1.0

1.5

0

1

Gene Expression

2

3

4

5

b

A3C score

Low

High

c

AC3C score

Low

High

d

A3C score

Low

High

p < 0.001

-1.5-

p < 0.001

p < 0.001

Pyrimethamine senstivity (IC50)

6-

NVP.BEZ235 senstivity (IC50)

Rapamycin senstivity (IC50)

1-

5-

-2.0-

0-

4.

-2.5-

-1-

·

3.

-3.0

·

:

i

·

-2-

2.

:

-3.5.

-3-

·

Low

High

Low

High

Low

High

[47,48]. The ultimate goal for immunotherapy is to adjust the TME to improve clinical outcomes. Potential biomarkers of response probably act as future directions for glioma immune-oncology. The current study provided a molecular target for a better therapeutic response to the combined management of glioma. Mainly based on comprehensive bioinfor- matic analyses, multiple levels of expression, survival rate, and biological functions were used to describe the effect of A3C in LGGs. Further detection revealed that A3C was highly correlated with immune response and expressed in immune cells. A3C is a reasonable pre- dictor of LGG prognosis, and it is a target biomarker that may assist in immunotherapy.

A3C has been studied in the treatment of HIV-1, which can escape from the immune system, and drugs targeting A3C may offer a novel management for

treating the virus [49]. Recently, studies reported that A3C plays a role in regulating the biological functions of cancer cells. Kanabe et al. [50] observed that the expression of A3C was lower in breast cancer samples compared with 40 paired normal samples. Upregulating A3C by lentivirus aggravated C - to - T mutational burden and fueled human pre-leukemia stem cell (pre-LSC) progress to acute myeloid leuke- mia stem cells [51]. In prostate cancer, Kawahara et al. [52] established a tissue proteome signature with 11 prostate-derived proteins, including A3C protein, for stratifying patients in clinical practice. A genome-wide association study of multiple myeloma, including 7319 multiple myeloma cases and 234,385 controls, found that A3C plays a role in defining multiple myeloma predisposition. The findings established novel views into the role of A3C in multiple myeloma biology [53].

a

b

Figure 12. (a) ScRNA analysis identified 15 subsets with cluster-specific genes and the subsets were annotated into 11 clusters representing different kinds of cells. They are annotated Neuro, Endothelial cell, B cell, T cell, Conventional dendritic cell 2b, cancer cell, Neural progenitor cell, macrophage, astrocyte, Oligodendrocyte and Unknown cell. (b) the expression of A3C could be found on almost all the cells, but mainly on B cells, T cells and macrophages. The A3C expression positive rate on B cells is higher than the others. And the average expression of A3C on T cells is the highest.

Neuro

10-

Endothelial cell

cell

B cell

Percent Expressed

5

B cell

Unknown

T cell

10

Macrophage

Oligodendrocyte

☒ 20

UMAP_2

Conventional dendritic cell 2b

☒ 30

0

Conventional dendritic cell 2b

☒ 40

Neuro

Identity

Unknown

Average Expression

Neural progenitor cell

Cancer cell

2.5

-5

2.0

1.5

Astrocyte

Neural progenitor cell

1.0

0.5

0.0

-10-

Macrophage

-0.5

Cancer cell

Astrocyte

-15-

Endothelial cell

Oligodendrocyte

-15-10 -5 0 5 10

UMAP_1

APOBEC3C

Features

The expression of A3C slid in gastric cancer when cytotoxin-associated gene A was overexpressed, trig- gering tumor formation [54]. In glioma, three studies reported constructing prognostic signatures using bioinformation, identifying A3C as an ingredient in the signature [18,19,55]. Indeed, A3C is a molecular biomarker brightly associated with cancer biology. However, few reports have explored the role or mechanism of A3C in affecting the outcome of LGGs.

A3C probably acts as an immune-related prognostic biomarker in LGGs TME. Little is known about the prognostic role of A3C in LGGs. The survival curves from this study revealed that a higher expression of A3C might improve the outcome of LGGs. LGG samples were stratified according to A3C-related genes from WGCNA, indicating a significantly different prognosis between these strata. Furthermore, the A3C score is an assessment tool calculated using DEGs from clusters; the tool indirectly proved the connection between A3C and disease prognosis. Moreover, chemotherapeutic sensitiv- ity results revealed that the A3C score was related to sensitivity to pyrimethamine, NVP.BEZ235 and rapamy- cin, improving the therapeutic outcome for LGGs. In addition to its prognostic role, A3C may also exert immu- nomodulatory functions. A previous study reported that A3C protein activation within immune cells aroused the anticancer immune response [16]. Wang et al. [55] estab- lished a prognostic signature in glioma using 15 immune- related genes, including A3C. Likewise, our multi-

dimensional biological information analysis results also indicated that A3C was associated with glioma immune response, confirming the prior studies. Interestingly, the scRNA analysis implied that A3C mainly expressed on immune cells, such as T cells, B cells and macrophages, which were identified as an essential component of TME of glioma. TME changes stem from insufficient intersti- tial perfusion, hypoxia, low pH, and increased interstitial fluid osmotic pressure, triggering the development of disorganized blood vessels [56-59]. The abnormal TME impairs the efficiency of immunotherapy and tumor vessel normalization. For example, antiangiogenic ther- apy may increase the potency of antitumor immune cells and improve the outcome [60-62]. The abnormal blood vessels, in turn, hinder the accumulation of T cells around the tumor tissue. The application of drugs targeting the A3C gene may reduce the aggregation of immune cells, then regulate the TME to influence the growth of glioma cells.

From the point of view of the mechanism of action, the immunotherapy has a good theoretical basis. This study provides another molecular biomarker that can be used to develop novel drugs to promote the effi- ciency of immunotherapy in LGGs. Although the sam- ple size is large enough, the bioinformatics analysis results should be further verified, and immunofluor- escence alone is insufficient. Furthermore, the A3C gene is mainly found in humans and primates. Therefore, in vivo experiments involve ethics and

high costs. In conclusion, the beneficial role of immu- notherapy in LGGs is a hotly debated topic. The dete- rioration of TME lowers the efficiency of immunotherapy. A3C was identified as a potential target to improve the efficiency of immunotherapy.

5 Conclusions

A3C is an immune-related prognostic biomarker in LGGs. Developing drugs to block A3C could enhance the efficiency of immunotherapy and improve disease survival.

Acknowledgments

We would like to acknowledge the researchers and patients who contributed to the TCGA, CGGA and the reviewers for their helpful comments to this paper.

Availability of data and materials

The data could be found from the UCSC Xena (https:// xena.ucsc.edu/), Genotype-Tissue Expression (https://com monfund.nih.gov/GTEx/), the TIMER database (https:// cistrome.shinyapps.io/timer/), the GSEA database (https://www.gsea-msigdb.org/gsea/index.jsp) and the CellMiner (https://discover.nci.nih.gov/cellminer/home. do).

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by Young Talent Development Plan of Changzhou Health Commission under Grant CZQM2021011; Major Scientific Project of Changzhou under Grant ZD202219. Changzhou Sci & Tech Program under Grant CJ20230072.

ORCID

Liang Shen D http://orcid.org/0000-0002-8342-4153

References

[1] Kiran M, Chatrath A, Tang X, et al. A prognostic signature for lower grade gliomas based on expression of long non-coding RNAs. Mol Neurobiol. 2019;56 (7):4786-4798. doi: 10.1007/s12035-018-1416-y

[2] Wang M, Li Q, Yu S, et al. Coronin 3 Promotes the development of oncogenic properties in glioma through the Wnt/B-Catenin signaling pathway. Onco Targets Ther. 2020;13:6661-6673. doi:10.2147/ OTT.S257001.

[3] Kim EH, Jo Y, Sai S, et al. Tumor-treating fields induce autophagy by blocking the Akt2/miR29b axis in glioblastoma cells. Oncogene. 2019;38 (39):6630-6646. doi: 10.1038/s41388-019-0882-7

[4] Vora P, Venugopal C, Salim SK, et al. The Rational development of CD133-targeting Immunotherapies for glioblastoma. Cell Stem Cell. 2020;26(6):832- 844.e6. doi: 10.1016/j.stem.2020.04.008

[5] Wick A, Desjardins A, Suarez C, et al. Phase 1b/2a study of galunisertib, a small molecule inhibitor of transforming growth factor-beta receptor I, in com- bination with standard temozolomide-based radio- chemotherapy in patients with newly diagnosed malignant glioma. Invest New Drugs. 2020;38 (5):1570-1579. doi: 10.1007/s10637-020-00910-9

[6] Gao Z-W, Huang Y-Y, Zhang J-Q, et al. Paeoniflorin elicits the anti-proliferative effects on glioma cell via targeting translocator protein 18 KDa. J Pharmacol Sci. 2021;145:115-121. doi: 10.1016/j.jphs.2020.10.004

[7] Reardon DA, Brandes AA, Omuro A, et al. Effect of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: The CheckMate 143 Phase 3 Randomized Clinical Trial. JAMA Oncol. 2020;6 (7):1003-1010. doi: 10.1001/jamaoncol.2020.1024

[8] McGranahan T, Therkelsen KE, Ahmad S, et al. Current state of immunotherapy for treatment of glioblastoma. Curr Treat Options Oncol. 2019;20 (3):24. doi: 10.1007/s11864-019-0619-4

[9] Constantin D, Dubuis G, Conde-Rubio MC D, et al. APOBEC3C, a nucleolar protein induced by genotox- ins, is excluded from DNA damage sites. FEBS J. 2022;289(3):808-831. doi: 10.1111/febs.16202

[10] Jarmuz A, Chester A, Bayliss J, et al. An anthropoid-specific locus of orphan C to U RNA-editing enzymes on chromosome 22. Genomics. 2002;79(3):285-296. doi: 10.1006/geno. 2002.6718

[11] Kim E-Y, Lorenzo-Redondo R, Little SJ, et al. Human APOBEC3 induced mutation of human immunode- ficiency virus type-1 contributes to adaptation and evolution in natural infection. PLOS Pathog. 2014;10 (7):e1004281. doi: 10.1371/journal.ppat.1004281

[12] Kitamura S, Ode H, Nakashima M, et al. The APOBEC3C crystal structure and the interface for HIV-1 Vif binding. Nat Struct Mol Biol. 2012;19 (10):1005-1010. doi: 10.1038/nsmb.2378

[13] McDonnell MM, Crawford KHD, Dingens AS, et al. APOBEC3C Tandem Domain Proteins Create Super Restriction Factors against HIV-1. MBio. 2020;11(2). doi: 10.1128/mBio.00737-20

[14] Anderson BD, Ikeda T, Moghadasi SA, et al. Natural APOBEC3C variants can elicit differential HIV-1 restriction activity. Retrovirology. 2018;15(1):78. doi: 10.1186/s12977-018-0459-5

[15] Bonvin M, Achermann F, Greeve I, et al. Interferon- inducible expression of APOBEC3 editing enzymes in human hepatocytes and inhibition of hepatitis B virus replication. Hepatology. 2006;43(6):1364-1374. doi: 10.1002/hep.21187

[16] Asaoka M, Patnaik SK, Ishikawa T, et al. Different members of the APOBEC3 family of DNA mutators have opposing associations with the landscape of breast cancer. Am J Cancer Res. 2021;11 (10):5111-5125.

[17] Zhang Y, Delahanty R, Guo X, et al. Integrative geno- mic analysis reveals functional diversification of APOBEC gene family in breast cancer. Hum Genomics. 2015;9(1):34. doi: 10.1186/s40246-015- 0056-9

[18] Zhang M, Wang X, Chen X, et al. Prognostic value of a stemness index-associated signature in

primary lower-grade glioma. Front Genet. 2020;11:441. doi: 10.3389/fgene.2020.00441

[19] Wang W, Li J, Lin F, et al. Expression and prognostic value of mRNAs in lower grade glioma with MGMT promoter methylated. J Clin Neurosci. 2020;75:45-51. doi: 10.1016/j.jocn.2020.03.037

[20] Pasciuto E, Burton OT, Roca CP, et al. Microglia require CD4 T cells to complete the fetal-to-adult transition. Cell. 2020;182(3):625-640.e24. doi: 10.1016/j.cell.2020. 06.026

[21] Governa V, Talbot H, Gonçalves de Oliveira K, et al. Landscape of surfaceome and endocytome in human glioma is divergent and depends on cellular spatial organization. Proc Natl Acad Sci U S A. 2022;119(9). doi: 10.1073/pnas.2114456119

[22] Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization classification of tumors of the central nervous System: a summary. Acta Neuropathol. 2016;131(6):803-820. doi: 10.1007/ s00401-016-1545-1

[23] Niemira M, Collin F, Szalkowska A, et al. Molecular signature of subtypes of non-small-cell lung cancer by large-scale transcriptional profiling: Identification of key modules and genes by weighted gene co-expression network analysis (WGCNA). Cancers (Basel). 2019;12(1):37. doi: 10.3390/cancers12010037

[24] Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinf. 2008;9(1):559. doi: 10.1186/1471-2105-9-559

[25] Zhang B, Wu Q, Li B, et al. m6A regulator-mediated methylation modification patterns and tumor micro- environment infiltration characterization in gastric cancer. Mol Cancer. 2020;19(1):53. doi: 10.1186/ s12943-020-01170-0

[26] Hänzelmann S, Castelo R, Guinney J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinf. 2013;14(1):7. doi: 10.1186/1471-2105-14-7

[27] Charoentong P, Finotello F, Angelova M, et al. Pan- cancer immunogenomic analyses reveal genotype-immunophenotype relationships and pre- dictors of response to checkpoint blockade. Cell Rep. 2017;18(1):248-262. doi: 10.1016/j.celrep.2016.12.019

[28] Yoshihara K, Shahmoradgoli M, Martínez E, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun. 2013;4(1):2612. doi: 10.1038/ncomms3612

[29] Ceccarelli M, Barthel FP, Malta TM, et al. Molecular profiling reveals biologically discrete subsets and pathways of progression in diffuse glioma. Cell. 2016;164(3):550-563. doi: 10.1016/j.cell.2015.12.028

[30] Reinhold WC, Sunshine M, Liu H, et al. CellMiner: a web-based suite of genomic and pharmacologic tools to explore transcript and drug patterns in the NCI-60 cell line set. Cancer Res. 2012;72(14):3499-3511. doi: 10.1158/0008-5472.CAN-12-1370

[31] Shankavaram UT, Varma S, Kane D, et al. CellMiner: a relational database and query tool for the NCI-60 cancer cell lines. BMC Genomics. 2009;10(1):277. doi: 10.1186/1471-2164-10-277

[32] Korsunsky I, Millard N, Fan J, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods. 2019;16(12):1289-1296. doi: 10.1038/ s41592-019-0619-0

[33] Guan J, Wang G, Wang J, et al. Chemical reprogram- ming of human somatic cells to pluripotent stem cells. Nature. 2022;605(7909):325-331. doi: 10.1038/ s41586-022-04593-5

[34] de Vlaminck K, Romão E, Puttemans J, et al. Imaging of glioblastoma tumor-associated myeloid cells using nanobodies targeting signal regulatory protein alpha. Front Immunol. 2021;12:777524. doi: 10.3389/fimmu. 2021.777524

[35] Huang S, Song Z, Zhang T, et al. Identification of immune cell infiltration and immune-related genes in the tumor microenvironment of Glioblastomas. Front Immunol. 2020;11:585034. doi: 10.3389/fimmu.2020. 585034

[36] Trevino AE, Sinnott-Armstrong N, Andersen J, et al. Chromatin accessibility dynamics in a model of human forebrain development. Science. 2020;367 (6476). doi: 10.1126/science.aay1645

[37] Elkjaer ML, Frisch T, Reynolds R, et al. Unique RNA signature of different lesion types in the brain white matter in progressive multiple sclerosis. Acta Neuropathol Commun. 2019;7(1):58. doi: 10.1186/ s40478-019-0709-3

[38] Müller S, Kohanbash G, Liu SJ, et al. Single-cell pro- filing of human gliomas reveals macrophage onto- geny as a basis for regional differences in macrophage activation in the tumor microenvironment. Genome Biol. 2017;18(1):234. doi: 10.1186/s13059-017-1362-4

[39] Gillen AE, Riemondy KA, Amani V, et al. Single-cell RNA sequencing of childhood ependymoma reveals neoplastic cell subpopulations that impact molecular classification and etiology. Cell Rep. 2020;32 (6):108023. doi: 10.1016/j.celrep.2020.108023

[40] Hovestadt V, Smith KS, Bihannic L, et al. Resolving medulloblastoma cellular architecture by single-cell genomics. Nature. 2019;572(7767):74-79. doi: 10. 1038/s41586-019-1434-6

[41] Zhong S, Zhang S, Fan X, et al. A single-cell RNA-seq survey of the developmental landscape of the human prefrontal cortex. Nature. 2018;555(7697):524-528. doi: 10.1038/nature25980

[42] Pombo Antunes AR, Scheyltjens I, Lodi F, et al. Single- cell profiling of myeloid cells in glioblastoma across species and disease stage reveals macrophage competi- tion and specialization. Nat Neurosci. 2021;24 (4):595-610. doi: 10.1038/s41593-020-00789-y

[43] de Vlaminck K, Romão E, Puttemans J, et al. Imaging of glioblastoma tumor-associated myeloid cells using nanobodies targeting signal regulatory protein alpha. Front Immunol. 2021;12:777524. doi: 10.3389/fimmu. 2021.777524

44] Neftel C, Laffy J, Filbin MG, et al. An integrative model of cellular states, plasticity, and genetics for glioblastoma. Cell. 2019;178(4):835-849.e21. doi: 10.1016/j.cell.2019.06.024

[45] Marinari E, Allard M, Gustave R, et al. Inflammation and lymphocyte infiltration are associated with shorter survival in patients with high-grade glioma. Oncoimmunology. 2020;9(1):1779990. doi: 10.1080/ 2162402X.2020.1779990

[46] Xie Y, He L, Lugano R, et al. Key molecular alterations in endothelial cells in human glioblastoma uncovered through single-cell RNA sequencing. JCI Insight. 2021. doi: 10.1172/jci.insight.150861

[47] Wang H, Xu T, Huang Q, et al. Immunotherapy for malignant glioma: current status and future directions. Trends Pharmacol Sci. 2020;41:123-138. doi: 10.1016/j.tips.2019.12.003

[48] Lim M, Xia Y, Bettegowda C, et al. Current state of immunotherapy for glioblastoma. Nat Rev Clin Oncol. 2018;15(7):422-442. doi: 10.1038/s41571-018-0003-5

[49] Bourara K, Liegler TJ, Grant RM, et al. Target cell APOBEC3C can induce limited G-to-A mutation in HIV-1. PLOS Pathog. 2007;3(10):1477-1485. doi: 10.1371/journal.ppat.0030153

[50] Kanabe BO, Ozaslan M, Aziz SA, et al. Expression patterns of LncRNA-GAS5 and its target APOBEC3C gene through miR-103 in breast can- cer patients. Cell Mol Biol (Noisy-le-Grand). 2021;67(3):5-10. doi: 10.14715/cmb/2021.67.3.2

[51] Jiang Q, Isquith J, Ladel L, et al. Inflammation- driven deaminase deregulation fuels human pre-leukemia stem cell evolution. Cell Rep. 2021;34:108670. doi: 10.1016/j.celrep.2020.108670

[52] Kawahara R, Recuero S, Nogueira FCS, et al. Tissue proteome signatures associated with five grades of pros- tate cancer and benign prostatic hyperplasia. Proteomics. 2019;19(21-22):e1900174. doi: 10.1002/ pmic.201900174

[53] Went M, Kinnersley B, Sud A, et al. Transcriptome- wide association study of multiple myeloma identifies candidate susceptibility genes. Hum Genomics. 2019;13(1):37. doi: 10.1186/s40246-019-0231-5

[54] Zhang B, Zhang X, Jin M, et al. CagA increases DNA methylation and decreases PTEN expression in human gastric cancer. Mol Med Rep. 2019;19:309-319. doi: 10.3892/mmr.2018.9654

[55] Wang J-J, Wang H, Zhu B-L, et al. Development of a prognostic model of glioma based on immune-re- lated genes. Oncol Lett. 2021;21(2):116. doi: 10.3892/ ol.2020.12377

[56] Qin B, Jiang M, Li X, et al. Oxygen nanocarrier broke the hypoxia trap of solid tumors and rescued trans- fection efficiency for gene therapy. J Nanob iotechnology. 2021;19(1):427. doi: 10.1186/s12951- 021-01144-4

[57] Thompson CB, Shepard HM, O’Connor PM, et al. Enzymatic depletion of tumor hyaluronan induces antitumor responses in preclinical animal models. Mol Cancer Ther. 2010;9:3052-3064. doi: 10.1158/1535-7163.MCT-10-0470

[58] Koyama S, Matsunaga S, Imanishi M, et al. Tumour blood vessel normalisation by prolyl hydroxylase inhibitor repaired sensitivity to chemotherapy in a tumour mouse model. Sci Rep. 2017;7(1):45621. doi: 10.1038/srep45621

[59] Li S, Zhang Q, Hong Y. Tumor vessel normal- ization: a window to enhancing cancer immunotherapy. Technol Cancer Res Treat. 2020;19:1533033820980116. doi: 10.1177/1533033 820980116

[60] Wu T, Dai Y. Tumor microenvironment and thera- peutic response. Cancer Lett. 2017;387:61-68. doi: 10. 1016/j.canlet.2016.01.043

[61] Stylianopoulos T, Munn LL, Jain RK. Reengineering the tumor vasculature: improving drug delivery and efficacy. Trends Cancer. 2018;4 (4):258-259. doi: 10.1016/j.trecan.2018.02.010

[62] Huang Y, Goel S, Duda DG, et al. Vascular nor- malization as an emerging strategy to enhance cancer immunotherapy. Cancer Res. 2013;73:2943-2948. doi: 10.1158/0008-5472.CAN- 12-4354