Abstract

   Glioma is one of the most common primary brain tumors, and is divided
   into low-grade and high-grade gliomas. Long non-coding RNAs have been
   increasingly implicated in the pathogenesis and prognosis of glioma.
   Here, we demonstrated that the long non-coding RNA TP73-AS1 is
   differentially expressed among gliomas with different
   clinicopathological features in The Cancer Genome Atlas (TCGA), Chinese
   Glioma Genome Atlas (CGGA), and GEO glioma datasets; high expression of
   TP73-AS1 was associated with poor clinical features, including age,
   stage, IDH mutation status, 1p/19q co-deletion status and overall
   survival. Measuring TP73-AS1 expression using real-time PCR showed the
   same result for 76 glioma tissue samples from our hospital. The
   infiltration levels of various immune cells in the tumor
   microenvironment were found to be significantly higher in patients with
   high expression of TP73-AS1. Taken together, our results suggest that
   TP73-AS1 has potential as a prognostic glioma biomarker. Moreover, the
   knowledge that TP73-AS1 affects the glioma immune microenvironment may
   provide new information for the immunological research and treatment of
   glioma.

   Keywords: glioma, TP73-AS1, immune, prognostic biomarker

INTRODUCTION

   Glioma is one of the most destructive tumors with high mortality rate
   [[46]1]. Therefore, it is urgent to identify prognostic markers and new
   therapeutic targets for glioma.

   Long-noncoding RNAs (lncRNAs) are a type of transcription product with
   no protein coding ability. Abnormal lncRNA expression has been related
   to the development of glioma [[47]2, [48]3]. For example, PAXIP1-AS1,
   as a long-noncoding RNA, was proved to facilitate progression of glioma
   by regulate the expression of KIF14 [[49]4]. In addition, silencing
   FOXD2-AS1 could inhibit the progression of glioma and promote apoptosis
   of glioma cells by regulating the microRNA-98-5p/CPEB4 axis [[50]5].
   The overexpression of lncRNA LOC101928963 contributes to the
   proliferation of glioma cells by regulating to PMAIP1 expression
   [[51]6]. The lncRNA TP73-AS1, also known as KIAA0495 and p53-dependent
   apoptosis modulator (PDAM), is located in human chromosomal band
   1p36.32 and plays a crucial role in various carcinomas. The
   downregulation of TP73-AS1 has been found in oligodendromas, in which
   it was confirmed that the knockout of TP73-AS1 in glioma cells induced
   cisplatin resistance [[52]7]. In contrast, other studies have shown
   that TP73-AS1 is upregulated in glioma tissues and that the knockdown
   of TP73-AS1 inhibits the proliferation and invasion of glioma cells
   [[53]8]. Therefore, the specific role of TP73-AS1 in gliomas is not
   clear. In particular, immunotherapy has proven to be a very promising
   treatment for gliomas [[54]9], but the relationship between TP73-AS1
   and immune cell infiltration in glioma tissue has not been studied so
   far.

   In this study, we used the CGGA RNA sequencing, TCGA low-grade and
   high-grade glioma dataset, and [55]GSE16011 dataset to investigate the
   potential roles of TP73-AS1 in glioma. Our study suggests that the
   TP73-AS1 could serve as a potential prognostic biomarker and may have a
   nonnegligible effect on the immune microenvironment composition of
   glioma, which may provide new insight for the immunological research
   and treatment of glioma.

RESULTS

The expression of TP73-AS1 is correlated with poor clinicopathological
features in gliomas

   To explore the difference in expression of TP73-AS1 between gliomas
   with different clinicopathological features, we examined the expression
   of TP73-AS1 in three datasets: TCGA, CGGA and [56]GSE16011. We found
   that the expression of TP73-AS1 increased with the increase of WHO
   grades ([57]Figure 1A–[58]1C). Age is a risk factor for glioma
   patients. We found that in the three datasets, the expression of
   TP73-AS1 in the older age group was significantly higher than that in
   the younger age group ([59]Figure 1D–[60]1F).

Figure 1.

   [61]Figure 1
   [62]Open in a new tab

   Expression levels of TP73-AS1 in gliomas with different
   clinicopathological features (A–C) Expression levels of TP73-AS1 in
   different WHO grades in the TCGA (A), CGGA (B), and [63]GSE16011 (C)
   datasets. (D–F) Expression levels of TP73-AS1 in different age groups
   in the TCGA (D), CGGA (E), and [64]GSE16011 (F) datasets.

   Considering the important influence of IDH mutation and 1p/19q
   chromosomal co-deletion on the prognosis of glioma patients, we
   evaluated the correlation between the expression of TP73-AS1 and the
   statuses of IDH and 1p/19q chromosome. We found the same result that
   TP73-AS1 expression was significantly downregulated in the IDH mutation
   group ([65]Figure 2A–[66]2F), and the 1p/19q chromosome co-deletion
   group in TCGA, CGGA and [67]GSE16011 ([68]Figure 2G–[69]2L).

Figure 2.

   [70]Figure 2
   [71]Open in a new tab

   Expression levels of TP73-AS1 in gliomas with different IDH and 1p/19q
   statuses. (A–C) Expression levels of TP73-AS1 for different statuses of
   IDH in all WHO grades in the TCGA (A), CGGA (B), and [72]GSE16011 (C)
   datasets. (D–F) Expression levels of TP73-AS1 for different statuses of
   IDH in all WHO grades in the TCGA (D), CGGA (E), and [73]GSE16011 (F)
   datasets. (G–I) Expression levels of TP73-AS1 for different statuses of
   1p/19q co-deletion in all WHO grades in the TCGA (G), CGGA (H), and
   [74]GSE16011 (I) datasets. (J–L) Expression levels of TP73-AS1 for
   different statuses of 1p/19q co-deletion in all WHO grades in the TCGA
   (J), CGGA (K), and [75]GSE16011 (L) datasets.

TP73-AS1 is a risk factor for the prognosis of gliomas

   The expression of TP73-AS1 is higher in glioma patients with a worse
   prognosis. We therefore explored the direct relationship between
   TP73-AS1 expression and the prognosis of glioma. We found that TP73-AS1
   was a significant prognostic risk factor, including for primary gliomas
   in the TCGA, CGGA, and [76]GSE16011, and recurrent gliomas in CGGA
   ([77]Figure 3A–[78]3D). We further analyzed different WHO grade of
   gliomas. We found that for WHO grade III gliomas, the high expression
   of TP73-AS1 indicated a poor prognosis for primary gliomas in the TCGA,
   CGGA, and [79]GSE16011 and for recurrent gliomas of CGGA ([80]Figure
   3I–[81]3L); however, the relation of high TP73-AS1 expression with poor
   prognosis was not obvious in WHO grade II ([82]Figure 3E–[83]3H) and
   WHO grade IV ([84]Figure 3M–[85]3P) gliomas.

Figure 3.

   [86]Figure 3
   [87]Open in a new tab

   Kaplan-meier analysis of TP73-AS1. (A–D) Kaplan-Meier analysis of OS
   for the expression of TP73-AS1 in all WHO grades in the TCGA primary
   glioma dataset (A), CGGA primary glioma dataset (B), CGGA recurrent
   glioma dataset (C), and [88]GSE16011 primary glioma dataset (D). (E–H)
   Kaplan-Meier analysis of OS for the expression of TP73-AS1 in WHO grade
   II in the TCGA primary glioma dataset (E), CGGA primary glioma dataset
   (F), CGGA recurrent glioma dataset (G), and [89]GSE16011 primary glioma
   dataset (H). (I–L) Kaplan-Meier analysis of OS for the expression of
   TP73-AS1 in WHO grade III in the TCGA primary glioma dataset (I), CGGA
   primary glioma dataset (J), CGGA recurrent glioma dataset (K), and
   [90]GSE16011 primary glioma dataset (L). (M–P) Kaplan-Meier analysis of
   OS for the expression of TP73-AS1 in WHO grade IV in TCGA primary
   glioma dataset (M), CGGA primary glioma dataset (N), CGGA recurrent
   glioma dataset (O), and [91]GSE16011 primary glioma dataset (P).

The methylation level of TP73-AS1 was negatively correlated with the
expression of TP73-AS1, a favorable prognostic factor

   In general, promoter methylation regulates gene expression. We analyzed
   the methylation levels of TP73-AS1 in gliomas with different
   clinicopathological features. We found that the methylation level of
   TP73-AS1 decreased with an increase in WHO grade in the TCGA database
   ([92]Figure 4A). IDH mutation ([93]Figure 4B) and 1p/19q co-deletion
   ([94]Figure 4C) led to high levels of TP73-AS1 methylation. Moreover,
   the methylation level of TP73-AS1 was decreased in the older age group
   ([95]Figure 4D). We confirmed that the methylation level of the
   TP73-AS1 promoter was negatively correlated with the expression level
   of TP73-AS1, suggesting that the methylation of the TP73-AS1 promoter
   led to a decrease in TP73-AS1 expression ([96]Figure 4E). Survival
   analysis showed that the methylation level of TP73-AS1 was a favorable
   prognostic factor, especially in WHO grade III gliomas ([97]Figure
   4F–[98]4I).

Figure 4.

   [99]Figure 4
   [100]Open in a new tab

   Analysis of the TP73-AS1 methylation level. (A) Methylation level of
   TP73-AS1 in different WHO grades in the TCGA dataset. (B) Methylation
   level of TP73-AS1 for different IDH statuses in the TCGA dataset. (C)
   Methylation level of TP73-AS1 for different 1p/19q statuses in the TCGA
   dataset. (D) Methylation level of TP73-AS1 for different age groups in
   the TCGA dataset. (E) Pearson’s correlation between the methylation
   level and mRNA expression of TP73-AS1. (F) Kaplan-Meier analysis of OS
   for the methylation level of TP73-AS1 for all WHO grades in the TCGA
   primary glioma dataset. (G) Kaplan-Meier analysis of OS for the
   methylation level of TP73-AS1 in WHO grade II in the TCGA primary
   glioma dataset. (H) Kaplan-Meier analysis of OS for the methylation
   level of TP73-AS1 in WHO grade III in the TCGA primary glioma dataset.
   (I) Kaplan-Meier analysis of OS for the methylation level of TP73-AS1
   in WHO grade IV in the TCGA primary glioma dataset.

Functional enrichment analysis of TP73-AS1

   To further study the function of TP73-AS1, we used the TCGA based
   database Linkedomics to perform a correlation analysis of TP73-AS1. Two
   heatmaps were constructed to illustrate the genes whose expression was
   most positively and negatively correlated with that of TP73-AS1
   ([101]Figure 5A). GSEA showed that TP73-AS1 probably participates in
   KEGG pathways, including those involving Staphylococcus aureus
   infection, complement and coagulation cascades, and Reactome pathways,
   including integrin cell surface interactions and immunoregulatory
   interactions between a Lymphoid and a non-Lymphoid cell. The genes
   whose expression was negatively related with that of TP73-AS1 were
   enriched in KEGG pathways, including those associated with the ribosome
   and nicotine addiction and Reactome pathways, including eukaryotic
   translation termination and cap-dependent translation initiation
   pathways ([102]Figure 5B–[103]5C).

Figure 5.

   [104]Figure 5
   [105]Open in a new tab

   Function and pathway enrichment analysis for TP73-AS1. (A) Heatmaps
   normalized using the Z-scores of the top 50 genes positively and
   negatively correlated with TP73-AS1. (B, C) Significant GSEA results of
   TP73-AS1, including KEGG pathways (B) and Reactome pathways (C). (D)
   Significant Gene Ontology terms of TP73-AS1. (E) Significant KEGG
   pathways of TP73-AS1.

   To further explore potential functional pathways associated with
   TP73-AS1, we conducted a functional pathway analysis based on the most
   positively correlated genes identified in the abovementioned analysis
   (Pearson r>0.5, P<0.01) using the “org.Hs.eg.db,” “clusterProfiler,”
   and “enrichplot” R packages. These genes were mainly associated with
   extracellular structure organization, extracellular matrix
   organization, collagen fibril organization, and positive regulation of
   receptor binding in GO (Gene Ontology) categories ([106]Figure 5D).
   KEGG pathway analysis indicated that the pathways in which the most
   highly-correlated genes were enriched involved focal adhesion,
   proteoglycans in cancer, ECM−receptor interactions, and leukocyte
   transendothelial migration ([107]Figure 5F).

Tumor immune microenvironment analysis

   The tumor microenvironment (TME), especially the tumor immune
   microenvironment, is important for the survival of tumor patients. To
   evaluate the populations of various immune cells, we employed
   single-sample gene set enrichment analysis (ssGSEA;
   [108]https://www.genepattern.org/modules/docs/ssGSEAProjection/4) to
   determine the relative infiltration level of 28 immune cells
   ([109]Figure 6A). We made two major observations: (1) The high TP73-AS1
   group had more complex immune cell components than the low TP73-AS1
   group, including cells exhibiting anti-tumor reactivity, such as
   activated CD4 T cells, activated dendritic cells, central memory CD4 T
   cells, central memory CD8 T cells, natural killer cells, and natural
   killer T cells, and cells involved in tumor suppression, such as
   immature dendritic cells, MDSCs, plasmacytoid dendritic cells,
   regulatory T cells, and type 2 T helper cells ([110]Figure 6B). (2) The
   correlation analysis revealed that the infiltration level of these two
   categories of immune cells were positively correlated ([111]Figure 6C).
   This result revealed a feedback mechanism that anti-tumor inflammation
   could promote the recruitment or differentiation of immunosuppressive
   cells. Correlation analysis showed that TP73-AS1 expression was
   positively correlated with the infiltration scores of most immune cells
   ([112]Figure 6D). To further verify the results, we used the Timer
   database ([113]https://cistrome.shinyapps.io/timer/) to analyze the
   infiltration scores of immune cells in glioma tissue. We came to the
   same conclusion, that the high TP73-AS1 group had more complex immune
   cell components ([114]Figure 6E) and that TP73-AS1 was positively
   correlated with the infiltration scores of most immune cells
   ([115]Figure 6F).

Figure 6.

   [116]Figure 6
   [117]Open in a new tab

   Tumor immune microenvironment analysis for TP73-AS1. (A) Heatmap
   normalized using the Z-scores of the 28 tumor-infiltrating immune cell
   populations. (B) The signature scores (ssGSEA scores) of 28
   infiltrating immune cells in the high and low TP73-AS1 expression
   groups. (C) Pearson’s correlation between the pro-tumor suppression
   score and anti-tumor immunity score. (D) The correlation heatmaps of
   the TP73-AS1 expression levels and infiltration scores (ssGSEA scores)
   of different immune cells. (E) The signature scores (Timer) of six
   infiltrating immune cells in high and low TP73-AS1 expression groups.
   (F) The correlation circle diagram of the TP73-AS1 expression and
   infiltration scores (Timer) of different immune cells.

Data analysis from the hospital

   We found that TP73-AS1 was differentially expressed in gliomas with
   different clinicopathological features in the TCGA, CGGA, and GEO
   glioma datasets. The overexpression of TP73-AS1 was associated with
   poor clinical features, including age, stage, IDH mutation status,
   1p/19q co-deletion status, and overall survival. RT-PCR was performed
   to detect the expression of the TP73-AS1 protein in 67 glioma tissues
   of different grades from our hospital. The results showed higher
   expression of TP73-AS1 in glioma samples of WHO grade IV than in glioma
   samples of WHO grade IV (p=0.002) ([118]Figure 7A), and high TP73-AS1
   expression predicted a poorer prognosis than low TP73-AS1 expression
   (P=0.048) ([119]Figure 7B); these findings agreed with our preliminary
   results. We also found that TP73-AS1 expression levels were
   significantly correlated with the WHO grade of gliomas (p <0.0001),
   IDH1 mutation (p=0.0035), and 1p/19q co-deletion state (p=0.0119)
   ([120]Table 1).

Figure 7.

   [121]Figure 7
   [122]Open in a new tab

   Expression and Kaplan-Meier analyses of TP73-AS1 in patients with
   glioma from the hospital. (A) The expression value of TP73-AS1 in
   glioma samples of patients from the hospital with different WHO grades.
   (B) Kaplan-Meier analysis of OS for the expression of TP73-AS1 in
   glioma samples of patients from the hospital with all WHO grades.

Table 1. Correlation between the expression of TP73-AS1 and
clinicopathological characteristics in glioma tissue.

   Characteristic           Expression level of TP73-AS1
                         n  Low (n)       High (n)        P-value
   Age (years)
   >50                   44   18             26           0.0588
   ≤50                   23   15              8
   Gender
   Male                  35   18             17           0.7096
   Female                32   15             17
   Tumor location
   Frontal lobe          30   15             15           0.4931
   Temporal lobe         21   12              9
   Others                16    6             10
   Tumor diameter
   <3cm                  30   16             14            0.411
   >3cm                  37   16             21
   KPS
   <80                   32   14             18           0.3889
   >80                   35   19             16
   Extent of resection
   Gross total resection 34   19             15           0.2706
   Subtotal resection    33   14             19
   IDH1
   Mutation              23   17              6           0.0035
   Wild                  44   16             28
   MGMT methylation
   Positive              41   23             18           0.1594
   Negative              26   10             16
   1p/19q codeletion
   Exist                 19   14              5           0.0119
   Without               48   19             29
   WHO grade
   II                    16   14              2           <0.0001
   III                   24   14             10
   IV                    27    5             22
   [123]Open in a new tab

DISCUSSION

   Malignant progression and high recurrence rate make glioma the most
   lethal primary brain tumor [[124]10, [125]11]. Disorders related to
   lncRNAs may contribute to the pathogenesis of glioma [[126]12–[127]14].
   In this study, we aimed to assess whether TP73-AS1 could function in
   the development of glioma.

   It was reported that TP73-AS1 is associated to aggressiveness in
   glioblastoma and promotes tumorigenicity of medulloblastoma cells
   [[128]15, [129]16]. To ascertain the role of TP73-AS1 in glioma more
   clearly, we collected TCGA, CGGA, and [130]GSE16011 glioma data to
   explore the clinicopathological features related toTP73-AS1. We found
   that level of TP73-AS1 increased with the glioma WHO grade and age of
   patients in all three datasets. IDH mutation and 1p/19q chromosome
   co-deletion have a huge influence on the prognosis of glioma patients
   [[131]17–[132]20]. TP73-AS1 expression was significantly downregulated
   in the IDH mutation group and 1p/19q chromosome co-deletion groups in
   the TCGA, CGGA, and [133]GSE16011, indicating that TP73-AS1 is a poor
   prognostic factor for glioma, which was consistent with previous
   reports [[134]15, [135]16]. Further analysis showed that TP73-AS1 is a
   poor prognostic factor for glioma, especially WHO grade III glioma in
   the TCGA, CGGA, and [136]GSE16011 datasets. In contrast, the
   methylation level of TP73-AS1 was found to be a valuable prognostic
   factor for glioma, especially for WHO grade III glioma.

   To further study the function of TP73-AS1, we divided the TCGA glioma
   samples into a high TP73-AS1 group and a low TP73-AS1 group, using the
   median expression of TP73-AS1 in the samples. The GSEA results showed
   that KEGG pathways, including those related to Staphylococcus aureus
   infection, complement and coagulation cascades, and Reactome pathways,
   including integrin cell surface interaction pathways and pathways
   related to immunoregulatory interactions between lymphoid and
   nonlymphoid cells, were significantly enriched in the high TP73-AS1
   group, while KEGG pathways, including those related to the ribosome and
   nicotine addiction, and Reactome pathways, including eukaryotic
   translation termination and cap-dependent translation initiation
   pathways, were significantly enriched in the low TP73-AS1 group.

   Immune cells in the tumor microenvironment are a vital element of tumor
   tissue [[137]21–[138]23]. Our results demonstrated that the high
   TP73-AS1 group was significantly associated with many immune-related
   signatures. By investigating the regulation of immune cell populations,
   we made two major observations: (1) The high-risk group had more
   complex immune cell components than the low-risk group, including cells
   exhibiting anti-tumor reactivity, such as activated CD4 T cells,
   activated dendritic cells, central memory CD4 T cells, central memory
   CD8 T cells, natural killer cells, and natural killer T cells, and
   cells with tumor suppression activity, such as immature dendritic
   cells, MDSCs, plasmacytoid dendritic cells, regulatory T cells, and
   type 2 T helper cells. (2) The correlation analysis revealed that the
   infiltration level of these two categories of immune cells were
   positively correlated. Correlation analysis based on two different
   analytical methods showed that TP73-AS1 was positively correlated with
   the infiltration scores of most of the immune cells. These results
   suggest that TP73-AS1 has a non-negligible effect on the immune
   microenvironment of glioma.

   Overall, our research indicated that TP73-AS1 is a reliable marker for
   predicting the overall survival of glioma patients and that TP73-AS1
   has a non-negligible effect on the immune microenvironment of glioma.

MATERIALS AND METHODS

Data collection

   The expression data and corresponding clinical information of glioma
   patients were obtained from (1) the TCGA dataset
   ([139]https://xenabrowser.net/), which included expression data and
   data on the clinicopathological features of 664 patients; (2) CGGA
   RNA-seq data ([140]http://cgga.org.cn/), which included expression data
   and data on the clinicopathological features of 984 patients; and (3)
   [141]GSE16011 ([142]https://www.ncbi.nlm.nih.gov), which included
   expression data and data on the clinicopathological features of 258
   patients. This study was approved by the ethics committee of the
   Zhejiang Cancer Hospital. Written informed consent was obtained from
   all participants in this study.

qRT-PCR

   TRIzol reagent (Life Technologies) and RevertAid First Strand cDNA
   Synthesis Kit (Thermo Scientific) were used to extract Total RNA and
   synthesize cDNA according to the manufacturer’s protocol. qRT-PCR
   analysis was performed using the standard protocol of the SYBR Select
   Master Mix kit. GAPDH was included as an internal reference gene. The
   primers used were as follows:

   TP73-AS1-F 5ʹ - CCGTGGTTAGCGTGGACATCAG – 3ʹ

   TP73-AS1-R 5ʹ - TGGGTGACAGGGCAAGACTCC – 3ʹ

   GAPDH-F 5ʹ - CATGAGAAGTATGACAACAGCCT – 3ʹ

   GAPDH-R 5ʹ - AGTCCTTCCACGATACCAAAGT – 3ʹ

Differential expression analysis and survival analysis

   To investigate the expression of TP73-AS1, we used the R packages
   ggplot2, ggpubr, and grid Extra in the R software (3.6.2) to study gene
   expression patterns in gliomas with different clinicopathological
   features. For survival analysis, the R packages survminer, survival,
   ggpubr, and gridExtra were employed. The age segmentation value of each
   dataset was the median age of patients in the data set.

Gene set enrichment analyses and enrichment analysis

   Data from the TCGA dataset were used for gene set enrichment analyses
   and enrichment analysis. GO analysis was performed using the EnrichGO
   function in the clusterProfiler R package, with the following
   parameters: ont = “all” pvalue-Cutoff = 0.05 and qvalue-Cutoff = 0.05.
   KEGG analysis was performed using the EnrichKEGG function of the R
   package “clusterProfiler,” with the following parameters: keyType =
   “kegg” pvalue-Cutoff = 0.05 and qvalue-Cutoff = 0.05. Gene set
   enrichment analyses were performed using the Linkedomics online dataset
   ([143]http://www.linkedomics.org/login.php).

Immune cell infiltration

   Single-sample GSEA was used to evaluate the infiltration level of 28
   immune cell types. Immune cell signatures were obtained from a
   published paper [[144]24]. The infiltration levels of immune cell types
   were represented by an enrichment score in the ssGSEA analysis of the
   GSVA R package. The ssGSEA score was normalized to unity distribution,
   with zero as the minimum and one as the maximum score for each immune
   cell type.

Ethics approval and consent to participate

   Additional informed consent was obtained from all participants whose
   identifying information was included in this article.

Data availability statement

   The datasets generated and/or analyzed during the current study are
   available from the corresponding author upon reasonable request.

ACKNOWLEDGMENTS