长链非编码RNA空间转录组分析揭示组织特异性和癌症中的功能

Spatial transcriptome analysis of long non-coding RNAs reveals tissue specificity and functional roles in cancer 长链非编码RNA空间转录组分析揭示组织特异性和癌症中的功能 1 first-author 1 1 1 1 1 1 1 1 2 corresp corresp Juan XU, xujuanbiocc@ems.hrbmu.edu.cn Juan XU, xujuanbiocc@ems.hrbmu.edu.cn Yongsheng LI, liyongsheng@hainmc.edu.cn Yongsheng LI, liyongsheng@hainmc.edu.cn liyongsheng@hainmc.edu.cn https://orcid.org/0000-0003-1914-0727 1 corresp Juan XU, xujuanbiocc@ems.hrbmu.edu.cn Juan XU, xujuanbiocc@ems.hrbmu.edu.cn xujuanbiocc@ems.hrbmu.edu.cn https://orcid.org/0000-0002-3709-4165 哈尔滨医科大学生物信息学科学与技术学院， 中国哈尔滨市， 150081 哈尔滨医科大学生物信息学科学与技术学院， 中国哈尔滨市， 150081 College of Bioinformatics Science and Technology, Harbin Medical University, Harbin 150081, China College of Bioinformatics Science and Technology, Harbin Medical University, Harbin 150081, China 海南医学院热带转化医学教育部重点实验室， 生物医学信息与工程学院， 海南省妇女儿童医学中心， 中国海口市， 571199 海南医学院热带转化医学教育部重点实验室， 生物医学信息与工程学院， 海南省妇女儿童医学中心， 中国海口市， 571199 Key Laboratory of Tropical Translational Medicine of Ministry of Education, College of Biomedical Information and Engineering, Hainan Women and Children’s Medical Center, Hainan Medical University, Haikou 571199, China Key Laboratory of Tropical Translational Medicine of Ministry of Education, College of Biomedical Information and Engineering, Hainan Women and Children’s Medical Center, Hainan Medical University, Haikou 571199, China Long non-coding RNAs (lncRNAs) play a significant role in maintaining tissue morphology and functions, and their precise regulatory effectiveness is closely related to expression patterns. However, the spatial expression patterns of lncRNAs in humans are poorly characterized. Here, we constructed five comprehensive transcriptomic atlases of human lncRNAs covering thousands of major tissue samples in normal and disease states. The lncRNA transcriptomes exhibited high consistency within the same tissues across resources, and even higher complexity in specialized tissues. Tissue-elevated (TE) lncRNAs were identified in each resource and robust TE lncRNAs were refined by integrative analysis. We detected 1 to 4684 robust TE lncRNAs across tissues; the highest number was in testis tissue, followed by brain tissue. Functional analyses of TE lncRNAs indicated important roles in corresponding tissue-related pathways. Moreover, we found that the expression features of robust TE lncRNAs made them be effective biomarkers to distinguish tissues; TE lncRNAs also tended to be associated with cancer, and exhibited differential expression or were correlated with patient survival. In summary, spatial classification of lncRNAs is the starting point for elucidating the function of lncRNAs in both maintenance of tissue morphology and progress of tissue-constricted diseases. 目的 2 长链非编码RNA（Long non-coding RNAs, lncRNAs）在维持组织形态和功能方面发挥着重要作用，其精确调控效应与其表达模式密切相关。然而，lncRNA在人类体内的空间表达模式还没有得到明确刻画。 方法 2 在本研究中，我们构建了5个完整的人类lncRNAs转录组图谱，覆盖了数千个正常或疾病状态下的主要组织样本。在不同资源的相同组织中，lncRNA转录组表现出高度的一致性，在特定组织中表现出更高的复杂性。在每个资源中都发现了组织升高（Tissue-elevated, TE）的lncRNA，并通过整合分析提炼出了稳定的TE lncRNA。我们在组织中检测到1～4684个稳定的TE lncRNA，其中TE lncRNA最多的是睾丸组织，其次是脑组织。TE lncRNA的功能分析表明，TE lncRNA在相应的组织相关通路中发挥重要作用。此外，我们发现稳健的TE lncRNA的表达特征使其成为区分组织的有效生物标记物；TE lncRNA也倾向于与癌症相关，表现出差异表达或与患者生存相关。 结论 2 综上所述，lncRNAs的空间分类是阐明lncRNAs在维持组织形态和组织限制性疾病进展中功能的重要起点。 空间表达 转录组 长链非编码RNA（lncRNA） 癌症 生物标志物 Spatial expression Transcriptome Long non-coding RNA Cancer Biomarker 1 Introduction 1 Long non-coding RNAs (lncRNAs), a class of non-coding RNAs in the human genome, control various crucial biological functions ( Xu et al., 2020 Xu et al., 2020 ). As a large group of RNA regulators, their precise regulatory effectiveness is closely related to the tissue context ( Lv et al., 2020 Lv et al., 2020 ), as well as their expression patterns in tissues under both healthy and diseased states. Moreover, spatial characterization of lncRNA expression across tissues is important, because tissue-constricted lncRNA expression and pathway regulation underlie human physiology, and their dysfunction often results in disease. These factors combined to make the understanding of the tissue context of lncRNA functions, disease pathophysiology, and lncRNA‍–‍disease associations particularly challenging. Identifying tissue-elevated (TE) lncRNAs across tissues is the first step in exploring their spatial expression patterns. Indeed, apart from a small number of genes only expressed in one tissue, more and more studies have observed an interesting group of protein-coding genes with TE expression in certain tissues or groups, which are either not expressed or expressed at lower levels in other tissues ( She et al., 2009 She et al., 2009 ; Burgess, 2015 Burgess, 2015 ; Uhlén et al., 2016 Uhlén et al., 2016 ; Aran et al., 2017 Aran et al., 2017 ; Wang et al., 2019 Wang et al., 2019 ). Thus, TE genes, including tissue-specific ones, can serve as biomarkers of specific biological processes or tissues in which they are expressed. To obtain a more comprehensive picture of the functions of lncRNAs, a multiple-tissue analysis is usually performed to look at their expression in a panel of tissues or organs. Indeed, it has been shown that disease genes, including lncRNAs, generally tend to be expressed in a limited number of tissues ( Li et al., 2020 Li et al., 2020 ). However, the spatial expression pattern of lncRNAs is still not comprehensively characterized across human tissues and cancers. Over the years, quantitative transcriptomic technologies such as RNA-sequencing (RNA-seq) and cap analysis of gene expression (CAGE) have been used to build genome-wide expression drafts of human genes, including both lncRNAs and protein-coding genes. The spatial patterns of protein-coding genes have been studied ( Rawal et al., 2021 Rawal et al., 2021 ). However, for lncRNAs, broader scale analyses had mostly focused on a single tissue or cell-type-resolved analysis. To the best of our knowledge, no broad-scale quantitative or integrative analysis of available lncRNA transcriptomes across healthy human tissues has been performed that would enable a comprehensive investigation of the spatial expression patterns of lncRNAs. Fortunately, several recent efforts to build large-scale expression profiles in samples representing most of the major human organs and tissues have been published. Well-accepted transcriptome studies on human tissues have been published by the Genotype-Tissue Expression (GTEx) consortium ( Uhlén et al., 2015 Uhlén et al., 2015 ), the Human BodyMap 2 (HBM2) ( Schroth, 2011 Schroth, 2011 ), the Human Protein Atlas (HPA) consortium ( Uhlen et al., 2010 Uhlen et al., 2010 ), the Functional Annotation of the Mammalian Genome (FANTOM) consortium ( Hon et al., 2017 Hon et al., 2017 ), and The Cancer Genome Atlas (TCGA) ( The Cancer Genome Atlas Research Network et al., 2013 The Cancer Genome Atlas Research Network et al., 2013 ), collectively comprising thousands of samples from all major human tissues. The former three were profiled from the normal physiological state in humans; in particular, the GTEx generated normal transcriptome with the greatest number of samples pooled from 30 healthy human tissues. On the other hand, TCGA provides a comprehensive transcriptome for various cancers. Four (GTEx, HBM2, HPA, TCGA) of these five expression profiles were derived from RNA-seq assemblies, and the last one (FANTOM) was derived from CAGE. Genes with accurate 5'-ends were identified. These five expression profiles are notable for representing independent surveys of human genes by distinct methods (RNA-seq vs. CAGE), in different tissues under distinct conditions (normal vs. disease). Therefore, these five large-scale expression profiles provide an opportunity to integrate data sets and further refine the spatial classification of lncRNAs in association with tissue context expression, and not just for protein-coding gene analysis. To improve our understanding of the spatial expression patterns of lncRNA across human tissues, we built a comprehensive spatial expression atlas of lncRNA in humans by re-analyzing a large-scale transcriptomic atlas and provided a comprehensive baseline map of lncRNA expression across the human body. Using integrative analysis of a lncRNA transcriptome covering 20 ‍519 samples across 38 tissues, we firstly assessed the relative expression of lncRNAs in each tissue in all publications. Next, robust TE lncRNAs in normal tissues were identified by analyzing four transcriptomes of normal tissues. We extended this analysis to investigate the expression feature of TE lncRNA in cancer, and discovered candidate clinically associated lncRNA markers by comparing transcriptional profiles of pathological samples to the normal transcriptome. Our integrative analysis substantially obtained the robust TE lncRNAs across the datasets. The comprehensive atlas of TE lncRNAs across multiple tissues and cancers provides data for a better understanding of lncRNA function and disease development and progression. 2 Materials and methods 1 2.1 Construction of a lncRNA and mRNA transcriptome across tissues 2 We used patients’ data acquired from publicly available datasets that were collected with patients’ informed consent. To conduct a systematic analysis of the transcriptome across multiple tissues, we used five widely available lncRNA and messenger RNA (mRNA) expression datasets (Table S1). From the GTEx ( Lonsdale et al., 2013 Lonsdale et al., 2013 ), we obtained the expression profiles for 8555 samples across 30 tissues. Expression of lncRNAs and protein-coding genes was measured by reads per kilobase of transcript per million mapped reads (RPKM). We obtained the raw RNA-seq data from two other tissue banks, HBM2 ( Schroth, 2011 Schroth, 2011 ) and HPA ( Rustici et al., 2013 Rustici et al., 2013 ; Uhlén et al., 2015 Uhlén et al., 2015 ). The HBM2 includes 16 samples across 16 tissues, and the HPA includes 95 samples across 27 tissues. The sequence reads were mapped to the human genome 19 using TopHat ( http://ccb. http://ccb. ‍jhu.‍edu/software/tophat/index.‍shtml) ( Kim et al., 2013 Kim et al., 2013 ), and the expression levels were calculated using Cufflinks v2.1.1 based on fragments per kilobase of transcript per million mapped reads (FPKM) ( Trapnell et al., 2012 Trapnell et al., 2012 ). In addition, we obtained gene and lncRNA expression profiles from 760 samples across 34 tissue types from the FANTOM project ( https://fantom. https://fantom. ‍gsc.riken.‍jp/5) ( Noguchi et al., 2017 Noguchi et al., 2017 ). Expression of genes and lncRNAs was measured by maximum counts per million (CPM). Moreover, we obtained the lncRNA and gene expression datasets generated by the TCGA Research Network ( http://cancergenome. http://cancergenome. ‍nih.‍gov). In total, 33 different TCGA projects representing different cancer types were analyzed. RNA-seq-based expression profile data were obtained via the R package “TCGAbiolinks” ( Colaprico et al., 2016 Colaprico et al., 2016 ). We downloaded the FPKM-based gene expression for 33 types of cancer. Based on the annotations in GENCODE ( https://www.gencodegenes. https://www.gencodegenes. ‍‍org) ( Derrien et al., 2012 Derrien et al., 2012 ), we divided the gene expression profiles into lncRNA and protein-coding gene expression for each cancer type. 2.2 Cluster analysis of tissues based on lncRNA expression 2 To investigate whether the expression of lncRNAs can distinguish different tissues, the top 1000 lncRNAs with higher variance were obtained from each data resource. The overlapping lncRNAs in four resources (GTEx, HBM2, HPA, FANTOM) were selected. We next performed uniform manifold approximation and projection (UMAP) dimension reduction with the R package “umap” ( https://github. https://github. ‍com/tkonopka/umap), using these shared top lncRNAs. Each sample was visualized in two dimensions and the samples were colored by tissue type and data resource. 2.3 Analysis of the complexity of the lncRNA transcriptome for each tissue 2 In each tissue, lncRNAs were first ranked in descending order based on their average expression levels in the tissue. For a given top number of lncRNAs ( k ), we defined the complexity ( C ) of the lncRNA transcriptome in tissue ( i ) as follows: <math display="block" id="M1"><mi>C</mi><mfenced separators="|"><mrow><mi>i</mi><mo>,</mo><mtext> </mtext><mi>k</mi></mrow></mfenced><mo>=</mo><mfrac><mrow><mstyle displaystyle="true"><munderover><mo largeop="true">∑</mo><mrow><mi>a</mi><mo>=</mo><mn mathvariant="normal">1</mn></mrow><mrow><mi>k</mi></mrow></munderover></mstyle><mrow><mi mathvariant="normal">e</mi><mi mathvariant="normal">x</mi><mi mathvariant="normal">p</mi><mo stretchy="false">(</mo><mi>a</mi><mo>,</mo><mtext> </mtext><mi>i</mi><mo stretchy="false">)</mo></mrow></mrow><mrow><mstyle displaystyle="true"><munderover><mo largeop="true">∑</mo><mrow><mi>a</mi><mo>=</mo><mn mathvariant="normal">1</mn></mrow><mrow><mi>n</mi></mrow></munderover></mstyle><mrow><mi mathvariant="normal">e</mi><mi mathvariant="normal">x</mi><mi mathvariant="normal">p</mi><mo stretchy="false">(</mo><mi>a</mi><mo>,</mo><mtext> </mtext><mi>i</mi><mo stretchy="false">)</mo></mrow></mrow></mfrac><mo>,</mo><mtext> </mtext><mi>k</mi><mo>=</mo><mn mathvariant="normal">1</mn><mo>,</mo><mtext> </mtext><mn mathvariant="normal">2</mn><mo>,</mo><mtext> </mtext><mn mathvariant="normal">3</mn><mo>,</mo><mtext> </mtext><mo>…</mo><mo>,</mo><mtext> </mtext><mi>n</mi><mo>,</mo></math> http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491782&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491778&type= 60.62133789 17.86466599 where n is the total number of lncRNAs in tissue i and exp ( a , i ) represents the average expression value of lncRNA a in tissue i . 2.4 Identification of robust TE lncRNAs by integration analysis 2 All the human TE lncRNAs were identified based on the spatial expression patterns of lncRNAs across different tissues, as has been done with coding genes in a previous study ( Uhlén et al., 2016 Uhlén et al., 2016 ). Collectively, three sub-categories of TE lncRNAs were further stratified to reflect increasing degrees of elevated expression in a particular tissue, including “tissue-specific (TS),” “tissue-enriched (TER),” and “tissue-enhanced (TEH)”: (1) TS lncRNAs were expressed only in a particular tissue; (2) TER lncRNAs had at least five-fold higher expression level in a particular tissue compared to other tissues; and (3) TEH lncRNAs had at least five-fold higher expression level in a particular tissue compared to the average expression levels in all other tissues. Similarly, we identified TE lncRNAs in each cancer type. We identified TE lncRNAs individually from each resource and integrated the TE lncRNAs from different datasets. To integrate datasets and further refine the robust TE lncRNAs in normal tissues, we required that the robust TE lncRNAs be identified in the same tissue in at least two resources. 2.5 Assessment of the power of TE lncRNAs to predict tissue origin 2 We used TE lncRNAs as molecular markers to identify each tissue. First, we normalized the expression of TE lncRNAs in each resource by Z -score. Then, the normalized expression profiles were subjected to CIBERSORT ( Newman et al., 2015 Newman et al., 2015 ) to predict the proportion in each tissue. 2.6 Enrichment analysis of cancer TE lncRNAs 2 To evaluate whether the TE lncRNAs identified in cancer patients were significantly enriched in the TE lncRNAs identified in corresponding normal tissues, we performed a hypergeometric test for each cancer type. The P -value was calculated by <math display="block" id="M2"><mi>P</mi><mo>=</mo><mn mathvariant="normal">1</mn><mo>-</mo><mi>F</mi><mfenced separators="|"><mrow><mi>x</mi><mo>-</mo><mn mathvariant="normal">1</mn><mo stretchy="false">|</mo><mi>N</mi><mo>,</mo><mtext> </mtext><mi>M</mi><mo>,</mo><mtext> </mtext><mi>n</mi></mrow></mfenced><mo>=</mo><mn mathvariant="normal">1</mn><mo>-</mo><mstyle displaystyle="true"><munderover><mo largeop="true">∑</mo><mrow><mi>i</mi><mo>=</mo><mn mathvariant="normal">0</mn></mrow><mrow><mi>x</mi><mo>-</mo><mn mathvariant="normal">1</mn></mrow></munderover></mstyle><mrow><mfrac><mrow><mfenced separators="|"><mrow><mfrac linethickness="0pt"><mrow><mi>n</mi></mrow><mrow><mi>i</mi></mrow></mfrac></mrow></mfenced><mfenced separators="|"><mrow><mfrac linethickness="0pt"><mrow><mi>N</mi><mo>-</mo><mi>n</mi></mrow><mrow><mi>M</mi><mo>-</mo><mi>i</mi></mrow></mfrac></mrow></mfenced></mrow><mrow><mfenced separators="|"><mrow><mfrac linethickness="0pt"><mrow><mi>N</mi></mrow><mrow><mi>M</mi></mrow></mfrac></mrow></mfenced></mrow></mfrac></mrow></math> http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491791&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491786&type= 72.81333160 15.66333294 , where N is the total number of lncRNAs, n and M are the numbers of TE lncRNAs identified in cancer patients and corresponding normal tissue, respectively, and x is the number of overlapping TE lncRNAs. We also calculated the odds ratio (OR) by <math id="M3"><mi mathvariant="normal">O</mi><mi mathvariant="normal">R</mi><mo>=</mo><mfrac><mrow><mi>x</mi><mo>/</mo><mi>M</mi></mrow><mrow><mi>n</mi><mo>/</mo><mi>N</mi></mrow></mfrac></math> http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491800&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=39491796&type= 16.25600052 7.28133297 . 2.7 Identification of clinically relevant lncRNAs in cancer 2 To identify clinically relevant lncRNAs in cancer, we detected both differentially expressed and survival-related lncRNAs. First, a t -test was used to identify differentially expressed lncRNAs in each cancer type. We only considered cancer types with more than five normal samples, and lncRNAs with an expression level of 0 in less than 30% of samples were included for subsequent analysis. The lncRNAs with fold changes of > 2 or < 0.5 and adjusted P -value of < 0.05 were identified as differentially expressed in each cancer ( Li YS et al., 2018 Li YS et al., 2018 ). To identify survival-related lncRNAs in cancer, we firstly ranked the cancer patients based on the expression of a specific lncRNA. The top-ranked and bottom-ranked 20% of patients were selected. The difference in the survival time of these two groups was analyzed with the log-rank test and lncRNAs with P -value of < 0.05 were identified as survival-related lncRNAs in cancer. 2.8 Functional enrichment analysis of lncRNAs 2 To predict the function of TE lncRNAs, we detected their co-expressed protein-coding genes via Pearson’s correlation coefficient ( R > 0.4 and P < 0.05). The hypergeometric test was used for detecting enriched Gene Ontology (GO) categories. Then, the enriched significant P -values were adjusted with the Benjamini and Hochberg’s methods, and finally, GO terms with adjusted P -value of < 0.05 were considered. 3 Results 1 3.1 High consistency exhibited by lncRNA transcriptomes within tissues 2 Analysis of the spatial expression of the lncRNA transcriptome across different tissues and organs would greatly improve our understanding of human biology and disease. Five independent datasets that covered distinct methods and conditions were obtained, and provided the opportunity to perform an integrative analysis of the spatial features of lncRNA expression. First, we discovered that approximately 30 tissues were available in each resource, except for HBM2, which included 16 tissues (Figs. 1 1 a and 1 1 b, Table S1). There were totally 9426 normal samples across 38 normal tissues and 11‍‍ ‍093 samples across 33 cancer types. Next, we collected a comprehensive lncRNA annotation, including Ensembl (GRCH37) ( Aken et al., 2017 Aken et al., 2017 ), GENCODE ( Derrien et al., 2012 Derrien et al., 2012 ), and an additional data of a previous study by Cabili et al. (2011) Cabili et al. (2011) . We compared the genomic coordinates of these lncRNAs to obtain non-redundant lncRNA annotation. If two lncRNA loci overlapped by more than 80%, we reserved these lncRNAs from GENCODE. Finally, 15 ‍646 lncRNAs were analyzed. We calculated the average expression of lncRNAs and mRNAs across all tissues and compared the expression distribution of lncRNAs and coding genes. To explore lncRNA expression patterns across this heterogeneous pool of data among different datasets and tissue types, we performed unsupervised clustering based on the expression of lncRNAs with high variation. We discovered that the sample groups of the same tissues from different data resources were clustered together ( Fig. 1 Fig. 1 c). At the same time, these groups of samples were not clustered together in the same way as in the dataset (Fig. S1a). Moreover, the frequent high similarity within the same tissues between the HPA (fresh-frozen tissues) and GTEx (postmortem tissues) indicated negligible effects in the sampling procedures. These results suggested that lncRNA expression exhibited high consistency within the same tissue and could distinguish different tissues more reliably than other factors, including the laboratory of origin. Thus, we are able to present a comprehensive resource of lncRNA expression across multiple tissues and conditions obtained by different methods. 3.2 Higher complexity expressed by lncRNA within specialized tissues 2 The evidence is clear that lncRNA plays a significant role in maintaining the morphology and function of tissues ( Cabili et al., 2011 Cabili et al., 2011 ; Li et al., 2014 Li et al., 2014 ). However, it is unclear how many human lncRNAs exist, where are they expressed, and in what quantities. First, we calculated the average expression of lncRNAs and mRNAs across all tissues and compared the expression distribution of lncRNAs and coding genes. Consistent with previous studies ( Eisenberg and Levanon, 2013 Eisenberg and Levanon, 2013 ; Lingadahalli et al., 2018 Lingadahalli et al., 2018 ), we found that lncRNAs showed lower expression levels in tissues obtained from all four projects ( Fig. 2 Fig. 2 a). The numbers of expressed lncRNAs in each tissue were also far less than that in protein-coding genes (Fig. S2). In an attempt to explore the tissue complexity and composition of the human lncRNA transcriptome, the first question we asked was how many lncRNAs were expressed in a specific tissue. As a result, we found 11‍ ‍035‍–‍13‍ ‍023 lncRNAs (at the expression threshold of 0.1 in GTEx, HPA, and HBM2) expressed in at least one of the 38 studied tissues ( Fig. 2 Fig. 2 b), suggesting that expression of approximately 80% of total lncRNAs seems to account for all basic and specialized functions in the studied tissues. The number of expressed lncRNAs in FANTOM was higher than that obtained from RNA-seq data at a similar proportion of expressed lncRNAs. Moreover, we found that the majority of lncRNAs were consistently expressed in all projects. When a relatively strict threshold of expression was set, we also found a similar proportion of expressed lncRNAs in the GTEx, HPA, and HBM2. Only approximately 1%‍‒‍9% of lncRNAs were detected in individual projects. This may be due to the difference in the detection of different tissues in these projects. We also obtained similar results when we used a relatively strict expression threshold to define the expressed lncRNAs (Fig. S3). The second question we asked was how the various tissues and the lncRNA transcriptome differed in composition and complexity. We calculated tissue complexity by measuring the proportion of the transcriptome contributed by the most abundant lncRNAs at different gradients ( Fig. 2 Fig. 2 c). Notably, the testis, brain, and kidney showed consistently high transcriptome complexity in all resources. These tissues expressed large percentages of lncRNAs in the genome, with a small fraction of the lncRNA pool contributing to the most highly expressed genes. On the other hand, the liver and muscle had a less complex transcriptome, expressing fewer genes in the genome. A large fraction of the transcriptome in these tissues contributed to the most highly expressed genes. For example, the top thousand most highly expressed genes contributed to 96.4% of the lncRNA population in muscle tissue in the HBM2, whereas they contributed less than 62.4% of the lncRNA in testis tissue in the GTEx. Similar trends in transcriptome complexity were reported from previous studies on protein-coding genes in mammals ( Jongeneel et al., 2005 Jongeneel et al., 2005 ; Ramsköld et al., 2009 Ramsköld et al., 2009 ; Zhu et al., 2016 Zhu et al., 2016 ), suggesting conservation of the tissue-controlled expression patterns. We proposed that this complexity in the expression pattern of lncRNAs might be explained in terms of not only the degree of specialization, but also the types of cells in each tissue. For example, the brain has a variety of cells specialized for equally important but different functions. As different cell types express different cell-controlled lncRNAs, tissue as a whole has a large collection of equally important TE genes expressed at comparable rates. 3.3 Spatial classification of lncRNAs based on tissue profiling 2 As described above, lncRNAs exhibit notably higher variation of expression among tissues when compared to protein-coding genes. However, knowledge about their expression patterns across tissues is still limited. We found that more than 80.61% of the lncRNAs were expressed in two or more tissues with a minimum expression value of 0.1. Moreover, 65.95% were expressed in five or more tissues. Thus, the comprehensive collection of a genome-wide lncRNA transcriptome would be an attractive way to establish a refined classification of lncRNAs with regard to their spatial expression patterns; the expression pattern of TE lncRNAs can be explained in terms of tissue complexity ( Salem et al., 2015 Salem et al., 2015 ). In this study, TE lncRNAs were detected only if their expression in one tissue was five-fold or higher compared to the expression values of the other tissues. Collectively, three sub-categories of TE lncRNAs were further stratified to reflect increasing degrees of elevated expression in a particular tissue, including TS, TER, and TEH ( Fig. 3 Fig. 3 a). Based on the definition, TE lncRNAs were identified based on analysis of the four resources of normal tissues ( Figs. 3 Figs. 3 b‒3e), and accounted for 56.47%‒75.38% of the lncRNAs. Notably, there were a higher number of TE lncRNAs in testis and brain tissues. In addition, about one-third of TE lncRNAs were the TS subtype in the GTEx project, and the majority of lncRNAs were TEH across other tissues ( Fig. 3 Fig. 3 b). For the other resources, there were more TEH lncRNAs in the HPA and FANTOM projects ( Figs. 3 Figs. 3 c‍–‍3e). As in the GTEx project, the testis was observed to contain the largest number of TE lncRNAs again, followed by brain, blood, and skin tissues. We also found that the proportion distributions of these three TE lncRNA subtypes in the four resources were highly similar ( Fig. 3 Fig. 3 f). For instance, approximately 44%‍–71% of the TS lncRNAs were shared among different resources. There were higher similarity scores between the GTEx and HPA projects for all three types of TE lncRNAs. Moreover, we obtained similar results when using the higher threshold of expression (Fig. S4). Overall, there was significant overlap in the spatial classification of lncRNAs based on these four independent datasets. 3.4 Closely correlation between robust TE lncRNAs and tissue-related functions 2 The above three types of lncRNA all exhibited high expression in specific tissues. Thus, we combined them and identified robust TE lncRNAs (Section 2.4). To integrate the TE lncRNAs across different resources, we next focused on the 34 tissues that were investigated in at least two projects. In total, we identified 1‍–‍4684 TE lncRNAs across these tissues ( Fig. 4 Fig. 4 a). Similar to our results above, the male reproductive tissue and the testis had by far the highest number of TE genes, followed by the brain. Moreover, the majority (40.7%) of TE lncRNAs in testis tissue were identified in more than two resources. Approximately 12.0% of the TE lncRNAs in testis tissue were observed in all resources. This indicated their robust spatial expression in the testis. Interestingly, these two tissues have been found to have the greatest number of TE protein-coding genes in both the HPA and GTEx ( Uhlén et al., 2016 Uhlén et al., 2016 ). However, the number of TE lncRNAs is greater than the number of protein-coding genes, indicating that lncRNAs have a more important role in these tissues. Functional enrichment analysis of the TER protein-coding genes in the HPA has been performed and the results showed links to the function of each tissue ( Uhlén et al., 2015 Uhlén et al., 2015 ). Similarly, we found that the TE lncRNA testis developmental related gene 1 ( TDRG1 ) was expressed in normal testis tissue in the four resources, and the expression level was higher than that in other tissues ( Fig. 4 Fig. 4 b) ( Jiang et al., 2011 Jiang et al., 2011 ). Functional analysis of their co-expressed genes by the Genomic Regions Enrichment of Annotations Tool (GREAT) ( http://great. http://great. ‍stanford.‍edu/public/html) revealed that they regulated testis-related biological functions, including spermatogenesis, male gamete generation, and spermatid differentiation ( Fig. 4 Fig. 4 c) ( Gan et al., 2016 Gan et al., 2016 ; Wang et al., 2016 Wang et al., 2016 ; Wei et al., 2018 Wei et al., 2018 ). Previous research has also shown that TDRG1 promoted development and migration of sperm and that over-expression might induce tumor development ( Chen et al., 2015 Chen et al., 2015 ). In addition, we found that the numbers of TE lncRNAs in normal brain tissues ranked third; this rank had priority over that based on TE protein-coding genes. An example is SRY-box transcription factor 2 overlapping transcript ( SOX2-OT ), which had enriched expression in the brain ( Fig. 4 Fig. 4 d). Functional analysis suggested that SOX2-OT was involved in chemical synaptic transmission and regulation, regulation of membrane potential, and neurotransmitter secretion ( Fig. 4 Fig. 4 e), and was associated with brain function. It is evident that the expression patterns of lncRNAs were likely to be associated with the phenotype consequences. Conversely, the cervix expressed the lowest number of TE lncRNAs, followed by blood vessel, vagina, duodenum, bone marrow, and then uterus, in increasing order. The uterus, cervix, and ovary have been found to have the least number of TE protein-coding genes; however, the blood vessel has not yet been investigated. What is the contribution of the TE lncRNAs to the transcriptome in different tissues? Although TE lncRNAs were relatively highly expressed in testis, brain, and liver tissues, these lncRNAs significantly contributed to the total cellular lncRNA pool (28.8%, 5.1%, and 1.7% of total lncRNAs, respectively). Conversely, in the cervix, blood vessel, and vagina, which expressed a small number of TE lncRNAs, these genes contributed only 0.0061%, 0.0061%, and 0.0246% of total cellular lncRNA, respectively. These findings indicated the wide variation in the number of genes and regulation of lncRNA expression that determine tissue variation. As described above, these TE lncRNAs showed high consistency across different projects. To determine whether expression profiles of the identified TE lncRNAs could distinguish different normal tissue types, we used the CIBERSORT tool to perform linear support vector regression ( Newman et al., 2015 Newman et al., 2015 ). The expression profiles of TE lncRNAs were used as the feature matrix. We found that the majority of tissue types were recalled based on the expression matrix of TE lncRNAs (Fig. S5). These results suggested that the TE lncRNAs we identified could have critical tissue-specific functions. 3.5 Trend of evalated expression of TE lncRNAs within the same tissues in cancers 2 It has been widely accepted that the identification and appropriate use of TE lncRNAs could provide important insights into disease mechanisms and tissue-specific therapeutic targets. Besides elevated expression in normal tissues, high disease-type specificity has been discovered ( Deras et al., 2008 Deras et al., 2008 ; Gupta et al., 2010 Gupta et al., 2010 ; Hung et al., 2011 Hung et al., 2011 ; Fu et al., 2019 Fu et al., 2019 ; Qu et al., 2022 Qu et al., 2022 ), and genes with disease-restricted expression are highly desirable both as markers and pharmacologic targets because selective expression imparts the specificity required for successful disease-context targeting approaches. To better understand the TE features of lncRNAs for each cancer type and to advance the discovery of clinically related lncRNAs, we also identified lncRNAs with especially elevated expression in the corresponding cancer samples that were absent or expressed weakly in other cancers. Based on lncRNA expression profiles of more than 10 ‍000 cancer patients across 33 cancer types from the TCGA project, we identified the TE lncRNAs in each cancer type. We found that 57.69% of lncRNAs were TE lncRNAs, 4.80% were TS lncRNAs, and there were high proportions of both TER (18.00%) and TEH (77.20%) lncRNAs, unlike the distribution in the normal tissue states (Fig. S6a). In addition, we identified a higher number of TE lncRNAs in acute myeloid leukemia (LAML) than in other solid tumors (Fig. S6b). In solid tumors, there were more TE lncRNAs in brain lower grade glioma (LGG), testicular germ cell tumor (TGCT), and liver hepatocellular carcinoma (LIHC). Interestingly, the corresponding normal tissues (brain, testis, and liver) also had many TE lncRNAs. Next, we further explored the association between TE lncRNAs within the same tissue under normal or cancer conditions. Significantly higher overlaps than in random conditions were discovered through both the OR index and hypergeometric test method across all tissues considered, which is a notable finding ( Fig. 5 Fig. 5 a). In particular, the highest number of TE lncRNAs was found in the brain vs. LGG and glioblastoma multiforme (GBM), followed by testis vs. TGCT (Table S2). This suggested that these TE lncRNAs in normal tissue were likely to exhibit high expression in a cancer state. By functional enrichment analysis of their co-expressed genes, we found that TE lncRNAs in brain tissue under both normal and cancer conditions were involved in axon development, neuron projection morphogenesis, and regulation of axonogenesis ( Fig. 5 Fig. 5 b). For example, SOX2-OT has been demonstrated to be up-regulated during central nervous system development, and is associated with Parkinson’s disease and a number of cancers ( Guo et al., 2021 Guo et al., 2021 ). SOX2-OT is elevated more than 47-fold in normal brain tissue and also has > 32 times higher expression than that in other cancers ( Fig. 5 Fig. 5 c). Similarly, TE lncRNAs in testis tissue under both normal and cancer conditions tend to regulate spermatogenesis and functions associated with DNA epigenetic modification ( Fig. 5 Fig. 5 d). Thus, we conjecture that the strong bias for the high expression of cancer lncRNAs might be related to more active functions during the progression of cancer. 3.6 Strong clinical associations exhibited by TE lncRNAs across cancers 2 Given the significant association of TE lncRNAs within tissues under normal and cancer conditions, a fundamental question is what fraction of TE lncRNAs in tumors is functionally tumorigenic or clinically valuable. To address this question, we focused on cancer TE lncRNAs that showed differential expression or correlations with patient survival. First, we identified the differentially expressed TE (DTE) lncRNAs in each cancer type with more than five normal tissues. A DTE lncRNA is defined by at least two-fold changes and P -value of < 0.05 ( t -test followed by multiple test correction using Benjamini-Hochberg’s method). All 15 of the analyzed cancer types contained DTE lncRNAs, ranging from 21 in ESCA to 256 in LIHC (Table S3). Interestingly, we found that cancer DTE lncRNAs were significantly enriched at the genome-wide level ( Fig. 6 Fig. 6 a, Table S3), indicating that these TE lncRNAs are likely to be associated with the development of cancer. Moreover, we found that a large number of DTE lncRNAs in most cancer types were likely to exhibit up-regulated expression in cancer tissues (Figs. 6 6 a and 6 6 b). For example, 92.97% (119 out of 128) DTE lncRNAs were up-regulated in prostate adenocarcinoma (PRAD). Most of the up-regulated DTE lncRNAs were cancer-constricted and up to 92.01% and 92.06% of DTE lncRNAs were up-regulated and down-regulated in only one cancer type, respectively ( Fig. 6 Fig. 6 c). Among these DTE lncRNAs, only 11.13% are reported to be related with cancers ( Fig. 6 Fig. 6 d), indicating that most DTE lncRNAs in cancers have not yet been described. For example, prostate cancer associated transcript 18 ( PCAT18 ) is elevated more than 16-fold in normal prostate tissue and also has more than 31 times higher expression in PRAD than in other cancers. Interestingly, it has been shown to be more highly expressed in blood in the case of prostate cancer. Tyrosine aminotransferase (TAT) antisense RNA 1 ( TAT-AS1 ) is another example; the expression level is elevated more than 33-fold in normal liver tissue and also more than 11 times higher in LIHC than in other cancers, but its association with disease has not been reported. Thus, these DTE lncRNAs are good candidates for cancer diagnosis as well as investigation of their function in cancer. On the other hand, lncRNA expression has been demonstrated to be associated with cancer patients’ survival ( Du et al., 2013 Du et al., 2013 ; Zhou et al., 2016 Zhou et al., 2016 ; Li YS et al., 2018 Li YS et al., 2018 ). We next explored the extent to which the TE lncRNAs were associated with the survival of patients in each cancer type. A survival-related TE (STE) lncRNA was defined by a P -value in survival analysisof < 0.05 (log-rank test). STE lncRNAs were found in all 33 of the cancers analyzed ( Figs. 6 Figs. 6 e‍–‍6g, Table S3). Interestingly, kidney renal clear cell carcinoma (KIRC) had 155 STE lncRNAs and also had the most DTE lncRNAs. Moreover, unlike DTE lncRNAs, the number of STE lncRNAs in six cancers was relatively small; for instance, no more than ten STE lncRNAs were found. These findings indicated that TE lncRNA had relatively high power as a diagnostic marker than a prognostic one. We also focused on risky STE lncRNAs, high expression of which is associated with poor survival, and found just 476 of them across cancer types. Among these risk STE lncRNAs, just 9.21% are reported to be related to survival, and these include rhabdomyosarcoma 2-associated transcript ( RMST ), MIR155 host gene ( MIR155HG ), DiGeorge syndrome critical region gene 9 ( DGCR9 ), and SOX2-OT . We found that high expression of surfactant associated 1, pseudogene ( SFTA1P ) and SLIT-ROBO Rho GTPase-activating protein 3 antisense RNA 2 ( SRGAP3-AS2 ) was associated with better survival of patients in lung adenocarcinoma (LUAD), and high expression of RP11-63A11.1 and RP11-536I6.2 was associated with better survival of patients in kidney renal papillary cell carcinoma (KIRP) (Fig. S7). These STE lncRNAs provided good candidates for cancer prognostic analysis. In summary, the association of clinically relevant TE lncRNAs with cancer points to a strategy of systematically exploring the relationship between pathology and the spatial expression patterns of lncRNAs across various human tissues. 4 Discussion 1 Gaining a more complete understanding of the relationship between the human genome and phenotypes will require an improved understanding of the nature of their specific or abnormal changes across tissues from health to disease. However, our ability is hindered by the lack of a complete portrayal of lncRNA with spatial expression patterns. In this study, we collected four large-scale expression profiles of lncRNA compiled for all the major normal tissues of the human body, which allowed a comprehensive identification of TE lncRNAs. The tissue-to-tissue correlation of lncRNA expression across these resources revealed high similarity within the same tissue types and dominant distinction among different tissues, rather than similarity linked to other factors such as laboratory of origin, sampling procedures, or detection technologies. Based on expression profiles of protein-coding genes, previous studies had reported high similarity within the same tissues after analyzing the HPA and GTEx ( Danielsson et al., 2015 Danielsson et al., 2015 ; Yu et al., 2015 Yu et al., 2015 ). Thus, the commonly used transcriptomes of normal tissues are valuable resources for lncRNAs, and not just for protein-coding gene analyses. Notably, analysis of the transcriptome by RNA-seq and CAGE is complementary, and it would be an attractive endeavor to integrate data obtained by these two approaches to refine gene expression models. Thus, our integrated analysis of these data resources provides the most comprehensive transcriptome of lncRNAs currently available, and allows the construction of a robust TE lncRNA map to explore the spatial expression patterns of lncRNAs. The precise actions of lncRNAs are frequently dependent on their spatial expression patterns. TE lncRNAs were consistently detected in a genome-wide manner from four independent datasets. More than half of human lncRNAs display TE expression patterns across the human body. Functional analysis of these TE lncRNAs is well in line with their function in the respective tissue or organ. Moreover, about 80.61% of TE lncRNAs are found in at least two tissues, which is compatible with their function pleiotropy across different tissues based on their regulation of different biological functions. Indeed, calculation of the degree of shared TE lncRNAs based on the Jaccard index makes it clear that cancer types originating from similar tissues tend to share TE lncRNAs (Fig. S8). These results showed that TE lncRNAs might be interesting starting points for further in-depth studies to gain a better molecular understanding of the cellular phenotypes that define the function of each respective tissue and organ. Combined with clinical studies, knowledge of TE lncRNAs provides a new means to generate hypotheses of the molecular basis of human disease. TE lncRNAs are expected to underlie many human diseases. Indeed, it has been shown that disease genes generally tend to be expressed in a limited number of tissues. Disentangling TE genes represented in large data compendia, along with their functions in specific tissues, offers means to address these challenges. To better understand the TE features of lncRNAs in disease states, we also identified TE lncRNAs in each cancer type. Interestingly, TE lncRNAs exhibited elevated expression in the corresponding cancer tissues compared to other cancer types, indicating that their higher expression might be a risk factor for disease. We used two criteria to prioritize the clinically related cancer lncRNAs (DTE lncRNAs and STE lncRNAs). The DTE lncRNAs tend to be up-regulated with cancer-constricted patterns, indicating their potential as diagnostic biomarkers. Similarly, TE lncRNAs tend to be related to patient survival in cancer-constricted patterns. Most of these DTE and STE lncRNAs have not yet been described, and could be good candidates for use in cancer diagnosis and prognostic analysis. Here, we provided an atlas of lncRNA marker candidates with elevated expression under both normal and disease conditions, and their clinical correlations ( Fig. 7 Fig. 7 ). In addition, an awareness of the tissues that express elevated lncRNAs could allow a lncRNA found in whole blood or serum to be used as a circulating biomarker for a specific disease. For example, expression levels of TE lncRNAs prostate cancer antigen 3 ( PCA3 ) and PCAT18 are elevated in testis tissue under both normal and cancer conditions, and are much higher in tumor tissue. Expression of PCA3 and PCAT18 was also found to be up-regulated in blood samples of prostate cancer patients ( Ren et al., 2013 Ren et al., 2013 ; Crea et al., 2014 Crea et al., 2014 ; Merola et al., 2015 Merola et al., 2015 ). The TE lncRNA HULC in liver tissue has also been proposed as a circulating biomarker for hepatocellular carcinoma, because it is up-regulated, and can be detected in blood ( Panzitt et al., 2007 Panzitt et al., 2007 ). Furthermore, TE lncRNAs are susceptible to drug interference, which may provide entry points for the design of novel and specific therapeutics. For convenience, we also added TE lncRNAs that are known to act as circulating markers to our online resource LncSpA ( Lv et al., 2020 Lv et al., 2020 ) using information from published databases, including Lnc2Cancer ( Gao et al., 2019 Gao et al., 2019 ), LncRNADisease ( Chen et al., 2013 Chen et al., 2013 ), and exoRBase ( Li SL et al., 2018 Li SL et al., 2018 ). 5 Conclusions 1 In summary, we used a framework to identify lncRNAs with spatial expression patterns based on an integrative transcriptome atlas of lncRNA across human tissues, and this framework can readily incorporate future lncRNA datasets to increase its utility. Spatial classification of lncRNAs is a powerful approach in the analysis of lncRNA function, and can be used as a starting point for elucidating the role of lncRNA in both the maintenance of tissue morphology and the progress of tissue-constricted diseases, crossing the gap between spatial expression patterns of lncRNAs and tissue physiology or pathology. References Aken BL , Achuthan P , Akanni W , et al. , 2017 . Ensembl 2017 . Nucleic Acids Res , 45 ( D1 ): D635 - D642 . https://doi.org/10.1093/nar/gkw1104 https://doi.org/10.1093/nar/gkw1104 Ensembl 2017 Aran D , Camarda R , Odegaard J , et al. , 2017 . Comprehensive analysis of normal adjacent to tumor transcriptomes . Nat Commun , 8 : 1077 . https://doi.org/10.1038/s41467-017-01027-z https://doi.org/10.1038/s41467-017-01027-z Comprehensive analysis of normal adjacent to tumor transcriptomes Burgess DJ , 2015 . Spatial characterization of proteomes . Nat Rev Genet , 16 ( 3 ): 129 . https://doi.org/10.1038/nrg3910 https://doi.org/10.1038/nrg3910 Spatial characterization of proteomes Cabili MN , Trapnell C , Goff L , et al. , 2011 . Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses . Genes Dev , 25 ( 18 ): 1915 - 1927 . https://doi.org/10.1101/gad.17446611 https://doi.org/10.1101/gad.17446611 Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses Chen G , Wang ZY , Wang DQ , et al. , 2013 . LncRNADisease: a database for long-non-coding RNA-associated diseases . Nucleic Acids Res , 41 ( D1 ): D983 - D986 . https://doi.org/10.1093/nar/gks1099 https://doi.org/10.1093/nar/gks1099 LncRNADisease: a database for long-non-coding RNA-associated diseases Chen HY , Sun J , He YQ , et al. , 2015 . Expression and localization of testis developmental related gene 1 (TDRG1) in human spermatozoa . Tohoku J Exp Med , 235 ( 2 ):‍ 103 - 109 . https://doi.org/10.1620/tjem.235.103 https://doi.org/10.1620/tjem.235.103 Expression and localization of testis developmental related gene 1 (TDRG1) in human spermatozoa Colaprico A , Silva TC , Olsen C , et al. , 2016 . TCGAbiolinks : an R/Bioconductor package for integrative analysis of TCGA data . Nucleic Acids Res , 44 ( 8 ): e71 . https://doi.org/10.1093/nar/gkv1507 https://doi.org/10.1093/nar/gkv1507 TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data Crea F , Watahiki A , Quagliata L , et al. , 2014 . Identification of a long non-coding RNA as a novel biomarker and potential therapeutic target for metastatic prostate cancer . Oncotarget , 5 ( 3 ): 764 - 774 . https://doi.org/10.18632/oncotarget.1769 https://doi.org/10.18632/oncotarget.1769 Identification of a long non-coding RNA as a novel biomarker and potential therapeutic target for metastatic prostate cancer Danielsson F , James T , Gomez-Cabrero D , et al. , 2015 . Assessing the consistency of public human tissue RNA-seq data sets . Brief Bioinform , 16 ( 6 ): 941 - 949 . https://doi.org/10.1093/bib/bbv017 https://doi.org/10.1093/bib/bbv017 Assessing the consistency of public human tissue RNA-seq data sets Deras IL , Aubin SMJ , Blase A , et al. , 2008 . PCA3: a molecular urine assay for predicting prostate biopsy outcome . J Urol , 179 ( 4 ): 1587 - 1592 . https://doi.org/10.1016/j.juro.2007.11.038 https://doi.org/10.1016/j.juro.2007.11.038 PCA3: a molecular urine assay for predicting prostate biopsy outcome Derrien T , Johnson R , Bussotti G , et al. , 2012 . The GENCODE v7 catalog of human long noncoding RNAs: analysis of their gene structure, evolution, and expression . Genome Res , 22 ( 9 ): 1775 - 1789 . https://doi.org/10.1101/gr.132159.111 https://doi.org/10.1101/gr.132159.111 The GENCODE v7 catalog of human long noncoding RNAs: analysis of their gene structure, evolution, and expression Du Z , Fei T , Verhaak RGW , et al. , 2013 . Integrative genomic analyses reveal clinically relevant long noncoding RNAs in human cancer . Nat Struct Mol Biol , 20 ( 7 ): 908 - 913 . https://doi.org/10.1038/nsmb.2591 https://doi.org/10.1038/nsmb.2591 Integrative genomic analyses reveal clinically relevant long noncoding RNAs in human cancer Eisenberg E , Levanon EY , 2013 . Human housekeeping genes, revisited . Trends Genet , 29 ( 10 ): 569 - 574 . https://doi.org/10.1016/j.tig.2013.05.010 https://doi.org/10.1016/j.tig.2013.05.010 Human housekeeping genes, revisited Fu PF , Zheng X , Fan X , et al. , 2019 . Role of cytoplasmic lncRNAs in regulating cancer signaling pathways . J Zhejiang Univ-Sci B (Biomed & Biotechnol) , 20 ( 1 ): 1 - 8 . https://doi.org/10.1631/jzus.B1800254 https://doi.org/10.1631/jzus.B1800254 Role of cytoplasmic lncRNAs in regulating cancer signaling pathways Gan Y , Wang Y , Tan ZY , et al. , 2016 . TDRG1 regulates chemosensitivity of seminoma TCam-2 cells to cisplatin via PI3K/Akt/mTOR signaling pathway and mitochondria-mediated apoptotic pathway . Cancer Biol Ther , 17 ( 7 ): 741 - 750 . https://doi.org/10.1080/15384047.2016.1178425 https://doi.org/10.1080/15384047.2016.1178425 TDRG1 regulates chemosensitivity of seminoma TCam-2 cells to cisplatin via PI3K/Akt/mTOR signaling pathway and mitochondria-mediated apoptotic pathway Gao Y , Wang P , Wang YX , et al. , 2019 . Lnc2Cancer v2.0: updated database of experimentally supported long non-coding RNAs in human cancers . Nucleic Acids Res , 47 ( D1 ): D1028 - D1033 . https://doi.org/10.1093/nar/gky1096 https://doi.org/10.1093/nar/gky1096 Lnc2Cancer v2.0: updated database of experimentally supported long non-coding RNAs in human cancers Guo YB , Liu YY , Wang H , et al. , 2021 . Long noncoding RNA SRY-box transcription factor 2 overlapping transcript participates in Parkinson’s disease by regulating the microRNA-942-5p/nuclear apoptosis-inducing factor 1 axis . Bioengineered , 12 ( 1 ): 8570 - 8582 . https://doi.org/10.1080/21655979.2021.1987126 https://doi.org/10.1080/21655979.2021.1987126 Long noncoding RNA SRY-box transcription factor 2 overlapping transcript participates in Parkinson’s disease by regulating the microRNA-942-5p/nuclear apoptosis-inducing factor 1 axis Gupta RA , Shah N , Wang KC , et al. , 2010 . Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis . Nature , 464 ( 7291 ): 1071 - 1076 . https://doi.org/10.1038/nature08975 https://doi.org/10.1038/nature08975 Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis Hon CC , Ramilowski JA , Harshbarger J , et al. , 2017 . An atlas of human long non-coding RNAs with accurate 5' ends . Nature , 543 ( 7644 ): 199 - 204 . https://doi.org/10.1038/nature21374 https://doi.org/10.1038/nature21374 An atlas of human long non-coding RNAs with accurate 5' ends Hung T , Wang YL , Lin MF , et al. , 2011 . Extensive and coord inated transcription of noncoding RNAs within cell-cycle promoters. Nat Genet , 43 ( 7 ): 621 - 629 . https://doi.org/10.1038/ng.848 https://doi.org/10.1038/ng.848 Extensive and coord Jiang XZ , Li DJ , Yang JF , et al. , 2011 . Characterization of a novel human testis-specific gene: testis developmental related gene 1 (TDRG1) . Tohoku J Exp Med , 225 ( 4 ):‍ ‍31 1 -31 8. https://doi.org/10.1620/tjem.225.3 https://doi.org/10.1620/tjem.225.3 11 Characterization of a novel human testis-specific gene: testis developmental related gene 1 (TDRG1) Jongeneel CV , Delorenzi M , Iseli C , et al. , 2005 . An atlas of human gene expression from massively parallel signature sequencing (MPSS) . Genome Res , 15 ( 7 ): 1007 - 1014 . https://doi.org/10.1101/gr.4041005 https://doi.org/10.1101/gr.4041005 An atlas of human gene expression from massively parallel signature sequencing (MPSS) Kim D , Pertea G , Trapnell C , et al. , 2013 . TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions . Genome Biol , 14 ( 4 ): R36 . https://doi.org/10.1186/gb-2013-14-4-r36 https://doi.org/10.1186/gb-2013-14-4-r36 TopHat2: accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions Klijn C , Durinck S , Stawiski EW , et al. , 2015 . A comprehensive transcriptional portrait of human cancer cell lines . Nat Biotechnol , 33 ( 3 ): 306 - 312 . https://doi.org/10.1038/nbt.3080 https://doi.org/10.1038/nbt.3080 A comprehensive transcriptional portrait of human cancer cell lines Li L , Eichten SR , Shimizu R , et al. , 2014 . Genome-wide discovery and characterization of maize long non-coding RNAs . Genome Biol , 15 ( 2 ): R40 . https://doi.org/10.1186/gb-2014-15-2-r40 https://doi.org/10.1186/gb-2014-15-2-r40 Genome-wide discovery and characterization of maize long non-coding RNAs Li SL , Li YC , Chen B , et al. , 2018 . exoRBase: a database of circRNA, lncRNA and mRNA in human blood exosomes . Nucleic Acids Res , 46 ( D1 ): D106 - D112 . https://doi.org/10.1093/nar/gkx891 https://doi.org/10.1093/nar/gkx891 exoRBase: a database of circRNA, lncRNA and mRNA in human blood exosomes Li YS , Li LL , Wang ZS , et al. , 2018 . LncMAP: pan-cancer atlas of long noncoding RNA-mediated transcriptional network perturbations . Nucleic Acids Res , 46 ( 3 ): 1113 - 1123 . https://doi.org/10.1093/nar/gkx1311 https://doi.org/10.1093/nar/gkx1311 LncMAP: pan-cancer atlas of long noncoding RNA-mediated transcriptional network perturbations Li YS , Jiang TTF , Zhou WW , et al. , 2020 . Pan-cancer characterization of immune-related lncRNAs identifies potential oncogenic biomarkers . Nat Commun , 11 : 1000 . https://doi.org/10.1038/s41467-020-14802-2 https://doi.org/10.1038/s41467-020-14802-2 Pan-cancer characterization of immune-related lncRNAs identifies potential oncogenic biomarkers Lingadahalli S , Jadhao S , Sung YY , et al. , 2018 . Novel lncRNA LINC00844 regulates prostate cancer cell migration and invasion through AR signaling . Mol Cancer Res , 16 ( 12 ): 1865 - 1878 . https://doi.org/10.1158/1541-7786.MCR-18-0087 https://doi.org/10.1158/1541-7786.MCR-18-0087 Novel lncRNA LINC00844 regulates prostate cancer cell migration and invasion through AR signaling Lonsdale J , Thomas J , Salvatore M , et al. , 2013 . The Genotype-Tissue Expression (GTEx) project . Nat Genet , 45 ( 6 ):‍ 580 - 585 . https://doi.org/10.1038/ng.2653 https://doi.org/10.1038/ng.2653 The Genotype-Tissue Expression (GTEx) project Lv DZ , Xu K , Jin XY , et al. , 2020 . LncSpA: lncRNA spatial atlas of expression across normal and cancer tissues . Cancer Res , 80 ( 10 ): 2067 - 2071 . https://doi.org/10.1158/0008-5472.CAN-19-2687 https://doi.org/10.1158/0008-5472.CAN-19-2687 LncSpA: lncRNA spatial atlas of expression across normal and cancer tissues Merola R , Tomao L , Antenucci A , et al. , 2015 . PCA3 in prostate cancer and tumor aggressiveness detection on 407 high-risk patients: a National Cancer Institute experience . J Exp Clin Cancer Res , 34 : 15 . https://doi.org/10.1186/s13046-015-0127-8 https://doi.org/10.1186/s13046-015-0127-8 PCA3 in prostate cancer and tumor aggressiveness detection on 407 high-risk patients: a National Cancer Institute experience Newman AM , Liu CL , Green MR , et al. , 2015 . Robust enumeration of cell subsets from tissue expression profiles . Nat Methods , 12 ( 5 ): 453 - 457 . https://doi.org/10.1038/nmeth.3337 https://doi.org/10.1038/nmeth.3337 Robust enumeration of cell subsets from tissue expression profiles Noguchi S , Arakawa T , Fukuda S , et al. , 2017 . FANTOM5 CAGE profiles of human and mouse samples . Sci Data , 4 : 170112 . https://doi.org/10.1038/sdata.2017.112 https://doi.org/10.1038/sdata.2017.112 FANTOM5 CAGE profiles of human and mouse samples Panzitt K , Tschernatsch MMO , Guelly C , et al. , 2007 . Characterization of HULC, a novel gene with striking up-regulation in hepatocellular carcinoma, as noncoding RNA . Gastroenterology , 132 ( 1 ): 330 - 342 . https://doi.org/10.1053/j.gastro.2006.08.026 https://doi.org/10.1053/j.gastro.2006.08.026 Characterization of HULC, a novel gene with striking up-regulation in hepatocellular carcinoma, as noncoding RNA Qu L , He XY , Tang Q , et al. , 2022 . Iron metabolism, ferroptosis, and lncRNA in cancer: knowns and unknowns . J Zhejiang Univ-Sci B (Biomed & Biotechnol) , 23 ( 10 ): 844 - 862 . https://doi.org/10.1631/jzus.B2200194 https://doi.org/10.1631/jzus.B2200194 Iron metabolism, ferroptosis, and lncRNA in cancer: knowns and unknowns Ramsköld D , Wang ET , Burge CB , et al. , 2009 . An abundance of ubiquitously expressed genes revealed by tissue transcriptome sequence data . PLoS Comput Biol , 5 ( 12 ): e1000598 . https://doi.org/10.1371/journal.pcbi.1000598 https://doi.org/10.1371/journal.pcbi.1000598 An abundance of ubiquitously expressed genes revealed by tissue transcriptome sequence data Rawal HC , Angadi U , Mondal TK , 2021 . TEnGExA : an R package based tool for tissue enrichment and gene expression analysis . Brief Bioinform , 22 ( 3 ): bbaa221 . https://doi.org/10.1093/bib/bbaa221 https://doi.org/10.1093/bib/bbaa221 TEnGExA: an R package based tool for tissue enrichment and gene expression analysis Ren SC , Wang FB , Shen J , et al. , 2013 . Long non-coding RNA metastasis associated in lung adenocarcinoma transcript 1 derived miniRNA as a novel plasma-based biomarker for diagnosing prostate cancer . Eur J Cancer , 49 ( 13 ): 2949 - 2959 . https://doi.org/10.1016/j.ejca.2013.04.026 https://doi.org/10.1016/j.ejca.2013.04.026 Long non-coding RNA metastasis associated in lung adenocarcinoma transcript 1 derived miniRNA as a novel plasma-based biomarker for diagnosing prostate cancer Rustici G , Kolesnikov N , Brandizi M , et al. , 2013 . ArrayExpress update-trends in database growth and links to data analysis tools . Nucleic Acids Res , 41 ( D1 ): D987 - D990 . https://doi.org/10.1093/nar/gks1174 https://doi.org/10.1093/nar/gks1174 ArrayExpress update-trends in database growth and links to data analysis tools Salem M , Paneru B , Al-Tobasei R , et al. , 2015 . Transcriptome assembly, gene annotation and tissue gene expression atlas of the rainbow trout . PLoS ONE , 10 ( 3 ): e0121778 . https://doi.org/10.1371/journal.pone.0121778 https://doi.org/10.1371/journal.pone.0121778 Transcriptome assembly, gene annotation and tissue gene expression atlas of the rainbow trout Schroth G , 2011 . RNA-Seq of human individual tissues and mixture of 16 tissues (Illumina Body Map) . BioStudies , E-MTAB-513. Retrieved from https://www.ebi.‍ac.‍uk/biostudies/arrayexpress/studies/E-MTAB-513 https://www.ebi.‍ac.‍uk/biostudies/arrayexpress/studies/E-MTAB-513 RNA-Seq of human individual tissues and mixture of 16 tissues (Illumina Body Map) She XW , Rohl CA , Castle JC , et al. , 2009 . Definition, conservation and epigenetics of housekeeping and tissue-enriched genes . BMC Genomics , 10 : 269 . https://doi.org/10.1186/1471-2164-10-269 https://doi.org/10.1186/1471-2164-10-269 Definition, conservation and epigenetics of housekeeping and tissue-enriched genes The Cancer Genome Atlas Research Network , Weinstein JN , Collisson EA , et al. , 2013 . The Cancer Genome Atlas Pan-Cancer analysis project . Nat Genet , 45 ( 10 ): 1113 - 1120 . https://doi.org/10.1038/ng.2764 https://doi.org/10.1038/ng.2764 The Cancer Genome Atlas Pan-Cancer analysis project Trapnell C , Roberts A , Goff L , et al. , 2012 . Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks . Nat Protoc , 7 ( 3 ): 562 - 578 . https://doi.org/10.1038/nprot.2012.016 https://doi.org/10.1038/nprot.2012.016 Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks Uhlen M , Oksvold P , Fagerberg L , et al. , 2010 . Towards a knowledge-based Human Protein Atlas . Nat Biotechnol , 28 ( 12 ): 1248 - 1250 . https://doi.org/10.1038/nbt1210-1248 https://doi.org/10.1038/nbt1210-1248 Towards a knowledge-based Human Protein Atlas Uhlén M , Fagerberg L , Hallström BM , et al. , 2015 . Tissue-based map of the human proteome . Science , 347 ( 6220 ): 1260419 . https://doi.org/10.1126/science.1260419 https://doi.org/10.1126/science.1260419 Tissue-based map of the human proteome Uhlén M , Hallström BM , Lindskog C , et al. , 2016 . Transcriptomics resources of human tissues and organs . Mol Syst Biol , 12 ( 4 ): 862 . https://doi.org/10.15252/msb.20155865 https://doi.org/10.15252/msb.20155865 Transcriptomics resources of human tissues and organs Wang DX , Eraslan B , Wieland T , et al. , 2019 . A deep proteome and transcriptome abundance atlas of 29 healthy human tissues . Mol Syst Biol , 15 ( 2 ): e8503 . https://doi.org/10.15252/msb.20188503 https://doi.org/10.15252/msb.20188503 A deep proteome and transcriptome abundance atlas of 29 healthy human tissues Wang Y , Gan Y , Tan Z , et al. , 2016 . TDRG1 functions in testicular seminoma are dependent on the PI3K/Akt/mTOR signaling pathway . OncoTargets Ther , 9 : 409 - 420 . https://doi.org/10.2147/OTT.S97294 https://doi.org/10.2147/OTT.S97294 TDRG1 functions in testicular seminoma are dependent on the PI3K/Akt/mTOR signaling pathway Wei JC , Gan Y , Peng DY , et al. , 2018 . Long non-coding RNA H19 promotes TDRG1 expression and cisplatin resistance by sequestering miRNA-106b-5p in seminoma . Cancer Med , 7 ( 12 ): 6247 - 6257 . https://doi.org/10.1002/cam4.1871 https://doi.org/10.1002/cam4.1871 Long non-coding RNA H19 promotes TDRG1 expression and cisplatin resistance by sequestering miRNA-106b-5p in seminoma Xu J , Shao TT , Song MX , et al. , 2020 . MIR22HG acts as a tumor suppressor via TGFβ/SMAD signaling and facilitates immunotherapy in colorectal cancer . Mol Cancer , 19 :‍ 51 . https://doi.org/10.1186/s12943-020-01174-w https://doi.org/10.1186/s12943-020-01174-w MIR22HG acts as a tumor suppressor via TGFβ/SMAD signaling and facilitates immunotherapy in colorectal cancer Yu NYL , Hallström BM , Fagerberg L , et al. , 2015 . Complementing tissue characterization by integrating transcriptome profiling from the Human Protein Atlas and from the FANTOM5 consortium . Nucleic Acids Res , 43 ( 14 ): 6787 - 6798 . https://doi.org/10.1093/nar/gkv608 https://doi.org/10.1093/nar/gkv608 Complementing tissue characterization by integrating transcriptome profiling from the Human Protein Atlas and from the FANTOM5 consortium Zhou CC , Yang F , Yuan SX , et al. , 2016 . Systemic genome screening identifies the outcome associated focal loss of long noncoding RNA PRAL in hepatocellular carcinoma . Hepatology , 63 ( 3 ): 850 - 863 . https://doi.org/10.1002/hep.28393 https://doi.org/10.1002/hep.28393 Systemic genome screening identifies the outcome associated focal loss of long noncoding RNA PRAL in hepatocellular carcinoma Zhu JH , Chen G , Zhu SB , et al. , 2016 . Identification of tissue-specific protein-coding and noncoding transcripts across 14 human tissues using RNA-seq . Sci Rep , 6 : 28400 . https://doi.org/10.1038/srep28400 https://doi.org/10.1038/srep28400 Identification of tissue-specific protein-coding and noncoding transcripts across 14 human tissues using RNA-seq Juan XU and Yongsheng LI designed the study. Kang XU, Xiyun JIN, and Ya LUO analyzed and interpreted the data. Kang XU, Haozhe ZOU, Dezhong LV, Liping WANG, and Yangyang CAI performed the cancer analysis. Kang XU, Limei FU, Tingting SHAO, and Yangyang CAI designed the figures. Kang XU, Juan XU, Yongsheng LI, and Tingting SHAO wrote and edited manuscript. All authors have read and approved the final manuscript, and therefore, have full access to all the data in the study and take responsibility for the integrity and security of the data. This work was supported by the National Natural Science Foundation of China (Nos. 31970646, 32060152, 32070673, and 32170676), the Hainan Province Science and Technology Special Fund (No. ZDYF2021SHFZ051), the Harbin Medical University Marshal Initiative Funding (No. HMUMIF-21024), the Marshal Initiative Funding of Hainan Medical University (No. JBGS202103), and the Heilongjiang Touyan Innovation Team Program. 10.1631/jzus.B2200206 1673-1581（2023）01-0015-17 徐康,靳喜云,罗芽等.长链非编码RNA空间转录组分析揭示组织特异性和癌症中的功能[J].浙江大学学报（英文版）（B辑：生物医学和生物技术）,2023,24(01):15-31. 徐康,靳喜云,罗芽等.长链非编码RNA空间转录组分析揭示组织特异性和癌症中的功能[J].浙江大学学报（英文版）（B辑：生物医学和生物技术）,2023,24(01):15-31. This work was supported by the National Natural Science Foundation of China (Nos. 31970646, 32060152, 32070673, and 32170676), the Hainan Province Science and Technology Special Fund (No. ZDYF2021SHFZ051), the Harbin Medical University Marshal Initiative Funding (No. HMUMIF-21024), the Marshal Initiative Funding of Hainan Medical University (No. JBGS202103), and the Heilongjiang Touyan Innovation Team Program. 2022-04-08 2022-07-13 2023-01-15 2023-02-06 2023-01-11 Copyright © Zhejiang University and Springer-Verlag GmbH Germany, part of Springer Nature 2022 2023 浙江大学出版社 Kang XU, Xiyun JIN, Ya LUO, Haozhe ZOU, Dezhong LV, Liping WANG, Limei FU, Yangyang CAI, Tingting SHAO, Yongsheng LI, and Juan XU declare that they have no conflict of interest. This article does not contain any studies with human or animal subjects performed by any of the authors. Compliance with ethics guidelines Supplementary information 1 Tables S1‍–‍S3; Figs. S1–‍S8 浙江大学学报（英文版）（B辑：生物医学和生物技术） 1 24