Abstract

   Shelterin proteins (TERF1, TERF2, TPP1, TINF2, POT1) protect telomeres,
   prevent unwarranted repair activation, and regulate telomerase
   activity. Alterations in these proteins can lead to cancer progression.
   This study uses an in-silico approach to examine shelterin in tumour
   samples across various cancers, employing mutation plots, phylogenetic
   trees, and sequence alignments. Network pharmacology identified TERF1
   as an essential shelterin protein and transcription factors RUNX1,
   CTCF, and KDM2B as potential biomarkers due to their interactions with
   miRNAs and shelterin proteins. We performed MCODE analysis to identify
   subnetworks of ncRNAs interacting with the shelterin proteins.
   Shelterin expression predicted patient survival in 24 cancer types,
   with TERF1, TERF2, TINF2, and POT1 significantly expressed in
   testicular, AML, prostate, breast and renal cancers, respectively, and
   TPP1 in AML and skin cancer. Spearman and Pearson's analyses showed
   significant correlations of TERF1 across cancers, with near-significant
   correlations for all five proteins in different cancer datasets like
   breast cancer, kidney renal papillary and lung squamous cell carcinoma,
   skin cutaneous melanoma, etc.,. Shelterin expression correlated with
   patient survival in breast, renal, lung, skin, uterine, and gastric
   cancers. Insights into TPP1-associated glycans highlighted glycosylated
   sites contributing to tumorigenesis. This study provides molecular
   signatures for further functional and therapeutic research on
   shelterin, highlighting its potential as a target for anti-cancer
   therapies and promising prospects for cancer prognosis and prediction.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s13062-024-00552-4.

   Keywords: Shelterin proteins, RNAs, Network pharmacology, Expression,
   Patient survival, Correlation

Introduction

   Cells that undergo malignant transformation face the challenge of
   evading replicative senescence and eventual cell death due to
   progressive telomere shortening that occurs with each cell division.
   Telomeres are protected by a specialized protein complex called
   shelterin or the telosome. This complex comprises proteins TERF1,
   TERF2, TPP1, POT1 and TINF2, which work together to prevent telomere
   degradation or fusion. Shelterin plays a critical role in facilitating
   the formation of T-loop structures, recruiting telomerase, and
   protecting telomeres [[26]1]. Moreover, the molecular components of
   telomerase act as additional factors that influence telomeres'
   replicative and protective functions. Often, these components work
   together with shelterin or other core telomere-binding proteins to
   orchestrate the intricate events that govern telomere maintenance and
   stability. When telomere protection is lost, it can lead to chromosome
   end-to-end fusions and a state of telomere crisis, characterized by
   extensive genome instability, chromothripsis, kataegis, and
   tetraploidization, all of which can promote cancer progression. Beyond
   its role in end-protection, shelterin also functions as a mechanism for
   sensing telomere length and regulates telomerase activity at the
   telomere. TERF1 and TERF2 are homodimers that bind to telomeric
   double-stranded DNA, while the POT1–TPP1 heterodimer binds and caps the
   telomeric 3′ tail. TINF2 connects the TERF1 and TERF2 dimers with
   POT1–TPP1, forming the complete shelterin complex (Fig. [27]1a)
   [[28]2]. Aberrant shelterin expression has been observed in various
   cancers like breast cancer [[29]3], adrenal cortical cancer [[30]4],
   colorectal cancer [[31]5], gastric cancer [[32]6–[33]8], glioblastoma
   [[34]9–[35]11], lung cancer [[36]12], prostate cancer [[37]13, [38]14],
   chronic lymphocytic leukaemia [[39]15–[40]17] and melanoma [[41]18]
   with mutations in some shelterin genes acting as cancer drivers.
   Paradoxically, cancer cells (CCs) require adequate shelterin to sustain
   their rapid proliferation, and increased levels of shelterin proteins
   such as TERF1, TERF2, and TINF2 are associated with tumour development.
   This complex relationship between shelterin expression and
   tumorigenesis necessitates further investigation (Fig. [42]1).

Fig. 1.

   [43]Fig. 1
   [44]Open in a new tab

   The components of the shelterin complex, telomerase and the specific
   functions of each shelterin component are highlighted in grey insets.
   Telomeric DNA is made up of both double-stranded and single-stranded
   DNA, and in vertebrates. This duality enables the single-stranded 3′
   tail to enter the double-stranded DNA region, forming a displacement
   loop (D-loop) and a telomere loop (T-loop). Shelterin complexes
   regulate T-loop development, which can prevent telomerase from
   accessing the 3′ tail. Once the telomere is opened, presumably during
   the S phase of the cell cycle, telomerase can connect to the 3′ tail
   using its RNA template and add telomeric repeats. CST then blocks
   telomerase activity, preventing excessive telomere lengthening

   Excessive telomere shortening due to telomerase mutations or severe
   telomere uncapping caused by shelterin dysfunction can trigger a DNA
   damage response (DDR) at chromosome ends, mistaking them for
   double-strand breaks. This can activate the nonhomologous end-joining
   pathway, leading to chromosomal end-to-end fusions, while increased
   homologous recombination might cause rapid telomere length changes and
   terminal deletions [[45]19]. Critically short telomeres fail to recruit
   enough shelterin to suppress checkpoint activation, a model supported
   by evidence that disrupting specific shelterin components like TERF1
   can induce a DDR even without telomere shortening (Fig. [46]2).
   Telomere dysfunction, whether due to critically short telomeres or
   uncapping, triggers a DNA damage response (DDR) through the activation
   of upstream kinases such as DNA-PK, ATM (ataxia-telangiectasia
   mutated), and ATR (ataxia-telangiectasia and Rad3 related) [[47]20].

Fig. 2.

   [48]Fig. 2
   [49]Open in a new tab

   Telomerase dysfunction leads to Cancer: When a p53/p21-dependent cell
   cycle arrest occurs, the damage can be 'repaired' through either the
   non-homologous end-joining (NHEJ) pathway, resulting in chromosomal
   end-to-end fusions or the homologous recombination (HR) pathway,
   leading to telomere length changes and terminal deletions. Both
   pathways induce chromosomal instability, potentially leading to the
   amplification of oncogenes and the loss of tumour suppressor genes,
   thus increasing the risk of cellular transformation and the initiation
   of cancer Cellular Degradation: Activation of the tumour suppressor
   proteins p53 and/or p21 can lead to apoptosis or senescence of cells,
   resulting in tissue degeneration and, ultimately, organ failure

   Most CCs express telomerase, a protein that maintains telomere length,
   which is inactive in most normal cells. CCs preserve telomere length by
   preventing telomere shortening, allowing them to bypass replication
   limits and divide indefinitely, a key characteristic of cancer. Changes
   in the structure and function of shelterin complex proteins can promote
   tumorigenesis and cancer progression. Understanding shelterin complex
   biology suggests that targeting shelterin proteins could be a novel
   strategy for cancer therapy.

   The messenger RNA (mRNA)-microRNA (miRNA)-long noncoding RNA (lncRNA)
   axis plays a crucial role in different cancer hallmarks, including
   inhibition of telomere shortening. This makes them potential biomarkers
   for predicting clinical outcomes, including survival in cancer
   patients. This study focuses on two types of noncoding RNAs (miRNAs and
   lncRNAs) that interact with differentially expressed metastatic mRNAs.
   The miRNAs, which are dysregulated in cancers, contribute to cancer
   hallmarks by modulating various pro- and anti-oncogenic properties.
   These small, conserved, endogenous RNA molecules [[50]21] are 19–25
   base pairs long and regulate gene silencing by binding
   post-transcriptionally to target mRNAs' 3'-untranslated (UTR) region.
   The lncRNAs are RNA transcripts typically longer than 200 nucleotides
   that are not translated into proteins [[51]22].

   Since time immemorial, plant-derived natural compounds and molecules
   have been exploited for their anti-oncogenic properties. Existing
   drugs, including chemotherapeutic agents, often have significant side
   effects, and prolonged use can lead to resistance in CCs. Minimizing
   side effects and preventing chemoresistance can be achieved through
   dosage adjustments or modifications of these drugs [[52]23]. Even with
   dosage modifications, chemotherapeutic agents have demonstrated the
   ability to cause cytotoxicity and result in tumour regression. However,
   only a limited number of drugs specifically target tumour tissues.
   Increasing research indicates that the development and recurrence of
   cancer are influenced not only by physiological factors but also by
   psychological and social aspects of a patient's life.

   Despite decades of extensive research aimed at developing novel
   therapies, effective treatment for cancer remains elusive due to the
   disease's complexity and heterogeneity. CCs often acquire drug
   resistance, and available treatments frequently come with severe side
   effects. Specifically, the study aims to discover novel alternative
   therapeutics and non-coding biomarkers, such as miRNAs and lncRNAs,
   that are significantly involved in maintaining telomeres for cellular
   immortality. This in-silico research may provide new targets for
   validation in experimental setups, potentially leading to more
   effective cancer treatments.

   In a nutshell, in this study, we aim at the detection of significant
   cliques, biomarkers and drugs derived from interaction networks of
   mRNA–miRNA–lncRNA–phytocompound interactions, followed by a multi-omics
   analysis of the shelterin proteins that can be useful and critical for
   an enhanced understanding of the telomerase activity in different
   cancers, which may lead to the identification of newer therapeutic
   strategies as well.

Methodology

Retrieval of significant mRNAs

   Various research and review articles have extracted information about
   the shelterin protein complex with keywords such as ‘shelterin
   proteins’, ‘shelterin protein complex’ and ‘telomeric shelterin
   proteins in cancers’ [[53]24]. Within this complex, TERF1 and TERF2
   bind to double-stranded telomeric DNA, while POT1 binds directly to
   single-stranded DNA. TINF2 and TPP1 serve as bridges between TERF1,
   TERF2, and POT1, stabilizing the entire complex. Although TERF1, TERF2,
   and POT1 directly bind to telomeres, TINF2 and TPP1 indirectly interact
   with telomeres through their associations with TERF1 or TERF2 [[54]25].

Retrieval of non‑coding RNAs

   In this study, we focus on two types of non-coding RNAs: miRNAs and
   lncRNAs. We retrieved common miRNAs interacting with shelterin proteins
   from the ENCORI using the RNA-RNA menu for miRNA-RNA
   ([55]https://starbase.sysu.edu.cn/), [[56]26] miRTarBase
   ([57]https://mirtarbase.cuhk.edu.cn/~miRTarBase/miRTarBase_2022/php/sea
   rch.php) using the search option and then searching for miRNAs using
   the ‘By Target Gene’ option [[58]27] and in TargetScan
   ([59]http://www.targetscan.org/vert_80/) [[60]28] database, the list of
   miRNAs can be downloaded by giving the input of the human gene symbol
   of the respective mRNA. The ENCORI database includes experimentally
   identified RNA–RNA and protein–RNA interaction networks derived from
   108 CLIP-Seq datasets across 37 independent studies. miRTarBase focuses
   on miRNAs, and their high-throughput validated miRNA-target
   interactions, particularly supported by CLIP-Seq verified data.
   TargetScan predicts miRNA targets by examining conserved 8-, 7-, and
   6-mer sites complementary to the miRNA seed region.

   For common lncRNAs interacting with differentially expressed genes
   (DEGs), we utilized two databases: lncBase v3 for retrieving
   interacting lncRNAs for miRNAs by entering the miRNA of interest and
   selecting the specific tissue types) (For eg. hsa-let-7a-5p in bone
   tissue) ([61]https://diana.e-ce.uth.gr/lncbasev3) [[62]29]. ENCORI was
   also used to identify interactions between miRNAs and lncRNAs under the
   ‘miRNA-Target’ drop-down menu.

Retrieval of interacting small compounds

   The drugs interacting with mRNAs are obtained from the Comparative
   Toxicogenomics Database (CTD) ([63]https://ctdbase.org/) by entering
   the gene name in the tab ‘Search by gene’ under ‘Chemical–Gene
   Interaction Query’ [[64]30]. CTD contains manually curated data on
   mRNA–drug interactions that affect various biological pathways
   implicated in diseases such as cancer. For information on drugs
   interacting with miRNAs, we utilized the Sm2miR database
   ([65]http://www.jianglab.cn/SM2miR/) [[66]31] using the miRNA name and
   species, which provides comprehensive data on drugs and small molecules
   that influence miRNA expression and miRNA-associated therapeutics. The
   lncRNA–drug interaction data is retrieved from the D-lnc database
   ([67]http://www.jianglab.cn/D-lnc/) [[68]32] by selecting ‘Homo
   sapiens’ under species and entering the lncRNA of interest in the
   ‘Search by lncRNA’ tab.

Generation of an mRNA–miRNA–lncRNA–drug interaction network, followed by hub
RNA identification, module detection, and transcription factor analysis

   To analyze the interactions within the complex network of
   miRNA–mRNAs–lncRNA–drugs, the Cytoscape v3.9.0 software [[69]33] was
   employed. This freely accessible platform allows the visualization of
   intricate networks involving multiple biological entities. Using the
   'Merge' attribute from the 'Tool' dropdown menu, smaller networks such
   as mRNA–mRNA, mRNA–lncRNA, mRNA–drugs, miRNA–drugs, lncRNA–drugs, and
   miRNA–lncRNA were consolidated. The Maximal Clique Centrality (MCC)
   ranking method from the 'cytohubba' plugin was then utilized to
   identify hub mRNAs within the interaction network. Additionally, the
   'MClique' plugin was used to detect cliques [[70]34].

   The MCODE plugin [[71]35] was applied to the interaction network to
   identify the top subnetworks with an MCODE score of 3 and above to
   discover significant transcription factors (TFs). The most prominent
   carcinogenic conditions were identified by analyzing the highly
   expressed hub genes (rank-wise: deep red denotes the most
   significance), with a cut-off degree of ‘30’ for significant genes from
   the topological parameters like ‘degree’, ‘betweenness’, ‘closeness’,
   ‘radiality’, ‘bottleneck’ and ‘eccentricity’. For mRNAs from the top
   two subnetworks identified by MCODE, interacting miRNAs were searched
   using the ENCORI database. Once the mRNAs and miRNAs were obtained,
   their interacting TFs were retrieved from databases like TRRUST
   ([72]https://www.grnpedia.org/trrust/) corresponding to mRNAs under the
   ‘Search’ tab in the ‘Search a gene in TRRUST database’ [[73]36] and
   TransmiR ([74]http://www.grnpedia.org/trrust) for miRNAs by exploring
   the ‘Search’ tab, giving the miRNA name and organism name ‘Homo
   sapiens’ [[75]37]. The subnetworks were merged to identify the hub TFs
   and the MClique plugin was employed on the interaction network, which
   involved mRNAs, miRNAs, and TFs. ChA3 TF analysis was conducted on all
   TFs obtained from TRRUST and TransmiR to validate the hub TFs
   identified through Clique analysis in MClique within Cytoscape. The
   significance of the TFs was further validated using ChA3 TF analysis
   [[76]38].

Multi-faceted analysis of the shelterin proteins

Gene expression profiling

   The TNMPlot database ([77]https://tnmplot.com/) [[78]39] offers
   expression profiles of shelterin proteins across 25 different types of
   cancer, utilizing nearly 57,000 samples from various RNA-seq and
   microarray datasets. It includes sensitivity/specificity plots based on
   expression cutoffs, illustrating the proportion of tumour samples with
   elevated expression compared to standard samples at different cutoff
   levels (minimum, first quartile, median, third quartile, maximum).
   These bar graphs provide clear insights into the clinical utility of a
   selected gene by highlighting its specificity to tumour cells, which is
   essential for identifying pharmacologically valuable targets. The
   expression profiles of these proteins are displayed cumulatively in box
   plots, and density graphs show the expression profiles of the shelterin
   proteins, comparing normal, tumour, and metastatic samples based on
   their log2 values.

Survival profiles related to the shelterin proteins

   Additionally, we performed a survival analysis of mRNAs using Oncolnc
   ([79]http://www.oncolnc.org/) [[80]40], an interactive platform that
   links cancer patient survival data from TCGA with expression levels of
   mRNA, miRNA, and lncRNA, presented in Kaplan–Meier plots. Tumor samples
   from 503 patients were divided into high- and low-expression groups and
   analyzed using the log-rank test. The statistical significance of the
   selected markers was validated with a p-value < 0.05.

Shelterin protein complex-associated mutation profiles

   Genetic mutations often impact more than just a single protein; they
   can influence the transcription of entire pathways. Comprehensive
   analysis of the mutational profiling of these proteins is carried out
   using Tumor Portal ([81]http://www.tumorportal.org), cBioPortal
   ([82]https://www.cbioportal.org/) [[83]41] and muTarget
   ([84]https://www.mutarget.com/) [[85]42].

   Tumor Portal facilitates the exploration of genes, cancers, mutations,
   and annotations by allowing users to delve into various tumour types,
   genes, and visual representations. cBioPortal provides mutational data
   based on the chromosomal positions of shelterin proteins. This
   integrative platform offers high-quality genetic profiles detailing
   various molecular alterations. This database identifies genetically
   altered mRNAs by examining samples from the TCGA dataset (Cell, 2015),
   revealing different chromosomal mutations, their positions, and base
   pair changes. In addition to the default layout, cBioPortal features
   supplementary bar plots adjacent to the heatmap, displaying the number
   of various modifications for each sample and gene. Mutational data from
   cBioPortal is then visualized using the web-based Circos tool
   ([86]https://circos.ca/software/download/circos/) [[87]43].

   The muTarget database encompasses 18 solid tumour subtypes and 7800
   patient samples, incorporating somatic mutations and RNA-seq gene
   expression data. Based on filters like ‘prevalence at least 3%’,
   ‘p-value cutoff < = 0.05’, ‘fold change cutoff = 1.44’ and ‘FDR =
   < 5%’, muTarget facilitates the identification of biomarkers and
   potential therapeutic targets across various types of solid tumours.

Gene correlation profiles

   Correlation profiles for shelterin proteins are analyzed using the
   TIMER database ([88]http://timer.cistrome.org/) [[89]44], employing
   Pearson and Spearman correlation coefficients. These coefficients are
   widely used to measure relationships between variables: the Pearson
   correlation coefficient evaluates linear relationships, while the
   Spearman correlation coefficient assesses monotonic relationships
   [[90]45, [91]46].

   In the TIMER database, the Gene-Corr module investigates the
   correlation between a gene of interest and a list of other genes across
   various cancer types. The resulting heatmap displays the
   purity-adjusted partial Spearman's rho value, indicating the strength
   of their correlation. Users can select the 'Purity Adjustment' checkbox
   to adjust the association for purity.

O-linked Glycan prediction

   Glycans play a critical role in cancer, serving as a potential source
   for developing new clinical biomarkers. Their ability to influence
   various stages of tumour progression and the biosynthetic pathways
   involved in glycan structures make them a promising target for cancer
   therapy. GlyGen is an open-source web interface that predicts
   shelterin-associated glycans by integrating glycoinformatics tools to
   explore glycoscience data. The GlyGen portal
   ([92]https://www.glygen.org/) [[93]47] currently includes N- and
   O-glycans from human, rat, and mouse species. In the database, from the
   drop down list of ‘Molecule’ the authors have chosen Glycan from the
   drop-down menu and under the ‘From ID type’ we need to select ‘PDB’ and
   then enter the PDB IDs of the shelterin proteins to retrieve the list
   of glycans corresponding to each one.

Functional enrichment analysis

   An automated functional and pathway-enrichment analysis was performed
   on the shelterin protein cluster using hypergeometric testing by the
   STRING database ([94]https://string-db.org/) [[95]48]. The database
   performed overrepresentation tests for gene ontology (GO) terms such as
   biological process (BP), cellular components (CC), Molecular Functions
   (MFs), enriched tissues, and Reactome pathways. Statistically
   significant enriched terms with a p-value < 0.05 were considered along
   with the overlapping genes against a statistical background of the
   entire genome. The false discovery rate (FDR) is a measure that
   describes how significant the enrichment is. The p-values corrected for
   multiple testing within each category using the Benjamini–Hochberg
   procedure.

Gene cooccurrence profiling of shelterin complex

   The STRING database retrieves the gene cooccurrence profile of the
   shelterin proteins across a wide range of taxonomy. In the STRING
   database, the lighter the red hue, the lesser the similarity or
   percentage of conservation of the proteins across the different
   taxonomies. Correlations of these presence/absence profiles can predict
   interactions. Clade coverage for groups of genomes collapsed in the
   phylogenetic tree, two distinct colours indicate the lowest and highest
   similarity observed within that clade.

   Figure [96]3 serves as a comprehensive guide, showcasing this study's
   systematic approach and methodology to achieve the research objectives.
   It emphasizes the sequential flow of tasks and the interconnections
   between different analysis components, ensuring a thorough
   understanding of the investigative process.

Fig. 3.

   [97]Fig. 3
   [98]Open in a new tab

   The fundamental analysis framework involved network pharmacology and
   multi-faceted analysis of the shelterin proteins

Results

Retrieval of interacting non-coding RNAs

   For the 5 shelterin proteins, 2056 interacting miRNAs are retrieved
   from TargetScan and ENCORI databases and 1275 lncRNAs from (mRNA–lncRNA
   interaction data) and ENCORI (miRNA–lncRNA interaction data) are
   considered for the study. 549 small molecules interacting with mRNA,
   miRNA and lncRNA were discovered from CTDbase, Sm2miR and D-lnc
   database, respectively.

Interaction network analysis using cytoscape v3.9

   In Cytoscape, six distinct interaction tables are individually loaded
   to encompass different types of interactions: mRNA–miRNA, miRNA–lncRNA,
   mRNA–lncRNA, mRNA–small molecules, miRNA–small molecules, and
   lncRNA–small molecules. These tables are then merged to form a
   comprehensive interaction network, as depicted in Fig. S1. This network
   visually represents mRNAs, miRNAs, lncRNAs, and drugs by pink hexagons,
   green ellipses, yellow triangles, and blue parallelograms, respectively
   (Fig. S1).

   The CytoHubba and MClique plugins were utilized to analyze the
   interaction network. CytoHubba identified the top 10 hub RNAs,
   including all five shelterin proteins and NEAT1, MALAT1, XIST, SNHG14,
   and KCNQ1OT1. Additionally, the MClique plugin revealed a total of 567
   cliques within the network: the two largest cliques contain four nodes
   each, while the remaining 565 cliques consist of three nodes each.
   Figure [99]4 illustrates the top 10 cliques. Examining the top 20
   cliques, significant mRNAs identified include TERF1 and TERF2 and
   notable lncRNAs include NEAT1. No redundant miRNAs were found among
   these top cliques. Regarding drugs, doxorubicin and diethylstilbestrol
   appear in the two most prominent cliques. Furthermore, bromocriptine is
   present in one of the cliques.

Fig. 4.

   [100]Fig. 4
   [101]Open in a new tab

   The top 10 cliques from the merged network. These cliques are retrieved
   using the MClique cytoscape plugin on the ‘Merged’ version for the
   interaction network of mRNA-miRNA-lncRNA-small compounds. The top 10
   cliques depict the most closely interacting coding and non-coding RNAs
   with various small molecules

Retrieval of sub-networks from the interaction complex network

   MCODE is implemented on the merged interaction network based on the
   haircut algorithm, considering parameters like a k core of 2, a node
   cut-off value of 0.2, and a maximum depth of 100. The clustering score
   was utilised to create sub-networks. Seven sub-networks were retrieved
   using the plugin with an MCODE score cut-off of 3. The sub-networks
   with MCODE scores of 3 and above are taken into consideration for this
   study. Table [102]1 lists the MCODE score of the seven sub-networks.

Table 1.

   The MCODE subnetworks with a score of 3 and above
   Subnetwork ranking MCODE score
   1                  6.154
   2                  3.918
   3                  3.36
   4                  3.429
   5                  3.429
   6                  3.25
   7                  3
   [103]Open in a new tab

Construction of TF–miRNA–mRNA interaction network and TF analysis

   The mRNAs involved in these sub-networks (Fig. [104]5a) are further
   exploited to study a mRNA–miRNA–transcription factor (TF) interaction
   network. For the three mRNAs involved in the seven MCODE subnetworks,
   19 interacting miRNAs, and 290 interacting TFs were retrieved. The
   merged network for mRNA–miRNA–TF is generated (Fig. [105]5b). Fig S2
   represents the hub RNA network with the top 20 RNAs depicting the
   highly potential (based on colour ranking:
   red > orange > mustard > yellow) RNA biomarkers (POT1, TERF1 and TPP1).
   The MClique plugin of cytoscape is utilised to reveal that the TFs,
   RUNX1, CTCF and KDM2B are present in the top three generated cliques
   (Fig. [106]5c). Further, the CheA3 TF analysis reveals that these three
   TFs from the cliques are present among the top 15 integrated mean ranks
   TF analysis clustergram (Fig. [107]5d).

Fig. 5.

   [108]Fig. 5
   [109]Open in a new tab

   The MCODE results and TF analysis. a. The seven subnetworks with MCODE
   score cutoff of 3. b. The merged network comprises the three mRNAs-19
   miRNAs-290 TFs c. The cliques generated from the network having the
   TFs. d. The CheA3 TF clustergram shows the significant TFs among the
   top 15

Multi-factorial analysis of the shelterin proteins

Expression profiling of the shelterin proteins

   The TNM plot represents the significance of enabling a real-time
   comparison of gene expression in ‘Normal’ Vs ‘Tumor’ of the shelterin
   proteins over a horizon of different cancers. TERF1, TERF2, TINF2, and
   POT1 are significantly expressed in testicular, Acute Myeloid Leukemia
   (AML), prostate and breast as well as renal cancers, respectively
   (Fig. [110]6). TPP1 is significantly expressed in AML and skin cancer.
   Additionally, Violin plots for individual cancers are portrayed as
   ‘Normal’ Vs ‘Tumor’ Vs ‘metastatic’ in Fig. S2 and listed in
   Table [111]2.

Fig. 6.

   [112]Fig. 6
   [113]Open in a new tab

   The box plots depict the expression analysis of the shelterin protein
   across 24 different cancers. TERF2 and TPP1 are most highly expressed
   in Acute Myeloid Leukemia (AML), followed by skin, prostate, breast,
   and renal malignancies, and oesophageal and ovarian cancers. TERF2 has
   its most expression in testicular cancer. POT1 seems to be expressed
   similarly in almost all cancers except hepatic cancer

Table 2.

   Compared expression profiles of shelterin proteins across 12 different
   cancers with p-values
    → Cancer Protein Breast ↓ Colon Intestine Kidney Liver Lung
   Oesophageal Oral Ovarian Pancreas Prostate Skin
   TERF1 2.95e-14 3.17e-36 4.48e-01 7.7e-07 1.65e-36 1.16e-22 5.87e-03
   2.39e-06 2.23e-07 1.06e-07 9.25e-05 4.19e-09
   TERF2 1.71e-03 6.7e-08 4.08e-02 1.37e-07 7.15e-07 3.25e-18 7.04e-03
   1.69e-02 1.27e-03 1.56e-07 4.15e-01 1.01e-09
   TINF2 7.4e-27 9.32e-04 6.83e-01 5.39e-04 1.01e-07 3.14e-51 7.04e-12
   9.54e-04 1.28e-06 7.38e-03 7.9e-04 4.8e-0.6
   TPP1 1.31e-01 7.16e-10 1.27e-03 2.18e-01 6.4e-42 7.53e-09 1.09e-21
   5.71e-07 1.62e-13 2.86e-02 1.91e-01 1.83e-31
   POT1 8.5e-07 6.55e-71 2.12e-05 6.46e-35 2.89e-31 6.21e-41 3.26e-22
   2.37e-08 4.67e-04 1.43e-09 4.6e-01 2.12e-18
   [114]Open in a new tab

Survival analysis of the shelterin proteins

   Unleashing an interactive platform, this innovation seamlessly weaves
   together survival data from TCGA with the intricate dance of mRNA,
   miRNA, and lncRNA expressions in cancer patients. Only markers with a
   resounding statistical significance, marked by a P-value < 0.05, made
   the cut. The shelterin proteins were assessed for survival analysis
   across various cancer datasets (Fig. [115]7).

Fig. 7.

   [116]Fig. 7
   [117]Open in a new tab

   Kaplan-Mier plots of the patient survival of the a. TERF1 b. TERF2
   c.TINF2 d. POT1

   The survival profiles of shelterin proteins in respective datasets are:
   TERF1 in BRCA (BReast CAncer gene), KIRP (Kidney renal papillary cell
   carcinoma), LUSC (Lung squamous cell carcinoma), SKCM (Skin cutaneous
   melanoma) and UCEC (Uterine Corpus Endometrial Carcinoma) datasets;
   TERF2 in LUSC, LGG (Low-grade glioma) and KIRC (Kidney renal cell
   carcinoma) datasets; TINF2 in LGG, BRCA and SKCM datasets; TPP1 in STAD
   (Stomach adenocarcinoma), SARC (Sarcoma) and HNSC (Head-Neck Squamous
   Cell Carcinoma) datasets and POT1 in LGG and SKCM datasets are found to
   be significantly associated with patient overall survival potential.
   Hence targeting these respective proteins in respective cancers has the
   potential of improving overall patient survival.

Mutation or polymorphic profiling of the shelterin proteins

   The mutational profiling of the shelterin proteins is assessed via
   three different datasets from three publicly available databases. The
   Tumor Portal database provides plots that annotate mutations associated
   with individual proteins as different legends. POT1 has a nearly
   significant mutation profile in Chronic lymphocytic leukaemia
   (Fig. [118]8a).

Fig. 8.

   [119]Fig. 8
   [120]Open in a new tab

   The mutation plots using a. POT1 is found to be significantly mutated
   in CLL. b. The circos plot represents the mutations associated with
   shelterin proteins at different chromosomal positions

   The circos plot for all the mutations associated with the shelterin
   proteins at different chromosomal positions (Fig. [121]8b) is listed in
   Table S1. The muTarget database represents mutations associated with
   cancers like bladder cancer, breast cancer (TCGA), cervix cancer, colon
   adenocarcinoma, head and neck cancer, renal clear cell carcinoma, renal
   papillary carcinoma, brain lower grade glioma, liver cancer, lung
   adenocarcinoma, lung squamous carcinoma, ovarian cancer, prostate
   cancer, sarcoma, multiple myeloma, melanoma, gastric cancer, Thyroid
   cancer and Uterine cancer that includes all somatic mutations (Fig.
   S3).

Gene-correlation

   In the TIMER database, the correlation among the shelterin proteins is
   determined based on the cut-off of p-value of 0.05. TERF1 is found to
   be significantly correlated with the highest number of cancers. All
   five proteins are found to be nearly significantly correlated as per
   DLBC (n = 48), PAAD (n = 179) and PCPG (n = 181) datasets. TERF1, TERF2
   and POT1 have a ‘near significant’ correlation in THCA (n = 509) and
   THYM (n = 120) datasets. TERF1 and POT1 are significantly correlated as
   per UVM (n = 80) datasets. TINF2 is negatively correlated based on the
   TGCT (n = 150) and LUSC (n = 501) datasets, while TPP1 is also
   negatively correlated as per the UVM dataset (Fig. [122]9).

Fig. 9.

   [123]Fig. 9
   [124]Open in a new tab

   The gene correlation heatmap represents the degree of correlation

Glycans

   The functional significance of O-linked glycosylation for the
   immunosuppression role of TPP1 has encouraged us to check and analyze
   the probable glycosylation sites in TPP1. This is important because the
   variations in glycosylated TPP1 can be a significant factor while
   designing cancer therapeutics against TPP1 for different cancers. Among
   the five shelterin proteins of interest, TPP1 is the only one with 22
   glycosylated domains (Fig. [125]10).

Fig. 10.

   [126]Fig. 10
   [127]Open in a new tab

   The glycosylated domains associated with the only shelterin protein,
   TPP1. The squares represent Nacetyl Hexosamine (GlcNAc), the circle
   galactose, the rhombus/diamond sialic acid, and the triangle fucose.
   The colour coding of each shape signifies a particular type of
   monosaccharide or alteration. Yellow and blue squares frequently
   represent N-acetylgalactosamine (GalNAc), whereas green circles
   typically represent mannose. α and β represent the anomeric
   conformation of the glycosidic bond

   The squares represent N-acetyl Hexosamine (GlcNAc), the circle
   galactose, the rhombus/diamond sialic acid, and the triangle fucose.
   The colour coding of each shape signifies a particular type of
   monosaccharide or alteration. Yellow and blue squares frequently
   represent N-acetylgalactosamine (GalNAc), whereas green circles
   typically represent mannose. α and β represent the anomeric
   conformation of the glycosidic bond. The symbols α and β indicate alpha
   and beta linkages, respectively. Numbers 2, 3, 4, and 6 represent the
   location of the glycosidic bond on the sugar ring, indicating how one
   sugar binds to another. For example, "α2" or "β4" indicate the link
   between specific carbon atoms of two sugars. Multipliers (×) represent
   the amount of a given monosaccharide in a glycan structure. '?' most
   likely denotes an unknown or unidentified component of the structure.
   For example, if it occurs adjacent to a form, it might indicate that
   the kind of monosaccharide or linkage is unclear. The letter 'p' is not
   conventional in glycan notation and may be context-specific. It might
   represent a phosphate group or another change if compatible with the
   diagram's context. Some glycan structures are branching. Glycans
   frequently have branched structures due to numerous connections, and
   branching is necessary for their function in biological systems.
   Branching patterns can potentially alter glycan function, notably in
   cell signalling and chemical recognition. In the case of TPP1, the
   first three are the only fully defined glycosylation sites. The first
   motif is Tn antigen with enzymes such as
   GALNT2,1,3,6,17,18,5,7,10,13,12,15,16,4,6,11,14 and 9. The second one
   has a composition of Hex7 HexNAc2 and enzymes like ALG3,14,1,12,2,9,13
   and DPAGT1. The last one has ALG11,14,1,2,13 and DPAGT1 as enzymes.

   G74724QE, G92275SC, G14669DU, G28681TP, G34989PA, G01485JJ and G20956ZV
   are O-glycans, while glycosphingolipids are G02815KT, G06110VR,
   G00912UN, G74724QE, G92275SC, G14669DU, G66538GV, G72291OX, G28681TP,
   G50282JC, G84349RE and G20956ZV. All but G57321FI and G93180LE are
   N-glycans.

Functional enrichment

   The InterPro database lists the functional gene set enrichment based on
   InterPro and PANTHER analysis. Figure [128]11 represents the PPI and
   Table S2 lists the detailed information regarding the same.

Fig. 11.

   [129]Fig. 11
   [130]Open in a new tab

   Functional enrichment analysis associated with the shelterin protein
   complex

Conservation profile of shelterin proteins

   Utilising the STRING database, the co-occurrence profiling of the
   shelterin proteins was performed across a wide array of taxa. It is
   seen that the sequences of the proteins are most conserved among the
   taxa: Glires (28 taxa), Cercopithecidae (12 taxa), Homininae (4 taxa)
   [Pan (2 taxa), Homo sapiens, Gorilla gorilla], Pongo abelii, Nomascus
   leucogenys, Platyrrhini (4 taxa). Only TPP1 was less conserved in
   Nomascus leucogenys (Fig. [131]12). The darker shade of red the
   cooccurrence profile shows, the more significant it is.

Fig. 12.

   [132]Fig. 12
   [133]Open in a new tab

   The co-occurrence heat-map visualisation of the shelterin proteins
   across different taxa

Discussion

   Shelterin is a well-known protein complex linked to cancer, crucial for
   understanding tumorigenesis and developing telomere-based therapies.
   Recent research suggests that shelterin components significantly impact
   cancer and ageing, with mouse models supporting this theory. However,
   the full complexity of telomere capping structures remains elusive.
   Little is currently understood about the regulation of shelterin
   components during development and in pathological conditions. Changes
   in several shelterin components, such as TERF2, TERF1, and TINF2, have
   been identified in human cancers, highlighting potential therapeutic
   targets to combat cellular ageing and telomerase activity in various
   cancers.

   Traditionally, therapies target individual signalling pathways or
   dysregulated proteins. This study focuses on the shelterin complex's
   role in mediating the function of various non-coding RNAs (ncRNAs)
   involved in cancer development. We employed network-based approaches to
   identify specific cliques involving the interactions of differentially
   expressed mRNAs, miRNAs, lncRNAs, and drugs/small molecules. These
   interactions provide a comprehensive view of the disease environment
   under the influence of interacting RNAs and drugs, suggesting new
   therapeutic strategies. The ncRNAs are often overlooked or treated as
   separate entities in disease mechanisms. However, they interact with
   mRNAs and play a crucial role in disease progression. For example, in
   the absence of a drug, dysregulated levels of ncRNAs can alter coding
   RNAs or protein expression levels, leading to disease. Using this
   network-based approach, we aim to understand the crosstalk between
   mRNAs and ncRNAs, paving the way for future therapeutic interventions
   considering the entire disease environment.

   As discussed previously, conventional drugs are linked to various side
   effects and the development of drug resistance in cancer cells (CCs).
   These complications arise partly due to interactions with approved
   medications. In this context, not only messenger RNAs (mRNAs),
   microRNAs (miRNAs), and long non-coding RNAs (lncRNAs) have emerged as
   promising therapeutic targets, but also three key transcription factors
   (TFs) identified among the top 15 in ChEA3 analysis have shown
   significant potential. Targeting these specific TFs could be crucial in
   combating cellular ageing across different types of cancers. Focusing
   on these non-coding RNAs (ncRNAs) and transcription factors makes
   fine-tuning conventional cancer drug dosages possible. This approach
   aims to minimize adverse effects and enhance the overall effectiveness
   of cancer therapies. The miR139-5p is a prognostic thyroid cancer
   marker involved in HNRNPF-mediated alternative splicing [[134]49]. The
   miR449a has evidence of being a prognostic biomarker in different
   cancers [[135]50]. The miR17-5p is associated with poor prognosis in
   prostate cancer [[136]51]. miR518d-5p utilizes mitochondrial activity
   to overcome liver tumour cell apoptosis [[137]52]. miR340-5p targets
   RhoA, inhibiting colon cancer cell viability and metastasis [[138]53].
   The miR5580-3p inhibits oral cancer cell immortality by suppressing
   LAMC2 [[139]54]. miR671-5p is involved in cellular immortality by
   targeting ALDH2 [[140]55].

   This study's network pharmacology approach reveals significant
   interactions between nine microRNAs (miRNAs) and the long non-coding
   RNA (lncRNA) NEAT1 with shelterin proteins. NEAT1 engages with TFs or
   shelterin complexes to activate the expression of oncogenes or cell
   cycle-regulating genes. NEAT1 promotes the process of EMT, which is
   linked to metastasis and cancer cell invasiveness. NEAT1 may affect the
   shelterin complex's ability to regulate telomerase activity. In certain
   malignancies, telomerase is activated to forestall telomere shortening,
   allowing limitless cell division. Interactions between NEAT1 and
   shelterin may impact this balance, resulting in telomere extension and
   the avoidance of cellular senescence, which is a crucial aspect of
   cancer cell longevity. If NEAT1 affects normal shelterin activity, it
   may cause impaired telomere function and chromosomal imbalances, which
   are common in cancer cells. Chromosomal instability can cause genetic
   diversity in tumors, which contributes to cancer growth and resistance
   to treatments. NEAT1 may regulate TERC (telomerase RNA component) and
   other telomere-associated parameters. The functions of NEAT1 in
   paraspeckle formation and shelterin in DNA damage response may overlap.
   NEAT1 regulates paraspeckles, which store stress-related substances.
   NEAT1 inhibitors could potentially increase the efficiency of
   telomerase inhibitors.

   Shelterin proteins are crucial for the protection and maintenance of
   telomeres, the ends of chromosomes, and their interaction with these
   non-coding RNAs (ncRNAs) suggests that the ncRNAs may modulate the
   function of shelterin proteins. The modulation of these ncRNAs in
   shelterin proteins is further substantiated by evidence showing that
   conventional cancer drugs such as doxorubicin and diethylstilbestrol
   influence these ncRNAs. This indicates a complex interplay where these
   drugs not only target cancer cells directly but also alter the
   expression and function of ncRNAs, which in turn can modulate the
   shelterin proteins.

   Significant transcription factors (TFs) like RUNX1, CTCF, and KDM2B
   have also been identified as potential biomarkers due to their
   interactions with miRNAs and shelterin proteins. These TFs play a
   crucial role in gene regulation and are involved in various cellular
   processes, including cell growth, differentiation, and apoptosis. The
   interactions of these TFs with miRNAs and shelterin proteins further
   highlight their importance in regulating telomere maintenance and
   cellular ageing. RUNX1, for example, regulates gene expression in
   hematopoiesis and has been implicated in several cancers. CTCF is a key
   regulator of chromatin structure and gene expression, playing a role in
   maintaining genome stability. KDM2B, a histone demethylase, is involved
   in chromatin remodelling and gene expression regulation
   [[141]56–[142]58].

   Point mutations in POT1, particularly those affecting its
   oligonucleotide/oligosaccharide-binding (OB) fold domain (POT1
   c.1432G > T (p.Cys476Phe) and c.1166C > T (p.Ala389Val), have been
   found in malignancies including chronic lymphocytic leukaemia (CLL),
   melanoma, and gliomas. Mutations in POT1 impair its capacity to bind
   telomeres, resulting in deprotected telomeres and high telomerase
   activity. POT1 interacts with telomeric Repeat-containing RNA (TERRA)
   to control telomere length. When POT1 is mutated, TERRA levels rise,
   resulting in telomere disruption, linked to various cancer types
   [[143]13, [144]59]. Deletions and point mutations (R425W) in TRF1 have
   been found in malignancies, including lung cancer, glioma, and acute
   myeloid leukaemia (AML). TRF1-deficient cells have weak telomeres prone
   to breaking and recombination, promoting mutation and cancer growth.
   RF1 mutations may attract NEAT1 and MALAT1 to create paraspeckles that
   control alternative splicing and RNA processing in response to telomere
   disruption. Mutations such as TRF2 T241A reduce its ability to prevent
   telomeres from being detected as DNA breaks. Point mutations (K280E) in
   TIN2 have been found in patients with dyskeratosis congenita (DC),
   which is linked to an elevated cancer risk. TPP1 mutations (D224A) are
   linked to family melanoma, aplastic anaemia, and glioblastoma [[145]24,
   [146]60–[147]67].

   By understanding these interactions, researchers can better grasp how
   ncRNAs and TFs contribute to telomere maintenance and the overall
   regulation of cellular ageing and cancer progression. This knowledge
   could lead to developing novel therapeutic strategies targeting these
   ncRNAs and TFs, potentially enhancing the efficacy of existing cancer
   treatments and reducing side effects by enabling more precise
   modulation of drug dosages.

   The expression profiles reveal that shelterin proteins are prominently
   present in several major types of cancer, including acute myeloid
   leukaemia (AML), prostate, breast, renal, testicular, and skin cancers.
   Analysis using violin plots, which depict the distribution of data
   points, indicates that shelterin proteins are significantly expressed
   not only in normal and tumour states but also in the metastatic states
   of these cancers.

   Furthermore, overall survival profiles indicate a strong association
   between the expression of all five shelterin proteins and patient
   survival across multiple cancer types, such as renal, breast, head and
   neck, lung, uterine, and stomach cancers. This correlation suggests
   that higher or altered levels of shelterin proteins can impact patient
   prognosis.

   Mutations or variations in shelterin proteins are known to disrupt
   various cancer-related pathways. These disruptions can lead to
   dysregulated cellular processes, including accelerated telomere
   lengthening, which supports the concept of cellular immortality—a
   hallmark of cancer progression. Therefore, specific mutations in
   shelterin proteins are significant contributors to cancer
   tumorigenesis.

   The conservation of shelterin proteins—a highly conserved complex
   across eukaryotic species—plays a crucial role in maintaining telomere
   integrity and chromosomal stability in normal cells. This evolutionary
   conservation ensures that the fundamental mechanisms of telomere
   protection and replication are maintained, preserving genomic stability
   across diverse cell types and organisms. The heatmap generated in the
   analysis highlights these overlaps, where darker shades represent genes
   involved in multiple cancers. Except for TINF2 in lung and testicular
   cancers and TPP1 in uterine cancers, shelterin proteins correlate
   positively across all 40 cancer types analyzed. The conservation of
   shelterin proteins is critical for telomere integrity and genomic
   stability in normal cells. These proteins, which include TRF1, TRF2,
   POT1, TIN2, TPP1, and RAP1, prevent telomeres from being identified as
   DNA damage, regulate telomerase activity, and prevent ineffective DNA
   repair mechanisms. This ensures controlled cell division in normal
   cells while also preventing premature senescence or apoptosis. However,
   in malignant cells, mutations or changed expression of shelterin
   components compromise telomere protection, resulting in genomic
   instability, a cancer hallmark. Mutations in proteins such as POT1 or
   TRF2 can cause excessive telomerase activity or telomere shortening,
   resulting in uncontrolled cell proliferation and chromosomal
   rearrangements that promote cancer. Thus, whereas shelterin complex
   conserved functions are critical for normal cell biology, their
   dysregulation in cancer drives oncogenesis.

   The study by Luo et al. (PMID: 33,497,432) [[148]68] provides a
   comprehensive pan-cancer analysis of shelterin proteins, investigating
   their expression patterns, mutations, and prognostic value across
   multiple cancer types. POT1, TRF1, and TRF2 were generally upregulated
   in most cancers. Frequent mutations in shelterin genes result in
   several cancers, including melanoma, lung adenocarcinoma, and
   colorectal cancer, especially POT1 mutations are prevalent. High
   expression of TRF2 and TIN2 was correlated with worse survival outcomes
   in lung cancer and glioblastoma, suggesting that shelterin protein
   levels can have prognostic significance. In this study, the
   comprehensive examination of shelterin proteins underscores their
   critical role in cancer biology. Their significant expression in
   various cancer states and their association with patient survival point
   to their potential as biomarkers for prognosis and therapeutic targets.
   The genetic correlations observed across different cancers suggest a
   common underlying mechanism involving shelterin proteins, further
   emphasizing their importance in cancer research and treatment
   strategies.

   Glycosylation, the most intricate post-translational modification of
   proteins, plays a crucial role in cancer pathogenesis and progression.
   Altered glycans on the surfaces of both tumour and host cells and
   within the tumour microenvironment have been identified as crucial
   mediators of critical events in these processes. According to Ogata et
   al., malignant cells are notably more enriched with highly branched
   complex-type N-linked sugar chains compared to their regular
   counterparts [[149]69]. TPP1's unique status as the only shelterin
   protein with 22 different glycosylation sites across various cancers
   highlights its potential significance in cancer biology. The
   modification of TPP1 through glycosylation likely plays a critical role
   in telomere maintenance, protein stability, cellular localization, and
   the aggressive characteristics of cancer cells, making it a promising
   focus for further research and therapeutic development.

Conclusion

   Current oncogenesis research is increasingly focusing on the shelterin
   complex. Various studies have identified the involvement of shelterin
   proteins in different tumours, including lymphocytic leukaemia,
   squamous cell carcinoma, non-small cell lung cancer, and breast cancer.
   Notably, the expression levels of certain shelterin components, such as
   TRF2, TRF1, and TIN2, have been found to be altered in many cancers.
   Some shelterin proteins also play crucial roles as either telomerase
   recruitment factors (POT1 and TPP1) or negative regulators of telomere
   length (TRF1 and TRF2). Understanding the role of the shelterin complex
   in telomere maintenance during tumorigenesis is essential. The data
   discussed underscore the potential of shelterin components as
   attractive targets for future anti-cancer therapies, offering promising
   prospects for prognosis and the prediction of cancer aggressiveness.
   Given their connection with telomere and telomerase biology, shelterin
   proteins could be pivotal in developing combined strategies and
   synergistic therapeutic interactions for cancer treatment.

Supplementary Information

   [150]Additional file 1^ (8.7MB, docx)

Author contributions

   Conceptualization: S.B. and S.C.; Methodology: S.C and S.B.; Formal
   analysis: S.C.; Data curation: S.C.; Writing—original draft: S.C.;
   Writing—review & editing: S.C and S.B.; Supervision: S.B.

Funding

   Open access funding provided by Vellore Institute of Technology.

Availability of data and materials

   The authors confirm that the data supporting the findings of this study
   are available within the article, and its supplementary and publicly
   available data sources have been mentioned in the manuscript. For ex.
   [151]https://tnmplot.com/, [152]http://www.oncolnc.org/,
   [153]http://www.tumorportal.org, [154]http://timer.cistrome.org/,
   [155]https://www.cbioportal.org/,
   [156]https://circos.ca/software/download/circos/

Declarations

Ethics approval and consent to participate

   This article contains no studies with human participants or animals
   performed by any authors.

Competing interests

   The authors declare no competing interests.

Footnotes

   Publisher's Note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

References