Long Non-Coding RNAs (lncRNAs) in Response and Resistance to Cancer Immunosurveillance and Immunotherapy

Long non-coding RNAs (lncRNAs) are critical regulatory elements in cellular functions in states of both normalcy and disease, including cancer. LncRNAs can influence not only tumorigenesis but also cancer features such as metastasis, angiogenesis and resistance to chemo-and immune-mediated apoptotic signals. Several lncRNAs have been demonstrated to control directly or indirectly the number, type and activities of distinct immune cell populations of adaptive and innate immunities within and without the tumor microenvironment. The disruption of lncRNA expression in both cancer and immune cells may reflect alterations in tumor responses to cancer immunosurveillance and immunotherapy, thus providing new insights into lncRNA biomarker-based prognostic and therapeutic cancer assessment. Here we present an overview on lncRNAs’ functions and underlying molecular mechanisms related to cancer immunity and conventional immunotherapy, with the expectation that any elucidations may lead to a better understanding and management of cancer immune escape and response to current and future immunotherapeutics.


Introduction
Cancer remains the second leading cause of mortality worldwide after cardiovascular diseases [1]. Cancer as a genetic disorder has its molecular base in the accumulation of multiple mutations, mostly in genes that control directly or indirectly cell survival and proliferation, as well as the inability of the host immune-surveillance mechanisms to recognize and kill the cancer cells [2][3][4]. Although novel chemo-and immuno-therapeutics have prolonged overall patient survival and quality of life, disease relapses frequently occur due to failure of long lasting anti-tumor responses [5][6][7]. Most tumor types, as they progress, may further acquire additional genetic alterations that confer to the formation of cancer phenotypes resistant to conventional treatment modalities and highly potent for immune-evasion [3]. The list of molecular mechanisms that have so far been identified as responsible for the acquisition of tumor resistance to endogenous immune-mediated cytotoxicity and conventional chemo-and immune-therapeutics is quite long, and it is constantly being enriched [8][9][10][11]. Given the complexity of these mechanisms and their tumor specificity that is often observed, the molecular basis of resistance and the ways to overcome it need to be better elucidated and explored on multiple levels. As such, there is enormous space for research on identifying novel molecular signatures of tumor unresponsiveness to external and internal apoptotic signals that may be putative candidates for therapeutic targeting.
A large part of the human genome is transcribed into long non-coding RNAs (lncR-NAs), which are defined as >200 nt long transcripts [12]. While more than 90% of lncRNAs lack protein-coding potential, it has been reported that some transcripts might encode phages, neutrophils and other innate immune cells via induction of "don't eat me" nals (e.g., CD47, CD73), while by downregulating chemokines and stress ligands tu cells may poorly respond to attraction of and recognition by NK cells. In addition, ponents of the humoral anti-tumor response such as the complement system-med cytotoxicity is counteracted by expression of neutralizing complement regulatory teins (e.g., CD46/55/59) [27]. The above mechanisms of resistance are summarized matically in Figure 1. Cancer cells escape host immunosurveillance by multiple mechanisms associated either with their phenotypic characteristics or with different immune cell populations of adaptive and innate immunity involved in the induction or suppression of a successful anti-tumor immune response. These mechanisms may include reduced expression of MHC class I, TAA, TSAs in cancer cells and/or upregulation of inhibitory immune checkpoint molecules on their surface. Alternatively, tumor cells may affect CD4+ and CD8+ T cell infiltration to TME by secreting factors, like TGF-β, while they can trigger the polarization of effector CD4+ T cells to Treg and/or Th2 phenotypes. In parallel, they can promote CTL exhaustion and apoptosis, accompanied by reduced CTL-associated expression of MHC class I/II complexes and cytotoxic activities. Concomitantly, the functions of innate immune cell populations including Mφ, NPs and NK cells are significantly diminished, either due to accumulation of suppressing molecules like Cancer cells escape host immunosurveillance by multiple mechanisms associated either with their phenotypic characteristics or with different immune cell populations of adaptive and innate immunity involved in the induction or suppression of a successful anti-tumor immune response. These mechanisms may include reduced expression of MHC class I, TAA, TSAs in cancer cells and/or upregulation of inhibitory immune checkpoint molecules on their surface. Alternatively, tumor cells may affect CD4+ and CD8+ T cell infiltration to TME by secreting factors, like TGF-β, while they can trigger the polarization of effector CD4+ T cells to Treg and/or Th2 phenotypes. In parallel, they can promote CTL exhaustion and apoptosis, accompanied by reduced CTL-associated expression of MHC class I/II complexes and cytotoxic activities. Concomitantly, the functions of innate immune cell populations, including Mϕ, NPs and NK cells are significantly diminished, either due to accumulation of suppressing molecules like CD47 and CD73, or lack of potent chemoattractors and stress-related ligands, or inefficient target recognition. Cancer cells may also negatively interfere with DC maturation and complement dependent cytotoxicity (CDC), mainly by expression of neutralizing complement regulatory proteins. The immunosuppressive effects in TME are further augmented by cancer cell-mediated MDSCs recruitment and polarization to immunosuppressive cell phenotypes. Abbreviations used: MHC I/II, Major histocompatibility complex classI/or classII; TAA, Tumor associated antigens; TSA, Tumor specific antigens; MDSCs, Myeloid-derived suppressor cells; DCs, Dendritic cells; Mϕ, Macrophages; NPs, Neutrophils; NK, Natural killers; TME, Tumor microenvironment; Tregs, T regulatory cells; Th2, Type 2 helper; CTL, Cytotoxic T lymphocyte; TGF-β, transforming growth factor beta.

LncRNAs and Cell-Mediated Anti-Tumor Immunity
The expression patterns of lncRNAs are precise, highly cell type-specific and contextspecific, as well as dependent on the stage of development and cell differentiation [31]. This is important for the smooth function of the immune system, which relies on the self-renewal and differentiation of hematopoietic stem cells (HSCs), where lncRNAs might serve as mediators between extracellular signals and the activity of transcriptional factors that determine the cell differentiation path and fate [32]. Common lymphocyte progenitor cells are differentiated into specific NK, B and T cell lineages, each expressing a different combination of lncRNAs. LncRNAs have been reported to play a significant role during the development, differentiation and activation of CD8+ T cells [33][34][35][36], CD4+ T cells [33][34][35][36][37] and NK cells [33,35,36].
Dysregulated patterns of lncRNA expression in tumor cells and different immune cell populations, including populations of both innate and adaptive immunities, have been directly and indirectly associated with the survival, activation and cytotoxic potency of effector immune cells involved in the anti-tumor responses. In the following paragraphs we discuss the pleiotropic effects of different lncRNAs in tumor response to innate and adaptive immunity, emphasizing on the nature of lncRNA transcriptome in the involved immune cell populations. • CD4+ T cell-expressed lncRNAs The lncRNAs small nucleolar RNA host gene 1 (lnc-SNHG1), epidermal growth factor receptor (lnc-EGFR) and insulin receptor precursor (lnc-INSR) have been reported to reduce the cytotoxic activities of CD8+ T cells within the TME, by accelerating the polarization of tumor infiltrated CD4+ T lymphocytes (TILs) into regulatory Tregs (CD4+/CD25+/Foxp3+ T cells) [38][39][40]. Specifically, in breast cancer (BC) models, upregulation of lnc-SNHG1 in CD4+ TILs promotes their differentiation into Tregs by interacting with miR-448, a negative regulator of indoleamine 2,3-dioxygenase (IDO). Induction of IDO by lnc-SNHGI-mediated inhibition of miR-448 results in increased CD4+ T cell differentiation into Tregs, and eventually in suppression of CTLs and induction of their apoptotic death [38]. Concomitantly, immune evasion of hepatocellular carcinoma (HCC) is supported by the lnc-EGFR, whose upregulation in tumor infiltrating CD4+ T cells also leads to their polarization into Treg cells. Lnc-EGFR binds to EGFR, thus sustaining its preservation and activation. EGFR activation in turn induces the AP1/NF-AT1 axis, which acts as a forward-feedback loop reinforcing lnc-EGFR expression and enhancing the transcription of EGFR and forkhead box protein P3 (FOXP3) genes. While their number increases, Tregs downregulate the action and propagation of effector T-cells within the TME, resulting in immunosuppression, toleration and progression of HCC cells [39]. On the same motif, abnormal overexpression of lnc-INSR in bone marrow infiltrated CD4+ T cells in pediatric T cell acute lymphoblastic leukemia (T-ALL) modulates the immune microenvironment and results in Treg increase and a decline in CTLs, thus allowing the tolerance of malignant cells in the bone marrow. Being co-localized with the INSR both in the membrane and the cytoplasm, lnc-INSR restrains the ubiquitination of INSR, thus sustaining INSR activation and stability. This leads to an enhanced activation of the PI3K/AKT-signaling pathway that promotes Treg cell differentiation and an immunosuppressive microenvironment [40].

• CD8+ T cell-expressed lncRNAs
In lung and breast cancer models overexpression of the NF-κB-interacting lncRNA (NKILA) in tumor-infiltrating CTLs and Th1 cells leads to their activation-induced cell death (AICD) through attenuation of NF-κB-dependent anti-apoptotic gene transcription. Briefly, NKILA transcriptional activity is significantly induced during the late-sensitive phase of T cell activation, after TCR stimulation triggers the recruitment of histone acetyltransferases to the NKILA promoter, which allow chromatin opening and binding of the transcriptional activator of NKILA, signal transducer and activator of transcription 1 (STAT1). While becoming upregulated, NKILA binds to NF-κB and represses its activity on downstream anti-apoptotic target genes, thus forcing effector CD8+ T cells to undergo AICD and tumor cells to escape CTL-mediated cytotoxicity [46]. In addition, a recent study exploiting the fundamental mechanisms by which CD8+ T cells are regulated in response to pathogens and potentially cancer, demonstrated evidence for a critical role of lncRNA Morrbid in CD8+ T cell expansion, survival and effector function by modulating the expression of the pro-apoptotic factor Bcl-2-like protein 11 (Bcl2111) and the activity of the PI3K-AKT signaling pathway [47]. Concomitantly, in colorectal cancer (CRC) xenograft mouse models induction of lncRNA GM16343 in IL-36-stimulated mouse CD8+ T cells increased their interferon gamma (IFN-γ) secreting ability in the TME and reduced tumor diameter, thus suggesting a critical role of lncRNA GM16343 in reinforcing the CD8+ T cell-mediated anti-tumor immune response [48].
• DC-expressed lncRNAs DCs, as professional APCs, play a pivotal role in regulating the balance between CD8+ T cell-mediated immunity and tolerance to tumor antigens. DCs activate CD8+ T cells by cross-presenting non-self-antigens, a process known as cross priming. Cross priming is thought to be critical in generating potent anti-tumor CTL responses [49]. LncRNA dysregulation in DCs can modulate their anti-tumor function by affecting their tumor infiltration, differentiation and metabolism, as well as their ability to cross-prime CD8+ T cell-mediated responses [50]. A characteristic example of a DC-associated lncRNA is the lnc-DC, which promotes DCs' differentiation and increases their ability to stimulate T cell activation by activating STAT3 [51]. Studies on lnc-DC silencing, upon hepatitis B virus (HBV) infection, showed that lnc-DC inhibits DCs' maturation and controls virally induced immune responses. The underlying mechanism of action includes lnc-DC-mediated reduction in IFN-γ, IL-6, IL-12 and tumor necrosis factor alpha (TNF-α) secretion and increase in IL-1β concentration in DCs, via targeting the TLR9/STAT3 signaling pathway [52]. Lnc-Dpf3 induction in DCs inhibits DC migration and glycolytic metabolism, by decreasing the hypoxia inducible factor 1 subunit alpha (HIF-1a)-mediated glycolysis [53].

MDSC-expressed LncRNAs
The expansion of myeloid-derived suppressor cells (MDSCs), a heterogenous group of immature immune cells from the myeloid lineage, has been associated with immuno- suppressive effects in TME and cancer progression [68,69]. Most of the tumor-derived MDSCs have lost their ability to differentiate into mature granulocytes, monocytes and macrophages, while they accelerate tumor progression by dramatically blocking T cellinduced antitumor responses and enhancing Tregs through release of Arg1, reactive oxygen species (ROS) and inducible nitric oxide synthase (iNOs) [68,[70][71][72]. Interestingly, recent studies in the peripheral blood of LC patients revealed that the percentage of MDSCs was negatively correlated with the percentage of Th1/CTL cells, thus suggesting that MDSCs might pose a suppressive effect not only to CTL responses, but also to CTL numbers [69,73]. Although, the involvement of lncRNAs in the regulation of the immunosuppressive function of MDSCs has recently begun to be exploited, early indications reveal a critical role for lncRNAs in MDSC expansion and function in cancer [71]. Upregulation of lncRNA C/EBPβ and C/EBP homologous protein (CHOP) in MDSCs induces expression of immunosuppressive factors, such as NO synthase 2, NADPH oxidase 2, Arg1 and cyclooxygenase-2, via interactions with CHOP and liver-enriched inhibitory Cells 2021, 10, 3313 9 of 26 protein (LIP), leading to activation of CCAAT-enhancer-binding protein (C/EBPB) and augmentation of tumor growth in many cancers [74]. Furthermore, lncRNAs pseudogene Olfr29-ps1 and retinal noncoding RNA 3 (RNCR3) regulate MDSC differentiation and promote their immunosuppressive phenotype by modulating the Olfr29-ps1/miR-214-3p/MyD88/M6A regulatory network [75] and RNCR3/miR-185-5p/Chop pathway, respectively [76]. LncRNA pseudogene Olfr29-ps1 also contributes to upregulation of both inflammatory and tumor-associated factors in the microenvironment of melanoma (MM) tumors [75]. Similarly, lncRNAs plasmacytoma variant translocation 1 (Pvt1) and RUNX1 overlapping RNA (RUNXOR) accelerated MDSC-mediated immunosuppression in LC in in vitro and in vivo models [77,78]. Conversely, other lncRNAs such as HOXA transcript antisense RNA myeloid-specific 1 (HOTAIRM1) and metastasis associated lung adenocarcinoma transcript 1 (MALAT1), have been reported to negatively affect the proportion of MDSCs in peripheral blood mononuclear cells (PBMCs) and downregulate their immunosuppressive activity in cancer [69,73]. High levels of HOTAIRM1 in LC patient-derived MDSCs could inhibit the development of MDSCs and the expression of MDSC-associated suppressive molecules, via induction of homeobox A1 (HOXA1), a molecule known to delay tumor progression and enhance the anti-tumor immune response by downregulating the immunosuppressive activity of MDSCs [69]. Accordingly, in the same tumor model, lncRNA MALAT1 was shown to play a direct role in inhibiting MDSC expansion, while accelerating CTL proportion within TME [73].
Overall, although for some of the aforementioned lncRNAs there is not yet direct association with a specific cancer type, their contribution in the regulation of immune cell properties and functions open new insights in exploring their role in the anti-tumor immunity.

Cancer Cell-Associated lncRNAs
A plethora of lncRNAs overexpressed in tumor cells may also modify the survival and activities of effector immune cells, via similar mechanisms to those described above. A list of such RNAs is shown in Table 2.
Overexpression of lncRNA SRY-box transcription factor 5 (lnc-sox5) in CRC cells leads to upregulation of IDO1 enzyme, which stimulates the differentiation of Tregs and eventually their infiltration into the tumor microenvironment. Tregs along with lnc-sox5 IDO1 upregulation may in turn suppress CD8+ T cell cytotoxicity and intensify their exhaustion by activation of Fas/FasL-mediated apoptosis [83]. Similarly, the long noncoding RNA for kinase activation (LINK-A), expressed in 25% of triple-negative breast cancer (TNBC) patients, is thought to be a negative regulator of APC and effector CD8+ T cell recruitment and infiltration within the TME. In addition, LINK-A induction has been associated with downregulation of β-2M and MHC-I expression in TNBC cells, via degradation of peptide loading complex (PLC), thus resulting in insufficient tumor antigen presentation by cancer cells and CTL unresponsiveness [84]. Moreover, three lncRNAs, myocardial infarction associated transcript (MIAT), LINC01297 and MYLK antisense RNA (MYLK-AS1), are included, together with four miRNAs (hsa-miR-200a-3p, hsa-miR-455-5p, hsa-miR-192-5p, has-miR-215-5P), in the underlying regulatory network of Ubiquitin Associated and SH3 Domain Containing B (UBASH3B), a protein which is involved in the infiltration of several immune cell types (macrophages, neutrophils, B and DCs, CD4+ and CD8+ T lymphocytes), in the TME of PC [85].
Recent findings provide evidence for a novel activity of tumor-derived exosomal lncR-NAs in interacting with immune cells, especially macrophages and Tregs, and regulating their immunological function. LncRNA small nucleolar RNA host gene 16 (lnc-SNHG16) detected in breast cancer-derived exosomes has been linked to CD73+ γδ1 Treg cell induction and consequently to TME immunosuppression. The underlying mechanism involves contact-free transmission of lnc-SNHG16 from the indicated exosomes to γδ1 T cells where it upregulates CD73 expression in a miR-16-5p/TGF-β1/Smad5-dependant manner [90]. In addition, overexpression of lncRNA ribonuclease P RNA component H1 (RPPH1) in CRC promotes tumor cell proliferation and metastasis by interacting with the epithelial to mesenchymal transition (EMT)-inducer tubulin beta 3 class III (TUBB3) and facilitating exosome-mediated macrophage M2 polarization [91]. Similarly, HCC-derived exosomal lncRNA TUC339 has been reported to regulate both macrophage activation and M1/M2 polarization [92]. The lncRNA X-Inactive Specific Transcript (XIST) is known for its regulatory role in breast cancer brain metastases (BCBM), yet its immunostimulating functions have only recently been discovered. Loss of XIST in mammary glands of mice not only promoted tumor cell proliferation, EMT and brain metastases, but also augmented secretion of exosomal miRNA-503, which triggered M1-M2 polarization of microglia with consequent induction of immune suppressive cytokines that inhibited T cell proliferation and function [93].
Expression of another BCBM regulator, the lncRNA associated with BCBM (lnc-BM), in BC cells has been associated with recruitment of macrophages in the brain via STAT3dependent induction of C-C motif chemokine ligand 2 (CCL2), resulting in IL-6 accumulation and feedback hyperactivation of the lnc-BM/JAK2/STAT3 loop and hence further BCBM enhancement [94]. Likewise, lncRNA lymph node metastasis associated transcript 1 (LNMAT1), known as regulator of lymph node metastasis, activates in bladder cancer (BLCA) the transcription of CCL2, which facilitates recruitment of macrophages into the tumor mass and promotion of lymphatic metastasis via vascular endothelial growth factor (VEGF)-C secretion [95]. The overexpression of one of the most-studied lncRNAs, HOX transcript antisense RNA (HOTAIR), is a hallmark in many cancer types, contributing among others in tumor resistance, to immune-mediated cytotoxicity [96]. HOTAIR, the antisense strand transcript of the HOXC locus, mainly functions as a gene silencing regulator by cooperating with chromatin-modifying complexes [97]. On this motif, high levels of HOTAIR in HCC positively correlate with recruitment of macrophages and MDSCs to the TME, through CCL2 induction [98]. Finally, potent inducers of M2 polarization have been reported to be the lncRNAs LIN00662, RP11-361F15.2 and GMA-AS1 expressed in HCC, osteosarcoma, NSCLC and BC, respectively, via different molecular mechanisms [56,57,99,100].

Regulation of T Cell Function-Associated Immune Checkpoints by lncRNAs in Cancer
The unbalanced expression of lncRNAs in immune and tumor cells may also modify tumor immunosurveillance, via mechanisms that involve stimulation or suppression of various 'breaks' during the 'cellular' immune cell activation (priming phase) and response (effector phase). Inhibitory receptors, also known as immune checkpoints, and their ligands can be found on a wide range of T lymphocytes and cancer types, respectively [106]. Overexpression of inhibitory receptors, including programmed cell death protein-1 (PD-1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) and T cell immunoglobulin mucin 3 (TIM-3), on T cells of cancer patients has been shown to counteract with co-stimulatory cell signals, leading to progressive loss of their activation potential and effector functions, exhaustion and eventually deletion [107][108][109]. LncRNAs affecting the expression of immune checkpoints are listed in Table 3 along with their underlying molecular mechanisms.
In this context, upregulation of the lncRNA nuclear-enriched autosomal transcript 1 (NEAT1) in PBMCs of HCC patients reduces CD8+ T cell survival and cytotoxicity, through modifications in the miR-155/Tim-3 axis. NEAT1 causes augmentation of Tim-3 expression via interaction with miR-155, a negative regulator of Tim-3. Tim-3 upregulation in turn results in loss of CD8+ T cell cytolytic activity and their death, allowing HCC cell immune escape, invasion and metastasis [110]. Similarly, the overexpression of lncRNA lnc-Tim3 in HCC-derived infiltrating CD8+ T lymphocytes appears to play a vital role in exacerbating their exhaustion, while preventing their apoptosis [111][112][113]. The underlying mechanism involves direct binding of lnc-Tim3 to Tim-3 that prevents HLA-B-associated transcript 3 (Bat3) attachment and downstream activation of the Lck/NFAT1/AP-1 signaling. Lack of appropriate signaling leads to nuclear localization of Bat3 and enhancement of p-300dependent p53 and RelA transcriptional activation of anti-apoptotic genes, including MDM2 and Bcl-2, therefore contributing to the survival of Tim-3+ exhausted CD8+ TILs [111].  Overexpression of lncRNA MALAT1 in diffuse large B cell lymphoma (DLBCL) causes effector T cell apoptosis and anergy, by upregulating PD-L1 expression on cancer cells. The underlying molecular mechanism involves direct binding of MALAT1 to miR-195, a suppressor of PD-L1 mRNA translation, thus allowing overexpression of PD-L1 on cancer cells. PD-L1 interacts with PD-1 and CD80 on CD8+ T cells, leading to the inhibition of their expansion, activation and cytotoxicity and consequently to tumor immune-escape [114]. Furthermore, MALAT1 exerts similar effects on PD-L1 expression in NSCLC cells by modulating miR-200a-3p function [115]. Apart from improving malignant cell propagation, migration and EMT, another lncRNA, the small nucleolar RNA host gene 14 (lnc-SNHG14), also protects DLBCL cells from immune-mediated cytotoxicity. Briefly, overexpressed lnc-SNHG14 in DLBCL acts as a miR-5590-3p sponge to facilitate de-repression of the zinc finger E-box binding homeobox 1 (ZEB1) translation. ZEB1 upregulation promotes the transcriptional activation of lnc-SNHG14 and PD-L1, which in turn inhibits the activation of CD8+ T cells and induces their apoptosis by binding to the PD-1 molecules [116].
Likewise, overexpression of lncRNA urothelial cancer associated 1 (UCA1) in intestinal gastric cancer has been associated with reduction in effector T cell cytotoxicity and viability, via PD-L1 upregulation, mediated by UCA1-dependant 'sponging' of its translational repressors miR26a/b, miR-214 and miR-193a [117]. In pancreatic cancer models, high levels of lncRNA LINC000473 reinforce tumor immunoescape by targeting another PD-L1 suppressor, miR-195-5p. PD-L1 upregulation in turn results in eliminated CD8+ T cell cytotoxicity and advanced apoptosis by induction of the pro-apoptotic protein Bcl-2associated X protein (BAX) and inhibition of the anti-apoptotic factor Bcl-2. LINC000473 further contributes to a decrease in the production of IFN-γ and IL-4 by CD8+ T cells, while it increases IL-10 secretion [118]. Contrarily, low expression of the lncRNA T cell leukemia/lymphoma 6 (TCL6) in breast cancer has been associated with worse prognosis in progesterone receptor (PR)-negative patients, attributed in part to tumor immune-escape, via regulation of immune-related pathways such JAK/STAT cascades. Specifically, TCL6 was positively correlated with tumor infiltration with immune cells, including CD8+ and CD4+ T lymphocytes, neutrophils, B and DCs, as well as with the expression of PD-1, PD-L1, PD-L2 and CTLA-4 immune checkpoint molecules [119].

Regulation of Anti-Tumor NK-Mediated Cytotoxicity by lncRNAs
The immune escape of cancer cells may also be promoted by alterations in the expression patterns of lncRNAs associated with NK cell activation and cytotoxic activity as well as with their potency to infiltrate the TME. Some of these lncRNAs are solely expressed by NK cells (listed in Table 1), while others by tumors (listed in Table 2).
Among several cancer types, HOTAIR overexpression in leukemia has been associated with tumor immune-escape through increased activation of the Wnt/β-catenin signaling pathway. Hyperactivation of this cascade in leukemic patients is considered a potent suppressor of both innate and adaptive immunity, as it is positively correlated with a reduction in NK cell activity, a decreased CD4+/CD8+ T subset ratio and a decline in cytokine release in peripheral blood, as well as in Ig production by B-lymphocytes [101]. In addition, in early-stage HCC, the expression patterns of a network of five lncRNAs, namely AATK-AS1, C10orf91, LINC000162, LINC00200 and LINC00501, that are known to act as competing endogenous RNAs (ceRNAs), positively correlate with good tumor prognosis and overall patients' survival, along with tumor infiltration by activated CD4+ memory T cells, NK and mast cells [102]. CeRNAs have been considered as the drivers in many diseases, including cancer, by interacting with miRNA response elements (MREs) found in both coding and non-coding RNAs, thus reducing the amount of miRNAs available to target mRNAs and relieving the miRNA repression on competing RNAs [16].
In contrast to IL-2-activated NK cells, which show high expression of lncRNA GAS5 and intense IFN-γ release, lncRNA GAS5 is downregulated in liver cancer patient-derived NK cells, contributing to reduced NK-mediated cytotoxicity and tumor immune-escape. The underlying mechanism involves direct interaction of lncRNA GAS5 with miR-544, a translational repressor of runt-related transcription factor 3 (RUNX3). Induction of RUNX3 promotes the transcription of the natural cytotoxicity receptor 1 (NCR1) gene and the expression of the encoded protein NKp46, an activating receptor that stimulates INF-γ secretion. Hence, downregulation of lncRNA GAS5 in the NK cells of liver cancer patients results in decreased natural cytotoxicity receptor NKp46 levels, a decline in INF-γ release and, eventually, in a reduction in NK cell cytotoxicity and CD107+ levels, thus facilitating tumor immune escape [79]. In this context, reduced expression of lncRNA RP11-222K-16.2 in NK cells of acute myeloid leukemia (AML) patients eliminates NK cell differentiation via cis-acting regulation of a nearby gene that encodes for the eomesodermin protein, a transcriptional activator of genes related to NK development [80]. Similarly, IFNβ-induced exosomal lncRNA EPH receptor A6-1 (linc EPHA6-1) enhances NK cytotoxicity, by acting as a competing endogenous RNA (ceRNA) for hsa-miR-4485-5p, leading to overexpression of the natural cytotoxicity receptor NKp46 [103], while lncRNA IFNγ antisense 1 (IFNG-AS1) is associated with increased IFN-γ secretion from NK cells, that enhances their activity [81]. Finally, lnc-CD56 positively regulates the expression of the NK surface marker CD56, thus promoting the polarization of CD34+ hematopoietic progenitor cells towards a NK cell phenotype [82].
Several lncRNAs may also interfere indirectly with NK anti-tumor activity and tumor immune escape by regulating the expression and shedding of MHC class I chain related A (MICA). MICA is one of the natural killer group 2, member D (NKG2D) ligands (NKG2DLs), produced by virus-infected or malignant cells and integrated into their cellular membrane. MICA can be recognized by NKG2D, an activating receptor located on the membrane of NK and other immune cells, while the MICA-NKG2D interaction can positively regulate host immune-surveillance against MICA-expressing cells [104]. Apart from the vast range of MICA polymorphisms, dysregulation of MICA expression in tumor cells might also help the later to evade immune-mediated cytotoxicity. For example, some tumor cells develop the capacity to discard MICA molecules off the cell surface using proteolysis, a process known as "MICA shedding". Various miRNAs and lncRNAs seem to be associated with the regulatory mechanisms of MICA expression, such as the lncRNA XXbac-BPG181B23.7 [104]. It is thus possible that changes in the expression patterns of lncRNAs like XXbac-BPG181B237 could contribute to the immune toleration of tumor cells by altering MICA presentation (Table 2).
In the same context, the circular lncRNA circ_0000997 was found to be a central regulator of hypoxia-induced resistance of pancreatic cancer (PACA) cells to NK-mediated killing, via promoting MICA shedding from tumor cell membranes. Briefly, the hypoxic microenvironment in PACA can induce circ_0000997 upregulation in malignant cells, which in turn sponges miR-153 and de-represses HIF-1α and a disintegrin and metalloproteinase domain 10 (ADAM10) translation. The elevated expression of HIF-1α and ADAM10 promote the shedding of MICA from the membrane of PACA cells (mMICA), thus increasing the levels of the soluble MICA (sMICA) molecules, which further bind to NKG2D receptors on NK cells. This interaction in turn, enhances the internalization and degradation of the receptor, hence triggering reduction in NK cell activity [105] (Table 2).

LncRNAs and Tumor Response to Immunotherapy
The dysregulated expression of a wide range of lncRNAs has been reported to affect not only the endogenous immune surveillance against cancer, but also the tumor response to exogenous immunotherapy. Although the relevant studies are still limited and mainly focused on Ab-mediated immunotherapy, it cannot be ruled out that lncRNAs might also be involved in the regulation of immune responses to adoptive cell transfer (ACT). A recent cohort study of 348 patients with bladder cancer and 71 patients with melanoma, correlated lncRNA-based immune subtypes that are associated with overall survival (OS) and patient response to cancer immunotherapy, aiming to produce a lncRNA score for multi-omic panels in precision immunotherapy. The findings revealed four distinct classes of lncRNA-based immune subtypes with significant differences in OS and response to immunotherapy, while the one with the greatest OS included the immune-active class which was characterized by the immune-functional lncRNA signature associated with high CTL infiltration [130]. Table 4 contains a list of lncRNAs reported to regulate tumor response to conventional Ab-mediated immunotherapy.
Among the mAbs currently used for Ab-based cancer immunotherapy, trastuzumab, a human epidermal growth factor receptor (HER2) inhibitor used for early-stage and metastatic HER2+ BC and GC treatment [131], compiles a long list of lncRNAs whose expression dysregulation seems to interfere with its action. Overexpression of lncRNA activated by TGF-β (lnc-ATB) in the SKBR-3 cell line and corresponding BC tissues has been associated with their resistance to trastuzumab, through a sponging effect on miR-200c that results in the upregulation of EMT-inducers ZEB1 and zinc finger protein 217 (ZNF-217) and the dysregulation of the TGF-β signaling [132]. In contrast, reduced levels of the lncRNA GAS5 in the same cell and tissue models enhanced tumor cell proliferation and resistance to trastuzumab via modulation of the miR-21/PTEN/mTOR axis [133]. LncRNA-small nucleolar RNA host gene 14 (lnc-SNHG14) is another example of lncRNA whose elevated levels in HER2+ BC cells and their secreted exosomes has been associated with diminished tumor response to trastuzumab. Although the exact underlying molecular mechanism of resistance is still unknown, the involvement of the apoptosis-related molecules Bcl-2/Bax is considered possible [134]. Primary or acquired resistance has been also reported for cetuximab, a therapeutic mAb that binds to EGF receptor and induces its degradation in CRC patients with metastatic disease. The presence of abnormally increased levels of lncRNA UCA1 in CRC cells and their secreted exosomes has been proposed as potential clinical biomarker for dictating tumor resistance to cetuximab [135]. In TNBC patients, the lack of hormonal receptors excludes the treatment option with anti-hormone inhibitors. Although the Ab-based immunotherapy with pembrolizumab, is currently employed as an alternative treatment option, TNBC patients showing resistance to anti-PD-1 treatment, are also unresponsive to pembrolizumab [136]. In these patients the overexpression of lncRNA LINK-A has been strongly correlated with increased resistance to PD-1 blockade by immune checkpoint inhibitors [84]. Similarly in patients with NSCLC, among other cancers, receiving anti-PD-1 immunotherapy, lncRNA RP11-705C-15.3 levels have been considered a putative prognos-tic factor for the treatment outcome [137]. Durvalumab, an antibody used for the treatment of NSCLC and BLCA, acts by blocking the interaction of PD-L1 with PD-1 and CD80. Patients treated with Durvalumab showed a higher survival rate, when they had augmented expression level of lncRNA RP11-291B-21.2, thus suggesting that RP11-291B-21.2 can be used as a potential biomarker of PD-1/PD-L1 blockage [138].

Cancer Stem Cells, lncRNAs and Tumor Immune Escape
The ability of cancer stem cells (CSCs) to evade immune destruction was initially proposed by the hierarchical model of cancer along with the cancer immune-editing theory [139]. The small subpopulation of CSCs consists of pluripotent and long-living cancer cells that are held responsible for tumor initiation, metastasis, recurrence and drug resistance [140][141][142]. Along with their ability to initiate neoplastic growth, regardless of cancer immunosurveillance, CSCs persist in latent tumors by forming specialized nicheconstrained cell reservoirs which, in contrast to niche-independent immunogenic daughter cells, are immune-privileged [143].
CSCs evade immune-mediated destruction and actively suppress innate and adaptive anti-tumor immune responses by expressing, diverse from non CSCs, intracellular, membrane-bound and soluble factors that allow them to survive in their niche and interact with a broad range of immune cells [144][145][146]. For example, CSCs in different tumor types show dysregulated expression of several anti-apoptotic and survival-inducing proteins such as Bcl-2, Bcl-xL and surviving. They also characterized by a distinct repertoire of secreted cytokines and other molecules, including IL-4, IL-6, IL-10, PGE2 and FasL that protect them from immune-mediated apoptosis triggered by effector T and NK cells, or by immunotherapeutic antibodies [143,[146][147][148][149][150][151][152][153]. In addition, the CSC-immune cell crosstalk along with CSC-secreted chemotactic factors and immunosuppressive cytokines, such as TGF-β, [143,154] advance the establishment of an immunosuppressive TME either by fostering tumor infiltration by a wide range of immunosuppressive immune cell subpopulations, including MDSCs, Tregs and TAMs, or/and by inhibiting the recruitment and activities of effector immune cells such as NK cells, CTLs and T cells (reviewed by Bayic and Lathia, 2021 [145] and Vahidian et al., 2019) [146]).
Studies on the multifaceted interactions between CSCs and the immune system have revealed that CSCs prevent the differentiation of naïve CD8+ T cells into CTLs during the CD8+ T cell priming by APCs, via downregulating the expression of CD80 (B7.1) and CD86 (B7.2) ligands responsible for secondary signals, while elevating levels of the inhibitory costimulatory ligand PD-L1 [155]. The priming of naïve CD8+ T cells by APCs may be further inhibited by CSC-mediated induction of CTLA-4 on T cells, a vital negative regulator of T cell response, leading to T cell exhaustion and dysfunction [155][156][157]. Accordingly, CSCs are able to suppress their recognition by activated CTLs through elimination of the surface expression of MHC I molecules [158] and inadequate presentation of TAAs or tumor specific antigens (TSAs), which are also less immunogenic than those of non CSCs [143,[146][147][148]. Finally, CSCs from different cancer types show abnormally elevated levels of inhibitory NK cell receptor ligands that protect them against destruction by innate anti-tumor immunity [143,144,146,147].
Although the direct impact of lncRNAs repertoire in the immunomodulatory potential of CSCs has not been directly exploited so far, it is, however, clear that the TME affects the lncRNA-regulated therapy resistance in CSCs [159]. Hypoxia, a major feature of TME, has been shown to promote cancer stemness and associated resistance to apoptosis by modulating the non-coding transcriptome, including the lncRNAs runt-related transcription factor 1 Intronic Transcript 1 (RUNX1-IT1) and hypoxia-associated lncRNA (HAL) in HCC and BC models, respectively [159][160][161][162]. In addition, the crosstalk of tumor cells with immunosuppressive immune cell components of TME, like TAMs, has been demonstrated in HCC to mediate cancer stemness and resistance to apoptotic signals through modulation of the lncRNA H19/miR-193b/MAPK1 axis [163].
Nevertheless, signaling pathways known as "hallmarks" for CSC functions, have now been identified as critical regulators of the tumor immune response. Among them the activation of Wnt/β-catenin signaling has been positively associated with cancer stemness, drug resistance and immune exclusion across several non-T cell-inflamed human tumors, thus suggesting a negative impact of this pathway in immune cell infiltration into TMEs [164][165][166][167][168]. Studies in diverse cancer types have revealed that CSC-associated lncRNAs are directly or indirectly involved in tumor resistance to CSC elimination by apoptosis, via regulating the Wnt/β-catenin cascade [159]. In this context, in GBM CSCs, down regulation of lincRNAp21, a negative regulator of β-catenin, suppresses CSC apoptosis [169], whereas in NSCLC cells the inhibition of lncRNA NEAT1, an lncRNA responsible for CSC enrichment, restrained the stemness traits and re-sensitized cells to apoptosis, via de-activation of the Wnt pathway [170]. Mechanistically, LncRNAs mediate the regulation of Wnt/β-catenin signaling, either at transcriptional level or by competing the action of pathway inhibitors [159].
Similarly, Notch signaling, elicits a CSC phenotype that contributes among others to CSC dormancy and resistance to immune destruction, via supporting an immunosuppressive microenvironment [171,172]. The Jagged-1-Notch4 axis is highly expressed in endocrine resistant breast CSCs and this activity has been correlated with the differentiation and polarization of macrophages to TAMs and eventually to decreased anti-tumor immune responses [173]. Notch-induced lncRNAs, such as LUNAR1, associates with reduced apoptosis of CRC cells and overall patient survival [174], while Notch-inducing lncRNAs, like prostate cancer-upregulated long noncoding RNA 1 (PlncRNA-1) function as an apoptosis inducer in glioma tumors [175]. Similarly, lncRNA antisense non-coding RNA in the INK4 locus (ANRIL) inhibits apoptosis and enhances progression of GC cells by targeting among others the Notch pathway [176,177].
In conclusion, although the literature lacks reports that demonstrate a direct link of lncRNAs involved in the regulation of CSC features, with CSC immune response or evasion, it is highly suggestive that signaling pathways important for both CSC functions and immune destruction may be downstream targets or upstream regulators of critical lncRNAs.

Future Perspectives
Early cancer diagnosis is critical for treatment success. Until recently, the era of cancer biomarkers was mainly enriched only by coding gene products. However, over the past few years, the focus on biomarker development has been shifted to previously thought "genetic debris", now known as non-coding RNAs, that have broadened the diagnostic and prognostic panels of a vast variety of human cancers. LncRNAs are key players in the regulation of cellular events and the pathophysiology of several diseases, including cancer and its associated characteristics [21]. LncRNAs are considered great candidates for diagnostic and prognostic biomarkers, given that their levels can be indicative of the stage of the disease, while they are easily detectable in body fluids, such as saliva, blood, plasma and urine [178][179][180]. LncRNAs can be also detected in exosomes or inside apoptotic bodies conjugated with RNA-binding proteins, thus avoiding RNase degradation [181,182]. In addition, their property to exhibit tissue-specific expression patterns makes them ideal targets for therapy [21]. Furthermore, the high sensitivity and specificity of lncRNAs undoubtedly make them important research tools for cancer diagnosis, prognosis and therapeutic targeting [183].
Although the translational potential of lncRNA targeting in clinic is still in its infancy, there are several quite novel strategies that can potentially be proved effective for specific lncRNA manipulation. These strategies mainly include post-transcriptional degradation of lncRNAs, modification of lncRNA genes and loss of lncRNAs' function [184]. The posttranscriptional degradation may be facilitated either by double-stranded RNA-mediated interference (RNAi) or single-stranded antisense oligonucleotides (ASOs) [184]. Although ASOs have already been used in clinical trials as a promising therapeutic strategy with good results, RNAi faces excessive difficulties in preclinical studies, due to ineffective delivery and low bioavailability [184,185]. Alternatively, modifications of lncRNA genes can be induced by application of the novel clustered regularly interspaced short palindromic repeats (CRISPR)-Cas system that is able to silence lncRNA expressing loci and block lncRNA transcription by merging Cas9 with transcriptional repressors that target the specific gene promoter [186,187]. Moreover, blocking of lncRNA-protein interactions or secondary/tertiary lncRNA formation structures can be achieved by small molecule inhibitors, or specific ASOs that can recognize and bind to lncRNAs [184].
Other major challenges that need to be overcome for applying therapeutic lncRNA targeting approaches in clinical practice, especially in the field of oncology and beyond, is the identification and characterization of the right lncRNA candidates, as well as the deep understanding of their role in cellular functions associated not only with initiation and progression of cancer, but also with individual characteristics of the tumor status, such as the acquisition of resistance to immunosurveillance, immunotherapy or chemotherapy [15]. It is also critical to select and characterize lncRNAs with crucial involvement in specific cancer type, subtype, or subpopulation like CSCs; however there is still a limited number of lncRNA-associated structural and functional data between normal and cancerous tissues [188].
Summarizing, over the last decade substantial effort has been made towards the identification of several lncRNAs that can potentially be implemented as diagnostic and prognostic biomarkers for both carcinogenesis and tumor resistance to endogenous and exogenous immune-mediated cytotoxicity. Nevertheless, the clinical application of lncRNAbased therapeutics is still in its infancy. Key challenges, including lncRNA specificity, delivery and tolerability issues, should be areas of extensive research. This, together with new technological advances, could enable the effective translation of newly developed lncRNA therapeutics.

Conflicts of Interest:
The authors declare no conflict of interest.