The tumour hypoxia induced non-coding transcriptome

(1)

Review

The tumour hypoxia induced non-coding transcriptome

Hani Choudhry

^a

, Adrian L. Harris

^b,

*, Alan McIntyre

^c,

**

aDepartment of Biochemistry, Faculty of Science, Center of Innovation in Personalized Medicine, King Fahd Center for Medical Research, King Abdulaziz University, Jeddah, Saudi Arabia

bMolecular Oncology Laboratories, Department of Oncology, Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, OX3 9DS, UK

cCancer Biology, Division of Cancer and Stem Cells, QMC, University of Nottingham, Nottingham, NG7 2UH, UK

A R T I C L E I N F O

Article history:

Available online 21 January 2016

Keywords:

miRNA lncRNA Hypoxia Non-coding RNA Cancer

A B S T R A C T

Recent investigations have highlighted the importance of the non-coding genome in regions of hypoxia in tumours. Such regions are frequently found in solid tumours, and are associated with worse patient survival and therapy resistance. Hypoxia stabilises the transcription factors, hypoxia inducible factors (HIF1α and HIF2α) which coordinate transcriptomic changes that occur in hypoxia. The changes in gene expression induced by HIF1α and HIF2α contribute to many of the hallmarks of cancer phenotypes and enable tumour growth, survival and invasion in the hypoxic tumour microenvironment. Non-coding RNAs, in particular microRNAs (miRNAs), which regulate mRNA stability and translation, and long-non- coding RNAs (lncRNAs), which have diverse functions including chromatin modiﬁcation and transcriptional regulation, are also important in enabling the key hypoxia regulated processes. They have roles in the regulation of metabolism, angiogenesis, autophagy, invasion and metastasis in the hypoxic microenvironment. Furthermore, HIF1α and HIF2α expression and stabilisation are also regulated by both miRNAs and lncRNAs. Here we review the recent developments in the expression, regulation and functions of miRNAs, lncRNAs and other non-coding RNA classes in tumour hypoxia.

Contents

1. Introduction ... 36

2. Hypoxic regulation of miRNAs ... 36

3. Regulation of the miRNA machinery in hypoxia ... 38

4. Regulation of HIF by miRNAs ... 38

5. Impact of miRNAs in hypoxia ... 41

5.1. The impact of HIF regulating miRNAs ... 41

5.2. Angiogenesis ... 42

5.3. Metabolism ... 42

5.4. Autophagy ... 42

5.5. Invasion and metastasis ... 42

5.6. Growth and survival ... 42

* Corresponding author. Molecular Oncology Laboratories, Department of Oncology, Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, OX3 9DS, UK. Tel.:+44 (0)1865 222457; fax:+44 (0)1865 222 431.

E-mail address:[email protected](A.L. Harris).

** Corresponding author. Cancer Biology, Division of Cancer and Stem Cells, QMC, University of Nottingham, Nottingham, NG7 2UH, UK. Tel.:+44 (0)115 8231307.

E-mail address:[email protected](A. McIntyre).

Molecular Aspects of Medicine 47–48 (2016) 35–53

Contents lists available atScienceDirect

Molecular Aspects of Medicine

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / m a m

(2)

6. Hypoxic regulation of lncRNAs and their impact on cancer biology ... 43

6.1. H19 ... 43

6.2. LincRNA-p21 ... 43

6.3. HINCUTs ... 45

6.4. lncRNA-LET ... 45

6.5. HOTAIR ... 45

6.6. WT1 ... 45

6.7. AK058003 ... 45

6.8. HIF2PUT ... 46

6.9. ENST00000480739 ... 46

6.10. EFNA3 lncRNA ... 46

6.11. lncRNA-UCA1 ... 47

6.12. RERT-lncRNA ... 47

6.13. linc-RoR ... 47

6.14. MALAT1 ... 47

6.15. aHIF1α ... 48

6.16. NEAT1 ... 48

7. Hypoxic regulation of other noncoding RNA classes ... 48

8. Summary and perspectives ... 49

Acknowledgements ... 50

References ... 50

1. Introduction

Regions of low oxygen (hypoxia) occur in solid tumours due to insuﬃcient vascularisation, and high tumour met- abolic and proliferative rates (Semenza, 2014). To survive, tumour cells need to adapt to this tumour micro- environmental stress and molecular adaption occurs through the stabilisation of the hypoxia inducible factor proteins HIF1α and HIF2α (Semenza, 2014; Shen and Kaelin, 2013).

HIF1α and HIF2α are constitutively expressed, however, in conditions of adequate oxygenation they are degraded. They are hydroxylated by the prolyl hydroxylases which require oxygen as a co-factor (Shen and Kaelin, 2013). Upon hy- droxylation, HIFs are ubiquitinated by von Hippel–Lindau (VHL) syndrome protein and degraded by the proteasome (Shen and Kaelin, 2013). In addition, the expression of HIFs are also regulated by growth factor signalling and a number of studies have shown their dependence upon growth factors (such as EGF and FGF-2) and MAPK and AKT signalling (Agani and Jiang, 2013; Feldser et al., 1999). Clinically, hypoxia is associated with metastasis, chemotherapy and radiotherapy resistance and worse survival (Multhoff et al., 2014;

Rebucci and Michiels, 2013). HIF1α and HIF2α heterodimerise with HIF1β (ARNT) and transcriptionally ac- tivate many genes involved in processes that contribute to the hallmarks of cancer and increase tumour survival in the hypoxic tumour microenvironment including: angiogenesis, metabolism, autophagy, invasion and metastasis (Brahimi-Horn et al., 2011; De Bock et al., 2011; Favaro et al., 2011; Semenza, 2014; Singleton et al., 2014). In addition to regulating protein coding RNA, it is clear that non-coding RNAs are also differentially expressed in hypoxia and that these play major roles in the hypoxic tumour microenvironment (Choudhry et al., 2014, 2015; Gee et al., 2014; Ivan and Huang, 2014).

Recent advances in human transcriptome analysis revealed that less than 2% of the transcriptional output encodes proteins and the remaining 98% encode different classes of

non-coding RNAs (Djebali et al., 2012). These non-coding RNAs can be categorized based on their length into small non-coding RNAs (<200 nucleotides), such as miRNA, piwiRNAs, snRNA, and tRNAs, and long non-coding RNAs (lncRNAs) (>200 nucleotides) such as MALAT1, NEAT1, and many antisense transcripts. Many of these non-coding RNAs have potential transcriptional, post-transcriptional, and epigenetic regulatory functions and are often deregulated in many diseases, including cancer. Alterations of the expression of these non-coding RNAs contribute to cancer formation and progression and has key roles in the hypoxic tumour microenvironment. Furthermore, many miRNA and a limited number of hypoxia responsive lncRNAs have been reported to play a regulatory role in the hypoxia/HIF pathway, which contributes to cancer development and metastasis. In this review, we will summarize the current knowledge regarding hypoxia-regulated miRNAs and lncRNAs and their impact on cancer biology. In addition, we will review the current literature regarding the effects of hypoxia on other non-coding RNA classes.

2. Hypoxic regulation of miRNAs

miRNAs are short 22 nucleotide duplexes that regulate mRNA stability and translation (Camps et al., 2014;

Nallamshetty et al., 2013). miRNA expression is changed in tumours compared to normal tissues and in a parallel with coding genes many oncogenic and tumour suppressive roles for miRNA have been identiﬁed (Lin and Gregory, 2015). Fur- thermore, differential miRNA expression and biogenesis is observed under hypoxic conditions (Choudhry et al., 2014;

Rupaimoole et al., 2014; van den Beucken et al., 2014). The biogenesis of miRNA and the processes by which they degrade mRNA and repress mRNA translation are complex (Camps et al., 2014; Nallamshetty et al., 2013). miRNA containing transcripts known as pri-miRNA are initially transcribed mostly by polymerase II under transcription factor regulation (Iorio and Croce, 2012). pri-miRNAs are

(3)

cleaved whilst still in the nucleus by the DROSHA and the Di Gerorge Syndrome critical region 8 (DGCR8) encoded protein. This cleavage generates a 70 nucleotide RNA known as a precursor miRNA (pre-miRNA) which is then ex- ported out of the nucleus into the cytoplasm by EXPORTIN5 (Camps et al., 2014; Nallamshetty et al., 2013). An alternative DROSHA independent miRNA biogenesis pathway exists, called the miRtron pathway, by which pri-miRNA are pro- cessed to pre-miRNAs without DROSHA cleavage (Iorio and Croce, 2012). Once in the nucleus, the pre-miRNA is cleaved to produce a 22-nucleotide miRNA duplex by the DICER- TRBP complex (Camps et al., 2014; Nallamshetty et al., 2013).

The miRNA then associates with the ARGONAUTE (AGO) proteins which recruit miRNAs and their target mRNAs to the RNA-induced silencing complex (RISC). RISC contains a number of proteins including AGO, DICER, PACT and TRBP.

miRNA target identification is defined by partial sequence complementarity (Iorio and Croce, 2012). miRNAs are pro- miscuous and tend to have many targets (Iorio and Croce, 2012). The miRNA-RISC complex then regulates target mRNAs by promoting their degradation or by translational repression (Camps et al., 2014; Nallamshetty et al., 2013). Based on thermodynamic properties, one strand -3p is designated the guide strand; however, the alternative pas- senger -5p strands can also act as functional guides and have significant functional relevance (Iorio and Croce, 2012).

pri-miRNA transcription is deregulated in tumours and is affected by the same mechanisms as mRNA expression, including: copy number variation, epigenetic regulation and regulation by transcription factor oncogenes such as MYC (Lin and Gregory, 2015). A number of studies have profiled the global changes in miRNA expression levels in hypoxia in tumours. These have identified many hypoxia- regulated miRNAs. As with hypoxia regulated gene expression, miRNAs have also been shown to be direct targets of HIF, identified through siRNA knockdown of HIF proteins and the identification of HRE in the promoter regions of these miRNAs (Devlin et al., 2011; Nagpal et al., 2015). Such HIF dependent miRNA include miR-210, miR- 191 and miR-21 (Devlin et al., 2011; Mace et al., 2013;

Nagpal et al., 2015).

A more global analysis of the specific role of HIF1α and HIF2α regulation of pri-miRNA expression identified a link between HIF binding sites identified by ChIP and upregulation of both pri-miRNAs and mature miRNAs in hypoxia (Camps et al., 2014). The miRNA that was most upregulated containing a HIF binding site was miR-210 (Camps et al., 2014). However, there was no correlation between the pri-miRNA and miRNA expression in hypoxia suggesting additional levels of post-transcriptional regulation (Camps et al., 2014). Indeed, the expression and maturation of miRNAs is also regulated by the miRNA machinery which is also modified in hypoxia and is discussed below (section 3). RNA seq analysis of miRNA expression in the breast cancer cell line MCF7 identified a total of 502 miRNAs, 41 of which were significantly upregulated in hypoxia and 28 of which were downregulated by hypoxia.

Time course analysis of hypoxic responses at 16, 32 and 48 hours revealed increases in the number of miRNA with differential expression with time (Camps et al., 2014). One hundred and eighty-ﬁve of the identiﬁed miRNAs were

within intronic regions of protein coding genes; however, no correlation was identiﬁed between the miRNA and host gene expression in hypoxia (Camps et al., 2014). For example, the miR-3140-3p host gene FBWX7 is downregulated in hypoxia whilst miR-3140-3p was up-regulated (Camps et al., 2014). Some of the miRNAs upregulated in this analysis correlated with a hypoxic gene expression signature in breast cancer including miR-210-3p, miR-27a-3p and miR-24-3p (Camps et al., 2014).

Studies have investigated the concordance of miRNA expression and global regulation of gene expression using miRNA target prediction software (MiRanda, PicTar, miRBase and TargetScan) (Guimbellot et al., 2009). These investigations did not identify a significant relationship between the global miRNA and mRNA expression (Guimbellot et al., 2009). However, given the considerable impact of HIF1α and HIF2α on mRNA expression profiles and the alternative effects of miRNAs on translational repression and transcript degradation, the lack of a significant relationship in the hypoxic milieu is hardly surprising. Furthermore, it could also be due to a lack of sensitivity of the target prediction software (Guimbellot et al., 2009).

Hypoxic signatures of miRNA expression have also been identified in a number of tumour types including glioblas- toma (Agrawal et al., 2014) and bladder cancer (Blick et al., 2015). A miRNA signature of hypoxia was identified in human breast and colorectal cancer cell lines (Kulshreshtha et al., 2007). This study identified hypoxic regulation of miR -23 -24 -26 -27 -103 -107 -181 -210 and -213. This study also identified that the majority of these identified hypoxia regulated miRNAs were overexpressed in a number of different tumour types via mining of miRNA published data sets (Kulshreshtha et al., 2007). However, there are some tissue specific differences in hypoxic miRNA regulation such as miR-155, miR-200a and miR-181b, which are up-regulated in hypoxia in a number of tumour types but not colon cancer (Guimbellot et al., 2009). A molecular subgroup of mela- noma regulated by transforming growth factor β1 (TGFβ1) has increased HIF1α and increased expression of miRNAs associated with hypoxia including miR-210, miR-218 and miR-224 (Hwang et al., 2014). Reduced expression of these miRNAs induced cell cycle arrest whilst increased expression induced increased expression of hypoxia regulated gene BNIP3 (Hwang et al., 2014). miRNAs have also been investigated by RNA sequencing in hypoxia (1% O224 hours) in endothelial cells (HUVECS) and a number of miRNAs were identified as differentialy expressed. Eighteen of these were validated by QPCR including miR-210 which is upregulated in hypoxic tumour studies and miR-150 which is downregulated in hypoxic tumour studies (Shen et al., 2013;

Voellenkle et al., 2012).

Additional HIF-independent or indirect regulations of miRNA also play a role in hypoxic changes in miRNA expression. For example, it has been noted that the well recognized hypoxymiR miR-210 contains many conserved transcription factor sites in its genomic vicinity including Oct-4 which is itself a hypoxic target (Devlin et al., 2011).

HIFs also regulate the expression of many transcription factors including TWIST1 which is known to regulate miR- 10b, an onco-miR which mediates metastasis (Shen et al., 2013). AKT signalling is activated in hypoxia and AKT2 37 H. Choudhry et al. / Molecular Aspects of Medicine 47–48 (2016) 35–53

(4)

signalling can increase NF-κB and CREB which can increase the expression of miR-21 (Shen et al., 2013).

Additionally, miR-424 is up-regulated in hypoxia by C/EBP-α /RUNX-1 transcription factors in endothelial cells and ischemic tissues undergoing vascular remodelling (Ghosh et al., 2010).Table 1shows a list of hypoxia-regulated miRNAs, their targets and functional impact.

3. Regulation of the miRNA machinery in hypoxia

Analysis of miRNA containing transcripts (pri-miRNAs) and mature miRNAs in hypoxia identiﬁed a lack of correlation suggesting additional post-transcriptional levels of modulation of mature miRNA levels (Camps et al., 2014).

Expression of genes which encode miRNA processing sub- unitsDDX5,XPO5,RAN,DICERandEIF2C2(encoding AGO2) were significantly reduced in hypoxia whilst EIF2C4 (encoding AGO4) was significantly increased. Analysis ofDICER expression in patient material datasets identified a significant inverse correlation betweenDICERexpression and a hypoxic breast cancer signature (Camps et al., 2014). HIF- dependent expression regulation of miRNA processing subunits was identified, including down-regulation of DICER and upregulation of AGO4 (Camps et al., 2014). Similarly in HUVECs, hypoxia reduces RNA and protein expression of many miRNA processing subunits including DICER, AGO1, AGO2, DGCR8, TRBP and EXPORTIN5 (Ho et al., 2012). The expression of DICER is also suppressed in hypoxia via oxygen sensitive epigenetic regulation of miRNA processing (van den Beucken et al., 2014). Hypoxic reduction of DICER expression is regulated by inhibition of the oxygen dependent H3K27me3 demethylases KDM6A/B, which supresses the DICER promoter (van den Beucken et al., 2014). This leads to reduced miRNA processing in hypoxia (van den Beucken et al., 2014). The authors highlight that reduced miRNA- processing results in a loss of miRNA repression including the miR-200 target ZEB1 resulting in stimulation of epithelial to mesenchymal transition (EMT) (van den Beucken et al., 2014). siRNA knockdown of DICER increased HIF1α expression in normoxia to levels similar to those found in hypoxia in the lung cancer cell line A549 (Yao et al., 2014).

Down-regulation of DROSHA has also been reported in hypoxia leading to dysregulation of miRNA biogenesis (Rupaimoole et al., 2014). DROSHA was negatively regulated by ETS1/ELK transcription factors and siRNAs targeting of these rescued DROSHA expression (Rupaimoole et al., 2014). The reduced miRNA biogenesis increased tumour progression which could be inhibited by siRNA knockdown of ETS1/ELK by siRNAin vivo(Rupaimoole et al., 2014). Fur- thermore, patient sample data analysis suggests that miRNA biogenesis is down-regulated in hypoxic tumours (Rupaimoole et al., 2014).

EGFR also suppresses the maturation of a speciﬁc subset of miRNAs, which are tumour suppressor like, in response to hypoxic stress (Shen et al., 2013). EGFR can phosphory- late argonaute 2 (AGO2) at Tyr393 which reduces AGO2 association with DICER and prevents miRNA processing of a subset of miRNAs from precursor miRNAs to mature miRNAs (Shen et al., 2013). The association between EGFR and AGO2 is enhanced in hypoxia as EGFR is retained in late endosomes where it co-localises with AGO2 and increases

the levels of AGO2 Tyr393 phosphorylation. A long loop structure of precursor miRNAs is important in inhibiting processing of precursor miRNAs by Tyr393 phosphorylated AGO2 to mature miRNAs. This produces a reduction in processing of a subcluster of long loop containing precursors including miR-31 and miR-192 but not of short loop precursors, for example miR-21 in hypoxic conditions (Shen et al., 2013). AGO2 Tyr393 phosphorylation is associated with worse overall survival in breast cancer patients and enables EGFR mediated survival and invasion in hypoxia (Shen et al., 2013).

4. Regulation of HIF by miRNAs

There is a complex network of miRNAs regulated by hypoxia that also affect the expression or stabilisation of HIF1α and or HIF2α either through direct binding of the 3’UTR of their mRNAs or indirectly through expression regulation of a regulatory unit of the HIFs such as VHL or a PHD.

Fig. 1is a schematic of the complex networks of HIF regulation by miRNAs.

A number of miRNAs that are down-regulated in hypoxia have been identiﬁed to directly target HIF1α and or HIF2α.

miR-199a is down-regulated in hypoxia and targets the 3’UTR of HIF1α and HIF2α in epithelial ovarian cancer iden- tiﬁed utilising 3’UTR luciferase reporter assays (Joshi et al., 2014). miR-18a is down-regulated in hypoxia in gastric carcinoma cell lines and microarray analysis revealed that HIF1α was a target. A luciferase reporter assay identiﬁed direct binding of miR-18a to the 3’-UTR of HIF1α in this setting (Wu et al., 2015). Similarly, HIF1α is a direct target of miR- 18a in basal type breast tumours (Krutilina et al., 2014).

Modulating miR-18a signiﬁcantly affected hypoxic gene expression patterns in both settings (Krutilina et al., 2014; Wu et al., 2015).

Conversely, a number of miRNAs that are increased in hypoxia have been identiﬁed to directly target HIF1α and/

or HIF2α or HIF1β. For example, miR-107 is signiﬁcantly upregulated in colorectal and breast cancer cell lines in response to hypoxia (Kulshreshtha et al., 2007). miR107 expression is inversely correlated with HIF1β expression in colon cancer and directly binds the 3’-UTR and reduces HIF1β expression (Yamakuchi et al., 2010). miR-155, which has an HRE in its promoter, is increased after exposure to hypoxia in colorectal cancer cell lines. miR-155 reduced expression and activity of HIF1α but not HIF2α in this setting, and miR-155 reduction sustained HIF1α expression in prolonged hypoxia. The authors proposed a model by which HIF1α and miR-155 act in concert to produce oscillatory expression of HIF1α in prolonged hypoxia (Bruning et al., 2011).

Similarly, miR-429 expression is up-regulated in hypoxia in endothelial cells (HUVECs) and its promoter region contains a HIF1α binding site (Bartoszewska et al., 2015). In endothelial cells, miR-429 reduces the stability of HIF1α mRNA, reducing HIF1α protein expression (Bartoszewska et al., 2015). In endothelial cells, the hypoxic mRNA and protein expression of HIF1α is dynamic, increasing to greater than normoxic levels within 2–4 hours before reducing below normoxic levels by 12–16 hours (Bartoszewska et al., 2015). The authors propose that, as in the colorectal cancer example with miR-155, miR-429 may produce

(5)

Table 1

Hypoxia-regulated miRNAs and their functional impact.

miRNA Cancer type Regulation HIF Target/Cancer impact Reference

Hypoxia differentially regulated miRNAs

miR-199a Epithelial ovarian cancer Downregulated under hypoxia — Targets HIF1α and HIF2α Joshi et al. (2014)

miR-18a Gastric carcinoma, basal type breast cancer

Downregulated under hypoxia — Targets HIF1α

Expression reduces invasion and increases apoptosis

Krutilina et al., (2014);Wu et al. (2015) miR-107 Colorectal cancer, breast cancer. Upregulated under hypoxia — Targets HIF1β Kulshreshtha et al. (2007);Yamakuchi et al. (2010) miR-155 Colorectal cancer, triple receptor

negative breast cancer, lung cancer, cervical cancer, nasopharyngeal cancer

Upregulated under hypoxia HRE in promoter

Targets HIF1α

Targets VHL and increases HIF1α and HIF2α Targets C/EBP reducing miR-143 Targets RHEB, RICTOR and RPS6KB2

Expression increases angiogenesis and glycolysis, reduces proliferation, induces G1cell cycle arrest, increases autophagy

Bruning et al. (2011);Kong et al. (2014);Wan et al.

(2014);Yao et al. (2014)

miR-429 Normal endothelial cells Upregulated under hypoxia. HRE in promoter

Targets HIF1α Bartoszewska et al. (2015)

miR-210 Normal activated T cells, breast cancer, hepatocellular carcinoma, many other tumour types and targets reviewed inDevlin et al. (2011) and Qin et al. (2014)

Upregulated under hypoxia HIF1α Targets HIF1α in activated T cells Targets GPDL1, and increases HIF1α

Targets ISCU and reduces TCA cycle and mitochondrial function increasing glycolysis

Targets VMP1 and increases metastasis Targets MNT and E2F3 and increases proliferation Targets CASP8AP2 and reduces apoptosis

Chan et al. (2009);Devlin et al. (2011);Favaro et al.

(2010);Kelly et al. (2011);Qin et al. (2014);Wang et al. (2014);Ying et al. (2011)

miR-143 Colorectal cancer, lung cancer Downregulated under hypoxia — Targets IGF-1R and reduces HIF1α. Targets HK2 Expression reduces xenograft growth, vascularisation and glycolysis

Qian et al. (2013);Yao et al. (2014)

miR-128 HeLa, glioma Downregulated under hypoxia — Targets p70S6K1 and reduces HIF1α

Expression reduces xenograft growth and vascularisation

Shi et al. (2012)

miR-199a Epithelial ovarian cancer Downregulated under hypoxia — Targets HER2 and reduces HIF1α Targets HIF1α and HIF2α

Expression reduces growth rate, vascularisation, migration and extracellular remodelling

He et al., (2013);Joshi et al. (2014)

miR-424 Normal endothelial cells and ischemic tissues

Upregulated under hypoxia C/EBP-α /RUNX-1

Targets Cullin2 and stabilises HIF1α. Ghosh et al. (2010)

miR-135b Multiple myeloma Upregulated under hypoxia — Targets FIH and increases HIF1α Umezu et al. (2014)

miR-101 HUVECs Upregulated under hypoxia — Targets Cullin3 reducing proteosomal degradation of

Nrf2 increasing HO-1 and VEGF Expression increases angiogenesis.

Kim et al. (2014)

miR-15b and miR-16

Nasopharyngeal cancer Downregulated under hypoxia — Both target VEGF. Madanecki et al. (2013)

miR-126 Endothelial cells Downregulated under hypoxia — Targets VEGF. Ye et al. (2014)

miR-21 Bladder cancer Upregulated under hypoxia HIF1α Direct targets not elucidated. Expression increases GLUT1, GLUT3, HK1, HK2. and glycolysis.

Gorospe et al. (2011);Yang et al. (2015)

miR-96 Prostate cancer Upregulated under hypoxia — Increases autophagy

At higher concentration targets ATG7 and reduces autophagy

Ma et al. (2014)

miR-137 Mouse brain tissue, SK-N-SH, HEK293, HeLa and mouse embryonic ﬁbroblast cell lines

Downregulated under hypoxia — Targets NIX and FUNDC1 Expression inhibits mitophagy

Li et al. (2014)

miR-191 Breast cancer Upregulated under hypoxia HIF1α Increases TGFβ2

Expression increases migration and invasion

Nagpal et al. (2015)

miR-100 Bladder cancer Downregulated under hypoxia. HIF1α Targets FGFR3

Expression decreases proliferation

Blick et al. (2013)

39H.Choudhryetal./MolecularAspectsofMedicine47–48(2016)35–53

(6)

oscillatory HIF1α expression (Bartoszewska et al., 2015). miR- 210 is a well described target of HIF1α and is robustly increased in many tumour types in response to hypoxia. Sim- ilarly, studies in activated T cells identiﬁed HIF1α regulation of miR-210; however, in activated T cells, miR-210 was found to target and reduce expression of HIF1α (Wang et al., 2014).

Whether this result can be repeated in tumour cells re- quires further investigation.

A number of miRNAs that are down-regulated in hypoxia have been identified to indirectly target HIF1α and or HIF2α, thus resulting in increased HIF protein in hypoxic conditions. In response to reduced expression of DICER in hypoxia, the conversion of pre-miR-143 to mature miR-143 is significantly reduced (Yao et al., 2014). IGF-1R is a direct target of miR-143, and increasing expression of miR-143 reduced HIF1α protein levels via reduced AKT signalling (Qian et al., 2013). miR-128 is significantly down-regulated in hypoxia in HeLa cells (Crosby et al., 2009). miR-128 levels are also decreased in glioma and decrease further with increasing tumour grade (Shi et al., 2012). p70S6K1 is a direct target of miR-128 and expressing miR-128 in glioma cell lines de-

creased p70S6K1 levels. p70S6K1 is a key downstream target of MTOR signalling and regulates a number of tumorigen- ic processes including angiogenesis and tumour proliferation (Shi et al., 2012). p70S6K1 also regulates HIF1α expression and overexpression of miR-128 reduced HIF1α protein expression inin vitroandin vivostudies (Shi et al., 2012).

miR-199a is downregulated in hypoxia and in epithelial ovarian cancer (Joshi et al., 2014). miR-199a overexpression reduced HIF1α expression (He et al., 2013). HER2 was iden- tiﬁed as a direct target of miR-199a. The miR-199a induced reduction in HIF1α expression could be rescued by overexpressing HER2 and activating AKT/MTOR/p70S6K1 signalling (He et al., 2013).

miR-210 expression is increased in hypoxia and miR- 210 stabilises HIF1α protein in HEK 293 and HeLa cells through repression of HIF1α hyperhydroxylation by reducing the expression of glycerol-3-phosphate dehydrogenase 1-like (Kelly et al., 2011). The data confirm thatglycerol-3- phosphate dehydrogenase 1-likeis a target of miR-210 (previously shown byFasanaro et al., 2009). The feedback mechanism by which HIF1α is stabilised, and which links Fig. 1. Regulation of HIF by miRNAs. HIF1α and HIF2α protein expression is regulated by miRNAs by direct targeting of HIF transcript or targeting of transcripts that regulate HIF protein expression and stabilisation. Red inhibition symbols define miRNA targeting and reduced expression of translation of the mRNA transcript which encodes the indicated protein. miRNA in blue boxes are downregulated in hypoxic conditions. miRNA in red boxes are upregulated in hypoxic conditions where those which also have red borderline are upregulated by HIF proteins binding directly to their promoter. miRNA in green boxes are not regulated by hypoxia but do regulate HIF1α and/or HIF2α. The figure is inspired by and additional information is fromHarris (2002). FIH;

factor inhibiting HIF, HER2; human epidermal growth factor receptor 2, IGF1R; insulin growth factor 1 receptor, MTOR; mechanistic target of rapamycin, P70S6k; P70S6 kinase, RBX1: ring-box 1, E3 ubiquitin ligase, VHL; von Hippel–Lindau protein, Ub; ubiquitin.

(7)

key molecular players in metabolism and hypoxia, has not been conﬁrmed independently, and its functional role is important to identify. miR-155 has an HRE in its promoter and is increased after exposure to hypoxia in colorectal cancer cell lines (Bruning et al., 2011). miR-155 is also frequently upregulated in multiple cancer types and its expression inversely correlates with VHL expression level in triple receptor negative breast cancer (TNBC) (Kong et al., 2014). miR- 155 directly targets VHL and increases HIF1α and HIF2α levels (Kong et al., 2014). miR-424 is up-regulated in hypoxia mediated by C/EBP-α /RUNX-1 in endothelial cells and ischemic tissues undergoing vascular remodelling but not in smooth muscle cells or tumour cell lines (Ghosh et al., 2010).

miR-424 targets Cullin2, a scaffolding protein essential for the assembly of the ubiquitin ligase complex which ubiquitinates HIF for degradation (Ghosh et al., 2010). In- creased miR-424 expression results in stabilisation of HIF1α and HIF2α in normoxia (Ghosh et al., 2010).

An interesting example of miRNA transmission in exosomes has been identiﬁed. miR-135b is up-regulated in exosomes released from multiple myeloma cells in response to chronic hypoxic exposure (Umezu et al., 2014).

miR-135b directly targets Factor inhibiting HIF (FIH) and miR-135b enriched exosomes directly suppressed FIH expression in endothelial cells and increased activity of HIF in these cells (Umezu et al., 2014). miR-135b was only expressed in cell lines after chronic exposure to hypoxia, after which levels were also maintained under normoxic conditions (Umezu et al., 2014). Conversely, FIH may be increased in hypoxia by an alternative mechanism. Dicer expression is reduced in hypoxia (Rupaimoole et al., 2014; van den Beucken et al., 2014). Investigation of global miRNA deple- tion achieved by knocking out DICER1 resulted in suppression of HIF1α transcription (Chen et al., 2014). When DICER1 was knocked out, FIH repression by miRNAs was lost, leading to reduced HIF expression. This highlights the regulation of FIH by miRNA in tumours (Chen et al., 2014).

Additional miRNAs which are not regulated by hypoxia, or which have not been investigated for their hypoxic expression, but which are often changed in tumours, have been identiﬁed to impact HIF1α and or HIF2α expression or stabilisation directly or indirectly. miR-138 expression is reduced in many tumour types and suppresses cell migration and invasion (Yeh et al., 2013). A study in ovarian cancer identiﬁed that miR-138 directly targeted SOX4 and HIF1α and that both of these regulated the anti-migratory and in- vasive phenotypes of miR-138 expression (Yeh et al., 2013).

miR-185 is downregulated in tumours compared to normal tissues and miR-185 overexpression by transfection with a mimic in colorectal cancer cell lines reduced HIF2α expression (Lu et al., 2014). miR-338-3p has reduced expression in hepatocellular carcinoma (HCC) tissue compared to adjacent normal liver tissue. miR338-3p directly targets the 3’UTR of HIF1α and reduces the expression of HIF1α and its regulated genes including VEGF and GLUT-1 (Xu et al., 2014). miR-145 expression inversely correlates with HIF2α expression. miR-145 overexpression reduced HIF2α and miR- 145 knockdown increased HIF2α levels. A putative miR- 145 binding site was identiﬁed in HIF2α and a luciferase reporter assay identiﬁed miR-145 direct regulation of HIF2α (Zhang et al., 2014). miR-519c downregulates HIF1α protein

levels in hypoxia in cell lines from lung and breast tumours through direct binding to the HIF1α 3’UTR. The levels of miR- 519c are not effected by hypoxia but are downregulated in response to HGF signalling (Cha et al., 2010). miR-183 expression levels are increased in high-grade gliomas. miR- 183 directly targets isocitrate dehydrogenase 2 (IDH2) reducing its RNA and protein levels (Tanaka et al., 2013).

The miR-183 induced reduction in IDH2 increased α-ketoglutarate levels and up-regulated HIF1α expression and stabilisation (Tanaka et al., 2013). The changes in the expression of these miRNA commonly found in tumours tend to result in increased HIF expression or stabilisation. It should be noted that many of the miRNAs considered above have targets in addition to the ones discussed here.

5. Impact of miRNAs in hypoxia

miR-210 is of particular importance in the ﬁeld of hypoxia miRNA research and has many identiﬁed targets (Devlin et al., 2011). Its many known direct targets regulate cell cycle, differentiation, apoptosis, translation, transcription metabolism and migration (Devlin et al., 2011).

5.1. The impact of HIF regulating miRNAs

The multiple processes upregulated in hypoxia through HIF transcriptionally increasing gene expression are by ex- tension affected by the regulation of HIF expression and stabilisation by miRNAs. The impact of miRNAs that target HIF will likely impact all the HIF regulated processes.

However, in the majority of cases the impact of only one of these processes has been investigated experimentally. Fur- thermore, individual miRNAs are likely to target a plethora of other mRNAs in addition to HIF. The effect of these miRNA on hypoxic cellular phenotype will depend on whether it positively or negatively regulates HIF expression and/or stabilisation, whether the miRNA itself is up-regulated or down-regulated in hypoxia and which additional mRNAs the miRNA targets.

For example, miR-199a which is downregulated in hypoxia and targets the 3’UTR of HIF1α and HIF2α directly also reduces HIF1α expression indirectly by targeting HER2 and reducing AKT/MTOR/p70S6K1 signalling (He et al., 2013;

Joshi et al., 2014). Functional investigations of the roles of miR-199a have identiﬁed that miR-199a overexpression reduced growth rate and vascularization of ovarian cancer xenograftsin vivo, an effect that was rescued by constitu- tive HIF1α or HER2 expression (He et al., 2013). A separate study identiﬁed that miR-199a inhibited migration of epithelial ovarian cancer cells, but not proliferation or clonogenic survival. miR-199a inhibited extracellular matrix remodelling through abrogation of HIF1α mediated hypoxic upregulation of LOX. Additionally, miR-199a reduced expression of VEGFA and EPO and reduced angiogenesis and increased necrosis inin vivostudies (Joshi et al., 2014). Sim- ilarly, miR-18a which is down-regulated in hypoxia directly binds to the 3’-UTR of HIF1α (Wu et al., 2015). miR-18a reduced invasion in breast cancer (Krutilina et al., 2014) and increased apoptosis and reduced invasion of gastric carcinoma cell lines in hypoxia (Wu et al., 2015). Additional miRNAs that are downregulated in hypoxia and negatively 41 H. Choudhry et al. / Molecular Aspects of Medicine 47–48 (2016) 35–53

(8)

regulate HIF expression or stabilisation include miR-143 and miR-128. Expression of both these miRNAs reduced xenograft growth and angiogenesisin vivovia reduced expression of VEGF (Qian et al., 2013; Shi et al., 2012). Conversely, miR- 155 which is upregulated in hypoxia by HIF (Bruning et al., 2011) targets VHL and experimental investigations identi- ﬁed that its expression increased angiogenesis, an effect inhibited by the re-expression of VHL (Kong et al., 2014).

A number of miRNAs that do not directly target HIF expression or stabilisation also impact hypoxic phenotypes by targeting the expression of genes that regulate these processes directly ; these are discussed below.

5.2. Angiogenesis

miR-101 expression is increased in hypoxia in HUVECs and tumour cell lines (Kim et al., 2014). miR-101 increases angiogenesis via indirect regulation of heme oxygenase-1 (HO-1) and VEGF. Investigations in HUVECs identiﬁed that miR-101 binds directly to the 3’UTR of Cullin 3 (CUL3) (Kim et al., 2014). This reduces proteosomal degradation of nuclear factor erythroid-derived 2-related Factor 2 (Nrf2) and results in its nuclear accumulation (Kim et al., 2014). Nrf2 increases the expression of VEGF and HO-1 (Kim et al., 2014).

miR-101 expression increased pro-angiogenic phenotypes inin vitroandin vivomodels, including a model of mouse hind-limb ischemia where injection of a lentiviral miR- 101 increased perfusion recovery (Kim et al., 2014). Both miR-20a and miR-20b directly target the 3’UTR of VEGF in addition to targeting HIF1α expression (Madanecki et al., 2013). Similarly, miR-15b and miR-16 which are downregulated in hypoxia in a nasopharyngeal cancer cell line also directly target the 3’UTR of VEGF (Madanecki et al., 2013). miR-126 expression is reduced in endothelial cells in hypoxia and miR-126 negatively regulated the expression of VEGF (Ye et al., 2014).

5.3. Metabolism

miR-143 directly targets hexokinase II (HK2), a key glycolytic enzyme. HK2 expression and glycolysis are increased in hypoxia (Yao et al., 2014). In response to reduced expression of DICER in hypoxia, the conversion of pre-miR- 143 to mature miR-143 is signiﬁcantly reduced. Conversely, the expression of miR-155 was increased in response to both hypoxia and decreased expression of DICER (Yao et al., 2014).

miR-155 suppresses miR-143 expression by targeting C/EBP which regulates transcription of miR-143. DICER knockdown increased expression of miR-155. DICER knockdown also increased glycolysis and increased expression of HK2 (Yao et al., 2014). miR-21 is up-regulated by hypoxia in various cancer cell lines, in HUVECs and in models of ischemia (Gorospe et al., 2011). miR-21 knockdown reduced glycolysis. miR-21 knockdown reduced the expression of a number of glycolysis genes including GLUT1, GLUT3, HK1 and HK2 (Yang et al., 2015). miR-21 expression was positively correlated with high expression of glycolytic genes, GLUT1, LDHA and HK1 in bladder cancer patient material (Yang et al., 2015). miR-21 knockdown also reduced phosphorylation of a number of signalling proteins of the MTOR pathway (including AKT and S6K). The direct targets of miR-

21 were not elucidated in this study (Yang et al., 2015). miR- 210 directly targets the mitochondrial iron sulfur scaffold protein ISCU (Chan et al., 2009; Favaro et al., 2010). The expression of miR-210 and ISCU were inversely correlated in breast cancer in which low levels of ISCU were associated with reduced recurrence-free survival (Favaro et al., 2010).

ISCU is required for the formation of iron sulphur clusters which are part of the active sites of many enzymes. Iron sulphur clusters enable maintenance of redox status. Iron sulphur clusters form an integral part of TCA cycle components such as succinate dehydrogenase and mitochondrial complex components I, II and III (Devlin et al., 2011; Favaro et al., 2010). miR-210 targeting of ISCU reduces the TCA cycle and mitochondrial function increasing glycolysis (Favaro et al., 2010).

5.4. Autophagy

miR-155 expression is increased in hypoxia and induced autophagy in cervical and nasopharyngeal cancer cell lines (Wan et al., 2014). A number of MTOR signalling pathway members, RHEB, RICTOR and RPS6KB2, are direct targets of miR-155. MTOR signalling activates the anabolic processes required for cell proliferation and inhibits catabolic processes including autophagosome formation (Wan et al., 2014). Therefore, miR-155 also reduces cell proliferation and induced G1phase cell cycle arrest (Wan et al., 2014). miR- 96 is increased in prostate cancer cells in hypoxia and stimulates autophagy by targeting mTOR ; however, higher expression reduced autophagy by directly targeting ATG7 (Ma et al., 2014). miR-137 is reduced in response to hypoxia and inhibits mitophagy (selective autophagy of mitochondria) without impacting autophagy (Li et al., 2014). miR- 137 reduces the expression of NIX and FUNDC1 which are mitophagy receptors and therefore reduces the interaction between LC3 and mitochondria. The reduction of miR- 137 in hypoxia increases the expression of NIX and FUNDC1 in these conditions and results in increased mitophagy (Li et al., 2014).

5.5. Invasion and metastasis

miR-210 is frequently increased in hepatocellular carcinoma (HCC) and promotes the migration and invasion of HCC cell lines (Ying et al., 2011). Vacuole membrane protein 1 (VMP1) is a direct target of miR-210 and its downregulation by miR-210 in hypoxia enables hypoxia induced HCC metastasis (Ying et al., 2011). miR-191 is increased in hypoxia, under HIF regulation, in breast cancer, and increased the expression of TGFβ2 in hypoxia, increasing migration and invasion inin vitroassays (Nagpal et al., 2015).

5.6. Growth and survival

miR-210 has a role in proliferation targeting MNT, a MAX network transcriptional repressor, and E2F3 (Qin et al., 2014).

Through its regulation of ISCU, miR-210 knockdown also reduced hypoxic survival and increased apoptosis in hypoxia (Favaro et al., 2010). miR-210 also regulates apoptosis through its regulation of Caspase 8 associated protein 2 (Qin

(9)

et al., 2014). Hypoxia reduces the expression of miR-100 in a HIF1α dependent manner (Blick et al., 2013). miR-100 regulates the expression of FGFR3, and thus as miR-100 levels reduce in hypoxia, FGFR3 levels increase (Blick et al., 2013). Expression of miR-100 in hypoxia reduced MAPK phosphorylation, a downstream signalling target of FGFR3, and decreased proliferation and spheroid growth of bladder cancer cell lines (Blick et al., 2013). Conversely, a study in rats identiﬁed that miRNA-15b expression is signiﬁcantly increased in response to ischemia re-oxygenation of the heart and hypoxia reoxygenation of cardiomyocytes (Liu et al., 2014). miR-15b increased apoptosis by reducing Bcl-2 protein which inhibits cytochrome c release from the mitochondria – a key stage in apoptosis (Liu et al., 2014).

6. Hypoxic regulation of lncRNAs and their impact on cancer biology

There is growing evidence of the role and function of lncRNAs in regulating health and diseases. LncRNAs present diverse regulatory functions, including chromatin modiﬁ- cation, genomic imprinting, and transcriptional interference and activation (Geisler and Coller, 2013). Many lncRNAs are aberrantly expressed in cancer and are involved in tumour formation and progression (Gutschner and Diederichs, 2012).

Our understanding of the regulation and functions of lncRNAs in response to different cellular stresses, including hypoxia, is still evolving. Recently, hypoxia has been shown to alter a number of lncRNAs, including NEAT1, H19, HOTAIR, MALAT1, and UCA1.

In response to hypoxia, lncRNAs are either up- or downregulated and regulate many key signalling path- ways, which contribute to tumour hallmarks(Fig. 2). The

regulatory role of lncRNAs in hypoxia can be broadly divided into two types(Table 2). Firstly, hypoxia/HIF regulated lncRNAs. Secondly, those lncRNAs such as aHIF-1α, linc- ROR, and lincRNA-p21 which can directly or indirectly regulate the HIF pathway. In this section, we will brieﬂy de- scribe the role of lncRNAs during hypoxia and/or in the HIF pathway and their impact on tumorigenesis.

6.1. H19

The imprinted maternally expressed transcript (H19) was among the first reported hypoxia-regulated lncRNAs. HIF- 1α-dependent H19 expression is significantly induced in hepatocellular and bladder carcinoma in response to hypoxia (Matouk et al., 2007). H19 knockdown in Hep3B xenograft models resulted in a significant reduction in tumour volume.

In follow up studies, Matouk et al.showed that H19 expression is dependent on the mutational status of the tumour suppressor p53. During hypoxia, H19 expression is not induced in p53 wild type (wt) cells, while it is signiﬁ- cantly induced in p53 null cells, indicating that p53 mutation status may inﬂuence lncRNA expression under hypoxia (Matouk et al., 2010). H19 overexpression causes the upregulation of genes involved in tumour angiogenesis, cell survival, and proliferation (Matouk et al., 2013, 2014). High expression of H19 promotes several cancers, including gastric (Yang et al., 2015), bladder (Luo et al., 2013), oesophageal (Huang et al., 2015), lung (Matouk et al., 2015), liver (Fellig et al., 2005), and breast cancer (Berteaux et al., 2005).

6.2. LincRNA-p21

Usingin silicoanalyses, Yang et al. identiﬁed lncRNAs with potential Hypoxia Response Elements (HREs) within their promoter (Yang et al., 2014). These lncRNAs include LincRNA- p21, lnc-ROR, MALAT1, NEAT1, HOTAIR, and A7. The expression of lincRNA-p21, a promoter-associated lncRNA, was the most induced under hypoxic conditions and was dependent on HIF1α in in HeLa cells (Yang et al., 2014).

LincRNA-p21 binds to both HIF1α and VHL, disrupting the VHL- HIF1α interaction and generating a positive feedback loop that further enhances HIF1α accumulation during hypoxia (Yang et al., 2014). In addition, exogenous transfection of HIF1α into lincRNA-p21 knocked down cells reversed the effect of lincRNA-p21 knockdown on glucose consumption, production of lactate, and expression of glycolytic proteins, suggesting that lincRNA-p21 promotes glycolysis to enhance ATP generation through HIF1α.

LincRNA-p21 overexpression promotes tumour growth in xenograft models (Yang et al., 2014) and enhances the sensitivity of colon cancer to radiotherapy (Wang et al., 2014).

Moreover, studies indicated that lncRNA-p21 plays an important role in the cis and trans regulation of gene expression and induces cell cycle arrest and apoptosis (Hall et al., 2015).

Although Yang et al. has shown a novel role of lncRNA in metabolism reprograming under hypoxia, there are a number of limitations which require further investigation (Yang et al., 2014). For example, hypoxic induction of lincRNA-p21 is not detected and reported in other published transcriptome and lncRNA analyses datasets in hypoxia. Since lincRNA-p21 modulates the Warburg effect, one could postulate that Fig. 2. The impact of lncRNAs in response to hypoxia on the hallmarks of

cancer.

43 H. Choudhry et al. / Molecular Aspects of Medicine 47–48 (2016) 35–53

(10)

Table 2

Hypoxia-regulated lncRNAs, lncRNAs regulating the HIF pathway, and their role in tumorigenesis.

lncRNA Cancer type Regulation HIF Cancer impact Reference

Hypoxia differentially regulated lncRNAs

NEAT1 Breast cancer Upregulated

under hypoxia

HIF2α • Cell survival

• ^Apoptosis

• Associated with poor clinical outcome in breast cancer

Choudhry et al. (2015)

MALAT1 Breast cancer Upregulated under hypoxia HIF1α • ^{Cell cycle}

• Tumour metastasis

• Angiogenesis

• Associated with poor clinical outcome in multiple cancers

Choudhry et al. (2014);Michalik et al. (2014)

HOTAIR Lung cancer Upregulated under hypoxia HIF1α • Cell viability

• ^Invasion

• ^Apoptosis

• Associated with poor clinical outcome in multiple cancers

Zhou et al. (2015)

H19 Hepatocellular carcinoma, bladder cancer

Upregulated under hypoxia HIF1α • Cell survival and proliferation

• Cell migration

• Tumour angiogenesis

Matouk et al. (2007, 2010)

UCA1 Bladder cancer Upregulated under hypoxia HIF1α • Cell proliferation

• Invasion and migration

• ^Apoptosis

• Association with clinical outcome of bladder cancer

Xue et al. (2014)

EFNA3 Breast cancer Upregulated under hypoxia HIF1α/

HIF2α

• ^{RNA sponge}

Gomez-Maldonado et al. (2015)

AK058003 Gastric cancer Upregulated under hypoxia — • Migration and invasion

Wang et al. (2014)

WT1 Myeloid leukaemia Upregulated under hypoxia HIF1α • Epigenetic regulation

• could regulate VEGF

McCarty and Loeb (2015)

HINCUT-1 (c.475) Colon cancer Upregulated under hypoxia HIF1α • Cell proliferation Ferdin et al. (2013)

LncRNAs regulate HIF pathway

aHIF-1a Multiple cancer types Upregulated under hypoxia HIF1α • Downregulates HIF1α Rossignol et al. (2002);

Thrash-Bingham and Tartof (1999);Uchida et al. (2004) lncRNA-LET Gallbladder cancer, squamous-cell

lung cancer, hepatocellular carcinoma and colorectal cancers

Downregulated under hypoxia • Regulates HIF1α level

• ^Apoptosis

• Cell invasion and metastasis

Ma et al. (2014);Yang et al.

(2013)

HIF2PUT Osteosarcoma • Upregulates HIF2α

• Cell migration and invasion

• Cancer stem cell proliferation

Wang et al. (2015)

Linc-ROR Hepatocellular carcinoma Upregulated under hypoxia • Upregulates HIF1α

• ^{RNA sponge}

• Cell-to-cell communications

• Cell survival

Takahashi et al. (2014)

LincRNA-p21 Cervical cancer, breast cancer Upregulated under hypoxia HIF1α • Upregulates HIF1α

• Enhances glycolysis

• Tumour growth

Yang et al. (2014)

ENST00000480739 Pancreatic cancer • Downregulates HIF1α

• Tumour invasion

• Cancer metastasis

• Association with clinicopathologic features and clinical outcome of pancreatic cancer

Sun et al. (2014)

RERT Hepatocellular carcinoma • Downregulates HIF

• Tumour growth

Zhu et al. (2012)

H.Choudhryetal./MolecularAspectsofMedicine47–48(2016)35–53

(11)

lincRNA-p21 interaction with HIF occurs in normoxia to maintain a constitutively regulation of HIF and this interaction is not exclusive to hypoxia. These ﬁndings should be investigated in a range of other cell types to conﬁrm wide- spread role of lncRNA-p21 regulation in metabolism reprograming.

6.3. HINCUTs

Ferdin et al. identiﬁed ﬁve highly induced lncRNAs under hypoxia, which are transcribed from ultra-conserved regions, named ‘hypoxia-induced non-coding ultra-conserved transcripts’ (HINCUTs) (Ferdin et al., 2013). These transcripts are overexpressed in HCT-116, and MCF-7 under hypoxia and in primary colorectal tumours. HINCUT-1 (c.475) is among the most highly induced lncRNAs under hypoxia and is transcribed as a retained intron of O-linked N-acetylglucosamine transferase (OGT) mRNA (Ferdin et al., 2013). HINCUT-1 suppression leads to the reduction of cell proliferation, with G2/M blockade. In addition, HINCUT-1 knockdown reduced OGT gene and protein expression, which was more evident under hypoxia. The authors also demonstrated that HINCUT-1 potentially presented an enhancer-like function in normoxia and hypoxia (Ferdin et al., 2013).

6.4. lncRNA-LET

lncRNA-Low Expressed in Tumour (lncRNA-LET) was found to be expressed at low levels in different types of tumours, including squamous-cell lung cancer, hepatocellular carcinoma (HCC), and colorectal cancer when compared to their heathy counterparts (Yang et al., 2013). Under hypoxia, lncRNA-LET expression is supressed due to activation of histone deacetylase 3 (HDAC3), which consequently decreases histone H3 and H4 acetylation levels on lncRNA- LET promoter. lncRNA-LET expression was inversely correlated with CA9 expression (hypoxia marker) in primary HCC. In addition, lncRNA-LET plays an important role in the accumulation and stabilisation of nuclear factor 90 (NF90) and HIF1α (Yang et al., 2013). Low expression of lncRNA-LET was signiﬁcantly correlated with tumour micrometastases and encapsulation in HCC, suggesting a role for lncRNAs in hypoxia-induced tumour metastasis. More- over,in vivoxenograft models showed that lncRNA-LET low expressing cells presented high tumour colonisation and invasion potential and metastases in mice compared to cells overexpressing lncRNA-LET (Yang et al., 2013).

In another study, Ma et al. reported a signiﬁcant downregulation of lncRNA-LET in gallbladder cancer tissues compared to matched normal tissues. In primary gallbladder cancer, the low expression of lncRNA-LET was associated with low differentiated histology, advanced tumour status, and nodal status when compared to patients with high lncRNA-LET expression (Ma et al., 2014). Multivariate analysis revealed that lncRNA-LET could be used as an independent prognostic marker for metastasis and death in patients with gallbladder cancer. lncRNA-LET was signiﬁ- cantly downregulated in gallbladder cancer cell lines under hypoxia. lncRNA-LET knockdown in SGC-996 cells pro- moted invasion under both hypoxic or normoxic conditions.

However, overexpression or knockdown of lncRNA-LET did

not alter cell proliferation under normoxic conditions, while, under hypoxia, lncRNA-LET overexpression signiﬁcantly reduced cell viability and lncRNA-LET suppression mark- edly increased viability (Ma et al., 2014). In addition, cell cycle analysis indicated that lncRNA-LET possibly inhibits gallbladder cancer cell proliferation through upregulating G0/G1 arrest and also apoptosis (Ma et al., 2014).

6.5. HOTAIR

HOX transcript antisense intergenic RNA (HOTAIR) is a well-known lncRNA located on chromosome 12 on the opposite strand of the Homeobox C Cluster (HOXC)gene locus.

HOTAIR is an oncogenic lncRNA and a negative prognostic factor for a range of cancer types, including non-small cell lung cancer (NSCLC), breast, cervical, and endometrial cancer (Zhang et al., 2014, 2015). Zhou et al.reported a signiﬁ- cant increase in the transcription of HOTAIR in NSCLC cells under hypoxia (Zhou et al., 2015). The induction of HOTAIR under hypoxia was dependent on HIF1α. Functional analysis after HOTAIR suppression suggests that HOTAIR increases cell viability and invasion and inhibits apoptosis of hypoxic NSCLC cells (Zhou et al., 2015). Moreover, HOTAIR acts by downregulating the expression of a number of tumour suppressor genes such asJAM2,PCDH10, andPCDHB5 (Croce, 2010; Kogo et al., 2011). Increased HOTAIR expression is found in metastatic breast cancer (Chisholm et al., 2012; Gupta et al., 2010), liver cancer (Ishibashi et al., 2013), colon cancer (Kogo et al., 2011), and NSCLC (Ono et al., 2014).

6.6. WT1

McCarty and Loeb reported the transcription of an antisense-oriented non-coding RNA (WT1 lncRNA) that over- laps with the intron 1 CpG island of the Wilms Tumor 1 (WT1) gene (McCarty and Loeb, 2015). WT1 is a transcription factor involved in haematopoiesis and kidney development (Cunningham et al., 2013; Kreidberg, 2010).

Alteration in WT1 expression results in the development of different cancers (Qi et al., 2015; Yang et al., 2007). McCarty and Loeb found that hypoxia causes demethylation of the intron 1 CpG island, which consequently leads to the expression of both WT1mRNA and WT1 lncRNA in acute myeloid leukaemia (AML) cell lines and in primary AML samples (McCarty and Loeb, 2015). In addition, they demonstrated that WT1 lncRNA suppression reduces the hypoxia-mediated upregulation of WT1 mRNA. Therefore, WT1 lncRNA expression is crucial for hypoxia-mediated WT1 expression. Moreover, HIF regulates WT1 expression during hypoxia through methylcytosine dioxygenase (TET2 and TET3) induction (McCarty and Loeb, 2015). This was the ﬁrst evidence of hypoxia-mediated epigenetic changes in DNA methylation inducing the expression of a lncRNA through cis-regulation.

6.7. AK058003

Using microarray on hypoxia-induced gastric cancer cell lines, Wang et al. identiﬁed a number of hypoxia responsive lncRNAs in gastric cancer. An intronic antisense lncRNA called lncRNA-AK058003 was among the most induced 45 H. Choudhry et al. / Molecular Aspects of Medicine 47–48 (2016) 35–53

(12)

lncRNAs upon hypoxia treatment in all examined gastric cancer cell lines (Wang et al., 2014). LncRNA-AK058003 expression was significantly increased in primary gastric cancer compared with adjacent benign tissues. Functional analysis showed that lncRNA-AK058003 is a positive metastatic regulator of gastric cancer by enhancing tumour migration and invasion bothin vivoandin vitro. The authors identified γ-synuclein (SNCG), a metastasis-associated protein, which is epigenetically downregulated in lncRNA- AK058003 suppressed cells (Wang et al., 2014). These data suggest that SNCG may act as a downstream target of lncRNA-AK058003 in gastric cancer. Furthermore, the expression of SNCG and lncRNA-AK058003 were positively correlated in primary gastric cancer. Under hypoxia, the upregulation of lncRNA-AK058003 promotes the induction of SNCG, which mediates hypoxia-induced metastasis (Wang et al., 2014). lncRNA-AK058003 was also found to be significantly highly expressed in breast cancer where it induced tumour proliferation, invasion, and metastasis (He and Wang, 2015).

6.8. HIF2PUT

Wang et al.identiﬁed cis-regulatory lncRNAs using bioinformatic analyses in osteosarcoma cancer stem cells (Wang et al., 2015). They predicted a possible association between lncRNAs and cancer-associated genes transcribed within a 10 kbp upstream or downstream region. They reported an antisense lncRNA (TCONS_00004241) transcribed upstream of the promotor region ofHIF2αmRNA. Thus, the lncRNA TCONS_00004241 was named HIF2α promoter upstream transcript (HIF2PUT) (Wang et al., 2015). HIF2PUT expression was co-regulated withHIF2αmRNA in primary osteosarcoma tumours and cancer cell lines. HIF2PUT suppression signiﬁcantly increases the expression rate of the cancer stem cell marker CD133 in MG63 cells and con- curred with elevated rate of spheroid-formation, cell migration, and invasion of MG63 cells. An opposite effect was observed with HIF2PUT overexpression (Wang et al., 2015). In summary, lncRNA-HIF2PUT suppresses cancer stem cell properties in osteosarcoma, partly by controllingHIF2α mRNA. However, the study did not evaluate the expression and role of lncRNA-HIF2PUT under hypoxia, when HIF2α is activated.

6.9. ENST00000480739

Through microarray analysis, Sun et al.profiled lncRNAs in pancreatic ductal adenocarcinoma (PDAC) and identified a novel lncRNA, ENST00000480739, which was expressed at significantly low levels in pancreatic cancer tissues when compared to its expression in adjacent normal tissues as well as in PDAC cell lines (Sun et al., 2014).

ENST00000480739 overexpression signiﬁcantly increases the expression of a downstream gene, osteosarcoma ampliﬁed- 9 (OS-9). Other studies indicated that OS-9 negatively regulates HIF1α by enhancing the interaction of HIF-1α and PHD2/3 (Baek et al., 2005). Sun et al.hypothesised that ENST00000480739 causes the suppression of HIF1α indirectly through regulating OS-9. Indeed, they showed that OS-9 suppression during hypoxia abrogates

ENST00000480739-induced downregulation of HIF1α (Sun et al., 2014). Moreover, ENST00000480739 overexpression signiﬁcantly decreases the expression of HIF1α target genes, includingMXI-1, PDGFC, FN1, andMMP28. These data indicate that ENST00000480739 may negatively modulate HIF1α activity by inducing OS-9 expression. In PDAC clinical samples, ENST0000048073 and OS-9 were co-regulated. As- sociation with clinicopathologic features revealed that ENST00000480739 expression was negatively correlated with the TNM stage and with lymph node metastasis (Sun et al., 2014). Additionally, following surgery, the survival rate of patients with PDAC, presenting a high expression of ENST00000480739 was signiﬁcantly higher than that of those who presented a low expression of ENST00000480739.

In a tumour mouse model, the metastatic pancreatic cancer SW1990 cells overexpressing lncENST00000480739 were less observed in the lungs of nude mice, compared to mice injected with SW1990 control cells (low expression of lncENST00000480739) which presented signiﬁcant lung metastases (Sun et al., 2014). Altogether, these data indicate the potential role of lncRNA ENST00000480739 in the induction of pancreatic cancer metastasis via indirect regulation of HIF.

6.10. EFNA3 lncRNA

Gómez-Maldonado et al.reported glycosyl phosphatidyl- inositol-linked membrane-bound ligand (EphrinA3) protein accumulation in hypoxia through upregulation of lncRNAs encoded by theEFNA3gene locus. Using publicly available expressed sequence tags data and 5′ and 3′-rapid amplifi- cation of cDNA ends (RACE), they found two novel TSSs in addition to EphrinA3 mRNA, suggesting the existence of other RNAs transcribed from the EFNA3 locus (Gómez-Maldonado et al., 2015). Both novel RNAs lack a functional protein reading frame and, accordingly, were defined as lncRNAs. Under hypoxia, these lncRNAs were strongly upregulated. However,EFNA3mRNA induction was only minimal. The induction of EFNA3 lncRNA in response to hypoxia depended on HIF, while that ofEFNA3mRNA did not. Interestingly, under hypoxia, overexpression of the EFNA3 lncRNA had a modest effect onEFNA3mRNA expression. However, it caused a significant increase in EFNA3 protein accumulation (Gómez-Maldonado et al., 2015). The authors proposed that EFNA3 lncRNA modulates the EphrinA3 protein accumulation through endogenous RNA competition.EFNA3is a target gene for hypoxia-inducedmiR- 210(Wang et al., 2013). It averts the translation ofEFNA3 mRNA by binding to the 3′-untranslated region. Therefore, EFNA3 lncRNA may promoteEFNA3mRNA translation by de- pleting EFNA3 target microRNAs, includingmiR-210(Wang et al., 2013). In addition, the cumulative expression of the EFNA3mRNA and lncRNA was significantly higher in clear cell renal cell carcinoma (ccRCC) tumours as compared with matched normal kidney cells. Moreover, usingin vivo xenotransplantation models, the authors demonstrated that the expression of eitherEFNA3mRNA or lncRNA induces tumour metastasis through efficient intravasation and ex- travasation of the vascular wall (Gómez-Maldonado et al., 2015). This study demonstrates that hypoxia induces HIF dependent EFNA3 lncRNA, which directly regulates EFNA3