Targeting liver cancer stem cells for the treatment of hepatocellular carcinoma

Abstract

Liver cancer is one of the most common malignant tumors and prognosis remains poor. It has been increasingly recognized that liver cancer stem cells (LCSCs) are responsible for the carcinogenesis, recurrence, metastasis and chemoresistance of hepatocellular carcinoma (HCC). Targeting LCSCs is promising to be a new direction for the treatment of HCC. Herein, we summarize the potentially therapeutic targets in LCSCs at the level of genes, molecules and cells, such as knockout of oncogenes or oncoproteins, restoring the silent tumor suppressor genes, inhibition of the transcription factors and regulation of noncoding RNAs (including microRNAs and long noncoding RNAs) in LCSCs at the genetic level; inhibition of markers and blockade of the key signaling pathways of LCSCs at the molecular level; and inhibiting autophagy and application of oncolytic adenoviruses in LCSCs at the cellular level. Moreover, we analyze the potential targets in LCSCs to eliminate chemoresistance of HCC. Thereinto, the suppression of autophagy and Nanog by chloroquine and shRNA respectively may be the most promising targeting approaches. These targets may provide novel therapeutic strategies for the treatment of HCC by targeting LCSCs.

Introduction

Liver cancer is one of the most common human malignancies in the world, it takes the second place of mortality, only next to lung cancer. Liver cancer includes several subtypes, hepatocellular carcinoma (HCC) is by far the most common worldwide, accounting for 78%, other subtypes such as bile-duct cancer (15%), hepatoblastoma and various liver sarcomas and carcinomas (7%).¹ The main risk factors for HCC are definite, including hepatitis B or C virus infection, alcohol abuse, intake of the fungal metabolite aflatoxin B1 and an emerging cause that named nonalcoholic fatty liver disease or nonalcoholic steatohepatitis (NASH).² In the past few decades, cancer stem cells (CSCs) have similar characteristics to normal stem cells, such as self-renewal and pluripotent activity, which have been found to be associated with the major malignant phenotypes of cancer, including recurrence, metastasis and chemoresistance.³ Liver cancer stem cells (LCSCs) which are identified by certain surface markers, are also known as hepatic cancer stem cells or liver tumor-initiating cells (T-ICs). Recently, the malignant behaviors of LCSCs have been increasingly recognized. They are identified to be responsible for the initiation, relapse, metastasis and chemoresistance of HCC.^3–6 Thus, targeting LCSCs may be a new strategy for the treatment of HCC. This paper reviews the potentially therapeutic targets against LCSCs at the level of genes, molecules and cells, including targets of drug resistance.

Targeting LCSCs at the level of genes

Targeting LCSCs at the genetic level includes knockout of oncogenes or oncoproteins, and restoration of the silent tumor suppressor genes (Table 1).

Table 1.

Targets at the level of genes.

Targets	Effect*	Mechanism	Species	References
Oncogenes or oncoproteins
BC047440	+	BC047440 increased the activation of NF-κB signaling.	Mouse	You and colleagues7
14-3-3ζ	+	Knockout of 14-3-3ζ improved IR-induced apoptosis by upregulating the expression of pro-apoptotic proteins.	Human	Lee and colleagues8
HSP90	+	Inhibition of HSP90 induced LCSC apoptosis by downregulating the HSP90 effector proteins.	Mouse	Yang and colleagues9
ANXA3	+	ANXA3 overexpression activated JNK pathway, leading to oncogenic events.	Human	Pan and colleagues10; Tong and colleagues11
Hepatitis B virus PreS 1	+	PreS1 upregulated CSCs-related genes including Klf4, Nanog, Sox2, Oct4, and c-Myc.	Human	Liu and colleagues12
SIRT1	+	SIRT1 regulated the transcriptional activity of Sox2, and was maintained by MEK1 signaling.	Human	Liu and colleagues13; Cheng and colleagues14
ADAM17	+	ADAM17 was a key point to activate Notch signaling pathway, silencing ADAM17 reduced the activity of Notch signaling pathway.	Human	Hong and colleagues15
Tumor suppressor genes
Shp2 and Pten	–	Shp2 and Pten worked together to promoted tumorigenesis of LCSCs by the upregulation of the proto-oncogene c-jun.	Human	Luo and colleagues16
C8orf4	–	C8orf4 impaired the self-renewal of LCSCs through inhibition of NOTCH2 signaling.	Human	Zhu and colleagues17
p53	–	Autophagy suppressed p53 which otherwise could be activated by PINK1 to downregulate the expression of Nanog.	Human	Liu and colleagues18
Numb	–	The phosphorylation of Numb destabilized p53 and promoted self-renewal of LCSCs in a Nanog-dependent manner.	Human	Siddique and colleagues19

^*+, promoting the stemness of LCSCs;–,suppressing the stemness of LCSCs.

ADAM17, a disintegrin and metalloproteinase-17; ANXA3, annexin A3; LCSC, liver cancer stem cell; PINK1, Pten-induced putative kinase 1; Pten, phosphatase and tensin homologue; Shp2, src-homology 2 domain-containing phosphatase 2; SIRT1, histone deacetylatase sirtuin 1.

Knockout or inactivation of the oncogenes or oncoproteins in LCSCs

In recent years, increasing genes are reported to have relations with the biological behaviors of LCSCs. The activation of oncogenes or overexpression of oncoproteins increase the proliferation and tumorigenicity of LCSCs. The functions of these oncogenes and oncoproteins which may be taken as targets against LCSCs are as followed.

BC047440

BC047440 (GeneBank accession number: BC047440) is a novel HCC-related gene, the full length of BC047440 cDNA was cloned from human HCC tissues, the activation of BC047440 contributed to development of HCC.²⁰ You and colleagues⁷ showed that BC047440 played an important role in maintaining stemness properties of LCSCs in the nude mouse model. The oncogenic effect of BC047440 depended on the increasing activation of nuclear factor kappa B (NF-κB), which was a critical nuclear transcription factor (TF) in tumor development. On the other hand, knockout of BC047440 resulted in tumorigenicity inhibition and hepatocyte-differentiation induction of LCSCs through enhancing the expression of hepatocyte nuclear factor 4α (HNF4α), which was important to regulate hepatocyte differentiation.⁷ Removal of the BC047440 gene may counteract the malignant behaviors of LCSCs.

14-3-3ζ

14-3-3ζ belongs to the 14-3-3 family, which is a group of evolutionarily highly conserved acidic proteins.²¹The abnormal expression of 14-3-3ζ has been detected in multiple cellular pathways in cancers. Lee and colleagues⁸ reported that 14-3-3ζ was upregulated in LCSCs after γ-irradiation (IR), it contributed to radiation resistance and survival of LCSCs. Knockout of 14-3-3ζ reduced the stemness properties of LCSCs and improved IR-induced apoptosis by upregulating the expression of pro-apoptotic proteins. The combination of radiotherapy and 14-3-3ζ silencing may be potential strategy for HCC therapy by targeting LCSCs.

HSP90

Heat-shock protein 90 (HSP90) can induce the heat-shock response as a chaperone protein.²² It facilitates cellular proteins folding and degradation, keeps protein stability under oxidative and heat stress. It also plays a critical role in the stabilization and activity of various oncogenic proteins.^22,23 Yang and colleagues⁹ found that HSP90 in CD90⁺ LCSCs was upregulated under hyperthermic condition in nude mice, the HSP90 inhibitor could sensitize CD90⁺LCSCs to hyperthermia and induce CD90⁺LCSCs apoptosis by downregulating the HSP90 effector proteins, which regulated cell survival and apoptosis. Thus, blockage of HSP90 may improve the thermotherapy sensitivity and reduce thermoresistance of LCSCs. HSP90 may become a promising target against LCSCs, especially in thermotherapy of HCC.

Annexin A3 (ANXA3)

ANXA3 is a group of Ca²⁺-dependent phospholipid-binding secretory proteins.²⁴ It contributed to the propagation and self-renewal of LCSCs.¹⁰ Upregulation of ANXA3 promoted the tumorigenicity of LCSCs through aberrant regulated c-Jun NH2-terminal kinase (JNK) pathway, the JNK inhibitor could suppress the oncogenic events caused by ANXA3 overexpression.¹¹ Thus, ANXA3 silencing and blockage of JNK pathway may suppress the malignant behaviors of LCSCs.

Hepatitis B virus PreS 1

Hepatitis B virus (HBV) infection is involved in HCC development. It is the initial step of HBV infection for the HBV envelope protein PreS1 to bind to the specific cellular surface receptor of hepatocytes.²⁵ Liu and colleagues¹² found that PreS1 acted as an oncoprotein in HCC development, it drove the expression of CSCs-related genes including Klf4, Nanog, Sox2, octamer 4 (Oct4), and c-Myc. PreS1 also upregulated CSCs-related markers, such as CD133, CD90 and CD117. Knockout of PreS1 counteracted the stemness properties of LCSCs, indicating that the suppression of Pres1 may be potential method for the therapy of HBV-related HCC through targeting LCSCs.

SIRT1

Histone deacetylatase sirtuin 1 (SIRT1) is class III histone deacetylase, taking a role in the regulation of cellular stress responses.²⁶ SIRT1 participates in the process of carcinogenesis by regulating lipid metabolism. In HCC, Liu and colleagues¹³ reported that SIRT1 promoted self-renewal and tumorigenicity of LCSCs by regulating the transcriptional activity of Sox2, which was essential TF for LCSCs. Knockdown of SIRT1 impaired the stemness capacities of LCSCs. On the other hand, the stabilization of SIRT1 was maintained by MEK1 signaling which belongs to the mitogen-activated protein kinase (MAPK) pathway.¹⁴ So, both SIRT1 and MEK1 signaling may be novel potential targets against LCSCs.

ADAM17

A disintegrin and metalloproteinase-17 (ADAM17), also known as tumor necrosis factor (TNF)-α converting enzyme (TACE), is important in processing single-spanning membrane proteins.¹⁵ Many proteins processed by ADAM17 including cytokines, receptors and growth factors are involved in cancer development. In HCC, it was revealed that ADAM17 contributed to radio resistance and was responsible for invasion and metastasis of CD133⁺ LCSCs. The mechanism was that ADAM17 was a key point to activate Notch signaling pathway, silencing ADAM17 reduced the radio resistance of LCSCs by suppressing Notch signaling.¹⁵ So, silencing ADAM17 may become novel method to improve the efficacy of radiotherapies and reduce the metastasis of HCC.

Restoring the silent tumor suppressor genes in LCSCs

In recent years, it has been recognized that the silencing of tumor suppressor genes caused by mutations, deletions, promoter inactivation or other epigenetic changes are relevant to malignant change in normal cells. Accordingly, it may become a potential strategy for HCC therapy to restore the expression of silent tumor suppressor genes in LCSCs.

Ptpn11/shp2 and Pten

Src-homology 2 domain-containing phosphatase 2 (Shp2) is an oncogenic tyrosine phosphatase, encoded by Ptpn11.²⁷ Shp2 mutation was detected in several types of leukemia.²⁸ The role of Shp2 on HCC development is controversial, it was demonstrated that Shp2 played a role as tumor suppressor in HCC on one hand,²⁹ but on the other hand it promoted HCC development,³⁰ and enhanced the invasion of LCSCs by activating β-catenin signaling.³¹

Phosphatase and tensin homologue (Pten), deleted from chromosome 10, is a classical tumor suppressor which could block the activation of PI3K/Akt signaling pathway.³² Pten and Shp2 work together to exert the inhibitory effect on LCSCs. It was identified that both deficiencies of Shp2 and Pten promoted tumorigenesis in LCSCs, Pten deficiency promoted Akt over-activation and Shp2 loss induced JNK activation, resulting in the upregulation of the proto-oncogene c-jun.¹⁶ Both silencing of Shp2 and Pten were related to poor prognosis in patients with HCC. So, Shp2 and Pten may become novel potential targets against LCSCs, but further research is needed.

C8orf4

C8orf4, also named thyroid cancer 1 (TC1), was cloned from thyroid cancer.³³ Overexpression of C8orf4 was reported being involved in tumorigenesis in several types of human cancers. It could promote clonogenicity of human lung cancer cells and also acts as an oncogene in breast cancer.^17,34,35 But, Zhu and colleagues¹⁷ found an inhibitory effect of C8orf4 in HCC, it impaired the self-renewal of LCSCs through inhibition of Notch2 signaling. It counteracted the nuclear translocation of Notch2 intracellular domain. By contrast, C8orf4 silencing drove the activation of Notch2 signaling, then improved self-renewal properties of LCSCs. Anyhow, C8orf4 was found to act oncogenic role in several other types of cancer, it needs further studies about the role of C8orf4 gene in targeted therapy of HCC based on LCSCs.

p53

As a classical tumor suppressor gene, p53 could restrain stem cells expansion by restricting self-renewal, inhibiting symmetric division and blocking the somatic/progenitor cells reprogramming into stem cells.^36,37 Liu and colleagues¹⁸ found that autophagy, a catabolic process of cells removing damaged organelles and protein aggregates, could positively regulate LCSCs by suppressing p53 which otherwise could be phosphorylated on mitochondria by Pten-induced putative kinase 1 (PINK1), a kinase associated with mitophagy, to downregulate the expression of CSC-related gene Nanog. So, both the restoration of p53 and blockade of autophagy may be promising methods to control LCSCs.

Numb

Numb is originally discovered in drosophila embryos as a cell fate determinant during sensory organ formation.³⁸ Accumulating evidences support that Numb acts as tumor suppressor in several types of cancer by inhibiting Notch signaling and epithelial-mesenchymal transition (EMT).³⁹ In pancreatic cancer, Numb was downregulated by Musashi2, a RNA-binding protein which was required for tumorigenesis as a translational repressor.^40,41 In HCC, Siddique and colleagues¹⁹ reported that Numb could conjunct with p53 forming the Numb-p53 complex to prevent the degradation of p53, the phosphorylation of Numb destabilized p53 and promoted self-renewal of LCSCs in a Nanog-dependent manner. Nanog phosphorylated Numb by enhancing the kinase activities of both Aurora A kinase (AURKA) and atypical protein kinase C zeta (aPKCζ) through the Nanog-AURKA-aPKCζ pathway. Thus, the Nanog-Numb-p53 signaling axis is important in the self-renewal and tumorigenesis of LCSCs, it may provide novel strategies for the treatment of HCC.

Targeting LCSCs at the level of molecules

Targeting LCSCs at the molecular level includes inhibition of the TFs, regulation of noncoding RNA (including microRNA and long noncoding RNA), decrease of markers and interruption of the essential signaling pathways of LCSCs (Table 2).

Table 2.

Targets at the level of molecules.

Targets	Effect*	Mechanism	Species	References
TFs
Twist	+	Twist is associated with EMT and self-renewal of LCSCs by regulating the CSCs marker CD24.	Human	Liu and colleagues42; He and colleagues43; Ren and colleagues44
HIF-1 and ELK3	+	ELK3 promoted the migration and invasion of LCSCs by targeting HIF-1α.	Human	Lee and colleagues45
KLF8	+	KLF8 facilitated the activation of the Wnt/β-catenin signaling pathway.	Human	Shen and colleagues46
Noncoding RNAs
MicroRNAs (miRNAs)
MiR-122	–	MiR-122 suppressed Wnt/β-catenin signaling pathway and inhibited glycolysis by targeting glycolytic genes PDK4.	Human	Song and colleagues47
MiR-152	–	MiR-152 directly binding to 3′ untranslated region and downregulating the expression of KIT which is a proto-oncogene.	Human	Huang and colleagues48
MiR-21	–	MiR-21 positively regulated the expression of LCSC markers CD13, EpCAM, CD90 and Oct4.	Human	Jiang and colleagues49
MiR-155	+	MiR-155 mediated TP53INP1 to regulate CSC phenotype, and enhanced the LCSC markers CD90, CD133 and Oct4.	Human	Liu and colleagues50
MiR-25	+	Knockdown of miR-25 increased apoptosis of LCSCs induced by TRAIL, via Pten/PI3K/Akt/Bad signaling pathway.	Human	Feng and colleagues51
MiR-200 family	+/–	MiR-429 decreased RBBP4 expression and resulted in the activation of Oct4. But, miR-200a suppressed the EMT phenotype of LCSCs.	Human	Li and colleagues52; Wang and colleagues53
MicroRNA let-7	–	Let-7a negatively regulating EMT and Wnt signaling pathway. Let-7c targeted PBX3 and suppressed the transcriptional activity of CSCs-related genes including CACNA2D1, EpCAM, Sox2 and Notch3.	Human	Jin and colleagues54; Han and colleagues55
MiR-1246	+	MiR-1246 activated the Wnt/β-catenin pathway through inhibiting the expression of Axin2 and GSK3β.	Human	Chai and colleagues56
Long noncoding RNAs (lncRNAs)
HULC and MALAT1	+	They cooperated to regulate the TRF2.	Human	Wu and colleagues57
LncDILC	–	LncDILC inhibited the autocrine IL-6/STAT3 signaling, and mediated the crosstalk between TNF-α/NF-κB signaling and IL-6/JAK2/STAT3 cascade.	Human	Wang and colleagues58
CUDR and H19	+	Pten depletion promoted the binding of CUDR to the oncogene CyclinD1, the CUDR-cyclinD1 complex then enhanced the H19 expression.	Human	Pu and colleagues59
HOTAIR	+	HOTAIR accelerated LCSC malignant proliferation through downregulating SETD2.	Human	Li and colleagues60
LncTCF7	+	LncTCF7 recruited the SWI/SNF complex to activation of Wnt signaling.	Human	Wang and colleagues61
LncSox4	+	LncSox4 recruited the TF Stat3 to the Sox4 promoter to trigger the expression of Sox4.	Human	Chen and colleagues62
LncBRM	+	lncBRM associated with BRM to trigger the BRG1/BRM switch and BAF, leading to activation of the transcriptional cofactors YAP1.	Human	Zhu and colleagues63
Lncβ-Catm	+	Lncβ-Catm associated with β-catenin and the methyltransferase EZH2, promoting the methylation and stability of β-catenin.	Human	Zhu and colleagues64
LncCAMTA1	+	LncCAMTA1 associated with CAMTA1 promoter to inhibit its transcription.	Human	Ding and colleagues65
LCSC biomarkers
CD133	+	The downregulation of CD133 decreased the level of NF-κB.	Human	Liu and colleagues66
ICAM-1	+	ICAM-1 is upregulated by Nanog, promoting the stemness of LCSCs.	Human	Liu and colleagues67
Signaling pathways
Wnt/β-catenin pathway	+	The Wnt/β-catenin pathway promoted the self-renewal and unlimited cell proliferation of CSCs.	Human	Chen and colleagues68; Kim and colleagues69; Seto and colleagues70
PI3K/Akt/mTOR pathway	+	HBV X protein facilitates AFP expression, which activates PI3K/Akt signal pathways.	Human	Zhu and colleagues71
Akt/GSK-3β/β-catenin pathway	+	Inhibition of the protein kinaseAkt reduced the self-renewal of LCSCs.	Human	Xu and colleagues72; Zhai and colleagues73; Kim and colleagues69
STAT3 signaling pathway	+	TAMs produced IL-6, activating STAT3 and elevating the cellular glucose uptake. TLR4 cooperated with STAT3 via Nanog to activate Twist1.	Human	Zhang and colleagues74; Wan and colleagues75; Uthaya and colleagues76
RAS/RAF/ERK pathway	+	Depleting MEK or reducing ERK1/2 phosphorylation suppressed the proliferation, invasion and migration of LCSCs. MEK maintained the stabilization of SIRT1 protein.	Human	Galuppo and colleagues77; Sun and colleagues78; Cheng and colleagues14
JNK signaling pathway	+	ANXA3 could enhance the activity of JNK pathway in CD133+LCSCs by upregulating the expression of c-MYC.	Human	Tong and colleagues11
Notch signaling pathway	+	The Notch signaling cascade associated with Wnt, MAPK and NF-κB signaling.	Human	Luo and colleagues79; Wang and colleagues80

^*+, promoting the stemness of LCSCs; –, suppressing the stemness of LCSCs.

BAF, BRG1-associated factor; BRM, Brahma; CAMTA1, the calmodulin binding transcription activator 1; CUDR, cancer upregulated drug resistant; DILC, downregulated in LCSCs; ELK3, Net/SAP-2/Erp; GSK3β, glycogen synthase kinase 3β; HIF-1, hypoxia-inducible factor 1; HOTAIR, HOX transcript antisense RNA; HULC, highly upregulated in liver cancer; ICAM-1, intercellular adhesion molecule 1; KLF8, Krüppel-like factor 8; LCSC, liver cancer stem cell; MALAT1, nuclear-enriched transcript 2 (NEAT2); NF-κB, nuclear factor-κB; RBBP4, Rb binding protein 4; STAT3, signal transducer and activator of transcription 3; TAM, tumor-associated macrophage; TF, Transcription factor; TLR4, Toll-like receptor 4; TP53INP1, tumor protein 55-induced nuclear protein 1; TRAIL, tumor necrosis factor-related apoptosis inducing ligand; TRF2, telomere repeat binding factor 2.

Inhibition of the key TFs in LCSCs

Twist

The Twist proteins belong to the highly conserved basic heli-loop-helix TF family; the Twist genes include Twist1 and Twist2.⁸¹ It is reported that Twist is associated with EMT and self-renewal of LCSCs by regulating the CSCs marker CD24, promoting the development of HCC.^42,82 Some phytochemicals show inhibitory effect on Twist signaling in LCSCs, such as casticin and 8-bromo-7-methoxychrysin (BrMC). Casticin is derived from Fructus Viticis (Chinese name, Manjingzi).⁴³ BrMC is a novel synthetic analogue of chrysin (5,7-dihydroxyflavone).⁴⁴ Both casticin and BrMC could inhibit EMT and the stemness of LCSCs by negatively regulating Twist.^43,44 Thus, Twist inhibitors might be therapeutic agents by targeting LCSCs.

HIF-1α and ELK3

Hypoxia-inducible factor 1 (HIF-1) is a basic-helix-loop-helix-PAS heterodimeric TF which mediates transcriptional responses in hypoxic cells.⁸³ HIF-1 is composed of two subunits, the HIF-1α and HIF-1β. HIF-1α is identified an important role in tumor development, including the regulation of oncogenes expression, cellular metabolism and proliferation, tumor metastasis and invasion.⁴⁵ Zou and colleagues⁸⁴ showed that the downregulation of HIF-1α inhibited the biological characteristics of LCSCs, such as self-renewal, migration, invasion.

The TF ELK3, also named Net/SAP-2/Erp, is part of the large family of ETS-domain TFs, belonging to the ternary complex factor subfamily.⁸⁵ ELK3 is associated with wound healing, angiogenesis, and tumor growth as a downstream of the RAS/ERK signaling pathway.⁸⁶ Lee and colleagues⁴⁵ showed that ELK3 promoted the migration and invasion of LCSCs by targeting HIF-1α. So, both HIF-1α and ELK3 might act as the therapeutic targets for HCC treatment.

KLF8

Krüppel-like factor 8 (KLF8) belongs to the KLF family of TFs as DNA-binding transcriptional regulators in many cellular processes.⁸⁷ KLF8 was highly expressed not only in LCSCs, but also in HCC tumors and distant migrated tissues.⁴⁶ Overexpression of KLF8 promoted the maintenance and chemoresistance of LCSCs, accordingly, played a role in HCC tumorigenesis. Mechanistically, KLF8 facilitated the activation of the Wnt/β-catenin signaling pathway.⁴⁶ The knockout of KLF8 gene showed markedly apoptosis of LCSCs. So, KLF8 could be taken as a novel target to suppress LCSCs.

Targeting noncoding RNAs in LCSCs

Noncoding RNAs include microRNAs (miRNAs) and long noncoding RNAs (lncRNAs). MiRNAs are a class of small noncoding RNAs which can post-transcriptionally regulate the expression of target genes and play roles as tumor suppressors or oncogenes in different tumorigenic progression.⁸⁸ Long noncoding RNAs (lncRNAs) are >200 bases long and have low or no protein-coding potential.⁸⁹ They are known to regulate splicing, recruit TFs, and regulate mRNA stability. Increasing studies demonstrate that both miRNAs and lncRNAs play important roles in LCSCs, they may be taken as therapeutic targets for the treatment of HCC.

Regulation of microRNAs in LCSCs

a. MicroRNA 122

MicroRNA 122 (miR-122) is a liver-specific miRNA, accounting for 70% of the total miRNAs in human liver.⁹⁰ MiR-122 was identified as a tumor suppressor in HCC.⁴⁷ Upregulation of miR-122 inhibited EMT, proliferation and invasion of hepatoma cells by suppressing the Wnt/β-catenin signaling pathway.⁹¹ Moreover, miR-122 inhibited glycolysis by targeting glycolytic genes PDK4, accordingly, suppressed the stemness properties and growth of CD133⁺ LCSCs.⁴⁷

b. MicroRNA 152

MicroRNA-152 (miR-152) is involved in diverse biological functions and induces apoptosis in HCC as a tumor suppressor.^48,92,93 In HBV-related HCC, the decrease of miR-152 contributed to the epigenetic inactivation of RIZ1, a candidate HCC suppressor gene which could be repressed by HBV X protein.⁹³ In LCSCs, miR-152 inhibited clonogenicity and growth of CD133⁺LCSCs by directly binding to 3’ untranslated region and downregulating the expression of KIT which is a proto-oncogene.⁴⁸

c. MicroRNA 21

MicroRNA 21 (miR-21) was identified as a pro-metastatic miRNA, which regulated the expression of metastasis-related proteins in HCC.⁹⁴ MiR-21 was upregulated in LCSCs.⁹⁵ Overexpression of miR-21 promoted the tumorigenesis, invasion and migration of LCSCs and positively regulated the expression of Oct4 and LCSC markers CD13, EpCAM and CD90.⁴⁹ MiR-21 played an oncogenic role in LCSCs which might be taken as a target.

d. MicroRNA 155

MicroRNA 155 (miR-155) could either be an oncogene or a tumor suppressor in several type of cancers.⁹⁶ In HCC, miR-155 was demonstrated to contribute to tumorigenesis.⁹⁷ MiR-155 facilitated the formation and self-renewal of LCSCs by mediating TP53INP1(tumor protein 55-induced nuclear protein 1) to regulate CSC phenotype.⁵⁰ Moreover, the overexpression of miR-155 enhanced the levels of Oct4 and LCSC markers CD90 and CD133, knockdown of miR-155 led to a decrease of LCSCs populations.⁵⁰

e. MicroRNA 25

Abnormal expression of MicroRNA 25 (miR-25) mediates tumorigenesis in many human malignant tumors.⁹⁸ In HCC, miR-25 played a role in promoting proliferation, invasion and migration of hepatoma cells.⁹⁹ It was overexpressed in LCSCs compared with non-LCSCs, knockdown of it increased apoptosis of LCSCs induced by TNF-related apoptosis inducing ligand (TRAIL), via the Pten/PI3K/Akt/Bad signaling pathway⁵¹. It indicates that both miR-25 and TRAIL may become novel anticancer agents by targeting LCSCs.

f. MicroRNA 200 family

MicroRNA 200 family (miR-200 family) consist of five evolutionarily conserved members in the human genome with miR-200b, miR-200a and miR-429 located in 1p36 cluster, and miR-200c, miR-141 located in 12p13 cluster.¹⁰⁰ MiR-429 was demonstrated as oncogene in HCC, its upregulation promoted the self-renewal, tumorigenicity and chemoresistance of EpCAM⁺LCSCs, it led to the activated transcription of Oct4 by E2F transcription factor 1 (E2F1) via reducing Rb binding protein 4 (RBBP4) expression, which otherwise could inhibit E2F1 transcriptional activity.⁵² Whereas, miR-200a acted as tumor suppressor, its upregulation suppressed the EMT phenotype of LCSCs by increasing the epithelial marker E-cadherin and decreasing the mesenchymal markers, fibronectin, vimentin, and N-cadherin.⁵³ Moreover, miR-200b/miR-200c/miR-429 subfamily was found to suppress HCC metastasis by inhibiting Rho/ROCK signaling pathway.¹⁰¹ However, the miR-200 family play indispensable roles in maintenance of LCSCs which still need further studies.

g. MicroRNA let-7

The let-7 family consists of 12 members: let-7a-1, -2, -3; let-7b; let-7c; let-7d; let-7e; let-7f-1, -2; let-7g; let-7i; MIR98, most of them are found to be downregulated in several cancers, restoration of normal expression prevents tumorigenesis.¹⁰² In HCC, let-7a and let-7c were identified as tumor suppressors by targeting LCSCs.^54,55 Let-7a suppressed the self-renewal of LCSCs by negatively regulating EMT and Wnt signaling pathway.⁵⁴ Let-7c targeted PBX3 which was essential for the maintenance of LCSCs and suppressed the transcriptional activity of CSCs-related genes including CACNA2D1, EpCAM, Sox2 and Notch3.⁵⁵

h. MicroRNA 1246

The upregulation of microRNA 1246 (miR-1246) was observed in several types of cancer, which was identified to play an oncogenic role.^56,103 In HCC, miR-1246 was demonstrated to promote migration and invasion of hepatoma cells.¹⁰³ Circulating miR-1246 has been shown to be a predictor of survival and tumor recurrence in HCC patients after liver transplantation.¹⁰⁴ MiR-1246 promoted stemness of LCSCs by activating the Wnt/β-catenin pathway through inhibiting the expression of Axin2 and glycogen synthase kinase 3β (GSK3β), which could induce the degradation of β-catenin.⁵⁶ Moreover, Oct4 acted as the direct upstream regulator of miR-1246 by direct promoter binding which drove β-catenin activation in LCSCs.

Regulation of lncRNAs in LCSCs

a. LncRNA highly upregulated in liver cancer and lncRNA MALAT1

Highly upregulated in liver cancer (HULC) is an lncRNA which was reported to act an oncogenic role in HCC progression.^57,105,106 MALAT1, also known as nuclear-enriched transcript 2 (NEAT2), is a highly conserved lncRNA involving in cell cycle control.^57,107 It was reported that the overexpression of HULC and MALAT1 promoted the proliferation of LCSCs, they cooperated to exert the oncogenic effect in LCSCs through regulating telomere repeat binding factor 2 (TRF2).⁵⁷

b. LncRNA downregulated in LCSCs

The absence of lncRNA downregulated in LCSCs (DILC) was identified in patients of HCC with poor prognosis, blockade of lncDILC significantly enhanced the expansion of LCSCs and then promoted the progression of HCC.⁵⁸ The mechanism was that lncDILC inhibited the autocrine interleukin (IL)-6/STAT3 signaling, and mediated the crosstalk between TNF-α/NF-κB signaling and IL-6/JAK2/STAT3 cascade, overexpression of lncDILC bound IL-6 promoter and blocked the IL-6 transcription, while lncDILC depletion enhanced IL-6 expression which induced by NF-κB in LCSCs.⁵⁸ So, restoration of lncDILC may be a novel method against LCSCs.

c. LncRNA cancer upregulated drug resistant and LncRNA H19

Cancer upregulated drug resistant (CUDR) is a novel lncRNA which was found overexpressed in doxorubicin-resistant sublines of human squamous carcinoma cells.¹⁰⁸ CUDR was also found overexpressed in many other tumors and promoted tumorigenesis.¹⁰⁹ In human HCC, CUDR cooperated with SET1A, component of histone methyltransferase complex, to promote tumor growth.¹¹⁰ In addition, CUDR could induce upregulation of HULC and β-catenin, and control human liver stem cells malignant differentiation, which was identified the possible origination of LCSCs.^109,59,111 CUDR and lncRNA H19 were found to have a combined action on LCSCs. H19 is encoded by a highly conserved imprinted gene and is essential for human tumor growth, its high expression was associated with poor clinical outcomes of patients with solid tumors.^112,113 In human HCC, H19 was reported to play a role in tumorigenesis of HCC.¹¹³ It stimulated angiogenesis and promoted the adhesion of CD90⁺ hepatoma cells to endothelial cell monolayer.¹¹⁴ It was demonstrated that Pten depletion promoted the binding of CUDR to the oncogene CyclinD1, the CUDR-CyclinD1 complex then enhanced the H19 expression by loading onto the promoter region of H19, which improved the cell telomerase activity and extended the telomere length in LCSCs, finally promoting LCSCs growth.⁵⁹

d. LncRNA HOX transcript antisense RNA

LncRNA HOX transcript antisense RNA (HOTAIR) acts as a transcriptional modulator in various fundamental biological activities, its overexpression is also related with poor prognosis of patients with HCC.¹¹⁵ Excessive HOTAIR promoted the malignant transformation of normal liver stem cells to LCSCs via inducing EMT.¹¹¹ HOTAIR also governed the oncogenic action of inflammatory related gene IKKα, IKKβ, and IKKγ in LCSCs.¹¹⁶ Most of all, it accelerated human LCSCs malignant proliferation through downregulating SETD2, which is an essential enzyme in DNA double-strand breaks.⁶⁰

e. Lnc RNA TCF7

LncTCF7 is a novel lncRNA which is highly expressed both in HCC and LCSCs, the protein-coding gene TCF7 acts as an upstream trigger to activate the Wnt signaling cascade.⁶¹ LncTCF7 recruited the SWI/SNF complex, a group of evolutionally conserved multi-subunits, to the promoter of TCF7 to initiate its expression, leading to activation of Wnt signaling and promoting self-renewal and tumorigenic capacity of LCSCs.⁶¹

f. LncRNA Sox4

LncSox4 is also a novel lncRNA which is recently discovered to be highly expressed in HCC and LCSCs.⁶² LncSox4 facilitated the self-renewal of LCSCs, mechanistically by targeting STAT3-Sox4 pathway, it recruited the TF STAT3 to the Sox4 promoter to trigger the expression of Sox4, which was required for the self-renewal of LCSCs.⁶²

g. LncRNA Brahma

LncBRM (gene symbol LINCR-0003) is recently identified to be highly expressed in LCSCs and is required for the oncogenicity and self-renewal of LCSCs.⁶³ The mechanism was that lncBRM associated with Brahma (BRM) to trigger the BRG1/BRM switch and the BRG1-associated factor (BAF) complex, leading to activation of the transcriptional cofactors YAP1 which was required for the self-renewal of LCSCs.⁶³

h. LncRNA β-Catm

Lncβ-Catm (gene symbol LINC00184) is a novel lncRNA which is recently found to be highly expressed both in LCSCs and HCC.⁶⁴ Lncβ-Catm contributed to LCSCs self-renewal by activating Wnt/β-catenin signaling. Mechanistically, it associated with β-catenin and the methyltransferase EZH2, promoting the methylation and stability of β-catenin, which resulted in the activation of Wnt/β-catenin signaling.⁶⁴

i. LncRNA CAMTA1

LncCAMTA1 (gene symbol RP11-312B8.1) is a novel lncRNA which is identified to be highly expressed both in LCSCs and HCC, promoting the stemness and tumorigenesis of LCSCs.⁶⁵ Mechanistically, lncCAMTA1 associated with the calmodulin binding transcription activator 1 (CAMTA1) promoter to inhibit the transcription of CAMTA1 which served as a tumor suppressor in HCC.⁶⁵

Decreasing LCSCs biomarkers

CD133

CD133 is an essential LCSCs biomarker which participates in the regulation of EMT, tumorigenicity and invasion of LCSCs. The downregulation of CD133 decreased the level of nuclear factor-κB (NF-κB), leading to inhibition of EMT and the stemness of LCSCs.⁶⁶ The expression of CD133 in HCC could be negatively regulated by Ikaros, a member of the Krüppel family, which interacted with the transcription repressor CtBP to form a complex directly bounding to the CD133 promoter.¹¹⁷

ICAM-1

Intercellular adhesion molecule 1 (ICAM-1) is a transmembrane molecule which is involved in many important cellular processes and also in the development of various human cancers, such as HCC, breast cancer and renal cancer.¹¹⁸ ICAM-1 is considered as a stem cell marker which is upregulated by Nanog, it is also high expressed in LCSCs, promoting the tumorigenicity and stemness of LCSCs.⁶⁷

Interruption of the essential signaling pathways in LCSCs

Several signaling pathways are essential in the development of LCSCs and HCC, which provide some promising targets against LCSCs for the treatment of HCC.

Wnt/β-catenin pathway

The Wnt/β-catenin pathway is responsible for the self-renewal and unlimited cell proliferation of CSCs. It has also been widely confirmed in maintenance of LCSCs. Chen and colleagues⁶⁸ revealed that constitutive expression of Wnt/β-catenin was detected in LCSCs, knockdown of it suppressed the cell phenotype of LCSCs. The Wnt/β-catenin inhibitor could impair the viability of LCSCs.^69,70,119 Moreover, some phytochemicals, such as casticin, triptolide and BrMC, also have been demonstrated to restrain the self-renewal and proliferation of LCSCs by suppressing Wnt/β-catenin signaling.^120,121

PI3K/Akt/mTOR pathway

The PI3K/Akt signal pathway plays a key role in tumorigenesis.¹²² In HCC, The PI3K/Akt/mTOR pathway promoted formation of LCSCs, leading to hepatocarcinogenesis.^123,124 The inhibitor of PI3K or mTOR could suppress the proliferation of LCSCs. Moreover, HBV contributes to the activation of the pathways, HBV X protein facilitates alpha fetoprotein (AFP) expression, which promotes the proliferation of LCSCs, by activating PI3K/Akt signal pathways.⁷¹

Akt/GSK-3β/β-catenin pathway

In HCC, the Akt/GSK-3β/β-catenin pathway promotes the proliferation and invasion of LCSCs.^72,73 Inhibition of Akt which is a protein kinase in multiple cellular processes, reduced the self-renewal and propagation of LCSCs.^73,122 However, increasing evidence shows that adverse effects including liver injury and inflammation, hyperglycemia and hyperinsulinemia accompany the complete deletion of Akt.¹²² Thus, the definite mechanisms of systemic inhibition of Akt or the pathways need further study.

STAT3 signaling pathway

Signal transducer and activator of transcription 3 (STAT3) plays an oncogenic role in cancer progression involving in the regulation of EMT and CSCs, it could be activated by Toll-like receptors, microRNA or cytokine receptors such as the IL-6 family of cytokines.¹²⁵ It was identified that IL-6 activated STAT3 and elevated the cellular glucose uptake, finally promoting the stemness of LCSCs.^74,75 On the other hand, TLR4 could also trigger the expression of Nanog, the latter directly interacted with STAT3 which could also phosphorylated by the leptin receptor (OB-R) pathways to promote the generation and invasion of LCSCs by upregulating Twist1.⁷⁶ Moreover, STAT3 meditated Sox4 expression to favor the activity of LnxSox4 as mentioned above.⁶²

RAS/RAF/MEK/ERK signaling pathway

Aberrant regulation of the RAS/RAF/MEK/ERK pathway has been detected in several malignancies including HCC.^126,127 Experimental research demonstrated that blockade of the RAS/RAF/ERK pathway suppressed the proliferation, invasion and migration of LCSCs by depleting MEK or reducing ERK1/2 phosphorylation.^71,77,78 Moreover, MEK signaling was identified to promote the self-renewal of LCSCs by maintaining the stabilization of SIRT1 protein which was involved in carcinogenesis.¹⁴

JNK signaling pathway

The JNK signaling pathway is involved in cancer development by regulating the tumor-initiating capacity of CSCs.¹²⁸ In HCC, increased JNK activity is associated with tumor proliferation.¹²⁶ The CSCs-related gene ANXA3 could enhance the activity of the JNK pathway in CD133⁺LCSCs by upregulating the expression of the proto-oncogene c-MYC, finally improving the self-renewal of LCSCs.¹¹

Notch signaling pathway

The Notch receptors consist of four members (Notch1-4) in mammals, which can combine with various ligands and be activated.¹⁷ The Notch signaling cascade participates in the progress of various cancers associating with Wnt, MAPK and NF-κB signaling.⁷⁹ In HCC, the activated Notch pathway was detected in LCSCs promoting the expression of CSC-related genes and maintaining the stemness of LCSCs.⁷⁹ Notch1 in LCSCs was identified to be a downstream of Wnt/β-catenin.⁸⁰ Notch2 which was activated in LCSCs could be suppressed by C8orf4.¹⁷

Targeting LCSCs at the level of cells

Targeting LCSCs at the cellular level includes inhibiting autophagy and application of oncolytic adenoviruses in LCSCs.

Blocking the process of autophagy in LCSCs

Autophagy is a conserved lysosomal degradation process during which the cytoplasmic components including macromolecules and organelles are degraded by the lysosome, accordingly maintaining the cellular homeostasis including anti-stress, immunity and antiaging.^129,130 The effect of autophagy varies in different phases of tumorigenesis, it suppresses malignant transformation in healthy cells, but reduces intracellular and extracellular stress of cancer cells to promote tumor development.¹³⁰ In healthy human liver, basal rates of autophagy maintain hepatocyte homeostasis by degrading misfolded proteins, protein aggregates and damaged mitochondria.¹³¹ Autophagy increased during the development of cirrhosis.¹³² Increased autophagy triggered the expression of hepatocyte growth factor (HGF) which bound with its receptor Met, leading to activation of JNK and STAT3 signaling to induce the formation of the Axin2⁺CD90⁺ LCSCs in liver cirrhosis.¹³³ In another way, mitophagy which selectively removes the mitochondria by autophagy, positively regulated LCSCs by suppressing the tumor suppressor p53 which otherwise could be activated by PINK1 on mitochondria to downregulate the expression of Nanog (Figure 1).¹⁸ Moreover, autophagy contributed to the initiation and invasion of HCC. In the tumor microenvironment, autophagy copes with hypoxia and nutritional deprivation, favoring the survival of LCSCs and the chemoresistance of hepatoma cells.¹³⁴ Chloroquine, the inhibitor of autophagy, showed inhibitory effect on the stemness of LCSCs.^133,134 It inhibited autophagic flux by decreasing autophagosome-lysosome fusion which formed autolysosomes.¹³⁵ Chloroquine may inhibit the formation of LCSCs by suppressing pSTAT3 and Nanog, and keep the p53 activity from being suppressed by autophagy. Thus, the application of chloroquine or other autophagy inhibitors in liver cirrhosis and HCC may be potential targeting approaches against LCSCs.

Figure 1.

Targeting LCSCs by suppression of autophagy and Nanog.

Autophagy positively regulated LCSCs by suppressing p53 and favoring the activity of Nanog. Mitophagy suppressed p53 which otherwise was activated by PINK1 on mitochondria to downregulate Nanog. Autophagy mediated HGF binding with its receptor Met to activate STAT3 which was also phosphorylated by IL-6, OB-R and TLR4. The pSTAT3 and Nanog directly interacted to trigger Twist1-CD24 axis. Nanog, the downstream of TLR4 and PreS1, suppressed p53 by phosphorylation of Numb and destabilization of Numbp53 complex via the Nanog-AURKA-aPKCζ pathway. Moreover, Nanog upregulated the level of ICAM-1, ABCB1 and ABCG2 which favored the self-renewal and chemoresistance of LCSCs. The autophagy inhibitor chloroquine could block the self-renewal of LCSCs. It inhibited autophagy by decreasing autophagosome–lysosome fusion which formed autolysosomes to degrade the cytoplasmic components. Chloroquine may block pSTAT3 and Nanog and keep the p53 activity from being suppressed by autophagy in LCSCs. On the other hand, the Nanog-targeting shRNA via lentiviral particles could efficiently block the expression of Nanog at both mRNA and protein levels by cleavage of Nanog mRNA. Nanog silence by shRNA which showed a long-term depletion of the targeted gene impaired both the self-renewal and the chemoresistance of LCSCs. Chloroquine and Nanog shRNA may provide potential LCSC-targeting approaches by the suppression of autophagy and Nanog respectively.

ABCG2, ATP-binding cassette G 2; ABCB1, ATP-binding cassette subfamily B member 1; AURKA, Aurora A kinase; aPKCζ, atypical protein kinase C zeta; HGF, hepatocyte growth factor; ICAM-1, intercellular adhesion molecule 1; LCSC, liver cancer stem cell; OB-R, the leptin receptor; PINK1, Pten-induced putative kinase 1; STAT3, signal transducer and activator of transcription 3; shRNA, short hairpin RNA; TLR4, Toll-like receptor 4.

Application of oncolytic adenovirus in LCSCs

In recent years, the antitumor use of viruses has been studied in depth. Oncolytic adenoviruses which are genetically modified, could selectively enter and spread inside tumors, show their cytotoxicity and suppression of tumors.¹³⁶ In HCC, it was reported that a new oncolytic adenovirus GD55 demonstrated a stronger killing effect on human LCSCs.¹³⁷ Moreover, Zhang and colleagues designed a novel oncolytic adenovirus which carried the tumor suppressor gene TSLC1 and targeted Wnt signaling pathway, the adenovirus inhibited growth and metastasis of LCSCs in vivo by mediating TSLC1 and suppressing Wnt signaling.¹³⁸ Therefore, oncolytic adenovirus may serve as a potentially therapeutic application by targeting LCSCs.

Eliminating the chemoresistance of LCSCs

Targeting LCSCs by depleting or inactivating the drug resistance genes or proteins and blocking the drug resistance-associated signaling pathways in LCSCs may be effective to eliminate its chemoresistance (Table 3).

Table 3.

Targets to eliminate the chemoresistance of LCSCs.

Targets	Mechanism	Species	References
Oncogenes or oncoproteins
ABCG 2	ABCG 2 is involved in drug efflux pumps, which is mediated by Oct4 and Nanog.	Human	Jia and colleagues139, Zhou and colleagues140
ZIC-2, PML and Oct4	ZIC-2 and PML act as upstream of Oct4 which has a positive association between ABCG2.	Human	Zhu and colleagues141; Tang al.142; Jia and colleagues139
Nanog	Nanog decreased the expression of ABCB1 and ABCG2 in LCSCs.	Human	Zhou and colleagues140
Laminin-332	Laminin-332 acted as part of the human LCSC niches, upregulated K19 expression, and downregulated phospho-histone H3 expression and induced phosphorylation of mTOR.	Human	Govaere and colleagues143
CHD4	CHD4 contributed to the repair of DNA damage in a PARP-dependent manner in EpCAM+LCSCs as a chromatin remodeling enzyme.	Human	Nio and colleagues144
MSI2	MSI2 upregulated Lin28A, which is critical CSC-related RNA-binding proteins.	Human	Fang and colleagues145
GRAMD1A	GRAMD1A regulated the transcriptional activity of STAT5.	Human	Fu and colleagues146
Dysadherin	Dysadherin might upregulated drug efflux pumps in LCSCs.	Human	Jiang and colleagues147
Signaling pathway
JNK signaling pathway	The number of LCSCs and phosphorylation of SAPK/JNK increased upon anticancer treatment.	Human	Kim and colleagues148
Akt signaling pathway	The inhibition of Akt signaling enhanced the sensitivity of LCSCs to sorafenib by upregulating ERK signaling, which was a primary target of sorafenib.	Human	Xu and colleagues72

ABCB1, ATP-binding cassette subfamily B member 1; ABCG2, ATP-binding cassette G 2; CHDs, chromodomain-helicase-DNA-binding proteins; GRAMD1A, GRAM domain-containing protein 1A; K 19, keratin 19; LCSC, liver cancer stem cell; MSI2, Musashi 2; Oct4, octamer-binding transcription factor 4; PARP, poly (ADP-ribose) polymerase; PML, promyelocytic leukemia; SAPK, phospho-stress-activated protein kinase; STAT5, signal transducer and activator of transcription 5.

Depletion or inactivating the drug resistance genes or proteins in LCSCs

ABCG2

The ATP-binding cassette (ABC)G2 is a member of the superfamily of ABC transporters drug efflux pumps, which lead to multidrug resistance in liver cancer.¹³⁹ ABCG2 is mediated by Oct4 for the chemotherapeutic resistance in hepatoma cells via a potential Oct4-Akt-ABCG2 pathway.¹⁴⁹ Both Oct4 and Nanog, another upstream of ABCG2, have a positive association with ABCG2, cooperating to favor the chemoresistance of LCSCs.^139,140 Thus, ABCG2 may become one of the most important targets for drug resistance of LCSCs.

ZIC-2–Octamer-binding transcription factor 4 (Oct4) axis and promyelocytic leukemia–Oct4 axis

Octamer-binding transcription factor 4 (Oct4) which is encoded by the Pou5f1 gene belongs to the POU (Pit, Oct and Unc) family of DNA-binding proteins.¹⁵⁰ As mentioned above, the overexpression of Oct4 favored the stemness of LCSCs by upregulating miR-1246 to drive β-catenin activation.⁵⁶ Its expression in LCSCs can be upregulated by PreS1, miR-21, miR-155 and miR-429/RBBP4/E2F1/Oct4 pathway.^12,49,50,52 It was also found that the zinc finger TF ZIC2 which was highly expressed in LCSCs acted upstream of Oct4, ZIC2 recruited the nuclear remodeling factor (NURF) complex binding to the promoter of Oct4, thereby initiating Oct4 activation.¹⁴¹Another protein upstream of Oct4 is the promyelocytic leukemia (PML) protein which was originally identified in acute promyelocytic leukemia.^151,142 PML took a role as tumor suppressor in multiple pathways, but it was found to preserve the activity of Oct4 gene in stem cells.^151,152 In LCSCs, suppression of PML decreased the level of Oct4, indicating that the PML favored the stemness of LCSCs via the PML/Oct4 axis.¹⁴² However, Oct4 overexpression is also involved in the acquisition of the drug-resistant phenotype of cancer cells.¹⁵³ It has a positive association between ABCG2, both of them contributed to drug resistance of CD90⁺CD133⁺LCSCs.¹³⁹ Oct4 is one of the most promising therapeutic targets for HCC treatment by targeting LCSCs.

Nanog

Nanog is a homeodomain-containing TF which plays a key role in the regulation of stemness acquirement and cancer development.¹⁵⁴ As mentioned above, Nanog is critical to the stemness of LCSCs, and can be activated by PreS1 and TLR4.^12,76 Its abnormal expression upregulated the level of ICAM-1.⁶⁷ The activated Nanog directly interacted with pSTAT3 to drive the expression of Twist1 which activated CD24.^42,76 Autophagy is associated with Nanog. It favors the activity of Nanog not only by activating the HGF/Met/STAT3 pathway, but also by suppressing p53, which otherwise downregulates the expression of Nanog.^18,133 On the other hand, Nanog suppresses p53 by phosphorylation of Numb and destabilization of the Numb-p53 complex via the Nanog-AURKA-aPKCζ pathway.¹⁹ Moreover, Nanog was found to be involved in chemoresistance of LCSCs; knockdown of it led to increased chemosensitivity to antitumor agents by decreasing the expression of ABCB1 (ABC subfamily B member 1) and ABCG2 in LCSCs which belong to the ABC drug efflux pumps (Figure 1).¹⁴⁰ It was found that Nanog silence by short hairpin RNA (shRNA) or small interfering RNA (siRNA) impaired both the self-renewal and the chemoresistance of LCSCs.^19,67,140 The Nanog shRNA via lentiviral particles could efficiently block the expression of Nanog at both mRNA and protein levels by the cleavage of Nanog mRNA. It showed a long-term depletion of the targeted gene. Thus, the suppression of Nanog via shRNA lentiviral particles or other gene therapy vectors may be potential LCSCs-targeting approaches.

Laminin-332

Laminin-332 belongs to the laminin family which is a kind of extracellular matrix proteins.¹⁴³ Laminin acts as part of hepatic progenitor cell (HPC) niches to maintain the phenotype of HPCs during human liver damage.¹⁵⁵ Govaere and colleagues¹⁴³ demonstrated that laminin-332 also acted as part of the human LCSCs niches, it supported stemness and chemoresistance of LCSCs under doxorubicin and sorafenib treatment by upregulating keratin(K)19 expression, downregulating phospho-histone H3 expression and inducing phosphorylation of mTOR. However, the effect of laminin-332 was decreased upon the inhibition of mTORC1 and mTORC2, and enhanced when mTORC1 was inhibited alone,¹⁴³ which need further studies.

CHD4

The chromodomain-helicase-DNA-binding proteins (CHDs) is part of the histone deacetylation (NuRD) complex protein, which was essential in transcriptional events of oncogenesis.¹⁵⁶ CHD4 was found to contribute to the repair of DNA damage in a poly (ADP-ribose) polymerase (PARP)-dependent manner in EpCAM⁺LCSCs as a chromatin remodeling enzyme.¹⁴⁴ The overexpression of CHD4 enhanced the chemoresistance of LCSCs to anticancer drugs, while inhibition of CHD4 by doubly suppressing PARP and histone deacetylase (HDAC) suppressed LCSC proliferation.¹⁴⁴ Thus, knockout of CHD4 may enhance the chemosensitivity of LCSCs which deserves further investigation.

Musashi 2

The Musashi family is a kind of evolutionarily conserved RNA-binding protein which comprises Musashi 1 and Musashi 2 (MSI2).⁴⁰ MSI2 is required in the tumorigenesis of several human cancers, and participates in inhibiting the tumor suppressor Numb and wildtype p53.^19,40 Fang and colleagues¹⁴⁵ found that MSI2 also contributed to the chemoresistance of LCSCs by upregulation of Lin28A, which is critical CSC-related RNA-binding proteins. Knockdown of MSI2 or Lin28A impaired the chemoresistance of LCSCs, moreover, MSI2 knockdown also reduce the level of Nanog, Oct4 and Sox2.¹⁴⁵ Thus, MSI2 may become new target to improve the chemosensitivity of LCSCs.

GRAM domain-containing protein 1A

GRAM domain-containing protein 1A (GRAMD1A) is a novel protein, the function of which has not been explored.¹⁴⁶ It is highly expressed in HCC and sustains the self-renewal and chemotherapy resistance of LCSCs by regulating the transcriptional activity of STAT5 (signal transducer and activator of transcription 5). However, other evidence showed that STAT5 could prevent the formation of aggressive HCC.¹⁵⁷ So, both GRAMD1A and STAT5 need more research.

Dysadherin

Dysadherin is a cell membrane glycoprotein which upregulates the production of chemokine and downregulates cell adhesion mediated by E-cadherin, accordingly, creating survival conditions for many types of human cancer.¹⁵⁸ In HCC, high expression of dysadherin sustained high resistance to chemotherapeutic drugs in LCSCs which might depend on upregulation of drug efflux pumps.¹⁴⁷ Knockdown of dysadherin showed increased sensitivity to chemotherapy agents and apoptotic cell death, as well as remarkable decreasing expression of stemness-related proteins.¹⁴⁷

Blockade of the drug resistance-associated signaling pathways in LCSCs

JNK signaling pathway

As mentioned above, the JNK signaling pathway is associated with the self-renewal and tumorigenicity of LCSCs.¹¹ JNK signaling was also involved in mutidrug resistance of HCC. It was found that the number of side population (SP) of cells and phosphorylation of phospho-stress-activated protein kinase (SAPK)/JNK increased upon anticancer treatment, the increase of SP cells which was considered as LCSCs could be blocked by JNK signaling inhibition, indicating that the activation of JNK may be responsible for drug resistance in HCC.¹⁴⁸

Akt signaling pathway

As mentioned above, the Akt signaling pathway is involved in the proliferation of LCSCs. Also, Akt signaling contributed to the chemotherapeutic resistance of LCSCs, the inhibition of Akt signaling enhanced the sensitivity of LCSCs to sorafenib by upregulating ERK signaling, which was a primary target of sorafenib.⁷²

Conclusion

In recent years, the important role of LCSCs in the initiation, relapse, metastasis and drug resistance of HCC has been identified. Targeting LCSCs is expected to be a promising approach for the treatment of HCC. In this review, we summarize potentially therapeutic targets which are key points of LCSCs at the genetic, molecular and cellular level. Moreover, we analyze the promising approaches to eliminate chemoresistance of LCSCs. These targets provide new opportunities for HCC treatment. However, it has been recognized that these key elements in the differentiation and self-renewal of LCSCs interacted with each other through a complex crosstalk. Thus, combined therapies by targeting LCSCs may enhance the therapeutic efficacy of HCC treatment. Moreover, although many of the targets mentioned above are highly expressed in LCSCs compared with normal liver cells or normal liver stem cells, but the systemic inhibition of these targets may result in injury of normal cells, how to save normal cells from damage in the clinical application is certainly a concern, which needs further research. In summary, targeting LCSCs could be a promising therapeutic strategy for the treatment of HCC.