Normal and disease-related biological functions of Twist1 and underlying molecular mechanisms

Abstract

This article reviews the molecular structure, expression pattern, physiological function, pathological roles and molecular mechanisms of Twist1 in development, genetic disease and cancer. Twist1 is a basic helix-loop-helix domain-containing transcription factor. It forms homo- or hetero-dimers in order to bind the Nde1 E-box element and activate or repress its target genes. During development, Twist1 is essential for mesoderm specification and differentiation. Heterozygous loss-of-function mutations of the human Twist1 gene cause several diseases including the Saethre-Chotzen syndrome. The Twist1-null mouse embryos die with unclosed cranial neural tubes and defective head mesenchyme, somites and limb buds. Twist1 is expressed in breast, liver, prostate, gastric and other types of cancers, and its expression is usually associated with invasive and metastatic cancer phenotypes. In cancer cells, Twist1 is upregulated by multiple factors including SRC-1, STAT3, MSX2, HIF-1α, integrin-linked kinase and NF-κB. Twist1 significantly enhances epithelial-mesenchymal transition (EMT) and cancer cell migration and invasion, hence promoting cancer metastasis. Twist1 promotes EMT in part by directly repressing E-cadherin expression by recruiting the nucleosome remodeling and deacetylase complex for gene repression and by upregulating Bmi1, AKT2, YB-1, etc. Emerging evidence also suggests that Twist1 plays a role in expansion and chemotherapeutic resistance of cancer stem cells. Further understanding of the mechanisms by which Twist1 promotes metastasis and identification of Twist1 functional modulators may hold promise for developing new strategies to inhibit EMT and cancer metastasis.

Introduction

Organogenesis, which involves morphogenesis and cytodifferentiation, requires intricate and delicate interplays among transcription factors, growth factors, cell surface molecules and extracellular matrix proteins. Post-fertilization, the multicellular blastula forms and undergoes gastrulation, resulting in mesoderm formation between the endoderm and the ectoderm. Twist1 was originally identified in Drosophila as one of the zygotic genes essential for mesoderm specification and subdivision into different tissue types and dorsal-ventral patterning during early embryo development^1,2,3. The name “Twist” was given to this gene on the basis of observations that Drosophila embryos lacking the Twist1 gene failed to gastrulate normally, produced no mesoderm and died at the end of embryogenesis with a 'twisted' appearance^1,2. The Twist1 gene encodes a transcription factor containing a basic helix-loop-helix (bHLH) domain⁴ and an amino-acid motif present in a protein family involved in the regulation of organogenesis^5,6,7. The critical roles of Twist1 in mesodermal development have been well illustrated by genetic studies. Human studies have shown that Twist1 gene mutations cause Saethre-Chotzen syndrome (SCS), an autosomal dominant inheritance disease characterized by a broad spectrum of malformations including short stature, craniosynostoses, high forehead, ptosis, small ears with prominent crus, and maxillary hypoplasia with a narrow and high palate^{8,9,10,11,12,13}. Gene-ablation experiments demonstrated that the heterozygous Twist1 null mice^14,15,16,17 manifest craniofacial and limb abnormalities resembling those in SCS patients. Recently, a number of studies have indicated that in addition to its essential roles in the development of multiple organs and systems, Twist1 also plays important roles in cancer metastasis^{18,19,20,21,22,23,24}.

This review will focus specifically on the biological roles of Twist1, with the overall objective of summarizing the remarkable progress toward our understanding of its structure, tissue/cell expression and biological functions. The review will also place particular emphasis on Twist1's roles in tumor initiation and progression, since the data concerning its involvement in cancer are newer and have attracted considerable attention in recent years. It is worth noting that Twist2 or Dermo1 shares many structural and functional similarities with Twist1, so we will also consider Twist2 when it is relevant to the discussion of Twist1 structure and function.

Molecular structures of the Twist1 gene and protein

Twist1 is a transcription factor that belongs to the bHLH family³. Structurally, bHLH proteins are characterized by the presence of a conserved domain containing a stretch of basic amino acids adjacent to two amphipathic α-helices separated by an inter-helical loop^5,6. The α-helices mediate the interaction of this protein with a second bHLH factor, leading to the formation of a dimer that binds to CATATG hexanucleotide sequences known as the Nde1 E-box. The E-boxes are present in the regulatory elements of many genes that are essential for various types of organogenesis⁶. The traditional classification categorizes the bHLH family into three subfamilies: class A, class B and class C⁵. The proteins in class A, which include E12, E47, HEB, E2-2 and Daughterless²⁵, are ubiquitously expressed in mammalian cells. Class B comprises bHLH proteins that have relative specificity in tissue expression and form dimers with class A molecules for binding to E-boxes. Class C molecules, consisting of the Myc proteins, do not form heterodimers with either class A or class B proteins. The Twist family, which has relative tissue specificity and forms heterodimers with E12 and E47, falls into class B²⁶.

Human Twist1 gene is mapped to 7q21.2 and contains two exons and one intron^10,27. The first exon contains an ATG site followed by an open reading frame encoding 202 amino-acid residues. The open reading frame is followed by a 45-bp untranslated portion in exon 1, a 536-bp intron and a second untranslated exon with two potential polyadenylation sites that are 65 and 415 bp from the 5′ end of exon 2. The molecular mass calculated from the amino-acid sequence of human Twist1 is approximately 21 kDa, with a theoretical isoelectric point of ∼9.6. The protein contains relatively more polar amino-acid residues in the region close to the NH₂-terminus and more nonpolar residues at the COOH-terminus where the bHLH domain is located. Thus, the NH₂-teminal portion of Twist1 appears to be more hydrophilic than the COOH-terminus. Human Twist1 protein shares 96% amino-acid sequence identity with mouse Twist1²⁸. It is worth noting that Drosophila Twist1 with 490 amino-acid residues is remarkably larger than human, mouse and Xenopus Twist1, which have 202, 206 and 166 residues, respectively.

The bHLH domains of Twist1 show a very high degree of conservation among a broad range of species, including human, mouse, frog, Drosophila, leech and Caenorhabditis elegans (Figure 1). Indeed, the bHLH domain of human Twist1 from residue¹⁰⁹ to residue¹⁶³ shares 100% homology with that of the mouse Twist1 (Figures 1 and 2). The functional importance of the bHLH domain has been well illustrated by the fact that point mutations in the bHLH domain result in SCS^9,10. The basic region, ¹⁰⁹Q–T¹²¹, of human Twist1 protein is the main domain responsible for binding DNA. The R118C mutation in this region diminishes Twist1 DNA-binding capability, and results in SCS (Figures 1 and 2)²⁹. Sequence homology between Twist1 and MyoD basic regions elicits the proposal that the R118C mutation can disrupt Twist1 interaction with the phosphodiester backbone of E-box DNA²⁹. In addition, the Loop-Helix II junction region of Twist1 is also involved in DNA binding. In agreement with this notion, human Twist1 with S144R or K145E mutation is unable to bind the CATATG Nde1 E-box²⁹. Mutation of S144 or K145 in Twist1 may prevent hydrogen bond formation between the phosphate group and the lateral amino-acid chain of Twist1 protein²⁹. In addition to using its bHLH domain to form heterodimers with E12 or E47 for DNA binding, Twist1 also interacts with MyoD, a bHLH transcription factor that regulates muscle differentiation (Figure 2)³⁰. The interaction of Twist1 with MyoD results in inhibition of MyoD and MEF2 functions, leading to the inhibition of muscle differentiation^30,31.

Figure 1

Comparison of amino-acid sequences among human (H), mouse (M), Xenopus (X), Drosophila (D), leech (L) and C. elegans (C) Twist1. The Twist1 amino-acid sequences from human (NP_000465.1), mouse (NP_035788.1), Xenopus (NP_001079352.1), Drosophila (NP_523816.2), leech (AAL05567.1) and C. elegans (AAC26105.1) have been aligned with the aid of the computer program Clustal W2 (Larkin et al., 2007). “^*” indicates the identical residues in all sequences in the alignment; “#” indicates conserved substitutions; “$” indicates semi-conserved substitutions; “-” indicates gaps in the alignment.

Figure 2

Molecular structure of the human Twist1 protein. The number of amino-acid residues for each structural domain is indicated. The regions that interact with other proteins are also indicated by solid lines. NLS1 and NLS2, nuclear localization signal sequences 1 and 2; bHLH, basic helix-loop-helix domain; WR, the tryptophan and arginine motif; CBP, cAMP-response element binding protein (CREB)-binding protein; PCAF, p300/CBP-associated factor, Runx2, Runt-related transcription factor; E12/E47, a bHLH transcription factor that forms dimer with Twist1; MyoD, a bHLH transcription factor that regulates muscle differentiation.

The WR motif, also known as the Twist box³², which is located between 20 and 55 amino acids COOH-terminal to the bHLH region, is highly conserved among vertebrates; the amino-acid sequences in this domain show 100% homology among human, mouse and Xenopus Twist1 (see the region from residue¹⁸⁰ to residue²⁰² in the human Twist1) (Figure 1). However, the WR domain is less conserved in the more ancient species. In fact, C. elegans Twist1 does not contain a WR dipeptide (Figure 1). While the function of this WR region is unclear, it is postulated to be required for Twist1 protein folding and activity¹². It has been demonstrated that the WR domain is required for the transactivation function of Twist1, and genetic mutations in the WR domain are associated with SCS of human patients^33,34,35.

Twist1 functions as a transcription factor in the cell nucleus. There are two nuclear localization signal (NLS) sequences, ³⁷RKRR⁴⁰ and ⁷³KRGKK⁷⁷, in the human Twist1 protein (Figures 1 and 2). The Twist1 protein with K76R mutation is capable of translocating into the nucleus by itself, while the Twist1 proteins with K38R, K73R or K77R mutation cannot translocate into the nucleus by themselves. However, the K38R, K73R or K77R Twist1 mutant is fully capable of forming heterodimers with E12 or TCF-4; these heterodimers can translocate into the nucleus despite the NLS mutations³⁶. These findings indicate that NLSs are functional and also that Twist1 can translocate into the nucleus with its heterodimer partners without using its own NLSs.

Furthermore, the N-terminus of Twist1 can interact with p300, cAMP-response element binding protein (CREB), CREB-binding protein (CBP) and p300/CBP-associated factor (PCAF) (Figure 2), resulting in inhibition of the acetyltransferase activities of these histone-remodeling enzymes³⁷. Since histone acetylation is usually coupled with transcriptional activation, the inhibition of p300, CBP and PCAF activities by Twist1 should repress gene expressions mediated by transcription factors that recruit these histone acetyltransferases. Finally, the C-terminus of Twist1 interacts with the DNA-binding domain of Runx2 to repress Runx2 function (Figure 2)¹⁷. Runx2 is a necessary transcription factor for osteoblast differentiation. During osteoblast development, relief of Runx2 inhibition by Twist1 is a mandatory event¹⁷. These findings suggest that Twist1 not only serves as a transcription factor to directly regulate its target genes, but also regulates other transcription factor/coregulator-mediated gene expression through interaction with other transcriptional regulators.

The human Twist2 protein contains 160 amino-acid residues and it shares 68% homology with human Twist1. The amino-acid sequences of their bHLH and Twist box domains are almost identical, which provides a structural base for their partially redundant biological functions.

The expression of Twist1 in normal tissues

Consistent with its major roles in organogenesis, Twist1 is primarily expressed in mesoderm-derived tissues. In Drosophila, the maternal protein Dorsal (dl) activates Twist1 transcription in the presumptive mesoderm^4,38,39. Dorsal, a Rel-containing sequence-specific transcription factor and the final maternal morphogen, forms a nuclear gradient that aids Twist1 expression in the presumptive ventral mesoderm. The expressed Twist1 protein forms a steep gradient across the presumptive mesoderm-neuroectoderm border in the early embryo. Specifically, during Drosophila embryogenesis, Twist1 is first detected at the cellular blastoderm stage, in the nuclei of ventral midline cells which are destined for mesoderm and mesectoderm formation⁴. During the early gastrulation period, Twist1 expression level is highly elevated in the mesodermal layer of the embryos, and is quantitatively similar among all of the cells that participate in ventral furrow invagination⁴. During the germ band elongation stage (the late gastrulation stage), Twist1 can be seen in cells of the whole germ band mesodermal layer and the anterior midgut primordium. During germ band retraction and later stages of Drosophila embryogenesis, Twist1 expression signals, though relatively weak, are seen within the mesodermal layer of the somatopleura and splanchnopleura. After birth, Twist1 can be seen in adult mesenchymal cells such as muscle stem cells, referred to as adult muscle precursors in Drosophila^40,41,42,43. In summary, during Drosophila embryogenesis, Twist1 accumulates at early stages (from the beginning of cellular blastoderm to germ band extension) in all cells of the presumptive mesodermal layer, but its expression decreases to relatively low levels in the mesodermal layer cells during later stages. This expression pattern is consistent with the critical roles of this zygotic gene in early embryogenesis. It should be noted that as a nuclear protein, Twist1 is localized in the nuclei at all developmental stages.

During mouse embryogenesis, Twist1 transcripts are first seen at embryonic day 7.5 (E7.5) in the anterior-lateral mesoderm underlying the head folds⁴⁴, in the primitive streak epiblast, and in scattered cells in the amniotic cavity⁴⁵. Then, Twist1 is predominantly and sequentially expressed in the somites, the neural crest-derived head mesenchyme, the first aortic arches, the lateral mesoderm, the second, third and fourth branchial arches, the anterior limb buds, and finally, the posterior limb buds^28,44,45,46. Twist1 expression in mouse dental mesenchyme and heart valves remains at least until E18, and the same is true for its expression in the mesenchyme underneath the epidermis and tongue epithelium⁴⁴. Twist1 expression in the developing limb^44,47,48 is uniform in the limb bud mesenchyme at E9.5, but becomes regionalized to a broad anterior (preaxial) and a narrow posterior (postaxial) mesenchymal domain, spanning most of the proximal-distal length of the bud at E10.5. By E11.5, Twist1 expression becomes confined to the marginal and proximal parts of the limb bud and is down-regulated in the core. Twist1 expression in the mesenchyme is generally stronger in the preaxial region than in the postaxial region of the E10.5 and E11.5 limb buds. At E12.5-E15.5, Twist1 expression localizes to the interdigital tissues and then the perichondrium of the phalanges⁴⁸. A general tendency is that Twist1 expression in the mouse embryo occurs first along a dorso-ventral gradient pattern until the headfold stage, and then moves along the rostro-caudal axis of the embryos as mesoderm cell layer- and neural crest cell-derived tissues develop.

After birth, Twist1 is expressed in the adult stem cells of the mesenchyme^43,49. Twist1 mRNA has also been detected in primary osteoblastic cells derived from newborn mouse calvariae⁵⁰ and in mouse brown and white adipocytes^51,52. This mouse Twist1 expression pattern is consistent with the role of Twist1 as a mesoderm-determining factor that regulates genes involved in the specification and differentiation of mesoderm and in the development of mesenchyme tissues throughout the mouse body.

While the expression patterns of Twist1 during embryonic development of Drosophila and mouse have been well studied, information regarding the expression profile of Twist1 during human embryogenesis is lacking. Wang et al. used various adult human tissues, as well as several cell lines originated from different normal human tissues to examine Twist1 expression pattern²⁷. Among the human tissues tested, the strongest signals for Twist1 mRNA were observed in the placenta, a tissue containing a large fetal portion and a small maternal portion. In particular, strong Twist1 signals were seen in the fetal portion of the placenta developed from the chorionic sac, which is mostly derived from the mesoderm. Intermediate strength signals were seen in the adult heart and skeletal muscle, which are also mostly derived from the mesoderm. Weak signals were found in the kidney and pancreas, while no signals were observed in the ectoderm-derived cells in the brain and endoderm-derived cells in the lung and liver. These findings indicate that in adult humans, Twist1 is preferentially expressed in mesodermally derived tissues. Among cell lines derived from different human tissues, Twist1 expression was found in WI-38 cells (fetal lung-derived fibroblasts), human peritoneal mesothelial cells and endometrial fibroblasts (both of which represent mesodermal cells derived from young adults) and human bone marrow-derived mesenchymal stem cells^27,53. Additionally, Twist1 expression is also observed in human white adipocytes⁵². Twist1 expression was not observed in human epithelial cells, although its mRNA was detected in both fetal and adult human skin fibroblasts²⁷.

In summary, as one of the zygotic genes essential for organogenesis, Twist1 is highly expressed in the mesoderm-derived embryonic mesenchyme. This is consistent with its roles in the development and specification of the tissues with a mesodermal origin. In postnatal tissues, Twist1 is primarily expressed in relatively quiescent adult stem cells located in mesoderm-derived mesenchymal tissues such as the muscle, adipose tissue and bone marrow.

The function of Twist1 in development and signaling pathways

Twist1 is a bHLH transcription factor essential for mesoderm development^1,2,4. Although especially important to the morphology and behavior of head mesenchyme cells that support morphogenesis of the cranial neural tube¹⁴, Twist1 is also an important regulator of many other biological processes. This is supported by the widespread expression of Twist1 and the various phenotypes associated with human Twist1 gene mutations and mouse Twist1 gene knockout¹⁷. Twist1 is believed to act upon a set of downstream target genes, through which it controls a variety of cellular events necessary for the proper formation of mesenchyme derivatives^14,54.

While previous research has shown the cruciality of Twist1 in mesoderm-associated organogenesis, the exact molecular mechanisms by which Twist1 controls mesenchymal tissue formation remain largely undefined²⁶. This section focuses on the regulatory roles of Twist1 in signaling pathways as well as its binding partners.

A number of studies have shown that Twist1 is involved in FGF signaling^55,56,57,58. In Drosophila and C. elegans, Twist1 induces expression of the FGFR homolog DFR1 and egl-15, respectively²⁶. Humans with features of the autosomal dominant SCS can carry mutations in the TWIST1, FGFR2 or FGFR3 genes⁵⁹, supporting the speculation that Twist1 is required for FGF signaling maintenance during morphogenesis progression. Additionally, the observation that calvaria cells isolated from an SCS patient with a Twist1 gene mutation had decreased FGFR2 levels⁶⁰ also lends support to this hypothesis.

Twist1 is postulated to perform its central regulatory roles in organogenesis at least partially via its control over FGF, BMP and perhaps also TGFβ signaling⁵⁶. Unlike most other bHLH proteins, Twist1 can form both functional homodimers (T/T) and heterodimers with E12 (T/E). The T/T homodimers and T/E heterodimers appear to have distinct activities and regulate expression of different gene sets. The relative levels of Twist1 and helix-loop-helix Id proteins determine the ratio between these dimers⁵⁶. Id proteins represent a third class of HLH proteins that lack the basic domain and, therefore, cannot bind to DNA. Id proteins preferentially dimerize with E proteins such as E12 and thus prevent them from forming functional heterodimers with Twist1⁶¹. Twist1 forms T/E dimers in the absence of Id proteins and forms T/T dimers with increased Id protein levels⁵⁶. Consistent with this belief, in the osteogenic fronts of the cranial sutures where Twist1 and Id1 are coexpressed, genes regulated by T/T dimers, such as FGFR2, are expressed; the Id protein here binds to E proteins, thus favoring the formation of T/T dimers in these places, elevating and expanding the expression of FGFR2. T/E-regulated genes, such as thrombospondin 1 (TSP-1), are expressed in the mid-sutures where only Twist1 is expressed; in such places, E proteins are not occupied by Id proteins and are thus available to form T/E heterodimers. In the sutures of Twist1^+/− mice, the ratio between these dimers is altered to favor an increase in homodimers. This results in an elevated expression of T/T-regulated genes⁵⁷. Additionally, T/E dimers have recently been shown to inhibit BMP signaling⁶², which is predominantly active in the osteogenic fronts of cranial sutures. Thus, the decrease of T/E dimers in the cranial sutures of Twist1^+/− mice (or SCS patients) also allows for an expansion of BMP signaling. BMP signaling induces Id expression⁵⁵, which would further promote T/T formation. This positive feedback loop is a likely cause for the premature closing of the sutures, leading to craniosynostosis that is one of the major clinical features of SCS. Studies also showed that Twist1 activity may have a similar effect on FGF signaling in the limb bud^{15,55,63,64,65}. In its regulation of the growth and differentiation of limb bud tissues, Twist1 appears to be also involved in the sonic hedgehog (SHH) pathways, in addition to the FGF signaling pathways. Along with the altered expression of FGFR2, FGF4, FGF8 and FGF10, Twist1 deletion in the limb bud also reduced the overall activities of genes involved in SHH signaling in the limb bud mesenchyme, which include Shh, Gli1, Gli2, Gli3 and Ptch^65,66. Twist1 also appears essential to Bmp4 expression in the apical ectoderm, as well as Alx3, Alx4, Pax1 and Pax3 activities in the mesenchyme^65,67.

Hand1 and Hand2 are also members of the Class B bHLH subfamily, to which Twist1 belongs. The balance between Twist1 and Hand2 within the developing limb may be critical for normal morphogenesis, and an increase in Hand2 relative to Twist1 may result in polydactyly⁶⁸. This notion is strengthened by the observation that Twist1-Hand2 double-heterozygous null mice are more phenotypically normal than mice heterozygous for only Twist1⁶⁸. These findings suggest that Twist1 and Hand2 may function antagonistically in organogenesis⁶⁹. However, it remains unclear whether Twist1 forms heterodimers with Hand2 and/or Hand1.

Runx2 is a critical osteoblast-differentiation transcription factor essential for bone formation⁷⁰. Twist1 is expressed in Runx2-expressing cells during early skeletal development, and expression of the osteoblast-specific gene Runx2 occurs only after Twist1 expression decreases¹⁷. Twist1 is believed to maintain mesenchymal cells in an undifferentiated state by negatively regulating Runx2^17,55. This belief is supported by the following observations: (1) double heterozygotes for Twist1 and Runx2 deletion had none of the skull abnormalities observed in Runx2^+/−; (2) Twist1 deficiency in mice led to premature osteoblast differentiation; and (3) Twist1 overexpression inhibits osteoblast differentiation. Twist1 is speculated to perform its antiosteogenic functions through interaction with the Runx2 DNA-binding domain, and the relief of Runx2 inhibition by Twist1 is a mandatory event preceding osteoblast differentiation¹⁷. The lack of sufficient Runx2 inhibition by Twist1 in Twist1^+/− mice is believed to be responsible for the premature differentiation of odontoblasts, leading to the formation of extensive pulp stones in the tooth⁷¹.

TNFα signaling induces both proapoptotic and NF-κB-mediated antiapoptotic pathways. The TNFα-activated NF-κB upregulates both Twist1 and Twist2 expression to prevent cells from apoptosis. Interestingly, Twist1 and Twist2, in turn, can interact with the p65 (RelA) subunit of NF-κB to repress NF-κB-mediated expression of cytokine genes to control inflammatory responses. The importance of this negative feedback loop was demonstrated by the phenotypes that Twist1 and Twist2 heterozygous mutant mice or Twist2 null mice exhibit elevated levels of proinflammatory cytokines and perinatal death from cachexia. These findings suggest that the Twist1 and Twist2-mediated negative feedback regulation plays an important role in preventing overactivation of NF-κB-mediated cytokine expression⁷².

In summary, both the pivotal role that Twist1 plays in mesenchymal development and the biological function of Twist1 within mesenchymal cell populations are well established. Although the exact mechanisms by which Twist1 functions in organogenesis are not entirely clear, Twist1 likely plays its critical roles by regulating a set of target genes that include those in the FGF and SHH signaling pathways. Additionally, Twist1 may also modulate the functions of the Hand proteins (Hand 1 and 2), Runx2 and NF-κB. These downstream target genes or interacting proteins of Twist1 are known to be involved in the development of various mesenchymal derivatives and diverse physiological functions.

Regulation of Twist1 protein stability

Regulation of protein stability is an important way to control its function. It has been shown that truncated Twist1 proteins derived from nonsense mutations were unstable, resulting in SCS⁷³. Interestingly, formation of Twist1/E47 heterodimers stabilizes Twist1 protein, while formation of Twist1/Id1 heterodimers destabilizes Twist1 protein. Thus, Twist1 overexpression can suppress bone morphogenetic protein (BMP)-induced osteoblast differentiation, and this inhibition can be overcome by Id1 expression through induction of Twist1 degradation⁶². Recently, it was reported that Twist1 protein stability is largely regulated by mitogen-activated protein kinase (MAPK)-mediated phosphorylation on S⁶⁸. The S⁶⁸ in Twist1 can be phosphorylated by p38, JNK and ERK1/2 MAPKs, and this phosphorylation prevents Twist1 protein from ubiquitination-mediated degradation⁷⁴. Accordingly, activation of MAPKs by an active Ras protein or TGFβ treatment significantly increases S⁶⁸ phosphorylation and Twist1 protein levels without altering Twist1 mRNA expression, while blocking of MAPK activities by either specific inhibitors or dominant-negative inhibitory mutants effectively reduces both S⁶⁸ phosphorylation and Twist1 protein levels⁷⁴. These findings may suggest a positive correlation among active MAPKs, Twist1 protein level, epithelial-mesenchymal transition (EMT) and invasiveness of cancer cells.

The expression and roles of Twist1 in cancer

In addition to its essential role in modulating mesenchymal tissues critical for organogenesis, Twist1 is also expressed in and associated with many types of aggressive tumors, including breast cancer¹⁸, hepatocellular carcinoma^75,76, prostate cancer^19,77, gastric cancer^78,79, oesophageal squamous cell carcinoma^80,81, bladder cancer^82,83 and pancreatic cancer⁸⁴. Twist1 plays multiple roles in cancer initiation, progression and metastasis. More specifically, Twist1 can override oncogene-induced cell senescence and apoptosis^85,86,87, increase cancer cell resistance to chemotherapy^23,88, enhance cancer stem cell (CSC) population^89,90,91, and facilitate cancer cell invasion and metastasis^{18,20,24,68,92,93,94,95}.

Many recent studies have highlighted the role of Twist1 in promoting cancer cell EMT and metastasis. Cancer metastasis consists of several steps: EMT, local invasion, intravasation, transportation in the circulation, extravasation, survival and proliferation at a secondary organ site, and formation of overt metastatic lesions⁹⁶. Twist1 enhances the ability of cells within a primary tumor to undergo a pathological EMT^18,24,68, similar to the role of Twist1 in development⁹⁷. EMT allows tumor cells to migrate away from the primary tumor, enter the lymphatic system and/or blood stream, and settle into secondary tumor sites¹⁸.

While Twist1 promotes cancer initiation, progression and metastasis, its expression patterns, functions and molecular mechanisms by which Twist1 affects different types of cancers may vary. In this section, we will discuss Twist1 expression and function in different types of cancers and potential pathways by which Twist1 may participate in the initiation and progression of cancers.

Twist1 and breast cancer

Yang et al.¹⁸ observed that four types of mouse- mammary tumor cell lines isolated from the same breast cancer displayed distinct abilities to metastasize in mice. By comparing the gene expression profiles, they discovered that (a) increased Twist1 expression correlates with breast cancer invasion and metastasis, (b) suppression of Twist1 expression by siRNA in the metastatic mammary-carcinoma cells specifically inhibits the cells' ability to metastasize from the mammary gland to the lung, and (c) expression of Twist1 in the epithelial cell lines results in loss of E-cadherin-mediated cell-cell adhesion, activation of mesenchymal markers and induction of cell motility; Twist1 binds to the E-box elements in the promoter region of E-cadherin and represses the transcriptional expression of this cell-cell adhesion molecule. Based on these findings, along with the known functions of Twist1 as a master regulator of embryonic morphogenesis, the authors postulated that Twist1 contributes to metastasis by promoting EMT in cancer progression¹⁸.

A number of studies have shown the associative relationships among Twist1, EMT, and breast cancer metastasis^{22,23,24,88,92,98,99,100,101,102}. STAT3 is known to be involved in breast cancer progression. Using an RNA interference (shRNA) approach, Ling and Arlinghaus⁹⁸ examined the effects of STAT3 knockdown on mammary tumor growth in mice. They found that Twist1 was eliminated in STAT3 knockdown cells; the proliferation rate of these cells remained the same, but the invasive capability of these cells was significantly reduced. These observations suggest that STAT3 enhances Twist1 expression in its promotion of breast cancer progression.

Subsequently, another study²² compared low invasive human breast cancer lines with highly invasive human breast cancer lines, and showed that activation of STAT3 (i.e., phosphorylation of Tyr⁷⁰⁵ in the STAT3 amino-acid sequence) increased Twist1 expression while inhibition of STAT3 significantly reduced Twist1 expression in the aggressive (more invasive) human breast cancer cell lines. The inhibition of STAT3 reduced migration, invasion, and colony formation of these more invasive cancer cells. This study also found that STAT3 binds directly to the second proximal STAT3-binding site on the human Twist1 promoter and activates Twist1 transcription. Based on the strong correlation between the levels of activated STAT3 (i.e., Tyr⁷⁰⁵ p-STAT3) and Twist1 at the late stages of breast cancer, this group postulated that activated STAT3 transcriptionally induces Twist1 expression, which subsequently promotes the migration, invasion and anchorage-independent growth of breast cancer cells. Together with another observation that Twist1 transcriptionally induces AKT2 (a serine/threonine kinase) to promote oncogenic functions⁸⁸, Cheng et al.²¹ proposed that STAT3, Twist1 and AKT2 form a functional signaling axis to regulate pivotal oncogenic properties of cancer cells. Recently, Eckert et al. demonstrated that PDGFRα is a direct target gene of Twist1 in breast cancer cells. The Twist1-induced PDGFRα activates the Src kinase to promote formation of invadopodia, which are specialized membrane protrusions for extracellular matrix degradation. Therefore, this invadopodia-mediated matrix degradation induced indirectly by Twist1 is another mechanism by which Twist1 promotes breast cancer metastasis¹⁰³.

Some microRNAs may be involved in mediating cancer metastasis. The microRNA-10b (miR-10b) is highly expressed in metastatic breast cancer cells, and it positively regulates cell migration and invasion¹⁰². The miR-10b transcription is directly regulated by Twist1, and miR-10b in turn inhibits translation of the mRNA-encoding homeobox D10, resulting in increased expression of the pro-metastatic gene RhoC. Moreover, the level of miR-10b expression in primary breast carcinomas is associated with clinical cancer progression. These findings suggest that Twist1 indirectly upregulates RhoC expression to increase breast cancer, cell invasion and metastasis.

Steroid receptor coactivator-1 (SRC-1) is a coactivator for nuclear hormone receptors such as estrogen and progesterone receptors and certain other transcription factors such as Ets-2 and PEA3^104,105. In breast cancer, SRC-1 expression positively correlates with poor prognosis¹⁰⁵. A recent study showed that in the virus-polyoma middle T breast cancer mouse model, SRC-1 specifically promotes breast cancer metastasis without affecting primary tumor growth¹⁰⁶. A subsequent investigation by the same group found that SRC-1 serves as a coactivator for the transcription factor PEA3 to enhance Twist1 expression, suggesting a molecular mechanism whereby SRC-1 promotes breast cancer invasiveness and metastasis by upregulating Twist1 expression⁹².

As already mentioned, MAPKs phosphorylate Twist1 and increase Twist1 protein stability in cultured cells⁷⁴. Examination of the invasive human breast ductal carcinomas further revealed that the levels of S⁶⁸ phosphorylation and Twist protein positively correlate with the JNK MAPK activities, which are significantly higher in progesterone receptor-negative and HER2-positive breast cancers. These findings suggest that JNK activation by multiple signaling pathways may substantially promote breast tumor cell EMT and metastasis via phosphorylation and stabilization of Twist1⁷⁴.

Twist1 and hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is a rapid-growth metastatic tumor. Tissue microarray and immunohistochemical staining of paired primary and metastatic HCC showed that Twist1 overexpression correlated positively with HCC metastasis and negatively with E-cadherin expression⁷⁵. Further studies on different HCC cell lines revealed that HCC cells with increased levels of Twist1 and decreased levels of E-cadherin have higher metastatic ability. This suggests that Twist1 suppresses E-cadherin expression and induces EMT changes, which are partially responsible for the increased HCC-cell invasiveness⁷⁵. Subsequent investigations also revealed that Twist1 upregulated vascular endothelial growth factor (VEGF) and N-cadherin expression in HCC, suggesting that Twist1 may also play an important role in HCC angiogenesis¹⁰⁷. More recently, Matsuo et al.¹⁰⁸ showed that Twist1 overexpression in HCC cells enhanced cell motility, while its knockdown reduced cell migration. Yang et al.⁷⁶ analyzed three major EMT regulators, Twist1, Snail and Slug, and found that Twist1 and Snail synergistically enhance HCC metastasis: co-expression of Snail and Twist1 reduced E-cadherin levels and worsened HCC prognosis more dramatically than did Twist1 expression alone. Slug expression, however, did not seem to affect the outcome. Additionally, Sun et al.¹⁰⁹ demonstrated that Twist1 enhanced the motility, invasiveness and vasculogenic mimicry formation of HCC cells through the suppression of E-cadherin expression and the upregulation of N-cadherin.

Twist1 and prostate cancer

The main cause of prostatic cancer-related death is metastasis. These cancer cells have a particular predilection for metastasis to bones, and androgen-independent metastatic prostate cancer poses an even greater challenge to the treatment of this disease. Kwok et al.¹⁹ showed that down-regulation of Twist1 in androgen-independent prostate cancer cells increased their sensitivity to anticancer drugs and suppressed their migration and invasion abilities, suggesting Twist1 inactivation as a potential strategy to control the growth and metastasis of these cells. Subsequent studies by the same group^77,110 correlated high levels of Twist1 expression with aberrant E-cadherin expression and bone metastasis. Twist1 might promote bone metastasis by enhancing the osteomimicry of prostate cancer cells and by modulating prostate cancer cell-mediated bone remodeling via regulation of the osteolytic metastasis-promoting factor, DKK-1. These results were in agreement with those from another study¹¹¹, which demonstrated that Twist1 knockdown reduced the expression level of N-cadherin and inhibited the migration rate of PC-3 prostate carcinoma cells. The regulation of N-cadherin by Twist1 requires an E-box element located within the first intron of the N-cadherin gene. A more recent study⁸³ revealed that in addition to promoting the hallmark changes associated with EMT, such as E-cadherin down-regulation and N-cadherin upregulation, Twist1 also enhances VEGF production in prostate and bladder cancers. Since VEGF is an angiogenic factor, its enhanced production may accelerate angiogenesis associated with the metastasis of these tumors.

Twist1 and gastric cancer

Immunohistochemistry and RT-PCR analyses revealed high levels of Twist1 expression in gastric cancer tissues, and the increase in Twist1 expression correlated with lymph node metastasis⁸². A subsequent study further demonstrated that gastric cancer cells stably transfected with Twist1 had greater migration and invasion abilities, and formed a greater number of cancer nodules in the abdominal cavity and liver of nude mice inoculated with the transfected cells⁷⁸. Moreover, Twist1 overexpression in these cells promoted the expression of Tcf-4's downstream target genes cyclin D1 and MMP-2, while its suppression reduced cyclin D1 expression and MMP-2 activity. These results suggest that Twist1 may promote gastric cancer cell migration, invasion and metastasis via the Wnt/Tcf-4 signaling pathway. Microarray analyses revealed that depletion of Twist1 in the HGC-27 gastric cancer cells increased the expression of NF1, RAP1A, SRPX, RBL2, PFDN4, ILK (integrin-linked kinase), F2R, ERBB3, and MYB, and decreased the expression of AKR1C2, FOS, GDF15, NR2F1, ATM, and CTPS, supporting that many Twist1-regulated genes are involved in the differentiation, adhesion and proliferation of gastric cancer cells¹¹². The same research group also discovered that the MGC-803 and HGC-27 gastric cancer cells with higher Twist1 levels exhibited higher invasive potential than did the BGC-823 and SGC-7901 gastric cancer cells with lower Twist1 levels. Twist1 overexpression in BGC-823 cells increased their migration and decreased their sensitivity to the arsenic oxide-induced cell death, while Twist1 ablation in MGC-803 and HGC-27 cells suppressed migration ability, increased cell apoptosis in response to arsenic oxide, and inhibited the cell cycle. Furthermore, Twist1 and p53 levels were negatively correlated, further supporting Twist1 as a critical regulator of gastric-cancer cell proliferation and migration⁷⁹.

Twist1 and other types of cancer

Immunohistochemical staining and RT-PCR analyses showed upregulation of Twist1 expression in primary oesophageal squamous cell carcinoma^80,113, while immunoblotting analysis revealed the elevation of Twist1 expression in the oesophageal squamous cell carcinoma cell lines⁸⁰. In addition, high Twist1 expression in oesophageal squamous cell carcinoma was significantly associated with greater metastasis risks in patients with oesophagectomy, suggesting a role of Twist1 upregulation in the development of distant metastasis of oesophageal squamous cell carcinoma⁸⁰. Another study showed a correlation between high Twist1 expression and low E-cadherin expression. In the group with preserved E-cadherin expression, the 5-year survival rate was better for patients with low Twist1 expression than for those with high Twist1 expression⁸¹.

Immunohistochemical staining analyses of cancerous and non-cancerous bladder tissues revealed significantly higher Twist1 expression in the cancer specimens. Among the cancer tissues, Twist1 expression was remarkably higher in the metastatic lesions than in the primary tumors⁸². Additionally, bladder cancers also exhibited a correlation between Twist1 elevation and E-cadherin reduction.

Satoh et al. observed that the presence of MSX2 correlates with the malignant behavior of pancreatic cancer cells, and that MSX2 enhanced the proliferation, migration and liver metastasis of pancreatic cancer cells via the induction of Twist1. When MSX2 was knocked down in pancreatic cancer cells, Twist1 was down-regulated. These findings indicate that MSX2-induced Twist1 expression plays a crucial role in pancreatic cancer progression by inducing changes consistent with EMT⁸⁴.

Elias et al. analyzed Twist1 expression in human gliomas and normal brains using RT-PCR, Northern blot, in situ hybridization and immunohistochemistry. They found Twist1 expression in a majority of human glioma-derived cell lines and human gliomas. The expression of Twist1 was also observed in embryonic and fetal human brain neurons, but not in the glia of the mature brain. The increased Twist1 expression accompanied the transition from low-grade to high-grade gliomas, and Twist1 overexpression in a human glioma cell line significantly enhanced tumor cell invasion. These findings support the roles for Twist1 in both early glial tumorigenesis and subsequent malignant progression¹¹⁴.

Epstein-Barr virus (EBV)-associated nasopharyngeal carcinoma is highly metastatic compared to other head and neck tumors. A study showed that the principal EBV oncoprotein, latent membrane protein 1 (LMP1), upregulates Twist1 to induce EMT, suggesting the contribution of Twist1 induction by the human viral oncoprotein LMP1 to the highly metastatic nature of nasopharyngeal carcinoma¹¹⁵. Furthermore, the study revealed that LMP1 regulates Twist1 through the NF-κB pathway.

Immunohistochemical staining analysis of head and neck cancer tissues showed that Twist1 expression was positively associated with cancer progression and lymph node metastasis¹¹⁶. Further analysis in this study revealed a positive correlation between Twist1 expression and the expression of CXCR4 and CCR7, suggesting that Twist1 may regulate CXCR4 and CCR7 expression in these cancer cells, which in turn promotes lymph node metastasis.

Loss of E-cadherin triggers peritoneal dissemination (detachment from the primary lesion, spreading in the abdominal cavity) of epithelial ovarian carcinoma, leading to an adverse prognosis for most patients with this cancer. A study showed that suppression of Twist1 expression in epithelial ovarian carcinoma cells changes the cellular morphology from a fibroblastic and motile phenotype to an epithelial phenotype, and inhibits their adhesion to mesothelial monolayers¹¹⁷. Furthermore, this investigation revealed that Twist1 down-regulation reduced the expression of MMP-2, membrane type 1 MMPs, and adhesion molecules CD29, CD44 and CD54. These findings suggest that reducing Twist1 expression suppresses the multistep process of peritoneal dissemination and may be a potential therapeutic strategy for the treatment of this carcinoma.

The role of Twist1 in cancer cell survival, immortalization and acquired chemoresistance

Oncogenic insults usually induce p53 and/or retinoblastoma (Rb) expression and result in cell apoptosis or senescence, which is a defensive barrier against cell transformation and tumor progression. Thus, tumorigenesis needs to protect cells from apoptosis or immortalize cells from senescence. Interestingly, both Twist1 and Twist2 were shown to inhibit oncogene-induced and p53-dependent cell death. Further analysis revealed that Twist might affect p53 indirectly through inhibition of ARF expression to modulate the ARF/MDM2/p53 pathway⁸⁵. The same study also demonstrated that Twist could promote colony formation of ras-transformed mouse embryo fibroblasts (MEFs) in soft agar, and Twist overexpression might enhance rhabdomyosarcoma formation by inhibiting myogenic differentiation⁸⁵. Similarly, Twist1 was found to be constantly overexpressed in neuroblastomas with N-Myc amplification, where this Twist1 overexpression was responsible for the inhibition of the ARF/p53 pathway involved in the Myc-dependent apoptotic response⁸⁶. Furthermore, Ansieau et al.⁸⁷ also demonstrated that Twist1 and Twist2 play a role in preventing H-Ras-induced premature senescence by abrogating the p53- and Rb-dependent pathways. Twist1 or Twist2 works cooperatively with H-Ras to transform MEFs and induce complete EMT and aggressive cell migration and invasion in mammary epithelial cells. These findings suggest that Twist proteins may facilitate potential tumorigenic cells to escape from safeguard programs and acquire invasive features.

Twist1 also plays a role in the acquired resistance of cancer cells to chemotherapy. Twist1 upregulation is associated with cellular resistance to taxol and vincristine, two microtubule-targeting anticancer drugs in nasopharyngeal, bladder, ovarian, and prostate cancers¹¹⁸. On the other hand, the chemotherapy-induced cancer cell apoptosis is counter-regulated by a subset of NF-κB-regulated genes. Twist1 is one of the major targets of NF-κB responsible for antagonizing chemotherapy-induced apoptosis, suggesting an important role of Twist1 in NF-κB-mediated cell survival and chemoresistance¹¹⁹.

Twist1 and cancer stem-like cells

Multiple lines of evidence have demonstrated a link between Twist1-induced EMT and stem-like cells. Mani et al. have shown that induction of EMT by expressing Twist1 or Snail in mammary epithelial cells increases stem-like cell population with high CD44 and low CD24 expression, while isolated mammary epithelial stem-like cells express endogenous EMT-inducing factors including Twist1, Snail, SIP1, Slug and FOXC2, and EMT marker genes⁸⁹. Another study has shown that Twist may repress CD24 expression to increase the CD44-high and CD24-low stem-like cell population⁹⁰. Furthermore, expression of Twist or Snail in HER2-transformed mammary epithelial cells also facilitates EMT and generates cancer stem-like cells that efficiently form mammospheres, soft agar colonies and tumors⁸⁹. Moreover, Battula et al. further demonstrated that Twist1- or Snail-induced EMT could convert human mammary epithelial cells to mesenchymal stem-like cells with the capacity to differentiate into multiple cell types, including osteoblasts, adipocytes and chondrocytes. This study also demonstrated that these EMT-derived cells, but not the control cells, have the ability to migrate towards tumor cells and wound sites, as mesenchymal stem cells do⁹¹.

Dysregulation of Twist1 expression and function in cancer

As described in preceding sections, several factors have been shown to upregulate Twist1 expression in cancers, including NF-κB in nasopharyngeal carcinoma and PEA3 and SRC-1 in breast cancer^92,115. Several other factors have also been shown to regulate Twist1 expression. Firstly, HIF-1α has been shown to mediate Twist1 expression under hypoxia condition in tumors. Stabilization of the hypoxia-inducible factor-1α (HIF-1α) transcription complex caused by intratumoral hypoxia, promotes tumor progression and metastasis, leading to treatment failure and mortality in several types of human cancers. HIF-1α can bind to the hypoxia-response element in the Twist1 proximal promoter to upregulate Twist1 expression, thus promoting EMT and metastastic phenotypes of cancers¹²⁰. This study also revealed that co-expression of HIF-1α, Twist1 and Snail in primary head and neck cancers correlated with metastasis and the worst prognosis. These results suggest a key-signaling pathway involving HIF-1α and Twist1 that promotes metastasis in response to intratumoral hypoxia. Interestingly, SRC-1 has been demonstrated as a coactivator of HIF-1α¹²¹, although it is unclear whether SRC-1 can enhance HIF-1α-mediated upregulation of Twist1 expression. Secondly, thrombin contributes to the malignant phenotype by promoting tumor metastasis. Thrombin may promote tumor progression by upregulating Twist1 to enhance angiogenesis¹²². Thirdly, Twist1 has been implicated in type I interferon (IFN)-induced suppression of TNFα expression and thus, TNFα-mediated inflammation¹²³. Type I IFNs activate Fc receptors and Toll-like receptors, leading to induction and activation of the Axl receptor tyrosine kinase and downstream Twist1 expression. Twist1 subsequently binds to the E-boxes in the TNFα promoter and represses NF-κB-dependent expression of the TNFα gene¹²³. Finally, Twist1 has been shown to be involved in ILK-mediated upregulation of HER2¹²⁴. Overexpression or activation of ILK may upregulate Twist1 expression, and Twist1 in turn upregulates Y-box binding protein-1 (YB-1) expression to enhance HER2 expression¹²⁴.

Twist1 has been shown to upregulate the expression of several target genes important for cancer progression. Twist1 upregulates Akt2 expression in breast cancer cells, which enhances cell migration, invasion and resistance to chemotherapy⁸⁸. Furthermore, Twist1 directly upregulates Bmi1, a polycomb-group protein that maintains stem cell self-renewal and is frequently overexpressed in human cancers. Twist1 targets Bmi1 expression and works with Bmi1 to promote tumor-initiating capability and EMT by repressing E-cadherin expression¹²⁵.

In addition to activation of target gene expression, Twist1 also promotes cancer cell EMT, migration, invasion and metastasis by repressing target gene expression. It is well established that Twist1 promotes EMT by repressing E-cadherin expression by associating with the E-cadherin promoter^18,24,126. Twist1 also represses CD24 expression to help generate CD44-high/CD24-low cancer stem-like cells⁹⁰. Moreover, Twist1 may enhance MMP activities to promote cancer cell invasion, and metastasis by repressing the expression of TIMP1, a key inhibitor of MMPs¹²⁷.

Molecular mechanisms responsible for Twist1-mediated repression of E-cadherin expression

A recent study has investigated the molecular mechanisms by which Twist1 represses the E-cadherin expression to promote EMT and cancer cell migration, invasion and metastasis²⁴. The authors purified and characterized the Twist1-associated protein complex. They discovered that Twist1 either directly or indirectly interacts with several components of the Mi2/nucleosome remodeling and deacetylase (Mi2/NuRD) protein complex, including metastasis-associated protein 2 (MTA2), Rb-associated protein 46 (RbAp46), Mi2 and histone deacetylase 2 (HDAC2). Twist1 recruits this gene repression protein complex to the E-cadherin promoter, resulting in repression of the E-cadherin promoter activity and E-cadherin expression (Figure 3). Among the components of the Mi2/NuRD complex, Mi2 harbors chromatin-dependent ATPase activity and facilitates nucleosome mobility through a sliding mechanism^128,129,130. The combined activities of HDAC and ATPase in the Mi2/NuRD complex result in the generation of densely packed, hypoacetylated nucleosomes for gene silencing¹²⁸. The RbAp46 and/or RbAp48 subunits were originally identified as proteins associated with the Rb tumor suppressor¹³¹, and they may function as structural proteins that provide interactive interfaces for other components of the Mi2/NuRD complex^128,132,133. The Mi2/NuRD complex also contains one of the MTA protein family members, MTA1, MTA2 or MTA3^{128,134,135,136}. Each MTA member modifies the functional specificities of the Mi2/NuRD complexes relevant to its upstream and downstream signaling pathways and molecular targets. Although different members of the MTA family direct the Mi2/NuRD complex to play distinct functions, the primary function of the Mi2/NuRD complex is to repress gene expression involved in many biological processes, including cancer initiation and progression. In agreement with the roles of the Mi2/NuRD complex in mediating Twist1-dependent repression of the E-cadherin promoter, knockdown of MTA2 or RbAp46 releases the Twist1-repressed E-cadherin promoter activity and endogenous E-cadherin expression in cancer cells. Knockdown of MTA2 or RbAp46 in the 4T1 mouse mammary tumor cells or the MDA-MB-435 human cancer cells also inhibits their migration, invasion and metastasis, just as knockdown of Twist1 does²⁴. These findings not only provide novel mechanistic and functional links between Twist1 and the Mi2/NuRD complex but also establish new essential roles for the components of the Mi2/NuRD complex in cancer metastasis.

Figure 3

Twist1 recruits Mi-2/NuRD complex to repress E-cadherin expression and promote EMT and metastasis. Twist1 forms heterodimer with E12, and interacts with the components of MTA2 (metastasis-associated protein 2), RbAp46 (Rb-associated protein 46) and Mi-2 in the Mi-2/NuRD protein complex. Consequently, the Mi-2/NuRD complex is recruited to the E-cadherin promoter by Twist1, resulting in suppression of E-cadherin expression and promotion of EMT and metastasis of breast cancer. Knockdown of Twist, MTA2 or RbAp46 in the complex inhibits breast cancer cell metastasis.

In summary, numerous studies have revealed Twist1 upregulation in a variety of cancers. It is relatively clear that Twist1 promotes cancer development by protecting cells from oncogene- and chemotherapy-induced apoptosis and senescence and enhances cancer invasion and metastasis by promoting EMT. However, the exact mechanistic pathways by which Twist1 regulates cancer initiation and progression remain largely unknown. In this section, we have summarized the possible pathways and molecules via which Twist1 may participate in tumor development, progression and metastasis (Figure 4). This illustration scheme is based on the available data reported in the English literature over the past 7 years. It should be noted that the potential upstream and downstream regulators of pathways in different contexts of cancer cells could vary. Further studies are required to define the nature of Twist1's involvement in a defined cancer type. As the studies on Twist1 continue, more information regarding the specific roles of this molecule and its involvement in signaling pathways during cancer progression will certainly emerge. A better understating of Twist1's roles in cancer progression is likely to have important clinical implications for both prognosis prediction and therapeutic targeting of different cancers.

Figure 4

Possible regulatory pathways by which Twist1 is involved in cancer progression. Factors that directly or indirectly upregulate Twist1 are listed on the left side. Targets and cellular functions that are directly or indirectly regulated by Twist1 are listed on the right side. SRC-1, steroid receptor coactivator-1; STAT3, signaling transducer and activator of transcription 3; MSX2, Msh homeobox 2; HIF-1α, hypoxia-inducible factor 1α; NF-κB, nuclear factor kappa B; LMP1, EBV latent membrane protein 1; ILK, integrin-linked kinase; IFN, type I interferon; Axl, Axl receptor tyrosine kinase; miR-10b, micro-RNA 10b; Wnt, wingless and Int; Tcf4, transcription factor 4; Mi-2/NuRD, nucleosome remodeling and deacetylase protein complex; YB-1, Y-box binding protein-1; TIMP1, tissue inhibitor of metalloproteinase-1; VEGF, vascular endothelial growth factor; DKK-1, dickkopt-related protein 1; RhoC, ras homolog C; MMP2, metalloproteinase 2; CXCR4, chemokine (C-X-C motif) receptor 4; CCR7, chemokine (C-C motif) receptor 7; CD29, integrin β1; CD44, CD44 antigen; CD54, inter-cellular adhesion molecule 1; Bmi1, BMI1 polycomb ring finger oncogene; HER2, human epidermal growth factor receptor 2; TNFα, tumor necrosis factor α; EMT, epithelial-mesenchymal transition.