Although the products of oncogenic fusion genes are diverse, they can primarily be classified into two groups, transcription factors and TKs. Several oncogenic fusion proteins are transcription factors and TKs. In fact, the products of fusion genes are diverse; some may be neutral, some may play less important roles in cellular phenotypes and some may cause cell death in which we can not see this type of the translocation. The translocations found in cancer, however, clearly have critical functions in tumorigenesis. Generally, transcription factors and TKs play more important roles in cellular phenotypes, and this may partly explain why many fusion proteins detected in human cancer are transcription factors and TKs. It should be noted, that these so-called oncogenic fusion proteins as transcription factors and TKs are already different from their functions of parental proteins in several aspects and they often acquire some new functions.

It is clear that the sites of DSBs are related to the functional consequences of fusion genes. DSBs are not random (42) and occur preferentially in large and evolutionarily conserved genes (43,44), fragile sites (45), transcription start sites (14,46,47) and euchromatin (48,49). The breakpoints do not usually occur in their functional domains if these genes are encoded for transcription factors or TKs, thus fusion proteins can still retain the activities of transcription factors or TKs (42). Several studies have shown that DSBs preferentially occur in euchromatin, consistent with a greater chance for translocation to occur in the sites with transcription activity (14,46,47). Following exposure to ionizing radiation, DSBs occur more often in euchromatin than heterochromatin, suggesting the highly compacted chromatin can prevent from radiation damage. From another point, euchromatin is relatively loose and has a lack of protective mechanism, so it is easily attacked by radiation (48,49). In addition, the mechanisms of DSB repair in euchromatin are also different from heterochromatin. Since the time for DSB repair in heterochromatin is longer than euchromatin (50,51), by extrapolation, the higher frequency of chromosomal translocations in euchromatin than in heterochromatin is reasonable.

The products of several oncogenic fusion genes function as transcription factors. In this group, each fusion protein consists of N-terminal partner fused to the DNA binding domain at the C-terminus (Fig. 2). For example, EWS-FLI fusion protein, a characterized protein in Ewing sarcoma, consists of N-terminal part of EWS, a member of the TET family at the N-terminus, and C-terminal part of FLI, a member of the ETS family, at the C-terminus (31). As a chimeric transcription factor, EWS-FLI fusion protein has different transcription functions compared to its parental transcription factor FLI (32), despite identical DNA-binding domain. This mistargeting is associated with 85% of Ewing sarcoma development (52).

The functions of the fusion proteins as oncogenic transcription factors are various. Some stimulate gene expression, such as TMPRSS2-ERG and TMPRSS2-ETV1. Whether the TMPRSS2-ETS are really fusion proteins is under debate. Some people consider that the TMPRSS2-ETS translocations are the expression of ETS under the influence of the TMPRSS2 promoter as the expression of MYC under the IGH regulatory elements in Burkitt lymphoma (2). In fact, the TMPRSS2-ETS translocations are very heterogeneous, both TMPRSS2 at the 5′-end and ETS at the 3′-end have different fusion forms which generate different fusion transcripts, including splice variants (26,29,53,54). In most cases, the TMPRSS2 promoter and first exon or first 2 exons are juxtaposed to the ETS exons, with deletion of the ETS promoter and first exon or first 2 exons (55). Therefore, the fusion genes are under the control of the androgen-regulated TMPRSS2 promoter, resulting in the high level expression of oncogenic ETS fusion genes. For example, TMPRSS2-ERG gene fusion is the most common among these translocations and some are composed of the TMPRSS2 promoter and the first exon at the 5′-end and the transcription factor domain of ERG at the 3′-end, resulting in a truncated ERG protein lacking TMPRSS2 as the TMPRSS2 exon 1 is noncoding and does not contain an ATG (53), some are composed of the TMPRSS2 promoter and the first 2 exons (exon 2 containing an ATG at 142) at the 5′-end and the transcription factor domain of ERG at the 3′-end (designed type VI), resulting in a true fusion protein containing the first 5 amino acids of the TMPRSS2 at the N-terminus and a slightly truncated ERG protein at the C-terminus (Fig. 2) (53). Androgen can stimulate the transcription of the TMPRSS2-ETS fusion since all TMPRSS2-ETS fusions retain the TMPRSS2 promoter which contains AR binding sites. In most cases, ETS retains DNA-binding domain, which can stimulate the transcription of target genes for cell growth, invasion and metastasis and promote prostate cancer progression (26,28).

Some inhibit gene expression, such as t(12;21)/ETV6(TEL1)-RUNX1(runt-related transcription factor 1, previously known as AML1), t(8;21)/RUNX1-RUNXIT1(ETO), t(15;17)/PML-RARα and inv(16)/CBFB-MYH11, which inhibit the transcriptional activity of genes required for normal differentiation of hematopoietic cells. Although these fusion proteins may not be sufficient to induce leukemia alone (56), they increase the developmental risk of acute leukemia in patients with these fusion proteins (4,34,57). These fusion proteins repress the functions of transcription via the different molecular mechanisms. For example, RUNX1-RUNXIT1 protein is found in ~13% of acute myeloid leukemia (AML) (58). In RUNX1-RUNXIT1 protein, the translocation deletes the transactivation domain but retains the runt homology domain (RHD) responsible for binding to DNA at N-terminus of RUNX1 (Fig. 2) (59). RUNX1-RUNXIT1 protein interferes with wild-type RUNX1-dependent transcription via RUNXIT1 recruiting the nuclear corepressor (N-CoR)-histone deacetylase (HDAC) complex (60). ETV6-RUNX1 protein is the most common abnormality in childhood ALL, occurring in ~25% (4). In ETV6-RUNX1 protein, the translocation deletes the ETS domain of ETV6, a member of the ETS family, but retains the runt domain of RUNX1 (Fig. 2). Similar to the RUNX1-RUNXIT1 protein as a dominant negative inhibitor of RUNX1, the ETV6-RUNX1 protein represses RUNX1-dependent transcription via ETV6 recruiting N-CoR-HDAC complex (61). RUNX1 targeting genes are required for normal hematopoietic cell development. PML-RARα protein is linked to the development of APL, a genetic distinct subtype of AML. This fusion protein is composed of most of the functional domains of RARα (including the RAR binding domain and the DNA binding domain) and the majority of PML, including dimerization domain (Fig. 2). As a transcription factor, wild-type RARα releases SMRT/N-CoR corepressor after binding retinoic acid (RA) and induces the transcription of target genes that promote cell differentiation. However, this fusion protein alters the sensitivity to physiological levels of RA and impairs the release of SMRT/N-CoR corepressor from RARα, therefore blocking the differentiation of promyelocytes (62). One of the mechanisms of pharmacologic levels of all-trans retinoic acid (ATRA) treatment APL promotes the release of the corepressor from RARα and recovers RA response (11). Arsenic trioxide (As₂O₃) is also used to treat APL by promoting degradation of PML-RARα protein (63).

There are some mistarget gene expressions, such as mixed lineage leukemia (MLL) fusions (64). MLL gene on 11q23 is often rearranged with other partner genes in ALL and AML, accounting for 8% of pediatric and 10% of adult ALL (4), 15–20% of pediatric AML and <3% of adult AML (65), or biphenotypic (mixed lineage) leukemias. The MLL gene encodes a complex DNA binding protein with histone H3 lysine 4 (H3K4)-specific methyltransferase activity, which positively regulates gene expression including HOX genes. MLL protein consists of multiple functional domains, including the AT-hooks, DNA methyltransferase homology domain that contains a CXXC zinc finger motif and trithorax PHD domains at the N-terminus, the transactivation domain (TAD) and SET domain that possesses H3K4 methyltransferase activity at the C-terminus. Post-translationally, taspase I cleaves MLL to generate two fragments (MLLN p300 and MLLC p180) that form a stable complex by direct interaction of the FYRN and FYRC domains (66). Unlike classical sequence-specific DNA-binding transcription factor, MLL regulates the expression of target genes via epigenetic mechanisms, such as DNA and histone methylation modification (66).

Chromosomal translocations lead to the fusion of 5′-end portion of MLL to one of >60 different partner genes, resulting in the formation of different fusion genes, such as MLL-AF4 (4q21), MLL-AF9 (9p22), MLL-ENL (19p13.3), MLL-AF10 (10p12), MLL-AF6 (6q27), MLL-ELL (19p13.1) (Fig. 3) (66,67). All MLL fusion proteins retain N-terminal AT-hooks, DNA methyltransferase homology domain, thus preserving DNA binding activity whereas the trithorax PHD domains, TAD and SET domains are always replaced by the partners. In these fusions, the original MLL H3K4 methyltransferase activity is replaced by the partners which play a critical role in MLL oncoproteins (68). Although MLL fusion proteins lose the activity of H3K4 methylation, these fusion proteins gain the activity of H3K79 methylation via recruiting the H3K79 methyltransferase hDOT1L which can cause dysregulation of whole genomic expression and is associated with MLL leukemogenesis (67,69). Since hDOT1L plays a key role in the development of MLL leukemia, hDOT1L is an ideal target for MLL leukemia. Several hDOT1L inhibitors are underdeveloped. In particular, EPZ004777, a specific hDOT1L inhibitor, seems to be a promising drug for leukemia with MLL gene translocation (70).

Since >60 MLL fusion proteins have been found (71), the functions of MLL fusion proteins are very different, and the functions of some MLL fusion proteins remain unclear or not fully understood. To date, we know that MLL oncoproteins induce leukemia through several pathways. First, MLL oncoproteins act as transcriptional regulators that can bind DNA and induce aberrant expression of leukemic stem cell target genes, such as HOX, MEIS1, WNT and RNA polymerase II. Among MLL target genes, transcription factor HOX genes are particular and essential for MLL leukemogenesis (72). MLL-ENL, MLL-ELL, MLL-AF4, MLL-AF9 and MLL-AF10 have been demonstrated to induce acute leukemia using this pathway (Fig. 3) (67,73–75). Second, MLL fusion partners provide a dimerization motif, such as AF1p/Eps15 and GAS7. The MLL dimerization/oligomerization proteins can recruit co-activators or basal transcriptional machinery to result in the aberrant expression of target genes for inducing acute leukemia (76,77). Third, MLL fusion partners increase the stabilization of MLL oncoproteins. For example, all MLL-AF4, MLL-AF9, MLL-ENL and MLL-ELL exhibit resistance to degradation mediated by the cell cycle ubiquitin-proteasome system (71).

Another group of oncogenic fusion proteins harbor activities of TKs. In this group, each translocation generates a different fusion protein consisting of N-terminal partner fused to the TK domain at the C-terminus (Fig. 4). TK domain in these fusion proteins is intact although they are the truncated proteins (2,78,79). For example, in ALCL, ALK breakpoints are located in the intron flanked by exons 16 and 17, and exons 17–26 encoding the intracytoplasmic kinase domain of ALK are intact (80). It is similar to that of RET in PTC (81). As the regulatory parts of kinase are often lost and replaced by unrelated sequences, the kinase activity of these fusion proteins is determined by the N-terminal partners. In most cases, the N-terminal partners supply domains that promote dimerization/oligomerization, allowing fusion kinase to be activated in the absence of physiological stimulating signals (79,81–85).

BCR-ABL fusion protein is linked to the development of CML and ALL (4). ABL protein has two isoforms, 1a and b. ABL1b contains a C14 myristoyl saturated fatty acid moiety covalently linked to the Cap region at the N-terminus and is expressed at higher levels than ABL1a, which is not myristoylated. The Cap region of ABL contains endogenous autoregulatory domain which can inhibit kinase activity via stabilizing SH3 and SH2 domains of ABL (86,87). This fusion protein is composed of the majority of BCR at the N-terminus and most of the functional domains of ABL except Cap domain at the C-terminus (Fig. 4), resulting in the disregulatory activation of BCR-ABL TK (88). In addition, oligomerization domain and GRB2-binding site at tyrosine 177 (Y177) in BCR partner are also essential for BCR-ABL-mediated CML (82,89). Imatinib/Gleevec^®, a specific BCR-ABL inhibitor, was the first molecular target drug approved by the US Food and Drug Administration (FDA) to treat patients with CML in 1996. Dasatinib and nilotinib, second generation inhibitors of ABL, have also been approved to treat patients with imatinib-resistant CML (90).

H4-RET fusion protein is the most common chromosomal translocation in PTC, and accounts for 60–70% of PTC. This protein consists of the N-terminal promoter and leucine zipper domain of H4 at the N-terminus and the TK domains of RET at the C-terminus (Fig. 4). RET lacks the signal peptide and transmembrane domain in this chimeric oncoprotein, thus the aberrant TK activity of RET fusion is controlled by H4 partner which provides an active promoter and dimerization domain for ligand-independent activation of the fusion protein (91).

Approximately 5% of NSCLCs have inv(2)(p21;p23), resulting in the formation of echinoderm microtubule-associated protein-like 4 (EML4)-ALK fusion gene (85). EML4-ALK protein consists of various length EML4 containing the coiled-coil domain at the N-terminus and the intracellular catalytic domain of ALK at the C-terminus (Fig. 4). As ALK lacks the extracellular and transmembrane domain in this oncoprotein, so EML4 partner constitutively activates the TK of ALK via the dimerization of EML4-ALK, involved in the carcinogenesis of NSCLC (85). EML4-ALK is most commonly detected in non-smokers with NSCLC. NSCLC with EML4-ALK has unique pathological and clinical features, such as Asian patients, younger, adenocarcinoma and lack of EGFR and KRAS mutations (92). Crizotinib, an ALK inhibitor, was recently approved by the FDA to treat patients with ALK-positive NSCLC (93).

Proto-oncogenes are brought into proximity with the new cis-regulatory elements, leading to their overexpression which is seen in several types of lymphoma and leukemia, particularly in B and T cell malignancies. This is because V(D)J recombination exists during B and T cell development, which generates antibody and T cell receptor (TCR) diversity. However, V(D)J recombination may also increase the risk of chromosomal translocation in the same regions, which may partly explain why chromosomal translocation frequently occurs in several types of lymphoma and leukemia. For example, the overexpression of oncogenes c-MYC, BCL-2, CCND1 and BCL-6 in B cell lymphomas may be associated with errors in V(D)J recombination (Fig. 1) (16,94–96), suggesting the mechanism of chromosomal translocations in these B cell lymphomas is similar.

In a subset of T cell ALL (T-ALL), chromosomal translocation can make proto-oncogenes under the control of TCR regulatory elements, resulting in the deregulated transcription of these proto-oncogenes, such as TLX1 (HOX11), TLX3 (HOX11L2), LMO1, LMO2, c-MYC, LYL1, T-cell acute lymphocytic leukemia-1/stem cell leukemia (TAL1/SCL), TAL2 and NOTCH1 (Fig. 5). TLX1 and TLX3 belong to homeobox transcription factors. LIM domain only (LMO) 1 and LMO2 belong to LIM transcription factors containing LIM zinc-finger motifs. c-MYC, LYL1, TAL1/SCL and TAL2 belong to basic helix-loop-helix (bHLH) transcription factors. NOTCH1, one of NOTCH family, is a transmembrane protein.

The TCR is composed of two different protein chains. In 95% of T cells, TCR consists of αβ chains, whereas in 5% of T cells, TCR consists of γδ chains. TCRα (TCRA) and δ (TCRD) chain genes are localized on 14q11.2, TCRβ (TCRB) and TCRγ (TCRG) loci are localized on 7q34 and 7p15, respectively. The breakpoints often occur in TCRA/D or TCRB. The t(11;14)(p13;q11) and t(7;11)(q34;p13) have been found in 3% T-ALL (97). Both translocations lead to LMO2 (11p13) under the control of the TCRD or TCRB locus, resulting in LMO2 overexpression which may be involved in the T-ALL development (98). Proto-oncogene TLX1 (T-cell leukemia homeobox 1, previously known as HOX11 or TCL3) on 10q24 is normally not expressed in T cells and its expression is often deregulated in T-ALL (99). This deregulated expression of TLX1 is related to t(7;10)(q34;q24) and t(10;14)(q24;q11) which account for 7% of childhood and 31% of adult T-ALL (97). These translocations make TLX1 under the control of the TCRB or TCRA locus, resulting in the overexpression of TLX1 which may contribute to T-ALL via blocking apoptosis of developing T cell in the thymus (100). TLX1 overexpression has also been demonstrated in the absence of a 10q24 rearrangement, suggesting that other mechanisms, such as epigenetic alterations, can lead to this aberrant expression of TLX1(101,102). The situation is similar to TAL1 (1p32). Approximately 7% of childhood T-ALL and 12% of adult T-ALL have t(1;14)(p32;q11), leading to deregulated expression of TAL1 under control of the TCRA/D loci (4). However, the overexpression of TAL1 in T-ALL also occurs in the absence of TAL1 rearrangement, suggesting that other mechanisms may influence the overexpression of TAL1(103).

NOTCH1 plays crucial roles in cell development, hematopoietic stem cell maintenance and T cell fate specification in the mature organism (104). NOTCH1 is regarded as an oncoprotein. In a low number of human T-ALL patients, they had t(7;9)(q34;q34.3) which results to fuse the 3′ end of NOTCH1 (9q34.3) to TCRB locus, leading to overexpression of a truncated NOTCH1 protein that lack the negative regulatory region (NRR) (105). NRR is NOTCH1 extracellular domain and responsible for preventing ligand-independent receptor activation.