- (2011) Volume 12, Issue 3
Sophie Gray1, Hardev S Pandha1, Agnieszka Michael1, Gary Middleton2, Richard Morgan1
1Postgraduate Medical School, Faculty of Health and Medical Sciences, University of Surrey;
2Royal Surrey County Hospital NHS Trust. Guildford, United Kingdom
The HOX genes are a family of homeodomain-containing transcription factors that determine cellular identity during development and which are subsequently re-expressed in many types of cancer. Some recent studies have shown that HOX genes may have key roles both in pancreatic development and in adult diseases of the pancreas, including cancer. In this review we consider recent advances in elucidating the role of HOX genes in these processes, how they may connect early developmental events to subsequent adult disease, and their potential both as diagnostic markers and therapeutic targets.
HOXB7 protein, human; HOXD13 protein, human; HXR9 peptide; Pancreatic Neoplasms
The development and maintenance of cellular identity is vital in both embryonic and adult tissues for normal organ function. Key to this is the establishment of stable transcriptional states within the cell, a process in which transcription factors have a key role. One group of transcription factors of particular note in this regard are the HOX genes, a family of homeodomaincontaining transcription factors that determine cellular and tissue identity by regulating specific transcriptional programmes [1, 2, 3]. Mammals have 39 HOX genes split between four groups of linked genes on different chromosomes, which are thought to have arisen from a series of duplication events. These groups are known as A, B, C and D, and the genes within each group are numbered from the 3’ most member (1) to the 5’ most member (13) [4]. Thus, for example the 3’ most member of the HOXA group is known as HOXA1 (Figure 1). Equivalently numbered genes in each group (e.g., HOXB4 and HOXD4) are referred to as paralogues and are thought to have arisen from a common ancestral gene as they represent the same position within the ancestral HOX cluster. Members of each group often share enhancer regions and are coregulated, giving rise, in part, to coordinated temporal and spatial expression patterns in the developing embryo whereby the 3’ most HOX genes are expressed earlier and more anteriorly than their more 5’ neighbours [5]. In this manner the HOX genes give rise to a pattern of overlapping expression domains along the anteroposterior axis that is key to defining the identities of cells along it [5].
Figure 1. The arrangement of HOX genes in the mammalian genome and their potential roles in pancreatic development and cancer.
The HOX transcription factors have a relatively limited specificity for binding to DNA, but this is enhanced by the binding of co-factors such as PBX and MEIS that increase DNA binding specificity and also modify transcriptional regulation [6, 7, 8]. These co-factors, as well as the HOX genes themselves, are highly conserved between animal phyla. The HOX genes also show considerable similarities between each other, especially paralogues and neighbouring members of the same group [1, 2, 3]. This has resulted in a high degree of functional redundancy, particular in respect to early developmental events although later, more organ specific development is generally dependent on a smaller number of HOX genes [9, 10]. The expression domains of HOX genes that are established during development are generally preserved in the adult [11], and there are a number of examples where adult HOX expression is required for the continuation of correct cell identities. This is most apparent where cells are turning over quickly, for example in the proliferation and differentiation of blood cells [1] and in the renewal of the endometrium [12]. The importance of HOX genes in these processes indicates a role in the promotion of cell proliferation and survival, in addition to maintaining cellular identity, and indeed HOXB4 is crucial for the continued proliferation of hematopoietic stem cells [1]. It is perhaps not surprising then that HOX genes are frequently deregulated in cancer, where their primary function also seems to be in promoting proliferation whilst preventing apoptosis [13, 14, 15, 16, 17].
The pancreas develops from endodermal cells in the future midgut region of the embryo, and is dependant on a number of inductive interactions between the endoderm and other tissues. The best characterised of these events is the development of the dorsal pancreatic primordia which is initiated in the endoderm by signalling from the overlying notochord through the secreted proteins activin and FGF-2. These signals are not spatially restricted to the notochord adjacent to the prepancreatic endoderm however, and the responsiveness of endodermal cells is presumably modified by pre-existing anteroposterior information in the endoderm. The HOX genes are key determinates of anteroposterior identity [2, 5, 18], and the expression patterns of HOX genes in the early endoderm suggests that HOXA4, HOXA5 and HOXB4 provide the spatial information needed to restrict the response to signals from the notochord [19]. Correspondingly, retinoic acid, which regulates HOX expression through binding to nuclear hormone receptors, is also known to have a key role in pancreatic development [20]. Furthermore retinoic acid is sufficient to drive embryonic stem cell differentiation towards functional insulin producing cells [21], and HOXA4, HOXA5, HOXB4 and HOXA1 (see below) are all activated directly by retinoic acid [22].
These early patterning events give rise to pancreatic progenitor cells which, in turn, are subdivided into exocrine and endocrine progenitors through the presence or absence of Notch signalling, respectively. In a number of early developmental processes Notch signalling is dependent on HOX transcription factors, together with the PBX co-factor [18, 23, 24], with the HOX1 paralogues (HOXA1 and HOXD1) mediating key transcriptional changes [18, 24]. Likewise HOXA1 expression is activated in pancreatic exocrine cells and is required for exocrine development, possibly by modulating TGF-beta signalling from the foregut mesoderm [25].
The endocrine progenitor cells are also defined by the expression of specific transcription factors, Pax-6 for alpha-cells and gamma-cells (glucagon and pancreatic polypeptide secreting, respectively), and Pax-4 for beta-cells and delta-cells (insulin and somatostatin secreting, respectively). Pax transcription factors are known to cross-regulate HOX target genes, at least in part through direct interactions with the latter [26], and also through the direct regulation of a number of HOX genes including HOXD4 [27]. Thus, HOX genes play a number of key roles in pancreatic development from the specification of early endodermal progenitor cells through to the determination of specific cellular subtypes within the maturing pancreas.
The deregulation of HOX genes in cancer is now well established, although in general rather less is known about their function [28]. For a number of malignancies, including melanoma [14], myeloma [13], and ovarian [15], renal [17], lung [16] and, indeed, pancreatic cancer [29], it has been shown that the HOX genes of paralogue groups 1 through 9 can promote cell survival by blocking apoptosis. In this respect many of these 27 HOX genes have a redundant or at least a highly overlapping function [13, 17, 29]. Targeting the antiapoptotic function of this group of genes has been achieved by antagonising the interaction between HOX proteins and the PBX co-factor. This approach exploits a highly conserved hexapeptide motif on PBX that is required for HOX binding [6, 7, 8]. The motif forms part of peptides such as HXR9 that act as competitive antagonists of HOX/PBX dimer formation can induce apoptosis both in vitro and in vivo, at least in part through the greatly elevated expression of cFos [13, 17, 29]. Disrupting HOX/PBX dimer formation in the pancreatic cancer derived cell line T3M4 also blocks cell proliferation [29], and reduces the expression of a number of cancer-related target genes by at least one order of magnitude. These include TMPRSS3, a transmembrane serine protease involved in tumour invasion and metastasis that is frequently over expressed in pancreatic cancer [30], and the S100 calcium binding protein P (S100P) which is involved in regulating the cell cycle and cell proliferation [31]. Interestingly HOXB2 and HOXA10 are also down regulated when HOX/PBX dimer formation is blocked, indicating a possible auto-regulatory pathway for these genes [29].
Other functions of HOX genes in pancreatic cancer may be more specific to particular members of the HOX family. Although to date there is only very limited data on HOX expression in normal pancreatic tissue of both the developing and adult pancreas, it would seem that generally these genes (HOXA1 [25], HOXA4 [19], HOXA5 [19], and HOXB4 [19]) are not up regulated in pancreatic cancer. Instead, other HOX genes are expressed, some of which have known oncogenic functions. Most notable amongst these is HOXB7 which has been shown to mediate epithelial to mesenchymal transition in breast cancer cells through the induction of bFGF [32], and to promote proliferation in oral cancer [33] and progression and metastasis in lung cancer [34]. HOXB7 is also over expressed in pancreatic cancer, both in primary pancreatic adenocarcinoma and in the pancreatic adenocarcinoma cells lines AsPc-1, BxPC-3, MiaPACA and PANC1 [35]. Two neighbouring genes of the HOXB cluster, HOXB5 and HOXB6, are also over-expressed in pancreatic cell lines as well as resected infiltrating pancreatic cancer tissue. Although the significance of this finding is unknown, it is noteworthy that forced over expression of HOXB6 in murine bone marrow is sufficient to immortalise a population of myelomonocytic precursor cells leading to acute myeloid leukemia [36]. Also up regulated in pancreatic cancer is HOXB2 [37], which was found to be present in 38% of pancreatic tumours. HOXB2 was shown to have prognostic value, as its expression is associated with nonresectable tumours, and when presented in resected tumours it is associated with poor survival. This finding may relate to the apparent ability of HOXB2 to promote metastasis in lung cancer, where its expression is also associated with a poor prognosis [38]. HOXB2 additionally promotes proliferation, as it is bound by the tumour suppressor protein p205 that is known to delay G2/M progression in dividing cells [39].
Other HOX genes are strongly down-regulated in pancreatic cancer. These include HOXD13, which is originally involved in the determination of the terminal digestive and urogenital tracts. HOXD13 expression is lower in tumour tissue compared to normal pancreatic parenchyma [40], and the absence of HOXD13 expression in tumours is associated with a significantly poorer prognosis, the 12-month survival rate for patients with HOXD13- tumours being 17.2% as compared to 79.8% for patients with HOXD13+ tumours. These findings imply that HOXD13 functions as a suppressor of metastasis, a characteristic shared by two further HOX genes, HOXB1 and HOXB3 [41]. Specific knockdown of either of these genes is sufficient to reduce the migration of pancreatic cancer cells in vitro and invasion and metastasis of pancreatic cancer in vivo, using zebrafish embryo xenotransplantation models. The same study also showed that HOXB1 and HOXB3 are suppressed by the microRNA miR-10a, the expression of which is significantly higher in metastatic pancreatic cancer [41]. HOX gene deregulation is also associated with premalignant pancreatic intraepithelial neoplasia (PanIN). HOXB2 was found to be expressed in 15% of early PanIN lesions [37], and as described above, is also associated with the fully malignant state, it is therefore possible that expression of HOXB2 in PanIN lesions may predict the development of cancer but this is yet to be shown conclusively. Also up regulated in PanIN is HOXA5 [42], which is of interest because other studies have suggested that its function is closer to a tumour suppressor rather than an oncogene as it activates the transcription of a number of key tumour suppressors, including p53 [43].
Although relatively little is still known of HOX expression and function in pancreatic cancer as compared to many other cancer types, it is already clear that HOX genes are of functional significance to the malignant phenotype. It is not therefore surprising to find that a number of HOX genes have prognostic value and may ultimately be of clinical relevance [38, 40]. More importantly still, HOX gene expression may prove to have diagnostic value for pancreatic cancer. The extremely poor survival rates for this disease are due in part to its predominantly late diagnosis; specific HOX genes expressed in, for example, circulating tumour cells, could potentially detect the presence of pancreatic cancer before the onset of symptoms. Additionally, a similar approach might also be of value in defining pancreatic cancer as a primary tumour in cases where the origin of metastasis is uncertain.
HOX genes are also potential therapeutic targets in pancreatic cancer. To date the only successful strategy for blocking HOX gene function is through interfering with the HOX/PBX interaction using a peptide mimic (HXR9) of the conserved PBX binding domain in HOX [13, 14, 15, 16, 17]. This is a novel strategy in pancreatic cancer, and is worthy of investigation as a possible therapeutic approach.
The authors have no potential conflict of interest
