Transcriptional Complexity of the HOXA9 Locus

Abstract

Evolutionarily conserved HOX genes play an important role during development and hematopoiesis. HOX protein products are transcription factors whose precise mechanism of action is still poorly understood. Regulation of HOX gene expression has been the topic of various studies. While alternative splicing and alternative promoter usage have been known to increase the number of transcripts across the HOX clusters, more recently high-throughput analyses have identified a number of new coding and noncoding RNA molecules whose function is not known. Here we review the transcriptome of the most studied HOX locus, HOXA9. Strict control of HOXA9 expression has been shown to play a critical role in hematopoiesis while aberrant expression has been shown important to the development of leukemia. However, it is still unclear how various transcripts from this locus are regulated and what specific role(s) each one of them plays.

Introduction

Drosophila HOM genes were identified in flies with homeotic mutations, in which a body segments, such as antennae, are converted to different segments, such as legs[1]. Subsequently, mammalian homologues of HOM genes were identified and named HOX genes. Both HOM and HOX genes play an essential role during embryogenesis by binding to DNA through a highly conserved 183-nt sequence called the homeodomain. Homeodomain containing proteins are transcription factors that play a role in determination of segment identity of the developing embryo along various body axes[2]. During evolution, cluster duplication gave rise to four mammalian HOX clusters located on different chromosomes, with a total of 39 HOX genes found in 13 different paralogous groups. In each cluster, the spatial and temporal pattern of expression of the HOX genes corresponds to their physical position within the cluster. The 3′ genes are expressed earlier than the 5′ genes, have a more anterior boundary of expression and have a higher sensitivity to transcriptional inducers such as retinoic acid[3]. More recent data has shown that HOX genes are also involved in hematopoietic cell development. Certain HOX genes are expressed in a lineage specific pattern with 5′ genes of the HOXA cluster being highly expressed in primitive hematopoietic progenitors and downregulated in more differentiated cells[4].

Analysis of the HOX cluster transcriptome shows an extensive repertoire of transcripts whose function is still poorly understood. Extensive splicing and alternative promoter usage adds an additional layer of complexity. It is proposed that some HOX genes share common cis-regulatory elements which help in the colinear pattern of HOX gene expression. Polycistronic messages have also been described for certain HOX genes[5]. In most of the knockout and overexpression studies only canonical forms of HOX transcripts have been studied, without truly discriminating between similar isoforms. In recent years, a number of noncoding transcripts have also been identified making the HOX clusters an even more complex system to decipher. Some of these are antisense transcripts which play a role in regulation of HOX gene expression[6]. Of special interest is the presence of microRNAs in the HOX cluster which may partially explain the importance of the whole HOX cluster in development and disease. Discovery of numerous noncoding and regulatory transcripts in the HOX clusters has raised additional questions regarding the role of each one of these transcripts.

HOXA9 Locus

The most studied HOXA cluster gene is HOXA9. Among 6800 genes tested, HOXA9 was shown to be the single most correlative marker of poor prognosis in patients with acute myeloid leukemia[7]. This finding confirms previous reports outlining the significant role that HOXA9 plays not only during normal hematopoiesis but also in leukemia. Hoxa9 deficient mice have impaired hematopoietic development[8]. Furthermore, the Hoxa9 locus is a common site of retroviral integrations in leukemia-prone BXH2 mice[9,10]. Resulting leukemias express high levels of Hoxa7 and Hoxa9 genes[9]. Overexpression of Hoxa9 and Meis1 also leads to immortalization of mouse bone marrow cells and leukemia development in mice[11]. It is not surprising that Hoxa9, as a transcription factor, regulates genes involved in stem cell biology[12]. While much work remains to be done on the downstream targets of HOXA9, studies have shown that HOXA9 is able to function both as transcriptional activator and repressor. The HOXA9 gene itself is a target of leukemic translocation involving chromosome 11(HOXA9) and chromosome 7 (NUP98) that encodes the NUP98-HOXA9 oncogenic protein[13,14]. NUP98-HOXA9 transformation of NIH-3T3 cells depends on HOXA9-mediated DNA binding, strengthening the idea that misregulation of HOXA9 targets is important in disease development[15].

The canonical HOXA9 transcript contains two exons: exon CD and the homeodomain-containing exon II (Fig. 1). This transcript is expressed in variety of embryonic and adult tissues[16]. A portion of exon CD can be alternatively spliced giving rise to a different new transcript with a shifted reading frame[17]. The splice site found in the first exon is present not only in humans, but in other species as well suggesting that this alternatively spliced transcript may play an important role in development and/or disease[16]. The frame shift causes a premature stop codon upstream of the homeodomain, leading to production of a non-homeodomain containing protein[17]. Absence of the homeodomain ablates DNA binding activity of the protein. Like canonical HOXA9, this alternatively spliced transcript is expressed in numerous tissues and its expression is conserved across species. The truncated protein is found in both the nucleus and cytoplasm which is in contrast to the homeodomain containing HOXA9 found exclusively in the nucleus. Like canonical HOXA9, the truncated protein binds CBP with high efficiency and the two isoforms may compete for CBP as well as for other HOXA9 interacting proteins[16]. If this model is accurate, the truncated protein may interfere with normal HOXA9 function as well as have other novel functions that have not been described yet.

Figure 1.

Schematic diagram of the HOXA9 locus and various cloned transcripts. Exons are represented with dark boxes on the top bar. Translocation sites for TCRβ, NUP98 and retroviral integration sites are labeled with arrows. Each transcript is labeled with an associated reference or EST accession number. Dark boxes on the bottom bar represent CpG islands.

Exon II is a particularly interesting region not only because it codes for the DNA binding domain, but because together with the 3′ UTR is the most shared region among various HOXA9 isoforms. Studies have shown existence of a number of transcripts that utilize exon II of the HOXA9 gene. One of these transcripts starts much more upstream from the canonical HOXA9, in the vicinity of the HOXA11 gene[18]. Other isolated transcripts utilize the previously described splice acceptor site within the CD exon and contain all of exon II[17, Popovic (unpublished)]. The majority of the transcripts presumably share the common 3′UTR as well. Studies that use 3′UTR tags for detecting expression levels of HOXA9 are not able to differentiate between the various isoforms and their data represent the “total” HOXA9 pool. More specific probes will need to be designed to dissect the pattern of expression for each one of the transcripts.

Another intriguing region in the HOXA9 locus is located approximately 4.2kb upstream of the canonical promoter and proximally 3′ of the HOXA10 gene. This region is very highly conserved across multiple species and encodes an additional HOXA9 exon designated AB. The AB exon is employed in a number of cloned transcripts as an alternative first exon in lieu of exon CD. While some RNAs are spliced to give AB-exon II transcripts, others use the internal CD splice acceptor site to form an AB-CD-exon II transcript[13,19, Popovic(unpublished)]. We have isolated two of these transcripts that differ only in 40bp of sequence used at the end of the AB exon. Both of the transcripts are found in bone marrow progenitor cells and sequence analysis predicts that they do not code for protein. It is still not understood the mechanism by which these RNAs are regulated or their precise transcription start site. Recently, Mainguy et al. described an antisense transcript transcribed from this region[18]. Interestingly, microRNA 196b is also located in this area and the majority of mir-196 targets tested so far are HOX genes themselves[20]. It is possible that the expression of sense and antisense transcripts regulates not only the transcription from this region but also help regulating expression of other HOX genes. This mode of regulation would aid in the collinear model of HOX gene expression.

Conclusions

There remain many unanswered questions regarding transcriptional regulation of the HOXA9 locus. It is clear that more detailed studies need to be performed to appreciate the difference in expression patterns of the various HOXA9 transcripts. Understanding the functional differences between all these transcripts will allow us to better comprehend what goes wrong in the diseases associated with HOXA9 misexpression. For a long time dogma considered the majority of the DNA sequence to be ‘garbage’, but with the discovery of noncoding RNA molecules, we began to realize the complexity of the transcriptional regulatory machinery. Proper expression of HOXA9 is evidently very important in normal development and disease. Different HOXA9 isoforms may possess alternative functions and thus play diverse roles in normal processes as well as in disease progression. Together with other HOX genes, HOXA9 is overexpressed in MLL-associated leukemias, however it is still unclear whether this is also true for all of the transcripts from the locus. MLL may differentially regulate expression levels of the various RNAs and in the presence of MLL fusion proteins the normal transcript balance may be altered. Also, the HOXA9 locus is a common site for chromosomal rearrangements with the TCRβ locus[21]. Patients with those types of translocations have high levels of HOX gene expression, including HOXA9. The chromosomal breakage occurs in the region between exons CD and AB and may influence misexpression of multiple transcripts. In efforts to develop successful therapies for treatment of leukemia and other diseases, we must understand the nature and function of the molecules we are trying to target. In the case of HOXA9, there is much work that remains to be done.