SEARCH

SEARCH BY CITATION

Keywords:

  • Lim-only protein;
  • Hox;
  • Pbx;
  • Hoxa2;
  • Hoxb2;
  • hindbrain patterning;
  • mouse

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The embryonic functions of Hox proteins have been extensively investigated in several animal phyla. These transcription factors act as selectors of developmental programmes, to govern the morphogenesis of multiple structures and organs. However, despite the variety of morphogenetic processes Hox proteins are involved in, only a limited set of their target genes has been identified so far. To find additional targets, we used a strategy based upon the simultaneous overexpression of Hoxa2 and its cofactors Pbx1 and Prep in a cellular model. Among genes whose expression was upregulated, we identified LMO1, which codes for an intertwining LIM-only factor involved in protein–DNA oligomeric complexes. By analysing its expression in Hox knockout mice, we show that Lmo1 is differentially regulated by Hoxa2 and Hoxb2, in specific columns of hindbrain neuronal progenitors. These results suggest that Lmo1 takes part in a Hox paralogue 2–dependent network regulating anteroposterior and dorsoventral hindbrain patterning. Developmental Dynamics 236:2675–2684, 2007. © 2007 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Homeoproteins of the Hox family are evolutionary conserved transcription factors that have been identified as master control proteins in embryonic development. Typically, misregulation or inactivation of Hox genes results in shifting the developmental identity of embryonic territories' development (reviewed in Maconochie et al.,1996; Kmita and Duboule,2003). In particular, in the arthropod and vertebrate segmented body, segments take on the developmental fate of more posterior or anterior body parts upon Hox mutations, a phenomenon that was termed homeosis more than a century ago.

As genetic architects in shaping the embryo, Hox genes control key molecular developmental programmes (reviewed in Maconochie et al.,1996; Graba et al.,1997). Thus, they are predicted to tune the balance between cell proliferation, differentiation, and death, integrate positional information provided by signalling molecules, and take part to cell-to-cell dialogue controlling cellular migration and cell fate commitment. In mammals, 39 Hox genes are distributed among 13 paralogy groups and are involved in patterning the axial skeleton, the hindbrain and spinal cord, cranial neural crest derivatives, branchial arches, and the limbs (Maconochie et al.,1996; Favier and Dolle,1997; Davenne et al.,1999). Finally, they also contribute to control several steps of organogenesis (Hombria and Lovegrove,2003) and recent data provided evidence that some fulfil specific roles at adulthood, like in the mammary gland (Chen and Capecchi,1999) or during hematopoiesis (van Oostveen et al.,1999). At the molecular level, the DNA binding specificity and affinity for target sites of Hox proteins is enhanced by the formation of trimeric complexes containing Pbx/exd and Prep/Meis/hth cofactors, members of the TALE (Three-Amino acid Loop Extension) group of homeoproteins (Mann and Affolter,1998). Such cofactors are also known to modulate the transcriptional activity of their Hox partners (Li et al.,1999).

While it is generally accepted that Hox proteins should take part in the control of a wide panel of genes, only a handful of their targets have been identified so far (Shen et al.,2000; Leemans et al.,2001; Bobola et al.,2003; Samad et al.,2004; Cobb and Duboule,2005; Kutejova et al.,2005; Lei et al.,2005; Martinez-Ceballos et al.,2005; Williams et al.,2005). In this study, we focused on Hox paralogue group 2 proteins and, in particular, on the genetic cascade downstream Hoxa2. Inactivation of Hoxa2 and its misexpression in several model animals have revealed that it plays a pivotal role in the patterning of neurons and cranial neural crest cells, in particular those colonizing the second branchial arch-, and in the regulation of chondrogenesis and osteogenesis (Gendron-Maguire et al.,1993; Krumlauf,1993; Rijli et al.,1993; Gavalas et al.,1997; Kanzler et al.,1998; Davenne et al.,1999; Jungbluth et al.,1999; Grammatopoulos et al.,2000; Pasqualetti et al.,2000; Ohnemus et al.,2001; Hunter and Prince,2002; Santagati and Rijli,2003; Dobreva et al.,2006; Oury et al.,2006; Massip et al.,2007). In addition, its expression pattern suggests that it may play some role in hematopoiesis (Vieille-Grosjean et al.,1992).

To improve our understanding of Hoxa2 function, we designed an approach to identify putative target genes up- or downregulated by Hoxa2. As Hox specificity is in part acquired through cofactor interaction, we co-expressed Hoxa2 with its cofactors Pbx1 and Prep in a cellular model and identified cDNA fragments representative of Hoxa2 regulated genes. Among the candidate targets, LIM domain only-1 (LMO1), previously known as Rhombotin-1, was further analyzed for its participation in a Hox–dependent regulatory cascade in vivo. This gene was first identified at a chromosomal translocation in a T-cell acute leukaemia (Boehm et al.,1988). Lmo1 is expressed in the developing mouse central nervous system (CNS) (Greenberg et al.,1990; Boehm et al.,1991; Foroni et al.,1992), as well as in the adult CNS in a cell-type specific fashion (Hinks et al.,1997). Recently, a null mutation of the Lmo1 gene was obtained in the mouse that did not cause overtly abnormal phenotypes, possibly due to functional redundancy with other members of the family, while combined inactivation of both Lmo1 and Lmo3 genes resulted in perinatal lethality (Tse et al.,2004).

Here, we show that Lmo1 transcripts are spatially restricted to distinct longitudinal columns of neuronal progenitors and neurons spanning the length of several rhombomeres (r) and overlapping with subsets of both Hoxa2 and Hoxb2 expression domains. Interestingly, Hoxa2 and Hoxb2 differentially regulate Lmo1 expression in a rhombomere-specific manner, as assessed by analysis in knockout mice. Thus, Lmo1 belongs to a Hox paralogue 2–dependent regulatory network involved in neuronal patterning in the hindbrain of the mouse embryo.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

COS7 Cells Are a Suitable Cellular System to Assay for Target Gene Transcriptional Response to Hoxa2

To design an assay to detect genes whose transcription is modulated by Hoxa2, we first aimed to identify a cell line devoid of Hoxa2 expression. Expression of Hoxa2 was assessed by RT-PCR on total RNA extracted from several cell lines, including embryonic stem (ES) cells, teratocarcinoma (EC) cells, embryonic fibroblasts (MEF), and transformed cells like the NIH3T3. The majority of the tested cell lines constitutively expressed Hoxa2 (Table 1), although at different levels. Hoxa2 appeared to be silent in murine EC P19 cells, but only at early culture steps. Hoxa2 was also silent in COS7 cells of primate origin, which were thus selected for this study.

Table 1. Semi-quantitative RT-PCR Detection of Hoxa2 Expression in Culture Cell Lines
Cell lineHoxa2
COS7
EC-P19++
EC-F9+++
ES-129SvEv++
MEF+++
Neuro2A+
NIH3T3++
HEK293+++
J774+

To evaluate the responsiveness of a known target enhancer upon Hoxa2 expression in these cellular contexts, an expression vector for Hoxa2 was cotransfected with a target reporter construct containing the well-characterized Hoxb1 autoregulatory enhancer (Popperl et al.,1995). This enhancer has been shown to be responsive to Hoxa1, Hoxb1, and to paralogue group 2 Hox proteins as well (Di Rocco et al.,1997). The Hoxb1 autoregulatory element (ARE) is made of three binding sites for heterodimers of Hox and Pbx, and contains a recognition site for the TALE homeoproteins Meis or Prep.

Cotransfection of an ARE-driven luciferase reporter (ARE-Luc) construct with an expression vector for Hoxa2 in COS7 cells did not provide significant reporter activation (Fig. 1). In contrast, a strong synergistic stimulatory effect was observed when Hoxa2 was cotransfected with Pbx1a and Prep1 expression vectors. As a control, a mutant Hoxa2 derivative, in which the crucial homeodomain residues Q50 and N51 were changed into alanines (Hoxa2(QN-AA)) to abolish binding activity (Gehring et al.,1994; Matis et al.,2001), did not stimulate the reporter activity, either alone or in combination with Pbx1a and Prep1.

thumbnail image

Figure 1. Hoxa2, Pbx1a, and Prep1 are active on the Hoxb1 autoregulatory enhancer in COS7 cells. A reporter construct (ARE-luc) consisting of the luciferase gene placed under the control of the Hoxb1 ARE enhancer was transfected in COS7 cells, alone or in combination with expression vectors for Hoxa2, Hoxa2 (QN-AA), Pbx1a, and Prep1 proteins. Values are expressed as ratios between luciferase activity and control β-galactosidase activity. Bars indicate the standard deviation of 3 independent experiments.

Download figure to PowerPoint

Thus, the high reporter transactivation mediated through the ARE enhancer upon Hoxa2, Pbx1a, and Prep1 co-expression indicated that COS7 cells could be a suitable system for the screening of putative target genes selectively modulated by Hoxa2 in combination with its cofactors.

Identification of LMO1 as a Candidate Target Gene of Hoxa2

Since strong Hoxa2-mediated transactivation of the ARE reporter occurred only if Pbx1a and Prep1 were also provided, we collected mRNA from COS7 cells transfected for the expression of these three proteins together. To identify genes that were regulated by Hoxa2, we compared the mRNA content of these cells to that of COS7 cells transfected with the defective Hoxa2(QN-AA) variant combined to Pbx1a and Prep1. The mRNA populations obtained from both cell samples were reverse transcribed and processed through several rounds of subtraction and PCR amplification, according to the Representational Difference Analysis (RDA) procedure (Hubank and Schatz,1994). For detection of upregulated genes, the cDNA population obtained from the Hoxa2 transfected cells (hereafter referred to as Hoxa2 representation), was titrated out by excess of cDNA synthesized from those cells transfected with the inactive Hoxa2(QN-AA) mutant (Hoxa2(QN-AA) representation) and the final differential amplification products corresponding to candidate target genes were subcloned and sequenced. Among those candidate target genes (Matis et al., unpublished data), we identified the LIM-Only-1 gene (LMO1) previously reported as Rhombotin-1 and known to be expressed in the developing hindbrain. Southern blots of the respective cDNA representations with a LMO1 probe confirmed that LMO1 mRNA species showed different abundance in both cell populations (data not shown).

Necessary Partnership Between Hoxa2 and TALE Cofactors for the Upregulation of LMO1

To verify that LMO1 reproducibly responded to Hoxa2 activity, COS7 cells were transfected with Hoxa2 or Hoxa2(QN-AA), together with Pbx1a and Prep1, mRNAs were further purified, and semi-quantitative RT-PCRs specific for the LMO1 gene were run and standardized to that for the β-actin gene. Upon Hoxa2, Pbx1a, and Prep1 transfection, LMO1 expression showed a 9.80 fold (± 0.29) increase with respect to the presence of the defective Hoxa2(QN-AA). These results are fully in agreement with the outcome of the RDA assays.

To evaluate the impact of the Pbx1a and Prep1 cofactors onto the transcriptional response of LMO1, transfections and RT-PCR assays were also performed with the Hoxa2 expression vector alone, or, conversely, only in the presence of Pbx1a and Prep1 vectors. Transfection of Pbx1a and Prep1 did not or only slightly modified LMO1 expression in the absence of Hoxa2 (Fig. 2). Similarly, LMO1 displayed only a partial, though significant, activation by Hoxa2 in the absence of the cofactors (Fig. 2). Such dependency on the simultaneous presence of the Pbx1a-Prep1 cofactors was confirmed further. A Hoxa2 variant, Hoxa2(WM-AA), displaying two amino acid substitutions in its Pbx interaction domain (or hexapeptide), was generated that was expected to abolish Hox–Pbx heterodimer formation (Phelan et al.,1995; Remacle et al.,2004). In all transfections, Hoxa2(WM-AA) behaved like a loss-of-function variant, i.e., the protein has lost its ability to transactivate the Hoxa2 target, either in the absence or in the presence of the Pbx1a and Prep1 cofactors (Fig. 2).

thumbnail image

Figure 2. Activation of the LMO1 gene by paralogy group 2 Hox proteins in synergy with Pbx1a and Prep1 cofactors. Semi-quantitative RT-PCRs were performed for the expression of LMO1 in COS7 cells (control) and in COS7 cells transfected for Hoxa2, Hoxa2(WM-AA), Hoxa1, Hoxb1, or Hoxb2 expression vectors. Values were normalized to the reference β-Actin gene expression. Bars indicate the standard deviation of 3 independent experiments.

Download figure to PowerPoint

Altogether, these data show that our approach allows us to efficiently select Hoxa2 target genes whose regulation relies on its synergistic partnership with TALE cofactors.

LMO1 Is Selectively Regulated by Paralogue Group 2, Though Not Group 1, Hox Proteins in Transfected Cells

Paralogue Hox proteins may display functional redundancy due to the high sequence conservation of their homeodomain (Greer et al.,2000; Tvrdik and Capecchi,2006). To evaluate if the observed LMO1 activation was specific to Hoxa2, or if other Hox proteins were able to activate it as well, we performed transfection experiments with distinct Hox expression vectors. We first assayed the ability of Hoxb2 for transactivation, as it is the closest relative to Hoxa2. We also tested that of Hoxa1 and Hoxb1 whose expression patterns partially overlap those of group 2 paralogues in the hindbrain. As shown in Figure 2, only the paralogy group 2 proteins were able to significantly enhance the expression of the target. Hoxb2 provided an LMO1 gene activation comparable to Hoxa2. In addition, as shown for Hoxa2, the Pbx1a and Prep1 cofactors synergized for the Hoxb2-mediated transactivation. In contrast, Hoxa1 and Hoxb1 did not stimulate LMO1 activity either on their own or in the presence of cofactors (Fig. 2).

Thus, these data indicate that Hoxa2 and Hoxb2 may display some redundancy in their ability to activate LMO1. Interestingly, Hox proteins of the first group of paralogy were unable to provide the same response under similar conditions.

The Expression of Lmo1 Is Differentially Regulated by Hoxa2 and Hoxb2 in the Developing Mouse Hindbrain

Among the target genes identified by RDA, Lmo1 held our attention because preliminary expression data provided evidence that this gene is expressed in the embryonic hindbrain (Greenberg et al.,1990; Boehm et al.,1991; Foroni et al.,1992), where Hoxa2 and Hoxb2 are known to play a critical role. In particular, expression of a lacZ reporter transgene driven by a portion of the Lmo1 promoter was prominent at the level of rhombomeres 2 and 4 in the hindbrain (Greenberg et al.,1990). To further correlate the Lmo1 expression pattern with those of Hoxa2 and Hoxb2 in the hindbrain, mouse embryos from E9.25 to E10.5 were subjected to whole-mount ISH using an antisense probe for Lmo1. The Lmo1 expression pattern was determined by the analysis of whole-mount as well as of transverse sections of stained embryos (Fig. 3).

thumbnail image

Figure 3. Expression pattern of Lmo1 in the early mouse embryonic development. A,B: Lateral view of whole stained embryos at 20–25 somites (E9.25) (A) and 28–33 somites (E9.75) (B). In A, the two black arrows show the ventral stripe expressing Lmo1 in r2 and r4. C,F,I,L,O, D,G,J,M,P, and E,H,K,N,Q: Transverse sections of stained embryos at 20–25 somites, 28–33 somites, and 35–42 somites (E10.5) stages, respectively; (C–E) at the level of r2, (F–H) r4, (I–K) r5, (L–N) the r5/6 boundary, (O–Q) r7. The black arrowheads show the ventral and/or dorsal stripes of Lmo1 expression, the red arrowheads show the medio-dorsal and medio-ventral stripes. In C to Q, dorsal is on top.

Download figure to PowerPoint

In the rhombencephalon, Lmo1 transcripts were detected caudally to the rhombomere (r)1/r2 boundary (Fig. 3A). Moreover, Lmo1 expression pattern was temporally dynamic. At E9.25, the staining was restricted to single ventro-lateral stripes in r2 and r4 (Fig. 3A,C,F; and data not shown), while additional medial and dorsal stripes of Lmo1 expression appeared later on between E9.75 and E10.5 (Fig. 3D,E,G,H,J,K,M,N,P,Q; see also flat-mount preparation in Fig. 4A). At E10.5, Lmo1 was expressed in ventral (i.e., medial on flat-mount preparation) columns on either side of the floor plate starting just rostral to the r1/r2 boundary and running through at least r6. Staining was more prominent in r2 and especially r4, where they contain the developing Vth (trigeminal) and VIIth (facial) nerve motor nuclei, respectively (Fig. 4A). More dorsally (i.e., laterally on flat-mount preparation), another set of stripes of strong Lmo1 expression extended from a sharp rostral limit at the r1/r2 border caudally to the r5/r6 boundary. In between these two main stripes, at least two additional columns of Lmo1 expression were evident that, nonetheless, displayed lower expression levels (Figs. 3E,H,K,N,Q, 4A). In the intermediate and dorsal stripes, Lmo1 transcripts were generally detected both in the ventricular zone and the mantle layer. In the ventral stripes, expression was instead mostly detected in the mantle layer of postmitotic progenitors (Fig. 3D,E,J,K,M–Q). In the spinal cord, expressing stripes were also observed albeit at lower levels than in the hindbrain (not shown). From E10.5 onward, the ventral-most labelling was fading down caudally to the r4 territory.

thumbnail image

Figure 4. Lmo1 expression is differentially altered in Hoxa2 and Hoxb2 mouse mutant backgrounds. E10.5 embryos of (A) wild type (WT), (B) Hoxa2−/−, (C) Hoxb2−/− genotypes were hybridized with a Lmo1 antisense probe and dissected for flat-mounting. The black arrows in B indicate the lack of Lmo1 expression in rhombomere 2 (R2) and 3 (R3) of Hoxa2−/− mutants. In C, the vertical dashed line accompanies the ventral decrease of Lmo1 expression in the R4 territory of Hoxb2−/− mutants, as compared to WT in A.

Download figure to PowerPoint

Previous analysis of the Hoxa2 and Hoxb2 expression patterns in the developing mouse hindbrain revealed a set of complementary, partially overlapping, and quantitatively distinct distributions both along the anteroposterior (AP) and dorsoventral (DV) axes (Davenne et al.,1999). Along the AP axis, Hoxa2 has an AP expression limit at the r1/r2 border, while Hoxb2 rostral expression boundary is just offset by one rhombomere at the r2/r3 boundary (Davenne et al.,1999). Dorsoventrally, Hoxa2 and Hoxb2 displayed overlapping, though quantitatively distinct, alternating domains of higher and lower expression organised in broad longitudinal columns running the length of the hindbrain (Davenne et al.,1999). Thus, the Lmo1 expression pattern overlapped with subsets of the Hoxa2 and Hoxb2 transcript distributions, suggesting that Lmo1 could be a paralogue group 2 Hox target in the mouse hindbrain.

We, therefore, characterized the expression pattern of Lmo1 in Hoxa2−/− (Rijli et al.,1993) and Hoxb2−/− (Davenne et al.,1999) knockout mouse embryos. We found selective changes along the AP and DV axes in mutants for both Hoxa2 and Hoxb2. At E10.5, Lmo1 expression in Hoxa2−/− mutants was selectively lacking in the r2-r3 portions of the dorsal columns normally extending from r2 to r6 (Fig. 4B, arrows). However, expression in the ventral columns did not appear to be affected (Fig. 4B). This observation implies that Hoxa2 directly or indirectly positively regulates Lmo1 expression in r2-r3. In contrast, in Hoxb2−/− mutants, a clear reduction of the ventral columns of Lmo1-expressing cells was observed specifically in r4, whereas the remainder of the expression pattern was apparently unchanged (Fig. 4C; see also coronal sections through r4 in Fig. 5).

thumbnail image

Figure 5. Depletion of Lmo1 expression in ventral r4 of the Hoxb2−/− mutant. Coronal sections of whole-mount hybridized E10.5 embryos. The Lmo1 hybridization signal is selectively reduced in the ventral portion of the r4 territory in Hoxb2−/− embryos (arrows), as compared to the wild type (WT).

Download figure to PowerPoint

Altogether, these observations show that Lmo1 takes part in the regulatory network under the control of Hoxa2 and Hoxb2 in vivo, i.e., in the developing hindbrain of the mouse embryo, thus supporting the results obtained in the cellular model.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We developed a cellular approach to identify genes regulated by the murine Hoxa2 protein. Taking advantage of the knowledge that Hox target specificity is enhanced by interaction with cofactors, we co-transfected Hoxa2 with Pbx1a and Prep1 proteins in COS7 cells and looked to the transcriptional response of putative target genes with respect to control cells transfected with an inactive Hoxa2(QN-AA) mutant. Among candidate target genes obtained by RDA and found to be differently represented in cDNA populations derived from these two cellular populations (Matis et al., unpublished data), we identified LMO1, which appeared to be up-regulated by a factor 9 to 10 by Hoxa2, Pbx1a, and Prep1. It might be argued that COS7 cells do obviously not reproduce the physiological context of Hoxa2 activity in mammalian development and adult life. Nevertheless, the LMO1 gene appeared to be regulated by Hoxa2, Pbx1a, and Prep1 but did not respond or slightly responded to Pbx and Prep or to Hoxa2 alone. In addition, it did not respond either to Hoxa2(WM-AA), Hoxa1, or Hoxb1, in the absence as well as in the presence of Pbx1a and Prep1, but it was activated by the Hoxb2 protein. Therefore, the LMO1 response to group 2 Hox proteins, Pbx1a, and Prep1 in COS7 cells appears to be quite specific.

Whether LMO1 is a direct or an indirect target of group 2 Hox proteins remains to be determined. LMO1 possesses two alternative first exons arising from the use of alternative promoters (Boehm et al.,1990). Nevertheless the proteins encoded by these two transcripts differ by only two N-terminal amino acid residues. Both LMO1 promoters are active during embryogenesis (Boehm et al.,1990; Greenberg et al.,1990; Foroni et al.,1992). We inspected the LMO1 promoter regions for sequence elements matching Hox-Pbx consensus binding sequences. This analysis was performed by running the Genomatix MatInspector software (http://www.genomatix.de, Quandt et al.,1995; Cartharius et al.,2005) followed by a visual analysis to identify sites specifically matching the consensus sequence for group 2 Hox responsive elements (Lampe et al.,2004; Samad et al.,2004; Ferretti et al.,2005; Tumpel et al.,2007). Four putative group 2 Hox-Pbx binding sites were found, three within a 1.4-kb sequence interval upstream of the more distal transcription start site and one between the two alternative promoters (data not shown). The roles of such sites in the Hoxa2- and Hoxb2-mediated regulation of LMO1 are currently under investigation.

LIM domains are zinc-binding motifs that mediate protein–protein interactions and are found in a wide variety of cytoplasmic and nuclear proteins (Schmeichel and Beckerle,1994). LMO1 belongs to the LIM-only proteins, which contain two or more LIM domains but otherwise no other recognized protein domains and which include the related LMO2, LMO3, and LMO4 proteins (Gill,1995). It is encoded by a gene that was first identified in humans at a chromosome translocation locus t(11;14) (p15;q11) causing T-cell acute lymphoblastic leukaemia (T-ALL; Boehm et al.,1988). In erythroid cells, the Lmo2 protein has been shown to be critical for erythroid differentiation (Warren et al.,1994) and it was found in oligomeric complexes involving the basic helix-loop-helix proteins Tal1 and E47, and the transcription factors Ldb1 and Gata1 (Wadman et al.,1997). In such complexes, Lmo2 and Ldb1 proteins would serve as non-DNA-binding bridging partners. Such an intertwining role for Lmo proteins was also confirmed for Lmo1, which appeared to interact with Ldb1 and Tal1 proteins in vivo as well (Wadman et al.,1994; Valge-Archer et al.,1998). These data suggest that the Lmo1 protein, as Lmo2, may play a role in hematopoiesis. Indeed, like Lmo2, Lmo1 is expressed during thymocyte development (Herblot et al.,2000). Moreover, misexpression of either Lmo1 or Lmo2 can lead to perturbation of T-cell differentiation and leukaemia (Herblot et al.,2000; Ferrando et al.,2002).

Hoxa2 is weakly expressed in the pre-B-lymphocytes but not in the T-cell lineage. On the other hand, Hoxb2 is not detected in either B- or T-cell lymphoblastic lineages (Magli et al.,1991; Vieille-Grosjean et al.,1992; Pineault et al.,2002). Since the misregulation of Lmo1 in the T-lineage has been reported to induce T-ALL leukaemia, it can be hypothesized that, as regulators of Lmo1, the aberrant expression of Hoxa2 or Hoxb2 in T-cell precursors may result in leukemogenesis as well. In that regard, it is noteworthy that chromosomal translocations involving the TCRB locus and targeting the HOXA locus are associated with T-ALL (Soulier et al.,2005). Such translocations have been reported to provoke a global overexpression of HOXA genes. Since misregulation of HOXA genes has also been reported for other translocation events leading to T-ALL, anchoring the TCRB-HOXA translocation cases to a typology of T-ALL strongly supports a major oncogenic role of HOXA genes. Strikingly, according to this typology, the subgroups of T-ALL associated with HOXA misregulation display a high average expression of genes characteristic of an arrest at Pre-T stages of the normal T-cell differentiation pathways (Soulier et al.,2005).

In the mouse developing central nervous system, Lmo1 expression pattern led to the hypothesis that this gene may play a role in brain development (Greenberg et al.,1990; Boehm et al.,1991; Foroni et al.,1992). Its recent inactivation in the mouse showed that, on its own, Lmo1 has no obligatory role (Tse et al.,2004). However, some degree of functional redundancy occurs with other members of the family, such as Lmo3. In fact, analysis of compound Lmo1/Lmo3 mouse mutants revealed perinatal lethality of newborn pups, in contrast to single mutants (Tse et al.,2004). Although a detailed molecular analysis of the developing hindbrain has not been performed in compound Lmo1/Lmo3 mutants (Tse et al.,2004), our expression study in the hindbrain of wild type as well as of Hoxa2−/− and Hoxb2−/− mutant mouse embryos strongly supports that Lmo1 is involved in a regulatory network controlling dorso-ventral patterns of neuronal development. In this respect, Lmo1 has been shown here to be expressed in a dynamic and spatially restricted fashion and to be active downstream of Hoxa2 and Hoxb2 in specific subpopulations of neuronal progenitors of the rhombencephalon, including in the developing facial motor nuclei of the VIIth nerve (Fig. 5). We, therefore, propose that Lmo1 may act as a downstream effector mediating Hoxa2- and Hoxb2-dependent neuronal patterning programmes in restricted cell populations.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Plasmid Construction

Hoxa2 expression vectors are based on the pCMX-PL1 plasmid (Matis et al.,2001). pGIH306 contains a 2.3-kb EcoRI- StuI genomic DNA fragment of Hoxa2, which encodes the Hoxa2 protein. The two-amino-acid substitution, involving the glutamine 50 and asparagine 51 of the homeodomain (Hoxa2(QN-AA)), was created by site-directed mutagenesis (mutagenic primers: 5′-GAAAGTGTGGTTTGCCGCTCGGAGAATGAAGC-3′). The substitution of the WM amino acids of the hexapeptide motif of Hoxa2 (Hoxa2(WM-AA)) was generated by mutagenic PCR (mutagenic primer: 5′- CCTGAGTATCCCGCG GCGAAGGAGAAGAAG-3′). Expression vectors for Pbx1a and Prep1 have been described by Remacle et al. (2002) and Goudet et al. (1999), respectively. The pAdMLARE contains the ARE-luc reporter and has been described elsewhere (Popperl et al.,1995; Di Rocco et al.,1997).

Cell Culture and Transient Transfection

COS7 cells were maintained in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% fetal bovine serum (Biowithaker), 100IU/ml of penicillin and 100 μg/ml streptomycin (Gibco), at 37°C in a humidified, 5% CO2 atmosphere.

For cotransfections aiming at transactivation of luciferase reporter genes, exponentially proliferating cells were trypsinized and 4.105 cells were plated per 35-mm culture dish. Cells were transfected by standard calcium phosphate precipitation procedure, 24 h after plating as described previously (Remacle et al.,2002). Each cotransfection experiment was performed with a constant amount of 10.2 μg of DNA containing 2 μg of each expression vector (Hoxa2 or Hoxa2(QN-AA), Pbx1a and Prep1), 4 μg of the pAdMLARE reporter vector, and 0.2 μg of lacZ internal standard plasmid (Remacle et al.,2002). Lysis and enzymatic activity dosages were performed with β-gal Reporter Gene Assay (chemoluminescent) kit (Roche) and the Luciferase Reporter Gene Assay (High Sensitivity) Kit (Roche).

For the Representational Difference Analysis (RDA) and RT-PCR experiments, exponentially proliferating cells were trypsinized and 5.105 COS7 cells were plated per 10-cm culture dish. Each transfection was performed with a constant amount of 24 μg of DNA containing 8 μg of each expression vector.

Representational Difference Analysis

Forty-eight hours after transfection, total RNA was isolated from cells by a guanidine isothiocyanate/ cesium chloride extraction. The mRNA was then purified on a oligo-dT-cellulose column of mRNA Purification Kit according to the manufacturer's protocol (Pharmacia). cDNA was synthesized by oligo(dT) priming using cDNA synthesis module as recommended by the manufacturer (Amersham). Double-stranded cDNA was synthesised from 5 μg poly(A)+-RNA and RDA was performed as described by Hubank and Schatz (1994). For each RDA experiment, two successive subtraction/amplification rounds were performed with a 1:100 (1st subtraction) and a 1:800 (2nd subtraction) excess of driver cDNA representation (Hubank and Schatz,1994). The final difference products were cloned into the pBluescript SK- vector (Stratagene), and sequenced.

Southern Blot on cDNA Representation

Ten micrograms of cDNA representation of Hoxa2-Pbx1a-Prep1 or Hoxa2(QN-AA)-Pbx1a-Prep1 transfected cells were electrophoresed in agarose gel and blotted to Porablot NY-Plus membranes (Macherey-Nagel). After UV-crosslinking, membranes were hybridised with [32P]-dCTP labelled DNA probes. The probes were cleaned on a ChromaSpin-10 (BD Biosciences). After overnight hybridization at 65°C in hybridization solution (5 ml Denhardt's solution 50×, 12.5 ml SSC 20×, 500 μl SDS, 800 μl hearing sperm DNA at 10 μg/ml and 31ml H2O), membranes were washed once in 2 × SSC for 5 min at 65°C, twice in 2 × SSC/0.1% SDS for 30 min at 65°C, and then twice in 0.2 × SSC/0.1% SDS for 20 min at 65°C. For quantitative analysis, the signals on the membranes were quantified using the Instant Phosphorimager for 2 hr.

RT and PCR Reactions

For RT-PCR amplification of target RNA, total RNA was isolated from cells with TRIZOL Reagent according to the manufacturer's protocol (Gibco), 48 hr after transfection.

Two micrograms of RNA were treated with DNAse I and subjected to reverse transcription (RT). cDNA synthesis was processed for 2 hr at 42°C using a Avian Myeloblastosis Virus Reverse Transcriptase (U.S. Biochemicals), after priming with random hexamers at a final concentration of 76 μg/ml (pd (N6), Pharmacia Biotech). Conventional PCR reactions were performed (Takara) with addition of 2 μCi of [32P]dCPT for semiquantitative PCR. As control for cDNA synthesis, the β-actin sequence was amplified. The endogenous LMO1 gene expression was recorded by PCR with specific primers: 5′-GATCAGCTTGCTGCAAGCAG-3′ and 5′-GATCAAGGACCGCTATCTGC-3′.

Whole-Mount In Situ Hybridization

CD1 mice were paired for mating overnight, after which females were checked for vaginal plugs. Day 0 of gestation was defined as the midpoint of the dark cycle during which copulation took place. Females were killed by cervical dislocation and the embryos at embryonic day (E) 9.25, 9.75, and 10.5 were dissected free of their membranes.

One probe recognizing the Lmo1 coding sequence was used in in situ hybridization (ISH). The sequence alignment between the coding sequence of Lmo1 and Lmo2 showed that they share 52% identity, the sequence alignment between the coding sequence of Lmo1 and Lmo3 showed that they share 58% identity, and the sequence alignment between the coding sequence of Lmo1 and Lmo4 showed that they share 40% identity. The Lmo1 coding sequence was obtained by RT-PCR from RNA extracted from E12.5 mouse embryos, and cloned in the pBluescript SK- (primers for amplification and cloning: 5′-GGA ATT CAT GGT GCT GGA CAA GGA GGA CGG-3′ and 5′-CCC AAG CTT GGT GCC ATT GAG ATG CCC CTC C-3′). Sense and antisense riboprobes were synthesized with T3 or T7 RNA polymerase in the presence of digoxigenin-UTP according to the manufacturer's recommendation (Roche and Stratagene).

Whole-mount ISH on embryos from E9.25 to E10.5 was performed essentially as described (Gofflot et al.,1998). After ISH staining, embryos were bleached by dehydration followed by a rehydration in PBT-ETOH solutions at 4°C. The embryos were then placed in a 25% glycerol solution for 10 min, and stored in 50% glycerol.

After observation and photography, the whole labelled embryos were processed for vibratome section at 50 μm as previously described (Gofflot et al.,2001).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Virginie Aerts for technical assistance in automated DNA sequencing and Lorenzo de Lichtervelde for computer-assisted promoter sequence analysis. C.M., S.R., and X.L. held a FRIA fellowship (FNRS, Belgium). F.O. is supported by a fellowship from the Association pour la Recherche sur le Cancer (ARC). X.L. is supported by a postdoctoral fellowship from the Fondation pour la Recherche Medicale (FRM). Work in the F.M.R. laboratory is supported by the Agence Nationale pour la Recherche (ANR), the Fondation pour la Recherche Médicale (FRM, “Equipe labelisée”), the Association pour la Recherche contre le Cancer (ARC), the Association Française contre les Myopathies (AFM), the Ministère pour la Recherche (ACI program), and by institutional funds from CNRS and INSERM.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES