Introduction

TLX2 (also known as HOX11L1, Ncx and Enx) is a homeodomain transcription factor playing a crucial role in the development of the enteric nervous system, as confirmed by three independent Tlx2−/− mouse models displaying an intestinal phenotype ranging from pseudo-obstruction1 to megacolon with giant enteric ganglia.2, 3 The phenotype of these latter animal models strongly resembles a human congenital disorder named intestinal neuronal dysplasia (IND), that has been described for the first time by Meier-Ruge in 1971.4 Though two different types of IND have been defined (IND type B; OMIM #601223 and IND type A; OMIM #243180), IND has become synonymous with IND type B, that is the most common type and is regarded as a malformation of the enteric nervous system ganglionated plexuses characterized by hyperplastic features. On the contrary, IND type A presents in neonates as bloody diarrhea with intermittent obstructive symptoms and is characterized by severe hypoplasia of adrenergic innervation in the gut wall. Despite formal diagnostic histopathological criteria for IND have been described,5 the diagnosis of IND remains not always reliable. For this reason, IND is still a subject of controversy.6, 7 The observation of a few familial clusters8, 9 and animal models1, 2, 3 provide robust pieces of evidence that IND is a real clinical entity. In particular, in addition to Tlx2−/− mouse models, a heterozygous endothelin B receptor-deficient rat demonstrated abnormalities of the submucous plexus similar to that observed in human IND.10 However, no mutation affecting the coding region and 2 kb of 5′-flanking region of TLX2 gene have been found in a set of IND patients9, 11 and, similarly, no mutations of the EDNRB gene have been detected in a small series of IND and Hirschsprung disease (HSCR)/IND patients.12

Although a role of TLX2 in human intestinal pseudo-obstruction and/or IND seems excluded, variants of upstream regulators of the gene may account for TLX2 incorrect or absent expression during the development of the intestinal neurons, and therefore for pathological conditions resembling the knockout mice phenotypes. Recently, TLX2 has been recognized by us as a PHOX2B target, presumably mediating the PHOX2B signal in the developing peripheral nervous system.13 PHOX2B is a homeobox gene crucial for the differentiation of the autonomic nervous system, whose mutations have been detected in Congenital Central Hypoventilation syndrome,14, 15 a neurocristopathy often associated with HSCR, the most frequent intestinal innervation defect in human.

In the present work, after demonstrating that TLX2 is also a transcriptional target of the PHOX2B paralogue PHOX2A gene, we have performed a screening analysis of these PHOX2 genes in a heterogeneous set of patients affected with impairments of the intestinal innervation including pseudo-obstruction and IND.

Methods

Constructs and transfections

Cloning of TLX2 5′-regulatory region and of both PHOX2A and PHOX2B wild-type cDNA sequences has already been reported.11, 16, 17 The PHOX2A mutant construct was generated by PCR starting from the wt corresponding construct, as described previously.13 Oligonucleotides used for site-directed mutagenesis were: 5′-GCCAAGGGCGCGGCGGGCGCCAAAAAGGG-3′ and 5′-CCCTTTTTGGCGCCCGCCGCGCCCTTGGCGCTGG-3′.

SK-N-BE neuroblastoma and HEK293 cells (105) were plated on 35 mm Petri dishes 1 day before transfection. Constructs under analysis and Renilla luciferase reporter plasmid pRL-CMV (Promega), used as a transfection efficiency control, were co-transfected using Fugene 6 (Roche). In particular, 160 fmoles of the expression constructs were mixed with 40 fmoles of constructs containing TLX2 regulatory region and 20 fmoles of pRL-CMV and added to 3 μl of Fugene 6. The empty pcDNA3.1 vector was co-transfected with the promoter–reporter construct as negative control.

PHOX2A binding study

Electrophoretic mobility shift assays (EMSA) were performed using SK-N-BE and IMR32 nuclear extracts (NE), prepared as described previously.13 Six micrograms of NE were incubated with the γ32P-labeled probe 5′-GGGGAAGGTAATGTAATTCCGGCCC-3′ for 20 min at room temperature in binding buffer (hepes 20 mM pH 7.9, glycerol 20%, EDTA 1 mM, KCl 50 mM, DTT 1 mM, PMSF 0.5 mM) with 2 μg of poly (dI-dC). For supershift assays, antibodies were incubated with the NE mix for 30 min on ice before adding the specific probe. A polyclonal antibody against a peptide corresponding to a sequence in the carboxy-termini of PHOX2A was produced in chicken egg yolk.18 Chromatin immunoprecipitation assay (ChIP) was performed using formaldehyde cross-linked and sonicated chromatin from IMR32 cells as already described.18 The TLX2 cell-specific promoter region was amplified by PCR using the primers 5′-CGGGAACCAGCAGGATGGAG-3′ and 5′-GAGAAGGGAGGTGGGGAAAGAC-3′ as already reported.13

Expression analysis

Endogenous levels of PHOX2A protein has been analyzed on NE by means of Western blot experiments using the above PHOX2A polyclonal antibody. The filter has been stripped and re-probed with an anti-Sp1 antibody (Upastate Biotechnology), used as a control for the integrity of the NE. The expression of PHOX2A protein after transfection, with the corresponding expression constructs has been confirmed on total lysates from transfected cells. A β-actin antibody (Sigma) has been used as control.

Patients and controls

A total of 22 patients with both sporadic and familial IND and 4 patients affected with chronic intestinal pseudo-obstruction were analyzed. Histochemical diagnosis of the IND cases, collected from 1991 to 2004, was performed both preoperatively and intraoperatively according to the reported criteria.5 The clinical diagnosis of chronic intestinal pseudo-obstruction was made following the criteria already reported.19

Eighty individuals of Italian origin were analyzed as controls. DNA was extracted from either blood samples or cell pellets as already described.9

The present study has been approved by the Gaslini Institute ethics committee (Genova, Italy).

Mutation screening

PHOX2A coding sequence was analyzed for mutations by PCR under the conditions reported in Table 1 followed by denaturing high-performance liquid chromatography (DHPLC) analysis (Transgenomics), according to instructions of the supplier. Samples showing anomalous chromatographic profiles were sequenced by using BIG DYE v1.1 and 3130 automated sequencer (Applied Biosystems).

Table 1 PCR conditions
Full size table

The 9 nt deletion thus identified was confirmed by cloning the PCR product obtained from patient's and father's DNAs in TA-cloning vector (Invitrogen) and subsequent sequencing of several bacterial colonies.

Results

Transactivation of the human TLX2 gene by PHOX2A

Starting from previous results demonstrating the capability of PHOX2B to specifically bind two ‘ATTA’ repeats located in the TLX2 promoter, thus inducing its transcription in neural crest-derived cells,13 we have focused on the possible role of its paralogue PHOX2A in the same pathway.

Co-transfection in SK-N-BE neuroblastoma cells of the PHOX2A expression construct, able to correctly induce expression of PHOX2A protein as demonstrated in Figure 1b, together with the TLX2 promoter construct, bearing the luciferase cDNA, showed an increased gene reporter activity of approximately two fold, in comparison with the empty pcDNA3.1 expression vector (Student's t-test P<0.05). As already observed for PHOX2B, specific mutations at the identified ‘ATTA’ repeats in the promoter sequence prevented the transactivation by PHOX2A (Figure 2a). To determine whether PHOX2A and PHOX2B could have either overlapping or opposite effects on TLX2 promoter, co-transfections of both expression constructs were carried out in SK-N-BE cells. The ability of the same PHOX2B expression construct to induce levels of PHOX2B protein has previously been demonstrated by Western blot analysis.17 As displayed in Figure 2a, after double co-transfection, the reporter gene resulted approximately 5.8-fold induced compared to the empty pcDNA3.1 expression vector. This activation level was not statistically different from that obtained transfecting PHOX2A or PHOX2B constructs alone,13 as shown by the Student's t-test.

Figure 1
figure 1

Analysis of proteins expression. PHOX2A expression has been assayed by Western blot analysis on NE to assess the endogenous protein level (a) on total lysates of transfected SK-N-BE (b) and HEK 293 (c) to confirm the functionality of all expression constructs used in the study. Empty=lysate from cells transfected with empty pcDNA 3.1 expression vector; 2A=lysate from cells transfected with wild-type PHOX2A expression construct; 2A mut=lysate from cells transfected with mutant PHOX2A expression construct; 2A+2B=lysate from cells transfected with both wild-type PHOX2A and PHOX2B expression constructs.

Full size image
Figure 2
figure 2

Transactivation of the human TLX2 gene by PHOX2A. (a) Luciferase activity displayed by either wt or ‘ATTA’ mutant TLX2 promoter–reporter construct measured on lysates from SK-N-BE cells co-transfected with both the empty pcDNA 3.1 and the pcDNA 3.1-PHOX2A expression construct; this latter in particular was assayed either alone or together with the pcDNA 3.1-PHOX2B expression construct. Values represent fold induction versus activity derived from transfection of the corresponding TLX2 promoter–reporter construct alone and have been obtained by three independent experiments performed in duplicate. (b) EMSA assays performed incubating NE from IMR32 and SK-N-BE cells with radiolabeled double-stranded oligonucleotide corresponding to ‘ATTA’ repeats in the TLX2 promoter sequence in the absence (lanes 1 and 3) or in the presence of the antibody α-PHOX2A (lanes 2 and 4). (c) ChIP assays performed on chromatin derived from IMR32 cells. Primers specific for TLX2 promoter sequence bearing ‘ATTA’ repeats were used to amplify DNA from complexes immunoprecipitated with an α-PHOX2A antibody. Chicken IgY and the antibody α-PHOX2B were used as negative and positive controls, respectively. Input=fragmented DNA before immunoprecipitation.

Full size image

The direct interaction between PHOX2A and the ‘ATTA’ repeats was investigated by EMSA. Previously, the specificity of the complex formed by IMR32 and SK-N-BE NE with a probe containing the wt PHOX2 binding site had already been proven by using wt and mutated cold competitors.13 In the current experiment, direct binding at the sequence under analysis has been demonstrated following pre-incubation of PHOX2A expressing IMR32 and SK-N-BE NE (Figure 1a) with a specific polyclonal antibody α-PHOX2A (Figure 2b, lanes 2 and 4). In particular, complex A was preserved and, in addition, a prominent, slightly supershifted band was formed, as already reported for PHOX2A binding to sequences in the dopamine β-hydroxylase promoter.20

To confirm that the interaction between PHOX2A and the TLX2 promoter occurs in vivo, we performed chromatin immunoprecipitation using formaldehyde cross-linked and sonicated chromatin from IMR32 cells and the antibody used in the EMSA experiments. IMR32 cells have been used in this case since they express both PHOX2A (Figure 1a) and PHOX2B,13 a circumstance that allowed to investigate whether in vivo binding of one of these two factors can exclude binding of the other one. Results of immunoprecipitated DNA amplification, represented in Figure 2c, showed that, similarly to the known interacting PHOX2B antibody used as positive control, the PHOX2A-specific antiserum effectively immunoprecipitated the TLX2 regulatory sequence bearing the ‘ATTA’ repeats, while the control assay performed with normal chicken IgY did not present any amplified DNA.

Screening of PHOX2A and PHOX2B coding sequences in patients

Starting from the observations that (1) the PHOX2 genes are regulators of the TLX2 gene expression and (2) the TLX2 gene is involved in the development of correct intestinal innervation, we have hypothesised a role of the PHOX2 genes in enteric nervous system disorders. To this end, we have undertaken the study of the coding region of PHOX2A and PHOX2B genes, in a series of 22 patients affected with IND and 4 patients affected with intestinal pseudo-obstruction. IND patients have been recruited according to criteria already described5 which, though hotly debated,21 are those currently used for the diagnosis of IND.22

The majority of these patients had already been tested for possible nucleotide changes in the coding and promoter sequences of the TLX2 gene, without finding any alteration.9, 11

We have performed a mutation screening of the three exons spanning the entire coding sequence of the PHOX2A gene by means of PCR amplification and DHPLC analysis (Table 1). In the exon 1 of an IND patient, we have found an already reported synonymous heterozygous nucleotide substitution c.156C>T of the Leu52 codon.23 Moreover, another IND patient presented a heterozygous G>A nucleotide change in the 5′-UTR, 19 bp upstream of the first ATG codon. This has resulted to be a common polimorphism, being detected also in 3 of 30 control individuals. The screening of exon 2 did not reveal any coding variant but a common polimorphism consisting of the insertion of a C nucleotide within a poly-C stretch at the distal end of intron 1, present in 6 of our patients and 5 controls among 30 individuals. Finally, a patient affected with IND showed a heterozygous in-frame deletion of 9 nt in the third exon, leading to the deletion of three amino acids close to the C terminus of the protein. In particular, this mutation determines the lack of a duplicated Glu-Ala-Ala peptide sequence, 10 amino acids downstream of the homeodomain. Analysis of the patient's parents has revealed that the deletion was inherited from the asymptomatic father (Figure 3). This deletion has resulted absent in 160 control alleles.

Figure 3
figure 3

PHOX2A 9nt in-frame deletion. In the upper box, sequence from patient's DNA is represented. Sequences of the single alleles, obtained after PCR cloning, are provided underneath.

Full size image

A mutation screening of the three exons spanning the entire coding sequence of the PHOX2B gene has been performed by PCR and direct DNA sequencing analysis, as already reported.15 Two patients affected with IND presented an already described synonymous heterozygous nucleotide substitution c.762C>A, involving the Ala254 codon.24, 25 A contraction of seven alanines in a stretch of 20 Ala residues, already reported in a patient affected with schizophrenia25 and rarely found in controls,14 was shown in another IND patient.

Functional analysis of the mutant PHOX2A

To assess a possible role of the newly identified PHOX2A mutation in impaired intestinal motility of the corresponding carrier, we assayed the ability of this variant to induce trans-activation of the regulatory region of TLX2.

To this end, an expression construct corresponding to PHOX2A containing the observed deletion was co-transfected together with a luciferase reporter construct containing the TLX2 regulatory region, into two different cell lines: SK-N-BE neuroblastoma cells, already used as a recipient to assay the PHOX2A and PHOX2B physiological capability of TLX2 transactivation,13 and HEK293, a cell line from embryonal kidney, thus presumably lacking of the expression of neural-specific factors, including PHOX2A (Figure 1a). In both cases, the mutant protein was expressed at the same level of the wt protein (Figure 1b and c) and the luciferase level induced by co-transfecting the TLX2 promoter with the mutant construct has resulted comparable to the value obtained with the corresponding wt construct (Figure 4).

Figure 4
figure 4

Functional analysis of the mutant PHOX2A. Luciferase activity driven by the TLX2 promoter sequence was measured on lysates from both SK-N-BE and HEK 293 cells co-transfected with expression plasmids containing wt or deleted PHOX2A. Values represent fold induction versus activity derived from transfection of the pGL3 basic vector together with the corresponding pcDNA3.1 expression vector.

Full size image

Discussion

Despite TLX2 has been proposed as a candidate gene for the IND by two independent research groups,2, 3 anomalies in its coding and promoter regions have been excluded in appropriate sets of patients.9, 11 Nonetheless, the phenotype of Tlx2−/− mice suggest that missed activation of the gene could induce an impairment in the intestinal development. In this light, the identification of genes acting upstream TLX2 during the ontogenesis could provide candidates likely involved in IND and/or in other genetic defects of intestinal innervation, in addition to possible diagnostic markers. Pursuing such a goal, we have already identified PHOX2B as a specific activator of the TLX2 transcription in cells of neural origin.13

Observations on PHOX2A and PHOX2B binding identical consensus sequences and increasing the dopamine β-hydroxylase promoter activity with comparable efficiency had suggested that PHOX2A and PHOX2B may be functionally redundant.26 By contrast, the generation of two knock-in mutant mice, in which Phox2a is replaced by the Phox2b coding sequence, and vice versa, indicates that Phox2a and Phox2b are not functionally equivalent, as only Phox2b can fulfil the role of Phox2a in the structures that depend on both genes.27 Consistently, PHOX2A is expressed, like TLX2 and PHOX2B, in enteric and cranial (VII, IX, X) nerve ganglia during embryogenesis.28 The PHOX2B involvement in a gene expression cascade including TLX2 and leading to enteric neuron differentiation does not surprise considering the already known absence of autonomic innervation displayed by the Phox2b−/− mice29 and the involvement of this gene in human neurocristopathies.14, 15 On the other hand, mice Phox2a−/− fail to develop the locus coeruleus, the anterior parasympathetic ganglia and the cranial sensory ganglia.30 Moreover, a few specific mutations of PHOX2A have been found associated with familial cases of Congenital Fibrosis of the Extra Ocular Muscle type II, a human disorder giving rise to strabismus.31 Nonetheless, a wider role of PHOX2A in the development of neural-crest cells toward autonomic neuronal lineages has recently been demonstrated. In particular, while, on one hand, the transcriptional components of the cAMP pathway, CREB and CBP, has been demonstrated to induce PHOX2A transcription, on the other, the cAMP-dependent protein kinase A (PKA) has been proven to regulate the activity of a Ser/Thr PP2A-like phosphatase responsible for the PHOX2A activation. PHOX2A thus results to be central in sympatho-adrenal cell specification.32 Since inhibition of PKA in murine enteric neurons causes lethal intestinal pseudo-obstruction,33 we are tempted to hypothesize that, in this case, impaired intestinal phenotype can derive from the lack of Tlx2 expression due to missed PKA-mediated Phox2a activation. Based on these observations, we decided to test whether PHOX2A, like its paralogue PHOX2B,13 may be involved in the transcriptional regulation of the human TLX2.

Despite a previous study had already defined a DNA sequence of 20 nucleotides located in the 5′ flanking region of murine Tlx2 as sufficient to maintain tissue-specific expression of the gene, without investigating the molecular details underlying such an activity and claiming for additional specific nuclear factors to control lineage-restricted expression,34 we have more recently identified PHOX2B as one of the factors able to induce TLX2 expression by binding a conserved tandem ‘TAAT/ATTA’ enhancer element, proximal to the above murine sequence, in the human TLX2 gene.13

Following the hypothesis that PHOX2A might play a similar role, co-transfection assays and EMSA experiments have confirmed that PHOX2A specifically enhances the reporter gene transcription, through the physical interaction with the human ‘ATTA’ sequences in the TLX2 promoter, previously identified as actively involved in PHOX2B-mediated activation.13 To determine whether the two paralogue PHOX2 transcription factors could have either additive or opposite effects on TLX2 promoter, co-transfections of PHOX2A and PHOX2B together were also carried out in neuroblastoma cells. In this case, the gene reporter activity did not result statistically different from that obtained transfecting each of the two expression constructs alone, thus excluding any possible antagonistic or additive effect. ChIP assay has confirmed that the interaction between TLX2 promoter and both PHOX2A and PHOX2B occurs in vivo, suggesting that these transcription factors play a similar role in IMR32 cells. Nevertheless, our study could not rule out that (a) PHOX2A and PHOX2B play distinct physiological roles in different body district or developmental stages; (b) the two factors interact with each other; and (c) the two factors interact with different proteins.

Since we have provided evidence that PHOX2A, like PHOX2B,13 is a regulator of TLX2 cell-specific expression, these two PHOX2 genes can be reasonably considered as candidates in the search for the genetic basis of IND and other defects of enteric innervation. Indeed, starting form the observation that Phox2b−/− mice lack the entire autonomic nervous system,28 PHOX2B has already been taken into consideration as a possible candidate in HSCR pathogenesis, though no mutation could be demonstrated in patients.24 IND and chronic intestinal pseudo-obstruction are very different from HSCR, since tissues from patients with these disorders do not display absence of ganglion cells, rather they may show morphological and functional alterations of enteric neurons. This is indeed the phenotype of the Tlx2−/− mice, suggesting that enteric nervous system defects can be accounted for by the TLX2 inability to work or to be expressed. After excluding abnormalities in the coding and promoter sequence of TLX2 in patients,9, 11 we have focused on its positive developmental regulators PHOX2A and PHOX2B. Their coding sequences have been investigated, thus identifying one inherited heterozygous in-frame PHOX2A deletion in a patient characterized by vesical dysfunction and megacystis, known to be rare associated anomalies in IND.35 Despite absence in a set of control individuals suggested a possible causative role, no impaired TLX2 transactivation in SK-N-BE and HEK293 cell lines could be demonstrated. Nonetheless, a functional effect of the examined mutation on still unknown targets involved in enteric neuron differentiation processes or a possible role of such a variant in the development of associated anomalies cannot be excluded.

In conclusion, in the present work we have identified PHOX2A as a novel TLX2 regulator, thus providing a new piece of information in the complicate pathways leading to the enteric neuron differentiation. Moreover, exclusion of a major causative role of the PHOX2 genes in the development of intestinal innervation defects different from HSCR confirms that these disorders still represent a challenge from both the clinical and the genetic point of view.