A number of transcription factors that contain zinc finger or homeodomain motifs are known to have evolutionary conserved functions in the spatial and temporal patterning of embryonic development. TheDosophilazfh1 (zfh1) encodes a protein with nine Cys2His2-type zinc finger motifs and a homeodomain that shows a complex pattern of expression in embryonic mesoderm and nervous system (Lai et al.1991). During ontogeny,zfh1 is first detected in the presumptive procephalic mesoderm and subsequently in the mesoderm anlagen during early gastrulation. In later developmental stages,zfh1 is expressed in a number of mesoderm-derived structures including the heart (dorsal vessel), support cells of the gonads, and segment-specific arrays of adult muscle precursors.zfh1 expression can also be detected in the majority of identified motor neurons of the developing central nervous system (CNS) (Lai et al.1991). A phenotypic analysis ofzfh1 mutant embryos reveals that this gene is required for the proper differentiation of a number of mesodermal tissues including the heart, muscle precursor (Lai et al.1993), and gonadal tissues (Broihier et al.1998) and for the segregation of muscle precursors (Lai et al.1993). Loss ofzfh1 activity also disruptsDrosophila visceral development with abnormal midgut constrictions (Lai et al.1993; Broihier et al.1998). In the heart, the loss ofzfh1 function results in a disrupted heart tube morphology, with the selective elimination of a subset of pericardial cells: the even-skipped positive pericardial cells (EPCs) (Su et al.1999). A recent report shows that, in the absence of EPCs, larval and adult heart functions are severely compromised (Fujioka et al.2005).

Recent genetic and evolutionary studies have suggested that the assembly of vertebrate organs, such as the heart, may be divided into functional and morphological units (Fishman and Olson1997). The underlying genetic mechanisms for each of these developmental modules may be evolutionarily conserved (Fishman and Olson1997). In an effort to search for a vertebrate homologue ofDrosophilazfh1 and to determine if the cardiac function is conserved within this gene family, we screened a Lamda Zap mouse heart expression library (Stratagene) using the conserved C-terminal zinc-finger regions ofzfh1. We isolatedδEF1-related transcription factors that we initially calledMzfh-1. A sequence analysis ofMzfh-1 revealed that it is identical to a previously described 5.3-kb transcript of Smad -interacting protein 1 (SIP1) in mouse (Verschueren et al.1999).SIP1 was initially identified by binding to Smad proteins mediated by the Smad-binding domain ofSIP1 (Verschueren et al.1999).SIP1 andδEF1 are two known vertebrate family members of the ZFHX1 family transcription factors. BothSIP1 andδEF1 show structural similarity with shared DNA binding sequence specificity to the E2-box-like motif CACCT(G) in vitro (Sekido et al.1994; Postigo and Dean1997; Remacle et al.1999; Verschueren et al.1999; Comijn et al.2001; van Grunsven et al.2001). However, the developmental functions of these two genes are dissimilar. While homozygote null mutantδEF1 mice exhibited severe T-cell deficiency of the thymus and skeletal defects of various lineages (Higashi et al.1997; Takagi et al.1998),SIP1-null mice die in utero around embryonic day (E) 9.5 due to heart dysfunction and with the embryo is unable to turn during development (Van de Putte et al.2003). In addition, the mutation ofSIP1 has also been associated with Hirschsprung diseases in humans (Cacheux et al.2001; Wakamatsu et al.2001; Yamada et al.2001; Zweier et al.2002; Garavelli et al.2003; Mowat et al.2003; Sztriha et al.2003a,b; Wilson et al.2003). The Hirschsprung disease is characterized by the absence of enteric neurons in the gastrointestinal tract and results in intestinal obstruction (Robertson et al.1997; Swenson2002). It is a complex disease often associated with other congenital abnormalities such as microcephaly, mental retardation, and heart disease (NIM 235730). A detailed analysis ofSIP1 expression patterns may help to reveal other potential developmental functions and regulation ofSIP1. In the present study, we have identified three other variants ofSIP1 in addition to the previously described 5.3-kbSIP1 transcripts. All fourSIP1 variants show differential temporal embryonic expression. The embryonic expression patterns ofSIP1 were studied in detail by in situ hybridization. The potential conserved genetic function ofSIP1 in heart development was further tested inDrosophila via transgene technology. Our results demonstrate the ability of the isolated mouse 5.3-kbSIP1 to restore the heart phenotype in theDrosophilazfh1 mutant embryos. These data suggest an evolutionarily conserved function of thezfh1 gene family in metazoan heart development.

In situ hybridization of tissue sections

BALB/c and C57BL/6×DAB/2 embryos, fetuses, and postnatal brains were fixed and embedded as described in Lyons et al. (1990). In brief, the embryos were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight, followed by dehydration and infiltration with paraffin. Serial sections (5–7 μm) were mounted onto gelatinized slides, deparaffinized in xylene, rehydrated, and treated with triethanolamine/acetic anhydride before probing with different cRNA probes. To distinguish among the ZFHX1 multigene family, we used antisense probes to the 3′ or 5′ untranslated regions of the mRNAs that are not conserved among the different genes. cDNAs were subcloned into bluescript vectors and their transcripts were synthesized according to the manufacturer’s conditions (Stratagene) and labeled with³⁵S-UTP (>1,000 Ci/mmol; Amersham). The cRNA transcripts larger than 1,200 nucleotides were subjected to alkaline hydrolysis to give a mean size of 100 bases for efficient hybridization. The following probes were used: (a) the 3′ end regions ofSIP1 (1,400 nucleotides that include 110 nucleotides of exon 9 and the entire exon 10) were linearized withHindIII and transcribed with T7 polymerase to produce an antisense probe. This probe should detect expression of all four transcripts; (b) the 5′ end region ofSIP1 was linerarized withEcoR1 and transcribed with T7 polymerase, yielding the 318 nucleotide antisense probe (264 nucleotides from exon 1 and 54 nucleotides from exon 2). Both of the 5.3- and 8.3-kb transcripts could be detected by this fragment; and (c) TheδEF1 3′ UTR region was linearized withEcoR1 and transcribed with T7 polymerase, producing the 1,800 nucleotide anti-sense probe. Hybridization and post-hybridization washes were carried out according to standard protocols (Wilkinson1998). The slides were dehydrated, dipped in Kodak NTB-2 emulsion, and exposed to BIOMAXMR X-ray films (Amersham) for 1–2 weeks at 4°C. The slides were developed in Kodak D-19, counterstained lightly with toluidine blue, and analyzed using both light and dark field illumination. The background signals only were detected with³⁵S-labeled sense control probes.

Northern blot analysis

Mouse multiple-tissue and developmental Northern blot membranes (Clontech, Palo Alto, CA, USA) were hybridized with the radiolabel probe E to determine tissue distribution and the presence of different isoforms ofSIP1. Seven different α[³²P]dCTP-radiolabel polymerase chain reaction (PCR) fragments, corresponding to the different regions ofSIP1, were used as probes to determine how the four isoforms ofSIP1 may be generated. TheSIP1-region-specific probes were generated via PCR using the following primer sequences:

The locations of the differentSIP1 probes are described in Fig. 2b. All PCR products are gel-purified using a gel extraction kit (Qiagen) and labeled with [α-³²P]dCTP using random priming methods.

Tissue blots were pre-hybridized for 16 h at 45°C in 50% formamide–5X Denhardt’ solution—1× SSPE [0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM ethylenediamine tetraacetic acid (pH 8.0)]–0.5% sodium dodecyl sulfate (SDS)–10 μg/ml of denatured salmon sperm DNA before hybridization with 200 ng random-primed [α-³²P]-labeled cDNA probes for 16 h at 45°C. Autoradiography was performed after three washes for 15 min each at 60 to 65°C in 0.2× SSC (1× SSC is 0.15 M NaCl and 0.015 M sodium citrate)–0.1% SDS.

cDNA library screening

Degenerate primers to the C-terminal zinc finger regions of chickenδEF1 were used to PCR-amplify a 240-bp fragment, which was used as [α-³²P]dCTP-labeled probe to screen an adult heart λZAPII (Stratagene) cDNA library. A total of 900,000 plaques from these libraries were screened using standard protocol. The sequences for the degenerate primers were: forward primer, 5′-AA(A/G)GCCTCCAA(A/G)CACAA(A/G)CAC-3′ and reverse primer, 5′-GCA(A/G)TAGGA(G/A)TA(G/C)CG(A/G)TG(A/G)TT-3′.

Fly stocks

We used a strongzfh-1² null allele stock (provided by Z. C. Lai) that lacks pericardial cells (Su et al.1999). Thetwi-GAL4 line (Greig and Akam1993) and 24B-GAL4 line (Brand and Perrimon1993) were combined to provide pan-mesodermal expression from stage 9 or 10 to stage 16 (Lockwood and Bodmer2002). A full-length 5.3-kbSIP1 cDNA was directionally cloned into theEcoR1 andXho1 sites ofDrosophila expression vector UAS promoter (PUAST) (Brand and Perrimon1993) and PUAST-SIP1 DNA was injected into fly embryos along with delta 2, 3 transposase. A full-lengthDrosophilazfh1 cDNA, a gift from Dr. Lai, was directionally into theNot1 andKpn1 sit of PUAST vector and was similarly injected into fly embryos along with transposase. Several stable fly transformation lines were obtained using standard germline transformation methods. A 24B-GAL4:zfh-1² line was made by standard meiotic recombination. For UAS/GAL4-induced ectopic expression, the embryos were collected at 29°C. In all experiments, we used a balancer chromosome carrying a P-LacZ transgene, P{TM3-ftz-lacZ}, to distinguish homozygous mutant embryos from the balancer-bearing siblings.

Whole-mount antibody staining

Antibody staining was performed using either horse radish peroxidase, in conjunction with biotinylated secondary antibodies in the Elite kit (Vector), or alkaline phosphatase with directly conjugated secondary antibodies. The embryos were fixed by gently shaking for 30 min in 5 ml of heptane and 5 ml of 0.5 M PBS with 0.25 ml of 37% formaldehyde. They were devitellinized and X-gal stained to identify the homozygouszfh-1² null mutants as previously described (Su et al.1999). After dehydration, staining with primary antibody and biotinylated secondary antibodies was conducted as described in Eldon and Pirrotta (1991). A double primary antibody staining procedure was carried out as in Patel (1994). The following antibodies were used in this study: rabbit anti-eve antibodies (1:3,000) (Frasch et al.1987), rabbit anti-Mef-2 antibodies (Lilly et al.1995), and mouse monoclonal anti-PC antibodies (Yarnitzky and Volk1995).

Characterization ofSIP1 variants in mice

In an attempt to isolate mammalianzfh1-like proteins, we used degenerate primers to amplify the C-terminal zinc fingers ofzfh1 and screened a mouse heart cDNA library. Two independent phage clones were isolated: one of them corresponded to the mouseδEF1 while the other corresponds to the previously described Smad interacting protein 1 (SIP1) in mouse (Verschueren et al.1999). To further characterize theSIP1 gene, we performed a Northern blot analysis using the 3′ endSIP1-sequence-specific region to determine its expression patterns in adult mouse tissue (Fig. 1a). Two different transcripts with molecular sizes of 8.3 and 5.3 kb were observed. Although the 5.3-kb transcript was prominently expressed in most of the adult mouse tissues including heart, brain, spleen, lung, skeletal muscles, kidney and testis, the longer 8.3-kb transcript was expressed strongly in adult mouse brain and lung tissues. A similar tissue expression of these two transcripts was also observed when the same 3′ end region of humanSIP1 sequences was used as a probe in the Northern blot analysis of adult human tissues (Fig. 1b). In addition, the 5.3-kbSIP1 transcripts were also detected in human pancreatic and placental tissues (Fig. 1b).

Expression patterns of differentSIP1 transcripts. Expression ofSIP1 in adult tissues.a Two differentSIP1 transcripts, 8.3 and 5.3 kb, are indicated witharrows. The C-terminal region ofSIP1 was used as a probe in Northern hybridization and 2 μg of poly A⁺ tissue RNA are loaded in each gel lane.b SimilarSIP1 transcripts are detected in multiple human tissues.c Dynamic expression of four differentSIP1 can be detected at various ontogenic stages of mouse embryos as indicated witharrows.d Structural scheme ofSIP1 and differentSIP1 probes (probeA–G) used in Northern hybridization and in situ hybridization.e FourSIP1 transcripts contain different coding regions as detected by Northern hybridization with variousSIP1 probes. Although the C-terminal region (probeF) detected all fourSIP1 transcripts, probeG (additional 3′ untranslated region) detected only the larger 8.3 kb transcripts. TheSIP1 DNA binding regions appear to be absent in the two smallerSIP1 transcripts as they were not detected by any of the probes containingSIP1 DNA binding regions

Full size image

We further analyze the expression of theseSIP1 variants in different stages of developing mouse embryos using the 3′ end region ofSIP1 (Fig. 1c). In addition to the 8.3- and 5.3-kbSIP1 transcripts, two smallerSIP1 transcripts, 2.7 and 1.9 kb, were observed in developing mouse embryos (Fig. 1c). TheseSIP1 transcripts were observed as early as 7 days post-coitum (d.p.c.) and were persistently expressed in 11-, 15-, and 17-day embryos (Fig. 1c). To determine how these differentSIP1 transcripts may be generated, we used different regions ofSIP1 as probes for Northern analyses in developing mouse embryos (Fig. 1d). The two smaller transcripts (2.7 and 1.9 kb) were not detectable by Northern blot analysis using probes corresponding to N-terminal zinc fingers, C-terminal zinc fingers, or homeodomain regions (Fig. 1d,e). In contrast, positive signals were observed for the two larger 8.3- and 5.3-kbSIP1 transcripts when these DNA binding regions were used as probes in Northern analyses (Fig. 1c,e). These data suggested that only the two larger transcripts (8.3 and 5.3 Kb) containSIP1 DNA binding regions (Fig. 1). After scrutinizing the existingSIP1 genomic and cDNA sequences, we noted an additional poly-A site (AATAAA) located at the 1.4-kb 3′ end downstream from the first poly-A additional site. To investigate if this poly-A site was used to generate any of theSIP1 transcripts, we used sequences adjacent to this poly-A site (Fig. 1d, probe G) as our probe for Northern blot analyses and found that only the larger 8.3- kbSIP1 transcript contains this sequence (Fig. 1c). These data suggested that the 8.3-kbSIP1 transcripts may be partially generated by alternative transcription termination/polyadenylation. Alternative polyadenylation has been observed forδEF1 ((Funahashi et al.1993).

Ubiquitous neural expression ofSIP1 transcripts in mouse tissue

SIP1 transcript expression was determined using the shared common 3′ end region which allows detection of allSIP1 transcripts in developing mouse embryos (Fig. 1c, Probe F). TheSIP1 transcripts were abundantly expressed in most embryonic mouse tissues, particularly in the developing nervous system.SIP1 were expressed in neural tube, notochord (Fig. 2a), and telencephalon (data not shown) at stage 9.5 d.p.c., and their persistent high expression was observed in all neuronal regions of the developing mouse telencephalon, as well as in other regions of neuroepithelium, neural tube, trigeminal ganglia, and dorsal root ganglia of stage 10.5 d.p.c. mouse embryos (Fig. 2b,c). The expression ofSIP1 was mostly observed in the ependymal layer of the neural tube and the ventricular zones of the brain, both regions of dividing neurons (Fig. 2d). By stage 14.5 d.p.c.,SIP1 expression was found in the cortex, striatum, thalamus, choroids plexus of the brain, and dorsal region of the spinal cord (Fig. 2f,g). In addition, theseSIP1 transcripts were also detected in the cerebella primordium, trigeminal nucleus, inferior olive, and acoustic ganglion of the cochlea of the inner ear (data not shown). TheSIP1 brain expression patterns were also observed in the neonatal mouse brain (data not shown).

SIP1 expressions in developing mouse embryos. The embryonic expression patterns ofSIP1 were determined using the C-terminal regions ofSIP1 (probeF).a Embryonic day (E) 9.5 transverse sections showing low-levelSIP1 in the neural tube (n) and notochord (thin arrow).b E10.5 day sagittal section showing strong expression ofSIP1 in neural tissues, dorsal root ganglia (small arrows), trigeminal ganglia (arrowheads), and myotomes (open arrow).c E11.5 day oblique sections showingSIP1 expression in other non-neural tissues and dorsal root ganglia (small arrow).d Sagittal section of E12.5 day embryo showing expression ofSIP1 in the ependymal layer of the neural tube and the ventricular zones of the brain, both regions of dividing neurons, and dorsal root ganglia (small arrow).e Transverse section of E12.5 days embryo showingSIP1 expression in umbilical cord (U) and blood vessels and weak positive expression ofSIP1 in developing endocardium (open arrow) in the heart.f Frontal section of E14.5 day embryo through the brain, eyes, and nasal sinuses showingSIP1 expression in the cortex (C), the mesenchyme around the whisker follicles (w) and retina (r).g E14.5 day midsagittal section showingSIP1 ubiquitous expression in different developing tissues including spinal cords (n), gut (g), and blood vessels (arrow).a, surrounding aorta;c, cortex;e, esophagus;f, frontonasal mass;g, gut;h, heart;I, limbud;j, jaw;t, telencephalon;m, mandibular arches;n, neural tube;p, lung;u, blood vessels of the umbilicus;x, liver;y, hindbrain;z, lung primordium.Scale bars, 800 μm. The tissue expression patterns ofSIP1 and δEF1 in developing mouse embryos are compared (h,i).SIP1 shows more robust expression thanδEF1 in developing trigeminal ganglia and dorsal root ganglia (arrows) and in the developing blood vessels (e,h). The expression ofδEF1 andSIP1 are overlapping in the CNS, PNS, and skeletal muscle (myotomes;open arrow) but not at the other sites whereSIP1 is expressed. A persistent high expression ofSIP1 is observed in neural tissues (h) whereδEF1 expression in the CNS and PNS has been reduced (i).n, neural tube;t, telecephalon;y, hindbrain. Dorsal root ganglia (small arrow) and myotomes (open arrow) (h,i).Scale bars, 800 μm

Full size image

The expression ofSIP1 occurred in other non-neural tissues including the heart, developing blood vessels (Fig. 2b), lung primodium (Fig. 2c), muscles (Fig. 2d,g), kidney, gonads (Fig. 2g), and yolk sac (Fig. 2a). In the heart,SIP1 was not detected in the ventricular cardiac myocytes but was evident in the endocardium of the developing heart values (Fig. 2e) and in the blood vessels surrounding the aorta (Fig. 2b,e), adjacent to the umbilical cord (Fig. 2e). TheSIP1 transcripts were highly expressed in the developing myotomes (Fig. 2b), skeletal muscle, and visceral smooth muscle (Fig. 2g). The expression ofSIP1 in liver tissue appears to be developmentally regulated such that it was expressed during early embryonic stages (E10.5 and E11.5 d.p.c.; Fig. 2c) but quickly diminished to a low/undetectable level by day 12.5 d.p.c. (Fig. 2d) and remained at undetectable levels in adult liver (Fig. 1a). In the frontal facial region of the developing mouse, theSIP1 transcripts were strongly expressed in the frontonasal mass, muscles of the jaw, nasal sinuses, mesenchyme around the whisker follicles, and eyes (Fig. 2f).

Comparison ofSIP1 andδEF1 embryonic expression patterns

SIP1 andδEF1 belong to a protein family ZFHX and both play important roles in normal mouse development. A null mutation ofδEF1 results in T-cell and skeletal patterning defects (Higashi et al.1997; Takagi et al.1998), whereas a mutation ofSIP1 produces Hirschsprung disease (Cacheux et al.2001; Wakamatsu et al.2001; Yamada et al.2001; Zweier et al.2002; Garavelli et al.2003; Mowat et al.2003; Sztriha et al.2003a,b; Van de Putte et al.2003; Wilson et al.2003). We compared the embryonic expression patterns ofSIP1 andδEF1 and found that both show overlapping expression patterns in the central nervous system (CNS), peripheral nervous system (PNS), and skeletal muscle tissues (Fig. 2h,i) (Takagi et al.1998). The expression patterns forSIP1 andδEF1 are drastically divergent within other internal organs. WhileδEF1 is prominently expressed in the developing limb buds (Takagi et al.1998),SIP1 showed no expression in the limb buds (Fig. 2b). The expression ofSIP1 can be detected in the heart, blood vessels (Fig. 2e), lung, distal region of the jaw, and frontonasal mass (Fig. 2g,h), butδEF1 expression was undetected in these regions (Fig. 2i) (Takagi et al.1998).

Sequence divergence amongSIP1 family members

The putative orthologousSIP1 genes from human, mouse, frog, fish, and fruit fly all exhibit a similar structural arrangement consisting of an N-terminal zinc-finger cluster (four zinc fingers), a C-terminal zinc-finger cluster (three zinc fingers), and an intervening Lim-like homeodomain (Fig. 3a). In addition, a single zinc finger is situated immediately upstream of the homeodomain. This single zinc finger occurs inSIP1 andDrosophilazfh1 but not in theδEF1 family member. Both N-terminal and C-terminal zinc finger clusters show high degrees of amino acid sequence conservation (Table 1) among different ZFHX family members. The C-terminal zinc fingers (78–100%) show a higher degree of sequence conservation than the N-terminal zinc fingers do (59–100%) and the homeodomain is the least conserved (35–98%) (Table 1) among these different ZFHX family members. In contrast, the single zinc finger, adjacent to the homeodomain, can only be found inDrosophilazfh1 andSIP1 family members. This additional zinc finger exhibits a higher degree of amino acid sequence divergence than the C-terminal zinc fingers (54–100%) with the conserved cysteins that are required to form the zinc-finger motif.

Sequence and structural comparison of ZFHX family members.a Structural organization ofSIP1,δEF1, andzfh1. All three ZFHX family members share similar protein structure with two terminal zinc-finger clusters, enclosing a center Lim-like homeodomain region. An additional single zinc finger, indicated with thearrow, can only be found inSIP1 andzfh1.b Amino acid neighbor-joining tree for ZFHX family members in which theDrosophila sequences were employed as outgroups. The amino acids for the full-length protein sequence were aligned using MEGA and a distance matrix was generated using the Kimura two-parameter models. The bootstrap percentages from 2,000 replications are shown on interior nodes

Full size image

Evolutionary relationships amongSIP1 family members

Phylogenetic analyses using available amino sequences of variousSIP1-like proteins from a variety of vertebrates and fromDrosophila were conducted (Fig. 3b). Note thatDrosophilazfh1 was sister to an exclusively vertebrate crown clade in whichSIP1 andδEF1 formed two reciprocally monophyletic and phylogenetically robust (BS=100) sister lineages.

Functional conservation ofSIP1 andzfh1 in fly heart development

To test the hypothesis of an evolutionarily conserved function betweenSIP1 and the flyzfh1 in development, we produced transgenic flies expressing the full-length 5.3-kbSIP1 cDNA in a vector containing the UAS promoter to be used in the Gal4 system (Brand and Perrimon1993). The heart-associated phenotype is well characterized inzfh1 null mutant embryos (Lai et al.1993; Su et al.1999) and we could evaluate the degree of functional conservation amongSIP1 andzfh1 by assessing the potency of mammalianSIP1 in restoring the heart phenotype ofzfh1 null mutant embryos when expressed in fly embryos. Transgenic flies containing the yeast Gal4, under the control of a pan-mesoderm promoter derived fromTwist (Greig and Akam1993), were used to express the 5.3-kbSIP1 transcripts in the mesoderm region of the flyzfh1 null mutant embryos. As shown in Fig. 4,zfh1 is required for the proper differentiation of a subset of pericardial cells, the Eve-positive pericardial cells (EPC) (Su et al.1999). Thezfh1 null mutant embryos exhibited two notable heart abnormalities. The EPC cells stained by anti-Eve antibodies were selectively absent from the linear heart tube inzfh1 null mutant embryos (Fig. 4b), whereas the contractile cardiac myocytes, distinguished byMef-2, were not affected in their cell numbers but showed an abnormal-alignment phenotype typical of cardiac morphogenesis defects (Qian et al.2005a,b). The distortion of the heart tubes ofzfh1 null mutant embryos is also demonstrated by the PC, pan-pericardial cell marker (Fig. 4e). The missing EPC and the heart tube misalignment phenotypes inzfh1 null mutant embryos were significantly completely restored when transgenic flies expressed the full-length mammalianSIP1 in mesoderm using the pan-mesodermal Gal 4 driver,Twist-24B (Fig. 4c,f,i). The total number and the location of EPCs inSIP1-rescuedzfh1 mutants were similar to the wild-type control embryos (Fig. 4a,c). Moreover, the heart tube morphology, as shown by anti-pericardin (PC) and anti-Mef-2 antibody staining, was indistinguishable in the wild type. These data strongly suggest that mammalianSIP1 can fulfill the function ofzfh1 inDrosophila cardiogenesis.

Conserved cardiac function ofSIP1 inDrosophila heart development. Stage 16Drosophila embryos were examined using pericardial cell markers (Eve,a–c;PC,d–f) and a myocardial cell marker (Mef-2,g–i). Wild-typeDrosophila embryos (a,d,g),zfh1 null mutant embryos (b,e,h), and embryos from Mzfh1 (SIP1) transgenic flies expressing the full-length 5.3 kbSIP1 cDNA with the mesodermal Twi-GAL4 and 24B-GAL4 drivers in azfh1 null mutant background (Twi;24B>Mzfh1;zfh1) (c,f,i). The cardiac phenotype ofzfh1 null mutant embryos can be rescued by using embryos from transgenic flies expressing full-lengthDrosophilazfh1 cDNA with pan-mesoderm Twi-GAL4 and 24B-GAL4 drivers inzfh1 null mutant background (Twi;24B>Dzfh1;zfh1) (panels 7j,k,l). The missing EPC cells inzfh1 is now present (j) and the heart tube is not distorted (k)

Full size image

Organogenesis is a complex, multiple-step process that often requires the coordinated function of multiple genes. Recent advancements in genetic and evolutionary studies have suggested that the assembly of functional organs, such as the heart, can be divided into both morphological and functional units (Fishman and Olson1997). The underlying genetic pathways for each of these modules may be conserved among diverse metazoan lineages, e.g., insects and vertebrates. It has been indeed suggested that the organogenesis of vertebrates can be viewed as the sum product of each of these separate genetic units. This present study provides evidence on how one member of the ZFHX gene family may have conserved evolutionarily functions in cardiogenesis.

Our initial screen for a mammalian homologue ofDrosophila zfh1 has enabled us to isolate mouseSIP1. Although aSIP1 transcript has been previously reported (Verschueren et al.1999; Bassez et al.2004), our study shows that additionalSIP1 transcripts are prominently expressed in adult and developing mouse embryos. Two differentSIP1 transcripts (8.3 and 5.3 kb) were observed in most adult tissues with varying degrees of expression levels. TheSIP1 transcripts were present in very early embryos and they were abundantly expressed in developing neuronal tissues. These data are consistent with previously described expression patterns forSIP1 (Cacheux et al.2001; Yamada et al.2001; Bassez et al.2004). In addition, two smallerSIP1 transcripts (2.7 and 1.9 kb) are observed only in developing embryos. AdditionalSIP1 transcripts have also been observed in mouse tissues (Bassez et al.2004); however, our data indicated that these two smallerSIP1 transcripts are developmentally regulated and can be detected only in developing embryos despite prolonged Northern blot exposure (2 weeks at −70°C; data not shown). This is in contrast to a previous study (Bassez et al.2004) where smallerSIP1 transcripts (3.5, 2.2, and 1.6 kb) were detected in adult tissues when oligonucleotide (60 mers) probes were used. We also preformed Northern hybridizations using RNAs, prepared from adult tissues of different mouse inbred lines (BABL/c; DB-1), and failed to detect these smaller transcripts using ourSIP1 probe (Fig. 2, probe F), suggesting that our divergent results, vis-à-vis the results of Bassez et al.2004), are unlikely due to tissues sources obtained from different mouse strains. Probe specificity and stringency differences may underlay these discrepancies; however, the presence of differentSIP1 transcripts in the developing mouse embryos emphasizes the inherent complexities associated with definingSIP1 mutations.

To further define these differentSIP1 transcripts, we used various regions ofSIP1, spanning the entire length of the open reading frame forSIP1, as probes for Northern hybridizations (Fig. 1d). Our results show that only the two larger transcripts appear to contain DNA binding regions such as homeodomain, N-terminal, and C-terminal zinc fingers (Fig. 1e), whereas the two smaller transcripts (2.7 and 1.9 kb) appear not to contain any of the DNA binding regions. Because we can detect all fourSIP1 transcripts with the 3′ end probe while the downstream probe of the potential polyadenylation site hybridizes with the largestSIP1 transcripts (8.3 kb), this fact suggests that this larger transcript may be generated in part by utilizing different 3′ polyadenylation sites (probe G in Fig. 1e). We currently do not know precisely how differentSIP1 mRNA species may be formed from their respective genomic sequences; they may be generated from alternative splicing of different exons, usage of alternative promoters and/or different transcription terminators/polyadenylation. Although the precise functions of theseSIP1 variants are currently unknown, alternative splicing for different mRNA and protein isoforms are important in regulating both physiological and biological gene functions (Stamm et al.2000; Xu et al.2002; Black2003; Stamm et al.2005). The present targetedSIP1 mutation (Higashi et al.2002; Van de Putte et al.2003) as well as those described for humanSIP1 mutations (Wakamatsu et al.2001; Yamada et al.2001; Zweier et al.2002; Garavelli et al.2003; Mowat et al.2003; Sztriha et al.2003a,b) likely disrupt both of the largerSIP1 transcripts. Deciphering the functions of individualSIP1 variants may be a key to understand the developmental roles ofSIP1 in development and also inSIP1-mediated disease manifestation.

In situ hybridizations demonstrated the early and prominent expression ofSIP1 in developing neural tissues, particularly in the dividing neurons, which is in accordance with the previously described pattern of expression ofSIP1 in early mammalian development (Espinosa-Parrilla et al.2002; Bassez et al.2004). The pan-neuronal patterns ofSIP1 expression in developing embryo implicate a potentially important role forSIP1 in the control of diverse neuronal cell functions. This could explain the observed microcephaly phenotype in which the brain fails to grow normally in patients carrying aSIP1 mutation (Mowat et al.2003). The expression ofSIP1 in neural crest derivatives, heart, and frontal facial region of the developing mouse also implicates roles forSIP1 in the proper development of these diverse tissues. Because the cephalic neural crest cells can give rise to facial primodia (Francis-West et al.1998), endocardiac cushions (Kirby and Waldo1995), and enteric neurons (Gershon1998; Burns et al.2000; Burns and Le Douarin2001), the severe migration defects displayed by cranial neural crest cells inSIP1 knockout mice (Van de Putte et al.2003) may play a causative role in the observed facial dysmorphism and cardiac- and Hirsprung-like syndrome displayed in human withSIP1 mutations. The TGFβ family members have been implicated in the development of neural-crest-derived cells into facial primodia (Francis-West et al.1998) and the differentiation of neural crest cells into sympathetic, adrenal, or enteric neurons (Shah et al.1996). The interaction betweenSIP1 and cytoplasmic SMADs (SMAD1, 2, 3, and 5) in mammalian cells may account for the observed phenotypes inSIP1 mutants (Amiel et al.2001; Cacheux et al.2001; Yamada et al.2001; Zweier et al.2002; Mowat et al.2003; Sztriha et al.2003a,b; Wilson et al.2003; Espinosa-Parrilla et al.2004).

We also comparedSIP1 andδEF1 embryonic expression patterns to investigate their different developmental functions. TheδEF1-deletion mouse mutants exhibit T-cell deficiency and skeletal defects (Higashi et al.1997; Takagi et al.1998) but no obvious nervous system anomalies. The presence ofSIP1 in the developing nervous system might plausibly have compensated the loss ofδEF1 activity in these tissues because these two proteins share a high degree of amino acid identity in their DNA binding regions (Fig. 3a). The diverged expression pattern forSIP1 andδEF1 in other internal organs may account for their different embryonic mutant phenotypes.

Our phylogenetic gene tree topology (Fig. 3b) was consistent with an evolutionary divergence ofSIP1 andδEF1 within the deuterostome/vertebrate lineage.Drosophilazfh1 may have evolved from a separate lineage of an ancestral gene; however, ‘SIP1’-like sequences from diverse invertebrate taxa, including echinoderms and chordates, are required to flesh out the metazoan evolutionary history of these proteins and to establish ifSIP1 andδEF1 are indeed vertebrate evolutionary innovations.

The prominent and early expression ofSIP1 in mammalian nervous system and its ubiquitous expression in mesodermal derivatives are similar to those expression patterns observed in theDrosophilazfh1. The deletion ofzfh1 inDrosophila results in a disrupted heart morphology with missing Eve-positive pericardial cells and, in parallel, congenital cardiac defects have also been observed in 9 out of the 21 patients withSIP1 mutation (Wilson et al.2003). In contrast, heart defects have not been observed in association withδEF1 deletion (Takagi et al.1998). These data suggest thatSIP1 may be a functional equivalent toDrosophilazfh1, a hypothesis that received significant support by our experimental demonstration that expression ofSIP1 inzfh1 fly mutant embryos can completely restore the ability to form eve-positive pericardial cells. We presently do not know ifδEF1 can also substitutezfh1 heart function, and it will be interesting for future studies to test if the cardiac function is indeed conserved among all Zfh family members. The genetic pathways underlying cardiac development is thought to have been conserved as the origin of the common ancestor of insects and vertebrates (Bodmer and Venkatesh1998). Members of the Nkx2.5 family, GATA factors, myogenic enhancing factors (Mef), and T-box family have been previously shown to play key roles in the conserved genetic pathways that regulateDrosophila and mammalian development (see reviews in Bodmer1993,1995; Fishman and Olson1997; Olson and Schneider2003; Miskolczi-McCallum et al.2005; Qian et al.2005a,b; Reim et al.2005; Stennard and Harvey2005). Our present data further underscores the cardiac function of Zfh family members in this conserved genetic pathway in heart development and the potential utility of using fruit flies to dissect the developmental functions ofSIP1.