Specification of the Drosophila CNS midline cell lineage: direct control of single-minded transcription by dorsal/ventral patterning genes

Abstract

The Drosophila CNS consists of a bilaterally symmetric group of neurons separated by a discrete group of CNS midline cells. The specification of the CNS midline cell lineage requires transcription of the single-minded gene. Genetic evidence suggests that a group of transcription factors, including Dorsal, Snail, Twist, and Daughterless:: Scute, is required for initial single-minded transcription. Comparison of the DNA sequences of the single-minded gene regulatory regions between two Drosophila species reveals conserved sequence elements. Biochemical studies using purified proteins indicate that a number of these conserved sequences represent binding sites for Dorsal, Snail, and Twist. In vitro mutagenesis combined with germline transformation indicates that these binding sites are required in vivo for single-minded mesectodermal transcription. These results show that single-minded transcription and, thus, CNS midline specification is directly controlled by dorsal/ventral patterning transcription factors. They also suggest a model in which multiple transcriptional activators function in a cooperative, concentration-dependent mode in combination with a transcriptional repressor to restrict single-minded transcription to the CNS midline precursor cells.

FIG. 1

Molecular genetics of sim mesectodermal transcription. Shown is a schematic cross-section of the Drosophila blastoderm embryo. Dorsal aspect is at the top. The different presumptive tissue anlage are dorsal ectoderm and amnioserosa (dea), neuroectoderm (ne), and mesoderm (mes). Specification of the mesectodermal lineage is correlated with expression of sim in the mesectodermal cells (filled). The genes shown to genetically influence initial sim transcription are shown along with their protein distributions. Dorsal positively (+) regulates sim transcription and Dorsal protein is distributed as a nuclear gradient with highest concentrations ventrally. Twi is also a positive regulator of sim and distributed as a gradient along the ventral region of the embryo. Snail is localized specifically in the mesoderm and represses (–) sim transcription. E-box binding proteins including Daughterless and Scute (Da::Sc) form an ubiquitously localized heterodimer that positively regulates sim transcription. Members of the Notch (N) signaling cell pathway positively regulate sim transcription although the relevant transcription factor has not been identified, nor is it known which cells are involved in sending the presumed signal.

FIG. 2

Sequence structure of the sim gene and transcripts. The exon-intron structure of the D. melanogaster sim gene is shown at the top. Numbered boxes correspond to the eight observed exons. Open boxes correspond to untranslated regions and closed boxes to coding sequence (exon 2 begins with a short untranslated region). The late promoter (P_L) drives embryonic midline precursor, midline glial, and muscle precursor expression (arrow points in the direction of transcription), and the early promoter (P_E), which lies within intron 1, drives initial mesectodermal and midline precursor gene expression. The letters A–E shown in exon 8 indicate five different polyadenylation sites identified in the cDNA clones shown below. Beneath the exon-intron structure diagram are the location of BamHI and EcoRI restriction enzyme sites, and the scale of the sim genomic region is shown below. Numbering of the sim gene starts at the beginning of exon 1 because the P_L start site of transcription has not been precisely defined. The extent of DNA sequence data obtained from the D. melanogaster sim gene is shown by two lines (Dm sequenced region), and the corresponding region sequenced in D. virilis is shown below (Dv::Dm compared region). Eight sim cDNA clones were isolated from embryonic cDNA libraries. Restriction map and DNA sequence analysis of these clones, including the λC1 clone previously reported, reveal the sequence structures shown. The polyadenylation site for each clone is shown using the letters (A–E) corresponding to the five different sites identified. The length in kb for each clone is indicated at right. The coding sequences of all eight clones overlap, and do not provide evidence for alternative Sim proteins. Clones S1, pC10, and pC11 correspond to P_L-derived transcripts, but no cDNA clone unambiguously corresponds to a P_E-derived transcript.

FIG. 3

Comparison of D. melanogaster and D. virilis sim early regulatory region DNA sequences. The DNA sequence of the D. melanogaster sim gene preceding and including exon 2 is shown. This region contains all sequences required for D. melanogaster mesectodermal transcription. Numbering begins at the 3′-most BamHI site (GGATCC) in intron 1 and the 5′-most G is assigned residue 10900. This is an approximate value because the entire gene has not been sequenced. Although the entire D. melanogaster sequence was determined from residue 10900, the sequence shown begins at residue 11521, because this was the extent of the D. virilis gene sequenced. Exon 2 is underlined, and the preceding 3048 bp of intron 1 is shown along with 60 bp of intron 2. Below the D. melanogaster sequence, the regions of similarity to the D. virilis sim gene are indicated. These conserved regions are referred to as smvs, and are indicated by the numbers 1 through 19. Sequence identities between the two DNA sequences are indicated by a dot and the D. virilis residue is shown when different from D. melanogaster. smv19 includes exon 2, and the sim mRNA initiator methionine is indicated as “Met” beginning at residue 14589. The 5′-positions of the P[sim-lacZ] deletion transgenes shown in Fig. 4 are indicated on the sequence as “2.8, 2.2, 1.6, and 0.9.”

FIG. 4

Sequences required for initial sim transcription are contained within a 2.8-kb region. Blastoderm or gastrulating embryos containing different P[sim-lacZ] transgenes were hybridized to a digoxygenin-labeled sim cDNA riboprobe and stained with AP-anti-digoxygenin and NBT. The P[sim-lacZ] strains were examined for the presence of mesectodermal lacZ-expressing stripes. (A) The location of the sim early regulatory region is shown below the genomic map along with the 5′ deletion fragments tested. Summary of the results indicating presence of mesectodermal lacZ transcription (MEC) is shown to the right. (B, a) Strains bearing the P[2.8sim] transgene show normal mesectodermal β-galactosidase stripes. (B, b–d) Further 5′ deletions that generated the P[2.2sim], P[1.6sim], and P[0.9sim] transgenes were tested and mesectodermal β-galactosidase expression was absent.

FIG. 5

Dot matrix comparison of the D. virilis and D. melanogaster sim early regulatory region gene sequences. D. melanogaster genomic DNA containing 3.0 kb of DNA 5′ to exon 2, exon 2, and 60 bp of intron 2 were compared to the same region of D. virilis. The D. melanogaster region contained 3307 bp of DNA whereas the D. virilis DNA was considerably larger, containing 4978 bp of DNA. Dot matrix analysis was carried out using GeneWorks with the window set at 30 and stringency at 49%. Overall, the two genes are conserved throughout the region, and the conserved sequences align in a linear fashion. The regions of significant sequence identity (smvs) number 19 (including exon 2), as indicated along side each diagonal of sequence conservation. The differences in size between the D. melanogaster and D. virilis sim early regulatory regions are due to the comparatively large size in D. virilis of three AT-rich regions found within this region (labeled “vat1–3” for D. virilis and “matl–4” for D. melanogaster that has an additional AT-rich region). The scale in kb is indicated along each axis. The D. melanogaster sequence is numbered according to Fig. 3 and the D. virilis sequence is numbered arbitrarily in kb with the 3′ end of the compared region labeled as “0.”

FIG. 6

Functional conservation of the D. virilis sim early regulatory region. The stretch of the D. virilis sim gene corresponding to the 2.8-kb D. melanogaster sim early regulatory region was sequenced and functionally assayed. At the top is a representation of the D. virilis sim gene showing the location of exon 2 (box with coding sequence filled and 5′ UTR unfilled) and the approximate location of PE. Below are the locations of the D. virilis smvs and the region sequenced. The D. virilis 4.6-kb NsiI-NcoI fragment was fused to lacZ in the CaSpeR-AUG-βgal P-element vector, introduced into D. melanogaster germline DNA, and assayed for mesectodermal transcription by hybridization to a lacZ RNA probe. The stage 7 embryo shows strong mesectodermal lacZ stripes. Ventral view is shown; anterior is to the left.

FIG. 7

Gel shift analysis identifies fragments of the sim early regulatory DNA region that contain Dorsal and Twi protein binding sites. (A) The 2.8-kb sim genomic DNA region was fractionated into 11 fragments (labeled A–K) that cover the entire interval and include smvs 1–19. Also shown is the position of the 5′ end of the 2.2-kb sim fragment fused to lacZ in P[2.2sim]. The region between 2.8 and 2.2 is required for initial sim transcription. (B) Purified baculoviral-produced Dorsal protein was incubated with ³²P-labeled fragments A–K (probe) and subjected to gel shift analysis. Fragments B and F show retarded DNA fragments. (Free) indicates unshifted DNA fragments. (C) Specificity of Dorsal binding is indicated by competition experiments. ³²P-labeled fragments B and F were incubated with Dorsal protein in either the absence (–) of competitor DNA, presence of 100 times molar excess of wild-type (wt) Dorsal high-affinity binding site oligonucleotide, or 100 times molar excess of Dorsal binding site oligonucleotide mutated (mut) within the binding site. (D) ³²P-labeled fragments A–K were incubated with GST-Twi protein and subjected to gel shift analysis. All fragments showed weak retardation although fragments B, F, and I showed the strongest binding. Binding was not observed when GST-Twi was absent (probe B is shown) or when only GST was added (data not shown).

FIG. 8

DNAseI footprint analysis reveals that Dorsal and Twi bind to sites in smv2 and 3. (A) sim fragment B that contains smv2 and 3 was ³²P-labeled on the antisense strand, incubated with protein, and subjected to DNAsel footprint analysis. The fragment was incubated with: lane 1, no recombinant protein; lane 2, 0.5 μg GST-Twi; lane 3, 1 μg GST-Twi; lane 4, 2 pg GST-Twi; and lane 5, 2 pg GST. GST-Twi protected (hatched lines) a region corresponding to smv2a that includes Twi binding site T1. Also protected was a region containing smv2b that contains the Sna E-box SE1 binding site that could be a binding site for Twi. Nucleotide positions within the sim early regulatory region are indicated to the right. (B) Fragment B was labeled with 32P on the sense strand, incubated with protein, and subjected to DNAsel footprint analysis. Lane 1 is a G/A ladder. The fragment was incubated with: lane 2, 2 pg GST; lane 3, 0.5 pg GST-Twi; lane 4, 1 pg GST-Twi; lane 5, 2 pg GST-Twi; lane 6, 2 pg GST-Twi + 5 μl Dorsal (Dl); lane 7, 5 μl Dorsal; and lane 8, no protein. Dorsal and GST-Twi failed to strongly protect any residues when used individually. However, when combined they strongly protected the Dorsal and Twi D1, D2, T2, and T3 sites in smv3.

FIG. 9

Deletion of smvs containing Dorsal, Twi, and Sna E-box binding sites results in loss of mesectodermal transcription. Mutations in the 2.8-kb sim early regulatory region were created, cloned into a P[lacZ] vector, and tested for mesectodermal lacZ transcription after introduction into germline DNA. The constructs tested are shown beneath a schematic of the 2.8-kb regulatory region with numbered smvs. Each construct listed with the smv deleted (D) is indicated by an “X” through the relevant number. P[2.8simΔSΔ16] deletes Sna sites S4, S5 (S), and smv 16. P[simB] is a construct with sim gene fragment B cloned into an enhancer tester P[lacZ] vector. P[simB2x16] contains fragment B and 2 copies of the SE6 and SE7 Sna binding sites. Presence of strong blastoderm mesectodermal (MEC) lacZ transcription is indicated to the right by (+), weak expression by (+/–), and the absence of expression by (–). The binding sites deleted in each construct are listed to the right. The bracketed sites listed for P[simB] and P[simB2x16] indicate binding sites that are included in the construct. Representative embryos for each construct are shown at the bottom.

FIG. 10

Summary of sim early regulatory region protein binding sites. Shown is 3.3 kb of the D. melanogaster sim early regulatory region. Exon 2 is indicated by a box and P_E by an arrow. The location of D/V transcription factor binding sites is indicated along with the CMEs The locations of the three major clusters of D/V regulatory protein bindings sites (DTC1, DTC2, and SC) are shown at the bottom.