The gcm-motif: a novel DNA-binding motif conserved in Drosophila and mammals

Abstract

In the Drosophila nervous system, the glial cells missing gene (gcm) is transiently expressed in glial precursors to switch their fate from the neuronal default to glia. It encodes a novel 504-amino acid protein with a nuclear localization signal. We report here that the GCM protein is a novel DNA-binding protein and that its DNA-binding activity is localized in the N-terminal 181 amino acids. It binds with high specificity to the nucleotide sequence, (A/G)CCCGCAT, which is a novel sequence among known targets of DNA-binding proteins. Eleven such GCM-binding sequences are found in the 5' upstream region of the repo gene, whose expression in early glial cells is dependent on gcm. This suggests that the GCM protein is a transcriptional regulator directly controlling repo. We have also identified homologous genes from human and mouse whose products share a highly conserved N-terminal region with Drosophila GCM. At least one of these was shown to have DNA-binding activity similar to that of GCM. By comparing the deduced amino acid sequences of these gene products, we were able to define the "gcm motif," an evolutionarily conserved motif with DNA-binding activity. By PCR amplification, we obtained evidence for the existence of additional gcm-motif genes in mouse as well as in Drosophila. The gcm-motif, therefore, forms a family of novel DNA-binding proteins, and may function in various aspects of cell fate determination.

Figure 1

The N-terminal region of GCM binds to DNA. (A) GCM fusion protein constructs tested for DNA-binding activity. Lanes 2–5 show a series of truncated GCM-maltose binding protein (MBP) fusions. The bottom part of the figure shows a schematic drawing of intact GCM. A nuclear localization signal (NLS) and the region rich in basic residues is represented by solid and hatched boxes, respectively. Vertical bars indicate the nine cysteine residues. The numbers above each construct indicate corresponding residues of GCM. MBP-lacZα was used as a control. DNA-binding activity is shown in the right column. (B) Gel-shift assays with GCM fusion proteins. A 100-bp genomic fragment A of repo (see Fig. 3A) was electrophoresed with one of the GCM fusion proteins. Lane numbers correspond to those shown in the left column of A. Two constructs, N243 and N181, showed DNA-binding activity, while C261 and M63 did not. DNA-protein complexes (B) and free probes (F) are indicated.

Figure 3

Eleven GCM-binding sites are present in the region upstream of the repo gene. (A) The GCM-binding sites. Positions of the GCM-binding sequence are indicated by solid ellipses (–11). The consensus binding sequences are clustered in the proximal upstream 4 kb, whereas none were found between the −4 kb to −7-kb region. The initiation site and the orientation of the repo gene are indicated by a large arrow, according to Halter et al. (12). Xiong et al. (11) reported that one more exon is located at −4 kb in this figure, but we could not find such sequences within the −7-kb region. Thick bars and thin bars respectively show fragments that bound or did not bind to GCM in gel-shift assays. A and B indicate the probes used in the gel-shift assays shown in Figs. 1 and 2. E, EcoRI; S, SalI; and X, XhoI. (B) Alignment of the GCM-binding sequences in the upstream of repo. The numbers correspond to those in A. Nucleotides identical to the consensus sequence are highlighted. Three sequences that are in opposite orientation to the others are indicated by “r.”

Figure 2

GCM specifically binds to an octamer sequence. (A) The GCM consensus-binding sequence. Forty-eight of the clones selected from a pool of random oligonucleotides on the basis of affinity to N243 were sequenced and aligned. Percentage values are the frequency of each base at each position. The most frequently present bases are highlighted and the derived consensus sequence is shown below. (B) Competition assays. Assays were performed using a 200-bp labeled DNA fragment B (see Fig. 3A) with protein N243 in lanes 2–10, and MBP-lacZα in lane 1 as a control. Unlabeled competitors were added to the reaction mixture in equal concentration to the labeled probe or in 10-, 100-, or 1000-fold molar excess. Absence of the competitor is indicated by minus signs. Addition of competitor a, which contains the consensus sequence, abolishes DNA-protein complex. Addition of competitor b, which lacks the consensus sequence, did not show competitive activity. Free probes (F) and DNA-protein complexes (B) are indicated.

Figure 4

Comparison of Drosophila and mammalian gcm-motif genes. (A) Sequence alignment of human GCMa (hGCMa), mouse GCMa (mGCMa), mouse GCMb (mGCMb), and Drosophila GCM (GCM). Amino acid residues shared by these four gene products are highlighted and the gcm-motif is shown below. Similar residues are shown by asterisks. The conserved seven cysteine and four histidine residues are indicated by #. Three absolutely conserved stretches of 9 or 10 amino acids are underlined (A–C). Gaps are indicated by dashes. (B) Sequence alignment of PCR products. mGCMa, a2, and b are amplified by RT-PCR, using mouse adult head poly(A)⁺ RNA as a template. dGCM2 is a product of PCR using Drosophila genomic DNA as template. Those PCR products have about 70% nucleotide identity to GCM.