GCM (glial cells missing)
GCM1 and GCM2 (glial cells missing 1 and 2) are the only human members of a small metazoan family that binds DNA through the GCM domain, a conserved 150-residue region with a fold unlike any other transcription factor class: two subdomains held together by structural zinc ions, reading the major groove with a five-stranded beta-sheet rather than an alpha-helix. They recognize an 8-bp GCM box, AT(G/A)CGGGT, as monomers, with no half-sites or spacing rules. Despite the name, the mammalian proteins are not gliogenic: GCM1 is required for placental labyrinth branching and trophoblast fusion, and GCM2 for parathyroid gland development, where loss-of-function and activating mutations cause hypoparathyroidism and hyperparathyroidism respectively.
The factors
GCM (glial cells missing) proteins are a small metazoan family with two human members, GCM1 (also GCMa) and GCM2 (GCMb); Drosophila has gcm and gcm2. They share a highly conserved 150-residue N-terminal region, the GCM domain, which carries all of the sequence-specific DNA-binding activity and sits next to a bipartite nuclear localization signal; the transactivation function maps to the divergent C-terminal half 9114061,12682016. There is no dimerization module — the GCM domain binds its site as a monomer — and no relationship to any other DNA-binding class 12682016. Expression is narrow. GCM1 is expressed in placental trophoblast, with lower levels in kidney and thymus; GCM2 is expressed in the developing parathyroid and, at lower levels, in a few brain regions 10888880,10910362,15353544. Despite the name, the mammalian proteins have no established role in gliogenesis: the glia-versus-neuron switch is the Drosophila function, and it was the fly gene that gave the family its name 15353544. With only two paralogues and highly restricted expression, GCM is one of the most self-contained families in this atlas.
The motif
The GCM box is an 8-mer, AT(G/A)CGGGT, determined by binding-site selection for Drosophila GCM and reproduced for both mammalian proteins 8962155,9114061,15353544. The models here agree closely: JASPAR, CIS-BP and the Vierstra archetypes all give ATGCGGGT, and family cohesion is 0.92. Note that two of the HOCOMOCO entries (GCM1.H14CORE.0.S.B, GCM2.H14CORE.0.SM.B) report the consensus as ACCCGCAT, which is the reverse complement of the same element — an orientation convention, not a different specificity. One short CIS-BP profile (M02020_2.00, TGCGGG) is simply a trimmed version of the same core.
The structural story is what makes this family unusual. The crystal structure of the murine GCMa GCM domain bound to its octameric site, at 2.85 A, revealed a fold with no precedent: two subdomains tethered together by one of two structural zinc ions, and a five-stranded beta-sheet that protrudes into the major groove perpendicular to the DNA axis 12682016. Almost every well-known DNA-binding class reads the major groove with an alpha-helix; GCM does it with a beta-sheet, and the zinc ions are structural rather than part of a classical zinc finger. That is why the recognized sequence looks like nothing else: the GC-rich CGGGT half is read by beta-strand side chains in a geometry not available to helical readers 12682016. Because the element is a single 8-mer with high information content and no half-site structure, there are no spacing rules to worry about and no composite arrangements to score. GCM1 and GCM2 are not distinguishable from sequence.
Where it acts
GCM boxes act in promoters and proximal enhancers of a small number of tissue-restricted genes. The best-documented GCM1 targets are the syncytin-1 (ERVWE1) promoter in human trophoblast, and Itga4, Rb1 and syncytin A in mouse placenta 12397062,18167345; GCM1 activity is regulated by cAMP/PKA signaling in trophoblast 16166624. GCM2 acts in parathyroid precursor cells 10910362. Because both factors are so restricted, the great majority of sequence matches genome-wide are unoccupied in any given cell type. An exact 8-mer of this composition is expected tens of thousands of times in the human genome, and GC-rich promoter sequence produces additional weak partial matches, so the motif is informative only when paired with the right cell type. The GCM box is not part of any named composite element.
Biology and disease
In Drosophila, gcm is a binary switch: cells that express it become glia, cells that do not become neurons 7553844,7553843. The mammalian phenotypes are entirely different. *Gcm1*-null mouse embryos fail to initiate chorioallantoic branching morphogenesis and die from a defective placental labyrinth 10888880; the fusion of trophoblast into syncytium depends on GCM1-driven syncytin expression 10888880,12397062. *Gcm2*-null mice have no parathyroid glands yet are viable with normal serum parathyroid hormone, which was first attributed to the thymus 10910362; later lineage and thymectomy work showed that the thymus is not an auxiliary source of serum PTH, and that thymic PTH comes mainly from small misplaced parathyroid clusters 21203493. In humans, loss-of-function GCM2 mutations, including substitutions that disrupt DNA binding by the GCM domain, cause autosomal recessive familial isolated hypoparathyroidism 15863676, while activating GCM2 variants cause familial isolated hyperparathyroidism 27745835 — a clean case of a transcription factor with disease alleles in both directions.
Reading a hit
A GCM box match is one of the more trustworthy family-level calls in this atlas. The fold is unique, the element is an 8-mer with high information content, and no other human family in this collection reads it: across the eight member motifs, similarity to the nearest non-GCM motif (0.67-0.80) sits well below within-family similarity. The evidence is also good — the HOCOMOCO entries rest on in vitro selection (codes S and SM), not on ChIP alone, so the claim "these proteins bind this sequence" is directly supported rather than inferred from co-occupancy.
What the match does not license is a call on which paralogue, or on occupancy. GCM1 and GCM2 are indistinguishable by motif; the discriminator is tissue, and outside trophoblast and parathyroid neither is expressed at a level that would make a site functional. Watch for two practical traps: databases report this motif in both orientations (ATGCGGGT and ACCCGCAT), so a scan that only checks one strand will miss half the sites; and permissive thresholds on GC-rich promoter sequence will produce partial matches to the CGGGT half that have little to do with GCM.
How this was worked out
Two groups independently show that Drosophila gcm is a binary switch: cells that express it become glia, cells that lack it become neurons.
Still open then: How GCM acted molecularly was unknown; its sequence resembled no known DNA-binding protein.
GCM is shown to be a sequence-specific DNA-binding protein with activity in its N-terminal 181 residues, binding (A/G)CCCGCAT; eleven such sites lie upstream of the glial gene repo, and human and mouse homologs share the conserved gcm motif.
Binding-site selection gives AT(G/A)CGGGT, the same element on the other strand, and maps DNA binding to an N-terminal domain beside a bipartite NLS and transactivation to the C-terminal half.
Mouse knockouts move the family away from glia: Gcm1-null embryos lack the placental labyrinth and syncytiotrophoblast, and Gcm2-null mice lack parathyroid glands yet keep normal serum PTH, which was attributed to the thymus.
Still open then: The thymic source of PTH was inferred from expression and ablation and later reinterpreted.
GCMa activates the human syncytin (ERVWE1) gene through two GCM boxes upstream of the HERV-W 5' LTR, linking the motif to trophoblast fusion.
The 2.85 A crystal structure of the murine GCMa GCM domain on its octamer reveals a new fold: two subdomains joined by a structural zinc ion, with a five-stranded beta-sheet inserted into the major groove.
Lineage and ablation experiments show that the thymus is not an auxiliary source of serum PTH: thymic PTH comes mainly from small misplaced parathyroid clusters, and the PTH made by medullary thymic epithelium is Gcm2-independent, revising the 2000 interpretation.
Recurrent gain-of-function GCM2 variants in a C-terminal inhibitory domain cause familial isolated hyperparathyroidism, giving GCM2 disease alleles in both directions alongside the recessive hypoparathyroidism mutations.
Still unresolved
- Whether GCM proteins use paired sites is unresolved: the structure and biochemistry show a monomer on one 8-mer, but the Vierstra archetype AC0550 is a 14-bp palindrome of two inverted GCM boxes (TGCGGGTACCCGCA), and it is not known whether this reflects cooperative binding without a dimerization module or only the avidity of two monomers in selection assays.
- Whether the low-level expression of mammalian GCM genes in brain has any function is open: the Drosophila glial switch has no established mammalian counterpart, and the knockout phenotypes characterized so far are confined to placenta and parathyroid.
The GCM box is a high-information 8-mer read by a structurally unique domain, so family-level assignment is reliable — nothing else in this atlas reads ATGCGGGT, and the HOCOMOCO models rest on in vitro selection rather than ChIP alone. GCM1 and GCM2 cannot be told apart from sequence; use tissue instead, since GCM1 is essentially trophoblast-restricted and GCM2 parathyroid-restricted, so most matches are unoccupied in most cell types. Two mechanical points: the element is reported in both orientations across databases (ATGCGGGT and its reverse complement ACCCGCAT), and permissive scoring on GC-rich promoter sequence yields partial matches to the CGGGT half that are not GCM sites.
References
- Schreiber J, Sock E, Wegner M. The regulator of early gliogenesis glial cells missing is a transcription factor with a novel type of DNA-binding domain. Proc Natl Acad Sci U S A 1997;94:4739-44. PMID 9114061
- Cohen SX, Moulin M, Hashemolhosseini S, et al. Structure of the GCM domain-DNA complex: a DNA-binding domain with a novel fold and mode of target site recognition. EMBO J 2003;22:1835-45. PMID 12682016
- Hosoya T, Takizawa K, Nitta K, et al. glial cells missing: a binary switch between neuronal and glial determination in Drosophila. Cell 1995;82:1025-36. PMID 7553844
- Jones BW, Fetter RD, Tear G, et al. glial cells missing: a genetic switch that controls glial versus neuronal fate. Cell 1995;82:1013-23. PMID 7553843
- Anson-Cartwright L, Dawson K, Holmyard D, et al. The glial cells missing-1 protein is essential for branching morphogenesis in the chorioallantoic placenta. Nat Genet 2000;25:311-4. PMID 10888880
- Günther T, Chen ZF, Kim J, et al. Genetic ablation of parathyroid glands reveals another source of parathyroid hormone. Nature 2000;406:199-203. PMID 10910362
- Yu C, Shen K, Lin M, et al. GCMa regulates the syncytin-mediated trophoblastic fusion. J Biol Chem 2002;277:50062-8. PMID 12397062
- Schubert SW, Lamoureux N, Kilian K, et al. Identification of integrin-alpha4, Rb1, and syncytin a as murine placental target genes of the transcription factor GCMa/Gcm1. J Biol Chem 2008;283:5460-5. PMID 18167345
- Baumber L, Tufarelli C, Patel S, et al. Identification of a novel mutation disrupting the DNA binding activity of GCM2 in autosomal recessive familial isolated hypoparathyroidism. J Med Genet 2005;42:443-8. PMID 15863676
- Guan B, Welch JM, Sapp JC, et al. GCM2-Activating Mutations in Familial Isolated Hyperparathyroidism. Am J Hum Genet 2016;99:1034-1044. PMID 27745835
- Hashemolhosseini S, Wegner M. Impacts of a new transcription factor family: mammalian GCM proteins in health and disease. J Cell Biol 2004;166:765-8. PMID 15353544
- Akiyama Y, Hosoya T, Poole AM, et al. The gcm-motif: a novel DNA-binding motif conserved in Drosophila and mammals. Proc Natl Acad Sci U S A 1996;93:14912-6. PMID 8962155
- Chang CW, Chuang HC, Yu C, et al. Stimulation of GCMa transcriptional activity by cyclic AMP/protein kinase A signaling is attributed to CBP-mediated acetylation of GCMa. Mol Cell Biol 2005;25:8401-14. PMID 16166624
- Liu Z, Farley A, Chen L, et al. Thymus-associated parathyroid hormone has two cellular origins with distinct endocrine and immunological functions. PLoS Genet 2010;6:e1001251. PMID 21203493
8 motifs in this family — JASPAR 2 · HOCOMOCO 2 · CIS-BP 2 · Vierstra 2
| motif | database | consensus | bits |
|---|---|---|---|
| AC0550:GCM:GCM | Vierstra | TgCgGGTaCCcGCA | 15.6 |
| GCM2.H14CORE.0.SM.BSMB | HOCOMOCO | ACCCGCAT | 13.2 |
| GCM1 (MA0646.2) | JASPAR | ATGCGGGTAC | 12.9 |
| GCM1.H14CORE.0.S.BSB | HOCOMOCO | ACCCGCAT | 12.9 |
| M04881_2.00 | CIS-BP | ATGCGGGT | 11.5 |
| GCM2 (MA0767.2) | JASPAR | ATGCGGGT | 10.8 |
| AC0549:GCM:GCM | Vierstra | ATGCGGGT | 9.6 |
| M02020_2.00 | CIS-BP | TgcGGG | 5.4 |
HOCOMOCO tags read as assay + quality grade: P ChIP-seq, S HT-SELEX, M methyl-HT-SELEX, G GHT-SELEX, I SMiLE-seq, B PBM; grade A means two or more assay types agreed. A P-only matrix is a record of where the protein was crosslinked, which need not be a sequence it binds itself.
The factors, elsewhere
These entries are written from the primary literature and every citation is checked against PubMed, but this one can still be out of date, too confident, or simply wrong — especially where a motif’s published site and its database matrices disagree. Corrections and additions are welcome, and they are the fastest way to make this page better for everyone.