A role for heterodimerization in nuclear localization of a homeodomain protein

Abstract

The A mating type genes of the mushroom Coprinus cinereus encode two families of dissimilar homeodomain proteins (HD1 and HD2). The proteins heterodimerize when mating cells fuse to generate a transcriptional regulator that promotes expression of genes required for early steps in sexual development. In previous work we showed that heterodimerization brings together different functional domains of the HD1 and HD2 proteins; a potential activation domain at the C terminus of the HD1 protein and an essential HD2 DNA-binding motif. Two predicted nuclear localization signals (NLS) are present in the HD1 protein but none are in the HD2 protein. We deleted each NLS separately from an HD1 protein and showed that one (NLS1) is essential for normal heterodimer function. Fusion of the NLS sequences to the C terminus of an HD2 protein compensated for their deletion from the HD1 protein partner and permitted the two modified proteins to form a functional transcriptional regulator. The nuclear targeting properties of the A protein NLS sequences were demonstrated by fusing the region that encodes them to the bacterial uidA (beta-glucuronidase) gene and showing that beta-glucuronidase expression localized to the nuclei of onion epidermal cells. These observations lead to the proposal that heterodimerization regulates entry of the active transcription factor complex to the nucleus.

Figure 1

Effect of deleting or adding predicted NLSs on b1–1–b2–3 heterodimer function. (A) Organization of the A42 and A6 mating type loci of C. cinereus. A42 and A6 contain representative members of three pairs of functionally redundant genes (a, b, and d), each pair encoding an HD1 and an HD2 protein. For simplicity, a nonfunctional pseudogene in the A42 locus is not shown. The diagonal arrow indicates one of two possible gene combinations that activate A- regulated development because their encoded proteins can heterodimerize when present in the same cell. (B) Wild-type and manipulated genes from which the NLSs were deleted (b1–1) or added (b2–3) were introduced into A6 and A42 host cells by DNA-mediated transformation and transformants were examined for signs of A- regulated development. unfused clamp cell indicates A-regulated development observed; simple septum indicates no A-regulated development observed. Predicted structural features of the HD1 and HD2 proteins are indicated by different motifs defined in the key.

Figure 2

Predicted NLSs are essential for b1–1–b2–3 heterodimer function and are protein-unspecific. (A) Organization of the partial null A locus of strain NAB1 that lacks both members of the b gene pair and in which the function of the b1–1–b2–3 heterodimer was assayed. (B) b2–3 and b1–1 were introduced by transformation into the NAB host and transformants screened for the development of clamp connections. Predicted proteins encoded by wild-type b1–1 and b2–3 (a), b1–1 ΔNLS and wild-type b2–3 (b), and b1–1 ΔNLS and modified b2–3 with added NLSs (c). (Bar = 10 μm.)

Figure 3

Subcellular localization of GUS activity directed by sequences from the A mating type proteins of C. cinereus. (A) Constructs introduced into onion cells. Coding regions were expressed under the control of the cauliflower mosaic 35S promoter and nopaline synthase terminator. (a) Control in which GUS was fused to the plant potyviral Nla protein that contains a bipartite NLS. (b) Control with GUS sequence alone. (c) GUS fused to amino acids 327–499 from b1–1. (d) GUS fused to b2–3. (B) GUS activity (a, c, e, and g) and 4′,6-diamidino-2-phenylindole-stained nuclei (b, d, f, and h) in transiently transformed onion epidermal cells. GUS activity localized to the nucleus by the Nla protein (a and b), GUS alone localized to the cytoplasm, (c and d) GUS activity localized to the nucleus by the predicted b1–1 NLSs (e and f), and GUS activity localized to the cytoplasm when fused to b2–3 (g and h). (Bar = 100 μm.)