Domain mapping on the human metastasis regulator protein h-Prune reveals a C-terminal dimerization domain

Abstract

The human orthologue of the Drosophila prune protein (h-Prune) is an interaction partner and regulator of the metastasis suppressor protein NM23-H1 (non-metastatic protein 23). Studies on a cellular breast-cancer model showed that inhibition of the cAMP-specific PDE (phosphodiesterase) activity of h-Prune lowered the incidence of metastasis formation, suggesting that inhibition of h-Prune could be a therapeutic approach towards metastatic tumours. H-Prune shows no sequence similarity with known mammalian PDEs, but instead appears to belong to the DHH (Asp-His-His) superfamily of phosphoesterases. In order to investigate the structure and molecular function of h-Prune, we expressed recombinant h-Prune in a bacterial system. Through sequence analysis and limited proteolysis, we identified domain boundaries and a potential coiled-coil region in a C-terminal cortexillin homology domain. We found that this C-terminal domain mediated h-Prune homodimerization, as well as its interaction with NM23-H1. The PDE catalytic domain of h-Prune was mapped to the N-terminus and shown to be active, even when present in a monomeric form. Our findings indicate that h-Prune is composed of two independent active sites and two interaction sites for the assembly of oligomeric signalling complexes.

Figure 1. Domain architecture of h-Prune and protein fragments used in this study

(a) Domain arrangement in h-Prune. A well-defined DHH domain is followed by a less well-conserved DHHA2 domain and a C-terminal CHD with a putative coiled-coil region and a proline-rich region. The constructs of h-Prune and its fragments which were expressed and purified in this study are indicated. (b) Alignment of h-Prune (370–453) with the cortexillin I coiled-coil region. The best-aligned region of h-Prune covers the predicted coiled-coil in the N-terminal section of the CHD subdomain, and a secondary structure prediction for h-Prune is indicated on top of its sequence. (c) Limited proteolysis of full-length h-Prune at 4 °C and 25 °C respectively. The protease used is detailed on top, and the fragments which were further analysed are indicated on the right-hand side.

Figure 2. Mapping of dimerization and catalytic domains

(a) Behaviour of full-length h-Prune and of h-Prune (1–393) in size-exclusion experiments. A logarithmic graph of a molecular-mass standards (MW) and the elution volumes of these proteins is shown. At the elution positions of the two h-Prune constructs, labels indicate the molecular masses for a monomeric and a dimeric species respectively. (b) Comparison of specific activities of full-length h-Prune (□) and the monomeric h-Prune (1–393) construct (▲). Shown is a double reciprocal plot used for the determination of K_m and V_max values [fit for full-length h-Prune, solid line; h-Prune (1–393), dotted line]. Both proteins show almost identical activity despite their different oligomeric states.

Figure 3. Interaction of h-Prune with NM23-H1

Pull down experiments were performed using His-tagged and phosphorylated NM23-H1 and either full-length h-Prune or h-Prune (1–393). (a) Lane 1, molecular-mass standards. Lanes 2–5, pull down with full-length h-Prune: lane 2, flow-through; lane 3, first wash; lane 4, second wash; lane 5, elution. Lanes 6–9, pull down with h-Prune (1–393); lane allocation is the same as for lanes 2–5. (b) As a control for non-specific interactions, NM23-H1 was incubated with BSA. Lane 1, molecular-mass standards; lane 2, flow-through; lane 3, wash; lane 4, elution. (c) Pull down experiment with tagged full-length h-Prune (lanes 1–5). GSK3β binds h-Prune in the presence of excess NM23-H1, indicating a non-competitive interaction. Lanes 6–9: control, showing that there is no non-specific binding of GSK3β to the resin. Lane 1, molecular-mass standards; lanes 2 and 6, flow-through; lanes 3 and 7, first wash; lanes 4 and 8, second wash; lanes 5 and 9, elution.

Figure 4. Model for the architecture of h-Prune-containing complexes

Homodimers of h-Prune are created, containing two separate and independent catalytic cores formed by the DHH and DHHA2 subdomains (blue). They are held together through interactions with the C-terminal CHD (green), which consists of a putative coiled-coil region (left), a proline-rich region (centre), and a helical extension (right). The two CHDs of the dimer also enable the binding of at least two partner molecules, such as NM23-H1 (yellow) and GSK3β (red), which can lead to the assembly of higher-order oligomeric structures. The various parts of h-Prune, as well as its interaction partners, are not drawn to scale, and the exact binding sites for GSK3β and NM23-H1 within the CHD are not yet known.