Structural Ties between Cholesterol Transport and Morphogen Signaling

The molecular details of how cholesterol exits lysosomes and is integrated into cellular and endoplasmic reticulum membranes remain unclear. Two proteins implicated in this exit process, the 13-transmembrane transporter NPC1 and secreted NPC2, are known to be mutated in Niemann-Pick type C (NPC) disease in humans, characterized by cholesterol accumulation. A recent X-ray crystallographic study in Cell (Kwon et al., 2009) proposes an ingenious “hand-off” mechanism between the cholesterol-scavenging NPC2 protein and the membrane-bound NPC1. An immunoglobulin-like fold in NPC2 opens to “swallow” a cholesterol molecule; the new finding is that the extracellular N-terminal domain of NPC1 has a helical structure that can also sequester cholesterol, but in an orientation opposite to that of NPC2. Neatly, a pocket-topocket interaction between cholesterolloaded NPC2 and NPC1 may move the cholesterol to the N-terminal domain, which is only a short pivot away from the transporter core of NPC1, enabling subsequent integration into the lysosomal membrane (Kwon et al., 2009). Using fold-recognition methods, we earlier predicted that the N-terminal cysteine-rich domain of Hedgehoginteracting protein 1 (HHIP1) has a Frizzled domain (FZD) fold (Bosanac et al., 2009). We have now used HHpred (Söding, 2005) to search the human proteome with a sequence profile derived from the HHIP1 alignment and have detected a distant but compelling sequence relationship to cysteine-rich modules of NPC1 and folate receptors (Supplemental Data available online); by transitivity, these latter domains should also share the FZD fold of HHIP1. These findings prompted us to make a detailed structural comparison between the cholesterol-bound N-terminal domain of human NPC1 (Kwon et al., 2009), the chicken riboflavinbinding protein docked to riboflavin (RBP; a paralog of folate receptors) (Monaco, 1997), and the FZD module of the mouse Frz8 (Frizzled) receptor (Dann et al., 2001). The Dali program for superimposing protein structures (Holm et al., 2008) reveals a highly significant structural homology between the similar length chains of NPC1 and RBP (224 and 212 amino acids, respectively; Z score = 7.6, rmsd = 3.7 Å over 135 aligned residues with 20% identity). However, the shorter 121 residue Frz8 structure is a more distant match to NPC1 (Z score = 4.1, rmsd = 3.4 Å for 96 aligned residues at 8% identity) and RBP (Z score = 2.0, rmsd = 4.2 Å with 83 matched residues at 12% identity). The shared core domain comprises four helices (Figure 1B, helices B–E) in a splayed-open bundle; RBP has replaced the kinked B helices of NPC1 and Frz8 with a hairpin loop. The overall low degree of sequence identity is offset by the striking concordance of cysteine residues that stitch together