Introduction

Elongation of glycan chains is often terminated by sialyltransferases. The terminal sialic acids of glycoproteins and gangliosides are sometimes further modified by O-acetylation, frequently at the 9-position [1]. O-Acetylation of sialic acids is remarkably tissue specific and developmentally regulated in a variety of biological systems. In fertilized eggs, ectopic expression of sialic acid-specific 9-O-acetylesterase consistently arrests development at the two-cell stage, suggesting that the O-acetylated sialoglycan chains may be involved in segmentation of the embryos [2]. In embryos, the late expression of 9-O-acetylesterase in specific organs causes morphological abnormalities [2]. O-Acetylation of ganglioside GD3 suppresses the proapoptotic activity of mitochondrial GD3 [3]. Despite its importance, however, the O-acetylation mechanism is poorly understood at the molecular and gene levels.

Acetylation of sialic acid occurs in the lumen of the Golgi apparatus (Fig. 1) [4]. The acetate donor, acetyl-CoA (Ac-CoA), is synthesized in the cytosol. Therefore, Ac-CoA must be transported into the lumen of the Golgi apparatus/endoplasmic reticulum to drive the acetylation reaction [4]. The recent cloning of the gene encoding the Ac-CoA transporter protein provides new clues for understanding the roles of Ac-CoA transport and metabolism during development and differentiation.

Fig. 1.
figure 1

Roles of ACATN (acetyl-CoA transporter). Gangliosides GD3 and GT3 are synthesized in the lumen of the Golgi apparatus via a reaction catalyzed by the sialyltransferase STSia8-I. Terminal sialic acids are further modified by acetylation at the 9-position. The addition of an acetyl group to sialic acid prevents subsequent sialylation. GD3 NeuAcα2–8NeuAcα2–3Galβ1–4GlcβCer; GT3 NeuAcα2–8NeuAcα2–8NeuAcα2–3Galβ1–4GlcβCer

Full size image

Identification of an Ac-CoA transporter

In 1997, an Ac-CoA transporter cDNA (AT-1) was isolated using the expression cloning method and COS-1 and HeLa cells possessing the precursor gangliosides GD3 and GT3 [5]. Expression of human Ac-CoA transporter in HeLa cells induces production of 9-O-acetylated GD3 and GT3 [4]. Rat and mouse Ac-CoA transporter cDNAs were subsequently cloned [6, 7] and the genomic structure and promoter regions of the mouse Ac-CoA transporter gene (slc33a1) were elucidated [8].

Because the transporter protein is tightly membrane bound and is unstable, the homogeneous protein has never been purified. Nonetheless, transmembrane hidden Markov model (TMHMM) analysis (www.cbs.dtu.dk/services/TMHMM-2.0/) [9] predicts that the amino acid sequence of human ACATN protein contains approximately 11 transmembrane-spanning domains. Mammalian ACATN proteins also contain a leucine-zipper motif. Immunofluorescence analysis shows that the 54-kDa protein localizes to the cytoplasm, presumably in endoplasmic reticulum. Incorporation assay of radiolabeled Ac-CoA into ganglioside fraction using semi-intact cells of stable ACATN transfectants indicates that the ACATN protein functions as an Ac-CoA transporter (Fig. 1).

Two major transcripts (3.3 kb and 4.3 kb) are present in all tissues (Table 1). In mouse, a single 3.0-kb transcript is detected [7] and ubiquitously expressed in adult tissues, including brain. The expression of Acatn is developmentally regulated; increased expression occurs during early embryonic stages.

Table 1. SLC33 acetyl-CoA transporter family
Full size table

The human and mouse genes have been mapped to chromosome bands 3q21.1 and 3E1-E3, respectively [10].

Phylogenetic features

Mouse and rat Acatn cDNAs are highly homologous to human ACATN cDNA, exhibiting 87% and 86% similarity, respectively, in their amino acid sequences. Homology searches of currently available nucleotide and protein databases for human ACATN identify three homologous proteins: (1) a putative transmembrane protein from Saccharomyces cerevisia (EMBL, accession no. Z36088) that has 560 amino acids and 31% shared identity; (2) a protein from Caenorhabditis elegans T26C5.3 (EMBL, accession no. Z50859) that has 632 amino acids and 48% shared identity; and (3) a protein from Drosophila melanogaster that has 525 amino acids and 48% shared identity. In the nematode C. elegans, two alternative spliced forms of the Acatn gene (T26C5.3a and T26C5.3b) are predicted. Escherichia coli and Haemophilus influenzae also express a homologous protein that bears little similarity to human ACATN (22% identity). AmpG, the homolog in E. coli, is a 53-kDa multitransmembrane protein required for recycling of murein tripeptide [11]. The presence of a N-acetylglucosminyl-β-1,4-anhydro-N-acetylmuramic acid is critical, because it is the substrate for AmpG permease [12]. This muramic acid is also involved in regulating β-lactamase induction [11]. There are several conserved amino acid residues in the N-terminal region of AmpG and mammalian ACATN proteins. At present, the role of AmpG protein in Ac-CoA metabolism is unknown.

Genetic manipulation of SLC33 proteins will enable us to identify their functional roles in vivo. For example, using C. elegans as a model animal, we have examined the effect of RNAi (RNA-mediated interference) on Acatn gene expression. Knocking out Acatn reduces brood size and shortens body length due to insufficient maturation of the gonadal system. It is also important to note that embryos bearing this Acatn knock-out occasionally die (unpublished observation).

Diseases and mutants

To date, no human diseases involving ACATN have been described. In C. elegans, we have isolated a deletion mutant allele of the Acatn gene (unpublished observation). Worms bearing this deletion mutant allele are viable, but their development is severely compromised. The gonadal system of these worms appears to be also severely compromised. Using mutant worms, we expect to identify new biological systems in which the Ac-Co-A transporter plays a key role.