N-terminal N-myristoylation of proteins: refinement of the sequence motif and its taxon-specific differences¹

Abstract

N-terminal N-myristoylation is a lipid anchor modification of eukaryotic and viral proteins targeting them to membrane locations, thus changing the cellular function of modified proteins. Protein myristoylation is critical in many pathways; e.g. in signal transduction, apoptosis, or alternative extracellular protein export. The myristoyl-CoA:protein N-myristoyltransferase (NMT) recognizes the sequence motif of appropriate substrate proteins at the N terminus and attaches the lipid moiety to the absolutely required N-terminal glycine residue. Reliable recognition of capacity for N-terminal myristoylation from the substrate protein sequence alone is desirable for proteome-wide function annotation projects but the existing PROSITE motif is not practical, since it produces huge numbers of false positive and even some false negative predictions.

As a first step towards a new prediction method, it is necessary to refine the sequence motif coding for N-terminal N-myristoylation. Relying on the in-depth study of the amino acid sequence variability of substrate proteins, on binding site analyses in X-ray structures or 3D homology models for NMTs from various taxa, and on consideration of biochemical data extracted from the scientific literature, we found indications that, at least within a complete substrate protein, the N-terminal 17 protein residues experience different types of variability restrictions. We identified three motif regions: region 1 (positions 1-6) fitting the binding pocket; region 2 (positions 7-10) interacting with the NMT’s surface at the mouth of the catalytic cavity; and region 3 (positions 11-17) comprising a hydrophilic linker. Each region was characterized by physical requirements to single sequence positions or groups of positions regarding volume, polarity, backbone flexibility and other typical properties of amino acids (http://mendel.imp.univie.ac.at/myristate/). These specificity differences are confined partly to taxonomic ranges and are proposed for the design of NMT inhibitors in pathogenic fungal and protozoan systems including Aspergillus fumigatus, Leishmania major, Trypanosoma cruzi, Trypanosoma brucei, Giardia intestinalis, Entamoeba histolytica, Pneumocystis carinii, Strongyloides stercoralis and Schistosoma mansoni. An exhaustive search for NMT-homologues led to the discovery of two putative entomopoxviral NMTs.

Introduction

Ongoing large-scale genome sequencing will further increase the imbalance between the number of genes described by sequence alone compared with the minority of proteins characterized functionally with biological, biochemical and/or biophysical techniques. This stimulates the development of computer-based approaches for sequence-structure and sequence-function assignments. In knowledge-based techniques, correlations between sequence patterns and biological features have to be established in advance and are transferred to uncharacterized sequences in a later step. Unfortunately, the information pointing to sequence-function associations is often scattered in electronic databases and in the scientific literature. Therefore, the extraction of the true protein sequence pattern from the heterogeneous data may become a time-consuming, independent scientific search, as we have experienced with the GPI-lipid anchor sequence pattern in animal proteins.1

Among the many known lipid modifications, N-terminal N-myristoylation of proteins is one of the best investigated from the experimental point of view.2, 3, 4, 5, 6, 7 Our tractate is dedicated to the refinement of the myristoylation sequence motif based on the meta-analysis of published data from electronic databases and from the literature. Especially, the issues of motif length and sequence context are addressed. We analyzed three major sources of data: (i) the sequences of experimentally verified substrate proteins; (ii) the kinetic data for oligopeptide myristoylation by the myristoyl-CoA:protein N-myristoyltransferase (NMT);2 and (iii) amino acid sequences, crystallographic structures8, 9, 10 and 3D homology models of NMTs for various species.

While NMTs seem to be ubiquitous among eukaryotes, the existence of isozymes and tissue-dependent activities11 complicates the interpretation of the actual, overlapping yet distinct12 substrate specificity between species.13, 14, 15 We will present data indicating that these specificities are due to differences in the substrate binding pocket, which are evolutionarily conserved and which separate lower and higher eukaryotic NMTs.

This text is organized as follows: after a short introduction into the biology of myristoylation, we summarize the status of experimental verification of NMT protein substrates and show that a description in terms of amino acid types is insufficient for recognition of the sequence motif encoding the capacity for myristoylation. Then, the results of searches for physical property patterns in the N-terminal region are presented. Finally, structural differences in the NMT binding pockets are analysed in context with biochemical data and we provide principal suggestions for the design of taxon-specific NMT inhibitors.

Although the existence of mechanisms with completely different substrate specificities becomes more and more evident,7 the most abundant form of myristoylation is catalyzed by the NMT11 that is absolutely dependent on the N-terminal glycine residue. This work focuses solely on this enzymatic activity. Here, we recall major facts highlighting the biological importance of this protein modification. The rare C₁₄ saturated fatty acid is linked most often cotranslationally16, 17via amide bond18 specifically to the N-terminal glycine residue19, 20 of a variety of eukaryotic and viral proteins. But myristoylation may also take place outside the translational context. The attachment of myristic acid to the N-terminal glycine residue of a protein from Dictyostelium discoideum was shown to occur post-translationally.21 Cleavage of the pro-apoptotic protein BID22 unveils a quondam internal glycine residue that is followed by a sequence motif recognized by the NMT. The lipid anchor targets BID to the mitochondrial membrane and, thereby, facilitates BID-induced release of cytochrome c, an important step in apoptosis. Although myristoylation has long been supposed to be irreversible, demyristoylating activity was observed in brain synaptosomes for MARCKS (myristoylated alanine-rich C kinase substrate).23

The attachment of the lipid moiety results in an increase of hydrophobicity that triggers membrane and protein association. Myristic acid represents less than 1% of all fatty acids in cells,24 but its specific length provides the possibility for reversible interactions with other proteins25 or membranes26 in contrast to highly stable associations facilitated by other, more hydrophobic lipid modifications. Myristoylation can be required but must not necessarily be sufficient for membrane anchoring, as known, for example, for the oncoprotein p60^v-src.27 Often, subsequent palmitoylation adds the missing hydrophobicity, but also a region rich in basic residues can mediate further attraction.28

Myristoylation is not always reduced to a simple anchoring function. The fatty acid can switch between folding back to a domain of the acylated protein itself and extending to the outside again controlled by the binding of Ca²⁺ as in recoverin.29 Other examples of myristoyl switches for reversible membrane association are MARCKS30 and HIV-1 Gag precursor.31

The myristoylated proteins were long considered to be restricted to intracellular compartments. Surprisingly, a hydrophilic acylated surface protein (HASP) without classical secretory sequence signals was shown to localize at the extracellular part of the plasma membrane of the parasite Leishmania.32 This new myristoylation/palmitoylation-dependent export mechanism does not seem to be limited to lower eukaryotes, as the same protein was transported to the extracellular side of the plasma membrane of transfected mammalian cells.

There are cases of myristoylation of other residues than glycine that are listed for the sake of completeness here. N-Myristoylation occurring not N-terminally (e.g. on internal lysine residues) has been observed for the α tumor necrosis factor precursor,33 the insulin receptor,34, 35 the μ immunoglobulin heavy chain,36 the lysozyme,37 the interleukin 1 α and β precursors38 and the subunit 1 of cytochrome c oxidase.39 Other occurrences of myristic acid include a lux-specific myristoyltransferase in luminescent bacteria,40, 41 fatty acid remodeling on GPI anchors42 and S-myristoylation.43, 44, 45