S-acylation (commonly known as palmitoylation) is a widespread post-translational modification that consists of the addition of a lipid molecule to cysteine residues of a protein through a thioester bond. This modification is predominantly mediated by a family of proteins referred to as PATs (palmitoyltransferases). Most PATs are polytopic membrane proteins, with four to six transmembrane domains, a conserved DHHC motif and variable C-and N-terminal regions, that are probably responsible for conferring localization and substrate specificity. There is very little additional information on the structure–function relationship of PATs. Swf1 and Pfa3 are yeast members of the DHHC family of proteins. Swf1 is responsible for the S-acylation of several transmembrane SNAREs (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptors) and other integral membrane proteins. Pfa3 is required for the palmitoylation of Vac8, a protein involved in vacuolar fusion. In the present study we describe a novel 16-amino-acid motif present at the cytosolic C-terminus of PATs, that is required for Swf1 and Pfa3 function in vivo. Within this motif, we have identified a single residue in Swf1, Tyr323, as essential for function, and this is correlated with lack of palmitoylation of Tlg1, a SNARE that is a substrate of Swf1. The equivalent mutation in Pfa3 also affects its function. These mutations are the first phenotype-affecting mutations uncovered that do not lie within the DHHC domain, for these or any other PATs. The motif is conserved in 70% of PATs from all eukaryotic organisms analysed, and may have once been present in all PATs. We have named this motif PaCCT (‘Palmitoyltransferase Conserved C-Terminus’).
- conserved motif
Protein palmitoylation, or S-acylation, the addition of a lipid molecule on cysteine residues through a thioesther bond, is involved in multiple cellular processes. In the case of hydrophilic proteins, such as Ras or G-proteins, palmitoylation serves to recruit them to the membrane, often in combination with prenylation. Since palmitoylation is reversible, it can dynamically regulate the localization and function of proteins. Many transmembrane proteins are also palmitoylated and this has been shown to regulate function, localization and stability [1–3]. Palmitoylation of transmembrane proteins occurs at cysteine residues often near or within the TMD (transmembrane domain) [4,5].
It has now become accepted that a family of proteins containing a 50 residue motif called DHHC-CRD (Asp-His-His-Cys cysteine-rich domain)  is involved in protein S-acylation ([7,8] and reviewed in ). There are over 20 predicted DHHC-CRD-containing proteins in the human genome and seven in the yeast Saccharomyces cerevisiae genome. They are mostly polytopic membrane proteins predicted to contain four to six TMDs. The membrane topology of the yeast PAT (palmitoyltransferase) Akr1 has been established experimentally and shown to match the in silico predictions . Subsets of substrates have been assigned to most of the yeast PATs [1,7,8,11–15], and a few mammalian PATs [11–13,16,17], indicating that they must have determinants for specific substrate recognition.
There is very little knowledge about the mechanism of protein palmitoylation and the way PATs function [18,19]. At the structural level, only a few residues in the conserved DHHC core have been mutated resulting in lack of function [7,8]. Mutations in the DHHC domain of human zf-DHHC9, the Ras PAT , have been linked to mental retardation , and residues within this region are also required for mouse HIP14 to produce oncogenic transformation . PATs are interesting targets for anticancer drugs owing to their important role in the subcellular localization of several oncoproteins . Apart from the DHHC motif, there are two other regions of sequence similarity present in all DHHC proteins , a DPG motif (Asp-Pro-Gly) next to TMD2 and a TTxE motif (Thr-Thr-Xaa-Glu) adjacent to TMD4 (Figure 1A) (the TMDs are numbered as in ). The role of these regions has not been addressed. The N- and C-termini of PATs are highly variable and probably confer localization and/or specificity towards the different sets of substrates .
Swf1 is a yeast DHHC protein involved in the palmitoylation of SNAREs (soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptors)  and possibly glycosyltransferases , suggesting that palmitoylation of proteins with single TMDs in yeast is mostly due to Swf1. The function of transmembrane SNARE palmitoylation is not clear, but in the case of the endosomal syntaxin Tlg1, it seems to protect it from degradation by the quality-control machinery. Non-palmitoylated Tlg1 is ubiquitinated and degraded in the vacuole . Unlike other yeast PATs, Swf1 has five predicted TMDs and thus the N-terminus is embedded in the membrane (Figure 1A). We initially focused our attention to the last 100 amino acids of the C-terminal region of Swf1, which are predicted to be facing the cytosol, and thus might represent a good candidate region to confer Swf1-specific characteristics to this PAT. In the present study, we identify a conserved motif at the C-terminus of Swf1, and within this motif, a residue, Tyr323, that is critical for Swf1 function. The motif, however, is not just conserved in Swf1 orthologues, but is also present in other yeast PATs, such as Pfa3, Pfa5 and Erf2.
We extended our studies to Pfa3, a yeast PAT localized to the vacuolar membrane [15,21,22] and predicted to have four TMDs. Pfa3 is required for the palmitoylation of Vac8 , a protein with several armadillo repeats that is also localized to the vacuolar membrane [23–25]. Vac8 is involved in vacuole fusion and vac8Δ cells have fragmented vacuoles [23,24]. Palmitoylation of Vac8 is important for its function [26,27]. We show that in the absence of the PaCCT (palmitoyltransferase conserved C-terminus) motif, the function of Pfa3 is abolished, probably because it is missorted to the vacuole lumen and subsequently degraded. Detailed bioinformatics analyses indicate that the motif is conserved in most PATs and is present almost exclusively within the C-termini of the DHHC protein family.
In silico analysis
An alignment of the conserved C-terminal regions in Swf1, Pfa3 and Erf2 was used to iteratively run hmmbuild, hmmsearch and hmmalign . As a result, we gathered an alignment of the putative motif from 720 proteins from the Uniprot database. These proteins were searched for the presence of the zfDHHC Pfam domain (PF01529), and all sequences which did not score above the noise cut-off according to Pfam  were discarded. Sequence redundancy was reduced to 60% using Jalview , to generate an alignment of 185 proteins. This alignment was used to build a HMM (Hidden Markov Model), which was used to search the Uniprot database. Hmmsearch runs using this HMM were performed with a highly specific cut-off score of 11.7 (the score corresponds to the lowest scoring DHHC protein for which empirical evidence of PAT activity is available, S. cerevisiae Swf1) and a more sensitive cut-off score of 7.3 (the score corresponding to the lowest scoring proteins with a minimum alignment coverage of 85% of the PaCCT HMM). Logos were generated with WebLogo , after reducing redundancy by eliminating sequences with 99% or higher similarity over the motif length using Jalview. Alignments were generated using ClustalX (version 2.05) , hmmalign  and manually curated using Jalview.
Phylogenetic tree reconstruction
For phylogenetic tree reconstruction, the data set used included a total of 68 PAT sequences from 12 Ascomycota organisms downloaded from the Swiss-Prot database (see Supplementary Table S1 at http://www.BiochemJ.org/bj/419/bj4190301add.htm), S. cerevisiae Akr2 was not included in this analyses, because it is only present in this organism. ProtTest v1.4  [implementing the AIC (Akaike Information criterion)] was used to estimate the most appropriate model of amino acid substitution for tree-building analyses. The best-fit model of protein evolution for the PAT protein family according to ProtTest corresponds to a JTT (Jones Taylor Thornton) +I+G+F model .
Tree reconstructions were performed by the NJ (neighbour-joining) method  using the MEGA v4.0 software package , using a JTT matrix and 1000 bootstrap pseudoreplicates, with the gamma-distribution model implemented to account for heterogeneity among sites. The shape parameter of the gamma distribution was estimated to be 1.76 using ProtTest. Support for each phylogenetic group was tested using 1000 bootstrap pseudoreplicates. Tree topology assessed by maximum parsimony was very similar to the NJ tree. Figures were generated using iTOL . Circular trees (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/419/bj4190301add.htm) display branch length, whereas rectangular trees (Figure 2) ignore branch length and have been stripped of branches with bootstrap values below 70%.
Plasmids and strains
The strains used in the present study were wild-type BY4742 from the EUROSCARF consortium, or derivatives containing complete deletions of SWF1 or PFA3. Swf1 and Tlg1 plasmids have been described previously . Swf1ΔPaCCT was constructed by PCR using oligo Swf1 01, which anneals to the SWF1 ATG region and oligo Swf1 13 (see Supplementary Table S2 at http://www.BiochemJ.org/bj/419/bj4190301add.htm for all oligonucleotide sequences). The PCR fragments were cloned BamHI–SalI into pjv29 which is a YcpLac33-based vector containing the TPI (triosephosphate isomerase) promoter, and PGK (phosphoglycerate kinase) terminator.
Pfa3 DNA was amplified from Euroscarf BY4742 genomic DNA, using oligos Pfa3 01 and Pfa3 02 and cloned BamHI–PstI in pjv97 [a YcpLac 33-based vector containing a TPI promoter, GFP (green fluorescent protein) and a PGK terminator]. Mutations were generated by ligation of two PCR-generated fragments, one obtained using oligos Pfa3 01 and Pfa3 04, and another generated using oligos Pfa3 05 and Pfa3 02. Oligonucleotides Pfa3 04 and Pfa3 05 introduce a silent XhoI site that allows in-frame ligation. The resulting fragment has the whole Pfa3 PaCCT domain deleted (amino acids Asn248 to Met263). A similar strategy was used to generate the point mutant, but oligonucleotide Pfa3 03 was used instead of oligonucleotide Pfa3 05, resulting in a mutation of Phe250 to alanine. The fragments were subcloned into pjv97 as above.
In vitro mutagenesis of Swf1
For the mutant N321A, oligonucleotides 17 and 18 were annealed to form double-stranded DNA. The oligonucleotides were designed to contain the desired mutation and overhanging MluI and KpnI sites, so that it could be cloned into these sites in Swf1. To generate the mutations Y323A, D324A and G326S the same strategy was used, annealing oligonucleotides 19 and 20 for Y323A, 21 and 22 for D324A, and 23 and 24 for G326S. To generate the mutation F328A, N331A and L332A, oligonucleotide pairs 26 and 27, 28 and 29, and 30 and 31 respectively were annealed to form double-stranded DNA and were cloned into the KpnI–SalI sites into either not tagged or N-terminal tagging vectors. The SalI site has been introduced into Swf1 plasmids after the stop codon.
Protein electrophoresis and Western blot analysis
Protein samples were prepared as described in . Monoclonal anti-GFP antibodies were from Roche (used at 1/2000 dilution). The blots were probed using secondary antibodies coupled to either IRdye680 or IRdye800 (Licor Biosciences) at a 1/20000 dilution, and then scanned using an Odyssey IR imager (Licor Biosciences).
Vacuole fragmentation assay
The vacuole fragmentation assay was carried out precisely as described in . Cells were imaged live, using an Olympus FV 1000 confocal microscope.
In silico identification of a conserved motif in the PATs C-terminal region
A ClustalX alignment of C-terminal regions from Swf1 orthologues showed that they are poorly conserved, apart from the TTxE motif and a region corresponding to the last 16 residues in yeast Swf1 (results not shown). This conserved region appeared to be present at the C-termini of other yeast PATs such as Pfa3 and Erf2. In these proteins, the motif is also in the C-terminal regions predicted to be cytosolic. A HMM was built for this putative motif as described in the Experimental section, which allowed us to align the C-terminus of many DHHC-containing proteins as shown in Figure 1(B). For the purposes of the present study we assume that all DHHC-containing proteins are PATs. Using a cut-off of 11.7 (score of yeast Swf1) when searching the Uniprot database we retrieved 607 proteins, of which 521 (86%) possessed a DHHC domain (Table 1). Figure 1(B) shows an alignment of representative PATs from several relevant organisms. The complete alignment of the motif present in PATs (retrieved using the HMM) was used to construct a LOGO which shows the consensus sequence for the motif (Figure 1C). We named this motif PaCCT (palmitoyltransferase conserved C-terminus). Within the motif, the positions that are most conserved were: position 3 (always aromatic amino acids) and position 11 (always asparagine residues); position 6 contains mostly glycine residues, and residues in position 7 are mostly polar residues; finally, residues in positions 15 and 16 have a tendency to be hydrophobic. Positions 1 and 2 are also relatively well conserved.
From the 86 non-DHHC proteins which scored above 11.7, eleven of them are possible PATs (owing to their predicted topology and presence of the PaCCT motif in the C-terminus), although their DHHC motifs are somewhat divergent and score low when searched with the zfDHHC Pfam HMM. In total, 40 of them possess no conserved domains as indicated by SMART searches and/or are protein fragments. In the remaining 35 proteins, which possess conserved domains, sequence similarity to the PaCCT motif is present within conserved domains (BEACH, MatK_N, ANF_receptor and zf-CHY) or in non-conserved regions. However, the motif is only present in very few members of their protein families; for instance, from 150 members of the zf-CHY protein family, the motif is detected only in two. Therefore the detection of the motif in non-PAT proteins may be an artefact, owing to the small size of the motif and moderate conservation. Moreover, when searching the manually curated database Swiss-Prot, the motif is detected in 111 PAT proteins and just one non-PAT protein.
When the search in Uniprot was carried out using a cut-off score of 7.3, we gathered 1293 proteins of which 594 (46%) scored significantly when searched with the Pfam zfDHHC HMM (Table 1), indicating that some PATs contain a slightly divergent PaCCT motif and thus were not detected using the more stringent 11.7 cut-off. The 7.3 cut-off also led to the detection of a greater number of non-PAT proteins (for a list of all proteins retrieved and their PaCCT motifs see Supplementary Table S3 at http://www.BiochemJ.org/bj/419/bj4190301add.htm).
When the 7.3 cut-off was used to search the Swiss-Prot database, we found that the PaCCT motif is conserved in 66% of fungal PATs (Supplementary Table S1), 67% of mammalian PATs (human, rat and mouse; see Supplementary Table S4 at http://www.BiochemJ.org/bj/419/bj4190301add.htm) and 73% overall (see Table 1).
Phylogenetic analysis of the PaCCT motif in Ascomycota
To analyse the occurrence of the PaCCT motif in certain subgroups of the PAT protein family and to determine its relation to PAT protein phylogeny, we reconstructed the phylogeny of the PAT family. The genomes chosen for the analyses belong to 12 members of the Ascomycota phylum (see Figure 2, Supplementary Table S1 and Supplementary Figure S1), because they have a well-documented phylogeny [39–41] and adequate evolutionary distance.
Figure 2 and Supplementary Figure S1 present NJ trees built with 68 PATs present in 12 Ascomycota organisms. These trees, as well as Supplementary Table S1, show the distribution of the PaCCT motif within the phylogeny of PAT proteins in fungi.
In the case of Akr1 proteins, the PaCCT motif is preserved in the early diverging Schizosaccharomyces pombe, in Pezizomycotina (which in turn contains Eurotiomycetes and Sordariomycetes subclasses), but not in Saccharomycotina, indicating the possibility that the PaCCT motif was lost/diverged early in the Saccharomycotina subphylum. With regard to Swf1 orthologues, the motif is present in all proteins. In Erf2 proteins, the motif can be found in S. pombe, the Eurotiomycetes subclass, and in only half of the Saccharomycotina subphylum members included in the present study. Pfa3 orthologues all present a PaCCT motif except for Neurospora crassa (S. pombe does not have Pfa3 or Pfa4 orthologues in the Uniprot database). With regard to Pfa4 orthologues, the motif is present in Pezizomycotina and half of the Saccharomycotina subphylum members included in the present study. Pfa5 orthologues all present the PaCCT motif, except for Candida glabrata. It is interesting to point out that all PATs present in the early diverging S. pombe and in Eurotiomycetes Emericella nidulans and Aspergillus fumigatus present a PaCCT motif (Figure 2 and Supplementary Table S1).
Deletion of the PaCCT motif abolishes Swf1 function
To test the importance of the PaCCT motif for the function of a DHHC protein in vivo, we investigated the effect of a C-terminal deletion of the PaCCT motif in S. cerevisiae Swf1, which in this protein comprises the last 16 residues of the C-terminus. Mutants in which the whole SWF1 gene has been deleted are unable to grow in YPD [1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose] medium containing 0.85 M NaCl or YP medium plus lactate as the sole carbon source [1,42]. The bases for these phenotypes are unknown, but they can be conveniently used for testing complementation of a swf1Δ strain in growth tests. These phenotypes can be complemented by SWF1-containing plasmids, indicating they are indeed due to lack of Swf1 function and not a secondary effect of the deletion (Figure 3 and ). A complementation test using SWF1ΔPaCCT shows that this construct is unable to complement the lack of growth of a swf1Δ strain in complete media (YPD) with 2% lactate as the sole carbon source (Figure 3), indicating that the last 16 amino acids of Swf1 are essential for its function in vivo.
Mutating the cysteine residue in the conserved DHHC motif has been shown to abolish palmitoylation activity for several PATs [7,8], and, although indirectly, also for Swf1 . A mutated version of Swf1 (Swf1-DHHA) was included in this growth test (Figure 3). This mutant was unable to grow in lactate, indicating that lack of growth in this medium is probably due to lack of Swf1 palmitoylating activity and not a secondary function of this protein.
Tyr323 is essential for Swf1 function and activity in vivo
To investigate the importance of individual residues present in the PaCCT motif, conserved positions were changed by In vitro mutagenesis and the phenotypes of the mutants were analysed in growth tests. Asn321, Tyr323, Asp324, Phe328, Asn331 and Leu332 which correspond to positions 1, 3, 4, 8, 11 and 12 in the LOGO (Figure 1C) respectively, were mutated to alanine. Gly326 (position 6) was mutated to serine because a mutation to alanine would have been conservative. Figure 3 shows that, although most mutations complemented the swf1Δ strain phenotype in lactate, Y323A did not, indicating that this tyrosine residue was crucial for Swf1 function in vivo.
To confirm that this non-complementation phenotype is due to lack of Swf1 palmitoylation activity, we studied the endosomal SNARE Tlg1, a known substrate of Swf1. This SNARE, when not palmitoylated, becomes ubiquitinated by the ubiquitin ligase Tul1 and it is subsequently delivered to the vacuole for degradation . This can readily be observed by analysing a fusion of Tlg1 to GFP. GFP is normally resistant to vacuolar proteolysis and thus delivery to the vacuole can be related to the appearance of a free GFP band in Western blot analysis. Figure 4 shows that although most of the label is in the form of full-length GFP–Tlg1 in wild-type or swf1Δ strains complemented with a wild-type SWF1 plasmid, swf1Δ strains, transformed with an empty vector, with Swf1ΔPaCCT or with mutant Y323A, display a clear increase in the level of free GFP, indicating that they are unable to palmitoylate Tlg1. The rest of the mutations do not have an effect on the degradation of Tlg1, in agreement with the growth tests results in lactate shown in Figure 3.
The PaCCT motif is required for Pfa3 function
To extend the validity of our observations, we analysed the influence of the PaCCT motif in another yeast PAT, Pfa3. This protein is involved in the palmitoylation of Vac8, a myristoylated protein that is involved in vacuolar fusion and inheritance . Smotrys et al.  have shown that in the presence of 2 mM DTT (dithiothreitol), a pfa3Δ strain has fragmented vacuoles and this phenotype is correlated with lack of Pfa3 palmitoylation activity, since a plasmid expressing a DHHS mutant version of this PAT cannot complement this phenotype . We have made use of this assay to investigate the role of the PaCCT motif in Pfa3 function. Two mutated versions of Pfa3 fused to GFP were constructed, one in which the whole PaCCT motif was deleted and another in which position 3 of the PaCCT motif, in this case Phe250, was mutated to alanine (see the Experimental section). Wild-type cells and pfa3Δ cells complemented with wild-type or mutant versions of Pfa3 were labelled with FM4-64  in the presence of 2 mM DTT, and then inspected for vacuole morphology. Figure 5 shows that pfa3Δ cells have extensively fragmented vacuoles and that Pfa3–GFP clearly complements this phenotype, whereas Pfa3ΔPaCCT does not. The cells expressing Pfa3 F250A display an intermediate phenotype (see the Discussion below).
Lack of PaCCT motif results in mislocalization of Pfa3 to the vacuole lumen
Pfa3–GFP has been shown to localize to the vacuolar membrane [15,21,22] and also to the vacuole lumen . In our hands, Pfa3–GFP is mostly localized to the vacuole membrane, although a fraction reaches the vacuole lumen (Figure 5B). However, the ΔPaCCT and F250A versions of Pfa3–GFP localized mostly to the vacuole lumen in pfa3Δ cells. The localization of these proteins was unaffected by the addition of DTT to the medium (results not shown).
Since delivery to the vacuole lumen usually results in protein degradation, the localization results should correlate with the total amount of protein present in a cell extract. Indeed, Western blot analyses of pfa3Δ cells expressing Pfa3–GFP or the mutated versions showed that the protein level is markedly decreased for the F250A mutant, and it is almost undetectable for Pfa3-ΔPaCCT (Figure 5C).
In silico analysis has allowed us to identify a motif that is clearly present in approx. 70% of all eukaryotic DHHC-containing proteins. The PaCCT motif is moderately conserved in PATs, but is always located at the cytoplasmic C-termini, after TMD4. The E-values for searching such a short motif are logically high. However, when taking into account other factors, such as repeated location within protein topology in a close subfamily of proteins such as PATs, this moderately conserved motif of only 16 amino acids can truly be considered a conserved motif present in PATs. In non-PAT proteins, sequence conservation is lower, there seems to be no relationship between proteins that possess the motif, and no conserved localization of the putative PaCCT motif within the domain topology of each protein family.
The distribution of the PaCCT motif in the closely related PAT proteins of the Ascomycota phylum may indicate that the motif was once present in all PAT proteins. In fact, some Ascomycota genomes (S. pombe, Emericella nidulans and Aspergillus fumigatus, see Supplementary Table S1) still possess a PaCCT motif in all its PATs. Posterior divergence or loss of the motif appears to have occurred independently for each PAT subfamily. The reason why some PATs have divergent/absent PaCCT motifs could be that sequence diversity does not hamper the function of the motif due to structural reasons thus precluding identification by sequence comparisons, or that some PATs have evolved different mechanisms to accomplish the PaCCT motif function, perhaps through interaction with a protein partner. Interestingly, the only PATs described to require a binding partner for activity are mammalian DHHC9  and its yeast homologue Erf2 . Unlike the other yeast PAT orthologues analysed (Figure 2 and Supplementary Table S1), whose conservation of the motif is either complete, or a speciation event can be accounted for motif loss/divergence, Erf2 displays a pattern of PaCCT motif conservation that does not directly correlate with gene phylogeny, suggesting that the motif may not be required for Erf2 function.
Complementation tests show that the PaCCT motif is required for the function of Swf1 in vivo. This is most likely due to lack of palmitoylation activity, since its substrate Tlg1 clearly shows the hallmarks of not being palmitoylated, when the PaCCT motif is altogether absent or its essential residue, Tyr323, is mutated. Degradation of Tlg1 in the vacuole is a reliable indicator of lack of Tlg1 palmitoylation . Tyr323 corresponds to position 3 in the PaCCT motif logo, which always contains an aromatic residue. It is interesting that mutation of the other two highly conserved positions, G326S and N331A, allows growth in lactate, and has no effect on Tlg1 palmitoylation. Nevertheless, we cannot exclude that more drastic mutations would result in visible phenotypes.
The PaCCT motif is also required for the function of Pfa3, as shown by lack of complementation of the fragmented vacuole phenotype, which indeed suggests that the motif has an important function at least for some PATs. Moreover, the mutation in position 3 of the PaCCT motif, F250A, also results in, at least, a diminished function.
Mutation of position 3 of the PaCCT motif (Swf1 Tyr323 and Pfa3 Phe250) is, to our knowledge, the first phenotype-changing mutation uncovered for any PAT, which does not lie within the DHHC motif. There are several reasons why this mutation would result in non-functional PATs. For instance, the proteins could be mislocalized or degraded. Attempts to compare Swf1 Y323A and Swf1 wild-type protein levels were unsuccessful, since we were unable to detect either the wild-type or the mutant versions of HA (haemagglutinin)-tagged Swf1 by Western blot analysis. Tagged versions of Swf1 have already proven very difficult to detect [15,21], and our polyclonal antibodies raised against the C-terminus failed to detect the endogenous protein (results not shown). We have previously shown that an epitope-tagged version of Swf1 was localized most prominently, but not exclusively, to the ER (endoplasmic reticulum), with some fluorescence present at the vacuole and in undefined puncta . These versions of Swf1, although expressed from a TPI-driven promoter, were also very difficult to detect, suggesting low levels of Swf1 protein in these conditions. Recently, Dighe and Kozminski  raised polyclonal antibodies against the Swf1 C-terminus, which allowed them to co-localize endogenous Swf1 with cortical actin patches and with actin cables. These antibodies might be useful for the study of Swf1 mutants. However, more work on the trafficking and localization of wild-type Swf1 might be required, before the effect of mutations on these processes can be evaluated with confidence.
The experiments carried out with Pfa3 are more informative regarding the putative function of the PaCCT motif since this protein is more easily detectable than Swf1, both by microscopy and Western blot analysis. This allowed us to observe that, unlike Pfa3–GFP, which localizes mostly to the vacuolar membrane, Pfa3Δ-PaCCT and the point mutant Pfa3-F250A are localized to the vacuole interior, where they are presumably degraded. Western blot analyses show that the degradation of the point mutant F250A is not as extensive as that of Pfa3-ΔPaCCT, which correlates with the intermediate phenotype observed for this mutant in the vacuole fragmentation assay. It is possible that the effect of the F250A mutation on the fragmented vacuole phenotype is underestimated, since the expression of these plasmids is driven by the TPI promoter, which results in, at least, a moderate overexpression.
Why Pfa3, and presumably Swf1, are less stable in the absence of the PaCCT motif is a question that will require further investigation. The motif is probably not a bona fide localization signal, since it is present in PATs with different subcellular localizations, such as Pfa3 (vacuole membrane) , Erf2 (ER)  and Swf1 (see above) . An interesting possibility is that the PaCCT motif is involved in protein–protein interactions resulting in increased stability. The fact that PATs might need a binding partner has been suggested  and, notably, Erf2 levels are reduced in the absence of its binding partner Erf4 .
Although there have been many advances in the identification of PATs and their substrates in recent years, information regarding structure, function and regulation of these proteins is still very scarce. The identification of the PaCCT motif represents a step forward in understanding S-acylation by PATs.
This work was supported by Secretaría de Ciencia y Tecnología, Universidad Nacional de Córdoba [grant number Res. 6908]; Agencia Nacional de Promoción Científica y Tecnológica [grant numbers PICT1239, PICT32937 (to H. J. F. M. and to J. V. T.)]. R. Q. is recipient of a fellowship from Agencia Nacional de Promoción Científica y Tecnológica. H. J. F. M. and J. V. T. are Career Investigators of Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET, Argentina).
We thank Mariana Ferrari, Dr José L. Barra and Dr Hugh Pelham for critical reading of the manuscript, helpful discussions and suggestions prior to submission.
Abbreviations: DHHC-CRD, Asp-His-His-Cys cysteine-rich domain; DTT, dithiothreitol; ER, endoplasmic reticulum; GFP, green fluorescent protein; HMM, Hidden Markov Model; JTT, Jones Taylor Thornton; NJ, neighbour-joining; PaCCT, palmitoyltransferase conserved C-terminus; PAT, palmitoyltransferase; PGK, phosphoglycerate kinase; SNARE, soluble N-ethylmaleimide-sensitive fusion protein-attachment protein receptor; TMD, transmembrane domain; TPI, triosephosphate isomerase; YPD, 1% (w/v) yeast extract/2% (w/v) peptone/2% (w/v) glucose
- © The Authors Journal compilation © 2009 Biochemical Society