The term “tail-anchored” (TA) proteins was introduced by the group of Tom Rapoport more than 10 years ago (Kutay et al. 1995) in a study on mammalian synaptobrevin, a protein of the soluble N-ethylmaleimide (NEM)-sensitive factor attachment protein receptors class (SNAREs). This term well describes the topology of proteins that are anchored to the membrane via a C-terminally located hydrophobic domain, followed by a short (or null) polar sequence. In the resulting topology, the bulk of the protein (functional, catalytic domain) faces entirely the cytosol, and the short C-terminal region is translocated into the lumen of the organelle of residence. Proteins with this topology are present in all eukaryotes and in bacteria (Kalbfleisch et al. 2007; Borgese et al. 2009), on the cytosolic face of any cellular membrane, and carry out a variety of functions, some of which are fundamental for cell metabolism and survival. Moreover, the TA protein class also includes several viral proteins (Brideau et al. 1998; da Fonseca et al. 2000; Schmidt-Mende et al. 2001; Koshizuka et al. 2002; Koshizuka et al. 2008).

The position of the TMD, which is located near to the C terminus, is responsible for the peculiar mechanism of insertion into the lipid bilayer. Indeed, the hydrophobic domain emerges from the ribosome only after translation has been completed. The polypeptide is therefore released in the cytosol, and necessarily, its insertion into the target membrane occurs posttranslationally (Fig. 1). This holds true also for TA proteins inserted into the endoplasmic reticulum (ER) membrane, whereas ER insertion of most integral membrane proteins is a co-translational process.

Fig. 1
figure 1

Posttranslational insertion of tail-anchored proteins. Tail-anchored proteins are translated on free ribosomes in the cytosol and cannot interact co-translationally with canonical translocation systems, such as SRP, because the C-terminal hydrophobic domain emerges from the ribosome only when the translation was completed. Once the protein is released in the cytosol, it can follow different pathway (see Fig. 3) to interact with membrane and to translocate the short polar sequence in the lumen of the target organelle

Full size image

Not all bilayers are able to support the posttranslational insertion of tail anchors (see below): TA proteins can be targeted to peroxisomes, chloroplasts, mitochondria, and the ER. Those resident of the Golgi complex, plasma membrane, vacuoles, and endosomes are first inserted into the ER bilayer and then travel along the secretory pathway until they reach their membrane of residence (Fig. 2).

Fig. 2
figure 2

How do tail-anchored proteins choose the correct target membrane? TA proteins translated on free ribosomes are released in the cytosol and can be addressed to a limited number of intracellular membrane: endoplasmic reticulum, chloroplasts outer envelope, mitochondrial outer membrane, peroxisomal membrane. TA proteins resident in the endomembrane system, downstream the ER are first inserted into the ER membrane and travel along the secretory pathway until reach their destination. Peroxisomal TA proteins can be address to these organelles directly from the cytosol or by passing through a specialized region of the ER (peroxisomal ER). ER endoplasmic reticulum, N nucleus, mt mitochondria, G Golgi complex, V vacuole, pex peroxisome

Full size image

There are three essential questions regarding TA proteins: (1) How do they choose their target membrane? (2) Which is the machinery involved in their post-translational translocation? (3) How do TA proteins maintain their localization in spite of membrane flow along the secretory pathway? In the first part of this review, these questions will be discussed, focusing on TA proteins in plants. In the second part, we will report an “in silico” analysis of Arabidopsis thaliana proteome, with the aim of creating a catalogue of putative plant TA proteins. These proteins will be classified both by function and localization. On the basis of our knowledge on TA protein biogenesis, in particular in plants, for each membrane, common features will be tentatively identified of its resident members of this protein class, with the intent of predicting the localization (and function) of uncharacterized TA proteins.

Results obtained to date on the machinery involved TA protein membrane insertion in mammalian and yeast cells will be discussed and summarized.

Finally, a recent work will be presented that propose the use of tail anchors in biotechnology as a strategy to increase the accumulation of recombinant antigens in transgenic plants.

How Do TA Proteins Choose Their Target Membrane?

Once a TA protein is released from free ribosomes, it must reach the correct target membrane. There is general consensus among scientists that the sorting and targeting seem to be governed by physical–chemical features of TA polypeptides rather than by defined sequence motifs (see reviews Borgese et al. 2003, 2007).

The fact that only a subset of intracellular membranes is able to accept TA protein insertion mainly depends on lipid composition, which determines fluidity and plasticity of the bilayer (Brambillasca et al. 2005). Non-acceptor membranes contain high levels of certain lipids, such as sterols, that increase bilayer rigidity and thus impair C terminus translocation (Brambillasca et al. 2005).

A useful model to study sorting and targeting of TA proteins is the electron carrier cytochrome b5 (cyt b5; D’Arrigo et al. 1993). Most of the studies have been performed on animal cells where cyt b5 is present in two isoforms, ER-b5 and MOM-b5, that reside in the ER and in the outer mitochondrial membrane (MOM), respectively (D’Arrigo et al. 1993). It has been demonstrated that the short C-terminal polar region determines targeting to the ER or MOM. This region carries a net negative charge in ER-b5, and reversal of this charge results in mistargeting of the mutated proteins to the MOM (Borgese et al. 2001). In contrast to animal cyt b5s, no plant isoform with a negative C terminus has been identified (see Table 2 in Borgese et al. 2001), suggesting that other mechanisms, instead of charge-based sorting, determine the localization in plants.

The targeting of four tung (Aleurites fordii) cyt b5 isoforms (Cb5-A, -B, -C, and -D) has been studied (Hwang et al. 2004). Mitochondrial targeting of Cb5-D is mediated by a combination of hydrophylic amino acids along one side of the TMD, an enrichment of branched β-carbon-containing residues in the medial portion of the TMD, and a dibasic -R-R/K/H-x motif in the C-terminal tail. By contrast, targeting to the ER depended primarily upon the overall length and hydrophobicity of the TMD, although an -R/H-x-Y/F- motif in the tail was also a targeting determinant (Hwang et al. 2004).

The Arabidopsis genome contains five putative TA cyt b5 isoforms, all with a positive luminal C terminus (see Fig 5a in Maggio et al. 2007). One of these isoforms is sorted to the ER and another to the chloroplast outer envelope (At5g48810/AtCb5-3 and At1g26340/AtCb5-6, respectively, Table 1; Maggio et al. 2007). In cells lacking chloroplasts, AtCb5-6 is targeted to mitochondria, indicating that there is a competition between the two organelles in capturing this cyt b5 isoform and that chloroplasts have the stronger affinity (Maggio et al. 2007). In search of differences between ER and COE Arabidopsis isoforms, it can be noticed that the hydrophobicity profiles of their TA are slightly dissimilar. There is a gradual increase of hydrophobicity in the first half of AtCb5-3 TMD (Table 1, At5g48810). Conversely, AtCb5-6 TMD starts with a sharp increase in hydrophobicity, which slightly decreases in the middle region (Table 1, compare At1g26340 and At5g48810). This could reflect differences in lipid composition of ER and COE. With regard to this point, Toc34, a TA component of the COE translocon, is able to insert itself in the lipid bilayer of the chloroplast envelope in the absence of helper proteins (Schleiff et al. 2001; Qbadou et al. 2003). The authors established that two positive charges in close proximity to the cytosolic end of the TMD dictate the topology of Toc34 (Qbadou et al. 2003). Importantly, the insertion of Toc34 is dependent on the lipid asymmetry present in the outer envelope and the presence of the non-bilayer lipids monogalactosyldiacylglyceride (MGDG) and phosphatidylethanolamine (PE) (Epand 1998; Qbadou et al. 2003). MGDG and PE, like cardiolipin in mitochondria, are a non-bilayer-forming lipid because they have a small polar headgroup relative to the diameter occupied by the two acyl chains. This small headgroup gives the lipid the shape of a cone when rotated along its long axes. Such cone-shaped lipids form inverted hexagonal phases characterized by high local curvature rather than bilayers. Non-bilayer property of membranes is essential for the function of the translocases (Rietveld et al. 1995; Epand 1998).

Table 1 Classification of different cytochrome b5 isoforms from A. thaliana
Full size table

Comparing the hydrophobicity profiles of the five Arabidopsis cyt b5s, it can be observed that AtCb5-2 and AtCb5-4 are similar to AtCb5-3, and therefore, we can hypothesize that they are ER isoforms. The profile of AtCb5-1 resembles that of AtCb5-6, but the amino acid composition of the C terminus is more similar to that of ER isoform. Notice that the ER and putative ER isoforms contain a tyrosine in the C-terminal polar region which is absent in the COE cyt b5. The analysis of putative phosphorylation site using Net-Phos 2.0 server (http://www.cbs. dtu.dk/services/NetPhos/), predicted that the C-terminal tyrosine of AtCb5-3 could be phosphorylated. The phosphorylation in the extreme C terminus could have regulatory implications in the mechanisms of targeting/insertion (Maggio et al. 2007).

The implementation of knowledge both on tail anchor structures and membrane lipid composition will contribute to elucidate the biogenesis of TA proteins in respect to intracellular membranes.

Posttranslational Insertion into the ER: More Than One Pathway

The first evidence of posttranslational insertion of proteins into ER membrane came from a study on rat liver microsomal ER-b5, which is considered the archetype of TA proteins. The results showed a preferential association of cyt b5 polyA+ messenger RNA (mRNA) with free ribosomes, despite the ER localization of the protein, and indicated that it is synthesized in a soluble form and only subsequently is inserted into the ER (Rachubinski et al. 1980). A few years later, it was shown that ER-b5 synthesized in vitro in a wheat germ cell-free translation system was able to tightly bind posttranslationally added dog pancreas microsomal membranes (Anderson et al. 1983). In the following years, several studies have contributed to acquire new information on TA protein biogenesis. In particular, the setup of increasingly sophisticated translocation assays has partially clarified the mechanism of posttranslational insertion (at least in mammalian and yeast cells).

According to current models, TA protein insertion into the ER membrane can follow three different pathways depending mainly on the physical–chemical features of the TMD; in certain cases, these pathways can also partially overlap.

The first pathway was described in the laboratory of Nica Borgese (Brambillasca et al. 2006) where the assay for TA protein translocation was also developed (Pedrazzini et al. 2000; Brambillasca et al. 2005). The authors used ER-b5 as a model and then extended their studies on other TA proteins. ER-b5 insertion does not depend on Sec61 channel and/or translocon accessory proteins (Yabal et al. 2003; Brambillasca et al. 2006) and can occur spontaneously, without protein assistance, across pure lipid vesicles (Fig. 3a; Brambillasca et al. 2006). At least another protein, protein tyrosine phosphatase 1B (PTP-1B), is able to follow this spontaneous pathway, indicating that the feature is not unique to cyt b5 (Brambillasca et al. 2006). What is the common characteristic between these two proteins? As we will illustrate below, other TA proteins are unable to insert spontaneously into the bilayer. The mild hydrophobicity of cyt b5 and PTP-1B TMDs can provide an explanation: TMDs with higher hydrophobicity could cause irreversible aggregation of the polypeptide immediately after its release from the ribosome and thus need interactions with molecular chaperones to avoid this.

Fig. 3
figure 3

Different pathways for TA proteins sorting. TA protein targeting to the ER membrane can follow three different pathways depending on the protein physical–chemical features; in certain cases, these pathways can also partially overlap. TA proteins which have TMDs with moderate hydrophobicity can remain in solution, forming micellae, and are therefore inserted into the membrane without assistance of molecular chaperones (a). TA proteins with a longer or more hydrophobic TMD need molecular chaperones, such as TRC40/Get3 or Hsp40/Hsc70, to avoid irreversible aggregation. A membrane receptor complex (Get1/Get2) has been identified in yeast, which recognized Get3-TA protein complex (b, c)

Full size image

The lipid composition of membranes is also important for an effective translocation of ER-b5 C terminus; indeed, cholesterol-loaded artificial vesicles impaired ER-b5 insertion completely, even if low concentration of sterol was used (Brambillasca et al. 2005). This is probably due to the increased order and thickness of the lipid bilayer caused by the sterols and can reflect the in vivo inability of TA proteins to insert into sterol-enriched membranes.

Our experimental evidence (unpublished) supports the hypothesis that the ability of spontaneous membrane integration is maintained also by plant ER-b5. As suggested by Maggio et al. (2007), the ability to insert directly in the lipid bilayer, without the help of translocons and protein machinery, could be a feature of ancestral proteins which were able to recruit lipids, contributing to the biogenesis of cellular membrane.

The second pathway involves a TMD recognition complex (TRC) and is ATP-dependent. This pathway is followed by a subset of TA proteins, such as the mammalian Sec61β and synaptobrevin, which have more hydrophobic TMDs, rendering them reliant on an incompletely characterized, ATP-dependent mechanism (High and Abell 2004; Abell et al. 2007; Stefanovic and Hegde 2007; Favaloro et al. 2008). The major player in the TRC pathway is the 40-kDa cytosolic factor TRC40/Asna1, the homologue of bacterial ArsA and yeast GET3 ATPases (Bhattacharjee et al. 2001; Shen et al. 2003). Mammalian Asna-1 has 27% homology to the bacterial ArsA, which is involved in arsenite transport (Kurdi-Haidar et al. 1996). However, the mammalian protein has little or no arsenite-stimulated ATPase activity and plays a different role from its distant bacterial homolog (Kurdi-Haidar et al. 1998). TRC40 binds directly TMDs with high hydrophobicity. The energy used to insert the protein into the bilayer comes from ATP hydrolysis (Fig. 3b). Moreover, the membrane insertion mediated by Asna1 is sensitive to NEM and oxidants, indicating that cytosolic redox conditions can influence the binding of Asna1 substrates (Favaloro et al. 2008). A recent study by Schuldiner et al. (2008) has shown that cytosolic GET3 recognition represents the key decision step for the insertion of TA proteins into yeast ER; loss of this factor can lead to mistargeting to mitochondria. Moreover, Get3–TA protein complexes are recruited by the Get1/Get2 receptor that resides on the ER membrane. The absence of Get1/Get2 causes cytosolic aggregation of Get3–TA complexes.

The third pathway of TA membrane insertion is stimulated by the chaperone Hsc70 in conjunction with Hsp40: this complex binds the TMD to avoid cytosolic aggregation (Fig. 3c; Abell et al. 2007; Rabu et al. 2008).

The mechanism of plant TA protein integration has not been investigated in detail yet, but the knowledge acquired on mammals and yeast combined with the large availability of proteomic data and tools that plant biologists can access (see http://www.arabidopsis.org; http://aramemnon.botanik.uni-koeln.de/) might quickly narrow the gap. Moreover, the analysis of A. thaliana knockout mutants can provide important information on the role played by individual TA proteins and by the components of the machineries that take care of their biogenesis in plant development and reproduction.

In the following part of this review, we will describe bioinformatic data on A. thaliana TA proteome that could be useful to elucidate the biogenesis of TA proteins in plants.

Putative TA Proteins in the Arabidopsis Proteome

An overall picture of the TA proteome in A. thaliana can cast light on the variety of functions of these proteins, reveal new functions, and help in identifying the target membranes. To this purpose, a first investigation of A. thaliana proteome has been performed using the tools provided by the site http://www.arabidopsis.org.

As a first step, the bulk protein search tool (http://www.arabidopsis.org/tools/bulk/protein/index.jsp) was used to extract putative proteins with a single TMD. It has been also fixed, as restricted by predicted protein characteristic, the intracellular location as other and undefined, to exclude proteins with predicted signal peptide (for co-translational translocation into the ER), chloroplast, or mitochondrial targeting signals. The option entire range was set for values of both isoelectric point (pI 0.00 to 14.00) and molecular weight (M W 0.00 to 1,000,000 Da) of the protein. Two lists of loci were obtained in this first step: one containing 877 loci coding for putative proteins with a single TMD and undefined localization (Electronic Supplementary Materials Table S1) and the second containing 338 loci coding for putative proteins with a single transmembrane domain and other localization (cytoplasm or not identified; Electronic Supplementary Materials Table S2). It should be noted that this first search did not provide any information about the position of the TMD. To restrict the two lists to TA proteins (which have the TMD near to the C terminus), each putative sequence was further analyzed one by one using the plant membrane protein database ARAMEMNON (http://aramemnon.botanik.uni-koeln.de/index.ep); sequences having no other putative TMD that might have escaped the first step selection, no farther than 60 residues from the C terminus and with average hydrophobicity more than 0.3 (calculation of the average hydrophobicity is based on the hydrophobicity scale published by Eisenberg et al. 1984 and is directly provided by ARAMEMNON—transmembrane detail window), were validated as TA protein. The 60 residues criterion was more restrictive than the limits of in vitro unassisted translocation of ER-b5 (85 residues) reported by Brambillasca et al. (2006). TA proteins with a C-terminal polar region longer than 35 residues could in theory be recognized by SRP and inserted co-translationally in vivo because in this case, the TMD would emerge from the ribosome before the end of translation (Borgese et al. 2003). Experimental evidence will be necessary to assess the posttranslational insertion of these proteins in vivo. Moreover, sequences that had a value of consensus prediction for localization in chloroplast, mitochondria, or secretory pathway higher than 8 were excluded as well. In the end, the list summarized in Table 2 was obtained, which contains 164 putative TA proteins.

Table 2 Putative and known tail-anchored proteins in A. thaliana
Full size table

These proteins were classified by putative function, once more using ARAMEMNON database. Table 2 shows nine different functional groups and confirms the variety of TA protein roles. Putative TA proteins with a C terminus longer than 35 residues are marked by an asterisk.

The most abundant group is constituted by SNARE polypeptides, which are involved in vesicle fusion to target compartments along the secretory pathway (Sanderfoot et al. 2000): from this analysis, 59 SNAREs (out of a total of 64 SNAREs entries) have a C-terminal putative TMD. The comparison of the data with the list of A. thaliana SNARE proteins published in a recent review (Lipka et al. 2007) indicates that 15 unclassified TA-SNARES encoded by the A. thaliana genome are revealed by the present analysis.

As mentioned above, TA proteins resident of the different compartments of the secretory pathway are first inserted into the ER and then traffic along the pathway until they reach their residence (Kutay et al. 1995). The maintenance of their localization, in spite of continuous vesicle traffic, is due predominantly to the length and hydrophobicity of their TMD (Pedrazzini et al. 1996; Bulbarelli et al. 2002). TA SNAREs were ordered based on the increasing average hydrophobicity value of their C terminus, and the results are listed in Table 3. When the established subcellular localization of known SNAREs is taken into account, it is rather clear that the more an Arabidopsis SNARE is distal from the ER along the secretory pathway, the more the hydrophobicity of its TMD increases. With the only exception of SYP132, all known SNAREs of the plasma membrane, tonoplast, and cell plate are in the second half of the list.

Table 3 Classification of TA SNARES by increasing hydrophobicity index of their C-terminal region, which include the TMD
Full size table

Another interesting group of putative TA proteins that has been identified is constituted by “transcription factor/DNA binding proteins” (Table 2). Membrane-bound transcription factors (MTFs) have been found in prokaryotes, yeast, animals, and plants (Brown et al. 1997; Hoppe et al. 2000; Kim et al. 2007; Seo et al. 2008). In plants, several MTFs were previously characterized (Kim et al. 2007; Chen et al. 2008). For example, the NAC MTF named NTM1 resides on the ER membrane: when the tail anchor of NTM1 is removed by proteolysis, the cytosolic portion, containing the NAC domain, is able to enter the nucleus and activates genes involved in cellular division (Kim et al. 2006). The transcription factor AtbZIP60 regulates ER stress response by shuttling from the ER membrane, where the stress signal is sensed, and the nucleus: its detachment from the ER seems to be mediated by proteolysis (Iwata et al. 2008). Therefore, signal transduction across intracellular bilayers, such as the ER membrane, seems to require less intracellular mediators in plants than animals because of the high availability of TFs which are directly bound to the membrane.

A number of components of translocation complexes of the ER and outer mitochondrial membrane (TOM) are TA proteins (Kalbfleisch et al. 2007). The present analysis indicates that only one component of the chloroplast outer envelope translocon (Toc34) has a TA topology, while seven TOM components are predicted to be TA proteins (Table 2). This observation could reflect differences in the biogenesis of chloroplasts and mitochondria. TA proteins of the ER translocon are Sec61β and Sec61γ (Table 2).

The TA proteome identified by this approach also includes putative TA enzymes with a variety of functions: the already reported five cytochrome b5 isoforms (Maggio et al. 2007), proteins involved in proteasomal degradation (AT1G17280 and AT1G33480), as well as two FKBP-like and one cyclophilin-like peptidyl-prolyl isomerases. Among these, PASTICCINO1 plays an important role in the control of plant development (Vittorioso et al. 1998).

A number of polypeptides having similarity of the N-terminal region with small heat shock proteins (sHsps) are also classified as TA proteins by the current bioinformatic analysis. One of them, RTM2, is involved in resistance of Arabidopsis to tobacco etch potyvirus (TEV) by blocking long-distance movement of the virus (Whitham et al. 2000). Several lines of evidence suggest that although the RTM2 N-terminal domain is related to sHSPs, RTM2 is unlikely to possess typical chaperone functions because it is not heat-inducible under conditions that stimulate the heat shock response (Whitham et al. 2000). The substrate of RTM2 is unknown, and several hypotheses on the mechanism of its action were postulated (Whitham et al. 2000). Certainly, the localization of membrane-bound plant sHSPs could contribute in the identification of real substrates, in understanding the role of these TA proteins in stress tolerance and development, and in unraveling their impressive multiplicity. Knowing sHSP substrates may also help understand whether the same sHSP plays the same role upon different conditions as well as recognize differences between plant species.

Protein containing zinc finger/RING finger domains are also putative TA and could have a role in protein–protein interactions on membrane surfaces. Finally, the screen identified TA proteins with various other functions, a number of TA proteins with unknown function, and some hypothetical TA proteins (Table 2—miscellaneous, unknown and hypothetical, respectively).

On the basis of the TMD properties and hydrophobicity profiles, it should be possible to predict the intracellular localization of TA proteins with as not yet identified functions.

Tail Anchors and Biotechnology

It has been recently shown that addition of a tail anchor improves the accumulation of HIV-negative factor (Nef), a promising target for vaccine development against HIV infection expressed in transgenic tobacco plants (Barbante et al. 2008). Nef is a cytosolic protein, but its accumulation levels in the cytosol or as a secretory protein (produced by adding a signal peptide) has been unsatisfactory (Marusic et al. 2007). In an effort to improve its accumulation, Nef was anchored to the cytosolic face of the ER membrane via the addition of the tail anchor of mammalian ER-b5. The chimeric protein (termed Nef-TA) has the expected TA topology. Nef-TA has a longer half-life and accumulates to higher levels (more than threefold) than its cytosolic counterpart. The half-lives of TA proteins are highly protein-specific and the turnover mechanism is, up to date, unknown. For example, the yeast TA protein Ubc6p is a very short-lived protein, but when the transmembrane domain (TMD) is removed, its half-life increases (Walter et al. 2001). On the contrary, when the TMD of the yeast TA long-lived protein Ubc4p was replaced with the TMD of Ubc6p, the half-life of the former decreased in spite of the fact that its localization on the ER membrane was maintained (Walter et al. 2001). Therefore, the half-life of TA proteins is strongly influenced by the structure of their TMDs. ER-b5 is a very long-lived protein, both in mammalian and plant cells (Borgese et al. 2001; Maggio et al. 2007; Pedrazzini et al. 2000). This is probably the reason why the cyt b5 tail anchor stabilizes Nef in the cytosolic environment. Other pharmaceutical proteins were fused to the TA domain of ER-b5, and also in those cases, accumulation of the chimeric proteins was improved (Alessandra Barbante and Emanuela Pedrazzini, unpublished data). Several GFP fusions to TA proteins have been expressed in plant cells (Lisenbee et al. 2003; Uemura et al. 2004); however, the different stabilities have not been compared. It will be interesting to produce GFP fusions to the tail anchors of different TA proteins located in the same compartment and compare stabilities. This would provide clues on the structural features that determine turnover. Besides the biotechnological implications, these results could cast light on the mechanisms of membrane protein degradation in plant cells.