OUP user menu

The secretory pathway: exploring yeast diversity

Marizela Delic, Minoska Valli, Alexandra B. Graf, Martin Pfeffer, Diethard Mattanovich, Brigitte Gasser
DOI: http://dx.doi.org/10.1111/1574-6976.12020 872-914 First published online: 1 November 2013

Abstract

Protein secretion is an essential process for living organisms. In eukaryotes, this encompasses numerous steps mediated by several hundred cellular proteins. The core functions of translocation through the endoplasmic reticulum membrane, primary glycosylation, folding and quality control, and vesicle-mediated secretion are similar from yeasts to higher eukaryotes. However, recent research has revealed significant functional differences between yeasts and mammalian cells, and even among diverse yeast species. This review provides a current overview of the canonical protein secretion pathway in the model yeast Saccharomyces cerevisiae, highlighting differences to mammalian cells as well as currently unresolved questions, and provides a genomic comparison of the S. cerevisiae pathway to seven other yeast species where secretion has been investigated due to their attraction as protein production platforms, or for their relevance as pathogens. The analysis of Candida albicans, Candida glabrata, Kluyveromyces lactis, Pichia pastoris, Hansenula polymorpha, Yarrowia lipolytica, and Schizosaccharomyces pombe reveals that many – but not all – secretion steps are more redundant in S. cerevisiae due to duplicated genes, while some processes are even absent in this model yeast. Recent research obviates that even where homologous genes are present, small differences in protein sequence and/or differences in the regulation of gene expression may lead to quite different protein secretion phenotypes.

Keywords
  • protein secretion
  • translocation
  • glycosylation
  • endoplasmic reticulum-associated degradation
  • protein folding
  • vesicle transport

Introduction

Protein secretion is essential for all living organisms. While bacterial secreted proteins have either functions in the cell wall or the exterior space, eukaryotes ‘secrete’ plenty of proteins that are targeted to intracellular compartments. The default pathway of eukaryotic protein secretion is the endoplasmic reticulum (ER) – Golgi pathway, initiated by translocation of the protein through the ER membrane. The minimum requirement for this process is a secretion signal peptide at the N-terminus of the nascent polypeptide. In baker's yeast Saccharomyces cerevisiae, 543 proteins are predicted to carry a signal peptide which accounts for 9.2% of the entire proteome. Further 57 proteins are predicted to be localized in the ER, the Golgi or the extracellular space, and additional 416 proteins are listed to be localized in the ER or Golgi in the Saccharomyces Genome Database (http://www.yeastgenome.org). A summary about the most important cellular decision points in the secretory pathway in S. cerevisiae has recently been provided by Hou et al. (2012). However, most of these 1016 proteins feature intracellular functions in the secretion pathway, in membranes, and in the cell wall.

Studying protein secretion in yeast is important for three main reasons. A number of human diseases, like Alzheimer's, Parkinson's, diabetes mellitus, atherosclerosis and ischemia, as well as liver and heart diseases (Yoshida, 2007) are related to protein folding and secretion. Yeasts can serve as valuable models to study the molecular background of such disorders (Coughlan & Brodsky, 2005). For pathogenic yeast species, a functional secretory pathway is crucial for virulence, and secreted proteins like hydrolytic enzymes are described as virulence factors (summarized, e.g., by Schaller et al., 2005; Sorgo et al., 2013). Yeasts also serve as hosts for heterologous protein production, accounting to approximately 20% of the current biopharmaceutical drug portfolio (Ferrer-Miralles et al., 2009). These, and the rapidly increasing requirement of industrial enzymes, are mainly produced employing the secretory pathway (Mattanovich et al., 2012).

Eukaryotic cells react to secretion stress caused by an overload of unfolded or misfolded proteins in the ER lumen by activation of the Unfolded Protein Response (UPR) pathway, which aims at restoring cellular homeostasis, for example, by induction of genes involved in protein folding and the ER-associated degradation pathway (ERAD). During ERAD, misfolded secretory proteins are retrotranslocated to the cytoplasmic side of the ER, polyubiquitinated, and delivered to the proteasome for degradation (UPR and ERAD were recently reviewed, e.g., by Malhotra & Kaufman, 2007; Vembar & Brodsky, 2008; Hoseki et al., 2010; Kohno, 2010; Smith et al., 2011). The yeast UPR transcription factor Hac1 regulates transcription of several hundred genes (see Travers et al. (2000) and Kimata et al. (2006) for S. cerevisiae; Graf et al. (2008) for Pichia pastoris; (Wimalasena et al. (2008) for Candida albicans).

Protein secretion is best studied in mammalian cells and S. cerevisiae, indicating that the core features of initial secretion steps are similar, while more differences exist for lateral processes. Single steps of the secretion pathway are comprehensively reviewed (as referenced later), featuring mammalian (mainly human) cells and S. cerevisiae, but a comprehensive overview of the current state of knowledge on protein secretion in yeast is missing. ‘Yeast’ usually refers to S. cerevisiae, not giving full credit of the huge genetic and physiological diversity of yeasts, both as research models and biotechnological production hosts. Genome sequencing has provided the research community with a wealth of information, greatly facilitating comparative studies, and technological development of non-conventional yeasts. As an example, Swennen & Beckerich (2007) studied vesicle transport of proteins in Yarrowia lipolytica on a genome scale in comparison with other yeasts. Genome sequence information also enabled the investigation of proteome subsets involved in secretion, as the intracellular interactomes of proteins destined to secretion in Y. lipolytica (Swennen et al., 2010) and P. pastoris (Pfeffer et al., 2012).

Apart from the fragmented literature overview of the processes involved in secretion, especially for other yeasts than S. cerevisiae, there is also some overlap to mammalian cells in the reviewed research concerning protein secretion-related pathways. Therefore, this review aims to provide a comprehensive overview of the protein secretion pathway in yeasts, differentiating between yeast specific functions and those of mammalian cells, and exploring the differences between S. cerevisiae and several other yeasts. To focus stringently on the secretion pathway as the route to export of proteins from the cell, we have not reviewed endocytosis and vacuolar targeting here. Saccharomyces cerevisiae will be used as a benchmark here, as by far most experimental evidence on the components and functions of the secretion pathway exist for this yeast. On the other hand, the fission yeast Schizosaccharomyces pombe has been intensely studied and is therefore included in this comparative study. This work was only possible due to the availability of annotated genome sequences of S. cerevisiae (Goffeau et al., 1996) and S. pombe (Wood et al., 2002).

Due to the fact that non-conventional yeasts like P. pastoris, Hansenula polymorpha, or Y. lipolytica are emerging platforms for applied research, especially for recombinant protein production (Mattanovich et al., 2012), it is substantial that the conclusions drawn from model species like S. cerevisiae are valid for such yeasts as well. Therefore, we compare in this review the secretory pathways of the respective yeasts and discuss existing gaps or discrepancies mainly based on a comparison of the presence of genes in the genome. As the phylogenetic distance between these hemiascomycete species is considerably high (Dujon et al., 2004), we chose to include also Kluyveromyces lactis as a closer relative to S. cerevisiae, even more as it is employed for protein production as well. Additionally two pathogenic yeasts were evaluated here, namely C. albicans and Candida glabrata, two Candida species of great biomedical interest due to their ability to cause diseases in human (Kim & Sudbery, 2011; Silva et al., 2012) although from evolutionary terms they are not closely related (Dujon, 2010). An overview about the evolutionary distance of all investigated yeast species is provided in Fig. 1, and the most important features of the investigated yeast species are summarized in Table 1.

View this table:
Table 1

Overview of the studied yeast genomes and proteomes

SpeciesStrainGenome size (Mbp)*Number of predicted proteins*Predicted secretome sizeInterestComments
Saccharomyces cerevisiae S288C12.165907156Model yeast, protein productionWhole genome duplication
Candida glabrata CBS 13812.345213121Opportunistic pathogenWhole genome duplication
Kluyveromyces lactis NRRL Y-114010.735085113Model yeast, protein production
Candida albicans WO-114.475752449PathogenMost strains are diploid; for clarity data of a haploid strain was analyzed
Hansenula polymorpha DL-18.864156Protein production, model yeast
Pichia pastoris GS1159.445033105Protein production, model yeast
Yarrowia lipolytica CLIB12220.556472299Protein production
Schizosaccharomyces pombe 972 h-12.595020112Model yeast, protein productionFission yeast
  • Data taken from http://www.ncbi.nlm.nih.gov/genome.

  • Data taken from Lum & Min (2011). The secretomes defined in that study include all proteins predicted to have a secretion signal peptide with a subcellular location predicted as extracellular, but not having a transmembrane domain or an ER targeting signal, as well as all manually curated secreted proteins. Data for H. polymorpha are not available.

  • Data derived from another, diploid strain of C. albicans.

Figure 1

Phylogenetic tree of the studied yeasts. Common taxonomy tree depicting the evolutionary relationship of the analyzed yeast species [adapted from Dujon (2010)]: branch lengths are not prop-ortional to evolutionary distances. The color code resembles the colors in Table S1.

As the nomenclature of yeasts is currently in a state of flux with several renamings in a row during the last years, as well as disputable splittings of species, we chose to use the established species names so that H. polymorpha is synonymous to both Pichia angusta and Ogataea parapolymorpha, and P. pastoris is synonymous to both Komagataella pastoris and K. phaffi.

The published genome sequences of the investigated yeasts were the basis of the comparison described here. The primary sources were the published genome sequences of C. albicans (Butler et al., 2009) and C. glabrata (Dujon et al., 2004), H. polymorpha (Ramezani-Rad et al., 2003), K. lactis (Dujon et al., 2004), P. pastoris (De Schutter et al., 2009; Mattanovich et al., 2009; Kuberl et al., 2011), and Y. lipolytica (Dujon et al., 2004), preferentially using the reference protein sequences in NCBI. Based on gene ontology, we selected S. cerevisiae genes annotated to ER, protein folding, glycosylation, ERAD, Golgi, soluble N-ethylmaleimide sensitive factor attachment protein receptors (SNAREs) and vesicle-mediated transport, to identify their homologs in the investigated genomes by reciprocal blast search, which was backed up by manual curation. Additionally all genes without homology to S. cerevisiae were searched for domains linked with functions in the secretory pathway using a HMMER search against the respective domain profiles in the protein family database Pfam. Signal peptides were predicted using SignalP, whereas TMHMM was employed to predict transmembrane domains.

In the following, we provide an overview of the consensus secretion pathway in yeast (Fig. 2), differentiated from higher eukaryotes, and scrutinize it in comparison with the genome sequences (Supporting Information, Table S1), and published literature of the above-mentioned yeasts.

Figure 2

Overview of the canonical protein secretion pathway in yeasts. During or after synthesis in the cytosol and eventual binding of cytosolic chaperones proteins are translocated to the ER lumen where they may by glycosylated. Folding is assisted by chaperones and enzymes. Misfolded proteins become target of the ER-associated degradation pathway which ends in proteasomal decay. Folded proteins are transported by COPII vesicles to the Golgi while COPI vesicles are responsible for retro-transport to cis-Golgi and ER. While most proteins that enter the secretion pathway remain in the cell, those destined for the exterior are transported by secretory vesicles to the cell membrane. Inserts: Fluorescence micrographs of Pichia pastoris cells illustrate the native relations inside yeast cells. Lower image: ER colored in green, and COPII vesicles in red. Upper image: ER in red, and Golgi in green.

Protein translocation across the ER membrane

The initial step of secretion is the transfer of a protein through the ER membrane. Translocation of proteins into the ER can occur either co-translationally (ribosome-coupled) where translation and translocation are directly linked, or post-translationally (ribosome-uncoupled) depending on the hydrophobicity and amino acid composition of the already fully translated signal peptide (Zimmermann et al., 2011). Our common knowledge is that both routes use the same translocation channel, the Sec61 complex, in combination with different channel partners. However, this is just a simplified point of view, as detailed later (Fig. 3).

Figure 3

Translocation. Co-translational translocation: while the synthesis of the protein on the ribosome is still proceeding, the signal sequence of the protein is recognized by the signal recognition particle (SRP). The ribosome-protein-SRP complex gets bound to the SRP receptor on the ER membrane. In the following, the nascent protein is inserted into either the Sec61 or the Ssh1 translocation complex that pushes it through the ER membrane. Post-translational translocation: The translocation of the nascent protein through the Sec61 translocation pore into the ER occurs after the process of translation has been completed in the cytosol. Several cytosolic and luminal proteins assist this process by stabilizing the unfolded polypeptide chain. Proteins which were not identified in all yeast genomes studied here are depicted in red.

Translocation mechanisms in S. cerevisiae were studied by several substrates relying either strictly on co-translational translocation (vacuolar dipeptidyl aminopeptidase B, secreted invertase) or on post-translational translocation (secreted alpha mating factor, vacuolar carboxypeptidase Y – CPY). On the other hand, the ER and Golgi resident proteins Kar2, Och1 or Ost1 were shown to be transported by either pathway. This double-tracked behavior might have evolved due to their important and often essential cellular functions. In this respect, it is tempting to speculate that the requirement of the signal recognition particle (SRP) for viability in Y. lipolytica and S. pombe (Brennwald et al., 1988; He et al., 1990) is due to the fact that one or more essential proteins of the secretory pathway has not evolved to use also the SRP-independent translocation mechanism. The preference for either pathway in S. cerevisiae was attributed to the hydrophobicity of the N-terminal signal peptide, which influences binding to the SRP (Ng et al., 1996), while for Y. lipolytica, the amino acid composition and conformation of the signal peptide seem to be the determinants of interaction with the SRP (Yaver et al., 1992).

Sec61 and Ssh1 – the translocon pore complexes

Saccharomyces cerevisiae contains two different translocon pores, the Sec61 and the Ssh1 complex. The Sec61 translocon pore consists of three protein subunits, Sec61, Sbh1, and Sss1 (Osborne et al., 2005), whereof Sec61 was shown to be the largest, spanning the ER membrane 10 times (Wilkinson et al., 1996; Rapoport, 2007). Sbh1 and Sss1 are single spanning membrane proteins and belong to the family of tail-anchored proteins. Sss1 is located at the channel periphery, where it is in contact with both halves of Sec61 and holds them together (Wilkinson et al., 2010). Sec61 as well as Sss1 show high sequence conservation and both subunits were shown to be essential for protein translocation and cell survival. Contrary, Sbh1 is not essential for protein translocation in S. cerevisiae (Rapoport, 2007). The translocon pore is most probably formed by dimers or trimers of the Sec61 complex. During post-translational translocation, Sec61 is part of the heptameric SEC complex (Sec61 complex plus Sec62, Sec63, Sec71, and Sec72), while the formation of the hexameric SEC′ complex, which consists of the Sec61 complex and Sec63, Sec71, and Sec72, is required for co-translational translocation (Jermy et al., 2006).

Additionally, the heterotrimeric Ssh1 (Sec sixty-one homolog) complex is present, which consists of the non-essential Sec61 homolog Ssh1, the Sbh1 homolog Sbh2 and the Sss1 subunit, which is shared between the two complexes (Finke et al., 1996). Despite being identified more than 15 years ago, the biological function and relevance of the Ssh1 complex remained rather undiscovered and undiscussed. Both Sec61 and Ssh1 complexes were shown to interact with the ribosomes (more precisely, the 28S ribosomal RNA of the large subunit) through cytosolic domains of Sec61 or Ssh1, respectively (Prinz et al., 2000). Initially the Ssh1 complex was thought to act primarily in co-translational translocation, as it does not interact with the Sec62/Sec63 complex (Finke et al., 1996). The observed defects in ERAD and translocation of the post-translational substrate CPY in a Δssh1-mutant might be explained by Ssh1-dependent translocation of luminal ER proteins required for correct CPY processing, or more likely due to an overload of the remaining Sec61 pore (Wilkinson et al., 2001). The authors showed that approximately 50–60% of analyzed co-translational substrates are indeed targeted to the Ssh1 complex under normal growth conditions, which are re-routed to the Sec61 complex when components of the Ssh1 complex are missing. As both complexes have approximately equal abundance (Finke et al., 1996), co-translational transport through the Sec61 complex might compete with post-translation translocation in the absence of the Ssh1-channel. This partially overlapping function was also confirmed by the notion that deletion of Ssh1 in sec61 mutants leads to a severely growth-impaired synthetic phenotype (Finke et al., 1996; Jiang et al., 2008). While the Sec61 complex accepts a wide spectrum of signal sequences, interactions with the Ssh1 complex are limited to more hydrophobic signal sequences including those of Kar2 and invertase (Wittke et al., 2002). Very recently, Spiller & Stirling (2011) have identified the first target protein (Sec71, an ER transmembrane protein) that is specifically directed to the Ssh1 complex.

According to our sequence analysis, all analyzed yeasts except S. cerevisiae and C. glabrata [which both underwent whole-genome duplication (WGD)] contain only a single homolog of the translocon subunit Sbh1/2, indicating that in pre-WGD species the two translocon pores share both their beta (Sbh) and gamma (Sss1) subunits and only differ in their selective alpha subunit (Sec61 or Ssh1, respectively). Most probably, this renders the single Sbh homolog essential for viability in these yeasts; however, implications on co- and post-translational transport remain to be studied.

Despite being essential for viability, the homologs of S. pombe and C. albicans Sec61 failed to complement the respective mutant in S. cerevisiae, which was explained by non-conserved amino acid substitutions within the cytoplasmic loop between TM helices 4 and 5 (Broughton et al., 1997; de la Rosa et al., 2004) indicating that cytosolic interactions are essential for Sec61 function. Based on our sequence alignment, these substitutions do not occur in the Sec61 proteins of the other analyzed species, which should therefore be able to complement each other.

Notably, the Sec61 homologs Ssh1 of Y. lipolytica and S. pombe do not align together with the Ssh1 protein cluster, but rather with the Sec61 cluster, again raising the question regarding functional complementation.

Co-translational translocation

Co-translational translocation into the ER lumen of S. cerevisiae involves the interplay of the nascent protein, the ribosome, the signal recognition particle (SRP), the signal recognition receptor (SR) and either the Ssh1 or the Sec61 translocon pore. SRP is a complex of six proteins (Srp14, Srp21, Srp54, Srp68, Srp72, and Sec65) and a 7S single RNA (SCR1) (Siegel & Walter, 1988; Brown et al., 1994). All protein complex members, except Srp54, assemble in the nucleus with the RNA, which provides a scaffold for the SRP. Once exported to the cytosol, the complex binds the residual subunit Srp54 (Ciufo & Brown, 2000). In turn, Srp54 recognizes and binds to the signal sequence of the newly synthesized polypeptide which emerges from the ribosome and is presented by the ribosome nascent chain complex (RNC), resulting in the SRP-RNC complex and GTP binding to Srp54. This interaction causes an arrest or pausing of translation, until the ribosome is bound to the translocon pore.

Co-translational protein import into the ER is regulated by the interplay of three GTPases, the SRP subunit Srp54 and the two subunits of the SR. GTP-containing Srp54 interacts with the α subunit of the SR, building a GTP-stabilized complex, which targets the SRP-RNC complex to the ER membrane. Association of the α subunit (Srp101) and the β subunit (Srp102) of SR is regulated by GTP binding to the membrane-bound β subunit. The Sbh1 or Sbh2 subunit of the respective translocon complex was shown to act as GEF (guanine nucleotide exchange factor) for SRβ in vitro (Helmers et al., 2003); however, it remains a point for discussion whether they also activate SR association in vivo (Jiang et al., 2008). When all three GTPases are in their GTP-bound state, the signal peptide is transferred to the translocon pore. After GTP hydrolysis, the SRP-SR complex dissociates from RNC, which remains bound to the Sec61 translocon (Wild et al., 2004). Deletion of any of SRP complex members leads to slow growth but not to cell death in S. cerevisiae. This is an indication that the complete SRP complex is not essential for co-translational translocation in S. cerevisiae (Brown et al., 1994), which is contrary to Y. lipolytica and S. pombe (Brennwald et al., 1988; He et al., 1990). So far, no information about the essentiality of the SRP is available for the two Candida species, H. polymorpha, and P. pastoris.

In our analysis, homologs of the S. cerevisiae protein Srp21, which is part of the SRP in the co-translational translocation, could not be found in S. pombe. Comparison of the S. cerevisiae Srp14 resulted in a very low sequence identity to the same protein of S. pombe, but the Srp14 domain could be identified. The lack of identification of orthologues of these two proteins in the genome of C. albicans previously (Braun et al., 2005) can be explained by their short sequence, as we can clearly identify Srp14 and Srp 21 in the genome of C. albicans.

Post-translational translocation

Post-translational translocation (SRP-independent) has been shown to occur in all kinds of organisms including mammalian cells, albeit to a different extent. In fast-growing organisms like yeast and bacteria, it is thought that translocation may not always keep pace with translation (Rapoport, 2007), while in higher eukaryotes, the post-translational mechanisms was initially thought to be a salvage pathway. Contrary to this assumption, distinct substrates for post-translational transport have recently been identified also in mammalian cells (Lakkaraju et al., 2012). Different preferences for co- and post-translational translocation have also been observed between different yeasts. In S. cerevisiae, approximately 30% of Sec61 complex is ribosome-associated, whereas in Y. lipolytica more than 70% of the complex was found associated with translating ribosomes (Boisramé et al., 1998).

After the release of the polypeptides from the ribosome, they have to remain in an unfolded or loosely folded form and prevented from aggregation prior to translocation. This is accomplished by binding to cytosolic chaperones Ssa1 and Ydj1, which are spontaneously released just before transport into the ER (Plath & Rapoport, 2000; Willer et al., 2003).

The heptameric SEC complex, which comprises of Sec61, Sbh1, Sss1, Sec62, Sec63, Sec71, and Sec72, is required for post-translational translocation across the ER membrane. While Sec62 and Sec63 are essential for cell survival, Sec71 and Sec72 (also known as Sec66 and Sec67) are not. Initially, the four additional members of the SEC complex have been identified as a separate tetrameric Sec62/Sec63 transmembrane complex (Brodsky & Schekman, 1993) that was shown to interact with the Sec61 complex. Instead of the SRP, Sec62 recognizes and binds to the signal peptide of nascent proteins destined for post-translational translocation. Binding of the alpha factor signal sequence takes place at Sec61 and Sec62 simultaneously and also involves Sec72 (Plath et al., 2004).

The ATPase activity of the ER luminal chaperone Kar2 was suggested to be the driving force of post-translational translocation, ‘pulling’ the nascent protein into the ER by a ‘ratcheting mechanism’ (Matlack et al., 1999).

Chaperones involved in translocation

The roles of Sec63 and Kar2 in co- and post-translational translocation have long been a point of discussion. It is now established that Sec63 fulfills at least three different roles. On one hand, Sec63 is binding Sec62 through its C-terminal acidic region, thus stabilizing the heptameric SEC complex, which is exclusive to the post-translational transport (Brodsky et al., 1995). On the other hand, the ER-located J domain of Sec63 is assumed to cooperate with Kar2 in gating the translocon pore, which is common to both pathways, but needs to be proven in vivo in yeast. Moreover, a role of Sec63 in assembly of the heptameric SEC and the hexameric SEC′ translocon complexes has been demonstrated, which is attributed to its cytosolic Brl domain (Jermy et al., 2006). Recently, Sec63 was also shown to be involved in co-translational transport via the Ssh1 pore (Spiller & Stirling, 2011).

Kar2 is also required in both processes (Brodsky et al., 1995; Young et al., 2001), yet the probable function as gate keeper of the translocon demonstrated for the mammalian Kar2 homolog BiP remains to be confirmed (Hamman et al., 1998). The ratcheting mechanism that pulls the nascent polypeptide into the ER involves not only Kar2, but also its co-chaperones Lhs1 and Sil1, with Lhs1 probably being the preferred nucleotide exchange factor (NEF) during the translocation process, as Δlhs1 but not Δsil1 are defective in translocation (Steel et al., 2004). Both Kar2 and Lhs1 function as molecular chaperones, and are assumed to bind sequentially to different regions of the same unfolded polypeptide in a coordinated manner, which is based on the reciprocal stimulation of their respective ATPase activities (de Keyzer et al., 2009).

The role of Sil1 in translocation during normal growth conditions is slightly unclear, as Sil1 binds to a different region on Kar2, and cannot bind substrates itself (Hale et al., 2010). In vitro, Sil1 promotes Kar2 recruitment by Sec63 and Sec63-mediated activation of Kar2 ATPase activity (Kabani et al., 2000). The observation that Δsil1 in S. cerevisiae is mainly defective in ERAD is in contrast to the situation observed in Y. lipolytica. The Y. lipolytica homolog Sls1 was co-immunoprecipitated with Kar2 and the ribosome-associated Sec61 complex thus indicating an involvement of Sls1 in co-translational translocation. Moreover, deletion of Sls1 had a strong impact on translocation of nascent secretory proteins (Boisramé et al., 1998). In S. cerevisiae, mutants with individual deletions of either Lhs1 or Sil1 are viable but exhibit UPR induction, while double deletion of these two genes is lethal (Tyson & Stirling, 2000). A homolog of Sil1 is missing in H. polymorpha and S. pombe; however, we do not assume constitutive UPR induction in these organisms.

Additionally, several uncharacterized ER membrane proteins that were reported to be involved in translocation in S. cerevisiae are either missing in all non-WGD species (Frt1 and Frt2) or present in less isoforms (Yet1-3), making their role in translocation more speculative or indicating that they might only be required for specific translocation substrates present in the close relatives of Saccharomyces.

Protein folding and maturation in the ER

Machineries of molecular chaperones: Hsp70s, J proteins and beyond

Molecular chaperones are present in all cellular compartments where de novo protein folding occurs (in yeast: cytosol, ER, and mitochondria). Each compartment has its distinct folding machinery in terms of localization; however, during several cellular processes such as translocation, chaperones from more than one compartment are involved. For sake of completeness also mitochondrial chaperones are listed in Tables 2 and S2, but are not discussed in the following. Cytosolic chaperones are covered by our review as they have been implicated also in several secretion-related processes.

View this table:
Table 2

J protein family

S. cerevisiae Description C. glabrata K. lactis C. albicans H. polymorpha P. pastoris Y. lipolytica S. pombe
Ydj1 (410 aa)Type I, cytosolXP_448143 (407 aa)XP_455231 (409 aa)EEQ43519 (393 aa)EFW96437 (402 aa)XP_002492146 (402 aa)XP_504839 (417 aa)NP_595428; MAS5 (407 aa)
Ydj1 (410 aa) Xdj1 (460 aa) Type I, cytosolXP_449302 (452 aa)XP_455941 (512 aa)EEQ46086 (439 aa)EFW95253 (435 aa)XP_002492708 (409 aa)XP_502347 (411 aa)NP_596309; XDJ1 (413 aa)
Apj1 (529 aa)Type I, cytosolXP_448159 (479 aa)XP_454306 (495 aa)EEQ43876 (539 aa)
Sis1 (353 aa)Type II, cytosolXP_446955 (349 aa)XP_453274 (354 aa)EEQ43679 (343 aa)EFW95876 (337 aa)XP_002492055 (346 aa)XP_503904 (368 aa)NP_588477; PSI1 (379 aa)
Djp1 (433 aa)/Caj1 (392 aa)Type II, DnaJ-X superfamily, cytosol XP_447789 (425 aa) XP_448815 (373 aa) XP_453663 (433 aa) XP_454145 (428 aa) EEQ42184 (508 aa) EEQ44942 (459 aa) EFW95570 (425 aa) EFW96294 (432 aa) XP_002492123 (474 aa) XP_002493650 (417 aa) XP_505317 (476 aa) NP_596098; YHXB (392 aa) NP_594337; YAY1 (355 aa)
Zuo1 (434 aa)Type III, cytosol, RACXP_446371 (433 aa)XP_455221 (446 aa)EEQ45197 (427 aa)EFW96419 (421 aa)XP_002491176 (446 aa)XP_499622 (429 aa)NP_596284; YOI1 (442 aa)
Swa2 (668 aa)Type III, cytosolXP_447105 (661 aa)XP_451983 (621 aa)EEQ43843 (773 aa)EFW97525 (568 aa)XP_002494187 (681 aa)XP_500799 (915 aa)NP_593480; UCP7 (697 aa)
Jjj1 (591 aa)Type III, cytosolXP_448027 (623 aa)XP_453932 (620 aa)EEQ43422 (576 aa)EFW98047 (530 aa)XP_002490651 (532 aa)XP_500925 (524 aa)NP_593763; MU185 (380 aa)
Jjj2 (584 aa)Type III, cytosolXP_447521* (455 aa)XP_451854* (653 aa)
Jjj3 (173 aa) type III, cytosol CSL-Zn finger XP_445373 (175 aa)XP_452330 (162 aa)EEQ45554 (149 aa)XP_002491742 (184 aa)XP_503631 (163aa)NP_594366; DPH4 (139 aa)
Cwc23 (284aa)Type III, nucleus, cytosolXP_447661* (296 aa)XP_452310* (270 aa)
Cwc23?? (284aa)DNAJ-RNA recognition motif (RRM)EEQ45013 (278 aa)XP_002489427 (274 aa)XP_504419 (280 aa)NP_587857; CWF23 (289 aa)
DNAJ (N-term)EEQ44873 (274 aa)EFW94367 (ca 300 aa)XP_002493845 (282 aa)XP_504032 (253 aa)NP_594359; YL39 (282 aa)
DNAJ (N-term)EEQ43274 (245 aa)EFW9815 (216 aa)XP_002491043 (201 aa)XP_503111 (204 aa)NP_595422; SPF31 (209 aa)
DNAJ (N-term)EFW95505 (470 aa)nuclear XP_002492134 (526 aa) cyto/mito
DNAJ (N-term), C–term TMEEQ46988§ (287 aa) TM?XP_002492479§ (303 aa)1 TMXP_505529§ (239 aa) 1 TM
DNAJ (N-term)XP_505671 (751 aa)
TRP–repeats (N-term) DNAJ (C-term)XP_504393 (488 aa) TRPNP_596790; DNJC7 (476 aa)
DNAJ (N-term)EEQ42297i (730 aa) nuclear
DUF3395, DNAJC11-superfamilyNP_587977** YCJ3 (642 aa)
DNAJ (N-term)NP_595124†† MU184 (551 aa)
DNAJ (N-term)NP_587986; YCJD1 (208aa) 1 TM
DNAJ (N-term)NP_595625‡‡ RSP1 (494 aa)
Mdj1 (512 aa)Type I, mitochondriaXP_447889 (522 aa)XP_453404 (471 aa)EEQ46384 (488 aa)EFW96456 (ca 500 aa)XP_002491863 (492 aa)XP_505333 (473 aa) NP_587824 MDJ1 (528 aa)
Mdj2 (146 aa)Type III, mitochondriaXP_445526 (143 aa)XP_454039 (146 aa)EEQ45952 (143 aa)EFW97058 (partial)XP_002491618 (133 aa)XP_501988 (110 aa)
Pam18 (169 aa)Type III, mitochondria XP_447753 (153 aa) 1 TM XP_454377 (163 aa) hydrophobic region Q59SI2 (157 aa) hydrophobic region EFW98124 (153 aa) hydrophobic region XP_002491775 (137 aa) hydrophobic region XP_500822 (148 aa) hydrophobic region NP_593445 (140 aa) SP, hydrophobic region
Jac1 (185 aa)Type III, mitochondriaXP_444785 (198 aa)XP_451212(190 aa)EEQ45481 (209 aa)EFW94312 (198 aa)XP_002493776 (215 aa)XP_502231 (203 aa) NP_594669 JAC1 (190 aa)
Jid1 (302 aa) 1 TM Type III, mitochondria (inner membrane, matrix) XP_445484 (300 aa) no TM XP_451905 (295 aa) 1 TM EEQ42088 EFW94416 (232 aa) 1 TM XP_002493458 (313 aa) 1 TM XP_504139 (361 aa) 1 TM NP_593700 YDJ1 (270 aa) 1 TM
Scj1 (378 aa) SP, KDEL Type I, ER XP_446132 (361 aa) SP, DDEL XP_452522 (368 aa) SP, DDEL XP_716745 (384 aa) SP, HDEL EFW97145 (ca 350 aa) KDE XP_002489518 (354 aa) SP, KDEL XP_503772 (361 aa) SP, RDEL NP_596697; SPJ1 (398 aa) SP, FDEL
Hlj1 (224 aa) 1 TM Type II, ER membrane, cytosolic J domain XP_446990 (232 aa) XP_454021 (231 aa), 1 TM EEQ47172 (331 aa) no TM, DUF1977 EFW96005 (330 aa), 2 TM, DUF1977 XP_002489727 (318 aa), no TM, DUF1977 XP_502307 (340aa), 1 TM NP_595587; HLJ1 (403 aa) 1 TM
Erj5 (296 aa) SP, 1 TM Type III, ER membrane, ER luminal J domain XP_445268 (297 aa) SP, 1 TM XP_451049 (277 aa) SP, 1 TM EEQ43604 (300 aa) SP, 1 TM EFW98152 (323 aa) SP, 1 TM XP_002489475 (299 aa) SP, 1 TM XP_505307 (287 aa) SP, 1 TM NP_594141; YKU3 (303 aa) SP, 1 TM
Sec63 (664 aa) 3 TM (1 Signal-TM) Type III, ER membrane, ER luminal J domain XP_446686 (668 aa) 3 TM (1 Signal-TM) XP_452636 (669 aa) 3 TM (1 Signal-TM) EEQ43507 (673 aa) 4 TM (1 Signal-TM) EFW95938 (670 aa) 3 TM (1 Signal-TM) XP_002493824 (664 aa) 3 TM (1 Signal-TM) XP_500184 (649 aa) 3 TM (1 Signal-TM) NP_595985; SEC63 (611 aa) 3 TM (1 Signal-TM)
Jem1 (646 aa) SP Type III, ER lumenXP_453346* (631 aa) SP
Jem1? (646 aa) SP Type III, ER lumenEFW96042 (ca 600 aa) XP_002492202 (625 aa) SP
Jem1? (646 aa) SP SP-TRP-DnaJ XP_446917 (656 aa) SP EEQ47388 (639 aa) SP XP_504057 (579 aa) SP
ER J domain (predicted) XP_501490 (384 aa) 3 TM (1 Signal-TM)
  • TRP, tetratricopeptide repeat domain.

  • Homologs found in Ashbya gossypii, Saccharomyces spp., Zygosaccharomyces rouxii, C. glabrata, Naumovozyma castellii, Lachancea spp.

  • Homologs found in Trichoderma spp., Aspergillus spp., Penicillium spp., Neurospora crassa, Candida spp., Cryptococcus spp., Pichia spp., Schizosaccharomyces spp.

  • No other homologs.

  • Homologs found in Debaryomyces hansenii, Candida spp., Pichia stipitis, Pichia guilllermondii.

  • Homologs found in Trichoderma spp., Aspergillus spp., Penicillium spp., Neurospora crassa.

  • Homologs found in Rhodutorula spp., Cryptococcus spp., Rhizopus spp., mammals.

  • No homolog found.

  • Homologs found only in S. pombe and Schizosaccharomyces japonicus.

Molecular chaperones of the heat shock protein 70 kDa (Hsp70) family are the key components of the chaperone network and are responsible for protein folding, protein degradation, protein translocation, and protein-protein interactions. Together with their co-chaperones Hsp70s assist in proper folding, prevent misfolding and aggregation, refold aggregated proteins, aid in translocation into ER and mitochondria, and prepare terminally misfolded proteins for degradation (Kampinga & Craig, 2010).

Folding is accomplished in an ATP-driven cycle of substrate binding and substrate release. While Hsp40/DnaJ co-chaperones stimulate ATP hydrolysis and stabilize substrate binding in doing so, nucleotide exchange factors (NEF) are required for exchanging ADP against ATP and subsequent substrate release, thereby recycling Hsp70 to its ‘interaction-competent’ form for new substrate binding.

According to the current model, the binding of unfolded polypeptide substrates starts with Hsp40/DnaJ co-chaperones recognizing their unfolded hydrophobic regions. The substrate is then handed over to the Hsp70 partner chaperone, in a manner involving ATP hydrolysis. In its ATP-bound state, unfolded substrates rapidly associate and dissociate from Hsp70. In its ADP-bound form, Hsp70 exhibits stabilized substrate binding until nucleotide exchange triggers substrate release. If the substrate has not attained its native form, repeated cycles of substrate binding and release are performed.

The sheer number of Hsp70s is rather limited in relation to the manifold functions they modulate in the cell. In S. cerevisiae, there are two types of cytosolic canonical Hsp70 chaperones (Ssa1-4 and Ssb1-2) and one type of canonical Hsp70 in the ER (Kar2). While Ssa proteins are the main cytosolic Hsp70s in all eukaryotes, fungi also possess the non-essential Hsp70 isoform Ssb. Together with the co-chaperones Ssz1 and Zuo1, Ssb1/2 build the ribosome-associated RAC complex required to chaperone nascent polypeptides upon their exit from the ribosome (Gautschi et al., 2001).

Further, Hsp70-related proteins with a C-terminal extension act as NEF for their chaperone activity, either belonging to the Hsp110 subfamily (Sse1-2 in the cytosol) or to the Hsp170 subfamily (Lhs1 in the ER). Additional NEFs with a totally unrelated structure exist in each compartment: the GrpE-like Fes1 in the cytosol and Sil1 in the ER (Kabani et al., 2000, 2002), as well as the Bag-like Snl1 in the cytosol (Sondermann et al., 2002). Despite Snl1 being tethered to the ER membrane, Snl1 and Fes1 do not seem to be involved in translocation or ERAD, but rather have a role in translation as indicated by their association with the ribosome (Verghese & Morano, 2012).

Reaction specificity of Hsp70 function is in most cases generated by the respective Hsp40/DnaJ co-chaperone. Besides their well-conserved J domain, Hsp40/DnaJ co-chaperones are quite divergent especially regarding their size and will thus be called J proteins in the following. The 70 amino acid J domain consists of four helices and a loop between helix 2 and 3 containing the highly conserved tripeptide histidine-proline-aspartate (HPD-motif) and is in most cases sufficient for stimulating ATPase activity of the corresponding Hsp70 partners (Sahi & Craig, 2007). In total, 22 J proteins are found in S. cerevisiae. Cytosolic localization is reported for 12 of them, whereas five each are connected to the ER or to the mitochondria, respectively (Table 2). J proteins are further divided into three subclasses, with Type I and Type II possessing an N-terminal J domain followed by a glycine/phenylalanine (G/F) rich region and an optional zinc finger-like domain in type I, while the J domain in type III J proteins can be anywhere in the protein. Classification of yeast J proteins is detailed in the study by Walsh et al., 2004, and will thus not be repeated here.

Some co-chaperones exhibit chaperone activity on their own, as reported, for example, for Ydj1. This has been attributed to their C-terminal client binding domains (G/F region and Zn-finger). Other J proteins mainly act in recruitment of Hsp70 to a special site of action, as demonstrated for the cytosolic J domain of the ER membrane protein Hlj1 that positions cytosolic Ssa to the cytosolic surface of the ER membrane during ERAD (Huyer et al., 2004). Usually, one Hsp70 interacts with a number of different J proteins to drive distinct cellular processes. Ssa1 reacts with all cytosolic J proteins except Zuo1, which is the only J family member described to interact with Ssb and Ssz1. In many cases, the fate of misfolded proteins is determined by the interacting J protein, not Hsp70.

A role for cytosolic Ssa and J proteins in aiding post-translational translocation of secretory precursors into the ER (and mitochondria) is proposed. Ngosuwan et al. (2003) confirmed that Ssa1 is required to maintain prepro-α factor in a translocation competent state, but could not actively refold aggregated prepro-α factor in vitro. This result has to be interpreted with caution as the in vitro experiments have been carried out without a NEF, and in vivo Ssa1-dependent refolding of aggregates of several non-native proteins has been reported (McClellan et al., 2005; Park et al., 2007). ATPase activity and/or ATP binding of Ssa1 is required for efficient translocation, but not needed to prevent aggregation. This is in line with the observation that interaction with unfolded substrate occurs in Ssa mutants defective in nucleotide binding (thus preventing aggregation of hydrophobic peptide stretches), however, subsequent substrate release is impaired. The main cytosolic J protein Ydj1 is not essential for stimulating translocation, but is able to prevent aggregation of prepro-α factor to maintain its translocation competence (Ngosuwan et al., 2003). Ydj1 is also required for refolding of denatured luciferase in cooperation with Ssa1 and the NEF Sse1/2 (Dragovic et al., 2006).

While Ssa is not required for the degradation of soluble secretory proteins such as mutant carboxypeptidase Y (CPY*) or α mating factor via ERAD (Taxis et al., 2003), ubiquitination of transmembrane ERAD substrates is defective in a ssa mutant (Han et al., 2007). Therefore, a role of Ssa and Ydj1 in recognition of misfolded protein domains is proposed, in order to keep misfolded proteins in a soluble state, and escort/deliver them to the ubiquitination machinery. Requirements of other co-chaperones to this process is not clear and may depend on the nature of misfolded protein [e.g. Hlj1, Cwc23, Jid1 and Hsp104 (Taxis et al., 2003); Ydj1 and Hlj1 during ERAD of membrane proteins (Huyer et al., 2004; Youker et al., 2004); Sti1 and Hsp90 (McClellan et al., 2005)].

Specialized functions of cytosolic J proteins in S. cerevisiae have been summarized by Sahi & Craig (2007). With respect to secretion, the role of the auxilin homolog Swa2 in uncoating clathrin-coated secretory vesicles (CCV) should be highlighted. Here, both Swa2 and Ssa1 interact with a correctly folded multimeric protein, and facilitate destabilization instead of promoting folding. The total number of J proteins in the different analyzed yeasts varies between 20 and 23, but their sequences are partly quite diverse. Altogether, we found 12 J proteins without homologs in S. cerevisiae whose functions remain to be identified (Table 2). Interestingly, the above-discussed J proteins as well as most of the ER-resident J proteins are conserved, thus highlighting their crucial involvement in the secretion process.

In addition to Hsp70 and J proteins, also Hsp90s, Hsp104, and small heat shock proteins are present in the cytosol. Up to now, Hsp90 and its associated co-chaperones (among others Sba1 and Sti1) have mainly been connected to folding of specific client proteins such as kinases and transcriptional regulators; however, Hsp90 cellular functions are still poorly understood. Large-scale screening approaches revealed a connection to secretion. By regulating the folding of several GTP-binding proteins, Hsp90 is linked to regulation of the secretory pathway. Additionally, physical and genetic interactions of Hsp90 with several multi-subunit components of the vesicular transport and protein trafficking pathway (e.g. TRAPP, COG, complexes involved in Golgi to MVB transport) have been established (McClellan et al., 2007). The authors speculate that the function of Hsp90 is to stabilize subunits during the controlled assembly and disassembly of these multidomain complexes. Hsp104 binds to the Ssa1p-Ydj1p complex in an ATP-dependent manner.

Contrary to S. cerevisiae having a stress-inducible and a constitutive isoform of each Hsp70 chaperone, all other analyzed yeasts contain just one isoform for Sse and Ssb chaperones, which probably have to fulfill the function also during cellular stress (see Table S2). Interestingly, overexpression of Sse1 was reported to be toxic in S. cerevisiae leading to retarded growth (Shaner et al., 2004), whereas no effect on cellular growth was observed when overexpressing this protein in P. pastoris (Gasser et al., 2007).

The number of Ssa′s varies between two (usually one being more similar to Ssa1/2 and the other having closer similarity to Ssa3/4) and four Ssa in Y. lipolytica. For S. cerevisiae, the presence of a single Ssa protein is sufficient for survival, confirming the high degree of functional overlap among them. In this line, also only one isoform of Hsp90 is present in the other yeasts, compared to Hsp82 and Hsc82 in S. cerevisiae.

Endoplasmic reticulum-resident chaperones

Endoplasmic reticulum-resident Kar2 and Lhs1 are characterized by a N-terminal signal peptide (present in all analyzed species) and a C-terminal ER retention signal (HDEL in S. cerevisiae, C. albicans, H. polymorpha, P. pastoris, and Y. lipolytica, but more divergent sequences in C. glabrata, K. lactis and S. pombe). Neither Hsp90s nor small Hsps with a predicted ER-localization can be found, contrary to mammalian cells.

In S. cerevisiae, there are five ER-resident J domain containing proteins, three of which are localized to the ER membrane (Sec63p, Hlj1, and Erj5) and two soluble luminal J domain chaperones (Scj1, Jem1). All of them (except Hlj1) cooperate with the ER Hsp70 chaperone Kar2, regulating its ATPase activity during distinct processes. Kar2 is participating in protein translocation across the ER membrane, protein folding in the ER lumen as well as targeting misfolded proteins to ERAD. While Sec63 is essential for the co- and post-translational translocation of nascent proteins into the ER (Brodsky et al., 1995), the other four ER Hsp40 co-chaperones are non-essential. Scj1 and Jem1 are both involved in keeping misfolded substrates soluble prior to re-translocation (Nishikawa et al., 2001). Additionally Scj1 assists Kar2 to fold and assemble proteins in the ER lumen (Silberstein et al., 1998), whereas Jem1 cooperates with Kar2 during mating to aid in nuclear membrane fusion. Jem1 has been reported to be membrane associated, and is proposed to recruit Kar2 to the ER surface of the membrane, while Hlj1 fulfills a similar function on the cytosolic side of the ER membrane. Erj5 is neither involved in translocation nor ERAD of the soluble substrate CPY*, but displays a functional overlap with Scj1 and Jem1. Although the function of Erj5 is not specified yet, it is assumed that loss of one luminal J protein is compensated by the transcriptional upregulation of the other ER-resident family members through UPR activation (Carla Famá et al., 2007).

Recently, Kar2 mutants defective in interaction with one distinct J protein have been identified (Vembar et al., 2010), which were also defective in distinct Kar2-mediated cellular processes (translocation, folding or ERAD, respectively). Also binding of the two Kar2 NEFs Sil1 and Lhs1 is mutually exclusive and involves different domains of Kar2. Thus, Kar2 mutants further contribute to our detailed understanding of co-chaperone requirement during certain processes.

Y. lipolytica contains an additional J protein with a predicted topology like Sec63, indicating the J domain to be ER-localized, but having a shorter C-terminal domain. The function of this protein is yet unknown (Table 2).

Interestingly, while most yeasts possess a Jem1-like protein, these proteins do not show sequence similarity outside of the Jem1-like J domain. Different subclasses exist for S. cerevisiae and K. lactis, H. polymorpha and P. pastoris, and Y. lipolytica, respectively. It remains to be studied if Jem1 function is conserved among the different yeast species (Table 2). In this respect, C. albicans Jem1 has been reported to have conserved function in ER protein folding, but does not complement the defect in membrane fusion during mating (Makio et al., 2008).

Oxidative protein folding – quality control and redox balance

Protein disulfide isomerases are enzymes responsible for the formation, rearrangement, and reduction of disulfide bonds within newly synthesized secretory proteins (Fig. 4a). PDI family members are characterized by having one or more thioredoxin-like domains (trx), which can either be catalytically active (a-domains) or inactive (b-domains). PDI family proteins differ in their number and arrangement of their trx-domains and in their active site motifs. The active site is usually composed of the amino acids WCXXC, often WCGHC. Detailed mode of action is described in the very comprehensive review by Hatahet & Ruddock (2009). An overview on the Pdi family members in the studied yeasts is provided in Table 3.

View this table:
Table 3

Protein disulfide isomerase (Pdi) family*

S. cerevisiae Description C. glabrata K. lactis C. albicans H. polymorpha P. pastoris Y. lipolytica S. pombe
Pdi1 (522 aa) CGHC-CGHC, HDEL a-b-b′-a′-c XP_445001 (523 aa) CGHC-CGHC, DDEL XP_452244 (527 aa) CGHC-CGHC, QDEL XP_716953 (560 aa) CGYC-CGHC, HDEL EFW96462 (515 aa) CGHC-CGHC, HDEL XP_002494292 (471 aa) CGHC-CGHC, HDEL XP_503481 (504 aa) CGHC-CGHC, DDEL NP_592871 (492 aa) CGHC-CGHC, ADEL
Eug1 (517 aa) CLHC-CIHS, HDEL a-b-b′-a′-c No oxidation possible, but disulfide shuffling XP_446874 (533 aa) CTHS-CQHS, KDEL
Mpd1 (318 aa) CGHC, HDEL a XP_449554 (304 aa) CGYC, NDEL XP_453179 (328 aa) CGYC, QDEL EEQ46452 (299 aa) CGYC, HDEL EFW97561, partial, CGYC,? XP_002489466 (298 aa) CGHC, HDEL XP_506015 (554 aa) CGHC RDEL NP_593653 (363 aa) CGYC ?
Mpd2 (277 aa) CQHC, HDEL a XP_447587 (240 aa) CSHC, VDEL XP_455206 (260 aa) CGYC, QDEL
Eps1? (701 aa) CPHC, TMD a°-a-TMD XP_448876 (668 aa) CSYC-CYEC, TMD XP_447795 (708 aa) CSHC-CPKC, TMD XP_453687 (706 aa) CHHC-CEDC, TMD EEQ47458 (737 aa) CHHC-CEDC, TMD EFW96199 (722 aa) CSHC-CPRC, TMD XP_002491323 (634 aa) CYHC-CYSC, TMD XP_503459 (617 aa) CGHC–CPHC, TMD NP_595525 (726 aa) CGAC-CDDC, TMD
NP_594172 (636 aa) CEDC, Signal-TM, TM
NP_588507 (561 aa) SP, TM
a°-a-D (ERp38) EEQ44186 (221 aa) CRHC-CKYC ? EWF94842 (369 aa) CSHC-CGHC XP_002489806 (369 aa) CSHC-CGYC, HQEL XP_501758 (364 aa) CGHC-CGHC, RDEL NP_593584 (359 aa) CGHC-CGHC ?
a–b EEQ45235 (342 aa) CKYC, HIEL EFW96744 (284 aa) CKHC, YRDL XP_002494218 (291 aa) CNYC, LREL
  • Ref Seq number, protein length, domain sequence and – if appropriate – the ER retention signal sequence or the existence of transmembrane domains (TM) are indicated.

Figure 4

Oxidative protein folding and reduction of misfolded proteins. (a) The nascent protein is captured by the chaperone Kar2, mediating folding. Sulfhydryl groups are oxidized by one of the protein disulfide isomerases (Pdi). Reoxidation of Pdi is mediated by ER oxidoreductin (Ero1) which in turn is transferring electrons to O2, thereby generating reactive oxygen species (ROS). (b) Misfolded proteins are reduced by Pdi activity before their retrotranslocation to the cytosol for degradation. Pdi is probably reduced by the action of glutathione. The increased binding of Kar2 to unfolded protein releases the unfolded protein response (UPR) sensor Ire1 which dimerises and initiates the UPR.

In S. cerevisiae, five members of the PDI family have been identified, thereof only Pdi1 is essential for viability. Pdi1 consists of an a-b-b′-a′ arrangement, with an N-terminal ER signal peptide, and the C-terminal ER retention signal HDEL. Eug1 is a catalytically inactive Pdi1-homolog featuring two WCXXS motifs, which has probably arisen during genome duplication (Bao et al., 2000; Nørgaard et al., 2001). Mpd1 and Mpd2 are two PDI proteins with only one catalytically active a domain; however, they seem to have different functions as the midpoint potential of their active sites differs significantly (Vitu et al., 2010). Indeed, Mpd1 may be involved in disulfide bond formation of glycosylated proteins as it cooperates with calnexin (Cne1) (Kimura et al., 2005), whereas no substrates have been described for Mpd2 so far. Additionally, S. cerevisiae possesses a membrane-bound PDI family member, Eps1, which is probably responsible for the presentation of ERAD substrates to the membrane-bound degradation machinery (for TM-ERAD substrates such as mutant Pma1, but not soluble CPY*). As the deletion of Eps1 leads to induction of the UPR due to the accumulation of misfolded proteins, a more general role of Eps1 in ERAD is assumed (Wang & Chang, 2003). This was confirmed by He et al. (2005) who observed insufficient ERAD of two recombinant soluble ERAD substrates in Δeps1.

After de novo transfer of disulfide bonds to nascent proteins, PDIs in yeast are re-oxidized by ER oxidase Ero1, a flavin containing enzyme, through dithiol–disulfide exchange. Electrons are passed further to the flavin cofactor FAD, and finally to molecular oxygen, resulting in the formation of H2O2 (Sevier & Kaiser, 2008). So far, no enzymes responsible for detoxification of reactive oxygen species (ROS) have been identified in the ER of yeasts. For mammalian cells, ER-resident peroxiredoxin (Tavender & Bulleid, 2010; Zito et al., 2010) and two glutathione peroxidases (Nguyen et al., 2011) have been discovered in the last few years, which are reported to fulfill this task, and making mammalian Ero1 non-essential (Bulleid & Ellgaard, 2011; Kakihana et al., 2012). Moreover, also H2O2 was shown to be able to re-oxidize Pdi in vitro (Karala et al., 2009), and it was recently confirmed that peroxide promotes disulfide bond formation also in vivo in mammalian cells (Margittai et al., 2012). Whether this is also occurring in the ER of lower eukaryotes remains yet to be established.

Ero1 activity is believed to be regulated through the ER redox conditions (Sevier et al., 2007), which are kept in balance by the ratio of oxidized to reduced glutathione (Appenzeller-Herzog, 2011). Although the ER redox state was determined to be similar for mammalian cells and yeast (Delic et al., 2010; van Lith et al., 2011), mammalian PDI is mainly reduced, while yeast Pdi1 mainly occurs in its oxidized form. Very recently, Pdi1 has been identified to be the main regulator of Ero1 activity in yeast, with reduced Pdi1 acting as Ero1 activator, and oxidized Pdi1 as inactivator (Kim et al., 2012). Contrary to mammalian cells, Pdi1 but not Ero1 seems to respond to the relative levels of reduced and oxidized glutathione in yeasts. In this line, plain overexpression of Ero1 did not lead to hyperoxidation of the yeast ER, while Pdi1 overexpression leads to more oxidizing ER redox conditions (Delic et al., 2012). Ero1 also interacts with the other PDI family members (Vitu et al., 2010). Besides for Pdi1, oxidative folding capability was reported for Mpd1 and Mpd2, but not for Eps1 (Kimura et al., 2005). PDI disulfide reductase activity is needed to reduce disulfides in misfolded proteins prior to ERAD (Fig. 4b). While distinct PDI players for this task have been identified in mammalian cells [ERdj5; (Ushioda & Nagata, 2011)], this function has been attributed to Pdi1 in yeast (Sakoh-Nakatogawa et al., 2009). As UPR and ERAD are closely interlinked, reduction of the ER redox state during UPR seems mainly devoted to keep PDI members in the reduced state, so that they can act as reductase preparing misfolded substrates for ERAD, or as isomerase to promote folding.

While Pdi1 and Mpd1 are present in all studied yeasts, Mpd2 is restricted to S. cerevisiae, K. lactis, and C. glabrata. All yeasts contain at least one PDI protein with a transmembrane domain (TMD), which are the supposed homologs of Eps1. Schizosaccharomyces pombe possesses three predicted PDI proteins with a TMD; however, only one of them contains a CXXC active site motif (see Table 3).

Other PDI family members found in the analyzed yeasts include a homolog of Aspergillus niger tigA/Neurospora crassa ERp38 (Jeenes et al., 1997), which is well conserved from plants to filamentous fungi and also to hemiascomycete yeasts, however, was probably lost during evolution in S. cerevisiae and closely related species such as C. glabrata and K. lactis. ERp38 consists of two N-terminal TRX-domains, and the C-terminal sequence is related to human ERp29. Tremmel et al. (2007) reported that ERp38 is binding to both PDI and BiP, and is found in a complex together with other folding catalysts in the ER of N. crassa. Apart from being induced by UPR, not much is known about this class of PDI proteins. This holds also true for another PDI family member, which has only been identified in C. albicans, H. polymorpha, and P. pastoris.

Peptidyl-prolyl isomerases (PPIases): role in folding and maturation of proteins?

Three different classes of PPIases, the FK506-binding proteins (FKBP), the cyclophilins, and the parvulins, are responsible for promoting the cis-/trans-isomerization of proline peptide bonds in nascent proteins, which has been reported to be a rate-limiting step in protein folding. Saccharomyces cerevisiae contains eight cyclophilin-type PPIases and four FKBPs, which are all non-essential for viability (Dolinski et al., 1997). Only the parvulin Ess1 is indispensable for cell growth. While the PPIase repertoires of 16 fungal genomes, including S. cerevisiae, K. lactis, Y. lipolytica, and S. pombe have been investigated in detail by Pemberton (2006) and Pemberton & Kay (2005), the functions of this diverse protein family is still largely unknown.

PPIases are ubiquitous folding catalysts with implicated functions in diverse cellular processes such as transcriptional regulation, mRNA splicing, cell cycle regulation, signal transduction, and trafficking [as summarized by Göthel & Marahiel (1999) and Wang & Heitman (2005)]. Specific target proteins have been identified for some of the S. cerevisiae cytosolic and nuclear PPIases such as Ess1 (recently reviewed by Shaw (2007), Fpr1 (Alarcon & Heitman, 1997), Fpr 4 (Xiao et al., 2006), and Cpr1 (Ansari et al., 2002). For the latter, recently a role in stress resistance was proposed (Kim et al., 2010), while Cpr6 and Cpr7 are interacting with the Hsp90 chaperones through their TPR (tetratricopeptide repeat) domains (Johnson et al., 2007; Zuehlke & Johnson, 2012). The presence of two TPR containing cyclophilins is conserved among all the studied yeasts (see Table S3) with the exception of S. pombe which contains only one homolog (SpCyp5).

For an overview on PPIase family members, see Table S3. It is remarkable that six families of PPIases that are represented in the non-conventional yeasts are not present in S. cerevisiae. While some of these additional cyclophilins have been studied in S. pombe (Pemberton & Kay, 2005), we could not find any reports about identified interaction partners, thus leaving the subject wide open for further research.

All studied yeasts contain PPIase proteins within their secretory pathway, ranging from a single signal peptide containing cyclophilin in S. pombe to three ER-resident family members (one soluble and one membrane-associated cyclophilin, one FKBP) in most budding yeasts and five secretory PPIases in S. cerevisiae (Table S3). Thereof, the ER-resident PPIase Crp5 is conserved among the studied yeasts, while its homolog Cpr2 is only present in S. cerevisae. Two cyclophilins carrying transmembrane domains (Cpr4 and Cpr8) were predicted to be localized within the secretory organelles of S. cerevisiae, with Cpr4 having homologs in K. lactis and the two Candida species. An analogous putatively membrane-bound enzyme is encoded in H. polymorpha and P. pastoris. Although it is tempting to speculate that this is related to the methylotrophic lifestyle, this will need more research for verification.

Stunningly, a combined knock-out of all five ER PPIases was viable in S. cerevisiae with no indications of growth impairment. However, a role in protein folding is indicated by the fact that they are all induced by the conditions leading to protein misfolding in the ER such as heat stress and tunicamycin treatment (Dolinski et al., 1997). Contrary to higher eukaryotic systems, where several specific substrates of secretory PPIases have been identified (reviewed by Göthel & Marahiel, 1999; Christis et al., 2008; Ferreira & Orry, 2012; Gollan et al., 2012), no specific target of yeast secretory PPIases have been reported so far, and no counterparts of the mammalian target proteins are present in the fungal species. While it is speculated that they might act as chaperones, a proof is missing.

Glycosylation – controling and conferring quality

The majority of all secreted proteins are glycosylated. As a first step of N-glycosylation the oligosaccharide precursor Glc3Man9GlcNAc2 is assembled on the lipid carrier dolichyl pyrophosphate both at the cytosolic and at the luminal sides of the ER membrane. This process, essentially mediated by enzymes encoded by the ALG genes, has been extensively reviewed by Burda & Aebi (1999). The oligosaccharide precursor is then transferred by the oligosaccharyltransferase (OST) complex to the asparagine residue of the N-glycosylation recognition sequence (Asn-Xxx-Ser/Thr) of the nascent protein during translocation. O-glycosylation on the other hand begins with the transfer of a mannose residue from dolichyl phosphate to a serine or threonine residue by protein O-mannosyl transferase (PMT) in the ER. The genes involved in the first glycosylation steps are depicted in Fig. 5. The initial ER-resident glycosylation functions are rather conserved, whereas the decoration of glycans with different, and differently linked, sugar mojeities varies significantly among yeast species. For an overview of glycan patterns, the reader is referred to Gemmill & Trimble (1999) and Goto (2007).

Figure 5

Early glycosylation. After translocation, proteins may receive either a mannose residue at a Ser/Thr residue by the protein O-mannosyl transferase (PMT) complex, or a Glc3Man9GlcNAc2 glycan at an Asn residue by the oligosaccharyltransferase (OST) complex. PMTs are dimers of either one member of the PMT1 and one of the PMT2 family, or dimers of Pmt4. The OST complex consists of 9 subunits. Mannose, or the core N-glycan, respectively, is recruited at the cytosolic side of the ER membrane and flipped into the ER luman on a dolichol anchor, as illustrated by Burda & Aebi (1999). Proteins which were not identified in all yeast genomes studied here are depicted in red.

Yeasts devoid of either N- or O-glycosylation are inviable. Apparently protein glycosylation (both N- and O-linked) confers multiple functions to secreted proteins, spanning from increased protein solubility to an impact on cell wall stability, osmotolerance, and budding (Gentzsch & Tanner, 1996; Willer et al., 2005; Goto, 2007). Additionally, virulence in C. albicans was reported to be related to O-glycosylation (Prill et al., 2005).

However, protein glycosylation plays another internal role for the eukaryotic cell as a signal for protein conformation. Most importantly N-glycosylation is employed for quality control of protein folding in the ER (Roth et al., 2010), as illustrated in Fig. 6.

Figure 6

Calnexin cycle. After initial N-glycosylation, two terminal glucose residues (indicated by green dots) are removed by glucosidases (Cwh41 and Rot2/Gtb1), leading to calnexin (Cne1) binding. Further glucose removal leads to calnexin release. Glucose may be re-added by UDP-glucose:glycoprotein glucosyltransferase (UGGT), closing the loop of the calnexin cycle. Removal of mannose residues by mannosidases (Mns1, Mnl1, Mnl2) is implicated with targeting to degradation, while correctly folded protein is released to further transport to the Golgi. Pathways and proteins which are not present in all yeasts are indicated in red.

The calnexin cycle and its components in mammalian cells and yeasts

After transfer of Glc3Man9GlcNAc2 to the protein, the terminal α-1,2 glucose is removed by glucosidase I (Cwh41 alias Gls1), whereas the second, α-1,3 linked glucose residue is removed by glucosidase II, consisting of two subunits in yeast, Rot2 (alias Gls2) and Gtb1. Interestingly, the Gtb1 homologs of P. pastoris, Y. lipolytica, and S. pombe carry a HDEL-like ER retention signal, while those of the other studied yeasts do not. Thus, the former three yeasts resemble more the human cells in this respect, where the function of the β-subunit was ascribed to retain glucosidase II in the ER (Trombetta et al., 1996). Further studies will be needed in future to investigate ER retention of this complex in yeasts lacking the HDEL-like signal on the β-subunit, and defining potential functions of Gtb1 apart from ER retention.

The ER membrane-bound lectin-chaperone calnexin (Cne1) binds specifically to Glc1Man9GlcNAc2. After the last α-1,3 glucose residue is removed by glucosidase II resulting in the Man9GlcNAc2 structure, the protein is released from Cne1 (Caramelo & Parodi, 2008). Mammalian cells possess an UDP-glucose:glycoprotein glucosyltransferase (UGGT) that adds one α-1,3 glucose residue to the A-branch of the mannose tree, enabling re-binding of calnexin so that the loop of the so called calnexin cycle is closed (Hebert et al., 1995). Re-entry of the calnexin cycle extends the available time for an individual protein molecule to fold correctly. UGGT acts as a folding sensor transferring glucose to partially misfolded glycoproteins (Trombetta & Parodi, 1992). Proteins with native conformation exit the calnexin cycle as they are not re-glycosylated. It is noteworthy that S. cerevisiae lacks the gene for UGGT, differently to other yeasts (Fernández et al., 1994; Babour et al., 2004). Correctly folded proteins are directed to the ER-exit sites (ERES). Mis- or unfolded proteins, however, remain in the ER. After release from calnexin, glycans may be further trimmed at the B and C branch by mannosidase 1 (Mns1), an ER-resident α-1,2 mannosidase resulting in Man8GlcNAc2 (Jakob et al., 1998) which is be further trimmed to Man7GlcNAc2 by Mnl1 [alias Htm1, (Nakatsukasa et al., 2001; Clerc et al., 2009)]. These mannose-trimmed glycoproteins are supposed to be subsequently directed to ERAD (van Anken & Braakman, Ruddock & Molinari, 2006; Anelli & Sitia, 2008). Based on more recent in vivo studies with S. pombe Stigliano et al. (2011) suggest that mannose trimming of the B and C branches rather redirects slowly folding protein to the calnexin cycle while these authors regard trimming of the A-branch to be responsible for ERAD direction at least in mammalian cells. Recently, Mnl2 was identified to cooperate with Mnl1 in mannose trimming prior to Yos9 binding (Benitez et al., 2011) which is the initial step toward ERAD.

Thus, calnexin appears as a branch point for routing proteins to Golgi, ERAD, or keeping them in the folding process. One can speculate that the distribution between these routes occurs on a basis of binding affinities of calnexin and UGGT (where it exists) to misfolded protein and the respective probability of each step including mannose trimming, keeping in mind that proteins may contain more than one N-glycan so that several potential calnexin binding sites may be present on one protein molecule.

O-Glycosylation and quality control

Ecker et al. (2003) have shown that O-glycosylation of a cell wall protein (Ccw5) precedes N-glycosylation and can prevent the latter by blocking the threonine residue of the recognition sequence. Their finding can explain that pmt4 deletion leads to the degradation of the endoprotease Axl2, which can be prevented by the N-glycosylation inhibitor tunicamycin (Sanders et al., 1999). It will be worthwhile to investigate in future whether O-glycosylation generally precedes N-glycan transfer and whether this may have a regulatory or quality control function. The QC role of O-glycosylation is less well understood than that of N-glycosylation. It has been reported that proteins are O-mannosylated in the ER when they misfold (Harty et al., 2001). The further fate is controversial, partly leading to ERAD (Hirayama et al., 2008) and partly to protection from terminal degradation, most probably due to enhanced solubility (Nakatsukasa et al., 2004). Goder & Melero (2011) suggested that the Pmt1/Pmt2 complex has a chaperone function in addition to its mannosyltranferase activity and can target misfolded proteins to ERAD.

Differences in glycosylation among yeast species

Some ER glycosylation functions related to quality control differ

The early steps of N-glycosylation are highly conserved among eukaryotes. The respective homologs in all studied yeasts are summarized in Table 4. The OST complex is vital for yeast (Knauer & Lehle, 1999), so that orthologs of all essential genes related to the OST complex are present in all investigated yeast genomes. Ost3 and Ost6 appear to be redundant in their function, which would explain why no homologous gene to OST6 was found in S. pombe and H. polymorpha (Table 4). Ost4 and Ost5 gave generally bad hits as these proteins are only 36, respectively, 90 amino acids long.

View this table:
Table 4

Initial steps of O- and N-glycosylation

S. cerevisiae Description C. glabrata K. lactis C. albicans H. polymorpha P. pastoris Y. lipolytica S. pombe
PMT1 Protein O-mannosyl transferase subunitXP_449091XP_454009EEQ47067EFW98111XP_002491100XP_503966NP_593237
PMT5 * Protein O-mannosyl transferase subunit
PMT2 Protein O-mannosyl transferase subunitXP_448088XP_455792EEQ44716EFW95619XP_002491148XP_502178NP_594135
PMT3 * Protein O-mannosyl transferase subunitXP_446007
PMT4 Protein O-mannosyl transferase subunitXP_449354XP_451701EEQ46011EFW96629CCA36946XP_503607NP_596807
PMT6 Protein O-mannosyl transferase subunitXP_448267XP_456061EEQ45029EFW96948XP_002494221
OST1 Member of the oligosaccharyltransferase complexXP_445659XP_455664EEQ44589EFW96251XP_002492969XP_505429NP_594536
OST2 Member of the oligosaccharyltransferase complexXP_445166XP_452796EEQ47512EFW95067XP_002491840XP_503704NP_593898
OST3 Member of the oligosaccharyltransferase complexXP_448080XP_452821EEQ46513EFW96759XP_002494049XP_502683NP_001018226
OST4 Member of the oligosaccharyltransferase complexXP_002999526XP_002999377EEQ47486???NP_593850
OST5 Member of the oligosaccharyltransferase complexXP_002999556XP_451054?????
OST6 Member of the oligosaccharyltransferase complexXP_446667XP_451584EEQ44676XP_002493390
STT3 Member of the oligosaccharyltransferase complexXP_444781XP_452719EEQ45610EFW94389XP_002490630XP_500973NP_595148
SWP1 Member of the oligosaccharyltransferase complexXP_447284XP_455412EEQ42806EFW96503XP_002489593XP_506062NP_587768
WBP1 Member of the oligosaccharyltransferase complexXP_448873XP_453681EEQ42119EFW95553XP_002491326XP_502492NP_588153
CWH41 Glucosidase IEEQ44917XP_454433EEQ44917EFW95603XP_002489866XP_505435NP_594106
ROT2 Glucosidase II α subunitXP_448526XP_455522EEQ46073EFW95624XP_002491714XP_500574NP_593490
GTB1 Glucosidase II β subunitXP_445896XP_455673EEQ43201EFW94731XP_002492388XP_504443NP_588052
CNE1 CalnexinXP_445424XP_455100EEQ45089EFW96881XP_002491218XP_500829NP_593612
§ UDP-glucose:glycoprotein glucosyltransferase EEQ44348EFW95337XP_002493168 XP_002489455XP_501771NP_595281
KRE5 §Required for β-1,6 glucan synthesisXP_445924XP_455189EEQ44348EFW95337?XP_002493168XP_501771NP_595281
MNS1 α -1,2 mannosidaseXP_449368XP_451726EEQ42796EFW96917XP_002491684XP_502939NP_594139
MNL1 α -1,2 mannosidaseXP_446361XP_451695EEQ45812EFW94311XP_002493127XP_502767NP_594434
MNL2 Putative mannosidaseXP_446049XP_455484XP_503061
  • In S. cerevisiae, Pmt5 can be replaced by Pmt1, and Pmt3 by Pmt2.

  • Ost4 and Ost5 are too short for distinct identification in several yeasts.

  • Ost6 is similar to, and obviously can be replaced by Ost3.

  • S. pombe UGGT and S. cerevisiae Kre5 are putative homologs; unambiguous identification of their respective homologs in other yeasts is difficult.

All studied yeasts contain a calnexin gene; however, the soluble homolog calreticulin seems to be specific for higher eukaryotes. The calnexin cycle is incomplete in S. cerevisiae and apparently also in K. lactis and C. glabrata due to the lack of UGGT. In S. pombe, an UGGT homolog has been described (Fernández et al., 1994). Homologous genes were found in Y. lipolytica (Babour et al., 2004) and a homolog was also identified here in P. pastoris, H. polymorpha, and C. albicans. All these yeast homologs contain a C-terminal HDEL signal for ER retention. Similarly plant and mammalian UGGTs contain potential ER retention signals. All yeast UGGTs are homologous to S. cerevisiae Kre5 which is involved in β-1,6 glucan formation (Table 4); however, the high similarity of the Y. lipolytica, P. pastoris, H. polymorpha, and C. albicans genes to S. pombe UGGT (Gpt1) especially at the C-terminus indicates that also among Saccharomycetales yeasts the full calnexin cycle is functional and was rather lost during evolution to close relatives of S. cerevisiae. The mannosidases Mns1 and Mnl1 are present in all investigated yeasts, indicating that the exit of the calnexin cycle toward ERAD is well conserved. An obvious homolog of Mnl2 was only identified in K. lactis, Y. lipolytica, and C. glabrata which indicates that it is not essential for other yeasts, which is supported by the fact that S. cerevisiae mnl2 deletion strains are viable as well.

O-glycosylation is initiated by the activity of the protein O-mannosyltransferase (Pmt) family (Table 4), consisting of seven enzymes in S. cerevisiae (Gentzsch & Tanner, 1996). Pmt's act as dimers, and obviously different dimers are responsible to glycosylate different proteins (Gentzsch & Tanner, 1997). In S. cerevisiae, the major Pmt proteins are subdivided in three subfamilies: The PMT1 family consists of Pmt1 and Pmt5, the PMT2 subfamily of Pmt2 and Pmt3, and subfamily PMT4 is represented by Pmt4 only (Girrbach & Strahl, 2003). Active dimers are formed by either a member of each subfamily PMT1 and 2, or homodimers of Pmt4. Pmt6 and Pmt7 are less well characterized but show a high degree of sequence identity to the other members of this protein family. In all studied yeasts, clear homologs of Pmt1, 2, and 4 could be identified, while the other Pmt's were less clearly found due to homology among the members themselves (Table 4). However, with at least one member of each of the three subfamilies, the described dimeric complexes can all be formed in all the yeast species.

Glycan maturation in the Golgi is highly variable

When proteins enter the yeast Golgi, the N-glycans obtain a single α-1,6 mannose by Och1 (Nakayama et al., 1992), an enzyme which is conserved among all studied yeasts. This mannose is the branching point of yeast-specific hypermannosylation, next obtaining α-1,6 mannose chains by sequential activity of two enzyme complexes, mannan polymerase I and II (M-Pol I and M-Pol II). M-Pol I consists of two subunits, Mnn9 and Van1 (Jungmann & Munro, 1998), both having homologs in all yeasts except for Van1 in S. pombe and C. albicans. Van1 and Anp1 (a subunit of M-Pol II) are highly similar, and it is tempting to speculate that the Anp1 homolog takes over the function of Van1 where this is not present. M-Pol II consists of five subunits, Mnn9, Anp1, Mnn10, Mnn11, and Hoc1 (Jungmann & Munro, 1998; Jungmann et al., 1999). All these subunits were identified in the genomes of the yeasts investigated here, with the exception that the best homolog of Hoc1 in S. pombe and C. albicans is their Och1 (Table 5). This comes not as a surprise as Hoc1 is homologous to Och1 in S. cerevisiae itself (as indicated in its short name), but it makes the clear annotation of these mannosyltransferases difficult. It should be noted, however, that both a Van1 homolog and a putative Hoc1 homolog were identified in diploid strains of C. albicans. Further research will be needed to clearly identify all players in the two M-Pol complexes of S. pombe and C. albicans.

View this table:
Table 5

Glycan modification in the Golgi

S. cerevisiae Description C. glabrata K. lactis C. albicans H. polymorpha P. pastoris Y. lipolytica S. pombe
OCH1 α -1,6 mannosyl transferaseXP_444841XP_456072EEQ44644EFW97215XP_002489596XP_505112NP_594852
MNN9 Member of mannan polymerase I and II complexXP_449325XP_454529EEQ44637EFW96256XP_002493508XP_500804NP_594753
VAN1 Member of mannan polymerase I complexXP_445071XP_453137EFW96183XP_002491707XP_503450
ANP1 Member of mannan polymerase II complexXP_448831XP_454954EEQ47368EFW98078XP_002492743XP_501421NP_595421
HOC1 Member of mannan polymerase II complexXP_445987XP_452168XP_716752EFW95452XP_002492846XP_505112
MNN10 Member of mannan polymerase II complexXP_448703XP_452892EEQ45426EFW98202XP_002492015XP_504403NP_594624
MNN11 Member of mannan polymerase II complexXP_446687XP_451277EEQ42919EFW96866XP_002492081XP_505534NP_594832 or NP_595579
KRE2 KTR1 KTR2 KTR3 KTR4 KTR5 KTR6 KTR7 YUR1 α -1,2 mannosyl transferase XP_447119 XP_449669 XP_445169 XP_449600 XP_447439 XP_449585 XP_452800 XP_451973 XP_454989 XP_454576 XP_451257 XP_454340 EEQ44249 EEQ44247 EEQ43617 EEQ45874 EEQ42772 EFW97151 EFW97153 EFW97152 EFW95981 EFW96872 EFW96941 EFW94338 EFW94808 EFW97940 EFW95344 XP_002492424 XP_002492423 XP_002492102 XP_002489479 XP_002490162 XP_002491999 XP_499811 XP_503435 XP_500395 XP_505884 NP_596168 NP_588253 NP_595938 NP_595123 NP_595614
MNN4 Putative regulator of KTR6XP_447196XP_452145EEQ44878EFW96861XP_002490538XP_503217
MNN1 MNT2 MNT3 MNT4 α -1,3 mannosyl transferase XP_447202 XP_445352 XP_445350 XP_447201 XP_445348 XP_445347 XP_445346 XP_445345 XP_446547 XP_445349 XP_445351 XP_454392 XP_453659 EEQ42201 EEQ46299 EEQ44597 EEQ46293 EEQ42685 EEQ43575 EFW94638 EFW96038
MNN2 MNN5 α -1,2 mannosyl transferase XP_447446 XP_449471 XP_449128 XP_454751 XP_452936 XP_456174 EEQ42223 EEQ43618 EEQ42162 EEQ45030 EEQ45574 EEQ47130 EFW97159 EFW95342 EFW98389 EFW95612 EFW94475 EFW96952 EFW97869 EFW96912 XP_002492593 XP_002490149 XP_002493020 XP_500042

Mannose branches with α-1,2 linkage are added by Mnn2 and Mnn5. Both genes could not be identified in S. pombe, which would explain why S. pombe adds α-1,2 galactose instead of mannose. Most other studied yeasts feature three or more homologs of the Mnn2/5 family, with especially high abundance of eight in H. polymorpha and six in C. albicans. The final elongation step in S. cerevisiae is the addition of α-1,3 mannose by the MNN1 family (Mnn1, Mnt2, Mnt3, and Mnt4). This family is highly abundant in C. glabrata (11 homologs) and C. albicans (six homologs), and present in K. lactis and H. polymorpha, but not in the other yeasts (Table 5) which is reflected also in the lack of α-1,3 mannans in most yeasts (Gemmill & Trimble, 1999).

The outer mannose residues in S. cerevisiae are often phosphorylated, mediated by the transfer of a mannosylphosphate to the glycan by Ktr6 (syn. Mnn6). Due to the high degree of similarity between the members of the KTR family and related proteins, no clear homolog to Ktr6 could be identified in any yeast, however, between four and 10 family members were found (Table 5). It has been reported that glycans in S. cerevisiae and C. albicans are phosphorylated but not those of K. lactis or S. pombe (Gemmill & Trimble, 1999; Jigami & Odani, 1999). Mnn4, the putative regulator of Mnn6, has a homolog in all studied yeasts except for S. pombe. While none of the four KTR family members of Y. lipolytica could be attributed to mannosylphosphate transfer, deletion of the Mnn4 homolog resulted in complete inhibition of the addition of acidic sugars to glycans, indicating a connection of this gene to phosphomannosylation (Park et al., 2011).

The first elongation step of O-glycans is a conserved addition of α-1,2 mannose residues by Kre2 (alias Mnt1), and its related enzymes of the KTR family, among which Ktr1 and Ktr3 have been shown to have complementary function to Kre2 (Lussier et al., 1999). Due to the high similarity of the KTR family members, we could not identify clear homologs to all family members but there is now doubt that several α-1,2 mannosyltransferases of this family exist in all studied yeasts – varying from four homologs in Y. lipolytica to 10 in H. polymorpha (Table 5).

The further glycan modifying steps show more diversity among the studied yeasts. Only S. cerevisiae and K. lactis contain all α-1,3 mannosyltransferase genes, and this type of mannose binding has not been identified in the other yeasts studied. The α-1,3 mannosyltranferases, members of the MNN1 family of S. cerevisiae (Lussier et al., 1999), are responsible for terminal capping of both N- and O-glycans (Romero et al., 1999). Their homologs are highly abundant in the two Candida species but totally absent in P. pastoris, Y. lipolytica, and S. pombe (Table 5). Thus, it is not surprising that α-1,3 mannosylation has not been described in the latter species (Goto, 2007); however, the reported absence of this linkage in H. polymorpha (Kim et al., 2004; Goto, 2007) and C. albicans (Goto, 2007) asks for re-evaluation based on the presence of respective genes. The α-1,3 mannosyl linkage of S. cerevisiae mannans has been described to be highly immunogenic (Ballou, 1990), leaving the question of the immunogenicity of other yeast mannans open for investigation.

Differently, C. glabrata, C. albicans, and P. pastoris link β-1,2 mannose to O- and N-glycans (Mille et al., 2008; Takahashi et al., 2012), a function encoded by the BMT gene family. Among the studied yeasts, the Candida species and P. pastoris harbor 4-8 BMT family homologs, respectively, while of the others only H. polymorpha carries one homologous gene, so that it is worthwhile to analyze for β-1,2 mannose linkage in this yeast as well. This sugar structure has been linked with virulence of C. albicans (Mille et al., 2008), but it should be noted that enzymes similar to the BMT family are encoded by many other yeasts and filamentous fungi, among them Pichia stipitis, Pichia guilliermondii, Aspergillus oryzae or Debaryomyces hansenii, which are partly used in food production since long time, so that a potential hazard should not be deducted per se from this glycan structure.

Schizosaccharomyces pombe glycans are characterized by terminal galactose residues which are additionally pyruvylated, a modification that rarely occurs in eukaryotic proteoglycans (Gemmill & Trimble, 1996). Kluyveromyces lactis has been shown to add terminal α-1,2 N-acetyl glucosamine in addition to α-1,3 mannose, a modification that has also been reported in S. cerevisiae (Yoko-o et al., 2003). These reports go in line with the identification of the α-1,2 N-acetylglucosaminyltransferase gene GNT1 in these two yeasts, as well as in the two Candida species. More research will be needed to characterize the structures and functions of different terminal proteoglycan modifications in different yeasts. However, these are not as central to protein secretion as the core glycans added in the ER.

Endoplasmic reticulum-associated protein degradation: a route to the proteasome

Additionally to protein folding and glycosylation, the ER also serves as a major location for protein quality control. After translocation into the ER, polypeptides have to be folded as a prerequisite for transport to the Golgi for further processing. Proteins that are terminally misfolded or fail to assemble cannot pass the ER quality control mechanism as they are recognized and targeted for degradation. This well-organized and conserved pathway is the ERAD, (Fig. 7), in which molecular chaperones and lectin-like enzymes are involved (Hoseki et al., 2010).

Figure 7

ER-associated protein degradation. Recognition of the misfolded protein depends on its composition. While misfolded glycosylated proteins are recognized by Yos9, Kar2 seems to lead non-glycosylated proteins to the ERAD machinery. For luminal proteins, Hrd1 is the major mediator of substrate ubiquitination during ERAD. It interacts with several other membrane-bound or membrane-associated proteins while ‘pushing’ the misfolded protein to the Cdc48 complex, which is connected to the proteasome. Proteins which are not present in all yeasts are indicated in red.

Recognition of misfolded proteins

In yeast, three different ERAD pathways have been described so far: ERAD-L (lumen), ERAD-M (membrane) and ERAD-C (cytosol) (Carvalho et al., 2006; Gauss et al., 2006), depending on the localization of the lesion in the protein. Specialized E3 ubiquitin ligases have a central role in routing of unfolded or misfolded proteins across the ER membrane toward the 26S proteasome in the cytosol (Sitia & Braakman, 2003; Vembar & Brodsky, 2008). Two E3 ligase complexes, Doa10 and Hrd1/Der3, are present in yeast. While transmembrane proteins with defects in the cytosolic domain are degraded via the Doa10 complex (ERAD-C), lesions in the luminal or transmenbrane domains are recognized by the Hrd1 complex (ERAD-L and ERAD-M) (Anelli & Sitia, 2008). We will concentrate on the players of the ERAD-L pathway, as this is the route of secreted proteins and is of major interest in this review. The Hrd1/Der3 complex additionally interplays with four other proteins: the luminal lectin Yos9, the putative channel component Der1, the transmembrane cofactor Hrd3, and Usa1, the regulator of the oligomeric state of Hrd1/Der3 (Hirsch et al., 2009). All these proteins participate in substrate recognition at the luminal side of the ER membrane and protein ubiquitylation and degradation at the cytosolic side.

Mns1, an α-1,2-mannosidase, is responsible for the removal of one mannose residue from the middle (B) branch in misfolded N-glycosylated ERAD-L substrates. A mannosidase-homology domain (MHD) has also been found in the yeast protein Mnl1 (syn. Htm1) where it was shown to be needed for substrate binding and further mannose trimming (removal of mannose from the C branch) (Anelli & Sitia, 2008; Wolf & Stolz, 2012). This glycan structure is then recognized by Yos9 as being misfolded. This protein was suggested to scan and bind proteins according to their trimmed glycan structure (Stolz & Wolf, 2010). Hrd3, an important interaction partner of Yos9, binds the potential ERAD-L substrate, also independently from Yos9 (Gauss et al., 2006). Kar2 escorts non-native polypeptides to the Hrd1/Der3 complex (van Anken & Braakman, 2005ab; Ruddock & Molinari, 2006; Anelli & Sitia, 2008). Before the misfolded protein is retro-translocated into the cytosol, Pdi1 reduces the disulfide bonds by interacting with Mnl1, a reaction that is supposed to be necessary for Pdi1 recruitment for ERAD-L (Sakoh-Nakatogawa et al., 2009). Glycoproteins that bear other mannose glycans are released for further refolding. The mechanism of ERAD of non-glycosylated proteins is only poorly understood. However, the lectin-like proteins are described to be involved also in the degradation of non-glycosylated proteins (Hoseki et al., 2010). Jaenicke et al. (2011) described recently that Yos9 assists in the HRD complex-dependent degradation of non-glycosylated proteins, as demonstrated for non-glycosylated mutants of Erg3 and CPY*. It has also been shown that Kar2 initiates ERAD-L of misfolded non-glycosylated proteins (Brodsky & McCracken, 1999; Nishikawa et al., 2001).

Boisramé et al. (2006) have already shown that Y. lipolytica exhibits two different Hrd1 homologs (XP_504928 and XP_503640), which are supposed to function in two different ERAD ways. In this study, we could determine a significantly higher similarity of XP_504928 to S. cerevisiae Hrd1. Our homology analyses of the yeast proteomes revealed that Usa1 was only present in S. cerevisiae, C. glabrata, and K. lactis. Der1 and its homolog Dfm1 are both missing in H. polymorpha. Candida albicans, P. pastoris, Y. lipolytica, and S. pombe exhibit only one protein with same sequence identity to both Der1 and Dfm1. Clear homologs of Yos9 with similar length (around 540 amino acids) could only be identified in C. glabrata, K. lactis, and Y. lipolytica, while P. pastoris, C. albicans, and S. pombe encode only a shorter protein (250–300 amino acids) with lower sequence identity. No homolog of Yos9 was found in H. polymorpha. The identity and function of Yos9 homologs in non-conventional yeasts is certainly a field for future research.

Retrotranslocation and degradation of misfolded proteins

The E3 ligase complex is also responsible for the retrotranslocation of ERAD-L substrates through the ER membrane to the cytosol. Several candidates have been suggested as potential retro-translocation channels. One of them is the Sec61 translocon complex (Schäfer & Wolf, 2009). Hrd1 was also proposed to be involved in retro-translocation of soluble proteins (Horn et al., 2009), in a reaction that requires the AAA-ATPase Cdc48 complex which consists of two ATPase domains and the proteins Cdc48, Ufd1, and Npl4 (Wolf, 2004). The homohexameric complex interacts with the ubiquitin-fold domain (UBX) of the membrane-associated protein Ubx2 (Hirsch et al., 2009), which enables the transfer of a misfolded multiubiquitylated protein to the Cdc48 complex (Römisch, 2006). Der1 was shown to co-immunoprecipitate with Ubx2 and to be a receptor for the Cdc48 complex as well (Schuberth & Buchberger, 2005). The polyubiquitination is catalyzed by Cue1-assembled Ubc7, and to a lesser content by Ubc6 and Ubc1 (Hiller et al., 1996; Friedlander et al., 2000; Pety de Thozée & Ghislain, 2006; Horn et al., 2009). Due to ATP hydrolysis, the complex undergoes a conformational change, which could act as a ‘molecular ratchet’ for the release of substrates from the membrane (Wolf, 2004; Hirsch et al., 2009). The function of the yeast proteins Der1 and Usa1 is still not well understood. Deletion of the ER-membrane anchored protein Der1 influenced only the turnover of soluble misfolded carboxypeptidase Y (CPY*) (Taxis et al., 2003). However, different studies have shown that Usa1 is needed for linking Der1 to the Hrd1/Der3 ligase (Carvalho et al., 2006; Horn et al., 2009), but on the other hand, it was not needed for the degradation of several proteins (Carvalho et al., 2006).

Polyubiquitylated misfolded proteins are further recognized by the 26S proteasome, which is (among eukaryotes) a conserved large protein complex composed of a 20S protease core particle and a 19S regulatory particle (Kim et al., 2011). In yeast, six proteins, Rpt1–6, were described to form a ring at the entrance of the proteasome degradation channel and thereby to control access in an ATP-dependent manner (Finley, 2009). The substrate delivery cofactors Rad23 and Dsk2 dock on the polyubiquitylated substrates and guide them from the ER membrane to their receptors Rpn10 and Rpn13 in the 19S proteasomal subunit (Medicherla et al., 2004; Finley, 2009) and may also shield the polyubiquitin chain from ubiquitin hydrolases (Hirsch et al., 2009). They are recognized via their ubiquitin-like (UBL) domain by their receptor proteins (Elsasser et al., 2004). Rpn1 was suggested as well to be a receptor of Rad23 and Dsk2 (Rosenzweig et al., 2012). N-glycans are removed by the N-glycanase Png1 that is escorted to the proteasome via the ubiquitin receptor Rad23 (Suzuki et al., 2001). This glycan trimming step is not essential for degradation, but seems to fasten the process (Suzuki et al., 2000).

Deubiquitinating enzymes (DUBs) Rpn11 and Ubp6 remove the polyubiquitin chain from the substrates. Rpn11 is a part of the 26S proteasome, whereas Ubp6 binds to its receptor Rpn1 of the proteasomal 19S base subunit (Kim et al., 2011). A third DUB enzyme, UCH37, which is not present in S. cerevisiae but was found in S. pombe and higher eukaryotes (Li et al., 2000; Yao et al., 2006) is a C-terminal hydrolase of the UCH family that cleaves ubiquitin chains into peptides (Kim et al., 2011). Our domain search analysis has shown that UCH37 is present also in P. pastoris (CCA40110) and K. lactis (XP_453197).

Vesicular transport

The transport of proteins between cellular compartments is mediated by vesicles that bud from one membrane and fuse to a defined target membrane. In this review, we describe the anterograde transport (ER to cis-Golgi) mediated by COPII vesicles, the retrograde transport (cis-Golgi to ER and trans- to cis-Golgi) mediated by COPI vesicles, the intra-Golgi transport, and the exocyst for the secretory vesicles (trans-Golgi to the plasma membrane). Insights into the sorting of protein cargo, vesicle budding, tethering, docking, and fusion to the target membrane are described for the different transport directions. An overview of the main proteins involved in this process is provided in Fig. 8.

Figure 8

Vesicular transport. The secretion of proteins is mediated by vesicles that bud at defined regions of the ER (tER sites) and move toward the plasma membrane. For every transport direction, the most important elements are listed in boxes: transporting vesicles, tethering complexes, Rab GTPases for vesicle docking, SM (Sec1/Munc18-like) proteins for SNAREs pairing and SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) for membrane fusion. The ER to Golgi transport is mediated by COPII vesicles (coat protein II), the TRAPPI tethering complex (transport protein particle I), the Rab Ypt1, the SM protein Sly1, and a group of SNAREs. The Golgi to ER transport is mediated by COPI vesicles (coat protein I), the tethering complex Dsl1 (depends on SLY1-20), the Rab Ypt1, the SM protein Sly1, and a group of SNAREs. The intra-Golgi transport is mediated by COPI vesicles, the tethering complex TRAPPII (transport protein particle II) with the Rab Ypt31/32, the tethering complex COG (conserved oligomeric Golgi) with the Rabs Ypt1 and Ypt6, the SM protein Sly1 and a group of SNAREs. Transport from the Golgi to the plasma membrane is mediate by the tethering complex exocyst, the Rabs Ypt31/32 and Sec4, the SM protein Sec1 and a group of SNAREs. Due to the still open debate on the cis- to trans-Golgi transport, this transport is simply illustrated by a wide arrow through the Golgi cisternae toward the plasma membrane without vesicles. Proteins which are not present in all yeasts are indicated in red.

En route to secretion: anterograde transport from ER to Golgi

Endoplasmic reticulum exit and COPII budding

The transport of correctly folded proteins from the ER to the cis-Golgi is mediated by COPII vesicles. Sorting and concentration of cargo proteins in COPII-coated vesicles take place at specialized ER membrane domains called ER exit sites (ERES) or transitional ER sites (tER) in a process that is well conserved in higher eukaryotes and some yeasts. tER sites can be considered as self-organizing structures. They are long-lived structures that form de novo, fuse together after collision, grow and shrink to maintain a steady state size (Bevis et al., 2002; Stephens, 2003). Several excellent reviews describe the organization and maintenance of ERES (Budnik & Stephens, 2009) and the process of assembly and organization of COPII vesicles (Duden, 2003; Hughes & Stephens, 2008).

Two proteins whose interaction seems to play a very important and conserved role in tER and COPII vesicles organization are Sec16 and Sec12 (Montegna et al., 2012). The peripheral ER membrane protein Sec16 recruits the guanine nucleotide exchange factor (GEF) Sec12 (Barlowe et al., 1993), which catalyzes the exchange of GDP with GTP on the small GTPase Sar1, the first member of COPII vesicles. The activated GTP-bound form of Sar1 can then associate with ER membranes and, through direct interaction with Sec23, recruits the heterodimer Sec23/Sec24 (Barlowe et al., 1994). Next, Sec23 acts as GTPase activating protein (GAP) of Sar1, and Sec24 captures the cargo proteins into the nascent vesicle. The following step is the interaction of the Sec23/Sec24 complex with the heterotetrameric Sec13/Sec31 complex that polymerizes to form the outer layer of the coat (Stagg et al., 2006). Sec13/Sec31 complexes assemble around the prebudding complex Sar1/Sec23/Sec24 forming a caged structure, which induces a bending at the ER membrane and the following budding of the COPII vesicles (Lee et al., 2005). Sec13/Sec31 complex also contributes at a further stimulation of the GTPase activity of Sar1 (Antonny et al., 2001). Recently, it has been described that also Sec24 together with Sec16 might regulate the GTPase activity of the COPII coat to prevent premature vesicle scission (Kung et al., 2012).

An efficient ER export requires an accurate and selective system that recruits the cargo proteins into COPII vesicles. It has been suggested that ERES might represent one of the first selection steps (Castillon et al., 2009). These authors observed in S. cerevisiae cells the presence of three distinct ERES populations carrying three different types of cargo proteins, like pro-α-mating factor, transmembrane proteins like aminoacid permeases or GPI anchored proteins, respectively. The second important selection step is the interaction of the cargo proteins with the COPII subunits. It has been described that Sec24 binds the majority of cargo proteins. The necessity of the cell to recognize a wide range of cargo proteins might explain why mammalian cells express four isoforms of Sec24 (Wendeler et al., 2007) and why S. cerevisiae expresses not only Sec24 with at least three independent cargo-binding sites (Miller et al., 2003), but additionally the Sec24 homologs Sfb2 (Kurihara et al., 2000) and Sfb3 (Miller et al., 2002).

However, not all proteins that have to be secreted interact directly with Sec24. Cells have developed other mechanisms to select and collect cargo proteins into COPII vesicles, like cargo receptors and cargo anchors. Cargo receptors are typically transmembrane proteins that recognize folded protein motifs and not short linear peptide signals. These receptors have a strong preference for correctly folded and assembled cargo proteins, suggesting a connection between sorting receptors and ER quality control system (Dancourt & Barlowe, 2009). Up to date, only few cargo receptors and their specificities have been identified (Herzig et al., 2012). Erv29, for example, binds the fully folded glycosylated pro-α-mating factor, and it is necessary for the packaging of this cargo into COPII vesicle (Belden & Barlowe, 2001). These authors suggest that Erv29 might also be important for the binding of other soluble secretory proteins. Differently, cargo anchors cannot bind the cargo directly and mediate their interaction with the COPII coat. One example is the p24 complex that mediates the export of GPI anchored proteins (Castillon et al., 2011). In yeast, four members of the p24 family proteins, Emp24, Erv25, Erp1, and Erp2, function in the p24 complex (Strating & Martens, 2009). Castillon et al. (2011) shows that only proteins with a correctly remodeled GPI anchor can be connected to the COPII coat through the p24 complex and that not correctly remodeled GPI anchored proteins which escaped the ER are transported back by the p24 complex itself. These data suggest a role of the p24 complex in the quality control of these specific cargo proteins.

Furthermore, analyses of a mammalian lectin transporter showed that binding and release of cargo proteins might require conformational changes of the cargo receptor, which might be induced by the different pH and Ca2+ concentration present in the ER and in the Golgi (Appenzeller-Herzog et al., 2004). An illustrative overview of sorting receptors involved in the early secretory pathway is described in the review of Dancourt & Barlowe (2010).

The yeast comparison performed here shows that the proteins involved in this first step of the secretory pathway are well conserved. Only few differences have been observed: Sfb2 is present only in S. cerevisiae and C. glabrata and an additional Sec23 is present in all the other yeasts apart from S. cerevisiae and K. lactis. Interestingly, we observed that C. glabrata has two copies each of Sec12, Sec13, and Sec23, which might be related to the high secretion efficiency of this yeast. Similarly, the presence of two paralogs of Sec16, Sar1, Sec23, and Sec31 in mammalian cells suggests that the system evolved to increase the complexity of the COPII coat. Yeast ER cargo receptors seem also to be well conserved, differences have been observed only for Emp47 and its homolog Emp46, which are considered to be cargo receptors for the ER export of N-linked glycoproteins (Sato & Nakano, 2002, 2003; Satoh et al., 2006). These authors describe that Emp46 and Emp47 lack the divalent cations binding sites which are normally present in calnexin, calreticulin and L-type lectins. Furthermore, they suggest that Emp46 might bind highly mannosylated glycoproteins in a Ca2+-independent process. The comparison among the different yeast species showed that only S. cerevisiae and C. glabrata have both proteins, while K. lactis, C. albicans, and P. pastoris, as well as mammalian cells have only one homolog. No homologs have been found in H. polymorpha, Y. lipolytica, and S. pombe. It would be interesting to understand whether the single Emp46/47 homolog binds to glycoproteins as S. cerervisiae Emp46 does, and why some yeasts do not need any homolog at all.

Moreover, we also found a group of genes present only in S. cerevisiae and Saccharomyces-related species which encode for the proteins Mst27, Mst28, Prm8, and Prm9. These proteins, which contain predicted transmembrane domains (Poirey et al., 2002), are usually present as heterodimer complexes, Mst27/Mst28 and Prm8/Prm9, and contain binding sites for COPI and COPII vesicles (Sandmann et al., 2003). Mst27 has been identified as multicopy suppressor of a sec21-3 mutant, which showed a block in the ER to Golgi transport of α mating factor precursor and CPY. It has been suggested that the overexpression of Mst27 might lead to an increased production of COPII vesicles or increased efficiency of the cargo packaging. They are not essential proteins and also the double deletions do not alter the growth phenotype (Poirey et al., 2002). Genome evolutionary analyses classified these genes as ‘dispersed duplications’ (Gordon et al., 2009) making it even more difficult to understand why these genes are still conserved in S. cerevisiae.

The transport protein particle TRAPPI and Rab Ypt1

After budding, vesicles have to move toward the target membrane proceeding through defined and progressive steps: tethering, docking, and fusion. A descriptive figure of the steps involved in vesicle budding and fusion is provided by Cai et al., (2007ab) and Szul & Sztul (2011). Tethering is mediated by tethering factors which can be long coiled-coil proteins as the yeast golgin Uso1 (Noda et al., 2007) and Bug1 (Gillingham & Munro, 2003; Behnia et al., 2007) or multisubunit complexes. Eight conserved tethering complexes have been identified in yeast. In this review, we describe the five complexes necessary for secretion: TRAPPI (ER to Golgi), Dsl1 (Golgi to ER), TRAPPII (intra-Golgi/endosome to trans-Golgi), COG (endosome to cis-Golgi), and exocyst (trans-Golgi to plasma membrane). However, we do not describe the remaining three complexes that are required for vacuolar protein sorting: CORVET (late-Golgi to endosome), HOPS (endosome to vacuole), and GARP/VFT (endosome to late-Golgi). The role of these multisubunit tethering complexes is well described in a recent review (Bröcker et al., 2010).

The TRAPPI complex, which contains the subunits Bet3, Bet5, Trs31, Trs23, Trs33, Trs20, and Trs85, mediates the interaction of COPII vesicles to cis-Golgi membrane through direct binding of Sec23. Since it has been shown that the sequential interaction of Sec23 with different binding partners is necessary to ensure the directionality of the anterograde transport, Sec23 is considered to be one of the major regulators of budding and fusion of COPII vesicles to the Golgi (Cai et al., 2007ab; Lord et al., 2011). These authors describe that during the process of coat assembly, the non-phosphorylated form of Sec23 interacts with Sar1-GTP, but once the budding process is complete, GTP is hydrolyzed and Sar1-GDP dissociates from the Sec23/Sec24 complex. Sec23 can then interact with the TRAPPI subunit Bet3 allowing the tethering of the vesicle to the Golgi membrane where the kinase Hrr25 phosphorylates Sec23, thereby inducing the release of the TRAPPI complex and allowing the fusion to the membrane. However, vesicles should also be uncoated to be able to fuse to the membranes but it is still unclear whether this process takes place before or after tethering. There are indications that that vesicles might keep their coat until they reach the Golgi (Lord et al., 2011).

The process of vesicle docking is mediated by another class of proteins named Rabs, which belong to the superfamily of small Ras-like GTPases. As described in the review of Hutagalung & Novick (2011), Rabs regulate membrane trafficking through the interaction with defined effectors. Rabs cycle between two conformational states: the cytosolic GDP bound inactive form and the GTP-bound active form which is located at the membrane. Newly synthesized Rabs needs to be geranylated to get inserted into the membrane. Rab escort proteins (REP) (Alory & Balch, 2003) are the proteins that bring Rabs to the geranylgeranyl transferases. In yeast, the REP Mrs6 brings the Rabs Ypt1 and Sec4 to the geranylgeranyltransferase GGTaseII which is formed by the subunits Bet2 and Bet4 (Jiang & Ferro-Novick, 1994). Next GDP is exchanged with GTP by a guanine exchange factor (GEF) (Barr & Lambright, 2010) to obtain the active form of Rab which can interact with the defined effectors. Rab inactivation is induced by GTPase activating proteins (GAP) (Rak et al., 2000). The recycling of inactive Rabs to the cytosol is mediate by GDP dissociation inhibitors (GDI) which are able to extract Rab-GDP from the membranes (Gilbert & Burd, 2001; Alory & Balch, 2003). When required, Rabs will be recruited again to the membranes by a GDF (GDI displacement factor), which is able to displace the GDI and allows the insertion of the Rabs into the membrane through a prenyl group (Dirac-Svejstrup et al., 1997).

The Rab required for ER to Golgi transport is Ypt1 (Jedd et al., 1995) and the TRAPPI complex is the GEF which activates Ypt1 in this vesicle transport (Wang et al., 2000; Cai et al., 2008). However, it has been described that the same Rab can be involved in different trafficking steps through the interaction with different effectors. Ypt1 is in fact also required for Golgi to ER transport and intra-Golgi transport (Cai et al., 2008; Kamena et al., 2008). It has been suggested that the Ypt1 integral membrane receptor might be formed by the three essential proteins Yos1, Yip1, and Yif1 (Matern et al., 2000; Heidtman et al., 2005). Yip1 and Yif1 are well conserved but Yos1 is missing in H. polymorpha and Y. lipoytica.

More than 60 Rabs proteins are present in mammalian cells, whereas only 11 Rabs have been identified in yeast (Pereira-Leal & Seabra, 2001; Liu & Storrie, 2012). Moreover, Pereira-Leal compared the genome of 26 fungi and observed that these fungi have similar sets of Rabs (Pereira-Leal, 2008).

Membrane fusion mediated by SNARE proteins

As mentioned previously, SNAREs are proteins necessary for the fusion of the vesicle at the target membrane. This family of small proteins is characterized by the presence of the SNARE motif, a 60–70 amino acid stretch which is organized in heptad repeats. Additionally, most of these proteins contain a single C-terminal transmembrane domain and many have N-terminal domains which are folded independently from the SNARE motif (Jahn & Scheller, 2006; Kienle et al., 2009). As long as SNARE proteins are monomeric, the SNARE motif is unstructured, but when they associate into a complex they form elongated four-helical bundles with a high degree of stability. The complex formation mechanically pulls the membranes closer together and results in the formation of a fusion pore (Jahn & Scheller, 2006). Previously it was thought that a distinct set of SNAREs is responsible for each of the trafficking steps but lately it has become clear that certain SNAREs play a role in several fusion steps and can take part in the formation of various complexes (Kienle et al., 2009). Functionally, SNARE proteins are separated into v (vesicle membrane)- and t (target membrane)-SNAREs reflecting the position of the protein, still one protein can sometimes act in a v- and other times in a t-position. A structural classification according to the amino acids in the core layer of the coiled-coil structure can be more useful. The subgroups Qa, Qb, and Qc contain glutamine and the subgroup R contains arginine in the center of the helix bundle. Complex formation is the key for the functionality of SNARE proteins and typically one protein of each subgroup takes part in such a complex. Q-SNAREs act on the target membrane side and R-SNARE on the vesicle membrane side, whereby also several exceptions exist (Gupta & Brent Heath, 2002; Jahn & Scheller, 2006; Alpadi et al., 2012). The interaction between SNAREs is supported by a different class of proteins, the SM (Sec1/Munc18-like) proteins, which bind the syntaxin domains present in many SNAREs. In their review, Südhof & Rothman (2009) describe a catalytic function of SM proteins in SNAREs assembly and speculate about their additional roles in the fusion process.

In S. cerevisiae, there are four SM involved in several trafficking steps, Sec1, Sly1, Vps33, Vps45, and the typical SNAREs set for the anterograde transport consists of Sed5, Bos1, Bet1, and Sec22. The fusion of the vesicles to the Golgi membranes is mediated by the interaction of the essential SM protein Sly1 with the syntaxin SNAREs Sed5 (Kosodo et al., 2002), Bet1 and Bos1. Bos1 is an essential v-SNARE localized at ER membranes, which interacts with Sed5 and together with Bet1 interacts with COPII vesicles, in a process mediated by the Rab Ypt1 (Lian & Ferro-Novick, 1993). Sec22 is not essential and can be substituted by the R-SNARE Ykt6, a multifunctional essential SNARE required for the anterograde transport, intra-Golgi trafficking and transport to the vacuole. It has to be localized at the membrane to be functional, and it has been described that its deletion leads to the accumulation of the p1 precursor of CPY and morphological abnormalities. Ykt6 homologs are highly conserved between many species including human: sequence identity of 47% between S. cerevisiae and human (McNew et al., 1997). After vesicle fusion, the SNARE complex is quickly disassembled in the presence of ATP by a combined action of the chaperone Sec18, a NSF (soluble N-ethylmaleimide-sensitive factor) homolog which is an ATPase, and its cofactor Sec17, which is an α-SNAP (soluble N-ethylmaleimide-sensitve factor attachment protein) homolog (Hong, 2005; Jena, 2011). It has been proposed that Sec17 and Sec18 might facilitate the recycling of SNAREs during the fusion events that occur at the different membranes.

In general, the membrane fusion process is highly conserved in all eukaryotes and there is also a high degree of conservation concerning the SNARE motifs. Still, the number and whole sequence conservation varies to some extent among yeasts. Saccharomyces cerevisiae contains 24 different SNAREs (Burri & Lithgow, 2004). Other yeasts generally have a slightly smaller set of SNAREs and mammalian cells contain around 36. In the transport step from the ER to the Golgi, the SNARE proteins are conserved in all yeast species, with the exception of Bos1, for which a homolog could not be detected in H. polymorpha.

Structural differences at tER sites and Golgi among yeasts, plants and higher eukaryotes

The comparison among higher eukaryotes and yeasts showed some interesting and functional differences at the structures involved in the ER to Golgi transport. Saccharomyces cerevisiae lacks discrete tER sites and cargo proteins leave the ER membrane from many small tER which are distributed on the overall ER (Rossanese et al., 1999; Castillon et al., 2009). Golgi compartments are present as individual cisternae scattered through the cytoplasm that only occasionally can associate with one another. Approximately 20 cisternae are present in S. cerevisiae cells and based on their different protein content they can be classified as cis, medial, trans or trans-Golgi Network (TGN) (Shorter & Warren, 2002; Lowe, 2011). Pichia pastoris contains a discrete number of tER sites and Golgi structures: two to five tER and two to five Golgi stacks, each containing three to four cisternae. Each tER site is juxtaposed to a polarized Golgi stack whose protein composition and distribution remind of the cis- and trans-Golgi markers of S. cerevisiae (Rossanese et al., 1999; Mogelsvang et al., 2003). New Golgi structures appear just after new tER sites and are always adjacent to the tER sites (Bevis et al., 2002). Similarly to mammalian and plant cells, P. pastoris Golgi stacks are surrounded and stabilized by a ribosome excluding matrix (Shorter & Warren, 2002; Mogelsvang et al., 2003; Staehelin & Kang, 2008). The Golgi cisternae seem to follow the maturation model and are ‘fenestrated’ with tubular extensions. Despite these similarities, it is known that in P. pastoris Golgi cisternae are separated from each other and not laterally interconnected to form a ribbon as in mammalian cells and Golgi organization is not mediated by microtubules. The fission yeast S. pombe lacks discrete tER sites (Vjestica et al., 2008) but similarly to mammalian cells and P. pastoris, the Golgi cisternae are organized in stacks. The presence of isolated cis- and trans-Golgi cisternae suggests anyway that these Golgi stacked structures are not so stable. Recent evolutionary studies show that ancestral unicellular eukaryotes possessed stacked Golgi (Mowbrey & Dacks, 2009) suggesting that S. cerevisiae lost this Golgi structure in favor of dispersed cisternae, but it is still not clear why this happened and which could be the advantages for S. cerevisiae. It is known that the presence of a defined number of tER sites in P. pastoris is a consequence of the interaction between Sec16 and Sec12. Interestingly, the combined expression of P. pastoris Sec16 and P. pastoris Sec12 in S. cerevisiae, but not of the single proteins, leads to a localization of the heterologous Sec12 to tER sites (Montegna et al., 2012). Pichia pastoris Sec16 is responsible for the recruitment of P. pastoris Sec12 at tER sites, suggesting a conserved role for the interaction between these proteins in the generation of COPII vesicles at the tER sites. This interaction is not conserved in S. cerevisiae and could explain why in this yeast, there are multiple tER sites and a lot of dispersed Golgi cisternae.

One of the first comparative study of Golgi structure in different yeasts showed that although all yeasts analyzed have tubular networks which should correspond to the Golgi compartment (Rambourg et al., 1995), the three-dimensional structures of these tubular networks are quite different. Only in H. polymorpha, P. pastoris, and S. pombe, the tubules are closely superimposed to each other in parallel arrays similarly to the stacked Golgi present in mammalian cells. Saccharomyces cerevisiae and Zygosaccharomyces rouxii have polygonal networks instead of arrays, and the third group of yeast, including K. lactis, C. albicans, and Candida parapsilosis, has arrays made out of only two to three layers. A more recent paper suggests that C. albicans might have a tubular structured Golgi similarly to mammalian cells (Isola et al., 2009).

The structural organization of the Golgi in mammalian cells and its role in protein trafficking require Golgi matrix proteins GRASP (Golgi reassembly stacking proteins) and golgins (Ramirez & Lowe, 2009; Lowe, 2011). Two GRASP proteins are present in vertebrates, GRASP65 and GRASP55, which play complementary and essential roles in Golgi stacking cisternae (Xiang & Wang, 2010). Deletion of both genes in Hela cells leads to a strong reduction of Golgi stacking and deletion of one of the two genes leads to a reduction in the number of cisternae per stack. GRASP65 interacts with the golgin GM130 which is involved in tethering at the cis-Golgi (Barr et al., 1998). However, the role of GRASP cannot be generalized to all eukaryotes. Plant cells have stacked Golgi but lack any GRASP homolog (Staehelin & Kang, 2008). Although the yeast GRASP homolog Grh1 is well conserved among the yeasts analyzed, not all yeasts manifest stacked Golgi. The role of Grh1 in P. pastoris and S. cerevisiae has been investigated in the study by Levi et al., 2010. The authors show that the deletion of Grh1 in P. pastoris does not alter the stacked Golgi structure, suggesting that other proteins might be necessary to keep the Golgi structure. Furthermore, the analysis of S. cerevisiae cells under glucose depletion, a condition that inhibits the secretory pathway with reduction in number of tER sites and increase in tER dimension, showed that Grh1 is not essential for the association between tER and cis-Golgi, suggesting a different role of this protein (Levi et al., 2010). Yeasts do not have a GM130 ortholog, but Behnia et al. found through in vitro assays that S. cerevisiae Grh1 forms a complex with the coiled-coil protein Bug1 (Behnia et al., 2007). The ability of Bug1 to interact with Sec23 and other proteins involved in the tethering processes suggests a role of the complex Grh1/Bug1 in the tethering of COPII. However, Bug1 is not conserved and a probable Bug1 homolog could only be observed for C. glabrata and K. lactis. Furthermore, GRASP proteins seem to be involved in the non-conventional secretion process, which will not be discussed in this review. It has been described that in Dictyostelium discoideum and P. pastoris (Kinseth et al., 2007; Manjithaya et al., 2010), the unconventionally secreted protein ACBP (Acyl Coenzyme A-binding protein) requires GRASP proteins for its secretion.

Back to start: retrograde transport from Golgi to ER

The transport of proteins from the Golgi to ER is mediated by COPI vesicles (Gaynor et al., 1998). COPI vesicles also mediate the intra-Golgi trafficking from trans- to cis-Golgi but, as reviewed recently for mammalian cells, there are structural differences among these two COPI vesicle populations (Popoff et al., 2011; Szul & Sztul, 2011). The COPI complex is composed of the Rab Arf1, which cycles between a cytosolic GDP bound inactive state and a GTP-bound active state anchored to the membranes (Spang, 2002) and the coatomer, a heptameric complex based on the subunits α, β, β′, γ, δ, ε and ζ (Cop1, Sec26, Sec27, Sec21, Ret2, Sec28, and Ret3) (Gaynor et al., 1998). The first step in COPI assembly is the interaction of Arf1 with cis-Golgi membranes. This process is initiated by the exchange of GDP with GTP by the GEF Gea1/2 (Peyroche et al., 1996). In vitro experiments and analyses of the COPI budding process in mammalian cells suggest that the nucleotide exchange induces conformational changes on Arf1 which allow the binding of Arf1-GTP to the Golgi membrane (Antonny et al., 1997). As described for mammalian cells, additional conformational changes might be necessary to allow the binding of Arf1 to the subunits of the coatomer (Zhao et al., 1997). In vitro studies suggest that the ability of the coatomer to interact with different protein partners could be connected to its ability to adopt different conformations in terms of shape and size (Faini et al., 2012). Once the COPI vesicles arrive at the ER membrane, Arf1 has to be inactivated to allow the release of the cargo proteins. Due to its low GTPase activity, the hydrolysis of GTP to GDP is facilitated by the GAP Glo3 (Yahara et al., 2006).

The selection of cargo proteins for COPI vesicles is mediated in S. cerevisiae by several retrieval systems (Gaynor et al., 1998). One system is based on the binding of the dilysine motif KKXX (Schröder-Köhne et al., 1998; Jackson et al., 2012). The observation that the dilysine ER retention signal of yeast Wbp1 functions also in mammalian cells suggests that this motif is well conserved (Cosson & Letourneur, 1994). Recently, Alisaraie & Rouiller (2012) identified binding sites for other defined peptide motifs in the γ and ζ subunits of the coatomer. Furthermore, cells have specific cargo receptors like Rer1 which bind Sec12 (Sato et al., 1995) and Erd2 which recognize the HDEL motif (Semenza et al., 1990; Semenza & Pelham, 1992; Townsley et al., 1994). While S. cerevisiae Erd2 has been reported to specifically accept only HDEL as ER retrieval sequence, the Erd2 receptors of other yeasts species seem to tolerate more variants of this motif including ADEL, DDEL, KDEL, QDEL, RDEL, and more. Especially C. glabrata and K. lactis, and the more distantly related S. pombe, have a low preference for HDEL (as can be seen for ER-resident proteins listed in Tables 2 and 3, and Tables S2 and S3). In mammalian cells, three KDEL receptors allow the recognition of more than 50 variants of the ER retention motif (Raykhel et al., 2007). Alanen et al. (2011) suggested that also the two amino acids upstream of the KDEL motif are important for receptor binding in mammalian cells. This should be investigated in yeasts as well, considering also the high diversity of the ER retention motif in different yeast species. Differently, an arginine-based motif is present on the cytosolic tail of membrane proteins and can control the ‘maturation’ of membrane protein complexes. If the complex is fully assembled, the sorting motifs are masked and the proteins can be exported to the cell surface, while if the complex is not fully assembled, it is transported back to the Golgi (Michelsen et al., 2005).

The transport of COPI vesicles from cis-Golgi to ER, but not the intra-Golgi transport, also requires the p24 complex (Aguilera-Romero et al., 2008; Strating & Martens, 2009; Szul & Sztul, 2011). p24 proteins can interact with Arf1-GDP and COPI subunits. However, the eight members of the p24 family in yeast are not essential proteins. The deletion of all eight p24 in S. cerevisiae is not lethal and leads only to a minor defect in the cargo transport without significant accumulation of unfolded proteins in the ER (Springer et al., 2000).

As previously described for COPII vesicles, also the fusion of COPI vesicles to the ER target membrane requires a tethering step which is mediated by the complex Dsl1. The tethering complex Dsl1 is localized at the ER and contains three subunits named Dsl1, Tip20, and Sec39, which are stably associated with three ER-localized SNAREs, named Use1, Ufe1, and Sec20 (Kraynack et al., 2005). These three proteins together with Sec22 make up the SNARE complex for the retrograde transport (Dilcher et al., 2003). The interaction of the tethering complex with COPI is mediated by the subunit Dsl1, but this is not a common mechanism to all yeasts compared. In S. pombe, no Dsl1 homolog has been found and it has been suggested that COPI binding should be mediated by the Tip20 subunit (Schmitt, 2010). The model described for the Dsl/SNARE complex suggests that Tip20 and Sec39 are kept at the ER membrane by Sec20 and Use1, whereas Dsl1 can expose its unstructured COPI-binding domain in a flexible way (Tripathi et al., 2009). The SNARE assembly in the retrograde transport to the ER is mediated by the interaction of the SM protein Sly1 with Sec20, Use1, and Ufe1, an interaction that seems to be important to protect Ufe1 from the ERAD degradation process (Braun & Jentsch, 2007). The direct interaction of the Rab Ypt1 with Ufe1 as well with Gea1/2 (Jones et al., 1999) supports the hypothesis that Ypt1 is also the Rab required for retrograde transport of COPI vesicles from cis-Golgi to ER. In the absence of Ypt1, the cells are still able to produce COPI vesicles but with reduced amounts of the cargo proteins Emp47 and Sec22 (Kamena et al., 2008). In mammalian cells, the Dsl1 equivalent complex, named Zw10, is involved in tethering at the ER membrane as well as in the capturing of microtubules by kinetochore during mitosis, a double function that seems to be common in ancestral eukaryotes (Schmitt, 2010). This suggests that the simple transport system present in yeast could be a consequence of the loss of chromosomes as organizing centres for the microtubules mediated transport. Gupta & Brent Heath (2002) compared SNAREs in the organisms S. cerevisiae, S. pombe, C. albicans, N. crassa, Aspergillus fumigatus, and Phanerochaete chrysosporium. They found that Ufe1, Sec20, Sec22, and Bet1 are present in all these organisms. In the yeasts comparison we performed, the H. polymorpha ortholog of Sec20 could not be identified. Except for C. glabrata and K. lactis, the degree of identity between S. cerevisiae and other Sec20 homologs is rather low, especially in the N-terminal cytosolic region (Weber et al., 2004). Sec20 has an effect on glycosylation in S. cerevisiae which was shown to be independent to that on protein secretion (Schleip et al., 2001). Furthermore, it was found that H. polymorpha Bet1 also contains an Use1 domain. This could indicate that this protein fulfills both functions but considering the incomplete state of the genome annotation of the organism, it is possible that some of the missing proteins will still be found at a later state.

Keeping proteins in the system: intra-Golgi trafficking

The transport protein particle TRAPPII

The transport from endosome to trans-Golgi and from trans- to cis-Golgi is mediated by the tethering complex TRAPPII, which is able to bind COPI, but not COPII vesicles. This complex contains all TRAPPI subunits except Trs85 and four additional subunits named Trs120, Trs130, Trs65, and Tca17 (Sacher et al., 1998). It is known that the TRAPPI complex is localized at the cis-Golgi and acts as GEF for Ypt1, which is necessary for the entry of vesicles into the cis-Golgi. It has been proposed that the subunits Trs120 and Trs130 might join the TRAPPI complex at the trans-Golgi inducing the formation of the TRAPPII complex and the switch of the GEF activity from Ypt1 to Ypt31, which is required for the exit from the Golgi (Morozova et al., 2006). The dynamic model proposed by the authors to explain the switch of the TRAPP complex is in agreement with the Golgi cisternal maturation model (Losev et al., 2006; Matsuura-Tokita et al., 2006). The essential subunits Trs120 and Trs130 have a dual important function; they are necessary for the GEF activity on Ypt31 and inhibit the GEF activity on Ypt1. The subunit Trs65 seems to mediate the dimerization of the TRAPPII complex, and it might be that TRAPPII uses the same catalytic site of TRAPPI. However, it is still not completely understood how the TRAPPII-specific subunits induce the changes on the COPII binding site in favor of a binding site for COPI vesicles (Yip et al., 2010). It seems that the two complexes are able to distinguish between the different coat proteins. However, COPI vesicles interact with several vesicle tethering complexes including Dsl1 and COG suggesting that there must be other components which are able to distinguish among the different compartments (Sacher et al., 2008). The observation that Trs65 binds the Arf-GEF Gea2 and the gamma subunit of the COPI coat and that Gea2 itself can bind the same COPI subunit suggests that TRAPPII is involved in the Arf1-GEF effector loop which is important to stabilize the TRAPPII complex at the Golgi membranes (Chen et al., 2011).

Tca17 has recently been confirmed to be a subunit of TRAPPII complex (Montpetit & Conibear, 2009; Choi et al., 2011). Tca17 acts together with Trs33 and Trs65 in promoting and stabilizing the TRAPPII complex. Tca17 is the only TRAPP subunit that is not conserved in all yeast species analyzed and is present only in C. glabrata, K. lactis, and P. pastoris. Most probably, in yeasts without Tca17 its role is taken over by the other TRAPPII subunits. Interestingly, Tca17 and Trs33 are conserved in higher eukaryotes, while Trs65 is present only in fungal species. Indeed, the human TRAPP complex is closely related to the yeast TRAPPII complex and up to date no equivalents of TRAPPI complex have been detected in mammalian cells (Scrivens et al., 2011). It has been described that mutations in Trs120 do not affect the integrity of cis- and trans-Golgi and that TRAPPII localizes the trans-Golgi/early endosome compartment which also contains Sec7 and Chs3. Furthermore, it has been suggested that Trs120 might be necessary to recycle proteins from the early endosome to the trans-Golgi (Cai et al., 2005). These proteins can be divided in three main groups: soluble vacuolar proteins like the CPY receptor, trans-Golgi resident proteins like Kex2, and integral plasma membrane proteins like the SNARE Snc1.

The conserved oligomeric Golgi complex

There is a second tethering complex that coordinates the retrograde transport from the endosome to the cis-Golgi, the conserved oligomeric Golgi complex (COG), which consists of eight subunits organized in two substructures, lobe A (subunits Cog1 to 4) and lobe B (subunits Cog5 to 8). While in mammalian cells, Cog1 and Cog8 are both important for the connection between the two lobes, in S. cerevisiae Cog1 has a more central role (Fotso et al., 2005). It has been described that the N-terminal region of Cog1 interacts with Cog2, Cog3, and Cog4, while its C-terminal region interacts with Cog8 (Lees et al., 2010). The deletion of the lobe B subunits does not affect yeast growth, while the deletion of Cog1 leads to severe growth defect and the deletion of Cog2, Cog3, or Cog4 is lethal in S. cerevisiae (Lees et al., 2010). In addition, the deletion of Cog1 causes a strong degradation of the lobe B subunits (Fotso et al., 2005). The lobe A subunits seem to be enough to provide a functional COG tethering process. It has been suggested that lobe A is more involved in cis-Golgi and lobe B in trans-Golgi. Surprisingly, homologs of Cog1 have been found only in C. glabrata, K. lactis, and S. pombe. Two more subunits of the COG complex are not well conserved: Cog7 which is missing in Y. lipolytica and S. pombe and Cog8 which is missing in H. polymorpha. Differently, from what was described by Fonzi et al., we could find a homolog of Cog2 also in C. albicans (Fonzi, 2009).

The main role of the COG complex in mammalian cells seems to be the sorting of glycosylation enzymes that must be correctly distributed within the Golgi stacks to allow the sequential and proper modification of glycoproteins. This means that acetylglucosaminyltransferase I and mannosidase II are usually present in cis/medial cisternae, while galactosyltransferase and sialyltransferase are mainly found in trans-Golgi. The role of COG subunits in regulating the position of glycosylation enzymes in the yeast Golgi compartments has not been yet completely investigated. However, it has been described that mutations of COG subunits lead to the accumulation of the precursor of CPY which lacks Golgi-specific sugar modification (Wuestehube et al., 1996). N-linked Golgi glycosylation of a secreted invertase was reduced by such mutations, as well as O-linked Golgi glycosylation on Hsp150 (Suvorova et al., 2002). Moreover, the COG complex seems to be involved in the retrograde transport of Och1 and Mnn1 (Bruinsma et al., 2004). Impaired glycosylation has been observed in cog3 mutants and might be a consequence of the absence of Och1 activity at cis-Golgi as a consequence of the incorrect localization of the glycosyl transferases. To understand the transport of Och1, sed5 and sft1 double mutants have been analyzed. Sed5 and Sft1 are, respectively, t- and v-SNAREs localized at the cis-Golgi and mediate the retrograde transport of cargo proteins to the cis-Golgi (Wooding & Pelham, 1998). The results suggest that Och1 is transported into retrograde vesicles from trans- to cis-Golgi. The COG complex known to function as tethering complex at cis-Golgi membranes might also act together with COPI vesicles to mediate the intra-Golgi retrograde transport of proteins like Och1 (Bruinsma et al., 2004). Furthermore, the phosphoprotein Sed5, additionally to its role in ER to Golgi transport, seems to be involved in maintaining the Golgi morphology. Mutations that do not allow the phoshorylation of Sed5 affect intracellular trafficking and lead to the accumulation of ordered stacked Golgi structures, which are typical for mammalian cells but not for S. cerevisiae (Weinberger et al., 2005).

Two distinct SNARE complexes exist within the Golgi. Sed5 forms a complex with Gos1, Sft1, and Ykt6 in the cis-Golgi, whereas Tlg2, Vti1, Tlg1, and Ykt6 form the SNARE complex for the trans-Golgi. Sft1 is essential for vesicle fusion in the trans- to cis-Golgi transport but interestingly, in agreement with a previous publication by Swennen & Beckerich (2007) it is missing in C. glabrata and P. pastoris. According to Kienle et al. (2009) Sft1 is also missing in Debaryomyces hansenii and several other species spread through the kingdom of fungi. Due to the absence of other parts of the intra-Golgi transport system, Fonzi (2009) proposes that C. albicans uses a different system of channeling proteins through this organelle. If this is true for the other species in which Sft1 is missing needs to be further elucidated. In mammalian cells, there is only one SNARE (GS15), which is similar to both Bet1 and Sft1p. For S. cerevisiae, it has been shown that Bet1 can take over the function of Sft1 (Gupta & Brent Heath, 2002) which might explain why Sft1 can be missing.

However, there are still open questions concerning the organization and regulation of the yeast Golgi apparatus (Suda & Nakano, 2011). Golgi enzymes needs to be transported back to cis-Golgi by COPI vesicles, but they lack known coatomer binding motifs. The ability of Vps74 to bind the cytosolic domain of cis- and medial-mannosyltransferases suggests that Vps74 might be involved in the Golgi retention of these enzymes (Schmitz et al., 2008; Tu et al., 2008). Vps74 is an effector of the phosphatidylinositol 4-kinase and seems to function together with the phosphatidylinositol 4-phosphatase Sac1, to restrict phosphatidylinositol 4-phosphate (PI4P) at the Golgi (Wood et al., 2012), suggesting that the sorting of resident Golgi enzymes is influenced by the Golgi lipid homeostasis. Besides that, it has been described that in mammalian cells phospholipid remodeling enzymes participate in the formation of vesicles and Golgi membrane tubules and that the lipids and phospholipids are important to generate membrane curvature in the process of vesicle budding (Ha et al., 2012).

Also in the intra-Golgi retrograde trafficking, the SNAREs assembly is mediated by the SM protein Sly1, which is able to selectively interact with Gos1 at the cis-Golgi and with Tlg2 at the trans-Golgi (Peng, 2005). In mammalian cells, the COG complex binds the SM protein Sly1 through the subunit Cog4 (Laufman et al., 2009). Furthermore, it has been shown that COG interacts with the long coiled-coil protein golgin, and it was suggested that this interaction might help the transfer of the vesicle from its budding site to the tip of the golgin toward the target membrane (Sohda et al., 2010). Like other tethering complexes also the COG complex interacts with the Rabs Ypt1 and Ypt6. It has been described that the overexpression of Ypt1 can suppress the defects associated with the loss of COG functions and that Cog3 interacts with the cis-Golgi t-SNARE Sed5 and several v-SNAREs involved in the transport to and from the cis-Golgi, as Gos1, Ykt6, and Sec22 (Suvorova et al., 2002). Moreover, the authors demonstrated that there is an interaction between the COG complex and the COPI vesicles. The subunit Cog2 interacts with the γ subunits of the coatomer and Arf1-GTP, suggesting a role of the COG subunits in the uncoating and fusion of COPI vesicles in the trans- to cis-Golgi trafficking. The interaction with Sed5 and Sec22 could be explained considering a role of the COG subunits in the recycling of these two proteins to the cis-Golgi. Furthermore, the combined loss of Ypt6 and one of the not essential COG subunits is lethal. The observation that the inactivation of the Rab Ypt6 leads to defects in protein glycosylation and in protein sorting at the trans-Golgi, suggests that this protein might be involved in the vesicular transport from the endosome to the Golgi and between the Golgi compartments (Luo & Gallwitz, 2003).

However, how exactly proteins which are synthesized at the ER are transported through the Golgi and further to the plasma membrane to be secreted is still under discussion. Two main models have been proposed to explain the intra-Golgi transport. The vesicular model suggests that Golgi cisternae are individual, long-lived and stable compartments that retain a defined set of resident proteins while cargo proteins move through the cisternae into vesicles. The cisternal maturation model suggests that Golgi cisternae are transient structures that form de novo, mature from cis to trans and then dissipate. In this model, cargo proteins move through the stacks within Golgi cisternae while resident proteins and cargo receptor are recycled back by the retrograde transport into COPI vesicles. Improved live cell imaging techniques demonstrated that yeast Golgi cisternae are dynamical compartment and that the distribution of resident membrane proteins changes over the time from cis to trans typical pattern, supporting the maturation model (Losev et al., 2006; Matsuura-Tokita et al., 2006; Glick & Nakano, 2009; Glick & Luini, 2011).

Way out of the cell: exocytosis

Post-Golgi secretory vesicles are tethered to the plasma membrane by the exocyst, an evolutionary well-conserved multimeric complex composed of the eight subunits Sec3, Sec5, Sec6, Sec8, Sec10, Sec15, Exo70, and Exo84 (Guo et al., 1999). The exocyst is localized at defined plasma membrane regions and it is necessary for a polarized membrane growth. In S. cerevisiae, it is localized at the tip of the growing bud and, during cytokinesis, at the mother-bud neck (He & Guo, 2009) and in S. pombe it is localized at the division septum during membrane scission (Wang et al., 2002). Involvement of the exocyst complex in polarized growth has also been described for mammalian cells, for example, in epithelial cells (Yeaman et al., 2004) and neurons (Hazuka et al., 1999). However, it has been described that the subunits of the exocyst complex are targeted to the plasma membrane with different mechanisms. The localization of Sec3 is actin independent, Exo70 is partially actin dependent, and the remaining six subunits, which are associated with the secretory vesicles, need actin cables to be delivered to the sites of exocytosis (Boyd et al., 2004). Rho GTPases are very important GTPases for the polarization of the actin cytoskeleton. They localize at the plasma membrane and recruit the exocyst subunits Sec3 and Exo70 to the site of polarization. Sec3 interacts with Rho1, Cdc42 and phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) (Guo et al., 2001) while Exo70 interacts with Rho3, Cdc42, and PI(4,5)P2 (Adamo et al., 1999, 2001; Robinson et al., 1999; Zhang et al., 2001). These data suggest that Rho GTPases and PI(4,5)P2 act together in controlling the exocyst function at the bud tip. The importance of phosphoinositides in the exocytic pathway is well described by Behnia & Munro (2005).

Secretory vesicles are transported from the TGN to the plasma membrane along actin filaments using the type V myosin motor Myo2 (Jin et al., 2011). Myo2 is recruited at newly formed secretory vesicles by the Rab Ypt31/32 which is known to be necessary for the formation of secretory vesicles (Jedd et al., 1997). Furthermore, Ypt31/32 together with PI4P, which is present in high amounts at the TGN and in small buds, recruits the GEF Sec2 that activates the Rab Sec4 (Ortiz et al., 2002; Lipatova et al., 2008). Myo2 is the first described example of myosin that needs phosphoinositide to bind its cargo. The activation of Sec4 is necessary to allow the polarized transport of vesicles to exocytic sites, indeed only Sec4 in its activated form is able to directly interact with Myo2 (Jin et al., 2011; Santiago-Tirado et al., 2011). Next is the interaction of Sec2 with the exocyst subunits Sec15, which leads to the release of Ypt31/32 and enables the binding of Sec15 to Sec4-GTP and Myo2 (Jin et al., 2011). The other exocyst subunits are recruited to the vesicles to allow their fusion to the plasma membrane. While the binding of Myo2 to Ypt31/32 and Sec4 seems to happen in two progressive steps, it is still not clear whether the binding of Myo2 with Sec4 and Sec15 happens simultaneously or in two separate steps. Sec4 appears to be negatively regulated by phosphorylation and the removal of these phosphate groups seems to be necessary to allow its interaction with Sec15 (Heger et al., 2011). The exocyst subunit Sec6 plays an important role in the formation of the SNARE complex at the plasma membrane together with the SM protein Sec1. If Sec6 is alone without the other exocyst subunits, it binds the tSNARE Sec9 preventing its interaction with the partner synthaxin Sso1 or the homolog Sso2, whereas if the entire exocyst complex is present, Sec6 binds Sec1 (Morgera et al., 2012) enabling the formation of the SNARE complex Sec9-Sso1. In contrast to the mammalian homolog Munc18, Sec1 binds strongly only to the fully assembled SNARE core complex consisting of Sso1, Sec9, and Snc1 and not Sso1 alone (Peng, 2005). However, Sec1 seems to be necessary before and after the assembly of the SNARE complex, suggesting a role in both vesicle docking and membrane fusion (Hashizume et al., 2009). The following step is the interaction of the Sec9/Sso1 with the v-SNARE Snc1 or Snc2 in a ternary complex which leads to the fusion of the membranes (Nicholson et al., 1998). This fusion step seems to be further supported by the binding of Sec4 with Sec9 (Jin et al., 2011). The polarized exocytosis is also mediated by the tomosyn proteins Sro7 and its homolog Sro77, which interact with Sec9 and Sec4, targeting the vesicles to the right position on the plasma membrane (Hattendorf et al., 2007). With the exception of Y. lipolytica, all selected yeast species contain only one Snc, one Sso, one Sec9, and one tomosyn homolog. Yarrowia lipolytica is the only species which has three Sso and two Snc proteins. Swennen & Beckerich (2007) argue that the additional copies reflect the high secretion efficiency and the capacity to induce hyphal growth. No homologs for the sporulation-specific SNARE Spo20 have been found which is consistent with previous results (Gupta & Brent Heath, 2002).

An additional protein involved in the secretion from the trans-Golgi is the essential protein Sec14. Sec14 is a phosphatidylcholine and phosphatidylinotisol transfer protein that regulates lipid metabolism and, in addition to its primary role in the endosome trafficking from the plasma membrane to the vacuole, seems to be responsible for the trans-Golgi export of specific proteins like Bgl2 and, to a certain extent, of Snc1 (Curwin et al., 2009).

The proteins involved in the exocytosis process are well conserved among the yeast species analyzed here.

Conclusions

A critical review of the state of knowledge of the protein secretion pathway in yeasts unveils the complex nature of these processes. Naturally most research has been carried out with S. cerevisiae. While analogies to mammalian cell processes may enhance our understanding where direct information from yeasts is missing, this bears the risk of some misinterpretations where biochemical functions of orthologous genes have changed. We suggest closing gaps rather by studying other yeasts which has become significantly more popular recently due to the development of efficient experimental tools and the availability of genome sequences. One pillar of this review is a thorough genomic comparison of the consecutive steps of protein secretion in eight different yeast species. We could identify significant differences which call for being investigated in detail in the future.

The recent years have brought many details of the translocation process to light. The translocon pore Ssh1 has returned into the research focus again as an alternative to the canonical translocon complex Sec61. The importance of post-translational translocation is well established for yeasts today and has been described also for mammalian cells recently. The primary role of protein glycosylation in folding quality control has received even more emphasis. Surprisingly one major function of this process, UDP-glucose:glycoprotein glucosyltransferase enabling the re-entry of the calnexin cycle is missing in S. cerevisiae while this gene is still available in other yeasts. For a thorough understanding of the folding process, it will be important to study the calnexin cycle in vivo in different yeasts. Golgi glycosylation depends mainly on mannosyl tranferases in S. cerevisiae, while other yeasts may use alternative glycoslyltransferases for the terminal decoration of their glycans. Especially the high abundance of α 1,2-mannosyl transferase homologs in the two methylotrophic yeasts, and of α 1,3-mannosyl transferases in the two pathogenic Candida species should be investigated for their function and regulation in future. The presence of additional ER-localized protein disulfide isomerase homologs esp. in H. polymorpha and P. pastoris calls for further research. Do they reflect the methylotrophic lifestyle of these yeasts, and do they act as reductases or oxidases? It is still not clear even for S. cerevisiae which Pdi family member is responsible for reducing proteins during ERAD. Significant differences were highlighted in the vesicle transport process with a number of genes being unidentified in several yeasts other than S. cerevisiae.

However, it should be noted that even the presence of a homologous gene does not necessarily mean that the biochemical process behind is similar. Gene regulation is significantly different among yeast species, thus tuning the available secretion machinery to the needs of the cell. As one example, ERAD and UPR are co-regulated in S. cerevisiae, while these processes are not linked on a regulatory level in Y. lipolytica (Boisramé et al., 2006). One interesting point in discussion is the potential feedback regulation of protein excretion (the final step of secretion) on the initial translocation process. Understanding this regulation would be of utmost importance for the application of yeasts as production hosts of secreted proteins or may enable a novel therapeutic strategy against pathogenic yeasts. On the one hand, specific inhibition of secretion of virulence factors is of utmost interest in case of pathogenicity. On the other hand, the ever growing demand of biopharmaceuticals and the need for large amounts of enzymes mainly in the upcoming biorefinery industry substantiate a huge demand for efficient protein production processes, which mostly rely on the secretion capacity of the production host. Thus, the importance of the complex process of protein secretion both for basic research and for application in an industry of extremely high value becomes obvious.

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Table S1. Secretory pathway.

Table S2. Protein prolyl isomerase family (PPI).

Table S3. Chaperones of the Hsp70 family & their NEFs.

Acknowledgements

Research on protein secretion in our laboratory is supported by the Austrian Science Fund (FWF), the Austrian Research Promotion Agency, the European Science Foundation (ESF), and by the Federal Ministry of Economy, Family and Youth (BMWFJ), the Federal Ministry of Traffic, Innovation and Technology (bmvit), the Styrian Business Promotion Agency SFG, the Standortagentur Tirol and ZIT—Technology Agency of the City of Vienna through the COMET-Funding Program managed by the Austrian Research Promotion Agency FFG. ABG was supported by the project ‘Competence Team Molecular Biology’ by the City of Vienna, MA27. Further support by Polymun Scientific GmbH, Biomin Research Center, Boehringer-Ingelheim RCV, Lonza AG, Biocrates Life Sciences AG, VTU Technology GmbH, and Sandoz GmbH is acknowledged. We thank the BOKU-VIBT Imaging Center for access to Leica fluorescence microscope devices. Verena Puxbaum kindly provided fluorescence micrographs of P. pastoris organelles.

Footnotes

  • Editor: Gerhard Braus

References