The unicellular organisms Escherichia coli and yeast Saccharomyces cerevisiae have roughly 4300 and 6600 genes, respectively, whereas the multicellular organism human contains 22,000–26,000 genes. One may feel that the number of the human genes is too small, given the complex architecture and function of human being. The limited number of genes in multicellular organisms may be due to a strong selective pressure; cells might have evolved not to increase the number of genes because more energy and time are necessary to carry extra genes and accurately transcribe and translate them. To prevent the increase in gene number, multicellular organisms might adopt a strategy to reuse certain proteins for very different cellular processes. Jeffery (1999) coined the term “moonlighting protein” to describe proteins with multiple roles. The list of moonlighting proteins continued to expand and now includes components implicated in membrane trafficking (reviewed by Copley, 2012; Royle, 2013).

The Dsl1 complex in yeast comprises Dsl1 (VanRheenen et al., 2001), Tip20 (Sweet and Pelham, 1993), and Sec39/Dsl3 (Mnaimneh et al., 2004; Kraynack et al., 2005). This complex is a member of the Complex Associated with Tethering Containing Helical Rods (CATCHR) family (Yu and Hughson, 2010) and has been implicated in tethering of Golgi-derived transport vesicles with the endoplasmic reticulum (ER) (Andag et al., 2001; Reilly et al., 2001; Andag and Schmitt, 2003; Zink et al., 2009; Diefenbacher et al., 2011). In 2004, using an immunoaffinity method we isolated from rat liver membranes a large complex containing syntaxin 18, an ER-associated soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) implicated in membrane fusion (Hatsuzawa et al., 2000), and found that the complex includes ZW10 and RINT1 (Hirose et al., 2004). Several lines of evidence suggest that ZW10 and RINT1 are the mammalian counterparts of yeast Dsl1 and Tip20, respectively, although the amino acid sequence identities between the yeast and mammalian proteins are very low (~14%). Intriguingly, both ZW10 and RINT1 had been discovered as cell cycle checkpoint proteins (Williams et al., 1992; Xiao et al., 2001). Nowadays, the mammalian Dsl1 complex is known to participate not only in membrane trafficking and cell cycle but also in other cellular processes including autophagy. In this review, we will focus on the moonlighting functions of the mammalian Dsl1complex, alternatively called the NRZ complex. Regarding the detailed structural and functional features of the Dsl1 complex and other CATCHR family complexes, we refer readers to some excellent reviews (Bröker et al., 2010; Brown and Pfeffer, 2010; Schmitt, 2010; Yu and Hughson, 2010; Bonifacino and Hierro, 2011; Spang, 2012; Hong and Lev, 2014).

In eukaryotic cells, communication between organelles in the secretory and endocytic pathways is mediated by membrane-bound vesicles that transit between organelles. Transport vesicles are formed at the donor compartment, traffic to their destination, lose their coat, and fuse with the acceptor compartment. Docking and fusion of transport vesicles with the target membrane involve an initial contact mediated by Rab GTPases and tethering factors, followed by SNARE-catalyzed membrane fusion.

Tethering factors, which not only facilitate long-range interactions between transport vesicles and the acceptor membrane but also coordinate SNARE complex assembly, can be classified into two groups, large proteins with extended coiled-coils and multisubunit protein complexes. The CATCHR complexes are a subfamily of the multisubunit tethering complexes consisting of the Dsl1, COG, exocyst, and GARP complexes (Yu and Hughson, 2010). Despite subtle sequence homology among their subunits, the CATCHR family tethers have strong three-dimensional structural homology.

The yeast Dsl1 complex, which comprises three subunits (Dsl1, Tip20, and Sec39/Dsl3) (Figure 1A), has been implicated in tethering of Golgi-derived transport vesicles on the ER (Andag et al., 2001; Reilly et al., 2001; Andag and Schmitt, 2003; Zink et al., 2009; Diefenbacher et al., 2011). Although all these proteins are rich in α-helices, their origins may be different; Dsl1 and Tip20 have a common α-helical fold, thus originating from an ancestral CATCHR protein (Tripathi et al., 2009), whereas Sec39/Dsl3 lacks the shared fold and thus appears to be derived from a non-related protein (Ren et al., 2009). The Dsl1 complex resides on ER membranes and binds to and regulates the assembly of the ER SNAREs Ufe1, Sec20, Use1/Slt1, and Sec22 (Figure 2A) (Kraynack et al., 2005; Ren et al., 2009; Diefenbacher et al., 2011). The Dsl1 complex has been predicted to form a 20-nm-tall tower from the ER surface, which can allow an interaction with COPI-coated vesicles (Ren et al., 2009). The COPI (α- and δ-COPs)-binding sites in Dsl1 have been defined to its central acidic region (Andag and Schmitt, 2003), which is located in the tip of the tower (Ren et al., 2009). An additional role of the Dsl1 complex is to assist uncoating of COPI-coated vesicles tethered on ER membranes. In this role, Dsl1 may block domains in COPI that drive repolymerization and the formation of large COPI aggregates (Zink et al., 2009).

While the roles of the yeast Dsl1 complex have been so far entirely limited to the tethering function on the ER, one study has suggested the involvement of the Dsl1 complex in peroxisome biogenesis. Previously, peroxisomes were recognized as an autonomous organelle, but recent studies have revealed a primary role of the ER in the de novo formation of peroxisomes (reviewed by Agrawal and Subramani, 2013). Searching for ER-associated proteins responsible for this role revealed that all Dsl1 complex components are required for peroxisome biogenesis (Perry et al., 2009). It was speculated that Dsl1 complex components may function as a tether for retrograde carriers from peroxisomes (Nagotu et al., 2010) or an anchor to recruit dynein for peroxisome biogenesis (Perry et al., 2009), but the mechanism by which Dsl1 complex subunits regulate peroxisome biogenesis should be elucidated in future studies.

The mammalian counterpart of the Dsl1 complex was identified by us (Hirose et al., 2004; Aoki et al., 2009) and was later called the NRZ complex for its subunit names, NAG (Sec39/Dsl3), RINT1 (Tip20), and ZW10 (Dsl1) (Civril et al., 2010) (Figure 1B and Table 1). Like the yeast Dsl1 complex (Sweet and Pelham, 1993; Kraynack et al., 2005; Ren et al., 2009; Tripathi et al., 2009; Diefenbacher et al., 2011), the NRZ complex is associated with the ER SNAREs syntaxin 18 (Ufe1), BNIP1 (Sec20), p31 (Use1/Slt1), and Sec22b (Sec22) (Figure 2B) (Hatsuzawa et al., 2000; Hirose et al., 2004; Nakajima et al., 2004; Aoki et al., 2009; Uemura et al., 2009). The yeast Dsl1 complex interacts with another SNARE, Ykt6 (Meiringer et al., 2011), but it is not clear whether or not the NRZ complex binds to the mammalian ortholog of this SNARE. The mechanism underlying SNARE complex assembly appears to be somewhat different between mammals and yeast. In mammals, the assembly of the ER SNAREs occurs in the absence of RINT1 (Arasaki et al., 2006), whereas yeast Tip20 plays a pivotal role in ER SNARE complex assembly (Kraynack et al., 2005; Diefenbacher et al., 2011). Moreover, the binding of Sec22b to syntaxin 18 in mammals creates high-affinity binding sites for BNIP1 and p31 (Aoki et al., 2008), whereas yeast Ufe1, Sec20, and Use1/Slt1 form a stable complex in the absence of Sec22 (Kraynack et al., 2005).

The Dsl1 complex is conserved in plants. Screening for Arabidopsis mutants that abnormally accumulate the precursors of storage proteins, 2S albumin and 12S globulin, in dry seeds identified a mutant, designated maigo2 (MAG2: At3g47700) (Li et al., 2006). MAG2 is the ortholog of RINT1 (Tip20). Affinity purification identified three MAG2-binding proteins, MIP1 (At2g32900), MIP2 (At5g24350), and MIP3 (At2g42700). MIP1 and MIP2 share sequence homology with ZW10 (Dsl1) and NAG (Sec39/Dsl3), respectively, (Li et al., 2013). MIP3 is a member of the Sec1 family domain-containing proteins, named SCFD2. SCFD2 is present in mammals, and may bind to ZW10 (Hutchins et al., 2010), although our original study failed to detect this protein in the syntaxin 18 immunoprecipitates (Hirose et al., 2004). Instead, we found Sly1/SCFD1 in the precipitates (Hirose et al., 2004), likely due to the direct binding of Sly1/SCFD1 to syntaxin 18 (Yamaguchi et al., 2002). The Arabidopsis Dsl1 complex is implicated in abscisic acid-mediated response to abiotic stresses. This response may be related to Dsl1 complex-mediated membrane trafficking between the ER and Golgi (Zhao et al., 2013). There is a homolog of MAG2, named MAG2-like (MAG2L: At1g08400) in Arabidopsis (Li et al., 2006), and MAG2L is partially redundant with MAG2 in response to environmental stresses (Zhao et al., 2013).

To preserve genetic information, the genomes of organisms must be accurately replicated and segregated before cell division. In eukaryotes, during mitosis and meiosis, sister-chromatid segregation occurs after all kinetochores form stable bipolar microtubule attachments. If not correctly attached to the spindle, kinetochores activate the spindle assembly checkpoint, leading to the block of cell cycle progression.

ZW10 (Zeste White 10) and ROD (Rough Deal) were discovered as proteins that are required for faithful chromosome segregation in Drosophila (Karess and Glover, 1989; Williams et al., 1992). Both proteins were found to be conserved in humans (Starr et al., 1998; Scaërou et al., 2001). At the onset of mitosis, these proteins redistribute from the cytoplasm to kinetochores (Williams et al., 1992; Scaërou et al., 1999) and recruit the core spindle checkpoint proteins Mad1-Mad2 to microtubule-unattached kinetochores (Buffin et al., 2005; Kops et al., 2005; Gassmann et al., 2010). Once all kinetochores become stably attached to the spindle, ZW10 and ROD are transported away from kinetochores by dynein (Howell et al., 2001; Wojcik et al., 2001; Basto et al., 2004). The removal of these proteins from kinetochores acts as a signal for the termination of the spindle checkpoint (Basto et al., 2000; Chan et al., 2000). ZW10 interacts with dynamitin, a subunit of the dynein-dynactin complex (Echeverri et al., 1996), thereby recruiting this motor to kinetochores (Starr et al., 1998). ZW10 and ROD form a large complex of 800 kDa, and immunoaffinity purification identified the third component, Zwilch (Williams et al., 2003), which encouraged Karess (2005) to call this complex RZZ for ROD-ZW10-Zwilch (Figure 1C). The binding of ROD and NAG to ZW10 is mutually exclusive, suggesting that the RZZ and NRZ complexes exist as distinct entities. Structural prediction revealed unexpected similarity between ROD and NAG. They share an N-terminal β propeller followed by an α solenoid, which is a characteristic structure of certain nucleoporins such as Nup133 and vesicle coat subunits such as clathrin heavy chain and α-COP (Civril et al., 2010). In Nup133, the C-terminal α solenoid structure is responsible for the localization at the nuclear pore complex in interphase and at kinetochores in mitotic cells, whereas the N-terminal β propeller domain interacts with CENPF, which can recruit dynein-dynactin via NudE or its homolog NudEL (Stehman et al., 2007; Bolhy et al., 2011). This protein–protein interaction chain (Nup133-CENPF-NudE or NudEL-dynein-dynactin) is reminiscent of the RZZ-mediated recruitment of dynein-dynactin to kinetochores. ROD binds to Zwilch through its N-terminal β propeller domain (Civril et al., 2010), as does Nup133 to CENPF.

Zwilch and RINT1, both of which are ZW10 binding partners, do not share any structural similarity, suggesting that Zwilch is not a descendant of the ancestral CATCHR protein (Civril et al., 2010). Moreover, in contrast to RINT1, Zwilch and ROD are not well conserved in eukaryotes (Schmitt, 2010), raising the possibility that the RZZ complex was generated after the occurrence of the NRZ complex accompanied by the acquisition of an additional function by ZW10. This is consistent with the view that the RZZ complex is a fairly recent add-on to the core spindle checkpoint complex (Vleugel et al., 2012). ROD and Zwilch always co-occur in Opisthokonta, and their co-occurrence correlates well with the presence of flagella and centrioles in Opisthokonta (Schmitt, 2010). The co-appearance of ROD and Zwilch might allow microtubules and their motor dynein to function in the termination of spindle checkpoint.

In Drosophila there is no NAG ortholog, and ROD may function as a substituent for NAG in membrane trafficking (Wainman et al., 2012). In Drosophila spermatocytes, ROD and ZW10 accumulate at the Golgi apparatus, as well as the ER in the case of ZW10, and these proteins are required for Golgi stack integrity (Wainman et al., 2012). The presence of ZW10 on both ER and Golgi membranes has also been reported in COS-7 (Varma et al., 2006) and rodent cells (Arasaki et al., 2007). Depletion of Drosophila RINT1, as well as mutations in the ZW10 and ROD genes, causes apparent alteration in Golgi morphology and reduces the number of Golgi stacks. In contrast, Zwilch is not associated with membranes and its mutation does not affect Golgi structure (Wainman et al., 2012). ZW10 and RINT1 are necessary for another membrane trafficking-related event, meiotic cytokinesis in spermatocytes, but not for mitotic one (Williams et al., 1992). In ZW10 or RINT1 mutant spermatocytes, regular central spindles and actomyosin rings can assemble, but furrow ingression halts prematurely due to a defective plasma membrane addition. On the other hand, neither ROD nor Zwilch is required for cytokinesis (Wainman et al., 2012).

Figure 3 shows the summary of protein–protein interactions between NRZ components. Many, but not all, interactions are mediated through putative coiled-coil regions. As described above, the NRZ complex, as well as the Dsl1complex, consists of two distinct structural units. ZW10 and RINT are derived from the CATCHR family ancestor, and NAG is from a different one. Because a CATCHR family member RINT1 interacts with another CATCHR family member COG1 (Arasaki et al., 2013), it is tempting to speculate that RINT1 orthologs in other eukaryotes may also interact with CATCHR family members. Indeed, a comprehensive analysis of protein–protein interactions in yeast demonstrated that Tip20 can interact with COG4 (Uetz et al., 2000). Like RINT1, ZW10, and its orthologs may also interact with CATCHR family members. Given that SCFD1 and SCFD2 bind to the mammalian COG4 (Laufman et al., 2009) and plant MAG2 (Li et al., 2013), respectively, it is worth examining the interactions between CATCHR family members and Sly1 family (SCFD) proteins. The currently available data suggest that the role of RINT1 in autophagy is passive; it anchors UVRAG until autophagy induction. However, as the ER is a major source of autophagosome formation (reviewed by Lamb et al., 2013), RINT1 and other NRZ complex subunits may have more active roles in autophagy.

Schmitt (2010) has proposed that the ancestor of animals and fungi probably expressed both sets of ZW10-interacting proteins (NAG- RINT1 and ROD-Zwilch). Rod and Zwilch encoding genes have been assumed to be lost independently in different phylogenetic branches, and Sec39 likely lost the N-terminal β-propeller domain in parallel with the acquisition of the acidic COPI-interacting region by Dsl1. We would like to suggest another possibility. The ancestor of NAG/Sec39 acquired a β-propeller domain with a C-terminal extension for yet unrevealed functions, and the occurrence of Zwilch might facilitate the appearance of ROD by gene duplication. In this context, it is important to identify proteins that interact with the β-propeller domain and/or C-terminal extension of NAG. Moreover, elucidation of the molecular mechanism by which NAG mediates the NMD pathway may provide a clue to the reason for the association/cooperation of CATCHR proteins with NAG.