Nvj1p resides in the outer nuclear membrane (ONM) and binds the vacuole membrane protein Vac8p to form nucleus-vacuole (NV) junctions in Saccharomyces cerevisiae. The induction of NVJ1 expression during starvation results in the sequestration of two additional binding partners, Tsc13p and Osh1p. Here, we map the domains of Nvj1p responsible for ONM targeting and partner binding. ONM targeting requires both the N-terminal signal anchor-like sequence and the topogenic membrane-spanning domain of Nvj1p. The N-terminal signal anchor-like sequence may anchor Nvj1p in the ONM by bridging to the inner nuclear membrane. A region encompassing the membrane-spanning domain is sufficient to bind Tsc13p. Osh1p and Vac8p bind to distinct regions in the cytoplasmic tail of Nvj1p. Overexpression of Nvj1p in trp1 cells causes a growth defect in low tryptophan that is rescued by additional copies of TAT1 or TAT2 tryptophan permeases. Conversely, nvj1trp1 cells grow faster than NVJ1+ trp1 cells in limiting tryptophan. Importantly, deleting the Osh1p-binding domain of Nvj1p abrogates the tryptophan transport-related growth defect of Nvj1p-overexpressing cells. Therefore, the Nvj1p-dependent sequestration of Osh1p negatively regulates tryptophan uptake from the medium, possible by affecting the trafficking of tryptophan permeases to the plasma membrane.

The endoplasmic reticulum (ER) is an interconnected membrane network composed of structurally distinct compartments, including the peripheral or cortical ER and the perinuclear ER, which surrounds the nucleus (reviewed in Voeltz et al., 2002). Although these compartments are topologically continuous and share a common lumen, each has unique properties and functions (reviewed in Levine and Rabouille, 2005; Voeltz et al., 2002). The perinuclear ER, which comprises the nuclear envelope (NE), consists of two concentric membrane sheets connected at fenestrae called nuclear pores. The inner nuclear membrane (INM) serves as an attachment sites for chromatin and, in higher eukaryotes, is associated with the nuclear lamina (reviewed in Hetzer et al., 2005). The outer nuclear membrane (ONM), like the rough ER, is studded with ribosomes but, uniquely, for example, contains cytoskeletal attachment sites for nuclear positioning (reviewed in Hetzer et al., 2005). The mechanisms responsible for creating and maintaining these structurally distinct compartments are poorly understood.

Accurate protein sorting to the nuclear envelope is critically important for nuclear function and cell survival. Mutations affecting the targeting and function of specific nuclear envelope proteins have been linked to several human diseases collectively termed nuclear envelopathies (reviewed in Somech et al., 2005). In general, proteins localize to the nuclear envelope through associations with resident nuclear proteins (reviewed in Holmer and Worman, 2001). Sorting to the INM occurs initially via energy-dependent interactions with the nuclear pore complex (NPC) (Ohba et al., 2004; Beilharz et al., 2003). Nuclear lamins exemplify an important class of peripheral INM proteins that are accompanied across the NPC by soluble receptors called karyopherins (reviewed in Hetzer et al., 2005). Likewise, tail-anchored INM proteins translocate through the NPC and insert post-translationally into the INM (Beilharz et al., 2003). However, integral INM proteins that are synthesized in the extranuclear rough ER are sorted to the INM probably by diffusion across the continuous membrane of the nuclear pore (Ohba et al., 2004). Saturable interactions with nuclear lamins, chromatin, or other nuclear proteins are thought to retain these proteins within the INM (Ostlund et al., 2006).

Mechanisms governing the sorting of ONM proteins are poorly characterized, as only a few cases are known. A WPP domain (tryptophan-proline-proline) at the N-terminus of a subset of plant proteins is necessary and sufficient to target the ONM, presumably by interacting with factors on the nuclear surface (Patel et al., 2004). Additionally, metazoan KASH-domain-containing proteins (Klarsicht, ANC-1, and Syne homology) are retained in the ONM through interactions across the perinuclear lumen with specific INM proteins (reviewed in Worman and Gundersen, 2006; Starr and Fischer, 2005). KASH domains physically bind the luminal SUN domains (Sad1p and UNc-84 homology) of SUN proteins, which are anchored to the INM by nuclear lamins (Crisp et al., 2006). Together, KASH and SUN proteins mediate the assembly of large nuclear envelope-associated complexes (LINC complexes) that physically link the cytoskeleton to nucleoplasmic components.

By analogy to KASH proteins, Nvj1p is a yeast membrane protein found exclusively in the ONM where it interacts with Vac8p on the vacuole membrane to form nucleus-vacuole (NV) junctions (Pan et al., 2000). In the absence of Vac8p, Nvj1p localizes evenly over the surface of the nuclear envelope but fails to escape into the cortical ER, even when overexpressed (Pan et al., 2000). Thus, an undetermined mechanism mediates the strict localization of Nvj1p to the ONM. Aside from Vac8p, Nvj1p also sequesters two proteins with roles in lipid homeostasis, Osh1p and Tsc13p, to the nuclear surface (Kvam et al., 2005; Kvam and Goldfarb, 2004). The fraction of Osh1p or Tsc13p present at NV junctions depends on the expression level of Nvj1p, which is upregulated in response to nutrient depletion. Tsc13p is an essential ER membrane protein involved in the synthesis of very-long-chain fatty acids (VLCFAs), which are important constituents of ceramides, sphingolipids, GPI anchors and unusual inositolphospholipids and, as such, play important roles in membrane fluidity, lipid raft biogenesis, and cell signaling (Eisenkolb et al., 2002; Kohlwein et al., 2001; Dickson, 1998). Osh1p, on the other hand, is a member of a large family of eukaryotic oxysterol-binding protein-related proteins (ORPs) that share a conserved oxysterol-binding domain, which forms a hydrophobic sterol-sensing cavity (Im et al., 2005). Yeast lacking the overlapping Osh protein family (Osh1p-Osh7p) show significant defects in sterol-dependent processes including endocytosis, vesicle transport, and homotypic vacuole fusion (Beh and Rine, 2004). Consistent with a role in sterol regulation, osh1-Δ cells exhibit a cold-sensitive tryptophan transport defect akin to erg6 mutants that are defective in ergosterol synthesis (Levine and Munro, 2001; Jiang et al., 1994). Ergosterol is required for trafficking of the high-affinity tryptophan permease, Tat2p, to the plasma membrane via detergent-resistant membrane domains (Umebayashi and Nakano, 2003). The localization of Osh1p is regulated by several targeting determinants, including a PH domain specific for Golgi membranes (Levine and Munro, 2001), a FFAT motif (FF in an acidic tract) that targets the ER through associations with Scs2p (Loewen et al., 2003), and an N-terminal ankyrin repeat domain that mediates localization to NV junctions (Levine and Munro, 2001).

NV junctions are sites of a starvation-induced autophagic process called piecemeal microautophagy of the nucleus (PMN) that degrades portions of the yeast nucleus in the hydrolytic vacuole lumen (Roberts et al., 2003). During PMN, the vacuole membrane envelopes a section of the nuclear envelope through invagination, forming a nuclear `bleb'. Ultimately, this tri-laminar PMN bleb, consisting of the inner- and outer-nuclear membranes and the vacuole membrane, is pinched-off as a vesicle and degraded in the vacuole lumen (Roberts et al., 2003). PMN appears to respond to nutrient starvation by scavenging and recycling nonessential nuclear components, including portions of the nuclear envelope and, often, underlying nucleolar pre-ribosomes (Roberts et al., 2003). Activities associated with Tsc13p and the Osh-protein family are required for efficient PMN biogenesis (Kvam et al., 2005; Kvam and Goldfarb, 2004).

In this study, we map the sequences of Nvj1p that mediate sorting to the ONM and associations with Vac8p, Tsc13p and Osh1p. The N-terminal signal anchor-like sequence of Nvj1p is not required for either targeting to the ER or correct orientation in the membrane, but is required for ONM anchoring, possibly by bridging to the INM. The membrane-spanning domain of Nvj1p mediates membrane insertion and also interacts with Tsc13p. Osh1p and Vac8p bind to non-overlapping regions in the cytoplasmic C-terminus of Nvj1p. Based on growth phenotypes associated with the Osh1p-Nvj1p interaction, we present evidence that Nvj1p functions as a negative regulator of tryptophan transport during nutrient depletion.

Saccharomyces orthologs of Nvj1p share four conserved domains

Nvj1p is an integral ER membrane protein that localizes exclusively to the ONM and sequesters proteins from diverse cellular pools to the nuclear surface. Since Nvj1p lacks known sequence motifs, we identified functionally constrained regions in Nvj1p by comparing orthologous protein sequences from six Saccharomyces species, including four `stricto senso' species similar to S. cerevisiae (S. paradoxus, S. mikatae, S. bayanus and S. kudriavzevii) and two more divergent species (S. castellii and S. kluyveri) (Cliften et al., 2001). These alignments revealed four conserved regions in Nvj1p (Fig. 1A,B). Region I at the N-terminus is characterized by a hydrophobic signal anchor-like sequence (Fig. 1B,C). Region II overlaps with the predicted membrane-spanning domain of Nvj1p (Fig. 1B,C). Region III is adjacent to the membrane-spanning domain, and region IV maps to the C-terminal end of Nvj1p (Fig. 1B). These conserved regions helped guide our efforts in mapping the putative functional domains of Nvj1p.

Region IV is the Vac8p-binding domain of Nvj1p

Protein-protein interactions between Nvj1p and Vac8p are required for the formation of NV junctions. These interactions are remarkable because they represent the only currently known inter-organelle junction apparatus in eukaryotes (Levine, 2004). Previously, we showed that Vac8p co-purifies with Nvj1p and genetically interacts with the C-terminal tail of Nvj1p (aa 261-321) by yeast two-hybrid analysis (Pan et al., 2000). These results position the Vac8p-binding domain of Nvj1p within the vicinity of region IV (Fig. 1A,B). To determine whether region IV is necessary for an association with Vac8p, we deleted a cluster of conserved residues at the C-terminal end of Nvj1p (aa 292-321; Fig. 1A,B) and analyzed the localization of the truncated reporter (Nvj1p(Δ293-321)-EGFP) in the presence and absence of Vac8p. Normally in VAC8+ cells, full-length Nvj1p-EGFP localizes to patches on the nuclear surface that co-localize with FM4-64-stained vacuole membranes, producing a yellow fluorescence characteristic of NV junctions (Fig. 2A) (Pan et al., 2000). In the absence of Vac8p, Nvj1p-EGFP diffuses evenly over the surface of nuclei but is restricted to the perinuclear ER (Fig. 2A) (Pan et al., 2000). As shown in Fig. 2A, Nvj1p(Δ293-321)-EGFP failed to localize in patches corresponding to NV junctions in VAC8+ cells but, instead, appeared evenly distributed over the nuclear envelope as in vac8-Δ cells (Fig. 2A). These data, in agreement with previous two-hybrid results (Pan et al., 2000), confirm that conserved residues at the immediate C-terminus of Nvj1p are necessary for binding Vac8p and creating NV junctions. This Vac8p-binding region is not required for the strict localization of Nvj1p to the nuclear envelope, a property that must map elsewhere in the protein. Finally, these results are consistent with our previous observation that the C-terminal 40 residues of Nvj1p fused to GFP partially localize to the vacuole membrane in a Vac8p-dependent fashion (Pan et al., 2000).

Fig. 1.

Nvj1p can be divided into four domains based on sequence homology across the Saccharomyces family. (A) Alignment of S. cerevisiae Nvj1p and its orthologs in closely related and divergent Saccharomyces species. Amino acid sequences were obtained from the Saccharomyces Genome Database (www.yeastgenome.org). The S. kluyveri ortholog of Nvj1p was identified through a tblastn search of contig 02.1344 (accession AACE02000001). Sequences were aligned using the ClustalX program (Thompson et al., 1997). Boxes were drawn around regions of sequence similarity in the Saccharomyces Nvj1p family. (B) Graphical view of residues in Nvj1p bearing similarity or identity across the Saccharomyces Nvj1p family. ClustalX conservation scores were calculated for each column of the alignment in 1A, and residues bearing similarity (50-74% homology, open circles) or near identity (75-100%, shaded diamonds) were aligned onto a linear map of Nvj1p. These residues sorted independently into four regions, as indicated. (C) Kyte-Doolittle hydropathy plot of the four conserved regions of Nvj1p. Region I at the N-terminus shows an area of considerable hydrophobicity, while region II coincides with the membrane-spanning domain of Nvj1p.

Fig. 1.

Nvj1p can be divided into four domains based on sequence homology across the Saccharomyces family. (A) Alignment of S. cerevisiae Nvj1p and its orthologs in closely related and divergent Saccharomyces species. Amino acid sequences were obtained from the Saccharomyces Genome Database (www.yeastgenome.org). The S. kluyveri ortholog of Nvj1p was identified through a tblastn search of contig 02.1344 (accession AACE02000001). Sequences were aligned using the ClustalX program (Thompson et al., 1997). Boxes were drawn around regions of sequence similarity in the Saccharomyces Nvj1p family. (B) Graphical view of residues in Nvj1p bearing similarity or identity across the Saccharomyces Nvj1p family. ClustalX conservation scores were calculated for each column of the alignment in 1A, and residues bearing similarity (50-74% homology, open circles) or near identity (75-100%, shaded diamonds) were aligned onto a linear map of Nvj1p. These residues sorted independently into four regions, as indicated. (C) Kyte-Doolittle hydropathy plot of the four conserved regions of Nvj1p. Region I at the N-terminus shows an area of considerable hydrophobicity, while region II coincides with the membrane-spanning domain of Nvj1p.

Region I is necessary for anchoring Nvj1p to the ONM

The N-terminus of Nvj1p plays an important role in sorting to the ONM, since the addition of tags onto its N-terminus causes the mis-localization of Nvj1p to extranuclear ER-like membranes that form ectopic junctions with vacuoles (Kvam and Goldfarb, 2004). The first ∼30 residues of the N-terminus (region I) comprise an imperfect hydrophobic sequence bordered by basic residues, reminiscent of a signal sequence (Fig. 1A,C) (Nielsen et al., 1997). Unlike signal peptide sequences, which are post-translationally cleaved, neural network algorithms predict that region I of Nvj1p is most likely not cleaved (Bendtsen et al., 2004).

To investigate whether region I is a sorting determinant, we constructed several truncations around its hydrophobic sequence. Unlike full-length Nvj1p, which localizes strictly to the ONM in both VAC8+ and vac8-Δ cells (Fig. 2A), a truncated reporter lacking the hydrophobic sequence of region I (Nvj1p(Δ1-26)-EGFP) spread into the cortical ER of vac8-Δ cells (Fig. 2B). Similar results were observed using a reporter that lacked the entire N-terminus of Nvj1p (Nvj1p(Δ1-86)-EGFP) (Fig. 2B). These data demonstrate that the N-terminus is not required for ER targeting but rather functions after membrane insertion. When expressed in VAC8+ cells, both truncated reporters formed atypical junctions between or along the surface of vacuolar lobes that appeared spatially detached from the nucleus in most cells (Fig. 2B). Thus, when bound to Vac8p, these N-terminally truncated reporters no longer escaped to the cortical ER but rather formed extranuclear ER junctions with vacuoles. Conversely, deleting the first six residues of Nvj1p did not significantly affect the targeting of Nvj1p(Δ1-6)-EGFP to the ONM in vac8-Δ cells, although a low level of cortical staining was sometimes visible (Fig. 2B). Moreover, Nvj1p(Δ1-6)-EGFP formed normal perinuclear NV junctions in VAC8+ cells rather than aberrant extranuclear junctions (Fig. 2B). Thus, residues upstream of the hydrophobic sequence in region I are not critical for ONM sorting. In total, these experiments directly implicate the hydrophobic signal anchor-like sequence of Nvj1p in proper NV-junction formation by targeting the ONM. Interestingly, a reporter containing the first 90 aa of Nvj1p fused to GFP localized to mitochondria (our unpublished results), which is consistent with the fact that region I of Nvj1p resembles the signal-anchor sequences of the outer mitochondrial membrane proteins Tom20p and Tim70p (Waizenegger et al., 2003).

Fig. 2.

Tripartite signals mediate the targeting of Nvj1p to NV junctions. (A) Region IV comprises the Vac8p-binding domain of Nvj1p. Full-length (FL) Nvj1p-EGFP and a C-terminal truncation (Δ293-321aa) lacking the most conserved residues of region IV were localized in vac8-Δ and VAC8+ (nvj1-Δ) cells. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained with Hoechst and FM4-64, respectively. Interactions between Nvj1p-EGFP and Vac8p on FM4-64-stained vacuoles create NV junctions, which appear yellow in the overlay. The C-terminal truncation of Nvj1p does not form NV junctions and instead localizes uniformly across the nuclear envelope. (B) Region I plays a role in anchoring Nvj1p to the perinuclear ER. N-terminal truncations preceding (Δ1-6aa) or succeeding (Δ1-26aa) the hydrophobic sequence of region I, or eliminating the N-terminus of Nvj1p entirely (Δ1-86aa), were localized in vac8-Δ and VAC8+ (nvj1-Δ) cells. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained as above. EGFP-tagged truncations lacking region I (denoted by a horizontal bar) escape into the cortical ER and form aberrant vacuole-associated membrane junctions in the cytosol (see arrows) that appear detached from the nucleus. (C) Region I and the membrane-spanning region of Nvj1p (region II) cooperatively sort to the perinuclear ER. Nvj1p fragments tagged on their C-terminus with EGFP were localized similarly to above. The membrane-spanning region of Nvj1p (residues 87-120, denoted by a shaded box) targets ER membranes but localizes exclusively to the perinuclear ER in conjunction with region I (denoted by a horizontal bar). Replacing the transmembrane region of Nvj1p (between residues 90-121) with the first transmembrane segment of Ste2p causes Nvj1p to spread into the cortical ER and form aberrant junctions with vacuoles in the cytosol (see arrows). Numbers in diagram correspond to amino acid position.

Fig. 2.

Tripartite signals mediate the targeting of Nvj1p to NV junctions. (A) Region IV comprises the Vac8p-binding domain of Nvj1p. Full-length (FL) Nvj1p-EGFP and a C-terminal truncation (Δ293-321aa) lacking the most conserved residues of region IV were localized in vac8-Δ and VAC8+ (nvj1-Δ) cells. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained with Hoechst and FM4-64, respectively. Interactions between Nvj1p-EGFP and Vac8p on FM4-64-stained vacuoles create NV junctions, which appear yellow in the overlay. The C-terminal truncation of Nvj1p does not form NV junctions and instead localizes uniformly across the nuclear envelope. (B) Region I plays a role in anchoring Nvj1p to the perinuclear ER. N-terminal truncations preceding (Δ1-6aa) or succeeding (Δ1-26aa) the hydrophobic sequence of region I, or eliminating the N-terminus of Nvj1p entirely (Δ1-86aa), were localized in vac8-Δ and VAC8+ (nvj1-Δ) cells. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained as above. EGFP-tagged truncations lacking region I (denoted by a horizontal bar) escape into the cortical ER and form aberrant vacuole-associated membrane junctions in the cytosol (see arrows) that appear detached from the nucleus. (C) Region I and the membrane-spanning region of Nvj1p (region II) cooperatively sort to the perinuclear ER. Nvj1p fragments tagged on their C-terminus with EGFP were localized similarly to above. The membrane-spanning region of Nvj1p (residues 87-120, denoted by a shaded box) targets ER membranes but localizes exclusively to the perinuclear ER in conjunction with region I (denoted by a horizontal bar). Replacing the transmembrane region of Nvj1p (between residues 90-121) with the first transmembrane segment of Ste2p causes Nvj1p to spread into the cortical ER and form aberrant junctions with vacuoles in the cytosol (see arrows). Numbers in diagram correspond to amino acid position.

The membrane-spanning domain of Nvj1p is also required for ONM targeting

As described above, deleting the first 86 residues of Nvj1p did not prevent targeting of this reporter to the ER (Fig. 2B). Since this deletion also retains C-terminal associations with the cytoplasmic vacuoles in VAC8+ cells, the N-terminus of Nvj1p is not required for properly orienting the protein in the membrane. Further truncation experiments revealed that a fragment of Nvj1p (aa 87-120) corresponding to the membrane-spanning domain of Nvj1p (region II; Fig. 1B,C) was sufficient to localize throughout the perinuclear and cortical ER network (Fig. 2C). Thus, region II of Nvj1p contains sufficient sequence information for targeting ER membranes rather than the plasma membrane or other organelles. Importantly, an N-terminal fragment of Nvj1p comprising both regions I and II (Nvj1p(Δ121-321)-EGFP) localized strictly to the nuclear envelope (Fig. 2C). These data illustrate that the hydrophobic N-terminal signal anchor-like sequence and the membrane-spanning domain of Nvj1p work in concert to mediate targeting to the ONM. To test whether the membrane-spanning domain of another protein can function in this regard, we replaced residues 90-120 of Nvj1p with a slightly modified version of the first membrane-spanning domain of Ste2p, which has been used extensively to study co-translational insertion into the ER (see Harley and Tipper, 1996). As shown in Fig. 2C, Nvj1p(Ste2pTM)-EGFP aberrantly spread into the cortical ER of vac8-Δ cells despite the presence of a native N-terminal domain (region I). Likewise, this chimeric reporter formed aberrant, vacuole-associated membrane junctions in VAC8+ cells (Fig. 2C), similar to those observed in cells expressing N-terminal truncations of Nvj1p (Fig. 2B). In summary, these data indicate that regions I and II function co-operatively to target Nvj1p into the ONM, and that the membrane-spanning domain of Nvj1p is adapted for this purpose.

Blocking the N-terminus of Nvj1p leads to ONM-expansion and separation from the INM

As described above, region I is required for the strict localization of Nvj1p to the ONM. It is possible that this domain functions to anchor Nvj1p in the ONM by attaching to the INM. By analogy, the ONM localization of KASH-domain-containing nesprin isoforms is mediated by interactions across the perinuclear lumen with SUN domain proteins (Crisp et al., 2006). A bridging model for Nvj1p is consistent with the observation that deleting its N-terminus results in the formation of extranuclear junctions in VAC8+ cells (Fig. 2B). Likewise, we previously reported that blocking the N-terminus of Nvj1p with a 3HA epitope promoted the formation of analogous extranuclear junctions (Kvam and Goldfarb, 2004). If our bridging model is true, then extranuclear ER-vacuole junctions might arise from sections of the ONM that have become detached from the INM.

To test this hypothesis we employed an N-terminally fused GFP-Nvj1p reporter that, like N-terminal truncations (Fig. 2B), spread into the cortical ER of vac8-Δ cells and promoted the formation of extranuclear ER junctions with vacuoles in VAC8+ cells (Fig. 3A). Anti-GFP immuno-EM was used to characterize the GFP-Nvj1p-labeled membranes associated with these extranuclear junctions. The example shown in Fig. 3B shows an extension of the ONM that is sandwiched between two vacuole lobes. The presence of GFP-Nvj1p in these membranes is confirmed by the presence of anti-GFP colloidal gold labeling (Fig. 3B, asterisks). A striking separation between the outer- and inner-nuclear membranes is apparent by the triangular expansion of the perinuclear lumen, which, as expected, is devoid of electron dense material such as ribosomes. These results are consistent with the hypothesis that the N-terminus of Nvj1p mediates ONM-localization through a direct or indirect physical attachment with the INM. A yeast two-hybrid screen for INM proteins that might interact with the N-terminal domain of Nvj1p failed to identify potential candidates (our unpublished results). It is also possible that the N-terminal domain of Nvj1p inserts directly into the lipid bilayer of the INM, by analogy to the insertion of the N-terminal amphipathic helix of Sar1p into the ER (Lee et al., 2005). Our results are also interesting because they show that zipper-like interactions between Nvj1p and Vac8p can promote membrane expansion in a nuclear envelope subdomain (see Campbell et al., 2006).

Region II of Nvj1p sequesters the ER membrane protein Tsc13p

We previously demonstrated Tsc13p, an ER membrane protein involved in VLCFA biosynthesis, is recruited into NV junctions through associations with Nvj1p (Kvam et al., 2005). Using confocal microscopy, we mapped the region of Nvj1p responsible for interacting with Tsc13p. Briefly, N- and C-terminal truncations of Nvj1p were tested for their ability to sequester Tsc13p-EYFP from the cortical ER to the nuclear envelope in nvj1-Δ cells. Sequences on either side of the membrane-spanning domain (region II) of Nvj1p proved dispensable for an association with Tsc13p (Fig. 4A). Tsc13p was efficiently sequestered to the nuclear envelope upon expression of an N-terminal fragment of Nvj1p (aa 1-120) lacking C-terminal residues proximal to region II (Fig. 4A). Likewise, expression of a C-terminal fragment of Nvj1p (aa 87-321) lacking residues preceding region II was sufficient to sequester Tsc13p into extranuclear, vacuole-associated membrane junctions (Fig. 4A), which arose as a consequence of the mis-targeting of this reporter (Fig. 2B). These data collectively narrowed down the Tsc13p-interaction domain to the membrane-spanning domain (region II) of Nvj1p. Alternatively, it was possible that both fragments contained separate Tsc13p-binding activity. However, we determined that region II alone is sufficient to co-immunoprecipitate Tsc13p-EYFP in vivo. As shown in Fig. 4B, Tsc13p-EYFP was efficiently co-immunoprecipitated from detergent-extracted cell lysates using a 3HA-tagged fragment of region II containing the membrane-spanning domain of Nvj1p and some bordering residues (Nvj1p(87-120)-3HA). Together, these results indicate that the sequences within region II are sufficient for binding and sequestering Tsc13p in the ER. By analogy, this interaction may occur between the membrane-spanning domains of these two proteins or possibly among residues very close to the plane of the membrane.

Fig. 3.

Blocking the N-terminus of Nvj1p promotes extranuclear junction formation by separation and extension of the ONM. (A) GFP-Nvj1p localizes throughout the perinuclear and cortical ER in vac8-Δ cells, and accumulates in extranuclear junctions between ER-like membranes and cytoplasmic vacuoles when bound to Vac8p. Vacuole membranes (red) were stained with FM4-64 and overlayed. Images were collected by Xiaozhou Ryan. (B) Immunogold localization of GFP-tagged Nvj1p by electron microscopy. GFP-tagged Nvj1p reporters were induced prior to cryofixation and immuno-EM analysis as described in Materials and Methods. Nvj1p-GFP, a functional reporter, localizes to NV junctions. Colloidal-gold-conjugated antibodies against GFP (asterisk) label the surface of nuclei at NV junctions. By contrast, GFP-Nvj1p (a non-functional reporter) forms extranuclear junctions with vacuoles. Here, colloidal-gold-conjugated antibodies against GFP (asterisk) label an extension of the ONM (black arrows) that is sandwiched between two vacuole lobes. The ONM is separated from the INM (denoted by white arrowheads), forming an extensive void in the perinuclear lumen. N, nuclei; V, vacuoles. Bars, 0.3 μm.

Fig. 3.

Blocking the N-terminus of Nvj1p promotes extranuclear junction formation by separation and extension of the ONM. (A) GFP-Nvj1p localizes throughout the perinuclear and cortical ER in vac8-Δ cells, and accumulates in extranuclear junctions between ER-like membranes and cytoplasmic vacuoles when bound to Vac8p. Vacuole membranes (red) were stained with FM4-64 and overlayed. Images were collected by Xiaozhou Ryan. (B) Immunogold localization of GFP-tagged Nvj1p by electron microscopy. GFP-tagged Nvj1p reporters were induced prior to cryofixation and immuno-EM analysis as described in Materials and Methods. Nvj1p-GFP, a functional reporter, localizes to NV junctions. Colloidal-gold-conjugated antibodies against GFP (asterisk) label the surface of nuclei at NV junctions. By contrast, GFP-Nvj1p (a non-functional reporter) forms extranuclear junctions with vacuoles. Here, colloidal-gold-conjugated antibodies against GFP (asterisk) label an extension of the ONM (black arrows) that is sandwiched between two vacuole lobes. The ONM is separated from the INM (denoted by white arrowheads), forming an extensive void in the perinuclear lumen. N, nuclei; V, vacuoles. Bars, 0.3 μm.

Fig. 4.

Region II of Nvj1p sequesters Tsc13p. (A) The membrane-spanning domain of Nvj1p (region II) is responsible for recruiting Tsc13p-EYFP. Full-length (FL) or N- and C-terminal truncations of Nvj1p were overexpressed in nvj1-Δ cells and tested for the ability to sequester Tsc13p-EYFP from cortical ER compartments to the perinuclear ER. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained with Hoechst and FM4-64, respectively. Regions flanking either side of the membrane-spanning domain of Nvj1p (denoted by a shaded box) proved dispensable for the sequestration of Tsc13p-EYFP to the nuclear envelope. Numbers correspond to amino acid positions. (B) The transmembrane region of Nvj1p (aa 85-120) is sufficient to co-immunoprecipitate Tsc13p-EYFP in vivo. Detergent-extracted lysates of cells expressing Tsc13p-EYFP and 3HA-tagged Nvj1p(85-120aa) or empty vector were prepared as described in Materials and Methods. Nvj1p(85-120aa)-3HA was immunoprecipitated (IP) with anti-HA conjugated agarose beads.

Fig. 4.

Region II of Nvj1p sequesters Tsc13p. (A) The membrane-spanning domain of Nvj1p (region II) is responsible for recruiting Tsc13p-EYFP. Full-length (FL) or N- and C-terminal truncations of Nvj1p were overexpressed in nvj1-Δ cells and tested for the ability to sequester Tsc13p-EYFP from cortical ER compartments to the perinuclear ER. Nuclear chromatin (blue) and vacuole membranes (red) were co-stained with Hoechst and FM4-64, respectively. Regions flanking either side of the membrane-spanning domain of Nvj1p (denoted by a shaded box) proved dispensable for the sequestration of Tsc13p-EYFP to the nuclear envelope. Numbers correspond to amino acid positions. (B) The transmembrane region of Nvj1p (aa 85-120) is sufficient to co-immunoprecipitate Tsc13p-EYFP in vivo. Detergent-extracted lysates of cells expressing Tsc13p-EYFP and 3HA-tagged Nvj1p(85-120aa) or empty vector were prepared as described in Materials and Methods. Nvj1p(85-120aa)-3HA was immunoprecipitated (IP) with anti-HA conjugated agarose beads.

Fig. 5.

Region III of Nvj1p is necessary and sufficient to bind Osh1p. (A) Deletion of the most conserved residues in region III abrogates the Nvj1p-dependent sequestration of GFP-Osh1p to the nuclear envelope. Full-length (FL) or N- and C-terminal truncations of Nvj1p were overexpressed in nvj1-Δ cells and tested for the ability to sequester GFP-Osh1p from soluble and Golgi-associated pools to the nuclear surface. Conserved residues in region III of Nvj1p (aa 130-176) proved necessary for GFP-Osh1p sequestration. Numbers correspond to amino acid positions. (B) Residues 130-177 of Nvj1p are sufficient to co-immunoprecipitate GFP-Osh1p in vivo. Lysates of cells expressing GFP-Osh1p and 3HA-tagged Nvj1p(130-177aa) or empty vector were prepared as described in Materials and Methods. Nvj1p(130-177aa)-3HA was immunoprecipitated (IP) with anti-HA conjugated agarose beads.

Fig. 5.

Region III of Nvj1p is necessary and sufficient to bind Osh1p. (A) Deletion of the most conserved residues in region III abrogates the Nvj1p-dependent sequestration of GFP-Osh1p to the nuclear envelope. Full-length (FL) or N- and C-terminal truncations of Nvj1p were overexpressed in nvj1-Δ cells and tested for the ability to sequester GFP-Osh1p from soluble and Golgi-associated pools to the nuclear surface. Conserved residues in region III of Nvj1p (aa 130-176) proved necessary for GFP-Osh1p sequestration. Numbers correspond to amino acid positions. (B) Residues 130-177 of Nvj1p are sufficient to co-immunoprecipitate GFP-Osh1p in vivo. Lysates of cells expressing GFP-Osh1p and 3HA-tagged Nvj1p(130-177aa) or empty vector were prepared as described in Materials and Methods. Nvj1p(130-177aa)-3HA was immunoprecipitated (IP) with anti-HA conjugated agarose beads.

Region III is sufficient and necessary to bind Osh1p

We previously showed that Nvj1p sequesters Osh1p into NV junctions in a Vac8p-independent fashion, most likely by direct binding (Kvam and Goldfarb, 2004). In order to identify the region of Nvj1p responsible for interacting with Osh1p, we expressed N- and C-terminal truncations of Nvj1p in nvj1-Δ cells and assessed their ability to recruit GFP-Osh1p from soluble and Golgi-associated pools to the nuclear envelope by confocal microscopy. As shown in Fig. 5A, the putative Osh1p-binding site mapped to a segment of Nvj1p (aa 120-177) adjacent to the membrane-spanning domain. Importantly, this area coincided with region III of Nvj1p (Fig. 1A,B) and, like the Vac8p-binding domain (region IV), is exposed to the cytoplasm. Saccharomyces orthologs of Nvj1p contain an especially well conserved sequence between residues 140-174 in region III (Fig. 1A,B). We created a myc-tagged Nvj1p mutant with an internal deletion that eliminated these conserved residues. Expression of this mutant, Nvj1p(Δ130-176)-myc, was confirmed by immunoblot (our unpublished data). As expected, overexpression of Nvj1p(Δ130-176)-myc failed to sequester GFP-Osh1p from surrounding cytoplasmic and Golgi compartments (Fig. 5A). This fact was not due to improper targeting of the Nvj1p(Δ130-176)-myc reporter, since a version tagged with GFP localized exclusively to the ONM, where it formed normal NV junctions in VAC8+ cells (our unpublished data). Moreover, expression of Nvj1p(Δ130-176)-myc was sufficient to sequester Tsc13p-EYFP from the cortical ER to NV junctions (our unpublished data). In biochemical pull-down experiments, GFP-Osh1p efficiently co-immunoprecipitated with an epitope-tagged fragment of region III encompassing residues 130-177 of Nvj1p (Nvj1p(Δ130-177)-3HA) (Fig. 5B). Together, these results show that a conserved sequence within region III is both necessary and sufficient for an interaction between Nvj1p and Osh1p.

Fig. 6.

Overexpression of Nvj1p, but not a mutant form lacking Osh1p-binding activity, confers a tryptophan transport-related growth defect. (A) Cells overexpressing Nvj1p, but not a mutant form lacking the Osh1p-binding domain, grow poorly on media containing limiting concentrations of tryptophan at low temperature. Cells harboring PCUP1-NVJ1, PCUP1-NVJ1(Δ130-176aa)-myc, or empty vector were induced and replica-plated onto media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan as described in Materials and Methods. (B) The cold-sensitive growth phenotype of Nvj1p-overexpressing cells is suppressed by low- and high-affinity tryptophan permeases. Cells harboring PCUP1-NVJ1 and either pTAT1, pTAT2, or empty vector were induced and replica-plated onto media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan as above. (C) NV junctions are not necessary to confer tryptophan transport defects in Nvj1p-overexpressing cells. Growth curves of vac8-Δ cells expressing PCUP1-NVJ1 (open squares), PCUP1-NVJ1(Δ130-176aa)-myc (black triangles), or empty vector (gray circles) were determined in liquid media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan at 25°C using a Bioscreen C Analyzer (see Materials and Methods).

Fig. 6.

Overexpression of Nvj1p, but not a mutant form lacking Osh1p-binding activity, confers a tryptophan transport-related growth defect. (A) Cells overexpressing Nvj1p, but not a mutant form lacking the Osh1p-binding domain, grow poorly on media containing limiting concentrations of tryptophan at low temperature. Cells harboring PCUP1-NVJ1, PCUP1-NVJ1(Δ130-176aa)-myc, or empty vector were induced and replica-plated onto media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan as described in Materials and Methods. (B) The cold-sensitive growth phenotype of Nvj1p-overexpressing cells is suppressed by low- and high-affinity tryptophan permeases. Cells harboring PCUP1-NVJ1 and either pTAT1, pTAT2, or empty vector were induced and replica-plated onto media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan as above. (C) NV junctions are not necessary to confer tryptophan transport defects in Nvj1p-overexpressing cells. Growth curves of vac8-Δ cells expressing PCUP1-NVJ1 (open squares), PCUP1-NVJ1(Δ130-176aa)-myc (black triangles), or empty vector (gray circles) were determined in liquid media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan at 25°C using a Bioscreen C Analyzer (see Materials and Methods).

NVJ1-overexpressing cells exhibit a cold-sensitive tryptophan transport defect related to Osh1p

Several studies have demonstrated that deleting OSH1 in tryptophan auxotrophs (trp1) produces a temperature-sensitive cell growth defect on media containing low concentrations of tryptophan (Loewen et al., 2003; Levine and Munro, 2001; Jiang et al., 1994). This growth defect is similar to that of erg6-Δ cells, which are defective in ergosterol biosynthesis, and is likely related to the sterol-dependent sorting of the high-affinity tryptophan permease Tat2p to the plasma membrane (Umebayashi and Nakano, 2003; Gaber et al., 1989). This sterol-sensitive step occurs during the sorting of Tat2p into detergent-resistant membrane domains (DRMs, or lipid rafts) within late Golgi or post-Golgi compartments (Umebayashi and Nakano, 2003; Gaber et al., 1989). Based on these results, we reasoned that the overexpression of Nvj1p might cause an analogous slow-growth phenotype by sequestering Osh1p into NV junctions and away from the trans-Golgi. Indeed, compared to control cells expressing an empty vector, NVJ1-overexpressing cells were hypersensitive to tryptophan limitation at low temperature (Fig. 6A). This growth defect was complemented by excess tryptophan or higher temperature (Fig. 6A), analogous to what was observed in osh1-Δ cells (Levine and Munro, 2001). Moreover, the growth defect of NVJ1-expressing cells was rescued by single copy CEN plasmids expressing either TAT1 or TAT2 tryptophan permeases under the control of their native promoters (Fig. 6B). By contrast, cells overexpressing Nvj1p(Δ130-176)-myc, which lacks Osh1p-binding activity (Fig. 5A), showed normal growth on low tryptophan media (Fig. 6A). These results were corroborated by monitoring cell growth rates at 25°C in liquid media containing low or high concentrations of tryptophan (Fig. 6C). The ability of Nvj1p to form NV junctions proved to be unimportant for these tryptophan transport-related growth phenotypes, since similar results were observed in vac8-Δ cells (Fig. 6C). In summary, these results demonstrate that the sequestration of Osh1p by Nvj1p causes a defect in tryptophan uptake similar to that previously reported for osh1-Δ cells.

NVJ1+ trp1 cells grow poorer in tryptophan-depleted media than nvj1-Δ trp1 cells

Osh1p localizes to both trans-Golgi membranes and NV junctions, but is found almost exclusively at NV junctions as Nvj1p levels rise by ectopic overexpression or as a consequence of nutrient depletion during late-log phase (Kvam and Goldfarb, 2004; Levine and Munro, 2001). Because high levels of Nvj1p affect cell growth by sequestering Osh1p (Fig. 6), we were interested in determining whether physiologically relevant levels of Nvj1p were active in this regard. For this purpose, we analyzed the growth of isogenic NVJ1+ trp1 and nvj1trp1 cells as a function of tryptophan limitation. At concentrations of tryptophan sufficient to slow the growth of NVJ1-overexpressing cells (15 μg/ml), the growth of NVJ1+ and nvj1-Δ cells at 25°C was indistinguishable (Fig. 7B). However, nvj1-Δ cells grew markedly better than NVJ1+ cells when the concentration of extracellular tryptophan became increasingly limiting (5 μg/ml) (Fig. 7A). This difference, albeit small, is significant and was consistently reproduced in several experimental trials. Thus, native levels of Nvj1p may be mildly deleterious for growth in tryptophan-depleted environments by affecting the localization and activity of Osh1p.

Fig. 7.

Native levels of Nvj1p are mildly deleterious for growth in tryptophan-depleted media. nvj1-Δ and its isogenic parental (NVJ1+) tryptophan auxotrophic strain (trp1) were grown in rich media to log-phase and shifted into SC media containing limiting (15 μg/ml) or depleted (5 μg/ml) concentrations of tryptophan. Cell growth was monitored at 25°C using a Bioscreen C Analyzer (see Materials and Methods). Error bars were calculated from two independent trials performed in duplicate.

Fig. 7.

Native levels of Nvj1p are mildly deleterious for growth in tryptophan-depleted media. nvj1-Δ and its isogenic parental (NVJ1+) tryptophan auxotrophic strain (trp1) were grown in rich media to log-phase and shifted into SC media containing limiting (15 μg/ml) or depleted (5 μg/ml) concentrations of tryptophan. Cell growth was monitored at 25°C using a Bioscreen C Analyzer (see Materials and Methods). Error bars were calculated from two independent trials performed in duplicate.

Nvj1p is a remarkable protein that mediates several novel phenomena, including the formation of the only known inter-organelle membrane junction apparatus (NV junctions), and a unique type of autophagy that pinches-off and degrades non-essential parts of the yeast nucleus (PMN). The key findings of this study include mapping the regions of Nvj1p responsible for sorting to the ONM and interacting with its binding-partners, Vac8p, Osh1p and Tsc13p. These proteins bind, either directly or indirectly, to discrete, non-overlapping regions of Nvj1p. The functional domains identified in this study map to four regions (I-IV) of Nvj1p that are conserved among the Saccharomyces family of NVJ1 orthologs. Some of these sequence regions are present in increasingly divergent Nvj1p-like proteins in more evolutionarily distant yeasts, including Kluveromyces waltii, Kluveromyces lactis, Candida glabrata and Ashbya gossypii (not shown). It should be noted that the genomes of all fungal species whose genomes have been sequenced, including Aspergillis nidulans and Neurospora crassa, encode well-conserved orthologs of VAC8, TSC13 and OSH1. Our results also reveal a novel role for Nvj1p in the regulation of tryptophan transport. These experiments are the first to expose a growth defect associated with NVJ1 expression. From our results, we propose that the sequestration of Osh1p by Nvj1p inhibits tryptophan uptake from the environment, possibly by modulating tryptophan permease-trafficking to the plasma membrane via a mechanism that was previously shown to require Osh1p (Levine and Munro, 2001; Jiang et al., 1994).

The sorting of eukaryotic proteins to the ONM is usually explained in terms of binding to resident proteins of the nuclear envelope, which begs the question of how the resident proteins themselves become localized. We show that the N-terminal hydrophobic domain (region I) and the membrane-spanning domain (region II) of Nvj1p are both required for efficient localization to the ONM. Region I contains a signal anchor-like sequence consisting of a relatively short, imperfect hydrophobic sequence flanked by positively charged residues. This sequence is not required for either the ER-targeting of Nvj1p or its correct orientation in the membrane, since N-terminal truncations of Nvj1p retain C-terminal associations with both Osh1p and Vac8p in the cytoplasm. The cytoplasmic C-terminal domain of Nvj1p (aa 121-321) is dispensable for ONM sorting, which rules out a role for Vac8p or Osh1p in this regard. However, Nvj1p-truncations lacking region I fail to localize strictly to the ONM and, instead, are found ubiquitously throughout the perinuclear and cortical ER. The membrane-spanning domain within region II plays an additional role in sorting to the ONM and, as discussed below, also interacts with Tsc13p. A 47-residue segment of region II is sufficient to target a GFP reporter to the ER network. We found that the important function of this domain in ONM sorting cannot be substituted by the first membrane-spanning domain of Ste2p, despite the fact that this chimera was efficiently expressed in the correct orientation in the ER. Thus, these results demonstrate that both regions I and II contribute specific information for ONM sorting.

NV-junction formation requires direct interactions between Nvj1p in the ONM and Vac8p on the vacuole membrane; however, the sorting of Nvj1p to the ONM is not dependent on Vac8p (Pan et al., 2000). We mapped the Vac8p-binding domain of Nvj1p to its C-terminus (region IV), which is consistent with previous two-hybrid results (Pan et al., 2000). When the sorting of Nvj1p is altered (either by blocking or mutating region I or through substitution of region II), Nvj1p spreads into the cortical ER and forms aberrant junctions with Vac8p-associated vacuole membranes. As shown in Fig. 3, these extranuclear junctions originate as extensions of the ONM. The N-terminal signal anchor-like sequence of Nvj1p, together with the membrane-spanning domain, may mediate a physical interaction across the perinuclear lumen with the INM (Fig. 8). Such a bridge would anchor Nvj1p in the ONM and prevent its escape into intermediate and cortical ER compartments. A physical connection to the INM would also explain how the ONM and INM move in concert into vacuole invaginations during PMN (Roberts et al., 2003). In support of this notion, blocking the N-terminus of Nvj1p with GFP promotes the separation of the inner and outer nuclear membranes, leading to expansion of the ONM through multiple zipper-like interactions between Nvj1p and Vac8p on the vacuole surface. Precedent for a nuclear envelope bridging apparatus comes from recent reports describing the LINC complex, which involves interactions between KASH-domain-containing nesprin isoforms in the ONM with SUN domain proteins in the INM (reviewed in Worman and Gundersen, 2006; Starr and Fischer, 2005). The ONM localization of nesprin requires that it be anchored to a SUN domain protein in the INM, since the depletion or the secretion of SUN domains into the ER lumen results in the mislocalization of nesprin to the bulk ER (Crisp et al., 2006).

In addition to mediating ONM localization, region II of Nvj1p also functions in sequestering Tsc13p within the ER. Our results are consistent with a model in which the membrane-spanning domain of Nvj1p interacts with one or more of the membrane-spanning domains of Tsc13p within the ER bilayer. Associations among membrane-spanning domains are known to stabilize certain integral membrane protein complexes (Li et al., 2003). Interactions between Nvj1p and Tsc13p, which are likely to be direct, may occur within the hydrophobic interior of the membrane or via residues immediately flanking their membrane-spanning segments. In this regard, it is interesting that the co-localization of Tsc13p with Nvj1p at NV junctions may depend on the availability of substrate long chain fatty acids (Kvam et al., 2005).

Finally, this report reveals that a conserved sequence in region III of Nvj1p is necessary and sufficient to bind Osh1p. The sequestration of Osh1p into NV junctions, first shown by Levine and Munro (Levine and Munro, 2001), is mediated by Vac8p-independent interactions with Nvj1p (Kvam and Goldfarb, 2004). Osh1p is particularly interesting because of its likely role in trafficking the high-affinity tryptophan permease, Tat2p. Tat2p is a constitutively expressed membrane permease that is routed to either the plasma membrane or vacuole depending on the concentration of tryptophan in the medium (Beck et al., 1999). A recent study revealed that Tat2p trafficking is dependent on its association with detergent-resistant membrane domains (DRMs) in post-Golgi compartments (Umebayashi and Nakano, 2003). In fact, several mutations disrupting the synthesis of DRM-associated lipids confer pleiotropic effects on tryptophan transport, including those affecting the synthesis of ergosterol (erg6-Δ, erg3-Δ, kes1-Δ), sphingolipids (elo2-Δ), GPI-anchor formation (gwt1-10), and specific phospholipids (cho1-Δ) (Okamoto et al., 2006; Umebayashi and Nakano, 2003; Chung et al., 2001; Nakamura et al., 2000; Hemmi et al., 1995; Jiang et al., 1994). Likewise, cells lacking Osh1p are defective in [3H]tryptophan uptake and osh1trp1 cells exhibit a severe growth defect on low tryptophan media (Levine and Munro, 2001; Jiang et al., 1994). Taken together, these published results strongly implicate Osh1p as a key player in the sterol-dependent trafficking of Tat2p to the plasma membrane.

Here, we report that overexpression of Nvj1p in trp1 auxotrophs causes a slow-growth phenotype similar to that reported for osh1-Δ cells (Levine and Munro, 2001). This growth defect is dependent on the sequestration of Osh1p, since cells overexpressing a mutant version of Nvj1p without an Osh1p-binding domain (region III) grew normally. The Nvj1p-dependent sequestration of Osh1p may provide a ready mechanism to downregulate tryptophan permease activity in response to nutrient depletion. In support of this notion, native levels of Nvj1p confer a moderating effect on cell growth in low-tryptophan media, presumably due to the steady (but physiologically significant) sequestration of Osh1p. As cells progress into late log phase, GFP-Osh1p becomes increasingly associated with NV junctions and less so at trans-Golgi compartments (Levine and Munro, 2001) (our unpublished results). It is during late log phase when high-affinity nutrient permeases such as Tat2p are routed to the vacuole and degraded (Schmidt et al., 1998). These observations are consistent with the hypothesis that Nvj1p is a negative regulator of Osh1p function and, as such, suggest that Nvj1p plays a novel role in sterol-dependent protein trafficking. Previous studies have also implicated the yeast homolog of mammalian VAMP-associated protein, Scs2p, in targeting Osh1p and other yeast ORPs to cellular membranes, although Scs2p is not strictly required for the localization of GFP-Osh1p to NV junctions (Loewen et al., 2003). Further work will be necessary to tease-out the roles of Nvj1p, Scs2p and other cellular factors in the complex control Osh1p function.

Fig. 8.

Molecular model of NV junctions. Nvj1p is restricted to the outer nuclear membrane, which is continuous with the perinuclear ER. The membrane-spanning domain (II) of Nvj1p interacts with Tsc13p in the ER membrane. An adjacent cytosolic region (III) associates with Osh1p. The C-terminal tail of Nvj1p (IV) binds to Vac8p, which is acylated to the vacuole membrane. The hydrophobic N-terminus of Nvj1p (I) may link the inner- and outer-nuclear membranes either by binding an unknown factor (X) in the INM (panel i), or by inserting directly into the INM by spanning across the perinuclear lumen (panel ii).

Fig. 8.

Molecular model of NV junctions. Nvj1p is restricted to the outer nuclear membrane, which is continuous with the perinuclear ER. The membrane-spanning domain (II) of Nvj1p interacts with Tsc13p in the ER membrane. An adjacent cytosolic region (III) associates with Osh1p. The C-terminal tail of Nvj1p (IV) binds to Vac8p, which is acylated to the vacuole membrane. The hydrophobic N-terminus of Nvj1p (I) may link the inner- and outer-nuclear membranes either by binding an unknown factor (X) in the INM (panel i), or by inserting directly into the INM by spanning across the perinuclear lumen (panel ii).

The faster growth of nvj1trp1 cells in limiting tryptophan suggests that Nvj1p may normally function as a finely tuned rheostat to control Osh1p activity, even during log phase when Nvj1p levels are normally low. Our results may also account for the increased survival of nvj1-Δ cells during stationary phase, which is a function of chronological aging (see Fabrizio et al., 2005). A recent genome-wide screen identified mutants in NVJ1 and a number of genes implicated in nutrient sensing and TOR signaling as some of the longest-lived in the deletion collection (Powers et al., 2006). Perhaps not coincidentally, TOR signaling has also been implicated in the degradation of high-affinity nutrient transporters (including Tat2p) in favor of lower-affinity transporters during nutrient stress (Edinger and Thompson, 2002; Beck et al., 1999; Schmidt et al., 1998).

In conclusion, we have mapped several functional domains of Nvj1p that mediate the recruitment of Vac8p, Osh1p and Tsc13p to the ONM, and revealed roles for both the N-terminal signal anchor-like sequence and membrane-spanning domain of Nvj1p in sorting to the ONM. Finally, we describe a functional link between the sequestration of Osh1p into NV junctions and the regulation of sterol-dependent protein trafficking to the plasma membrane. This mechanism may have broader implications for the control of ORP function in higher cells.

Yeast strains and growth conditions

Yeast strains used in this study were based in the YEF473 genetic background (trp163 leu21 ura3-52 his3200 lys2-801) (Bi and Pringle, 1996). Deletion of NVJ1 and VAC8 in YEF473 was described elsewhere (Pan et al., 2000; Pan and Goldfarb, 1998). Unless otherwise indicated, cells were cultured at 30°C in YPD, synthetic complete media (SC), or standard dropout media containing 2% glucose (Sherman, 1991).

Plasmids

Plasmids for the expression of NVJ1, NVJ1-GFP or GFP-NVJ1 under CUP1 promoter control (PCUP1-NVJ1, PCUP1-NVJ1-GFP, PCUP1-GFP-NVJ1) were described previously (Pan et al., 2000). Nvj1p truncations were constructed by PCR-amplifying fragments of NVJ1 corresponding to the indicated amino acid positions with primers that introduced EcoRI and HindIII sites. These fragments were ligated into the EcoRI and HindIII sites of pEGFP-C-FUS, which was constructed from pGFP-C-FUS (Niedenthal et al., 1996) by replacing GFP with a PCR-amplified ClaI-EGFP-SalI fragment generated from pEGFP (Clontech, CA). To express non-fluorescent versions of these truncations, NVJ1 fragments were subcloned from pEGFP-C-FUS into the PGAL1/10 plasmid pESC(HIS) (Stratagene) using EcoRI and ClaI sites. For immunoprecipitation experiments, the FLAG epitope of pESC(HIS) was replaced with the triple HA epitope from pGTEPI (Tyers et al., 1992) using SpeI and SacI sites, and PCR-amplified fragments of NVJ1 were cloned upstream of 3HA using EcoRI and SpeI sites. Deletion of residues 130-176 of Nvj1p was accomplished by fusing two PCR-amplified fragments of NVJ1, corresponding to residues 1-129 and 177-321-myc of PGAL1-NVJ1-myc (Kvam and Goldfarb, 2004), into the multiple cloning site of the PCUP1 plasmid, pRK2 (Pan et al., 2000). To replace the membrane-spanning region of Nvj1p with that of Ste2p, an XholI-substituted chimeric NVJ1/STE2 reverse primer (5′-ccgctcgaggacaatcaaagtcaaagcagctgcaccacatatgacaccaaacataatggccaattcgctggattgtctgg-3′), corresponding to residues 84-90 of Nvj1p and residues 52-68 of Ste2p [including the mutation R58I used by Harley and Tipper (Harley and Tipper, 1996)], was used to PCR-amplify a fragment corresponding to the first 90 residues of NVJ1. This chimeric fragment was ligated to a second NVJ1 PCR fragment (corresponding to residues 121-321) using introduced XholI sites, and cloned into pEGFP-C-fus. Plasmids expressing chimeric or truncated version of NVJ1 were confirmed by sequencing. Plasmids for expression of the tryptophan permease genes TAT1 and TAT2 (pTAT1 and pTAT2) were generously supplied by M. Hall (Schmidt et al., 1994). Plasmids for the constitutive expression of GFP-Osh1p (pRS416-GFP-OSH1) or the inducible expression of Tsc13p-EYFP (PCUP1-TSC13-EYFP) were also described previously (Kvam et al., 2005; Levine and Munro, 2001).

Growth assays for Nvj1p-overexpressing and nvj1-Δ cells

Growth sensitivity on solid media was assayed using previously described methods for tryptophan uptake (Levine and Munro, 2001). Briefly, log-phase trp1 cells harboring the indicated PCUP1 plasmids were induced for 3 hours with 0.1 mM CuSO4. Three OD600 units of culture were then collected by centrifugation. Cell pellets were resuspended in 1 ml of fresh media containing 0.1 mM CuSO4, diluted serially by tenfold, and plated onto selective media containing limiting (15 μg/ml) or excess (40 μg/ml) concentrations of tryptophan using a multi-pronged replicator. Plates were incubated at 25°C or 37°C for 72-96 hours. For growth-curve analyses in liquid media, log-phase trp1 cells harboring the indicated PCUP1 plasmids were induced for 3 hours with 0.1 mM CuSO4. A total of 0.25 OD600 units of culture were collected by centrifugation and inoculated into 5 ml of SC media containing limiting (15 μg/ml) or excess (40 μg/ml) tryptophan and 0.1 mM CuSO4. Approximately 350 μl of culture were pipetted into triplicate wells of a Honeycomb 2 cuvette multiwell plate (MTX Lab Systems, Vienna, VA). Cells were grown at 25°C in a Bioscreen C Analyzer (MTX Lab Systems) programmed to measure the optical density (600 nm) of the Honeycomb 2 plate every 20 minutes, with medium shaking for 10 seconds prior to each reading. Data points were collected using EZExperiment software (MTX Lab Systems). Triplicate readings were averaged for each strain. A similar protocol was used to measure the growth of nvj1-Δ and its parental trp1 strain in SC media containing 5 μg/ml or 15 μg/ml of tryptophan.

Immunoprecipitation and immunoblotting

Cells harboring PGAL1/10-NVJ1(130-177aa)-3HA, PGAL1/10-NVJ1(85-120aa)-3HA, or empty vector and co-expressing either pRS426-GFP-OSH1 or PCUP1-TSC13-EYFP were cultured in selective media containing 2% raffinose to log-phase. Protein expression was induced with 2% galactose or, where appropriate, 0.1 mM CuSO4 for 3.5 hours. Approximately 50 OD600 units of culture were collected by centrifugation. Cell pellets were weighed and overlayed with an equivalent volume of acid-washed 425-600 μm glass beads (Sigma) and 2 volumes of extraction buffer (EB) (40 mM Tris-HCl pH 7.5, 100 mM NaCl, 2 mM DTT, 0.5 mM EDTA) containing Complete™ protease inhibitors (Roche, Mannheim, Germany) and 2 μg/ml pepstatin A, aprotinin, and leupeptin (Sigma). Cells were vortexed five times for 2 minutes, intermittent with 2 minute incubations on ice. Lysate suspensions were solubilized with 1% NP-40, vortexed, and cleared by centrifugation (500 g for 10 minutes). After collecting the supernatant, glass beads were washed with 2 volumes of EB, vortexed, and cleared by centrifugation. Combined supernatants were incubated with 30 μl of goat IgG-agarose for 15 minutes with rotation to remove non-specific peptides. After brief centrifugation, the total protein concentration of the cleared lysate was determined by Bradford assay. Approximately 3 mg of total protein was transferred into MicroSpin Columns (Amersham) and the final volume was brought up to 0.5 ml with EB when necessary. Lysates were incubated with 30 μl of goat anti-HA conjugated agarose at 4°C overnight with rotation (Santa Cruz Biotechnology). Bead complexes were isolated by centrifugation (500 g for 1 minute), washed four times with HNTG (20 mM HEPES pH 7.5, 0.15 M NaCl, 0.1% Triton X-100, 10% glycerol), and eluted into a fresh tube with 35 μl of 2× protein gel sample loading buffer (100 mM Tris-HCl pH 6.8, 2% SDS, 20% glycerol, 2% 2-mercaptoethanol, 0.1% bromophenol blue). Eluates were boiled for 5 minutes, and 10-15 μl of sample were analyzed by 8% and 18% SDS-PAGE. GFP-Osh1p and Tsc13p-EYFP were probed by Immuno blot with polyclonal BD Living Colors™ A.v. peptide antibodies (Clontech). HA-tagged Nvj1p fragments were probed with mouse anti-HA monoclonal antibodies (Santa Cruz Biotechnology). All immuno-probed proteins were detected using alkaline phosphatase-coupled goat anti-rabbit IgG (Santa Cruz Biotechnology) or alkaline phosphatase-coupled goat anti-mouse IgG (Zymed) and developed colometrically

Cell imaging and confocal microscopy

Unless otherwise indicated, cells were induced for specific reporters at log-phase. EGFP-labeled truncations of Nvj1p were expressed from pEGFP-C-FUS by culturing in media lacking methionine for 3 hours. Non-fluorescent truncations of Nvj1p were expressed from pESC(HIS) by shifting cells from raffinose- to galactose-containing media and culturing for 3 hours. Where indicated, Tsc13p-EYFP was co-expressed for 3 hours upon addition of 0.1 mM CuSO4 to the media. Vacuoles were stained with FM4-64 as described previously (Pan and Goldfarb, 1998). Nuclear chromatin was stained immediately prior to microscopic analysis with 5 μM Hoechst reagent H-1398 (Molecular Probes). Confocal microscopy was performed on a Leica TCS NT microscope equipped with a 100X Fluorotar lens and UV, Ar, Kr/Ar, and He/Ne lasers (Leica Microsystems, Chantilly VA). Images were processed using Adobe PhotoShop 5.0 (Adobe Systems, CA). Panels reflect the localization phenotypes observed in all cells overexpressing the indicated reporters.

Immunoelectron microscopy

Wild-type cells harboring PCUP1-NVJ1-GFP or PCUP1-GFP-NVJ1 were grown to log phase and induced for 1 hour with 0.1 mM CuSO4. Cells were high-pressure frozen, freeze substituted, sectioned, and stained as previously described (Giddings et al., 2001). GFP-tagged Nvj1p was detected by immuno-EM using affinity purified rabbit polyclonal anti-GFP antibody and gold-conjugated anti-rabbit secondary antibody. Serial thin sections were viewed with a CM10 electron microscope (Philips Electronic Instruments, Mahwah, NJ) and images were captured with a GATAN digital camera.

We thank T. H. Giddings and J. B. Meehl for their expert electron microscopy, and Xiaozhou Ryan for the confocal images in Fig. 3. We also thank Rebecca Gilson for assistance with tryptophan uptake assays, Michael Hall for providing pTAT1 and pTAT2 plasmids, Tim Levine for providing pRS416-GFP-OSH, and Bill Burke for helpful comments on the manuscript. This study was supported by US National Institutes of Health RO1 grant GM67838 (to D.S.G.).

Beck, T., Schmidt, A. and Hall, M. N. (
1999
). Starvation induces vacuolar targeting and degradation of the tryptophan permease in yeast.
J. Cell Biol.
146
,
1227
-1238.
Beh, C. T. and Rine, J. (
2004
). A role for yeast oxysterol-binding protein homologs in endocytosis and in the maintenance of intracellular sterol-lipid distribution.
J. Cell Sci.
117
,
2983
-2996.
Beilharz, T., Egan, B., Silver, P. A., Hofmann, K. and Lithgow, T. (
2003
). Bipartite signals mediate subcellular targeting of tail-anchored membrane proteins in Saccharomyces cerevisiae.
J. Biol. Chem.
278
,
8219
-8223.
Bendtsen, J. D., Nielsen, H., von Heijne, G. and Brunak, S. (
2004
). Improved prediction of signal peptides: SignalP 3.0.
J. Mol. Biol.
340
,
783
-795.
Bi, E. and Pringle, J. R. (
1996
). ZDS1 and ZDS2, genes whose products may regulate Cdc42p in Saccharomyces cerevisiae.
Mol. Cell. Biol.
16
,
5264
-5275.
Campbell, J. L., Lorenz, A., Witkin, K. L., Hays, T., Loidl, J. and Cohen-Fix, O. (
2006
). Yeast nuclear envelope subdomains with distinct abilities to resist membrane expansion.
Mol. Biol. Cell
17
,
1768
-1778.
Chung, N., Mao, C., Heitman, J., Hannun, Y. A. and Obeid, L. M. (
2001
). Phytosphingosine as a specific inhibitor of growth and nutrient import in Saccharomyces cerevisiae.
J. Biol. Chem.
276
,
35614
-35621.
Cliften, P. F., Hillier, L. W., Fulton, L., Graves, T., Miner, T., Gish, W. R., Waterston, R. H. and Johnston, M. (
2001
). Surveying Saccharomyces genomes to identify functional elements by comparative DNA sequence analysis.
Genome Res.
11
,
1175
-1186.
Crisp, M., Liu, Q., Roux, K., Rattner, J. B., Shanahan, C., Burke, B., Stahl, P. D. and Hodzic, D. (
2006
). Coupling of the nucleus and cytoplasm: role of the LINC complex.
J. Cell Biol.
172
,
41
-53.
Dickson, R. C. (
1998
). Sphingolipid functions in Saccharomyces cerevisiae: comparison to mammals.
Annu. Rev. Biochem.
67
,
27
-48.
Edinger, A. L. and Thompson, C. B. (
2002
). Akt maintains cell size and survival by increasing mTOR-dependent nutrient uptake.
Mol. Biol. Cell
13
,
2276
-2288.
Eisenkolb, M., Zenzmaier, C., Leitner, E. and Schneiter, R. (
2002
). A specific structural requirement for ergosterol in long-chain fatty acid synthesis mutants important for maintaining raft domains in yeast.
Mol. Biol. Cell
13
,
4414
-4428.
Fabrizio, P., Li, L. and Longo, V. D. (
2005
). Analysis of gene expression profile in yeast aging chronologically.
Mech. Ageing Dev.
126
,
11
-16.
Gaber, R. F., Copple, D. M., Kennedy, B. K., Vidal, M. and Bard, M. (
1989
). The yeast gene ERG6 is required for normal membrane function but is not essential for biosynthesis of the cell-cycle-sparking sterol.
Mol. Cell. Biol.
9
,
3447
-3456.
Giddings, T. H., O'Toole, E. T., Morphew, M., Mastronarde, D. N., McIntosh, J. R. and Winey, M. (
2001
). Using rapid freeze and freeze-substitution for the preparation of yeast cells for electron microscopy and three-dimensional analysis.
Methods Cell Biol.
67
,
27
-42.
Harley, C. A. and Tipper, D. J. (
1996
). The role of charged residues in determining transmembrane protein insertion orientation in yeast.
J. Biol. Chem.
271
,
24625
-24633.
Hemmi, K., Julmanop, C., Hirata, D., Tsuchiya, E., Takemoto, J. Y. and Miyakawa, T. (
1995
). The physiological roles of membrane ergosterol as revealed by the phenotypes of syr1/erg3 null mutant of Saccharomyces cerevisiae.
Biosci. Biotechnol. Biochem.
59
,
482
-486.
Hetzer, M. W., Walther, T. C. and Mattaj, I. W. (
2005
). Pushing the envelope: structure, function, and dynamics of the nuclear periphery.
Annu. Rev. Cell Dev. Biol.
21
,
347
-380.
Holmer, L. and Worman, H. J. (
2001
). Inner nuclear membrane proteins: functions and targeting.
Cell. Mol. Life Sci.
58
,
1741
-1747.
Im, Y. J., Raychaudhuri, S., Prinz, W. A. and Hurley, J. H. (
2005
). Structural mechanism for sterol sensing and transport by OSBP-related proteins.
Nature
437
,
154
-158.
Jiang, B., Brown, J. L., Sheraton, J., Fortin, N. and Bussey, H. (
1994
). A new family of yeast genes implicated in ergosterol synthesis is related to the human oxysterol binding protein.
Yeast
10
,
341
-353.
Kohlwein, S. D., Eder, S., Oh, C. S., Martin, C. E., Gable, K., Bacikova, D. and Dunn, T. (
2001
). Tsc13p is required for fatty acid elongation and localizes to a novel structure at the nuclear-vacuolar interface in Saccharomyces cerevisiae.
Mol. Cell. Biol.
21
,
109
-125.
Kvam, E. and Goldfarb, D. S. (
2004
). Nvj1p is the outer-nuclear-membrane receptor for oxysterol-binding protein homolog Osh1p in Saccharomyces cerevisiae.
J. Cell Sci.
117
,
4959
-4968.
Kvam, E., Gable, K., Dunn, T. M. and Goldfarb, D. S. (
2005
). Targeting of Tsc13p to nucleus-vacuole junctions: a role for very-long-chain fatty acids in the biogenesis of microautophagic vesicles.
Mol. Biol. Cell
16
,
3987
-3998.
Lee, M. C., Orci, L., Hamamoto, S., Futai, E., Ravazzola, M. and Schekman, R. (
2005
). Sar1p N-terminal helix initiates membrane curvature and completes the fission of a COPII vesicle.
Cell
122
,
605
-617.
Levine, T. (
2004
). Short-range intracellular trafficking of small molecules across endoplasmic reticulum junctions.
Trends Cell Biol.
14
,
483
-490.
Levine, T. and Rabouille, C. (
2005
). Endoplasmic reticulum: one continuous network compartmentalized by extrinsic cues.
Curr. Opin. Cell Biol.
17
,
362
-368.
Levine, T. P. and Munro, S. (
2001
). Dual targeting of Osh1p, a yeast homologue of oxysterol-binding protein, to both the Golgi and the nucleus-vacuole junction.
Mol. Biol. Cell
12
,
1633
-1644.
Li, G., Yan, Q., Oen, H. O. and Lennarz, W. J. (
2003
). A specific segment of the transmembrane domain of Wbp1p is essential for its incorporation into the oligosaccharyl transferase complex.
Biochemistry
42
,
11032
-11039.
Loewen, C. J., Roy, A. and Levine, T. P. (
2003
). A conserved ER targeting motif in three families of lipid binding proteins and in Opi1p binds VAP.
EMBO J.
22
,
2025
-2035.
Nakamura, H., Miura, K., Fukuda, Y., Shibuya, I., Ohta, A. and Takagi, M. (
2000
). Phosphatidylserine synthesis required for the maximal tryptophan transport activity in Saccharomyces cerevisiae.
Biosci. Biotechnol. Biochem.
64
,
167
-172.
Niedenthal, R. K., Riles, L., Johnston, M. and Hegemann, J. H. (
1996
). Green fluorescent protein as a marker for gene expression and subcellular localization in budding yeast.
Yeast
12
,
773
-786.
Nielsen, H., Engelbrecht, J., Burnak, S. and von Heijne, G. (
1997
). Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage sites.
Protein Eng.
10
,
1
-6.
Ohba, T., Schirmer, E. C., Nishimoto, T. and Gerace, L. (
2004
). Energy- and temperature-dependent transport of integral proteins to the inner nuclear membrane via the nuclear pore.
J. Cell Biol.
167
,
1051
-1062.
Okamoto, M., Yoko-o, T., Umemura, M., Nakayama, K. and Jigami, Y. (
2006
). Glycosylphosphatidylinositol-anchored proteins are required for the transport of detergent-resistant microdomain-associated membrane proteins Tat2p and Fur4p.
J. Biol. Chem.
281
,
4013
-4023.
Ostlund, C., Sullivan, T., Stewart, C. L. and Worman, H. J. (
2006
). Dependence of diffusional mobility of integral inner nuclear membrane proteins on A-type lamins.
Biochemistry
45
,
1374
-1382.
Pan, X. and Goldfarb, D. S. (
1998
). YEB3/VAC8 encodes a myristylated armadillo protein of the Saccharomyces cerevisiae vacuolar membrane that functions in vacuole fusion and inheritance.
J. Cell Sci.
111
,
2137
-2147.
Pan, X., Roberts, P., Chen, Y., Kvam, E., Shulga, N., Huang, K., Lemmon, S. and Goldfarb, D. S. (
2000
). Nucleus-vacuole junctions in Saccharomyces cerevisiae are formed through the direct interaction of Vac8p with Nvj1p.
Mol. Biol. Cell
11
,
2445
-2457.
Patel, S., Rose, A., Meulia, T., Dixit, R., Cyr, R. J. and Meier, I. (
2004
). Arabidopsis WPP-domain proteins are developmentally associated with the nuclear envelope and promote cell division.
Plant Cell
16
,
3260
-3273.
Powers, R. W., III, Kaeberlein, M., Caldwell, S. D., Kennedy, B. K. and Fields, S. (
2006
). Extension of chronological life span in yeast by decreased TOR pathway signaling.
Genes Dev.
20
,
174
-184.
Roberts, P., Moshitch-Moshkovitz, S., Kvam, E., O'Toole, E., Winey, M. and Goldfarb, D. S. (
2003
). Piecemeal microautophagy of the nucleus in Saccharomyces cerevisiae.
Mol. Biol. Cell
14
,
129
-141.
Schmidt, A., Hall, M. N. and Koller, A. (
1994
). Two FK506 resistance-conferring genes in Saccharomyces cerevisiae, TAT1 and TAT2, encode amino acid permeases mediating tyrosine and tryptophan uptake.
Mol. Cell. Biol.
14
,
6597
-6606.
Schmidt, A., Beck, T., Koller, A., Kunz, J. and Hall, M. N. (
1998
). The TOR nutrient signaling pathway phosphorylates NPR1 and inhibits turnover of the tryptophan permease.
EMBO J.
17
,
6924
-6931.
Sherman, F. (
1991
). Getting started with yeast.
Meth. Enzymol.
194
,
3
-21.
Somech, R., Shaklai, S., Amariglio, N., Rechavi, G. and Simon, A. J. (
2005
). Nuclear envelopathies-raising the nuclear veil.
Pediatr. Res.
57
,
8R
-15R.
Starr, D. A. and Fischer, J. A. (
2005
). KASH `n Karry: the KASH domain family of cargo-specific cytoskeletal adaptor proteins.
BioEssays
27
,
1136
-1146.
Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. and Higgins, D. G. (
1997
). The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools.
Nucleic Acids Res.
25
,
4876
-4882.
Tyers, M., Tokiwa, G., Nash, R. and Futcher, B. (
1992
). The Cln3-Cdc28 kinase complex of S. cerevisiae is regulated by proteolysis and phosphorylation.
EMBO J.
11
,
1773
-1784.
Umebayashi, K. and Nakano, A. (
2003
). Ergosterol is required for targeting of tryptophan permease to the yeast plasma membrane.
J. Cell Biol.
161
,
1117
-1131.
Voeltz, G. K., Rolls, M. M. and Rapoport, T. A. (
2002
). Structural organization of the endoplasmic reticulum.
EMBO Rep.
3
,
944
-950.
Waizenegger, T., Stan, T., Neupert, W. and Rapaport, D. (
2003
). Signal-anchor domains of proteins of the outer membrane of mitochondria: structural and functional characteristics.
J. Biol. Chem.
278
,
42064
-42071.
Worman, H. J. and Gundersen, G. G. (
2006
). Here come the SUNs: a nucleocytoskeletal missing link.
Trends Cell Biol.
16
,
67
-69.