(Received for publication, May 8, 1997, and in revised form, June 11, 1997)
From the Departments of Cell and Molecular Biology, The Scripps Research Institute, La Jolla, California 92037
The GTPase Ran/TC4 and the 14-kDa protein nuclear transport factor 2 (NTF2) are two of the cytosolic factors that mediate nuclear protein import in vertebrates. Previous biochemical studies have shown that NTF2 binds directly to the GDP-bound form of Ran/TC4 and to proteins of the nuclear pore complex that contain phenylalanine-glycine repeats. In the present study we have used molecular genetic approaches to study the Saccharomyces cerevisiae homologue of NTF2. The scNTF2 gene encodes a protein that is 44% identical to the human protein. We found that deletion of the scNTF2 gene is lethal and that repression of NTF2p expression by a regulatable promoter results in gross structural distortions of the nuclear envelope. In a screen for high copy number suppressors of a scNTF2 deletion, the only gene we isolated other than scNTF2 itself was GSP1, the S. cerevisiae homologue of Ran/TC4. Furthermore, we found that high levels of Ran/TC4 can relieve the requirement for NTF2 in a mammalian-permeabilized cell assay for nuclear protein import. These data suggest that certain of the nuclear protein import functions of NTF2 and Ran/TC4 are closely linked and that NTF2 may serve to modulate a transport step involving Ran/TC4.
Exchange of macromolecules between the cytoplasm and nucleus is
specified by signals encoded within transported proteins. The signals
that specify nuclear protein import, which are called nuclear
localization signals (NLSs),1
usually consist of short stretches of amino acids enriched in basic
amino acid residues (1). A group of cytosolic factors that mediate
nuclear import of proteins containing these basic-type NLSs has been
identified. These include the subunit of the NLS receptor (
importin/
karyopherin; Refs. 2-4) and its
subunit (p97/
importin/
karyopherin; Refs. 5-7), the GTPase Ran/TC4 (8, 9), and
the homodimeric protein nuclear transport factor 2 (NTF2/p10; Refs. 10
and 11). According to current working models (1, 12), the pathway for
import of proteins with basic-type NLSs begins in the cytoplasm, where
the
subunit of the receptor binds to the NLS of a protein destined
for import. This complex then binds to the cytoplasmic surface of the
nuclear pore complex (NPC) via the
subunit, an event that is
referred to as the initial binding or docking step of transport.
Subsequently, the substrate-receptor complex is delivered to the
central channel of the NPC and then translocated into the nucleoplasm
(events collectively referred to as the translocation step; Ref. 13).
The latter processes very likely involves Ran/TC4 and NTF2.
A number of binding interactions of Ran/TC4 and NTF2 have been
described by in vitro biochemical studies, and these
associations can provide a framework for studying the precise functions
of these factors in nuclear import. The GTP-bound form of Ran/TC4 can
bind to three distinct proteins implicated in nuclear import: the subunit of the NLS receptor (14, 15), RanBP1 (16), and the NPC protein
RanBP2/Nup358 (17, 18). The GDP form of Ran/TC4 can bind to NTF2 (14,
19) as well as to a complex containing RanBP1 and the
subunit of
the NLS receptor (20). NTF2 itself can bind directly to multiple
proteins of the NPC containing FXFG repeat motifs, including
p62 (10, 19, 21). It has become clear that elucidating the mechanism of
nuclear protein import will require understanding how the cytosolic
transport factors interact with each other and with the NPC at specific transport steps and how these interactions are coupled to the GTPase
cycle of Ran/TC4 (1, 12).
The mechanisms of nuclear protein import are predicted to be conserved between vertebrates and Saccharomyces cerevisiae. Basic-type NLSs within proteins such as the glucocorticoid receptor and the SV40 large T antigen are fully functional in yeast, and the primary structures of the cytosolic transport factors show 40-80% identity when human and yeast proteins are compared. The basic architecture of the NPC appears to be conserved as well (22). We have, therefore, used a molecular genetic approach to characterize NTF2 in the budding yeast S. cerevisiae. We have found that the scNTF2 gene is required for viability and that depletion of scNTF2p leads to nuclear structure defects similar to those observed with mutants of certain nuclear pore complex proteins. We carried out a genetic selection for multicopy suppressors that would compensate for loss of scNTF2p. This analysis revealed that the S. cerevisiae homologue of Ran/TC4 is a multicopy suppressor of a scNTF2 deletion strain, thus providing the first in vivo evidence that certain functions of NTF2 and Ran/TC4 are linked. Furthermore, we have used recombinant proteins to show that Ran/TC4 can relieve the requirement for NTF2 in a permeabilized cell import assay in mammalian cells. Taken together, these data support the hypothesis that NTF2 serves to modulate a transport step(s) involving Ran/TC4.
To generate the plasmid used for methionine-regulated expression of scNTF2p, the scNTF2 ORF was amplified by PCR and cloned into the unique BamHI site immediately downstream of the MET3 promoter in the plasmid pRS405-MET3 (generously provided by Drs. Steve Haase and Steve Reed, The Scripps Research Institute). The MET3-scNTF2 fragment was then removed by digestion with NotI and HindIII and cloned into pRS315 (23) to generate LEU2-marked construct pMET-scNTF2. The plasmid for galactose-regulated expression of scNTF2 was constructed by PCR amplification of the scNTF2 ORF and subsequent cloning into the unique EcoRI site of yCPG2 (provided by Drs. Steve Haase and S. Reed).
Yeast StrainsAll yeast strains used in this study are
derivatives of the homozygous diploid strain 15D (MATa/MAT,
ura3/ura3, leu2/leu2, trp1/trp1, his2/his2, ade1/ade1, provided by
Drs. Steve Haase and Steve Reed). A single copy of the
scNTF2 locus was disrupted in 15D via a PCR-based method to
generate BPY1 (scNTF2/
scNTF2:: URA3) and BPY3
(scNTF2/
scNTF2::TRP1). Specifically, to
generate BPY1, PCR was used to generate a URA3 fragment
flanked by sequences homologous to flanking regions of the
scNTF2 gene such that upon integration, the entire ORF of
scNTF2 would be deleted. The oligos used were
5
-TAAGGAACCCAGGTTTTAATACTATTATCTTTATAATGGCTTTTCAATTCAATCATC-3
and
5
-AAATACATGTTTCTGTGGTGACTTAAAAAATCCTTAAGCAGATTCCCGGGTAATAACTG-3
. After PCR amplification, the fragment was gel-purified and used to transform the diploid strain 15D to Ura+. To confirm
correct integration of the construct at the scNTF2 locus,
whole cell PCR was performed with flanking primers (primer 1 = 5
-CAAGGTGAGACTTAGGCTGATAAG-3
, primer 2 = 5
-GCCTTATACATCGTTAGCTAAGC-3
) or one flanking primer (primer 1) and
one URA3-internal primer, primer 3 = 5
-GCTGACATTGGTAATACAGTC-3
). Isolates in which both PCR reactions
produced fragments of the size expected for integration at
scNTF2 were judged to be deletion isolates. The same method was applied to replace the scNTF2 ORF with TRP1
in BPY3. The primers used to generate the PCR product for
TRP1 integration into the scNTF2 locus were
5
-CCTAAGGAACCCAGGTTTTAATACTATTATCTTTATAATGGGCATTGGTGACTATTGAGC-3
and
5
-AAATACATGTTTCTGTGGTGACTTAAAAAATCCTTAAGCAGGCAAGTGCACAAACAATAC-3
. Strain BPY4 (
scNTF2::TRP1
containing pMET3-scNTF2) was generated by transforming BPY3
with the pMET3-scNTF2 plasmid followed by sporulation and dissection to
yield a Leu+, Trp+ haploid. The mating type of
this strain was not determined. A similar procedure was used to
generate a
scNTF2 haploid carrying the plasmid
pGAL10-scNTF2. Propagation of yeast strains was performed using
standard media recipes as described in Ref. 24.
The screen for multicopy suppressors was carried out in the strain BPY4. Log phase yeast cells were transformed by using polyethylene glycol and lithium acetate (25) with a S. cerevisiae genomic library constructed in YEp24 (26) and plated onto synthetic medium containing 5 mM methionine and lacking uracil. We plated 1% of the transformation onto media lacking methionine and determined that this procedure yielded a total of ~108 Ura+ transformants. The 480 Ura+ colonies that grew under the condition of scNTF2p repression were replated onto methionine-containing media. Whole cell PCR was used to screen the 480 colonies for the presence of the library plasmids that contained the scNTF2 gene.
Electron MicroscopyLog phase yeast cells were harvested up to 7 h after the addition of methionine and processed for electron microscopy essentially as described by Byers and Goetsch (27). We used both Spurr and Epon resins, the latter of which yielded slightly better preservation of membranes in our experiments.
Other MethodsThe permeabilized cell assay used to quantitate nuclear protein import in HeLa cells was described previously (10). The reporter molecule for nuclear protein import was fluorescein isothiocyanate-labeled bovine serum albumin-conjugated with the SV40 large T antigen NLS (10). Pretreatment of HeLa cell cytosol with p62-Sepharose and preparation of recombinant NTF2 and Ran/TC4 have also been described previously (8, 10). The antibody to recombinant human NTF2 was generated in rabbits, affinity purified on NTF2-Sepharose, and used at a final concentration of 2 µg/ml. Anti-serum to the GSP1 protein (28) was used at a dilution of 1:3000 and was kindly provided by Dr. Pierre Belhumeur (University of Montreal). Immunoblots were developed with peroxidase-labeled secondary antibodies and enhanced chemiluminescence.
Sequence from the S. cerevisiae genome project revealed
an open reading frame on chromosome V (cosmid 9537) that is highly related to the sequence of NTF2, a cytosolic factor that facilitates protein transport into the nucleus in mammalian cells (10). The
predicted ORF encodes a 125-amino acid protein that is 44.4% identical
and 61.5% similar to the 127-amino acid human NTF2 protein (Fig.
1A). The central region of the
protein (Phe-12-Pro-76) is 58% identical, whereas the primary
structures of the amino and carboxyl termini are unrelated except for
residues Asn-116-Leu-121. The sequence relationship together with the
similarities in size and isoelectric point suggest the yeast ORF
encodes the S. cerevisiae homologue of NTF2 and is hereafter
referred to as scNTF2p. Evidence that these proteins are functional
homologues was obtained in recent work showing that the human cDNA
encoding NTF2 can substitute for the yeast gene (29).
The scNTF2p Is Required for Viability
We assessed the
requirement for the scNTF2p in cell viability by standard gene
replacement methods in the yeast strain 15D. To generate an
scNTF2 deletion allele, we first used PCR primers that
included sequences complementary to the 5 and 3
regions of the
scNTF2 gene to amplify the URA3 gene (Fig.
1B). The diploid strain 15D was then transformed with the
linear PCR product. Stable Ura+ transformants were screened
for integration of the URA3 gene by PCR using
oligonucleotides that flank the scNTF2 locus (primers 1 and
2; Fig. 1C). The DNA template from one Ura+
transformant yielded PCR products with sizes of ~560 and ~1300 base
pairs. The smaller product is the size expected from amplification of
the scNTF2 locus, whereas the larger product is consistent with the replacement of the scNTF2 ORF with the
URA3 gene. This was confirmed by showing that a PCR reaction
with a primer that flanks scNTF2 (primer 1) and a primer
within the URA3 gene (primer 3) produces a ~700-base pair
fragment.
The heterozygote deletion strain for scNTF2
(scNTF2/ scNTF2::URA3,
denoted BPY1) displayed no obvious defects in growth, morphology, or
thermosensitivity (data not shown). Sporulation of the BPY1 strain and
subsequent analysis of >40 tetrads revealed a 2:0 segregation
(live:dead), demonstrating that the scNTF2 gene is required
for viability (Fig. 1D). We confirmed that viable segregants
were Ura
by their failure to grow in the absence of
uracil. These data are in agreement with those obtained using other
strains of S. cerevisiae (PSY853, Ref. 29; W303, data not
shown, Ref. 19).
As a first step toward
analyzing the function of scNTF2p in vivo, we examined the
effects of scNTF2p depletion in haploid deletion strains containing
plasmids with the scNTF2 gene under the control of different
regulatable promoters. Initial experiments were carried out with
scNTF2p expression regulated by the GAL10 promoter. We
observed that colony growth in haploid deletion strains carrying the
GAL10 plasmid was slowed substantially but not completely inhibited (data not shown), suggesting that expression from this plasmid is leaky. We chose the MET3 promoter as an
alternative for achieving repressible scNTF2p expression. Transcription
driven by this promoter can be strongly repressed by growing cells in the presence of millimolar concentrations of methionine (30). In a
haploid deletion strain (BPY4) containing scNTF2 under the control of the MET3 promoter, transcription from the scNTF2
plasmid in the absence of methionine restored growth of this strain to the same level as the WT strain (Fig. 2,
Methionine). In contrast, substantial growth inhibition
resulted from including 5 mM methionine in the media, an
effect that was manifest as an approximately 100-fold difference in
viability as compared with the WT strain (Fig. 2,
+Methionine). The BPY4 colonies that did survive selection in the presence of methionine were also visibly smaller than colonies composed of WT cells.
Methionine-repression of the scNTF2 gene in the BPY4 strain
resulted in distortion and apparent fragmentation of the nucleus such
that the nuclear DNA appeared as small 4,6-diamidino-2-phenylindole dihydrochloride staining domains within these cells (data not shown).
Surprisingly, the distorted nuclei present in the mutant cells
accumulated a NLS-containing version of green fluorescent protein.2 However, it was not
possible to quantititively compare the nuclear import rate in these
mutant cells to the rate in wild type cells, due to the altered nuclear
morphology of the mutant cells. We note that two mutant alleles of
scNTF2 have been isolated that show a temperature-sensitive
defect in protein import (29), providing direct evidence that scNT2p is
involved in nuclear import.
To further characterize the altered nuclear morphology associated with
depletion of the scNTF2p, we grew the WT and BPY4 strains in the
presence of methionine for 7 h and processed the cells for
thin section electron microscopy. Cells from the BPY4 strain (Fig.
3B) displayed grossly
distorted nuclear envelopes that often took the form of large
intranuclear invaginations. These structures were never observed in WT
cells (Fig. 3A). Some of the mutant cells contained areas
with apparent discontinuities or holes in the nuclear envelope
(arrowheads, Fig. 3B). We also observed the juxtaposition of two double pore-containing membranes within a single
nucleus and the alignment of NPCs (arrows, Fig.
3D). Interestingly, similar nuclear morphology phenotypes
have been reported for yeast strains deficient for certain NPC
proteins. For example, repression or overexpression of the nucleoporin
Nup170p in a POM152 null background results in structural
alterations of the nuclear envelope (29) that are nearly identical to
the changes observed by simple repression of scNTF2p expression
reported here. These observations suggest that scNTF2p is involved in
the ongoing maintenance of nuclear structure. This could be indirectly
due to the need for scNTF2p in nuclear protein import or could reflect
a direct involvement of scNTF2p in some feature of nuclear
architecture.
A Screen for Multicopy Suppressors of scNTF2
We carried out a multicopy suppressor screen to identify components that allow cells to grow in the absence of NTF2p, with the goal of identifying functionally linked components in the nuclear protein import pathway. The screen relied on the efficient repression of scNTF2p expression obtained with the MET3 promoter (30). The relatively small number and size of colonies that grew in the presence of methionine (Fig. 2) suggested that the selection scheme would have a background of <1%. This type of screen could select for genes that functionally overlap with scNTF2 or that act in parallel functional pathways.
We transformed the BPY4 strain with a high copy number S. cerevisiae genomic library (26) and plated the cells on selective media containing 5 mM methionine. From this we obtained nearly 500 colonies that grew in the presence of methionine in both the initial and secondary plating steps. Since we expected that the scNTF2 gene would represent a significant proportion of the putative suppressors, we employed a PCR screen with primers that flanked the scNTF2 gene to rapidly identify library clones containing scNTF2. Since the genomic scNTF2 ORF in the BPY4 strain had been replaced with the TRP1 gene, amplification of this locus generated a product whose larger size distinguished it from the product amplified from the scNTF2 gene on a library plasmid. The scNTF2 ORF under the control of the MET3 promoter was not detected in this screen due to its lack of complementarity with the PCR primers used. A total of 480 isolates were examined by this procedure. Two-thirds of the colonies (322 of 480) were determined to be scNTF2 by this criteria, demonstrating the validity of our selection regimen. Furthermore, the strong selection for scNTF2 suppressors strongly suggests that the methionine-induced repression of transcription from the pMET-scNTF2 plasmid (together with cell doubling) leads to depletion of cellular scNTF2p.
Three of the 158 clones that did not contain the scNTF2 gene
were analyzed further. We sequenced approximately 400 base pairs at
each end of these clones and learned all three were derived from an
approximately 7-kilobase segment of chromosome XII. This region of
chromosome XII is notable as it includes the locus for GSP1,
the S. cerevisiae homologue of Ran/TC4 (28). These three clones also contained the gene for a GTPase of unknown function, termed
GUF1 (31). We determined by deletion analysis that the suppressor activity can be ascribed to the portion of the insert that
includes the GSP1 gene. The BPY4 strain carrying the
pMET-scNTF2 plasmid was transformed with the original library plasmid
or the deletion variants as depicted in Fig.
4A. A dilution series of each
culture was then applied to solid medium prepared without or with
methionine. In this assay, we found that growth in the presence of
methionine could be rescued by the full-length library insert or
the portion containing GSP1, whereas a deletion construct containing the GUF1 locus failed to rescue (Fig.
4A).
We analyzed 27 additional candidate suppressor clone candidates from the 158 methionine-resistant colonies by Southern blotting and PCR and determined that 25 of these 27 clones contained the GSP1 gene. The remaining two clones failed to rescue growth on methionine when reintroduced into yeast and likely represent the background in this screen. These data indicate that the vast majority of the clones that rescued growth on methionine are GSP1.
Our finding that cells depleted of an essential gene product (scNTF2p)
can be rescued by a gene that is unrelated in sequence (GSP1) is surprising. One explanation is that in the
presence of methionine the scNTF2p was still expressed at low levels,
and in this background overexpression of the GSP1p suppressed the growth defect. Alternatively, it was possible that the scNTF2p was
virtually absent from the mutant strain, and overexpression of the
GSP1p compensated for this loss. To address this issue, we tested
whether overexpression of GSP1p would allow a complete loss of the
pMET-scNTF2 plasmid in the BPY4 (scNTF2) strain. We transformed the
BPY4 strain with pURA-GSP1 and maintained transformants in continuous
culture for 48 h. This was done in the absence of uracil to
maintain the pURA-GSP1 plasmid and in the presence of methionine and
leucine to repress transcription and to allow for loss of the LEU
plasmid pMET-scNTF2, respectively. The cultures were plated at low
density on medium lacking uracil, grown for three days, and
replica-plated onto media lacking leucine to score for the
presence/absence of the pMET-scNTF2 plasmid. Approximately one-third of
the colonies were Ura+ and Leu
(data not
shown), indicating that overexpression of GSP1p can, in fact,
compensate for the complete loss of scNTF2p in vivo. We also
found that the growth rate of the strain carrying pGSP1 and lacking
pMET-scNTF2 was essentially identical to the WT strain (Fig.
4B). GSP1p expression in this strain was elevated 1.73-fold relative to WT cells carrying only a plasmid to confer growth in the
absence of uracil (Fig. 4C). It is not surprising that a
higher level of GSP1p expression is not achieved, considering that
GSP1p in WT cells is already a very abundant cellular protein. Taken
together, these data establish unequivocally that modulating the level
of GSP1p relieves the requirement for what is otherwise an essential
gene.
Our finding that a high level of the S. cerevisiae homologue of Ran/TC4 can rescue growth in a mutant
yeast strain that lacks the scNTF2 gene prompted us to
examine the requirements for these proteins using an assay that
measures nuclear protein import in permeabilized mammalian cells.
Specifically, we sought to determine if the addition of recombinant
Ran/TC4 could serve to biochemically complement cytosol depleted of the
NTF2 protein. We prepared NTF2-depleted cytosol by passing a high speed
extract of HeLa cells over p62-Sepharose (10). This treatment removes
>96% of cytosolic NTF2 as determined by quantitative immunoblotting,
but it does not remove Ran/TC4 from cytosol (Ref. 10 and data not
shown). We measured the nuclear protein import supported by this
cytosol as a function of Ran/TC4 concentration both in the absence and
presence of NTF2 (Fig. 5A). The addition of purified recombinant NTF2 to NTF2-depleted cytosol stimulated transport 2.3-fold, an effect that was altered only slightly
by adding low concentrations (<2 µg/ml) of Ran/TC4 protein. However,
the addition of Ran/TC4 (25 (µg/ml) to the NTF2-depleted cytosol
(Fig. 5A, lower curve) restored transport to
levels that are comparable to those obtained in the presence of both
NTF2 and Ran/TC4 (Fig. 5A, upper curve). These
data show that the requirement for NTF2 protein in the nuclear protein
import assay can be relieved by supplementing the transport reaction
with recombinant Ran/TC4 protein. We note that low levels of NTF2
contributed by the depleted cytosol and the permeabilized cells
together with the Ran/TC4 in the depleted cytosol probably accounts for
the transport obtained without supplementing the reaction with
recombinant NTF2 (Fig. 5B).
Our initial characterization of scNTF2p in the budding yeast S. cerevisiae has provided valuable insight on the role of this protein in cell function. We have found that this protein is required for cell viability, in agreement with observations from other laboratories (19, 29). Moreover, we observed that depletion of scNTF2p leads to aberrant nuclear morphologies characterized by gross invaginations of the nuclear envelope and juxtaposition of NPCs in stacked nuclear membranes, similar to the morphological phenotypes obtained with mutants of certain NPC proteins (32). Most significantly, we have demonstrated that the scNTF2 null mutant can be rescued by GSP1, the S. cerevisiae homologue of Ran/TC4. This provides the first in vivo evidence that certain functions of NTF2 and Ran/TC4 are linked. These linked functions probably relate to nuclear protein import, since both Ran/TC4 and scNTF2 are involved in nuclear import in yeast (29, 33) as well as in higher eukaryotes (12).
Our present genetic analysis does not unequivocally demonstrate a direct interaction between GSP1 and scNTF2, and it is formally possible that GSP1 rescues the scNTF2 null mutant by an indirect mechanism. However, addition of recombinant Ran/TC4 to digitonin-permeabilized HeLa cells depleted of NTF2 results in a rescue of nuclear protein import. Taken together, these observations suggest that the rate of nuclear protein import is closely tied to the cellular concentrations of NTF2 and Ran/TC4. The finding that GSP1p levels were elevated only 1.73-fold in scNTF2 null mutants was surprising given that the GSP1 gene was carried on a high copy number (2 µm) plasmid. This may be explained by the fact that overexpression of GSP1 is slightly toxic in the strains used in this study (Fig. 4A). Therefore, under the conditions of the genetic selection, growth was inhibited both by the repression of scNTF2p expression and by the overexpression of GSP1p from the library plasmid.
In vertebrates, both Ran/TC4 and NTF2 are required for transport event(s) that occur during movement of a substrate-NLS receptor complex through the NPC. There is good evidence that during the transport process the GTP form of Ran/TC4 binds to RanBP2, a putative initial substrate/NLS receptor docking site at the NPC and that RanGAP-stimulated hydrolysis at this site is involved in the forward progress of a transport complex (34, 35). After the transport complex leaves RanBP2, it is not known whether nucleotide exchange on Ran/TC4 and additional rounds of Ran-GTP hydrolysis are required to drive translocation through the nuclear pore (15, 35).
The role of NTF2 in nuclear protein import is even less well defined. NTF2 binds directly to several FXFG-type repeat proteins of the NPC as well as to the GDP form of Ran, suggesting it could play a role in modulating the interaction of transport complexes with several discrete NPC sites (14, 91, 21). Our observations that a high level of Ran/TC4 can relieve the requirement for NTF2 in cell viability (in yeast) and nuclear protein import (in mammalian cells) suggests that NTF2 may regulate a protein-protein interaction(s) that involves Ran/TC4. For example, dissociation of a substrate-NLS receptor complex from RanBP2 during its transfer to another NPC protein could be promoted by the binding of either NTF2 or Ran-GTP to RanBP2, and in the absence of NTF2 this could occur more rapidly with an increased concentration of Ran/TC4. Alternatively, NTF2 may contribute to binding or dissociation reactions promoted by RanGDP, if the latter itself acts as a soluble factor to promote a step(s) of nuclear import (36). Our observation that NTF2 is dispensable under certain conditions argues against the hypothesis that NTF2 promotes GDP/GTP exchange on Ran/TC4 during nuclear import (19).
In summary, we have demonstrated that the scNTF2 gene encodes an essential protein that is linked to some function(s) of GSP1, the S. cerevisiae homologue of Ran/TC4. The ability of a high level of Ran/TC4 to relieve the requirement for an otherwise essential gene product in vivo as well as in vitro places constraints on possible functions for NTF2 in nuclear protein import. We favor a role for NTF2 in regulating protein-protein interactions that involves Ran/TC4 at the nuclear pore complex.
We thank Drs. Steve Haase, Peter Kaiser, and Steve Reed for gifts of strains and plasmids and for helpful advice. We thank Drs. Janet Leatherwood and Paul Russell for instruction and use of their tetrad dissection apparatus. We thank Dr. David Goldfarb for examining our strains for potential protein import defects and Dr. Pierre Belhumeur for providing the antiserum to GSP1. We also thank Konrad Zeller for technical assistance.