Subcellular localization of the hypusine-containing eukaryotic initiation factor 5A by immunofluorescent staining and green fluorescent protein tagging.

Abstract

Eukaryotic initiation factor 5A (eIF-5A) is the only protein in nature that contains hypusine, an unusual amino acid residue formed posttranslationally by deoxyhypusine synthase and deoxyhypusine hydroxylase. Although the eIF-5A gene is essential for cell survival and proliferation, the precise function and localization of eIF-5A remain unclear. In this study, we have determined the subcellular distribution of eIF-5A by indirect immunofluorescent staining and by direct visualization of green fluorescent protein tagged eIF-5A (GFP-eIF5A). Immunofluorescent staining of the formaldehyde-fixed cells showed that eIF-5A was present in both the nucleus and cytoplasm. Only the nuclear eIF-5A was resistant to Triton extraction. Direct visualization of GFP tagged eIF-5A in living cells revealed the same whole-cell distribution pattern. However, a fusion of an additional pyruvate kinase (PK) moiety into GFP-eIF-5A precluded the nuclear localization of GFP-PK-eIF-5A fusion protein. Fusion of the GFP-PK tag with three different domains of eIF-5A also failed to reveal any nuclear localization of the fusion proteins, suggesting the absence of receptor-mediated nuclear import. Using interspecies heterokaryon fusion assay, we could detect the nuclear export of GFP-Rev, but not of GFP-eIF-5A. The whole-cell distribution pattern of eIF-5A was recalcitrant to the treatments that included energy depletion, heat shock, and inhibition of transcription, translation, polyamine synthesis, or CRM1-dependent nuclear export. Collectively, our data indicate that eIF-5A gains nuclear entry via passive diffusion, but it does not undergo active nucleocytoplasmic shuttling.

Full text

Page 1

Journal of Cellular Biochemistry 86:590–600 (2002)
Subcellular Localization of the Hypusine-Containing
Eukaryotic Initiation Factor 5A by Immunofluorescent
Staining and Green Fluorescent Protein Tagging
David Li-En Jao and Kuang Yu Chen*
Department of Chemistry and Chemical Biology, Joint Graduate Program in Cell and Developmental
Biology, Rutgers, The State University of New Jersey, Piscataway, New Jersey 08854-8087
Abstract
amino acid residue formed posttranslationally by deoxyhypusine synthase and deoxyhypusine hydroxylase. Although
the eIF-5A gene is essential for cell survival and proliferation, the precise function and localization of eIF-5A remain
unclear. In this study, we have determined the subcellular distribution of eIF-5A by indirect immunofluorescent staining
and by direct visualization of green fluorescent protein tagged eIF-5A (GFP-eIF5A). Immunofluorescent staining of the
formaldehyde-fixed cells showed that eIF-5A was present in both the nucleus and cytoplasm. Only the nuclear eIF-5A
was resistant to Triton extraction. Direct visualization of GFP tagged eIF-5A in living cells revealed the same whole-cell
distribution pattern. However, a fusion of an additional pyruvate kinase (PK) moiety into GFP-eIF-5A precluded the
nuclear localization of GFP-PK-eIF-5A fusion protein. Fusion of the GFP-PK tag with three different domains of eIF-5A
also failed to reveal any nuclear localization of the fusion proteins, suggesting the absence of receptor-mediated nuclear
import. Using interspecies heterokaryon fusion assay, we could detect the nuclear export of GFP-Rev, but not of GFP-
eIF-5A. The whole-cell distribution pattern of eIF-5A was recalcitrant to the treatments that included energy depletion,
heat shock, and inhibition of transcription, translation, polyamine synthesis, or CRM1-dependent nuclear export.
Collectively, our data indicate that eIF-5A gains nuclear entry via passive diffusion, but it does not undergo active
nucleocytoplasmic shuttling. J. Cell. Biochem. 86: 590–600, 2002.
Eukaryotic initiation factor 5A (eIF-5A) is the only protein in nature that contains hypusine, an unusual
? 2002 Wiley-Liss, Inc.
Key words: eIF-5A; hypusine; GFP-tagging; subcellular localisation
Eukaryotic initiation factor 5A (eIF-5A) is
the only protein in nature that contains a
hypusine residue. This unusual amino acid
formed posttranslationally through the action
of deoxyhypusine synthase and deoxyhypusine
hydroxylase [review by Park et al., 1993, 1997;
Chen and Liu, 1997; Chen and Jao, 1999].
Disruption of either eIF-5A or deoxyhypusine
synthase gene in yeast leads to a lethal
phenotype [Schnier et al., 1991; Sasaki et al.,
1996]. Inhibition of deoxyhypusine synthase
activity in mammalian cells causes growth
arrest [Jakus et al., 1993; Park et al., 1994], cell
death [Tome et al., 1997], or tumor differentia-
tion [Chen et al., 1996]. In addition, hypusine
formation exhibits a striking attenuation in
senescent cells [Chen and Chen, 1997a], but a
marked increase in virally-transformed cells
[Chen and Chen, 1997b].
Although, eIF-5A is essential for cell survival
and proliferation, its physiological function is
unclear. Due to the lack of a clear correlation
with general protein synthesis, eIF-5A may
not be a bona fide translation initiation factor
[Kang and Hershey, 1994]. The suggestion that
eIF-5A may serve as a target protein for Rev or
Rex [Ruhl et al., 1993; Katahira et al., 1995]
has been questioned because a direct pro-
tein interaction cannot be established in vitro
[Henderson and Percipalle, 1997; Mattaj and
Englmeier, 1998; Lipowsky et al., 2000]. Based
on the finding that eIF-5A can bind to RRE
? 2002 Wiley-Liss, Inc.
Grant sponsor: National Cancer Institute, NIH (United
States Public Health Services); Grant number: RO1
CA49695.
*Correspondence to: Dr. Kuang Yu Chen, Department of
Chemistry, 610 Taylor Road, Rutgers, The State University
of New Jersey, Piscataway, NJ 08854-8087.
E-mail: KYCHEN@rutchem.rutgers.edu
Received 23 January 2002; Accepted 30 April 2002
DOI 10.1002/jcb.10235

Page 2

and U6 and that eIF-5A recognizes synthetic
RNA in a sequence-dependent manner, we
proposed that eIF-5A may function as an RNA-
binding protein [Liu et al., 1997; Xu and Chen,
2001].ThefindingthateIF-5Acomplementsthe
defectinmRNAdecayphenotypesofatempera-
ture-sensitiveyeastmutant[ZukandJacobson,
1998] indirectly suggests that eIF-5A may be
involved in the mRNA metabolism.
The intracellular localization of eIF-5A has
been previously studied, but with conflicting
results [Ruhl et al., 1993; Shi et al., 1996].Since
the information of protein localization can pro-
vided important clues on protein function, we
decided to examine the subcellular distribution
of eIF-5A not only by in vitro immunofluores-
cent staining, but also by direct visualization of
eIF-5A using the GFP-tagging technique. GFP
is an autofluorescent protein derived from the
jellyfishAequoreavictoriaandGFPtagginghas
been widely used as a marker for protein
targeting and localization in aliving cell [Tsien,
1998]. Moreover, fusion proteins of GFP with
truncated eIF-5A allowed us to determine pos-
sible presence of domains for nuclear import
and export.
MATERIALS AND METHODS
Materials
All chemicals were purchased from Sigma
(St.Louis,MO)unlessotherwiseindicated.Poly-
clonal antibody against recombinant human
eIF-5A was raised in chicken and affinity-
purified by recombinant eIF-5A proteins. The
purifiedantibodydoesnotcross-reactwithother
cellular proteins [Chen and Jao, 1999]. Rabbit
anti-GFPantibodywaspurchasedfromClontech
(Palo Alto, CA). Anti-calnexin monoclonal anti-
body was purchased from StressGen (Victoria,
BC). Anti-human CRM1 antiserum was kindly
provided by Minoru Yoshida [Kudo et al., 1998].
Cell Culture
Human epithelial carcinoma cell (HeLa),
COS-7(greenmonkeykidneycell),andNIH3T3
(mouse embryonic fibroblast) were cultured in
Dulbecco’s modified Eagles’s medium (DMEM)
containingwith10%fetalbovineserum(FBS)at
378C. For transient transfection, cells grown on
glass coverslips in 35-mm dishes were trans-
fected with 1-mg plasmid DNA using GenePOR-
TERTMtransfection reagent (Gene Therapy
Systems, San Diego, CA).
Molecular Cloning
Human eIF-5A cDNA was cloned into Hin-
dIII-cut pEGFP-C1 (Clontech) to generate
pGFP-5A. GFP-pyruvate kinase (GPK) tag was
constructedasfollows:thecDNAcorresponding
to chicken muscle PK from codon 17 to the
carboxyl terminal end was PCR-amplified from
the pMyc-PK plasmid [Siomi and Dreyfuss,
1995]. The amplified PK cDNA fragment was
then cloned between SspBI and XhoI sites
of pEGFP-C1 and pGFP-5A to generate pGPK
and pGPK-5A, respectively. For the construc-
tion of GFP- or GPK-tagged eIF-5A deletion
mutants, PCR-generated cDNAs correspond-
ing to amino acid sequences 1–83, 84–154,
and 41–120 of human eIF-5A were substituted
for the full-length eIF-5A cDNA in pGPK-5A
to create pGPK-5A(1–83), pGPK-5A(84–154),
and pGPK-5A(41–120), respectively. To con-
struct pGPK-5A (K50R), the lysine residue at
codon 50inthehumaneIF-5Agene ofpGPK-5A
was mutated to arginine using TransformerTM
Site-Directed Mutagenesis Kit (Clontech). The
pGPK-NLS was generated by replacing the
HindIII-Myc-PK-EcoRI fragment in pMyc-PK-
NLS [Michael et al., 1995] with NheI-GPK-
EcoRI fragment obtained from pGPK. The Eco
RI ends were ligated first and then the HindIII
and NheI ends were blunted by T4 DNA poly-
merase followed with ligation. To create pGPK-
5A-NLS, the primers 50-CCAAAAGCTTAATG-
GCAGATGACTTGGACTTCGAG-30(sense)and
50-TTAAAGCTTCCTTTTGCCATGGCCTTGA-
TTGC-30(antisense) were used to amplify the
human eIF-5A cDNA by PCR. After digestion,
the eIF-5A cDNA was cloned into the HindIII
site of pGPK-NLS to generate pGPK-5A-NLS.
To generate pGPK-Rev-NLS and pGPK-M10-
NLS, the Rev and Rev M10 cDNA fragments
were amplified from pCsRevsg25GFP and pCs
RevM10BLsg25 [Afonina et al., 1998]. After di-
gestion,thecorrespondingfragmentwascloned
into the EcoRI site of pGPK-NLS to gene-
ratepGPK-Rev-NLS
respectively.
and pGPK-M10-NLS,
Immunostaining and Fluorescence Microscopy
Cells were fixed either with methanol or 4%
paraformaldehyde in PHEM buffer (60 mM
PIPES, pH 6.9, 25 mM HEPES, 10 mM EGTA,
2mMMgCl2)for15minandpermeabilizedwith
0.5% Triton X-100 for 10 min. For detergent
extraction prior to fixation, the cells were
Subcellular Localization of eIF-5A591

Page 3

incubated first with 0.2% of Triton X-100 in
PHEM buffer for 1 min at room temperature
before fixation. The cells were then incubat-
ed with the blocking solution containing 10%
bovine serum albumin (BSA), 0.1% Tween-20,
and 0.02% sodium azide in PHEM buffer for 30
min. The cells were probed with anti-eIF-5A
(1:20 dilution), anti-calnexin (1:200 dilution),
anti-CRM1 (1:200 dilution), or anti-GFP (1:100
dilution) antibodies for 1 h at room tempera-
ture, followed by washing with the blocking
solution. The treated cells were incubated with
appropriate secondary antibodies (1:100 dilu-
tion) coupled with fluorescein isothiocyanate
(FITC) or Cy3 fluorophores for 30 min. The
cells were then washed in PHEM buffer exten-
sively,mountedinPHEMbuffercontaining10%
glycerol, 2.5% 1,4-diazabicyclo-[2.2.2] octane
(DABCO), and analyzed with an Olympus
BH-2 fluorescence microscope equipped with a
digital camera (MDS120, Kodak). To observe
GFP fusion proteins in living cells, cells were
seeded onto glass coverslips and maintained
at 378C with full humidity and 5% CO2. The
coverslip was removed from the incubator and
mounted upside-down onto a hanging-drop
slide containing DMEM about 24–48 h post-
transfection for visualization. The cells were
observedimmediatelyafterthemountingunder
the fluorescence microscope at 258C.
Interspecies Heterokaryon Assay
The interspecies heterokaryon assay was
performed as described [Pinol-Roma and Drey-
fuss, 1992]. Briefly, 24 h after transfection,
the transfected HeLa cells were seeded with a
fivefold excess of untransfected mouse NIH3T3
cells. The following day, cells on the coverslip
were washed with DMEM and then inverted
ontoadropofpolyethyleneglycol3350(50%w/w
in Earl’s balanced salt solution) for 2 min. The
coverslip was rinsed three times with DMEM
and returned to heterokaryon growth medium
(DMEM, 10% FBS, 20% water, 100 mg/ml cyclo-
heximide, and 1% penicillin and streptomycin)
for 3 h before fixation for fluorescence micro-
scopy. To block new protein synthesis, the cul-
ture was incubated for 3 h in the presence of
50 mg/ml cycloheximide and 30 min in the
presence of 100 mg/ml cycloheximide before
fusion. All solutions were pre-warmed to 378C.
In order to distinguish the 3T3 mouse nuclei
from the HeLa nuclei, cells were stained with
propidiumiodide(PI,0.5mg/ml)andobservedby
the rhodamine filter set. Mouse nuclei display a
speckled pattern after PI staining. In addition,
they are less transparent than human nuclei
after PI staining.
Western Blot Analysis
Transfected COS-7 cells growth in 35-mm
dishesweretrypsinized,washedinPBS(50mM
sodium phosphate, pH 7.5, and 150 mM NaCl),
and lyzed in 500 ml of lysis buffer containing
50mMHepes,pH7.5,150mMNaCl,1%NP-40,
5 mM EDTA, and a panel of protease inhibitors
(Roche Molecular Biochemicals, Indianapolis,
IN). Cell lysates were clarified by centrifuga-
tion at 14,000g for 10 min. The supernatant
was subjected to SDS–PAGE and Western
blot analysis using antibodies against eIF-5A
or GFP (1:1,000 dilution). The proteins were
detected by enhanced chemical luminescence
method (Amersham Pharmacia Biotech, Piscat-
away, NJ).
RESULTS
Indirect Immunofluorescent Staining
of Endogenous eIF-5A
The immunofluorescent staining of eIF-5A in
cells was performed under two fixation condi-
tions,onewithformaldehydeandtheotherwith
methanol. As shown in Figure 1A, the fluores-
cent signal in cells fixed with formaldehyde
displayed a whole-cell distribution pattern and
appeared to be more prominent at the site that
resemblesthestructuretermedannulatelamel-
lae.Incontrast,Figure1Cshowsthatthefluores-
cent signal was apparent only in the cytoplasm
of cells fixed with methanol, although with an
intense staining near the perinuclear region.
As a control, Figure 1E shows that no signal
was observed when the pre-absorbed anti-
body was used for staining. The corresponding
phase-contrast images of cells were included
to indicate the integrity of cells after fixation
(Fig. 1B,D,F). This result suggests that the
controversy about the eIF-5Alocalization in the
literature [Ruhl et al., 1993; Shi et al., 1996]
could be due to different fixation protocols used.
Although, formaldehydeand methanol areboth
commonly used in immunofluorescent stain-
ing,theycanproducedifferentstainingpatterns
in some cases. For example, the nuclear pre-
sence of plakophilin [Mertens et al., 1996] and
feline immunodeficiency virus, Vif [Chatterji
et al., 2000] can be demonstrated only by
592Li-En Jao and Yu Chen

Page 4

formaldehyde fixation, but not by methanol
fixation.Formaldehydefixationisamoregentle
methodinpreservingcellularstructurebecause
of the formation of cross-linked networks of bio-
polymers [Solomon and Varshavsky, 1985]. In
contrast,methanolfixationtendstocollapsethe
nuclear structure, which may randomize the
availability of the epitope to the antibody or
lead to leaking of certain nuclear proteins
[Schimenti and Jacobberger, 1992; Chatterji
et al., 2000].
Direct Visualization of GFP-Tagged eIF-5A
in Living Cells
Tofurtherconfirm thewhole-cell distribution
pattern of eIF-5A that we observed in the
formaldehyde-fixed cells, we took advantage of
the approach of GFP-tagging, which allows
direct observation of tagged protein in vivo.
Figure 2A shows the expression level of green
fluorescent protein tagged eIF-5A (GFP-eIF-
5A)(Mr¼45kDa)inthetransientlytransfected
COS-7 cells as compared to that of endogenous
eIF-5A.Usingyeastplasmidshufflingassay,we
have shown that GFP-eIF-5A fusion protein is
fully functional (Chatterjee and Chen, unpub-
lisheddata).Figure2BshowsthattheGFP-eIF-
5A in the living cells exhibited a whole-cell
distribution pattern. We then compared the
pattern with that obtained by immunostaining
using anti-GFP antibody. While GFP-eIF-5A
displayed a whole-cell distribution in the for-
maldehyde-fixed cells (Fig. 2C), it was apparent
onlyinthecytoplasminthemethanol-fixedcells
(Fig. 2D). This result clearly indicated that
formaldehyde fixation produced a staining pat-
tern most closely resembles that obtained by in
vivo GFP-tagging. We, therefore, concluded
that the whole-cell distribution pattern prob-
ably represents more closely to the true locali-
zation of eIF-5A in living cells.
Does eIF-5A Co-Localize With Calnexin
or CRM1?
Earlier studies have suggested that eIF-5A
co-localizes either with calnexin, a resident
protein of endoplasmic reticulum [Shi et al.,
1996] or CRM1, a general nuclear export recep-
tor [Rosorius et al., 1999]. Since protein locali-
zation revealed by immunofluorescent staining
can be influenced by fixation protocols, we de-
cided to re-examine the possible co-localization
Fig. 1.
munofluorescence with two different fixation protocols. Immu-
nofluorescence (left-hand panel) and corresponding phase-
contrast images (right-hand panel) of COS-7 cells fixed with
4% paraformaldehyde (A and B; E and F) or methanol (C and D)
are shown. Fixed cells were treated with the purified anti-eIF-5A
antibody (1:20). FITC-conjugated rabbit anti-chicken immuno-
globulin (IgG) was used as the secondary antibody to visualize
the protein. As a control, fixed cells were also probed with the
anti-eIF-5A antibody that was pre-absorbed with recombinant
human eIF-5A proteins (E and F). Bar: 25 mm. [Color figure can
be viewed in the online issue, which is available at www.
interscience.wiley.com.]
Subcellular localization of eIF-5A by indirect im-
Fig. 2.
ently transfected COS-7 cells. (A) Western blot analysis of GFP-
eIF-5A in the transfected cells. (B) Direct visualization of GFP-
eIF-5A. (C) Indirect immunofluorescent staining of GFP-eIF-5A
in the transiently transfected cells (fixed with 4% paraformalde-
hyde). (D) Indirect immunofluorescent staining of the same cells
in methanol. Bar: 25 mm. [Color figure can be viewed in the on-
line issue, which is available at www.interscience.wiley.com.]
Expression and localization of GFP-eIF-5A in transi-
Subcellular Localization of eIF-5A593

Page 5

of eIF-5A with calnexin and CRM1. The top
panels of Figure3show that both formaldehyde
and methanol fixation revealed the same cyto-
plasmic localization of calnexin with the char-
acteristic punctate staining pattern (Fig. 3B,F).
This staining pattern of calnexin was clearly
different from that of eIF-5A (Fig. 3A,E). There
appeared to be certain overlap of calnexin and
eIF-5A in the merged images (Fig. 3C,G). This
partial overlap is likely to be fortuitous due to
the whole-cell distribution nature of eIF-5A.
However, other biochemical approach is needed
to determine whether there is any functional
association of eIF-5A with ER or other struc-
tures such as microtubule organization center
(MTOC).
The bottom panels of Figure 3 show that
CRM1 was found to be in the nucleus of both
the formaldehyde- and methanol-fixed cells,
(Fig.3J,N).Doubleimmunofluorescentstaining
showsasubstantialoverlapofeIF-5AandCRM1
signals inthenuclear areaoftheformaldehyde-
fixed cells (Fig. 3K), and very little overlap was
observed in the methanol-fixed cells (Fig. 3O).
In the formaldehyde-fixed cells, the signals of
both eIF-5A and CRM1 were prominent at
the site corresponding to annulate lamellae.
Although,thefunctionalroleofannulatelamel-
lae is unclear [Meier et al., 1995], the presence
of CRM1 in annulate lamellae has been demon-
strated [Fornerod et al., 1997]. These results
suggest that eIF-5A does not co-localize with
Fig. 3.
and CRM1 (bottom panels) by double immunofluorescence
microscopy. COS-7 cells were fixed with formaldehyde or
methanol and then stained with anti-eIF-5A and anti-calnexin
antibodies (top panel) or anti-eIF-5A and anti-CRM1 anti-
bodies (bottom panel). FITC-fluorescence images revealed the
distribution of eIF-5A (A, E, I, and M). Cy3-fluorescence images
Staining of eIF-5A and calnexin (top panels) or eIF-5A
revealed the distribution of calnexin (B and F) or CRM1 (J and
N). Double immunofluorescence images revealed the merged
images of eIF-5A and calnexin (C and G) or eIF-5A and CRM1
(K and O). Phase contrast images of the corresponding fixed
cells are shown in (D), (H), (L), and (P). Bar: 25 mm. [Color figure
can be viewed in the online issue, which is available at www.
interscience.wiley.com.]
594Li-En Jao and Yu Chen

Page 6

calnexin, but may co-localize with CRM1 in
the nucleus.
How Does eIF-5A Gain Nuclear Entry?
The existence of eIF-5A in the nucleus poses
thequestiononhoweIF-5Againsnuclearentry.
Because of the relatively small size, either
eIF-5A or GFP-eIF-5A can enter the nucleus
through passive diffusion. Since globular pro-
teinswithsizelargerthan60kDacannotdiffuse
across the nuclear pore barrier [Ohno et al.,
1998], the fusion of PK moiety to a small
polypeptide has been used to minimize the con-
cern of passive diffusion [Siomi and Dreyfuss,
1995]. Thus, to distinguish the receptor-medi-
ated import from passive diffusion, we have
constructed GPK-tagged eIF-5A chimera pro-
teins and examined their localization in intact
cells. Figure 4A shows that the expression
level of GPK-eIF-5A in the log phase cells as
compared to that of the endogenous eIF-5A.
Figure 4B shows that GPK-eIF-5A, with a size
of about 98 kDa, was localized exclusively in
the cytoplasm in the intact cells. This result
suggests stronglythat eIF-5A may gain nuclear
entry via passive diffusion.
Does eIF-5A Contain Nuclear Localization
Signal (NLS)?
If indeed eIF-5A enters the nucleus via dif-
fusion process, it would imply that eIF-5A does
not contain a functional NLS. Alternatively,
the NLS may be dormant due to masking and
thus non-functional. To differentiate these pos-
sibilities, we have constructed chimeras with
GPKtagfusedtotheN-terminaldomain(amino
acids 1–83), C-terminal domain (amino acids
84–154), and the central region (amino acids
41–120) of eIF-5A. Figure 5A shows the expres-
sionlevelsofthesedifferentchimeraproteinsin
thetransfectedcells.Although,GPK-eIF-5A(1–
83) protein cannot be recognized by our anti-
eIF-5A antibody, its expression was confirmed
by Western blot using anti-GFP antibody
(Fig. 5A). Although all three chimera proteins
were expressed, none of them was detected in
the nucleus, suggesting a lack of NLS in any
of these regions (Fig. 5B–D). Furthermore, we
found that mutation of hypusine site did not
affect the subcellular distribution of GPK-eIF-
5A (Fig. 5E), suggesting that the hypusine
residue is not involved in the nuclear import.
We do not think that GPK tag itself could affect
the import activity of eIF-5A, since addition
of a small NLS segment derived from hnRNP
K to the GPK-eIF-5A led to nuclear import
(see below). Collectively, these results suggests
that receptor-mediated nuclear import acti-
vity of eIF-5A was either absent or very
Fig. 4.
ently transfected COS-7 cells. Visualization was carried out at
room temperature. (A) Western blot analysis of GPK-eIF-5A in
the transfected cells as detected by anti-eIF-5A antibody. (B)
Direct visualization of GPK-eIF-5A in COS-7 cells. Bar: 25 mm.
[Color figure can be viewed in the online issue, which is
available at www.interscience.wiley.com.]
Expression and localization of GPK-eIF-5A in transi-
Fig. 5.
eIF-5A domain chimera proteins in living cells. (A) Western blot
analysis of various GPK-eIF-5A fusion proteins using anti-eIF-5A
or anti-GFP antibody. Note that GPK-eIF-5A(1–83) could not be
detected by the anti-eIF-5A antibody, since the epitopes of this
antibody are located to the C-terminal portion of eIF-5A.
However, it can be detected by anti-GFP antibody. (B) Direct
visualization of GPK-eIF-5A (aa l–83). (C) Direct visualization
of GPK-eIF-5A(aa 41–120). (D) Directvisualization of GPK-eIF-
5A (aa 84–154). (E) Direct visualization of GPK-eIF-5A (KSOR).
[Color figure can be viewed in the online issue, which is
available at www.interscience.wiley.com.]
Expression and subcellular localization of GPK-tagged
Subcellular Localization of eIF-5A 595

Page 7

weak and that the nuclear localization of endo-
genous eIF-5A was most likely due to passive
diffusion.
Nuclear Export Activity of eIF-5A
The human eIF-5A contains a leucine-rich
region at the C-terminal domain (amino acids
89–102), with a sequence similar to the nuclear
exportsignal(NES)inHIV-1RevandInfluenza
A NS1 [Liu et al., 1997]. Thus, it is still possible
that eIF-5A may contain a functional NES and
thus can engage nucleocytoplasmic trafficking.
To determine whether this is the case, we per-
formed an interspecies heterokaryon assay.
We first used Rev and Rev mutant M10, whose
NES is mutated, to demonstrate the feasibility
of the assay. HeLa cells were first transfected
withanexpressionplasmidencoding GPK-Rev-
NLS fusion protein (i.e., GPK-Rev fused at its
carboxylterminustoaclassicNLSofhnRNP K)
orGPK-M10-NLS.The transfected human cells
were then fused with mouse NIH 3T3 cells to
form heterokaryons. The heterokaryons were
fixedandthenucleocytoplasmictransportprop-
ertiesoftheGPKfusionproteinswereexamined
under the fluorescence microscope. Figure 6
showsthatGPK-Rev-NLSshuttledbetweenthe
heterologous nuclei (Fig. 6A,B), whereas the
NES-mutated GPK-M10-NLS protein was re-
tained in the human nuclei (Fig. 6D,E). Under
thesamecondition,theGPK-eIF5A-NLSfusion
protein was retained in the human nuclei and
did not exhibit any nuclear export (Fig. 6G,H).
This result suggested that eIF-5A does not
contain a functional NES for active nucleocyto-
plasmic shuttling.
Fig. 6.
NLS, GPK-M10-NLS, or GPK-5A-NLS and then fused with mouse 3T3 cells as described under Materials
and Methods. (A), (D), and (G) are the intrinsic GFP signals of GPK-Rev-NLS, GPK-M10-NLS, and GPK-5A-
NLS, respectively. (B), (E), and (H) are the PI staining of the same cells in (A), (D), and (G), respectively. (C),
(F), and (I) are the phase-contrast images in (A), (D), and (G), respectively. Arrows denote the mouse nuclei.
Bar: 10 mm. [Color figure can be viewed in the online issue, which is available at www.interscience.
wiley.com.]
Heterokaryon analysis of the nuclear export of eIF-5A. HeLa cells were transfected with GPK-Rev-
596 Li-En Jao and Yu Chen

Page 8

Nuclear eIF-5A is Resistant to
Detergent Extraction
Treatment of cells with non-ionic detergent
beforefixationhasbeenemployedtodistinguish
two populations of antigens, one soluble and
the other presumably associated with certain
cytoskeletal structures [Bravo and MacDonald-
Bravo, 1987]. We have adopted this method to
determine whether the cytoplasmic and the
nuclear presence of eIF-5A can also be distin-
guished by Triton X-100 extraction. Figure 7
showsthatwhileTritonX-100removedtheeIF-
5A signal in the cytoplasm, the eIF-5A signal in
nucleus remained strong (Fig. 7A). This is to be
contrasted with GFP, which also exhibited a
whole-cell distribution pattern after transfec-
tion (data not shown). The GFP signal, both
in the nucleus and cytoplasm, was completely
eliminated by Triton X-100 treatment (Fig. 7C).
These results suggest that the cytoplasmic eIF-
5A may be soluble, whereas the nuclear eIF-5A
may be associated with certain structural
components in the nucleus.
Subcellular Distribution of eIF-5A Under
Different Conditions
Changes in the subcellular distribution of a
target protein under different pharmacologi-
cal or physiological conditions can yield clues
onproteinfunction.Wehaveexaminedwhether
the eIF-5A localization may be perturbed
by treatments, such as heat shock, energy
depletion, cytoskeleton disruption, polyamine
depletion, and inhibition of protein synthesis,
transcription, or CRM1-dependent nuclear
export. The results were summarized in Table I,
which indicated that none of the treatments
significantlyalteredthesubcellularlocalization
of endogenous eIF-5A or transfected GFP-eIF-
5A. Thus, the global distribution of eIF-5A
inside the cell appeared to be recalcitrant to
many conditions that severely alter the meta-
bolic and physiological state. Nevertheless,
we still cannot exclude the possibility that our
detection method based on immunostaining or
GFP-tagging may not be sensitive enough to
detect a small fraction of eIF-5A pool that
undergoes changes in distribution.
DISCUSSION
In the present study, we have employed both
the in vitro and in vivo approach and demon-
strated that eIF-5A is present in the cytoplasm
andnucleus(Figs.1and2).TheuseofGFP-and
GPK-tagging for direct visualization of marked
eIF-5A in living cells provided the in vivo data
suggesting that eIF-5A enters the nucleus via
passive diffusion (Fig. 2 vs. 4). The absence of
detectable NLS activity in various GPK-tagged
chimeras (Fig. 5) substantiates the notion that
passive diffusion is the mechanism for eIF-5A
to gain nuclear entry. Furthermore, heterokar-
yon assay results indicate that eIF-5A does not
undergo any appreciable nucleocytoplasmic
shuttling (Fig. 6).
The whole-cell distribution pattern of eIF-5A
is consistent with the notion that eIF-5A may
functionasanRNA-bindingprotein.Itisknown
that many RNA-binding proteins are distribu-
ted in the cytoplasm and nucleus [Shen et al.,
2000]. The lack of any appreciable nuclear
signal in the methanol-fixed cells could be attri-
buted to either the randomization of the avail-
able epitope or the leaking. Similar findings
have been reported for other nuclear proteins
after methanol fixation [Schimenti and Jacob-
berger, 1992; Chatterji et al., 2000].
ThateIF-5AmayfunctionasanRNA-binding
protein is supported by the following evidence:
(i) motif analysis revealed that it has a bi-
modular structure similar to RNA-binding pro-
teins such as Rev and NS-1 [Liu et al., 1997];
(ii) two X-ray diffraction studies of the archae
Fig. 7.
patterns of eIF-5A and GFP in the COS-7 cells. COS-7 cells were
treated with 0.2% Triton X-100 before fixation as described
under Materials and Methods. (A) Immunofluorescent staining
of the Triton-extracted COS-7 cells using anti-eIF-5A antibody.
(B) Phase-contrast image in (A). (C) Immunofluorescent staining
of the Triton-extracted GFP-transfected COS-7 cells using anti-
GFP antibody. (D) Phase-contrast image in (C). Bar: 25 mm.
[Color figure can be viewed in the online issue, which is
available at www.interscience.wiley.com.]
Effect of detergent extraction on the immunostaining
Subcellular Localization of eIF-5A597

Page 9

eIF-5A precursor show that it is composed of
two domains, one containing many positively
chargedresiduesandtheother issimilartocold
shockdomainofCspA,linkedbyaflexiblehinge
[Kim et al., 1998; Peat et al., 1998]. Both the
positively-charged hypusine and the cold shock
domain have the potential to interact with
nucleic acids; (iii) we have shown that eIF-5A
iscapableofbindingtoRNAinvitroandthatthe
bindingrequireseitherdeoxyhypusineorhypu-
sine[Liuetal.,1997];and(iv)morerecently,we
havedemonstratedbySELEXmethodthat eIF-
5A recognizes RNA in a sequence-dependent
manner [Xu and Chen, 2001]. The notion that
eIF-5A may function as an RNA binding pro-
tein is also consistent with the finding that
an expression of TIF51A gene, which encodes
eIF-5A, complements the impaired mRNA
decay phenotypes of a yeast mutant [Zuk and
Jacobson, 1998].
Although, eIF-5A is termed as initiation
factor, the evidence is still weak. In fact, it fails
to show any significant effect on the translation
withphysiologicalmRNAinvitro[Kemperetal.,
1976; Benne and Hershey, 1978]. Further, de-
pletion of eIF-5A in a conditional yeast mutant
onlyinhibitstheproteinsynthesisby30%[Kang
and Hershey, 1994]. It is still possible that
eIF-5A may be required for translation of
selective mRNAs [Kang and Hershey, 1994].
If this is the case, eIF-5A may interact with a
small class of mRNA in vivo. In this regard, it
is interesting to note that the consensus se-
quence identified by SELEX is present in over
400 human EST sequences [Xu and Chen,
2001]. It is quite possible that some of these
sequenceswillserveasthephysiologicaltargets
of eIF-5A.
Although, eIF-5A has been suggested to
target HIV-Rev and to mediate its nuclear
export [Ruhl et al., 1993], no direct binding
between eIF-5A and Rev could be demon-
strated so far [Henderson and Percipalle,
1997;Lipowskyetal.,2000].Ontheother hand,
two recent studies have demonstrated the
interaction of eIF-5A with the nuclear export
receptor, CRM1 [Rosorius et al., 1999] and
exportin4 [Lipowsky et al., 2000]. It remains to
be investigated, however, whether these inter-
actions may be related to the binding of eIF-5A
to RNA. Nevertheless, both findings are con-
sistentwiththenuclearlocalizationofeIF-5Aas
shown in the present study.
The presence of eIF-5A in the nucleus poses
the following questions: (i) what is the physio-
logical function of the nuclear eIF-5A? (ii) is the
functionofnucleareIF-5Adifferentfromthatof
the cytoplasmic eIF-5A? and (iii) how the level
of nuclear eIF-5A is regulated? The possible
interactionofeIF-5Awithnuclearexportersuch
as CRM1 [Rosorius et al., 1999] or exportin4
[Lipowskyetal.,2000]wouldimplicatearolefor
eIF-5A in nuclear transport. However, despite
the high binding affinity between exportin4
TABLE I. Effects of Various Pharmacological and Physical Treatments on the Subcellular
Distribution of eIF-5A in COS-7 Cells
Effect of treatmentsConditions of treatmentsa
Localizationb
Inhibition of polyamine biosynthesis5 mM a-difluoromethylornithine (DFMO)
5 mM methylglyoxal bis(guanyl-hydrazone) (MGBG)
40 mM N1-guanyl-1,7-diaminoheptane (GC7)
Incubated for 48 h at 378C
10 mM deoxyglucose
20 mM CCCP
10 mM sodium azide
Incubated in dialyzed serum-containing medium for 3 h at 48C
100 nM leptomycin B for 2 h at 378C
200 nM leptomycin B for 9 h at 378C
50 mg/ml puromycin for 16 h at 378C
100 mg/ml cycloheximide for 4 h at 378C
5 mg/ml actinomycin D for 4 h at 378C
5 mg/ml cytochalasin B
5 mg/ml cytochalasin D
5 mg/ml colchicine
Incubated for 2 h at 378C
Incubated at 448C for 3 h
N, C
Depletion of energy source N, C
Inhibition of CRM1-dependent nuclear export N, C
Inhibition of protein synthesisN, C
Inhibition of RNA synthesis
Disruption of cytoskeleton
N, C
N, C
Heat stress N, C
aCells at logarithmic growth were treated under various conditions as listed and then processed for determining the localization of
eIF-5A.
bSubcellular distribution of eIF-5A was determined by either in vitro indirect immunofluorescent staining in all cases or direct
visualization of GFP-tagged eIF-5A in the cases of inhibition of nuclear export and protein synthesis. N, present in the nucleus;
C, present in the cytoplasm.
598Li-En Jao and Yu Chen

Page 10

and eIF-5A, we could not detect any appreci-
able nucleocytoplasmic shuttling of eIF-5A by
the heterokaryon assay (Fig. 6C). In contrast,
we have no problem to demonstrate the nuclear
export of Rev in the same assay (Fig. 6A,B).
The nuclear export of eIF-5A was previously
demonstrated in vitro by either microinjection
or cell permeabilization using exogenously
added eIF-5A [Rosorius et al., 1999; Lipowsky
et al., 2000]. Since both methods could result
in very high concentration of eIF-5A in the nu-
cleus, we suspect that in our study, the nuclear
level of GFP-eIF-5A may not be high enough
to trigger receptor-mediated export. Alterna-
tively, it is possible that the degree of receptor-
mediated nuclear export, if present, is too small
to be detected by the assay method. Although,
the specificity and high affinity of eIF-5A-
exportin4interactionsuggeststhatthisbinding
has to be functional, exportin4 is absent in the
yeastSaccahromycescerevisiae[Lipowskyetal.,
2000]. However, we noted that the concentra-
tion of eIF-5A in HeLa cell is about 10-fold
higher than that in the yeast (data not shown).
In light of this, it will be of interest to know
whether exportin4 is evolved to endow eIF-5A
with expanded functions or simply to prevent
any untimely accumulation of eIF-5A in the
nucleus. Our finding that nuclear eIF-5A is
resistant to detergent extraction (Fig. 7) sug-
gests that it may interact with some structural
components in the nucleus. This would also
suggest that the chemical potential of the nu-
clear eIF-5A could be significantly lower than
thatofthecytoplasmiceIF-5A.Ifthisisthecase,
it will make sense to have some mechanism in
place to prevent the accumulation of nuclear
eIF-5A.
In conclusion, the whole-cell distribution pat-
tern of eIF-5A is consistent with the notion
that eIF-5A functions as an RNA binding pro-
tein [Liu et al., 1997; Xu and Chen, 2001]. In
addition, our study revealed: (i) eIF-5A enters
nucleus via passive diffusion; (ii) there is an
apparent lack of detectable nucleocytoplasmic
shuttling; (iii) the nuclear eIF-5A is resistant
to non-ionic detergent extraction; and (iv) the
subcellular distribution of eIF-5A is recalci-
trant to various pharmacological treatments
(Table I). Future research on the function of
eIF-5A should accommodate these findings.
However, the nuclear presence of eIF-5A will
have to be reconciled with the apparent di-
lemma that archaea have no nuclei, although
clearly eIF-5A is essential for the survival of
archaea [Jansson et al., 2000].
ACKNOWLEDGMENTS
We are grateful to Dr. Gideon Dreyfuss,
University of Pennsylvania for pcDNA3-myc-
PK and Myc-PK-NLS; Dr. George N. Pavlakis,
NCI-Frederick Cancer Research and Develop-
ment Center for pCsRevsg25GFP and pCs
RevM10BLsg25; Dr. Barbara Wolff, Norvatis
Research Institute, for leptomycin B. The work
was supported in part by United States Public
Health Services Grant RO1 CA49695 awarded
by the National Cancer Institute, NIH.
REFERENCES
Afonina E, Stauber R, Pavlakis GN. 1998. The human
poly(A)-binding protein 1 shuttles between the nucleus
and the cytoplasm. J Biol Chem 273:13015–13021.
Benne R, Hershey JW. 1978. The mechanism of action of
protein synthesis initiation factors from rabbit reticulo-
cytes. J Biol Chem 253:3078–3087.
Bravo R, MacDonald-Bravo H. 1987. Existence of two popu-
lations of cyclin/proliferating cell nuclear antigen during
the cell cycle: Association with DNA replication sites.
J Cell Biol 1549–1554.
Chatterji U, Grant CK, Elder JH. 2000. Feline immunode-
ficiency virus Vif localizes to the nucleus. J Virol 74:
2533–2540.
Chen ZP, Chen KY. 1997a. Dramatic attenuation of
hypusine formation on eukaryotic initiation factor 5A
during senescence of IMR-90 human diploid fibroblasts.
J Cell Physiol 170:248–254.
Chen ZP, Chen KY. 1997b. Marked elevation of hypusine
formation activity on eukaryotic initiation factor 5A in v-
HA-RAS transformed mouse NIH3T3 cells. Cancer Lett
115:235–241.
Chen KY, Jao DL. 1999. Chemistry of hypusine formation
on eukaryotic initiation factor 5A in biological systems.
J Chin Chem Soc 46:727–734.
Chen KY, Liu AY. 1997. Biochemistry and function of
hypusine formation on eukaryotic initiation factor 5A.
Biol Signals 6:105–109.
Chen ZP, Yan YP, Ding QJ, Knapp S, Potenza JA, Schugar
HJ, Chen KY. 1996. Effects of inhibitors of deoxyhypusine
synthase on the differentiation of mouse neuroblastoma
and erythroleukemia cells. Cancer Lett 105:233–239.
Fornerod M, van Deursen J, van Baal S, Reynolds A, Davis
D, Murti KG, Fransen J, Grosveld G. 1997. The human
homologue of yeast CRM1 is in a dynamic subcomplex
with CAN/Nup214 and a novel nuclear pore component
Nup88. EMBO J 16:807–816.
Henderson BR, Percipalle P. 1997. Interactions between
HIV Rev and nuclear import and export factors: The Rev
nuclear localization signal mediates specific binding to
human importin-b. J Mol Biol 274:693–707.
Jakus J, Wolff EC, Park MH, Folk JE. 1993. Features of
the spermidine-binding site of deoxyhypusine synthase
as derived from inhibition studies. Effective inhibition by
bis- and mono-guanylated diamines and polyamines.
J Biol Chem 268:13151–13159.
Subcellular Localization of eIF-5A599

Page 11

Jansson BP, Malandrin L, Johansson HE. 2000. Cell cycle
arrest in archaea by the hypusination inhibitor N(1)-
guanyl-1,7-diaminoheptane. J Bacteriol 182:1158–1161.
Kang HA, Hershey JWB. 1994. Effect of initiation factor
eIF-5A depletion on proteinsynthesis and proliferationof
Saccharomyces cerevisiae. J Biol Chem 269:3934–3940.
Katahira J, Ishizaki T, Sakai H, Adachi A, Yamamoto K,
Shida H. 1995. Effects of translation initiation factor eIF-
5A on the functioning of human T-cell leukemia virus
type I Rex and human immunodeficiency virus Rev
inhibited trans dominantly by a Rex mutant deficient in
RNA binding. J Virol 69:3125–3133.
Kemper WM, Berry KW, Merrick WC. 1976. Purification
and properties of rabbit reticulocyte protein synthesis
initiation factors M2Ba and M2Bb. J Biol Chem 251:
5551–5557.
Kim KK, Hung L-W, Yokota H, Kim R, Kim S-H. 1998.
Crystal structures of eukaryotic translation initiation
factor 5A from Methanococcus jannaschii at 1.8 A˚reso-
lution. Proc Natl Acad Sci USA 95:10419–10424.
Kudo N, Wolff B, Sekimoto T, Schreiner EP, Yoneda Y,
YanagidaM, HorinouchiS, Yoshida M. 1998.Leptomycin
B inhibition of signal-mediated nuclear export by direct
binding to CRM1. Exp Cell Res 242:540–547.
Lipowsky G, Bischoff FR, Schwarzmaier P, Kraft R, Kostka
S, Hartmann E, Kutay U, Gorlich D. 2000. Exportin 4:
A mediator of a novel nuclear export pathway in higher
eukaryotes. EMBO J 19:4362–4371.
Liu YP, NemeroffM, YanYP, Chen KY. 1997.Interactionof
eukaryotic initiation factor 5A with the human immuno-
deficiency virus type 1 Rev response element RNA and
U6 snRNA requires deoxyhypusine or hypusine modifi-
cation. Biol Signals 6:166–174.
Mattaj IW, Englmeier L. 1998. Nucleocytoplasmic trans-
port: The soluble phase. Annu Rev Biochem 67:265–306.
Meier E, Miller BR, Forbes DJ. 1995. Nuclear pore complex
assembly studied with a biochemical assay for annulate
lamellae formation. J Cell Biol 129:1459–1472.
Mertens C, Kuhn C, Franke WW. 1996. Plakophilins 2a
and 2b: Constitutive proteins of dual location in the
karyoplasm and the desmosomal plaque. J Cell Biol
135:1009–1025.
Michael WM, Choi M, Dreyfuss G. 1995. A nuclear export
signal in hnRNP A1: A signal-mediated, temperature-
dependent nuclear protein export pathway. Cell 83:415–
422.
Ohno M, Fornerod M, Mattaj IW. 1998. Nucleocytoplasmic
transport: The last 200 nanometers. Cell 92:327–336.
Park MH, Wolff EC, Folk JE. 1993. Hypusine: Its post-
translational formation in eukaryotic initiation factor 5A
and its potential role in cellular regulation. BioFactors
4:95–104.
Park MH, Wolff EC, Lee YB, Folk JE. 1994. Antiprolifera-
tive effects of inhibitors of Deoxyhypusine synthase.
J Biol Chem 269:27827–27832.
Park MH, Lee YB, Joe YA. 1997. Hypusine is essential for
eukaryotic cell proliferation. Biol Signals 6:115–123.
Peat ST, Newman J, Waldo GS, Berendzen J, Terwiliger
TC. 1998. Structure of translation initiation factor 5A
from Pyrobaculum aerophilum at 1.75 A˚resolution.
Structure 6:1207–1213.
Pinol-Roma S, Dreyfuss G. 1992. Shuttling of pre-mRNA
binding proteins between nucleus and cytoplasm. Nature
355:730–732.
Rosorius O, Reichart B, Kratzer F, Heger P, Dabauvalle
M-C, Hauber J. 1999. Nuclear pore localization and nu-
cleocytoplasmic transport of eIF-5A: Evidence for direct
interaction with the export receptor CRM1. J Cell Sci
112:2369–2380.
Ruhl M, Himmelspach M, Bahr GM, Hammerschmid F,
Jaksche H, Wolff B, Aschauer H, Farrington GK, Probst
H, Bevec D, Hauber J. 1993. Eukaryotic initiation factor
5A is a cellular target of the human immunodeficiency
virus type 1 Rev activation domain mediating trans-
activation. J Cell Biol 123:1309–1320.
Sasaki K, Abid MR, Miyazaki M. 1996. Deoxyhypusine
synthase gene is essential for cell viability in the yeast
Saccharomyces cerevisiae. FEBS Lett 384:151–154.
Schimenti KJ, Jacobberger JW. 1992. Fixation of mammal-
ian cells for flow cytometric evaluation of DNA content
and nuclear immunofluorescence. Cytometry 13:48–59.
Schnier J, Schwelberger HG, Smit-McBride Z, Kang HA,
Hershey JWB. 1991. Translation initiation factor 5A and
its hypusine modification are essential for cell viability
in the yeast Saccharomyces cerevisiae. Mol Cell Biol
11:3105–3114.
Shen EC, Stage-Zimmermann T, Chui P, Silver PA. 2000.
The yeast mRNA-binding protein Np13p interacts with
the cap-binding complex. J Biol Chem 275:23718–23724.
Shi X-P, Yin K-C, Zimolo ZA, Stern AM, Waxman L. 1996.
The subcellular distribution of eukaryotic translation ini-
tiation factor, eIF-5A, in cultured cells. Exp Cell Res
225:348–356.
Siomi H, Dreyfuss G. 1995. A nuclear localization domain
in the hnRNP A1 protein. J Cell Biol 129:551–560.
Solomon MJ, Varshavsky A. 1985. Formaldehyde-mediated
DNA-protein crosslinking: A probe for in vivo chromatin
structures. Proc Natl Acad Sci USA 82:6470–6474.
Tome ME, Fiser SM, Payne CM, Gerner EW. 1997. Excess
putrescine accumulation inhibits the formation of modi-
fied eukaryotic initiation factor 5A (eIF-5A) and induces
apoptosis. Biochem J 328:847–854.
Tsien RY. 1998. The green fluorescent protein. Annu Rev
Biochem 67:509–544.
Xu A, Chen KY. 2001. Hypusine is required for a sequence-
specific interaction of eukaryotic initiation factor 5A with
post-SELEX RNA. J Biol Chem 276:2555–2561.
Zuk D, Jacobson A. 1998. A single amino acid substitution
in yeast eIF-5A results in mRNA stabilization. EMBO J
17:2914–2925.
600Li-En Jao and Yu Chen