ARL4, an ARF-like Protein That Is Developmentally Regulated and Localized to Nuclei and Nucleoli*

ADP-ribosylation factors (ARFs) are highly conserved ; 20-kDa guanine nucleotide-binding proteins that participate in both exocytic and endocytic vesicular trans-port pathways via mechanisms that are only partially understood. Although several ARF-like proteins (ARLs) are known, their biological functions remain unclear. To characterize its molecular properties, we cloned mouse and human ARL4 ( mARL4 and hARL4 ) cDNA. The appearance of mouse ARL4 mRNA during embryonic development coincided temporally with the sequential formation of somites and the establishment of brain compartmentation. Using ARL4-specific antibody for immunofluorescence microscopy, we observed that endogenous mARL4 in cultured Sertoli and neuroblastoma cells was mainly concentrated in nuclei. When expressed in COS7 cells, ARL4-T34N mutant, predicted to exist with GDP bound, was concentrated in nucleoli. Yeast two-hybrid screening and in vitro protein-interac-tion assays showed that hARL4 interacted with impor-tin- a through its C-terminal NLS region and that the interaction was not nucleotide-dependent. Like ARL2 and -3, 50 Dulbecco’s modified LipofectAMINE gently modified C -galactosidase the two-hybrid assay. five different recombinant importin- a in the C-terminal The in vitro interaction experiments using GTPase-defec-tive (ARL4-Q79L) and GTP-binding defective (ARL4-T34N) of ARL4, as as ARL4-dC, confirmed that interaction of hARL4 with importin- a is not GTP-dependent and does require its C-terminal NLS.

The Ras superfamily of ϳ20-kDa guanine nucleotide-binding proteins (or GTPases) contains more than 100 gene products that have been grouped into five subfamilies, i.e. the Ras, Rab, Rho, Ran, and ARF families. The ARF 1 (ADP-ribosylation factor) family, which is most different from the other groups, comprises at least six ARFs and six ARF-like (ARL) proteins (reviewed in Refs. 1 and 2). Although ARFs and ARLs are very similar in amino acid sequences, most ARLs, but not ARL1, apparently do not activate the cholera toxin ADP-ribosyltransferase. They differ also from ARFs in having demonstrable GTPase activity and different conditions that favor guanine nucleotide binding. Both ARFs and ARLs are widely distributed in eukaryotic organisms from yeast to human, consistent with evolutionary conservation of their biological functions.
ARFs play an important role in intracellular membrane trafficking, although there remains much to be learned. The biological functions of ARLs are still unclear, although some are expressed in a tissue-and/or differentiation-specific pattern (3)(4)(5)(6)(7). ARLl was localized in the Golgi complex of normal rat kidney cells (8) and Saccharomyces cerevisiae (9), consistent with a function in vesicular trafficking. Unlike the lethal phenotype of double null alleles of arf1 and arf2, however, knockout of the yeast ARL1 gene was not lethal (9). Expression of rat ARL4 was reported to be cell differentiation-dependent (4). Its role in adipocyte metabolism and sperm production was suggested (10) and nuclear localization of transiently expressed protein was demonstrated (11).
To obtain additional clues to its physiological role(s), we investigated the expression, subcellular localization, and biochemical properties of ARL4. As reported here, mouse ARL4 (mARL4), which is abundant in testis, is developmentally regulated during mouse embryogenesis. The mARL4 mRNA appears transiently, progressing in a rostro-caudal direction in day 8.5 to day 10.5 embryos, which coincides temporally with the appearance of somitomeres in the same locations. Endogenous mARL4 in Sertoli (TM4) and neuroblastoma (Neuro 2A) cells was mainly localized in nuclei. When expressed in COS7 cells, ARL4-T34N, a mutant predicted to be GDP-bound, was localized to nucleoli. This may be the first report of a small GTPase localized in nucleoli in a nucleotide-dependent manner. Data from yeast two-hybrid and in vitro protein interaction analyses revealed that hARL4 interacted with the NLSreceptor, importin-␣, through a bipartite nuclear localization signal (NLS) in its C-terminal region and that this interaction was not nucleotide-dependent. To our knowledge, no other small GTPase has been implicated in the regulation of somite development. The specific spatial and temporal expression of mouse ARL4 mRNA in the central nervous system during later embryonic stages suggests that ARL4 might also be involved in neuronogenesis or cortical histogenesis. Thus, ARL4 may have a physiological role(s) in vertebrate somite formation, and central nervous system differentiation, as well as in the early events of gametogenesis.

EXPERIMENTAL PROCEDURES
Isolation of Mouse and Human ARL4 cDNA-Mouse ARL4 cDNA was synthesized by polymerase chain reaction (PCR) from a mouse gt11 cDNA library. PCR-based cloning methods were used to obtain cDNA segments, from which a composite sequence of the full-length coding region was assembled (12). A probe composed of degenerate oligonucleotides (ARL-R1 and ARL-R2)( Table I) corresponding to part of the consensus sequences WDVGGQE and KLRPLWK in human and rat ARL4 was used to screen a mouse gt11 cDNA library (CLON-TECH) in the one-site-specific PCR to capture 3Ј and 5Ј ends of mouse ARL4 cDNA as described previously (9). All PCR products were purified, subcloned, and sequenced by the dideoxy chain termination method (13). The nucleotide sequence of mouse ARL4 has GenBank accession number U76546. Human ARL4 cDNA was isolated by the same procedures. The identical nucleotide sequence was deposited undery GenBank accession number U73960.
Northern Analyses-Blots with RNAs from adult mice and mouse embryos at several stages of development (CLONTECH) were processed for hybridization with mARL4-specific probes as described previously (12). A blot with samples of poly(A) ϩ RNA (2 g) from testes of 5-, 10-, 15-, 20-, 25-, 30-, and 60-day-old rats was kindly provided by Dr. Ian Okazaki (National Institutes of Health). TM4 (Sertoli) and MA10 (Leydig) cells were grown and harvested as described (14,15). Total RNA was extracted and reverse-transcribed using a reverse transcriptase kit (Life Technologies, Inc.). The identity of the RT-PCR product was confirmed by Southern blotting with an mARL4 cDNA probe.
In Situ RNA Hybridization-mARL4 RNA probes were prepared from pCRII-mARL4 constructed by PCR amplification of the adult mouse ARL4 cDNA and cloning of the product into the pCRII vector. In situ hybridization using digoxigenin-labeled RNA probes was performed as described by Cheng et al. (16). Briefly, 14-m cryosections of day 12.5, 14.5, 17.5, or 19.5 mouse embryos were incubated serially with 6% H 2 O 2 , 1 ϫ PBT (PBS with 0.1% Tween 20), protease K, and 4% paraformaldehyde for fixation. Prehybridization was performed at 70°C for 2 h, followed by overnight hybridization at 70°C with the digoxigenin-labeled probe (1 g/ml) in hybridization buffer (50% formaldehyde, 5 ϫ SSC, and 1% SDS containing yeast tRNA (50 g/ml) and heparin (50 g/ml)). After three stringent post-hybridization washes with 50% formaldehyde, 2 ϫ SSC, slides were incubated (2 h, room temperature) with anti-digoxigenin antibody conjugated with alkaline phosphatase (Roche Molecular Biochemicals) diluted 1:2000, and colorized by incubation with substrate nitro blue tetrazolium chloride/5bromo-4-chloro-3-indolyl phosphate. Inspection and photography were performed with an Olympus microscope. In situ RNA hybridization with whole mounted embryos was performed as described by Cheng and Flagnan (17).
Expression and Purification of Recombinant Proteins-The entire open reading frame of human ARL4 was obtained by PCR, using primers that incorporated unique NdeI and BamHI sites, respectively, at the initiating methionine and 6 base pairs downstream from the stop codon.
For the preparation of the His-tagged fusion protein, the hARL4 PCR product was cloned into the expression vector pET15b (Novagen), yielding pET15b-His-hARL4. For the nonfusion protein, the hARL4 PCR product was digested with NdeI and BamHI, purified, and annealed to expression vector pT7/Nde (18), yielding pT7-ARL4, which was used to transfect BL21(DE3) (9). Cell pellets were harvested and His-tagged fusion protein was isolated on Ni 2ϩ -NTA resin (Qiagen, Chatsworth, CA) by standard methods. The purity of the His-tagged hARL4 was assessed by SDS-PAGE.
Generation of ARL4 Antisera and Immunoanalyses-Rabbits were immunized with keyhole limpet hemocyanin-conjugated synthetic peptide LRNSLSLSEIEKLLAMGC (peptide B), corresponding to residues 138 -154 of hARL4. Antibodies (ARL4-B) were affinity purified on immobilized, recombinant hARL4. Western analysis and immunoprecipitation were performed according to the procedures of Harlow and Lane (19).
Fractionation by Differential Centrifugation-Nuclear (N), crude cytosol (C), and membrane (M) fractions were prepared as described previously (20,21). Briefly, confluent TM4 cells were scraped and homogenized in HES buffer (20 mM HEPES, pH 7.4, 1 mM EDTA, 250 mM sucrose) plus 1 mM phenylmethylsulfonyl fluoride and a mixture of protease inhibitors (leupeptin, aprotinin, chymostatin, antipain, and pepstatin, each 1 g/ml) at 4°C by 10 strokes in a ball-bearing homogenizer. The cell lysate was centrifuged at 400 ϫ g for 10 min to eliminate unbroken cells, nuclei, and cell debris. The supernatant was centrifuged (150,000 ϫ g, 1 h) at 4°C to generate cytosolic (C) and membrane (M) fractions. To obtain the nuclear fraction, cell pellet containing unbroken cells, nuclei, and cell debris was dispersed in 1 ml of Tris-buffered saline, transferred to a microcentrifuge tube, and centrifuged for 15 s in a Microfuge. Tris-buffered saline was removed and the pellet was suspended in 400 l of cold buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) by gentle pipetting in a yellow tip. The cells were allowed to swell on ice for 15 min, after which 25 l of a solution containing 10% Nonidet P-40 were added and the tube was vigorously vortexed for 10 s. The homogenate was centrifuged for 30 s in a Microcentrifuge, and the nuclear pellet (N) was collected (21).
Cell Culture and Transient Transfection-Sertoli TM4 cells were grown at at 37°C on glass coverslips (18-mm diameter) in 12-well dishes for 16 h before processing at room temperature. Mouse neuroblastoma cells, Neuro 2a (ATCC: CCL-131), were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.), supplemented with 10% fetal bovine serum, 2 mM glutamine, and 100 units/ml each of penicillin and streptomycin. The cells were subcultured by trypsinization (0.05% (w/v) trypsin, with 1% EDTA), and plating in growth medium in a humidified 5% CO 2 incubator at 37°C every 2 to 3 days. The cDNA fragments of hARL4 were fused in-frame to the C terminus of GFP by subcloning into the EcoRI and SalI sites of pEGFP-C2 (CLON-TECH). COS7 (ATCC: CRL-1651) cells were seeded on coverslips 16 h before transfection with the aid of LipofectAMINE (Life Technologies, Inc.). Freshly prepared solution A (2 g of plasmid DNA in 50 l of Dulbecco's modified Eagle's medium) and solution B (6 l of Lipo-fectAMINE in 50 l of Dulbecco's modified Eagle's medium) were gently mixed for 30 min at room temperature, added to 400 l of Dulbecco's modified Eagle's medium, and incubated with cells for 6 h at 37°C. Additional growth medium with 20% fetal bovine serum (500 l) was then added without removing the transfection mixture. Medium was replaced with fresh growth medium the day after transfection, and cells were harvested 30 to 36 h later for analysis.
Indirect Immunofluorescence Staining and Immunohistochemistry-  Cells were fixed with 4% paraformaldehyde in PBS-Ca 2ϩ -Mg 2ϩ (0.6 mM CaCl 2 and 0.5 mM MgCl 2 in 1 ϫ PBS) for 15 min, incubated with 0.1% Triton X-100 and 0.05% SDS in PBS-Ca 2ϩ -Mg 2ϩ for 4 min, and in the same buffer containing 0.2% bovine serum albumin for an additional 15 min, followed by incubation with primary antibodies; i.e. affinity purified anti-hARL4-peptide, mouse anti-p58 (Sigma), mouse anti-␤-COP (Sigma), or mouse anti-C23 (nucleolin, Santa Cruz) in the same blocking solution for 40 min. After three washes with PBS-Ca 2ϩ -Mg 2ϩ , cells were incubated with secondary antibody, Alexa 594-conjugated antirabbit IgG antibody, or Alexa 488-conjugated anti-mouse IgG antibody (Molecular Probes), washed three times with PBS-Ca 2ϩ -Mg 2ϩ , mounted on Mowiol (supplemented with Hoescht 33258), and examined with a Zeiss Axiophot equipped for epifluorescence according to standard procedures (22). Primary antibodies previously depleted of anti-ARL4 activity by incubation with purified recombinant hARL4 were used as control. Immunohistochemistry was performed using an avidin-biotin system, the ABC Vectastain Elite kit (Vector Laboratories, Burlingame, CA). Representative, mouse embryos were fixed in 4% paraformaldehyde and processed for OCT block preparation. 5-mm cryostat sections of coronally positioned 14-dpc embryos were collected. After several rinses with phosphate-buffered saline, pH 7.4 (PBS), endogenous peroxidases were quenched by adding 6% hydrogen peroxide in PBS for 15 min at room temperature. After washing with PBS, sections were treated with 20% normal goat serum for 30 min, washed with PBS, incubated with the anti-ARL4 antibody for 1 h at room temperature, and then washed with PBS. Biotinylated goat anti-rabbit antibodies diluted 1:200 in PBS were added to sections for 1 h at room temperature, followed by washing with PBS and addition of 0.05% 3,3Ј-diaminobenzidine tetrahydrochloride. Color development was monitored and stopped by dilution with water. Sections were dried, mounted, and inspected by light microscopy (Olympus Co.). Yeast Two-hybrid Screen and Assay-Yeast strains (L40), plasmids (pBTM116 and pVP16), and library for the yeast two-hybrid screen were obtained from Dr. H. Shih. The genotype of the S. cerevisiae reporter strain L40 is MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::lexA-lacZ (23). Yeast strains were grown at 30°C in rich medium (1% yeast extract, 2% Bacto-peptone, 2% glucose) or in synthetic minimal medium with appropriate supplements. Full-length wild type hARL4 cDNA was generated using 5Ј (sense) primer ARL4A, and 3Ј (antisense) primer ARL4B (Table I). Replacement of Thr 34 with Asn (T34N) or Gln 79 with Leu (Q79L) were accomplished using a two-step PCR technique as described (22). The 5Ј (sense) mutagenic primer ARL4C and the antisense mutagenic primer ARL4D were used to generate hARL4-Q79L. The 5Ј (sense) mutagenic primer ARL4E and the antisense mutagenic primer ARL4F were used to generate hARL4-T34N. The point mutation is underlined in oligonucleotide sequences (Table I). To generate ARL4-dC (NLS-deleted mutant) with deletion of 11 C-terminal amino acids (positions 190 through 200) that include the putative nuclear localization signal (10), PCR amplification was used with primer ARL4A and the mutagenic 3Ј-end antisense oligonucleotide primer ARL4G. Plasmids pLexA-ARL1, pLexA-ARL3, pLexA-ARL4, pLexA-ARL4-Q79L, pLexA-ARL4-T34N, and pLexA-ARL4-dC, constructed, respectively, by inserting a PCR-generated fragment of the ARL1, ARL3, ARL4, ARL4-Q79L, ARL4-T34N, or ARL4-dC cDNA into the EcoRI site of the pBTM116 plasmid, were used to express the ARL as a fusion protein with the DNA-binding domain of LexA.
In Vitro Interaction of ARL4 and Importin-␣-Recombinant human ARL4 mutants (ARL4-Q79L, ARL4-T34N, and ARL4-dC) were PCR amplified from the NdeI site upstream of the initiator methionine codon. PCR fragments were ligated into the pT7Blue Blunt vector (Novagen). The pT7Blue (ARL4s) plasmids were digested with NdeI and BamHI and the ARL4 fragments were ligated in-frame to the pET-15b expression vector (Novagen). To synthesize importin-␣ as an N-terminal GST fusion protein, importin-␣ cDNA was amplified by PCR using primers containing NotI sites at both 5Ј-and 3Ј-ends. After ligation into pT7Blue Blunt vector, the NotI/NotI fragment was subcloned into the same sites of GST fusion vector pGEX-4T1 (Amersham Pharmacia Biotech) to generate pGEX-importin-␣. Transformed Escherichia coli strain BL21 were grown and recombinant protein was prepared as described previously (9). The soluble fraction of E. coli expressing GST or GST-importin-␣ was incubated with glutathione-Sepharose 4B beads (Amersham Pharmacia biotech) for 30 min at room temperature. The beads were then washed four times with PBS, and finally suspended in PBS. GST and GST-importin-␣ immobilized on glutathione-Sepharose beads were quantified by SDS-PAGE with Coomassie Blue staining.
Each hARL4 construct was expressed in E. coli, and 750 l of each soluble fraction were incubated at 4°C for 1 h with 10 g of GST or GST-importin-␣ immobilized on glutathione-Sepharose beads. After washing five times with ice-cold washing buffer (25 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol), beads plus 25 l of SDS sample buffer were boiled for 10 min and a 15-l sample of each supernatant was subjected to SDS-polyacrylamide gel electrophoresis FIG. 1. ARL4 mRNA in mouse tissue and embryos and rat testis. Blots containing poly(A) ϩ RNA from adult mouse tissues (A) or whole embryos at different developmental stages (B) were hybridized with a random-primed, 32 P-labeled mARL4 cDNA probe. Hybridization with a ␤-actin probe was a control for sample loading. C, a blot containing poly(A) ϩ RNA (2 g) from prepubertal (5 and 10 days), early pubertal (15 days), midpubertal (20 and 25 days), late pubertal (30 days), and adult (60 days) rat testis was hybridized with a random-primed, 32 P-labeled rat ARL4 cDNA probe and a GAPDH cDNA as control. in 12% gel along with a 7.5-l sample of each original soluble fraction. Bound proteins were analyzed by immunoblotting.
Expression and Detection of Myristoylated ARL4 -To produce myristoylated proteins, BL21(DE3)-competent bacteria were co-transfected with pT7-ARL4 and pACYC177/ET3d/yNMT, which encodes yeast (S. cerevisiae) N-myristoyltransferase (25), and selected for both ampicillin and kanamycin resistance. [9,10-3 H]Myristic acid (1 mCi/ml in ethanol, Amersham Pharmacia Biotech) was added to a final concentration of 30 Ci/ml and the cultures were further treated as described (26). Proteins separated by SDS-PAGE were transferred electrophoretically to Immobilon-P membranes (Millipore). Incubation with indicated antibodies was carried out in phosphate-buffered saline, pH 7.4, containing 0.1% Tween 20 and 5% dried skim milk at room temperature for 60 min. Bound antibodies were detected with the ECL system (Amersham Pharmacia Biotech) according to the manufacturer's instructions.
CTA-catalyzed ADP-ribosylation and Nucleotide Binding Assay-Samples (5 g) of purified His-tagged-hARL4 or -yARF1 fusion proteins were tested for their ability to stimulate cholera toxin-catalyzed auto-ADP-ribosylation (27). Binding of GTP␥S to purified recombinant hARL4 was determined by a filtration method (22) with minor modification (9). Unless otherwise specified, 3 g of His-tagged hARL4 were incubated at 30°C in 20 mM HEPES, pH 7.5, 100 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 0.5 mM MgCl 2 , and 4 M [ 35 S]GTP␥S (Amersham Pharmacia Biotech, Ͼ1000 Ci/mmol) containing bovine serum albumin, 20 g/ml, without or with the indicated lipids, in a final volume of 50 l. Duplicate or triplicate samples were transferred to 2 ml of ice-cold 20 mM Tris-Cl, pH 7.4, 100 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol before rapid filtration on 0.45-m HA filters (Millipore, Bedford, MA). Nucleotide bound to the fusion protein was quantified by scintillation counting.

RESULTS
Developmentally Regulated Expression of mARL4 -mARL4 mRNA was much more abundant in testis than in other adult mouse tissues (Fig. 1A). A prominent ϳ1.4-kilobase transcript along with a lesser one of ϳ4.0-kilobase was observed in mouse tissues. On Northern blot analysis of mouse embryos at different stages of development, the level of mARL4 mRNA was highest on embryonic day 7 and was 90% lower by day 11 (Fig.  1B). Northern blot analysis of whole testis preparations collected from prepubertal (5 and 10 days), early pubertal (15 days), mid-pubertal (20 and 25 days), late pubertal (30 days), and adult (60 days) rats revealed a major rat ARL4 mRNA of 1.4 kilobases that was first detectable in mid-pubertal testis and was more abundant in the mature 30-and 60-day-old testis possessing fertilizing activity (Fig. 1C). In situ hybridization with frozen sections of testis and epididymis from 3-and 6-month-old mice identified mARL4 RNA mainly localized in the basal region of seminiferous tubules (i.e. spermatogonia/ spermatocytes and Sertoli cells); it was not detected in the androgen-producing Leydig cells (data not shown), as reported by Jacobs et al. (10). Similarly, in cultured cell lines, RT-PCR showed that Sertoli cells (TM4), but not Leydig cells (MA10), had mARL4 transcripts, consistent with the results of in situ hybridization, although combined RT-PCR and Southern analysis did reveal small amounts of mARL4 mRNA in the Leydig cell line (data not shown).
In Situ Hybridization Analysis of ARL4 RNA in Mouse Embryos-Abundant mARL4 mRNA was detected in 7-dpc (days post-coitus) mouse embryos (Fig. 1B). In situ hybridization of mouse embryos at 8 -10 dpc and tissue sections of embryos at later stages was performed to determine whether mARL4 is expressed in a tissue-specific manner during embryonic development. mARL4 mRNA was prominent at the earliest time point analyzed, day 8.5 at the 10 -12 somite stage ( Fig. 2A), localized specifically in pairs of somites and at the junction of forebrain and midbrain ( Fig. 2A, red and black arrows, respectively). In 9.0-dpc embryos, mARL4 mRNA was present in more caudal somites, coinciding with the sequential formation of new somites (Fig. 2B, red arrow). At the 25-29-somite stage (day-9.5 embryos), mARL4 mRNA was concentrated at the junction of midbrain and hindbrain (Fig. 2C, arrowhead) and in the caudal somites (Fig. 2C, red arrow); it had disappeared from the first 10 -12 pairs of somites and the forebrain-midbrain junction.
Later in embryonic development (Fig. 3, A and D), mARL4 mRNA was detected in the neuronal stoma around the central canal, the dorsal and ventral horns of the spinal cord, and the root ganglia (day 14.5 embryo). It was still present in the spinal cord and root ganglia of day 16.5 embryos (Fig. 3, B and E). In day 14.5 to 17.5 embryos, the brain parenchyma contained less mARL4 mRNA than did the spinal cord. A prominent zone of mARL4 mRNA was detected, however, in the neopallial cortex of the day 17.5 embryo (Fig. 3C). mARL4 mRNA decreased with maturation and was undetectable in the spinal cord of the neonatal mouse, although trace amounts were still discernible in the brain parenchyma (data not shown).
Subcellular Localization of ARL4 -To identify the intracellular location of the ARL4 protein, we generated ARL4-specific antibodies against a unique peptide sequence (residues 138 -154) of human ARL4. Predicted amino acid sequences of human and mouse ARL4 differ only at position 14 (data not shown). The affinity-purified antibodies proved to be sensitive and specific for detection of both human and mouse ARL4 proteins. Immunoblotting with this antiserum detected ARL4 in low nanogram amounts, while no reaction was detected with 100 ng of other recombinant ARLs (Fig. 4A). To assess the subcellular distribution of ARL4 in mouse Sertoli TM4 cells, homogenates were fractionated by differential centrifugation. Nuclear (N), membrane (M), and cytosol (C) fractions were separated and ARL4, proliferating cell nuclear antigen, and tubulin (cytoplasmic marker) in subcellular fractions were identified by Western blot analysis (Fig. 4B). Using affinity-purified anti-ARL4 antibodies, endogenous ARL4 was detected in the nuclear fraction, whereas immunoprecipitated ARL1 was in the cytoplasmic fraction (Fig. 4B). ARL3 was not detected in the TM4 cells. The specific ARL1, ARL3, and ARL4 antibodies used did not cross-react with other ARLs. Immunodetection of ARL4 was abolished by prior incubation of the antiserum with recombinant ARL4 (data not shown). By immunofluorescence microscopy, endogenous mARL4 of Sertoli TM4 and neuroblastoma (Neuro 2a) cells were distributed mainly over the nucleus in a fine punctate pattern (Fig. 4C). No immunoreactivity was detected after incubation of antibody ARL4-B with purified recombinant ARL4 (Fig. 4C, e) or with preimmune serum (not shown). Nuclei were stained with the DNA-binding dye H33258 (Fig. 4C, b, f, and j) and Golgi with anti-␤-COP antibodies (Fig. 4C, c, g, and k).
We have tried, thus far without success, to demonstrate nuclear localization of ARL4 in mouse fetal tissues. In the day 14 embryo, only weak staining was observed in the spinal cord and root trunks by immunohistochemistry, and none in the adjacent tissues derived from myotome and sclerotome (Fig.  4D). This is consistent with the low level of ARL4 mRNA in the day 11 mouse embryo (Fig. 1B), and makes it difficult to establish mARL4 localization in nuclei.
Two-hybrid Interaction of Importin-␣ with ARL4 C-terminal Nuclear Localization Signal NLS-To identify molecules that might act as downstream effectors of ARL4, we used plasmid pLexA-ARL4-Q79L to express the putatively constitutively active mutant of ARL4 (ARL4-Q79L) as bait in a yeast two-hybrid screen of human testis cDNA library (23). Plasmids that were associated with ␤-galactosidase production were identified from a screen of approximately 4 ϫ 10 6 colonies. The DNA sequence of each library insert was determined and eight distinct genes were chosen for further analysis. One of these, importin-␣ (karyopherin ␣2; accession number NM_002266) (28,29), was further characterized to support the observations on ARL4 localization. Five different fragments (residues 1-530, 135-530, 152-350, 233-530, and 235-530) of importin-␣ (total of 530 amino acids) interacted with pLexA-ARL4-Q79L. Because ARL4 contains a putative bipartite NLS (K 188 RRKMLRQQKKKR 200 ) at its C terminus (Fig. 5A), we constructed an NLS-deleted ARL4 mutant (ARL-dC) to test whether interaction of hARL4 and importin-␣ is dependent on this sequence. We also constructed wild type ARL4 and ARL4-T34N (predicted to be GDP-bound mutant) to test whether interaction of hARL4 and importin-␣ is nucleotide-dependent. All of the LexA fusion ARL proteins were expressed in yeast and detected by antibodies against LexA or against specific ARLs (Fig. 5B). In the yeast two-hybrid assay, transformants containing interacting proteins that transactivate two reporter genes, HIS3 and LacZ, exhibit ␤-galactosidase activity and can grow on minimal medium lacking histidine. As illustrated in Fig. 5C, LexA-ARL4, LexA-ARL4-Q79L, and LexA-ARL4-T34N, but not LexA-ARL4-dC, LexA-ARL1, or LexA-ARL3, interacted with the Gal4AD-importin-␣ fusion protein and activated the reporter genes.
To confirm that the interactions between ARL4 and importin-␣ are direct, an in vitro GST pull-down assay was used. Recombinant His-tagged hARL4 and its mutants were produced in E. coli, and the soluble bacterial proteins were incubated with immobilized GST-importin-␣ in vitro. As shown in FIG. 4. Immunolocalization of endogenous mARL4 in Sertoli TM4 and neuroblastoma Neuro-2a cells. A, specificity of antibody against ARL4. ϳ100 ng of the indicated purified recombinant Histagged ARL (a and b) and the indicated amounts of purified His-tagged mARL4 (c) were subjected to SDS-PAGE in 12.5% gels. Proteins were transferred to nitrocellulose and reacted with (a) anti-His-tag or (b and c) anti-ARL4 antibodies, followed by detection using the ECL system (A  and B). B, subcellular distribution of mARL4. Nuclear (N), membrane (M), and cytosolic (C) fractions of Sertoli TM4 cells were prepared as described under "Experimental Procedures." Equivalent amounts (from total homogenate) of each fraction were analyzed by electrophoresis and immunoblotting using specific antibodies against ARL4, ␣-tubulin (cytosolic marker), or proliferating cell nuclear antigen (PCNA) (nuclear marker). The lowest panel shows results of immunoprecipitation using specific anti-ARL1 serum (7); positions of ARL1 and the light chain of IgG are marked. Positions of protein standards (kDa) are on the left. C, immunolocalization of ARL4. Sertoli TM4 or neuroblastoma Neuro-2a cells on glass coverslips, treated as described under "Experimental Procedures," were incubated with primary antibodies, the affinity-purified anti-hARL4 peptide (a and i), or the same antibodies after incubation with purified hARL4 (e) and mouse anti-␤-COP (c, g, and k). Coverslips were mounted on Mowiol (supplemented with Hoescht 33258, b, f, and j) and inspected with a Zeiss Axiophot equipped for epifluorescence. Phase-contrast microscopy (d, h, and l) views are on the right. D, immunohistochemical localization of mARL4 protein in coronal section of 14-dpc mouse embryo. Immunohistochemistry was performed using an avidin-biotin system, the ABC Vectastain Elite kit (Vector Laboratories, Burlingame, CA) as described under "Experimental Procedures." The spinal cord (arrowheads), root trunks (arrows), and sclerotome (stars) are indicated. Fig. 5D, recombinant GST-importin-␣, but not GST, adsorbed significant amounts of ARL4, ARL4-Q79L, and ARL4-T34N, whereas no binding of ARL4-dC or ARL1 was detected. Binding of ARL4 and the two mutants (expected to exist largely in GTPor GDP-bound forms) was not grossly different.
Nucleolar Localization of ARL4-T34N Mutant-Although transient expression of hARL4 had shown its nuclear localiza- FIG. 5. Interaction of hARL4 with importin-␣ in the two-hybrid system. A, diagramatic comparison of wild type (WT) hARL1, hARL3, and hARL4 and three hARL4 mutants with the total number amino acids on the right. Residues 188 to 200 contain the bipartite NLS of ARL4. ARL-dC lacks 11 of those amino acids (190 to 200) at the C terminus. Positions of the mutations Q79L and T34N are also indicated. B, expression of LexA-ARL fusion proteins. Yeast reporter strain L40 was transformed with the indicated LexA construct, pBTM116 (LexA only), or pLexA-lamin. Samples (ϳ20 g) of cell lysates were subjected to SDS-PAGE in a 10% gel. Proteins were transferred to nitrocellulose and reacted with anti-LexA antibodies (upper panel), or specific anti-ARL1 (16), anti-ARL3 (see Footnote 2), or anti-ARL4 antibodies as indicated (lower panel), followed by detection using the ECL system. Positions of protein standards (kDa) are on the left. C, interaction of hARL4 and mutants with importin-␣ in the two-hybrid system. The yeast reporter strain L40 was co-transformed with pACT2-importin-␣ and the indicated pLexA-ARL construct or pLexA-lamin. Co-transformants were plated on synthetic histidine-containing medium lacking leucine, tryptophan, uracil, and lysine (Hisϩ plate, upper panel). Colonies from Hisϩ plates were assayed for ␤-galactosidase activity by a filter assay to test for specificity (lower panel). Colonies from Hisϩ plates were also patched on HisϪ selective plates lacking histidine, leucine, tryptophan, uracil, and lysine (His-plate, middle panel). D, in vitro interaction of GST-importin-␣ with various hARL4 constructs. Recombinant His-tagged ARL4, ARL4-Q79L, ARL4-T34N, ARL4-dC, and ARL1 were synthesized in E. coli, and 750 l of soluble fraction from each batch of cells, as indicated, were incubated with 10 g GST or GST-importin-␣ immobilized on glutathione-Sepharose beads at 4°C FIG. 6. Transient expression of hARL4 and its mutants in COS7 cells. A, localization of hARL4 and mutants in COS7 cells. Cells transfected with GFP fusion constructs of ARL4, ARL4-Q79L, ARL4-T34N, and ARL4-dC were grown on glass coverslips, fixed with formaldehyde, incubated with primary antibodies, mouse anti-p58 (Golgi marker) (d, h, l, and p), or mouse anti-nucleolin (nucleolar marker) (t). Coverslips were mounted on Mowiol (supplemented with Hoescht  33258, b, f, j, n, and r) and inspected with a Zeiss Axiophot equipped for epifluorescence. Phase-contrast microscopy (a, e, i, m, and q) and GFP-ARL4 fluorescence (c, g, k, o, and s; green) are shown. B, subcellular distribution of ARL4. Nuclear fraction (N), membrane (M), and cytosolic (C) fractions of COS7 cells were prepared as described under "Experimental Procedures." Equivalent amounts (from total homogenate) of each were analyzed by immunoblotting analysis using specific antibodies against ARL4, ␣-tubulin (cytosolic marker), and proliferating cell nuclear antigen (PCNA) (nuclear marker).
for 1 h. After washing five times with ice-cold washing buffer, 25 l of SDS sample buffer was added to each batch of beads, followed by boiling for 10 min. From each preparation, a sample (15 l) was subjected to SDS-polyacrylamide electrophoresis in 12% gel. A sample (7.5 l) of the indicated E. coli soluble fraction, in the first lane, is followed by GSTbound and GST-importin-␣-bound samples in each group of three lanes. tion (11), we wanted to examine further whether the subcellular localization of ARL4 was dependent on GTP or GDP binding. COS7 cells transiently expressing GFP-ARL4, -ARL4-Q79L, -ARL4-T34N, and -ARL4-dc were inspected by fluorescence microscopy (Fig. 6A). ARL4 appeared to be located in nuclei and partially in nucleoli (Fig. 6A, a-c, arrows). ARL4-T34N, but not ARL4-Q79L, appeared to be more concentrated in nucleoli (Fig. 6A, i-k, arrows). The nucleolar localization of ARL4-T34N was confirmed by coimmunostaining with nucleolin, a marker for nucleoli. ARL4-T34N was colocalized with nucleolin, confirming its nucleolar localization (Fig. 6A, q-t). ARL4-dC, not seen in nuclei, was in part co-localized with the Golgi marker p58 (Fig. 6A, o and p). Subcellular distribution of transiently expressed hARL4 and its mutants in homogenates of COS7 cells was also examined. ARL4, ARL4-Q79L, and ARL4-T34N were detected mainly in the nuclear fraction with very little in the membrane fraction (Fig. 6B). ARL4-dC, however, appeared mainly in the membrane fraction, confirming the results of fluorescence microscopy.
Functional Properties of ARL4 Protein-The hARL4 fusion protein, like ARL2 and ARL3, failed to stimulate auto-ADPribosylation of the cholera toxin A1 protein (data not shown). It did bind GTP␥S in a concentration-dependent manner that reached a steady state within 60 min at 30°C. Phospholipids that increased GTP␥S binding by hARF1, however, markedly decreased binding by hARL4 (Fig. 7). To determine whether hARL4 could be myristoylated, hARL4, hARL1, and yARL3 were co-expressed in E. coli with yeast N-myristoyltransferase. As shown in Fig. 8, hARL4 was myristoylated as was hARL1, which has been reported (9). yARL3, previously shown not to be myristoylated (26), served as a negative control. Thus, the biological function of ARL4, like those of ARFs and other proteins, may be influenced by myristoylation. DISCUSSION We have characterized the expression, subcellular localization, and biochemical properties of a highly conserved small GTPase, ARL4. RNA blot hybridization revealed more than one species of ARL4 mRNA, which might serve different biological functions in different cells. The existence of multiple mRNAs has been reported also for human ARL1 (7), ARL3 (5), and ARFs (reviewed in Refs. 1 and 2). Our data clearly indicate that the expression of mARL4 mRNA is developmentally regulated and are consistent with involvement of the protein in early events of spermatogenesis, somitogenesis, and the embryogenesis of the murine central nervous system. By indirect immunofluorescence and biochemical techniques, we show that localization of ARL4 in nucleoli is influenced by nucleotide binding.
ARL4 mRNA, which was detected in numerous adult organs, was most abundant in the spermatogonia and/or Sertoli cells of adult testis. No reactivity was detected in the epididymis by in situ hybridization, suggesting that ARL4 is not involved in the activation of motility or the maturation of spermatids induced by the ciliated epithelium of epididymis. The mRNA for rat ARL4, which differs in only one amino acid from mouse ARL4, was reported to be abundant in testis and 3T3-L1 adipocytes (10).
In mouse embryo, the pattern of appearance of ARL4 mRNA proceeding in a rostro-caudal direction, coincided temporally with the appearance of somites during embryonic days 8.5-10.5. Somites form in a pairwise fashion within the presomitic mesoderm following gastrulation. In the mouse embryo, somite pairs are laid down in a rostro-caudal progression with a total of 65 somite pairs formed during embryogenesis. Somitogenesis is the basis of the segmented body plan and precursor to the axial skeleton, the dermis of the back, and all striated muscles of the adult body. Proteins proposed to participate in somitogenesis include c-hairy-1, notch/delta, and the eph family (30 -33). The function of c-hairy-1 is suggested to be that of a molecular clock determining vertebrate segmentation as a result of transient waves of expression in chicken presomitic mesoderm that move rostrally with a periodicity corresponding to the time required to form one somite (90 min). A caudalrostral wave of c-hairy-1 expression is repeated during the formation of each somite.
The notch/delta family was initially identified as a group of genes encoding cell-surface proteins that define neuronal cell fate and subsequently demonstrated to participate in prefiguring somite units. Notch/delta is expressed in a metaric pattern in the presomitic mesoderm, thereby establishing boundaries of each somite segment during somitogenesis (39 -42). Unlike c-hairy-1 or notch/delta mRNAs, mARL4 mRNA was evenly distributed in each somite pair. Its appearance moves caudally as the embryo develops, as does the pattern of eph family signaling in somitogenesis (34). No other GTPase has apparently been implicated in somitogenesis. It will be important to  8. Expression of myristoylated hARL4. Recombinant hARL4, hARL1, and yARL3 were synthesized in E. coli co-expressing yeast N-myristoyltransferase and grown in medium containing [ 3 H]myristic acid. Samples of bacterial proteins (ϳ20 g) were subjected to SDS-PAGE in a 15% gel. Upper panel, gel fixed in 10% acetic acid and 45% methanol for 30 min, incubated in Amplify (Amersham Pharmacia Biotech) for 20 min, dried, and exposed to Hyper-film (Amersham Pharmacia Biotech) for 41 h at Ϫ80°C. Lower panel, nitrocellulose blot of the three recombinant proteins after SDS-PAGE was reacted with antibodies against ARL4, hARL1 (16), or yeast ARL3 (34); and the ECL system was used for visualization. investigate a possible link between the eph receptor tyrosine kinase family and the ARL small GTPase family in signal transduction pathways that determine cell fate during embryogenesis.
mARL4 mRNA was detected in the embryonic central nervous system at the earliest time examined (8.5-dpc mouse embryo), around the junction between the forebrain and midbrain. In 9.5-dpc mouse embryos, that earlier concentration of mARL4 had disappeared, and mARL4 was, instead, localized at the midbrain-hindbrain junction (Fig. 2). We interpret this change to mean that mARL4 may play a role in the establishment of central nervous system compartmentalization analogous to its function in the segmentation of somitogenesis. Indeed, Ephrin-A1, Ephrin-B2, and the receptor EphA4 are expressed in an iterative manner in the developing somites and in a gradient along the anterior-posterior axis of the developing midbrain (16,34).
At later stages of embryonic development, mARL4 was found in specific central nervous system, structures, which included the neopallial cortex (future neocortex), ventricular zone of the cerebrum, and the spinal cord. The expression zones exactly matched those of neuronal cell bodies (perikaryon), which contain abundant rough endoplasmic reticulum, ribosomes, vesicles, and inclusions that are thought of as regions of mRNA concentration. A more comprehensive study will be required to define precisely the subcellular localization of mARF4 in individual cells.
Immunofluorescence microscopy and subcellular fractionation analyses revealed that endogenous mARL4 of Sertoli TM4 and neuroblastoma Neuro-2a cells was localized mainly in nuclei, in a punctate pattern (Fig. 4). Although, transiently expressed GFP-ARL4 fusion protein was reported to be localized in nuclei (11), we found that some of the transiently expressed GFP-ARL4 was localized to nucleoli. Moreover, nucleolar concentration of ARL4-T34N appeared to depend on its GDP-bound conformation, since ARL4-Q79L was not similarly concentrated. The function of the nucleolus as a factory for assembling ribosomal subunits is well established, but one of the more surprising findings of the past decade is the discovery of a variety of macromolecules in the nucleolus with no apparent ribosomal function (reviewed in Refs. 36 and 37). The nucleolus also seems to play a role in nuclear export, sequestering regulatory molecules, modifying small RNAs, assembling ribonucleoprotein, and controlling aging. In these novel events, the nucleolus serves as a privileged site for both recruitment and exclusion of regulatory complexes. The nucleolus may serve as a "sequestration center" for proteins that are to be kept inactive. This notion raises the possibility that some of the many proteins that have been localized to the nucleolus may be stored there in anticipation of eventual release. Inter-estingly, a p34 cdc2 homolog was localized to the nucleoli of neurons and glia in the mitotically quiescent murine central and peripheral nervous systems (38). Recently, three cell-cycle regulators, Cdc 14, Mdm2, and Pch2, have been identified whose activity is regulated by sequestration in nucleolus (reviewed in Ref. 39). Furthermore, nucleolar localization of the tumor suppressor protein p19ARF, with concomitant sequestration of the p53 inhibitor Mdm2, is disrupted by tumorassociated mutations and may be key for p53 activation (reviewed in Ref. 36). To our knowledge, ARL4 may be the first small GTPase reported to be localized in nucleoli. Although, the physiological significance of the presence of ARL4 in nucleoli is not understood, it will be important to determine whether ARL4 can participate in the regulation of nucleolar sequestration of proteins.
Of eight proteins that interacted with hARL4 in the yeast two-hybrid screening, one that interacted with ARL4-Q79L, but not ARL4WT or ARL4-T34N, in the two-hybrid system is the Sec7-domain of a known guanine nucleotide-exchange protein. 2 It will be interesting to learn whether this guanine nucleotide-exchange protein, although believed to have an extranuclear distribution, can translocate into nuclei to activate ARL4. Six of the proteins that interacted with ARL4 have a nuclear localization and one of these is involved in the dynamics of the inner nuclear membrane and lamina. 2 The small GTPase Ran, which plays a key role in nuclear transport, was recently reported to function also in mitosis by regulating microtubule nucleation and/or growth (40). The nuclear envelope of higher eukaryotes is a dynamic structure that breaks down during mitotic prometaphase and reforms during anaphase and telophase (41). During nuclear envelope breakdown, the nuclear lamina and pore complexes disassemble, and the nuclear membranes vesiculate. During reassembly, nuclear membranes are targeted to the daughter chromosomes where they fuse to enclose the chromatin. The nucleus then grows by protein import through newly assembled pore complexes and the fusion of additional vesicles. Our studies demonstrated that ARL4, like ARFs, can be amino terminally myristoylated, consistent with a function dependent upon its reversible association with specific intracellular membranes that is influenced by myristoylation as well as guanine nucleotide binding. It has been suggested that a non-ARF GTPase is required for nuclear fusion and mitotic membrane disassembly (42), and we also speculate that ARL4 may have such a role in novel nuclear membrane trafficking and/or signaling cascades during embryonic development.