We describe here a novel, evolutionarily conserved set of predicted G-proteins. The founding member of this family, TbNOG1, was identified in a two-hybrid screen as a protein that interacts with NOPP44/46, a nucleolar phosphoprotein of Trypanosoma brucei. The biological relevance of the interaction was verified by co-localization and co-immunoprecipitation. TbNOG1 localized to the trypanosome nucleolus and interacted with domains of NOPP44/46 that are found in several other nucleolar proteins. Genes encoding proteins highly related to TbNOG1 are present in yeast and metazoa, and related G domains are found in bacteria. We show that NOG1 proteins in humans and Saccharomyces cerevisae are also nucleolar. The S. cerevisae NOG1 gene is essential for cell viability, and mutations in the predicted G motifs abrogate function. Together these data suggest that NOG1 may play an important role in nucleolar functions. The GTP-binding region of TbNOG1 is similar to those of Obg and DRG proteins, which, together with NOG, form a newly recognized family of G-proteins, herein named ODN. The ODN family differs significantly from other G-protein families, and shows several diagnostic sequence characteristics. All organisms appear to possess an ODN gene, pointing to the biological significance of this family of G-proteins.
The nucleolus is well known to be the site of rDNA transcription, rRNA processing and ribosome biogenesis. A group of nonribosomal proteins, including nucleolin, p120/Nop2p, fibrillarin and NO38/B23, are thought to participate in nucleolar biogenesis in higher eukaryotes in part by serving as recruiting stations for other proteins (Tuteja and Tuteja, 1998; Zatsepina et al., 1999; Dundr and Olson, 1998; Gustafson et al., 1998). Consistent with such a role, these proteins possess several conserved domains, including poly D/E stretches and RGG repeats, that bind to nucleic acid or to other nucleolar proteins. In NO38/B23, the poly D/E is essential for binding to histone H2A and H2B (Kleinschmidt et al., 1985). In nucleolin, it may be involved in the interaction with the protein kinase CK2α (Li et al., 1996). RGG repeats are present in fibrillarin, p120/Nop2p and nucleolin. In nucleolin, the RGG domain mediates the interaction with several ribosomal proteins (Bouvet et al., 1998) and also participates in binding to nucleic acids (Heine et al., 1993; Ginisty et al., 1999). The RGG region of p120/Nop2p interacts with rRNA (Gustafson et al., 1998). Interestingly, proteins with these conserved domains are multifunctional. Nucleolin is a good example: it is involved in numerous steps of ribosome biogenenesis, from rRNA synthesis and processing, through the formation of pre-ribonucleoparticles, to nucleolar- cytoplasmic trafficking, and has additional functions unrelated to ribosome biogenesis (Tuteja and Tuteja, 1998; Ginisty et al., 1999).
Similar domains are found in NOPP44/46, the only T. brucei nucleolar protein characterized to date (Das et al., 1996). It consists of four regions: unique (U; aa 1-96), junction (J; aa 97-168), acidic (A; aa 169-217) and repeat (R; aa 218-312). The R region, with its characteristic RGG repeats, is sufficient for interaction with nucleic acids (Das et al., 1998). The moderately acidic J region and poly D/E containing A region are involved in the interaction with a protein kinase(s). The A region also interacts with the RNA binding protein p34/37 (Zhang et al., 1998) to form a trimolecular complex with 5S RNA (Pitula et al., submitted). The U region is sufficient for nuclear and nucleolar localization, despite the lack of a typical nuclear localization signal (NLS) (Das et al., 1998). It displays weak yet significant sequence similarity with a number of nucleolar proteins, such as nucleolar histone deacetylases and some immunophilins (Das et al., 1998; Lusser et al., 1997), which also lack NLSs. Despite the presence of conserved domains, NOPP44/46 does not show overall homology to any specific protein. The presence of multiple interaction domains, coupled with the high abundance of NOPP44/46, led us to propose that it may function as a scaffolding protein in the nucleolus, thereby facilitating the interaction of other nucleolar molecules.
To further explore the role of NOPP44/46, we used a two-hybrid screen to identify its molecular partners. As reported here, among the molecules identified was a predicted G- protein, TbNOG1. TbNOG1 contains the typical motifs involved in GTP-binding, plus a large C-terminal extension. Closely related sequences are present in other eukaryotes and Archea. NOG sequences also have significant homology with previously identified Obg and DRG G-proteins, and together they form a new family of G-proteins. Like TbNOG1, the Saccharomyces cerevisae and human NOG1 proteins are nucleolar and the S. cerevisiae homologue is essential. The striking conservation of amino acid sequence between distantly related species and essential nature of S. cerevisae NOG1 suggests an important function for NOG1 in an as yet unknown regulatory pathway in the nucleolus.
MATERIALS AND METHODS
Organisms, media and general procedures
Escherichia coli XL1-Blue was used as a plasmid host. S. cerevisiae strains were grown in YEPD (1% yeast extract, 2% BactoPeptone and 2% glucose) or synthetic complete medium (C medium) (Sherman et al., 1981) lacking appropriate supplements to select for plasmids. For activation of GAL1 regulated constructs, 2% galactose and 2% raffinose were used in lieu of glucose. Yeast were transformed by the lithium acetate method (Gietz et al., 1992). KAC plates (1% potassium acetate, 0.1% yeast extract, 0.1 mg/ml adenine, and 0.1 mg/ml uracil) were used for sporulation. The strains employed include the C306 diploid (MATα/MATa, trp1/trp1, ura3-52/ura3-52, leu2-3, 112/leu2- 3, 112, his3D/his3D, can1/+, cyh2/+, lys5/+, ade6/+), the haploid strain C296-21-3h1 (B. Jensen, unpublished strain), EGY191 (MATα, his3, trp1, ura3, lexA2ops-LEU2) and EGY48, which differs only by the presence of six rather than two operators (Estojak et al., 1995). Procyclic form T. brucei TREU667 and the transgenic 29-13 procyclic form line (Wirtz et al., 1999) were cultured in SDM-79. The latter cells co-express T7 RNA polymerase and the tetracycline repressor (TetR). COS cells were cultured in Dulbecco’s modified Eagle medium.
All primers used, along with the relevant restriction sites, are shown in Table 1. Unless otherwise noted, for polymerase chain reaction (PCR) cloning the products were digested with the listed enzymes and ligated to similarly cleaved vectors. All constructs were checked by a combination of restriction enzyme digestion and DNA sequence analysis.
Two-hybrid system for NOPP44/46-interacting proteins
The U (aa1-96) and UJ (aa1-168) and UJA (aa1-217) regions of NOPP44/46 were PCR amplified from T. brucei genomic DNA using 5′ primer NOPP-2S and 3′ primers NOPP-96A, NOPP-168A or NOPP-217A. The NOPP44/46 fragments cloned into pEG202. The resulting bait plasmids encode the LexA DNA binding domain (aa1- 202) fused to U, UJ, or UJA.
The yeast strain EGY191 harboring the β-galactosidase reporter plasmid p1840 (GAL1-LexA1op-lacZ and URA3 gene) and the LexA- UJ bait plasmid, were transformed with a T. brucei genomic library. This library, gift of Drs U. Goeringer and A. Souza, was generated from a Sau3A partial digest of T. brucei genomic DNA (IsTaR 1) that was cloned into the BglII site of pJG4-5. The libarary sequences are fused to a hemagglutinin (HA)-tagged B42 activation domain and expression is controlled by the GAL1 promoter. Colonies expressing proteins that interacted with LexA-UJ were identified by galactose- dependent growth in medium lacking Leu and confirmed by β- galactosidase activity, and prey plasmids were rescued.
For mapping interactions, coding regions of TbNOG1 and its fragments TbNOG1(1, 425) and TbNOG1(1, 334) were amplified from genomic DNA using 5′ primer TNOG-1S and 3′ primers TNOG- 655A, TNOG-425A or TNOG-334A. XhoI-digested PCR products were ligated with XhoI-digested pJG4-5. The PCR products were also for cloning into pTbmyc2 (see below). The G1 mutant GKS>AAA (aa 181-183) was constructed by site-directed mutagenesis of the full length TbNOG1 bait plasmid using the Stratagene Quick Change site- directed mutagenesis kit and primers TNOG1-G1S and TNOG1- G1AS. Interactions were tested in strain EGY48.
DNA library screening and sequencing
A portion of the original clone 4 was used as a probe to screen procyclic and bloodstream λZap II cDNA libraries. Positive plaques were purified and the plasmids containing cDNAs were recovered according to the manufacturer’s protocol (Stratagene). Sequences of both strands of two cDNAs were determined following the creation of nested deletions (Henikoff, 1987).
Expression of epitope-tagged TbNOG1 in T. brucei
The PCR amplified TbNOG1 fragments generated above were restricted with XhoI and XbaI and ligated with similarly digested pTbmyc2 (Das et al., 1998), which places a double myc tag at the C terminus. Coding regions of TbNOG1-myc and TbNOG1(1, 334)- myc were re-amplified from these constructs using primers TNOG- 1S-H and Myc-A and cloned into pLew79 (Wirtz et al., 1999), replacing the luciferase gene. The resulting plasmids express tagged TbNOG1 sequences from a PARP promoter that is regulated by the Tet operator. The phleomycin resistance gene is expressed separately from a T7 promoter on the same plasmid. pLew79 derived plasmids were linearized with NotI within the rDNA intergenic sequence and transfected into the T. brucei 29-13 cell line as described (Anderson et al., 1998) for selection with phleomycin (10 μg/ml). For the induction, Tet was added to 10 μg/ml.
Immunoprecipitations and immunofluorescence analysis of epitope-tagged proteins in T. brucei
The preparation of cell lysates, immunoblotting, immunoprecipitation, and immunofluorescence analysis were performed as described (Das et al., 1996). Primary antibodies were mouse anti-myc 9E10 (IgG1) ascites fluid and purified mouse anti- NOPP44/46 antibody ID2 (IgG2a) (Parsons et al., 1994). They were used for immunoprecipitations at 0.75 μl/mg and 0.75 μg/mg cellular protein, respectively, and complexes were collected with Protein G- or Protein A-Sepharose. For western blot analysis, anti-myc was used at 1:1000 dilution and anti-NOPP44/46 was used at 1 μg/ml. Secondary antibodies were 125I-labeled goat anti-mouse IgG or Protein A. For immunofluorescence microscopy, primary antibodies were used at 10-fold higher concentrations. They were revealed with fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG1 and Texas red (TXRD)-conjugated goat anti-mouse IgG2a at 10 μg/ml. All fluorescently-tagged secondary antibodies were obtained from Southern Biotechnology, Inc. Cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI).
Expression of recombinant TbNOG1 and production of antisera
The TbNOG1 gene was excised from the plasmid p413-GAL1-HA- TbNOG1 (J. Park, unpublished plasmid) using SpeI and XhoI. The resulting fragment, which encodes HA-tagged TbNOG1, was ligated into the NheI and XhoI sites of the pRSetA expression vector. Following transformation into BL21(DE3)pLysS cells, HA-TbNOG1 synthesis was induced for two hours at 37°C with 1 mM isopropylthio-β-D-galactoside. Cell pellets were resuspended in 1× PBS, 10 mM imidazole, pH 7.4, and 100 μg/ml lysozyme, and lysed by freeze/thaw followed by sonication. HA-TbNOG1 was in the inclusion body fraction. This fraction was separated by SDS-PAGE and the appropriate band excised. The isolated material was used to immunize rabbits (six injections of 0.5 mg, the first in complete Freund’s adjuvant and the rest in incomplete Freund’s adjuvant). The antiserum was used at a 1/2000 dilution in immunoblot analyses, using approximately 3×106 cells per lane. As a control, the same blots were probed with antibody to the 56 kDa isozyme of phosphoglycerate kinase, at a 1/2000 dilution (Parker et al., 1995).
Expression of epitope-tagged human NOG1 in COS cells
HsNOG1 (accession 4191616) was PCR amplified from HeLa S3 cDNA library (Clontech) using the primers HNOG1-1S and HNOG1-633A and cloned into pFLAG-CMV-2 (Kodak). The resultant plasmid encodes the full-length HsNOG1 with the FLAG tag at the N terminus.
For transfection, cells were grown as monolayers on glass coverslips. After fixation in methanol/acetic acid (1:1, v/v) for 10 minutes, cells were permeabilized with 0.2% Triton X-100 in PBS. The nucleolus was stained with mouse anti-NO38/B23 (1:500 dilution, gift of P. K. Chan) and revealed with TXRD-conjugated rabbit anti-mouse IgG (10 μg/ml). Tagged HsNOG1 was revealed by FITC-anti-FLAG M2 at 50 μg/ml. Cells were counterstained with DAPI.
Expression of epitope-tagged ScNOG1 in yeast
The ScNOG1 open reading frame (ORF) plus 293 bp downstream of the stop codon was amplified usingTurbo Pfu polymerase and primers SNOG1-1SX and SNOG1-3′K. The restricted PCR product was ligated into p414-GAL1-HA-TbNOG1 (B. Jensen, unpublished plasmid) replacing the TbNOG1 gene. The resulting plasmid, pGAL- HA-ScNOG1, a double HA tagged-ScNOG1 under the control of the GAL1 promoter. The ligation reaction was directly transformed into C306nog1Δ (see below), since it proved difficult to isolate the ScNOG1 gene in E. coli.
For immunofluorescence, the transformed cells were grown overnight to mid-log phase in C medium lacking Trp supplemented with raffinose. The cells were washed with succinate buffer (95 mM, pH 6) containing 0.16% yeast nitrogen base without amino acids or ammonium sulphate, and resupended in C medium lacking Trp and containing galactose and raffinose to induce HA-ScNOG1. After 2-2.5 hours the cells were fixed with 3.7% formaldehyde in PBS for 1 hour, and processed for immunofluorescence as described (Winey et al., 1991). The HA epitope was detected using rat anti-HA 3F10 (Boehringer Mannheim) at 0.5 μg/ml and detected with TXRD- conjugated goat anti-rat IgG at 5 μg/ml. Nop1p was detected using a mouse anti-Nop1p (Aris and Blobel, 1988), kindly supplied by Dr J. Aris, at 40 ng/ml and detected with FITC-goat anti-mouse IgG1 at 0.5 μg/ml.
Generation and complementation of a NOG1 deletion in S. cerevisae
The deletion disruption of ScNOG1 was obtained by transformation of a PCR-synthesized marker cassette with ScNOG1 flanking homology regions into the diploid strain C306. The KanMX marker, which specifies resistance to G418, was amplified from pBS-KanMX (B. Jensen, unpublished plasmid) with the primers SNOG1-Δ5 and SNOG-Δ3. The PCR product was used to directly transform the yeast strain C306 with selection on YEPD containing 0.2 mg/ml G418. Proper integration was verified by PCR using primers SNOG-5′ and SNOG-3′. The resulting strain was referred as to C306 NOG1/nog1Δ.
To rescue gene disruptions, the ScNOG1 ORF was amplified using primers SNOG1-1SE, and SNOG1-647A. The digested PCR fragment was ligated into p414-GAL1 (Mumberg et al., 1994) and directly transformed into C306 NOG1/nog1Δ. The resultant diploid, C306 NOG1/nog1Δ/pGAL1-ScNOG1, was sporulated and dissected onto medium containing galactose or glucose to induce or repress expression of the plasmid-encoded ScNOG1. All viable segregants were tested for their genotype at the chromosomal NOG1 locus by plating on G418.
Generation and functional analysis of mutations in the GTP-binding domain of ScNOG1
The transcriptional terminator of ScNOG1 was amplified from strain C296-21-3h1 with the primers SNOG-3′SX and SNOG-3′AK and cloned into p414-GAL1 to yield p414-GAL1-NOG1term. XbaI and XhoI were used to cleave the plasmid between the promoter and terminator. DNA fragments containing mutations in either the G1 or G3 motifs of ScNOG1 were generated using the Turbo-Pfu (Stratagene) in stitching PCR. For the GKS to AAA mutation at residues 179-181 (G1 motif), the M13 reverse primer and the mutagenic SNOG1-G1S primer were used to generate a fragment spanning the GAL1 promoter region through the mutated residues using pGAL-HA-ScNOG1 as template. Similarly, primers SNOG1-G1A and SNOG-3′AK were used to generate a fragment spanning the mutated residues through the terminator region. PCR products were mixed together to serve as template for a final round of PCR using the M13 reverse primer and SNOG-3′AK primers. The mutation KTD to ATL at residues 279-281 was similarly performed using the mutagenic primers SNOG1-G3S and SNOG-G3A. Following cotransformation into yeast with the digested p414-GAL1-NOG1term DNA, the plasmid-encoded NOG1 mutant genes were amplified and sequenced.
The C306 NOG1/nog1Δ diploids harboring the mutant or wild-type HA-ScNOG1 plasmids or vector, p414-GAL1-NOG1term, were sporulated. The progeny were plated on galactose plus raffinose C medium lacking Arg and Trp and containing canavanine (60 μg/ml, Sigma) and cycloheximide (10 μg/ml, Sigma) to select for haploid segregants. To test for the presence of the nog1::KAN allele, individual segregants were replica plated to galactose plus raffinose medium lacking Trp and to the same medium containing 400 μg/ml G418.
Two-hybrid screening with the UJ region of NOPP44/46
To detect potential molecular partners of NOPP44/46, we conducted an interaction screen using the yeast two-hybrid system. We screened a genomic library since T. brucei genes are densely packed and rarely contain introns. To minimize false positives, the bait plasmid contained the U to J region (aa1-168) of NOPP44/46 (Das et al., 1998), but lacked the highly charged A and R domains. Additionally, we used a stringent system, where only two operators were present next to the selectable marker (LEU2) (Estojak et al., 1995). From 2 million independent transformants (equivalent of approximately 40 haploid genomes), we identified 34 clones that activated the reporters in combination with LexA-UJ but not with LexA alone or LexA-lamin. The plasmids mapped into five major groups. Some features of each predicted protein as characterized to date are summarized in Table 2. Analysis of the predicted protein sequence of partial DNA sequences of clones 1 and 2 revealed no obvious similarities with sequences in the databases. Clone 3 displayed a high degree of homology with a member of the variant surface glycoprotein family (major plasma membrane proteins in T. brucei). It was therefore deemed a non-physiological positive. Clone 5 was related to a predicted protein of unknown function in Arabadopsis thaliana. Clone 4 displayed a high degree of sequence similarity with hypothetical proteins in Saccharomyces cerevisiae, Caenorhabditis elegans, and humans.
cDNA, structure and interaction map of clone 4
To obtain a complete clone 4 cDNA, we screened both bloodstream and procyclic cDNA libraries and chose the longest clone from each library for sequencing. The combined sequence comprised 3934 nt plus a poly(A) tail (Fig. 1a). The presence of the spliced leader (the sequence trans-spliced to form the 5′ end of all nuclearly-encoded mRNAs in T. brucei) (Parsons et al., 1984), together with the poly(A) tail, indicates that the sequence represents the full-length mRNA. Interestingly, structural heterogeneity was found at the 3′ end: relative to the stop codon, the poly(A) tail was attached at nt 64 in the procyclic clone and at nt 1711 in the bloodstream stage clone. Whether this represents a stage-regulated difference is unclear.
The ORF encodes a polypeptide of 655 aa residues with a calculated molecular mass of 74.8 kDa (Fig. 1b). The predicted protein has a pI of 9.8, with 16.3% basic and 12.5% acidic residues. PSORT (Nakai and Horton, 1999), identifies a bipartite NLS bipartite at residues 131-144 and two predicted NLSs at residues 395-401 and 511-518 (Fig. 1b).
Codons 176-293 of the ORF contain the three consensus motifs, designated here as G1, G2 and G3, that define a G- protein (Fig. 1c). The G1 motif, GXXXXGK(S/T), is found in most proteins that bind purine nucleoside triphosphates, including G-proteins, protein kinases and ATPases (Walker et al., 1982). The G2 motif, DXXG, is situated close to the G1 motif in the tertiary structure and is thought to be involved in the conformational change between the GTP- and GDP-bound forms (Bourne et al., 1991; Seeburg et al., 1984). The G3 motif, NKXD, determines the specificity for guanine. Based on the presence of well-characterized GTP-binding motifs in the required order, we predict that the clone 4 gene encodes a G- protein. Since the N-terminal region (aa1-334) contains the GTP-binding motifs, it is referred as to the G-domain. As shown below, the protein is localized to the nucleolus and the gene is therefore named TbNOG1 (T. brucei nucleolar G-protein 1).
To map the interaction of TbNOG1 and NOPP44/46, we tested of various regions of TbNOG1 and the UJA, UJ, and U domains of NOPP44/46 in the two-hybrid system (Fig. 2). The original two-hybrid clone encodes TbNOG1(426, 655), which interacted with the UJA and UJ regions of NOPP44/46 but not with the U region. Interestingly, TbNOG1(1, 425) and TbNOG1(1, 334), which do not overlap with TbNOG1(426, 655), interacted with all of the NOPP44/46 constructs, including the U region. The same results were obtained with the full-length TbNOG1. Together, these data indicate that the N-terminal 334 aa, which contains the G domain, also contains a binding site for the U region of NOPP44/46, while the C- terminal 230 aa contains a binding site for the J region. These experiments do not rule out additional sites of interaction.
We also tested the ability of the NOPP44/46 fragments to interact with TbNOG1 that contained mutations in the G1 motif. Side chains of residues within this motif are involved binding the β-phosphate of GTP and coordination with the magnesium ion (Kjeldgaard et al., 1996). In this experiment, the conserved G1 residues GKS were mutated to AAA, residues that would be incapable of mediating the aforementioned interactions. The mutant protein was still capable of interacting with U, UJ, and UJA fragments.
Expression of myc-tagged TbNOG1 protein in T. brucei
We were unable to isolate stable T. brucei transfectants constitutively expressing myc-tagged full-length TbNOG1, or its truncated versions (aa 1-425, aa 1-334 and aa 255-655). We therefore turned to the Tet-inducible system for the expression of myc-tagged TbNOG1 (Wirtz et al., 1999). Sequences encoding myc-tagged TbNOG1 were cloned into pLEW79, where expression is driven by the PARP promoter and controlled by an adjacent Tet operator (Wirtz et al., 1999). Following electroporation into T. brucei procyclic forms, stable transfectants were isolated in the absence of Tet. In induced, but not uninduced, cells western blot analysis with anti- myc detected a single protein with the molecular mass of approximately 83 kDa (Fig. 3a). This size approximates that expected for the 74.8 kDa protein plus 3.4 kDa tag. Induction of TbNOG1-myc did not alter the expression of NOPP44/46 (Fig. 3a).
TbNOG1-myc is nucleolar and interacts with NOPP44/46 in T. brucei
To investigate the subcellular localization of TbNOG1, the transfectants were processed for immunofluorescence. Fig. 3b shows the fluorescence patterns of cells grown in the absence or presence of Tet and then stained with anti-myc to disclose the location of TbNOG1-myc. Staining with DAPI and anti- NOPP44/46 revealed the nucleus and nucleolus respectively. In uninduced cells, no signal was observed with anti-myc. Upon addition of Tet to induce TbNOG1-myc, anti-myc detected a region within the nucleus that is indistinguishable from the anti-NOPP44/46 stained nucleolus (see Fig. 3b, overlay). Thus, TbNOG1 is a nucleolar protein in T. brucei.
To provide biochemical evidence that TbNOG1 interacts with NOPP44/46 in T. brucei, we carried out co- immunoprecipitation experiments (Fig. 3c). In cells expressing TbNOG1-myc, the tagged protein was immunoprecipitated with anti-myc. As detected by western blot analysis with anti- NOPP44/46, the precipitates also contained NOPP44/46. As expected, anti-myc immunoprecipitates from uninduced cells did not contain NOPP44/46, demonstrating that co- precipitation is mediated by TbNOG1. Conversely, anti- NOPP44/46 immunoprecipitates from induced cells contained TbNOG1-myc as detected with anti-myc antibodies. The colocalization and coprecipitation studies confirm that the interaction observed in yeast reflects a biologically relevant interaction in the parasite.
The two-hybrid interaction mapping (Fig. 2) demonstrated that the G-domain of TbNOG1 interacts with the U region of NOPP44/46. We also expressed a double myc-tagged version of this fragment in T. brucei using the Tet-inducible system. Immunofluorescence analysis demonstrated that TbNOG1(1, 334)-myc is nucleolar (Fig. 4a). As shown in Fig. 4b, NOPP44/46 was coprecipitated with TbNOG1(1, 334)-myc in induced cells. In the reciprocal experiment, the tagged G domain coimmunoprecipitated with NOPP44/46. Phosphorimaging analysis showed that about three-fold more NOPP44/46 was associated with full-length NOG1-myc than with the G-myc domain (normalized to the amount of immunoprecipitated myc-tagged protein).
TbNOG1 is expressed in bloodstream and procyclic stage T. brucei
To assess the expression of TbNOG1 in the parasite life cycle, HA-tagged TbNOG1 was expressed in E. coli, purified, and used to raise antisera. The antiserum detected a protein of apparent molecular mass of 81 kDa in immunoblot analyses (Fig. 5). The protein was observed in the actively dividing procyclic (insect midgut) and mammalian bloodstream slender forms, as well as in the non-dividing stumpy bloodstream forms. Normalization of of the signals to those seen with the constitutively expressed 56 kDa phosphoglycerate kinase (PGK-A) indicates that the levels of NOG1 are not highly regulated during parasite development. Immunofluorescence analysis using anti-TbNOG1 revealed a nucleolar distribution (data not shown).
TbNOG1 is a member of a novel family of G-proteins
A BLASTp search (Altschul et al., 1997) of the predicted protein data banks showed that sequences highly related to TbNOG1 are found in diverse eukaryotes, including Homo sapiens and S. cerevisiae (Fig. 6). The G domain (aa 1-334) displayed a high degree of sequence homology (62-71% identity), while the C-terminal half was much less conserved (25-31% identity). Interestingly, the G-domain also showed over 50% similarity with proteins in members of the Archaea. In the archeal proteins, the C-terminal extension is missing. All the proteins shown in Fig. 6 were found fortuitously by genomic or cDNA cloning and thus are functionally uncharacterized. Apart from the GTP-binding region, TbNOG1 displayed no obvious sequence similarities with the well-characterized G-proteins of the Ras, EF-2 and heterotrimeric G protein families, indicating that it is a member of a distinct family of G-proteins.
The effector (E) motif, also called Switch I, is not involved in GTP-binding but is thought to participate in the interaction with effector molecules. Unlike the G1-G3 motifs, which are shared among almost all G-protein families, this motif is family-specific. The E motif is located between the G1 and G2 motifs, and contains a conserved threonine (Sprang et al., 1997). In the sequences closely related to NOG1, a conserved threonine containing motif is found in an appropriate position. This motif, YAFTT, occurs in TbNOG1 at position 199-203 (Fig. 1b). To identify the constellation of proteins related to the NOG1 G-domain, we scanned the protein databases with the pattern (G(X)4GKS(X)15YXFTT) corresponding to the G1 and E motifs. The screen identified 42 sequences, all of which also contain G2 and G3 motifs, indicating they are all predicted G- proteins. Analysis of the regions containing GTP-binding motifs (a subset is shown in Fig. 7a) identified several residues that are conserved within these proteins but not with other G- proteins. These include Pro residues at position 3 in the G1 and G2 motifs, and Leu or Ile at position 5 in the G2 motif.
Alignments of the regions spanning the G motifs were used to calculate a phylogenetic tree. As shown in Fig. 7b, the 42 proteins clustered into three groups, one of which contained all the NOG1 related proteins. Within this cluster, the phylogenetic tree was very similar to that seen when the tree was built with the entire sequences of NOG1 proteins (data not shown). The other two clusters contain the known Obg and DRG proteins (Okamoto et al., 1997; Maddock et al., 1997; Sazuka et al., 1992), as well as some unrecognized homologues. This analysis shows that NOG1, Obg and DRG G regions are closely related to each other, but only distantly related to the Ras family of G-proteins. EF-2 and heterotrimeric G-proteins were even less related. The close evolutionary relatedness, and the striking conservation of family-specific E motif and other residues clearly indicate that Obg, DRG and NOG1 are sister groups, forming a novel, large family of G-proteins. We propose this family be designated as an ODN for its prototype subfamily constituents.
Human and yeast NOG1 homologues localize to the nucleolus
Humans possess two predicted proteins with overall homology to NOG1, while S. cerevisae encodes a NOG1-related protein at the YPL093w locus (named herein ScNOG1). To determine whether nucleolar localization is a common feature of eukaryotic NOG proteins, we expressed epitope-tagged versions of ScNOG1 and one of the human proteins (named herein HsNOG1). FLAG-tagged HsNOG1 was expressed transiently in COS cells (Fig. 8a). Anti-FLAG staining revealed that tagged HsNOG1 was found in several discrete patches within the nucelus. These regions corresponded to the regions staining strongly for the nucleolar protein B23. Because B23 shuttles between the nucleolus and cytoplasm (Finch et al., 1993), some cytoplasmic staining is also observed. This result indicates that HsNOG1, like TbNOG1, localizes to the nucleolus.
The coding region of ScNOG1 was tagged at its 5′ end with sequences encoding two HA epitopes and the resulting plasmid (pGAL1-HA-ScNOG1) was transformed into diploid yeast. Following galactose induction for two hours, the fluorescence pattern of HA-ScNOG1 shows the characteristic nucleolar crescent shape at the edge of the nucleus (Fig. 8b). The staining overlaps that of the nucleolar marker Nop1p (Aris and Blobel, 1988), albeit being fainter and perhaps slightly more diffuse. Upon prolonged induction, HA-ScNOG1 was also found in the nucleus and, eventually, cytoplasm. No staining was seen without induction. These data indicate that ScNOG1 is nucleolar when expressed at low levels.
ScNOG1 is an essential gene
We examined the effect of disruption of the NOG1 gene in S. cerevisae, an organism more amenable to genetic manipulation than trypanosomes. One copy of ScNOG1 in the diploid strain C306 was replaced with the KanMX marker module, which confers resistance to G418, to generate the strain C306 NOG1/nog1Δ. A plasmid expressing ScNOG1 under the control of a GAL1 promoter was then transformed into this strain. The transformants were sporulated, and dissected onto medium containing glucose to repress expression of plasmid- encoded ScNOG1. Subsequent analysis of the dissected tetrads grown on glucose showed a 2:2 segregation of viable to nonviable spores (Fig. 9). Microscopy revealed that the dead spores were able to germinate and divide only once. The haploid colonies failed to grow on medium containing G418 indicating that they received the wild-type gene and that the dead spores must have received the nog1::KanMx deletion allele. These data are consistent with a recently released genome-based analysis that suggested that ScNOG1 is essential (Winzeler et al., 1999). When tetrads were dissected on medium containing galactose to induce expression of plasmid borne ScNOG1, the tetrads yielded either three or four viable colonies. This rescue indicates the loss of viability was caused by disruption of ScNOG1 gene. Thus ScNOG1 is an essential gene.
The predicted G motifs are required for ScNOG1 function
To assess the functional significance of the signature GTP- binding motifs, we tested the ability of ScNOG1 mutated in either the G1 or G3 motifs to complement the ScNOG1 deletion. In the G1 mutant, the conserved residues GKS (aa 179-181) were mutated to AAA. In the G3 mutant, the conserved residues KTD (279 to 281) were mutated to ATL. The plasmids were transformed into the diploid C306 NOG1/nog1Δ and the expression of the wild-type and mutant HA-tagged proteins was verified by immunoblot analysis using anti-HA (Fig. 10a). The wild-type tagged ScNOG1 and G1 mutant were expressed at similar levels, while expression of the G3 mutant protein was somewhat lower. Immunolocalization using anti-HA demonstrated that, like wild-type ScNOG1, the G1 mutant protein resided in the nucleolus (Fig. 10b). Thus the mutant protein was functional with respect to localization.
The diploids containing the mutated ScNOG1 genes were sporulated and individual haploid segregants were replica plated onto medium plus and minus G418. Only segregants with the chromosomal nog1 deletion (nog1::KanMx) can survive the drug. To survive, cells must also have a (plasmid- encoded) functional NOG1 gene. As shown in Table 3, colonies with the genotype C306 nog1Δ that contained the wild-type HA-ScNOG1 plasmids were viable, but no colonies were obtained when we attempted to cover the deletion with the G1 (Scnog1-g1) or G3 (Scnog1-g3) mutant constructs. Thus mutation of the G1 or G3 motifs abrogates NOG1 function, strongly suggesting that GTP/GDP binding is a required function of this molecule. The recovery of colonies bearing the wild-type construct was lower than theoretically expected, suggesting that the plasmid borne HA-ScNOG1 may not be completely equivalent to the wild-type chromosomal copy. Such a finding could be due to the presence of the tag or to altered expression levels resulting from the use of the GAL1 promoter.
Interaction of NOPP44/46 and a nucleolar G-protein, NOG1
T. brucei NOPP44/46 does not show overall homology to any known protein, yet each of its domains has significant similarity with modules that participate in intermolecular interactions in other nonribosomal nucleolar proteins (Tuteja and Tuteja, 1998; Zatsepina et al., 1999; Dundr and Olson, 1998; Gustafson et al., 1998). Due to the presence of these domains and the relatively high abundance of NOPP44/46, we previously hypothesized that this protein provides a docking site that brings other nucleolar proteins into proximity. To gain insight into this possibility, we searched for proteins that directly interact with NOPP44/46 using the two-hybrid system. One of the proteins thus identified was TbNOG1, a nucleolar predicted G-protein. A search of the literature indicates one other potential nucleolar G-protein. This protein, encoded by the human Ngp-1 gene, has the typical G1 and G2 motifs (Racevskis et al., 1996). However, the identified G3 motif is amino to the G1 motif, a highly unusual situation.
Using the two-hybrid system and coimmunoprecipitation studies, we demonstrated that sites within the G-domain and C-terminal extension of TbNOG1 interact with the U and J regions of NOPP44/46 respectively. As noted above, sequences related to the U and J regions are present in a number of nucleolar proteins. We speculate that proteins with U- or J-like domains may interact with their corresponding NOG1 homologues. The multiple binding sites for conserved domains of nucleolar proteins raise the possibility that some but not all interactions with NOG1 are modulated by the GDP/GTP ratio. Our two-hybrid analysis of the Tbnog1-g1 mutant protein suggests that the interaction with NOPP44/46 is not fully restricted to the GTP-bound form of NOG1. This finding suggests that the interaction between these two proteins may be of a constitutive nature or may be more subtly altered by the ratio of GTP to GDP. NOG1’s interactions with other proteins may be directly related to ribosome biogenesis, but additional possibilities exist as well. For example, the function of a number of proteins, including CDC14 (Shou et al., 1999) and MDM2 (Zhang and Xiong, 1999; Tao and Levine, 1999), have been shown to be regulated by nucleolar sequestration. Moreover, functions involved in RNA modification and gene silencing (Pederson, 1998a; Pederson, 1998b; Smith et al., 1999), aging, exit from mitosis, cell proliferation and tumor suppression have also been associated with the nucleolus (Garcia and Pillus, 1999; Zhang and Xiong, 1999; Tao and Levine, 1999).
A newly recognized family of conserved G-proteins, ODN
Several features distinguish the NOG subfamily from other well-characterized G-proteins (Kjeldgaard et al., 1996). The sequence similarity of TbNOG1 with those proteins is restricted to the GTP-binding motifs. However, the entire amino half of TbNOG1, including the G motif region, displays a high degree of sequence similarity with the NOG-like proteins of distantly related organisms. In these proteins, the GTP binding motifs are located internally, whereas in most G- proteins the G1 motif resides within the N-terminal 50 amino acids. Additionally, eukaryotic NOG proteins contain an additional, moderately conserved domain of 200 amino acids (Kjeldgaard et al., 1996).
Until now, the relationship between Obgs and DRGs had not been recognized. However, the identical family-specific E motif and the unusual conservation of several amino acid residues in the GTP-binding region indicate that Obg, DRG and NOG are sister groups, forming a novel family of G- proteins that we call ODN. Archea and eukaryotes possess both DRG and NOG1 proteins, while eubacteria possess Obg proteins. The only eukaryotic Obg-like sequence we have identified is from C. elegans. These data suggest that most eukaryotic ODN proteins are derived from archeal ancestors, a hypothesis that can be assessed as more eukaryotic genomes are completed. The presence of ODN G-protein sequences in eubacteria, Archea, and eukaryotes indicate that the ODN progenitor gene is truly ancient, being present in their common ancestor. ODN sequences are widely distributed among phyla and were detected in every genome that has been completely sequenced, pointing to strong selective pressure to maintain this family of G-proteins. The essential nature of Obg in eubacteria (Okamoto et al., 1997; Maddock et al., 1997; Kok et al., 1994) and NOG1 in S. cerevisae as presented here corroborate the importance of the ODN family.
Sequence and function of ODN proteins
The presence of typical GTP-binding motifs sequentially arranged within 120 residues strongly suggests that NOG1 binds GTP. Indeed, these motifs in NOG1 are almost identical to those in Bacillus subtilis (Fig. 7) and Caulobacter crescentus Obg-like proteins, the only ODN family members that have been shown biochemically to bind GTP (Lin et al., 1999; Welsh et al., 1994). Attempts to express soluble NOG1 in E. coli under a variety of conditions were unsuccessful, precluding detailed biochemical studies and [32P]GTP gel overlay assays are unsuitable for larger G-proteins (Huber and Peter, 1994). However, our analysis of mutants in the G1 and G3 motifs of ScNOG1 shows that these motifs are essential for functional complementation, buttressing the argument that NOG1 function involves GTP/GDP binding.
The ODN family of G-proteins shows specific, unique sequence characteristics. All members possess a Pro residue within the G1 motif (Fig. 7a), corresponding to Gly-12 in H- Ras. Gly-12 of Ras is mutated in several types of human cancers (Hamer et al., 1991) and interestingly, proline is the only substitution that is not oncogenic (Seeburg et al., 1984). Conversely, in Streptomyces coelicolor, an Obg with a Pro to Val mutation does not support cell viability (Okamoto and Ochi, 1998). The substitution of Gln-61 of the G2 motif with Leu in Ras family of G-proteins (e.g. RanQ69L and RhoQ63L) abolishes intrinsic GTPase activity (Bischoff et al., 1994; Renshaw et al., 1996). All the members of the ODN family contain Leu or Ile at this position, implying that low GTPase activity seen in the Obgs of B. subtilis and C. crescentus (Welsh et al., 1994; Lin et al., 1999) may be a characteristic of this family. These two Obgs also have low affinity for GTP compared to Ras-like G proteins (micromolar versus nanomolar) and exchange is several logs faster.
Recently, Obg has been shown to be required for stress activation of σB, a transcription factor that enables expression of the B. subtilis stress regulon (Scott and Haldenwang, 1999) and to associate with ribosomes in this organism (Scott et al., 2000). Additionally, Obg is necessary for the transition from vegetative growth to sporulation in B. subtilis and S. coelicolor (Okamoto et al., 1997; Kok et al., 1994). Since fluctuations in cellular GTP content are thought to be the primary initiator of morphological development in these cells (Okamoto et al., 1997; Kok et al., 1994), it has been proposed that Obg acts as a sensor for intracellular GTP/GDP ratio and thereby for cell growth status set by nutrient availability. This hypothesis is compatible with the biochemical analysis of Obg, which indicates that the molecule has characteristics expected for a direct sensor: fast exchange, similar micromolar affinities for GTP and GDP, and a low rate of hydrolysis. Intracellular GTP levels are linked to growth in multiple ways. For example, in bacteria, promoter activity is the rate-limiting step in ribosome synthesis (Nomura et al., 1984; Gaal et al., 1997) and efficient initiation requires a high concentration of GTP (Gaal et al., 1997). In eukaryotes, the cellular GTP levels have also been thought to be responsible for the regulation of rRNA synthesis (Grummt and Grummt, 1976). Additionally, localization of NO38/B23 into the nucleolus is dependent on the cellular GTP levels (Finch et al., 1993). These observations raise intriguing possibilities for the role of a nucleolar G protein such as NOG1, which could sense the cellular energy status and use that information to aid in the integration of nucleolar functions. It is interesting to recall that the class II transcriptional transactivator protein CIITA of mammals is also a GTP- binding protein with low hydrolysis rates (Harton et al., 1999).
Trypanosome, yeast and human NOG1 proteins are localized to the nucleolus of their respective organisms, clearly implying that nucleolar localization is a common characteristic of these proteins in eukaryotes. The findings that NOG1 proteins are evolutionarily conserved and that ScNOG1 is essential for viability suggest that these proteins may play a crucial role in nucleolar functions. A global survey of yeast mRNA expression shows that NOG1 transcript levels decrease dramatically as the cells switch from glycolytic to fermentative metabolism (DeRisi et al., 1997). This shift is accompanied by a major decrease in transcription of genes involved in ribosome biogenesis (Neuman-Silberberg et al., 1995). We are currently exploring the effects of NOG1 on rDNA transcription and rRNA processing in S. cerevisae.
We thank Drs E. Wirtz and G. Cross for providing pLew79 and 29- 13 cells, Dr R. Brent for providing two-hybrid vectors and strains, and Drs U. Goeringer and A. Souza for providing the T. brucei genomic two-hybrid library. We are grateful to Dr Q. Wang for testing the interactions of Tbnog1-g1. We also thank Drs J. Aris and P. Chan for antibodies, Drs S. Yarfitz and B. Hedlund for assistance in bioinformatics, and Dr Nicole Doria-Rose for assistance with COS cell transformation. This work was supported in part by NIH AI31077 and NIH S10 RR11865-01. B.J. was supported by NIH 1T32 AI07509-01.