A Deubiquitinating Enzyme That Disassembles Free Polyubiquitin Chains Is Required for Development but Not Growth inDictyostelium *

Although cell differentiation usually involves synthesis of new proteins, little is known about the role of protein degradation. In eukaryotes, conjugation to ubiquitin polymers often targets a protein for destruction. This process is regulated by deubiquitinating enzymes, which can disassemble ubiquitin polymers or ubiquitin-substrate conjugates. We find that a deubiquitinating enzyme, UbpA, is required for Dictyostelium development.ubpA cells have normal protein profiles on gels, grow normally, and show normal responses to starvation such as differentiation and secretion of conditioned medium factor. However,ubpA cells have defective aggregation, chemotaxis, cAMP relay, and cell adhesion. These defects result from low expression of cAMP pulse-induced genes such as those encoding the cAR1 cAMP receptor, phosphodiesterase, and the gp80 adhesion protein. Treatment ofubpA cells with pulses of exogenous cAMP allows them to aggregate and express these genes like wild-type cells, but they still fail to develop fruiting bodies. Unlike wild type, ubpAcells accumulate ubiquitin-containing species that comigrate with ubiquitin polymers, suggesting a defect in polyubiquitin metabolism. UbpA has sequence similarity with yeast Ubp14, which disassembles free ubiquitin chains. Yeast ubp14 cells have a defect in proteolysis, due to excess ubiquitin chains competing for substrate binding to proteasomes. Cross-species complementation and enzyme specificity assays indicate that UbpA and Ubp14 are functional homologs. We suggest that specific developmental transitions inDictyostelium require the degradation of specific proteins and that this process in turn requires the disassembly of polyubiquitin chains by UbpA.

Modification of proteins by ubiquitin is involved in many cellular processes including cell cycle progression, signal transduction, ligand-inducible endocytosis of cell-surface proteins, mating type switching in yeast, and elimination of damaged proteins (reviewed in Refs. 1 and 2). Attachment of polyubiquitin chain(s) to a protein frequently serves to target the mod-ified protein for proteolysis by the 26 S proteasome. The ubiquitin molecules in these chains are most often linked to one another by isopeptide bonds between the C terminus of one ubiquitin and the ⑀-amino group of lysine 48 of the next ubiquitin.
Protein ubiquitination is reversible. Deubiquitination is catalyzed by specialized thiol proteases, called deubiquitinating or DUB 1 enzymes, which hydrolyze the amide bond between the C-terminal Gly of ubiquitin and primary amino groups of the substrate protein (3). One of the two known classes of DUB enzymes is the ubiquitin-specific processing protease or UBP class. All members of the UBP family contain two short consensus sequences, the Cys and His boxes, which are likely to help form the active site (4,5). Several additional short sequences are also moderately well conserved (3). The DUB family is large. For example, 17 genes encode DUB enzymes in the yeast Saccharomyces cerevisiae, of which 16 are in the UBP class (2). However, little is known about their physiological functions.
One role of DUBs is the regulation of protein degradation by the 26 S proteasome. Mammalian 26 S proteasomes contain a DUB enzyme that specifically removes ubiquitin molecules from the distal ends of polyubiquitin chains attached to protein substrates (6). It may function to rescue inappropriately ubiquitinated proteins from degradation or may facilitate movement of substrates within the protease complex. Two UBP enzymes from yeast have been shown to have general roles in proteasome-mediated proteolysis in vivo. The Doa4 enzyme appears to cleave ubiquitin chains from proteins already committed to degradation by the proteasome (5). On the other hand, Ubp14 has recently been found to be the major yeast DUB that disassembles free ubiquitin chains (7). Both Doa4 and Ubp14 may function by preventing competition between products generated (at least in part) by the proteasome (ubiquitinated protein remnants and unanchored ubiquitin chains, respectively) and polyubiquitinated protein substrates of the protease.
Other DUBs may have more specialized functions. The one demonstrated example of a DUB enzyme that is crucial for the development of a multicellular eukaryote is the Drosophila fat facets UBP enzyme (5). In addition to having a maternal effect lethal phenotype, mutant fat facets (faf) flies have a specific defect in eye development characterized by extranumerary photoreceptor cells. Interestingly, partial loss-of-function alleles of a proteasome subunit gene suppress the faf developmental defect, suggesting that the Faf protein normally functions to antagonize the degradation of one or more key regulators of eye development (8).
The simple eukaryote D. discoideum is an excellent model system because its growth is independent of development. As the vegetative amoebae starve, they enter an initial stage of development wherein they stop dividing and begin secreting the cell density sensing factor, CMF (9 -13). When there is a high concentration of starving cells, as indicated by high levels of CMF, the cells enter another stage of development where relayed pulses of cAMP cause an increase, from a basal level, of the transcription of genes encoding proteins such as the cAMP receptor, cAR1, and a phosphodiesterase (PDE) that causes extracellular cAMP levels to return to a base-line level in the interval between the pulses of cAMP (see Refs. 14 -17 for review). The pulses of cAMP also function as a chemoattractant that causes the cells to aggregate (reviewed in Ref. 18). The aggregate, which is held together by adhesion proteins such as gp80, subsequently develops into a fruiting body consisting of a mass of spore cells supported on a column of stalk cells (see Refs. 19 -24 for review). The entire developmental process takes only 24 h. In Dictyostelium, ubiquitin genes have been identified, and ubiquitin mRNA species were shown to be developmentally regulated, suggesting a role for ubiquitin in development (25,26). In addition, a ubiquitin-conjugating enzyme, UBC1, has been implicated in Dictyostelium development (27).
Here we demonstrate that the D. discoideum UbpA protein is a deubiquitinating enzyme, that it is a functional homolog of yeast Ubp14 and shares the highly restricted substrate specificity found previously for Ubp14 (7) and mammalian isopeptidase T (1,28,29), and that the UbpA enzyme is crucial for development. We used a random mutagenesis protocol (30) and isolated ubpA as a mutant with defective aggregation. Mutants lacking UbpA grow and enter the CMF secretion stage of development normally, but these cells fail to reach the stage where pulses of cAMP normally increase the transcription of genes required for further development, such as those encoding cAR1, PDE, and gp80. It is intriguing that such a specific developmental defect results from the loss of a DUB enzyme with a very general but nonessential function in ubiquitin-dependent proteolysis. These findings suggest that there is a central role for the ubiquitin-proteasome pathway in setting the proper expression level of specific regulatory factors controlling progression between stages of Dictyostelium development.

EXPERIMENTAL PROCEDURES
Growth and Manipulation of Dictyostelium, Yeast, and Escherichia coli Cells-D. discoideum Ax4 cells were grown as described previously (31). The uracil-lacking strain DH1 (a gift of Peter Devreotes, Johns Hopkins University. Baltimore), a pyr5-6 derivative of Ax4, was grown in HL5 medium (32) supplemented with 20 g/ml uracil. Cells were developed on filters (31) or starved in shaking cultures as described (13). Shaking cultures that were pulsed with cAMP received pulses to 30 nM every 6 min between 1 and 6 h of starvation. Low cell-density assays for prestalk and prespore cells were carried out as described by Clay et al. (33). Time lapse videomicroscopy, motility assays, and trypan blue exclusion were done as described in Yuen et al. (13).
Disruption, Cloning, and Sequencing of ubpA-Restriction enzyme-mediated integration (REMI) was carried out following Kuspa and Loomis (30)  The largest cDNA clone was 3 kb in length and contained an open reading frame from one end through to a poly(dA) region at the other end. RACE-PCR was used to obtain additional sequence at the 5Ј-end (10,36). First strand cDNA synthesis from 5 g of total RNA was carried out using the gene-specific antisense primer 5Ј-GATTCTGCAT-TCCAACTGACG-3Ј with the Life Technologies, Inc. SuperScript preamplification system following the manufacturer's protocol. Primers and dNTPs were removed by using a Microcon-100 microconcentrator. A poly(dC) tail was added to the cDNA using terminal deoxynucleotidyltransferase (U. S. Biochemical Corp.). The dC-tailed cDNA was amplified by PCR, carried out for 30 cycles (1 min at 94°C, 1 min at 48°C, 2 min at 72°C), with the 5Ј-RACE Anchor Primer (Life Technologies, Inc.) and a nested gene-specific primer, 5Ј-ACTTTATCTCTTCATCC-3Ј, using the GeneAmp kit (Perkin-Elmer). Product (0.1 l) from this PCR was used as template for nested amplification, carried out for 17 cycles (1 min at 94°C, 1 min at 49°C, 1 min at 72°C), with the Universal Amplification Primer (Life Technologies, Inc.) and the antisense primer, 5Ј-CCACCATTCTCTATAGTTGG-3Ј.
The 5Ј-RACE fragment extended the cDNA sequence an additional 135 bp, which included the first 4 bp of the open reading frame. The cDNA sequence matched the sequence obtained from the genomic clones except for the 5Ј-end of the cDNA (exon 1; Fig. 1A), which was not present on the genomic fragment of pM7E, suggesting that a deletion occurred during the cloning of this plasmid. This was confirmed by PCR of genomic DNA from wild-type cells using a primer specific for a region upstream of the presumptive deletion and a gene-specific (exon 2) antisense primer. The sequence of the genomic fragment obtained by PCR matched the cDNA and 5Ј-RACE sequence. Plasmid pM7E thus contains a deletion of 350 bp including exon 1.
Construction of Plasmids for the Expression of UbpA and Ubp14 -For ubpA cloning, the ubpA coding sequence was amplified by PCR from a gt11 cDNA clone that contained the entire ubpA open reading frame. PCR was carried out for 30 cycles (45 s at 95°C, 45 s at 45°C, 3 min at 72°C) using a AmpliTaq (Perkin Elmer, Branchburh, NJ)/Pfu enzyme (Stratagene) mix (37) and the primers 5Ј-CGGATCCAATGGAATTAT-TCCCAGAATTAAAAAATATTAAAGTACC-3Ј and 5Ј-GCTCTAGAAAT-TAAAATTTAATTTAGTTGTC-3Ј, which correspond to the 5Ј-and 3Јends of the ubpA coding region, respectively. The PCR product was digested with BamHI and XbaI (restriction sites which were built into the 5Ј-and 3Ј-primers) and subcloned into BamHI, XbaI-cut pYES2 (Invitrogen, Carlsbad, CA), a yeast expression vector, to generate pYEU1. The ubpA insert of pYEU1 was removed by digestion with BamHI, blunt-ended using Klenow, and digested with XbaI, and then subcloned into similarly cut pDXA-3H, a Dictyostelium expression vector (38). For expression in E. coli, the complete ubpA coding sequence, contained on a gt11 cDNA clone, was amplified by 30 PCR cycles (45 s at 95°C, 45 s at 55°C, 3 min at 74°C) using PrimeZyme (Biometra, Tampa, FL) and the primers 5Ј-CGGGATCCTAATGGAATTATTCCCA-GAATTAAAAAATATTAAAGTACC-3Ј and 5Ј-CGGGGTACCTAAAT-TAAAATTTAATTTAGTTGTC-3Ј. The PCR product was blunt-ended using Klenow, digested with BamHI, and cloned into pGEX-3 (Amersham Pharmacia Biotech) that was cut with EcoRI, blunt-ended using Klenow, and digested with BamHI to generate pGEX-ubpA. To express the yeast ubp14 in Dictyostelium, the ubp14 cDNA obtained by digesting pGEX-UBP14 (7) with XhoI, blunt-ended using Klenow, followed by digesting with BamHI, was cloned into similarly cut pDXA-3H.
Determination of Isopeptidase Activity in E. coli Extracts-To induce expression of GST-Ubp14 and GST-UbpA, JM101 bacterial cells transformed with pGEX-UBP14 and pGEX-ubpA plasmid DNAs, respectively, were grown to an A 600 of 0.7 in LB ϩ 100 g/ml ampicillin at 30°C. After addition of 1 mM isopropyl-1-thio-b-D-galactopyranoside and incubation at 30°C for 3 h, cells (100 ml) were collected by centrifugation and resuspended in 3 ml of phosphate-buffered saline (PBS). The suspension was sonicated with a microtip attachment until clarified; 1/20 volume of 10% Triton X-100 was added, and the suspension was gently mixed by inversions of the tube. Cell debris was removed by centrifugation at 14,000 ϫ g for 10 min, and the extracts were adjusted to 0.5 mM EDTA, 10 g/ml aprotinin, 5 g/ml pepstatin A. Expression of fusion proteins was checked by PAGE and Coomassie Blue staining. Isopeptidase activity was assayed in 50 mM Tris-HCl, pH 7.3, 2 mM dithiothreitol in a total volume of 20 l following Amerik (7). Reaction mixtures contained 2 l of bacterial extracts and 0.5 g of Ub-Ub or Ub-Ub⌬GG. After 1 h at 37°C, aliquots were removed and analyzed by anti-ubiquitin immunoblotting.
RNA Analysis-RNA isolation, electrophoresis, and transfer to Duralon UV membrane (Stratagene) were performed as described (31). A 250-bp PvuII-EcoRI fragment (described above) and a 400-bp BspDI fragment from genomic DNA clones pM7E and pM7N, respectively, were used to prepare ubpA-specific DNA probes labeled with [ 32 P]dCTP by random hexamer-primed DNA synthesis (Amersham Pharmacia Biotech). Additional probes were prepared using the inserts of the following genomic or cDNA clones. Jakob Franke (Columbia University, New York) provided plasmid pDE-0.9, containing the 0.9-kb BstBI-EcoRI fragment of the cyclic nucleotide phosphodiesterase cDNA (39). Chi-Hung Siu (University of Toronto, Toronto, Canada) provided the gp80 clone which carried a 1.0-kb EcoRI cDNA insert (40). Alan Kimmel (National Institutes of Health, Bethesda) provided the cAR1 clone which carried a 1.3-kb EcoRI cDNA insert. Hybridization was performed at 60°C in 0.125 M Na 2 HPO 4 , pH 7.2, 0.25 M NaCl, 5% SDS, 1 mM EDTA, 10% PEG (M r 8,000). Blots were washed with 50 mM Na 2 HPO 4 , pH 7.2, 0.5% SDS twice at room temperature for 30 min each and then at 60°C for 30 min. Molecular size standards were from the Life Technologies, Inc., 0.24 -9.5-kb RNA ladder.
Antibody Production and Purification-An anti-UbpA polyclonal serum was prepared using a bacterially expressed UbpA fusion protein.
The primer 5Ј-CGGGATCCACCAACAGAAAAATCACG-3Ј and the antisense primer 5Ј-CGGGATCCAAGATGGACGAG-3Ј were used in a PCR to amplify cDNA encoding a 438-residue segment of UbpA (residues 13-450; Fig. 1). PCR was carried out with Pfu enzyme for 35 cycles (1 min at 94°C, 1 min at 45°C, 4 min at 74°C). The primers added BamHI sites at the ends of the amplified region. Following digestion with BamHI, the 1.3-kb PCR product was cloned in frame into the BamHI site of the expression vector pET15b (Novagen, Madison, WI). After the construct was verified by DNA sequencing, protein was expressed in the bacterial strain BL21(DE3)pLysS (Novagen) and purified on His-Bind resin (Novagen) following the manufacturer's protocol. The recombinant protein was isolated further by 15% SDS-PAGE followed by electroelution of the excised protein-containing gel slice and lyophilization (33). A rabbit was then immunized by Cocalico Biologicals, Inc. (Reamstown, PA) with 30 g of fusion protein in complete Freund's adjuvant injected into the popliteal lymph nodes followed 30 days later with 150 g in incomplete Freund's adjuvant injected subcutaneously. Serum was collected 10 days after the boost, and antibodies were purified with an EZ Sep kit (Amersham Pharmacia Biotech). Monospecific antibodies against the recombinant UbpA protein were obtained from the crude antibody preparation by blot affinity purification essentially as described by Talian et al. (41); elution of the antibody bound to the antigen/polyvinylidene difluoride strip was done by use of low pH shock. Affinity purified antibody was immediately neutralized, diluted 4-fold with 50 mM Tris, pH 7.9, 150 mM NaCl, 0.05% Tween 20, 0.05% NaN 3 (buffer A), then concentrated and stored at 4°C at a 1:10 dilution relative to the starting antiserum volume.
Western Blots-Dictyostelium cells harvested at various developmental time points were directly solubilized with 2% SDS sample buffer and heated at 100°C for 5 min. For all Western blots, the protein from 3 ϫ 10 5 cells per lane was resolved on 15 or 17.5% polyacrylamide-SDS gels and was transferred to polyvinylidene difluoride membrane following Towbin et al. (42). For the analysis of crude fractionated subcellular samples, 10 9 vegetative Ax4 cells were first collected by centrifugation and resuspended to 3 ϫ 10 8 cells/ml in ice-cold MESES buffer (20 mM MES, pH 6.5, 1 mM EDTA, 0.25 M sucrose). Cells were then homogenized with a tight-fitting Dounce and centrifuged at 3,000 ϫ g for 5 min to pellet nuclei. Post-nuclear supernatants (5 ml) were centrifuged at 10,000 ϫ g for 10 min, and the resulting supernatant was spun again at 200,000 ϫ g for 30 min. Blots were blocked with 5% low-fat powdered milk in PBS (8 mM Na 2 HPO 4 , 1 mM KH 2 PO 4 , pH 7.4, 0.14 M NaCl, 3 mM KCl) for 30 min, washed 3 min with buffer A, and then incubated for 1.5 h with affinity purified UbpA antibody diluted 1:50 in buffer A at 25°C. After a brief wash with buffer A, blots were incubated with 0.230 Ci/ml 125 I-protein A (Amersham Pharmacia Biotech) in 20 mM Tris, 150 mM NaCl, 0.05% Tween 20, 1% bovine serum albumin, 0.05% NaN 3 (buffer B) for 2.5 h at 25°C and then washed with buffer B five times (10 min each). Autoradiography was done using a Cronex lightning plus amplifying screen (NEN Life Science Products) and Kodak XAR-5 film.
To detect ubiquitin, Western blots were blocked for 1 h in PBS, 0.1% Tween 20 and then incubated with a 1:2000 dilution of anti-ubiquitin antibodies in PBS, 1% Tween 20, 1% Nonidet P-40, 0.1% SDS for 1 h. To enhance resolution of low molecular mass proteins, a Tricine gel system (43) was used. Proteins were transferred to Immobilon-P membranes (Millipore); the filters were then boiled in water and incubated with anti-ubiquitin antibodies. All blots used anti-ubiquitin antibodies that had been affinity purified on a ubiquitin affinity resin and were provided by C. Pickart or S. Swaminathan. Bound antibody was detected with the ECL Western blotting kit (Amersham Pharmacia Biotech) following Pampori et al. (44) using horseradish peroxidase-conjugated donkey anti-rabbit antibodies (Amersham Pharmacia Biotech). Molecular size standards were from the Life Technologies, Inc., 10-kDa protein ladder or ubiquitin chain standards.
Determination of Isopeptidase Activity in Yeast Extracts-Overnight cultures of ubp14⌬ cells transformed with pUBP14 or pYEU1 plasmids were centrifuged and resuspended in 3 volumes of disruption buffer (20 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 , 1 mM EDTA, 5% glycerol, 1 mM dithiothreitol, 0.3 M ammonium sulfate, 200 g/ml aprotinin, 100 g/ml pepstatin). Cells were mixed with 4 volumes of chilled glass beads and vortexed five times for 45 s each, leaving cells on ice for 1 min between vortexings (35). The extracts were centrifuged, and supernatants were stored at Ϫ20°C as 50% glycerol solutions. Assays of isopeptidase activity were conducted at 37°C in buffer containing 50 mM Tris-HCl, pH 7.3, 2 mM dithiothreitol in a total volume of 20 l. Reaction mixtures contained 2 l of yeast extract (160 ng of protein) and 1 g of Ub-Ub. 3-l aliquots were removed at different times and analyzed by antiubiquitin immunoblotting.
Immunofluorescence Staining-All operations were at room temperature unless specified. Vegetative DH1 and ubpA cells were resuspended in PBM (20 mM KH 2 PO 4 , pH 6.1, 1 mM MgCl 2 , 10 M CaCl 2 ) and allowed to settle and attach to 18 ϫ 18 mm glass coverslips for 15 min. After a brief wash with PBM, cells were fixed with 95% ethanol for 10 min on ice, air-dried, and placed in PBST (PBS, 0.1% Tween 20) for 15 min. Prior to use, antibodies were diluted 1:200 in PBST and preadsorbed as follows. Six coverslips containing confluent, fixed ubpA cells were consecutively incubated with 60 l of antibody for 15 min each. Experimental coverslips were then incubated for 1 h with the preadsorbed antibody, washed twice (5 and 15 min) with PBST, and incubated with 1:200 fluorescein isothiocyanate-conjugated goat anti-rabbit (Cappell/Organon Teknika, Durham NC) for 1 h. Slides were then washed as above and mounted as described previously (45). Cells were examined and photographed with a Microphot FX (Nikon, Melvillie, NY) and Kodak Tmax p3200 film (Kodak, New Haven, CT) using a ϫ 60 oil immersion objective.
Assays for cAMP Relay Response, Chemotaxis, and EDTA-resistant Cell-Cell Adhesion-The cAMP relay response was measured following Van Haastert (46). Briefly, 5 ϫ 10 6 cells in 0.1 ml were stimulated with 10 M 2Ј-deoxy-cAMP (a functional cAMP analog) in the presence of 10 mM dithiothreitol. At 0 and 3 min after stimulation, the cells were lysed, and total cAMP was measured using the Amersham Pharmacia Biotech cAMP [ 3 H] assay system. For cAMP binding assays, cells starved on filters or in shaking cultures were harvested, washed in phosphate buffer, resuspended to 1 ϫ 10 8 cells/ml in ice-cold phosphate buffer, and assayed for [ 3 H]cAMP binding in 90% ammonium sulfate (47). Chemotaxis was determined by the single spot small population assay using a modification of Soll and Mitchell (48). Cells starved for 4 h were washed twice in MES-PDF (8.2 mM MES, pH 6.5, 20 mM KCl, 0.6 mM MgSO 4 ), and the final pellet was vortexed into a dense suspension. Small droplets of cells were placed on 2% agarose containing 0 or 1 M cAMP. Diameters of the droplets were measured using a Nikon microplot Fx with a ϫ 4 objective and a video system (allowing the drop diameter to be measured on the monitor) just after plating (zero time) and at 3 h. EDTA-resistant cell-cell adhesion was monitored as described by Soll and Mitchell (48).

RESULTS
UbpA Has Similarity to a Deubiquitinating Enzyme-A mutant generated by an insertion into the ubpA gene was found in a screen of REMI (restriction enzyme-mediated integration) transformants as cells that never aggregated when grown on bacteria and then allowed to starve on agar plates. When grown in axenic culture and then starved on filters, wild-type cells begin to visibly aggregate (ripple stage) 6 h after starvation. Under identical conditions, ubpA cells required at least 24 h to form ripples. After 3 days, ubpA cells formed a small number of aberrant structures but never formed fruiting bodies.
Genomic ubpA DNA was cloned from the mutant strain (designated M7) in two fragments. Plasmid pM7E contained an EcoRI fragment containing the insertion site and plasmid pM7N contained an NsiI fragment from one side of the insertion (Fig. 1A). The pM7E plasmid was linearized with EcoRI and used to transform a fresh wild-type host strain. The resulting gene disruption by homologous recombination successfully recapitulated the delayed aggregation phenotype.
A probe prepared from the region flanking the insertion site was used to screen a cDNA library, and several cDNA clones were isolated. These clones were sequenced as were the genomic fragments from pM7E and pM7N. Sequence analysis revealed a single long open reading frame interrupted by one intron (Fig. 1A). The deduced amino acid sequence predicts a protein of 837 residues with a molecular mass of 95 kDa (Fig.  1B). The sequence has 44% identity (63% similarity) to human isopeptidase T (ISOT1), and 29% identity (50% similarity) to the yeast deubiquitinating enzyme Ubp14 (1,7,29,49). The regions of similarity include the Cys and His boxes, which are found in all members of the UBP family. These data therefore suggest that UbpA is a UBP enzyme.
UbpA Is a Deubiquitinating Enzyme-To test the potential deubiquitinating activity of UbpA, the protein was expressed in bacteria as a glutathione S-transferase (GST) fusion protein.
Enzymatic activity was assayed in bacterial cell extracts using either a Lys-48-isopeptide bond-linked diubiquitin substrate (Ub-Ub) or a derivative of this substrate that lacks the two C-terminal glycine residues of the unanchored ubiquitin moiety (Ub-Ub⌬GG). Extracts prepared from GST-UbpA-expressing E. coli cells efficiently cleaved the diubiquitin substrate to ubiquitin monomers (Fig. 2). The activity was comparable to that of extracts from bacterial cells expressing a GST-Ubp14 fusion. Recently, yeast Ubp14 and its human homolog isopeptidase T were shown to have very similar substrate specificities (7). Most strikingly, neither enzyme can cleave the truncated diubiquitin, Ub-Ub⌬GG. We conclude that the Dictyostelium UbpA enzyme, yeast Ubp14, and human isopeptidase have closely related or identical substrate specificities, with activity apparently restricted to ubiquitin chains with a normal, unanchored C terminus.
UbpA Is a Functional Homolog of the Yeast Ubp14 Enzyme-The finding that UbpA is a deubiquitinating enzyme prompted us to compare intracellular levels of ubiquitin and ubiquitinprotein conjugates in wild-type and ubpA cells by anti-ubiq- uitin immunoblot analysis. At all stages of development, ubpA cells were found to contain small ubiquitinated species that were absent in wild-type cells, the most prominent of which comigrated with unanchored Lys-48-linked ubiquitin oligomers ( Fig. 3 and data not shown). In addition, ubpA cells contained increased levels of Ͼ80-kDa ubiquitin-containing species. Yeast ubp14 cells show a similar accumulation of ubiquitin oligomers (7). However, the yeast mutant does not display an obvious accumulation of high molecular mass ubiquitinated species, and the unanchored ubiquitin tetramer accumulates to a higher relative level than any other species.
The high degree of similarity in sequence and in substrate specificity between UbpA and Ubp14 and the comparable accumulation of unanchored ubiquitin chains in Dictyostelium ubpA and yeast ubp14 cells led us to investigate whether the Dictyostelium enzyme could functionally replace Ubp14 in yeast cells. The ubpA gene was subcloned into the yeast expression vector pYES2, and the resulting plasmid was transformed into yeast ubp14 cells. ubp14 cells have a number of characteristic phenotypic abnormalities, including hypersensitivity to amino acid analogs and defects in the degradation of a variety of ubiquitin system substrates (7). Anti-ubiquitin immunoblot analysis of ubp14 cells expressing ubpA revealed that the levels of unanchored ubiquitin chains were reduced to wild-type levels (Fig. 4A). In addition, these cells grew indistinguishably from wild-type cells on plates containing the arginine analog canavanine (Fig. 4B). We previously found that cell-free extracts prepared from ubp14 mutants display a greatly reduced activity against the Lys-48-linked diubiquitin test substrate (but not against Ub-Ub⌬GG) relative to wildtype cells, indicating that Ubp14 is the major yeast DUB capable of cleaving unanchored ubiquitin chains in yeast cells (7). Transformation of ubp14 cells with the UbpA expression plasmid restored activity against diubiquitin in cell-free extracts to wild-type levels (data not shown).
We also investigated whether the yeast ubp14 could restore wild-type development in ubpA cells. Expression of the ubp14 cDNA downstream from the Dictyostelium act15 promoter complemented the aggregation defective phenotype of ubpA cells; fruiting bodies formed but they were fewer in number and somewhat aberrant when compared with wild-type (data not shown). However, the phenotype of these cells was similar to ubpA cells that contained the ubpA cDNA expressed downstream from the act15 promoter. These data and those of the previous section led us to conclude that the D. discoideum UbpA and S. cerevisiae Ubp14 enzymes are functional homologs.
UbpA Is Present in Vegetative and Developing Cells-The temporal transcription pattern of ubpA was examined by Northern analysis. A 2.9-kb ubpA mRNA was present in vegetatively growing cells (0 h), increased in level during early development to a maximum at 7.5 h, and continued to be present through the remainder of development (Fig. 5A). ubpA RNA was not present in ubpA cells (data not shown). Immunoblot analyses indicated that a 95-kDa UbpA protein was present in wild-type cells (Fig. 5B). The pattern of UbpA protein expression in wild-type cells was similar to the mRNA pattern. A relatively high level of the protein accumulated in growing cells; after the onset of starvation, UbpA levels initially dropped, then increased to maximum levels at 10 -15 h, and continued to be present throughout the remainder of development (Fig. 5B). The protein was not detected in the ubpA mutant or when preimmune serum was used in immunoblot analyses of either wild-type or mutant cells (data not shown).

FIG. 4. Expression of Dictyostelium UbpA in yeast ubp14⌬ cells restores the wild-type phenotype.
A, anti-ubiquitin immunoblot analysis of wild-type and ubp14 mutant cells containing various expression plasmids. Accumulation of unanchored ubiquitin chains is suppressed to wild-type level in ubp14 mutants carrying the UbpA-expressing plasmid pYEU1. Positions of free ubiquitin and unanchored ubiquitin chains are indicated. B, growth on the media containing 1.0 mg/ml canavanine sulfate. Wild-type and ubp14 mutant cells carrying the indicated plasmids were streaked onto canavanine plates and incubated for 3 days at 30°C. As in A, wild-type yeast MHY501 cells carried the vector YEplac195 and the ubp14 strain MHY840 carried YE-plac195, YEplac195UBP14, or pYEU1. Cells were grown in media containing 3% galactose.
Expression of ubpA RNA and protein in the parental DH1 strain was similar to that seen in the wild-type cells (Ax4) (data not shown).
To determine the subcellular location of UbpA, vegetative cells were lysed and extracts were fractionated by differential centrifugation, and the fractions were examined by anti-UbpA immunoblotting. UbpA was present in the supernatant after a low speed spin to remove nuclei, and a medium speed spin to pellet large organelles. The UbpA protein was still present in the supernatant when the second supernatant was centrifuged for 6 ϫ 10 6 g/min (data not shown). The subcellular and cell type distribution of UbpA was further characterized by immunofluorescence analysis of fixed cells. Vegetative and developing DH1 cells showed a diffuse staining throughout the cytoplasm and nucleus, and all cells were stained (data not shown). Only a weak, apparently nonspecific staining was observed for vegetative ubpA mutant cells. No staining of either wild-type or mutant cells was seen with preimmune sera. The fractionation and immunofluorescence data thus suggest that UbpA is a soluble protein present in all cells throughout growth and development.
Growth and Initial Responses to Starvation Appear to be Normal in ubpA Cells-We next examined the general phenotype of growing and starving ubpA cells. When grown axenically, DH1 parental and ubpA cells doubled at essentially the same rate. When grown on bacterial lawns, isolated ubpA cells formed plaques that increased in diameter at approximately the same rate as DH1 cells. Vegetative and starved ubpA cells were indistinguishable from wild-type cells by phase-contrast light microscopy, and trypan blue staining showed that lack of UbpA does not affect the viability of cells. Using time lapse videomicroscopy, we determined that both growing and starved ubpA cells have normal motility. As in wild-type cells, cell division ceases in mutant cells upon starvation (data not shown).
We found that ubpA cells secrete normal amounts of biologically active CMF (data not shown). In addition, ubpA cells were able to express both the CP2 prestalk and the SP70 prespore markers within 18 h when starved at low cell density in the presence of CMF and then cAMP and that there was a normal percentage of the two cell types. This indicated that starved ubpA cells are able to differentiate into the proper initial ratio of cell types, sense CMF, sense the cAMP added at 6 h, and to express mRNAs and proteins many hours after starvation.
Expression of cAMP Pulse-induced Genes Is Altered in Cells That Lack UbpA-Since all the initial responses to starvation that we examined are apparently normal in ubpA cells, we examined whether there was a successful transition to the next developmental stage. This stage is marked by chemotaxis to cAMP, cAMP relay, and acquisition of EDTA-resistant cell-cell adhesion. To assay for chemotaxis, we used the single spot small population method. The ability of ubpA cells to chemotax is somewhat impaired (Table I). We also examined the ability of cells to relay a pulse of cAMP. In the parental DH1 cells, cAMP accumulation in response to a pulse of a cAMP analog was similar to that previously seen for wild-type cells (50), whereas ubpA cells exhibited a 3.4-fold less cAMP production. Also during early aggregation, EDTA-resistant cell-cell binding sites begin to appear, enabling the cells to form tight multicellular aggregates (51). The formation of these binding sites is mediated by gp80, a cell-surface glycoprotein. EDTA-resistant cell-cell adhesion in wild-type cells begins after 3 h and peaks at roughly 12 h (see Ref. 54; Fig. 6). For ubpA cells, EDTAresistant cell-cell adhesion only first appears at 24 h and reaches a peak at 36 h (when the cells are aggregating) (Fig. 6). The data thus suggest that during the time when wild-type cells are aggregating, ubpA cells show an abnormally low level of both chemotaxis to cAMP and cAMP relay and undetectable EDTA-resistant cell-cell adhesion.
Northern blot analysis was then used to determine whether the above defects correlated with lowered expression of cAR1, PDE, and gp80. In wild-type cells, cAR1 and PDE mRNAs are present during growth and early development, whereas a basal level of gp80 transcription is initiated soon after the onset of starvation. The levels of all three mRNAs greatly increase during the cAMP pulse-mediated aggregation and then decline as development proceeds (see Refs. 52-54; Fig. 7A). In ubpA cells, all three mRNAs were also expressed at a low level during  These data suggest that the defective aggregation of ubpA cells is associated with a failure to increase expression of some aggregation-specific mRNAs that normally occurs in response to pulses of cAMP. UbpA Is Required for Generating, but Not Sensing, Pulses of cAMP-Two possible explanations for why the ubpA cells do not up-regulate cAMP pulse-induced genes are either the cells are insensitive to cAMP pulses or they are not producing the pulses. To determine if ubpA cells can respond to pulses of cAMP, cells were starved in shaking culture in the presence or absence of exogenous pulses of cAMP. After 6 h, the cells were harvested and plated on non-nutrient agar. As shown in Fig. 8, ubpA cells starved in the absence of pulses do not aggregate, whereas the same cells starved in the presence of exogenous cAMP pulses formed aggregates that developed to (but not beyond) the early culminate stage. DH1 parental cells starved in the presence or absence of cAMP pulses formed fruiting bodies ( Fig. 8 and data not shown). The ability to aggregate in response to exogenous cAMP pulses was paralleled by an enhanced accumulation of aggregation-specific mRNAs under the same conditions (Fig. 7B). Binding of cAMP to both DH1 and ubpA cells was increased similarly as well (Fig. 9A); therefore, the extra cAR1 mRNA induced in pulsed ubpA cells was translated, resulting in an up-regulation of the cAR1 receptor.
Finally, we directly examined whether the defective aggregation of ubpA cells was due to a defect in the ability of these cells to generate pulses of cAMP (Fig. 9B). The parental DH1 cells starved in the absence of exogenous cAMP pulses will generate a relay pulse of cAMP in response to a cAMP stimulus; this response increases for cells exposed to cAMP pulses. ubpA cells have a significantly attenuated capacity for constitutive and stimulated cAMP production, but when exposed to exogenous pulses of cAMP, the cAMP product increases strongly but is still significantly reduced from wild-type levels (Fig. 9B). This suggests that ubpA cells have a partial defect in cAMP-induced cAMP production, which can account for their failure to aggregate normally. DISCUSSION In this report, we have characterized the Dictyostelium UbpA protein, a new member of the UBP family of deubiquitinating or DUB enzymes, and we have shown that deletion of UbpA from cells, while of little consequence during growth, prevents cell aggregation and fruiting body development. UbpA bears strong sequence similarity to the yeast Ubp14 and mammalian isopeptidase T enzymes and shares with these well characterized enzymes a specificity for unanchored polyubiq-uitin chains. Moreover, we demonstrate by cross-species complementation that UbpA and Ubp14 are in fact functional homologs. This is the first time a DUB enzyme with a well characterized substrate specificity has been directly implicated  (0), at 3 h of development, and at 7, 11, and 14 h when development of DH1 had reached the ripple, tight aggregate, and early culminate stages, respectively. In addition, RNA was prepared from ubpA cells at 24, 36, and 42 h when the mutant cells were at the ripple, loose aggregate, and tipped aggregate stages, respectively. RNA (5 g) was resolved by electrophoresis through 1.2% agarose/formaldehyde gels, blotted to nylon membranes, and hybridized with a probe specific for cAR1 cAMP receptor (cAR1), cAMP phosphodiesterase (PDE), or transcripts encoding the cell-cell adhesion molecule gp80. B, total RNA was prepared from parental DH1 and mutant ubpA cells that were starved for 6 h in shaking culture in the presence (P) or absence (S) of exogenous pulses of cAMP. RNA (5 g) was resolved by electrophoresis through 1.2% agarose/formaldehyde gels, blotted to nylon membranes, and hybridized with a probe specific for cAR1, PDE, or gp80. in multicellular development. A central question raised by these data is how a deubiquitinating enzyme whose activity is apparently restricted to free ubiquitin chains can be crucial to a specific stage (or stages) in Dictyostelium development.
The UbpA Enzyme Is Responsible for Polyubiquitin Chain Disassembly in Vivo-UbpA belongs to the Ubp14/isopeptidase T subfamily of DUBs. There are three known isozymes of isopeptidase T in humans, and UbpA is over 40% identical to the two that have been studied in vitro, IsoT-1 and IsoT-2 (1,29). Although the conservation with yeast Ubp14 is weaker, sequence identity (ϳ29% overall) extends over the entire lengths of the two proteins (7). Whereas all UBPs share conserved sequences in the Cys and His boxes and a few short stretches between these two motifs, there is little further similarity among them as a group, particularly in the generally large regions N-terminal to the Cys box (3).
The conservation of primary sequence among members of the Ubp14/isopeptidase T subfamily is paralleled by their comparable substrate specificities. A dramatic reduction in the rate of disassembly of a Lys-48-linked diubiquitin molecule by both isopeptidase T (3) and Ubp14 (7) is observed when the last two glycines are deleted from the proximal ubiquitin monomer (Ub-Ub⌬GG). The same block to disassembly is also observed with UbpA ( Fig. 2) but not observed with other DUB enzymes (e.g. see Ref. 7). Most telling, cross-species complementation experiments verify that UbpA and Ubp14 are functional homologs (Fig. 4); analogous data had been obtained when human isopeptidase T was expressed in yeast ubp14 cells (7). The functional interchangeability of Dictyostelium, human, and yeast enzymes indicates that the more detailed mechanistic analysis done in yeast cells can be used to infer how other members of the Ubp14 subfamily participate in the ubiquitin pathway in their respective cell types.
Consistent with this last point, we observe a comparable profile of ubiquitin-containing species in yeast ubp14 (7) and in Dictyostelium ubpA mutants (Fig. 4). The bulk of these species comigrate with Lys-48-linked ubiquitin chain standards and, in the case of ubp14 cells, were shown to share the same isoelectric point as free ubiquitin monomer, indicating that the majority of ubiquitin conjugates that accumulate abnormally in these mutants are unanchored ubiquitin chains. Interestingly, ubpA cells, unlike yeast ubp14 cells, accumulate ubiquitinated material of very high molecular mass in addition to the easily identified ubiquitin oligomers (Fig. 3). These larger species may be either very large free polyubiquitin chains or extensively ubiquitinated proteins (or both). If the former were true, it might reflect a more active or more processive polyubiquitin chain synthesizing machinery in Dictyostelium than in yeast, resulting in greater average chain lengths. This view is consistent with the observation that yeast ubp14 cells accumulate particularly high levels of ubiquitin tetramer and only small amounts of larger species (7), whereas this bias is not seen in the Dictyostelium mutant.
On the other hand, it is very likely (see below) that, like yeast ubp14 cells, the ubpA mutant is impaired in proteasome-medi- ated protein degradation, which can lead to the accumulation of high molecular mass polyubiquitinated proteins. A more severe proteolytic defect in the Dictyostelium mutant than in the corresponding yeast mutant might account for the difference in conjugated accumulation between taxa. However, this would predict a considerable growth impairment in ubpA cells given that proteasomes are essential for viability, and this was not observed. Alternatively, yeast might have a more potent protein deubiquitination potential than Dictyostelium, resulting in more rapid deubiquitination of the polyubiquitinated substrates that fail to get targeted to the proteasome. These various explanations are not mutually exclusive, and in any event, the similarities in unanchored polyubiquitin chain accumulation between the yeast and Dictyostelium mutants remain striking.
A Specific Developmental Defect Is Associated with Loss of UbpA-The absence of UbpA affects neither the growth rate of Dictyostelium cells nor the gross appearance or motility of vegetative cells. UbpA does not appear to be required for sensing starvation since ubpA cells exhibit certain normal responses to starvation such as CMF secretion, a decrease in motility, the ability to sense CMF, initiation of a basal level of gp80 transcription, and differentiation into CP2-positive prestalk, SP70-positive prespore, and null cells. Even though the absence of UbpA delays aggregation for several days, UbpA is not required for several aspects of development that occur 10 -15 h after onset of starvation, such as sensing high continuous levels of cAMP, or prestalk, and prespore protein expression.
In growing wild-type cells, certain cAMP pulse-induced genes, such as cAR1, are expressed at low levels (53). This presumably allows the cells to have a small amount of the cAMP pulse signal transduction machinery in place when they encounter starvation conditions, so that when a high level of CMF informs the cells that there is a local high density of starved cells (11,12), the cells can begin to relay pulses of cAMP. A bootstrap process then permits the pulses of cAMP to up-regulate the cAMP signal transduction system to a level that allows the cells to aggregate. The bootstrap routine in Dictyostelium cells is such that a small random pulse of extracellular cAMP can trigger cells to relay the pulse and to produce more of the proteins required to sense and relay the cAMP pulses. Each time a cell sees a cAMP pulse during the bootstrap phase of development, it will respond by generating a slightly greater cAMP pulse, which provides an enhanced signal that causes nearby cells to increase production of the cAMP relay machinery, and so forth.
Like wild-type cells, ubpA cells have an initial low level of cAR1 transcription, and their partial ability to chemotax toward cAMP and relay a pulse of cAMP indicates that they do have a functional cAMP signal transduction mechanism. However, the cAMP pulse-induced increase in the accumulation of mRNAs encoding proteins such as cAR1, PDE, and gp80 does not occur. The ability of exogenously added pulses of cAMP to rescue this bootstrap process in the ubpA cells so that they are able to increase the accumulation of pulse-induced mRNAs, increase cell-surface cAR1, increase the cAMP pulse relay, and aggregate suggests that the lack of UbpA does not cause a block in the signal transduction machinery that senses the pulses of cAMP and increases the expression of the pulse-induced genes. The rescue of the various responses in ubpA cells by pulsing them with cAMP, and the observation that ubpA cells have a poor cAMP relay efficiency, thus indicates that these cells do not generate sufficiently large pulses of extracellular cAMP to elicit a normal bootstrap response. Because there is some relay of cAMP pulses in the ubpA cells, one would then predict, and we indeed observe, that the bootstrap process will be present but slower, resulting in the cells eventually aggregating. This suggests that one function of UbpA is to permit starved cells to generate a large pulse of cAMP in response to sensed cAMP early in the aggregation phase of development.
Providing exogenous pulses of cAMP to starving ubpA cells makes the cells competent to aggregate when plated onto nonnutrient agar; however, these cells do not develop beyond the aggregate stage. This result and the fact that UbpA protein accumulates to a high level just prior to and during aggregation suggest that ubpA cells are also deficient in at least one other developmental transition. Alternatively, UbpA may be required only for aggregation, but ubiquitin conjugates that accumulate in the ubpA mutant may interfere with development beyond the aggregate stage. It is possible that UbpA functions throughout development; but, in its absence, ubiquitin conjugates accumulate and block development only at specific stages.
Polyubiquitin Chain Metabolism, Proteolysis, and Development-The primary, if not exclusive, physiological role of Ubp14 in yeast is the disassembly of unanchored ubiquitin chains (7). In principle, such ubiquitin chains can be generated by DUB enzymes during 26 S proteasome action on ubiquitinprotein conjugates and/or by release from protein substrates independent of the proteasome. In addition, ubiquitin chains can be synthesized de novo from free ubiquitin monomer (see Ref. 7 and references therein). Free ubiquitin chains can inhibit ubiquitin-dependent proteolysis in vitro, presumably by competing for binding to specific polyubiquitin-binding sites on the 26 S proteasome (28,55). Our recent work in yeast strongly supports the idea that such inhibition can occur in vivo as well. Degradation of multiple ubiquitin system proteolytic substrates is inhibited in ubp14 mutants, suggesting that Ubp14 functions to keep polyubiquitin levels low enough to avoid inhibition of the proteasome (7).
We have shown here that yeast Ubp14 and Dictyostelium UbpA are functional homologs with strong similarities in sequence and in substrate specificity. Hence, UbpA is also very likely to be important for maximal rates of ubiquitin/proteasomedependent proteolysis of a broad range of substrates. Knowledge of the enzymatic specificity of the UbpA/Ubp14/isopeptidase T family of enzymes argues against the alternative hypothesis that UbpA regulates the turnover of specific protein substrates by selective protein deubiquitination. In the one other example of a UBP enzyme clearly implicated in development, the Drosophila Faf protein, the genetic evidence implied that Faf acts to antagonize proteasomal degradation of specific proteins by deubiquitinating them (8). Unlike UbpA and Ubp14, little is known about the substrate specificity of Faf.
Despite the striking accumulation of unanchored ubiquitin chains in yeast ubp14 mutants, the proteolytic defect in these cells in relatively mild. For the ubiquitin system substrates tested, inhibition was only ϳ2-5-fold (7). It is reasonable to assume that similarly modest but significant proteolytic defects will be obtained in the Dictyostelium ubpA mutant inasmuch as the yeast and Dictyostelium mutants have similar ubiquitin chain accumulation profiles and neither has a growth defect. This would imply that development in D. discoideum is extremely sensitive to even minor inhibition of the ubiquitin system. The yeast sporulation pathway, which is also induced by starvation, can be regarded as an analog of the Dictyostelium developmental pathway leading to fruiting body formation. Homozygous ubp14 ⌬ yeast diploid cells are unable to sporulate normally at 30°C (but sporulate at near wild-type frequency at 25°C), indicating that this process is also sensitive to small perturbations of ubiquitin-dependent proteolysis (7). Progression through the bootstrap phase and probably additional stages of Dictyostelium development may require overcoming inhibition by specific negative regulators or "checkpoint" factors. In the simplest model, regulated ubiquitin-dependent degradation of such factors would allow the proper coordination of the molecular and cellular events required for development.
In support of the above ideas, it was found recently that a Dictyostelium strain carrying a mutant allele of the prtC gene, which encodes a C2-like subunit of the 20 S proteasome, also suffers from specific developmental defects (see GenBank TM accession number U60168), although not at the aggregation stage. Drosophila melanogaster is known to have at least three variants of another ␣-type 20 S proteasome subunit, and two of these are specifically expressed in the testes during spermatogenesis (56). One can anticipate additional examples of ubiquitin system involvement, particularly the participation of UbpA-like enzymes, in specific developmental processes. In this regard, it is interesting that at least three isoforms of isopeptidase T exist in humans, two of which derive from tissue-specific alternative splicing and one encoded by a separate gene (see (7).