Growth Retardation in Mice Lacking the Proteasome Activator PA28γ*

The proteasome activator PA28 binds to both ends of the central catalytic machine, known as the 20 S proteasome, in opposite orientations to form the enzymatically active proteasome. The PA28 family is composed of three members designated α, β, and γ; PA28α and PA28β form the heteropolymer mainly located in the cytoplasm, whereas PA28γ forms a homopolymer that predominantly occurs in the nucleus. Available evidence indicates that the heteropolymer of PA28α and PA28β is involved in the processing of intracellular antigens, but the function of PA28γ remains elusive. To investigate the role of PA28γ in vivo, we generated mice deficient in the PA28γ gene. The PA28γ-deficient mice were born without appreciable abnormalities in all tissues examined, but their growth after birth was retarded compared with that of PA28γ+/− or PA28γ+/+ mice. We also investigated the effects of the PA28γ deficiency using cultured embryonic fibroblasts; cells lacking PA28γ were larger and displayed a lower saturation density than their wild-type counterparts. Neither the expression of PA28α/β nor the subcellular localization of PA28α was affected in PA28γ−/− cells. These results indicate that PA28γ functions as a regulator of cell proliferation and body growth in mice and suggest that neither PA28α nor PA28β compensates for the PA28γ deficiency.

The proteasome with an apparent sedimentation coefficient of 20 S is a protein-killing machine equipped with a variety of catalytic centers that presumably contribute to the hydrolysis of multiple peptide bonds in single polypeptide substrates by a coordinated mechanism (see Ref. 1 and references therein). The 20 S proteasome is a barrel-like particle with a molecular mass of ϳ750 kDa, appearing as a stack of four rings made up of two outer ␣ rings and two inner ␤ rings. Tertiary structural analysis indicates that the center of the ␣ ring is almost closed, thus preventing penetration of substrates into the interior of the ␤ ring on which the proteolytically active sites are located (2). Presumably because of this structural feature, the 20 S proteasome exists as a latent form in cells. The latent proteasome is activated fully by a recently identified endogenous protein, the proteasome activator PA28 (also known as the 11 S regulator) (3,4). PA28 differs from another well known proteasome regulator designated PA700. The latter is made up of multiple subunits of 25-110 kDa and associates with the 20 S proteasome to form the 26 S proteasome, a eukaryotic ATP-dependent protease complex with a molecular mass of ϳ2 MDa (reviewed in Refs. [5][6][7][8]. The 26 S proteasome is responsible for the degradation of a wide variety of cellular proteins tagged with a polyubiquitin chain that functions as a degradation signal. Extensive studies conducted during the past decade have established that the 26 S proteasome plays a critical role in various biological processes by regulating the levels of cellular proteins rapidly, timely, and/or irreversibly (9,10).
The native PA28 is a protein complex with a molecular mass of ϳ200 kDa that binds directly to both ends of the 20 S proteasome to form a football-like structure (5,11). Binding of PA28 greatly stimulates multiple peptidase activities of the 20 S proteasome. However, the football-like proteasome lacks the ability to degrade large protein substrates, suggesting that PA28 may cooperate with the 26 S proteasome in a sequential proteolytic pathway (2,3). Indeed, our recent work indicates the existence of a "hybrid-type" proteasome (12), with PA28 attached to one end and PA700 to another end of the 20 S proteasome. Similar to the 26 S proteasome, this type of proteasome appears to function as an ATP-dependent protease. 1 PA28 is composed of two subunits, named PA28␣ and PA28␤, that share ϳ50% amino acid identity (13). These subunits assemble into a heterohexamer (␣3␤3) with alternating ␣ and ␤ subunits (14) or a heteroheptamer (␣3␤4 or a mixture of ␣3␤4 and ␣4␤3) (15). Cloning of PA28␣ and PA28␤ cDNAs revealed that the ␣ and ␤ subunits are structurally similar to a nuclear protein, termed Ki antigen, which was initially identified with autoantibodies found in sera of patients with systemic lupus erythematosus (13). Ki antigen associates reversibly with the 20 S proteasome, indicating that it is a genuine component of the PA28 protein family (16). Therefore, Ki antigen was renamed PA28␥ (16). Intriguingly, PA28␥ appears to form a homopolymer, presumably PA28(␥) 6 or PA28(␥) 7 (16). Upon * This work was supported in part by grants-in-aid for Scientific Research from The Ministry of Education, Science, Sports, and Culture of Japan, The Human Frontier Science Promotion Organization, and The Naito Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ § To whom correspondence should be addressed: Dept. of Molecular Oncology, Tokyo Metropolitan Inst. of Medical Science, 3-18-22 Honkomagome, Bunkyo-ku, Tokyo 113-8613, Japan. Tel./Fax: 81-3-3823-2237; E-mail: tchiba@rinshoken.or.jp. stimulation with IFN-␥, 2 the protein levels of both PA28␣ and PA28␤ increase markedly in a variety of cells, whereas that of PA28␥ remains unchanged or is slightly decreased (reviewed in Refs. 17 and 18). Furthermore, PA28␣ and PA28␤ are located primarily in the cytoplasm, whereas PA28␥ occurs in the nucleus without appreciable localization in the cytoplasm (19,20). Thus, the two forms of the PA28 complex differ in terms of the responsiveness to IFN-␥ and subcellular localization.
Proteasomes are the central enzymes responsible for the generation of major histocompatibility complex (MHC) class I ligands (reviewed in Refs. 17, 18, and 21). Recently, Groettrup et al. (22) found that, together with the MHC-encoded LMP subunits of the proteosome, the IFN-␥-inducible form of PA28 enhances the generation of class I binding peptide by altering the cleavage pattern of the proteasome. More recently, this form of PA28 was found to enhance the induction of virusspecific killer T cells in vivo (23) and to favor production of dominant class I ligands in vitro (24,25). These findings suggest that the IFN-␥-inducible PA28 plays an important role in the generation of MHC class I ligands (17). In contrast, no information is available concerning the biological or biochemical function of the PA28␥ homopolymer. To gain insights into its biological function, we created mice lacking the PA28␥ gene (Psme3). PA28␥ was not essential in mouse development, but its absence resulted in retardation of cell proliferation and body growth.

EXPERIMENTAL PROCEDURES
Construction of the PA28␥ Targeting Vector-A 12-kb NotI fragment containing the entire coding region of the PA28␥ gene was obtained from a 129/SvJ mouse genomic library (Stratagene) as previously reported (26,27). The targeting vector consisted of the 6.2-kb genomic DNA containing the 1.2-kb neomycin resistance gene derived from pMC1neopoly(A) (see Fig. 1A). A 1.1-kb diphtheria toxin gene (DT-A) derived from pMC1DT-3 was attached to the 5Ј end of the PA28␥neomycin construct.
Production of PA28␥ Null Mice-The targeting vector was linearized with SalI and transfected by electroporation into TT2 embryonic stem (ES) cells (courtesy of Dr. S. Aizawa, Kumamoto University). ES cells were selected in 200 g/ml G418 (Life Technologies, Inc.). Colonies of ES cells with homologous recombination events were identified by polymerase chain reaction (PCR) amplification of a 2.4-kb fragment using the primer pair derived from the 5Ј-upstream region of the PA28␥ gene (5Ј-CACTGCCTAATTGTTGAAAGAGGAGATGCTGTC-3Ј) and from the neomycin gene (5Ј-GTCTTTTTATTGCCGATCCCCTCAGAAGAAC-TC-3Ј). To verify the results of PCR screening, genomic DNAs prepared from the PCR-positive ES clones were digested with EcoRI and BamHI, transferred to a nylon membrane (Amersham Pharmacia Biotech), and then hybridized with the probe specific for the 5Ј-upstream region of the PA28␥ genes (see Fig. 1A). Expected fragment sizes for wild-type and mutant PA28␥ genes were 5.0 and 2.0 kb, respectively (see Fig. 1B).
ES cells heterozygous for the targeted mutation were microinjected into 8 cell-stage ICR mouse embryos and then transplanted into uteri of pseudopregnant BDF1 mice. Chimeric males were crossed to C57BL/6J females. Germline transmission was identified by Southern blot analysis. Progeny containing the mutant PA28␥ allele were intercrossed to obtain PA28␥-deficient mice. Southern, Northern, and Western blot analyses confirmed disruption of the gene.
Northern Blot Analysis-Total RNA was isolated from newborn mice using TRIZOL™ (Life Technologies, Inc.). Ten g of total RNA was electrophoresed on a 1% agarose gel and transferred to a nylon membrane. Full-length mouse PA28␥ cDNA (26) was labeled with 32 P using random primers and used as a probe to detect PA28␥ mRNAs.
Preparation of MEFs and Cell Culture-MEFs were obtained from 2 The abbreviations used are: IFN-␥, interferon-␥; DMEM, Dulbecco's modified Eagle's medium; ES, embryonic stem; FBS, fetal bovine serum; FITC, fluorescein isothiocyanate; MHC, major histocompatibility complex; MEF, mouse embryonic fibroblast(s); PCR, polymerase chain reaction; PBS, phosphate-buffered saline; kb, kilobase. A single probe to the 5Ј end of the PA28␥ gene (shown as the 5Ј probe) was used to identify recombinant alleles at 2 kb and wild-type alleles at 5 kb, following EcoRI and BamHI digestion of genomic DNA. B, Southern blot analysis of DNA from mice obtained by intercross-breeding of heterozygous mice is shown. Mice from inter-cross-breeding were genotyped using tail DNA digested with EcoRI and BamHI. C, Northern blot analysis of total RNA prepared form PA28␥ ϩ/ϩ , PA28␥ ϩ/Ϫ , and PA28␥ Ϫ/Ϫ mice. Full-length mouse PA28␥ cDNA (26) was 32 P-labeled using random primers and used as a probe. D, Western blot analysis of extracts prepared from PA28␥ ϩ/ϩ , PA28␥ ϩ/Ϫ , and PA28␥ Ϫ/Ϫ mice. The blot was probed with the anti-PA28␥ antibody. Flow Cytometric Analysis-For cell size measurement, MEFs were trypsinized, washed with ice-cold PBS, and their forward scatter was measured by FACScan (Becton Dickinson) as soon as possible. For cell cycle analysis, MEFs at logarithmic growth were pulse-labeled with 10 M BrdUrd (Sigma) for 30 min, trypsinized, and fixed in 70% ethanol. For continuous labeling, G 1 synchronized cells were trypsinized, replated at a density of 1.5 ϫ 10 6 cells per 10-cm dish in DMEM with 10% FBS containing 65 M BrdUrd, trypsinized after 3, 6, 9, 12, and 24 h, and fixed in 70% ethanol. Fixed cells were kept at Ϫ20°C until analysis. Cells were treated with 2 N HCl to denature the DNA, followed by neutralization by sodium borate. Cells were then incubated with fluorescein isothiocyanate (FITC)-conjugated anti-BrdUrd antibodies (PharMingen) and counterstained with propidium iodide containing RNase (20 g/ml). The stained cells were analyzed by FACScan. Cell debris and fixation artifacts were gated out, and G 1 , S, and G 2 /M populations were quantitated using CellQuest software. In each experiment, a similar number of events was analyzed.
Immunohistochemistry-PA28␥ ϩ/ϩ and PA28␥ Ϫ/Ϫ MEFs were cultured on a slide glass, fixed with 4% paraformaldehyde/PBS, quenched with 50 mM NH 4 Cl, permealized with 0.2% Triton X-100/PBS for 30 min at room temperature, and blocked with Blockace™ (Yukijirushi) overnight at 4°C. Cells were then incubated with a primary antibody solution containing 0.1% Tween 20 for 1 h, washed in Triton X-100/PBS, and incubated with a secondary FITC-conjugated anti-rabbit antibody. As a control, nonimmune rabbit sera were used. Cells were embedded in SlowFade (Molecular Probes) mounting medium.

Targeted Disruption of the PA28␥ Gene Creates a Null Mutation-
The targeting construct was designed to delete the genomic fragment extending from exon 2 to exon 8 of the PA28␥ gene (Fig. 1A). PCR screening of 600 colonies resistant to G418 revealed that three ES cell clones underwent homologous recombination. Southern blot analysis with the probe external to the targeting construct confirmed the expected recombination events. Microinjection of PA28␥ ϩ/Ϫ ES cells into 8 cell-stage ICR embryos produced multiple chimeric males, and germline transmission was demonstrated for chimeras derived from one ES clone. Intercross of the PA28␥ ϩ/Ϫ mice produced offspring at predicted Mendelian ratios ( Table I), suggesting that the PA28␥ deficiency did not result in mortality during embryogenesis or fetal development. Southern blot analysis of DNAs extracted from the PA28␥ ϩ/Ϫ and PA28␥ Ϫ/Ϫ mice confirmed the disruption of the PA28␥ gene (Fig. 1B). Inactivation of the PA28␥ gene was also confirmed by Northern and Western blot analyses (Fig. 1, C and D). The expression levels in PA28␥ ϩ/Ϫ heterozygotes were nearly one-half of those in PA28␥ ϩ/ϩ homozygotes.
Lack of PA28␥ Results in Small Body Size-Following weaning, the PA28␥ Ϫ/Ϫ mice weighed less than sex-matched littermate controls. Representative growth curves in two pairs of such siblings are shown in Fig. 2A. A cohort of 177 consecutive intercross-breeding was observed up to 12 months of age. During this period, PA28␥-deficient mice were fully viable and fertile.
Analysis of the progeny from intercross-breeding showed that, at 24 weeks of age, the mean body weight of PA28␥ Ϫ/Ϫ mice was significantly smaller than that of PA28␥ ϩ/ϩ mice in both sexes (p Ͻ 0.01, see Fig. 2B and Table I). The mean body weight was intermediate in PA28␥ ϩ/Ϫ mice. To examine whether visceral organ size is proportional to the body weight we measured the wet weight of the heart, liver, spleen, thymus, kidney, and brain. The ratio of organ to body weight was consistently smaller in PA28␥ Ϫ/Ϫ mice than in wild-type mice (data not shown).
Growth Properties of PA28␥ Ϫ/Ϫ MEFs in Vitro-MEFs were prepared from PA28␥ ϩ/ϩ and PA28␥ Ϫ/Ϫ embryos 13.5 days postcoitum, and their growth properties were examined in vitro. At passage 4, PA28␥ Ϫ/Ϫ MEFs showed a significantly slower growth rate and a lower saturation density than their wild-type counterparts (Fig. 3A). Single MEF size was measured using flow cytometry as forward scatter intensity. PA28␥ Ϫ/Ϫ MEFs were slightly larger than wild-type ones (Fig.  3B), consistent with the observation that they showed lower saturation densities.
Entry into the S Phase Is Impeded in PA28␥ Ϫ/Ϫ MEFs-To investigate the cause of growth retardation, proliferating MEFs at the logarithmic phase were pulse-labeled with Br-dUrd and analyzed by flow cytometry. As indicated in Fig. 4A, PA28␥ Ϫ/Ϫ cells showed an increase in the number of G 1 phase cells and a decrease in the number of S phase cells relative to wild-type cells. To confirm this observation, MEFs synchronized at the G 1 phase were continuously labeled with BrdUrd. The percentage of cells that entered the S phase was significantly smaller in PA28␥ Ϫ/Ϫ cells than in wild-type cells (Fig.  4B). These results indicate that entry into the S phase is slightly impeded in PA28␥ Ϫ/Ϫ MEFs. Expression of PA28␣ and PA28␤ in PA28␥-deficient Cells-We next examined whether the absence of PA28␥ influences the expression of PA28␣ or PA28␤. Western blot analysis using MEF extracts showed that the levels of neither PA28␣ nor PA28␤ are affected by the deficiency of PA28␥ (Fig. 5A). Under normal conditions, PA28␣ and PA28␤ are co-localized predominantly in the cytoplasm, whereas PA28␥ occurs mainly in the nucleus (19,20). However, PA28␣ has two possible nuclear localization signals (27). Therefore, it was of interest to examine whether the PA28␣/␤ complex can enter the nucleus in the absence of PA28␥ and compensate for the deficiency of PA28␥. As shown in Fig. 5B, immunohistochemical analysis of PA28␥ ϩ/ϩ MEFs indicated that PA28␥ is located predominantly in the nuclei excluding the nucleoli (upper right panel) as reported previously (19,20). No significant staining was observed in PA28␥ Ϫ/Ϫ cells (lower right panel). On the other hand, the subcellular distribution of PA28␣ was essentially the same between PA28␥ Ϫ/Ϫ and PA28␥ ϩ/ϩ MEFs (upper and lower left panels). Thus, the PA28␣/␤ heteropolymer is unlikely to take over the function of PA28␥ in PA28␥-deficient cells.

DISCUSSION
In this report, we generated PA28␥-deficient mice by gene targeting (Fig. 1) and found that PA28␥ affects body size, cell growth, and cell proliferation (Figs. 2 and 3). Flow cytometric analysis of MEFs (Fig. 4) revealed that the proportion of cells that entered the S phase was significantly lower in PA28␥ Ϫ/Ϫ cells than in wild-type cells. The PA28␥ homopolymer can associate with the 20 S proteasome (16) and stimulate the latent proteasome activity strongly (28). Thus, the phenotype of PA28␥-deficient mice raises the possibility that PA28␥ might be involved in the degradation of nuclear proteins regulating cell cycle progression. For example, loss of PA28␥ may delay the degradation of cyclin-dependent kinase inhibitors such as p21 cip1 and p27 kip1 , the degradation of which is required for G 1 to S transition in the cell cycle (29). Therefore, we examined the protein levels of p27 in PA28␥ Ϫ/Ϫ and PA28␥ ϩ/ϩ MEFs by Western blot analysis but could not find any obvious difference (data not shown).
A series of recent studies has established that the PA28␣/␤ complex is involved in the generation of MHC class I ligands (reviewed in Refs. 17, 18, and 22). Therefore, we investigated whether PA28␥-deficient mice show any abnormalities in their immune system. The cellularity and size of the spleen and thymus did not show any obvious difference between PA28␥ Ϫ/Ϫ and wild-type mice (data not shown). Furthermore, disruption of the PA28␥ gene did not alter the CD4/CD8 ratios in thymocytes and splenocytes, the T/B cell ratio in peripheral lymphocytes, or the expression level of cell surface MHC class I molecules (data not shown). These results indicate that PA28␥ is unlikely to play a pivotal role in the MHC class I antigen processing/presentation system. This is consistent with the observation that PA28␥ has been found in invertebrates that do not have the adaptive immune system (26). However, the present study does not rule out the possibility that PA28␥ plays a specialized role in antigen presentation such as the processing of antigens in the nucleus.
Because PA28␥ localizes almost exclusively in the nucleus and is highly conserved between vertebrates and invertebrates, we supposed that PA28␥ might have an essential role in maintaining cellular activities. Contrary to this initial expectation, PA28␥ turned out to be dispensable. Subcellular localization of PA28␣, and hence, that of the PA28␣/␤ complex did not change significantly between PA28␥-deficient and wild-type cells (Fig.  5). Thus, it appears unlike that dispensability of PA28␥ ensues from the compensation provided by the PA28␣/␤ complex. A better understanding of the biological functions of the PA28 family proteins might be gained by creating mutant mice lacking all three members of this family. Such work is in progress in our laboratories.