Evidence for the Direct Involvement of the Proteasome in the Proteolytic Processing of the Aspergillus nidulans Zinc Finger Transcription Factor PacC*

The 72-kDa zinc finger transcription factor PacC, distantly related to Ci/Gli developmental regulators, undergoes two-step proteolytic processing in response to alkaline ambient pH. “Signaling protease” cleavage of PacC72 removes a processing-inhibitory C-terminal domain, making its truncated PacC53 product accessible to a second “processing” protease, yielding PacC27. Features of the processing proteolysis suggested the proteasome as a candidate protease. We constructed, using gene replacements, two missense active site mutations in preB, the Aspergillus nidulans orthologue of Saccharomyces cerevisiae PRE2 encoding the proteasome β5 subunit. preB1K101A is lethal. Viable preB2K101R impairs growth and, like its equivalent pre2K108R in yeast, impairs chymotryptic activity. pre2K108R and preB2K101R active site mutations consistently shift position of the scissile bonds when PacC is processed in S. cerevisiae and A. nidulans, respectively, indicating that PacC must be a direct substrate of the proteasome. preB2K101R leads to a 2–3-fold elevation in NimE mitotic cyclin levels but appears to result in PacC instability, suggesting an altered balance between processing and degradation. preB2K101R compensates the marked impairment in PacC27 formation resulting from deletion of the processing efficiency determinant in PacC, further indicating direct proteasomal involvement in the formation of PacC27. Deletion of a Gly-Pro-Ala-rich region within this processing efficiency determinant markedly destabilizes PacC. Arg substitutions of Lys residues within this efficiency determinant and nearby show that they cooperate to promote PacC processing. A quadruple Lys-to-Arg substitution (4K→R) impairs formation of PacC27 and leads to persistence of PacC53. Wild-type PacC53 becomes multiply phosphorylated upon alkaline pH exposure. Processing-impaired 4K→R PacC53 becomes excessively phosphorylated.

The eukaryotic 26 S proteasome complex, composed of 20 S core particles (CP) 2 and 19 S regulatory particles, is a multichambered protease ideally suited for the rapid, selective, and complete degradation of cellular proteins (1). However, the view that the proteasome acts exclusively in proteolytic destruction was challenged by the finding that the transcription factor NF-B p50 is generated from its 105-kDa precursor by proteasome-mediated proteolytic processing (2). Since then, the list of substrates processed (i.e. partially proteolyzed to yield a functional form) by the proteasome has increased sufficiently to indicate that this mechanism is not uncommon. Additional, relevant examples are the Drosophila melanogaster zinc finger transcription factor Ci (3) and its metazoan homologue Gli3 (4,5).
The molecular mechanisms mediating the proteolytic processing of NF-B p105 and Ci/Gli3 are remarkably complex. Proteins are usually (but not exclusively) targeted to the 26 S proteasome after polyubiquitination of one or more of their Lys residues (1). NF-B p105 is polyubiquitinated in different regions, which involves at least two ubiquitin ligases. Following signal-dependent phosphorylation of p105, its C-terminal domain recruits a multisubunit E3 through the WD40/F-box protein ␤-TrCP, promoting p105 ubiquitination and proteasome-mediated processing but additionally leading to its degradation (6,7). Thus, the final output is determined by a delicate balance between complete degradation and processing (6,8). Notably, a recent report demonstrates that NF-B p105 can be processed post-translationally by purified proteasomes in the absence of polyubiquitination, reflecting its intrinsic tendency to be targeted to the proteasome. It has also been reported that NF-B p105 is processed co-translationally (9). Finally, a glycine-rich repeat acting as a proteasome stop signal has been shown to play a key role in processing (8, 10 -12) but is dispensable for NF-B p105 processing in yeast (13).
Ci, the transducer of the hedgehog signal, is synthesized as a 155-kDa precursor, which can be processed to a 75-kDa repressor. In the presence of hedgehog signaling, processing is prevented, and Ci 155 translocates into the nucleus to activate hedgehog-responsive genes (3). The E3 component WD40/Fbox protein Slimb (the fly orthologue of ␤-TrCP) binds phos-phorylated target motifs in Ci to promote such processing (14,15). Hedgehog signaling prevents Ci 155 processing to Ci 75 but promotes Ci 155 conversion to a labile form of the precursor (16), reflecting the delicate balance between processing and degradation inherent to proteasome-mediated processing.
What determines whether a protein is processed rather than destroyed by the proteasome? Substrates need to be actively translocated into the catalytic chamber, and such translocation involves energy-dependent substrate unfolding (1). Matouschek and colleagues (12,17) elegantly demonstrated that substrates are processively unraveled from degradation signals mediating substrate recognition by multichambered proteases and established the requirements for proteolytic processing by the proteasome. These include, in addition to a degradation signal targeting the substrate to the proteasome, the presence of a tightly folded (thus recalcitrant to unfolding) domain in the direction of proteasome movement that leads, when encountered by the proteasome, to release of a partially degraded substrate molecule (12). Release is favored by the presence of a low complexity sequence preceding the folded domain in the direction of proteasome movement (12). The NF-B p105 glycinerich region mentioned above is one such low complexity sequence (12). The N-terminally located Rel homology domain in NF-B p105 and the zinc finger region in Ci 155 play the key role of the tightly folded domain in their respective proteins (12).
Previous work (18,19) highlighted the similarities between the zinc finger regions of Ci/Gli and PacC, the Aspergillus nidulans transcription factor mediating regulation of gene expression by ambient pH in fungi (20,21). 674-residue PacC contains an N-terminal DNA binding domain with three canonical C 2 H 2 zinc fingers (22) and undergoes two-step proteolytic activation removing ϳ425 residues from the C terminus in response to alkaline ambient pH (22)(23)(24). The 72-kDa PacC 72 translation product is converted to 53-kDa PacC 53 by a "signaling" protease. PacC 53 is the substrate of a second proteolytic processing reaction, which results in the 27-kDa final product PacC 27 (Fig.  1A). The signaling protease is almost certainly the cysteine protease PalB (25,26). The subject of this work is the identity of the "processing" protease mediating the second step.
PacC 72 is held in a "closed" conformation, inaccessible to the processing protease, by intramolecular interactions involving three regions, denoted A, B, and C (schematically depicted in Fig. 1A). pH-responsive signaling protease cleavage removes interactive region C, thus disrupting intramolecular interactions and making the resulting C-terminally truncated product PacC 53 accessible to the processing protease (27). Therefore, PacC 53 meets one of the requirements for a proteasome substrate, with exposure of unstructured regions that the proteasome needs to initiate polypeptide translocation (28).
Indeed, several arguments strongly suggested the proteasome as the likely processing protease. (i) Extensive mutational analyses have failed to identify the PacC-processing protease, which would agree with the essential role of the proteasome. (ii) The processing protease is conserved in S. cerevisiae, despite the fact that the yeast PacC orthologue Rim101p undergoes a single step "signaling protease cleavage" proteolytic activation (29,30). (iii) Resembling Ci 155 processing, PacC processing is independent of sequence at the processing cleavage site (12,31). (iv) The C terminus of PacC 27 (32) (this work) is heterogeneous, as expected from the model of processing by the proteasome of Matouschek and co-workers (12). (v) The site(s) of processing is dictated by sequence or structure determinants that are remote from the processing site (32). (vi) A processing efficiency determinant has been mapped to a region located C-terminal to the processing site, which would be exposed after signaling protease cleavage (24). (vii) The zinc finger regions of PacC and Ci share distinctive structural features (18,20), and indeed the C-terminal fingers (those that the proteasome would first encounter) share, in addition to the zinc finger fold, high amino acid sequence identity (19). Thus, we explored the possible role of the proteasome in PacC processing by exploiting the amenability of S. cerevisiae and A. nidulans to classical and reverse genetic manipulation. Our data demonstrating that partial loss-of-function mutations affecting the active site of the orthologous proteasome ␤5 subunits Pre2p/PreB characteristically alter the specificity of the processing reaction provide strong evidence that the proteasome is indeed the PacC-processing protease.
Media for A. nidulans Adjusted to Different Ambient pH Conditions-Cultures for protein extraction and Western analysis were made in Aspergillus fermentation medium as described (26). In pH shift experiments, the pH of the cultures before shifting mycelia to alkaline conditions (pH ϳ8.4) was ϳ4.3. Overnight (14 -16-h) cultures in relatively uniform pH conditions were made after adding 100 mM NaH 2 PO 4 plus 100 mM NaCl for acidic conditions (starting and final pH values were 5.5 and ϳ5.1-5.4, respectively) and 100 mM Na 2 HPO 4 for alkaline conditions (starting and final pH values were 8.0 and ϳ7.2-7.4, respectively). Harvesting of mycelia and protein extraction have been described (26).
Yeast Techniques-S. cerevisiae strains with active site missense mutations pre3 T20A , pup1 T30A , pre2 K108A , pre2 K108R , and pre2 T76S single mutations as well as pre3 T20A pup1 T30A and pre2 K108R pre3 T20A double mutations in genes encoding the proteasome proteolytic subunits (36) were isogenic to the WCG4a strain (MATa leu2-3,112 ura3-⌬5 his3-11,15 canS GAL2 and were obtained from Prof. Dr. D. Wolf (University of Stuttgart, Germany). pRS416-Myc3-PacC14 was made by inserting a pacC cDNA fragment encoding PacC residues 5-492 (thereby equivalent to the pacC c 14 product) with three copies of the Myc epitope fused to its N terminus into pRS416, using HindIII and BamHI restriction sites. Strains were transformed with pRS416-Myc3-PacC14 following a LiAc-based procedure (37) and cultured in minus uracil-selective YM4 minimal medium. Whole cell yeast samples for immunoblotting were prepared from 4 A 600 units of yeast in log phase following an alkaline lysis procedure as described by Stimpson et al. (38).
Western Blot Analyses-50 g of A. nidulans protein per sample (as determined by the Bradford assay) were used unless otherwise indicated. Myc 3 -PacC, NimE-GFP, and actin were detected with the anti-c-MYC mouse monoclonal antibody 9E10 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at 1:1,000, a mouse monoclonal antibody anti-GFP mixture (clones 7.1 and 13.1; Roche Applied Science) at 1:1,000, and a mouse anti-actin monoclonal antibody (clone C4; ICN Biomedicals Inc.) at 1:25,000, respectively. Peroxidase-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Laboratories) was used as secondary antibody at a 1:4,000 dilution. Peroxidase activity was detected using Amersham Biosciences ECL. If required, chemiluminiscence was quantified using a TDI 06-LAS3000Q Chemiluminiscence Image Analyzer with Fujifilm Multi Gauge software.
pacC Gene Replacements-A. nidulans pacC mutations were all constructed by gene replacement using plasmid pSpacC900 as described (26) with the correction that the triple MYC epitope and linker sequence preceding the major PacC initiator Met is MAEQKLISEEDLNGEQKLISEEDLNGEQKLIS-EEDLNGT. Missense and deletion mutations were introduced using Stratagene's QuikChange methodology. MAD265 (yA2 pabaA1 biA1 pyrG89 ⌬pacC::pyr-4 argB2 was used as recipient strain. A. nidulans transformants were selected in the presence of 12.5-25 mM sodium molybdate. Those carrying the expected gene replacement were identified by Southern analysis and meiotically crossed to remove the pyrG89 mutation as described (26). Myc-tagged pacC alleles segregating in crosses were additionally genotyped by PCR. The gene-replaced pacC alleles used in this work are listed in Table 1.
PCR Genotyping of Myc 3 -Conidiospores were taken from the surface of colonies using a toothpick and suspended in 0.1 ml of breaking buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 1 mM EDTA, and 10 mM Tris-HCl, pH 8) in screwcap 1.5-ml tubes that were vortexed for 30 s after the addition of 150 mg of 0.45-mm glass beads. Tubes were incubated for 30 min at 70°C with occasional vortexing before the addition of 0.2 ml of phenol-SEVAG (phenol/chloroform/isoamyl alcohol, 25:24:1). The two phases were mixed by vortexing for 5 min before recovering the aqueous phase after centrifugation for 5 min at 14,000 rpm in an Eppendorf microcentrifuge. 0.5-l samples were used in 20-l PCRs with Taq polymerase, using primers that amplified a fragment including pacC Met codon 5, such that a shift was clearly observed by agarose gel electrophoresis if this fragment contained the in-frame insertion of the sequence encoding Myc 3 .
preB Gene Replacements-A genomic DNA fragment containing 0.35 kb of flanking DNA on either side of the preB gene was PCR-amplified (with Pfu polymerase) and cloned in pBS-SK ϩ using NotI and SpeI ends. A BglII fragment containing the Aspergillus fumigatus pyrG gene was inserted downstream of preB, followed by a 1.5-kb fragment corresponding to preB 3Ј-untranslated region. Mutant versions encoding PreB K101A and PreB K101R were constructed using mutagenic PCR with Stratagene's QuikChange and appropriate oligonucleotides. 5,525-bp NotI-ClaI preB-pyrG fum -3Ј-untranslated region preB wild-type or mutant DNA fragments were used to transform the nonhomologous end-joining pathway-deficient A. nidulans pyroA4 pyrG89 nkuA⌬::argB ϩ argB2 strain by selecting transformants on pyrimidine-deficient Aspergillus complete medium (39). Primary transformants were purified on the same medium by mycelial transfer (to ensure heterokaryon recovery) before proceeding to homokaryon purification from conidiospores as described (40) (also see "Results"). preB2 and preB0 homokaryosis were confirmed by Southern blot analysis and by direct sequencing of the preB2 mutation following PCR amplification.
Assessment of Proteasome Proteolytic Activities-Mycelia were harvested by filtration, washed, frozen in liquid nitrogen, and lyophilized. Lyophilized cells were broken using a Fast Prep cell disruptor FP120 and a 0.5-mm ceramic bead with a 10-s pulse at a setting of 4. The resulting powder was resuspended in 0.5 ml of Tris-HCl, pH 8.2, 2 mM ATP, 5 mM MgCl 2 , 1 mM dithiothreitol, and 20% glycerol (v/v). 0.7 ml of 0.45-mm glass beads was added to the suspension, which was shaken for 20 s at a setting of 6 in the Fast Prep. Crude extracts were clarified by centrifugation for 30 min at 16,000 ϫ g and 4°C. Proteasome proteolytic activities were determined using these crude extracts by monitoring cleavage of fluorogenic peptide substrates essentially as described (41,42), using N-succinyl-LLVY-7-amido-4-methylcoumarin (N-succinyl-LLVY-AMC), t-butoxylcarbonyl-LRR-AMC (Boc-LRR-AMC), and benzyloxycarbonyl-LLE-AMC (Z-LLE-AMC) as substrates for chymotrypsin, trypsin, and PGPH activities, respectively. Fluorescence readings of released AMC were made using 380 and 460 nm excitation and emission wavelengths, respectively, and were converted to pmol of AMC by using a standard curve.
Bacteriophage -Phosphatase Treatment-Treatment with -phosphatase was carried out using 100 g of protein with incubation for 30 min at 37°C in 100-l reaction mixtures in -phosphatase buffer (New England Biolabs) with 2 mM MnCl 2 and 400 milliunits of -phosphatase (Biolabs, Northbrook, IL). Where indicated, 10 mM Na 3 VO 4 was used as phosphatase inhibitor. Mock incubations were carried out in the absence of enzyme and inhibitor.

RESULTS
Heterogeneity of PacC 27 -pacC900, constructed by gene replacement and phenotypically indistinguishable from the wild type, encodes a PacC protein with three copies of the Myc epitope attached to its N terminus (26). Previous work using electrophoretic mobility shift assays (32) clearly indicated that the C terminus of PacC 27 , located at residues ϳ252-254, is heterogeneous. Using pacC900 in Western blots (Fig. 1B), we reproducibly detected four different C termini for PacC 27 , corresponding to polypeptides that we denoted a-d (Fig. 1B). Predominating forms a and b seemingly correspond to the major form(s) of PacC 27 detected in electrophoretic mobility shift assays and thus differ in ϳ5 residues (see Fig. 3B in Ref. 32). PacC 72 and PacC 53 also showed some heterogeneity ( Fig. 1; see below).
Processing Specificity of PacC Is Altered in a Class of S. cerevisiae Proteasome Active Site Mutants-Data from S. cerevisiae showed that proteasome inhibitors such as MG132 are inefficiently taken up by fungi (43), a problem that would be exacerbated in A. nidulans, because its greater metabolic versatility compared with yeast would increase the likelihood that these compounds are catabolized, further enhancing their ineffectiveness in vivo.
To address the possible role of the proteasome in generating PacC 27 , we initially followed a transgenic approach using S. cerevisiae, taking advantage of a mutant PacC with a nonsense mutation in codon 493 (the product of the pacC c 14 allele). This C-terminally truncated version of PacC resembles PacC 53 (whose C terminus lies within residues 493-500) and is therefore processed to PacC 27 in an ambient pH-independent manner (23,24). The pacC c 14 product is processed ex vivo in S. cerevisiae (32), allowing us to use extant viable mutants impairing the yeast proteasome, following the approach reported by the Maniatis group for their analysis of proteolytic processing of NF-B p105 (2,13).
The proteasome is essential in yeast. The barrel-shaped 20 S CP is a stack of four heteroheptameric rings, with the two central rings formed by related ␤-subunits and the outer distal rings formed by related ␣-subunits (44). Since ␣ and ␤ subunits contribute to the overall architecture of the CP, deletion of any single subunit in the 20 S CP, with the single exception of Pre9p, is lethal.
We constructed a plasmid driving expression of Myc 3 -PacC-(5-492) in S. cerevisiae and used a panel of yeast proteasome mutants to analyze PacC 53 processing (Fig. 2). In the wild type, FIGURE 1. Nomenclature of PacC forms. A, the short term response of PacC to ambient pH alkalinization is shown. The Western blot illustrates the twostep processing of PacC in a pacC900 strain when cells cultured under acidic (pH 4.5) conditions are shifted to alkaline (pH 8.4) conditions. pacC900 was constructed by gene replacement and is phenotypically wild-type (26). It encodes a wild-type PacC protein tagged with three copies of the Myc epitope preceding Met5, which represents the major translation initiation site. The 72-kDa translation product PacC 72 is converted to the 53-kDa intermediate PacC 53 by the "signaling" protease. PacC 53 is converted to the final, 27-kDa product PacC 27 by the "processing" protease. A-C, three interactive regions preventing accessibility of the processing protease to PacC 72 . Region C is removed by the signaling protease, which cleaves a peptide bond located inside the signaling protease box (small box). By disrupting intramolecular interactions, removal of region C makes PacC 53 accessible to the signaling protease. The two ovals in PacC 72 indicate two PalA binding sites, and the circle indicates one nuclear localization signal located within the zinc finger region. Details of this model are given in Ref. 21. B, PacC forms in cells cultured for 16 h under moderately acidic (pH 5.4) and alkaline (pH 7.2) conditions. Note that PacC 53 is barely detectable under alkaline conditions and that moderately acidic conditions (H ϩ ) allow detectable processing to PacC 27 . The four forms of PacC 27 are indicated. The Myc 3 epitope is attached to the N terminus of the protein, and thus these forms reflect C-terminal heterogeneity. The C terminus of b is located at approximately residues 252-254 (32).  53 and PacC 27 is included as control. ␣, ␤, and ␥ denote the three forms of PacC 27 seen in wild-type and mutant yeasts. The Western blots displayed show results typical of at least two independent experiments for each construct. In addition, we confirmed that there were no significant differences among different clones involving the same construct, as illustrated by the two different clones (clones 1 and 2) shown for pre3 T20A pre2 K108R .
we detected a prominent band corresponding to Myc 3 -PacC-(5-492) and a processed band with the mobility of PacC 27 (Fig.  2, lanes 2 and 10; an A. nidulans extract containing PacC 53 and PacC 27 was loaded as control in lane 1). Pre3p T20A and Pup1p T30A substitutions inactivating the PGPH-and trypsin-like proteolytic centers, respectively, had no significant effect on Myc 3 -PacC (5-492) processing (Fig. 2, compare lanes 3 and 4  with lanes 2 and 10). The double mutant (Fig. 2, lane 5) had no effect either. In marked contrast, substitutions involving Pre2p Lys 108 had a marked effect. In single mutants, both Pre2p K108A and K108R led to a distinctive pattern of misprocessing characterized by the appearance of two abnormally high mobility forms of PacC 27 (denoted ␤ and ␥ in Fig. 2), apparently at the expense of the major wild-type form (indicated by ␣). These abnormal ␤ and ␥ forms were absent in the Pre2p T76S mutant (Fig. 2, lane 8) (Thr 76 is the N-terminal, catalytic residue in the mature subunit). Although Pre2p substitutions involving Lys 108 lead to markedly deficient chymotrypsin-like activity, Ser functionally substitutes for Thr in the N terminus of mature Pre2p (36). Thus, these data strongly suggest that PacC 27 originates from proteasome-mediated processing of Myc 3 -PacC-  and that Pre2p plays a key role in determining the actual size (i.e. the C terminus) of the processed product, since impairing the Pre2p catalytic activity shifts the position(s) of the scissile peptide bond(s). This would be consistent with previous data (and our working model) according to which the position of the scissile bond(s) would be determined by their relative distance from an N-terminal sequence or structure determinant. In agreement, the double pre3 T20A pre2 K108R mutation led to the characteristic misprocessing resulting from Pre2p K108R but additionally led to a marked reduction of form ␥, indicating that processing to this polypeptide involves the proteolytic activity of Pre3p (Fig. 2, lanes 6 and 9, representing two independent clones).
A preB Mutation Resulting in K101A Substitution Is Lethal-In view of the above results, we set out to obtain loss-offunction mutations in the A. nidulans gene encoding the proteasome ␤5 subunit. We identified chromosome II-located AN3932.3 (available on the World Wide Web) as the PRE2 orthologue, which we denoted preB, and confirmed its predicted exon-intron structure by cDNA sequencing (supplementary Fig. S1). preB encodes a 296-residue polypeptide showing high amino acid sequence identity with Pre2p (supplementary Fig. S2; similarity is lost in the region of the propeptide, which is released upon proteolytic activation). As expected, key residues involved in proteolysis are invariant. The equivalents in the PreB proenzyme of yeast Pre2p Thr 76 (the N-terminal catalytic Thr after proenzyme proteolytic activation) and Lys 108 (Lys 33 in the mature protein) are Thr 69 and Lys 101 , respectively.
In S. cerevisiae Pre2p, K108A substitution has a more extreme phenotype than K108R (36,45). We constructed preB alleles encoding mutant proteins carrying the corresponding single residue substitutions by gene replacement (supplementary Fig. S1), using an nkuA⌬ recipient strain, deficient in the nonhomologous end-joining recombination pathway (46). Our preB gene replacement protocol uses transforming linear fragments carrying A. fumigatus pyrG as a selective marker (pyrG encodes orotidine 5Ј-phosphate decarboxylase) and involves a double recombination event. One cross-over must occur at the preB 3Ј-untranslated region to ensure pyrG integration into the genome. In contrast, the second cross-over can take place either upstream (leading to a replacement by the mutant allele) or downstream (leading to a wild-type allele) of the mutated preB Lys 101 codon (supplementary Fig. S1). We performed three separate transformation experiments with preB1 K101A , preB2 K101R , and wild-type preB0 alleles, respectively.
Transformants obtained with the preB0 (i.e. preB ϩ ) control grew normally on minus-pyrimidine medium and were phenotypically stable upon subsequent subculturing. They carried the expected recombination event, as determined by Southern blot hybridization and diagnostic PCR (data not shown). Although the replaced wild-type allele was denoted preB0 to underscore that, unlike the "true" wild type, it contains A. fumigatus pyrG downstream of the preB open reading frame, the preB0 and preB ϩ strains were shown to be phenotypically indistinguishable (see below). In contrast, two classes were noticeable among preB K101A and preB K101R transformants (supplemental Fig. S1). Transformants in one class resembled those in the single preB0 class and were shown to contain the wild-type preB gene. Transformants in the second class grew slowly (preB2 K101R ) or very slowly (preB1 K101A ) compared with the wild type. Slowly growing colonies were shown to be heterokaryons whose cells contain recipient (preB ϩ , pyrimidine-requiring) and transformed (preB1 K101A or preB2 K101R , pyrimidine-nonrequiring) nuclei.
A. nidulans conidiospores contain a single nucleus and thus the heterokaryotic state cannot propagate through conidia. In contrast to preB0 or preB2 K101R transformants, conidiospores corresponding to slowly growing preB K101A transformants were able to form colonies only when plated on medium containing pyrimidines (i.e. under conditions supporting growth of the recipient strain) but not on selective medium lacking pyrimidines (supplemental Fig. S1), suggesting that preB K101A ::pyrG fum conidia were not viable (40).
To characterize the terminal phenotype of the preB K101A :: pyrG fum allele, we examined the ability of conidiospores developed from heterokaryotic colonies to germinate. Control preB0 conidia were fully germinated after 24 h irrespective of the presence or absence of exogenous pyrimidines. In contrast, conidia from a strain carrying preB K101A ::pyrG fum in heterokaryosis were indistinguishable from the pyrG89 control in that they formed hyphae in synthetic medium supplemented with pyrimidines but did not give rise to germ tubes after 24 or 72 h at 25°C in the absence of pyrimidine supplementation (supplemental Fig. S3). The absence of the expected class of mutant preB conidiospores that would be unable to germinate irrespective of the presence or absence of pyrimidines strongly suggests that a strain carrying preB K101A ::pyrG fum in heterokaryosis does not give rise to viable preB K101A ::pyrG fum conidia, possibly because this mutation abrogates conidiospore development. (In A. nidulans, conidiospore development involves singly nucleated cells from its very early stages (47).) In any case, these data demonstrate that the preB K101A ::pyrG fum allele is lethal and can only be maintained in heterokaryosis. NOVEMBER 30, 2007 • VOLUME 282 • NUMBER 48

JOURNAL OF BIOLOGICAL CHEMISTRY 34739
A preB Mutation Resulting in K101R Substitution Is Viable and Impairs the Proteasome Chymotryptic Activity-Homokaryotic pyrimidine independent clones carrying the preB K101R ::pyrG fum gene replacement event (as determined by Southern analysis; data not shown) were obtained from conidia collected from primary transformants, indicating that this mutation is not lethal. Direct sequencing of PCR products confirmed that these strains were carrying the missense mutation, leading to K101R substitution, which we denoted preB2, in homozygosis. Compared with the wild type, preB2 leads to reduced growth at 37 and 42°C, but its debilitating effect on growth was very conspicuous at 30°C (supplemental Fig. S4) and nearly fully prevented growth at 25°C (supplemental Fig.  S5). Notably, preB2 strains were markedly hypersensitive to molybdate toxicity and could not grow at all at 37°C on alkaline pH plates or on media containing the aminoglycoside antibiotic neomycin, conditions under which the wild type grows reasonably well. In contrast, they were not differentially affected by DTT, high osmolarity, or sodium stress (supplemental Fig. S4).
Enzyme assays using fluorogenic peptide substrates specific for each of the activities of the proteasome demonstrated that preB2 mutant extracts were markedly deficient in proteasomal chymotryptic activity, which was determined to be ϳ20% of that in the isogenic preB0 wild type (Fig. 3A). In contrast, their trypsin and PGPH activities are possibly somewhat increased (Fig. 3A), resembling the situation reported for pre2 K108R mutants in S. cerevisiae (36).
To confirm that proteasome-mediated proteolytic degradation was impaired by PreB K101R , we determined the steady state level of the prototypic proteasome substrate NimE, one of the two B type cyclins in A. nidulans (48), in mutant and wild-type cells. We used a gene encoding a NimE-GFP fusion protein (kindly donated by Stephen Osmani) that was crossed into the preB2 genetic background to demonstrate, using Western blots, that the NimE steady-state level in the mutant was increased 2-3-fold relative to the wild type (Fig. 3B). In marked contrast, levels of actin or of the cytoplasmic enzyme hexokinase were unaffected by the mutation. NimE localizes to the nucleoplasm and the spindle pole body in interphase nuclei (Fig. 3C). In agreement with the increase in the steady state level of NimE-GFP, we observed a 2-fold increase in nuclear NimE-GFP fluorescence in the preB2 mutant relative to the wild type (Fig. 3, C and D). We conclude that as far as the degradative  role of the proteasome on this particular substrate is concerned, the preB2 mutation results in partial loss of function, which correlates with loss of chymotryptic activity.
preB2 Results in a Marked Decrease in the Steady State Levels of PacC and Affects the Fidelity of Processing to PacC 27 -To address the involvement of the proteasome in PacC processing, we crossed pacC900 into the preB2 mutant background. Unexpectedly, PacC was barely detectable in preB2 extracts when equal amounts of wild-type and preB2 protein extracts were analyzed by Western blotting (Fig. 4A). Detection of PacC processing at equal sensitivity required a considerable increase in the amount of mutant preB2 protein extracts loaded on Western blots (Fig. 4B). Digital and quantitative chemiluminiscence analyses confirmed that the steady state of PacC in the mutant is 3-5 times lower than in the wild type (not shown). Since Northern blot analyses demonstrated that preB2 does not result in a significant reduction in pacC transcription (supplemental Fig. S6), these experiments suggest that preB2 leads to marked destabilization of PacC.
Western analyses of PacC in mutant preB2 cells revealed that the kinetics of the two-step processing of PacC 72 to PacC 53 and PacC 27 were not greatly affected (Fig. 4B). However, a closer inspection of the PacC processing pattern revealed two striking effects of preB2 on the processing protease reaction. One was the abnormally high levels of PacC 27 relative to PacC 72 seen under acidic pH conditions, under which processing is barely detectable in preB ϩ cells (Fig. 4, A and B). However, the most conspicuous effect of preB2 is on the fidelity of the processing protease reaction. As noted above, PacC 27 is heterogeneous and involves four major polypeptides, denoted a, b, c, and d in decreasing size order (Fig. 1B). In the wild type, PacC 27 b largely predominates, PacC 27 a follows in abundance, and forms c and d are very minor (Fig. 4C). In marked contrast, this highly reproducible pattern is altered by preB2, such that c and d become very prominent under persistent alkaline ambient pH conditions, apparently at the expense of PacC 27 a and b (Fig. 4C). The effects on the fidelity of processing that we observed in the A. nidulans preB2 mutant are markedly similar to those seen in the corresponding S. cerevisiae pre2 K108R mutant, although the actual electrophoretic mobilities of the minor bands increased in prominence are clearly different in yeast and Aspergillus (Fig.  4D). The fact that the fidelity of PacC processing is altered by a single residue change affecting the active site of the A. nidulans proteasome ␤5 subunit constitutes evidence that PacC is a direct substrate of the proteasome and that this is indeed the processing protease.
pacC910 Rescues the Low Levels of PacC Resulting from the preB2 Mutation by Favoring Its Proteolytic Processing-Previous work established that residues 266 -407 contain a processing efficiency determinant (PED). We constructed by gene replacement pacC910, a Myc-tagged version of pacC⌬10 encoding a PacC mutant deleted for the PED (24) for comparison with pacC900. In agreement with our published data using electrophoretic mobility shift assays, a major proportion of PacC910 from acidic ambient pH conditions is detected as the (mutant) primary translation product, with a minor proportion seen as PacC 27 (Fig. 5A, lane 1). Since the deleted region 266 -407 includes the nearly entire interactive region B (including critical Leu 340 ; see also below) and part of region A, PacC910 must be in the open conformation and thus accessible to the processing protease under any ambient pH condition (24,27). However, since deletion of the PED markedly reduces processing, the primary PacC910 translation product predominates under acidic conditions. Upon shifting cells to alkaline conditions, PacC910 is converted to mutant PacC 53 , which, unlike the situation in the wild type, remains stable over time and is not converted to PacC 27 (24) (Fig. 5A). Notably, PacC910 protein levels were 4 -6 times higher than in the wild type (data not shown) (Fig. 5A, compare lanes 1-7 with lane 8). This is attributable, at least in part, to elevated levels of pacC transcription in the mutant, as determined by Northern analyses (supplemental Fig. S6).
Since preB2 and pacC910 have opposing effects on PacC levels, we constructed a pacC910 preB2 mutant for comparison with a preB ϩ strain. In pH shift experiments, overall PacC levels in pacC910 preB ϩ single mutant cells were 4 -6 times higher than in the wild-type (pacC900 preB ϩ ) and hence some 20 -30 times higher than in the pacC900 preB2 single mutant (Figs. 4 and 5). Compared with the wild type (pacC900), pacC910 leads to a ϳ3-fold increase in overall PacC levels in the preB2 background (Fig. 5B) (data not shown). However, unlike the situation in the preB ϩ background, mutant (lacking the PED)

Direct Involvement of the Proteasome in PacC Processing
PacC910 53 accumulates 30 min after the shift but becomes barely detectable at later time points, apparently without conversion to PacC 27 (Fig. 5B). Most notably, in stark contrast with the pacC910 single mutant, where the full-length PacC910 largely predominates under acidic conditions (Fig. 5A, lane 1), PacC 27 is the almost exclusive PacC form in pacC910 preB2 cells and the one that accounts for the elevated levels relative to the pacC900 preB2 single mutant (Fig. 5B). pacC910 preB2 cells show the typical preB2 PacC 27 pattern characterized by con-spicuous high mobility bands PacC 27 c and PacC 27 d (Fig. 5B). From experiments in Figs. 4 and 5 we conclude that deletion of the PED largely precludes PacC processing in the context of the wildtype proteasome, whereas in the context of the preB2 proteasome active site mutant, deletion of the PED favors a processing pathway apparently at the expense of a degradative pathway. We therefore conclude that the proteasome is the PacC-processing protease.
Epistasis Relationships between pacC and preB Alleles-Molybdate sensitivity growth tests carried out with the pacC910 preB2 double mutant revealed that pacC and preB interact genetically. Acidity-mimicking loss-of-function mutations in pacC or in any of the pal genes typically preclude growth at alkaline pH and result in hypersensitivity to molybdate and resistance to neomycin (supplemental Fig. S4, lanes 10  and 11). pacC910 has a neutrality-mimicking phenotype (32) characterized by wild-type growth at alkaline pH and marked resistance to molybdate and neomycin (supplemental Fig. S4,  lane 8). preB2 prevents growth on alkaline pH medium and results in hypersensitivity to molybdate and neomycin (supplemental Fig. S4, compare preB2 with preB0 or preB ϩ strains,  lanes 4 -6). Notably, although pacC910 does not suppress preB2 on alkaline pH or neomycin plates, it does strongly suppress its molybdate hypersensitivity. To address whether this suppression is pacC910-specific, we used gene-replaced, Myc 3 -tagged versions of the gain-of-function alkalinitymimicking (and thus molybdate-hypertolerant) mutations pacC c 14 and pacC c 69, respectively, denoted pacC c 14900 (26) and pacC c 69900 (this work). Both suppressed the preB2 molybdate hypersensitivity phenotype to a substantial extent ( Fig. 6A) but had no effect on its neomycin or alkaline pH sensitivity (data not shown). Thus, three pacC mutations leading to increased tolerance to molybdate suppress the preB2 molybdate-hypersensitive phenotype, showing that pacC and preB interact genetically and strongly, indicating that pacC plays a role downstream of or in concert with preB in determining tolerance to molybdate.
Alkalinity-mimicking pacC c gain-of-function mutations result in pH-independent processing to PacC 27 , because their mutant products are synthesized in the processing proteaseaccessible conformation (23,24,32) (Fig. 6B, preB ϩ lanes). Notably, preB2 did not prevent processing of the pacC c 14900 and pacC c 69900 products to PacC 27 but led to the characteristically altered PacC 27 pattern typical of preB2 mutants and to a ϳ4-fold reduction in the levels of PacC, resembling the situation with pacC910 (Fig. 6B, note the different loadings in preB ϩ and preB2 lanes). (We used here cells continuously cultured under moderately acidic (where a small but detectable amount of signaling occurs) or alkaline conditions rather than the pH shift protocol, since pacC c 14900 and pacC c 69900 result in PacC FIGURE 6. Effects of preB2 on mutant PacC proteins committed to pH-independent processing. A, pacC910, pacC14900, and pacC69900 suppress the marked molybdate hypersensitivity resulting from the preB2 mutation. Strains were grown for 2 days at 37°C in appropriately supplemented minimal medium (SC) with or without 12.5 mM Na 2 MoO 4 , as indicated. B, the indicated strains were cultured for 16 h under moderately acidic (pH ϳ 5.5) and alkaline (pH 7.8) conditions before proceeding to protein extraction and Western blot analyses of Myc-tagged PacC forms. Note that moderately acidic pH does not fully prevent pH signaling in pacC900 preB ϩ or preB2 cells (see also Fig. 1B). Since pacC14900 and pacC69900 translation products are processed independently of ambient pH, the very similar results obtained from independent cultures grown at two different pH values for these mutants provide a direct indication of the reproducibility of the assays. In addition, data shown for pacC14900 preB2 are typical of four independent experiments. processing at any ambient pH.) The increase in PacC 27 levels that takes place in preB2 pacC910, preB2 pacC c 14900, and preB2 pacC c 69900 double mutants relative to the single preB2 mutant, which is most noticeable under pH 5.5 conditions (Fig.  6B, preB2 lanes), possibly underlies their increased tolerance to molybdate.
In summary, preB2 K101R results in a characteristic pattern of PacC 27 processing with the wild type and with three significantly different mutant PacC substrates: one (pacC910) lacking the PED (residues 266 -407); a second (pacC c 14900) carrying a C-terminal truncation removing processing-preventive interactive region C (23, 24, 27); and a third (pacC c 69900) carrying the L340S substitution that disrupts processing-preventing intramolecular interactions (27,32). These data further implicate the proteasome as the processing protease.
Deletion Analysis of a Gly-Ala-Prorich Sequence within the PED-Our attempts to delimit further the PED by deletion analysis were largely unsuccessful, possibly indicating that it includes different regions, each contributing partially to the efficiency of the processing protease reaction in addition to their role in the intramolecular PacC interactions (data not shown). However, one interesting conclusion came from deletion of a region within the A. nidulans PacC PED, including residues 296 -325, which is very rich in Pro, Ala, and Gly (Fig.  7A). In other fungal homologues, this region is markedly rich in Gly residues (27). A Gly-rich region has been proposed as a stop signal preventing the complete degradation by the proteasome of the NF-B p105 processed products p50 and p52 (8, 10 -12). It has also been reported that Gly-Ala repeats impair proper substrate unfolding by the proteasome (49). To address the possible role of this region, we constructed pacC913, encoding a Myc-tagged PacC deleted for residues 296 -325. In pH shift experiments, PacC913 ⌬296 -325 appears unstable upon shifting cells to alkaline conditions (Fig. 7B). In agreement, pacC913 cells cultured under moderately acidic or alkaline conditions showed markedly reduced levels of PacC compared with wild type (Fig. 7C). Thus, deletion of this Gly-Ala-Pro-rich sequence appears to be required for normal stability of PacC under conditions activating the pH signaling pathway.
Substitution of Lys Residues in the PED and an Adjacent Region by Arg Largely Prevent Processing Proteolysis-One feature that the PED could potentially provide to promote efficient PacC processing would be one or more Lys residues whose ubiquitination targets the PacC substrate to the proteasome. The 266 -407 region contains potentially ubiquitinable Lys 267 and Lys 337 residues. Lys 267 belongs to a putative bipartite nuclear localization sequence composed of two triplets of basic residues, which additionally include Lys 252 and Lys 253 (Fig. 8A). Simultaneous substitution of all six basic residues in this bipartite sequence by Ala (pacC906;  Fig. 6B. Under alkaline pH conditions, PacC 53 (indicated by an asterisk) is detectable in the pacC904 mutant but not in the wild type, and pacC904 largely prevents processing to PacC 27 . The primary translation product (arrow) is detectable in pacC14904 but not in pacC14900 cells, indicating that the four Lys-to-Arg substitutions impair its pH-independent processing to PacC 27 . The Western blots shown are typical of six independent experiments for pacC904 and three experiments for pacC14904. E, the shifted electrophoretic mobilities seen for PacC 53 bands result from PacC 53 phosphorylation, as indicated by their sensitivity to -phosphatase treatment. Samples are pacC904 protein extracts. Mock lane, no phosphatase added; inhibitor, phosphatase treatment in the presence of sodium vanadate. Note that the number of phosphates per molecule appears to increase with the time that cells are exposed to ambient alkaline pH, as shown by the conspicuous decrease in untreated PacC 53 mobility seen between the 15 and the 240 min time points (see also F). F, pacC904 mutant PacC 53 undergoes more extensive phosphorylation than the wild type. Samples correspond to 60 min time points. Note that although all wild type and mutant shifted mobility bands are sensitive to -phosphatase treatment, the pacC904 mutant shows one additional, phosphatase-sensitive band (arrows) of markedly reduced electrophoretic mobility, which is absent from the wild type. 1) demonstrated that none of them is involved in PacC processing (data not shown). Thus, we constructed the genereplaced allele pacC901 encoding a mutant PacC carrying the conservative yet nonubiquitinable K337R substitution and found that this mutant protein was converted to PacC 27 slightly less efficiently than the wild type (data not shown). We thus considered the possibility that Lys 267 could potentially compensate for the absence of Lys 337 and constructed pacC902 encoding a Myc-tagged PacC carrying the double K267R/K337R substitution. Indeed, this double substitution significantly impairs PacC processing (Fig. 8B). This prompted us to test the effect in processing of K252R/K253R substitution alone (pacC903) and in combination with the double K267R/K337R substitution (pacC904). In agreement with the above Ala-scanning mutagenesis data, K252R/ K253R had no significant effect on processing (Fig. 8B). In marked contrast, the quadruple Lys to Arg substitution largely prevented processing in pH shift experiments (Fig.  8C) and resulted in detectable mutant PacC 53 accumulation and a marked reduction in the steady state level of PacC 27 in cells from overnight cultures under alkaline conditions (Fig. 8D).
It is notable that no increase in either PacC 72 or PacC 53 commensurate with the marked decrease in PacC 27 resulting from pacC904 was detected. Since pacC900 and pacC904 show no evident difference in their transcriptional response to alkaline pH (as determined by Northern blots), 3 this suggests that the mutant forms might be unstable, although this possibility has not been addressed here. In summary, the negative, additive effect on processing of Arg substitutions involving four Lys residues located within or adjacent to the PED would agree with (although it does not unequivocally demonstrate) a requirement for their ubiquitination in PacC processing, with Lys 337 playing the most prominent role.
PacC 72 can be phosphorylated (50). The aggregate electrophoretic mobilities of wild-type pacC900 and mutant pacC904 PacC 53 decrease over time with exposure to alkaline conditions (Fig. 8C). They consist of several closely spaced bands, with the lower mobility PacC 53 bands predominating at late time points, apparently at the expense of faster mobility bands prominent at 15 min (Fig. 8E, untreated lanes). -Phosphatase treatment shifted the mobility of all relatively slow bands toward the fastest mobility band, demonstrating that they all result from phosphorylation (Fig. 8, E and F). (Note that acquisition of negative phosphate group charges results in reduced electrophoretic mobility). The multiple -phosphatase-sensitive PacC 53 bands predominating at late time points (Fig. 8E, untreated lanes) therefore result from multiple phosphorylation events on the same molecule.
An even lower mobility PacC 53 band, very prominent at late time points in the pacC904 mutant but not in the wild type (Fig.  8F, corresponding to 60 min time points; this band is indicated by arrows), clearly contributes to the relatively lower PacC 53 aggregate electrophoretic mobility of the mutant (Fig. 8C), suggesting that the mutant PacC 53 is excessively phosphorylated.
Sensitivity of this band to -phosphatase treatment (Fig. 8F) shows that it indeed results from PacC 53 phosphorylation. We conclude that wild-type PacC 53 is phosphorylated and that markedly reduced PacC 53 processing resulting from pacC904 correlates with excessive phosphorylation of the quadruple Lys-to-Arg-substituted mutant PacC 53 .
The Quadruple Lys to Arg Substitution Impairs Conversion of PacC 53 into PacC 27 -pacC c 14900 (see above) encodes a Myctagged gene-replaced version of pacC c 14 (see above). As reported previously, PacC14900 is efficiently processed to PacC 27 irrespective of ambient pH (26), and indeed no primary translation product was detectable in protein extracts from cells cultured under either acidic or alkaline conditions (Figs. 6 and 8D). In contrast, the quadruple Lys to Arg substitution led to detection of the truncated PacC14904 primary translation product (Fig. 8D), strongly suggesting that it impairs PacC 53 conversion into PacC 27 .

DISCUSSION
The fungal pH-regulatory protein PacC undergoes two-step proteolytic processing in response to alkaline ambient pH. Previous work indicated involvement of calpain-like protease PalB in the signaling protease step converting PacC 72 to PacC 53 and suggested involvement of the proteasome in the proteolytic processing reaction leading to formation of PacC 27 .
We used a genetic approach to obtain evidence that PacC 27 results from proteasome-mediated processing. In S. cerevisiae, substitutions of Lys 108 (Lys 33 in the mature Pre2p subunit) by Ala or Arg shift about two-thirds of the major processed form toward alternative PacC 27 products shortened at their C termini. This effect is specific for Pre2p, since inactivation of the catalytic Thr residues in Pre3p and Pup1p (accounting for the PGPH and trypsin activities of the proteasome, respectively) have no effect and because substitution of Pre2p N-terminal Thr by Ser, which reduces the ability of the proteasome to degrade proteins (51), also reduces the efficiency of processing to PacC 27 . Involvement of other proteasome subunits in Pre2p K108R shifting of the site of proteolysis to other scissile bonds is likely, because one such abnormal cleavage is impaired by additional inactivation of Pre3p. A. nidulans PreB Lys 101 is the equivalent of yeast Pre2p Lys 108 . We demonstrate that preB1 (K101A) is lethal, but preB2 (K101R) resembles its yeast counterpart in that it shifts the specificity of PacC processing toward otherwise minor cleavage sites, leading to a processing pattern that is remarkably reminiscent of that resulting from the corresponding yeast pre2 K108R allele. Thus, equivalent active site mutations in S. cerevisiae and A. nidulans PRE2/preB orthologues leading to partial loss of proteasome chymotryptic activity do not prevent processing but shift specificity from the major processing site toward normally minor (in A. nidulans) or barely detectable (in S. cerevisiae) sites, which implies that PacC is a direct substrate of the proteasome.
Substrates are recruited to the proteasome by polyubiquitindependent or -independent signals. Productive engagement additionally requires the presence of an unstructured region in the substrate where the protease begins unfolding (28,52). Unfolding is required for transfer into the catalytic chamber and precedes degradation, which proceeds sequentially from this initiation site (12,17,28,49,53). Processing, rather than complete degradation, results from release of partially proteolyzed substrates when the proteasome encounters a stop transfer signal involving a tightly folded domain, generally preceded by a low complexity sequence (12,49,53). Our results can be rationalized in the context of this model and the lack of sequence specificity of the proteolytic activities of the proteasome (44). Since the processed form PacC 27 retains the N-terminal zinc-finger domain, proteolysis must proceed toward the N terminus, and the actual peptide bonds accessible to the active sites of the proteolytic ␤ subunits will be dictated by their relative distance from the zinc finger domain. In the context of the wild-type proteasome, at least four bonds are scissile, although cleavage of peptide bonds involving the C-terminal residue in PacC 27 b and, to a lesser extent, that in PacC 27 a is strongly favored. Impairment of PreB markedly shifts preference from major to minor cleavage sites PacC 27 c and PacC 27 d, in striking similarity to the situation in yeast, suggesting that PacC 27 a and -b are attributable to the chymotryptic activity. In conjunction with yeast data, this additionally suggests that PreB impairment increases accessibility to the active sites of other proteolytic ␤ subunits.
Proteasome Active Site Mutations in A. nidulans-preB1 (K101A) and preB2 (K101R) are the first reported active site proteasome mutations in A. nidulans (although a ts mutation in the gene encoding an outer ring ␣-subunit has been described (54)). preB1 K101A results in inviability, which demonstrates that, as expected, the proteasome is essential. preB2 K101R allows viability and results in deficient proteasome chymotryptic activity. In S. cerevisiae, substitution by Ala of the corresponding Lys 108 is viable but impairs growth to a greater extent than substitution by Arg, possibly as a result of impaired proteasome maturation. Since propeptide autoprocessing is coupled to proteasome assembly (36,45,55), this difference appears to reflect that ␤5 K108R is fully autoprocessed, albeit less efficiently, whereas ␤5 K108A is only partially autoprocessed (45). Hence, that preB1 K101A results in lethality, whereas preB2 K101R allows viability is unsurprising.
The preB2 Proteasome Results in a Marked Reduction in PacC Levels-Steady-state levels of the wild type and the three tested mutant PacC proteins were markedly reduced in preB2 strains (Figs. 4 -6). This was unexpected, because preB2 results in elevated levels of a B-type mitotic cyclin, as expected from partially deficient proteasome-mediated proteolysis.
Since preB2 does not significantly reduce pacC transcription in wild type (nor, incidentally, in pacC910 cells) (supplementary Fig. S6), the reduced PacC levels due to the proteasome mutation in the wild-type (pacC ϩ ) and in three different pacC mutant strains are unlikely to be due to reduced synthesis. Hence, assuming that steady-state levels represent, at least in part, an assessment of stability, PacC appears unstable in preB2 cells. Two highly speculative, not mutually exclusive and, as yet, untested possibilities could explain this. One would be that cleavage events leading to PacC 27 a and -b, impaired by preB2, would facilitate release of the polypeptide chain from the proteasome, whereas cleavage by the other catalytic ␤-subunits would favor degradation. In the second, PreB K101R would lead to structural changes in the proteasome affecting accessibility and/or orientation of the PacC substrate to/at the catalytic chamber. The second possibility is supported by the growth defect of S. cerevisiae Pre2p K108R mutants, which probably results from impaired proteasome maturation (36,45,55). A. nidulans preB2 leads to marked cryosensitivity (supplementary Fig. S5), which would be consistent with a partial defect in the assembly of the proteasome complex.
The occurrence of four relatively close cleavage sites at the C terminus of PacC 27 most likely reflects the stochastic nature of the processing reaction, where the processing site depends on a distance dictated by a stably folded domain. The inability of the proteasome to unfold this domain cannot be absolute, and one unwanted lateral effect of proteasome-mediated processing is, inevitably, some complete degradation. In a model substrate, about one-third of the proteasome-mediated proteolytic events give rise to a processing product rather than to complete degradation (49). Hence, it seems plausible that the preB2 proteasome mutation showing altered specificity also alters the efficiency of processing to the benefit of degradation, particularly if it leads to changes in the orientation of the PacC substrate within the proteasome CP.
pacC910 and the Role of the Processing Efficiency Determinant-Additional support for a direct role of the proteasome in PacC processing came from the finding that the processing-impairing pacC910 mutation significantly ameliorates the low steady state levels of PacC resulting from preB2, but paradoxically, it does so by increasing levels of PacC 27 . Hence, PacC910 is recalcitrant to the processing protease in preB ϩ cells but leads to relatively elevated levels of the processed PacC 27 product(s) in preB2 cells. Since the pattern of cleavage sites of these products is characteristic of preB2 mutants, these results are strong evidence that the proteasome is the processing protease.
Conversion of pacC910 PacC 53 , lacking the PED, to PacC 27 is blocked both in preB ϩ and in preB2 backgrounds (Fig. 5). Thus, the PED contains one or more essential elements for PacC 53 processing to PacC 27 . These might include a recruitment signal and a loosely folded domain, where the proteasome is engaged. Interactive regions A and B within the PED (Fig. 1A) are most likely unstructured in PacC 53 and would provide a suitable domain(s) for productive proteasome engagement (28,52). Proteins are usually recruited to the proteasome after their polyubiquitination (1). Thus, we investigated the role in processing of Lys 337 and Lys 267 within the PED. A quadruple Lysto-Arg substitution involving these residues plus otherwise processing-silent Lys 252 and Lys 253 largely prevents processing to PacC 27 , which would agree with a hypothetical role of these Lys residues in ubiquitin-mediated recruitment of PacC to the proteasome. However, our attempts to detect ubiquitinated PacC have been unsuccessful thus far. One possibility is that the role that these Lys residues play in processing does not involve ubiquitination, and PacC is recruited to the proteasome in an ubiquitin-independent manner (see Ref. 56 and references therein). One notable example of ubiquitin-independent proteasome recruitment is that of NF-B p105, whose processing to p50 can be mediated by the proteasome CP in the complete absence of ubiquitination (8). Possibly basal and inducible mechanisms exist for induction of p105 processing. Ubiquitin-independent recruitment would suffice for basal levels of processing, whereas signal-induced activation would be most efficient because of the reinforcing effect that polyubiquitination would have on "basal" substrate recruitment to the proteasome (8). We note that such basal processing would explain the detectable level of signal-independent processing that occurs in the pH regulatory system (21) and that might underlie the low levels of processing seen with PacC910.
One feature of the currently accepted model for proteasomemediated processing is the presence of a low complexity region preceding the tightly folded domain that interferes with the molecular mechanisms pulling the substrate into the catalytic chamber, decreasing further the likelihood that the domain is unfolded (and thus fully degraded) by the proteasome (12,49). This is an important feature but not an essential one, since NF-B p105 is processed in S. cerevisiae in the absence of the GRR (13). We deleted a Gly-Ala-Pro-rich region within the PED located ϳ40 residues C-terminal to the major processing cleavage site and found that it leads to some impairment of processing but, most notably, results in apparent destabilization of PacC.
PacC53 Undergoes Multiple Phosphorylation-PacC 53 and PacC 72 are multiply phosphorylated (this work and, for PacC 72 , see Ref. 50). We show here that the extent of phosphorylation of PacC 53 increases with time of exposure to the alkaline pH signal and that the quadruple Lys substitution impairing PacC 53 processing to PacC 27 additionally results in its excessive phosphorylation (Fig. 8), as expected if ubiquitination were coupled to phosphorylation. Such a sequential mechanism has been reported for NF-B p105 (6) but, most notably, for the processing of Ci/Gli transcription factors, where a phosphorylation cascade promotes ␤TrCP recruitment to mediate ubiquitination and subsequent proteolytic processing (5,(57)(58)(59)(60). The remarkable similarities with the zinc finger DNA binding domains and the mechanisms of regulation of Ci/Gli have been underlined previously (20).