A TBlastN comparison of USP25 sequence against nr database at the National Center for Biotechnology Information (NCBI) server revealed a homologous USP sequence on PAC clone pDJ356d6 (GenBank accession number AC002036) located at 11q23. Specific probes designed following this genomic sequence were used to screen 10⁶ recombinant phages of a human fetal brain cDNA library (Clontech). Six positive clones were isolated, subcloned and sequenced. Four clone sequences were chimeric and were thus discarded. Two overlapping cDNA clones, 5A11 (2.3 kb) and 3A11 (1.5 kb), showed an exact match with chromosome 11 PAC clone pDJ356d6. Evidence that clones 5A11 and 3A11 did not encode the full-length cDNA was based on the protein alignment between the deduced amino acid sequence and USP25, the mRNA size obtained from northern analysis when probed with cDNA clone 5A11, and the absence of the strictly conserved UBP cysteine domain in either clone 5A11 or 3A11.

Three different strategies were followed to isolate the 5' end of USP28 cDNA. First, a chromosome 11-specific forward primer (5.2 cr11F) was designed from the TBlastN analysis of USP25 (Figure 1). This sequence corresponded to the 5'-terminal sequence of the predicted exon encoding the cysteine domain in the chromosome 11 PAC DNA. A reverse oligonucleotide, 11race4, was also designed on the basis of the 3A11 cDNA clone sequence (Figure 1). An RT-PCR reaction using heart mRNA as template was subsequently performed with the forward and reverse primers. The expected 958 bp amplified fragment was subcloned in pUC18. Its sequence overlapped with 3A11 and 5A11 cDNA clones, thus allowing the characterization of an additional 908 bp. The overall cDNA sequence, although still incomplete, contained the cysteine domain, together with all other UBP reported signatures.

Second, 10⁶ recombinant phages from kidney and placenta cDNA libraries (Clontech) were screened using three different 5'-USP28-specific probes obtained after PCR amplification with primers deduced from the TBlastN comparison to USP25 (Figure 1). Five different positive clones were isolated: four from kidney and one from placenta. All inserts were subcloned in pBLUESCRIPT SK+ (Stratagene). Sequence analysis revealed a further 318 bp, although the cDNA was still incomplete.

Third, a 5' RACE experiment was performed on 10⁹ phages from the kidney cDNA library with two specific reverse primers (11race6 and 11race7) (Figure 1) and two λgt10 vector-derived forward primers (GR1, GR2). The comparison with the PAC clone PDJ105h16 sequence (htgs database) allowed us to identify the first presumptive methionine. However, as no further in-frame stop codons were detected, the use of another initiation codon could not be ruled out. The search for a putative TATA box in the upstream genomic sequence rendered no positive results.

No cDNA clones with a poly(A)⁺ tail could be isolated, but BlastN comparisons against the dbEST database allowed the identification of an overlapping EST (AI337094) containing the polyadenylation signal.

USP28 cDNA (accession number AF266283) is at least 3,624 bp long and encodes 1,077 amino acids with a relative molecular mass of 122.4 kDa. The gene is organized in 25 exons and exon-intron boundaries were determined after comparison with the human htgs database (Table 1). Average exon length was 130 bp, the shortest being 71 bp and the longest 229 bp. Nucleotide and amino acid identitiesbetween USP28 and USP25 (Figure 2) are 55.77% and 51.36%, respectively. USP28 is more similar to USP25 than to any other known UBP. Besides, exon-intron boundaries between USP28 and USP25 are highly conserved (Figure 2). The homology between USP28 and USP25 with other UBP family members appears to be confined to the reported conserved domains.

Northern analysis for USP28 was performed using the cDNA clone 5A11 as probe (Figure 3). A single transcript of approximately 4.5 kb was identified in heart and skeletal muscle after short film exposure. At longer exposures a transcript of the same size was detected in all assayed tissues, albeit at a very low level. To compare the amount of RNA poly(A)⁺ loaded in each lane an actin probe was used as a control.

Isoforms generated by alternative splicing have been described for some vertebrate UBPs [13,15]. Sequence analysis of the isolated USP25 cDNA clones revealed four alternatively spliced exons. Three of those generated an in-frame stop codon in some human fetal brain cDNA clones and were located between exons 3 and 4, 10 and 11, and 17 and 18 (exons numbered according to [14]. The fourth alternatively spliced product corresponded to exon 19 and added 32 amino acids to the protein product (exon 19 has been renamed 19 b in this work). It had been previously identified in a single I.M.A.G.E consortium neuroepithelium cDNA clone (id: AA209364 [16]). In order to verify all these putative exons, specific sets of primers (Table 2) were designed for RT-PCR amplification on 24 different tissues (Multiple Tissue cDNA Panels, Clontech).

The additional sequences between exons 3-4 and 17-18 were not further observed in any tissue tested. However, the alternativesequence between exons 10 and 11 (named exon 10b) was clearly detected and verified by sequencing in testis (Figure 4a), where both transcripts (exons 10-11 and 10-10b-11) had been amplified at similar levels. Exon 10b was also detected, albeit faintly, in small intestine, spleen and peripheral leukocytes. This exon encompasses 124 bp and generates a stop codon in the open reading frame (ORF) at nucleotide 1,453. Therefore, this transcript would produce a 361 amino acid truncated protein where the cysteine, but not the histidine UBP domain, would be preserved.

RT-PCR experiments performed with specific primers, 321F-321R (located in exons 17 and 19b, respectively) and N1-121R (in exons 18 and 20, respectively) (Table 2), allowed us to visualize three mRNA isoforms. The shortest, already identified in many expressed sequence tags (ESTs), showed exon 18 directly fused to exon 20 and was detected in all the 24 different tissues tested (Multiple Tissue cDNA Panel, Clontech). The second isoform contained exons 18-19b-20, was of intermediate size and corresponded to the reported USP25 [14]. This isoform was detected in all the assayed tissues except fetal and adult muscle and heart (Figure 4a). Finally, the longest isoform was only detected in fetal and adult muscle and heart mRNAs and included a new exon (named 19a) which was always fused to 19b (18-19a-19b-20) in these tissues (Figure 4a). Exon 19a contained 114 bp that added 38 amino acids to the reported USP25 protein.

Primers 5A11.2 and 5A11.1 from USP28 (Table 2), located respectively in exons 16 and 24, were used to verify the homolog of the USP25 muscle and heart isoform (exons 18-19a-19b-20). RT-PCRs were performed in different tissues and the amplified products were subcloned and sequenced. Two USP28 sequences were detected in adult and fetal muscle, heart and brain. The shortest fragment corresponded to the reported USP28 cDNA containing exon 19b (Figure 4b). The longest contained a new exon (named 19a) (Figure 4b), which introduced 62 amino acids to the reported ORF. The tissue-specific exon 19a from USP25 (38 amino acids) and USP28 (62 amino acids) showed 41.6% amino acid identity. In contrast to USP25 exon 19b, the exon 19b of USP28 (sharing 53.1% amino acid identity) was not alternatively spliced, as it was always present in the transcripts from all tissues assayed.

To test the ubiquitin-specific protease activity of USP28, USP25 and the muscle and heart USP25 isoform (iUSP25), these enzymes were synthesized as inducible fusion proteins in the pGEX-4T-1 vector (pGEX-USP25; pGEX-iUSP25 and pGEX-USP28). Escherichia coli XL1blue cells were cotransformed with each construct and a plasmid expressing the fusion protein Ub-M-β-gal, a synthetic substrate for UBP. Immunoblot analysis using anti-β-galactosidase (β-gal) antibodies showed an efficient cleavage of ubiquitin from the Ub-M-β-gal fusion protein (Figure 5, lanes 5, 6 and 7). As expected, cells with the control pGEX-4T-1 vector failed to cleave Ub-M-β-gal (lane 4). The endogenous E. coli XL1-blue β-galactosidase (lacking the α peptide, and thus producing a shorter protein) was also detected (lanes 2 and 8).

As USP25 is located on chromosome 21, its overexpression could be presumed in Down syndrome (DS) patients. In order to assess the expression level of USP25 in DS versus control samples, we analyzed eight independent fetal brain samples (four trisomic and four disomic) by real-time quantitative PCR. An average of 1.7-fold overexpression of USP25 was shown in trisomic versus disomic samples.

To screen cDNA libraries from human fetal brain, placenta and kidney (Clontech), specific primers of the USP-like region on chromosome 11 were designed (Figure 1) after comparison of USP25 cDNA against AC002036, a PAC that contained genomic DNA from chromosome 11. Approximately, 10⁶ phages of each library were plated and transferred onto Nylon membranes (HybondN, Amersham-Pharmacia Biotech).

Probes were labeled by random hexamer priming with [α-³²P]dCTP. A solution containing 50% (v/v) formamide, 5× SSC, 5× Denhart's, 0.1% (w/v) SDS and 100 μg/ml denatured salmon sperm was used to pre-hybridize and hybridize the filters for 2 h and 18 h, respectively. Washes were performed as follows: two washes in 2× SSC/0.1%SDS at 65°C for 10 min and two times in 1× SSC/0.1%SDS at 65°C for 10 min. Filters were exposed during two days with double screen in order to amplify the positive signals. Positive clones were subcloned into pBLUESCRIPT SK+ (Stratagene), sequenced in an ABI377 automatic sequencer (Applied Biosystems) and analyzed at the NCBI BLAST server [26]. As only truncated cDNA clones were initially isolated, additional screenings were needed to identify the full-length cDNA.

The cDNA clone 5A11 (from nucleotides 1,473 to 3,618, Figure 1) was used to probe a human multiple tissue northern blot (Clontech). Hybridization and washes were carried out according to the manufacturer's protocol. A control actin probe was used to assess the amount of RNA poly(A)⁺ loaded in each lane.

A set of primers corresponding to exon sequencesof USP25 and USP28 (Table 2) (USP25 primers numbered according to [14]) were designed for RT-PCR experiments on 24 samples from different tissues (Multiple Tissue cDNA Panels, Clontech). Amplification was performed in a total volume of 25 μl containing: 2.5 μl of template cDNA, 200 μM dNTPs, 5 pmol of each primer, 1× Taq Platinum buffer, 1.5 mM MgCl₂ and 1U Taq Platinum (GIBCO-BRL). After denaturing at 94°C for 3 min, two-step PCR was carried out for 35 cycles at 94°C for 30 sec and 58°C for 30 sec. Final extension was for 5 min at 72°C. Bands differing from the expected size were subcloned in pBLUESCRIPT SK+ (Stratagene) and subsequently sequenced in an ABI377 automatic sequencer (Applied Biosystems). Sequences were analyzed using the BLAST software at the NCBI [26,27].

The ubiquitin-specific protease activity of USP25, USP25-specific heart and muscle isoform, which contains the additional exon 19a (USP25 isoform), and USP28 was determined as described [10,11]. The three corresponding cDNAs were cloned in-frame in pGEX-4T-1 (Amersham Pharmacia Biotech) downstream from the glutathione-S-transferase (GST) coding region. Plasmid pACY184 Cm^r expressing Ub-Met-β-gal protein fusion (pAC-M-β-gal) was kindly provided by M. Hoschtrasser. E. coli XL1blue bacteria were co-transformed with pAC-M-β-gal and either pGEX-4T-1-USP25 or pGEX-4T-1-USP25 isoform or pGEX-4T-1-USP28. Amp^r and Cm^r colonies were grown and induced for 3 h with isopropyl-β-thiogalactopyranoside (final concentration 1 mM). Total protein extracts were analyzed by western blotting with anti-β-galactosidase rabbit polyclonal antibody (Cappel).

Total RNA from 5 mg of fetal brains of Down and control samples (three samples each) were obtained using ABI PRISM 6700 Automated Nucleic Acid workstation. RT-PCRs were produced according to Taqman Reverse transcription reagents (Roche Molecular Systems). Quantitative PCRs were performed using the Universal Master Mix (Applied Biosystems) following the manufacturer's specifications. The real-time amplification was analyzed by the ABI Prism 7700 Sequence detection system. USP25 primers and the Taqman probe (using FAM as reporter and TAMRA as quencher) were designed according to the Primer Express software. Forward primer: 5'-GATGAAAGGTGTCACAACATAATGAAA-3'; reverse primer: 5'-CCACTCCTCATATTCCTCCAAGTTT-3'; TaqMan probe: 5'-TCAAGCCAAACTGGAAATGATAAAACCTGAAGA-3'. To normalize the USP25 quantitative determinations, GAPDH (Applied Biosystems) was used as endogenous control. The standard deviation of the disomic and trisomic samples was 0.08 and 0.13, respectively. The overexpression (1.7 fold) of USP25 in trisomic fetal brains shows statistical significance according to the Mann-Whitney test (P < 0.05).