Apicomplexan UCHL3 retains dual specificity for ubiquitin and Nedd8 throughout evolution

Ub protease activity is present in T. gondii-infected human foreskin fibroblasts

Enzymes that remove the post-translational modifiers Ub, Nedd8, SUMO and ISG15 can be assayed in cell lysates by using epitope-tagged electrophilic derivatives of these proteins (Borodovsky et al., 2001; Hemelaar et al., 2004). This strategy was used to identify proteases from a variety of sources (Borodovsky et al., 2002; Hemelaar et al., 2004; Kattenhorn et al., 2005; Artavanis-Tsakonas et al., 2006; Misaghi et al., 2006). To generate the probe, these small-protein modifiers are equipped with an electrophile such as vinylmethylester (VME) or vinylsulfone (VS) at their C-terminus. When a DUB attacks the modified Ubl, a thioether bond is formed between the active-site cysteine of the target protease and the probe. The epitope tag on the probe can be used to detect these interactions immunochemically or can be employed for retrieval of the target protease. The probes used in this study are based on the mouse sequences of Ub, Nedd8 and SUMO1. Mouse and T. gondii Ub differ in only one amino acid (E16D) while Nedd8 and SUMO exhibit approximately 50% identity between the two species.

To determine if T. gondii-infected human foreskin fibroblasts (HFFs) contain non-host proteins either contained within or secreted from T. gondii with deubiquitinating activity, we prepared lysates of HFFs 48 h after infection with RH strain T. gondii tachyzoites and reacted the lysates with HA-tagged Ub-VME. Immunoblotting revealed the presence of several bands unique to the infected sample, the most prominent being a polypeptide of about 40 kDa as well as a number of polypeptides with a molecular mass greater than 150 kDa (Fig. 1A). The presence of these specific bands prompted their identification.

Identification of Ub-VME-reactive proteins from T. gondii

We scaled up the DUB labelling reaction by using cell lysates from 3 × 10⁸ HFFs 48 h after infection with T. gondii RH at a multiplicity of infection (moi) of 5, followed by incubation with HA-Ub-VME. We chose not to separate parasites from cells so that DUBs potentially secreted by the parasites into the host cell could also be detected (Misaghi et al., 2006). The reacted material was immunoprecipitated, separated by SDS-PAGE and the polypeptides were visualized by silver staining (Fig. 1B). Polypeptides specific to the T. gondii-infected sample were excised, trypsinized and subjected to tandem mass spectrometry (MS/MS). Four T. gondii-specific polypeptides were identified, all of which corresponded to annotated putative C-terminal Ub hydrolases with ToxoDB Accession No. 80.m00082, 55.m05059, 31.m00905 and 55.m05062 (Fig. 1B). blast searches revealed the closest mammalian homologues for these enzymes to be USP7, USP4, UBP5 and UCH-L3 respectively. All genes contain multiple introns. USP7, USP4 and UBP5 have ambiguous initiation and termination sites as extracted from the available database. RACE PCR analysis of the USP4 homologue (55.m05059) revealed a misannotation in ToxoDB. The C-terminal 71 amino acids, highlighted in blue in Fig. 1C, are absent from the predicted open reading frame (ORF) for this protein. A blast search of the sequence of our 3′ RACE PCR product against the Toxoplasma genome reveals that this stretch includes two exons on chromosome VIIb, 14725 base pairs downstream of the predicted termination of the protein. The correct sequence is shown in Fig. 1C, together with the extent of peptide coverage obtained by MS/MS. Moreover, we find a peptide that maps to this missing stretch of the protein by mass spectrometry. Sequence alignment with HsUSP4 shows a 27% identity and 40% similarity between the proteins, with all catalytic residues conserved (data not shown). The UBP5 and USP7 sequences have not been verified and are shown in their currently annotated state in Fig. S1.

We decided to further investigate the UCH-L3 homologue TgUCHL3 (Fig. 1D), as the mammalian enzyme has been well characterized (Johnston et al., 1999; Misaghi et al., 2005) and provided us with a well-defined backdrop against which to describe its parasite counterpart. We were particularly interested in the ability of UCH-L3 to both deubiquitinate and deNeddylate target proteins (Wada et al., 1998; Hemelaar et al., 2004) and to determine whether this feature has been conserved throughout evolution.

UCHL3s from T. gondii and P. falciparum also have deNeddylating activity

To verify the specificity of the deubiquitinating activity, as well as test for deNeddylating and possible deSUMOylating activity of TgUCHL3, we cloned the ORF of TgUCHL3 from T. gondii cDNA. We further generated a catalytically inactive mutant by substituting the putative active-site cysteine with an alanine. Both constructs were translated in vitro and reacted with Ub-VME, Nedd8-VS and SUMO-VS.

TgUCHL3 reacted with both the Ub and the Nedd8 probes, as seen from the 9 kDa shift in electrophoretic mobility (Fig. 2A), but failed to be modified by the SUMO probe (data not shown). Nedd8 and Ub reactivities were abolished by pre-incubation of extracts and of TgUCHL3 with the alkylating agent N-ethylmaleimide (NEM), confirming the nature of this enzyme as a cysteine protease. This was further supported by the inability of the cysteine to alanine mutant to be modified by either probe. Interestingly, T. gondii Nedd8 and SUMO both share a similar degree of conservation to their human counterparts. Thus, we conclude that adduct formation is not an occurrence that applies across the board to all electrophilic Ubl derivatives.

To determine whether the homologous gene product from P. falciparum also reacts with both Ub-VME and Nedd8-VS, we cloned the ORF of PfUCHL3 (NP_702465) from total 3D7 mRNA by reverse transcription polymerase chain reaction (RT-PCR), as this gene contains multiple introns. We noted a difference in the annotated sequence from the Plasmodium database versus the gene product we cloned. Comparison of the intron donor/acceptor sites with the genomic sequence of this gene validates our cloned sequence, which is further supported by sequence alignment with other UCHL3 homologues (Fig. 3A). PfUCHL3 was in vitro translated and reactivity with both the Ub as well as the Nedd8 probes was demonstrated (Fig. 2B). PfUCHL3 also failed to react with a SUMO probe (data not shown). Again, inclusion of NEM and mutation of the catalytic cysteine to an alanine abolished probe modification. We concluded that both UCHL3s from T. gondii and P. falciparum, which are only 32% identical and 42% similar, retain dual specificity towards Ub and Nedd8, as described for human UCHL3 (Hemelaar et al., 2004).

TgUCHL3 localizes to both the nucleus and cytoplasm of T. gondii

We next wanted to investigate the subcellular localization of TgUCHL3 in the tachyzoite stage of T. gondii. To this end, we infected HFFs with T. gondii RH stably expressing TgUCHL3 C-terminally fused to YFP and incubated infected cells with a nuclear dye immediately before imaging. Using a spinning-disk confocal microscope we obtained z-stack images of HFFs infected with parasites. In some cases a single parasite was present within the parasitophorous vacuole, whereas in others the parasites had multiplied. An example of the latter is shown in Fig. 3. We found that TgUCHL3 localized to both the nucleus and the cytoplasm of T. gondii and was not secreted into the parasitophorous vacuolar space, This distribution is not surprising, as other DUBs have been localized both nuclearly and cytoplasmically (Soboleva et al., 2005). Furthermore, the function of TgUCHL3 as a deNeddylating enzyme, which likely regulates degradation of cell cycle factors as seem in mammals (Chiba and Tanaka, 2004), supports a nuclear localization.

DeNeddylating activity of UCHL3 is conserved throughout evolution

To gain insight into the origin of the dual specificity of UCH-L3, we performed a phylogenetic analysis of the sequence of UCHL3 in several species. Notably, we were unable to find any UCHL3 orthologues in species like Theileria parva, Entamoeba histolytica and Babesia bovis, whereas we only found two UCHL3-like pseudogenes in Tetrahymena thermophila. We aligned the amino acid sequences of UCHL3 in multiple organisms, ranging from parasites to mammals (Fig. 4A). This alignment shows the presence of two similar large-amino-acid insertions in the sequences of T. gondii and wild cabbage Brassica oleracea UCHL3 (BoUCHL3). We next deduced the phylogenetic relationship of the aligned UCHL3 sequences by maximum likelihood and parsimony analyses. As shown in Fig. 3B, the results of both analyses are consistent with the phylogenetic relationship of the species under study. Notably, the sequence of T. gondii UCHL3 clusters only loosely with the sequences of fellow apicomplexan Cryptosporidium parvum and P. falciparum UCHL3, in spite of the close phylogenetic relationship between T. gondii and C. parvum. On the other hand, the sequence of BoUCHL3 robustly clusters with that of Arabidopsis thaliana UCHL3 (AtUCHL3), also a plant, which suggests that the insertions in TgUCHL3 and BoUCHL3 are not phylogenetically related. Finally, we created a homology model of the structure of TgUCHL3 using HsUCHL3 as a template (Misaghi et al., 2005). This structure shows that the large insertion in the sequence of TgUCHL3 maps to the surface of the protein and does not affect the interaction site with ubiquitin (Fig. 5). However, an analysis of allowed conformations suggests that this insertion may interact with ubiquitin outside the canonical interaction site (Fig. 5).

Cells and parasites

RH strain T. gondii were grown in confluent HFFs maintained in DMEM, 10% fetal calf serum (FCS), penicillin/streptomycin and fungizone (Roos et al., 1994). To generate T. gondii RH that stably express TgUCHL3 as a C-terminal YFP fusion protein, tubTgUCHL3YFP/sagCAT (see below) was transfected into the tachyzoites by electroporation as previously described (Roos et al., 1994) and selected with 2 μM chloramphenicol.

Labelling and immunoblot

Human foreskin fibroblasts were cultured in DMEM with 10% FCS, 4 mM glutamine, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. Forty-eight hours after infection with T. gondii strain RH at an moi of 5, the cells were lysed in 100 mM Tris/HCl pH 7.4, 20 mM NaCl, 70 mM KCl, 12 mM MgCl₂, 1 mM phenylmethylsulfonylfluoride, 1 mM dithiothreitol, 0.5% NP-40 for 20 min at 4°C. Cell debris was removed by centrifugation and 25 μg of total lysate protein was diluted with 2 volumes 50 mM Tris/HCl pH 8, 100 mM NaCl and reacted with 0.2 μg of HA-Ub-VME for 1 h at room temperature. To block reactive sulfhydryl groups of cysteines, the lysate was pre-incubated with 10 mM NEM for 15 min at room temperature. Lysates were separated by gel electrophoresis and proteins were transferred to a PVDF membrane, blocked with 5% milk before being probed with 1:1000 anti-HA-HRP antibody (Roche, Mannheim, Germany). After six successive washes in PBS-Tween proteins were visualized by chemiluminescence.

Identification of labelled DUBs using mass spectrometry

Human foreskin fibroblasts (3 × 10⁸) were lysed 48 h after infection with T. gondii strain RH at an moi of 5. The samples were processed as described above except that 4 mg of total lysate protein was reacted with 40 μg of Ub-VME. Gel bands were excised and digested with trypsin. The resulting peptides were analysed by liquid chromatography/MS/MS and the data correlated against the Tg10x_31-Draft3_TIGR database using SEQUEST. All proteins reported were identified by the presence of two or more peptides.

Cloning and mutagenesis

Toxoplasma gondii total RNA was isolated from unsynchronized tachyzoite RH culture using Trizol reagent. cDNA was prepared using the SMART RACE cDNA amplification kit (Clontech, Mountain View, CA). TgUCHL3 coding cDNA (55.m05062) was amplified by PCR using ToxoDB to design the primers TgUCHL3For: 5′-GAGATCTGGATCCCCATGGAGGGGAAGAGC and TgUCHL3Rev: 5′-GCTCTAGAGCCTACGCGGCCTTGGCGTC. The resulting PCR products were digested with BamHI and XbaI and ligated into the multicloning site of pcDNA3.1(+) to yield pcDNA3TgUCHL3. This construct was used as a template to generate the cysteine to alanine mutant of TgUCHL3 using Stratagene's QuikChange II site-directed mutagenesis kit (La Jolla, CA) and the primers 5′-GTCGGTAATGCAGCCGGCACTGTGGCCCTGCTCC and 5′-GGAGCAGGGCCACAGTGCCGGCTGCATTACCGAC. In order to generate TgUCHL3 C-terminally fused to YFP the gene was re-cloned into the BglII and AvrII sites of tubYFP-YFP/sagCAT (Gubbels et al., 2003). This yielded the vector tubTgUCHL3YFP/sagCAT.

Total mRNA from 3D7 P. falciparum was prepared as described in Kyes et al. (1999). Using the PlasmoDB prediction for the coding sequence of Pf14-0576, primers were designed against the 5′-CGGAATTCATGGCAAAGAATGATATTTGGAC and 3′-CCGCTGAGTATAATATCAAAGTTATCGTTTGG ends with EcoRI and XhoI linkers respectively, for directional cloning. Following Pfu amplification, the product was digested and cloned into pcDNA 3.1(+) in frame with a C-terminal HA epitope tag. In order to generate the cysteine to alanine mutant the same strategy as described above was used with the primers 5′-TACATTCCAAACTCAGCTGGAACCATAGCCTTG and 5′-CAAGGCTATGGTTCCAGCTGAGTTTGGAATGTA.

In all cases competent Escherichia coli were transformed with the plasmids and resulting colonies were grown for DNA isolation and sequence verification. Sequencing of PfUCH24 revealed a slightly different coding sequence than that predicted by PlasmoDB. The cloned amino acid sequence is shown in Fig. 3A.

Both 5′ and 3′ cDNA ends for TgUSP4 were amplified by RACE PCR using the Invitrogen GeneRacer kit and primers 5′-GAAGCCAGACGCGCAAGTGGCTGAGC and 3′-CCCCAGCTGCGGCTCTCACTTCTGC. Nested RACE products (using the same gene-specific primers but nested universal primers) amplified with TripleMaster polymerase mix (Eppendorf) were TOPO cloned into plasmid pCR2.1 (Invitrogen) and sequenced.

In vitro translation and labelling

Template plasmid DNA (200–300 ng) of either TgUCHL3 or PfUCHL3 as the wild type or the cysteine to alanine variant was added to 25 μl of rabbit reticulocyte lysate TNT T7 Quick Coupled Translation/Transcription Sytem mastermix (Promega, Madison, WI). The reactions were supplemented with 1 μCi of ³⁵S-methionine (Perkin Elmer) each and incubated at 30°C for 45 min. Each sample mixture was then reacted with 0.2 μg of the probe as described above, subjected to SDS-PAGE (10% polyacrylamide) and analysed by autoradiography.

Live microscopy

Human foreskin fibroblasts were infected with T. gondii RH stably expressing tubTgUCHL3YFP/sagCAT. Twenty hours after infection 1 μM DRAQ5 (Alexis Biochemicals, San Diego, CA) was added for 5 min and the cultures were imaged on a spinning disk confocal microscope (Perkin Elmer Ultraview RS, Boston, MA) with a Prairie 3 watt laser with AOTF (Prairie Technologies, Middleton, WI). The system incorporated a Nikon TE2000-U inverted microscope using a Nikon 100× 1.4NA DIC lens and Nikon type A immersion oil (Melville, NY). Cells were maintained at 37°C with 5% supplemental CO₂ in room air using a Solent scientific chamber (Segensworth, UK), which fully enclosed the microscope stage area. The camera used in these experiments was the Hamamatsu Orca ER camera (model number C4742-95-12ERG). Metamorph software (Molecular Devices, Sunnyvale, CA) was used for acquisition.

Phylogenetic analysis

The sequences of UCHL3 from 12 organisms including metazoans, fungi, plants, apicomplexans and Trypanosoma brucei were aligned automatically with clustalx version 1.8 (http://bips.u-strasbg.fr/en/Documentation/ClustalX/) and then manually with genedoc version 2.6. (http://www.psc.edu/biomed/genedoc). The programs used for phylogenetic inference are included in the Phylip package version 3.6 (http://evolution.genetics.washington.edu/phylip/getme.html). After producing 100 bootstrapped replicates of the original alignment with the bootstrap program, the most parsimonious tree was calculated with the protpars program. A second tree was constructed by maximum likelihood analysis with the proml program. The consense tree was calculated after both analyses with the consense program, using the maximum likelihood algorithm and forcing a 60% minimum bootstrap score to display a branch. TbUCHL3 was used to root the trees.

Homology modelling

The sequence of TgUCHL3 was manually threaded over the structure of HsUCHL3 (Misaghi, 2005) using DeepView v3.7. The resulting alignment was used to build a model of the tertiary structure of TgUCHL3 with Swiss-Model (http://swissmodel.expasy.org/). Residues 30–48 of the resulting model, corresponding to a large insertion in TgUCHL3 sequence, were rebuilt with DeepView. We analysed multiple possible conformations of this insertion according to similar sequences in proteins of known structure.