PfUCH37 is mainly cytoplasmic
To investigate PfUCH37 localisation and protein interactions, we generated a transgenic parasite line harboring an extra episomally-encoded gene copy with a carboxy-terminal HA epitope tag (Figure 1A). Parasites were transfected with the plasmid and grown continuously under blasticidin pressure to maintain the episome. Expression of the tagged protein was confirmed by immunoblot (Figure 1B) and localisation was probed by immunofluorescence (Figure 1C). Unlike the distribution of UCH37 in mammalian cells which is mainly nuclear and a bit cytoplasmic 19,20, PfUCH37 was found to be more cytoplasmic than nuclear.
PfUCH37 associates with the proteasome
The subcellular distribution of PfUCH37 alongside the lack of identifiable INO80 homologs would suggest that, in Plasmodium, this enzyme lacks the defined nuclear function it has in higher eukaryotes. To probe this question, wild type 3D7 and PfUCH37-HA-expressing parasites were lysed using 1% Triton X-100, immunoprecipitated using anti-HA resin and captured proteins were eluted with HA peptide before being subjected to LC-MS/MS mass spectrometry.
Samples were run in triplicate and data was processed using MaxQuant and Perseus. A two-sample Student’s t-test was performed to identify proteins co-immunoprecipitated with PfUCH37. The results of this analysis were used to generate a volcano plot (Figure 2A), which shows 26 proteins as significantly enriched through interaction and co-immunoprecipitation with PfUCH37.
Among the 26 proteins that were significantly enriched in the PfUCH37-HA test samples were 19 proteasomal subunits (Figure 2B). These were mainly subunits of the 19S regulatory particle lid and base, although one beta subunit of the 20S core particle was pulled down as well (Figure 2C). All known subunits of the 19S regulatory particle co- immunoprecipitated with PfUCH37, except for subunit Rpn15. Six other non-proteasomal proteins were also identified; among these were N- ethylmaleimide-sensitive fusion protein and parasitophorous vacuole membrane protein S16, an integral surface membrane protein of sexual stage parasites 21 which is particularly well-expressed at the onset of gametocytogenesis 22. A steroid dehydrogenase with putative- 3-hydroxyacyl-CoA dehydrogenase, oxidoreductase, and fatty acid biosynthetic processing activity was also identified 23 as was a putative importin-7 protein involved in GTPase binding and protein import into the nucleus as a nuclear transport receptor or through association with the importin-beta subunit KPNB1 24. A conserved Plasmodium protein of unknown function (PF3D7_1206800) and the DNA replication licensing factor MCM2 were the final two non-proteasomal proteins to be identified.
In humans, UCH37 interacts with the DEUBAD (DEUBiquitylase ADaptor) domains of Rpn13 or NFRKB. Protein sequence and structural homology searches did not identify DEUBAD domains in any putative interactors identified. We also generated structural models and tested the likelihood of interaction with PfUCH37 using AlphaFold 3 25.
We used the interaction of human UCH37 and ADRM1 as a reference for the accuracy of the prediction 5 which yielded a pTM score of 0.79 and ipTM of 0.83. The interaction scores for the rest of the proteomic hits were well below this score (Figure 3) indicating that they interact with PfUCH37 indirectly, or were captured non-specifically. The only one yielding an appreciably confident score was the interaction of PfUCH37 with PfRpn13. Although the prediction for a PfUCH37 and PfS16 complex resulted in a relatively high pTM score of 0.62, the ipTM score suggested a lack of confidence in the predicted interaction.
PfUCH37 and PfRpn13 interact in a similar manner to the orthologous human proteins
Since the interaction between PfUCH37 and the proteasome was most likely given our analysis and the defined interaction of the orthologous proteins in mammalian cells, we set out to biochemically validate this result.
In human and mouse cells, UCH37 is known to interact with the proteasome by binding Adrm1/Rpn13. Specifically, The carboxy-terminal regions of both Rpn13 and UCH37, which contain KEKE motifs, play a crucial role in their physical interaction and assembly of the 26S proteasome. A KEKE motif is defined as a sequence longer than 12 amino acids, devoid of W, Y, F, or P residues, composed of more than 60% lysine (K) and glutamic acid/aspartic acid (E/D), and lacking five consecutive residues that are positively (K, R, H) or negatively (D, E) charged. These motifs are recognized for their role in mediating protein-protein interactions 26. We observed that PfUCH37 exhibits a KE-rich regions at its C-terminal ends (Figure 4A). Although the C-terminal KE-rich region of PfUCH37 does not fit the previously defined KEKE motif, it is abundant in lysine and glutamic acid residues. Therefore, we used Alphafold 3 25 to dock models of these two proteins and determine the most likely conformation. The ipTM score was 6 which is relatively confident and the proteins do appear likely to interact via their KEKE domains (Figure 4B). Overlaying the predicted Plasmodium complex interface with the solved crystal structure of the human proteins indicates that the interactions occur in a very similar manner. However, amino acid sequence alignment of the interaction domains of each protein revealed a number of differences (Figure 4A). The structural conservation of the DEUBAD domains in PfRpn13 and HsRpn13 is evident in their comparable binding modes to UCH37. In both species, the core DEUBAD domains interact with the C-terminal tail of UCH37. In human UCH37, the amphipathic helix α11 in the C-terminal tail is clasped by the DEUBAD domains and further stabilized by helix α12, forming an extensive hydrophobic interface 5. AlphaFold predicts a similar interaction for PfUCH37. Thus, we hypothesized that the binding of the Plasmodium DEUBAD domain also requires these helices, as a human UCH37 variant lacking them (UCH-L5Δα11-12) fails to interact with RPN13 5.In HsUCH37, GLU300 forms critical hydrogen bonds with ARG309 and TYR313 in Rpn13. However, in PfUCH37, this residue is replaced by SER440, which cannot form similar bonds with the DEUBAD domain. To identify the amino acids involved in the protein-protein interaction and compare the PPI interface between these two species, we used PDBePISA 200727. The results, summarized in Supplementary Data, revealed differences in both the interacting residues and the types of bonds formed. Given that UCH37 and Rpn13 are essential proteins in Plasmodium falciparum, these differences could be targeted for the design of specific small-molecule inhibitors of the PfUCH37-Rpn13 interaction. Nonetheless, further crystal structure analysis is necessary to validate the accuracy of the AlphaFold predictions.
The interaction between PfUCH37 and PfRpn13 requires an intact KEKE domain
With convincing in silico evidence that the association of PfUCH37 with the proteasome is mediated via a direct interaction with the PfRpn13 subunit, we next set out to confirm this interaction in vitro. We generated expression constructs of HA-tagged PfUCH37 and FLAG-tagged PfRpn13, codon-optimized for expression in human cells. Like many P. falciparum proteins, PfUCH37 contains an intrinsically disordered, poly-asparagine-rich region (between amino acid positions 250-299). We anticipated that the presence of the poly-asparagine-rich region might adversely affect recombinant expression of the protein, thus we also generated an HA-tagged construct lacking this region (PfUCH37dN) (Figure 5A). Previous studies have demonstrated the dispensability of poly-asparagine repeats for protein function, and we have shown this to be true for PfUCH37 where deletion of this region does not affect the folding of the catalytic domain, nor does it negatively impact enzyme activity against either Ub-AMC or Nedd8-AMC 17. Moreover, AlphaFold 3 structure and protein-protein interaction prediction showed the fold of the PfUCH37dN mutant is identical to that of the full length enzyme, and that both the C-terminal KEKE motif and the catalytic cysteine of PfUCH37dN are accessible for interaction with Rpn13 and ubiquitin, respectively (data not shown).
These constructs were co-transfected into HEK293 cells and the interaction between PfUCH37 and PfRpn13, initially identified through mass spectrometry, was validated using an IP-western blot approach. As expected, the expression of wild-type UCH37 was very low in HEK293 cells whereas the poly-asparagine deletion (UCH37dN) significantly improved expression (Figure 5B). Nevertheless, IP of both full-length and UCH37dN proteins co-precipitated PfRpn13 as demonstrated by anti-FLAG blot. This interaction was further validated by performing the reciprocal experiment, IP of PfRpn13 followed by anti-HA blot to detect PfUCH37 (Figure 5B).
To test whether this interaction is dependent on the c-terminal KE-rich region of PfUCH37, we generated a truncation construct missing both the poly-asparagine repeat and the last 20 amino acids aligning with amino acids 313-329 of the human enzyme known to be critical for binding Rpn13 (Figure 5A). As suspected, deletion of this C-terminal portion entirely ablated the ability of UCH37 to co-precipitate with PfRpn13 (Figure 5C).
Association with PfRpn13 increases catalytic activity of PfUCH37
PfUCH37 is thought to liberate the proteasome from difficult to process ubiquitinated substrates by trimming Lys48-linked Ub chains from their distal end 11,28. As such, its enzymatic action positively regulates proteasomal degradation by preventing proteasome stalling, and also allows recycling of ubiquitin by cleaving unanchored Lys48-linked polyUb chains. Generally, UCH-family DUBs are known to release only small adducts or unfolded polypeptides from the C-terminus of ubiquitin; they do not hydrolyze protein conjugates such as di-ubiquitin. Several Human UCHs such as Uch-L1, Uch-L3, Uch-L5 and Yeast Yuh1 exhibit this characteristic 3,29–31. Specifically, previous studies have shown that Uch37 alone, or in the Uch37-Adrm1 and Uch37-Adrm1-S1 complexes, also fails to hydrolyze significant amounts of the Lys48-linked di-ubiquitin. We attempted to determine the linkage-cleavage specificity of PfUCH37 by testing its ability to cleave a panel of di-Ub linkage variants, however, as seen for its human counterpart 3, the enzyme appeared unable to hydrolyse di-Ub (data not shown).
By binding the C-terminus of UCH37, Rpn13 is able to pull away its autoinhibitory tail and enable access to the enzyme’s active site. As such, UCH37 is more catalytically active in the presence of Rpn13 or whole proteasomes 1,3. We tested whether this was also the case for PfUCH37 by subjecting it to a Ub-amido-methyl-coumarin (AMC) activity assay. In this assay, enzymes able to hydrolyse ubiquitin will engage with the substrate and cleave the AMC which will fluoresce when released. Fluorescence will accumulate and can be measured in real time, revealing the DUB enzyme’s kinetics. HA-PfUCH37dN and FLAG-PfRpn13 were expressed in 293 cells and purified via capture with anti-HA and anti-FLAG resin, respectively followed by peptide elution. Presence of each protein was validated by immunoblot (Figure 6A) and quantified by BCA. The activity of PfUCH37dN was assessed by adding 10 µL of 250 nM enzyme to 10 µL of 250 nM Ub-AMC substrate in reaction buffer. The reactions were conducted in the presence of 10 µL of Rpn13 at a 1:1 molar ratio or with 10 µL of reaction buffer, resulting in a final reaction volume of 30 µL.Fluorescence measurements were taken every minute for 3 hours until all substrate was depleted, as judged by a plateau in fluorescence emission. Similarly to what has been observed for the human enzyme, PfUCH37 activity was higher in the presence of PfRpn13, with Ub-AMC substrate becoming depleted more quickly (Figure 6B). Although PfRpn13 does not enhance Uch37 cleavage of di-ubiquitin (data not shown), it markedly triggers Uch37-catalyzed hydrolysis of ubiquitin-AMC (Fig 6C). Therefore, in Plasmodium, Rpn13 not only recruits Uch37 to the proteasome but also amplifies its deubiquitination activity.