Proteomics Uncovers Novel Components of an Interactive Protein Network Supporting RNA Export in Trypanosomes

T. cruzi Dm28c epimastigotes were maintained in axenic culture in liver infusion tryptose (LIT) medium at 28 °C (). Procyclic forms of T. brucei Lister 427 and RNA interference T. brucei 29-13 line were maintained in SDM-79 medium supplemented with 10% fetal bovine serum at 28 °C (). For T. brucei 29-13, medium was also supplemented with G418 (15 μg/ml) and hygromycin (50 μg/ml).

Anti-GFP nanobodies were expressed and purified (). Antibodies (Sigma) or anti-GFP nanobodies were coupled to Dynabeads M-270 epoxy (Life technologies) (). T. cruzi cell extracts were prepared by cryomilling with a Planetary Ball PM100 (RETSCH, UK) and affinity purification of GFP-tagged proteins (). Pullout of TcSub2::GFP was achieved using a buffer containing 20 mM HEPES pH 7.4, 10 μM CaCl₂, 1 mM MgCl_2, 10 mM sodium citrate, 0.1% CHAPS (w/v), and protease inhibitors (COMPLETE Mini Protease inhibitor cocktail tablet, Roche); for TceIF4AIII, TcMago, TcNTF2L, and TcMex67 a buffer with 20 mM HEPES pH 7.4, 10 μM CaCl₂, 1 mM MgCl₂, 50 mM sodium citrate, 0.1% Triton X-100 (v/v), and protease inhibitors (COMPLETE Mini Protease inhibitor cocktail tablet, Roche); and for TcFOP and TcAPI5, a buffer with 20 mM HEPES pH 7.4, 10 μM CaCl₂, 1 mM MgCl₂, 50 mM sodium citrate, 0.1% Triton X-100 (v/v), and 10% glycerol (v/v). Aliquots of purified complexes were separated on 4 to 12% NuPAGE Novex Bis-Tris precast gels (Life technologies) and stained with SilverQuest Silver Staining (Life technologies).

For TcFOP, TcAPI5, TcNTF2L, and TcHYP, the complex was analyzed by mass spectrometry. T. cruzi cell extracts were prepared using 5 × 10⁸ parasites lysed in 1 ml of citrate buffer (50 mM sodium citrate, 20 mM Hepes pH 7.4, 1 mM MgCl₂, 1 μM CaCl₂, 0,5% Triton X-100, protease inhibitor (COMPLETE Mini Protease inhibitor cocktail tablet, Roche) by sonication. Pullout of TcAPI5::GFP, TcFOP::GFP, TcNTF2L::GFP, and TcHYP::GFP was achieved by affinity purification of GFP-tagged proteins (). For TcFOP and TcAPI5 pullout, 10% glycerol was added to citrate buffer. The complexes were eluted using 15 μl of elution buffer (2% SDS, 20 mM Tris-HCl pH 8.0) at 72 °C for 20 min. Aliquots of purified complexes were separated on 13% SDS-PAGE and stained with Silver Staining protocol.

Initially, TcSub2 and TceIF4AIII complexes were affinity purified and, for those analyses, three different technical replicates were performed for each TcSub2::GFP or TceIF4AIII::GFP complex. In each experiment, two samples were compared: TcSub2::GFP or TceIF4AIII::GFP pullout with a wild-type pullout (control). Next, samples were processed and identified by LC-MS/MS as previously described. Proteins were compared between TcSub2 and TceIF4AIII complex with the control samples, and only proteins identified in at least two replicates and with emPAI index >0.1 were considered for analyses.

For TcMex67, TcAPI5, TcFOP, TcNTF2L, and TcHYP complexes, different replicate numbers were processed or reprocessed in different experiments to increase the reliability and the coverage. Three biological replicates were used for TcMex67 analyses. For TcAPI5, seven replicates were used: three biological replicates, but technical triplicates were performed for two biological replicates. For FOP, TcNTF2L, and TcHYP, five technical replicates were used for each protein: three biological replicates, but technical replicates were performed for one biological replicates. In each replicate, two samples were compared: target pullout and wild-type pullout (control). The TcMex67, TcAPI5, TcFOP, TcNTF2L, and TcHYP data analysis was performed based on the averages of all replicates and comparing the protein intensities between the target and control samples. The fold change ≥2 was considered to select the proteins.

For TcSub2 and TceIF4AIII samples, mass spectrometry was performed at the FingerPrints Proteomics Facility at the University of Dundee (Scotland, UK). Samples were purified and desalted by electrophoresis into 4 to 12% NuPAGE Novex Bis-Tris precast gel (Life technologies), silver-stained, excised, and dehydrated in a SpeedVac (Thermo Scientific). The dehydrated gel was digested with trypsin (Modified Sequencing Grade, Roche) for 18 h at 30 °C. Tryptic peptides were extracted from the gel with 95% Acetonitrile/5% Formic acid (v/v) and captured with an Acclaim PepMap 100 system (C18, 100 μm × 2 cm) and fractionated on a C18 Easy-Spray PepMap RSLC (75 μm × 50 cm) (Thermo Scientific) columns with an Ultimate 3000 RS LCnano system (Thermo Scientific) coupled to an LTQ OrbiTrap Velos Pro (Thermo Scientific). For identification of proteins, the Mascot Search Daemon (Version 2.4.1) (Matrix Science) (http://www.matrixscience.com/) was used with a custom T. cruzi protein sequence library of 54,610 sequences from five different strains (CL Brener Esmeraldo-like, CL Brener non-Esmeraldo-like, Sylvio, Dm28c, and Marinkellei, downloaded on February 4, 2015 from Uniprot). This strategy increased coverage and hence the likelihood of identification of peptides. MS tolerance allowed up to two missed and/or nonspecific tryptic cleavages, and peptide mass tolerance and fragment mass tolerance were set as ±10 ppm and ±0.6 Da, respectively. Carbamidomethylation (C) was set as a fixed modification, and Acetyl (N-term), Dioxidation (M), Glu->pyro-Glu (N-term E), Oxidation (M) were included as variable modifications. FDR was calculated using a reverse translated proteome. This target-decoy strategy uses along with the target database a decoy database (randomized database, reverse target database used in maxquant). The number of matches to this decoy database equals the number of random matches (false positives) obtained in the target database and is used to calculate local or global FDRs. The relative abundance of proteins was estimated using the exponentially modified protein abundance index (emPAI score), determined in a peptide mixture based on the coverage of peptides from the identified protein () with a 1% FDR and significance p value ≤0.05.

TcMex67, TcAPI5, TcFOP, TcNTF2L, and TcHYP samples were processed at the mass spectrometry facility at the Carlos Chagas Institute/FIOCRUZ-PR (RPT02H PDTIS/Instituto Carlos Chagas—Fiocruz Paraná, Brazil). Samples were purified by electrophoresis into SDS–polyacrylamide gels; lanes were excised and dehydrated with 100% ethanol and dried in a vacuum centrifuge. After reduction with 10 mM DTT, 50 mM ammonium bicarbonate (ABC), samples were alkylated with 50 mM iodacetamide, 50 mM ABC. Gels were washed with 50 mM ABC and dehydrated with 100% ethanol, and this step was repeated. Then, gels were incubated with 12.5 ng/ml trypsin (Promega, V5113), 50 mM ABC at 37 °C for 18 h. After trypsinization, trifluoroacetic acid (TFA) was added to a final concentration of 0.5%. Peptides were extracted from the gel matrix through incubation twice with 30% MeCN, 3% TFA, and twice with 100% MeCN. The extracted peptide solution was concentrated in a vacuum centrifuge to remove MeCN and desalted with homemade C18 spin columns and analyzed by LC-MS/MS in a Thermo Scientific Easy-nLC 1000 system coupled to a LTQ Orbitrap XL ETD. Peptide separation was carried out in a 30 cm (75 μm inner diameter) fused silica in-house packed column with reverse-phase ReproSil-Pur C18-AQ 1.9 μm resin (Dr Maisch GmbH). Chromatography runs were performed with a flow rate of 250 nl/min from 5 to 40% MeCN in 0.1% formic acid, 5% DMSO in a 120 min gradient. The mass spectrometer operated in a data-dependent mode to automatically switch between MS and MS/MS (MS²) acquisition. Survey full-scan MS spectra (at 300–2000 m/z range) were acquired in the Orbitrap analyzer with resolution of 60,000 at m/z 400 (after accumulation to a target value of 1,000,000 in the C-trap). The 12 most intense ions were sequentially isolated and fragmented in the linear ion trap using collision-induced dissociation at a target value of 30,000. The “lock mass” option was enabled in all full scans to improve mass accuracy of precursor ions (). Peaklist picking, protein identification, quantification, and validation were done using the MaxQuant platform (version 1.5.5.1) (), which includes the algorithm Andromeda () for database searching. Default parameters of the software were used for all analysis steps, unless stated otherwise. Proteins were searched against a “decoy database” prepared by reversing the sequence of each entry of the T. cruzi CL Brener protein sequence database (containing 19,242 protein sequences, downloaded on August 10, 2016 from Uniprot) and appending them to the forward sequences. This database was complemented with frequently observed contaminants (porcine trypsin, Achromobacter lyticus lysyl endopeptidase, and human keratins) and their reversed sequences. Search parameters specified an MS tolerance of 20 ppm, an MS/MS tolerance of 0.5 Da, and full trypsin specificity, allowing for up to two missed cleavages. Carbamidomethylation of cysteine was set as a fixed modification, and oxidation of methionine and N-terminal acetylation (protein) were allowed as variable modifications.

For validation of the identifications, a minimum of seven amino acids for peptide length and two peptides per protein was required. In addition, an FDR threshold of 1% was applied at both peptide and protein levels. When transforming peptide identifications into protein identifications, similar protein sequences (e.g., isoforms) present in the database that could not be distinguished by the experimentally detected peptides were grouped and referred to as protein groups. Protein quantification was performed using a label-free approach, where peptides eluting from each LC run are detected as three-dimensional features—retention time versus signal intensity (extracted ion chromatogram, XIC) versus mass/charge—aligned and compared across runs, as previously described ().

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (https://www.ebi.ac.uk/pride/) partner repository with the dataset identifier PXD028072.

For TceIF4AIII phylogenetic analysis, database searches were done using BLASTp () and ortholog proteins identified by reciprocal best hit (RBH). Protein sequences from representative species were obtained from the NCBI database and included proteomes from Saccharomyces cerevisiae (fungi), Homo sapiens (metazoa), Dictyostelium discoideum (amoebozoa), Arabidopsis thaliana (Viridiplantae), Plasmodium falciparum (Chromalveolata), Toxoplasma gondii (Chromalveolata), T. cruzi (Excavata), T. brucei (Excavata), and Leishmania major (Excavata). For calculating the similarity and identity among amino acid sequences (supplemental Table S1, sheet2), the Needle program from the EMBOSS package was used (), which allows a global alignment between two protein sequences. Phylogenetic analysis was performed using maximum likelihood (ML) by FastTree 2.1 using WAG model (). The phylogenetic tree was drawn using FigTree v1.4 program (http://tree.bio.ed.ac.uk/software/figtree/). TcHYP ortholog proteins were identified by RBH criteria and synteny evaluation. A total of 108 positions with more reliable alignment obtained by PSI-coffee () composed the final dataset. The tree was rooted using the Bodo saltans sequence. An evolutionary tree of TcHYP (TcCLB.506435.150) orthologs was inferred by using MrBayes 3.2.6 (). Markov Chain Monte Carlo (MCMC) was run for 1,000,000 generations, sampling trees every 1000 generations. An ML tree (JTT +F + G + I) was also inferred using MegaX () with 1000 bootstrap replicates. For TcAPI5, TcFOP, and TcNTF2L, T. cruzi CL Brener Non-Esmeraldo-like sequences were obtained from TriTrypDB version 33 and used as queries for BLASTp against a database containing 121 eukaryotic predicted proteomes (supplemental Table S2) with an e-value threshold of 0.1. All hits identified were subject to reverse BLASTp using an e-value threshold of 0.1. Orthology was predicted based upon by RBH criteria being the original T. cruzi query sequence. Additional hits were identified using sequences obtained from B. saltans and Dictyostelium purpureum. Predicted orthologous sequences were aligned using MUSCLE (), trimmed manually, and both ML and BA phylogenetic trees were constructed using PhyML () and MrBayes (), respectively. PhyML 3.0 was used with default settings, and a bootstrap of 1000 and MrBayes 3.2.6 was used with an MCMC generation of 800,000, Γ shape rate variation with four discreet categories and a sampling frequency of 1000 with the first quarter as burn-in. MrBayes trees were run using the CIPRES Science Gateway portal (). Proteins were identified by accession number in the TriTryp database (). Protein domains were searched using InterProScan () or HMMscan from HMMER () looking for hidden Markov models (HMM) available in Pfam database 27.0 version ().

Interactome visualization was performed using Cytoscape (available in https://cytoscape.org) from selected data from Tables 1 and 2. A static representation was generated from Cytoscape and processed using Affinity Designer (available in https://affinity.serif.com/en-gb/).

The open reading frames of TcAPI5, TcFOP, TcNTF2L, and TcMago were amplified by PCR using specific oligonucleotide primers as shown in supplemental Table S3. T. cruzi Dm28c genomic DNA was used as template. The PCR products were cloned into the pDONR221 vector from Gateway technology (Invitrogen) and further recombination into the pDEST17 vector (Invitrogen) to produce his-tagged recombinant protein, according to the manufacturer’s protocol. Expression of recombinant protein was induced in Escherichia coli BL21 (DE3) by addition of 1 mM IPTG and incubation for 3 h at 37 °C. His-tagged protein was purified by affinity chromatography on Ni-NTA resin (Qiagen) under denaturing conditions and used to inoculate mice to produce polyclonal antibodies, according to Inoue et al. (2014) (). Animals were immunized by intraperitoneal injection with approximately 50 μg of the his6-tagged protein in Freund's complete adjuvant (Sigma) for the first inoculation. For the three booster injections, we used 20 μg of the recombinant protein in Alu-Gel (Serva) administered at 2-week intervals. Antiserum was obtained 5 days after the last booster injection and by blood collection via cardiac puncture. The protocol was approved by Ethics Commission on Animal Use (CEUA) from FICRUZ, protocol 47/12-3, license LW15/13; project title: "Characterization of trypanosome proteins."

Proteins were separated by gel electrophoresis (SDS-PAGE) and transferred to a nitrocellulose membrane (Hybond C, Amersham Biosciences). The membrane was blocked with 0.1% Tween 20 and 5% milk in phosphate-buffered saline (PBS). Primary antibodies were diluted in blocking solution at the following concentrations: mouse anti-TceIF4AIII (diluted 1:500) (), rabbit anti-Protein A (Sigma-Aldrich, diluted 1:40,000), rabbit anti-TcSub2 (diluted 1:1000) (), mouse anti-TcAPI5 (diluted 1:500), mouse anti-TcFOP (diluted 1:500), mouse anti-TcNTF2L (diluted 1:500), mouse anti-TcMago (diluted 1:500), rabbit anti-GFP (kindly provided by Dr Stenio Fragoso Perdigão, diluted 1:500), mouse anti-GFP (Roche, diluted 1:500), rat anti-HA (Roche, diluted 1:500). Antibodies were incubated with the membrane for 1 h. The membrane was washed three times in 0.1% Tween 20 in PBS. Bound antibodies were detected by fluorescence with IRDye whole IgG secondary antibody (LI-COR). Relative quantification of Western blots was performed using ImageJ software (available in https://imagej.nih.gov/ij/).

The coding sequences of TceIF4AIII, TcMago, TcAPI5, TcHYP, TcFOP, and TcFOPlΔFOP, a mutated version of TcFOP without the FoP domain, cloned previously into pDONR221 (Invitrogen) were recombined pTcGW () for N-terminal eGFP or PTP tagging. Similarly, the coding sequence of TcNTF2L was cloned into pTcGW1.1 () for C-terminal eGFP/PTP tagging. T. cruzi epimastigotes were transfected with these plasmids, and stable lines were selected by adding 500 μg/ml G418 to the culture medium. Cell lines carrying TcSub2 tagged at N-terminal with GFP and TcMex67 tagged at C-terminal with PTP were described in Serpeloni et al. (2011) () and Kugeratski et al. (2015) (), respectively. In. T. brucei, cell lines expressing HA-tagged or GFP-tagged TbMago were generated by in situ tagging with the pMOTag4H or pMOTag4G plasmid, as described previously in Oberholzer et al. (2006) (). The parasites were analyzed by FACSAria II (BD) in the flow cytometry facility at Carlos Chagas Institute–Fiocruz/PR. GFP-positive parasites were cloned using single cell precision mode into 96-well plate containing 100 μl of LIT medium/well and the selection antibiotic.

The endogenous and tagged proteins were localized by indirect immunofluorescence assays, as described by Serpeloni et al. (2011) (). The parasites were incubated with specific antibodies for 1 h at 37 °C and then washed with PBS and incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG or Alexa Fluor 594-conjugated goat anti-rabbit IgG antibodies (Invitrogen, 1:600 dilution), as appropriate, for 1 h. DNA was stained by incubation with 5 μg/ml DAPI for 15 min. Slides were analyzed by fluorescence microscopy (Nikon 80i), and images were captured with a CoolSnap PROcf (Media Cybernetics) camera and analyzed with Image Pro-Plus v. 4.5.1.22 (Media Cybernetics). Images were also obtained by inverted microscopy (Leica DMI6000B) associated with deconvolution software Leica AF6000 (microscope facility RPT07C PDTIS/Carlos Chagas Institute–Fiocruz Paraná).

The coding sequences of orthologs of TcFOP (TbFOP, Gene ID Tb927.6.1470), TcAPI5 (TbAPI5, Gene ID Tb927.7.2240), TcNTF2L (TbNTF2L, Gene ID Tb927.10.2240), and TcHYP (TbHYP, Gene ID Tb927.04.3060) from T. brucei were analyzed by RNAit () to identify specific target nucleotide sequences for RNA interference. The oligonucleotides predicted by RNAit were used to amplify a DNA fragment (supplemental Table S3) for cloning into the p2T7-177 vector (). A total of 10 μg insert-containing vector was linearized with NotI and transfected into procyclic forms of T. brucei 29-13. Transfected parasites were selected by the addition of 5 μg/ml phleomycin to the medium, and RNAi was induced by adding 2 μg/ml tetracycline to log phase parasite cultures.

To analyze the distribution of poly(A)⁺ RNA in T. brucei, a digoxigenin-conjugated oligo(dT) probe was used as previously protocol described in Inoue et al. (2014) (). Probe binding was detected by indirect immunofluorescence analysis, as described above, with mouse monoclonal anti-digoxigenin antibody (Sigma-Aldrich, 1:300 dilution) and Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Invitrogen, 1:600 dilution).

T. brucei total RNA was extracted using TRIzol Reagent (Invitrogen). To remove residual DNA, the DNase treatment was performed with RQ1 (RNA Qualified) RNase-Free DNase (Promega) according to the manufacturer's instruction. RNA quantification was performed with NanoDrop One Spectrophotometer (Thermo Fisher Scientific), and 1 μg of RNA was run on a formaldehyde gel to check the integrity of RNA. To prepare templates for quantitative PCR, cDNA was synthesized from 1 μg of RNA by ImProm-II Reverse Transcription System (Promega), using random primers (Invitrogen. Real-time PCR was performed in triplicate with the SYBERSelect Master Mix (Applied Biosystems) on LightCycle 96 system (Roche).

qPCR analysis for TbAPI5, TbFOP, TbNTF2L, and TbHYP was normalized using the data obtained by amplification of 7SL RNA. supplemental Table S3 lists the primers used for qPCR, PCR conditions were as follows: preincubation at 95 °C for 10 min, followed by 65 cycles of denaturation at 95 °C for 20 s, annealing at 55 °C for 20 s, and extension at 72 °C for 20 s, one cycle of melting at 95 °C for 60 s, 40 °C for 60 s, 65 °C for 1 s, and 97 °C for 1 s. Expression levels were calculated by Pfaffl’s method ().

The TbSub2 RNAi lineage, previously obtained in Serpeloni et al. (2011) (), was used to isolate RNA samples. The induction of RNAi was performed in triplicate by adding 2 μg/ml tetracycline to log phase parasite culture. Total RNA from 1 × 10⁹ cells was isolated using RNeasy Mini Kit (Qiagen) at 24- and 48-h postinduction. RNA was treated with DNAse1 on-column using a RNase-Free DNase Set (Qiagen) following the manufacturer protocol. The integrity was determined by nanoelectrophoresis using an Agilent 2100 Bioanalyzer (Agilent) and quantified by spectrophotometry using a NanoDrop One Spectrophotometer (Thermo Scientific). Approximately 5 μg RNA was sent to BGI services (China) for transcriptome sequencing using Illumina HiSeq 2000 technology (Agilent). The transcriptome data have been submitted to SRA at NCBI with the accession number SUB10190938. We used https://www.bioinformatics.babraham.ac.uk/projects/fastqc and MultiQC for quality control, and analysis was done with Snakemake script. We performed differential expression between each pair of conditions, using edgeR (specifically, exactTest), selecting only coding sequences with at least ten counts in at least one condition (mean across replicates). We used sleuth for differential expression of Salmon data, which simulates technical replicates and improves accuracy. There are 3736 transcripts “differentially expressed” across all three conditions (that is changed in at least one condition) out of 8485 total transcripts analyzed.

To identify components of the T. cruzi mRNA maturation machinery, we used T. cruzi Dm28c ectopically expressing N-terminal GFP fusions of TcSub2 () and TceIF4AIII () as bait. We used cryomilling in liquid nitrogen to preserve dynamic and low-affinity protein–protein interactions (). Complexes were affinity-captured using anti-GFP nanobodies conjugated to magnetic beads (). Cell grindates were directly thawed into buffers to determine optimum conditions for preservation and isolation of complexes. Following analysis by SDS-PAGE (Fig. 1A), where we confirmed the enrichment of the tagged-proteins, TcSub2 and TceIF4AIII isolates were analyzed by mass spectrometry and protein identifications filtered according to exponentially modified protein abundance index (emPAI). Proteins from three independent experiments with emPAI >0.1, and identified in at least two of the three replicates, are listed in the supplemental material (supplemental Table S4). In addition to proteins known to be involved with RNA metabolism, many additional proteins were recovered among the significant identifications (Table 1).

There was considerable coincidence in the proteins interacting with TcSub2 and TceIF4AIII, together with specific interactions (Table 1). The coincident identifications suggest that TcSub2 and TceIF4AIII associate as a supercomplex, further supported by copurification of TcSub2 and TceIF4AIII. The TcSub2/TceIF4AIII cohort includes orthologs of PRP19-like, Dbp1, Alba1-3, and orthologs of DRBD2 and 3, all of which have defined functions in mRNA metabolism (, , , , ). Significantly, components of TREX or TREX-2 were not identified, in accordance with their apparent absence from trypanosomatid genomes (). However, proteins associated with chromatin and splicing were identified, including TSR1, TSR1IP, U2AF26, and SR34 (, , , ), which interact with quality control (QC) components and the NPC.

In addition, proteins involved in ribosome biogenesis, SDA1 () and NOP5 (), as well as RNA granule associators, SCD6 (, ) and TcDHH1 (, ), were also detected, indicating an interplay of transcription/mRNA export with other RNA metabolism pathways. The domain architectures of several hypothetical proteins were predicted and include representatives containing an FOP domain (TcFOP), a dual API5 domain (TcAPI5), and an NTF2-like domain (TcNTF2L), while others were identified as core components of the EJC (Fig. 1B). Phylogenetic reconstructions demonstrated that these proteins are restricted to the trypanosomatids and eubodonids and hence are likely kinetoplastid-specific mRNA export factors (Fig. 2 and supplemental Fig. S1). Among them, only NTF2L has been already described in trypanosomes as a component of mRNA export, both associated to EJC complex () and to nuclear peripherical granules ().

For TcNTF2L, while the N-terminal NTF2-like domain is broadly conserved, the overall architecture is clearly kinetoplastid-specific. Other proteins correspond to the human Ran GTPase-activating protein-binding protein 1 and 2 (Q13283 and Q9UN86) and contain an RNA recognition motif (RRM) in the C-terminal region absent from the kinetoplastid NTF2L proteins (supplemental Table S2, sheet 3). Phylogenetic reconstructions for NTF2L (Fig. 2) recovered two major clades, one containing only kinetoplastid NTF2L proteins and the other all other sequences, consistent with unique pathways for mRNA export ().

To provide orthogonal evidence for the identified interactions between TcAPI5, TcFOP, and TcNTF2L with TcSub2 and TceIF4AIII, we raised antibodies to TcAPI5, TcFOP, and TcNTF2L. Western blots confirmed TcAPI5 and TcFOP as present in the TcSub2 isolates, while TcNTF2L was present in TceIF4AIII pullouts, confirming the assignments (supplemental Fig. S2).

We previously described the T. cruzi ortholog of eIF4AIII (), and phylogenetic tree indicated that TceIF4AIII grouped into the eIF4AIII clade (supplemental Fig. S3). eIF4AIII is a core component of the metazoan EJC and functions in multiple pathways of mRNA metabolism (), while the S. cerevisiae ortholog, Fal1, is involved in 40S-ribosomal-subunit biogenesis (). Additional proteins related to the mammalian EJC core were identified here; TcMago is represented in the TceIF4AIII/TcSub2 cohort, while TcY14 was identified only in TcSub2 isolations. A protein containing a Barentsz/BTZ domain, named as TcBarentsz-like (TcBTZL), was also present among the TcSub2 interactors.

To confirm the interaction between TcMago and TcSub2/TceIF4AIII, we created T. cruzi parasites expressing TcMago::GFP. The fusion protein localized mainly in the nucleus (Fig. 3A), and the T. brucei ortholog tagged with HA or GFP displayed a nuclear localization (supplemental Fig. S4), as described in Tryp Tag database (http://tryptag.org). TcMago localizes with TcSub2 in the nucleus (Fig. 3, A and B), and Western blot analysis of TcMago::GFP immunoprecipitations (Fig. 3D) identified both TcSub2 and TceIF4AIII (Fig. 3E), robustly confirming that conserved core EJC proteins interact with components of the trypanosome mRNA export pathway.

We selected four proteins identified here and which are restricted to kinetoplastids for further characterization. TcFOP, TcAPI5, and TcHYP are also localized to the nucleus, like TcSub2 (Fig. 4, Aand B and supplemental Fig. S7A) while TcNTF2L is cytoplasmic, like TceIF4AIII (Fig. 4C). To investigate the physical interactions of these proteins, we created lines expressing TcFOP, TcAPI5, TcNTF2L, and TcHYP tagged with GFP (supplemental Figs. S5 and S7A). We confirmed that TcSub2 is present in immunoprecipitates from both TcFOP::GFP (Fig. 5B) and TcAPI5::GFP, while TceIF4AIII is indeed present in TcNTF2L::GFP pullouts (Fig. 6B). To ensure that the GFP tag was not affecting TcNTF2L location, we used antibodies to TcNTF2L and observed a similar cytoplasmic localization to TcNTF2L::GFP (supplemental Fig. S5E). Altogether, these data also provide robust support for the presence of kinetoplastid-specific proteins TcFOP, TcAPI5, TcNTF2L, and TcHYP in association with conserved components within trypanosomatid mRNA export pathways.

The FOP domain of TcFOP is divergent from canonical domains (supplemental Fig. S6). In metazoa, FOP contains a conserved LDXXLDAYM UAP56-binding motif (UBM) that is duplicated in the C-terminal region (X = any amino acid) (, ), but in the trypanosomatid FOP, the UBMs are separated by an RGG motif. We asked if these are functional for binding TcSub2, the trypanosome ortholog of the binding partner UAP56. Deletion of the C-terminal UBM from TcFOP (FOPΔC) altered interactions with TcSub2 (Fig. 5B), decreasing TcSub2 recovery in pullouts by ∼50%, indicating that despite divergence, the FOP domain does mediate interaction with TcSub2 (Fig. 5C).

Mex67, a major mRNA export factor, has been partially characterized in T. brucei (, , ). Using T. cruzi cells expressing TcMex67 tagged with PTP () and the same conditions as the TceIF4AIII pullout (Fig. 7A), we confirmed interactions with several proteins including Mtr2 and Ran previously identified in T. brucei (, ) (supplemental Table S5). Although TceIF4AIII itself was not detected, several proteins common to the TceIF4AIII and TcSub2 interactomes were found, including PABP2, RNA helicase Hel67, and DRBD2, together with TcFOP and TcNTF2L (Fig. 7B). This further supports the presence of kinetoplastid-specific proteins as core components of the trypanosome mRNA export machinery.

Analysis of the kinetoplastid-specific protein interactions (supplemental Table S6) confirmed the presence of several conserved components of the processing, export, quality control, and translation systems (Table 2). These include TcSub2, ALBA, DRBD2, NOP, Mago, TceIF4AIII, TceEF1a, and TcPABP2. Most proteins identified with TcFOP, TcAPI5, TcNTF2L, and TcHYP were also identified as TcSub2 and TceIF4AIII interactors. RNA granule proteins were also identified, including TcDhh1, TcSCD6, and TcPABP2, reinforcing connectivity between mRNA export and the cytoplasmic processing machinery. Further, we identified PUMILIO proteins, RNA-binding proteins involved in controlling gene expression (, , ). The intersection of the interacting proteins recovered with the six isolated proteins was processed using a Venn Diagram tool (https://bioinformatics.psb.ugent.be/cgi-bin/liste/Venn/calculate_venn.htpl), and those proteins recovered with three or more protein complexes plotted (Fig. 8). Overall, these data indicate an unexpected high level of connectivity between multiple aspects of mRNA processing and transport.

We analyze the roles of TcFOP, TcAPI5, and TcNTF2L by silencing their T. brucei orthologs. Quantitative PCR confirmed the efficiency of RNAi against all targets (Fig. 9A). Knockdown of TbAPI5 and TbNTF2L affected proliferation, while knockdown of TbFOP had little proliferative impact (Fig. 9B). None of the proteins appear involved in global export of mRNAs since accumulation of mRNA in the nucleus by polyA FISH was not seen (Fig. 9C). Two different controls (without probe and RNAse treatment) confirmed the specificity of probe signal (supplemental Fig. S8). Additionally, we also analyzed TbHYP knockdown and obtained a similar phenotype to TbAPI5, TbFOP, and TbNTF2L with no obvious inhibition of global mRNA export (supplemental Fig. S7G). Hence, these new proteins may be mediating the processing of a subset of RNA molecules or that their functions are redundant.

TbSub2 is the sole protein analyzed whose absence clearly blocked mRNA export (). The impact of TbSub2 knockdown on the transcriptome was restricted and given a supposed central role in mRNA maturation, surprising. Only seven genes demonstrated highly significant increases in abundance (Fig. 10), of which three correspond to 28S and 16S rRNAs (Fig. 11) and the remaining four are protein-encoding. The top protein hits Tb927.7.6960, Tb927.5.680, and Tb927.4.910, with the fourth (Tb927.9.7290) corresponding to a VSG pseudogene. Tb927.7.6960 is an E2 ubiquitin ligase with a Ubc2 domain, while Tb927.5.680 and Tb927.4.910 are hypothetical proteins, both kinetoplast-restricted; based on localization at TrypTag Tb927.5.680 is likely endolysosomal. Similarly, few genes were found to have decreased associated transcript levels, and there was no obvious functional or other feature uniting the cohort (supplemental Table S7). The possibility that TbSub2 is involved in rRNA processing is intriguing as it is associated with mRNA export and quality control in other organisms, but the presence of several nucleolar proteins in our interactome would be consistent with an expanded role or close associations between proteins of the mRNA and rRNA export pathways.