Characterization of the Late Endosomal ESCRT Machinery in Trypanosoma brucei

Location of ESCRT machinery in trypanosomes

The localizations of TbVps23 and TbVps4 were investigated using BSF cell lines in which the endogenous proteins were epitope tagged by homologous recombination into the native chromosomal alleles. Three cell lines were generated: i) single tagged TbVps23-HA; ii) double tagged TbVps23-HA/Ty-TbRab7; and iii) double tagged TbVps23-HA/TbVps4-Ty [note that the order of epitope-to-orf indicates tagging at N- or C- termini]. We first examined TbVps23 because it had been previously shown to localize within the endosomal system (33). TbVps23-HA typically presented as a prominent structure in the post-nuclear region closely aligned with, but distinct from, the endogenous lysosomal marker p67 (Fig. 1A). Careful examination of raw non-deconvolved images (Fig. S1) also revealed considerable diffuse cytoplasmic staining for TbVps23, which has both membrane-bound and free cytoplasmic pools (33). The dominant signal proximal to the lysosome is reminiscent of the late endosomal marker TbRab7 (25), and double staining for TbVps23-HA and Ty-TbRab7 confirmed prominent colocalization of these proteins (Fig. 1B). TbVps23-HA also strongly colocalized with TbVps4-Ty (Fig. 1C). As with TbVps23-HA, there was also diffuse TbVps4-Ty staining throughout cells (Fig. S1) consistent with recruitment to membranes from a cytoplasmic pool. These results indicate the mammalian ESCRT machinery homologues TbVps23 and TbVps4 are predominantly associated with the TbRab7+ late endosome in BSF trypanosomes. Given the lack of evidence for a morphologically defined MVB in trypanosomes, and conversely the experimental precedent for the TbRab7+ late endosome (14, 25), we will refer to this compartment hereafter as the late endosome.

TbVps4 RNAi silencing

To investigate the function of trypanosomal ESCRT homologues, we silenced TbVps4 using an inducible stem-loop dsRNA construct. We first chose TbVps4 because of its crucial role in the terminal steps of the MVB pathway in mammalian cells. BSF growth was affected by 24 hrs of TbVps4 silencing and cell death occurred by 48 hrs (Fig. 2A). Subsequent analyses were done at 20 hrs because there were no defects in gross cell morphology as observed by light microscopy (data not shown). At that time qRT-PCR revealed that TbVps4 message levels were reduced 46.5±1.1% (Fig. 2B). However, the steady-state localizations of the lysosomal lumenal thiol protease TbCatL and membrane glycoprotein p67 were not altered as determined by fluorescence microscopy (Fig. 2C). These results indicate that while TbVps4 is essential in BSF trypanosomes, short-term TbVps4 silencing does not affect steady state lysosomal morphology.

TbVps4 silencing affects TbVps23 localization

In mammalian cells, Vps4 disassembles the terminal ESCRT machinery from sites where ILVs have invaginated, consequently disruption of Vps4 function results in localized build up of ESCRT machinery (26, 27). To determine if TbVps4 functions in a similar manner, TbVps23 localization was investigated after specific silencing. As seen in control cells (Fig. 1A), following knockdown of TbVps4 prominent TbVps23-HA localization was apparent in close proximity to the lysosome (Fig. 3A). However, there was a dramatic increase in the localized TbVps23-HA signal in the TbVps4 silenced cells, which is best seen in non-deconvolved images (Fig. 3B). Quantification of raw unprocessed images revealed that localized TbVps23-HA fluorescence intensity was increased 2.9±0.1 fold after TbVps4 silencing (Fig. 3C). Control immunoblotting experiments indicated that steady state TbVps23-HA levels were unaltered by TbVps4 silencing (Fig. S2). These results show that disruption of TbVps4 function in BSF trypanosomes leads to significant accumulation of TbVps23 from the cytoplasmic pool to the late endosome, consistent with conservation of function relative to the mammalian ESCRT machinery.

TbVps4 silencing has differential effects on biosynthetic and default lysosomal trafficking

The role of TbVps4 in biosynthetic trafficking to the lysosome was investigated by pulse-chase analysis using soluble TbCatL and membrane-bound p67 as endogenous reporters. As previously shown (25), TbCatL was synthesized in control cells as a 53 kDa precursor that was cleaved over time to a 44 kDa mature form upon lysosomal arrival (Fig. 4A). A similar pattern of processing was seen after TbVps4 silencing (Fig. 4B), although elevated amounts of TbCatL were consistently observed in the final media fraction (∼5%). Quantification (Fig. 4C) revealed that there was a small, but statistically significant reduction in the rate and extent of TbCatL processing corresponding to the increase increase in secreted forms. These results suggest that TbVps4 silencing has a modest but real effect on the biosynthetic sorting of soluble TbCatL in a pre-lysosomal compartment, presumably the late endosome.

Newly synthesized native p67 is first synthesized as the 100 kDa (gp100) ER glycoform (Fig. 5A, left) in control cells, as previously reported (19, 25). After processing to a 150 kDa Golgi intermediate (gp150), p67 is cleaved into quasi-stable fragments (gp42, gp32) upon arrival in the lysosomal. In TbVps4 silenced cells there was no defect in disappearance of the gp100 ER form or in subsequent appearance of the gp150 Golgi form. However, conversion of gp150 to the lysosomal gp42 and gp32 fragments was significantly delayed. These results show that TbVps4 silencing does not affect p67 trafficking in the early secretory pathway, but does significantly delay lysosomal delivery of p67.

The p67 C-terminal cytoplasmic domain contains two dileucine motifs that are necessary and sufficient for lysosomal delivery in procyclic trypanosome. These signals are completely dispensable for lysosomal targeting in BSF cells (19, 20). This situation implies the existence of a cryptic default pathway to the lysosome operative in BSF trypanosomes. Silencing of TbRab7 has no effect on this pathway suggesting that it may not include the late endosome (25). The possible role of the trypanosomal ESCRT machinery in this default route was investigated by pulse-chase analysis with TbVps4 RNAi cells constitutively expressing modified p67 reporters in which either the cytoplasmic domain (CD), or the transmembrane (TM) and cytoplasmic domains, were replaced with a triple HA tag (p67ΔCD-3×HA and p67ΔTM-3×HA respectively). As expected, in control cells this default reporter was processed in a manner that is indistinguishable from native p67, including the orderly kinetic appearance of all glycoforms (Fig. 5B). Furthermore, as was seen previously with TbRab7 silencing, processing of p67ΔCD-3×HA was completely unaffected by TbVps4 knockdown. Likewise, turnover of the soluble p67ΔTM-3×HA default reporter was unaffected by TbVps4 silencing (Fig. S3). These results imply that the trypanosomal ESCRT machinery does not participate in default lysosomal trafficking of membrane proteins, or alternatively that such trafficking does not proceed via the late endosome. However, the results with native p67 and TbCatL are intriguing. Whereas TbRab7 silencing had no effect on normal trafficking of these proteins (25), TbVps4 knockdown significantly impeded delivery of both to the lysosome. These results suggest that the TbESCRT machinery does function in normal biosynthetic trafficking of endogenous lysosomal proteins.

TbVps4 silencing reduces the kinetics of endocytic trafficking to the lysosome

The role of TbVps4 in endocytic trafficking was investigated using tomato lectin (TL) and transferrin (Tf) as markers for receptor-mediated endocytosis. TL binds poly-N-acetyllactosamine (pNAL) structures on flagellar pocket glycoproteins (19, 42) and Tf binds to a unique trypanosomal transferrin receptor (TfR) (43, 44). Both proteins are taken up by clathrin-mediated endocytosis and are delivered to the lysosome via the endosomal system (11, 45). In control and TbVps4 silenced cells, both internalized Tf and TL co-localized prominently with the lysosomal marker p67 (Fig. 6A & B, respectively). To quantitatively investigate the kinetics of uptake and lysosomal delivery, endocytosis of pH-sensitive TL:FITC was monitored by flow cytometry. As seen previously (25), there was a biphasic decrease in fluorescent intensity in control cells as the probe was first rapidly internalized, and then more gradually transited endosomal compartments of decreasing pH, finally reaching the terminal lysosome (Fig. 6C). Silencing of TbVps4 significantly reduced the rate of endocytic trafficking of TL:FITC, most dramatically in the second transitory phase. These results suggest that TbVps4 silencing prevents neither the initial uptake nor the final lysosomal delivery of endocytic cargo, but it does significantly decrease the rate of transport through the endosomal system.

ESCRT machinery and turnover of ISG65

ISG65 is an endogenous type I membrane glycoprotein that has distinct cell surface and internal endosomal pools (37-39). ISG65 can be modified by ubiquitinylation of defined lysine residues in its cytoplasmic domain and, as in mammalian cells, recognition of these epitopes by the ESCRT machinery is believed to mediate endosomal localization and subsequent turnover. In support of this proposal, mutation of the critical lysine residues blocked ubiquitinylation resulting in increased cell surface localization (39-41), and RNAi knockdown of TbVps23 reduced ISG65 turnover in an acidic compartment (but see below), presumed to be the lysosome (33).

Consistent with these reports we observed prominent ISG65 staining in the post-nuclear endosomal region of permeablized control cells (Fig. 7A, top left). However, following TbVps4 silencing, no gross alteration was seen in ISG65 localization (Fig. 7A, top right), and no apparent increase in cell surface ISG65 was observed in non-permeabilized cells (data not shown). These data suggest that TbVps4 silencing does not qualitatively affect the steady-state location of ISG65, but more qualitative analyses would be required for definitive conclusions.

The mode of ISG65 turnover was next assessed. In both control and TbVps4 silenced cells, treatment with the lysosomal cysteine protease inhibitor FMK024 led to prominent colocalization of ISG65 with p67 (Fig. 7A, bottom). The apparent loss of signal from the endosomal region is relative only, resulting from the dramatic accumulation of un-degraded ISG65 in the lysosome. Repeated attempts to measure ISG65 turnover by standard pulse-chase analysis failed due to poor signal after immunoprecipitation with anti-ISG65 (data not shown). Therefore, cycloheximide-chase analysis (39) was performed (Fig. 7B). In control cells, after 4 hours of cycloheximide treatment to inhibit protein synthesis, steady state ISG65 was reduced 54.1±5.2%. This turnover was substantially reversed by concurrent treatment with FMK024. However, after TbVps4 silencing, by 4 hours of cycloheximide treatment the ISG65 signal was reduced by 92.8±1.5%. Again turnover was blocked by FMK024. These results confirm that ISG65 is normally turned over in the lysosome in both control and TbVps4 silenced cells. Unexpectedly however, TbVps4 silencing markedly enhanced ISG65 degradation (∼2-fold), suggesting that trafficking to the lysosome from the endosomal pool is actually increased when ESCRT function is blocked.

Our finding that disruption of TbVps4 function accelerates ISG65 degradation is contrary to the aforementioned report that silencing of the ESCRT I component TbVps23 blocks ISG65 turnover (33). We therefore reinvestigated TbVps23 function by RNAi silencing. First we precisely replicated the dual opposing promoter construct of Leung et. al (33). Multiple clones were tested without success in reducing cell growth (data not shown). It should be noted that Leung et. al (33) achieved only a transient ∼two-fold decrease in growth rate despite an ∼80% reduction in specific TbVps23 message levels. We next tested an equivalent stem-loop TbVps23 dsRNA construct. In this case, cell growth was completely halted at 24 hour of induction (Fig. 8A), although TbVps23 message levels were only reduced by 55.1±2.0% (Fig 8B). These results indicate that TbVps23 is essential for proliferation of BSF trypanosomes. ISG65 turnover was analyzed at the 24-hour time point because there were no obvious defects in cell morphology as judged by light microscopy (data not shown). Despite the strong growth phenotype, turnover was not affected (Fig. 8C). Collectively, these results strongly indicate that silencing the terminal ESCRT machinery in BSF trypanosomes actually enhances the lysosomal turnover of ubiquitinylated membrane proteins. Possible reasons for the discrepancy of these results with the prior work of Leung et. al (33) are discussed below.

Finally, to further investigate the route of ISG65 trafficking, we followed turnover in a previously characterized TbRab7 RNAi cell line (25). Again ISG65 decreased ∼50% after 4 hours of cycloheximide treatment (Fig. 9). Contrary to expectation however, TbRab7 ablation actually accelerated turnover. The steady state level of ISG65 was reduced to ∼40% of control cells after 28 hr of silencing (T0), and loss of ISG65 during the subsequent ‘chase’ was twice as fast (Tet- ∼2-fold vs Tet+ ∼4-fold decrease at T4). This finding contrasts markedly with the complete disruption of endocytic trafficking to the lysosome caused by TbRab7 silencing (25). Possible explanations are discussed below.

Parasite cultures

Bloodstream form (BSF) Lister 427 strain T. brucei brucei were grown in HMI-9, as previously described (20, 32). Tetracycline-responsive single-marker BSF cell lines (50) used for inducible expression experiments were maintained in the same media supplemented with tet-free serum (Atlanta Biologicals, Lawrenceville, GA).

Reporter and RNAi constructs

All TbVps4 and TbVps23 constructs are derived from T. b. brucei genes Tb927.3.3280 and Tb11.01.5840, respectively (http://tritrypdb.org). A TbVps4 dsRNA construct was generated using the pLEW100.v5X:Pex11 stem-loop vector (25). Briefly, a 406 bp region (nts 319-724) of the TbVps4 orf, chosen using RNAit (51), was PCR amplified with nested terminal 5' BamHI/XhoI and 3' XbaI/AscI sites. The product was sequentially inserted upstream of the Pex11 stuffer using XhoI/AscI, and then downstream in the opposite orientation using BamHI/XbaI, generating a tetracycline-responsive stem-loop dsRNA plasmid. In an identical manner, a TbVps23 stem-loop construct was created using a 436 bp region (nts 589-1024) of the TbVps23 orf. A second TbVps23 inducible dsRNA construct was created using the dual T7 promoter vector p2T7Ti (52). This construct is identical to that used previously to knockdown TbVps23 (33) and targets the same coding sequence as the above stem-loop construct. The resultant plasmids were linearized with NotI for transfection.

The p67ΔTM-3×HA and p67ΔCD-3×HA default trafficking reporters used here for constitutive over-expression were previously created in pXS6^puro (25). These are codons 1-616 (p67ΔTM) and 1-639 (p67ΔCD) of the p67 open reading frame (Tb927.5.1810), each with an in-frame C-terminal 3× HA tag (1×: YPYDVPDYA). These constructs were linearized with NotI for transfection.

An in situ TbVps23 C-terminal 3×HA tagging construct was prepared using the pXS6^puro:p67ΔCD-3×HA plasmid. First, the entire TbVps23 ORF was PCR amplified with flanking 5′ HindIII and 3′ XhoI sites and inserted into restricted pXS6:p67ΔCD-3×HA, thereby replacing the p67ΔCD orf with the TbVps23 orf in frame with the C-terminal 3×HA tag. Next, the TbVps23 3′ intergenic region (nts +1 to +448 relative to the TbVps23 stop codon) was PCR amplified with flanking 5' PacI and 3' SacI sites and inserted downstream of the selectable marker. Finally, the entire construct, containing (5′-3′) TbVps23-3×HA, the aldolase intergenic region, the puromycin resistance cassette, and the TbVps23 3′ intergenic region, was excised with HindIII/SacI for transfection. An in situ TbVps4-Ty C-terminal tagging construct was generated using the pXS6^neo plasmid. The 5′ targeting sequence was made by PCR amplifying the 3′ end of the TbVps4 ORF (nts 813-1332) with a single in-frame 3′ Ty epitope tag (EVHTNQDPLD) and flanking 5′ ClaI and 3′ XmaI sites. The 3′ targeting sequence was made by PCR amplification of the downstream intragenic region (nts +1 to +556 relative to the TbVps4 stop codon) with flanking 5' PacI and 3′ SacI sites. These fragments were inserted sequentially into pXS6^neo using the corresponding restriction sites. The entire construct, containing (5′-3′) TbVps4-Ty, the aldolase intergenic region, the neomycin resistance cassette, and the TbVps4 3′ intergenic region, was excised with ClaI/SacI for transfection. The N-terminal Ty-TbRab7 construct is described elsewhere (25).

All newly constructed plasmids were sequence verified. Linearized plasmids and excised tagging constructs were introduced into the appropriate host cells by electroporation as described previously (53, 54), and clonal cell lines were derived by limiting dilution and selection with the appropriate antibiotic. The TbRab7 RNAi cell line is described elsewhere (25). Induction of TbVps4, TbVps23 and TbRab7 dsRNA in clonal RNAi cell lines was achieved with 1 μg ml^-1 of tetracycline.

Antibody, secondary and blotting reagents

Rabbit anti-TbCatL, rabbit anti-BiP, and mouse monoclonal anti-p67 have been described elsewhere (32, 55). Rabbit anti-ISG65 was the kind gift of Professor Mark Carrington (University of Cambridge). It was generated to recombinant protein corresponding to the entire mature extracellular domain (codons 15-355) (56). Mouse anti-Ty was from the UAB Hybridoma Facility (Birmingham, AL) and rabbit anti-HA was from Sigma-Aldrich (St. Louis, MO). Alexa Fluor-conjugated goat secondary antibodies were purchased from Molecular Probes (Eugene, OR). The following ligand conjugates were used for endocytosis assays: transferrin:biotin (Tf:Bio, Molecular Probes); and tomato lectin:biotin and tomato lectin:fluorescein (TL:Bio & TL:FITC, Vector Laboratories, Burlingame, CA). Streptavidin:Alexa Fluor 488 (SA:A488, Molecular Probes) was used for secondary detection of TL:Bio. Horseradish peroxidase (HRP)-conjugated anti-rabbit IgG was from KPL (Gaithersburg, MD).

Radiolabeling and immuoprecipitation

Pulse-chase metabolic radiolabeling with [³⁵S]Met/Cys (Perkin Elmer, Waltham, MA) and subsequent immunoprecipitation of specific radiolabeled proteins from cell lysates were performed as described previously (19). Pulse labeling times were 15 min for p67 and 10 min for TbCatL trafficking; chase times are indicated in the figures. Immunoprecipitations were fractionated by SDS-PAGE and dried gels were analyzed by phosphorimaging using a Typhoon FLA 9000 with native ImageQuant Software (GE Healthcare, Piscataway, NJ). Bands were quantified by volume analysis on areas of interest after subtracting background signal from equivalent unlabeled areas.

ISG65 turnover

Cells were incubated with cycloheximide (100 μg ml^-1) to inhibit de novo protein synthesis. As indicated the lyosomal thiol protease inhibitor FMK024 (morpholinourea-phenylalanine-homophenylalanine-fluoromethyl ketone; 20 μM; MP Biomedicals, Aurora, OH) was included prior to and during cycloheximide treatment. Whole cell lysates were fractionated by SDS-PAGE, transferred using an Owl semi dry apparatus (Thermo Fisher Scientific) to an Immobilon-P transfer membrane (Millipore Corp., Bedford, MA), and blocked in a 5% milk solution. Membranes were incubated 1 hr with anti-ISG65 (1/50,000), washed, and incubated with anti-rabbit IgG:HRP (1/10,000, 1 hr). Blots were stripped and reprobed with rabbit anti-BiP. The final washed blots were visualized on X-ray film using Pierce SuperSignal West Pico substrate (Thermo Fisher Scientific), as described elsewhere (57). Film was digitized to 16-bit grayscale using a transparency scanner and bands quantified using the gel volume analysis function of ImageJ 1.46 (NIH, Bethesda, MD).

Quantitative Real Time PCR

The levels of mRNA knockdown during TbVps4 and TbVps23 RNAi were quantified using primers that amplify nts 191-261 of TbVps4 and 324-402 of TbVps23. Single-strand cDNA was synthesized with an anchored oligo(dT) primer using Superscript III Reverse Transcriptase (Invitrogen, Carlsbad, CA) and 10 mg of total RNA as template (Qiagen RNeasy Mini Kit, Valencia, CA). Real time analysis was performed using the StepOne Real-Time PCR System (Applied Biosystems, Carlsbad, CA) with diluted cDNA and the Power SYBR Green PCR Master Mix (Life Technologies, Carlsbad, CA). Post-amplification melting curves confirmed a single dominant product. Specific mRNA levels were determined relative to the reference gene TbGPI8 in triplicate (means ± sem) and were calculated using the StepOne software v2.2.2.

Epifluorescence microscopy

For most immunofluorescence (IFA) microscopy of epitope tags, cells were methanol/acetone-fixed as previously described (32). For imaging with anti-ISG65 cells were formaldehyde fixed/detergent permeablized as described in (12). Cells were stained with primary and secondary antibodies diluted in blocking buffer and mounted as described in (54). Cells were also stained with DAPI (0.5 μg ml^-1) to reveal nuclei and kinetoplasts. Serial image stacks (0.2 micron Z-increment) were collected with capture times from 100-500 msec (100× PlanApo, oil immersion, 1.4 na) on a motorized Zeiss Axioimager M2 stand equipped with a rear-mounted excitation filter wheel, a triple pass (DAPI/FITC/Texas Red) emission cube, differential interference contrast (DIC) optics, and an Orca ER CCD camera (Hamamatsu, Bridgewater, NJ). Images were collected with Volocity 6.1 Acquisition Module (Improvision Inc., Lexington, MA) and individual channel stacks were deconvolved by a constrained iterative algorithm, pseudocolored, and merged using Volocity 6.1 Restoration Module. Unless otherwise stated all images presented are summed stack projections of merged channels. The xyz pixel precision of this arrangement has been validated in (54) (see Figure S1 therein). Quantification of TbVps23 fluorescence signal from raw, non-deconvolved images was performed using Image J 1.46. The singular post-nuclear punctate region of TbVps23 signal was selected and the mean gray value of that region was calculated.

Endocytosis assays

Standard binding and uptake assays have been previously described (16). Briefly, for TL uptake washed cells (10⁷ ml^-1) were incubated (30 min, 37°C) with TL:Bio (5 μg ml^-1) in serum-free HMI9 with 0.5 mg ml^-1 BSA, washed, and then incubated in fresh media for 20 min to chase ligand into the terminal lysosome. Tf uptake was performed in the same manner with Tf:Bio (5 μg ml^-1), but FMK024 (20 μM) was also included for a 30 min pre-incubation before and during the subsequent uptake and chase periods. In each case cells were processed as described above for IFA and internalized ligands were detected with SA:A488. The kinetics of endocytosis and lysosomal delivery was determined with pH-responsive FITC-conjugated tomato lectin (TL:FITC). Cells were allowed to bind TL:FITC (5 μg ml^-1, 5°C, 30 min), washed into fresh medium, and incubated at 37°C to chase bound ligand into internal compartments. At the indicated chase times labeled cells were analyzed on a LSR II flow cytometer (BD Biosciences, San Jose, CA) using native FACSDiva acquisition software and FlowJo analysis software (Tree Star Inc., Ashland, OR), collecting 10,000 events per data point, as previously described (16). Live cells were analyzed with DAPI (5 μg ml^-1) staining to exclude dead cells during acquisition.

Statisical analyses

Statistical analyses were performed with Prism 4 software (GraphPad Software, Inc., San Diego CA). Biological replicates (n) are indicated where appropriate.