The trypanosome UDP-glucose pyrophosphorylase is imported by piggybacking into glycosomes where unconventional sugar nucleotide synthesis takes place

Glycosomes are peroxisome-related organelles of trypanosomatid parasites containing metabolic pathways usually present in the cytosol of other eukaryotes, such as glycolysis and biosynthesis of sugar nucleotides. UDP-glucose pyrophosphorylase (UGP), the enzyme responsible for the synthesis of the sugar nucleotide UDP-glucose, is localised in the cytosol and glycosomes of the bloodstream and procyclic trypanosomes, despite the absence of any known peroxisomal targeting signal (PTS1 and PTS2). The questions we addressed here are (i) is the unusual glycosomal biosynthetic pathway of sugar nucleotide functional and (ii) how the PTS-free UGP is imported into glycosomes? We showed that UGP is imported into glycosomes by piggybacking on the glycosomal PTS1-containing phosphoenolpyruvate carboxykinase (PEPCK) and identified the domains involved in the UGP/PEPCK interaction. Proximity ligation assays revealed that this interaction occurs in 3-10% of glycosomes, suggesting that these correspond to organelles competent for protein import. We also showed that UGP is essential for growth of trypanosomes and that both the glycosomal and cytosolic metabolic pathways involving UGP are functional, since the lethality of the knock-down UGP mutant cell line (RNAiUGP) was rescued by expressing a recoded UGP in the organelle (RNAiUGP/EXPrUGP-GPDH). Our conclusion was supported by targeted metabolomic analyses (IC-HRMS) showing that UDP-glucose is no longer detectable in the RNAiUGP mutant, while it is still produced in cells expressing UGP exclusively in the cytosol (PEPCK null mutant) or glycosomes (RNAiUGP/EXPrUGP-GPDH). Trypanosomatids are the only known organisms to have selected functional peroxisomal (glycosomal) sugar nucleotide biosynthetic pathways in addition to the canonical cytosolic ones. Importance Unusual compartmentalization of metabolic pathways within organelles is one of the most enigmatic features of trypanosomatids. These unicellular eukaryotes are the only organisms that sequestered glycolysis inside peroxisomes (glycosomes), although the selective advantage of this compartmentalization is still not clear. Trypanosomatids are also unique for the glycosomal localisation of enzymes of the sugar nucleotide biosynthetic pathways, which are also present in the cytosol. Here we showed that the cytosolic and glycosomal pathways are functional. Like in all other eukaryotes, the cytosolic pathways feed glycosylation reactions, however the role of the duplicated glycosomal pathways is currently unknown. We also showed that one of these enzymes (UGP) is imported into glycosomes by piggybacking on another glycosomal enzyme (PEPCK), which are not functionally related. The UGP/PEPCK association is unique since all piggybacking examples reported to date involve functionally related interacting partners, which broadens the possible combinations of carrier-cargo proteins being imported as hetero-oligomers.


UDP-glucose pyrophosphorylase (UGP) has a dual glycosomal and cytosolic localisation 134
Previous studies on the UGP subcellular localisation revealed that the protein is associated with 135 glycosomes of the BSF (7), despite the absence of any predicted peroxisomal targeting signal 136 (PTS1/PTS2). We raised an anti-UGP immune serum to confirm this unique glycosomal 137 localisation of UGP in PCF by western blotting analyses of glycosomal and cytosolic fractions 138 prepared by differential centrifugation, using as controls antibodies against glycosomal (NADH-139 dependent fumarate reductase, FRDg) and cytosolic (enolase, ENO) proteins. The anti-UGP 140 immune serum detected a 55 kDa protein corresponding to the predicted size of UGP (theoretical 141 MW: 54.5 kDa), in both glycosomal and cytosolic fractions (Fig. 1A). This dual localisation was 142 further confirmed by digitonin titration as UGP was released together with the cytosolic protein 143 at low concentrations of detergent and the UGP signal increased with the digitonin concentration 144 required to release the glycosomal marker (Fig. 1B). The increased signal at higher digitonin 145 concentrations suggests that the total amount of UGP in the glycosomes is at least equivalent to 146 that in the cytosol. We also addressed the UGP subcellular localisation in BSF by performing 147 hypotonic lysis, which released cytosolic proteins while glycosomal proteins remained in the 148 cellular pellet, as evidenced by the glycosomal aldolase and cytosolic enolase markers (Fig. 1C). 149 UGP is similarly distributed over the two compartments in BSF, as observed for PCF (Fig. 1C). showing that UGP was no longer detected in the Δpepck glycosomes, while the protein was still 158 present in the total cell extracts (Fig. 1D). Importantly, re-expression of the PEPCK gene in the 159 PEPCK null background (Δpepck/ EXP PEPCK.i cell line; .i stands for tetracycline-induced) 160 rescued the glycosomal localisation of UGP (Fig. 1D). These data suggest that import of UGP 161 into the glycosomes depends on the presence of PTS1-containing PEPCK, potentially by the so-162 called piggybacking mechanism not reported so far in trypanosomatids (30). In this context, 163 UGP would be co-transported with PEPCK, which is imported into the glycosome via its PTS1. 164 To confirm this dual subcellular localisation of UGP, we produced cell lines expressing a MYC-165 tagged UGP under the control of tetracycline in both the parental and the Δpepck backgrounds 166 (Fig. 1E, right panel). Immunofluorescence analyses showed a clear cytosolic pattern in the 167 tetracycline induced EXP UGP-MYC.i and Δpepck/ EXP UGP-MYC.i cell lines (Fig. 1E, left panel). 168 A signal co-localising with the glycosomal marker aldolase was detected for the EXP

Determination of critical parts for PEPCK-UGP interaction 214
To investigate which part of UGP and PEPCK interacts with its piggybacking partner, truncated 215 versions of each protein were expressed in the parental or Δpepck cell lines, respectively. Since 216 PEPCK form homodimers (33), the truncated PEPCK proteins were expressed in the Δpepck cell 217 line to prevent heterodimer formation. UGP is reported to be monomeric (7, 22) and was only 218 detected as monomer in native gel analyses (Fig. S2), therefore the native and recombinant 219 proteins will not directly interact. We expressed in the parental background the recombinant 220 UGP with the 10xTY tag either at the N-terminal or the C-terminal ends of UGP ( EXP TY-UGP 1- to note that the UGP coding sequence used for the UGP XXX-485 -TY constructs was recoded from 230 position 165 to 337 amino acids to become resistant to the RNAi construct (see below), which 231 was useful to confirm the correct insertion of the recombinant fragment in the UGP locus (Fig.  232   S3). The UGP 124-485 -TY truncated protein was no more imported inside the glycosomes (Fig.  233   3B), while glycosomal import of the UGP 1-124 -TY, UGP 1-173 -TY and UGP 1-226 -TY proteins was  234 not affected (Fig. 3C), suggesting that the N-terminal domain up to the 123 amino acid position 235 contains residues interacting with PEPCK. The truncated recombinant UGP missing (UGP 66-485 -236 TY) or containing only (UGP 1-66 -TY) the N-terminal 66 residues were imported into 237 glycosomes, although with a lower efficiency compared to the parental cell line, suggesting that 238 key residues of the PEPCK-binding site are located on either side of position 66 ( Fig. 3B-C). The 239 presence of the PEPCK binding site in the N-terminal extremity of UGP may explain the low 240 glycosomal import of the TY-UGP 1-485 recombinant protein (Fig. 3A). 241

242
We performed a similar analysis to determine the PEPCK region involved in UGP glycosomal 243 import by expressing truncated versions of recombinant PEPCK using the pLew100 vector. The 244 PEPCK was truncated from its N-terminal extremity in order to maintain the C-terminal PTS1 245 required for glycosomal import of both PEPCK and UGP. Unfortunately, none of the truncated 246 PEPCK peptides were detectable by western blotting in total cell extracts, probably due to 247 protein instability. To resolve this stability issue, the truncated PEPCK peptides were fused to 248 the C-terminal extremity of the green fluorescent protein (eGFP) and used to produce four 249 different cell lines (Fig. 4A). We determined the glycosomal import of UGP in these 250 Δpepck/ EXP eGFP-PEPCK XXX-525 cell lines by western blot analyses of glycosomal and cytosolic 251 fractions. As mentioned above, UGP is no more detected in glycosomes isolated from the 252 parental Δpepck mutants (Fig. 4B). The glycosomal import of UGP was not affected in the 253 absence of the N-terminal first 140 and 180 residues of PEPCK (Δpepck/ EXP eGFP-PEPCK 140-525 254 and Δpepck/ EXP eGFP-PEPCK 180-525 cell lines), while deletion of the N-terminal first 214 and 321 255 residues abolished glycosomal import of UGP, which remained exclusively in the cytosolic 256 fractions (Fig. 4C). This suggested that the 34-residues peptide between amino acids positions 257 180 and 214 of PEPCK is required for UGP import into glycosomes. Importantly, none of the 258 eGFP-PEPCK truncations have PEPCK activity, indicating that the import of UGP is not related 259 to PEPCK activity inside the glycosomes (Fig. 4D) 260

261
The UGP protein is essential for T. brucei 262 The stem-loop RNAi strategy was used with the conditional pLew100 vector to address the role 263 of UGP in the procyclic trypanosomes. Two RNAi UGP cell lines obtained from individual 264 transfections (H10 and E4) showed a strong reduction of growth 7 days after tetracycline 265 induction, indicating that UGP is essential for PCF viability (Fig. 5A, top panel). For both RNAi 266 cell lines, the growth rate of the parental strain was restored 18 days post-induction, 267 concomitantly with the re-expression of the native UGP (Fig. 5A, lower panel). This re-268 expression of RNAi targeted genes is often observed for trypanosome essential genes (29). It is 269 noteworthy that the UGP expression was barely detectable in the non-induced RNAi UGP-H10 total 270 cell extracts. Western blotting analyses of enriched glycosomal fractions, which proved to be 271 more sensitive than on total cell extracts, showed that UGP expression was reduced by ~30-fold 272 compared to the parental cells without any significant effect on growth (Fig. 5B, left panel). This 273 suggests that UGP activity is present in large excess in parental PCF. The distribution of UGP 274 between glycosomal and cytosolic compartments is not affected by this ~30-fold reduction (Fig.  275 5B). After five days of induction, UGP was no more detectable in the glycosomal fractions and 276 was reduced by ~2-fold in the cytosol (Fig. 5B). These small amounts of UGP were not 277 sufficient to sustain growth of PCF. 278 To determine whether UGP is also required for growth of the procyclic trypanosomes in the 279 insect-like glucose-free conditions, the parasites were grown in the absence of glucose, as 280 described before (34). The growth of the RNAi UGP.i and Δpepck/ RNAi UGP.i cell lines is similar 281 regardless of the amounts of glucose in the medium (Fig. S4), indicating the UGP is probably 282 also essential in the insect vector, which is considered to be free of glucose (35). In addition, the 283 subcellular distribution of UGP in the parental cells is not affected by the absence of glucose 284 RNAi construct (rUGP) was fused at its 3'-extremity with a 3xMYC tag followed by different 294 glycosomal targeting peptides (PTS1), namely the last C-terminal 12 residues of glycosomal 295 FRDg (rUGP-FRDgPTS1), the full-length PTS1-containing glycosomal glycerol-3-phosphate 296 dehydrogenase (GPDH) gene (rUGP-GPDH), and the full-length PTS1-containing glycosomal 297 phosphoglycerate (PGKc) gene (rUGP-PGKc*). Since glycosomal expression of PGK is lethal 298 for the PCF trypanosomes (36), the codon of the lysine residue (K215) essential for the PGK 299 enzymatic activity (37) was replaced by the alanine codon. These recombinant proteins were 300 conditionally expressed in the parental cell line and their distribution between the glycosomal 301 and cytosolic compartments was determined by digitonin titration (Fig. 6A). The rUGP-302 FRDgPTS1 and cytosolic enolase proteins showed the same cytosolic profiles, which implies 303 that the extended FRDg PST1 motif is not sufficient for glycosomal import of UGP. In contrast, 304 the rUGP-PGKc* recombinant protein is mostly associated to the glycosomes, but a minor part 305 remained in the cytosol. Finally, the rUGP-GPDH (~100 kDa) and the glycosomal FRDg

The Δugp/ EXP rUGP-GPDH cell line is viable 326
Considering that UGP is an essential protein, knock-out mutants were produced in two cell lines 327 expressing tetracycline-inducible recombinant UGP, i.e. glycosomal/cytosolic rUGP (recoded 328 UGP followed by a MYC tag) and glycosomal rUGP-GPDH. The The specific activity of UGP (ratio between enzymatic activity and relative amount of proteins 357 detected by western blotting) in the cytosolic fractions of EXP rUGP.i is ~3.5-times lower than in 358 the parental WT cells, suggesting that the C-terminal MYC tag affects UGP activity (Fig. 8C). 359 Similarly, the native UGP shows a specific activity in the glycosomal fraction 4-times lower than 360 in the cytosolic fraction of the parental cells, which suggests that the glycosomal sequestration of 361 UGP affects its activity by a yet unknown mechanism. We also confirmed that the coupling 362 enzyme (UDP-Glc dehydrogenase) used in the UGP activity assays was not affected by the 363 presence of the same amounts of the glycosomal or cytosolic samples (273 versus 245 mU.mg -1 364 of protein, respectively). The activity of the recombinant rUGPs, which is ~30 times more 365 expressed in the EXP rUGP.i than the native UGP, was not affected in glycosomes, as the enzyme 366 specific activity is similar in the glycosomal and the cytosolic fractions (Fig. 8C). It is also 367 noteworthy that the UGP activity was detected in the cytosol of the RNAi UGP/ EXP rUGP-GPDH.i, 368 while the native UGP is not detectable by western blot (Fig. 8B) and the recombinant rUGP-369 GPDH is exclusively glycosomal (Fig. 5C). This could be due to the rupture of a few 370 glycosomes during the grinding step designed to primarily disrupt the plasma membrane. 371 To confirm the role of UGP subcellular localisation in UDP-Glc production, we used mass-372 spectrometry based metabolomics to determine the intracellular amounts of G6P, G1P and UDP-373 Glc (Fig. 8D), as well as other metabolites as controls ( previously reported for procyclic trypanosomes (110 to 540 uM) (38) (Fig. 8D). UDP-Glc is no 382 longer detectable in the RNAi UGP.i cell line (Fig. 8D), which shows that UGP is the only enzyme 383 producing UDP-Glc in PCF trypanosomes. It is also of note that UDP-Glc is detected in non-384 induced RNAi UGP cells at levels similar to parental cells, despite the ~30-fold reduction of UGP 385 protein levels (Fig. 6B), which shows that PCF trypanosomes express a large excess of UGP.

Trypanosomes and cell cultures 504
The procyclic form of T. brucei EATRO1125.T7T (TetR-HYG T7RNAPOL-NEO) was cultured 505 at 27°C in SDM79 medium containing 10% (v/v) heat inactivated fetal calf serum, 5 μg.ml -1 506 hemin (64), hygromycin (25 µg.ml -1 ) and neomycin (10 µg.ml -1 ). Alternatively, the cells were 507 cultivated in a glucose-free medium derived from SDM79, called SDM79-GlcFree ( The UGP gene (Tb927.10.13130) was cloned using the In-Fusion cloning system (Clontech) in 519 the HindIII-NdeI restriction sites of pLew100-X-MYC, which was designed for expression of 520 recombinant protein tagged at the C-terminal extremity with 3 MYC epitopes (modified from 521 (66)). The EATRO1125.T7T parental cell line, Δpepck (29), TY-PEPCK and Δpepck/TY-522 PEPCK cell lines were transfected with the pLew100-UGP-MYC tetracycline inducible plasmid 523 and cells were selected in SDM79 containing phleomycin (5 μg.ml -1 ). The UGP gene was also in 524 situ tagged at the N-terminal or C-terminal extremities (67). Briefly, the DNA sequence 525 encoding 10xTY1 tag and BLA resistance cassette was amplified from the pPOTv7-10xTY1 526 vector using long primers (see Table S1) that incorporate a 5' overhang of 80 nt homologous to 527 UGP gene and its UTR.  (Table S2), washed with PBS and incubated for 45 min with 620 secondary antibodies (Table S2) Table S2. 648 Revelation was performed using the Clarity western ECL Substrate as described by the 649 manufacturer (Bio-Rad). Images were acquired and analysed with the ImageQuant LAS 4000 650 luminescent image analyser. 651 652 UGP activity assay 653 The UGP activity in total lysates and aliquots of glycosomal and cytosolic fractions was 654 measured as previously described (22). For normalisation of the UGP activities, the malic 655 enzyme activity was determined on the total cell extracts and the cytosolic fractions, as described 656 before (75). For normalisation of the UGP activities in glycosomal extracts, the glycerol kinase 657 activity was determined as described before (76). The PEPCK activity was measured in total 658 lysates as previously described (Hunt and Köhler, 1995). 659