Metabolic selection of a homologous recombination mediated loss of glycosomal fumarate reductase in Trypanosoma brucei

The genome of trypanosomatids is rearranged at the level of repeated sequences, where serve as platforms for amplification or deletion of genomic segments. We report here that the PEPCK gene knockout (Δpepck) leads to the selection of such a deletion event between the FRDg and FRDm2 genes to produce a chimeric FRDg-m2 gene in the Δpepck* cell line. FRDg is expressed in peroxisome-like organelles, named glycosomes, expression of FRDm2 has not been detected to date, and FRDg-m2 is a non-functional cytosolic FRD. Re-expression of FRDg significantly impaired growth of the Δpepck* cells, while inhibition of FRDg-m2 expression had no effect, which indicated that this recombination event has been selected in the Δpepck* cells to eliminate FRDg. FRD activity was not involved in the FRDg-mediated negative effect, while its auto-flavinylation motif is required to impair growth. Considering that (i) FRDs are known to generate reactive oxygen species (ROS) by transferring electrons from their flavin moiety(ies) to oxygen, (ii) intracellular ROS production is essential for the differentiation of procyclic to epimastigote forms of the parasite and (iii) the fumarate reductase activity is not essential for the parasite, we propose that the main role of FRD is to produce part of the ROS necessary to complete the parasitic cycle in the tsetse fly. In this context, the negative effect of FRDg expression in the PEPCK null background is interpreted as an increased production of ROS from oxygen since fumarate, the natural electron acceptor of FRDg, is no longer produced in glycosomes.


Abstract 29
The genome of trypanosomatids is rearranged at the level of repeated sequences, where serve as 30 platforms for amplification or deletion of genomic segments. We report here that the PEPCK gene 31 knockout (Δpepck) leads to the selection of such a deletion event between the FRDg and FRDm2 32 genes to produce a chimeric FRDg-m2 gene in the Δpepck* cell line. FRDg is expressed in 33 peroxisome-like organelles, named glycosomes, expression of FRDm2 has not been detected to date, 34 and FRDg-m2 is a non-functional cytosolic FRD. Re-expression of FRDg significantly impaired 35 growth of the Δpepck* cells, while inhibition of FRDg-m2 expression had no effect, which indicated 36 that this recombination event has been selected in the Δpepck* cells to eliminate FRDg. FRD activity 37 was not involved in the FRDg-mediated negative effect, while its auto-flavinylation motif is required 38 to impair growth. Considering that (i) FRDs are known to generate reactive oxygen species (ROS) by 39 transferring electrons from their flavin moiety(ies) to oxygen, (ii) intracellular ROS production is 40 essential for the differentiation of procyclic to epimastigote forms of the parasite and (iii) the fumarate 41 reductase activity is not essential for the parasite, we propose that the main role of FRD is to produce Tb927.2.4210) and pyruvate phosphate dikinase (PPDK, EC 2.7.9.1, Tb927.11.3120) have a redundant 73 function for the essential gluconeogenesis from proline (17). In glucose-rich conditions, PPDK and 74 PEPCK work in the opposite direction to produce pyruvate and PEP, respectively, in addition to ATP.

75
This pathway is also essential to maintain the glycosomal redox balance (18). Glycosomes are 76 specialized peroxisomes, which harbour the 6 or 7 first glycolytic steps (19). Because of the 77 impermeability of the glycosomal membrane to bulky metabolites, such as cofactors and nucleotides, 78 ATP molecules consumed by the first glycolytic steps (steps 1 and 2 in Fig S1) Fig 1A). Similarly, the GPDH signal was strongly reduced (15.2-fold) in the 110 Δppdk/Δpepck/ RNAi GPDH.i mutant. This analysis also showed that expression of FRDg and FRDm2 111 were 6.5-fold decreased and 10-fold increased, respectively, in the Δppdk/Δpepck/ RNAi GPDH.i cell 112 line, while expression of FRDm1 was not affected (Fig 1A). In contrast, expression of the three FRD 113 isoforms remained unaffected in the three other mutant cell lines (PXD020185 dataset on the 114 ProteomeXchange Consortium). This FRD expression pattern was confirmed by Western blotting 115 using immune sera specific to FRDg (αFRDg) and FRDm2 (αFRDm2), in addition to the αFRD 116 immune serum produced against the conserved FRDg central domain, which is 100% and 71% 117 identical with FRDm2 and FRDm1, respectively (Fig 1B-C). The αFRD antibodies recognized two 118 proteins in both the parental and Δppdk/Δpepck/ RNAi GPDH.i cell lines, including the ~130 kDa FRDm1 119 isoform (Fig 1D-E). As previously reported, the second isoform expressed in the parental cell line 120 (~120 kDa) was recognized by the αFRDg, while no signal corresponding to FRDm2 was detected 121 5 using αFRDm2 (24). In contrast, the ~115 kDa protein expressed in the Δppdk/Δpepck/ RNAi GPDH.i cell 122 line was recognized by αFRDm2, but not αFRDg. This suggests that the mutant cell line switched 123 from FRDg to FRDm2 expression, although the apparent size of the detected FRDm2 isoform was 124 higher than the theoretical one (~115 versus 94.8 kDa). Coomassie staining, Western blotting (Fig 1E)

143
In conclusion, these data showed that a recombination event occurred between the FRDg and FRDm2 144 genes in the Δppdk/Δpepck/ RNAi GPDH mutant to generate a FRDg-m2 chimeric gene coding for a FRD 145 chimeric protein slightly smaller than FRDg (theoretical molecular weights: 120.6 versus 123.5 kDa, 146 respectively). This DNA rearrangement event was present on both alleles of the locus, since the wild-147 type FRDg and FRDm2 amplified DNA fragments (Fig. 2C) and the endogenous FRDg protein (Fig. 148 1D) were not detectable in the mutant cell line.

150
The homologous recombination event occurs in wild type cells. To further study this DNA 151 rearrangement event, we used PCR with primer pairs designed for amplification of the central FRD 152 domain of the FRDg (g5 and g3 primers), FRDm2 (m5 and m3 primers) and FRDg-m2 (g5 and m3 153 primers) genes (see Fig 3A). As expected, the FRDg-and FRDm2-specific DNA fragments were 154 amplified from the parental EATRO1125 cell line but not from the Δppdk/Δpepck/ RNAi GPDH genomic 155 DNA (Fig 3B-C), confirming the loss of the wild-type FRDg/FRDm2 locus in the mutant cell 156 population. Also in agreement with the Southern blot data, the FRDg-m2 specific fragment was 6 also very weakly PCR-amplified from the parental EATRO1125 cell line, which suggests that the 159 recombination event stochastically occurred in the wild-type cells (Fig 3B-C). Moreover, a 1.5kb PCR 160 product was obtained using parental DNA and the g3 and m5 primers; we suggest that the template 161 was a circularized version of the deleted fragment (Fig 3A-C

168
We compared the FRDg-m2 copy number in the two lines by semi-quantitative PCR using different 169 amounts of genomic DNA and the g5-m3 primer pair. Primers specific for a control gene (fructose-170 1,6-bisphosphatase, Tb927.9.8720) were used for normalisation ( Fig 4A). The results showed that the

286
The flavinylation of all endogenous and expressed FRD isoforms and mutants was directly assessed by 287 in gel detection of flavin fluorescence at 526 nm. Using denaturing gels and boiling of protein 288 samples, only covalently linked flavin was detected (Fig 8C-D). For the glycosomally expressed FRD 289 mutant proteins, flavinylation correlates with growth phenotype (Fig 9). This confirmed that covalent 290 flavinylation is required for FRD activity in vivo in trypanosomes (Fig 8A). can be distinguished, provided that one is uniformly 13 C-enriched. To achieve this, we used U-13 C-304 proline (25). The parental trypanosomes incubated in the presence of glucose and proline produced 305 71% acetate, 18% succinate and 11% alanine, as excreted end products ( Table 1). As expected, 306 10 succinate was no longer produced from glucose in the Δpepck cell line (20), while the amounts of 307 proline-derived succinate were not affected (see Fig S1). The metabolic patterns of the Δpepck and 308 Δpepck* cell lines were similar, indicating that succinate excreted from proline in the PEPCK null 309 background was produced not in the glycosomes, but in the mitochondrion by succinyl-CoA 310 synthetase as previously proposed (25). Our interpretation of these data is consistent with the 311 maintenance of the level of succinate production from proline in the Δpepck*/ OE FRDg.i mutant, in 312 which expression of FRDg is 4.5 times higher than in the EATRO1125.T7T cell line (WT) ( Table 1).

313
As expected, no difference was also observed in the Δpepck*/ RNAi FRDg-m2.i (Table 1). These data 314 suggest that the observed growth phenotype is probably the consequence of the absence of succinate 315 production within the glycosomes, which may lead to an increase production of ROS by FRDg.

372
How to explain the growth phenotype due to overexpression of FRDg? Several observations support 373 the view that FRDg, as already observed for FRD from other organisms, can produce reactive oxygen 374 species (ROS), known to be toxic at high concentrations. For instance, FRD is a major contributor to 375 ROS formation in Bacteroides fragilis exposed to oxygen (34). ROS are formed by autoxidation when 376 redox enzymes accidentally transfer electrons to oxygen rather than to their physiological substrates.

377
This enzymatic promiscuity is well illustrated by the E. coli aspartate:fumarate oxidoreductase which 378 was conventionally named aspartate oxidase since oxygen is used as electron acceptor in aerobic 379 conditions (35). However, this enzyme can also transfer electrons to fumarate, which is certainly its 12 natural substrate in the anaerobic conditions encountered in the intestine by E. coli (36). This 381 autoxidation activity of FAD-dependent redox enzymes is due to the solvent accessibility of the flavin 382 moiety, which is situated at the protein surface in order to interact with soluble substrates, as described 383 for the E. coli FRD in the absence of its natural substrate, i.e. fumarate (37). The notion that oxygen 384 and fumarate compete for electrons provided by FAD was also reported for the T. brucei FRD since 385 fumarate inhibited hydrogen peroxide formation with the same affinity as it stimulated NADH-386 dependent FRD activity (K i = 16 versus 20 µM) (38). Thus, abolition of glycosomal succinate 387 production in the Δpepck background, which is probably due to a significant reduction of the 388 glycosomal amounts of fumarate (see Fig S1), might stimulate autoxidation activity of FRDg (Fig 10).

389
This hypothesis is consistent with the absence of growth phenotype for the Δpepck*/ OE FRDg-∆2-9 cell 390 lines, which lost the covalently bound flavin required to transfer electron to the acceptor (Fig 8B).