PPDK and PEPCK are essential for gluconeogenesis from proline

PCF trypanosomes incubated in glucose-free and proline-rich conditions are able to maintain significant levels of (glycolytic) sugar-phosphates, thought at lower extent than cells incubated with glucose and proline (S1 Table). This means that proline is able to support the synthesis of the sugar-phosphates by a gluconeogenesis process. According to the current view of the central carbon metabolism of the PCF trypanosomes, PPDK and PEPCK are the only enzymatic steps that can produce PEP, the precursor of gluconeogenesis, in proline catabolism (steps 16 and 17 in Fig 1A). These two enzymes have been shown to contribute to glucose catabolism, however, their roles in glucose-depleted conditions have not been investigated so far [12, 22]. To investigate this poorly explored gluconeogenic metabolic pathway from proline, we have determined by mass spectrometry the incorporation levels of ¹³C atoms from uniformly [¹³C]-enriched proline ([U-¹³C]-proline) into key gluconeogenic intermediates of parental and knock-out (Δppdk, Δpepck and Δppdk/Δpepck) PCF cell lines. Cells were incubated in PBS containing 2 mM [U-¹³C]-proline in the presence or absence of the same amounts of non-labelled glucose. Incorporation of ¹³C into glycolytic intermediates was quantified by IC-MS/MS and the values for selected glycolytic metabolites are shown in Fig 2. Incorporation of ¹³C atoms into hexose phosphate glycolytic intermediates from [U-¹³C]-proline is very low in the presence of glucose (3.1% on average), compared to cells incubated in the absence of glucose (94.1% on average). The same labelling pattern was also observed for the intermediates of the pentose phosphate pathways, i.e. 6-phosphogluconolactone and sedoheptulose 7-phosphate. This confirms that proline feeds gluconeogenesis and the pentose phosphate pathway in the absence of glucose, while the presence of glucose down-regulates conversion of proline into metabolite intermediates of these pathways, as we previously observed [21]. Incorporation of at least two ¹³C atoms into the same hexose phosphates is more reduced in the Δppdk mutant than in the Δpepck mutant (13.8% versus 49.3% on average, respectively), suggesting that PPDK might contribute more than PEPCK to gluconeogenesis from proline (Fig 2). More importantly, this ¹³C-incorporation is abolished in the Δppdk/Δpepck double mutant (not detectable in Fig 2), although proline consumption is increased by 2-fold [12, 21]. This demonstrates that PPDK and PEPCK are the only enzymes significantly contributing to PEP production from proline catabolism and have a redundant function for the essential production of glycolytic intermediates from proline. These data also confirm that, under the conditions pertaining in the cell, pyruvate kinase (EC 2.7.1.40, step 18 in Fig 1) irreversibly converts PEP into pyruvate, even in the absence of PEPCK and PPDK.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. IC-MS/MS analysis of intracellular metabolites after isotopic labelling with [U-¹³C]-proline.

The EATRO1125.T7T parental (WT), Δpepck, Δppdk and Δppdk/Δpepck cell lines were incubated for 2 h in PBS containing 2 mM [U-¹³C]-proline with 2 mM glucose (+G, only the parental cells) or without glucose (-G) prior to metabolite extraction. The figure shows enrichment of key glycolytic intermediates at 0 to 6 carbon positions (m0 to m6) with ¹³C expressed as percentage of all corresponding molecules (MID; Mass Isotopomer Distribution). For abbreviations, see Fig 1: S7P, sedoheptulose 7-phosphate; 2/3PG means that 2-phosphoglycerate (2PG) and 3-phosphoglycerate (3PG) are undistinguished by IC-MS/MS. The asterisks (*) in the F6P column for the Δpepck mutant highlights that the absolute amount of this metabolite being particularly low in this incubation conditions, solely the m0 and m3 isotopomer were detected.

https://doi.org/10.1371/journal.ppat.1007502.g002

Glycerol is an alternative gluconeogenic carbon source

To explore whether blocking gluconeogenesis from proline affects parasite growth, we developed an SDM79-derived glucose-free medium (SDM79-GlcFree) containing less than 1 μM glucose, as determined by NMR spectrometry and supplemented with 50 mM N-acetyl D-glucosamine, a competitive inhibitor of glucose transport [21]. The growth of the parental EATRO1125.T7T PCF cell line is moderately affected by the absence of glucose with a doubling time increased by ~20% in the SDM79-GlcFree compared to the same medium supplemented with 10 mM glucose (Fig 3A). Similarly, growth of the Δppdk/Δpepck cell line is not compromised in the absence of glucose, suggesting that another gluconeogenic carbon source can be used by the PCF in addition to proline (Fig 3A). The abolition of PEP production from gluconeogenic amino acids (including proline), as demonstrated above, strongly supports the view that glycerol might be a possible alternative to proline that would feed gluconeogenesis in the Δppdk/Δpepck cell line. To address this hypothesis, we have measured by IC-MS/MS the [¹³C]-incorporation into glycolytic intermediates of the parental PCF incubated with [U-¹³C]-glycerol. In this experiment, most hexose phosphate glycolytic intermediates (83.2% on average) are fully [¹³C]-enriched after 2 h of incubation with [U-¹³C]-glycerol as the only carbon source (S1 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Functional analysis of Δppdk/Δpepck cell lines.

This figure represents growth curves of the Δppdk/Δpepck cell line (panel A) and tetracycline-induced Δppdk/Δpepck/^RNAiGK.i, Δppdk/Δpepck/^RNAiGPDH.i and Δppdk/Δpepck/^RNAiTIM.i triple mutants (panels B-D) grown in SDM79-GlcFree medium containing 10 mM glucose (+G) or not (-G). In glucose-free conditions, the SDM79-GlcFree medium was supplemented with 50 mM N-acetyl-D-glucosamine that inhibits uptake of residual glucose. Cells were maintained in the exponential growth phase (between 10⁶ and 10⁷ cells ml^-1). The insets show western blot analyses of the parental (WT), Δppdk/Δpepck (ΔΔ) and tetracycline-induced (.i) and non-induced (.ni) triple mutants with the immune sera indicated in the right margin (ASCT, acetate:succinate CoA-transferase; ENO, enolase). The cross in panel B means that the mutant ultimately dies.

https://doi.org/10.1371/journal.ppat.1007502.g003

Blocking gluconeogenesis from both proline and glycerol affects growth of the PCF trypanosomes in glucose-free conditions

In order to abolish all possible gluconeogenic pathways, the expression of the key enzymes of glycerol conversion pathway was successively down-regulated in the Δppdk/Δpepck null background, i.e. glycerol kinase (GK, EC 2.7.1.30, Tb927.9.12550-Tb927.9.12630, step 32), glycerol-3-phosphate dehydrogenase (GPDH, EC 1.1.1.8, Tb927.8.3530, step 33) and triose phosphate isomerase (TIM, EC 5.3.1.1, Tb927.11.5520, step 23). We hypothesised that the abolition of gluconeogenesis from both proline and glycerol, in the Δppdk/Δpepck/^RNAiGK, Δppdk/Δpepck/^RNAiGPDH and Δppdk/Δpepck/^RNAiTIM triple mutants, should lead to the death of parasites grown in glucose-free conditions. Indeed, the growth of the tetracycline-induced Δppdk/Δpepck/^RNAiGK cell line (Δppdk/Δpepck/^RNAiGK.i) in SDM79-GlcFree medium stopped after 10 days of tetracycline induction and cells ultimately died two weeks later, while the addition of glucose restored their growth (Fig 3B). Similarly, the growth of the Δppdk/Δpepck/^RNAiTIM.i mutant was strongly affected by the absence of glucose, but this was not lethal to the cells (Fig 3D). However, the growth of the Δppdk/Δpepck/^RNAiGPDH.i cells appears to be only moderately affected by the absence of glucose, although GPDH was not detected by western blotting anymore (Fig 3C), probably because the glycerol 3-phosphate/dihydroxyacetone phosphate shuttle could bypass the GPDH step (see Fig 2 in [5] for this pathway).

To determine whether the glycerol metabolism is abolished in the Δppdk/Δpepck/^RNAiGK.i cell line that shows the strongest growth alteration in SDM79-GlcFree medium, quantitative analyses of ¹³C-enriched end products excreted from [U-¹³C]-glycerol metabolism were performed by proton NMR spectrometry [6]. It is noteworthy that (i) the use of ¹³C-enriched glycerol as carbon source allows us to distinguish between the products excreted from metabolism of [U-¹³C]-glycerol and an uncharacterised internal carbon source [6, 23], and (ii) the succinate excretion is abolished in the Δppdk/Δpepck/^RNAiGK.i mutant (Fig 4), as previously observed for the Δppdk/Δpepck and Δpepck cell lines [12, 22], since the succinate branch is interrupted in the absence of PEPCK (see Fig 1B). The Δppdk/Δpepck/^RNAiGK.i mutant shows a 15-fold decrease of the total end products excreted from glycerol breakdown (acetate, alanine, pyruvate and lactate) compared to the parental Δppdk/Δpepck mutant, while glucose metabolism is not affected (Fig 4). This confirms that the growth arrest observed for the Δppdk/Δpepck/^RNAiGK.i is due to the abolition of glycerol metabolism that is required to feed gluconeogenesis in the PPDK/PEPCK null background in glucose-free conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Proton (¹H) NMR analysis of end products excreted from glucose and [U-¹³C]-glycerol metabolism.

The Δppdk/Δpepck and tetracycline-induced (.i) Δppdk/Δpepck/^RNAiGK cell lines were incubated for 6 h in PBS/NaHCO₃ buffer containing 4 mM glucose or 4 mM [U-¹³C]-glycerol. Panel A shows the quantitative data of excreted end products from glucose or [U-¹³C]-glycerol metabolism expressed as nmol of acetate, pyruvate, alanine and lactate excreted per h and per mg of protein. Panel B shows one representative NMR spectrum (out of 3 spectra and ranging from 1.2 to 4 ppm) of end products excreted from [U-¹³C]-glycerol metabolism of the Δppdk/Δpepck and Δppdk/Δpepck/^RNAiGK.i cell lines. The resonance massif corresponding to unconsumed [U-¹³C]-glycerol as indicated (from 3.3 to 3.8 ppm) and resonances indicated by asterisks correspond to [¹³C]-enriched pyruvate and acetate. The resonances labelled 1 and 2 correspond to non-[¹³C]-enriched succinate and acetate, respectively, produced from the catabolism of an unknown internal carbon source [6, 23].

https://doi.org/10.1371/journal.ppat.1007502.g004

FBPase is not essential for PCF trypanosomes in glucose-free in vitro conditions

Gluconeogenesis from glycerol and proline leads to the production of fructose 1,6-bisphosphate (F1,6BP), that is converted into fructose 6-phosphate (F6P) by the fructose-1,6-bisphosphatase (FBPase, EC 3.1.3.11, step 25 in Fig 1A), a well-known gluconeogenic enzyme that is, however, not involved in glycolysis. To confirm the gluconeogenic role of FBPase in PCF, both alleles of the single copy FBPase gene (Tb927.9.8720) were replaced by the blasticidin and puromycin markers (Δfbpase) and FBPase expression was conditionally down-regulated by RNAi (^RNAiFBPase). Both the Δfbpase and ^RNAiFBPase.i cell lines are viable in the absence of glucose, although with a similarly increased doubling time compared to the parental cells (18.4 h and 18.5 h, respectively, compared to 13.7 h) (Fig 5B and 5C). These data suggest that, although the absence of FBPase affects parasite growth in glucose-free conditions, an alternative enzyme or pathway could substitute for the FBPase reaction, in such a way that proline and/or glycerol would contribute to G6P production even in the absence of FBPase. Re-expression of a rescue ectopic copy of the FBPase gene in the Δfbpase background using the tetracycline-inducible pLew100 vector (Δfbpase/FBPase.i cell line) restored the WT growth (Fig 5D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Analysis of FBPase mutants cell lines.

Panel A shows a PCR analysis of genomic DNA isolated from the parental (WT) and Δfbpase cell lines. The 1 to 10 lanes of the gel picture correspond to different PCR products described in the lower panel with circled numbers. As expected, PCR amplification of the FBPase gene (lane 10) was only observed in the parental cell line, while BSD and PAC PCR-products were observed only in the Δfbpase mutant (lanes 1–2 and 7–9). As control, both the 5'- and 3'-untranslated regions (UTRs) are PCR amplified from the parental and Δfbpase samples (lanes 3–6). Panels B-D show the growth curves of the Δfbpase (B), ^RNAiFPBase.i (C) and Δfbpase/FPBase.i (D) mutant cell lines, together with the parental (WT) cell line, incubated in SDM79-GlcFree medium containing 10 mM glucose (+G) or not (-G). Western blot analyses with the immune sera indicated in the right margin of the parental (WT), Δfbpase, the tetracycline-induced (.i) and non-induced (.ni) ^RNAiFPBase and/or Δfbpase/FPBase.i mutant are shown in the insets.

https://doi.org/10.1371/journal.ppat.1007502.g005

To confirm the role of FBPase in gluconeogenesis, we determined by IC-MS/MS the incorporation levels of [U-¹³C]-proline into glycolytic intermediates of the WT, Δfbpase and Δfbpase/FBPase.i rescue cell lines (Fig 6). The incorporations of [¹³C]-enriched atoms from proline into triose phosphates (PEP, 2/3PG, 1,3BPG and Gly3P, see Fig 1 for abbreviations) and F1,6BP are equivalent in the parental, Δfbpase and Δfbpase/FBPase.i cell lines. However, the incorporations of [¹³C]-enriched atoms in F6P (the product of the reaction catalysed by FBPase), as well as in metabolites produced from F6P, i.e. G6P, mannose 6-phosphate (M6P) and 6-phosphogluconolactone (6PG) and sedoheptulose 7-phosphate (S7P), are strongly reduced in the Δfbpase mutant compared to the parental cell line. Indeed, the relative amounts of non-enriched F6P, G6P, M6P and 6PG and S7P in the Δfbpase cells are increased by 6.5-, 4.6-, 6.9, 7.2- and 5.8-fold, respectively. As expected, incorporation of ¹³C-carbons into hexose phosphates is restored in the Δfbpase/FBPase.i rescue cell line (Fig 6). These data clearly demonstrate that gluconeogenesis from proline, although strongly reduced, still occurs in the absence of FBPase, which implies the existence of an alternative step to FBPase producing G6P from triose phosphates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. IC-MS/MS analysis of intracellular metabolites of the Δfbpase mutant incubated with [U-¹³C]-proline.

The parental (EATRO1125.T7T), Δfbpase and Δfbpase/FBPase.i cell lines were incubated for 2 h in PBS containing 2 mM [U-¹³C]-proline. Enrichment of key glycolytic intermediates at carbon positions 0 to 6 (m0 to m6) is expressed as percentage of all the corresponding molecules. For abbreviations, see Figs 1 and 2.

https://doi.org/10.1371/journal.ppat.1007502.g006

To order to highlight the alternative to FBPase, the FBPase activity was determined on the parental, Δfbpase and Δfbpase/FBPase.i (rescue) cell lines. This enzymatic activity was not detectable on total cellular extracts, but was detectable in the glycosomal fractions of the parental cell line at a low level compared to the glycerol kinase activity (14.8 ±4.5 versus 11,100 ±2,300 nmol min^-1 mg^-1 of proteins) (Fig 7A). Interestingly, this activity is not abolished in the Δfbpase mutant, although decreased by 3.8-fold. Reintroduction of the FBPase gene in the Δfbpase cell line only partially restored the FBPase activity (7.1 ±2.3 versus 14.8 ±4.5 nmol min^-1 mg^-1 of proteins), which is consistent with the lower level of FBPase expression in the rescue cell line compared to that of the parental cells (Fig 7A). Altogether, these data strongly support the view that an unknown glycosomal enzyme is responsible for approximately one fourth of the glycosomal FBPase activity. Incidentally, the T. brucei genome contains a single gene potentially coding for a sedoheptulose-1,7-bisphosphatase (SBPase, Tb927.2.5800), which belongs to the same superfamily as the FBPases. Western blot analysis of the glycocomal and cytosolic fractions of the parental trypanosomes with the anti-FBPase and anti-SBPase immune sera demonstrated the glycosomal localisation of these two proteins (Fig 7B), which is in agreement with previously published proteomic analyses of glycosomal fractions [24, 25].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. The glycosomal FBPase activity is not abolished in the Δfbpase mutant.

Panel A shows the FBPase activity (top panel) and western blotting analysis (lower panel) of glycosomal fractions of the parental (WT), Δfbpase and Δfbpase/FBPase.i (rescue) cell lines. The FBPase activities were normalised with the glycerol kinase activities. In panel B the glycosomal localisation of FBPase and SBPase was confirmed by western blotting analyses on glycosomal and cytosolic fractions, using control immune sera against glycosomal (PPDK) and cytosolic (enolase) markers.

https://doi.org/10.1371/journal.ppat.1007502.g007

FBPase is essential for the establishment of mature infections in the insect fly salivary glands

We then attempted to determine whether the absence of FBPase would affect parasite virulence in vivo in their tsetse fly vectors. To perform such experimental infections, a cell line bearing a single FBPase allele (Δfbpase-/+), a FBPase null mutant (Δfbpase), and two distinct constitutive add-back rescue cell lines (Δfbpase/FBPase-1 and 2) were generated in the AnTat1.1E pleomorphic strain [26]. Indeed, this strain has maintained the ability to efficiently colonise salivary glands and to produce infectious metacyclic forms, as opposed to parasites with the EATRO1125 genetic background [27]. Two strategies were developed to produce the Δfbpase/FBPase rescue cell lines, i.e. the addition of a FBPase ectopic copy under the control of the PARP promoter in the pLew100 vector (Δfbpase/FBPase-1), and the in-situ re-insertion of one FBPase gene under the control of its endogenous UTRs (Δfbpase/FBPase-2). Batches of about 50 teneral male tsetse flies were artificially fed through a silicone membrane with cultured PCF parasites of either one of the four cell lines in parallel (the parental wild-type, the Δfbpase-/+, the Δfbpase and the Δfbpase/FBPase-1 or -2 cell lines). Flies were dissected after one month in order to quantify the infection rates per organ, the parasite densities per organ and the proportion of parasite morphotypes as previously described [28]. A total of 784 flies were dissected with similar results obtained with the two series: 338 flies out of 4 biological replicates with the Δfbpase/FBPase-1 panel (Fig 8A) and 446 flies out of 6 replicates with the Δfbpase/FBPase-2 panel (Fig 8B). First, the midgut infection rates were comparable in all conditions (33% on average, ranging from 19% ±27% in Fig 8A to 45% ±19% in Fig 8B for the null mutants). Second, although the salivary gland infection rates were on average 16% for the wild-type cells, 7% for the Δfbpase-/+ parasites and 3% for the Δfbpase/FBPase rescue cell line, no null mutant parasites were seen in any salivary glands and this observation was statistically significant (p = 0.011 by ANOVA Tukey add-hoc post-test at 95% confidence when comparing %SG+ in Δfbpase versus wild-type). This demonstrates that FBPase is essential for the second part of T. brucei cyclical development, i.e. the establishment of mature infections in the insect salivary glands. The lower salivary gland infection rates observed with the Δfbpase/FBPase cell lines could be explained by the additional round of transformation and the longer time spent in culture during the generation of this cell line that would both have affected its virulence. It is also noteworthy that the in-situ rescue approach is apparently more adapted than the use of the pLew100, probably because the PARP promoter used for driving gene expression in the pLew100 vector may be naturally less efficient in the short epimastigotes, attached epimastigotes and/or in the metacyclic stages during the late steps of the cyclical development as compared to the procyclic and mesocyclic midgut stages.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. The deletion of the FBPase gene prevents parasites to colonise the tsetse salivary glands.

In this experiment, batches of 50 Glossina morsitans morsitans teneral males were artificially fed with either the parental AnTat1.1E wild-type, the Δfbpase-/+ (one allele deleted), the Δfbpase (null mutant) or the Δfbpase/FBPase-1 and -2 (rescue cell lines) PCF cell lines in culture medium as previously described [28]. Four weeks after the infective meal, a total of 790 flies were dissected to assess the presence of parasites in their midgut and salivary glands by microscopic examination. Midgut and salivary gland infection rates (in % ±SD) are presented for each cell line as the mean of 4 independent biological replicates done with the rescue cell line expressing the FBPase through the pLew100 vector (panel A), and of 6 independent biological replicates done with the rescue cell line expressing the FBPase from an ectopic FBPase copy in the FBPase locus (panel B). The proportion of flies developing a salivary gland infection are: 12/96 for the parental AnTat1.1E, 0/60 for Δfbpase, 3/104 for Δfbpase-/+ and 1/84 for Δfbpase/FBPase-1 in panel A and 33/163 for the parental AnTat1.1E, 0/91 for Δfbpase, 8/65 for Δfbpase-/+ and 6/125 for Δfbpase/FBPase-2 in panel B. Below the graphs are shown western blot analyses of the FBPase expression using the glucose-6-phosphate dehydrogenase (G6PDH) as loading control. (C) The proportions of cells in the different morphotypes found in the proventriculus and midgut were compared after dissection of flies infected with either one the four different cell lines from at least two independent biological replicates. The proportion of trypomastigotes and epimastigotes is indicated in grey and black respectively. 692, 373, 432, and 280, cells isolated from 6, 2, 3 and 2 flies have been analysed for the parental AnTat1.1E, Δfbpase, Δfbpase-/+ and Δfbpase/FBPase-2 cell lines, respectively. Panel D shows growth of the AnTat1.1E and Δfbpase cell lines in the SDM79-GlcFree medium containing from less than 20 μM to 100 mM proline, using the Alamar Blue assay. The proline concentration values on the x axis correspond to the amounts of proline added to the proline-depleted SDM79-GlcFree medium, which contains less than 20 μM proline.

https://doi.org/10.1371/journal.ppat.1007502.g008

Assuming that (i) the parasite densities in the midgut, estimated by microscopic observations, were similar for all the cell lines and that (ii) no obvious motility defect was detected among these cell lines neither in vitro nor in vivo, we reasoned that the absence of FBPase may have impaired parasite differentiation, especially the first morphotype switch occurring in proventricular parasites. Therefore, parasites were isolated from the anterior midgut and proventriculus of infected flies, and stained with a DNA marker (DAPI) and an axonemal marker Mab25 [29] in order to determine the proportions of cells in each morphotype (trypomastigote versus epimastigote) according to the cell lines (Fig 8C). Nevertheless, no difference was detected between cell lines that were all displaying on average 80% of proventricular parasites in the trypomastigote morphotype and 20% of epimastigotes (280 to 692 cells out of 2 to 6 flies per cell line). This demonstrates again that FBPase is not essential for the trypanosome cyclical development in the midgut, while it is required at least for the initial colonisation of salivary glands.

We reasoned that these differences in behaviour could be the consequence of possible low proline concentrations in the oesophagus and/or the salivary glands of the tsetse, which would not be sufficient for feed gluconeogenesis in the absence of FBPase. To test this hypothesis, we estimated the growth of the parental and Δfbpase cell lines as a function of proline concentrations in the SDM79-GlcFree medium, using the Alamar Blue assay. As expected these two cell lines failed to grow under less than 20 μM proline, this amino acid being the primary carbon source used by the parasite to feed its central carbon metabolism (Fig 8D). Growth of the parental cells is considerably improved by addition of at least 20 μM proline, the parental cells growing only 2-times slower in the presence of 0.5 mM compared to 5 mM proline (concentration in the SDM79-GlcFree medium). In contrast the Δfbpase cells are barely growing in the presence of 0.5 mM proline. This analysis showed that the Δfbpase mutant requires ~30-times more proline to grow at the same speed as the parental cells (Fig 8D), which may provide a rational explanation of the observed difference in behaviour of these two cell lines in vivo if proline concentrations are higher in the midgut compared to the oesophagus and/or the salivary glands. Unfortunately, the concentrations of proline in the tsetse organs are unknown. It is noteworthy that, immunofluorescence analyses using an immune serum against a bona fide glycosomal protein, aldolase, showed no difference in the number of glycosomes per cell between the parental, Δfbpase and Δfbpase/FBPase cell lines (S2 Fig).

In order to deepen the study of the in vivo role of FBPase, variations of the FBPase expression during the parasite cycle of wild-type trypanosomes were scrutinised by immunofluorescence analysis with an anti-FBPase antibody (Fig 9A). Assuming that the observed fluorescence is directly correlated to the amount of expressed proteins accessible to the antibody, all stages were seen to express FBPase in their glycosomes, yet in variable amounts (Fig 9A). The maximum signal intensities (Fig 9B) and the total amounts of fluorescent signal per surface unit (Fig 9C) were measured in a total of 454 individual cells isolated from more than 12 flies, further normalised to the values obtained on PCF cells from the same fly after removal of the background values, and finally plotted by stage in percentage of the fluorescent signals from PCF. The lowest maximum fluorescent signal intensities (Fig 9B) and total amounts of fluorescence per surface unit (Fig 9C) were observed in proventricular epimastigotes (DE, LE and SE) whereas the highest values were detected in salivary glands metacyclic forms (MT), and the differences between the two groups were statistically significant by ANOVA (p = 0.002, <0.001, <0.0001 and <0.0001 between MT Vs. DE, LE, SE and AE respectively by Tukey ad-hoc post-test on the maximum intensities, and p = 0.002 and <0.0001 between MT Vs. LE and AE for the total fluorescence per surface unit). The naturally lower FBPase expression in proventricular epimastigotes is in accordance with the absence of phenotype of the Δfbpase null mutants in terms of morphotype switching ability (Fig 9A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. FBPase is expressed in all parasite cycle stages yet at variable levels.

(A) Wild-type AnTat1.1E cultured bloodstream (cBSF) and procyclic (cPCF) forms, as well as parasites isolated from infected tsetse flies were fixed in methanol and stained with the anti-FBPase antibody (white) and DAPI (cyan). PC: procyclic trypomastigote, MS: mesocyclic trypomastigote, DE: dividing epimastigote, LE: long epimastigote, SE: short epimastigote, AE: attached epimastigote, and MT: metacyclic trypomastigote. (B-C) A region of interest was drawn on the phase picture of each individual cell and used for the quantification of the fluorescence provided by the anti-FBPase labelling. After background removal, the fluorescent signals were normalised (in %) to that obtained in PCF treated in the same experiment. The maximum fluorescence intensity (B) and total amount of fluorescence (C) corresponding to the FBPase expression levels were plotted according to the parasite stage. The total number of cells analysed from 3 independent experimental infections is indicated below the histogram bars for each stage. The different shades of grey represent groups of stages statistically identical between each other and different from those in the other groups according to a two-tailed ANOVA test with Tukey ad-hoc post-tests at 95% confidence for inter-group comparisons. * indicates the stage statistically different from all others with p<0.005.

https://doi.org/10.1371/journal.ppat.1007502.g009

FBPase is essential for production of metacyclic forms in proline rich-conditions

The fly passage phenotype observed may be due to an impairment of the parasite developmental program per se or to defects in migration, and/or to mobility defects during parasite migration in the tsetse and/or to non-adapted interactions with the tsetse environments. To distinguish between these possibilities, we tested the impact of the FBPase deletion in an in vitro model mimicking the parasite development upon inducible over-expression of the RNA binding protein 6 (RBP6), a central regulator of fly-stage differentiation [30]. The stage-specific morphology of induced cells, based on cell size and shape, as well as the relative position of the kinetoplast to the nucleus were monitored over the differentiation kinetics [31]. In PCF trypomastigotes the kinetoplast is at the posterior part of the cell, whereas in epimastigotes (EMF) the kinetoplast migrates to the opposite side of the nucleus and is found at the anterior part of the cell or in close proximity to the nucleus. Metacyclic trypomastigotes (MF) are smaller than EMF and PCF with the kinetoplast at the very end of the rounded posterior tip and a bloodstream form-like flagellar shape. In addition, we used two distinct stage-specific marker proteins for staging, i.e. EP procyclin (EP) for PCF [32] and calflagin for MF [33]. Calflagins are proteins localising to lipid rafts in the flagellar membrane and their amounts in bloodstream forms and MF are ten-fold enriched compared to that in PCF. Four days after the tetracycline-induction of RBP6 over-expression, ~30% of the cells (RBP6.i) had a characteristic MF morphology and after six days MF subpopulation reaches a peak at 55% (Fig 10A, left). Flow cytometry of RBP6.i cultures showed that EP expression was maintained in all cells during the first two days post induction and only then decreased proportionally to the fraction of the remaining PCF population (Fig 10A, middle). EP expression has previously been described in early salivary gland EMF [34]. The fraction of cells expressing calflagin constantly increased over time and attained the same percentage as the MF subpopulation after four and six days (Fig 10A, right). Previous reports showed that EP was not expressed in MF and that calflagins were detectable at basal levels in PCF but significantly up-regulated in MF [35, 36]. Additional evidence for developmental changes in these cultures is provided by RBP6 expression levels that decreased over time of the experiment (Fig 10C) and by a cell proliferation arrest (Fig 10D) in successfully differentiating cultures. The Δfbpase RBP6.i mutant, in which RBP6 was conditionally over-expressed in the Δfbpase background, was analysed in parallel under exactly comparable conditions (Fig 10B). Most importantly, the RBP6 expression levels as determined by quantitative western blotting (Fig 10C) were similar in the Δfbpase and parental backgrounds over the entire time and did not decrease after 8 days. Induced Δfbpase RBP6.i cells developed into EMF after two days but completely failed to differentiate into MF and consequently the fraction of EMF remained stable at around 50% for the duration of the experiment. EP and calflagin expression levels were not different between cell lines all over the time course and the population growth was maintained (Fig 10D). As the EP-positive population did not decrease, it is likely that the developmental blockage occurred in early EMF. Together, we can conclude that FBPase is absolutely essential for development in culture and hence this strong phenotype is cell autonomous.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 10. FBPase is essential for metacyclogenesis in culture.

RBP6-induced differentiation kinetics is shown for the parental EATRO1125.T7T RBP6.i cell line (A) and the Δfbpase RBP6.i mutant (B). RBP6 over-expression was induced at time 0 by addition of 1 μM tetracycline. Developmental stages were morphologically scored by cell size, shape and the relative position of the kinetoplast to the nucleus within the cell (n >100 cells per time point, biological triplicate, SEM). Expression of the stage-specific proteins EP and calflagin was quantified at the single cell level by flow cytometry and gated into two populations with low or high fluorescence. The left panels show histograms of morphotypes with PCF in yellow, EMF in blue and MF forms in red. The middle and right panels present morphotype fractions from the histogram (solid lines, data points colour matched to histogram) together with the fractions of stage marker expression levels (cells with high fluorescence/total number of cells, dashed lines) for EP and calflagin, respectively. (C) Western blot showing the RBP6 over-expression levels upon tetracycline induction of the indicated cell lines. PFR was used as loading control and the Δfbpase knock out was verified by anti-FBPase. (D) Growth curves of the RBP6.i and Δfbpase RBP6.i populations upon induction of RBP6 over-expression. C and D are from a representative replicate of the three experiments shown in A and B.

https://doi.org/10.1371/journal.ppat.1007502.g010

Trypanosomes and cell cultures

The PCF of T. brucei EATRO1125.T7T (TetR-HYG T7RNAPOL-NEO) and AnTat1.1E [26] were cultivated at 27°C in the presence of 5% CO₂ in SDM79 medium containing 10% (v/v) heat-inactivated fetal calf serum and 3.5 mg ml^-1 hemin [48] or in a glucose-depleted medium derived from SDM79, called SDM79-GlcFree. This SDM79-GlcFree medium consists of a glucose-depleted SDM79 medium, containing 20% (v/v) heat-inactivated fetal calf serum, in which parental cells were cultured during 72 hours in order to consume the glucose coming from the serum and then diluted with the same volume of glucose-depleted SDM79 medium without serum to finally obtain SDM79-GlcFree. Glucose depletion was verified by NMR spectrometry analyses (with a detection threshold of glucose ≤ 1 μM) and to prevent import of residual glucose, a non-metabolised glucose analogue inhibiting glucose transport (N-acetyl-D-glucosamine, 50 mM) was added in the medium [21]. The same protocol was followed to produce SDM79-GlcFree and proline-depleted medium, except that the starting glucose-depleted SDM79 medium containing 100 μM proline, instead of 5 mM, was diluted with one volume of proline/glucose-depleted SDM79. Considering the rate of proline consumption of PCF trypanosomes, we estimated that the proline-depleted SDM79-GlcFree medium contains less than 20 μM proline. To estimate the cell density by the Alamer Blue assay, cells diluted in 200 μl of proline-depleted SDM79-GlcFree medium in which 1 μM to 100 mM proline was added by serial two-fold dilutions at a 2 x 10⁶ cell density, were incubated for 48 h at 27°C in microplates, before adding 20 μl of 0.49 mM Alamar Blue (Resazurin). Measurement of fluorescence was performed with the microplate reader Fluostar Optima (BMG Labtech) at 550 nm excitation wavelength and 600 nm emission wavelength.

Gene knockout and rescue cell lines

The Δppdk mutant cell line was obtained by replacing both alleles of the PPDK gene (Tb927.11.3120, http://www.genedb.org/genedb/tryp/) by two different plasmids encoding the hygromycin resistance gene and the T7 RNA polymerase gene for the first one, and the neomycin resistance gene and the tetracycline repressor gene under the control of the T7 RNA polymerase promoter for the second one, as described before [49]. Replacement of the PEPCK gene (Tb927.2.4210) by the blasticidin (BSD) and puromycin (PAC) resistance markers via homologous recombination was described before (Δpepck cell line) [22]. The Δppdk/Δpepck double mutant was also described before [12]. Replacement of both alleles of the FBPase gene (Tb927.9.8720) by the blasticidin (BSD) and puromycin (PAC) resistance markers via homologous recombination was performed with DNA fragments containing a resistance marker gene flanked by the FBPase UTR sequences. Briefly, the pGEMt plasmid was used to clone an HpaI DNA fragment containing the BSD or PAC resistance marker gene preceded by the FBPase 5'-UTR fragment (729 bp) and followed by the FBPase 3'-UTR fragment (694 bp). The PAC resistance marker replaced one allele of the FBPase gene via homologous recombination and BSD resistance marker replaced the second allele. The EATRO1125.T7T parental cell line, which constitutively expresses the T7 RNA polymerase gene and the tetracycline repressor under the control of a T7 RNA polymerase promoter for tetracycline inducible expression (TetR-HYG T7RNAPOL-NEO) [50], was used to generate the FBPase knockout cell line. Transfection and selection of drug-resistant clones were performed as reported previously [11]. Transfected cells were selected in SDM79 medium containing hygromycin (25 μg ml^-1), neomycin (10 μg ml^-1), blasticidin (10 μg ml^-1) and/or puromycin (1 μg ml^-1). The selected cell line TetR-HYG T7RNAPOL-NEO Δfbpase::BSD/Δfbpase::PAC is called Δfbpase. To produce the add-back Δfbpase/FBPase rescue cell lines, a FBPase ectopic copy under control of the PARP promoter in the pLew100 expression vector, which contains the phleomycin resistance gene (kindly provided by E. Wirtz and G. Cross) [51], was introduced in the Δfbpase null background. The pLew100-FBPase plasmid was generated by introduction the full length FBPase gene in the HindIII and BamHI restriction sites of the vector (Genecust). The Δfbpase/FBPase cell line was selected in SDM79 medium containing phleomycin (5 μg ml^-1) in addition to the 4 other antibiotics used to select the Δfbpase cell line.

For experimental infections of Δtsetse flies, the Δfbpase null mutant was generated in the pleomorphic AnTat1.1E PCF strain, using the same approach described above. Then, two distinct strategies were used to produce the add-back Δfbpase/FBPase rescue cell lines in the AnTat1.1E Δfbpase null background. The Δfbpase/FBPase-1 cell line expresses a FBPase ectopic copy under control of the PARP promoter in the pLew100 expression vector as described above for the EATRO1125 Δfbpase/FBPase cell line. The Δfbpase/FBPase-2 cell line was generated by in situ re-insertion of one FBPase gene under the control of its endogenous UTRs. The 3.5 kb DNA fragment for in situ re-insertion consists on the FBPase coding sequence preceded by its 5' intergenic region (722 bp) and followed by its 3' intergenic region (662 bp), the phleomycin resistance gene (375 bp) and a second copy of the FBPase 3' intergenic region. The 3.5 kb DNA fragment flanked by two HpaI restriction sites was introduced into the pBluescript vector (Genecust) and the resulting plasmid was digested by HpaI before transfection of the AnTat1.1E Δfbpase cell line. The Δfbpase/FBPase-1 and Δfbpase/FBPase-2 cell lines were selected in SDM79 medium containing blasticidin (10 μg ml^-1), puromycin (1 μg ml^-1) and phleomycin (5 μg ml^-1).

Inhibition of gene expression by RNAi

Accession numbers of genes targeted by RNAi are as follows; fructose-1,6-bisphosphatase (FBPase, Tb927.9.8720), glycerol kinase (GK, Tb927.9.12550-Tb927.9.12630), glycerol-3-phosphate dehydrogenase (GPDH, Tb927.8.3530) and triosephosphate isomerase (TIM, Tb927.11.5520). RNAi-mediated inhibition of gene expression was performed in the EATRO1125.T7T PCF by expression of stem-loop “sense-antisense” RNA molecules of the targeted sequences [50, 52] using the pLew100 expression vector [51].

To inhibit by RNAi the expression of the FBPase gene, a 597-bp fragment of the FBPase gene (from position 152 to 749) was introduced in the pLew100 vector to produce the pLew-FBPase-SAS plasmid. Briefly, a PCR-amplified 716-bp fragment, containing the antisense FBPase sequence with restriction sites added to the primers, was inserted into the HindIII and BamHI restriction sites of the pLew100 plasmid. The separate 615-bp PCR-amplified fragment containing the sense FBPase sequence was then inserted upstream of the antisense sequence, using HindIII and XhoI restriction sites (XhoI was introduced at the 3'-extremity of the antisense PCR fragment). The resulting plasmid pLew-FBPase-SAS contains a sense and antisense version of the targeted gene fragment, separated by a 89-bp fragment, under the control of a PARP promoter linked to a prokaryotic tetracycline operator. The same strategy was used to produce the pLew-GK-SAS, pLew-GPDH-SAS and pLew-TIM-SAS plasmids designed to inhibit the expression of the GK, GPDH and TIM genes, respectively. The size of the GK, GPDH and TIM targeted sequences are 617 bp (from position 460 to 1077), 622 bp (from position 218 to 840) and 558 bp (from position 67 to 625), respectively. The ^RNAiFBPase mutant was generated by transfecting the EATRO1125.T7T parental cell line with the pLew-FBPase-SAS plasmid and selection in glucose-rich SDM79 medium containing hygromycin (25 μg ml^-1), neomycin (10 μg ml^-1) and phleomycin (5 μg ml^-1). The triple mutant cell lines (Δppdk/Δpepck/^RNAiGK, Δppdk/Δpepck/^RNAiGPDH and Δppdk/Δpepck/^RNAiTIM) were obtained by introducing the relevant plasmid in the Δppdk/Δpepck cell line, after selection in glucose-rich SDM79 medium containing hygromycin (25 μg ml^-1), neomycin (10 μg ml^-1), phleomycin (5 μg ml^-1), blasticidin (10 μg ml^-1) and puromycin (1 μg ml^-1). Aliquots were frozen in liquid nitrogen to provide stocks of each line that had not been cultivated long term in medium. Induction of RNAi cell lines was performed by addition of 1 μg ml^-1 tetracycline.

Preparation of glycosomal fractions and enzymatic activities

Subcellular fraction enriched in glycosomes was prepared by differential centrifugation as described in [53], after homogenising the cells with silicon carbide as grinding material. Briefly, 2 x 10⁹ cells were washed once in 10 ml of STE (25 mM Tris, 1 mM EDTA, 250 mM sucrose pH 7.8). After centrifugation, the pellet was resuspended in 0.15 ml of homogenisation buffer STE (STE supplemented with ‘Complete EDTA-Free’ protease-inhibitor cocktail, Roche Applied Science, Mannheim, Germany) and ground in a pre-chilled mortar with 1.5 gr of wet-weight silicon carbide per gr of cell pellet. The cells were microscopically checked for at least 90% disruption. The cell lysate was diluted in 7 ml of homogenisation buffer, centrifuged at 1,000 g and then at 5,000 g for 10 min each, at 4°C. The supernatant was centrifuged at 33,000 g for 10 min at 4°C to yield the cytosolic fraction and the glycosome-enriched pellet, which was washed once with 1 ml of STE, centrifuged at 33,000 g for 10 min at 4°C before resuspension in 0.1 ml of STE.

The FBPase activity of aliquots of the glycosomal fractions was measured following the reduction of NADP⁺ at 340 nm in the presence of 100 mM triethanolamine (pH 7.6), 2 mM MgCl₂, 0.1 mM EDTA, 0.3 mM NADP⁺, 10 mM F1,6BP, 25 μg ml^-1 glucose-6-phosphate isomerase and 25 μg ml^-1 glucose-6-phosphate dehydrogenase. The glycerol kinase activity was measured at 340 nm via oxidation of NADH according to published procedures [54].

Western blot analyses

Total protein extracts (5 x 10⁶ cells), or glycosomal and cytosolic fractions, of parental (AnTat1.1E, EATRO1125.T7T) or mutant PCF of T. brucei were separated by SDS-PAGE (10%) and immunoblotted on TransBlot Turbo Midi-size PVFD Membranes (Bio-Rad) [55]. Immunodetection was performed as described [55, 56] using as primary antibodies, the rabbit anti-FBPase (1:1,000, gift from P. Michels, Edinburgh, UK), the rabbit anti-SBPase (1:250, gift from P. Michels, Edinburgh, UK), the rabbit anti-GK (1:5,000, gift from P. Michels, Edinburgh, UK), the rabbit anti-G6PDH (1:1,000, gift from P. Michels, Edinburgh, UK), the rabbit anti-TIM (1:1,000, gift from P. Michels, Edinburgh, UK), the rabbit anti-PFR (1:10,000), the rabbit anti-enolase (1:100,000, gift from P. Michels, Edinburgh, UK), the rat anti-PEPCK (1:1,000, gift from T. Seebeck, Bern, Switzerland) [22], the rabbit anti-GPDH (1:1,000) [57], the rabbit anti-hsp60 (1:10,000) [58], the rabbit anti-PPDK (1:1,000) [11], anti-EP mouse monoclonal TBRP1/247 (1:500) (Biozol), anti-calflagin (1:1,000, gift of D. Engman, Chicago, USA [59]) and the rabbit anti-RBP6 (1:1,000, gift from C. Tschudi, New Haven, USA) [30]. Anti-rabbit or anti-rat IgG conjugated to horseradish peroxidase (Bio-Rad, 1:5000 dilution) was used as secondary antibody. Revelation was performed using the Clarity western ECL Substrate as described by the manufacturer (Bio-Rad). Images were acquired and analysed with the ImageQuant LAS 4000 luminescent image analyser.

In vitro differentiation of PCF expressing RBP6

The RBP6 gene was amplified by PCR and cloned via HindIII/BamHI into the pLew100v5b1d vector (pLew100v5 modified with a blasticidin resistance gene BSD). The Δfbpase cell and its parental EATRO1125.T7T cell line were transfected with the pLew100v5-RBP6 linearised with NotI in pools to generate the Δfbpase RBP6 cell line with the genotype TETR::HYG T7RNAP::NEO Δfbpase::BLAS/Δfbpase::PAC RBP6^Ti::BLE and RBP6, respectively. In vitro differentiation experiments were done as described in [30] in SDM79 medium without glucose in the presence of 50 mM N-acetyl-D-glucosamine and 10% (v/v) heat-inactivated fetal calf serum. The faster kinetics of differentiation observed in our experiments as compared to that described in [30] is likely due to the different parental backgrounds (Lister 427 29:13 versus EATRO1125.T7T) or to the different media (Cunningham’s medium [60] versus modified SDM79) that were used in the two labs.

Flow cytometry analysis of stage-specific surface marker

2 × 10⁷ cells were harvested and fixed overnight with 2% paraformaldehyde in PBS at 4°C, washed thrice with PBS and resuspended in 500 μl PBS. Cells were incubated with monoclonal mouse anti-procyclin EP (TBRP1/247, 1:500) in 1% BSA. For calflagin detection 3 × 10⁷ cells were permeabilised with 0.2% NP-40 for 5 min prior incubation with polyclonal mouse α-calflagin (1:1,000) [59] in 1% BSA. Alexa Fluor 488-conjugated goat antibodies were used as secondary antibodies. Samples were analysed with a FACSCalibur cell analyser (Becton Dickinson) and data were evaluated with the FlowJo 8.8.6 software.

Mass spectrometry analyses of [¹³C]-incorporation into cellular metabolites

EATRO1125.T7T parental and mutant cell lines grown in SDM79 medium were washed twice with PBS and resuspended in PBS containing [U-¹³C]-proline (or [U-¹³C]-glycerol) with or without the same amount of non-labelled glucose. The cells were incubated for 2 h at 27°C before being collected on filters by fast filtration preparation (2 x 10⁷ cells per filter) for mass spectrometry analysis, as described before [22]. Metabolites were analysed by ion-exchange chromatography coupled with tandem mass spectrometry (IC-MS/MS) using the method described by Bolten et al. [61]. Retention time on the column and multiple reactions monitoring (MRM) transition of each analysed metabolite are shown in Table 1 of [21]. The ¹³C mass isotopomer distribution of intracellular metabolites was determined from relevant isotopic clusters in the IC-MS/MS analysis, according to Kiefer et al. [62]. ¹³C mass isotopomer distribution measurements were performed using a triple quadrupole mass spectrometer (4000Qtrap, Applied Biosystems). To obtain [¹³C]-labeling patterns (¹³C isotopologues), isotopic clusters were corrected for the natural abundance of isotopes other than ¹³C, using the in-house software IsoCor (available at MetaSys) [63].

Analysis of excreted end products from glucose and glycerol metabolism by proton NMR

1 x 10⁸ T. brucei PCF were collected by centrifugation at 1,400 g for 10 min, washed once with phosphate-buffered saline (PBS) and incubated in 5 ml of PBS supplemented with 2 g l^-1 NaHCO₃ (pH 7.4). Cells were maintained for 6 h at 27°C in incubation buffer containing 20 μmol of D-glucose or [U-¹³C]-glycerol (4 mM). The integrity of the cells during the incubation was checked by microscopic observation. The supernatant was collected and 50 μl of maleate solution in D₂O (10 mM) was added as internal reference. H-NMR spectra were performed at 500.19 MHz on a Bruker Avance III 500 HD spectrometer equipped with a 5 mm cryoprobe Prodigy. Measurements were recorded at 25°. Acquisition conditions were as follows: 90° flip angle, 5,000 Hz spectral width, 32 K memory size, and 9.3 sec total recycle time. Measurements were performed with 64 scans for a total time close to 10 min 30 sec. Resonances of the obtained spectra were integrated and metabolites concentrations were calculated using the ERETIC2 NMR quantification Bruker program.

Tsetse fly maintenance, infection and dissection

Tsetse flies (Glossina morsitans morsitans) were maintained, infected and dissected at the Institut Pasteur as previously described [28]. Teneral males were collected 24 h to 48 h post-eclosion and artificially fed through a silicone membrane with 6–9 x 10⁶ parasites ml^-1 in SDM79 medium supplemented with 10% FCS for their first meal. Flies were then maintained in Roubaud cages for one month at 26°C and 60% hygrometry and fed twice a week with mechanically defibrinated sheep blood. Flies were starved for at least 48 h before being individually dissected 28 days after ingestion of the infected meal. Salivary glands were first rapidly dissected into a drop of PBS. The whole tsetse alimentary tract was then dissected and arranged lengthways for assessment of parasite presence. The proventriculus and anterior midgut were physically separated from the posterior midgut in a distinct PBS drop. Tissues were dilacerated to allow parasites to spread in PBS, parasites were recovered and treated for further experiments no more than 15 min after dissection.

Immunofluorescence analysis

For immunofluorescence, parasites were rapidly allowed to settle on poly-lysine coated slides until drying. Cells were fixed for 10 sec in methanol at -20°C and re-hydrated in PBS during 10 min. Slides were then incubated for 45 min at 37°C with the anti-FBPase primary antibody diluted at 1:10 in PBS with 0,1% bovine serum albumin. Slides were washed in PBS and incubated for 30 min at 37°C with the appropriate anti-rabbit secondary antibody coupled to an Alexa 488 fluorophore (Invitrogen). Slides were then stained with DAPI for visualisation of their kinetoplast and nuclear DNA contents and mounted under coverslip with Prolong antifade reagent (Invitrogen). IFA experiments were repeated on trypanosomes issued from 2 to 4 flies and from 6 distinct experimental infections. Slides were finally observed with a DMI4000 epifluorescence microscope (Leica) and images were captured with a Horca 03G5 camera (Hamamastu).

Measurements and analyses

Normalisation of signals was carried out by parallel manipulation of min/max signals in ImageJ (NIH). Statistical analyses and plots were performed with XLSTAT 2015.4.01 (Addinsoft) and Excel 2011 (Microsoft), respectively. Statistical analyses include ANOVA with Tukey ad-hoc post-tests for inter-group comparison with 95% confidence.