Probing the Role of Compartmentation of Glycolysis in Procyclic Form Trypanosoma brucei RNA INTERFERENCE STUDIES OF PEX14, HEXOKINASE, AND PHOSPHOFRUCTOKINASE*

Trypanosoma brucei and related organisms contain an organelle evolutionarily related to peroxisomes that se-questers glycolysis, among other pathways. We have shown previously that disruption of protein import into this organelle, the glycosome, can be accomplished through RNA interference (RNAi)-mediated knockdown of the peroxin PEX14. Decreased PEX14 in turn leads to cell death, which, at least in the procyclic stage, can be triggered by the presence of glucose. Here we show that fructose, which is taken up and metabolized by procyclic form T. brucei , and glycerol, which interfaces with the glycosomal glycolytic pathway, are also toxic during PEX14 RNAi. Earlier computer modeling studies predicted that glycolysis would be toxic to T. brucei in the absence of glycosomal compartmentation because of the intrinsic lack of feedback regulation of the parasite hexokinase and phosphofructokinase. To further test this hypothesis, we performed double RNAi, targeting hexokinase and PEX14. Knockdown of hexokinase rescued PEX14 knockdown cells from glucose toxicity, even though glycosomal proteins continue to be mislocalized to the cytosol. Knockdown of phosphofructokinase was benign in the absence of glucose but toxic in the presence of glucose. When PEX14 and phosphofructokinase RNAi constructs. cell equivalents/sample electrophoresis gels

Trypanosoma brucei subspecies cause African sleeping sickness in humans and nagana in cattle. The human disease is fatal if untreated, and the animal disease continues to compromise agricultural development in Africa. T. brucei and related protozoa (kinetoplastids and bodonids) are the only organisms known to sequester glycolysis within a cellular organelle. This organelle, the glycosome, is evolutionarily related to peroxisomes. Glycosomes and peroxisomes share a number of metabolic functions, such as the ␤-oxidation of fatty acids and etherlipid biosynthesis (1). In addition, targeting sequences that route proteins to these organelles are closely related (2), as are the biogenesis proteins, which are known as peroxins (PEXs) 1 (3)(4)(5)(6). The hallmark of the glycosome, in distinction to peroxisomes, is the presence of the Embden-Meyerhof segment of the glycolytic pathway (see Fig. 1 for schematic). Metabolism of glucose to either 1,3-bisphosphoglycerate or 3-phosphoglycerate (depending on the developmental stage and species) occurs in the glycosome, and further conversion to pyruvate can occur in both the glycosome and cytosol. Glycosomes and peroxisomes are closed systems in which ADP/ATP and NAD/ NADH must be kept in balance internally, and transporters are required for the transfer of substrates and products (7,8). Glycosomes are essential for the survival of both bloodstream (mammalian) and procyclic (insect midgut) form T. brucei (4,9).
Glycolysis and the glycosome have received considerable attention because the mammalian stages of T. brucei rely exclusively on glycolysis for the generation of ATP. However, the procyclic stage does not require glucose and primarily uses amino acid catabolism to generate ATP (10). In glycolysis, for each molecule of glucose that is metabolized, two ATPs are initially invested and then four are generated, for a net yield of two ATPs (Fig. 1). The net ATP is generated in the cytosol, whereas ATP balance within the glycosome is maintained by multiple enzymes for which the interplay has not been established (11). Hexokinase (HK) mediates the first reaction in glycolysis by phosphorylating glucose in an ATP-dependent manner, and, following isomerization, phosphofructokinase (PFK) generates fructose 1,6 bisphosphate. In most organisms, the activity of these enzymes is tightly regulated. However, this is not the case for the kinetoplastid homologues, and in particular HK is not inhibited by glucose 6-phosphate (12). This lack of regulatory control creates an automatic feed forward mechanism in kinetoplastids, sometimes referred to as "turbo glycolysis" (13). Computer modeling studies, done on data derived from bloodstream form parasites, suggest that the role of compartmentation of glycolysis within the glycosome is to keep the activities of HK and PFK sequestered from cellular pools of substrates (e.g. ATP) and that this is essential to protect the cell (14,15). Mislocalization of these unregulated kinases to the cytosol might lead to runaway phosphorylation of hexoses (because of the availability of the cytosolic pool of ATP), ATP depletion, and cell death.
We have previously identified and examined the role of one of the T. brucei glycosome biogenesis proteins, PEX14 (4). This peroxin is an essential component of the receptor-docking complex, which is involved in the targeting of matrix proteins to the glycosome and translocation of these matrix proteins across the glycosomal membrane. We demonstrated by RNAi that PEX14 is essential for the proper localization of glycosomal matrix proteins and that its depletion results in the death of both bloodstream and procyclic form parasites. In addition, we showed that one specific cause of this lethality in procyclic stage parasites was glucose. Here we extend our analysis to demonstrate that several sugars or components of the glycolytic pathway have a similar lethal phenotype in cells where PEX14 has been depleted by RNAi. In addition, we demonstrate the specific role played by glycolysis in this phenotype by construction of T. brucei double RNAi cell lines that knock down both PEX14 and either HK or PFK expression.

MATERIALS AND METHODS
Growth of Parasites-T. brucei strain TREU667 (16,17) and 29-13 (16) procyclic form parasites were maintained in the JRH Biosciences formulation of SDM-79 supplemented with 2 g/liter sodium bicarbonate, 3.75 mg/liter hemin, and 10% fetal calf serum. G-418, hygromycin, and phleomycin were added to a final concentration of 15, 50, and 2.5 g/ml, respectively. This formulation does not contain glycerol, unlike our previous formulation of SDM-79. In addition, NaH 2 PO 4 ⅐ H 2 O was increased from 126 to 954 mg/liter, glutamine was increased from 320 to 513 mg/liter, and serine was reduced from 71 to 10 mg/liter. We also obtained a special order medium from JRH Biosciences that further lacked glucose and glucosamine (SOGG medium). When cells were grown in SOGG, unless otherwise noted, we used serum that was dialyzed against a 40-fold volume of phosphate-buffered saline (dialysis tubing was extensively washed to remove glycerol). RNAi experiments employed the strain 29-13, which contains integrated copies of T7 RNA polymerase and the tetracycline (Tet) repressor, which allows Tetregulated expression of introduced genes. Cell densities were determined using a Z1 Coulter counter, and growth curves were performed using duplicate or triplicate cultures. Cultures in log phase were diluted to 10 6 cells/ml to begin the experiment, and RNAi was induced by the addition of 4 g/ml Tet. Tet was added every 24 -48 h, and cultures were kept below 2 ϫ 10 7 cells/ml.
Plasmid Construction-RNAi constructs were based on the plasmid pZJM, which contains a cloning site between opposing T7 promoters under the control of the Tet operator (18). This plasmid can be integrated into the transcriptionally silent ribosomal DNA intergenic spacer. Amplification of the relevant fragments from T. brucei genomic DNA used primers containing HindIII sites. For HK, the forward primer was 5Ј-AAGCTTCAATATCCTCGAACACATC-3Ј, and the reverse primer was 5Ј-AAGCTTCTCGAGGAAGCCCTTCGTCCACCGG-3Ј. There are two HK loci in T. brucei (19), HK1 and HK2, which differ only by 1 bp in the 526-bp region used for RNAi (nt 18 -543 of the 1416-nt open reading frame). For PFK, primers 5Ј-AAGCTT-GAATCTCGCAGCCGCGTTACG-3Ј (forward) and 5Ј-AAGCTTGAA-GAGGATGTTGACACCG-3Ј (reverse) were used to amplify nt 10 -582 of the 1461-nt PFK open reading frame. The PCR fragments were cloned into pGEM-T Easy and sequenced. Cloned fragments were then ligated into the vector pZJM-PEX14 (4) following digestion of both vector and insert with HindIII (see Fig. 3), yielding both orientations (tail-to-tail (A) or tail-to-head (B)). The A orientation plasmid was digested with XhoI, taking advantage of a site in the reverse primer (HK) or the naturally occurring site within the cloned fragment, 5 bp upstream of the reverse primer (PFK), to release the PEX14 fragment. Religation then generated constructs targeting only the glycolytic enzyme.
For the PFK strains, PFK and PEX14 transcript levels were determined by Northern analysis on RNA samples (10 g) prepared as above. Samples were separated on a 1% formaldehyde-agarose gel (NuSieve: agarose, 3:1) and transferred to nylon membranes (Nytran N, Schleicher & Schü ll). Following prehybridization in ULTRAHyb (Ambion), RNA probes that were prepared using the Promega Riboprobe-T7 kit were added to 10 6 cpm/ml. The probes detected both the native transcripts and those produced by the RNAi constructs. Blots were washed twice with 0.1ϫ SSC (15 mM NaCl/1.5 mM sodium citrate, pH 7.0) and 0.1% SDS for 1 h at 65°C.
Western Analysis-Cells were washed in 1ϫ phosphate-buffered saline three times and resuspended in SDS sample buffer. Approximately 5 ϫ 10 6 cell equivalents/sample were separated by electrophoresis on 10% acrylamide gels (Cambrex PAGEr). Samples were transferred to nitrocellulose membranes (Protran) and probed with rabbit anti-HK or anti-PFK (a gift from Paul Michels), which was prepared against purified T. brucei HK1 or PFK (20). The second step was protein A (1:5000) coupled to horseradish peroxidase, detected using the ECL Plus system (Amersham Biosciences). After quantitating the HK signal, blots were stripped and incubated with anti-PEX14 (1:100) (4) and anti-NOG1 (1:5000) (21) as a loading control. Image analysis was done using the Amersham Biosciences storm PhosphorImager system and software.

PEX14 RNAi Induction Is Lethal with Various Sugars or
Glycerol-In our previous work, we established an RNAi procyclic form cell line in which the expression of double-stranded RNA corresponding to the 5Ј half of PEX14 was induced by Tet. We demonstrated that induced expression of PEX14 RNAi led to mislocalization of glycosomal matrix proteins and was lethal for parasites when glucose was present in the medium (4). This phenotype was evident at 4 days following the induction of RNAi in procyclic forms, apparently requiring the dilution of PEX14 through cell division and protein turnover. We wanted to determine what concentration of glucose was lethal and whether other sugars or other components of our medium related to glycosomal metabolism might be lethal as well. Although procyclic form cultures are normally maintained in SDM-79 medium, the previous experiments used a modified RPMI-based medium. We therefore obtained a special formulation of SDM-79 that lacked glycerol, glucose, and glucosamine.
When RNAi was induced, the PEX14 RNAi parasites in SOGG medium supplemented with 10% non-dialyzed fetal calf serum died after 4 days (results not shown). This concentration of fetal calf serum yielded a final concentration of 356 M glucose. Growth of the PEX14 RNAi cell line was tested in SOGG medium using dialyzed serum (yielding a final concentration of 9 M glucose) and supplemented with different concentrations of glucose ( Fig. 2A). In the absence of additional glucose, induced cultures showed only a slight growth defect, with one-half the cell numbers of their uninduced counterparts by day 11. In the presence of 100 M glucose, uninduced cultures grew somewhat better than the induced cultures. Higher levels of glucose (1 and 10 mM) were toxic to induced cultures. The difference in cumulative cell number by day 11 for these cultures was approximately 2 orders of magnitude. As we found previously in the RPMI-based medium, if cells were grown in the absence of glucose for several days during PEX14 RNAi induction, and then glucose was added back to 10 mM, the lag period was eliminated, and the culture died within 24 h (data not shown).
Galactose is not taken up by T. brucei (23,24), and, as expected, its presence had no effect on the growth of induced parasites (Fig. 2B). In contrast, a lethal phenotype was observed in induced cultures containing 10 mM fructose and 10 mM glycerol (Fig. 2, B and C). Glycerol was additionally tested at various concentrations ranging from 100 M to 10 mM. The addition of 100 M glycerol to induced cultures reduced the number of viable parasites 10-fold, whereas 1 and 10 mM glycerol reduced them 100-fold (data not shown). 2-Deoxyglucose is phosphorylated by HK but inhibits glucose-6-phosphate isomerase and is not further metabolized (Fig. 1). It arrested growth almost immediately, both in the presence and absence of RNAi induction (Fig. 2C). This effect was also seen in another cell line, TREU667, and was ameliorated by increasing amounts of glucose in the medium (data not shown). parasites suggest that sequestration of glycolysis within the glycosome may be essential to protect the cell from the potentially lethal effects of unregulated HK and PFK in the cytosol (15). We tested this hypothesis with procyclic parasites in vivo by taking advantage of the mislocalization phenotype that arises when PEX14 is depleted. We created double RNAi constructs in T. brucei with portions of PEX14 and either HK1 or PFK. The 5Ј end of HK1 was amplified from T. brucei genomic DNA and cloned into the PEX14 RNAi construct in two orientations with respect to PEX14 (Fig. 3). The double-stranded RNA should target transcripts for both HK genes (HK1 and HK2) because the two differ by only one nt in the region employed. An RNAi construct targeting HK1/2 mRNA (but not PEX14) was constructed as well. These constructs were stably transfected into T. brucei 29-13. The levels of HK1/2 and PEX14 mRNA were evaluated by reverse transcription and real-time PCR on day 2 after induction of RNAi (Fig. 4). HK1/2 mRNA levels in induced HK1 and HK1-PEX14 A and B RNAi cell lines were lowered to ϳ15% of the levels of the corresponding uninduced cell line. In the original PEX14 RNAi line, HK1/2 mRNA levels did not change upon induction. Similarly, PEX14 mRNA levels were reduced in the PEX14 and the two HK-PEX14 double RNAi cell lines upon addition of Tet but remained unchanged in the HK knockdown. In the PEX14 knockdown, PEX14 mRNA was ϳ25% of the uninduced control, and in the double RNAi mutants, it was 35% of the control. Thus, knockdown of both mRNA target molecules was accomplished by induction of one RNAi construct with two gene fragments.
HK Knockdown Rescues PEX14 Knockdown-To test whether depletion of HK could rescue the PEX14 knockdown, we examined the growth of cultures of HK, HK-PEX14 B, and PEX14 RNAi cell lines in the presence of glucose. At the same time, we assessed the levels of the relevant proteins as well as compartmentation of glycosomal proteins. Immunoblotting of cell lysates from several time points revealed that following induction of RNAi in the HK-PEX14B line, both PEX14 and HK protein levels were reduced to levels comparable with those seen by real-time PCR analysis (Fig. 5A) (HK1 and HK2 both have molecular masses of 51 kDa). We also assessed when mislocalization of glycosomal proteins occurred, because it is presumably linked to the detrimental phenotype. Immunofluorescence analysis was performed using an anti-glycosome antibody that recognizes three different glycosomal matrix proteins (pyruvate phosphate diki-nase, aldolase, and glyceraldehyde-phosphate dehydrogenase) (22). The HK RNAi knockdown showed the punctate pattern typical of uninduced cells at all time points (day 8 is shown in Fig. 5B). In contrast, none of the PEX14 RNAi cells had the normal punctate pattern of staining 4 days after induction (Fig. 5B). By this time the population had stopped growing (Fig. 5C). The population began to increase again around day 11. We have shown previously that this escape is accompanied by re-expression of normal levels of PEX14 (4). The slower growing HK-PEX14 double RNAi cell line showed a longer phenotypic lag for mislocalization. Little mislocalization of glycosomal proteins was seen at day 4, but a strong mislocalization pattern was observed at day 8, when only about 5% of the cells demonstrated the punctate phenotype (Fig. 5B). Of the remaining cells, approximately one-third showed a diffuse staining pattern, whereas two-thirds had both diffuse and punctate characteristics. By 11 days postinduction, only 2% of cells had a punctate pattern of staining, two-thirds had diffuse cytosolic staining, and one-third had both diffuse and punctate characteristics. During the time period when glycosomal proteins were accumulating in the cytosol of the HK-PEX14 RNAi cell line, there was no slowing of proliferation (Fig. 5C), and microscopic analysis showed no evidence of significant cell death. Taken together, these data show that a decrease in HK rescues the PEX14 knockdown from glucose toxicity.
Effects of Silencing of PFK and PEX14 -The second ATPrequiring step in glycolysis is mediated by PFK within the glycosome. We therefore generated RNAi constructs targeting PFK and PEX14 mRNA as well as one targeting PFK alone. mRNA levels in the corresponding single and double RNAi lines were evaluated by Northern analysis. As in the case of the HK mutants, significant reductions of both PFK mRNA and PEX14 mRNA (Fig. 6A) were seen in the double RNAi constructs, whereas only the expected knockdown was observed in the single RNAi lines. Immunoblot analysis of induced cultures of PFK and PFK-PEX14B showed that the proteins were similarly depleted (Fig. 6, C and E). On day 8, PFK levels were reduced to 9 and 25% of control levels in the single and double knockdown lines, respectively. On day 8, PEX14 was decreased to 10% of control levels in the double knockdown line.
We tested the effects of PFK depletion on growth in various media. In contradistinction to the result obtained with the knockdown of HK, knockdown of PFK mRNA resulted in sensitivity to glucose; growth ceased and cells began to die (Fig.  6B). RNAi-induced knockdown of PFK would create a bottleneck in glycolysis at the same position as 2-deoxyglucose with respect to ATP consumption, which we demonstrated above to be toxic to the parasites. The effects of PFK RNAi take longer to appear than the effects of 2-deoxyglucose, at least in part because the RNAi-induced phenotype requires that the preexisting PFK be diluted to low levels. In contrast to glucose, the addition of glycerol to the medium was beneficial whether or not PFK RNAi was induced (Fig. 6B).
Because the knockdown of PFK alone was so deleterious in the presence of glucose, it is not surprising that knockdown of PFK did not rescue PEX14 RNAi for growth in glucose (data not shown). We predicted that PFK depletion would not rescue the PEX14 RNAi line for growth in glycerol, because glycerol interfaces with the glycolytic pathway well below the reaction mediated by PFK (Fig. 1). The PFK-PEX14 double RNAi cell line was evaluated in the presence of 1 mM glycerol for growth (Fig. 6D). On day 4, growth of the population ceased, and mislocalization of glycosomal proteins to the cytosol was observed by immunofluorescence (data not shown). These results suggest that the lethal phenotype caused by the loss of glycosomal compartmentation mediated by knockdown of PEX14, at least in the case of glycerol, occurs downstream of PFK. DISCUSSION The studies reported here shed additional light on the role of glycosomal compartmentation in T. brucei and the danger of unregulated glycolysis. The first step in the metabolism of glucose is phosphorylation by HK. Most reactions involving glucose flow from this initial event (e.g. glycolysis, pentose phosphate pathway, nucleotide sugar biosynthesis). The rescue of the otherwise lethal PEX14 RNAi by co-reduction of HK via the double RNAi construct suggests an important role for HK in inducing the lethal phenotype. It is possible that the unregulated activity of HK causes the lethal phenotype in the PEX14 RNAi cell line through a depletion of cellular ATP and a concomitant accumulation of phosphorylated hexoses, as suggested by the computer models of Bakker et al. (15). It is also possible that a reduction in HK mRNA levels reduces flux through the glycolytic pathway, preventing the accumulation of a toxic intermediate further downstream in the pathway. Other workers have shown that the removal of glucose from the medium or the knockdown of HK affects the expression of surface glycoconjugates (19) and resistance to the adenosine analog tubercidin (26), suggesting that glucose metabolism has more far-reaching effects on the cell. In any case, the rescue from PEX14 knockdown via depletion of HK appears essentially identical to deletion of glucose from the growth medium.
The results for the PFK knockdown, although not entirely discordant with the suggestions of the computer modeling, are still a somewhat unexpected twist. Here it seems quite clear that a lethal phenotype results when glucose is present, even in the context of an intact glycosome. This phenotype could result because trypanosomal HK, unlike other HKs, is not inhibited by accumulating glucose 6-phosphate that would normally shut off glycolysis when PFK is depleted (see Fig. 1). One result would be the consumption of intraglycosomal ATP and increased glycosomal hexose phosphates. The specific block in glycolysis caused by 2-deoxyglucose is at glucose-6-phosphate isomerase. With respect to ATP consumption in glycolysis and accumulation of hexose phosphates, this inhibition is at the same place as the PFK knockdown. Taken together, these results suggest that sequestration of the pathway within the glycosome does not provide complete protection from the dangerous design of glycolysis. These results highlight the consequences of futile initiation of glycolysis in the presence of an intact glycosome, even in a stage when the parasite does not even require glycolytic metabolism (27). How this knockdown of PFK impacts flux in other pathways in trypanosomes remains to be studied, although it could increase flow through the pentose phosphate pathway as it does in Escherichia coli (28). The question remains as to whether the depletion of intraglycosomal ATP and accumulation of intraglycosomal hexose phosphates are sufficient to cause cell death or whether toxicity results from an interplay with other, possibly cytosolic, molecules. We hasten to point out that these results do not rule out the possibility that the effects of glucose toxicity result from a downstream process in the PEX14 RNAi mutant that has lost glycosomal compartmentation.
The observation that the PEX14 knockdowns are not rescued from glycerol toxicity by co-reduction of PFK supports the idea that at least one problem with the PEX14 RNAi mutant lies downstream of HK and PFK. The parasites possess a glycerol kinase that is normally sequestered within the glycosome (29) (Fig. 1). The forward reaction for this enzyme (glycerol ϩ ATP 3 glycerol 3-phosphate ϩ ADP) is kinetically preferred (but normally opposed in glycosomal metabolism) and has the possibility of consuming cellular ATP, just as proposed for the hexose kinases. It will be interesting to determine whether this enzyme is involved in the lethal effect of glycerol during PEX14 RNAi.
We have found that upon prolonged growth in SOGG medium, growth of parasites induced for PEX14 RNAi begins to slow compared with their uninduced controls. These data indicate that although glucose and glycerol may mediate immediate toxic effects when PEX14 is reduced in this cell line, it is likely that the lack of a functional glycosome has other deleterious consequences. Examples of other pathways in the glycosome that might be impacted by loss of compartmentation include ether-lipid synthesis or ␤-oxidation of fatty acids (1). Additionally, protection against cellular stresses may also be affected because a member of the glutathione-dependent peroxidase family (30) and certain superoxide dismutases (31) appear to be glycosomal. The functions of these pathways in procyclic form T. brucei have received little attention. We expect that in other trypanosomatids, such glycosomal pathways might prove to be more obviously important than in T. brucei. For example, knock-out of ether-lipid biosynthesis in Leishmania causes a specific defect in the establishment of intracellular parasitism (32). Using RNAi in T. brucei, we are now able to begin to experimentally address the importance of sequestering glycolysis within the glycosome. These insights allow us in turn to focus on a more general analysis of the role of compartmentation of other biological processes within this organelle.