Specialized Fatty Acid Synthesis in African Trypanosomes: Myristate for GPI Anchors

See allHide authors and affiliations

Science  07 Apr 2000:
Vol. 288, Issue 5463, pp. 140-143
DOI: 10.1126/science.288.5463.140


African trypanosomes, the cause of sleeping sickness, need massive amounts of myristate to remodel glycosyl phosphatidylinositol (GPI) anchors on their surface glycoproteins. However, it has been believed that the parasite is unable to synthesize any fatty acids, and myristate is not abundant in the hosts' bloodstreams. Thus, it has been unclear how trypanosomes meet their myristate requirement. Here we found that they could indeed synthesize fatty acids. The synthetic pathway was unique in that the major product, myristate, was preferentially incorporated into GPIs and not into other lipids. The antibiotic thiolactomycin inhibited myristate synthesis and killed the parasite, making this pathway a potential chemotherapeutic target.

Trypanosoma bruceicauses human sleeping sickness and livestock disease in Africa. The bloodstream form of this parasite covers its surface with ∼107 identical molecules of a glycosyl phosphatidylinositol (GPI)–anchored variant surface glycoprotein (VSG). The VSG GPI contains the fatty acid myristate (14:0) as its lipid moiety (1), and fatty acid remodeling reactions incorporate these myristates into the GPI precursor (2). Despite the large requirement for myristate, this fatty acid is not abundant in host bloodstreams (3), and the trypanosome was believed to be unable to either synthesize fatty acids de novo or to shorten longer fatty acids by β oxidation (4). This dilemma (5) provided a rationale for our investigation of fatty acid metabolism in trypanosomes. In studies on the cultured bloodstream form of T. brucei(6), we found that the elongation of laurate (12:0) to myristate was highly efficient, whereas the elongation of myristate to palmitate (16:0) was inefficient. We wondered why trypanosomes have a mechanism for the elongation of laurate, as there is little laurate in host bloodstreams (3) or in the parasite (7). We considered the possibility that trypanosomes can actually synthesize fatty acids de novo and that the elongation of laurate is a step in this pathway.

To test this hypothesis, we established a trypanosome cell-free fatty acid synthesis system (8). We found maximal activity in a membrane fraction. Fatty acid synthesis requires a primer [often acetyl–coenzyme A (CoA)], malonyl-CoA as a two-carbon unit donor, and reduced nicotinamide adenine dinucleotide phosphate (NADPH) as a reducing agent. We used [14C]malonyl-CoA as a radiolabel. As shown in an autoradiograph of a thin-layer chromatography (TLC) plate (Fig. 1A), we discovered synthesis of 14C-labeled fatty acids in bloodstream-form trypanosomes that depends on butyryl-CoA rather than acetyl-CoA as a primer.

Figure 1

Fatty acid synthesis in a cell-free system. (A) Bloodstream cell membranes were incubated with 71 μM [14C]malonyl-CoA ([14C] Mal-CoA), 200 μM acetyl-CoA (Ac-CoA) or butyryl-CoA (But-CoA), and 2 mM NADPH for 20 min. O, origin; F, solvent front; FA, fatty acids. The species marked with an arrow is probably β-ketohexanoic acid; other unmarked species were not characterized. (B) Time course of fatty acid synthesis. Lipid extracts were partitioned in chloroform/methanol/water (8:4:3), and radioactivity in the lower organic phase, containing fatty acids but little malonyl-CoA (as verified by TLC), was measured. Circles, bloodstream form; squares, procyclic form. (C) Lipids extracted at 30 min from (B) were analyzed for fatty acid chain length. Methyl ester derivatives were prepared and separated on reverse-phase-18 high-performance TLC (HPTLC) (Analtech) (21) with modifications to maximize extraction of medium chain fatty acids (22). Numbers at left indicate the number of carbon atoms as follows: 8, caprylate; 12, laurate; 14, myristate; 18, stearate.

We then compared an extract from bloodstream trypanosomes to one from procyclic forms. Procyclic forms, which reside in the insect vector, are known to synthesize fatty acids (9). Indeed, we found that the rate of [14C]malonyl-CoA incorporation in procyclic extracts was 5.3 times higher than that in bloodstream-form extracts (Fig. 1B). Evaluation of the chain length of the de novo–synthesized fatty acids indicated that myristate was the predominant product in bloodstream extracts, whereas procyclic extracts synthesized more hydrophobic species (Fig. 1C). We found that the major product in bloodstream trypanosomes was insensitive to catalytic hydrogenation or silylation (10), which suggests that this product was not an α,β-unsaturated or β-hydroxylated species that comigrated with myristate on our TLC system. All these results show that bloodstream trypanosomes can synthesize fatty acids and that the major product is myristate.

To examine whether de novo–synthesized myristate can serve as a substrate for GPI remodeling (Fig. 2A), we initiated GPI biosynthesis in the cell-free system by adding uridine 5′-diphospho-N-acetylglucosamine (UDP-GlcNAc) and guanosine 5′-diphosphate (GDP)–[3H]mannose (11). As expected, we found accumulation of glycolipid θ, the first intermediate of the remodeling pathway (at time 0, Fig. 2B) (2). When we added myristoyl-CoA to initiate remodeling, glycolipid θ was rapidly converted to glycolipid A" and then to θ′, A, and C (2) (Fig. 2B). When we replaced myristoyl-CoA with a combination of butyryl-CoA, malonyl-CoA, and NADPH, we observed a similar conversion of θ to myristoylated GPIs. Remodeling depended on the addition of all three substrates for fatty acid synthesis (6). Additionally, we found incorporation of radioactivity from [14C]malonyl-CoA into GPIs, and subsequent treatment with phosphatidylinositol-specific phospholipase C released radiolabeled diacylglycerol as expected (6).

Figure 2

Incorporation of de novo–synthesized myristate into GPIs. (A) Scheme for fatty acid remodeling. Open rectangles, myristate; vertical lines, fatty acid longer than myristate; horizontal lines, glycerol; diamonds, phosphate; circles, inositol; black rectangles, remainder of core glycan. Glycolipid A is the precursor transferred to VSG. Glycolipid C is inositol-acylated and is in equilibrium with A (23). (B) For GPI core glycan synthesis, membranes were incubated with 1 mM UDP-GlcNAc and 4 μCi/ml GDP-[3,4-3H]mannose (DuPont, 20 Ci/mmol) for 5 min, followed by a 5-min chase with 1 mM nonradioactive GDP-mannose. MnCl2, inhibitory to remodeling, was then removed by centrifugation. Myristoyl-CoA (50 μM) or a mixture of 50 μM butyryl-CoA, 50 μM malonyl-CoA, and 2 mM NADPH (concentrations are final) was then added to initiate remodeling. For extraction of GPIs, an n-butanol-water partition was conducted after chloroform/methanol/water (10:10:3) extraction. A portion of the TLC plate [Rf (migration relative to solvent front), 0.18 to 0.62] is shown.

To confirm our in vitro findings, we next determined whether trypanosomes can synthesize myristate in vivo and then use it for GPI remodeling. We found inefficient labeling of trypanosome lipids with [3H]acetate or [14C]butyrate, due apparently to poor uptake of the precursors into the cell (6). To circumvent this problem, we used the medium chain fatty acids [3H]caprylate (8:0) and [3H]laurate (12:0) as metabolic precursors for fatty acid synthesis, because these more hydrophobic species can efficiently diffuse through a lipid bilayer (12). Although trypanosomes incorporated less [3H]caprylate and [3H]laurate relative to [3H]myristate (possibly due to less efficient conversion to acyl-CoAs), the pattern of lipid labeling by these precursors resembled that of [3H]myristate (Fig. 3A) (13). A major fraction of radioactivity (49 to 79%) was incorporated into GPIs [that is, VSG (6) and glycolipids A and C (Fig. 3A)]. This preferential labeling of GPIs is remarkable, considering that free GPIs constitute substantially less than 1% of total cellular phospholipids (14). In contrast, [3H]palmitate (16:0) did not label GPIs but instead labeled conventional phospholipids (Fig. 3A). To verify that the [3H]caprylate or [3H]laurate was elongated to myristate before incorporation into GPIs, we analyzed the chain length of total radioactive fatty acid incorporated into the cell. Nearly all of the [3H]caprylate and [3H]laurate was converted to myristate, whereas elongation of myristate or palmitate was not detectable (Fig. 3B). Thus, the de novo–synthesized myristate is preferentially incorporated into GPIs.

Figure 3

Metabolic labeling with [3H]fatty acids in infected blood. (A) Lipid extracts loaded on each lane of an HPTLC plate (Merck) contained 5000 dpm and corresponded to 1.0 × 107 cells per lane for [3H]caprylate, 2.9 × 105cells per lane for [3H]laurate, 1.6 × 105 cells per lane for [3H]myristate, and 4.4 × 105 cells per lane for [3H]palmitate. PL, phospholipids. (B) Chain length analysis (as described in Fig. 1C) of the lipid extracts shown in (A). Standard (STD) lanes indicate authentic saturated [3H]fatty acid methyl esters with indicated chain length. Aberrant migrations of [3H]caprylate-labeled glycolipid A [(A), asterisk] and methyl myristate [(B), asterisk] are due to excessive cell equivalents loaded onto these lanes. Upon subsequent TLC purification, these species comigrated with authentic glycolipid A and methyl myristate (6).

We next examined the effect of the fatty acid synthetase inhibitors cerulenin and thiolactomycin (Fig. 4A). Cerulenin inhibits the condensation step of both eukaryotic and bacterial fatty acid synthetases (15), usually with a median inhibitory dose (IC50) less than 100 μM. At first glance, cerulenin seemed to have little effect on the cell-free synthesis of [14C]malonyl-CoA–labeled fatty acids at concentrations up to 1 mM (Fig. 4B). However, with increasing cerulenin concentration, trypanosomes accumulated [14C]caprate (10:0) rather than the normal end product myristate (Fig. 4C), which suggests a block in the final steps of myristate synthesis. Next we tested thiolactomycin, an antibiotic that inhibits the condensation step of the bacterial (but not eukaryotic) fatty acid synthetases (15). Thiolactomycin effectively inhibited trypanosomal fatty acid synthesis in vitro with an IC50 of ∼150 μM (Fig. 4B). Furthermore, thiolactomycin inhibited cell growth in culture in the same concentration range as that needed to inhibit fatty acid synthesis in vitro (Fig. 4D), which is consistent with the possibility that the in vivo target of thiolactomycin is indeed the fatty acid synthetase.

Figure 4

Inhibition of fatty acid synthetase by cerulenin and thiolactomycin. (A) Structures of the drugs. (B) Effect of cerulenin (squares) and thiolactomycin (circles) on in vitro fatty acid synthesis. The cell-free system was preincubated with cerulenin (Sigma) or with thiolactomycin for 10 min at 37°C before substrate addition and another 10 min of incubation. Incorporation into lipids was measured as in Fig. 1B. (C) Effect of cerulenin on the product chain length, measured as in Fig. 1C. A portion of the TLC plate (Rf, 0.30 to 0.84) is shown. (D) Effect of thiolactomycin on live cells. Bloodstream-form T. brucei 427 (24) in HMI-9 medium (25) were challenged in duplicate with 0 (white circles), 62.5 (black squares), 125 (black circles), 250 (white squares), or 375 μM (triangles) thiolactomycin.

All these results demonstrate that bloodstream trypanosomes express a fatty acid synthetase. The enzyme is atypical of eukaryotic enzymes in that it produces myristate rather than palmitate, is membrane-associated rather than cytosolic, and is sensitive to thiolactomycin. This pathway resolves the dilemma that trypanosomes require more myristate than is available for salvage from the host bloodstream (5). Using the rate of myristate synthesis in the cell-free system (Fig. 1B), we estimate that de novo synthesis can account for at least 25% of the total requirement for myristate, with the rate in vivo potentially being much higher. Preferential incorporation of newly synthesized myristate into GPIs (Fig. 3A) is especially striking because syntheses of bulk phospholipids and GPIs both occur in the endoplasmic reticulum (ER). In the related organismLeishmania mexicana, GPI biosynthesis appears to occur in a subcompartment of the ER (16). The preferential use of de novo–synthesized myristate in GPI remodeling may imply the presence of a similar ER subcompartment in trypanosomes that contains the pathways for myristate synthesis and GPI myristoylation, but not that for conventional phospholipid synthesis.

Antitrypanosomal drugs are desperately needed. Because GPI myristoylation does not occur in mammals, this pathway could be a chemotherapeutic target. In fact, some myristate analogs are selectively toxic to trypanosomes (17). We now find that thiolactomycin may affect GPI remodeling by inhibiting the synthesis of myristate. Thiolactomycin has low toxicity in mice (18) and protects them against some bacterial infections (19), making this compound a promising lead for antitrypanosomal drug development. Finally, a recent finding that thiolactomycin inhibits growth of the malaria parasite (20) raises the possibility that fatty acid synthesis can be a versatile drug target in diverse parasitic infections.

  • * To whom correspondence should be addressed. E-mail: penglund{at}


View Abstract

Navigate This Article