Comparative lipidomic profiling of African trypanosomes reveals metabolic distinctions impacting pan-species drug development

Abstract

African trypanosomes are a group of pathogens that cause disease in humans (human African trypanosomiasis; HAT) and animals (African animal trypanosomosis; AAT) across sub-Saharan Africa, leading to profound impacts on human health as well as economic instability. Achieving sustainable disease control requires a One Health approach, yet the development of urgently required pan-species chemotherapeutics is severely hindered by a lack of unified biological understanding of the causative species. A major bottleneck in drug discovery is the distinct in vitro requirements of the clinically relevant species. Whilst Trypanosoma brucei is typically cultured in FBS-supplemented culture, T. congolense, the primary cause of AAT, requires the use of goat serum. The differing serum supplementation requirements of these two trypanosome species point to metabolic distinctions, which may be important considerations in developing experimental systems to enable the identification and design of novel, pan-species therapies. In this study, untargeted LC-MS lipidomics analyses were conducted to determine the relative lipidomic profiles of T. brucei and T. congolense bloodstream form parasites. The use of a novel medium formulation that permits effective in vitro culture of both species revealed that their lipidomic profiles are distinct. Notably, ether phospholipids, key molecules in T. brucei lipid biology, are relatively absent in T. congolense. Furthermore, the T. brucei lipidome is enriched with shorter chain saturated lipids, whilst T. congolense exhibits a preference for longer-chain polyunsaturated fatty acids (PUFAs). These observations indicate that there are significant differences in the ways these parasites synthesise and remodel their lipid complement, highlighting an evolutionary divergence between the species that likely carries implications for host-pathogen interactions as well as trypanosome membrane biology. Crucially, defining the species-specific lipid dependencies provides the foundation for pan-species culture systems, thereby removing a critical barrier to the discovery and design of novel therapeutics effective against all African trypanosome species.

Author summary

African trypanosomes are a group of neglected parasites that can infect both humans and animals. The resulting diseases cause severe health impacts and billion-dollar economic losses across sub-Saharan Africa, where the diseases are prevalent. Controlling these diseases requires a One Health approach whereby both the human and animal diseases are targeted simultaneously. However, the development of new drugs is hindered by a lack of understanding of the biological commonalities and differences between the main disease-causing trypanosome species. Furthermore, the human-infective (Trypanosoma brucei) and primary animal-infective (Trypanosoma congolense) species require completely different conditions to survive in the laboratory. In this work, we aimed to determine whether differences in the ways these species use lipids explains their differing laboratory culture requirements. Using a newly designed media formulation, it was possible to culture mammalian-infective forms of both parasite species under identical conditions, enabling direct comparison of their lipidome - a complete inventory of the different fats and lipids the cells contain. We demonstrate that the T. brucei lipidome significantly differs from that of T. congolense, and that T. congolense shows a preference for longer, more unsaturated lipids. These differences are likely to underlie species-specific differences observed during host infections, and likely explains why T. brucei exhibits differential susceptibility to certain drugs and metabolic inhibitors compared to T. congolense. Furthermore, our work demonstrates that understanding the lipid biology of protozoan parasites aids in optimisation of laboratory culturing conditions, thereby facilitating further research into these understudied pathogens, including the development of new therapies aimed at all disease-causing trypanosome species.

Introduction

African trypanosomes are a group of parasitic hemoflagellates that pose a significant risk to the health of humans (Human African Trypanosomiasis; HAT) and animals (African Animal Trypanosomosis; AAT), representing a challenge that requires a comprehensive One Health approach. HAT is prevalent across sub-Saharan Africa, and caused by subspecies of Trypanosoma brucei (T. b. rhodesiense and T. b. gambiense). With case numbers decreasing to <1,000 annually, the disease is currently targeted for elimination of transmission by WHO [1]. In stark contrast, AAT remains a major hindrance to the development of sustainable agriculture across sub-Saharan Africa. Unlike HAT, AAT is primarily caused by three species of African trypanosome; Trypanosoma congolense, T. vivax and T. brucei [2]. It is estimated AAT causes in excess of three million cattle deaths per year, with up to 90 million at risk of disease [3]. New treatment options to treat the disease are sorely needed. Historically, AAT has been largely neglected by drug development endeavours, although this has changed recently with the ongoing development of human and veterinary benzoxaborole compounds as potential novel trypanocides [4,5]. Consequently, no new drugs for AAT have been introduced in over 50 years, leading to increasing reports of parasite drug resistance to the two main chemotherapeutics (isometamidium chloride and diminazene aceturate) [3].

Establishing control of AAT infection is critical to protect economic interests (estimated economic losses to AAT amount to US$ 4.5 billion [6]) and to remove a significant barrier to the development of sustainable agriculture in the affected areas. Furthermore, joint management of AAT animal reservoirs is essential for eradicating human cases. There is growing interest in pan-species therapies as such strategies could streamline control across different hosts [7]. However, this requires detailed knowledge of all trypanosomiasis-causing species.

An impediment to developing drug treatments for AAT has been the relative difficulty of studying the different causative parasite species in a laboratory setting. Although T. brucei has been widely studied, in part due to its role in causing HAT, it was only recently that mammalian-infective bloodstream form (BSF) T. congolense became more widely studied under in vitro conditions [8–10], and there is still no established method for the in vitro culture of BSF T. vivax. Furthermore, tools for the genetic manipulation of T. congolense have only recently become available, enabling the functional study of this species on a molecular level [8].

Whilst in vitro cultivation of T. brucei has been a laboratory mainstay for many decades [11], culture of T. congolense has proved far more challenging, with a key difference being a requirement for fresh goat serum (GS) supplementation of culture medium, in contrast to the more commonly used and inexpensive foetal bovine serum (FBS) for T. brucei, the latter in common with most other eukaryotic in vitro culture models [9,12]. The use of GS as serum supplement significantly hinders the ability to carry out direct comparative analyses of T. congolense and T. brucei, including their co-culture. It also raises a key question - how do these sera differ biochemically, and what are the key components T. congolense requires from GS that FBS cannot provide?

In previous work, we generated a comprehensive overview of the T. congolense metabolome [10], including a comparative analysis with T. brucei. Of particular note were differential responses to inhibition of fatty acid (FA) synthesis, with T. congolense exhibiting significantly reduced sensitivity to an inhibitor of acetyl-CoA synthetase and Orlistat, a known inhibitor of, amongst other targets, FA synthase [10,13]. These data suggest profound differences in FA biosynthesis, and likely, composition, between T. brucei and T. congolense. However, these studies were still carried out with the caveat that both species were cultured in different sera, the sole source of lipids under in vitro conditions.

In this study, we sought to analyse the metabolic signatures of the sera used to culture trypanosomes, and the BSF-stage parasites themselves, in order to generate a comparative analysis of lipid metabolism in T. congolense and T. brucei BSFs. Employing a new media formulation that supports the proliferation of both T. brucei and T. congolense BSFs, it was possible to establish that the lipidome of T. congolense is distinct from that of T. brucei, even when both species are subjected to the same in vitro media conditions. This points to a fundamental divergence in their lipid metabolism, which could have key implications for drug development and host-pathogen interactions. Targeted GC-MS analysis indicates that this divergence extends to the fatty acid content. Overall, results indicate it may be possible to fine-tune fatty acid provision to permit further culturing and/or co-culturing options for these trypanosome species, giving us greater insight into evolutionary divergences in lipid metabolism in these closely related pathogen species.

Results

Trypanosome proliferation in vitro is heavily influenced by serum source

Previous studies have emphasised serum source as an important factor for the successful cultivation of animal trypanosomes in vitro [9,11]. Notably, in vitro medium formulations described for T. congolense contain a goat serum (GS) supplement (20%), in contrast to T. brucei, which is typically supplemented with 10% FBS [11]. To date, no biological reason for this discrepancy has been validated. To characterise the viability of T. brucei and T. congolense in differing serum supplements, growth curves were carried out in HMI-11 (T. brucei) or HMI-93 (T. congolense) supplemented with 10% and 20% v/v of different animal sera (Fig 1).

Fig 1. Effect of serum supplement on trypanosome in vitro viability.

Cultures of T. brucei (A and B) and T. congolense (C and D) were seeded and cell density counted by haemocytometer every 24 hr in the presence of 20% (A and C) or 10% (B and D) serum. A total of five sera were tested: goat serum (GS), foetal bovine serum (FBS), adult bovine serum (ABS), horse serum (HS) and calf serum (CS). Cells were passaged every 48 hr, and cumulative cell density over a period of 6 days is shown. Doubling times were calculated and statistically compared (BH-corrected t-test) to the relevant control (10% FBS and 20% GS for T. brucei and T. congolense, respectively), and are presented in S1 Fig.

As expected, T. brucei grows best in FBS, supplemented at either 20% or 10% (Fig 1A and 1B; doubling times shown in S1 Fig). Cells remained viable in goat serum and horse serum, albeit with a reduced doubling time (S1 Fig). Adult bovine serum and calf serum did not support T. brucei culture, with no cell division observed after 2–3 days of culture (doubling times could therefore not be calculated for these samples).

In contrast, T. congolense was, in general, viable in all sera tested (Fig 1C and 1D; doubling times shown in S1 Fig). However, long-term culture in FBS (up to 20 passages) was not successful. It was also determined that reduction of any serum to 10% (instead of 20%) led to a reduction in T. congolense doubling time (S1 Fig).

The function of cell culture serum is to provide additional (undefined) nutrients vital to supporting the proliferation of cells in vitro. Serum contains sugars, such as glucose, amino acids and micronutrients such as inorganic ions and vitamins, which are supplied to considerable excess in basal media formulations. This includes IMDM, which is a main component of HMI-11 and HMI-93 [9,11]. Metabolomics analysis to assess potential differences in the small, polar molecule complement of FBS and GS revealed many differences in polar metabolites (Fig 2A). Of the 482 metabolites detected, 285 were significantly different (BH-corrected t-test; S1 Table), of which the top 50 are shown in Fig 2A. Biologically relevant metabolites are shown in Table 1. These metabolites were mainly annotated as amino acids, such as L-aspartate (Log₂FC, GS/FBS: -2.81), L-glutamate (Log₂FC: -1.99) and L-arginine (Log₂FC: 2.12), carbohydrates including glucose (Log₂FC: -3.56) and D-glucosamine (Log₂FC: -3.84), and nucleotides such as uracil (Log₂FC: -3.77) and inosine (Log₂FC: 5.17). However, as mentioned above, many of these metabolites are included in commercially formulated culture media, and are therefore unlikely to influence cell viability. Lipids are often poorly detected on LC-MS metabolomics platforms using HILIC chromatography. However, peaks putatively annotated as phosphatidylethanolamines (PEs), and in addition several phosphatidylcholines (PCs) and phosphatidylinositols (PIs), were detected via LC-MS, and determined to be present only in goat serum (S1 Table). Furthermore, L-carnitine and O-acetylcarnitine were shown to be significantly lower in FBS relative to goat serum (Table 1). Altogether, many of the putatively identified metabolites significantly altered between GS and FBS appeared to be lipid-like in nature (S1 Table). This indicated that lipids may be a key influencer of cell culture viability, as this metabolite class is only provided to cells in serum supplements.

Fig 2. Adaptation of T. brucei and T. congolense to a universal medium.

A) LC-MS metabolomics analysis of GS (lilac) and FBS (green) was carried out and fold changes as well as t-tests calculated for each detected metabolite. The 50 most significant differences are shown, highlighting widespread differences in polar metabolites between the two sera. B) Cell viability of T. congolense in FBS-supplemented culture with the addition of 2.5% (AMII). Regardless of percentage (v/v) serum, AMII drastically improved cell proliferation, even compared to the goat serum control. C) Addition of AMII does not impact T. brucei proliferation compared to the FBS controls. Taken together, these results led to a new medium formulation – HMI-AMII supplemented with 10% FBS and 2.5% AMII. D) Drug sensitivity analysis of T. congolense (Tc; orange: HMI-93; gold: HMI-AMII) and T. brucei (Tb; dark blue: HMI-11; light-blue: HMI-AMII) to an acetyl-CoA synthetase inhibitor (ACS-In), showing that in both species-specific medium, as well as HMI-AMII, T. congolense is less sensitive to acetyl-CoA synthetase inhibition. E) Drug sensitivity analysis of T. brucei and T. congolense (same colour scheme as D) to Orlistat, a lipophilic inhibitor of lipid synthesis. Similar to ACS-In, T. congolense is less sensitive to this compound in both species-specific medium and HMI-AMII, compared to T. brucei.

Table 1. Biologically relevant metabolites with significant variation in abundance between FBS and GS. Log₂ fold-change (FC) analysis (GS/FBS) is summarised alongside their associated P-values (t-test) and false discovery rate (FDR)-adjusted P-values. Positive values indicate higher abundance in GS, negative values indicate higher abundance in FBS. The full dataset is available in the supplementary results (S1 Table).

Metabolite	Log2FC	P-value	FDR
Cholate	19.78	5.49E-07	3.83E-06
Inosine	5.17	4.77E-06	2.24E-05
3-(4-Hydroxyphenyl)pyruvate	4.06	3.88E-02	6.26E-02
L-Carnitine	2.26	8.07E-07	5.19E-06
O-Acetylcarnitine	2.14	1.99E-06	1.08E-05
L-Arginine	2.12	8.33E-06	3.40E-05
Pyruvate	1.43	1.56E-03	3.57E-03
Imidazole-4-acetate	1.42	2.42E-05	8.53E-05
L-Citrulline	1.28	9.56E-04	2.30E-03
L-Asparagine	1.24	2.37E-03	5.20E-03
L-Phenylalanine	-1.05	7.07E-06	2.99E-05
Hypoxanthine	-1.07	1.10E-04	3.26E-04
Citrate	-1.09	5.32E-05	1.69E-04
L-Glutamine	-1.09	3.21E-05	1.10E-04
Adenine	-1.33	8.20E-04	2.00E-03
Deoxyuridine	-1.69	1.48E-06	8.60E-06
(R)-Lactate	-1.90	1.14E-09	7.24E-08
D-Glucose	-3.56	1.02E-07	1.22E-06
Uracil	-3.77	5.17E-10	4.99E-08
Succinate	-4.95	5.94E-08	8.68E-07

To probe this further, viability of both species was assessed when grown in medium supplemented with FBS and a commercially available lipid-loaded BSA supplement (albuMAX II, AMII; 2.5% v/v; Fig 2B and 2C). To ensure consistency, a single HMI-11-based medium – termed HMI-AMII – was used for both species (note that standard HMI-11 and HMI-93 differ only in serum requirements and in the concentrations of L-cysteine, hypoxanthine and 2-mercaptoethanol; S2 Table). Control cultures were maintained in their standard species-specific media (HMI-93 for T. congolense, Fig 2B; HMI-11 for T. brucei, Fig 2C). Notably, it was not possible to compare growth in AMII to BSA-only controls, as this supplement is pre-coupled in a proprietary manner, and BSA is known to vary depending on batch and origin.

In HMI-AMII supplemented with 20% and 10% FBS, T. congolense grew at rates comparable to that of the goat serum control (mean doubling time of 10.03 hr, 9.5 hr and 9.1 hr for 20% FBS, 10% FBS and 20% GS, respectively), suggesting the addition of the BSA supplement was key to proliferation of this species in FBS-supplemented medium in vitro (Fig 2B). Cell viability of T. brucei was not impacted by lipid supplementation, indicating that HMI-AMII medium supplemented with 10% FBS and 2.5% AMII (HMI-AMII) was sufficient for the culture of both trypanosome species (Fig 2C).

To determine whether the altered medium conditions impacted T. brucei and T. congolense lipid metabolism, parasite sensitivity to a metabolic inhibitor of acetyl-CoA synthetase (ACS-In) and Orlistat (inhibitor of FA synthase) were carried out in HMI-93 (T. congolense), HMI-11 (T. brucei) and HMI-AMII (both species). Consistent with previous findings [10], T. congolense remained less sensitive to both inhibitors in HMI-AMII (Fig 2D and 2E). However, comparisons between HMI-AMII and species-specific culture medium (HMI-93 and HMI-11 for T. congolense and T. brucei, respectively) revealed clear medium-dependent shifts with regards to inhibitor sensitivity. T. brucei exhibited increased resistance to ACS-In when transferred from HMI-11 (EC₅₀: 10.90 ± 0.24 µM) to HMI-AMII (EC₅₀: 19.37 ± 0.63 µM; Fig 2D), indicating reduced sensitivity under HMI-AMII conditions. This is consistent with the additional exogenous lipids supplied by AMII that reduce parasite reliance on de novo FA synthesis and elongation. Reduced sensitivity was also observed with respect to Orlistat (EC₅₀’s: 38.39 ± 1.57 nM and 68.13 ± 7.6 nM in HMI-11 and HMI-AMII, respectively; Fig 2E), likely due to the aforementioned reasons. In contrast, T. congolense exhibited increased sensitivity to both ACS-In (EC₅₀’s: 82.07 ± 0.85 µM and 58.42 ± 0.21 µM in HMI-93 and HMI-AMII, respectively; Fig 2D) and Orlistat (24.03 ± 0.68 µM and 0.86 ± 0.06 µM in HMI-93 and HMI-AMII, respectively; Fig 2E) when cultured in HMI-AMII. This suggests that whilst HMI-AMII supports cell proliferation, it does not fully reproduce the lipid environment provided by 20% GS, because exogenous lipids differ in abundance or composition. The parallel shift in ACS-In sensitivity supports the view that this is not solely an Orlistat-specific effect, although altered free drug bioavailability is also likely to contribute, particularly for the highly lipophilic Orlistat [14]. Together, these data indicate that susceptibility to lipid-metabolism inhibitors is strongly shaped by the extracellular nutrient environment. They also suggest that, under HMI-AMII conditions, T. congolense retains some dependence on endogenous lipid metabolism despite remaining less sensitive than T. brucei overall.

Comparative analysis of the T. brucei and T. congolense lipidomes

Serum supplements are, typically, the single source of exogenous lipid precursors for in vitro culture systems. Therefore, the apparent requirement for exogenous lipids in T. congolense may indicate differences in how this species obtains and utilises the host’s lipids it requires for cell survival compared to T. brucei. This suggests that T. congolense exhibits a greater reliance on lipid scavenging compared to T. brucei.

Interrogation of the T. brucei and T. congolense genomes suggests that T. congolense shares the majority of enzymes linked to lipid metabolism, including all four fatty acid elongases (ELO1–4; S3 Table), as previously described [10]. However, no orthologue was found in either of two independent T. congolense genome assemblies (shotgun or PacBio, released in 2016 and 2019, respectively) for the two T. brucei genes encoding putative fatty acid desaturases, Tb927.10.7100 and Tb11.v5.0580. The protein encoded by Tb11.v5.0580 was recently found to desaturate long chain FAs (20:4 and 22:4) resulting in the generation of docosapentaenoic acid (22:4) and docosahexaenoic acid (22:6), and has been hypothesised to play a role in BSF differentiation to stumpy-form [15]. However, it is difficult to speculate on precise lipid metabolic distinctions using genomic information alone, particularly as many of these predicted enzymes remain to be experimentally validated in either trypanosome species.

In order to compare differences in lipid composition further, untargeted LC-MS lipidomics analysis was performed on T. brucei (cultured in HMI-AMII) and T. congolense BSFs (cultured in HMI-93 with 20% GS as well as HMI-AMII), in order to directly compare parasite lipidomes when grown under identical conditions (with one caveat that T. brucei is grown at 37°C, and T. congolense at 34°C). Even under these circumstances, T. brucei and T. congolense bloodstream forms displayed distinct lipidomic profiles (Fig 3).

Fig 3. Overview of comparative LC-MS lipidomics analysis of T. brucei and T. congolense.

A) PCA plot of three sample groups: TbFBS, TcoGS and TcoFBS. The first and second principal components are plotted with 95% confidence regions shown, highlighting separation of species (PC1) and culture conditions (PC2). Data were Log₂ transformed and auto-scaled prior to PCA loading. B) Volcano plot showing significantly altered (p < 0.05 & Log₂FC > 2 [Red], < -2 [Blue]) lipid species when comparing T. brucei culture in FBS and T. congolense cultured in GS. A total of 46 lipids were significantly altered under these criteria. C) Volcano plot highlighting the 42 lipid species that were significantly altered when comparing T. brucei and T. congolense grown under the same in vitro conditions (10% FBS, 2.5% AMII). D) Volcano plot showing the 22 lipid species that were significantly altered when comparing T. congolense cultured in GS and FBS, 2.5% albuMAX II.

LC-MS lipidomics analysis resulted in the identification of 209 lipids across the three sample groups (Fig 3; S4 Table). PCA analysis showed clear separation of the three sample groups, with PC1 (explaining 63.5% of variation) separating primarily by species, whilst PC2 (explaining 31.1% of variation) separated primarily by serum content (Fig 3A). Statistical analyses (BH-corrected t-tests) were applied to three comparisons - TbFBS vs TcoFBS, TbFBS vs TcoGS and TcoFBS vs TcoGS - highlighting a total of 46, 42 and 22 significantly altered (Log₂ fold change [FC] >2 & p < 0.05; Fig 3B-3D) lipid species in the respective comparisons. Notably, there were no clear patterns in the 22 lipid species that were significantly altered between the two T. congolense sample groups, suggesting that the switch from HMI-93 to HMI-AMII did not lead to widespread lipidomic changes (Fig 3D).

Given the data was normalised to cell number, within-feature comparisons and patterns across groups were analysed in more detail (rather than absolute quantification), with a particular focus on lipids significantly altered in the aforementioned two-way analyses. The most abundant group of lipids present across the sample groups were glycerophospholipids (phosphatidylcholines, PCs; phosphatidylethanolamines, PEs; lyso-phosphatidylcholines, LPCs; phosphatidylinositols, PIs; 61.7% of identified lipids, S4 Table).

PCs, in particular short chain PCs, were consistently more abundant in T. brucei (S4 Table), compared to T. congolense cultured in both HMI-AMII and HMI-93 (Fig 4A). In contrast, for all significantly altered PCs recovered, none exhibited differences between the two T. congolense sample groups. Notably, longer chain PCs with higher levels of unsaturation (e.g., PC 36:6, 40:8 and 40:9) were more abundant in T. congolense, with a switch to FBS leading to reduced levels of PC 40:9 in particular (Fig 4A). These data indicate that T. congolense exhibits a preference for longer acyl chain PCs.

Fig 4. PC and PE profiles in T. brucei and T. congolense.

Forest plots were generated for each lipid classes to show fold changes of lipids that were significantly altered in one of the two-way comparisons. A) Nine PCs were significantly altered, with T. brucei showing elevated abundance of 6 PCs compared to T. congolense in both conditions. PCs did not vary between the two T. congolense sample groups, regardless of serum supplement. B) There were also 9 PEs significantly altered between the three sample groups. In this case, shorter PEs were generally more abundant in cells grown in FBs, regardless of species. C) A total of 11 PC-Os were significantly altered, in each case the significant change in abundance was a species-specific difference. Shorter chain PC-Os were generally more abundant in T. brucei, whilst longer chain PC-Os were more abundant in T. congolense. D) Nine PE-O lipids were significantly altered in this data. Similar to PC-Os, the PE-Os showed species-specific differences in abundance, with very little variation between the two T. congolense sample groups. Where multiple isomers were detected, these are indicated with an underscore and letter. E) Stacked bar plot showing each carbon chain length as a percentage of the total PC pool, highlighting a reduction in 36-carbon chain lipids in T. congolense. F) A stacked bar plot showing carbon chain length abundances for PEs, this time highlighting a serum-dependent change, with T. congolense cultured in GS exhibiting almost exclusively 42-carbon length lipids. G) A stacked bar plot for PC-O, highlighting a species-specific difference, with T. congolense exhibiting a substantial increase in 38-carbon chain length PC-O lipids. H) PE-O lipids exhibited a very similar species-specific different to PC-Os, with T. congolense exhibiting large amounts of 38-carbon lipids in this pool.

In general, abundances of significantly altered PEs were higher in T. brucei, regardless of chain length (Fig 4B). Notably, T. congolense cultured in GS-supplemented media exhibited lower abundance of almost all PEs identified, compared to both FBS sample groups. This was with the exception of long chain unsaturated PEs, in particular PE 42:6 and 42:8 (Fig 4B), mirroring the PC lipid class that suggested a preference for longer acyl chain lipids for T. congolense.

The most prominent species-related difference in these data was found in ether-linked glycerophospholipid abundance (PC-O and PE-O; Fig 4C and 4D, respectively). For both lipid classes, T. congolense exhibited significantly reduced abundance compared to T. brucei (S2A and S2B Fig), regardless of culturing conditions. A pattern was observed here, whereby T. brucei exhibited significantly increased abundance of shorter chain PE-Os and PC-Os (e.g., PE-O 36:x and PC-O 36:x; Fig 4C and 4D, respectively), whilst both T. congolense samples showed elevated levels of longer-chain lipids in both classes, in particular PC-O 38:2, 40:x and 42:5 (Fig 4C; shown on a logarithmic scale in S2C Fig) and PE-O 40:3 (Fig 4D). PC-O and PE-O lipid abundances were only significantly altered in inter-species comparisons, with the two T. congolense samples exhibiting very similar profiles for ether-linked glycerophospholipids (Fig 4C and 4D). Notably, T. brucei exhibited elevated abundance of a PC-O 38:3 putatively annotated as PE(O-20:0/18:3) (38:3_b), whilst an isomer annotated as PE(O-18:0/20:3) (38:3_a) was elevated in T. congolense (Fig 4D), indicating differences in lipid remodelling between the two species. Finally, T. congolense cultured in GS exhibited a high abundance of several LPCs (S2A Fig), when compared to either species cultured in FBS. In particular, ether-linked LPCs (20:0, 20:1 and 20:2) were significantly elevated in abundance (S2D Fig). Notably, the total LPC content in each sample group remained below 0.4% of the total PC pool (S4 Table), confirming that elevated abundances observed in T. congolense GS samples reflects a true biological difference instead of an artifact resulting from lipase activity during lipid extractions.

To further determine whether T. congolense does indeed exhibit a preference for longer chain lipids compared to T. brucei, abundances of each carbon chain length were summed (regardless of chain saturation) for each lipid class and graphed as a percentage of the total abundance of that lipid class (Fig 4E, 4F, 4G and 4H). Whilst the PC pool in T. brucei was > 40% 36-chain lengths, this was reduced in T. congolense under both GS and FBS-supplemented conditions, with T. congolense instead exhibiting increased abundance of 38 and 40-carbon chain lipids (Fig 4E). The PE pools of T. congolense and T. brucei were strikingly similar under FBS-supplemented conditions, dominated by 42-carbon chain lipids (~60%). In contrast, when cultured with GS-supplementation, the T. congolense PE pool was almost exclusively 42-carbon chain length (Fig 4F). For both PC-O (Fig 4G) and PE-O (Fig 4H), a strong species-specific difference was observed, with T. congolense exhibiting a high abundance of longer chain lipids (in particular 38-carbon chains), regardless of culturing conditions.

There were fewer clear species- or condition-specific patterns in other lipid classes, in particular sphingolipids (sphingomyelins, SM) and PIs. For lipids that were significantly altered in two-way comparisons, these were generally higher in abundance in T. congolense, especially when parasites were cultured in the same medium (Fig 5A). Similar to above, longer chain SMs (42:x and 44:2) were highly abundant in T. congolense (Fig 5A), whereas the T. brucei SM pool was almost exclusively made up of 40-carbon chain SMs (Fig 5B).

Fig 5. PI and SM profiles, and lipid uptake ratios in T. brucei and T. congolense.

A) Forest plot to show abundance changes in 12 SMs and 2 PIs that were significantly altered in at least one of the two-way comparisons. In general, these data highlighted a reduction of longer-chain SMs in T. brucei compared to T. congolense. B) Stacked bar plot showing each carbon chain length as a percentage of the total in the SM pool. Unlike glycerophospholipids, differences in the SM pool were not as drastic, although T. congolense exhibited a small increase in 32- and 36-carbon chain SM lipids, compared to T. brucei. C) Stacked bar plot showing overall chain length composition of the PI pools in T. brucei and T. congolense. Both species showed highest abundance of 40-carbon chain length lipids in this class, with no significant variation between the three sample groups tested. D) Global XY scatter plot comparing the relative uptake ratios of individual lipid species in T. brucei (TbFBS vs FBS) and T. congolense (TcoFBS vs FBS). The solid diagonal line represents the line of identity (x = y), indicating equivalent relative lipid accumulation between the two species. While shared lipid pools cluster along this line of identity, divergence is evident in several lipids. E) Targeted relative enrichment factor for 18 representative lipids. A base line of 1.0 (indicated by the vertical dashed line) indicates likely passive equilibration with the host serum, as observed universally for neutral storage triglycerides (TGs). Values deviating from 1.0 highlight species-specific accumulation. For example, T. brucei exhibits selective accumulation of PE-O 38:3_b and PC-O 36:x. In contrast, T. congolense exhibits selective accumulation of longer chain lipids such as PC-O 40:2 and PE-O 40:3. The full dataset is presented in S5 Table.

The remaining lipid classes identified were neutral lipids, cardiolipins and cholesterol esters (S2B Fig). Two cardiolipins were significantly altered in a species-specific manner, with CL 76:12 being highly abundant in T. brucei, and in contrast, CL 76:8 being highly abundant in T. congolense under both GS and FBS-supplemented conditions (S2B Fig).

It was recently shown that T. congolense is highly resistant to inhibitors of fatty acid synthesis, leading to the hypothesis that this species preferentially scavenges lipids from the exogenous environment. To evaluate this hypothesis, the sera used to culture T. brucei and T. congolense were also analysed by LC-MS lipidomics (S4 Table). Spearman rank correlations were subsequently performed to compare the parasite lipidomes with their respective sera lipidomes (S3 Fig, S5 Table). However, correlations of different lipid classes between the parasites and sera were only moderate to weak (with the exception of storage lipids - MG, DG and TG). This lack of global correlation indicated that neither T. brucei nor T. congolense scavenge lipids indiscriminately. Therefore, to quantify whether host lipids are selectively scavenged, individual relative uptake ratios (lipid abundance within the parasite relative to the serum) were calculated for each lipid species detected in both the parasites and the matched serum samples, with a focus on FBS (Fig 5D and 5E; S5 Table).

When plotted together, the majority of lipids showed similar relative uptake ratios between T. brucei and T. congolense, mapping closely to the line of identity (x = y; Fig 5D). This is highlighted by the aforementioned neutral storage lipids, in particular TGs, with uptake ratios for both species of approximately 1.0 (Fig 5E). However, there were several lipids that deviated significantly from the identity line, highlighting species-specific differences in lipid uptake. The majority of lipids exhibiting this divergence were glycerophospholipids (Fig 5E).

To further characterise species-specific lipidomic differences, a previously published comparative transcriptomics dataset [10] was analysed with a focus on lipid enzymes (S4A and S4B Fig). Notably, these data were generated in HMI-11 and HMI-93 (supplemented with 20% GS) for T. brucei and T. congolense, respectively [10]. Firstly, RNAseq data was mapped to glycerophospholipid metabolism (S2C Fig), where a bottleneck was observed at ether-lipid initiation. Transcripts corresponding to dihydroxyacetone phosphate acyltransferase (DAT) were not detected in T. congolense samples (S3 Table), and furthermore, transcripts corresponding to alkyl‑DHAP synthase (ADP) were reduced in this species, compared to T. brucei. This bottleneck could explain the marked reduction in ether-glycerophospholipid abundance in T. congolense, although the reasons for lack of activity in the pathway are currently unknown.

Within glycerophospholipid headgroup assembly pathways, there was a reduction in transcript abundance corresponding to PE synthesis in T. congolense (ECT & EPT; S2C Fig). In contrast, transcripts corresponding to PC synthesis (CCT & CPT; S2C Fig) exhibited upregulation in this species, compared to T. brucei, supporting the lipidomic differences observed, where PE species were generally lower in abundance in both T. congolense sample groups (Fig 4). Of further note, transcripts associated with PLA1 were higher in T. brucei (S2C Fig), supporting higher LPC production in this species. Increased levels of LPC 22:5 and 22:6 could also be explained by exogenous lipid bioavailability [16,17].

Sphingolipid and long-chain FA metabolism were subsequently analysed (S2D Fig). In general, transcripts corresponding to these pathways were most abundant in T. brucei, with key exceptions being ELO4 and ACS5, suggesting species-specific channelling of acyl-CoA into sphingolipid and glycerophospholipid synthesis. These transcriptomic differences could contribute to the distinct chain-length distributions seen in the corresponding lipidomes. Taken together, species-specific differences in previously published transcriptomics data support the species-specific differences observed in lipidomics analyses carried out in this study.

Comparative analysis of T. brucei and T. congolense GC-MS fatty acid profiles

Overall, the above results point towards differences in the fatty acyl composition of T. brucei lipids relative to T. congolense. To examine this further, targeted GC-MS fatty acid analysis was conducted for the three sample groups explored above: T. brucei in HMI-AMII, T. congolense in HMI-93 and T. congolense in HMI-AMII (Table 2; raw data in S6 Table). It is known that T. brucei BSFs contain high levels of C18:0, C18:1 and C18:2 fatty acids, which is also evident from the current analysis. Additionally, T. brucei also display substantial enrichment of C22:4, C22:5 and C22:6 PUFAs relative to their hosts (or cell culture serum) [16], indicating the parasites either have a mechanism for selectively acquiring these fatty acids from their environment or input significant resources to de novo biosynthesise them. Although BSF T. congolense are shown to contain similar fatty acid species to T. brucei, their relative abundance differs – the ratio of fatty acids C22:4 and C22:6 increases relative to levels of C18:2, the most abundant fatty acid in both trypanosome species (Table 2). There is a drop in relative cellular levels of C22:6 when T. congolense parasites are moved from HMI-93 to HMI-AMII, suggesting supplies of this fatty acid are supported by exogenous sources, and that abundance of this PUFA is reduced in the latter supplement. The most significant species-specific differences were observed for C16:0, C18:0, C18:3, C20:3 and C22:5 (Table 2). Finally, T. congolense exhibited increased abundance of C16:0 compared to T. brucei, irrespective of culturing conditions.

Table 2. Comparative GC-MS fatty acid analysis of T. brucei and T. congolense. The culture medium used for each sample group is indicated in the column title (HMI-AMII: 10% FBS, 2.5% AMII; HMI-93: 20% goat serum). Abundance of each FA is shown as a percentage of total detected FA.

Fatty acid	T. brucei (HMI-AMII)	T. congolense (HMI-AMII)	T. congolense (HMI-93)
(C15:0)	0.00	0.10	0.25
(C16:0)	1.37	14.44	12.86
(C17:0)	0.39	0.65	0.72
(C18:0)	20.78	14.52	14.62
(C18:1)	8.55	8.50	5.58
(C18:2)	27.43	26.06	25.94
(C18:3)	2.78	5.39	4.88
(C20:0)	0.00	0.34	0.03
(C20:2)	0.00	0.10	0.28
(C20:3)	2.24	0.96	0.00
(C20:4)	0.62	0.85	1.04
(C22:4)	5.88	6.13	4.30
(C22:5)	19.82	15.33	16.28
(C22:6)	10.13	6.63	13.22

Discussion

African trypanosomes cause both human and animal disease. Whilst sub-species of T. brucei are primarily responsible for HAT, AAT is more complex, with three causative species (T. brucei, T. congolense and T. vivax). Management of HAT and AAT therefore requires a comprehensive One Health approach. This is hindered as the majority of our biological knowledge on African trypanosomes derives solely from T. brucei. Yet emerging data suggests that these species differ significantly at the genome [18], transcriptome [19], proteome [20] and metabolome [10] level. Furthermore, this divergence likely underlies important phenotypes relevant to infection biology, disease, and crucially, disease control. Whilst in vitro culture and genetic manipulation of T. congolense has become routine over the past decade, there remain notable and peculiar differences compared to T. brucei and, indeed, many other axenic eukaryotic in vitro systems. For example, T. congolense is only viable with a GS supplement instead of FBS [9]. Secondly, unlike most eukaryotes, T. congolense BSF are only viable at 34°C, rather than 37°C [9]. The reasons underlying these peculiarities are currently unknown, but they place limitations on carrying out comparative inter-species analyses under in vitro conditions, as well as hindering pan-species drug development.

In this study, in vitro culture of T. congolense was optimised by testing various mammalian-derived sera. T. brucei, was not viable in CS or ABS, whilst T. congolense proliferation was not viable in FBS, as previously reported. Given that serum is the primary source of lipids, proliferation in HMI-11 was subsequently tested with the addition of an alternative lipid source - AMII. Interestingly, the addition of AMII to HMI-11 supported T. congolense proliferation to a greater degree than HMI-93 (containing 20% GS), indicating that exogenous presence of lipids is important for T. congolense viability, although further work is required to understand how this translates to the situation in vivo. It is currently unclear whether the apparent lipid requirement of T. congolense influences host-adaptation, or restricts host specificity (e.g. T. congolense is known to infect many non-ruminants). Importantly, the HMI-AMII formulation enables the in vitro culture of both T. brucei and T. congolense in the same medium (this was not previously possible), allowing for their direct comparison.

Given the growth phenotypes observed in this study, a comparative lipidomic characterisation was carried out between the two trypanosome species. Even when grown under the same media conditions these two parasite species display distinct lipidomic profiles. Most lipid metabolic genes that have been identified and functionally characterised in T. brucei to date have orthologues in the T. congolense genome. Therefore, despite this apparent genomic similarity, functional divergence has occurred between the two species in terms of lipid metabolism. Indeed, previous analyses showed that T. congolense is less sensitive to specified inhibitors (e.g., Orlistat & acetyl-CoA synthetase inhibitors) of fatty acid synthesis than T. brucei, indicating that T. congolense preferentially scavenges lipids from its host environment, in lieu of biosynthesis [10].

This study reveals two important aspects of lipid metabolism in trypanosomatids. Firstly, T. congolense appears to have a preference for longer chain polyunsaturated lipids, compared to T. brucei, with notable enrichment observed among glycerophospholipids (for example, PC 40:8, PC 40:9, PE 42:8, PI 34:2 & PI 38:6). Whilst it could be hypothesised that this is a result of increased scavenging, correlation between the T. congolense lipidome and host serum lipids is low, similar to T. brucei. Instead, it is likely that T. congolense elongates and remodels lipids to a greater extent, compared to T. brucei, a mechanism supported by increased transcript abundance of ELO4 and ACS5. Glycerophospholipids, in particular PC and PE, are key components of cell membranes [21], and increased abundance of these long chain PUFAs could result in increased fluidity, flexibility and elasticity of the plasma membrane [22,23]. Further studies are required to determine the biophysical differences between the T. brucei and T. congolense plasma membranes.

Secondly, T. congolense exhibits very low abundance of ether-linked glycerophospholipids, for reasons currently unknown. Finally, even in shared medium, intrinsic species-specific lipidomic differences exist. T. brucei retains much larger ether-linked glycerophospholipid pools and the greatest total SM abundance, whereas T. congolense produces comparatively fewer ether lipids.

The biosynthesis of ether glycerophospholipids is known to be initiated at the glycosomes of T. brucei, where the precursor dihydroxyacetone phosphate (DHAP) is converted to cytoplasmic alkyl-glycerophosphate through a number of enzyme catalysed reactions. This alkyl-glycerophosphate can then pass to the ER for use in glycerophospholipid biosynthesis [24]. Currently, the wider functional significance of the high levels of ether lipids typically observed in T. brucei is unknown. The enzyme dihydroxyacetonephosphate acyltransferase (TbDAT) that underpins ether lipid biosynthesis was shown to be dispensable in procyclic forms [25], which would suggest that ether lipids themselves are non-essential, at least in this life cycle stage. RNAi of ethanolamine phosphotransferase (EPT; the enzyme that catalyses ether-type PE production) in procyclic T. brucei led to diacyl-type PE becoming dominant, which points to ether lipids being largely derived through de novo biosynthesis rather than host lipid scavenging in this species. However, the picture is less clear in bloodstream form trypanosomes. The relative absence of ether lipids in T. congolense could indicate an overall decrease in de novo lipid biosynthetic activity relative to T. brucei. Indeed, transcriptomics data indicating a reduction in both DAT and the subsequent step in ether GPL biosynthesis, alkyl-dihydroxyacetonephosphate synthase, support this hypothesis.

Sphingomyelin is a major sphingolipid in African trypanosomes, functioning in plasma membranes, as well as being involved in the active trafficking of GPI-anchored variant surface glycoproteins [26]. Both the T. brucei and T. congolense SM pools are dominated by 34:1, albeit with subtle differences in other detected SMs. For example, T. congolense exhibits increased SM 32:1, 36:1 and 36:2 abundance, indicating to some extent, a propensity for longer-chain lipids, similar to that observed for PCs. This is supported by transcriptomics data that highlight increased ACS5 and ELO4 transcript abundance in this species, with both enzymes acting on long acyl chain [17,27].

Functionally, these data highlight potential differing membrane strategies. With increasing sphingolipid carbon chain length, it has been shown that membrane thickness increases, and that degrees of their unsaturation lead to increased disorder [28]. This may account for observed decreases in lateral diffusion rates within membranes holding long chain sphingolipids. Interdigitation of sphingolipids between the two membrane leaflets also increases with increased carbon length and unsaturation. Thus, the differing sphingolipid profile presented by T. congolense (presence of longer-chain SMs with increased desaturation) may indicate that membrane microdomain structures, such as lipid rafts, differ from those occurring in T. brucei. Indeed, although it was beyond the scope of this current work to perform targeted analysis of the T. congolense sterol profile, sterol species present in these parasites may also differ, given the well-established co-regulation of sterols and sphingolipids in eukaryotes [29].

Several limitations temper these conclusions. Firstly, T. congolense is cultured at 34°C, compared to 37°C for T. brucei. Whilst a 3°C reduction in temperature is likely to impact the lipidome - eukaryotic cells are sensitive to minor temperature fluctuations [30], it more likely reflects the tissue tropism of T. congolense, which binds to vascular endothelial cells that leads to peripheral sequestration, likely within the dermal and subcutaneous microvasculature where temperatures are significantly lower than the core body temperature [31]. Both trypanosome species infect mice, which maintain a core temperature of 37°C, and it would therefore be of interest to carry out further lipidomics analyses to ascertain whether the observations made in vitro are reflective of the in vivo situation.

A universal problem with commercial serum supplements is batch-to-batch variation. In this case, differences attributed to goat serum versus FBS may partially reflect batch variation rather than true species-specific serum lipidome profiles. Comprehensive multi-omics characterisation of standard commercial animal sera such as those carried out for human serum [32] would therefore be of great benefit to standardising in vitro lipidomics and metabolomics.

In summary, this study further underscores fundamental biological differences between T. brucei and T. congolense, specifically regarding lipid metabolism. It confirms that significant divergence exists in the way these species acquire, synthesise and remodel their lipidomes, with T. congolense exhibiting a distinct preference for very long-chain PUFAs. While the functional implications of these findings remain to be fully elucidated, they likely provide a foundation for key aspects of T. congolense biology, such as intravascular cytoadhesion. Ultimately, these species-specific lipid profiles likely underlie phenotypic divergence in the in vivo biology and in host-pathogen interactions of these two parasite species.

Materials and methods

Cell culture

Bloodstream form (BSF) T. brucei strain Lister 427 were routinely maintained in HMI-11 [11] supplemented with 10% FBS, and incubated at 37°C, 5% CO₂. BSF T. congolense strain IL3000 were routinely maintained in HMI-93 [9] supplemented with 20% goat serum, and incubated at 34°C, 5% CO₂. For comparative lipidomic analysis, both species were maintained in HMI-11 supplemented with 10% FBS and 2.5% w/v AlbuMAX II (AMII; Thermo Scientific) at their respective optimal incubation conditions for 1 week (2–3 passages) prior to lipid extraction. Medium formulations are provided in S2 Table.

For growth curves, cells were seeded at a seeding density of 2 × 10⁴ cells/mL or 5 x 10⁴ cells/mL for T. brucei or T. congolense, respectively, in 2 mL cultures in a 24-well plate. Cells were counted daily using a haemocytometer and passaged every two days, or when cell density exceeded 1 × 10⁶ cells/mL. Cumulative cell density was calculated by taking dilution factors (culture volume ÷ cell seeding volume) into account. Subsequent cell counts were multiplied by these dilution factors to generate cumulative cell density values that were plotted.

Doubling times were calculated using Microsoft Excel. Cell densities were log-transformed using the LN() function and a linear regression was performed using the SLOPE() function to calculate the exponential growth rate constant (k day^-1). Doubling time was then calculated using the formula DT = ln(2)/k × 24. Where cell viability was significantly impacted, doubling time was impossible to determine. These sample groups are labelled as “ND”.

Various serum supplements were used in this study: Goat serum (GS; Gibco, Cat #: 16210072, Lot #: 2426173), Foetal Bovine Serum (FBS; Gibco, Cat #: 10270106, Lot #: B2915971RP), Adult Bovine Serum (ABS, Capricorn, Cat #: ABS-1B, Lot #: CP20–3340), Horse Serum (HS; Capricorn, Cat #: HOS-1A, Lot #: CP22-5442) and Calf Serum (CS; Capricorn, Cat #: CS-1A, Lot #: 2534176).

Drug sensitivity assays

Sensitivities of T. brucei and T. congolense to Orlistat (Cambridge Bioscience) and Acetyl-CoA synthetase inhibitor (Merck) were determined using a standardised Alamar Blue assay, as described previously [10,33]. Two-fold serial dilutions of the compounds (including a drug-free control) were prepared in 96-well opaque flat-bottomed plates (100 µL/well; Greiner One). Parasites were cultured in their standard media (HMI-11 and HMI-93 for T. brucei and T. congolense, respectively) or HMI-AMII for at least two passages prior to harvesting at log-phase of growth, and distributed across the wells (100 µL/well) at final concentrations of 2 × 10⁴ and 2.5 × 10⁵ cells/mL for T. brucei and T. congolense, respectively, in their corresponding culture media. Plates were incubated for 48 h at the relevant growth conditions for each species, after which 20 μL of 125 mg/mL (0.49 mM) resazurin dye in PBS was added to each well, and the plate further incubated for 24 h. The endpoint fluorescence at λ_em544 nm/λ_exc590 nm was determined for each well of the plate using a Cytation 5 imaging reader (BioTek), and the EC₅₀ of each drug was calculated using non-linear regression fitted to a sigmoidal dose-response curve in GraphPad Prism (v10).

LC-MS analysis of serum

LC-MS analysis was carried out as previously described [10]. Hydrophilic interaction liquid chromatography (HILIC) was carried out at the Mass Spectrometry Facility within the MVLS Shared Research Facilities (University of Glasgow, UK), using a Dionex UltiMate 3000 RSLC system (Thermo Fischer Scientific) coupled to a ZIC-pHILIC column (150 mm × 4.6 mm, 5 µm column, Merch Sequant). The column was maintained at 30°C and samples were eluted with a linear gradient (20 mM ammonium carbonate in water and acetonitrile) over 26 minutes with a flow rate of 0.3 mL/minute.

Sample injection volume was 10 µL and samples were maintained at 4°C before injection. A Thermo Orbitrap Exactive (Thermo Fischer Scientific) was used to generate mass spectra, and was operated in polarity switching mode with the following settings: Resolution: 50,000; AGC: 106; m/z range: 70–1,400; sheath gas: 40; auxiliary gas: 5; sweep gas: 1; probe temperature: 150°C; capillary temperature: 275°C. Samples were run in both positive and negative polarity with the following ionisation: source voltage +4.5 kV, capillary voltage +50 V, tube voltage +70 kV and skimmer voltage +20 V for positive mode; source voltage -3.5 kV, capillary voltage -50 V, tube voltage -70 V and skimmer voltage -20 V for negative mode. Mass calibration was performed for each polarity immediately prior to each analysis batch. The calibration mass range was extended to cover small metabolites by inclusion of low-mass contaminants with the standard Thermo calmix masses (below m/z 1400), C₂H₆NO₂ for positive ion electrospray ionisation (PIESI) mode (m/z 76.0393) and C₃H₅O₃ for negative ion electrospray ionisation (NIESI) mode (m/z 89.0244). To enhance calibration stability, lock-mass correction was also applied to each analytical run using these ubiquitous low-mass contaminants. A set of authentic standards was run prior to the sample set.

RAW spectra were converted to mzXML files (mzML files for fragmentation data) using XCMS for untargeted peak detection [34]. The resultant files were further processed using mzMatch [35] for peak matching and annotation, resulting in a tabular output that was analysed using IDEOM with default settings [36]. Statistical analyses were carried out using MetaboAnalyst (v6.0) [37].

Lipid extraction

Lipid extracts were prepared using the Folch extraction method [38]. 1 × 10⁸ cells were rapidly quenched using a dry ice/EtOH bath, then centrifuged at 1300 × g, 10 min, 4°C. Cell pellets were resuspended in a minimal volume of media and transferred to 1.5 mL eppendorf tubes, and were centrifuged at 2500 × g, 5 min, 4°C. The pellets were then washed 1 × 1 mL PBS (centrifugation 2500 × g, 5 min, 4°C). Pellets were resuspended in 50 µL PBS, and transferred to a glass vial containing 800 µL 2:1 CHCl₃:MeOH. The vial was centrifuged, then 150 µL H₂O was added to induce biphasic partition. Vials were then vortex mixed for 1 hr at 4°C. Samples were centrifuged at 1000 × g, 10 min and the organic (lipid-enriched phase) was transferred to a fresh glass vials. 10 µL of each extract was contributed to a pooled QC vial, for use in downstream mass spectrometry (MS) analysis. Samples were dried under nitrogen and stored at -80°C prior to analysis. For cell culture serum samples, extraction procedure was as described above except 100 µL of serum was mixed directly with 800 µL 2:1 CHCl₃:MeOH and 100 µL H₂O.

LC-MS lipidomic analysis

Immediately prior to analysis, dried samples were resuspended in 500 µL MeOH. 100 µL of each sample was then mixed with 100 µL MeOH 5 mM ammonium formate, and samples were centrifuged at 11,000 × g, 5 min. The supernatant (~200 µL) was then transferred to an instrument sample vial.

Untargeted lipidomics analysis was performed using a binary HPLC (Accela, Thermo Fisher Scientific) coupled to an electron spray ionization (ESI) and orbitrap mass analyser (Exactive, Thermo Fisher Scientific). Separation was conducted using a Hypersyl GOLD C18 column, 100 × 2.1 mm 1.9 nm particle size (Thermo Fisher Scientific) maintained at 50°C. Lipids were eluted over 25 minutes at constant flow rate 400 μL min^-1 using the solvent system A – H₂O 10 mM ammonium formate, 20 mM formic acid; B – 9:1 isopropanol:acetonitrile 10 mM ammonium formate 20 mM formic acid, with gradient: 65% A, 35% B (0.5 min); 35% A, 65% B (4 min); 0% A, 100% B (19 min); 65% A, 35% B (21.1 min). Water and acetonitrile were HPLC grade and obtained from Thermo Fisher Scientific, IPA was LC-MS grade (Hypergrade LiChrosolv, Merck). Ammonium formate and formic acid were both LC-MS grade and obtained from Sigma Aldrich. Mass spectra were acquired in both positive (POS) and negative (NEG)

Raw data was imported into Progenesis QI v2.4 (Non-Linear Dynamics), where automatic alignment and peak picking was performed. Relative fold quantification was performed by the software using all ion normalization, followed by data filtering based on ion intensity (>10,000), ANOVA score (<0.05), fold change (>2) and a coefficient of variance threshold of 30%. Lipidmaps and HMDB database searches were performed to assign lipid annotations. Lipid annotations were assigned based on expected retention time characteristics, mass error thresholds (<5 ppm), isotope similarity (>80) and expected adduct formation, with reference to retention times and adduct formations observed for a lipid standard mix used to monitor instrument mass calibration. This was performed for both positive and negative ionisation modes.

Tab-delimited output from Progenesis was further filtered manually, and then analysed using MetaboAnalyst (v6.0) [37]. Briefly, missing values were estimated using left-censored data estimated, a low variance filter (interquartile range) was then applied, and data were log₂ transformed and auto-scaled prior to statistical analyses.

Where two peaks were detected and annotation as the same lipid species, pearson correlation was applied. When r > 0.9, peaks were summed and reported as one lipid; when r < 0.9 lipids were individually reported as “a” and “b”.

Fatty acid methyl ester (FAME) preparation

Dried total lipid extract, prepared using the Folch method from 1 × 10⁸ cells, was resuspended in 850 µL methanol and 150 µL MeOH:H₂O (85:15) 8% (v/v) HCl. Samples were then heated at 65°C overnight. Dried samples were then resuspended in 1 mL 1:1 H₂O:Hexane and the organic layer containing FAMEs was transferred to a new glass vial. Samples were dried under nitrogen and stored at -80°C prior to analysis.

GC-MS fatty acid analysis

FAMEs samples were analysed via gas chromatography mass spectrometry (GC-MS). Gas chromatography was carried out on a Rtx-2330 60 m length × 0.25 mm inner diameter × 0.20 µm film thickness column (Restek, P/N: 10726) installed in a Trace Ultra gas chromatograph (Thermo Fisher Scientific). Helium carrier gas at a flow rate of 1.4 mL/min was used. 1 µL of derivatized sample was injected into a Programmable Temperature Vaporiser (PTV) injector using a 1 in 10 split injection. Injection temperature was 80°C before increasing to a transfer temperature of 220°C at a ramp rate of 14.5°C/min. Oven temperature programme was as follows: 80°C (3.16 min hold); ramp rate 30°C min^-1 to 175°C; ramp rate 3°C min^-1 to 210°C; 30°C min^-1 to 265°C (10 min hold). Eluting peaks were transferred through an auxiliary transfer temperature of 200°C into a ITQ900-GC mass spectrometer (Thermo Scientific) with scanning mass range set to 60–400 m/z. Targeted analysis with comparison to compound FAMEs standards peak areas were extracted using TraceFinder v3.3. Standards and unknowns were determined by using custom R scripts using XCMS 3.8.2, for peak picking, CAMERA peak grouping package, version 1.42.0 and metaMS 1.22.0.

Supporting information

S1 Fig. Trypanosome doubling times when cultured in different sera.

A) Doubling times for T. brucei. B) Doubling times for T. congolense. In both panels, mean doubling time is shown (in hrs) inside the bars. Statistical difference was calculated for each condition compared to the control (10% FBS and 20% GS for T. brucei and T. congolense, respectively) via a BH-corrected t-test. Statistics: ns = not significant; * = < 0.05; ** = < 0.01; *** = < 0.001. Abbreviations: ND, not determined; FBS, foetal bovine serum; GS, goat serum; ABS, adult bovine serum; HS, horse serum; CS, calf serum, aMAX, albuMAX II lipid-loaded BSA.

(TIFF)

S2 Fig. Analysis of lipidomics data.

A) Overall abundance of ether-linked PCs (PC-Os) in T. brucei (FBS), T. congolense (FBS) and T. congolense (GS). Overal abundance was calculated by summing the peak intensities of all detected PC-Os, highlighting a significant reduction in both T. congolense sample groups. B) Similarly, T. congolense exhibits a significant reduction in ether-linked PEs (PE-Os) compared to T. brucei. C) T. brucei exhibits higher abundance of PC-O 36 lipids, compared to T. congolense, regardless of culturing conditions of the latter (left panel). Notably, the degree of increased abundance reduces with increasing desaturation of the 36-carbon chain. Contrastingly, T. congolense possesses elevated levels of PC-O 38 lipid species compared to T. brucei, although T. brucei abundance is higher when the 38-carbon chain is desaturated. The third panel shows 40 + carbon chain length PC-Os, with T. congolense exhibiting higher abundance of lipids in this category. D) Forest plots highlighting 6 lysophospholipids that were significantly altered in at least one of the two-way comparisons between the three sample groups. E) Forest plots showing the remaining 8 lipids from various classes that were significantly altered in at least one of the two-way comparisons, including a diglyceride, two cardiolipins, two ceramides and three cholesterol esters.

(TIFF)

S3 Fig. Global correlation analysis of parasite and serum lipidomes.

XY scatter plots were generated to display comparisons for (A) TcoGS vs GS, (B) TcoFBS vs FBS, and (C) TbFBS vs FBS. Individual lipid species are colour-coded into 9 lipid classes (indicated in the legend). Only moderate-to-weak correlations were detected (Spearman; S5 Table), with the exception of neutral lipids (>0.9).

(TIFF)

S4 Fig. Comparative analysis of lipid metabolism in T. brucei and T. congolense at the transcriptome level.

A) RNAseq pathway analysis of glycerophospholipid metabolism. Data were derived from a previous comparative study of T. brucei and T. congolense [10]. Enzyme names are shown in italics, and dots representing each enzyme are coloured in blue (higher mRNA abundance in T. brucei), white (equal abundance) or red (higher mRNA abundance in T. congolense). Full enzyme names are listed in S3 Table. B) RNAseq pathway analysis of sphingolipid metabolism. Data were derived as mentioned above. Full enzyme names are listed in S3 Table.

(TIFF)

S1 Table. Metabolomics analysis of FBS and goat serum.

A total of 482 metabolites were putatively identified in one or both sera. Column A: putative identities of the metabolites; columns B-E: raw peak intensities for four technical replicates of goat serum; columns F-I: raw peak intensities for four technical replicates of FBS; columns J-Q: Log₂-transformed peak intensities for goat serum (J-M) and FBS (N-Q). The transformation was carried out on peak intensities + 1 to ensure that undetected metabolites (peak intensity = 0) returned a value of 0; column R: Log₂ fold change between goat serum and FBS, calculated by subtracting the mean Log₂ FBS from the mean Log₂ goat serum values. Column S: t-test result for each Log₂FC, calculated using Metaboanalyst; column T: FDR-adjusted P value (t-test). Abbreviations: GS: goat serum; FBS: foetal bovine serum; Log₂FC: Log₂ fold change; FDR: false discovery rate.

(XLSX)

S2 Table. Medium formulations for HMI-93, HMI-11 and HMI-AMII.

Column B: component name; column C: component molecular weight (if applicable); Columns D-E: concentrations in mg/L and µM for HMI-93; columns F-G: concentrations in mg/L and µM for HMI-11; columns H-I: concentrations in mg/L and µM for HMI-AMII. Rows 3–44 are components of Iscove’s Modified Dulbecco’s Medium (IMDM). Rows 45–52 are required for trypanosome culture.

(XLSX)

S3 Table. Lipid metabolic enzyme comparison between T. brucei and T. congolense.

Genes encoding lipid enzymes were previously identified in the TREU 927 reference genome [17]. These genes were used to find orthologues in two T. congolense genomes: the shotgun assembly (2016) and the more recent PacBio assembly (2019). These data were combined with transcriptomics data generated previously [10]. This analysis also included an Orthofinder [39] analysis of the T. brucei and T. congolense proteomes. Where multiple gene IDs are listed in one row, Orthofinder was unable to determine definitively which gene is the true syntenic orthologue, and therefore, the orthologues are grouped together.

(XLSX)

S4 Table. Lipidomics results and analysis.

Microsoft Excel spreadsheet containing all detected and annotated lipids as well as the final dataset used to generate all figures. Separate worksheets are filtered to show only the significant hits obtained via three two-way comparisons: TbFBS vs TcoGS, TbFBS vs TcoFBS and TcoGS vs TcoFBS.

(XLSX)

S5 Table. Parasite and serum lipidome correlation and uptake ratio analysis.

Microsoft Excel spreadsheet containing the Spearman Rank values for 9 lipid classes (PC, PE, ether lipids, lysophospholipids, PI, neutral lipids, sphingolipids, CE & CL) to test for correlations between parasite lipidomes and that of their respective sera. A second worksheet contains uptake ratios for each lipid (parasite abundance vs serum abundance) for TbFBS and TcoFBS compared to FBS.

(XLSX)

S6 Table. GC-MS analysis of T. brucei and T. congolense.

Raw GC-MS data including retention times and detected masses used to generate Table 2.

(XLSX)

Data Availability

All data generated or analysed during this study are included in this published article and its Supplementary Information files.

Funding Statement

E.D. was supported by a Scottish Metabolomics Network Training grant, a Wellcome ISSF feasibility grant, and funding from the Scottish Universities Life Sciences Alliance. P.S. is supported by a BBSRC Discovery Fellowship (BB/X009807/1). This study was partially supported by a BBSRC grant awarded to L.M. (BB/S00243X/1) and M.B. (BB/S001034/1). The Roslin Institute is supported through core funding from the BBSRC (BS/E/D/20002173; BBS/E/RL/230002C). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.