De Novo Assembly of the Trypanosoma congolense Genome Reveals an Organization Influenced by Antigenic Variation but Distinct from Trypanosoma brucei Authors: Marija Krasiļņikova, Jane C Munday, Dario Beraldi, Stephen D Larcombe, Guy R Oldrieve, Craig Lapsley, Liam J Morrison, Keith R Matthews, Richard McCulloch Published: 2026 ABSTRACT AbstractAntigenic variation allows pathogens to evade mammalian adaptive immunity through the continuous change in exposed antigens. In African trypanosomes, antigenic variation involves changes in expressed Variant Surface Glycoproteins (VSGs). Understanding of VSG expression control and change amongst African trypanosomes is most advanced in Trypanosoma brucei. In the important animal trypanosome, Trypanosoma congolense, incomplete genome assembly has held back understanding of the mechanics of antigenic variation. Here, we have used long-read DNA sequencing and Hi-C DNA interaction analysis to provide a telomere-to-telomere assembly of the T. congolense genome. This assembly reveals a genome comprising 12 diploid chromosomes, one tetraploid chromosome, and more than 100 small chromosomes. With this assembly we reveal several features of VSG organization and expression that differ from T. brucei. The majority of the T. congolense VSG archive, estimated at ∼1,500 genes, localizes to subtelomeres in 12 of the 13 large chromosomes, but these loci are notably smaller than are found in T. brucei. Furthermore, transcriptome analysis suggests expression of VSGs across the T. congolense subtelomeres, which are not separated within the nucleus from non-VSG chromosome regions, suggesting that there is no dedicated VSG expression site. Strikingly, one chromosome contains approximately 40% of the VSG archive and is largely transcriptionally silent, potentially acting as the major reservoir of new VSG variants. Finally, we show that VSG expression can be detected from multiple small chromosomes. In summary, the new genome assembly provides a platform for understanding a potentially unusual operation of VSG expression and switching in T. congolense. African trypanosomes share with each other the expression of Variant Surface Glycoprotein (VSG) on their cell surface, drawing from very large archive of VSG genes in a process called antigenic variation. Despite this shared VSG biology, we have little understanding of the mechanics or dynamics of VSG expression in the important animal trypanosome, Trypanosoma congolense. Here, we provide an assembly of the Trypanosoma congolense genome and map VSG expression, revealing substantial differences in the organization of VSG genes and their expression relative to the best-studied African trypanosome, Trypanosoma brucei. Comparing the genomes of two closely related African trypanosomes therefore reveals shared and divergent architecture and provides preliminary evidence for differing strategies for antigenic variation via a common antigen, the VSG. Introduction African trypanosomes, including Trypanosoma brucei, Trypanosoma congolense, and Trypanosoma vivax, rely on antigenic variation to evade elimination by the adaptive immune response of mammals (McCulloch et al. 2017). In each case, antigenic variation is thought to involve periodic changes in expression of Variant Surface Glycoprotein (VSG) (Bangs 2018; Silva Pereira et al. 2022). Our understanding of this critical survival mechanism is most advanced in T. brucei (Faria et al. 2022; Barcons-Simon et al. 2023), whose genome has been shaped by the need to express one VSG in a cell at a time (Borst 2002; Glover et al. 2013; Cestari and Stuart 2018) and to house a store of silent VSG genes that can be expressed in the course of an infection (Berriman et al. 2005; Marcello and Barry 2007b). Expression of one VSG/cell in the mammal is dictated by selective transcription of a single VSG from one of ∼15 bloodstream VSG expression sites (BESs), all of which are found abutting the telomeric ends of many T. brucei chromosomes (Hertz-Fowler et al. 2008). Most of the T. brucei genome is found on 11 megabase chromosomes (∼1 to 5 Mbp size range), each of which is separated into two compartments: a stable, diploid core that is unusually heavily transcribed due to virtually all genes being expressed from multigenic transcription units (Daniels et al. 2010); and less stable, transcriptionally silent subtelomeres that separate the chromosome cores from telomeric VSG BESs (Berriman et al. 2005). The subtelomere compartment contains arrays of ∼2,000 silent VSGs and VSG pseudogenes (Marcello and Barry 2007a; Cross et al. 2014) and its total size approaches that of the core (Muller et al. 2018; Krasilnikova et al. 2025). Moreover, subtelomeres are notably variable between T. brucei strains (Callejas et al. 2006), between chromosome homologs (Cosentino et al. 2021), and over time in culture, in particular after mutation of the homologous recombination (HR) factors RAD51 or BRCA2 (Trenaman et al. 2013; Krasilnikova et al. 2025). Such variability appears consistent with Hi-C evidence for greater interactivity of chromosome subtelomeres relative to the cores (Muller et al. 2018). At least part of this subtelomere variability likely stems from recombination of silent VSGs into BES to change the expressed VSG during antigenic variation. Such recombination is not limited to the VSG archive in the megabase chromosomes, since T. brucei also possesses many sub-megabase chromosomes, often named mini (∼50 to 150 kbp in size) or intermediate-chromosomes (∼150 to 700 kbp). Telomere-proximal VSG BES and silent VSGs are found on many of these chromosomes, thus expanding the VSG archive (Wickstead et al. 2004), though some also contain other genes, which are expressed (Krasilnikova et al. 2025). A shared feature of sub-megabase chromosomes is 177 bp repeats (Wickstead et al. 2004), which appear to act as origins of DNA replication (Krasilnikova et al. 2025) and recruit many kinetochore components (Carloni et al. 2026). Our understanding of the organization of the T. congolense genome and how it is influenced by antigenic variation is substantially less advanced than that of T. brucei. Fewer pulsed field gel analyses of T. congolense have been conducted than for T. brucei, meaning the total number of T. congolense chromosomes is not known, though there is evidence for numerous submegabase chromosomes (Majiwa et al. 1985a; Gibson and Borst 1986; Garside et al. 1994; Abbas et al. 2018). Though recombination of T. congolense VSGs has been linked to expression (Majiwa et al. 1985b), it is unclear if such recombination is the driver of immune evasion. Indeed, expression of a VSG from a minichromosome has been documented without recombination (Majiwa et al. 1986). Such a finding is consistent with long-read PacBio sequencing that revealed potential BES facsimiles on several T. congolense minichromosomes (which lack 177 bp repeats) (Abbas et al. 2018). However, though such putative minichromosomal BESs comprise telomeric VSGs and some upstream conserved sequences (Abbas et al. 2018), they are unlike T. brucei BESs in that they lack detectable expression site associated genes (ESAGs), known RNA Pol I promoters or 70 bp repeats (Hertz-Fowler et al. 2008), and no work has documented VSG transcription from them. Importantly, no work has asked if individual T. congolense cells express just a single VSG at a time, or has explored the balance between transcription changes and recombination in switching VSG expression. More widely, the organization of the VSG archive (Jackson et al. 2012) in the T. congolense genome is unknown, adding to uncertainty about the operation of antigenic variation. It is clear that T. congolense has the capacity to encode many VSGs but in contrast to T. brucei, where VSGs are separated into a- and b-types (based on cysteine residue distribution) (Carrington et al. 1991) that undergo seemingly unconstrained recombination, T. congolense contains only b-type VSGs (Silva Pereira et al. 2022) that fall into multiple defined phylotypes, with recombination considered to be largely limited to within a phylotype (Jackson et al. 2012; Silva Pereira et al. 2018). Whether such a constraint on VSG recombination is based on VSG gene sequence or aspects of genome location or transcription is unknown, since the locations of VSGs across the full T. congolense genome has not been documented. In this regard, no work has documented sequences around the telomeres of all chromosomes, and so it remains unclear if canonical VSG BESs are present in the T. congolense genome but have escaped detection. Here, we have addressed many of these questions around how antigenic variation has shaped the T. congolense genome by using long-read Oxford Nanopore Technology (ONT) sequencing allied to Hi-C to generate a de novo assembly of the T. congolense IL3000 genome. Using this approach, we show that the T. congolense genome comprises 13 “megabase” chromosomes, all of which were resolved as haplotypes and assembled telomere-to-telomere (T2T). Many more submegabase chromosomes were similarly fully assembled. 12 of the 13 megabase chromosomes are diploid, while one is tetraploid in wild type cells but reduces in ploidy in the absence of the HR factor RAD51. We show that the organization of VSGs in the genome is quite different from that of T. brucei: VSG-rich subtelomeres in T. congolense megabase chromosomes are notably smaller than in T. brucei, appear not to be transcriptionally silent, and do not show Hi-C evidence for being a distinct nuclear compartment to the core. Furthermore, though a silent VSG archive is present, it appears limited to a single chromosome pair that is largely untranscribed and whose ploidy is reduced after loss of BRCA2 or RAD51. Results and Discussion Assembly of the T. congolense IL3000 Genome To attempt to assemble the T. congolense IL3000 genome, we prepared DNA from wild type, RAD51 null mutant and BRCA2 null mutant bloodstream cells (Fig. S1) growing in culture and compared several sequencing and assembly approaches (Fig. S2). Based on a combination of genome completeness scores determined by QV (Rhie et al. 2020) and BUSCO (Tegenfeldt et al. 2025), as well as contiguity of the haplotype-resolved sequences, the most complete assembly of the largest chromosomes was generated by verkko (Rautiainen et al. 2023), combining ONT long read data and Hi-C reads. In contrast, the more numerous smaller chromosomes were best assembled using hifiasm (Cheng et al. 2024). Thus, a combination of the two assemblies provided a description of the whole nuclear genome. The resulting combined assembly, following gap closure, comprised 28 contigs >850 kbp (range 0.88 to 3.88 Mbp), all of which displayed telomeric repeats at each contig end (TTAGGG/CCCTAA) (Fig. 1a). In addition, two intermediate-sized T2T contigs (236,893 bp and 415,598 bp) and 145 T2T <100 kbp contigs were also assembled, with both of these sets of sequences containing 369 bp repeats previously shown to be characteristic of the small chromosomes of this Trypanosoma species (Abbas et al. 2018). Whether or not the 369 bp repeats provide the same DNA replication origin (Krasilnikova et al. 2025) and putative centromere functions (Carloni et al. 2026) as the 177 bp repeats in T. brucei is untested. A haplotype-resolved assembly of the 13 largest chromosomes in the T. congolense genome and synteny with T. brucei megabase chromosomes. a) All assembled chromosomes of T. congolense IL3000 above 850 kbp in length are shown, comparing two haplotypes for chromosomes 2 to 13 and four haplotypes for chromosome 1. Genes and pseudogenes predicted using Companion (Haese-Hill et al. 2024) are shown in two shades of gray (reflecting their coding strand), and VSG family genes are highlighted (Pfam PF00913 for a-type and Pfam PF13206 for b-type VSGs). An unresolved sequence gap is highlighted by an asterisk. b) Mapping of the newly assembled T. congolense chromosomes onto T. brucei brucei Lister 427 megabase chromosomes; T. brucei subtelomeres are underlined (genome version 2018 v.12) (Rabuffo et al. 2024). Sequence analysis of the 28 > 850 kbp contigs revealed that these represent haplotype-resolved sequences of 13 distinct chromosomes (Fig. 1a, Fig. S3). 12 of the chromosomes are diploid, while one (chr 1, the smallest) is tetraploid in wild type cells. In 12 of the chromosomes directional gene clusters could be discerned (Fig. 1a, Fig. S4), consistent with the expected multigenic transcription seen in T. brucei and wider kinetoplastids (Jackson et al. 2008; Daniels et al. 2010). In one copy of chromosome 3 a gap in the assembly could not be closed, which was centered on the repetitive splice leader gene array. In each chromosome a single locus was detected that comprised AT-rich tandem repeats, which was syntenic between the chromosome homologs but differed in size across all chromosomes (24.7 to 128 kbp; median 53.1 kbp, mean 55.3 kbp). It seems plausible that this feature represents a single centromere on each chromosome, consistent with what is seen in T. brucei (Obado et al. 2007; Obado et al. 2011; Muller et al. 2018), but experimental evidence is needed to test such a prediction (Akiyoshi and Gull 2014). To understand how related the large chromosomes of T. brucei and T. congolense are, we performed synteny analysis between the 11 megabase chromosomes of T. brucei and the 13 predicted chromosomes of T. congolense (Fig. 1b). In no case was there complete synteny between any chromosome pairs, and instead each T. brucei chromosome core was a composite of at least two T. congolense chromosomes. Thus, there has been a significant genome rearrangement since the two species emerged in evolution. In addition, there was limited evidence of sequence conservation of the T. brucei subtelomeres with sequences in T. congolense (Fig. 1b), which appears consistent with previous analysis that indicated the VSGs of the two species are distinct in sequence and recombination (Jackson et al. 2012). Though the size range of these 13 large T. congolense chromosomes (0.88−3.88 Mbp) is smaller than the megabase chromosomes of T. brucei, the ability to map much of the core genome content between them suggests they represent the “megabase” chromosome complement of the genome. For 12/13 of the megabase chromosomes, a clear distinction from all the smaller chromosomes (see below) was the lack of discernible 369 bp repeats. The sole exception is chromosome 4, where the single AT-rich, tandemly repeated region also contains an array of 369 bp repeats. This observation may suggest that the smaller chromosomes (see below), many of which mainly contain VSGs, evolved from this megabase chromosome. However, the unusual composition of chromosome 4 (discussed below), may indicate that it arose by aggregation of minichromosomes and addition of a putative centromeric AT-rich locus to ensure its segregation. An Ancestral Tetraploid Chromosome in T. congolense Aneuploidy is common across the kinetoplastida (Reis-Cunha et al. 2018), having been documented in Leishmania (Downing et al. 2011; Rogers et al. 2011) and T. cruzi (Reis-Cunha et al. 2015), but appears scarce in T. brucei (Almeida et al. 2018). Much of such aneuploidy, at least in Leishmania, arises apparently stochastically during growth (Sterkers et al. 2014), but recent work has shown that one chromosome, conserved across a wide range of trypanosomatids, is stably polyploid (Reis-Cunha et al. 2024; Greif et al. 2026). In T. brucei there is no such amplified chromosome, which has instead evolved as a duplication within chromosomes 4 and 8 (Jackson 2007; Marques et al. 2015; Reis-Cunha et al. 2024), thus preserving diploidy across the megabase genome. As our assembly predicted that chromosome 1 in T. congolense is found in four copies, we asked if this chromosome corresponds to the ancestrally duplicated chromosome by searching for orthology in L. major and in T. brucei (Fig. 2a). Synteny analysis showed that the core of T. congolense chromosome 1 best corresponds to chromosome 31 of L. major, which is stably polyploid, and best synteny for this chromosome with T. brucei was seen on chromosomes 4 and 8, with some additional overlap with chromosome 10. Thus, T. congolense retains a complete chromosome that is tetraploid, suggesting an organization more like the majority of trypanosomatids and distinct from T. brucei, which may therefore be unique in absorbing such raised ploidy into diploidy through duplication of the ancestral chromosome. In this regard, further analysis of the T. vivax genome may be warranted (Reis-Cunha et al. 2024), as well as more closely related species such as T. simiae and T. suis. Loss of HR undermines stability of tetraploid chromosome 1 and VSG-rich chromosome 4 in T. congolense. a) Sequences best matching L. major chromosome 31, as well as T. brucei chromosomes 10, 8, and 4 (T. brucei brucei Lister 427 2018 v.12), are shown for the four copies of chromosome 1 (tco01a-tco01d). b) Chromosome copy number is shown for T. congolense IL3000 chromosomes 1 and 4 (tco01 and tco04) in wild type (WT) cells and in RAD51 or BRCA2 null mutants, before (P0) and after prolonged in vitro growth (P22); chromosome copy number was estimated from normalized read depth coverage of short reads across putative coding sequences. Loss of Homologous Recombination Leads to Ploidy Reduction of Select T. congolense Chromosomes Analysis of chromosome copy number in T. congolense RAD51 and BRCA2 mutants provided insight into genome stability (Fig. 2b). Whereas chromosome 1 was found to be stably tetraploid in wild type cells after growth in vitro, a decrease in ploidy was seen in the two mutants: one copy of chromosome 1 was lost in each of the mutants. Additionally, a copy of chromosome 4 was lost in the RAD51 mutants. In contrast, no change in ploidy was seen for any other chromosome in either mutant (Fig. S5). These data suggest that functional HR is needed to maintain polyploidy of chromosome 1 for reasons that remain unclear, but it would be of interest to explore this intersection in other trypanosomatids (Reis-Cunha et al. 2024). Loss of chromosome 4 in HR mutants does not relate to polyploidy and instead may reflect VSG archive organization (see below). Arrangement and Expression of the VSG Archive Differ in T. congolense and T. brucei To examine the organization of VSG genes in the T. congolense genome, we localized all potential VSGs by identifying genes that fulfilled the following criteria: annotated by Companion (Haese-Hill et al. 2024) as VSG, and contained the b-VSG Pfam domain PF13206. In total, 1,529 putative VSGs were predicted in this way, of which Companion predicted 24.1% to be pseudogenes, comparable to that reported previously (Jackson et al. 2012) and substantially fewer than the predicted ∼80% of T. brucei VSGs thought to be pseudogenes (Marcello and Barry 2007a; Cross et al. 2014). VSGs were found on all megabase chromosomes (Fig. 1a), and this part of the repertoire represents most (70.2%) of the T. congolense VSG archive. Genes containing Pfam domain PF00913 (a-VSGs in T. brucei) are not thought to encode VSGs in T. congolense but to be PAGs (procyclin-associated genes) or encode transferrin receptors (Silva Pereira et al. 2022); these genes were less numerous than VSGs (Fig. S6a) and were found in all the T. congolense megabase chromosomes (Fig. 1a, Fig. S6a). Given the separation of T. congolense VSGs into discrete phylotypes (Silva Pereira et al. 2018), we asked if these were spatially separated across the megabase chromosomes, potentially explaining their recombinational isolation (Jackson et al. 2012), but each chromosome contained a range of phylotypes (Fig. S6b). Arrangement of the T. congolense VSG archive across the megabase chromosomes was strikingly different from T. brucei in three ways. First, in contrast to the large VSG-rich subtelomeres found in each T. brucei megabase chromosome (median 26.7%, mean 31.1% of chromosome length) (Muller et al. 2018; Krasilnikova et al. 2025; Rabuffo et al. 2024), 12/13 T. congolense chromosomes contained notably smaller telomere-proximal, VSG-rich regions: only 0.36% to 7.55% of predicted protein-coding genes and pseudogenes in each chromosome were VSGs (Fig. 1a, Fig. S6a). Second, the entire length of one chromosome (4) contained VSGs (Fig. 1a): in total 533 VSGs were housed here, amounting to 41.2% of the total archive and representing 29% and 32% of all putative genes on the two chromosome homologs. A further measure of the abundance of VSGs in chromosome 4 is seen when considering only genes and pseudogenes that have identifiable Pfam domains: the VSG B Pfam domain PF13206 accounts for 347/399 and 298/331 of any detected Pfam in the two homologs (Fig. S6c). Indeed, so unusual is this organization, and so unlike what is seen in T. brucei, it was impossible to derive a synteny map between T. congolense chr 4 and any T. brucei chromosome (Fig. 1b). For reasons that are unclear, most of the VSGs on chr 4 are encoded on the same strand (Fig. S4). Third, analysis of sequence proximal to all 56 telomeres detected in the 28 haplotypes of the 13 chromosome homologs failed to detect any feature consistent with a T. brucei VSG BES: though VSGs were found directly adjacent to the telomere tract in 31/56 cases (see below, Fig. 3), these were not preceded by a conserved set of genes (based on Pfam analysis of all putative genes and pseudogenes 70 kbp upstream of telomeric repeats), and we were unable to detect 70 bp or 50 bp repeats (characteristic elements found upstream of, respectively, T. brucei VSGs (Liu et al. 1983; Marcello and Barry 2007a) and BESs (Sheader et al. 2003; Krasilnikova et al. 2025)) anywhere in the T. congolense genome assembly. Thus, the telomeric ends of the T. congolense megabase chromosomes are not locations of loci resembling a VSG BES, which is the sole site of VSG transcription in T. brucei cells in the mammal (Hertz-Fowler et al. 2008). Nonetheless, such sequence analysis cannot exclude that T. congolense VSG transcription occurs from these chromosomes in a different form of expression site, perhaps more akin to the smaller VSG expression sites used by T. brucei in metacyclic life cycle forms in the tsetse (Graham et al. 1999; Ginger et al. 2002; Hutchinson et al. 2021). Gene expression from the 13 largest chromosomes in T. congolense. a) Bulk RNAseq data is shown mapped to haplotype A copies of the 13 largest T. congolense chromosomes, with VSG family Pfam-containing genes and AT-rich tandem repeat regions highlighted.; RNAseq was generated from P0 wild-type cells and analysed using Salmon (output in TPM). b) A more detailed look at expression from the subtelomeric regions of both copies of chromosome 4 (tco04). c) As in b), a more detailed look at expression levels at subtelomeric regions of haplotype A of chromosomes 7, 6, and 10 (tco07a, tco06b, and tco10a). To understand if and how VSGs might be expressed in the megabase chromosomes, we mapped RNA from recently cloned wild type bloodstream form T. congolense IL3000 to the new genome assembly: Fig. 3a shows mapping to a single homolog of each chromosome, while Fig. S7 shows mapping to the two homologs of each chromosome. Consistent with the presence of multigenic transcription units, RNAseq signal was seen across most of the non-VSG “central” regions of 12 of the 13 chromosomes; the only regions without detectable read mapping corresponded to the single putative centromeric AT-rich locus in each chromosome. Strikingly, in each of these chromosomes, RNAseq reads mapped to VSGs, whether distal or proximal to the telomere (Fig. 3c). Applying a conservative expression threshold of minimum 5 transcripts per million (TPM), 309/540 (57.2%) of the putative VSGs on 12/13 chromosomes were considered expressed. Chromosome 4 was again a notable variant: here, there was little evidence for RNAseq mapping across most of the chromosome (Fig. 3a and b, Fig. S7) and, applying the same threshold, only 21/533 (3.94%) VSGs were expressed, with the clearest RNAseq signal being telomere-proximal. Whether this single chromosome acts as a silent VSG reservoir comparable to that spread across all subtelomeres of T. brucei and thereby provides a source of new VSG variants by recombination, remains to be seen. Equally, what purpose(s) is served by the many other unexpressed non-VSG genes in chromosome 4 is unclear. Taken together, these data reveal a striking divergence from what has been extensively documented regarding VSG expression dynamics in T. brucei growing in vitro. Whereas in wild type bloodstream form T. brucei a single VSG transcript and protein, corresponding to a VSG gene located in a telomeric BES, hugely predominates in a population (DuBois et al. 2012; Glover et al. 2016; Briggs et al. 2018; Faria et al. 2019; Mehnert et al. 2021; Escrivani et al. 2023; Keneskhanova et al. 2025), we find no such evidence for predominant expression of one VSG in T. congolense. Instead, we see expression of many VSGs and no predominant localization within the megabase chromosomes. These data could be explained by levels of VSG recombination to an unidentified VSG expression site (or sites) that surpasses what is seen in wild type T. brucei, or by lack of selective expression of one VSG (or a small number of VSGs) in T. congolense. Hi-C Provides no Evidence for Distinct Core and Subtelomere Genome Compartments in T. congolense Assembly of the T. congolense genome was informed by Hi-C interaction data, and so we next analysed these data in more detail to ask if the genome of T. congolense is organized within the nucleus like that of T. brucei. Figure 4a shows Hi-C interaction matrices of inter- and intra-chromosomal DNA interactions for all megabase chromosomes; these data support the assembly and arrangement of these larger chromosomes. To investigate whether large-scale genome compartmentalization can be seen in T. congolense, as described in T. brucei (Muller et al. 2018; Rabuffo et al. 2024) and T. cruzi (Diaz-Viraque et al. 2023; Greif et al. 2026), we generated correlation matrices, along with 1 to 10 eigenvector data and insulation scores, and measured contact decay, from the Hi-C data (Fig. 4b, Figs. S8–S10). Distinct genomic compartments could not be clearly discerned in T. congolense from any of these analyses, an observation that aligns with the above RNAseq data (Fig. 3, Fig. S7), which failed to reveal a clear difference in transcription between the subtelomeric, VSG-rich parts of the chromosomes and in the VSG-denuded chromosome cores. Hi-C provides no evidence for large-scale compartmentalization within the 13 largest T. congolense chromosomes. a) Corrected Hi-C matrix for haplotype A and haplotype B of the 13 largest T. congolense chromosomes at 50 kbp resolution (interaction matrices not corrected for distance decay). b) Examples of correlation matrices, along with insulation scores and 4 eigenvector values, for four chromosomes at 25 kbp resolution. Putative genes are highlighted in black or gray, depending on coding strand; VSG genes are highlighted in green and putative centromeres in light blue. PC1-4 plotting range is −0.1 to +0.1, insulation score plotting range is −2 to +2. PC denotes principal component, and IS denotes insulation score. Two T. congolense genomic elements did, however, stand out in the Hi-C analyses: the AT-rich locus on each chromosome, and sites of gene direction change (Fig. 4, Figs. S9 and S10). In T. brucei, centromeres have been shown to act as pronounced boundaries within the megabase chromosomes (Muller et al. 2018), and the interaction profile around the AT-rich locus on each T. congolense chromosome (Fig. 4b, Fig. S9) may be consistent with these being centromeres. However, we note that high conservation of the AT-rich repeats in these loci may confound the analysis due to multimapping sequencing reads. While no transcription initiation- or termination-associated factors, including histone variants, have been mapped to the T. congolense genome, many of the loci inferred to be the ends of directional gene clusters, by virtue of change in common gene orientation, showed shared interaction patterns (Fig. 4b, Fig. S9). This observation shares a similarly with what has been observed at transcription start sites in T. brucei (Rabuffo et al. 2024), suggesting transcriptional hubs are a broad feature of trypanosomatid nuclei, driven by the needs of widespread polycistronic transcription. Notably, such interaction profiles were less clear across chromosome 4 (Fig. 4b), consistent with RNAseq evidence for a lack of gene expression in the VSG-rich core. Organization of and VSG Expression From the Abundant Small Chromosomes in T. congolense The above discussion focuses on the megabase chromosomes of T. congolense, but the assembly additionally predicted 147 telomere-to-telomere chromosomes ranging in size from 27 to 416 kbp (Fig. 5). Based on their gene content, size, chromosome organization and repeat content, these smaller chromosomes can be split in three groups. In the first group, consistent with previous reports (Abbas et al. 2018), the 141 smallest chromosomes (27 to 55 kbp) contain a central core of tandemly repeated 369 bp repeats, flanked by one or more putative coding sequences; 135/141 of these chromosomes contain 1 to 5 putative VSGs (Fig. 5a). The second group, comprising four slightly larger (64 to 74 kbp) and less symmetric chromosomes, each has a large and small gene coding region (40 to 48 kbp) flanking the 369 bp repeat region (Fig. 5a), and no detected VSGs. The third group comprised two notably larger chromosomes (237 and 416 kbp) that contain an additional AT-rich region containing a 32 bp repeat adjacent to the 369 bp repeat region (Fig. 5b). Together, these submegabase chromosomes house approximately 20% of VSGs in the genome assembly (Fig. 5c). Based on Pfam content analysis, the two largest submegabase chromosomes contain predominantly putative genes with no detected Pfam domains, as well as numerous VSG A and VSG B family genes (Fig. 5d). This variable composition of submegabase chromosomes in T. congolense warrants comparison with a recent reevaluation of the small chromosomes of T. brucei (Wickstead et al. 2004; Krasilnikova et al. 2025), in that in both species they are not solely repositories of VSGs but contain other genes and, furthermore, their evolution has involved retention of likely centromeric elements to allow DNA replication and/or provide mitotic stability. Assembly and expression of the abundant small chromosomes in the T. congolense genome. a) Examples of T. congolense IL3000 minichromosomes and their gene expression from bulk RNAseq data in wild type cells, with putative genes, VSGs and repeat elements highlighted. b) as in A but for two predicted intermediate chromosomes. c) The distribution of putative VSG genes across the different chromosome groups (megabase: chromosomes >850 kbp; intermediate: 200 to 500 kbp; and minichromosomes: <100 kbp). The genome assembly also contains contigs that are not telomere-to-telomere assemblies, and these are categorized as “unassigned”. d) Gene content of the two intermediate chromosomes shown in b), based on Pfam analysis of translated putative gene and pseudogene sequences. In T. brucei, no work has documented VSG transcription from minichromosomes (Zomerdijk et al. 1992), whereas a previous report suggested sequence elements in some T. congolense submegabase chromosomes that might allow telomere-adjacent transcription (Abbas et al. 2018), including conserved noncoding element 1 (CNE1, Fig. 5a). To test this proposed difference in function, Fig. 5a and b shows RNAseq mapping to the submegabase chromosomes assembled here. These data reveal that T. congolense small chromosomes display VSG expression: using the same TPM cutoff of minimum 5 TPM, 121/239 (50.63%) putative VSGs located on these chromosomes are expressed across 97/135 VSG-containing small chromosomes. Furthermore, at least in the two largest small chromosomes, transcripts were detected from nontelomeric VSGs (Fig. 5b). As for the megabase chromosome VSG archive, the expression we detect of multiple minichromosomal VSGs may reflect frequent recombinogenic activation to an unknown locus, but may also indicate transcription in situ, which would represent a further the disparity between T. brucei and T. congolense VSG expression dynamics. Conclusions A combination of long-read ONT sequencing and Hi-C DNA interaction analysis has revealed the genome of T. congolense to be composed of 12 diploid chromosomes, one tetraploid chromosome, and more than 100 small chromosomes. Superficially, this organization compares with the similarly sized genome of T. brucei and indicates that both genomes have been shaped by the need for antigenic variation, in that VSGs are the main gene found on most small chromosomes, and a large number of VSGs localize to the subtelomeres of most of the larger chromosomes. However, a deeper analysis reveals several differences, which indicate deviation in the way VSG expression is organized in the two parasites: telomere-to-telomere assembly of all chromosomes confirms previous suggestions (Abbas et al. 2018) that T. congolense does not contain T. brucei-like VSG expression sites; VSG-rich subtelomeric regions of the largest chromosomes in T. congolense are notably smaller than seen in T. brucei; VSG expression in T. congolense is not limited to telomere-adjacent genes, consistent with a lack of subnuclear compartmentalization between chromosome cores and subtelomeres; a single T. congolense chromosome acts as a potential silent reservoir of VSGs; and VSG expression can be detected from multiple small chromosomes. Thus, this genome assembly provides a comprehensive platform to dissect the operation of host immune evasion by VSG expression in T. congolense. Methods Parasite Culture, Generation of Mutant Lines Bloodstream form T. congolense strain IL3000 (Bienen et al. 1991) was used for all experiments (originally isolated at the International Livestock Research Institute (Nairobi, Kenya) and derived from infected mouse blood, received from Theo Baltz, University of Bordeaux). Parasites were cultured at 34 °C with 5% CO2 in modified HMI-93 (Hirumi and Hirumi 1991; Coustou et al. 2010) media, with 10% goat serum (Gibco) and 20 g/L albumax II (Thermo-Fisher). Cells were routinely cultivated in 6 or 24 well plates and a P1000 pipette tip or 5 mL stripette, as appropriate, was used to flush the cells adhered to the bottom of the well prior to passage into a new well, or for hemocytometer counting of cells. For large volumes, 75 cm2 flasks were used, and a 10 mL or 25 mL stripette used to flush the parasites sticking to the surface of the flask. To generate RAD51 (TcIL3000.11.8740) and BRCA2 (TcIL3000_1_180) null mutant parasites, plasmids pJM-RMC-09 (TcoRAD51) and pJM-RMC-10 (TcoBRCA2) were synthesized by swapping the UTRs of RAD51 and BRCA2 into pJM-RMC-03 (BSD-resistance; Addgene plasmid # 248661; https://www.addgene.org/248661/), using restriction enzyme cloning. Briefly, UTRs were PCR-amplified from T. congolense IL3000 gDNA with Phusion polymerase (NEB UK, M0530S; see Table S2 for all primer sequences) and ligated into pGEM-T Easy Vector (Promega UK, A1360). Correct amplification of the UTRs was confirmed by Sanger Sequencing (Eurofins Germany) with primer M13F. To insert the 5′ UTRs, pJM-RMC-03 and both pGEM-T Easy with 5′ UTR plasmids were digested with NotI-HF and HindIII-HF (NEB UK, R3189S and R3104S) and ligated with T4 ligase (Promega UK, M1801). To insert the 3′ UTRs, the resulting plasmids and pGEM-T Easy 3′ UTR plasmids were digested with NsiI-HF and XhoI (NEB UK, R3127S and R0146S) and ligated together. Correct insertions were confirmed by Sanger Sequencing (Eurofins Germany; 5′UTR confirmation: JM015; 3′UTR confirmation: SP6). The hygromycin-resistance plasmids pJM-RMC-19 (RAD51) and pJM-RMC-20 (BRCA2) were generated by restriction enzyme cloning; briefly, plasmids pJM-RMC-09 and pJM-RMC-10 and pyrFEKO-HYG (a gift from George Cross; Addgene plasmid # 24020; https://www.addgene.org/24020/) were digested with BglII and XbaI (NEB UK, R0144S, and R0145S), with the backbones minus antibiotic ligated to the hygromycin resistance gene from pyrFEKO-HYG using T4 ligase. Correct integration of the new antibiotic resistance gene was checked using Sanger Sequencing (primer JM042). To generate the null mutant parasites, the relevant plasmids (pJM-RMC-09, −10, −19 or −20) were linearized using NotI-HF and XhoI (NEB UK). Around 3-4 × 107 T. congolense cells were flushed, counted and spun at 1250 g for 10 min. They were then resuspended in 100 µL Tb transfection buffer (Schumann Burkard et al. 2011), mixed with 10 µg of the relevant digested DNA and electroporated in an Amaxa Nucleofector II (Lonza) using program Z-001. Parasites were resuspended in 20 mL of fresh media and immediately plated out at 1:5, 1:10 or 1:20 dilutions in 24 well plates, and with the remaining pooled cells plated in 6 well plates. After around 18 h, antibiotic selection was commenced, with final concentrations of 0.25 μg/mL hygromycin B (Roche, 10843555001) or 0.5 μg/mL blasticidin (Invivogen, ant-bl-1), depending on the linearized plasmid used. Successful integration of the antibiotic-resistance cassettes was assessed by PCR across the relevant locus: gDNA was extracted using a Qiagen DNeasy Blood and Tissue Kit, and Phusion polymerase used to amplify target fragments, (Fig. S1). RT-qPCR of the double antibiotic-resistant clones confirmed RNA loss: RNA was extracted using a Qiagen RNeasy Kit, cDNA produced using Superscript IV (Thermo Fisher), and an Applied Biosystems 7500 Fast machine used with Fast Sybr Green master mix (Thermo Fisher), all according to the manufacturer's instructions. The lines generated were designated TcoRAD51−/− clone H2B2 and TcoBRCA2−/− clone H4B1. The TcoBRCA2−/− H4B1 line had the antibiotic resistance cassettes removed by the transient expression of pLEW100cre (a gift from John Donelson via George Cross; tryps.rockefeller.edu/Plasmids/pLEW100cre.txt) and induction of Cre-recombinase using 1 µg/ml tetracycline (Merck, T7660) as per Kim et al. (Kim et al. 2013), to generate TcoBRCA2−/− floxed clone 2 (fx2). Parasites were cloned in a 24-well plate (at 1:10 dilution, straight after transfection) and selected using 100 µg/mL ganciclovir (Merck, G2536) 16 h post transfection. Clones were checked for sensitivity to blasticidin and hygromycin. Cloning of Parasites T. congolense IL3000 WT, TcoRAD51−/− clone H2B2 and TcoBRCA2−/− fx2 parasites were recloned, by dilution of each line to 0.5 cells/well in a single 24 well plate. Plates were monitored for growth and at least three clones per line were selected for growth. The starting clones (P0) were stabilated as soon as possible after cloning (once a density of ∼1-2 × 106/mL had been achieved in a well of a 6 well plate; around 22 to 28 generations from a single cell; depending on the line). Stabilates were made using a final concentration of 15% glycerol in HMI-93 media. The three clones were then sub-passaged 2 to 3 times per week, growing in 6 or 12 well plates to a density around 2-3 × 106/mL, before being diluted to around 5 × 104/mL at each passage. After 11 passages, stabilates were made of each clone (P11 samples; ∼80 to 90 further generations from the starting culture), followed by a further 11 passages, and stabilates made of the surviving clones (P22 samples; ∼80 to 90 generations from the P11 samples). Three clones for each line survived to passage 22; two per line were used for further analysis (WT IL3000 clones 1 and 2; TcoRAD51−/− clones 2 and 5; TcoBRCA2−/− clones 1 and 2). Nucleic Acid Extraction and Sequencing Stabilates of each clone at each passage time point were removed from liquid nitrogen and recovered in 6 mL media in one well of a 6 well plate. Once growing, these parasites were transferred to four 75 cm2 filter top culture flasks, with 40 mL HMI-93 media and positioned horizontally to allow attachment of the parasites on the back surface. Parasites were grown to a density of between 2 and 5 × 106/mL, before processing. For RNA extraction, 6 × 107 cells were pelleted from each individual flask (giving 4 replicates), and RNA extracted from each pellet using a Qiagen RNeasy Extraction kit, following manufacturer's instructions, and including the on-column DNase-digestion steps. RNA libraries were prepared from the three samples with the highest RNA concentration using an Illumina TruSeq stranded mRNA prep kit and an Illumina NextSeq2000 was used to perform paired-end sequencing (2 × 100 bp). The libraries were sequenced by the Molecular Analysis facility at MVLS Shared Research Facilities (Glasgow University). For ONT sequencing, high molecular weight genomic DNA was extracted from each of the six clones at both P0 and P22 using the Monarch HMW DNA Extraction Kit for cells and blood (NEB UK), pooled from the same flasks as used for the RNA extractions, with lower density flasks refilled with media and grown for a further 2 to 3 d. Around 1.5–2 ×108 cells were pelleted by centrifugation, washed with PBS, and stored at −80 °C before DNA preparation. During library preparation, ONT Ligation Sequencing Kit (SQK-LSK114) was used as per manufacturer's instructions, and sequencing was performed on a GridION Mk1 device using R10.4.1 MinION flow cells. The same high molecular weight gDNA was also submitted to BGI Genomics for short read sequencing using DNBseq paired-end sequencing (2 × 100 bp), with 30 M reads per sample. Six samples were sequenced: IL3000 WT clone 2 P0 and P22; TcoRAD51−/− clone 5 P0 and P22, and TcoBRCA2−/− clone 1 P0 and P22. For PacBio sequencing, six C57BL/6J mice were infected with WT IL3000. At day 7 post infection, all six were culled and their blood combined. The parasites were purified through a DE52 column (Lanham and Godfrey 1970), washed with PBS-G and pelleted by centrifugation at 600 g for 10 min, washed in PSG buffer, pelleted again and stored at −80 °C. DNA was extracted from thawed pellet using the Monarch HMW DNA Extraction Kit for cells and blood kit, using the manufacturer's instructions. High molecular weight DNA was submitted to the Earlham Institute for library preparation (using Express Template Prep Kit), and sequencing using PacBio Sequel II SMRT cell. For Hi-C, IL3000 WT clone 1 was used, five further passages from the starting clone P0 culture described above (around 40 generations). The Proximo Hi-C (Microbe) Kit from Phase Genomics (Seattle, USA) was used, following the manufacturer's instructions. Two samples were prepared, from 6.5 × 108 cells and 5 × 108 cells respectively. The samples were combined after the DNA purification steps of the protocol and one library made using both. An Illumina NextSeq2000 was used to perform paired-end sequencing (2 × 100 bp), by the Molecular Analysis facility at MVLS Shared Research Facilities (Glasgow University). ONT Data Preprocessing ONT whole genome sequencing data was basecalled using dorado (v.0.9.1, https://github.com/nanoporetech/dorado) duplex sup and samtools (v.1.19.2) (Danecek et al. 2021) was used to split the output reads into simplex, duplex and de-duplexed simplex reads for further processing based on the dx tag value. Chopper (v.0.9.0) (De Coster and Rademakers 2023) was used to remove ONT control sequence (DCS) reads. Dorado correct was used to correct the simplex data, which was used for de novo genome assembly downstream. Genome Assembly and QC A combination of two de novo genome assemblies of T. congolense IL3000 bloodstream-form cells was used to create the genome version presented here. Verkko (v.2.2.1) (Rautiainen et al. 2023) was used to assemble the haplotype-resolved megabase chromosomes of the parasite and hifiasm (v.0.25.0) (Cheng et al. 2026) de novo assembly was used for the smaller chromosomes, as these were not assembled by verkko. For both assemblies, the same sets of data were used: Hi-C reads, dorado-corrected ONT simplex data (provided as “–hifi” in verkko) and uncorrected ONT simplex data (as “–nano” in verkko and “–ul” in hifiasm). A number of other genome assembly approaches were tested with the available data as well (Fig. S2). Before merging the two assemblies, a number of checks were performed on the hifiasm assembly and between the two assemblies to prevent duplicate sequences in the final merged assembly. As our motivation for merging the assemblies related specifically to retaining minichromosomes, and any potential intermediate chromosomes, the hifiasm assembly was filtered to keep only sub-800 kb contigs (as the larger “megabase” chromosomes in this parasite are above this size range). This subset of sequences was initially checked for duplicate sequences within the assembly, followed by hifiasm versus verkko reciprocal comparison, using minimap2 (ava asm5), megablast and nucmer (maxmatch). This approach identified a hifiasm contig mapping to a part of tco08a that was removed; additionally, 6 verkko contigs were also removed prior to merging with the hifiasm assembly as they represented duplicate, mostly fragmented, sequences. For quality control (QC), a number of metrics were used to evaluate the generated genome assemblies (Table S2): BUSCO (v.5.8.3) (Manni et al. 2021) was used to determine and compare genome completeness (euglenozoa_odb10 database used); consensus quality (QV) was assessed using merqury (v.1.3) (Rhie et al. 2020) and meryl (v.1.4.1) (k-mer size 17), and QUAST (v. 5.3.0) (Mikheenko et al. 2023) was used to evaluate general genome metrics (gene detection using glimmer, rRNA detection enabled). Telomere-to-telomere (T2T) assessment was carried out by finding telomeric (“TTAGGG”) repeats using seqtk telo (v.1.4) (https://github.com/lh3/seqtk) and homologous chromosomes (both within an assembly and between assemblies) were identified using nucmer and promer (mummer suite, v. 3.23) (Kurtz et al. 2004) with the “–maxmatch” flag. The same approach (promer with “–maxmatch”) was used, reciprocally, to find synteny with T. brucei and L. major genomes. Genome Assembly Gap Closure Details The verkko genome assembly of the larger 13 chromosomes contained 4 gaps on different contigs (tco01c, tco01d, tco03a, tco04a), and one contig only had one telomeric repeat-capped end (tco03b). 3/4 of the gaps and the missing chromosome end were resolved as a result of the approaches detailed below. tco01c and tco01d. For two copies of tetraploid chromosome 1, the best supporting read evidence suggested these were not genuine gaps, but rather misassembled regions with overlaps in both flanking sequences, as confirmed using ONT read mapping and blastn. In both cases the gap sequence (NNN) and the overlapping sequences flanking it were resolved (de-duplicated) based on the coordinates of blastn output. tco04a. For the gap in chromosome 4, an ungapped version of this chromosome was identified in a previous assembly attempt using hifiasm (assembly “1' in Fig. S2) and confirmed using HiFi reads that showed mapping spanning the whole region, with minimum sequence identity 98% and best mapping in the assembly corresponding to this contig; the region was transposed into tco04a to resolve the gap. tco03b and tco03a. Both copies of chromosome 3 were incomplete: tco03a contained a gap in what was later identified as the splice leader region (SL RNA locus), whereas tco03b was missing a TTAGGG-containing contig end (and was therefore not telomere-to-telomere). For tco03b, an ONT read (9dac27fc-34fe-4e27-a8a6-9a2c9a94ee2f) was identified as containing telomeric repeats and mapping to the corresponding arm of tco03b, and blastn confirmed that this read mapped to this region better than anywhere else in the assembly (based on sequence identity and alignment length). This read was used to complete the tco03b contig. For tco03a, despite extensive attempts, an unresolved gap remains, as no ONT or HiFi reads or, indeed, previous assembly attempts provided enough support for gap closure. Some reads suggested that, similar to tco01c and tco01d, there is no genuine gap at the SL RNA locus; however, the read mapping was conflicting. Thus, we reduced the size of the N-containing region in the tco03a gap but did not resolve it. Genome Annotation Putative genes and pseudogenes were annotated using Companion (https://companion.gla.ac.uk/) (Haese-Hill et al. 2024) with the following settings: reference protein alignment enabled, RATT conserved gene transfer on species level enabled, taxon ID 5654. Repetitive genome elements were identified using Tandem repeats finder (TRF) (v.4.09.). Repeat motifs (369, 32, 70, and 50 bp) were located or searched using fimo of meme suite of tools (v. 5.5.9) (Grant et al. 2011) with the P-value cutoff of 10−9. Putative VSG family genes were identified based on the presence of the Pfam domains and companion gene annotation: VSG genes, Pfam domain PF13206; other VSG family genes, Pfam domain PF00913 (identified in translated predicted gene and pseudogene sequences using hmmsearch (v.3.4) (hmmer.org), and only significant hits were retained. For this, gene and pseudogene DNA sequences were translated using bedtools getfasta (v.2.31.1) (Quinlan 2014) and seqkit translate (v.2.8.2) (Shen et al. 2024). To further profile VSG diversity, hmm profiles of described VSG phylotypes (Silva Pereira et al. 2019) were used to detect phylotype-specific protein signatures in the translated gene and pseudogene sequences (hmmsearch was used, as above, retaining only significant hits). Pfam domain searches across the translated genes and pseudogenes were performed using pfam_scan (https://github.com/aziele/pfam_scan), utilizing InterPro Pfam domain motif data (https://www.ebi.ac.uk/interpro/download/pfam/) and Pfam-defined cutoffs for the individual motifs. Genome-wide sequence composition was assessed using bedtools make windows and bedtools nuc (v.2.31.1) (Quinlan 2014) in 1 kbp windows; this was also used to identify AT-rich repetitive regions found on the larger chromosomes. Chromosome Copy Number Analysis Chromosome copy number was estimated utilizing short read whole genome sequencing data similarly to previously published work (Damasceno et al. 2025). Briefly, paired-end reads were mapped to the genome assembly using bwa mem (v.0.7.19) (https://arxiv.org/abs/1303.3997) and samtools (v.1.22) was used to sort, index and remove duplicate reads from the resulting alignment files. BamCoverage from the deepTools package (v.3.5.5) (Ramirez et al. 2016) was used to calculate read depth across the genome (normalization using RPKM, 50 bp bins, reads centered). The resulting bedgraph files were used in a custom python script that extracts coverage across all annotated genes and pseudogenes and calculates the mean coverage per gene; this was then normalized against the median coverage of all genes in the genome. Per-chromosome copy number was computed as the median of these normalized values for a given chromosome. RNAseq Analysis For analysis of VSG expression, short-read Illumina RNA-seq was performed from one starting (P0) clone (see above; i.e. from a population that had been grown for ∼22 to 28 generations after cloning). RNAseq sequencing data was trimmed using Trim Galore! (v.0.6.10, paired end settings, fastQC output enabled) (https://github.com/FelixKrueger/TrimGalore) and salmon quant (v.1.10.3., automatic library type detection enabled) (Patro et al. 2017) was used to quantify the RNAseq data in transcripts per million; for running salmon quant, salmon index was run in decoy-aware mode as recommended by its developers, by running the generateDecoyTranscriptome.sh script. Both genes and pseudogenes, as predicted using Companion annotation, were used as the transcriptome. Hi-C Data Analysis HiCExplorer (v.3.7.6) (Wolff et al. 2020) was used to analyse the Hi-C data, as per the developers’ recommendations (https://hicexplorer.readthedocs.io/en/latest/content/example_usage.html). Briefly, the genome was split into two haplotypes (A and B), which were analysed separately; bwa mem with the recommended settings was used to map the reads (-A 1 -B 4 -E 50 -L 0) and samtools was used to change the output file format to bam. Following this, hicBuildMatrix was used to generate matrices at 5 kbp resolution, and was subsequently merged using hicMergeMatrixBins to generate 10 kbp, 25 kbp, 50 kbp, and 100 kbp resolution matrices. HicCorrectMatrix was used to remove outlier values (ICE, thresholds used –2.0, 5.0). HiCExplorer hicPCA was used to generate Pearson and Obs/Exp matrices, as well as retrieve eigenvectors 1 to 10 across the 5 resolution corrected matrices. Topologically associated domains (TADs) and genome-wide insulation score were called using hicFindTADs (FDR correction, –minDepth 3 bins, –maxDepth 10 bins) and hicPlotDistVsCounts was used to investigate the distribution of contacts with distance between the two haplotypes at 25 kbp, 50 kbp and 100 kbp resolutions. Supplementary Material