Genome Sequence Data Reveal Complex and Variable Ploidy in the Amoebozoan Acanthamoeba castellanii

Abstract

Acanthamoeba castellanii is a free-living amoeba that is emerging as a model organism for the study of eukaryotic microbiology. This species is one of the most widely studied members of the Amoebozoa and is both an important grazer in soil communities and an opportunistic human pathogen; A. castellanii is thus of evolutionary, ecological, and biomedical significance. Despite its potential as a lab workhorse, the genome biology of A. castellanii is complex and poorly understood. Polyploidy is a common feature of many amoebozoan genomes, and members of the genus Acanthamoeba are no exception; they appear to be not only polyploid, where genome copy number is inflated beyond the conventional haploid and diploid states, but also aneuploid, i.e. with inter-chromosomal copy number variation. To better understand aneuploidy in A. castellanii and how it may vary over time and between closely related strains, we analyzed nanopore and Illumina sequence datasets from several wild-type and mutant A. castellanii lines, with a focus on quantifying single nucleotide polymorphism and structural variant allele frequencies across chromosome-scale scaffolds. Sequence depth of coverage was also considered. Our findings suggest that intragenomic chromosome copy number is highly variable in Acanthamoeba, consistent with a considerable degree of aneuploidy, but is predominantly stable in laboratory culture and on the evolutionary scale of the species or genus.

Significance.

Acanthamoeba castellanii is becoming an important model organism for basic and applied research. However, its apparent polyploidy and aneuploidy has the potential to complicate the interpretation of results that depend on knowledge of gene copy number. In this study, we reveal the complex nature of ploidy in this organism by analyzing long- and short-read sequence data. Our results provide a reference point against which genomic and experimental data from A. castellanii can be interpreted, and guide future efforts aimed at more precisely characterizing how the organism regulates its genome and chromosome copy number.

Introduction

Acanthamoeba castellanii is a phagotrophic and free-living amoeba, commonly found in soil and freshwater environments (Neff 1957; Page 1967). A. castellanii is well-known both for its propensity to host bacterial endosymbionts and viruses (Hall and Voelz 1985; Horn and Wagner 2004; Schmitz-Esser et al. 2008; Iovieno et al. 2010) and as an opportunistic pathogen of humans (Lorenzo-Morales et al. 2015). For various biological and experimental reasons, it is favored as a model organism by molecular and cell biology researchers. However, aspects of the nuclear biology of A. castellanii are complex and poorly understood, most notably its chromosome number and ploidy. To address this, several approaches have been employed. These include using cytological methods to count chromosomes (Pussard 1972), measuring nuclear DNA content (Byers 1986), and pulsed-field gel electrophoresis (PFGE) to explore A. castellanii karyotypes (Rimm et al. 1988; Matsunaga et al. 1998). From these experiments, relatively little was learned to any degree of certainty. From nuclear DNA content, Byers (1986) was able to infer the now-widely-cited estimate of 25n for A. castellanii's ploidy. The PFGE experiments were able to place a lower bound of 17 distinct chromosomes for the Neff strain (Rimm et al. 1988) and showed the karyotypes of 28 Acanthamoeba strains across 12 species to be heterogeneous, even within species (Matsunaga et al. 1998). Based on variable band intensities, the authors proposed that Acanthamoeba genomes are aneuploid.

A draft genome assembly for A. castellanii was published in 2013 (Clarke et al. 2013), but while useful for gene discovery, it was produced using Illumina short-read technology and thus was highly fragmented. More recently, a combination of nanopore long-read and Hi-C technology allowed near-complete resolution of all chromosomes from two Acanthamoeba strains into scaffolds (Matthey-Doret et al. 2022), which makes it possible to revisit the question of ploidy in this organism. These assembled chromosomes serve as a predicted karyotype to a level of detail that has thus far been impossible to discern in PFGE experiments, while the amplification-free library preparation method allows relative quantitation of the chromosomes.

With high throughput sequencing, data on allele frequencies across chromosomes can also be collected. Long-read data provide two independent types of allele frequency data, the more common single nucleotide polymorphism (SNP) allele frequencies and structural variant allele frequencies. Short-read data provide an additional set of SNP frequencies for cross-referencing. Here we sought to shed light on the genome organization and ploidy of Acanthamoeba using the rich sequence data sets underlying the long-read, chromosome-scale assemblies of the wild-type Neff and C3 strains (Matthey-Doret et al. 2022) and transformed clones sequenced to study the fate of artificial transgenes (Colp et al. 2025).

Results

Long-read Allele Frequency Analysis

Using the Acanthamoeba castellanii strain Neff assembly (Matthey-Doret et al. 2022) ploidy was first explored with ploidyNGS (Augusto Corrêa dos Santos et al. 2017) v3.0.0. This program plots a histogram of the absolute counts for each possible allele frequency (in steps of 0.01) detected in sequencing read sets mapped to a reference genome, and the overall shape of the histogram can be used to infer ploidy. PloidyFrost (Sun et al. 2023) was not employed due to the fact that the upper limit of utility is stated to be octoploid (the software was also not compatible with our computing environment).

When the A. castellanii strain Neff genome was analyzed with ploidyNGS, a very high level of background noise was observed across the possible range of allele frequencies and there were no clear peaks (Fig. S1). Given the possibility that this complex pattern was due to aneuploidy, the analysis was repeated by dividing the assembly into its constituent scaffolds and performing the same mapping and plotting analyses for each scaffold individually. This approach produced interpretable but variable signals across the chromosome-sized scaffolds (Fig. 1).

Fig. 1.

PloidyNGS SNP frequency plots from the 12 longest scaffolds (chromosomes) of Acanthamoeba castellanii strain Neff. Plots were generated using nanopore reads mapped to individual reference genome scaffolds with minimap2. Mapping coverage was capped at 200×. The ploidy signal is clearly varied across these 12 scaffolds from the same strain. a-l) Scaffolds 1 to 12, respectively. Scaffold lengths are indicated in the respective plot titles.

Interpretation of these data, including the types to be introduced below, requires a distinction between “ploidy signal” and actual ploidy. Here ploidy signal refers to the lowest ploidy level that would be consistent with the observed allele frequency data. For example, peaks at allele frequencies 0, 0.5, and 1 are consistent with diploidy, but it is formally possible that two identical sets of five chromosomes (10 in total) could yield this same signal. This is important because we do not have absolute quantitation of any of the A. castellanii chromosomes and must default to identifying the lowest ploidy level consistent with the plots as the “ploidy signal”, while recognizing the possibility that the true copy number is a multiple of the base number. This is also why the ploidy levels inferred here are lower than those of Byers (1986) who inferred a global ploidy of 25n from nuclear DNA content.

Figure 1 shows ploidy signal across the 12 largest scaffolds of the wild-type Neff assembly (ranging from 1.47 to 2.54 Mbp in size). Real data do not always cleanly conform to the above-mentioned patterns expected for different ploidy levels, but most of these plots can be interpreted. For example, the plots for scaffolds 1, 2, 3, 5, 7, and 8 (Fig. 1a-c, e, g, and h) appear to represent haploid signal, with peaks near the extremes of the plot but not between them. The peaks for Scaffold 1, 5, 7, and 8 (Fig. 1a, e, g, and h) do tail out further toward the middle of the allele frequency range, but no distinct peaks can be observed along those curves that would change the interpretation.

The plots for Scaffolds 4, 9, 10, and 12 (Fig. 1d i, j, and l) all closely resemble what one would expect for a diploid chromosome, with peaks at the extremes and a smooth peak centering around an allele frequency of 0.5. The plots for Scaffolds 6 and 11 are more difficult to interpret (Fig. 1f and k). Here there are allele frequencies represented somewhere between the two peaks at the extremes, but these additional peaks are not well resolved. For Scaffold 11, the peaks from the extremes of the plot tail down toward the centre, and there is a small inflection in the corresponding curve around allele frequencies 0.33 and 0.67 which could represent a triploid signal. The plot for Scaffold 6 is the least clear; here there could be additional peaks near 0.33 and 0.67 once again, or potentially 0.4 and 0.6 which may correspond to pentaploidy. Regardless of data interpretation, these examples illustrate how to infer ploidy signal in general and across different scaffolds within a genome. In this case, they are consistent with aneuploidy in Acanthamoeba.

Having found support in the long-read sequencing data for aneuploidy in the wild-type A. castellanii Neff genome, other independent read sets from different cultures were analyzed in the same way to ask several related evolutionary questions. These additional cultures included (i) another wild-type culture of Neff, maintained and sequenced separately by collaborators at the Institut Pasteur, (ii) a wild-type culture of A. castellanii C3, also maintained and sequenced at the Institut Pasteur, and (iii) four monoclonal isolates of Neff derived from wild-type cultures in the Archibald Lab at Dalhousie University, transformed by the same plasmid, and then individually bottlenecked into clonal cultures. Three of these were transformed with linearized pGAPDH-EGFP in the same experiment, while one was transformed with a circular form of the same plasmid in a separate experiment (Colp et al. 2025). Interestingly, the results showed variation in the inferred ploidy profiles across all chromosome-level scaffolds within and between the different datasets.

Two additional lines of evidence support the allele frequency-based inferences from the long-read data: Illumina-based ploidyNGS allele frequency plots and structural variant allele frequency information from long-read data, defined in this case as indels of at least 50 bp. When visualizing long reads mapped to the A. castellanii reference genome assemblies, clear instances of structural variation could be observed in the mapped reads and coverage plots, where subsets of reads were missing segments of DNA ranging from a hundred to a few thousand base pairs, with an associated, cleanly delineated drop in depth of coverage. Importantly, structural variants on the same chromosome were found to share approximately the same allele frequency (where alleles correspond to the presence or absence of the indel) when they came from the same dataset, suggesting that at least some of the signal is biological. The long-read structural variant calling program Sniffles 2 (Smolka et al. 2024) was used to make structural variant calls on all chromosomes independently using each distinct long-read dataset, such that each SNP-based ploidyNGS plot across the range of chromosomes and datasets had a corresponding structural variant-based Sniffles 2 histogram of allele frequencies (Fig. S2; the process of reconciling these different data types is described in the Supplementary Material under the Results heading).

Short-read SNP Frequency Analysis

To generate the ploidyNGS plots using Illumina data, the same general approach was used as for the long reads, with appropriate adjustments such as the specific choice of read mapping program. For a subset of the isolates, this meant the possibility of comparing three different sources of allele frequency information to arrive at a consensus best inference for each chromosome in each of those isolates. Given the noise and/or uncertainty in some of these plots by themselves, having multiple independent lines of evidence proved valuable.

Depth of Coverage Analysis

Ploidy inferences based on allele frequency-based analysis were augmented by consideration of sequencing coverage depth across all strains, using the same read sets as previously described. The approach employed by Rogers et al. (2011) was used here to summarize depth of coverage across the scaffolds, with the median depth across all nucleotide positions being reported for each scaffold. All absolute coverage data are plotted in Figs. S3–S9. The coverage depth was then normalized to allow meaningful ploidy interpretation. This was done to be able to place the read depth information on a scale useful for ploidy inference, while also permitting comparisons of read depth across isolates and between sequencing technologies. While most strains had their chromosome with the least sequence depth of coverage set as tetraploid for normalization (see Materials & Methods), the least-covered chromosome was set as pentaploid for LT9 because this was the lowest ploidy level that could not be rejected by cross-referencing with allele frequency data. A summary plot of normalized coverage data is shown in Fig. S10. The “ploidy signal” caveat made above for allele frequency data also applies here, but in a slightly different way; coverage depth analysis allows for relative quantitation of the chromosomes, but it is formally possible that the true copy number is a multiple of the inferences made here.

Even when normalized, the depth of coverage data are difficult to interpret alone, but considered together with allele frequency data, this information greatly facilitated ploidy inference. When one of these two data types yielded an ambiguous result, the other constrained the range of possibilities enough to allow ploidy level to be inferred for a given scaffold in each strain. The primary goal was not to precisely define the ploidy of each chromosome across all isolates, but to assess the variability in ploidy signal across chromosomes and isolates. The ploidy signals were inferred and recorded based on all available information but not necessarily unambiguously recorded for any portion of the data individually; this is because it was sometimes necessary to cross-reference up to all four different types of data to arrive at a reasonable inference. Accordingly, we considered the normalized sequence coverage as a useful summary of the data. While the normalized coverage values and the inferred ploidy levels are presented in similar formats, we were careful not to directly compare them and risk placing undue emphasis on the weight of normalized coverage depth relative to the other three data types.

Based on the four different lines of evidence, inferences of the ploidy signal were compiled for the 30 largest chromosomes across all A. castellanii Neff isolates, recognizing that some had more information available than others. Only the 30 largest scaffolds were considered because for the five smallest chromosome-scale scaffolds, the plots were too sparsely populated to discern meaningful patterns. For the C3 strain, Circos plot data generated previously (Matthey-Doret et al. 2022) were used to determine which scaffolds were homologous to the 30 largest Neff scaffolds. In many cases this could not be determined unambiguously due to apparent or suspected chromosomal rearrangements disrupting chromosome-wide synteny. In these cases, longer alignments were prioritized when matching scaffolds between the strains. The goal of this matching was to assess whether homologous sequences had retained similar behavior with respect to ploidy between the two strains, but sequences were not further broken down into sub-sequences to compare ploidy signal because their replication and therefore copy number cannot be separated from the rest of the sequence to which they are physically joined. As a result, the C3 and Neff comparison broadly respects homology with some exceptions.

Ploidy inferences for all A. castellanii isolates considered in this study are summarized in a clustered column graph where each cluster represents a chromosome (Fig. 2). While the data are imperfect, noticeable consistency across the (wild-type and mutant) Neff cultures was observed. For each scaffold, a single inferred ploidy level—or more commonly a narrow range of levels—was shared by all examined strains, giving rise to an overarching pattern of ploidy signal variation across all scaffolds.

Fig. 2.

Inferred ploidy levels for the 30 largest scaffolds (chromosomes) in multiple isolates of A. castellanii strain Neff, and the homologous scaffolds in A. castellanii strain C3. Two separate wild-type (WT) cultures of A. castellanii strain Neff were examined (“WT Dal” and “WT Pasteur”), as well as three clonal Neff cultures transformed with linearized pGAPDH-EGFP (“LT6”, “LT8”, and “LT9”), and one clonal culture transformed with circular pGAPDH-EGFP (“Clone 1”). Wild-type A. castellanii strain C3 was also included. Ploidy levels were inferred based on nanopore- and Illumina-based SNP frequency plots made with ploidyNGS, nanopore-based structural variant allele frequency plots, and median depth of coverage information from both nanopore and Illumina data. The C3 data are plotted here alongside the Neff chromosomes to which they are homologous based on whole genome alignment (Matthey-Doret et al. 2022) rather than according to their own chromosome numbering. The Neff cluster and single C3 column are set apart for clarity.

Across the examined strains, ploidy inferences were not necessarily congruent with the majority for every scaffold, but in some strains they consistently were. Specifically, the wild-type Neff culture from Dalhousie and Clones LT6 and LT8 can all frequently be observed to align with the majority when inspecting each cluster of columns in Fig. 2, and when they do deviate it is by a minimal amount. Others, such as C3, the wild-type Neff culture from the Institut Pasteur, and Clone LT9, each independently show more deviation while Clone 1 falls somewhere in the middle of the two extremes in this respect. Those that deviate more nevertheless tracked the same ploidy level trends across their respective chromosomes as the more congruent strains—they did so either at a higher baseline ploidy level and/or with more variation from the others. Only Clone LT9 showed a higher baseline for inferred ploidy, while ploidy levels in C3 and wild-type Neff from the Institut Pasteur simply varied further from the central tendency among all isolates in both directions. It is unclear to what extent the higher inferred ploidy of many LT9 chromosomes (as apparent in Fig. 2) can be explained by setting a pentaploid baseline for normalization. When comparing the four data types for all scaffolds from this isolate, there is no obvious reason to doubt this choice, so Clone LT9 may well have higher chromosome copy numbers than the other isolates across the board. However, if the pentaploid reference point was set in error or Scaffold 27, the one used as a reference point, was not representative of either its true relative abundance or of its allele frequencies, it would in fact manifest as inflated ploidy level inferences for all chromosomes, similar to what can be observed in Fig. 2.

There were also some scaffolds that demonstrated considerably less inter-strain diversity in ploidy level, most notably scaffolds 5, 21, and 23, and others that varied quite a bit across isolates, such as scaffolds 4, 12, 14, 25, 26, 29, and 30 (Fig. 2). It is unclear to what extent ploidy level inference is more variable in the latter four of these examples due to being shorter and therefore having a smaller sample of heterogeneous sites carrying ploidy signal.

Discussion

The goal of this study was to use a wealth of new sequence data from multiple Acanthamoeba castellanii isolates to improve our understanding of the organism's ploidy. Our results do not provide absolute quantification of the chromosomal copy number, but they do reveal patterns of variation in effective ploidy signal across isolates, as well as between chromosomes within a given isolate. This information helps provide bounds on the range of possible ploidy signals for individual A. castellanii chromosomes, as well as insight into how such levels can vary over laboratory and evolutionary timescales.

The SNP- and indel-based approach employed herein was not without limitations, even when cross-referenced against depth of coverage data. First, it is important to note that all these results stem from amoeba cultures that have been growing in laboratory conditions for hundreds of generations. Such artificial conditions are a known risk factor for inducing artifacts into evolutionary studies. Future investigations applying our approach to environmental isolates could help disentangle genuine evolutionary patterns in this species and genus from artifacts associated with extended propagation in the lab. In any case, resolving peaks on the SNP frequency plots proved challenging, and as the signal in the data moves toward a higher ploidy level, it becomes difficult to distinguish heterozygous alleles with a 1:n-1 ratio from homozygous alleles. As a result, it can be difficult to make precise inferences at any ploidy level. Even apparently haploid and diploid signals, which one might expect to be easiest to identify, require careful assessment. This is exemplified in Fig. S2c (Scaffold 9 of A. castellanii strain Neff, Clone LT9) in the case of apparent diploid signal. Apparent haploidy, meanwhile, could appear when analyzing polysomic chromosomes that are too homogeneous to derive any ploidy signal. Ploidy levels may also be systematically underestimated because without a high degree of structure in the allele frequency plots, estimating the signal will tend toward less granular interpretations of exactly where peaks appear. For example, without extremely narrow peaks, an actual allele frequency of 0.083, consistent with alleles in a 1:11 ratio across 12 chromosome copies, could be largely indistinguishable from an allele frequency of 0.1, consistent with alleles in a 1:9 ratio across 10 chromosome copies. In this case, one would likely tend toward the more “conservative” inference of 10 chromosome copies.

Another issue is that the three wild-type datasets are not from clonal isolates, so inferences may be less precise due to mixed signals in the data from diverging clones within the same culture. Importantly, each of the transformed Neff cultures analyzed herein was established from a single cell (Colp et al. 2025) and the data from these clonal isolates still suggest a complex ploidy state. This result supports the idea that this complexity is not simply an artifact of within-culture clonal heterogeneity; at the time the clonal isolates were established, any initial heterogeneity would have necessarily been intragenomic because these isolates were derived from a single cell. However, it is possible that additional heterogeneity arose within these cultures after their establishment. Even though we cannot pinpoint ploidy levels precisely, there is still value in considering overall trends in the data, and what those trends tell us about the dynamic nature of ploidy in A. castellanii.

We inferred as many as ten unique ploidy levels across the 30 largest scaffolds of individual isolates (Fig. 2), and a lack of peak resolution in the allele frequency plots may be masking even higher numbers. Inferences ranging from tetraploid to tetradecaploid (14-ploid) signal were made across all the isolates and chromosomes investigated. This speaks to the existence of considerable variability in chromosomal copy number. If we accept that there genuinely is a substantial amount of variation in copy number across the chromosomes of any given A. castellanii cell, then it is worth considering what the biological advantage(s) of such variation might be. It is possible that gene dosage plays a role (see below). Environmental factors may also influence the generation of aneuploidy. For example, in various fungi, exposure to antifungal compounds and other stresses such as nutrient scarcity can result in karyotype changes, presumably as part of an adaptive strategy (Gilchrist and Stelkens 2019; Tsai and Nelliat 2019). Host colonization or infection has also been observed to have effects on the fungal karyotype (Kakade et al. 2023). As a ubiquitous microbe, A. castellanii may well encounter similar environmental stresses, and could conceivably vary its karyotype in response. The organism's facultative pathogenicity may also induce infection-related karyotype changes. And as noted, laboratory culturing constitutes a dramatic shift in environment with additional or diminished stressors, as well as smaller population sizes that are subjected to frequent bottlenecking. Some or all of these factors could have contributed to the results observed here on our cultured isolates.

The simplest explanation for the varied copy number is that the differential amplification of different chromosomes is stochastic and of no overarching importance. However, in this case, there must be some short- to medium-term stability with respect to genome contraction and amplification, because across the isolates studied here, ploidy inferences were largely consistent for any given chromosome and for the overall karyotype (Fig. 2). It is unclear the extent to which minor deviations shown by some isolates for each chromosome are biological in nature or artifacts of the noisy nature of the data; the lack of a systematic way to interpret these data is a confounding factor. Further insight into the minor variations embedded within the overall trend will require a more precise approach to ploidy detection.

At first glance, there appears to be some correlation between evolutionary relatedness and ploidy signal, as careful inspection of the data in Fig. 2 reveals that strain C3 is qualitatively more different from the Neff isolates than they are from one another. That said, on a chromosome-by-chromosome basis, the inferred ploidy signals in C3 still fall well within the range collectively demonstrated by the various Neff isolates. To some extent, it also appears that the range of variation in ploidy signal across the C3 chromosomes is similar to the variation within any given Neff isolate, as is the pattern of which chromosomes are more or less abundant. However, the actual ploidy values deviate more between C3 and the ensemble of Neff isolates than they do among the respective Neff isolates. This suggests that the divergence between the karyotype of two strains is correlated with their evolutionary distance. The genetic divergence between the Neff isolates studied here, accrued over months-to-years in laboratory culture, is considerably smaller than the distance between all the Neff strains and C3, but is arguably still real. The approximately one-day generation time of A. castellanii in optimal culture conditions and the collective influence of population size and frequent bottlenecking can serve to accelerate evolutionary processes. Accordingly, we observed ploidy variation among Neff isolates that had only begun to diverge from the beginning of the experiment, albeit with less variation among them than to the C3 strain, which stems from a separate environmental isolate.

The parasitic kinetoplastid genus Leishmania is an informative example of how we might think about Acanthamoeba aneuploidy moving forward. In Leishmania it is thought that aneuploidy is a constitutive feature, but the genome is fundamentally 2n (Sterkers et al. 2011). In most Leishmania species where aneuploidy has been studied, including L. major, L. donovani, and L. amazonensis, the range of chromosome copy number is much more restricted than in A. castellanii (Sterkers et al. 2011; Dumetz et al. 2017; Valdivia et al. 2017; Patino et al. 2020; Black et al. 2023). Leishmania chromosomes are most often disomic or trisomic, with occasional monosomy and one permanently tetrasomic chromosome. Aneuploidy in Leishmania is generally referred to as “mosaic aneuploidy”, referring to the fact that a given population can have a diversity of karyotypes represented at any given time. Some data suggest that upon founding of a population, there is an initial diversification of karyotypes followed by a shift toward a subset becoming more dominant (Black et al. 2023). A correlation has been reported between gene expression and chromosome copy number in Leishmania donovani; under drug treatment or environmental stress, there appears to be an increase in copy number of the chromosomes bearing the genes that are upregulated in response to the change in conditions (Dumetz et al. 2017). Perhaps a similar process explains, at least in part, the observed aneuploidy in A. castellanii. However, it is worth noting that this observation in L. donovani involves only a few chromosomes in the aforementioned stress response, while A. castellanii appears to exhibit considerable variation in copy number across all chromosomes.

One study of aneuploidy in L. major used fluorescent in situ hybridization (FISH) to study the karyotype dynamics of individual cells directly (Sterkers et al. 2011). Aneuploidy was seen to be variable among clones and strains, and even after isolating a clone, mosaicism reappeared within the population, although the ploidy state generally resembled the parent strain. If aneuploidy functions similarly in Acanthamoeba, this could suggest that even in the clonal isolates analyzed here, which showed a less convoluted signal than the wild-type strains that were not bottlenecked, there could be a small amount of mosaicism emerging that was not strong enough to be detected by the methods employed herein.

In their FISH study of L. major, Sterkers et al. (2011) also observed a high rate of asymmetric chromosome allocation during mitosis, where, for example, three copies of a given chromosome were allocated to one daughter cell, and two copies to another. They hypothesized that this resulted from poor control of DNA replication, where one chromosome might be replicated an extra time, or one may fail to be replicated, resulting in one daughter cell receiving an extra copy or lacking a copy, respectively. Such a process could be at least partly responsible for generating aneuploidy in A. castellanii. If the replication defects were infrequent, it would also explain the apparent short-term stability of the ploidy state. In L. major, it was generally observed that chromosomes only reached trisomy, that is, three copies of a given chromosome, with very rare tetrasomy, while in Acanthamoeba, higher copy numbers are inferred herein for many chromosomes. To achieve this result, presumably the asymmetric replication and segregation of chromosomes would have to persist over more generations and/or be biased toward replication of extra chromosome copies rather than failure to replicate existing ones.

Overall, the aneuploidy studies of Leishmania serve as a useful framework for future studies aimed at understanding Acanthamoeba aneuploidy. Establishing reliable FISH protocols in Acanthamoeba would provide a very powerful tool to make more sense of its ploidy state and its dynamic nature.

Conclusions

We have capitalized on a wealth of sequence data for strains of A. castellanii to infer appreciable variation in ploidy across the chromosomes within a given clone, as well as across clones. Although we were unable to address the question of just how polyploid Acanthamoeba may be, we did discover what appears to be prominent and dynamic aneuploidy. There is still much to learn about aneuploidy and polyploidy in this organism, and how it manages the varying state of its ploidy. Comparing these observations in Acanthamoeba with those gleaned from Leishmania and other microbial eukaryotes will hopefully yield ideas for experiments aimed at testing specific hypotheses about the precise mechanisms that generate aneuploidy in this important yet enigmatic amoeba.

Materials & Methods

We analyzed nanopore sequence data generated by two recent genomic studies on A. castellanii (Matthey-Doret et al. 2022; Colp et al. 2025). This includes multiple nanopore read sets of the wild-type A. castellanii Neff strain and a read set from the C3 strain. Sequence data from one Neff clonal isolate transformed with circular pGAPDH-EGFP, and three clonal isolates transformed with linearized pGAPDH-EGFP were also analyzed (see below for details on transformation).

Culturing Conditions

Culturing of the wild-type and mutant strains of A. castellanii were described in Matthey-Doret et al. (2022) and Colp et al. (2025). A. castellanii strain C3 and the strain Neff culture from the Institut Pasteur were grown on amoeba culture medium (2% Bacto tryptone, 0.1% sodium citrate, 0.1% yeast extract), supplemented with 0.1 M glucose, 0.1 mM CaCl₂, 2.5 mM KH₂PO₄, 4 mM MgSO₄, 2.5 mM Na₂HPO₄, and 0.05 mM Fe₄ (P₂O₇)₃ at 20 °C. All cultures of strain Neff from Dalhousie University were grown at room temp in Neff base medium with additives (ATCC Medium 712; 0.75% yeast extract, 0.75% proteose peptone, 2 mM KH₂PO₄, 1 mM MgSO₄, 1.5% glucose, 0.1 mM ferric citrate, 0.05 mM CaCl₂, 1 µg/mL thiamine, 0.2 µg/mL D-biotin, and 1 ng/mL vitamin B₁₂). After transformation, cultures underwent selection with 10 µg/mL G418 until stable and clearly growing, followed by 50 µg/mL G418 for routine maintenance.

The method developed by Peng et al. (2005) was used for transformation with the plasmid pGAPDH-EGFP (Bateman 2010). Detailed information on the transformation protocol and treatment of all clonal isolates is available online at protocols.io (Colp 2018). Briefly, the method is based on the QIAGEN SuperFect reagent and requires the addition of 8 to 12 μg of plasmid per million cells. This was done with circular plasmid for the transformation that gave rise to “Clone 1” and with linearized plasmid for the transformation yielding clones “LT6”, “LT8”, and “LT9”. Once the transformed cultures had recovered and the populations stabilized, clonal isolates were established by picking single cells.

For Neff Clone 1, exploratory experiments were done on a mixed population of transformants; the mixed transformant culture had undergone approximately 15 passages at roughly 4-week intervals before individual clones were isolated, and Clone 1 itself underwent approximately 10 more passages at 4-week intervals between its isolation and the generation of the sequence data presented here. In the case of clones LT6, LT8, and LT9, single-cell isolates were established immediately after the culture had recovered from transformation. They then each underwent three to five culture passages at roughly 4-week intervals between isolation and the generation of the sequence data presented here.

Inferring SNP Allele Frequencies From Long- and Short-Read Sequence Data

Wild-type Neff nanopore reads were mapped against the entire wild-type Neff assembly with minimap2 (Li 2018) v2.24 and allele frequencies of SNPs were plotted using ploidyNGS (Augusto Corrêa dos Santos et al. 2017) v3.0.0. The assembly was then divided into its individual scaffolds, and the wild-type nanopore reads generated in the Archibald Lab were mapped against each of the 35 largest scaffolds independently. PloidyNGS plots were then generated for each scaffold, of which only the 30 largest were sufficiently populated to be selected for analysis. Read mapping and plotting were performed as above. The ploidy-guessing feature of ploidyNGS failed to output meaningful information for A. castellanii, presumably due to the noisy and complex nature of the data.

To explore the stability of the observed patterns over laboratory timescales (i.e. between wild-type Neff and transformed clonal isolates) and evolutionary timescales (between Neff and C3), nanopore reads from “Clone 1”, “Clone LT6”, “Clone LT8”, “Clone LT9”, and the Institut Pasteur wild-type Neff culture were also mapped against the separated wild-type Neff scaffolds and ploidyNGS plots were generated for each scaffold using the reads from each of these data sets. The reads from the Institut Pasteur C3 culture were mapped against the C3 wild-type genome assembly instead, but the 35 largest scaffolds were still analyzed independently. For C3, 35 scaffolds were analyzed to completely account for homology to the 30 largest Neff scaffolds.

Illumina short reads were previously generated for a subset of the isolates mentioned above for the purposes of polishing nanopore data (Matthey-Doret et al. 2022; Colp et al. 2025). This includes Illumina reads from the Archibald Lab and the Institut Pasteur for the wild-type Neff culture, and from the Institut Pasteur for the C3 culture. Among the transformed cultures, “Clone LT6” and “Clone LT9” were sequenced with Illumina. For these isolates, Illumina reads were mapped against the reference genome using HISAT2 (Kim et al. 2019) v2.2.1 and SNP allele frequencies were plotted with ploidyNGS v3.0.0. As before, all read sets were mapped against the wild-type Neff assembly, except for C3 reads which were mapped against the C3 assembly.

Structural Variant Calling and Allele Frequency Plotting From Nanopore Reads

The structural variant caller Sniffles 2 (Smolka et al. 2024) was used to detect structural variants in the nanopore read mapping data described above. Sniffles was configured to detect insertions or deletions of at least 50 bp, and this was also how we defined “structural variants” for the purposes of this study. Sniffles plot was used to visually summarize variant calling data, including plotting allele frequencies of structural variants.

Depth of Coverage Analysis

Long (nanopore) reads were mapped for all isolates to their respective reference genome (the wild-type Neff reference sequence for all except the C3 reads) using minimap2. For those isolates where Illumina data existed, short (Illumina) reads were mapped to their respective reference genomes using HISAT2. It was not necessary to separate the scaffolds for this analysis; while ploidyNGS collapsed data from all scaffolds into the same plot, depth of coverage can be parsed on a scaffold-by-scaffold basis from read mapping data by default. From the mapped reads, per-site read depths were reported by samtools depth and the median was calculated for each scaffold within each isolate.

Next, the absolute read depth values for each scaffold in each isolate were normalized to represent multiples of the putative haploid coverage. Tying coverage to real quantities (chromosome copy number) served to place the values on a scale more relevant to ploidy inference. The goal was to reduce the read depth values to the lowest terms that were compatible with the allele frequency information. To this end, the first step was to identify the lowest coverage scaffolds in each isolate and determine the lowest ploidy level they could be assigned without being clearly rejected by the allele frequency data for those particular scaffolds. Then, their coverage was divided by this assigned ploidy level to determine an estimated haploid coverage value, and the coverage of all other scaffolds was divided by the estimated haploid coverage. Scaffold 27 worked well for this in the first isolate, the wild-type Neff culture from Dalhousie, so it was also used in all other isolates where its coverage was consistent with having the lowest or joint-lowest ploidy level. This was not the case in the wild-type Neff culture from the Institut Pasteur so Scaffold 2 was used instead.

The normalized coverage data and allele frequency data of all types were then considered together to make ploidy inferences. Importantly, both normalized coverage and signal from allele frequency plots were treated as two separate hypotheses for explaining the apparent karyotypes of these isolates, while treating neither as representing a definitive answer. We explored plausible alternative interpretations of the data in both cases until they were consistent with one another. Normalized coverage estimates of scaffolds (other than the aforementioned reference scaffold) were sometimes rejected based on allele frequency data, but multiples of these estimates were supported. In those cases, to respect the relative abundances of all chromosomes, the entire set of normalized coverage estimates were multiplied by the same factor. For example, this could mean changing the lowest inferred ploidy in an isolate from diploid to tetraploid and doubling all other inferences if another of its chromosomes could not be assigned as tetraploid but was consistent with octoploidy. Conversely, a hypothetical chromosome strongly appearing to be diploid from allele frequency data alone was not inferred to be as such if it was covered more than twice as deeply as a clearly triploid chromosome. Most isolates were normalized based on an inferred lowest ploidy level of tetraploid, but for Clone LT9 this could not be reconciled with the allele frequency data and a lowest inferred ploidy level of pentaploid was used instead.

Supplementary Material

Supplementary material is available at Genome Biology and Evolution online.

Author Contributions

M.J.C. and J.M.A. designed the study. M.J.C. collected and analyzed the data. M.J.C. and J.M.A. interpreted the results. M.J.C. wrote the manuscript with revisions by J.M.A.

Funding

This research was supported by a grant from the Gordon and Betty Moore Foundation (GBMF5782). M.J.C. was supported by graduate student scholarships from the Natural Sciences and Engineering Research Council of Canada and Dalhousie University.

Data Availability

The sequencing datasets analyzed in this study can be found through SRA on NCBI. For wild-type Neff reads sequenced at Dalhousie University: SRX4620962, SRX4620963, SRX4620964, SRX4620965 with nanopore; SRX4625411 with Illumina. For wild-type Neff reads sequenced at the Institut Pasteur: SRX12218490 with nanopore; SRX12218479 with Illumina. For C3 reads: SRX12218489 with nanopore; SRX12218478 with Illumina. For Clone 1: SRX7813525. For Clone LT8: SRX27968414. For Clone LT6: SRX27968413 with nanopore, SRX27968419 with Illumina. For Clone LT9: SRX27968415 with nanopore, SRX27968420 with Illumina.