Single-cell genome sequencing of protozoan parasites

Abstract

Despite considerable genetic variation within hosts, most parasite genome sequencing studies focus on bulk samples composed of millions of cells. Analysis of bulk samples is biased toward the dominant genotype, concealing cell to cell variation and rare variants. To tackle this, single-cell sequencing approaches have been developed, and tailored to specific host-parasite systems. These are allowing the genetic diversity and kinship in complex parasite populations to be deciphered, and for de novo genetic variation to be captured. Here, we outline the methodologies being used for single-cell sequencing of parasitic protozoans, such as Plasmodium and Leishmania spp., and how these tools are being applied to understand parasite biology.

Why do we need single-cell DNA sequencing for parasites?

Parasitic infections often contain substantial pathogen genetic diversity. This can be in the form of multi-species infections, or genetic diversity arising within a single species [1–7]. The genetic diversity present can influence clinically relevant phenotypes (i.e. drug or vaccine response), reveal the transmission events resulting in an infection (i.e. single or multiple mosquito bites), and identify signatures of ongoing adaptation from patterns of genetic variation (see Glossary). Both mathematical models and experimental host-parasite systems show multiplicity of infection (or MOI), the presence of multiple genetically distinct parasites within a host, can drive the evolution of virulence, host fitness and drug resistance [8–10] and can increase disease severity [9]. These consequences result from competition between different haplotypes of parasites within a host. While model systems of intra-host competition, such as the rodent malaria parasite Plasmodium chabaudi, show that highly virulent parasites can be rapidly selected when coinfections of genetically distinct parasites occur [11], it has been more challenging to understand how intra-host diversity can drive pathology in human infections.

When we sequence the genomes of parasites infecting a host, we typically assay a mixture of genetically distinct pathogens. Typing genetic markers or bulk genome sequencing data provides limited information on the number, relative abundance and identity of haplotypes within an infection. Individually sampling each cell, however, provides information on each cell of the mixture and allows the genetic diversity within a single host to be deconvoluted, for rare genotypes to be detected and for de novo mutations to be identified. To date such approaches have been applied in an extremely limited number of settings, for instance we lack any basic description of genetic diversity at the single-cell level from any vector-parasite system.

In the following sections, we will (1) outline the methodologies commonly used to isolate, amplify and sequence genetic material of single-cell parasite, (2) highlight how these methods can be used to address meaningful biological questions related to parasite life history traits, and (3) describe the future prospects and directions for parasite single-cell sequencing. While we refer to many parasite species in this review, we have focused in greater details on blood stage malaria parasites, as they constitute one of the most studied pathogens at a single-cell resolution.

Isolating a single-cell parasite

The initial step in a single-cell sequencing project is to identify and isolate the cell of interest from a complex sample [12]. The main challenges with isolating a single cell from a heterogenous population is to maintain the purity and the integrity of the targeted cell. In this context, we define purity as the extent of contamination from DNA from either another cell or the environment, and integrity to be the extent of damage to the cell itself [13]. There are multiple ways to isolate a cell, and different approaches may be more suited to specific experimental systems than others. As details on available single-cell isolation platforms have been previously reported [12–15], we will focus on methods more widely used for single-cell isolation of parasites including limiting dilution cloning, fluorescent activated cell sorting and microfluidic sorting. The protocols we discuss below all rely upon live parasites. The development of approaches which can capitalize on fixed or frozen cells will enable access to archived material, and minimize biosafety concerns.

Cloning by limiting dilution

Many experimental studies require a clonal parasite isolate to minimize the impact of genetic variation on phenotypic measurements. Limiting dilution cloning is well established and widely used in infectious disease research for this purpose ([2, 16–18]). This approach uses standard pipetting techniques to isolate cells by dilution. The success of isolating a single cell is based on statistical probability. To maximize the probability of capturing one cell per well, a typical concentration of 0.25 cells per well is commonly used [2]. An increase of this concentration increases the chance to capture multiple cells per well. For instance, the statistical probability to obtain doublets (2 cells per well) based on Poisson’s distribution is 8% and 16% for cell concentration of 0.5 and 0.9 respectively [13].

Early on, limiting dilution cloning has been used to quantify Leishmania parasitemia [16, 17] and to capture phenotypic diversity in malaria parasite infections [18]. In the genomic era, limiting dilution methods have been used to dissect complexity of infection [2], and measure mutation rates in malaria parasites [19–21]. Cloning by limiting dilution has the advantage of preserving cell integrity and allowing future phenotypic characterization of clonal cell lines. However, this approach is labor intensive and time consuming, and is consequently cumbersome for large scale experiments. Critically, this approach can only be applied to organisms where a robust laboratory culture system is in place and cultures can be expanded from a single cell. Even in situations where limiting dilution cloning is feasible, the genetic diversity captured may not reflect the genetic diversity of an infection accurately, and additional mutations may accrue in each genome captured. The limitations of this method spurred the adoption of other, high throughput, single-cell approaches.

Fluorescence activated cell sorting

Fluorescence activated cell sorting (FACS) is a widely used tool in single-cell applications. For FACS-based isolation, a cell is fluorescently tagged, typically using either a cell dye, or a fluorescently labelled antibody. Cells are then passed through a FACS machine where the fluorescence of a cell, and other morphological features (like size and granulometry), can be evaluated by laser excitation and detection (Figure 1A). Cells which fit a specific criterion can be sorted into individual tubes or wells of a microplate. Given the widespread availability of FACS, this approach is typically possible without the need for specialized equipment to be obtained. The approach can be highly selective, as fluorescently tagged cells for a given phenotype can be sorted precisely. However, the set-up and validation of such approaches could be a barrier for their implementation, and biosafety protocols must be independently considered for each system. Cell sorting is not guaranteed to be free of potential contaminants and guidelines to minimize and quantify risk must be developed (Box 1) [3, 6].

Figure 1. Schematic view of single-cell sorting by FACS and 10X Genomics.

(A) The FACS workflow. A fluidic system transports particles or cells in a stream to the laser beam where each particle is excited. The resulting light signals from the excitement are directed to an appropriate detector and converted to an equivalent electronic signal that can be processed by a computer. From the computer software, the analyzed cells are grouped and gated on a data acquisition plot. After identification and selection, cells of interest are sorted by the deflection plate which applies electric charge to only the droplet of interest. Uncharged droplet pass into waste. Parasites which have readily accessible stages which are free of host cell DNA such as Plasmodium, or Cryptosporidium make using FACS approaches for single-cell isolation very appropriate. Staining the parasite DNA or some cell surface marker will provide a clear distinction between target cells from other cells in the sample. FACS-based approaches are scalable to 10s–100s of cells. (B) The 10x Genomics Chromium single-cell workflow. The 10X Genomics Chromium system performs a massive partitioning and barcoding of single cell using gel bead in emulsion (GEM) in a microfluidic chip. Each gel bead is labeled by a unique barcode. After the break emulsion, Barcoded fragments are pooled and attached to standard Illumina adaptors. As all the cell injected into the system are isolated an enrichment of parasitic cells may be needed to reduce the capture of host or non-target cells. Microfluidic approaches are scalable to 100s–10,000s of cells.

Box 1. Sterility processes and minimization of contamination risks.

Sample preparation, single-cell isolation and DNA amplification each require strict protocols for minimizing the risk of external contamination. For malaria parasites, these protocols include: (1) a dedicated room (lab) for reagent preparation, free from high molecular weight malaria parasite DNA. Here all materials for FACS isolation and MDA are prepared in a PCR hood with sterile pipettes and filtered tips. During preparation, experimenters wear a face mask and shield, gloves and sterile gown that are uniquely opened and used in this dedicated room. No experimenter who has been in an area where malaria parasite DNA is handled in the previous 12 hours is permitted in this room. (2) After sterile reagents preparation, parasite samples are prepared and stained in a Biosafety cabinet class II and transported for FACS isolation. To minimize environmental contamination, the benchtops and all materials are cleaned and sterilized using 80% Ethanol, and the FACS machine cleaned. The system is flushed and a sample containing sterilized water is analyzed between parasites samples. (3) MDA is performed on FACS isolated single cell in a dedicated PCR hood, with dedicated handling equipment and thermocycler.

Quality control is critical to ensure protocols are effective at minimizing contamination. Prior to FACS isolation fluorescent beads are analyzed and sorted into 96 well plates in triplicate (288 wells). Each well is manually checked by microscopy to estimate the efficiency and purity of the cell sorter, with a target rate lower than 0.5% of wells showing multiple beads. Mixtures of clonal parasites with known genotypes (i.e. HB3 and 3D7) are also analyzed and single-cell sorted to test contamination across the pipeline (Figure I). Single-cell samples of this mixture are library prepared, sequenced and analyzed to estimate the purity of the sorting system and to evaluate the contamination probability during sample processing.

Figure I (in Box 1). Plate sorted mixtures result in pure genome sequences.

(A) Parasites with known genotypes (A and C) are mixed, analyzed and sorted by FACS. After whole genome amplification, library preparation and sequencing, the bioinformatic analysis show unique genotype for each sorted single cell showing the absence of contamination between cells. (B) Contamination between cells can be detected by the genome size analysis. Mapping the genotypes will enable distinguishing a contamination from different genotypes (C) A single locus of the mitochondrial genome is shown for 3 3D7 and 3 Hb3 cells. All the reads at each locus match either 3D7 or Hb3 with no cross-contamination. This was consistent across all reliably called loci. In the lower panel an example of a mixed cell is shown.

FACS-based methods are particularly appropriate for most species of malaria-infected red blood cells (RBCs) as infected RBCs contain parasites DNA and therefore can be isolated from the uninfected RBCs that lack of DNA. This approach has been successfully used to remedy the labor-intensive limiting dilution method [3, 6, 22]. FACS-based methods have been optimized for both Plasmodium vivax and P. falciparum in order to sort single-cell parasites for whole genome amplification and sequencing (see Amplifying the genome of single cell section) [3, 6]. FACS-based approaches enable single-cell sequencing of P. vivax, a parasite where routine laboratory culture is not feasible.

10X Genomics

Recently, microfluidics approaches have become popular high throughput tools for single-cell genomic studies. These approaches essentially perform limiting dilution cloning on a grand scale and typically combine cell isolation with whole genome amplification in a closed system, allowing for strict control of environmental contamination. While there are multiple microfluidics platforms for single-cell isolation, the 10X Genomics Chromium system discussed here is widely used [23]. Here, single cells are captured in nanoliter droplets containing DNA-barcoded beads (Figure 1B). These DNA barcodes ‘tag’ individual cells and can be deciphered from next generation sequencing reads. This system is capable of processing thousands of cells in parallel, and can be used to capture transcripts, genetic information or epigenetic information from a cell. The platform is simple and easy to use, and is coupled with dedicated, user friendly, bioinformatics software. Compared to FACS based approaches, the 10X Genomics Chromium system reduces dramatically the required equipment from single-cell isolation to sequencing. However, 10X Genomics users cannot determine which cells are collected prior to the downstream analysis, this can be problematic when the target cells are only a small proportion of all cells as is the case with malaria infections. In fact, at low parasitemia, many red blood cells are not infected while the infected RBCs are the target cells. Parasite enrichment is usually required prior to analysis in order to increase the probability for the 10X Genomics Chromium to capture a high percentage of infected RBCs [24]. Consequently, it is currently challenging to develop bespoke protocols for specific host-parasite systems. Additionally, 10X approaches typically generate only sparse coverage of individual genomes. The method is optimized for analysis of thousands of cells in a sequencing run and, coupled with inefficiencies in the protocol, the depth of coverage typically obtained by a sequencing run is much less than 1X. While this is suitable for the analysis for CNVs, detecting other forms of variation is a greater challenge.

Amplifying the genome of a single cell

The DNA present in an individual cell is insufficient for direct genome sequencing, or targeted genotyping. The amount of genomic DNA present in a single bacterial or human cell is in the order of 1fg-1pg [25], while the minimum input for standard genome sequencing library construction is ~100ng of DNA. In order to produce enough biological material for sequencing, whole genome amplification (WGA) is typically required. Degenerate oligonucleotide primer polymerase chain reaction (DOP-PCR) was the first developed method for WGA. This method uses degenerate oligonucleotide primers to densely cover a genome with random priming sites for initiating in vitro DNA replication. This technique produces highly repeatable DNA replication profiles, resulting in low noise for detecting copy number variation (CNV). However, the thermostable polymerase required by PCR results in a relatively high per-base error rate [26]. An alternative to PCR based approaches emerged in multiple-strand displacement amplification (MDA) [27]. MDA uses random hexamer priming to amplify DNA using the Phi29 DNA polymerase. This enzyme is able to amplify DNA under isothermal conditions (typically 30°C), enabling a very low per-base error rate. However, there is a far more pronounced amplification bias with MDA, limiting its use for CNV quantification. The desire to combine the low amplification noise from DOP-PCR with the low per-base error rate of MDA spawned a number of hybrid protocols. These protocols contain both PCR and isothermal amplification steps, and include PicoPlex (Takara) and multiple annealing and looping-based amplification cycles (MALBAC) [28, 29]. As each of these WGA methods have differing read coverage uniformity, error rates and degree of reagent contamination the choice of amplification method to use depends strongly on the downstream application [25]. As studies increasingly require a detailed picture of genetic variation in a single cell, further development of parasite specific protocols which maximize uniformity and coverage will be highly desirable.

Dedicated WGA protocols for parasites are now being developed which tackle specific challenges of different species. Imamura et al. [30] compared two WGA methods PicoPLEX (Takara) and RepliG (Qiagen) for their ability to preserve aneuploidy levels and SNPs in Leishmania donovani and L. braziliensis. This study suggested that PicoPLEX is the most suitable to assess chromosomal aneuploidy. Liu et al. have optimized a WGA protocol based on MALBAC designed to tackle the extreme nucleotide bias (~80% AT-rich) of the Plasmodium falciparum genome [31]. This can successfully amplify the parasite genome and generate data with a coverage (79% of genic regions) that allows small CNVs to be detected in single cells. This highly promising approach will shed new light on understanding mechanisms of drug resistance and environmental adaptation. For P. falciparum, WGA protocols based on multiple displacement amplification (MDA) prior to genome sequencing have been developed [6]. By targeting parasites in the latter stages of DNA replication using FACS sorting, Trevino et al. [3] optimized this method to reach a result with near complete capture of the P. falciparum genome (90.7%).

Applications of single-cell sequencing in parasite biology

Single-cell sequencing can capture the complete genome sequence of an individual cell with high accuracy. Single-cell approaches applied to protozoan pathogens to date (Table 1) have addressed major questions regarding genetic diversity and adaptive mutations. We outline below major applications for which single-cell sequencing can be used.

Table 1.

List of single-cell genomics applications on protozoan parasites.

Year and goal of study	Species	Nb of cells	Methods	Polymorphisms detected
2014: Unraveling the complexity of P. vivax and P. falciparum infections [6].	P. falciparum and P. vivax	260 cells	FACS + MDA + genotyping/WGS	Targeted panel and genome-wide SNPs
2015: Intra-host population structure of the Apicomplexan Eimeria tenella [73].	E. tenella	36 cells	FACS + MDA + Genotyping	Targeted SNP panel
2016: Protocol optimization to dissect cryptosporidium infection with closely related parasite species or subtypes [5].	C. parvum	10 cells	FACS + MDA + WGS	de novo mutations
2017: Protocol optimization for high-resolution single-cell genomics of malaria parasites [3].	P. falciparum	48 cells	FACS + MDA + WGS	SNPs
2020: Protocol optimization of single-cell amplification for CNV detection and analysis [31].	P. falciparum	27 cells	CellRaft + opt.MALBAC + WGS	SNPs and CNVs
2020: Deciphering malaria parasites within host diversity and the complexity of an infection [22].	P. falciparum	485 cells	FACS + MDA + WGS	SNPs
2021: Estimating the number of individual karyotypes of Leishmania donovani [74]	L. donovani	1,560 cells	10X Chromium	Aneuploidy and CNVs
2020: Validation of single-cell genomics for Leishmania parasites [30].	L. donovani and L. braziliensis	28 cells	FACS + PicoPLEX/MDA + WGS	Aneuploidy and CNVs
2020: in vitro mutation accumulation in malaria parasite [63].	P. falciparum	40 cells	FACS + MDA + WGS	SNPs and small indels
2021: Characterization of de novo mutation and population survey from clinical samples of malaria parasites [75].	P. vivax and P.falciparum	406 cells	FACS + MDA + WGS	SNPs

Capturing genome-wide haplotypes

Single-cell sequencing provides a direct estimate of the number of haplotypes. A haplotype is the set of polymorphisms carried by a strand of DNA inherited from a single parent. Haploid organisms (such as asexual stage malaria parasites) carry a single, genome-wide haplotype, while diploid organisms (such as leishmania parasites) carry two (and entamoeba and giardia carry four). In malaria infections intra-host parasite genetic diversity is derived from two sources of variation: (i) the parasite haplotypes already present in the population of infected cells (also known as standing variation), and (ii) the mutations which arise on these haplotypes after meiosis (also known as de novo variation). In this instance, MOI is equal to the number of unique haplotypes infecting its host. Genetic diversity and MOI are typically higher in high transmission areas [32, 33], though complex features of the biology and lifecycle of a parasite including host immunity, population structure, host genetics and specific transmission events shape the diversity of individual infections [34]. Capturing haplotypes in host-parasite systems where meiosis happens in the host of interest (the definitive host, i.e. malaria parasite mosquito stages, or toxoplasma feline stages) will not inform on MOI, though could be used to better understand the process of inbreeding and recombination in these settings. A major unexplored area is the application of single-cell sequencing to vectors to better characterize diversity, and relatedness.

In the context of malaria, there are multiple tools to estimate the number of haplotypes present in an infection from bulk genetic data (i.e. COIL, the REAL McCOIL) [35, 36], though these typically do not reconstruct haplotypes. The exception is DEploid [37], a statistical model to jointly estimate the number and identity of haplotypes present within a malaria infection, and their abundance from bulk sequencing data. While this is a major advance in computational genomics, inference is limited to five or fewer haplotypes, and accuracy is dependent upon the complexity of an infection, inbreeding and relatedness, and the abundance of a haplotype within an infection [22, 38]. Conversely, single-cell sequencing provides a direct estimate of the number of haplotypes. In a haploid organism each single-cell sequence is a haplotype, the number of haplotypes present can be estimated by simply counting the number of unique sequences (after accounting for sequencing error and de novo mutation). Statistical approaches developed for capturing ecological diversity, such as rarefaction, can be applied to assess whether all unique haplotypes have been captured [6, 39]. A survey of polyclonal infections of P. falciparum by single-cell sequencing from Malawi suggests inbreeding, highly related parasites and highly complex infections (i.e. >5 haplotypes) are common in a high transmission setting [22].

Characterizing transmission and population structure of individual infections

Parasitic organisms encompass a range of population structures, driven by the local prevalence of infection, immunity, genetic diversity and the opportunity and predilection for outcrossing [40]. In complex malaria infections, parasitic genetic diversity can come from (i) bites from multiple mosquitoes, each infected by one parasite haplotype, from (ii) a single bite from a vector infected with multiple parasites haplotypes, or from a mixture of (i) and (ii). Individual parasite haplotypes, such as those obtained by single-cell sequencing, provide information on the relative timing of infection of haplotypes within an infection with closely related parasites being infected by the same mosquito bite. Distinct haplotypes within an infection can colonize an individual through different routes (Figure 2). In malaria infections, these routes can be classified as superinfection (where diversity is the result of independent bites of infected mosquitoes) or co-transmission (where a single mosquito transmits different genotypes of parasite during a single bite). The routes are not independent, both superinfection and co-transmission can drive the diversity present within a single host in different ways. It is expected that co-transmission will result in closely related parasites being inoculated into a host while the superinfection will result in the inoculation of more distantly related parasites [41]. When the complexity of an infection increases, bulk analysis approaches such as DEploid struggle to deconvolute within host diversity and the possible routes of parasite transmission. Disentangling how all genotypes present in a single host are related can provide important information on transmission intensity and the chance of outcrossing.

Figure 2. The pattern of genetic diversity within a malaria patient is driven by the mode of transmission of the parasites.

(A) A superinfection of unrelated genotypes is generated when two mosquitoes each bearing genetically different genotypes (filled circles and open circles) transmit parasites to a single individual. (B) A co-transmission of related genotypes occurs when in a mosquito, distinct genotypes go through sexual recombination generating multiple related genotypes that the mosquito transmits to an individual.

Using malaria parasite transmission as an example, we demonstrate how the haplotypes present within an infection can distinguish between superinfection and co-transmission (Figure 2). In Plasmodium species, sexual recombination occurs in the mosquito vector. When parasites do not pass through the same mosquito, no recombination of haplotypes can occur, and the haplotypes are distinct (Figure 2A). Haplotypes which have recombined in the mosquito share ~50% of their genome identical by decent (IBD) with each other (Figure 2B). Both processes can contribute to the genetic diversity within a single infection.

Using single-cell genomics, Nkhoma and colleagues [22] resolved the complexity of P. falciparum infections with high resolution. This study was based in a high transmission setting (an estimate entomological inoculation rate of 183 bites per person per year [42]) where the opportunity for superinfection is high. Despite this, most infections contained related parasites which were likely transmitted by a single mosquito bite. Infections showing co-transmission, or co-transmission plus superinfection were dominant, with only a single simple superinfection identified. Even in high transmission areas, co-transmission plays a major role in the generation and maintenance of within-infection genetic diversity. The application of single-cell approaches to regions with different degrees of malaria endemicity will allow us to understand the differing role co-transmission and superinfection play.

Characterizing functional, low frequency and de novo mutations

As we mention above, the genetic variation of an intra-host parasite population can be split into standing variation inoculated into a host, or de novo variation arising during an infection (Figure. 3A). Novel variation introduced by de novo mutation is of particular interest as it may allow the emergence of novel traits (such as escape from the immune system or drug resistance). Parasitologists and epidemiologists are often interested specific mutations, such as those driving drug resistance. While genome sequencing is critical to (i) identify genetic mutations responsible for drug resistance, (ii) track their spread [43–51] and (iii) quantify within-host mutation allele frequency [52, 53]; single-cell sequencing can add critical information by characterizing drug resistance haplotypes.

Figure 3. Within host genetic variation is from standing or de novo variation.

(A) In plasmodium species, sexual reproduction occurs in the mosquito vector. In presence of two distinct genotypes, sexual recombination will generate newly related genotypes. The transmission of these parasites to an individual will lead to both standing and de novo variations. (B) Calling deletions using read depth data. For each panel the read depth profiles of each single cell are shown in rows. The top three rows (blue) are from 3D7 cells, the next three rows (pink) are from Hb3. The remaining rows are from a single infection (MAL15) containing multiple genotypes. A UPGMA tree of 1-pairwise allele sharing is shown on the left, with closely related samples colored identically. Where read depth drops to zero (indicating a deletion) the profile is shown in grey. Left: two independent deletions affecting PHISTa/PHISTb in the parasite genome with both genes in Hb3 deleted and a single gene deleted in two closely related lineages of MAL15. Right: a deletion between the CLAG3.1 and CLAG3.2 genes in two closely related lineages from MAL15.

Unlike viruses and bacteria, drug resistance is rarely acquired by de novo mutations in malaria infections; instead, it is selected, inherited, and transmitted [54, 55]. When multiple drug resistance mutations are present, the process of genetic recombination in the mosquito midgut can either combine or separate resistance into individual genomes (Figure. 3B). This approach has been applied to identify multidrug resistant parasite within single infections of P. falciparum and P. vivax [6]. In addition to drug resistance, the genetic background carrying other functional genetic mutations can be defined. In Figure 3B, we examine single-cell sequencing data from Trevino et al [3] to show the genes involved in host cell remodeling (PHIST) [56, 57] and nutrient uptake (CLAG3.1/3.2) [58] which are deleted in specific lineages within a single infection. These deletions are present in multiple related genetic backgrounds and are an example of standing variation, rather than de novo mutation.

High throughput sequencing technologies offer exciting opportunities to witness evolution firsthand. Almost all studies of natural parasite populations have focused on standing variation [59–61]. However, characterization of mutations at the population level is insufficient for understanding the genetic basis of parasite adaptation. It is necessary to identify de novo mutations and the extent to which adaptation to an individual host is acting. Within a population of cells, mutations arise frequently. The mutation rate of the P. falciparum genome is in the order of 3.3×10⁻¹⁰ per base-pair per generation [21, 62]. To put this in context, a patient with ~1% parasitemia carries ~10¹⁰ parasites, meaning every 48 hours every single base pair in the 23Mb parasite genome will mutate at least three times over. Each of these mutations will arise in a single genome and be undetectable by bulk genome sequencing. By sequencing a single parasite genome, you enrich for the mutations found in that genome. In experimental culture populations, single-cell sequencing has been used to discover mutations selected during long term culture adaptation of P. falciparum [63], and karyotypes underlying drug resistance in Leishmania donovani [30]. It is an ongoing effort to bring these approaches to natural parasitic infections to capture de novo mutations. These may be critical to understanding how parasites can adapt to an individual host.

Accurately capturing de novo low frequency mutations based on next generation sequencing platforms is a major challenge [64]. In bulk samples analysis, most rare mutations (those present at <1% frequency) cannot be detected by any means. Single-cell sequencing is emerging as an alternative to capture rare but important mutations. While these approaches have yet to be broadly applied to parasitic infections, capture of de novo mutations from single-cell sequencing data is being powerfully applied to understand genetic variation in human gametes,[65] and cancer progression. Within a growing tumor, multiple lineages of cancerous cells can develop. These may each bear distinct phenotypes impacting drug response, growth and clinical pathology. For instance, single-cell sequencing of cancer cells shows drug resistance occurs through adaptative selection from pre-existing resistant genotypes within a tumor [66]. Mutations which drive and distinguish clinically relevant lineages can be identified. These are revealing distinct differences between cancers, for instance those where multiple adaptive lineages have emerged and show signatures of competition with one another (known as clonal interference). In laboratory experiments, we are able to capture similar signatures in P. falciparum [63]. While such experiments require careful design to ensure sufficient numbers of cells have been sequenced to detect rare variation, exploration for intra-host adaptation and its constraints are another major potential application of single-cell sequencing in parasitic diseases.

Future directions: Multi-omic profiling of single cells

In pathogens, studies are mainly focused on genetic or transcriptomic information analyzed separately. Sequencing of both the genome and transcriptome from the same cell enables deconvolution of haplotype-specific transcriptomes from mixed infections and can capture the impact of de novo mutation on gene expression. Joint profiling of DNA and RNA will tell us about the effect of multiple genotypes of parasites within an infection and the functional role of mutations on parasites adaptation. Mapping the genetic and transcriptomic diversity of a pathogen within an infection will allow correlation of genotypes to specific gene expression programs. This simultaneous analysis will help us to understand if similar genotypes within an infection share the same transcriptional patterns. From the genetic diversity determined by single-cell genome sequencing, we can integrate information from single-cell transcriptomes to map eQTLs across different haplotypes and link genetic variations to gene regulatory network differences [67]. In fact, this analysis at the single-cell level will deconvolute the role of rare haplotypes that may be shadowed when using the bulk analysis approach. In addition, we can study mechanisms of adaption (i.e., drug resistance, immune system and/or vaccine escape) controlled by genetic variation in individual genotypes.

Approaches that permit both genome and transcriptome sequencing from a single cell have been developed for other organisms including DR-seq (gDNA-mRNA sequencing) [68] and G&T-seq (genome and transcriptome sequencing) [69, 70]. DR-seq uses specific primers to enrich single-cell DNA and RNA together and separate the sample for independent amplification of the gDNA and cDNA. G&T-seq physically separates gDNA and RNA after cell lysis and before amplification, library preparation and sequencing. G&T-seq uses a modified Smart-Seq2 [71] protocol for the mRNA amplification, whereas the genomic DNA can be amplified by any genomic amplification method of choice. Consequently, the G&T technique can be readily optimized for parasite-specific protocols. Bioinformatic approaches for inferring polymorphisms from single-cell RNAseq are an alternative strategy; these are appealing as they can be applied to existing platforms. However, the genome-wide coverage of single-cell RNAseq data is likely to challenge the identification of de novo mutations or distinguish highly inbred parasite genomes robustly [72].

Concluding remarks

Single-cell genomics is a powerful tool for understanding parasite biology and host-parasite interactions. As we outline, there are a dizzying array of platforms to obtain genetic information from single cells. None of these approaches are a one-size-fits-all and the approach adopted should be driven by the relevant biological questions. Despite the array of options, single-cell approaches are constrained by technical challenges. When dealing with single cells, it is critical to ensure that robust protocols are in place to minimize contamination from the environment, and from other target cells. These protocols must be validated to ensure their effectiveness. In the present review, we have demonstrated that single-cell sequencing can access many aspects of genetic variation which cannot be directly observed using bulk approaches. With single-cell genomics methods, we can now (i) directly phase the haplotypes from an infection and estimate the relatedness between parasitic cells, (ii) capture any mutations, potentially related to drug resistance and other clinically relevant phenotype and arising on the same genetic background, and (iii), identify any de novo mutations segregating at low frequency in the parasite population. As we move forward, the technical feat of single-cell sequencing is revealing fundamental biology of protozoan parasites (see Outstanding Questions), how they interact with their respective hosts and pave the way for the discovery of new therapeutic targets and approaches.

Outstanding Questions.

• How can multiple types of omic and phenotypic data be integrated from single cells?

• How does the genetic variation present within a single infection impact disease outcome and progression?

• Single-cell sequencing can resolve intratumor evolution. Can we use single-cell approaches to understand within host adaptation for protozoan parasites?

• Can single-cell approaches be applied to parasite-vector interactions to understand the link between transmission and genetic diversity?

Highlights.

• In many host parasite systems, an infected host may carry several different pathogen genotypes. Genetic variation is associated with mechanism of evolutionary adaption.

• Single-cell genomics has emerged as a powerful method to decipher genetic diversity and complexity of pathogens within an infection and allow to estimate relatedness between haplotypes. Consequently, this led to a better understanding of pathogen transmission and adaptation.

• With a single-cell approach, in addition to estimating the preexisting genetic variants within an infection, it is also possible to track de novo mutations, which allow us to evaluate the relative contribution of each type of variation.

• Finally, combination of single-cell genomics with single-cell transcriptomics from the same cell will bring new light on understanding phenotypic adaptation traits driven by genetic variation.

Glossary

Bulk samples

A sample of cells obtained from a host, or other environment.

Coinfection

When a patient is infected by multiple pathogens, each via the same infection event.

Copy number variation (CNV)

A mutational class where a segment of the genome is either deleted or amplified altering the number of copies of the region.

Genetic variation

Pre-existing genetic diversity present within a host or population.

Multiplicity of infection (MOI)

The number genetically distinct pathogen genotypes within a single host.

Relatedness

The distance between two individuals in terms of the number of meiosis which separate them. For instance, siblings are more related to one another than second cousins.

Superinfection

When a patient is infected by multiple pathogens, each via independent infection events.

Haplotypes

The set of alleles carried on a same chromosome and inherited from a single parent.