How colonization bottlenecks, tissue niches and transmission strategies shape protozoan infections

Abstract

Protozoan pathogens such as Plasmodium spp., Leishmania spp., Toxoplasma gondii and Trypanosoma spp. are often associated with high-mortality, acute and chronic diseases of global health concern. For transmission and immune evasion, protozoans have evolved diverse strategies to interact with a range of host tissue environments. These interactions are linked to disease pathology, yet our understanding of the association between parasite colonization and host homeostatic disruption is limited. Recently developed techniques for cellular barcoding have the potential to uncover the biology regulating parasite transmission, dissemination, and the stability of infection. Understanding bottlenecks to infection and the in vivo tissue niches that facilitate chronic infection and spread has the potential to reveal new aspects of parasite biology.

Understanding protozoan population dynamics

Parasitic protozoans including Plasmodium spp., Leishmania spp., Toxoplasma gondii and Trypanosoma spp. cause the debilitating diseases of human global health concern ranging from malaria, leishmaniasis, toxoplasmosis, sleeping sickness to Chagas disease [1]. The success of these eukaryotic pathogens can be linked to their multifaceted life cycles, which include transmission between multiple eukaryotic or insect host species (Figure 1) [2]. To facilitate transmission, protozoan parasites have evolved strategies to traffic to tissue niches distal to the site of primary infection, such as the skin or small intestine [3,4]. Establishing residency in these distal sites is often linked to changes in parasite gene expression, parasite differentiation, and chronic infection of permissive tissue niches. For example, the conversion of T. gondii tachyzoites into bradyzoites within neuronal tissue and skeletal muscle, and Trypanosoma brucei and P. falciparum immune evasion via surface antigen variation [5–8]. Infection and inflammation in chronic niches like the cardiac muscle (T. cruzi) or the brain (T. gondii) can have profound negative consequences for tissue homeostasis. To protect these tissues, elaborate host barrier structures have evolved to limit the potential for pathogen colonization while facilitating nutrient and immune cell accessibility [9]. The complexity of protozoan parasite life cycles, coupled with limited ‘druggability’ of many chronic tissue niches has also limited the success of therapeutic tools to combat protozoan infections [10]. The need for targeted treatment options and a deeper understanding of how parasite localization influences fitness has stimulated scientific interest in understanding how the host environment shapes the population dynamics of protozoan infections [1].

Figure 1. Selective pressures and stochastic bottlenecks shape population structure of parasites during infection.

Parasites encounter stochastic pressures first in a novel environment with mostly negative consequences on parasite population structure and genetic diversity. Within the host, the parasite encounters further selective pressure that dictate which niches parasites occupy due to resource availability or as safe havens.

During an infection, parasite population dynamics are influenced by contrasting pressures – those that are selective, and those that are stochastic (Figure 1). Host selective pressures can include nutrition, drug treatment, immune polarization regulated by host genotype, and environmental inputs. Distinct from these selective mechanisms are stochastic bottlenecks. These typically have a physical basis, such as a transmission bottlenecks or an endothelial barrier, that constricts the parasite population irrespective of an individual parasite’s genetic fitness [11]. Whole-genome CRISPR knockout screens have revealed parasite genes conferring fitness in vivo. This approach is now being combined with host genetic tools and drug treatments to understand the parasite biology that facilitates adaptation to host selective pressures [12–16]. By contrast, stochastic bottlenecks have been less studied. Mechanisms typically used by pathogens to overcome selective pressures, include recombination or horizontal gene transfer, and occur at the level of individual genomes. In this regard, stochastic population bottlenecks can have an outsized impact on the long-term population structure of a species by dramatically impacting population size and having non-selective impact upon total population genetic diversity. Appreciating the magnitude, location and timing of stochastic bottlenecks should clue researchers into the infection niches where gene expression and host effector mechanisms are the most critical for the outcome of infection. This review will explore the ways in which the obligate intracellular parasite T. gondii and the largely extracellular parasite T. brucei interact with host tissue barriers, how these barriers influence colonization, and the tools available to study parasite population dynamics within these sites.

Protozoan traversal of host barriers and tissue niche colonization

Toxoplasma gondii can infect a remarkably wide range of euthermic intermediate hosts, with acute infection resulting in flu-like symptoms that typically resolve. However, chronic infection is thought to be lifelong and immune suppression can trigger parasite recrudescence and potentially life-threatening toxoplasmosis. T. gondii is orally infectious, acquired through the ingestion of meat containing bradyzoite tissue cysts, or food and water contaminated with oocysts shed from the feline definitive host (Figure 2) [17]. This obligate intracellular parasite invades the small intestine and can be visualized replicating within enterocytes as early as five days post-infection [18,19]. It is not known whether T. gondii infects enterocytes by traversing the F-actin-rich enterocyte brush border directly, by disrupting tight junctions and invading the basolateral side of these cells, or by first infecting M-cells in the villi crypts [20,21]. While this is an important early niche for the parasite, epithelial damage and recognition of commensal microbiota plays an important role in the activation of a local, protective immune response [22–24].

Figure 2. Toxoplasma gondii and Trypanosoma brucei confront multiple dissemination barriers to establish infection in the host.

(A) Felidae species support the sexual stage of T. gondii development, shedding oocysts into the environment. Warm-blooded intermediate hosts consuming oocysts or bradyzoite tissue cysts. (B) Following ingestion T. gondii invades the small intestinal epithelium. Parasites grow in endothelial cells and use immune cells as ‘trojan horses’ to access the vasculature and disseminate through the host. (C-D) T. gondii enters tissues through vascular barrier, including the blood brain barrier, skeletal muscle and cardiac muscle the major sites of chronic infection and transmission. (E) T. gondii can be vertically transmitted to fetuses following traversal of the placental barrier. (1) Trypanosoma spp. are transmitted by bites of the Tsetse fly. Trypomastigotes are transferred into the dermis as the fly pool feeds, disrupting blood and lymph vessels with its proboscis. (2) Trypanosoma migrate extracellularly to the vasculature then (3) disseminate into a range of tissues, including adipose depots, which may be sites of VSG recombination and chronic infection. Skin and blood colonization may facilitate transmission to the Tsetse fly. (4) Late-stage Trypanosome infection can lead to spinal fluid infection and transfer of into the brain via the choroid plexus causing host-maladaptive, lethal inflammation.

In the first week after infection, T. gondii is found in stromal cells of the lamina propria, as well as infiltrating monocytes, neutrophils, and dendritic cells [25]. Activated innate immune cells can restrict T. gondii growth via cell-autonomous immunity, however, naive monocytes and dendritic cells play a critical role in trafficking T. gondii to other tissues, including lymph nodes, spleen, adipose depots, liver, lung, skeletal muscle, cardiac muscle, and the central nervous system [26–28]. By four weeks post-infection, parasites are cleared to levels that fall below PCR or histological detection in most tissues other than the skeletal muscle, cardiac muscle and central nervous system, however, human organ transplant data indicates that many tissues harbor sufficient parasite load to induce infection of Toxoplasma-negative transplant recipients, including lung, kidney, heart, liver [29]. To access these tissues, T. gondii must traverse the tissue-vascular barrier which generally consists of endothelial cells in close apposition to pericytes which also interface with tissue stromal cells or fibroblasts (Figure 2) [30].

For most tissue sites the mechanism of trans-endothelial migration has not been explored, the exception being the blood brain barrier (BBB). The brain contains the highest chronic parasite load per gram of tissue [31], making it a critical site for transmission to the feline definitive host. Neural infection is associated with parasite differentiation into bradyzoite cysts, a program that is regulated epigenetically and transcriptionally, and associated with slower parasite growth and immune evasion [32]. Immune suppression can lead to severe visual impairment, brain damage, and death [31]. At the BBB three mechanisms of parasite entry have been proposed: transcellular infection of endothelial cells [33], paracellular trafficking of free parasites [26,34,35] or paracellular trafficking within extravasating immune cells (Figure 2) [36,37]. The precise anatomical site(s) of parasite entry into the brain and the function the cerebrospinal fluid barrier and the meningeal arachnoid barrier are currently unknown [38,39]. Both central nervous system infection and gastrointestinal infection and have been intense areas of interest, however, little is known about how effective anatomical barriers are at limiting parasite entry and stable niche colonization, or how they regulate dissemination to other tissues.

T. brucei is an extracellular parasite transmitted by bites from tsetse flies, causing human African trypanosomiasis and nagana disease in livestock. Two main subspecies cause human African trypanosomiasis: T. brucei rhodesiense is associated with acute disease that lasts days to weeks and T. brucei gambiense is associated with chronic illness that lasts weeks to months. Tsetse flies are pool feeders, that use their proboscis to lacerate blood vessels and feed on blood and lymph fluid in the mammalian host. As a result, T. brucei is deposited in the dermal layer of the skin [40]. Parasites are observed in connective tissue associated with collagen bundles [41], which can develop into chancres as early as 2 days post-infection [42,43]. The first wave of macrophage, neutrophil, NK cell, and T cell infiltration, is critical to control parasite load. However, T. brucei can remain proliferative in the skin, eliciting a secondary neutrophil response and immunological environment that is beneficial for the parasite in the early stage of the infection [44]. During the later stages of the infection the skin contains quiescent, transmissible parasites which can differentiate into the insect procyclic form, supported by evidence where teneral tsetse flies fed on mice with low to undetectable parasitemia became infected [45]. Whether this population represents a tissue-resident pool from the initial infection or a secondary reseeding population from circulation is unclear. Together, these data indicate the dermal environment can pose both a barrier to systemic colonization by T. brucei entry as well as a potential long-term reservoir for transmission.

Blood parasitaemia can be detected as early as one day post-infection, marking the early hemolymphatic stage of trypanosomiasis. Hydrostatic pressure and chemokine sensing have been proposed to underpin parasite migration from the bite site to lymph and blood vessels although the precise mechanism is not clear [46]. This migratory event may represent an initial stochastic bottleneck that parasites must overcome to enter the circulation. As the infection progresses, T. brucei accesses the brain parenchyma, crossing the BBB via the choroid plexus [47–49]. If untreated, parasite growth is poorly controlled causing meningoencephalitis and host death [43]. In contrast to T. gondii infection, the brain is not a transmission niche for T. brucei [47]. This indicates that the selective pressures shaping the entry of T. gondii and T. brucei into the brain are likely distinct. In agreement with this, CNS infection by T. brucei frequently results in pathological inflammation [50,51] whereas T. gondii infection is controlled and well-tolerated by the host [52].

T. brucei CNS infection also disrupts the host sleep-wake cycle which has been hypothesized to benefit T. brucei by giving the insect vector longer to feed and therefore increasing the likelihood of parasite transmission [53].

Historically, T. brucei was thought to traffic from the blood or lymphatic system directly to the CNS, however, recent studies have discovered that parasites cross the vascular and/or lymphatic barriers to persist within a range of tissues including adipose tissue [54,55], heart [56], lung [57], ovaries [58], testes [59] and spleen [60]. Recent studies evaluating VSG diversity have revealed that extravascular niches may seed subsequent rounds of blood infection by clones that have recombined their VSG locus and are resistant to circulating antibodies raised against the primary infection VSG type [61]. The majority of parasites isolated from the blood are quiescent [62], suggesting that the replicative niche for VSG switching is extravascular [63]. Occupying these niches may facilitate host immune evasion and limit contact with therapeutic agents that cannot efficiently penetrate deep tissues. In addition to the skin, adipose tissue is colonized early in infection and contains a high load of proliferative parasites. Adipose parasites present a distinct morphology that is intermediate to the characteristic ‘slender’ and ‘stumpy’ forms found in blood, and it has been suggested that this unique morphology may facilitate T. brucei migration between cell-cell junctions [54,55,64]. Adipose-resident parasites also exhibit metabolic shifts that may reflect niche specific-nutrient availability and parasite persistence [55]. The levels of available non-esterified fatty acids are higher in the adipose tissue than in the blood [65]. Although T. brucei can synthesize lipids, it scavenges host lipids for several essential metabolic pathways including biosynthesis of glycosylphosphatidylinositol, a membrane anchoring moiety appended to procyclins and VSGs. To satisfy this major nutritional requirement, T. brucei has been shown to hijack the adipose T cell response to liberate fatty acids [66], a response that ultimately leads to adipose wasting and cachexia in chronically infected mice. Thus, adipose residency may benefit the parasite by providing both nutritional resources, and a tolerogenic immune environment for persistent infection. These emerging models highlight a complex interplay of selective and stochastic events, and underscore the impact that bottlenecks imparted by the infection discrete tissues have on the composition of parasites available for persistent infection transmission.

Tools to dissect stochastic bottlenecks protozoan infections

Traditional genetic approaches have been used with great success to dissect selective pressures. However, when considering population dynamics within an organism, the use of genetic mutants and pooled mutant libraries have been avoided as they purposefully generate heterogeneously fit or virulent populations, which would be expected to result in barcode loss unrelated to any stochastic process. The impact of stochastic bottlenecks have historically been less tractable, and mostly unstudied in the case of protozoan parasite infections. Progress within viral and prokaryotic infection disciplines has benefited from the cross-pollination of concepts and approaches typically employed by population geneticists. Lineage tracing methods, such as cellular barcoding, are notable examples, recently reviewed in relation to human disease by Sankaran and colleagues [67]. Cellular barcoding uses naturally occurring variation or experimentally introduced unique molecular identifiers to ‘barcode’ cells of interest [68–70]. When applied experimentally, exogenous unique selectable markers, sequences or tags can serve as ‘alleles’ from which changes in frequency within genetically complex populations can be quantified. This has been used in lineage tracing [71,72], functional profiling [73–75] and investigating spatiotemporal population dynamics of pathogens [76–78] (Table 1).

Table 1.

Studies investigating population dynamics using cellular barcoding in vivo.

Organism	Barcoding technique	Integration via	Quantification method	Number of barcodes	References
Borrelia burgdorferi	WITS carrying two 20 bp DNA tag	Insertion into linear lp25 plasmid	PCR	7	[79]
Escherichia coli	WITS carrying 20 bp DNA tag (STAMPR)	Homologous recombination	NGS (Illumina)	~1100	[80]
	WITS carrying 40 bp DNA tag	Tn7 mediate integration	qPCR	10	[81]
Listeria monocytogenes	WITS carrying ~30 bp DNA tag (STAMP)	Phage integrase-mediated recombination	NGS (Illumina)	200	[82,83]
	WITS carrying 40bp DNA tag	Phage integrase-mediated recombination	qPCR	20	[84]
Pseudomonas aeruginosa	WITS carrying ~30 bp DNA tag (STAMP)	Integrase-mediated recombination	NGS (Illumina)	~4000	[85]
Salmonella	WITS carrying 40 bp DNA tag	Lambda-Red recombination	qPCR	8	[76,86–89]
	WITS carrying 40 bp DNA tag	Lambda-Red recombination	rtqPCR	7	[90–93]
	WITS or MITS carrying 40 bp DNA tag	Lambda-Red recombination	NGS (Illumina)	8	[94,95]
	WITS carrying 21 bp DNA tag (STAMP)	Lambda-Red recombination	NGS (Illumina)	85	[77]
Streptococcus pneumoniae	Two strains carrying “OVA” or “AVO” peptides each	Janus-cassette mediated recombination	rtPCR	2	[96]
Toxoplasma gondii	WITS carrying 6 bp DNA tag	CRISPR-Cas9 mediated homologous recombination	NGS (Illumina)	96	[97]
Trypanosoma brucei	WITS carrying 6 bp DNA tag	CRISPR-Cas9 mediated homologous recombination	NGS (Illumina)	96	[97]
	Genetically homogenous cells carrying 40 bp DNA tag	Homologous recombination	NGS (Roche)	8	[98]
Vibrio cholerae	WITS carrying 30 bp DNA tag (STAMP)	Homologous recombination	NGS (Illumina)	~500	[99,100]
Yersinia pseudotuberculosis	WITS carrying 40 bp DNA tag	Transformation with plasmid containing tag	Southern blot	33	[101]

MITS= mutant isogenic tagged strains. bp= base pairs. qPCR= quantitative PCR. rtqPCR= real-time quantitative PCR.

A recent advance has been the use of wild-type isogenic tagged strains (WITS). Here, a unique molecular identifier is inserted into a neutral locus therefore having no effect on cell fitness [76]. The WITS approach has been widely used in bacterial infection studies, such as dissecting stable niche colonization by Salmonella [87]. One limitation of the original methodologies relates to the small number of barcoded strains that can be identified, leading to coarse-grain understanding of stochastic pressures impacting within-post population dynamics. Next-generation WITS methods seek to overcome this limitation by combining a greater number of markers with next-generation sequencing for quantitative analysis of highly complex barcoded strain libraries, and population genetic mathematical analyses. In principle, the higher the number of genetic markers (i.e. barcodes) used, the greater the resolution for discerning the width of the bottleneck [102]. Exemplifying this, sequence-tagged analysis of microbial populations (STAMP) and its successor, STAMPR, quantify the relative abundance of individuals in a tagged isogenic population [78,99]. The method applies mathematical equations derived from population genetic theory and next generation sequencing (NGS) to accurately estimate bottleneck sizes within the host, while also providing information on the spatiotemporal dynamics. For these methods, the bottleneck width, or the founding population size (N_b), is calculated using initial sample size (i.e. number of sequences) and sample size at a given sampling time. It is derived from several population genetics approaches, including effective population size, N_e [103] and equations from Krimbas and Tsakas [104]. STAMPR improved upon the assumption that changes in allele frequency correspond to stochastic, homogenous movement of populations through a bottleneck. It incorporates considerations of complex colonization patterns such as repeated bottleneck events and heterogenous growth rates after a bottleneck event. As an alternative approach, the genetic relatedness of populations can be determined by calculating their genetic distance from one another, using the chord distance equations such those defined by Cavalli-Sforza [105]. Studying colonization using STAMP and STAMPR revealed complex dissemination patterns in a host in a temporal manner initially in bacterial species including Vibrio cholerae [99], Salmonella [77], extraintestinal E. coli [78], and Listeria monocytogenes [82]. However, there are limitations to the effectiveness of this approach at low numbers of barcodes, which may be problematic when studying host-pathogen interactions where a low inoculation dose of pathogen is required. The application of this approach to protozoan parasites has the potential to reveal dynamics of the population structure within a host.

Barcoding methods have only recently been applied to study population dynamics in protozoan infections. Wincott et al. developed an approach to label T. gondii or T. brucei with 96 unique DNA barcodes and monitor population dynamics in tissue culture and in vivo [97]. This study challenged the assumption that the CNS barrier systems function as a stringent bottleneck to T. gondii entry as majority of the barcodes identified in the inoculum were represented mouse brains one month after infection. Although this study was designed to introduce 96 barcodes at an equal ratio, natural variation in relative abundance was observed in the inoculum. Intriguingly, low abundance barcodes in the inoculum were able to become highly abundant in the brain at chronic infection, further supporting the stochastic nature of parasite access to the CNS and niche establishment [97].

Most studies evaluating T. brucei colonization of extravascular spaces have relied on bioluminescent or fluorescent reporters to quantify parasite load. However, the spatiotemporal relations of the populations within each niche has been difficult to assess until recently. In one pioneering study, eight barcoded strains of T. brucei were used to assess the population dynamics of parasite transfer to tsetse flies, following a blood meal and transmission to the murine host [98]. This study indicated that both tsetse flies and mammalian bite recipients can be colonized by more than one founder allelic type of T. brucei [98]. As only eight barcoded strains T. brucei were used, it is likely that the increased number of cellular barcodes would reveal new aspects of host colonization. The approach recently developed by Wincott et al. allows for at least 96 unique barcodes to be simultaneously incorporated within a neutral locus, and should provide greater resolution of the magnitude of stochastic events impacting T. brucei’s within-host population dynamics during colonization [97]. These studies provide the foundations for exploring parasite diversity within discrete tissue niches, the inter-relationship of each niche over time, and host response mechanisms that regulate niche accessibility.

Future perspectives on cellular barcoding of protozoans

T. gondii and T. brucei survive within the host environment despite encountering immunological and physical barriers as seen in the intestinal epithelial barrier and BBB for T. gondii, and in the blood, skin, and BBB for T. brucei. Cellular barcoding approaches can facilitate quantification of the magnitude of the population bottleneck encountered at these barrier sites and how residence in protected tissue spaces regulates parasite virulence strategies are largely open questions (considered within Outstanding questions). Less is known about how this biology operates at sites like the placenta, which facilitates vertical transmission of T. gondii and T. brucei, and barrier sites in definitive hosts (feline species for T. gondii and the Tsetse fly for T. brucei). Such approaches can be used to understand the ecology and evolution of co-infection dynamics between parasite species or strains and the host environment [106]. The relevance of the scenario is nicely illustrated for T. brucei, where field samples indicate that the insect vector is typically co-infected with different trypanosome species [107,108]. The use of independently barcoded populations of these parasites would allow for the co-infection dynamics to be ascertained with high spatiotemporal resolution, and provide insight into the impact of co-infections on parasite biology and disease pathology.

The insertion of unique sequences into the genome, a requisite for cellular barcoding, can now be readily and widely achieved with other protozoan parasites. In one study, a library of uniquely barcoded P. falciparum parasites was used to study the relationship between fitness and drug resistance [13]. In Leishmania spp., LeishGEdit incorporated a barcoding strategy into a functional screen for flagellar mutants, which could be adapted for within-host population genetic studies [109]. The method developed for the cellular barcoding of T. gondii and T. brucei [97], could potentially be applied to other trypanosome species such as T. congolense where CRISPR-Cas9 genome editing was recently adapted [110].

The technical application of lineage tracing approaches to study host colonization by these parasites is still in its infancy. For example, the use of more advanced recorder-type barcoding strategies [111]. These lineage recorders introduce additional Cas9-generated ‘scars’- insertions or deletions that then allow the descendants of a particular cell to be determined. These recorders would add a further level of temporal detail to population genetic studies during an infection, allowing new questions to be posed. For example, allowing the comprehensive dissection of parasite recrudescence following a medical or disease-driven compromise in the immune system of the host. When a host becomes immunocompromised, where does the recrudescent population originate from? What is the persister-source for this infection, and then armed with these data are we better able to design improved therapeutic intervention strategies that improve infection outcomes? As generational counters, the use of these recorder-type systems would also provide insight into other basic aspects of parasite biology, such as the relative rate of the replication. Much of what we know comes from knowledge gained from in vitro cell culture of these two parasites. In the case of T. gondii, what is the in vivo growth rate? How is this affected by different tissue niches? Bradyzoites are eponymously described in reference to their slower growth rate relative to faster growing tachyzoites, but does this generalization hold for the entire organism, or is growth rate tuned to the specific nutritional state of the niche environment – as would seem to be the case for adipose-residence T. brucei - and how does this then feed into ideas of persistence?

Technological advances in whole tissue imaging (e.g. light sheet microscopy) [112], tools for multi-parameter imaging in situ imaging (e.g. digital spatial profiling) [113] and spatial proteomics and transcriptomic tools (e.g. 10x/Visium, lightSeq, AutoSTOMP, DeepVisual Proteomics) [114–119] have the potential to enhance our understanding of the spatial distribution of host-pathogen interactions [120]. A recent study by Quintana et al. assessed localization, phenotype, and transcriptional signatures of T. brucei in the mouse brain relative to host cells using 10x/Visium spatial transcriptomics. They found that “slender” and “stumpy” morphology T. brucei were present in distinct brain regions and that neural infection led to the recruitment and expansion of follicular-like T cells in the brain parenchyma [7]. One limitation of tissue imaging and spatial-omic methods is that they typically assess the parasite as a homogenous population without considering the relationships between individual parasites in space and as they disseminate over time. Merging cellular barcoding with imaging techniques has the potential to teach us about host pathogen interactions. For example, if a niche is occupied by a single parasite clone does this make that site resistant or permissive to future colonization events? What is the parasite or local host gene expression programs that facilitate dominant infection by a clone or clearance of less successful lineages? How does abundance or localization of a parasite clone relate to successful transmission? Combining burgeoning technologies in this way has the potential to provide unique insight into potential colonization bottlenecks at host barrier sites, revealing how and where various pathogens enter tissue niches and the host and parasite gene expression programs that contribute to pathogen dissemination and/or clearance.

It is apparent that there remains much to be understood about how stochastic bottlenecks shape protozoan parasite infections, and hopefully with tools to quantify population dynamics, colonization and tissue niche bottlenecks can finally be widened.

Figure 3. Models for trans-vascular dissemination of T. gondii and T. brucei.

Three models have been proposed for T. gondii to traverse the vasculature. (1) Trojan horse: consists of infected immune cells directly transporting T. gondii into tissue niches during extravasation. (2) Paracellular: extracellular T. gondii migrates between endothelial cells that have been compromised or have increased permeability (vasodilation). (3) Transcellular: T. gondii infects and replicates within endothelium then egress directly into the tissue parenchyma. T. brucei is thought to traverse the endothelium by penetrating and subsequently rupturing of the endothelial cells, then invading the extravascular space between cells.

Highlights.

• Stochastic bottlenecks encountered during infection can have a disproportionately large influence on successful parasite colonization.

• The relationship between parasite infection of discrete tissue niches and long-term persistence or transmission is not fully understood.

• Cellular barcoding combined with spatial proteomics and transcriptomics has the potential to connect gene expression phenotypes with infection outcomes during colonization.

Glossary

BBB

Blood-brain barrier

CNS

Central nervous system

GPI

Glycosylphosphatidylinositol

IP

Intraperitoneal

NGS

Next generation sequencing

PMBCs

Peripheral blood mononuclear cells

STAMP

Sequence-tagged analysis of microbial populations

TEM

Transmission electron microscopy

VSG

Variant surface glycoproteins

WITS

Wild-type isogenic tagged strains