In the case of malaria, numerous studies have generated omics data during life cycle transitions (Table 1). These include the cataloging of genes (genomics), mRNA transcripts (transcriptomics), translated protein (proteomics), metabolites (metabolomics), and translational repression/de-repression of transcripts as the parasite transitions through its life cycle stages. The genome of the Plasmodium parasite was initially published in 2002 (Gardner et al., 2002), and has been refined since. Initial transcriptomes and proteomes of P. yoelii and P. berghei Liver Stages have been generated, which have provided lists of the components involved in liver stage development, and further revealed the requirement of fatty acid synthesis from both the parasite and host during liver infection (Tarun et al., 2008; Albuquerque et al., 2009; Vaughan et al., 2009).

Insights that originate from transcriptomic analysis of Liver Stage infection reveal that Type I interferons and ER stress are systematically upregulated during liver stage infection and can modulate the level of liver stage infection (Liehl et al., 2014; Miller et al., 2014; Inacio et al., 2015; Kaushansky and Kappe, 2015a). Additional information can be obtained by monitoring changes in the parasite and host during infection under different environmental conditions. These, and related datasets can inform models that predict causality and cellular outcomes. The goal of this effort would be to identify networks of parasite and/or host factors that facilitate the development or demise of the parasite during its infection of the liver.

A major goal of host-pathogen studies is to elucidate specific interactions that dictate success or failure of the pathogen. While transcriptomics and proteomics can catalog changes that occur in the infected cell, a list of alterations alone does not provide mechanistic insight, and is unsatisfying to most cell biologists. Databases of protein interactions in many organisms, including humans, are becoming highly populated (Hein et al., 2015; Huttlin et al., 2015). Yet, many immunopurification—mass-spectrometry (IP-MS) based approaches to study protein-protein interactions do not meet the standard of quantitative and comprehensive. As datasets become larger, statistical tools can be used to predict which interactions are more likely to be specific, compared to commonly identified (abundant or promiscuous) proteins (Mellacheruvu et al., 2013). The pitfalls of qualitative and low throughput data have been partially overcome in model organisms such as yeast, where whole-genome GFP tag libraries have been generated and used in IP-MS experiments, although even these datasets remain incomplete, and are error prone (Ghaemmaghami et al., 2003; Huh et al., 2003; Mellacheruvu et al., 2013).

More sophisticated approaches designed to distinguish between bona fide and spurious interactions are being developed and applied. For example, Isotopic Differentiation of Interactions as Random or Targeted (I-DIRT) and variants (Tackett et al., 2005; Selbach and Mann, 2006; Trinkle-Mulcahy et al., 2008; Byrum et al., 2012; Trinkle-Mulcahy, 2012) exploit isotopic labeling and immunopurification to distinguish between interactions that occur before cell lysis, from those interactions that are introduced during the purification process. While these approaches improve confidence in interactions, they are neither widely adopted, nor have they been applied in genome scale studies.

Since the initial discovery of liver stage parasites by microscopy in 1948 (Shortt and Garnham, 1948), imaging has been an invaluable tool of malaria research. However, most common imaging methods are neither quantitative nor comprehensive, limiting their capacity to inform modeling approaches. This is particularly troubling for applying a systems biology approach, as cellular outcomes are what we aim to predict, but are often poorly defined. Traditional imaging also falls short of reaching the temporal resolution necessary to elucidate the dynamic cellular processes during invasion and throughout liver stage infection.

A number of new imaging modalities enhance our ability to increase resolution, quantification and throughput. A comprehensive review of the advances made in increasing throughput, quantification and resolution in the imaging field is outside the scope of this review, we will highlight some examples that are particularly relevant to malaria research. One example, correlated light microscopy and electron microscopy (CLEM) combines fluorescence microscopy with electron microscopy, thereby increasing the throughput of monitoring rare events like liver stage infection at EM-level resolution (van Rijnsoever et al., 2008), and has already been applied to monitor liver stage development (Grutzke et al., 2014). Intravital imaging (IVM) has been adapted for malaria research and facilitates analysis of live tissue with microscopic resolution to reveal cellular responses that closely mimic in vivo infection, both spatially and temporally (Pittet and Weissleder, 2011; De Niz et al., 2017). Additional instrumentation, such as the Lattice Light Sheet Microscope (Betzig et al., 2006), enhances temporal and spatial resolution, with applications in both in vivo and in vitro systems, which could enable a more quantitative assessment of cellular outcomes.

An essential component of systems biology is the experimental testing of predictions made by modeling efforts. This testing is greatly assisted by the capacity to perform genetic perturbations. Indeed, one of the major shortfalls of employing systems biology is that testing predictions is largely performed by single candidate-based approaches, and thus often fails to recapitulate the complexity of the system. In many cases, it remains difficult to determine if the model is incorrect, or if reductionist approaches cannot fully capture the emergent properties associated with a complex system. New genome-editing approaches, like CRISPR/Cas, can assess multiple perturbations in combination, in both mammalian and parasite genomes, which will facilitate testing more complex models (Cong et al., 2013; Mali et al., 2013; Ghorbal et al., 2014; Wagner et al., 2014; Lu et al., 2016).

In addition to evaluating individual or groups of gene candidates for function, new genome-editing approaches also have the ability to globally evaluate both host and parasite genes. Whole genome CRISPR/Cas9 knockout screens are now common in mammalian cells (Cong et al., 2013; Mali et al., 2013) and can be adapted to the Plasmodium genome. The Plasmodium Genetic Modification Project (PlasmoGEM), a new community resource from the Wellcome Trust Sanger Institute, aims to produce new tools for the genetic modification of malaria parasites at genome scale. This resource has already demonstrated that two-thirds of P. berghei genes contribute to normal blood stage development (Bushell et al., 2017). Subsequent studies should not only focus on the role of parasite genes in other life cycle stages, but also interrogate the role of host genes during each stage of parasite development.

During hepatocyte invasion, the parasite attaches to the host cell, at least partially through circumsporozoite protein (CSP), which interacts directly with highly sulfated proteoglycans (HSPGs) on the cell surface to trigger CSP cleavage, inducing the sporozoite to switch to an invasive state (Table 2A) (Coppi et al., 2007, 2011). Thrombospondin-related anonymous protein (TRAP) is also involved in this process (Kappe et al., 1999; Matuschewski et al., 2002; Morahan et al., 2009). Additionally, Plasmodium proteins P36 and P52 play a role in invasion, parasitophorous vacuole membrane (PVM) formation, and protecting the host against apoptosis (Ishino et al., 2005; van Dijk et al., 2005; Ploemen et al., 2012). How each of these factors works in concert to facilitate productive invasion of the hepatocyte remains unknown.

A collection of host factors have also been described to impact parasite infection (Table 2B). Scavenger Receptor B1 (SRB1) and the tetraspanin CD81 both play roles in cholesterol-rich microdomain formation and are critical for hepatocyte invasion (Silvie et al., 2003; Rodrigues et al., 2008; Yalaoui et al., 2008a; Valacchi et al., 2011). More recently, it has been described that CD81 and SRB1 are involved in invasion in different species; CD81 is required for P. yoelii and P. falciparum infection, but appears to be dispensable for P. berghei and P. vivax infection. SRB1 plays a more substantial role in P. vivax and P. berghei infections (Silvie et al., 2003; Manzoni et al., 2017). It remains unknown if either protein makes contact with the sporozoite, although it has been suggested that SRB1 might directly engage the parasite, whereas CD81 indirectly impacts infection (Yalaoui et al., 2008b; Manzoni et al., 2017). The receptor tyrosine kinase EphA2 is also critical for hepatocyte infection, at least in part by engaging the parasite protein P36 (Kaushansky et al., 2015). While each of these factors contributes to the infection process, how they work in concert, and how changes in one invasion factor impacts another remain unknown. New approaches that integrate biochemical information and omics datasets are well-suited to merge with existing candidate-based research to create a more comprehensive view of the molecular components required for hepatocyte invasion (AlQuraishi et al., 2014; Gujral et al., 2014). The capacity to integrate biochemical data into a more global framework also paves the way for the identification of molecules or networks that could be targeted for intervention.

Once the parasite has taken up residence in the hepatocyte, the sporozoite dedifferentiates over the course of 12 h in rodents, or 2–3 days in humans. This process results in a rounded trophozoite, which is characterized by dramatic changes in the parasite including the disassembly of molecular and cellular structures and the expulsion of invasion machinery (Bano et al., 2007) (reviewed in Kaushansky and Kappe, 2015b; Vaughan and Kappe, 2017). Following dedifferentiation, schizogony begins, which involves the massive replication the genome, and takes place over the course of 2–10 days, depending on the Plasmodium species. During this time, cellular structures including lysosomes and late endosomes sequester around the parasitophorous vacuole membrane and associate with the tubovesicular network (Lopes da Silva et al., 2012; Grutzke et al., 2014). The unfolded protein response is triggered, which promotes endoplasmic reticulum stress and the survival of the Liver Stage parasite (Inacio et al., 2015; Kaushansky and Kappe, 2015a). The most dramatic change, however, is the replication of the liver stage schizont, which produces tens of thousands of merozoites (membrane-bound, haploid, red blood cell invasive forms) that eventually invade erythrocytes during blood stage development. During this process, some parasites survive, and re-wire their host cells to resist certain types of apoptotic stimuli, while others succumb to host cell apoptosis or alternative cell death stimuli (Leiriao et al., 2005; van de Sand et al., 2005; Kaushansky et al., 2013a,b; Douglass et al., 2015). While these dramatic cellular changes have been qualitatively observed, they are rarely monitored quantitatively.

Some molecular determinants have been linked to Liver Stage survival and development. For example, Plasmodium proteins Upregulated in Infectious Sporozoites (UIS) UIS3 and UIS4 have been hypothesized to play an active role in host nutrient acquisition, in part because of the demonstration that UIS3 associates with the host Liver Fatty Acid Binding Protein (L-FABP) (Mikolajczak et al., 2007; Blume et al., 2011; Slavic et al., 2011; Favretto et al., 2013) and localizes both proteins to the PVM (Mueller et al., 2005). Fatty acids of both host and parasite origin, including host phosphatidylcholine, have been demonstrated to be required for optimal liver stage development (Mazumdar et al., 2006; Vaughan et al., 2009; Itoe et al., 2014). How each of these components specifically contributes to the observed cellular changes, and how each factor co-opts host defenses remains unknown. A more quantitative assessment of the cellular changes that occur, matched to molecular information, will enable the development of models that describe networks of host-parasite interactions required for development of the liver stage parasite.

Intracellular pathogens must exit their host cell in order to propagate and survive. The precise strategies they use directly impact their ability to disseminate within a host, transmit to new hosts, and engage or avoid host immune responses. Despite these important functions of exit, detailed investigations into the mechanisms governing Plasmodium exit have been lacking. This process is important not only for our basic understanding of liver stage development, but also for immunity. This is illustrated by the finding that the most potent stimulus of the immune system is elicited by parasites that develop through the liver stage and exit, but cannot undergo replication within the blood stage (Bijker et al., 2013).

Egress from hepatocytes occurs through the rupture of the PVM, followed by destabilization of the actin cytoskeleton, to allow the budding of merozoites from the host cell through the formation of merozoite-filled vesicles (merosomes). These structures are surrounded by a membrane of host origin (Graewe et al., 2011), and have been hypothesized to shuttle merozoites into the bloodstream to begin blood stage infection (Burda et al., 2017). This process inhibits exposure of phosphatidylserine (PS) on the outer surface of the cell, thereby simultaneously ensuring migration of parasites to the bloodstream and protection from host immune responses (Sturm et al., 2006; Tarun et al., 2006; Baer et al., 2007). Recent methodological advances have developed a platform for quantifying exit events (Stanway et al., 2009). This quantification, and a global assessment of molecular changes that occur during exit, will drive the development of models that describe networks of host-parasite interactions that underlie the exit process. Importantly, these networks could then be used to predict host and parasite determinants of dissemination to the blood stream, and the ability to engage or avoid host immune responses.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.