Background

Malaria caused by Plasmodium vivax presents a significant public health burden on endemic countries and can impede developing economies. The Plasmodium parasite exhibits a complex lifecycle, including a rapid population expansion in the Anopheles mosquito vector, in the host liver, and then in the host bloodstream – the stage causing symptomatic malaria. Unique to P. vivax is its ability to produce dormant forms in the liver, termed hypnozoites (Krotoski et al., 1982). When a human is infected with P. vivax from an infectious mosquito bite, hypnozoites persist long after the initial infection is cleared. They can then resume growth and cause a relapsing blood infection. Hypnozoites cannot be killed by most antimalarial drugs, and the only drugs with hypnozonticidal activity are hemolytic (Chu et al., 2017). Until recently, discovery and development of new hypnozonticidal drugs have been slowed by inadequate methods to culture P. vivax liver stage parasites, including hypnozoites, for weeks or even a month in vitro (Figure 1A). This protocol report describes in detail the different assay versions, improvements, and modifications used to generate data for several published and anticipated reports describing P. vivax liver stage compound screening and drug development.

Previously-described platforms to maintain hepatocytes infected with liver stage parasites in vitro include complex culture methodologies such as extracellular matrix overlays (Dembele et al., 2014), micropatterned co-cultures (March et al., 2013; Gural et al., 2018), microwell devices (Maher et al., 2020), and hepatic spheroids (Chua et al., 2019). The first report of a medium throughput platform capable of long-term hepatocyte culture and robust Plasmodium infection using standard 384-well microtiter plates was described in 2018 (Roth et al., 2018). Termed the version 1 (v1) assay herein, this format included immunofluorescent (IF) staining (Schafer et al., 2018) and high content imaging (HCI) to quantify growth and persistence of P. vivax liver stages within primary human hepatocytes and was able to produce dose response and single point screening data. However, the v1 assay suffered from poor curve fitting due to the lack of replicate wells for characterizing well-to-well variability (Figure 1B). Nonetheless, this report describes the potent effect of ionophores on hypnozoites, making them useful tool compounds for studying hypnozoite biology and technical positive controls for screening small molecules. The utility of ionophores as control compounds is important because the only drugs that clinically prevent relapse, the 8-aminoquinolines (8AQs) tafenoquine and primaquine, do not exhibit rapid hypnozoite killing in vitro (Gural et al., 2018; Chua et al., 2019; Maher et al., 2020) and therefore are not ideal positive controls for screening assays. The v1 assay also demonstrated the different chemosensitivity of immature hypnozoites (day 1-4 post-infection) versus mature hypnozoites (by day 5 post-infection) using the phosphatidylinositol 4-kinase inhibitors, underpinning the concept that P. vivax liver stage assays can be performed in a ‘prophylactic’ or ‘radical cure’ mode depending on when treatment is applied (Zeeman et al., 2016; Gural et al., 2018; Roth et al., 2018) (Figure 1A).

To address the shortcomings of the v1 assay, an improved version 2 (v2) assay was developed (Maher et al., 2021), which includes an expanded dose response (from 8 to 12 points), testing from a higher concentration (from 10 µM to 50 µM), ionophore controls for normalization, and a direct comparison of potency and cytotoxicity to better understand selectivity (Figure 1C). However, two specific deficiencies were noted in the v2 assay. First, when assaying a series of new hypnozonticidal compounds, we frequently obtained a dose response curve with a partial plateau around 50% inhibition. We hypothesized this could be the result of an active compound not having enough time to clear killed hypnozoites from the wells, resulting in dead or dying hypnozoites being aberrantly counted as viable during HCI. By extending the assay 4 days (to 14 days post-seed or 12 days post-infection), we found better clearance of dying hypnozoites that had been treated 7 days earlier (Figure 1D). Second, this platform is unique in that it includes metabolically active hepatocytes, a potential liability when attempting to discover and develop unoptimized small molecules. We performed a series of experiments to characterize the effect of adding 1-aminobenzotriazole (ABT, which has been shown to inhibit cytochrome P450 enzymes) to culture media prior to compound treatment, in an effort to clarify the activity of parent versus possible reactive metabolites against liver stage parasites (Ortiz de Montellano and Mathews, 1981). These improvements led to the version 3 (v3) assay detailed herein (Figure 1D).

Figure 1. Overview of assay modes, versions, and example data. A. In prophylactic assays, cultures are treated shortly after sporozoite infection of hepatocytes, in order to assess activity against immature hypnozoites and early schizonts. In radical cure assays, cultures are treated at 5 days post-sporozoite infection to allow for hypnozoites to mature. The v1 and v2 radical cure assays share the same timecourse, but feature different plate maps and normalization protocols. The v3 radical cure assay includes an extended endpoint (12 days post-infection) and addition of 1-aminobenzotriazole (ABT) to the medium on treatment days. The v4 assay includes a further extended endpoint of 20 days post-infection and addition of a PI4K (phosphatidylinositol 4-kinase) inhibitor (“PI4Ki”) to eliminate schizonts present at 9-10 days post-infection. “Seed” indicates hepatocyte seed, “Infect” indicates infection with sporozoites, “Media” indicates media change, “Treat” indicates compound treatment via the pin tool. The endpoint of all assays is fixation (“Fix”) prior to immunofluorescence staining and high content imaging. B. Example data from a v1 radical cure assay for two hit compounds with activity against hypnozoites. C. Example data from a v2 radical cure assay for a compound with nonselective activity against hypnozoites. The highest three doses of compound show a decrease in hepatic nuclei, indicating that cytotoxic conditions (red circle), not hypnozonticidal activity, are killing hypnozoites (top, red arrow) at those concentrations. Bars represent standard deviation (SD). D. Example data from v2 (orange) and v3 (blue) radical cure assays for two developmental compounds with activity against hypnozoites. The v2 curves show partial top plateaus that become complete inhibited with the longer v3 assay. Bars represent SD. Images and data are adapted fromMaher et al. (2021), used with permission. Figure 1A was adapted fromMaher et al. (2021) and reproduced under a Creative Commons 4.0 license (http://creativecommons.org/licenses/by/4.0/).

The activity of 8AQs against mature hypnozoites has only recently been demonstrated in vitro (Dembélé et al., 2020; Maher et al., 2021). In these reports, both primaquine and tafenoquine are made more potent by co-treatment with the blood schizonticide chloroquine, which is currently co-administered with tafenoquine clinically (Llanos-Cuentas et al., 2014). We observed non-clearance of 8AQ-treated hypnozoites in our v2 and v3 assay formats; thus, we further extended the assay endpoint (22 days post-seed or 20 days post-infection) in order to enable a longer time for dead hypnozoite clearance, as well as observe schizont formation following quiescence as a hypnozoite (Maher et al., 2021). Termed “reactivation in vitro” (Markus, 2020), this is accomplished by addition of PI4Ki at day 9 post-infection—2 days after test compound treatment—to selectively kill schizonts and not hypnozoites. When 20-day cultures are quantified by HCI, schizonts observed most likely originated from forms that were hypnozoites 10 days sooner, during the selective PI4Ki treatment. These modifications resulted in the version 4 (v4) assay described herein (Figure 1A).

While several reagents and steps are critical to the protocol’s success, perhaps the most fundamentally important reagent is the specific production lot of cryopreserved human hepatocytes. Each production lot exhibits different properties (viability, plateability, purity, etc.) that are, in large part, the result of the process used at the company producing the lot. As a result, lots from some production protocols result in cells that are not supportive of P. vivax and are otherwise difficult or impossible to culture in vitro. We have formed a lot screening and selection process which results in validated lots for P. vivax liver stage studies. The donor selection process should be used first to identify a lot suitable for further assay setup and validation (Table 1).

Table 1. Donor Selection Cascade and Criteria

Cascade Step	Summary	Notes
1-Preselection	-Viability > 75% -Cryoplateable -Excess quantity vials available -Pre-tested for microbial contamination -8 × 106 cells/vial -Pure hepatocyte population	Approximately 40% of harvested human livers are cryoplateable, while 80% of livers contain microbial contamination
2-Donor Screen	-Obtain 10-30 donor lots -Culture for 3 weeks -Monitor for Stability and Contamination -Infect with sporozoites of same case, quantify relative liver stage infection efficiency	Approximately 10% of lots tested are noticeably contaminated due to microbial bioburden, while only 5% of lots tested produce robust liver stage parasite numbers
3-Confirmation	-Buy 10-20 vials of the best lot from step 2 -Perform low-throughput drug assays, look for consistency from thaw to thaw and infection to infection -Run a standard set of compounds to chemically validate the lot	Reproducibility from vial to vial can be poor, so ensure all vials perform in the same manner. In this stage, 10% of lots failed the reproducibility test

This protocol includes several technologies or facilities which are often considered ‘infrastructure,’ including an ACL2 insectary, automated liquid handling systems, high-content imaging systems, and drug discovery data management software. It is outside of the scope of this manuscript to provide a full protocol for designing an ACL2 insectary, as well as training scientists how to work in arthropod containment: the knowledge base needed takes months to teach and years to master, but primers are available in the literature (American Committee Of Medical Entomology American Society Of Tropical Medicine And Hygiene, 2019). Likewise, there are many different commercially available systems for automated liquid handling, high content imaging, and drug data management. In this protocol, we describe our use of specific resources (i.e., a Biomek 4000 for automated liquid handling, a Lionheart FX high content imager, and CDD Vault for drug data management) as one example of a complete workflow, but it is outside the scope of this protocol to fully describe the basics of using each. Rather, these systems come with extensive instruction documentation and training modules to enable use. Obtaining patient isolate blood samples infected with P. vivax parasites requires an Institutional Review Board (IRB)-approved Human Subjects Protocol and a medical worker experienced in venipuncture. The elements of a Human Subjects approval are specific to each institution and the governing policies of endemic countries; thus, obtaining Human Subjects approval and demonstrating safe venipuncture are outside the scope of this protocol. Lastly, drug discovery and development is a voluminous field, and a full understanding of this protocol would require a basic understanding of drug discovery and HCI terms and methods.