Identification of Plasmodium falciparum proteoforms from liver stage models

The chimeric mouse studies were performed at Princeton University. Animals were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, and all protocols (number 1930) were approved by the Institutional Animal Care and Use Committees (IACUC). All facilities are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) International and operate in accordance with the NIH and U.S. Department of Agriculture guidelines and the Animal Welfare Act.

FNRG mice were generated and transplanted as previously described [24, 25]. Female mice between 6 and 10 weeks of age were injected with approximately 1.0 × 10⁶ cryopreserved adult human hepatocytes. Primary human hepatocytes were obtained from BioIVT (Westbury, NY). FNRG mice were cycled on NTBC (Yecuris Inc, Tualatin OR) supplemented in their water to block the build-up of toxic metabolites. FNRG mice were maintained on amoxicillin chow. Hepatocyte engraftment was monitored by ELISA for human albumin.

Levels of human albumin in mouse serum were quantified by ELISA; 96-well flat-bottomed plates (Nunc, Thermo Fischer Scientific, Witham MA) were coated with goat anti-human albumin antibody (1:500, Bethel) in coating buffer (1.59 g Na₂CO₃, 2.93 g NaHCO₃, 1L dH₂O, pH = 9.6) for 1 h at 37 °C. The plates were washed four times with wash buffer (0.05% Tween 20 (Sigma Aldrich, St. Luis MO) in 1× PBS) then incubated with superblock buffer (Fisher Scientific, Hampton NH) for 1 h at 37 °C. Plates were washed twice. Human serum albumin (Sigma Aldrich, St. Luis MO) was diluted to 1 µg/mL in sample diluent (10% Superblock, 90% wash buffer), then serial diluted 1:2 in 135 µL sample diluent to establish an albumin standard. Mouse serum (5 µL) was used for a 1:10 serial dilution in 135 µL sample diluent. The coated plates were incubated for 1 h at 37 °C, then washed three times. Mouse anti-human albumin (50 µL, 1:2000 in sample diluent, Abcam, Cambridge, UK) was added and plates were incubated for 2 h at 37 °C. Plates were washed four times and 50 µL of goat anti-mouse-HRP (1:10,000 in sample diluent, LifeTechnologies, Carlsbad, CA) was added and incubated for 1 h at 37 °C. Plates were washed six times. TMB (100 µL) substrate (Sigma Aldrich, St. Luis, MO) was added and the reaction was stopped with 12.5 µL of 2 N H₂SO₄. Absorbance was read at 450λ on the BertholdTech TriStar (Bad Wildbad, Germany).

The chimeric FNRG mice were infected with 1 × 10⁶ freshly dissected P. falciparum NF54 sporozoites through tail vein injection. Seven days after inoculation mice were sacrificed, chimeric livers were removed, and livers were placed in OCT (optimal cutting temperature) media and immediately frozen at − 80 °C. Mouse liver sections were stained with either anti-CSP Cat #MRA-183A (1:100 BEI resources, Manassas, VA) or anti-P. falciparum HSP70 LifeSpan Biosciences, Inc (Seattle, WA) (1:50) followed by anti-mouse secondary antibody (1:200) or anti-rabbit secondary antibody (1:100) with either Hoechst stain (1:2000) or DAPI.

Sporozoite-infected chimeric mouse livers preserved in OCT media were thawed and washed 30 mL of 1× PBS by centrifugation at 1000×g for 5 min. After centrifugation, the supernatant was removed and replaced with 1 mL of 10% Acetic acid. Infected livers were subjected to dounce homogenization. Liver lysates were centrifuged at 5000×g for 5 min and the supernatant (containing proteoforms) was harvested. The proteoform-containing eluents were immediately treated with 1 mL of 1 M Tris pH 7.5 to neutralize the acetic acid and stabilize proteoforms. Eluents from infected liver lysates were centrifuged through spin filters with a Microcon 10 kDa mass cut-off (Millipore, Burlington, MA) at 10,000×g for 15 min. The retentate was discarded and the flow through was harvested and subjected to desalting over a reverse-phase C8 macrotrap column (Michrom Bioresources, Auburn, CA) and lyophilized via speedvac.

Primary human hepatocytes were cultured as described by Zou et al. [26]. Briefly, primary human donor hepatocytes were purchased from BioIVT, Inc (Baltimore, MD). 200,000 viable hepatocytes from three different human donors were plated per well. The hepatocytes were plated on LabTek^R (ThermoFisher, Watham, MA) chamber slides and inoculated with 100,000 P. falciparum NF54 freshly dissected sporozoites. Following inoculation hepatocytes were washed every 24 h with 1× phosphate-buffered saline and fresh media was provided. PHH cells were fixed and stained with anti-P. falciparum HSP70 LifeSpan Biosciences, Inc (Seattle, WA) (1:50), anti-rabbit secondary antibody (1:100), DAPI, and Evans Blue at 72 and 196 h after inoculation.

Sporozoite-infected hepatocytes were washed in 500 µL of 1× PBS by centrifugation at 1000×g for 5 min. After centrifugation the supernatant was removed and replaced with 500 µL of 10% Acetic acid to liberate proteins and protein fragments. Hepatocyte lysates were centrifuged at 1000×g for 5 min and the supernatant (containing proteoforms) was harvested. The proteoform-containing eluents were immediately treated with 500 µL of 1 M Tris pH 7.5 to neutralize the acetic acid and stabilize proteoforms. Eluents from infected hepatocytes were centrifuged through spin filters with a Microcon 10 kDa mass cut-off filter (Millipore, Burlington, MA) at 10,000×g for 15 min. The retentate was discarded and the flow through was harvested and subjected to desalting over a reverse-phase C8 chromatography column and lyophilized.

The desalted and lyophilized proteoforms were subjected to MudPIT (multi-dimensional-protein-identification-technology) as described previously [27]. Briefly, proteoforms were loaded onto a strong cation exchange column packed in-line with C8 reversed phase chromatography material. The proteoforms were electrosprayed into a ThermoFisher Q Exactive Plus mass spectrometer (ThermoFisher, Bremen, Germany). MS1 resolution was set to a resolution of 70,000. The top 15 ions were selected for fragmentation. MS2 resolution for fragmented proteoform spectra was set to a resolution of 17,500. Proteoforms were sequenced using a dynamic exclusion setting duration of 30 s to identify lower abundance proteoforms.

For monoculture samples, a human P. falciparum (strain NF54) UniProt concatenated database was generated and loaded onto the Galaxy server and searched using TD Portal. For humanized mouse samples, a human-mouse P. falciparum (strain NF54) UniProt concatenated database was generated and loaded onto the Galaxy server and searched using TD Portal. For both human and chimeric mouse samples proteoforms were reported at a 5% FDR cut-off. Host and parasite proteoforms with spectral matches above and below the 5% FDR are included in Additional file 2: Table S1 and Additional file 3: Table S2.

Image processing was performed using Fiji and Nikon elements software (Nikon, Minato, Tokyo, Japan). The area quantification tool was utilized to determine schizont size. The dot identification tool was utilized for merozoite quantification by using it in the DAPI channel and limiting it to the schizont area.

Statistical analysis was performed using Graphpad Prism Software (Graphpad, La Jolla, CA). A nonparametric t test was performed. P values less than 0.01 were considered statistically significant.

While several laboratories have reported gains in malaria liver stage model development, these studies have yet to explore the repertoire of host and parasite proteoforms generated during schizont maturation. Given past reports from human and animal models of live sporozoite challenge, it is clear that attenuated sporozoites can induce a population of CD8+ and CD4+ T cells targeting liver stage antigens [7]. The generation of these T cell populations has formed the basis for a model in which hepatocytes present malaria antigens on MHC class I. Also possible, but largely unexplored, is the idea that developing schizont proteoforms are digested by host or parasite proteases in a manner that does not resemble MHC Class I peptides (Fig. 1). These non-MHC Class I proteoforms could exist as random fragments with no potential for interaction with histocompatibility receptors or these fragments could be on-path for additional digestion and subsequent MHC Class I presentation (Fig. 1).

A typical bottom up mass spectrometry approach to identify malaria proteoforms from liver stage models would digest host and parasite fragments with trypsin prior to mass spectrometry analysis. Digestion of the host and malaria proteoforms prior to analysis obfuscates the structure of the proteoform from the liver tissue. To identify malaria polypeptides in their intact form from malaria liver stage models requires an unbiased approach capable of matching candidate mass spectra from non-digested host and parasite proteoforms. To capture native host and parasite proteoforms from liver stage models, this study utilized a top down mass spectrometry approach.

Towards identification of liver stage malaria proteoforms this study first analysed parasite growth characteristics in state-of-the-art experimental liver stage models including PHHs grown in vitro and human liver chimeric (FNRG) mice [26, 28‐30]. Primary human hepatocytes were transplanted into FNRG mice which after several weeks became highly engrafted as seen by ~ 1 × 10⁴ µg/mL human albumin level in the serum of transplanted animals (Additional file 1: Figure S1). The chimeric FNRG mice were subsequently injected with P. falciparum NF54 sporozoites to form liver stage parasites. PHHs were inoculated with P. falciparum NF54 sporozoites and subsequently stained at different time points with anti-P. falciparum-HSP70 antibodies. Seventy-two hours after sporozoite inoculation of PHHs, parasite schizonts were apparent and smaller than the hepatocyte nucleus (Fig. 2a). Eight days later the parasite schizonts exceeded the size of the hepatocyte nucleus, but merozoite formation was disorganized and the overall growth of the in vitro schizont was stunted (Fig. 2b). In contrast, infected hepatocytes from human liver chimeric mice inoculated with P. falciparum NF54 sporozoites via tail vein injection and stained with anti-P. falciparum-CSP had large, organized schizonts filled with merozoites that dwarfed the size of the hepatocyte nucleus 7 days after inoculation (Fig. 2c). The size of schizonts in human liver chimeric FNRG mice and PHH cultures were quantified (Fig. 3a). Schizonts from engrafted FNRG mouse livers were statistically larger (p = 0.0002) ranging from 260 to 385 nm², while schizonts from PHH mono-cultures were 162 nm² to 212 nm² in size. In addition, there were statistically more numerous merozoites (p < 0.0001) per schizont in FNRG humanized mice ranging from 20 to 50 versus 5 to 18 in PHH mono-cultures (Fig. 3b). These results suggest that while the in vitro PHH model supports schizont development during the first few days after sporozoite inoculation, the chimeric humanized mouse liver model is a more supportive conduit for late liver stage development. This could be due to the fact that the latter human hepatocytes have a transcriptional profile more similar to what is observed in the human liver because they are in the three-dimensional context of the murine liver.

Mild acid elution (10% acetic acid) and molecular weight spin columns were used to extract low molecular weight proteoforms from intact chimeric human/mouse liver tissue and PHHs infected with P. falciparum NF54 sporozoites. The use of 10% acetic acid will result in only acid-soluble proteoforms being identified as a possible limitation of the approach. Due to the aforementioned developmental capacity of the chimeric human/mouse model, livers were harvested 7 days after infection with P. falciparum NF54 sporozoites. PHHs were harvested 96 h after infection prior to the apparent disruption in schizont development observed in Fig. 2b. Soluble, acid-extracted proteoforms were subjected to multi-dimensional chromatography and tandem mass spectrometry followed by bioinformatics analysis using TDPortal and TDViewer [23]. Proteoform MS/MS spectra were searched against a concatenated tripartite Human-Mouse-NF54 database and scrambled decoy database to identify spectral matches from humanized mouse experiments whereas proteoform spectra from human hepatocyte monocultures were searched against a concatenated Human-NF54 database with a matched scrambled decoy database. Top-down bioinformatics tools were employed to: (1) identify fragmented proteoform species isolated in their native state and (2) to accurately control the FDR associated with mass spectrometry identification of proteins and proteoforms as described recently by Leduc et al. [23].

Using TDPortal at a 5% proteoform FDR cutoff, a total of 5343 unique proteins were identified from three different infected human chimeric mouse livers. Host and parasite proteoform spectra identified from these studies that did not pass the 5% FDR cut-off are included in Additional file 2: Table S1 but were not analysed here. In Additional file 2: Table S1 and Additional file 3: Table S2 a global Q-value is included that is the equivalent of an individual FDR value for each proteoform [31]. From three biological replicate human primary hepatocyte (monoculture) samples a total of 1339 total unique proteins were identified. The length distribution of host (Fig. 4a) and parasite proteoforms (Fig. 4b) was analysed from infected chimeric human mouse livers and PHHs. Results showed a mean amino acid length of 24.4 (standard deviation of 11.86), and a median length of 29 for host proteoforms from humanized mouse livers and a mean length of 26.8 (standard deviation of 25.3), and a median length of 17 from primary human hepatocytes. Parasite proteoforms had a mean length of 22.5 amino acids (standard deviation of 6.0), and a median length of 22 from humanized mouse livers and a mean length of 16.4 (standard deviation of 5.9) amino acids, and a median length of 15 in primary human hepatocytes. These results suggest that both host and parasite proteoforms from liver stage models are within the expected size range of MHC Class I peptides (8–12 amino acids), MHC Class II peptides (12–31 amino acids), and randomly degraded proteoform fragments.

To identify host and P. falciparum proteins that were sequenced across different biological replicates of infected humanized mouse livers and infected primary human hepatocytes sample sets were analysed using venn diagrams. Among three humanized infected mouse liver samples 5343 total host proteins (Fig. 5a) and 190 total P. falciparum proteins were identified (Fig. 5b). Conserved proteins from infected chimeric mouse livers, (representing those that were sequenced in two or more samples), numbered 1930 (36% of total) among host proteins (Fig. 5a) and 20 (10.5% of total) among P. falciparum proteins (Fig. 5b). Among infected primary human hepatocytes, a total of 1339 total Human proteins (Fig. 5c), 39 total P. falciparum proteins (Fig. 5d), 573 (42% of total) conserved Human proteins (Fig. 5c), and two (5% of total) conserved P. falciparum proteins were identified across biological replicates (Fig. 5d). The four proteins identified among all infected humanized mouse livers were W7KN90 (an uncharacterized protein), W7K8P5 (26S proteasome regulatory subunit), W7JYB7 (Actin-2), and W7K9G1 (DNA polymerase epsilon catalytic subunit A) (Table 1). Among infected primary human hepatocyte samples a single protein (W7K7Q9), encoding an uncharacterized protein, was sequenced in all three biological replicates (Table 2).

Conserved P. falciparum proteoforms sequenced from infected chimeric mouse livers ranged in length from 9 to 29 amino acids. The C-score, indicating relative level of proteoform characterization among conserved proteoforms ranged from 3 to 1059 (Table 1) in the infected humanized mouse samples. The C-score utilizes native proteoform mass data (precursor ion information) and proteoform fragmentation data to pair the best match against the target proteomes (Human and/or Mouse, and P. falciparum NF54 in this case) [31]. C-scores > 40 indicate extensive characterization, while C-scores between 3 and 40 are identified but only partially characterized [31]. Conserved P. falciparum proteoforms sequenced from infected primary human hepatocytes ranged in length from 10 to 19 amino acids with a C-score ranging from 24 to 274 (Table 2).

Infected chimeric humanized mouse livers contained large mature liver forms. Consistent with the late liver stage two different merozoite surface proteoforms derived from MSP8 and MSP7-like proteins were identified (Table 1) from the infected mouse liver samples. Overall, a strong representation (90%) of proteins expressed at blood stages from the chimeric liver samples was observed whereas 65% of proteins were expressed at sporozoite stages. Fifteen-percent of proteins identified from the chimeric mouse livers contained PEXEL domains.

Among proteins identified in the primary human monoculture samples- both are expressed in sporozoite stages. Proteoforms derived from PTEX150 are expressed in both sporozoite and liver stages. While none of the monoculture proteins contained PEXEL domains, PTEX150 is a structural component of the translocation system that moves parasite molecules from the PVM to the host cytoplasm [32, 33].

Among the proteoforms sequenced from the monoculture and chimeric humanized mouse samples the Immune epitope database and analysis resource (IEDB) MHC Class I immunogenicity tool was used to test which proteoform sequences had the highest likelihood of inducing T cell responses [34]. Proteoforms with the highest IEDB immunogenicity score were derived from the MSP7-like protein identified in humanized mouse samples containing an IEDB score of 0.966. Consistently sequenced proteoforms from chimeric humanized mouse samples had IEDB scores ranging from (− 1.14 to 0.966) (Table 1). Monoculture IEDB scores were moderate and ranged from (0.219 to 0.552) (Table 2).