Methods ethics statement

Use of human blood and serum was approved by the Walter and Eliza Hall Institute of Medical Research Human Ethics Committee under approval numbers 19-05VIC-13 and HREC21/6. All animal procedures were approved by the Walter and Eliza Hall Institute of Medical Research Animal Ethics Committee under approval numbers 2017.014 and 2019.013.

Parasite culture and transfection

Asexual blood stage cultures of P. falciparum were grown in O⁺ erythrocytes (Lifeblood Australia) and maintained in Roswell Park Memorial Institute (RPMI) RPMI 1640 medium, Glutamax, HEPES (ThermoFisher) supplemented with 50 μg/ml hypoxanthine, 20 μg/ml gentamicin, 0.25% (w/v) Albumax II^TM (ThermoFisher), and 5% (v/v) heat-inactivated human serum (Lifeblood, Australia. Agreement number 21-06VIC-01). The parasite cultures were maintained in microisolator boxes gassed with malaria gas mix (94% N₂, 1% O₂ and 5% CO₂) and grown at 37 °C. In this study, the inducible gametocyte producing (iGP2) parasite line was used as the parental line. To stop commitment to gametocyte production, the parasites were cultured in the presence of 2.5 mM D-(+)-glucosamine hydrochloride. Parasite lines containing the human dihydrofolate reductase (hdhfr) gene were maintained on media containing 5 nM WR99210 (WR) or blasticidin deaminase on 11 μM blasticidin S HCl (BSD). The parasitemia of the cultures were monitored by Giemsa-stained thin blood smears, viewed on a microscope at 1000 x magnification. Transfections were performed on late stage segmented schizonts⁷⁰. The Lonza Amaxa Basic Parasite Nucleofector Kit 2 and Amaxa Nucleofector device was used for tranfections, using programme U-33. A full list of the cell lines generated in this study is shown in the Reagents and Resources table. If cell lines were shown to be non-clonal when generated, they were cloned by limiting dilution and individual clones were selected and verified.

Construction of the MGET-GFP reporter plasmid

To generate the MGET-GFP reporter line, we amplified the promoter region of the mget by PCR (PF3D7_1469900) from NF54 gDNA using the SL316/ SL317 oligonucleotides and cloned this fragment into the NotI/NheI restriction sites of the pkiwi003-p230p-bsfGFP plasmid⁷¹ (Supplementary Data 1). The resulting pkiwi-p230p–mget_proGFP plasmid was designed to be integrated into the p230p locus in the genome (PF3D7_0208900). As p230p is required for male gametocytes, we replaced the 5’ and 3’ homology sequences of p230p with sequences from the non-essential pfs47 gene (PF3D7_1346800). The 5’ and 3’ homology arms of Pfs47 were amplified from NF54 gDNA by PCR using the oligonucleotides SL360/ SL361 and SL362/ SL363, respectively and cloned sequentially into the XhoI/ HindIII and SacI/ NotI restriction sites of the pkiwi-p230pmget_pro-GFP plasmid. The final plasmid pkiwi-Pfs47-mget_pro-GFP does not contain a drug selection cassette, allowing for marker-free cell lines.

Appropriate CRISPR-Cas9 guide sequences were selected using the online software CHOPCHOP⁷². To generate the CRISPR-Cas9 guide plasmid for targeting the Pfs47 locus, the complementary guide oligonucleotides (SL364/ SL365) were annealed and Infusion cloned (Takara) into the pDC2-DHFR-Cas9-P230p plasmid, generating the pDC2-DHFR-Cas9-Pfs47. The dhfr coding sequence was replaced with the blasticidin S deaminase gene (bsd) to produce pDC2-BSD-Cas9-Pfs47 for use in cell lines that already had plasmids containing the hdhfr resistance cassette.

Integration of the repair templates into the Pfs47 locus was verified by PCR using the SL 367/ SL346 and SL366/SL347 primers to test for integration and SL 366 and SL367 to test for the presence of wildtype locus. All oligonucleotide and plasmid sequences are detailed in Supplementary Table S1A, B.

All oligonucleotides used are listed in Supplementary Data 5.

Construction of the gene knockout plasmids

The ∆PfGID1, ∆PfGID2, ∆PfGID7, ∆PfGID8, ∆PfGID9, ∆PfYPEL5, ∆GD1, and ∆PfDPL gene knockout plasmids were created by modifying the p1.2-RON3-HA-mNeonGreen plasmid. The 5’ and 3’ homology regions (HR) were PCR amplified from NF54 gDNA using the oligonucleotides outlined in Supplementary Table S1A, and the fragments sequentially cloned into the NotI/ XmaI (5’ HR) and EcoRI/ KasI (3’HR) restriction enzyme sites of the p1.2-RON3-HAmNeonGreen plasmid. The resulting gene KO plasmids were linearised and ready for transfection. This plasmid has a hdhfr drug selection cassette. A full list of the gene KO plasmids and sequences are shown in Supplementary Table S1B.

The CRISPR-Cas9 guide sequences were selected, annealed and inserted into the BtgZI digested pUF1-Cas9G plasmid using InFusion cloning (Takara). The guide sequences used for each gene can be found in Supplementary Table S1 A. The full list of the KO cell lines generated, including the parent line used, is shown in the Reagent and Resource table. The deletion of the target locus was confirmed by PCR using gene and vector specific primers Supplementary Table S1 A.

All oligonucleotides used are list in Supplementary Data 5.

Construction of the epitope tagging plasmids

The PfGID1-HA, PfGID7-HA, PfGID8-HA, PfGID8-Flag-StrepII, PfGD1-HA-mScarlet and PfDPL-HA targeting plasmids were created by modifying the p1.2-RON3-HA-mNeonGreen plasmid. A synthetic gene sequence was designed that contained a region of homology followed by the remaining downstream coding sequence that had been recodoned (Genscript). This 5’ HR recdoned sequence was cloned into the NotI/ XmaI restriction enzyme sites of the p1.2RON3-HA-mNeonGreen plasmid. The 3’ homology regions for each construct were amplified by PCR from 3D7 gDNA and cloned in EcoRI/ KasI sites of the plasmids. The p1.2 GID8Flag-StrepII and p1.2- GD1-HA-mScarlet plasmids were created by replacing the HAmNeonGreen and mNeonGreen sequence in the plasmid with synthesised sequences encoding for Flag-StrepII and mScarlet, respectively. The p1.2-GID8-HA, PfGID8-Flag-StrepII and GD1-HA-mScarlet vectors were further modified by replacement of the dhfr coding sequence with the blasticidin S deaminase gene (bsd). Gene-specific guides were cloned into the pUF1Cas9G as described above.

For the creation of the GD1-overexpression plasmid, a synthetic gene consisting of recodoned full-length GD1-3xHA (Genscript) was cloned into the NheI/ PstI sites of the pkiwi-Pfs47mget_pro-GFP plasmid described above. Next, the promoter sequence of the sexual stage-specific gene Pfs16 (PF3D7_0406200) was amplified by PCR (oligonucleotides SL506/ SL508) and cloned into the SacII/ NheI enzyme sites to create the pkiwi-Pfs47-Pfs16_proGD1-3xHA plasmid.

The full list of oligonucleotides, and plasmid constructs can be found in Supplementary Data 5.

Gametocyte cultures and stage distribution assays

Cultures containing a majority ring stage iGP2 asexual parasites were sorbitol synchronised and returned to the culture at 3% parasitemia without glucosamine in the media to trigger gametocyte conversion (day 2). The next day (day 1), the culture contained trophozoite and schizont stage parasites. The culture media was replaced with fresh media that did not contain glucosamine and placed in a shaking incubator set at 40 rpm overnight to reduce multiple infections in the culture. On day 0, gametocyte rings and asexual rings will be present. To remove asexual stage parasites, the culture media was supplemented with 62 mM N-acetyl-D glucosamine (NAG) for the remaining 12 days of development. Gametocyte development was monitored using microscopy of Giemsa-stained thin blood smears. Gametocyte morphology was visually scored based on previous criteria.

Stage distribution assays were established as described above, and smears were taken daily for the 12 days of the assay. Gametocyte stage was assessed from day 3 onwards by visually scoring the stage of 50–100 gametocytes chosen at random. The percentage of each lifecycle stage was then calculated, and the mean and standard error of the mean from three experiments were plotted.

Gametocyte parasitemia was assessed on day 6. The mean and the standard error of the mean from three experiments was plotted and significance tested by performing an ordinary one-way ANOVA test (Kruskal-Wallis test) in Graphpad Prism.

Gametocyte culture for transmission

Gametocyte cultures for transmission were initiated as described above and grown in RPMI 1640 medium, Glutamax, HEPES (ThermoFisher), supplemented with 50 μg/mL hypoxanthine and 10% (v/v) heat-inactivated human serum (Lifeblood, Australia). A total of 45 mL of culture was used and divided equally into each well of a six-well plate. The media was replaced daily from day 2 until day 12 using a slide warmer set to 37 °C to ensure the gametocytes remained warm. On day 12 of the gametocyte culture, the stage V gametocyte numbers were counted.

Exflagellation assays

A 50 μL sample of culture containing day 12 gametocytes was harvested and transferred to a 1.5 mL microcentrifuge tube. The microcentrifuge tube was kept at 37 °C to prevent premature activation. A 50 μL aliquot of pre-warmed ookinete medium (in RPMI 1640 medium, Glutamax, HEPES (ThermoFisher) supplemented with, 50 μg/mL hypoxanthine, 10% (v/v) heat-inactivated human serum (Lifeblood, Australia) and 100 μM xanthurenic acid) was then added to the microcentrifuge tube and mixed well with a pipette. The mixture was left at room temperature for 15 min and then transferred to the hemocytometer. The number of exflagellation centres and the number of red blood cells in the sample were then counted and expressed as a percentage of exflagellation centres⁷³. Graphpad Prism was used to perform an ordinary one-way ANOVA test (Kruskal-Wallis test to determine significance).

Standard membrane feeding assays

Day 3 post-emergent female Anopheles stephensi mosquitoes were glucose starved for 2-3 h prior to performing standard membrane feeding assays. Between 30–40 mosquitoes were used for each cell line per experiment. Standard membrane feeding assays (SMFAs) were performed using glass water jacket feeders, with pre-stretched parafilm (Sigma-Aldrich) stretched across the large opening in the feeder. The feeder was plumbed into a circulating water bath set to 38 °C via plastic tubing, mounted on top of the mosquito cages and allowed to reach temperature.

While the feeder was warming, the blood meal was prepared by diluting the day 12 stage V gametocytes with uninfected red blood cells to obtain a 500 μL sample at 0.4% parasitemia. This infected red blood cells sample is then mixed with 500 μL of heat-inactivated human serum to obtain a 50% haematocrit solution. All steps are performed, ensuring the temperature of the blood meal does not drop below 37 °C. Immediately, the prepared blood meal was transferred into the glass feeder, and mosquitoes were allowed to feed for 45 minutes. The mosquitoes were left to rest after feeding for another 20-30 minutes. The mosquitoes were then knocked-down with CO₂, and the fully engorged mosquitoes were sorted into a clean container.

Midgut dissections and oocyst counts

Seven days post-bloodmeal, infected mosquitoes were anaesthetised by placing mosquitoes at 20 °C for 5 min and transferred into a Petri dish containing 80% ethanol. Mosquito midguts were dissected from each group and stained with 0.1% mercurochrome for 15 min before being mounted on a glass slide, and the oocysts per midgut were counted by microscopy. Experiments were performed three times. The number of oocysts per midgut and the percentage of midguts containing oocysts (prevalence) was recorded. The mean and standard error of the mean are plotted. Statistical significance was determined by performing an ordinary one-way ANOVA test (Kruskal-Wallis test) in Graphpad Prism.

Generation of Pfg377 antibodies

The sequence corresponding to amino acids 666 to 1146 of Pfg377²⁹ was subcloned into the pET-28a(+)-TEV plasmid to express an N-terminal His-tag followed by a TEV protease cleavage site. Pfg377 was expressed in E. coli BL21 (DE3) cells as an inclusion body and solubilised in 50 mM Tris pH 8, 6 M guanidine HCl. Pfg377 was purified via NiNTA agarose (Qiagen) resin, eluted in 10 mM Tris pH 4.5, 100 mM NaH₂PO₄, 8 M urea and then dialysed against 20 mM Tris pH 8, 20 mM NaCl. Soluble Pfg377 was further purified via a 1 mL HiTrap Q HP column (Cytiva) followed by size exclusion chromatography (S200 Increase 10/300 GL, Cytiva) pre-equilibrated in PBS. Polyclonal antibodies against Pfg377 (R2313) were generated in a rabbit at the WEHI Antibody Facility through three rounds of immunisation with recombinant Pfg377. All procedures were approved by the Walter and Eliza Hall Institute of Medical Research Animal Ethics Committee.

Fluorescence activated cell sorting (FACS) analysis assays

500ul of infected red blood cells (iRBC) were harvested by centrifugation at 500 x g, and the supernatant was discarded. The iRBC pellet was resuspended in 1:10000 Hoechst 33342 (ThermoFisher Scientific) in 1xDPBS. The sample was incubated in the dark at 37 °C for 30 min. The sample was centrifuged at 500 x g, the supernatant discarded, and the sample washed three times in 1 ml of DPBS. The samples were resuspended in 1 ml of DBPS and analysed for the presence of parasites using an LSR II Flow cytometer (Becton Dickinson). The (FACS) gating strategy is shown in Supplementary Fig. 8.

Immunofluorescence assays

Infected red blood cells (iRBC) were harvested by centrifugation at 500 x g, and the supernatant was discarded. The iRBC pellet was washed once in at least 10 volumes of 1 x DPBS, before being fixed in DPBS containing 4% (v/v) paraformaldehyde and 0.0075% (v/v) glutaraldehyde for 30 min at room temperature. The sample was centrifuged at 500 x g, the supernatant discarded, and the fixed iRBCs washed once in 5 volumes of DPBS. The fixed iRBCs were then permeabilised with 0.1% TX100 (v/v) in DPBS for 15 minutes and again centrifuged and washed. All antibody incubation steps were performed in 3% (w/v) BSA in DPBS. Rat α-HA (3F10), 1:300, Merck), rabbit α-Pfg377 (1:1000), rabbit α-GAP45 (R728, 1:1000)⁷⁴ and rabbit α-EXP2 (1:1000)⁷⁵ primary antibodies were incubated with the sample for 1- 2 h at room temperature, washed as described above and the secondary antibodies added to the iRBC and incubated for 1 hour. The native GFP and mScarlet fluorescence were visualised directly, without antibody labelling. goat α-rabbit IgG-conjugated Alexa fluor 488 or 594 (ThermoFisher) antibodies were used. After the secondary antibodies, the samples were stained with 2 µg/mL DAPI in PBS (ThermoFisher) for 10 min and washed a final time before a small sample of the iRBC pellet was placed on a glass slide and mounted under a coverslip and sealed with nail polish before proceeding to imaging.

A Leica Stellaris 8 confocal microscope equipped with a white light laser (WLL) and 405 nm diode laser was used to image the samples. DAPI was excited with the 405 nm diode laser, and wavelengths of 499 nm and 590 nm were selected to excite GFP and Alexa594, respectively. All images were acquired using a 1.4 NA 63x HC PL APO CS2 oil immersion objective. A consistent pixel size of 58 nm was set for all images to satisfy the optimal sampling. DAPI signal was collected on a HyD X detector over a range of 430-494 nm, GFP was collected on a HyD S over a range of 510 – 583 nm, and Alexa594 was collected on a HyD X over a range of 600-750 nm. Both HyD X detectors were operated in digital mode with a gain of 10 V, and the HyD S operated in analogue mode with a gain of 2.5 V. Z-stacks were set to capture a full parasite infected RBC with a system optimised sampling frequency. Between 40–50 parasites were imaged for each biological replicate, with single parasites imaged where possible to simplify the downstream analysis.

Image analysis

All images were analysed using a custom-written Fiji⁷⁶ macro to segment and quantify the mean fluorescence intensity in individual parasites. For all data with GAP45 labelling, the GAP45 channel was used for the mask creation for segmentation. For Pfg377 data, the mask was created using a summed image of both Pfg377 and MGET-GFP signals. For all data the masked channel was projected via a maximum intensity projection then had a gaussian blur, of radius 2, applied. This image was then auto-thresholded using a Li thresholding method before being converted to a mask. A fill holes binary operation was applied followed by three times erosion to account for uneven labelling across different parasites. Finally, a watershed separation was used to separate touching parasites. This mask was then used to measure the mean intensities of MGET-GFP, Pfg377 or GAP45 within the masked areas between the autothreshold values. Data were plotted for all conditions, normalised to the mean intensity for MGET-GFP parasites. All data were plotted using a web app for quantitative comparison of unpaired data⁷⁷.

Western blotting

Parasite cultures were harvested at the desired lifecycle stage, centrifuged at 500 x g and the supernatant discarded. Infected RBC pellets were then lysed with 5 pellet volumes of 0.03% saponin and incubated on ice for 15 min. The samples were centrifuged at 16,000 x g, the supernatants discarded, and the pellets washed 3 time with 1xPBS. The pellets were then resuspended in PBS containing Bolt LDS sample buffer and Bolt sample reducing agent (ThermoFisher). The samples were heated at 95 °C for 10 min and separated on 4–12% Bis-Tris gels (ThermoFisher) using 1 x MOPS buffer for 30 minutes. Gels were transferred to nitrocellulose membranes using the iBLOT 2 platform. Membranes were blocked in 3% skim milk (w/v) PBS for a minimum of 1 h prior to the addition of antibodies. Primary and secondary antibody dilutions were made in 3% (w/v) skim milk/PBS. Primary antibody incubations were performed either overnight at 4 °C or at room temperature for 1 h all secondary incubations were performed at room temperature for 1 h. The membranes were washed three times, 10 min each with PBS-Tween (PBS/0.05% (v/v) Tween 20) after antibody incubations. The following HRP-conjugated primary antibodies were used: mouse α–FlagM2-HRP conjugate (1:500, Merck) Rat α-HA-HRP conjugate (3F10, 1:300, Merck). A rabbit α -HSP 70 antibody (1:1000)⁷⁸ was used as a loading control and was detected using goat α-rabbit antibodies conjugated to HRP (1:1000, Merck). Blots were incubated with Biorad Clarity Western ECL substrate (Bio-Rad) for 5 minutes before being imaged on a ChemiDoc (Bio-Rad).

RNA sequencing

RNA was isolated from the PfGID1, 2, 7, 8 and 9 knockouts and parental iGP2 parasites at day 3, 6, 9 and 12 of gametocyte (five biological replicates for each cell line and time point were performed) development using the Isolate II RNA mini kit (Bioline). The RNA concentration of each sample was calculated using the QuantiFluor RNA system (Promega), and RNA quality was determined using high-sensitivity RNA screen tape analysis (Agilent). The RNA concentrations were then normalised to 1 ng/µL by dilution with the FlexDrop automated dispenser (Revvity), and then re-quantified. 1.1 ng of total RNA was reverse transcribed using SuperScript II (Invitrogen) and processed using an adapted CelSeq2 protocol. Briefly, barcoded samples with unique molecular identifiers (UMIs) were pooled after first- strand cDNA synthesis and second-strand synthesis performed using the MEGAscript T7 transcription kit (ThermoFisher). Amplified RNA was reverse transcribed using a tagged random hexamer without fragmentation, followed by sequencing on a NextSeq 2000 P3 flow cell (Illumina). Technical replicates were performed for each RNA sample.

RNA-seq analysis

RNA-seq reads were mapped to the P. falciparum reference genome (P. falciparum 3D7, PlasmoDB v66) and ERCC spike-in sequences using the Subread aligner (v2.16.0)⁸² and assigned to genes using scPipe (v2.2.0)⁷⁹ with gene annotations also taken from the 3D7 P. falciparum reference genome (PlasmoDB v66). Gene counts were exported as UMI-level and read-level count matrices by scPipe. Following comparisons of the UMI-level and read-level data, the read-level data was used for subsequent analysis. All analyses were performed in R (v4.3.3)⁸⁰ using Bioconductor (v3.18), with differential expression analyse using edgeR (v4.0.1)⁸¹ and limma (v3.58.1). Technical replicates with small library sizes were excluded (17 / 240 technical replicates with library size n = 116 biological replicates with a median of 5,456,385 reads/sample. The ‘filterByExpr’ function from edgeR⁸² was applied to determine which genes had sufficiently large counts to be retained in a differential expression analysis (5313 / 5687 genes), and library sizes were normalised using the trimmed mean of M-values (TMM) method was applied to determine which genes had sufficiently large counts to be retained in a differential expression analysis (5313 / 5687 genes), and library sizes were normalised using the trimmed mean of M-values (TMM) method⁸³.

The initial differential expression analyses used a design matrix consisted of a single factor specifying the ‘cell line’-‘timepoint’ combination of each sample. The ‘voomLmFit’ function from edgeR^84,85 was used to transform count data to log2-counts per million (logCPM), estimate voom precision weights, estimate sample weights, and fit limma linear models while allowing for loss of residual degrees of freedom due to exact zeros. Samples were blocked by ‘cell line replicate’ to account for any correlation of the gene expression measurements taken from the same replicate over time, although the average correlation was low (0.02)⁸⁶. Analysis conducted included comparisons of the same cell line between different days and different cell lines on the same day as required. In all comparisons, differentially expressed genes (DEGs) were defined as those with a Benjamini-Hochberg adjusted P-value 87. The results of the DE analyses were interactively explored using Glimma (v2.12.0)⁸⁸ and visualised as heatmaps using the pheatmap package (v1.0.12). The DEG lists and their gene ontology information are available in Supplementary Table S3 A-T., and fit limma linear models while allowing for loss of residual degrees of freedom due to exact zeros. Samples were blocked by ‘cell line replicate’ to account for any correlation of the gene expression measurements taken from the same replicate over time, although the average correlation was low (0.02)⁸⁶. The DEG lists and their gene ontology information are available in Supplementary Table S3 A-T.

Two types of gene set analyses were performed on the DEGs based on their gene ontology. Published transcript lists were used for these analyses and included lists of the top 100 male and female transcript lists²⁸ and lists of genes involved in translational repression and mRNA binding proteins⁵². The ‘kegga’ function from limma was used to perform over-representation analyses (i.e., hypergeometric test) of DEG lists, separately for up- and down-regulated DEGs. The ‘cameraPR’ function from limma was used to perform competitive gene set tests, accounting for inter-gene correlation, using the empirical Bayes moderated t-statistic. Heatmaps of the log-fold changes and ‘barcode plots’ of the empirical Bayes moderated t-statistics were used to visualise the expression of these genes in each comparison, while the ‘fry’ method was applied to perform a self-contained test of each gene set.

We also considered a time course analysis where the design matrix included the ‘cell line’ and their interactions with a cubic spline fit of ‘timepoint’ for each sample, again using the ‘voomLmFit’ function from edgeR with sample weights and blocking on ‘cell line replicate’. Using this, we tested the following comparisons:

1.Any changes in gene expression over time (within cell line),

a.∆PfGID1KO, e.g.,

i. Empirical Bayes moderated F-test of the coefficients ∆PfGID1:X1, ∆PfGID1:X2, and ∆PfGID1:X3, where X1, X2, and X3 are the 3 terms of the spline fit.

2.Any changes in gene expression over time in any of the KOs, i.e.,

a.Empirical bayes moderated F-test of the coefficients ∆PfGID1:X1, ∆PfGID1:X2, ∆PfGID1:X3, ∆PfGID2:X1, ∆PfGID2:X2, ∆PfGID2:X3,

∆PfGID7:X1, ∆PfGID7:X2, ∆PfGID7:X3, ∆PfGID8:X1, ∆PfGID8:X2, ∆PfGID8:X3, ∆PfGID9:X1, ∆PfGID9:X2, and ∆PfGID9:X3, where X1, X2, and X3 are the 3 terms of the spline fit.

In all comparisons from the time course analysis, DEGs were defined as those with a

Benjamini-Hochberg adjusted P-value 87. The time course DEG lists are available in Supplementary Tables S3 A-T. The results of the time course DE analyses were visualised using customised plots made with ggplot2 (v3.4.4) as well as heatmaps made using the pheatmap package (v1.0.12). Owing to the complexities of grouping the resulting DEGs into ‘similar’ patterns, as compared to the simple ‘up’ or ‘down’ grouping in the non-time course DE analysis, no gene set tests were performed of the DEGs resulting from the time course analysis.

GD1 protein immuno-precipitation and RIP-Seq

Day 6 gametocytes from the GD1-HA parasite lines were harvested (three replicate experiments) and used to conduct RNA Immunoprecipitation Sequencing (RIP-Seq) to identify the mRNAs bound to GD1. For each replicate, two 30 mL culture dishes containing 5% gametocytes were harvested, and the media was discarded. The pellets were lysed in 0.15% (w/v) saponin in 1 × Dulbecco’s phosphate-buffered saline (DPBS) on ice for 15 minutes, before centrifugation at 16,000 x g for 5 minutes. The pellets were washed three times with 1x DPBS, with centrifugation at 16,000 x g for 5 min between washes. The washed pellets were then resuspended in 10 x the pellet volume of extraction buffer (1% NP-40 (ThermoFisher), 150 mM HEPES, and 150 mM NaCl (pH 7.0), supplemented with Complete

Protease Inhibitor Cocktail (Merck)). Samples were sonicated on ice using a Bandelin Sonoplus Sonicator UW3200 at 20% amplitude for 30 s, employing a 1 s on/off pulse cycle. Following sonication, samples were incubated overnight at 4 °C on a rotating platform. The next day, lysates were clarified by centrifugation at 16,000 x g for 20 min at 4 °C, and the resulting supernatant was collected for downstream processing.

Unenriched ‘input’ samples were prepared by reserving 10% of the starting lysate prior to immunoprecipitation, this sample acts as a control. The remaining lysate was subjected to immunoprecipitation (IP) by adding 20 μL of a-HA magnetic beads (Merck) to the cleared supernatant and incubating at 4 °C for 2 h with gentle rotation. Beads were then collected using a DynaMag-2 magnetic rack (ThermoFisher) and sequentially washed, five times with 1 mL of extraction buffer, five times with 1 mL of 0.5 x extraction buffer, and five times with 1 mL of nuclease-free water to reduce nonspecific binding.

For RNA extraction, 100 μL of the ‘input’ sample was mixed with 1 mL of TRIzol reagent (ThermoFisher). The immunoprecipitated RNA was extracted by directly resuspending the washed beads in 1 mL of TRIzol. Both the input and IP samples were then mixed with 200 μL of chloroform, vortexed, and centrifuged at 13,000 x g for 15 min at 4 °C to separate the aqueous phase. RNA was precipitated from the aqueous phase using isopropanol, and the RNA pellets were washed with 70% ethanol, air-dried, and resuspended in nuclease-free water for downstream processing. DNase treatment was performed using the Turbo DNA-Free DNase kit (ThermoFisher) following the manufacturer’s protocol. DNA-free RNA quantification was performed using the Qubit 2.0 fluorimeter (ThermoFisher) using the High Sensitivity RNA reagent (ThermoFisher), and the RNA integrity was assessed on the 4100 Agilent Tapestation System using a High Sensitivity RNA kit.

Input and IP RNA sample libraries were generated using the Illumina Stranded Total RNA-seq Kit (Illumina), including the rRNA depletion. Final libraries were diluted in Nuclease-free water, and the library quality and integrity were assessed using a D5000 DNA kit on a 4100 Agilent Tapestation System (Agilent). Samples were run with 150 paired-end sequencing on the Miniseq platform (Illumina).

The quality of raw sequencing reads was assessed using FASTQC software with default settings. Low-quality reads were filtered and trimmed using Trimmomatic v0.39⁸⁹ applying a sliding window of 4 nucleotides, a minimum average PHRED64 quality score of 20, trimming of leading and trailing sequences by 3 nucleotides, and a minimum read length of 50 nucleotides. The resulting trimmed paired-end reads were then aligned to the P. falciparum 3D7 (PlasmoDB v68) genome using Subread⁹⁰ with default parameters. Gene-wise read counts were subsequently obtained using featureCounts⁹¹ with the P. falciparum 3D7 68 GFF annotation file, applying the -S option to ensure strand specificity and enable accurate quantification of strand-specific reads. To identify enriched transcripts from RIP-Seq data, raw read counts were obtained from both RIP and control (input) samples and merged with transcript length information. Genes with low expression (≤ 5 reads in more than one replicate per condition) and low composite read coverage (√(RIP₁ × RIP₂ × RIP₃ + Control₁ × Control₂ × Control₃) ≤ 10) were filtered out to ensure robust quantification. Transcript expression was normalised using Transcripts Per Million (TPM), and median TPM values were calculated for both RIP and control groups. Enrichment was determined as the log₂-transformed ratio of median TPM values (RIP over control), with values constrained to finite numbers. A two-component Gaussian mixture model (GMM) was fitted to the log₂ enrichment ratio distribution to classify transcripts as enriched or non-enriched based on the log odds ratio (LOD) of posterior probabilities. Transcripts with LOD > 0 were considered enriched. A histogram overlayed with GMM component density curves visualised the enrichment distribution, while a log₁₀-scaled TPM scatter plot highlighted select transcripts of interest. All data processing and visualisation were performed using R with the ggplot2, mixtools, and ggrepel packages. Gene ontology enrichment for the RIP-Seq enriched genes was performed using STRING v12. We visualised interaction networks at a medium confidence score of 0.4, with continuous lines for direct interactions and dashed lines for indirect associations and clustered the network using the Markov Cluster Algorithm with a default inflation score of 3.

Enrichment analysis was run in STRING v12.0 (string-db.org) using the built-in Reactome and Gene Ontology (Biological Process) libraries. The background was the organism-wide Plasmodium falciparum 3D7 proteome. P values were adjusted by Benjamini–Hochberg; terms with FDR (q) 

To identify if GD1-associated transcripts were translationally repressed, we integrated transcriptomic (RNA-Seq) and proteomic datasets from day 6 wild-type and ∆PfGID knockout (KO) gametocytes, focusing on genes with robust mRNA levels (up-regulated or stable transcript level) but low protein abundance that become significantly up-regulated at the protein level in the ∆PfGID.

Immunoprecipitation assays

Five biological replicates were performed for each parasite line and condition. For each replicate, 3 × 30 mL dishes of cultures at 5–10% parasitemia were harvested by centrifugation at 500 x g, and the resulting pellets were lysed using 0.15% (w/v) saponin in 1 x DPBS). Following lysis with saponin, the iRBC pellets were washed three times in DPBS, with centrifugation at 16,000 x g between washes. These pellets were then incubated with 10x the pellet volume of extraction buffer (1% NP40 (Thermo Fisher Scientific), 150 mM HEPES, 150 mM NaCl, pH 7) supplemented with Complete Protease Inhibitors (Merck). These samples were then sonicated on ice using a Bandelin Sonoplus Sonicator UW3200 with pulses set at 20% amplitude for 30 s with a 1 s on/off cycle. The sonicated samples were incubated overnight on a roller at 4 °C. The samples were centrifuged at 16,000 x g for 20 min, and the supernatant was collected. 20 μL of anti-HA magnetic beads (Merck SAE0197-1ML) were added to the supernatant and allowed to bind at 4 °C for 2 h. Beads were collected using a DynaMag2 magnetic rack and washed 5 times with 1 mL of extraction buffer, followed by five 1 mL washes of H₂0 water. Protein was eluted from the beads by adding 150 mL of 0.5% (w/v) SDS in H₂0 prewarmed to 95 °C and incubated on the beads for 2 min. Samples were either examined by western blotting or mass spectrometry.

Sample preparation of IP samples for mass spectrometry analysis

Eluates from immunoprecipitations were prepared for LC-MS/MS analysis using the Filter Assisted Sample Preparation (FASP) method⁹² with the following modifications. Briefly, eluates were added to a Vivacon® 30 kDa MWCO (Sartorius, VN01H22), and proteins were solubilised and reduced in 6 M urea (Wako, Japan) containing 10 mM Tris-(2carboxyethyl)phosphine (TCEP, Sigma-Aldrich) for 30 min, alkylated with 50 mM iodoacetamide (Sigma-Aldrich), then digested with 1 μg trypsin (SoLuTrypsin, Sigma Aldrich) in 50 mM ammonium bicarbonate (AmBic) (Sigma-Aldrich) and incubated overnight at 37 °C. Peptides were eluted with 50 mM AmBic in two 40 μL sequential washes and acidified in 1% formic acid (FA, final concentration). Peptides were lyophilised to dryness using a CentriVap (Labconco) before being reconstituted in 30 µL of 0.1% formic Acid (FA)/2% acetonitrile (ACN), ready for mass spectrometry analysis.

Sample preparation for mass spectrometry-based global proteomics analysis

Five biological replicates were performed for each parasite line and condition. Saponin lysed infected RBC pellets were lysed in 100 µL of preheated (95 °C) buffer (2.5% SDS in 100 mM Tris-HCl, pH 8.5). DNA was hydrolysed by adding 2 µL neat Trifluoroacetic acid (TFA) (Sigma), and lysates were neutralised to pH 8.5 by adding 40 µL of 1 M Tris-HCl. Protein concentration was determined using Pierce™ BCA Protein Assay Kit following the manufacturer’s instructions. Cell lysates (20 µg protein per replicate) were transferred to 0.5 mL LoBind Deep Well plate (Eppendorf) prepared for mass spectrometry analysis using the modified SP3 protocol⁹³ with some modifications. Briefly, samples were subjected to simultaneous reduction and alkylation with a final concentration of 10 mM Tris (2carboxyethyl) phosphine (TCEP) and 40 mM 2-chloracetamide followed by heating at 95 °C for 10 min. Prewashed magnetic PureCube Carboxy agarose beads (20 µl, Cube Biotech) were added to all the samples along with acetonitrile (ACN,70% v/v final concentration) and incubated at room temperature for 20 min. Samples were placed on a magnetic rack, the supernatants discarded, and the beads were washed twice with 70% ethanol and once with neat ACN. ACN was evaporated entirely from the tubes using a CentriVap (Labconco) before the addition of digestion buffer (50 mM Tris-HCl, pH 8) containing 1 µg each of enzymes Lys-C (Wako, 129–02541) and SOLu-Trypsin (Sigma-Aldrich, EMS0004). Trypsin-LysC on-bead digestion was performed with agitation (400 rpm) for 1 h at 37 °C on a ThermoMixer C (Eppendorf). Following digestion, the samples were transferred to pre-equilibrated AttractSPE tips C18 (Tips-C18.T3.200.96 AFF, Affinisep) for sample clean-up. The eluates were lyophilised before being reconstituted in 0.1% FA/2% ACN, ready for mass spectrometry analysis.

DDA-based mass spectrometry analysis of IPs and global samples

Reconstituted peptides (IPs: IGP2, GID8 and DiCre.WT, DiCre.WT.Rapamycin, HA.DPL, HA.DPL.Rapamycin) were run in a data-dependent (DDA) mode on an Orbitrap Eclipse™ Tribrid mass spectrometer interfaced with a Neo Vanquish UHPLC System. Peptides were loaded onto a C18 fused silica column (inner diameter 75 µm, OD 360 µm × 15 cm length, 1.6 µm C18 beads) with an integrated emitter tip (IonOpticks) using pressure-controlled loading with a maximum pressure of 1500 bar and an EASY-nLC source and electrospraying directly into the mass spectrometer. We then employed a linear gradient of 3% to 30% of solvent-B at 400 nl/min flow rate (solvent-B: 80% (by vol) acetonitrile) for 20 min and 30% to 40% solvent-B for 10 min and 35% to 99% solvent-B for 5 min which was maintained at 90% B for 10 min and equilibrated the column high pressure for 2 min comprising a total of 45 min run with a 30 min gradient in a data dependent (DDA) mode. MS1 spectra were acquired in the Orbitrap (R = 120 k; normalised AGC target = standard; MaxIT = Auto; RF Lens = 30%; scan range = 350–1500; profile data). Dynamic exclusion was employed for 30 s, excluding all charge states for a given precursor. Data-dependent MS2 spectra were collected in the Orbitrap for precursors with charge states 2-7 (R = 30k; HCD collision energy mode = fixed; normalised HCD collision energies = 30%; scan range first mass = 120 m/z; normalised AGC target = 200%; MaxIT = 45 ms).

DIA-based mass spectrometry analysis of IPs and global samples

Reconstituted peptides (IPs: CPSF, DiCre, GID7; Global: WT, GID1, GID2, GID8 and GID9) were separated by reverse-phase liquid chromatography on a 15 cm C18 fused silica column with an integrated emitter tip (IonOpticks, ID 75 µm, OD 360 µm, 1.6 µm C18 beads) using a custom nano-flow HPLC system (Thermo Ultimate 300 RSLC Nano-LC, PAL systems CTC autosampler). The HPLC was coupled to a timsTOF Pro (Bruker) equipped with a CaptiveSpray source. Peptides were loaded directly onto the column at a flow rate of 600 nL/min with buffer A (99.9% Milli-Q water, 0.1% FA) and eluted at 400 nL/min on a 30 min linear analytical gradient of increasing buffer B (90% ACN, 0.1% FA) from 2 to 34%. The timsTOF Pro (Bruker) was operated in diaPASEF mode using Compass Hystar 5.1. The settings on the TIMS analyser were as follows: Lock Duty Cycle to 100% with equal accumulation and ramp times of 100 ms, and 1/K0 Start 0.6 V.·/cm2 End 1.6 V·s/cm2, Capillary Voltage 1400 V, Dry Gas 3 l/min, Dry Temp 180 °C. The DIA methods were set up using the instrument firmware (timsTOF control 2.0.18.0) for data-independent isolation of multiple precursor windows within a single TIMS scan. The method included two windows in each diaPASEF scan, with window placement overlapping the diagonal scan line for doubly and triply charged peptides in the m/z – ion mobility plane across 16 × 25 m/z precursor isolation windows (resulting in 32 windows) defined from m/z 400 to 1200, with 1 Da overlap, and CID collision energy ramped stepwise from 20 eV at 0.8 V·s/cm2 to 59 eV at 1.3 V·s/cm2.

Reconstituted peptides (IPs: WT, HA-tagged and WT, GD1-HA in WT cells, GD1-HA in GID7KO cells and global proteomics: IGP, GID1, GID2, GID7) were separated using reverse-phase liquid chromatography on a 15 cm C18 fused silica column with an integrated emitter tip (IonOpticks, ID 75 µm, OD 360 µm, 1.6 µm C18 beads) using a nano-flow HPLC system (Thermo Ultimate 3000 RSLC Nano-LC) that is coupled to a Orbitrap Eclipse Tribrid Mass Spectrometer (Thermo Scientific) using Easy nLC source and electro sprayed directly into the mass spectrometer. Peptides were loaded directly onto the column at a flow rate of 600 nL/min with buffer A (99.9% Milli-Q water, 0.1% FA) and eluted at 400 nL/min on a 30 min linear analytical gradient of increasing buffer B (90% ACN, 0.1% FA) from 2 to 34%. Data was acquired in a data-independent (DIA) mode. MS1 spectra were acquired in the Orbitrap (R = 120 k; normalised AGC target = 300%; MaxIT = custom; RF Lens = 40%; scan range = 375–1500; profile data). Dynamic exclusion was employed for 30 s, excluding all charge states for a given precursor. MS2 spectra were collected in the Orbitrap (R = 30k; first mass = 120 m/z; normalised AGC target = 100%; MaxIT = 54 ms).

Reconstituted peptides (global proteomics: WT, GD1.KO and IGP2.WT, ∆DPL) were separated using reverse-phase liquid chromatography on a 15 cm C18 fused silica column with an integrated emitter tip (IonOpticks, ID 75 µm, OD 360 µm, 1.6 µm C18 beads) on a Neo Vanquish HPLC system coupled to a Thermo Fisher Orbitrap Astral. Peptides were analysed on a 25 min linear analytical gradient of increasing buffer B (80% ACN, 0.1% FA) from 2 to 34%. Data was acquired in a data-independent (DIA) mode. The MS1 settings were as follows: Orbitrap resolution 240,000; scan range m/z 380–980; AGC target 500%. The DIA parameters were as follows: isolation window: custom; HCD collision energy: 25%; scan range: m/z 145–1450; maximum injection time: 3 ms; AGC target: 800%.

Data processing and statistical analysis

Raw DDA MS data (IPs: IGP2, GID8 and DiCre.WT, DiCre.WT.Rapamycin, HA.DPL and HA.DPL.Rapamycin) were searched using MaxQuant 2.0.1.0, incorporating the Andromeda search engine⁹⁴. The analysis was conducted against the P. falciparum 3D7 Reference proteome FASTA database obtained from UniProt (October 2021) and a separate reverse decoy database using a strict trypsin specificity, allowing up to 2 missed cleavages. The minimum required peptide length was set to 8 amino acids. Modifications: Carbamidomethylation of Cys was set as a fixed modification, while N-acetylation of proteins and oxidation of Met were set as variable modifications. First search peptide tolerance was set at 20 ppm, and main search set at 20 ppm (other settings left as default). Maximum peptide mass [Da] was set at 4600. The “match between runs” option in MaxQuant was used to transfer identifications made between runs based on matching precursors with high mass accuracy. Peptide-spectrum match and protein identifications were filtered using a target-decoy approach at an FDR of 1%. Label-free quantification (LFQ) quantification was selected, with a minimum ratio count of 2. Peptide-spectrum match scores and protein identifications were filtered using a target-decoy approach at an FDR of 1%.

Raw DIA data (IPs: CPSF, DiCre, GID7 and WT, HA-tagged and WT, GD1-HA (in WT cells) and GD1-HA (in GID7KO cells) and global proteomics: WT, GID1, GID2, GID8, GID9 and IGP, GID1, GID2, GID7) were analysed using DIA-NN 1.8⁹⁵ in library-free mode. diaPASEF acquired files were searched against reviewed sequences from P. falciparum 3D7 Reference proteome obtained from Uniprot (October 2021) with the following settings: trypsin specificity, peptide length of 7–30 residues, cysteine carbidomethylation as a fixed modification, variable modifications set to N-terminal protein acetylation and oxidation of methionine, the maximum number of missed cleavages at 1. Mass accuracy was set to 10 ppm for both MS1 and MS2 spectra and match between runs (MBR) enabled, and filtering outputs set at a precursor q-value 

Raw DIA data (global proteomics: WT, GD1.KO and IGP2.WT, ∆DPL) were analysed on Spectronaut version 19.2⁹⁶ using BGS factory settings of direct DIA analysis. A combined Human and P. falciparum Uniprot reference proteome (2021) was used for database searching.

All information relating to immuno-precipitation and global proteomics are available in Supplementary Data 6.

Data processing and statistical analysis of IPs

Data processing and statistical analyses were performed in R (v 4.2.0 – v4.4.0). Protein groups flagged as false hits, contaminants, reverse sequences, or ‘only identified by site’ were removed. Protein quantitative intensities were log2-transformed prior to downstream processing. Data quality was assessed by inspecting global log-intensity distributions, the number of detected precursors and proteins per sample and the extent and pattern of missing values. Sample-level structure and potential technical effects were further evaluated using principal component analysis (PCA) and relative log expression (RLE) diagnostics. Proteins were then filtered to retain those quantified in at least 60% of samples within at least one experimental group.

To reduce systematic technical variation, intensities were normalised across samples using cyclic loess normalisation implemented in limma (v3.60.4). Because the IP control and treated groups exhibited markedly different global abundance profiles, cyclic loess normalisation was performed within experimental groups to avoid borrowing information across conditions. No missing-value imputation was performed except for two experiments (IPs: CPSF, DiCre, GID7 and WT, HA-tagged) where missing values were imputed by applying the Barycenter approach (v2-MNAR) implemented in msImpute (v.1.7.0).

Statistical analyses comprised differential abundance testing for pairwise contrasts and multi-group comparisons (ANOVA framework) using limma. For differential analyses, linear models were fitted and moderated using empirical Bayes shrinkage, and p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure, with proteins considered significant at FDR-adjusted p-value 

Data processing and statistical analysis of global proteomics

Data processing and statistical analyses were performed in R (v 4.2.1 – v4.4.2). Proteins lacking proteotypic precursors and those with q-values > 0.01 were excluded. Protein quantitative intensities were log2-transformed prior to downstream processing. Data quality was assessed by inspecting global log-intensity distributions, the number of detected precursors and proteins per sample and the extent and pattern of missing values. Sample-level structure and potential technical effects were further evaluated using principal component analysis (PCA) and relative log expression (RLE) diagnostics. Proteins were then filtered to retain those quantified in at least 60% of samples within at least one experimental group. To reduce systematic technical variation, intensities were normalised across samples (please refer to supplementary Data X for the normalisation strategy implemented in limma). No missing-value imputation was performed for most studies except for one (WT, GID1, GID2, GID8 and GID9) where missing values were imputed by applying Barycenter approach (v2-MNAR) implemented in msImpute (v.1.7.0).

Statistical analyses comprised (i) global longitudinal/trajectory modelling using maSigPro (v1.78.0), where applicable, (ii) unsupervised hierarchical clustering (hclust) to group proteins by similarity in their longitudinal trajectories over time, and (iii) differential abundance testing for pairwise contrasts and multi-group comparisons (ANOVA framework) using limma. Statistical analyses comprised differential abundance testing for pairwise contrasts and multi-group comparisons (ANOVA framework) using limma (v3.52.4). For differential analyses, linear models were fitted and moderated using empirical Bayes shrinkage, and p-values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure, with proteins considered significant at FDR-adjusted p-value 

Protein extraction and digestion of samples for K-GG analysis

Five biological replicates were performed for each parasite line and condition, with parental iGP2 parasites used as a control. Saponin-isolated parasites were pelleted and snap frozen. Protein extraction was performed by adding pre-heated SDC buffer (1% sodium deoxycholate (SDC), 10 mM tris(2-carboxyethyl)phosphine (TCEP), 40 mM chloroacetamide (CAA), 75 mM Tris-HCl at pH = 8.5) to isolated-parasite pellets, resuspended and heated at 85 °C for 10 min. Samples were allowed to cool to room temperature before the addition of universal nuclease (ThermoFisher) and incubated for 5 minutes before centrifugation, and the clarified lysates were collected. Protein concentrations were determined using the BCA assay, and proteins were digested with trypsin/Lys-C mix overnight at 37 °C with a 1:50 enzyme to protein ratio. The digestion was stopped by adding three volumes of 1% trifluoroacetic acid (TFA) in isopropanol and loaded onto SDB-RPS cartridges (Strata™-X-C, 100 mg, Phenomenex Inc.). Columns were activated and pre-equilibrated with 3 mL of 30% methanol (MeOH)/1% TFA and washed with 3 mL of 0.2% TFA. Samples were loaded and washed twice with 3 mL 1% TFA in isopropanol and once with 3 mL 0.2% TFA/2% acetonitrile (ACN). Peptides were eluted twice with 2 mL of 1.25% ammonium hydroxide (NH₄OH)/80% ACN and diluted with water to a final ACN concentration of 30%. The eluates were snap-frozen in liquid nitrogen and lyophilised overnight.

Crosslinking of K-GG antibody

The kGG antibody—licensed to Cell Signalling Technology (PTMScan Ubiquitin Remnant Motif (K-epsilon-GG) Kit #5562) —bound beads were washed three times with 100 mM sodium tetraborate (pH 9.0) and then crosslinked for 30 minutes at room temperature with 0.5 mL of 20 mM dimethylpimelimate in 100 mM sodium borate (pH = 9.0). The cross-linking reaction was then quenched by adding 200 mM ethanolamine (pH = 8.0) and washed twice before being incubated for 2 h after the addition of 200 mM ethanolamine (pH = 8.0). The beads were washed three times with IAP buffer (50 mM MOPS, pH 7.2, 10 mM Na₂HPO₄, 50 mM NaCl).

K-GG peptide enrichment and LC-MS/MS sample preparation

K-GG peptides were resuspended in 0.5 ml of cold IAP buffer (50 mM MOPS pH 7.2, 10 mM Na2HPO4, 50 mM NaCl) and incubated with 2.5 µL of packed cross-linked K-GG antibody-bead conjugate (corresponding to 31 μg of antibody per sample) for 2 h at 4 °C with end-over-end rotation. Beads were transferred to glass-fibre filter stage tips and washed twice with 1 mL of IAP buffer and three times with cold water via centrifugation between each wash.K-GG peptide. The glass-fibre filter stage tips were then stacked onto SDB-RPS stage tips. The SDB-RPS stage tips were preactivated and equilibrated with the addition of 60 µL of isopropanol, 60 µL 80% ACN and 100 µL 0.2% TFA with centrifugation occurring between each addition.

The kGG peptides were eluted off the beads and directly captured onto the SDB-RPS stage tips via two separate TFA (100 µL of 0.15 %) elution and centrifugation steps. Peptides were then desalted by washing the SDB-RPS stage tips with 0.2% TFA/2% ACN, prior to elution with 60 µL 80% ACN/2.5% NH₄OH directly into level 3 SureStart 0.2 mL mass spectrometry vials (ThermoFisher). Peptides were Speedvac dried, resuspended in 10 µL of 0.1% FA/ 2% CAN and 4 µL injected into the mass spectrometer.

LC-MS/MS measurements of k-GG peptides

Peptides were loaded on a 25 cm IonOpticks column, which was maintained at 50 °C using a column oven. A Neo Vanquish UHPLC system (ThermoFisher) was directly coupled online with the mass spectrometer (Eclipse ThermoFisher), and peptides were separated with a binary buffer system of buffer A (0.1% formic acid (FA)) and buffer B (99.9% acetonitrile plus 0.1% FA), at a flow rate of 400 nL/minute. The gradient started at 2% B and increased to 34% in 48 min before increasing to 60% within 5 min and then reaching 90% in 1 min and held for 5 min prior to returning to 2% and re-equilibrated. The mass spectrometer was operated in positive polarity mode with a capillary temperature of 275 °C.

The DIA methods consisted of an MS1 scan (m/z = 350-1650) with an AGC target of 2.6 × 10⁶ and a maximum injection time of 60 ms (R = 120,000). DIA scans were acquired at R = 30,000, with an AGC target of 3 × 10⁵, “auto” for injection time and a default charge state of 3. The spectra were recorded in profile mode, and the stepped collision energy was 25, 27.5, 30% normalised collision energy. 38 non-uniform DIA segments were set to achieve an average of 6 data points per peak.

Raw data processing and analysis of k-GG peptides

MS raw files were processed using DIA-NN 1.8.1. Library-free searching with kGG variable modification enabled with a maximum of two modifications and two-missed cleavages was performed with the P. falciparum 3D7 proteome (Uniprot UP000001450), with MBR enabled and Robust LC quantification.

Precursors were averaged and filtered to retain kGG peptides that were observed in at least 3/5 biological replicates for at least one condition. Peptide intensities variance-stabilisation normalised, imputed using a mixed MLE/MinProb approach and limma differential expression analysis performed using the DEP R package. Significance testing of log2-transformed intensities was performed, and Benjamini-Hochberg corrected p-values 

Statistical analysis

The computational and statistical analyses were executed using the specified open-source software tools in conjunction with the outlined procedures.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.