HLA polymorphisms shape divergent outcomes of Toxoplasma and Plasmodium infection in Eastern Indian HbE/β-thalassemia cohort

Abstract

Genetic disorders such as HbE/β-thalassemia (HBT), though deleterious, may undergo positive selection in regions of high parasitic-endemicity, shaping infection outcomes. By integrating NGS and Sanger sequencing from 61 HBT patients and 50 healthy controls, with imaging-based ex-vivo infection-assays and biochemical analysis, we demonstrate, that individuals carrying the HLA-A*33 allele in an Eastern Indian HBT-cohort show significant protection against Toxoplasma gondii infection compared to A*33-negative counterparts. Importantly, this protective phenotype was independent of transfusion frequency. Infection assays in peripheral blood mononuclear cells (PBMCs) confirmed that while parasite entry was unaffected, intracellular replication was markedly restricted in A*33-positive PBMCs, correlating with enhanced CD8⁺ IFN-γ⁺ responses relative to susceptible HLA genotypes. SPR binding studies using recombinant A*33 and Toxoplasma-derived antigenic-peptide SAG2C, revealed that elevated IFN-γ production is linked higher binding affinity of A*33 with SAG2C, leading to improved antigen presentation. Interestingly, A*33 did not confer protection against Plasmodium falciparum, a closely related apicomplexan parasite sharing common ancestry and intracellular lifestyle with Toxoplasma. Instead, NGS-based HLA profiling in this HBT patient-cohort identified a potential negative association between HLA-C*07 and Plasmodium-infection, validated through infection studies in HLA-null K562 cell lines expressing C*07 or A*33. C*07 exhibited strong binding affinity to the Plasmodium-specific MSP3 peptide but not to SAG2C, indicating pathogen-specific antigen recognition with minimal cross-reactivity. These findings highlight the functional interplay between host HLA diversity and apicomplexan antigen specificity, suggesting that selective immune advantages may contribute to the persistence of otherwise deleterious HBT genotypes in Apicomplexan-endemic populations

HLA Polymorphisms shape the divergent outcomes of Toxoplasma and Plasmodium infection in an Eastern Indian HbE/β-thalassemia cohort.

Introduction

β-thalassemia is a prevalent monogenic disorder affecting millions worldwide, characterized by reduced or absent adult hemoglobin (HbA)¹ synthesis, resulting in chronic hemolytic anemia. An estimated 1.5% of the global population (~80–90 million) carries the β-thalassemia trait. In India, the carrier rate is ~2.78%², with substantial regional variation. Importantly, India also reports high prevalence of HbE, a hemoglobin variant common in South Asia,^3,4. Co-inheritance of HbE with β-thalassemia significantly worsens clinical outcomes, resulting in a broad phenotypic spectrum from mild to transfusion-dependent disease⁵. Many patients face lifelong transfusion needs and associated complications, including iron overload, oxidative stress, organ failure, and infection risks, which remain major causes of mortality despite improved blood safety measures.

In regions where hemoglobinopathies and certain infections represent overlapping areas of co-existence, these conditions may interact in complex ways influencing the outcome and clinical manifestations of HbE/β-thalassemia,^6,7. For instance, sickle cell anemia (SCA) provides protection against Plasmodium falciparum in endemic areas, contributing to positive selection of SCA trait. Similarly, the high prevalence of hemoglobinopathies like α-thalassemia, β-thalassemia, and SCA compared to other monogenic disorders reflect their protective advantage against malaria in regions of co-occurrence. Therefore, high frequency of HbE/β-thalassemia in South-East Asia coinciding with high Apicomplexan⁸ burden highlights the need to explore key host genetic factors or blood transfusion frequencies in affecting infection susceptibility.

Among these, HLA alleles play a central role in modulating immune responses and have been linked to disease outcomes in infections, cancer, autoimmune disorders, and hematological conditions^8–10. The highly polymorphic HLA region (6p21) encodes over 250 genes, including classical class I (A, B, C) and class II (DP, DQ, DR) loci. Evidence suggests HLA polymorphisms¹¹ influence malaria susceptibility and are shaped by local selective pressures,^12,13. Epidemiological data also suggest that regions where HbE mutations and infectious diseases co-occur may also drive natural selection of specific HLA alleles,^14,15.

This study investigates the interplay between HLA genetic architecture and infection susceptibility in HbE/β-thalassemia patients from Eastern India cohort, a region burdened by both hemoglobinopathies and Apicomplexan infections. By combining NGS sequencing with cell biology and biochemistry of HbE/β-thalassemia patients and healthy controls, this study identifies key alleles HLA-A*33, C*07, and DQB1*03 that can modulate host immune responses by differentially presenting antigenic peptides of Toxoplasma and Plasmodium, independent of transfusion dependence. These findings underscore the role of HLA-mediated CD8+ responses in shaping host susceptibility towards Apicomplexans, offering insights into allele-specific immune protection by understanding how these associations can aid identifying at-risk individuals, improving overall disease management.

Methodology

Sample collection and classification of the HbE/β-thalassemia patients based on disease severity

To study the phenotypic heterogeneity of the HbE/β-thalassemia, we have collected 71 HbE/β-thalassemia patient samples from N.R.S Medical College & Hospital, Kolkata under the expert supervision of physicians. Clinical and hematological data of participants were collected from medical records or by physical examination. Further, 50 healthy individual samples to serve as Control samples was also collected.

HLA haplotying of the patients using high-resolution NGS and Sanger sequencing method

Genomic DNA was extracted from the collected blood samples using Nucleopore Genetix kit. All gDNA samples (71 HbE/β-thalassemia patients and 10 Healthy Controls) were sequenced by high resolution Illumina-sequencing method (MedGenome Labs). Five HLA loci (A, B, C, DQB1 and DRB1) were sequenced with 150X sequencing depth. Additional 40 control samples were sequenced using the bi-directional Sanger sequencing method.

qRT-PCR and High-Resolution-Melting-Curve Analysis (HRM) to detect the presence of Apicomplexan (Plasmodium and Toxoplasma) infection in HbE/β-thalassemia patient blood

Total genomic DNA extracted from blood samples as per user protocol was used as a template to detect the presence of Apicomplexan (Plasmodium and Toxoplasma). qRT-PCR-HRM assay was performed using an QuantStudio 5 real-time-PCR systems (Applied Biosystems™). The melting curve plots were generated and analysed using QuantStudio^TM Design & Analysis Software v1.5.2 to determine the average melting temperature (T_m) for Plasmodium (Tm = 77.079 ± 1) and Toxoplasma (Tm = 85.247 ± 1). Fresh 10 samples were checked for parasite presence using the same method, and the HLA allelic information was confirmed using PCR. All the primer sequences for RT-qPCR are enlisted in Table 1.

Table 1.

Oligos for Apicomplexan infection and HLA detection via q RT-PCR

Marker	Label	Sequence (5’→3’)
18S rRNA (Plasmodium spp.)	FP	GCTCTTTCTTGATTTCTTGGATG
	RP	AGCAGGTTAAGATCTCGTTCG
rep-529 (Toxoplasma gondii)	FP	GACTACAGACGCGATGCC
	RP	CTCTTCAATTCTCTCCGCCAT
EMP1 (Plasmodium spp.)	FP	AAGAAAACGAATTATTTGGGACA
	RP	AGAAACATCAGTATTCAACGTT
GRA16 (Toxoplasma gondii)	FP	GTGATTCCACCTGACGGATTAG
	RP	GTCGAAAGCACCTTGAAGCTTC
HLA-A*33	FP	ACAACCAGAGCGAGGCCGGTG
	RP	CCCAAGGCTGCTGCCGGTGTG
HLA-A*24	FP	AACCCTCCTCCTGCTACTCTCG
	RP	TCTGGATGGTGTGAGAACCTGGC

Cell culture

Confluent monolayers of human foreskin fibroblasts (HFFs), were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium (GIBCO), supplemented with 10% FBS (GIBCO) and 100 U PenStrep/Gentamycin (GIBCO). The cells were left to adhere for 24 h at 37 °C and 5% CO₂ before infection. K562 cells (HLA-null)^16,17 were used for the HLA overexpressing studies as previously reported^18,19. They were routinely cultured in the similar method, as mentioned above. Isolated PBMCs²⁰ were also cultured in the similar method.

Toxoplasma culture and ex-vivo infection model

Toxoplasma gondii (ΔKu80) and GFP-ΔKu80 were maintained in the same conditions as mentioned above. PBMCs from the HbE/β-thalassemia patients/healthy controls were infected with freshly egressed tachyzoites at a MOI of 1:10, and used for respective experiments.

Plasmodium culture and Synchronization

Plasmodium falciparum 3D7 cultures were maintained by the candle jar method at 37 °C as described previously at 5% haematocrit, with 0.5% albumax (Invitrogen, Carlsbad, CA, USA) and 0.005% hypoxanthine supplemented RPMI 1640 medium. Parasite cultures were synchronized at the ring stage by treatment with 5% sorbitol (Sigma, St. Louis, MO, USA).

In vitro infection

For in vitro infection, K562-HLA-A*33/24 cells were infected with freshly egressed tachyzoites at a MOI of 1:10 for different time points, as per the specific experimental requirement.

For in vitro infection, K562-HLA-C*07 cells were infected with RBCs containing P. falciparum 3D7 parasites (late schizonts) at a 5% parasitemia, for 6 h, 12 h, 24 h, 36 h and 48 h, thereafter fixed in Methanol and stained with Geimsa stain accordingly, to count the total number of infected cells and calculate the infectivity index.

Red/green invasion assay

Freshly egressed tachyzoites were used to infect the A*33, A*24 PBMCs to assess the early invasion by the parasites. The infected cells were spun down at 1000 rpm for 2 min, to facilitate closer interaction of the parasites with cells. The cells were then kept in the 37 °C-water bath for 15 min. Next the cells were transferred to the CO₂ incubator and kept there for 35 mins. Subsequently the cells were fixed in 4% Paraformaldehyde/0.05% Glutaraldehyde for 5 min, and taken for microscopy.

Immunofluorescence Assay (IFA) and confocal microscopy

For IFA, the following antibodies were used - rabbit anti-GAP45 and anti-SAG1. All images were captured using Olympus confocal microscopy (FV3000) and Leica Stellaris 5 Dmi8 demonstration unit. Final image analysis and processing were done with ImageJ. To improve the distinction of different fluorescent signals, pseudo-colors were assigned to fluorescence images, ensuring better differentiation between various cellular components.

Cytokine measurement

IFN-ℽ levels were measured in the infected A*33, A*24 PBMCs supernatant, at different time points post-infection, by ELISA (R&D Systems) following the manufacturer’s manual.

Immunostaining and flow-cytometry based cell binding assay

All flow-cytometric experiments were carried out in a BD-LSRFortessa™ instrument, and analyses were done using FlowJo™ software version 10.6.0 (Becton, Dickinson and Company, Ashland, OR; 2019). To study proliferation of HLA-I expressing cells, and CD8+and CD4 + T-cells, isolated PBMCs (A*33+ and A*33-) infected with Tachyzoites, were stained with anti-CD4 (PerCP-Cy5.5/APC) (BioLegend) (1:100 dilution), anti-CD8 (FITC) (Thermo) (1:50 dilution) and anti-MHC-I (HLA-I) (FITC) (Abclonal) (1:100 dilution) antibodies. Cells were then washed, fixed, and analyzed by flow-cytometry for binding of HLA-I, CD4+ and CD8 + T-cells.

Apoptosis assay

Infection mediated apoptosis in PBMCs (A*33 + , A*33-) was detected by Annexin V Conjugates for Apoptosis Detection staining (Invitrogen). The infected cells were also Giemsa stained and counted at 10X resolution using the Leica bright field microscope to assess the number of viable cells left.

Induction of differentiation in K562 cells to serve as model for P. falciparum infection

Erythroid differentiation was performed in the presence of Recombinant-human-EPO (3IU/ml) for 72 h. Post differentiation, FACS was carried out to check the expression of CD235a (Glycophorin-A), which is essential for the entry of P. falciparum parasites into the cells.

Cloning and expression of his-tag-HLA-C*07, mCherry-A*33 and mCherry-A*24 alleles in K562 cells

pcDNA3.1/C*07/His-tag construct was transfected into K562 cells using Lipofectamine™ 2000(Invitrogen) and screened with 1000 µg/ml G418 disulfide(A1720-Sigma) for over 7 days. The positive clones were confirmed by RT-qPCR and flow-cytometry with anti-MHC-I (HLA-I) antibody for C*07 surface expression. The full length of mCherry was amplified from pcDNA3.1- and cloned into the A*33-pcDNA3.1 backbone plasmids using the NEB-Gibson Assembly Master-Mix (Lot #10229799). Positive clones were confirmed by RT-qPCR, IFA and immunoblot for mcherry-HLA-A*33/24 expression.

Cloning and purification of recombinant C-terminal His-tag-A*33 and His-tag-A*24 proteins

The full length of 6X His-Tag was amplified from the 2.8-lic-HisTag plasmid and cloned into the A*33, A*24 backbone plasmid using the NEB-Gibson Assembly Master Mix. His-Tag/HLA construct was purified from the K562 cells, as per previously used protocols^21,22. The purified protein fractions were then run on a 12% SDS gel to assess their purity, and further protein expression was confirmed by immunoblotting using anti-HisTag antibody.

Immunoblotting analysis

Immunoblot directed against anti-MHC-Class-I (HLA-I), anti-mCherry rabbit polyclonal antibody, anti-His-Tag rabbit was performed as per previously reported protocol²³.

Structural Modeling of A*33, C*07 and Peptide Selection for Docking Studies

Since HLA-Class-I Heavy Chain α covers the entire binding groove domain^24,25 and IMGT/HLA database primarily contains Chain A’s exact sequences, it is the only chain considered for the alignment and structure prediction using Alphafold2, as previously reported^26–28. A total of 15 peptides derived from the protozoan parasite Toxoplasma were docked with both the A*33 and the A*24 using HADDOCK. The binding interactions of 2 peptides (PLA2, SAG2C) with A*33 and A*24 were analysed using Discovery-Studio 2024 (BIOVIA DS, San Diego:Dassault Systèmes, 2024) to identify key residues involved in binding to assess stability of the complexes. These key amino acids were mutated by the other 19 amino acids and modeled using Alphafold2. These mutated peptides were again docked with the HLA-A*33 and A*24 using HADDOCK with the same protocol discussed above.

For the Plasmodium protective allele binding study, two experimentally determined structures of Plasmodium protective allele HLA-C*07, PDB IDs 5VGE and 6PAG, were first evaluated to determine the most suitable template for peptide docking. The RMSD values indicated that 5VGE exhibits a substantially lower RMSD (0.805 Å) compared to 6PAG (1.357 Å), demonstrating a closer structural resemblance to the expected class I HLA binding conformation. Therefore, 5VGE was selected for HLA-C*07 for all docking analyses.

A set of peptide sequences derived from Plasmodium falciparum proteins was provided for computational screening, based on their immunogenicity from previous reports^29,30. Each peptide was docked with both HLA-C*07 (using 5VGE) and HLA-A*33 using the same HADDOCK workflow employed previously. The HADDOCK scores were compared between HLA-C*07 and HLA-A*33 to identify peptides showing preferential binding to HLA-C*07.

Surface plasmon resonance for protein-peptide interaction study

The interaction of A*33, A*24 with wild-type/mutant SAG2C was analyzed using Biacore-T200. A*33 protein was immobilized on a CM5 chip (Cytiva) and different concentrations of SAG2C as an analyte (1–200 μM) were injected at a flow rate of 30 μl/min for 120 s. The results were analyzed using BIA-evaluation-Software 2.0 and Graph Pad Prism 8.0.

The interaction of C*07, A*33 with wild-type MSP3 (analyte) was analyzed in the same process as mentioned above.

Statistical analysis

The statistical analysis was carried out using GraphPad Prism 8 software. The results were presented as Mean ± SD and one-way ANOVA was performed to compare the study groups. Odds ratio (ORs) and 95% confidence intervals were calculated as statistical differences regarding HLA frequencies. p < 0.05 were considered to be statistically significant. All experiments were repeated at least three times and in triplicates. The comparison between two groups were performed by two-tailed Student’s t test. To investigate the association between HLA alleles and parasitemia, Binomial logistic regression analysis was carried out. Infection status (Toxoplasma/Plasmodium) was coded in Binary outcome (1 = present, 0 = absent) and HLA alleles were used as predictor variable. For each type of infection, Logistic regression models were fitted using a generalized linear model (GLM) with a logit function³¹. Statistical analyses were performed in R (version 4.5.2) using the glm function.

Statistics and reproducibility

All experiments were repeated at least three times and in triplicates. To study the phenotypic heterogeneity of the HbE/β-thalassemia, 71 patients were stratified into three different disease severity groups (mild, moderate and severe) as per Mahidol Scoring, a standard scoring method used to grade clinical severity of HbE/β-thalassemia patients. Further 50 healthy control samples were also collected. All the primary findings on population/group reported in this work is validated by in depth mechanistic experiments with statistical associations which also confirms that sample size don’t serve as a constrain for the conclusion drawn from this work. Further, this study is a specific cohort (Eastern-India) based study and whether these allelic functions translate to other cohorts as well or not, warrants further investigation.

Reporting summary

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

Results

Toxoplasma and Plasmodium Infection positivity among HbE/β-thalassemia patients is independent of blood transfusion

To investigate the potential occurrence of Apicomplexan infection in relation to transfusion dependency, 71 HbE/β-thalassemia (HBT) patients were recruited. Based on disease severity and transfusion requirements³², 15 patients were classified as transfusion-dependent thalassemia (TDT), and 46 as non-transfusion-dependent thalassemia (NTDT). High-resolution melting (HRM) curve analysis was used to detect Plasmodium and Toxoplasma DNA in patient blood (Fig. 1.A.i–ii.) (Supplementary Fig 1.A.i–ii.). Among 71 asymptomatic HBT patients, 29 (40.84%) were Plasmodium-positive and 14 (19.71%) were Toxoplasma-positive, whereas none of the 50 healthy controls (HC) shown presence of Toxoplasma or Plasmodium DNA (Fig. 1.B.i–ii.). Importantly, percentage of infection, particularly observed for Plasmodium in our HbE/β-thalassemia cohort, although high, is not rare as previous reports have also demonstrated similar high positivity cases in different pockets of endemic populations^33,34. Moreover, as population stratification can potentially confound the detection of parasitic infection in the healthy control (HC) group, we therefore assessed genetic stratification between the HbE/β-thalassemia (HBT) and HC groups using Multiple Correspondence Analysis (MCA), with the aim of ruling out any bias arising from differences in HLA allele frequencies. The MCA plot (Supplementary Fig 1E) illustrates the distribution of individuals and the overall variation within the categorical HLA dataset. Individuals from both the HbE/β-thalassemia (HBT) and healthy control (HC) groups predominantly cluster near the origin, indicating largely shared and common HLA allele patterns and thereby arguing against the presence of stratification-based bias between the groups, at least in terms of genetic makeup. Co-infection with both pathogens was rare (n = 3), confirming assay specificity as well. Surprisingly, this infection prevalence did not correlate with transfusion dependence (Fig. 1.C.i–ii.). Instead, Plasmodium positivity was significantly higher in NTDT patients (Fig. 1.C.ii.), suggesting additional host factors may influence infection susceptibility. Given previous associations between HLA polymorphisms, transfusion-related outcomes, and parasitic susceptibility, next-generation sequencing (NGS) of class I/II HLA loci was performed on 61 QC-passed HBT samples and 10 HC samples were performed, identifying 92 alleles (Fig. 1.D) (Supplementary Data 2). Five alleles (A*33, C*07, DQB1*03, DQB1*02, DRB1*07) were shortlisted based on occurrence in ≥33% of the cohort (p < 0.0001), which were further validated in 40 additional HC samples via Sanger sequencing (Fig. 1.E.) (Supplementary Data 3). Among the top hits, HLA-A*33 showed a significant negative association with Toxoplasma infection. In Binomial logistic regression analysis assessing Toxoplasma infection in HbE/β-thalassemia, protective allele HLA-A*33 is identified to be significantly associated with reduced risk of infection (OR = 0.10, 95% CI = 0.005–0.60 and p = 0.034). The allele appears to reduce 90% odds of infection, leading to the protective immunity. In contrast, susceptible allele A*24(OR = 1.68 and p = 0.47) and disease neutral allele A*02(OR = 1.40, and p = 0.67) show no significant association with Toxoplasma infection (Supplementary Table 1). Among A*33-positive individuals, only 27.28% were Toxoplasma-positive, while 51.51% were Toxoplasma-negative HBT patients, and 21.21% were HC. A*33 carriers were significantly more likely to be Toxoplasma-negative [OR = 0.51, CI = 0.30–0.90, p = 0.02] and showed enrichment compared to HC [OR = 2.47, CI = 1.39–4.35, p = 0.002] (Fig. 1.F.i.) (Supplementary Table 2), indicating a protective role of A*33.

Fig. 1. Transfusion dependence is not linked with higher prevalence of Toxoplasma/Plasmodium infection among HbE/β-thalassemia patient.

A.i., B.i. qRT-PCR based HRM curve analysis for Toxoplasma rep-529 detection in HbE/β-thalassemia patients (n = 61) and Healthy Controls (n = 50), respectively. (A.ii., B.ii) qRT-PCR based HRM curve analysis for Plasmodium 18S rRNA detection in HbE/β-thalassemia patients (n = 61) and Healthy Controls (n = 50), respectively. C.i–ii. Co-relation analysis between Toxoplasma and Plasmodium infection positivity with Transfusion Dependency in HbE/β-thalassemia patients. D NGS (150X) based raw sequence read of HbE/β-thalassemia patients (HL_001, HL_002 and HL_003) aligned using a FASTA-like algorithm. The dotted box indicates presence of HLA-A*33 in HbE/β-thalassemia patient sample HL_002. E Representative Sanger Sequence Chromatogram of Healthy Control DNA denoting HLA-A*33. The red box marks the position of the substitution. F.i–iii Percentage frequency of HLA alleles (A*33, C*07, DQB1*03) from Toxoplasma and Plasmodium infection, respectively. F.iv. Percentage frequency of HLA allele (DQB1*02) showing susceptibility towards Plasmodium infection. G.i–ii Genomic DNA PCR based detection of protective HLA alleles (A*33, C*07) in HbE/β-thalassemia patients (n = 9). G.iii qRT-PCR based HRM curve analysis for Toxoplasma rep-529 detection in HbE/β-thalassemia patients (n = 9). G.iv qRT-PCR based HRM curve analysis for Plasmodium 18S rRNA detection in HbE/β-thalassemia patients (n = 9). p ≤ 0.05 is marked as *p ≤ 0.01 is marked as ** and p ≤ 0.001 is marked as ***.

For Plasmodium, protective associations were identified with HLA-C*07 and HLA-DQB1*03. In Binomial logistic regression analysis assessing Plasmodium infection in HbE/β-thalassemia, HLA-C*07 emerged as protective allele (OR = 0.29, 95% = 0.078-0.93, and p = 0.44), suggesting a significant reduction in odds of Plasmodium infection in patient cohort having this allele. No significant associations were observed in patients having susceptible allele HLA-A*33(OR = 1.25 and p = 0.70) or disease neutral allele HLA-A*02(OR = 0.91 and p = 0.88) (Supplementary Table 3). Among C*07 carriers, 56.41% were Plasmodium-negative, 23.08% were positive, and 20.51% were HC. Plasmodium infection was less likely among C*07 carriers [OR = 0.41, CI = 0.23–0.71, p = 0.002] versus HC [OR = 2.66, CI = 1.52–4.65, p = 0.0007] (Fig. 1.F.ii.) (Supplementary Table 2). Similarly, DQB1*03 carriers were significantly more likely to be Plasmodium-negative [OR = 0.56, CI = 0.33–0.97, p = 0.04] and enriched relative to HC [OR = 2.27, CI = 1.28–3.94, p = 0.005] (Fig. 1.F.iii.) (Supplementary Table 2). In contrast, HLA-DQB1*02 was positively associated with Plasmodium infection [OR = 1.79, CI = 1.05–3.09, p = 0.03] (Fig. 1.F.iv.) (Supplementary Table 2). To validate these associations, 10 new HBT samples were tested. Seven A*33-positive samples (Fig. 1.G.i.) were all Toxoplasma-negative (Fig. 1.G.iii.), while both C*07-positive (Fig. 1.G.ii.) samples were Plasmodium-negative (Fig. 1.G.iv.). Of the seven C*07-negative samples, five were Plasmodium-positive (Fig. 1.G.iv.), reinforcing the potential protective roles of A*33 and C*07. These findings suggest specific HLA alleles may modulate susceptibility to Apicomplexan infections in HBT patients, independent of transfusion dependency, providing insight into host genetic contributions to infection outcomes.

A*33 restricts intracelluar proliferation of Toxoplama leading to prolonged persistence of infected host

To evaluate the role of HLA-A*33 in restricting Toxoplasma infection, ex vivo assays were performed using PBMCs from HbE/β-thalassemia patients with A*33+ and A*24+ (most prevalent allele in our Toxoplasma infected cohort) genotypes. A*24+ PBMCs showed a progressive increase in intracellular tachyzoite load, while A*33+ cells exhibited a sharp decline in parasite count post 24 hours (h), indicating effective restriction of replication and clearance (Fig. 2.A.i–iii.). Similar assays in healthy A*33+ and A*24+ PBMCs confirmed this protective phenotype (Fig. 2.B.i–iii.). To determine whether reduced tachyzoite burden in A*33+ PBMCs was due to impaired invasion or reduced intracellular proliferation (Fig. 2.C.i–ii.), invasion assays were conducted (Fig. 2.C.iii.). Comparable levels of intracellular (red-GAP45) and extracellular (green-SAG1) parasites were observed in A*33+ and A*24+ PBMCs from both patients and healthy individuals, indicating no significant difference in initial invasion efficiency (Fig. 2.D.i–iv.). Further, significantly declined tachyzoites per parasitophorous vacuole in A*33+ PBMCs were observed between 24 h and 72 h post infection (P.I.), while, A*24+ PBMCs showed increased parasite proliferation with large PV showing 16 or more tachyzoites (Fig. 2.E.i–ii.). Finally, HLA-null K562-cells expressing mCherry-tagged A*33 or A*24 (Supplementary Fig 1.B.i.-iii.) confirmed these findings. A*33-K562 displayed significantly reduced parasite load at all-time points (Fig. 2.F.i–iii., 2.G.i–iii.), invasion efficiency remained comparable for A*33-K562 and A*24-K562 (Supplementary Fig 1.C.i–ii.). Collectively, these results demonstrate that A*33 confers a conserved protective effect against Toxoplasma by restricting intracellular replication, not invasion. Interestingly, apart from limiting tachyzoite proliferation, A*33+ PBMCs remained intact till 120 h P.I., unlike A*24+ PBMCs, which showed gradual lysis 72 h P.I (Supplementary Fig 1.D.i–ii.) (Supplementary Video SV1–SV4). Annexin V-FITC based apoptosis assay further supported this observation, revealing higher late apoptosis (68.6%) and necrosis (11.6%) in A*24+ PBMCs vs. A*33+ at 48h–72 h P.I. (Fig. 3.A–D.), indicating A*33 allele protects against Toxoplasma-induced cell death,^35–37.

Fig. 2. A*33 hinders successful intracellular replication of Toxoplasma tachyzoites.

A.i. Representative Giemsa stain images of A*33 and A*24 infected HbE/β-thalassemia patient PBMCs at 24 h, 72 h and 120 h P.I. Black arrow represents Toxoplasma tachyzoites. N denotes host nucleus. Scale bar 10 μM. A.ii–iii. Intracellular Toxoplasma tachyzoite count and Infectivity Index for 24 h, 72 h and 120 h P.I. Each dot represents count from 100 infected HbE/β-thalassemia patient PBMCs (n = 6). B.i Representative Giemsa stain images of infected A*33 and A*24 Healthy Control PBMCs at 24 h, 72 h and 120 h P.I. Black arrow represents Toxoplasma tachyzoites. N denotes host nucleus. Scale bar 10 μM. B.ii–iii. Intracellular Toxoplasma tachyzoite count and Infectivity Index for 24 h, 72 h and 120 h P.I. Each dot represents count from 100 infected Healthy Control PBMCs (n = 6). C.i–ii. Scheme showing different possibilities of Toxoplasma tachyzoite infection in PBMCs (i) Defective/Delayed Invasion (ii) Successful Invasion but Disruptive/Delayed Replication. C.iii. Schematic representation of differential antibody staining based Invasion assay: SAG1-Extracellular (Green) GAP45-Intracellular (Red) SAG1 + GAP45-Extracellular (Yellow). D.i. Representative confocal images of HbE/β-thalassemia patient PBMCs infected with Toxoplasma tachyzoites using a red/green invasion assay. External tachyzoites are stained green, internal tachyzoites are stained red, host nuclei are DAPI stained-blue. D.ii. Representative confocal images of Healthy control PBMCs infected with Toxoplasma tachyzoites using a red/green invasion assay. External tachyzoites are stained green, internal tachyzoites are stained red and host nuclei are DAPI stained-blue. D.iii–iv. Quantification of % invasion (Total Red tachyzoites/Total Green tachyzoites*100) using the red/green assay. Data is represented as Mean ± SD from 3 independent experiments. E.i–ii. Growth kinetics of Toxoplasma tachyzoites in A*33 and A*24 PBMCs evaluated by counting number of parasites per parasitophorous vacuole (PVs). Percentage of PVs with 2, 4, 8, 16 and >16 parasites at 24 h, 72 h P.I. was plotted. F.i. Representative Giemsa stain images of K562-A*33 Transfected (T) and K562-A*24 Transfected (T) infected cells at 24 h, 72 h and 120 h P.I. Black arrow represents Toxoplasma tachyzoites. N denotes host nucleus. Scale bar 10 μM. Enlarged view of F.i. showing infected cells. F.ii–iii. Infectivity Index and Intracellular Toxoplasma tachyzoite count for 24 h, 72 h and 120 h P.I. Each dot represents count from 100 infected A*33 (T) and A*24 (T) cells (n = 3). G.i. Confocal images of GFP-tagged Toxoplasma tachyzoite infected mCherry-K562-A*33 (T) and mCherry-K562-A*24 (T) cells at 24 h and 72 h P.I. marked with pseudo-colors. Magenta (pseudo) color represents the tachyzoites, green color represents Transfected A*33/A*24-T (HLA-T). In merged panel, white arrow demarcates the tachyzoites. Scale bar 10 μM. G.ii–iii. Infectivity Index and Intracellular Toxoplasma tachyzoite count for 24 h and 72 h P.I. Each dot represents count from 100 infected A*33 (T) and A*24 (T) cells (n = 3). Each experiment was performed in 3 biological replicates and graphical data are represented as Mean ± SD. p ≤ 0.05 is marked as *p ≤ 0.01 is marked as ** and p ≤ 0.001 is marked as ***. Adobe Illustrator have been used to frame (C).

Fig. 3. Toxoplasma infected A*24 PBMCs undergoes Apoptosis.

A.i–iv. Contour Plot representing Flow cytometry evaluation depicting early apoptosis in different groups; A*33 UI, A*33 I; A*24 UI, A*24 I (from left to right) 24 h P.I. B.i–iv. Contour Plot representing Flow cytometry evaluation depicting late apoptosis in different groups; A*33 UI, A*33 I; A*24 UI, A*24 I (from left to right) 48 h P.I. C.i–iv. Contour Plot representing Flow cytometry evaluation depicting necrosis in different groups; A*33 UI, A*33 I; A*24 UI, A*24 I (from left to right) 72 h P.I. D.i–iii. The percentage (%) of live, early apoptotic, late apoptotic and necrotic PBMCs after Toxoplasma infection between the aforementioned groups; 24 h, 48 h and 72 h P.I. respectively (from left to right). Each experiment was performed in biological replicates and graphical data are represented as Mean ± SEM.

A*33 is linked with increased HLA-I presentation and CD8 mediated IFN-ℽ response in Toxoplasma infected host

Since A*33 restricted tachyzoite replication and apoptosis, and CD8-dependent IFN-γ is known to limit intracellular Toxoplasma proliferation,^38–42, it was hypothesized that A*33 enhances antigen presentation, triggering a stronger CD8-mediated IFN-γ response against Toxoplasma infection. To check this, IFN-γ levels were measured in supernatants of infected A*33 and A*24 PBMCs (Fig. 4A). A*33 PBMCs showed a steady increase in IFN-γ, peaking before it plateaued around 120 h, likely due to parasite clearance. In contrast, A*24 PBMCs showed persistently low IFN-γ (Fig. 4A). This response aligned with HLA-I expression: while almost similar at 24 h (Fig. 4.B.ii.), A*33 PBMCs showed significantly increased HLA-I at 72 h), unlike A*24 PBMCs (Fig. 4.B.iii–iv). This increased HLA-I expression was supported by, higher CD8 and CD4 activation in A*33 PBMCs at 72 h P.I. (Fig. 4.C.ii–ii., D.ii–ii.). Notably, CD8 + T-cells, key IFN-γ producers, were more abundant than CD4+ subset in A*33+ PBMCs at 72 h (Fig. 4.E.), supporting the conclusion that A*33 promotes CD8-driven IFN-γ responses through enhanced HLA-I antigen presentation, aiding in effective parasite clearance.

Fig. 4. Toxoplasma infected A*33 PBMCs show significantly high HLA-I expression and CD8 dependent IFN- γ mediated pro-inflammatory response.

A IFN-γ level in the supernatant of Toxoplasma infected A*33 and A*24 PBMCs 24 h, 72 h, 120 h and 168 h P.I. Data is represented as Mean ± SD from 3 independent experiments. B.i. Schematic representation of experimental design for quantifying HLA-I expressing cells. B.ii–iii Representative Flow cytometry-based scatter plot showing enrichment of FITC-HLA-I expressing cells in Infected A*33 and A*24 PBMCs in comparison to the uninfected control group, 24 h and 72 h P.I., respectively. B.iv Bar graph representing Flow cytometry data from 24 h and 72 h P.I., respectively, as in B.i., B.ii. are represented. C.i–ii Representative Flow cytometry-based scatter plot showing enrichment of FITC-CD8 + T cells in Infected A*33 and A*24 PBMCs in comparison to the uninfected control group, 72 h P.I. D.i–ii Representative Flow cytometry-based scatter plot showing enrichment of APC-CD4 + T cells in Infected A*33 and A*24 PBMCs in comparison to the uninfected control group, 72 h P.I. E Bar graph representing Flow cytometry data from 72 h P.I., as in C.i–ii, D.i–ii are represented. Data is represented as Mean ± SD from 3 independent experiments. p ≤ 0.05 is marked as *p ≤ 0.01 is marked as ** and p ≤ 0.001 is marked as ***. Adobe Illustrator have been used to frame (B.i.).

Ser13 residue in Toxoplasma Antigenic peptide SAG2C interacts with antigen binding pocket of A*33

Since A*33 (protective) and A*24 (susceptible) show differential (HLA-I) expression during Toxoplasma infection and (HLA-I) stability depends on peptide binding, it is likely that A*33 presents Toxoplasma antigens more efficiently, leading to CD8 + T cell activation (Fig. 4Ci–ii) and parasite clearance. HLA polymorphisms are known to affect antigen-binding capacity. Sequence alignment with HLAs having available PDB structure showed A*33 shares 95.1% similarity with A*11 in global alignment (Supplementary Fig 2A, Table 2), more than with A*24. Homology based predicted structures confirmed greater 3D structural overlap of A*33 with A*11, particularly within the antigen binding pocket (residues 86–107), showing 86.3% similarity vs. 72.7% with A*24 (Fig. 5Ai–ii). Ramachandran analysis validated stereochemical structure of A*33 model generated with A*11 template, with 83.7% residues in core and 7.2% in allowed regions (Supplementary Fig 2B). HADDOCK-based docking of A*33 and A*24 structures with 15 selected Toxoplasma peptides (Fig. 5C, Table 3) identified PLA2 (−98.7 ± 4.7 kcal/mol) and SAG2C (−104.2 ± 6.5 kcal/mol) as top candidates (Fig. 5Di–iv), due to stronger binding with A*33 and known immunogenicity in clinical infections,^43–45. Prior studies show that even single-point mutations in peptide can alter their HLA specificity,^46,47. Amino acids residues in PLA2 and SAG2C interacting with antigen binding pocket of A*33 (Table 4) were subjected to unbiased in silico single-point substitutions with all 19 alternatives. Mutating key Ser57 in PLA2 yielded affinities similar to wild-type, suggesting stabilization through neighboring interactions, making PLA2 more mutation-resilient. In contrast, replacing SAG2C Ser13 with Thr (S13T) disrupted binding with A*33 (−69.7 ± 2.3 kcal/mol), as did Arg21→Gln (R21E: −75.4 ± 3.4 kcal/mol). Combining both mutations (S13T + R21Q) showed no further drop in affinity (−76.7 ± 5.1 kcal/mol) (Fig. 5Ei–iii., Supplementary Table 4) confirming Ser13 as critical residue for SAG2C-A*33 stability. Eukaryotic expression and purification of recombinant A*33 and A*24 from K562 showed monodisperse SEC peaks (Fig. 6Ai–ii), further confirmed by SDS-PAGE and anti-His western blot (~42 kDa) (Fig. 6Bi–ii). SPR analysis with immobilized A*33 and WT SAG2C peptide (analyte) showed a concentration-dependent binding response peaking at ~150 RU at 200 µM, while A*24 showed minimal response (~15 RU) at the same concentration (Fig. 6Ci–iii). Further, S13T-mutant SAG2C peptide showed markedly reduced binding with A*33 (Fig. 6C.iv), confirming that the structural features of A*33’s antigen-binding pocket enable specific interaction with SAG2C.

Table 2.

Sequence similarity and root mean square deviation (RMSD) of non-hydrogen atoms for different A*33 models

Predicted Model	Template	Sequence Similarity (%)	Binding Site Similarity (%)	RMSD (Å)
Model 1	HLA-A*24 (PDB ID- 7JYV)	93.2	72.7	0.020
Model 2	HLA-A*11 (PDB ID- 7S8Q)	95.1	86.3	0.140

Fig. 5. High affinity binding of HLA-A*33 to Toxoplasma antigenic peptides SAG2C and PLA2.

Ai–ii Superimposition of binding sites in A*33 (AlphaFold2 predicted) modeled structures using A*11 (green) and A*24 (purple) as template, respectively, with A*33 modeled without a template as the reference (red). Dotted lines highlight the binding site (residues 86–107), with A*11 showing the maximum similarity in binding site residues (86.3%). B Superimposed structures of variations in the binding site amino acids A*33 (green) (residues 86–107) and A*24 (purple) (residues 62–83) in panels (i–xii). C Heatmap showing the differential binding HADDOCK score of 15 Toxoplasma antigenic peptides with A*33 and A*24. D.i–iv Representative images of the interactions (black dotted lines) docked poses of Toxoplasma peptides PLA2 and SAG2C with A*33 and A*24, respectively. A*33 is depicted in green color, while A*24 is shown in purple color. Binding affinity is depicted pink. E.i–iii Representative images of the interactions (black dotted lines) docked poses of the mutated Toxoplasma peptide SAG2C with A*33 (i.) Ser13 to Thr13 [S13T], (ii.) Arg21 to Glu21. [R21E], (iii.) Ser13 to Thr13 [S13T] and Arg21 to Glu21 [R21E]. Binding affinity is depicted pink.

Table 3.

Peptide sequence and binding affinity values (kcal/mol) as computed from HADDOCK for the 15 peptides docked with A*33 and A*24 alleles

Peptides	Peptides Sequence	Binding Affinity with A*33	Binding Affinity with A*24
ABCt	FLSDRLFHV	–102.3 ± 6.1	–124.1 ± 11.2
DAGK	RMIHIVLSV	–98.6 ± 6.5	–114.0 ± 2.1
GOT1	FLDRALLTL	-83.7 ± 3.3	–85.5 ± 2.7
GRA5	AVVSLLRLL	–59.3 ± 1.5	–64.3 ± 8.5
GRA6	AMLTAFFLR	–73.2 ± 3.9	–86.6 ± 9.0
LMBR1	ALLALLIYA	–114.5 ± 6.4	–136.4 ± 3.7
MIC1	VLLPVLFGV	–71.2 ± 7.3	–73.5 ± 7.0
PIG-P	FLFFFLWAV	–82.9 ± 7.3	–132.1 ± 7.2
PLA2	FVGSYLVIV	–98.7 ± 4.7	–88.5 ± 4.5
RON8	FLNDKLWKV	–70.4 ± 2.2	–87.2 ± 26.3
SAG1	KSFKDILPK	–62.4 ± 4.6	–80.9 ± 9.0
SAG2C	SMFWPCLLR	–104.2 ± 6.5	–79.3 ± 14.2
SRS52A	SSAHVFSVK	–64.8 ± 0.9	–82.4 ± 2.1
TER1P	FLADLLHSV	–94.8 ± 0.5	–106.2 ± 1.8
VPTA	FVFEFIPQV	–115.0 ± 6.4	–130.4 ± 14.4

The standard error in the binding affinity values is also indicated.

Table 4.

List of Residue-wise interactions of the selected peptide with A*33 and A*24

Peptides	Interactions with A*33		Interactions with A*24
PLA2		HLA------Toxo		HLA-----Toxo
	Hydrogen Bond:	Thr104 – Ser57	Hydrogen Bond:	Thr73 – Val55
	Carbon H-Bond:	Thr104 – Gly56		Thr73 – Ser57
	Hydrophobic Interactions:	Ile97 – Val55	Hydrophobic Interactions:	Ala69 – Val60
		Val100 – Val55		His70 – Val55
		Val100 – Tyr58
	Binding Affinity:	−98.7 ± 4.7 kcal/mol	Binding Affinity:	−88.5 ± 4.5 kcal/mol
PLA2	Hydrogen Bond:	Arg86 – Ser13	Hydrogen Bond:	Arg79 – Arg21
		Gly103 – Arg21		Thr73 – Leu20
	Carbon H-Bond:	Arg89 – Ser13	Carbon H-Bond:	Gly65 – Trp16
		Asn90 – Ser13	Hydrophobic Interactions:	Ala69 – Leu19
		Gln96 – Pro17		Lys68 - Trp16
	Polar Interaction (Amide Groups):	Arg106 – Arg21		Ala69 – Trp16
	Hydrophobic Interactions:	Ala93 – Met14
		Ala93 – Phe15
		Arg99 – Cys18
		Val100 – Cys18
		Val100 – Tyr16
	Binding Affinity:	−104.2 ± 6.5 kcal/mol	Binding Affinity:	−79.3 ± 14.2 kcal/mol

Fig. 6. SPR analysis shows differential interactions between HLA-A*33 and HLA-A*24 with SAG2C peptide.

A.i–ii SEC profile of purified A*33-His Tg (elution volume- 71.2 mL) and A*24-His Tg (elution volume- 74 mL) in Superdex 200 column, showed that it elutes as a monomer, respectively. The black dotted line shows the molecular weight standards. B.i SDS-PAGE gel image of purified A*33-His Tg [~42 kDa] and A*24-His Tg [~42 kDa] is shown with the molecular weight standards (kDa) labeled on the left lane. B.ii Western blot of whole cell lysate showing high expression of A*33-His Tg and A*24-His Tg (anti-His [~42 kDa]) in transfected K562 cells, respectively, while no visible expression was observed in Untransfected (UT) cells, 168 h post transfection. C.i Concentration-dependent significant increase in response (RU) [~150 RU] indicating strong binding, was observed when different concentrations of SAG2C (in μM) were injected as analyte over A*33 immobilized surface. C.ii Concentration-dependent minimal increase in response (RU) [~15 RU]) indicating relatively weaker binding, was observed when different concentrations of SAG2C (in μM) were injected as analyte over A*24 immobilized CM5 sensor chip surface. C.iii Distinct decrease in SPR binding response was observed when similar concentration (100 μM) of SAG2C peptide was injected on the A*33 and A*24 immobilized CM5 sensor chip surface. C.iv A similar sharp decrease in SPR binding response compared to wild-type SAG2C (100 μM) was observed when SAG2C mutant (100 μM) was injected on the A*33 immobilized CM5 sensor chip surface.

Ectopically expressed HLA-C*07-K562 restricts Plasmodium proliferation

Since A*33 displayed a protective role against Toxoplasma, the role of C*07 previously identified to negatively correlate with Plasmodium infection (Fig. 1Fii) was further evaluated to determine if such protective role is restricted only to A*33 or could be attributed as a generalized mechanism in context of other Apicomplexan infection. C*07 was expressed in HLA class-I-null K562-cells⁴⁸ (Fig. 7A) and confirmed by flow cytometry (Fig. 7Bi–iii). For Plasmodium infection model, C*07-K562 cells were erythroid-differentiated using EPO^48–51, showing robust CD235a+ (glycophorin-A) expression (Fig. 7A, Ci–ii), allowing in vitro infection assays, with previously reported Plasmodium susceptible allele, A*33-K562⁵² as control (Supplementary Fig 1A, B). As our NGS revealed no significant association of the same, we still proceeded to check A*33 as a susceptible experimental control. While early-stage parasites were observed in both C*07 and A*33 expressing K562 at 6 h P.I., with comparable ring-stage parasite at 12 h P.I. (Fig. 7Di–ii), major differences emerged as infection progress. While C*07-K562-cells halted parasite development at the ring stage, blocking progression to trophozoite/schizont stages and clearing the infection by 48 h (Fig. 7Diii), A*33-K562-cells, in contrast, supported continued parasite maturation, leading to host cell damage and nuclear fragmentation at 48 h P.I. (Fig. 7Diii). Subsequently, the infectivity index was significantly higher in A*33-expressing cells, while C*07-expressing cells effectively restricted parasite growth (Fig. 7Div). These findings suggest that C*07 confers protection against Plasmodium infection by impairing progression to the mature stages, highlighting HLA specificity as a key factor in determining outcome of Apicomplexan infection.

Fig. 7. C*07 stymies successful expansion of ring stage Plasmodium in spite of similar infectivity in early stages of infection.

A Schematic representation showing Plasmodium infection in C*07 transfected-erythroid differentiated K562 cells. Bi–iii Representative Flow cytometry-based scatter plot showing significant proliferation in population of FITC-HLA-I cells, 72 h and 168 h post C*07 transfection (T) in K652 cells. Ci–ii FACS plot showing different proportions of FITC-CD235a+ erythroid cells 72 h post EPO treatment in K562 cells. Di Representative Giemsa stain images of C*07 (T) and A*33 (T) infected cells at 6 h and 12 h P.I. Black arrow represents Plasmodium parasites. N denotes host nucleus. Scale bar 5 μM. Dii Bar graph plot of Infectivity Index for 6 h and 12 h P.I. Each dot represents count from 100 infected C*07 (T) and A*33 (T) cells (n = 6). Diii Representative Giemsa stain images of C*07 (T) and A*33 (T) infected cells at 24 h, 36 h and 48 h P.I. Black arrow represents Plasmodium parasites. N denotes host nucleus. Scale bar 5 μM. Div Bar graph plot of Infectivity Index for 24 h and 36 h P.I. Each dot represents count from 100 infected C*07 (T) and A*33 (T) cells (n = 6). Data is represented as Mean ± SD from 3 independent experiments. p ≤ 0.05 is marked as *p ≤ 0.01 is marked as ** and p ≤ 0.001 is marked as ***. Adobe Illustrator have been used to frame (A).

Arg79 residue in Plasmodium Antigenic peptide MSP3 interacts with antigen binding pocket of C*07

Screening of the P. falciparum peptide dataset revealed noticeable differences in binding preference between HLA-C*07 and HLA-A*33 (Supplementary Table 5). Among all the examined peptides, the MSP3-derived peptide “GNNQIDSTLKDLVEELSK” showed the most favorable binding with HLA-C*07. HADDOCK scoring demonstrated a substantial difference in affinity: HLA-C*07 (5VGE): –97.8 ± 4.6 kcal/mol and for HLA-A*33: –69.2 ± 8.5 kcal/mol (Fig. 8Ai–ii). This strong difference in binding scores clearly indicates a preferential interaction for binding of the MSP3 peptide with HLA-C*07. Eukaryotic expression and purification of recombinant C*07 and A*33 from K562 showed monodisperse SEC peaks (Fig. 8Bi), which was further confirmed by SDS-PAGE and anti-His western blot (~42 kDa) (Fig. 8Bii–iii). SPR analysis with immobilized C*07 and MSP3 peptide (analyte) showed a concentration-dependent binding response peaking at ~60 RU at 200 μM, while A*33 showed minimal response (~12 RU) at the same concentration (Fig. 8Ci–iii). This confirms the specific and stronger interaction of MSP3 with C*07’s antigen-binding pocket.

Fig. 8. High affinity and differential binding of HLA-C*07 to Plasmodium antigenic peptide MSP3.

Ai–ii Representative images of the interactions (black dotted lines) docked poses of Plasmodium peptide MSP3 with C*07 and A*33, respectively. C*07 is depicted in pink color, while A*33 is shown in green color. Binding affinity is depicted pink in both. Bi SEC profile of purified C*07-His Tg (elution volume- 77 mL) in Superdex 200 column, showed that it elutes as a monomer. The black dotted line shows the molecular weight standards. Bii SDS-PAGE gel image of purified C*07-His Tg [~42 kDa] is shown with the molecular weight standards (kDa) labeled on the right lane. Biii Western blot of whole cell lysate showing high expression of C*07-His Tg (anti-His [~42 kDa]) in transfected K562 cells, respectively, while no visible expression was observed in Untransfected (UT) cells, 168 h post transfection. Ci Concentration-dependent significant increase in response (RU) [~60 RU] indicating strong binding, was observed when different concentrations of MSP3 (in μM) were injected as analyte over C*07 immobilized surface. C.ii Concentration-dependent minimal increase in response (RU) [~12 RU]) indicating relatively weaker binding, was observed when different concentrations of MSP3 (in μM) were injected as analyte over A*33 immobilized CM5 sensor chip surface. C.iii Distinct decrease in SPR binding response was observed when similar concentration (100 μM) of MSP3 peptide was injected on the C*07 and A*33 immobilized CM5 sensor chip surface.

Discussion

The management of HbE/β-thalassemia, a disorder arising from compound heterozygosity of the HbE mutation and β-thalassemia, is particularly challenging due to its pronounced clinical heterogeneity². Accounting for approximately 15–20% of severe thalassemia cases, the disease is marked by asymptomatic to severe anemia, repeated blood transfusion dependence, and other co-morbid conditions, making it very challenging for clinicians to manage the disease. While the core genetic mutations are well-documented, phenotype-modifying factors such as infections and immune components have largely been overlooked although they are critical factors in defining disease severity. In this regard, a higher proportion of HbE/β-thalassemia patients from regions of co-endemicity have been shown to exhibit Plasmodium positivity^53,54, and prior studies suggest that endemic protozoan infections like Plasmodium and Toxoplasma exert selective pressures influencing the disease’s clinical⁵⁵. Interestingly, in this study, initial testing for infection positivity (Plasmodium and Toxoplasma) among a cohort of HbE/β-thalassemia patients from Eastern India revealed no clear correlation with the frequency of blood transfusions. Paradoxically, increased transfusion frequency was associated with a decline in infection positivity (Fig. 1), despite patients being asymptomatic and transfusion protocols being strictly followed. This indicates that transfusion-related infection risk,^56,57 may not solely depend on frequency, challenging earlier assumptions^58,59.

Consistent with this notion, Next-Generation-Sequencing data of individual HbE/β-thalassemia patient samples, indicated protective effects against common Apicomplexan Plasmodium and Toxoplasma infections, among HbE/β-thalassemia patients to be linked with genetic architecture of their Human Leukocyte Antigen (HLA) C*07, DQB1*03, and A*33. In contrast, increased susceptibility to Plasmodium infections was found to be associated with the presence of the DQB1*02 allele in these patients. Notably, the relatively high prevalence of Plasmodium positivity observed in our cohort may also reflect co-morbid factors intrinsic to HbE/β-thalassemia, including splenic dysfunction. Reports from the Centers for Disease Control and Prevention indicate that malaria prevalence, when asymptomatic infections are included, can reach ~55% positivity in high-burden regions such as Madagascar⁶⁰. Consistent with this, studies from the Indian subcontinent have highlighted a substantial discrepancy between reported and true malaria burden; for example, in Gujarat, clinically diagnosed cases accounted for 24%, whereas microscopy-based confirmation revealed a prevalence of 52%, reflecting a 28% underestimation^61,62. Similarly, high parasitaemia rates (~54.6%) have been reported among children in parts of Uganda, likely influenced by age-related immunity³⁴. Together, these observations underscore that although HLA variation may serve as a critical determinant, infection burden in HbE/β-thalassemia likely reflects a convergence of immunogenetic background, disease-associated co-morbidities, and regional epidemiology.

Ex vivo infection assays further validated the allelic associations. PBMCs carrying the A*33 allele, which correlated negatively with Toxoplasma infection, allowed parasite invasion but effectively inhibited intracellular proliferation, leading to clearance, unlike A*24 PBMCs, which showed higher parasite load leading to host cell lysis (Figs. 2 and 3). Mechanistic studies further demonstrated upregulated HLA-I expression (Fig. 4) and increased IFN-γ-producing CD8 + T cell subsets in A*33 infected PBMCs as compared to A*24 (Fig. 4), suggesting enhanced antigen presentation and immune activation.

Despite minor differences in HLA primary sequences, even single nucleotide variations in the antigen-binding pocket can impact immune responses⁶³. Structural flexibility of HLA has been previously identified as critical factor, making host cells either permissible or non-permissible to protozoan infection^64,65. Our study highlights Toxoplasma antigenic peptides SAG2C and PLA2, which are dominantly represented in clinical infection^66,67, bind more strongly to A*33 than to A*24 (Fig. 5). SPR analysis confirmed higher affinity of SAG2C to recombinant A*33, with this interaction disrupted upon altering key binding residues (Fig. 6), providing direct biophysical evidence linking allele-specific peptide binding to functional immunity.

Importantly, A*33-mediated protection appears specific to Toxoplasma, as Plasmodium falciparum infection studies implicated C*07, not A*33, in protective responses (Fig. 7). While A*33 has been recently associated with Plasmodium susceptibility in certain populations⁵², our NGS data showed no significant association of A*33 within HbE/β-thalassemia cohort of Eastern India, though experimental data suggested mild susceptibility. This discrepancy may reflect complex interactions between HbE and infection susceptibility, possibly contributing to balancing selection and regional HLA enrichment.

It is also important to mention that these findings reflect a significant geographical coherence. In Eastern India, we observed high prevalence of both toxoplasmosis (21.2%) and HbE/β-thalassemia (3.92–23.9%), which is in coherence with the correlation analysis from multiple publicly available genetic databases, like INDIGEN⁶⁸, which highlights A*33 as a prevalent allele in this region. Apart from this, in other parts of India, like the southern region, high prevalence of A*33 is again in coherence with the prevalence of both HbE/β-thalassemia trait (4.6%)² and Toxoplasma infection (37.3%). On the contrary, in areas where both HbE/β-thalassemia trait (0.04–0.13%)² and Toxoplasma infection is significantly low (8.8%)⁶⁹, like that in the western part of India, lower prevalence of A*33 is observed (Supplementary Table 6). Hence, whether these allelic functions translate to other cohorts as well or not, warrants further investigation. However, it should also be noted, one of the primary limitations of this study is the relatively small sample size. A limited number of participants reduces the statistical power of the analysis and increases the likelihood of Type II errors^68,69, potentially obscuring subtle biological differences. Small cohorts may also inflate effect sizes and limit the robustness of the findings, and make them specific for the area of the study. Moreover, while our data strongly implicate HLA-mediated antigen presentation as a key determinant of infection outcomes (Figs. 6C, 8A, C), contribution of additional factors, including splenomegaly, splenectomy, age-related immunity, and broader host defence pathways in dictating infection outcome should also be acknowledged. Indeed, recent studies implicating molecules such as SOD3 in regulating interferon-γ and interleukin signaling highlight that HLA function operates within a broader immunological network⁷⁰.

In summary, our findings reveal allele-specific host-pathogen dynamics that likely confer selective fitness advantages in the context of HbE/β-thalassemia, as illustrated in the graphical abstract. These insights provide a mechanistic framework for understanding HLA distribution in endemic regions such as India, which bears a disproportionate burden of neglected tropical diseases. By linking parasite-derived virulence peptides, MHC class I presentation, and disease phenotype, this study underscores the clinical and evolutionary relevance of immunogenetic diversity and supports the rationale for personalized management strategies in genetically and epidemiologically complex settings.

Ethical statement

The purpose of the study was explained to participants and written informed consent was obtained prior to the study. The ethical approval for the study was obtained from both Indian Institute of Technology, Kharagpur as well as NRS Medical College prior to collection. (Approval Nos. IIT/SRIC/DR/2019 dated 06.11.2019, IIT/SRIC/DEAN/2022 dated 12.08.2022; No/NMC/3483 dated 12.07.2019; IIT/SRIC/SAO/2017 dated 11.12.2017; No/NMC/4154 dated 03.08.2016).

Data sharing statement

The sequencing data are publicly available and has already been deposited in the SRA database (NCBI) with BioProject accession number PRJNA1251014 (https://www.ncbi.nlm.nih.gov/sra). All uncropped and unedited image of gels/blots are provided as Supplementary File. All source data relating to all Figs. (1–8) of this study are presented in the Supplementary data 1 excel file.

Author contributions

S.B. conceptualization, sample collection, experiments, analysis, MS preparation. M.R. sample collection, analysis, MS preparation. S.S helped with bioinformatics analysis and modeling studies. DKR helped in performing experiments related to Plasmodium spp. S.M. helped with statistical analysis. S.P. and B.D. conceptualization and clinical insights. P.C.S. and G.M. helped in conceptualization and project administration. M.G. did bioinformatics analysis and supervision, MS review and editing. DG helped with statistical analysis, MS review and editing. M.K.B. supervision of with Plasmodium experiments, MS review and editing. T.K.D. helped in sample collection, conceptualization, clinical insights and MS review. NC. and B.M. conceptualization, conceived and directed the project and prepared the MS with input from all the authors. All authors have read and approved the final version of the manuscript.

Peer review

Peer review information

Communications Biology thanks Guillaume Butler-Laporte and the other, anonymous, reviewers for their contribution to the peer review of this work. Primary Handling Editors: Tobias Goris and George Inglis. A peer review file is available.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-10222-y.