CAR-macrophage therapy: a novel strategy to overcome antimalarial drug resistance by targeting PfEMP1-mediated adhesion

Abstract

Antimalarial drug resistance poses a critical threat to global malaria control efforts. Despite the continuous development of novel antimalarial compounds, the emergence of drug resistance remains inevitable, highlighting the urgent need for paradigm-shifting therapeutic approaches. Here, we propose an innovative chimeric antigen receptor-macrophage (CAR-M) cell therapy that circumvents traditional small-molecule limitations by harnessing the innate phagocytic capacity of macrophages. This strategy exploits the specific adhesive interactions between Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1) and host receptors (CD36, ICAM-1/CD54, EPCR/CD201) to enable targeted recognition and elimination of infected erythrocytes. By engineering macrophages with chimeric antigen receptors directed against PfEMP1-binding domains, we establish a cell-based immunotherapy platform that provides sustained anti-parasitic activity independent of conventional drug susceptibility profiles. This approach represents a fundamental departure from chemical-based interventions by providing a potentially resistance-proof therapeutic modality for drug-resistant malaria.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12936-026-05915-y.

Key points

Strategic Innovation: Circumventing the inevitable resistance to chemical compounds by employing cell-based immunotherapy that targets an essential parasite virulence mechanism

Target Specificity: Exploiting PfEMP1-mediated adhesion to multiple host receptors (CD36, CD54, CD201) as stable recognition epitopes

Biological Advantage: Harnessing the intrinsic phagocytic capacity of macrophages enhanced through synthetic receptor engineering

Therapeutic Durability: Establishing a potentially resistance-proof platform given the essential role of PfEMP1 in parasite survival and pathogenesis

Supplementary Information

The online version contains supplementary material available at 10.1186/s12936-026-05915-y.

Introduction

Malaria remains one of the most devastating infectious diseases worldwide, with an estimated 282 million cases and 610,000 deaths reported in 2024 [1], predominantly affecting sub-Saharan Africa. Antimalarial drug resistance continues to be a global health challenge. Plasmodium falciparum, the most lethal malaria parasite species, has systematically developed resistance to nearly all deployed antimalarial drugs, from chloroquine and sulfadoxine-pyrimethamine to the current frontline artemisinin-based combination therapies (ACTs) [2]. The emergence of artemisinin-resistant strains in Southeast Asia and recent reports of resistance markers in Africa signal an impending therapeutic crisis [3, 4].

The conventional paradigm of antimalarial drug development-centered on the identification of novel chemical entities with anti-parasitic activity-faces an inherent and persistent limitation. Regardless of the initial potency of a newly developed compound, the evolutionary capacity of P. falciparum inevitably leads to resistance through genetic mutations, gene amplifications, or altered drug metabolism [5]. As a result, the introduction of each new antimalarial agent is eventually followed by the selection and spread of resistant parasite strains, progressively diminishing the drug's clinical utility. This predictable pattern of resistance emergence underscores the urgent need for a fundamentally different therapeutic strategy, one that moves beyond traditional chemical entity screening and instead leverages host–pathogen interactions or harnesses host immunity to circumvent the parasite's evolutionary capabilities.

Recent advances in cellular immunotherapy, particularly CAR-T (Chimeric Antigen Receptor T-cell) therapy for cancer, have demonstrated the transformative potential of engineered immune cells [6–8]. However, the unique biology of malaria-wherein parasites reside within erythrocytes and sequester in microvasculature-requires adaptation of this technology to macrophage platforms [9, 10].

Macrophages serve as the body's primary phagocytic cells and naturally clear senescent or damaged erythrocytes [11, 12]. Engineering these cells with chimeric antigen receptors (CAR-M cells) offers several advantages: (1) enhanced specificity through synthetic receptor recognition, (2) potent phagocytic capacity, (3) ability to traffic to the sites of parasite sequestration, and (4) potential for long-term engraftment and sustained therapeutic effect.

PfEMP1 (P. falciparum erythrocyte membrane protein 1) family members are major surface antigens on infected erythrocytes and mediate adhesion to multiple human receptors including CD36, intercellular adhesion molecule 1 (ICAM-1, also known as CD54), and endothelial protein C receptor (EPCR, also known as CD201) [13–19]. The invariable surface presentation of PfEMP1 on infected red blood cells (iRBCs)-despite its antigenic variation via var gene switching-combined with its expression regardless of parasite drug resistance status, makes it an ideal and stable target for immunotherapeutic intervention [20].

PfEMP1 expression and surface presentation vary with parasite genotype and intraerythrocytic stage, influencing adhesion phenotypes, sequestration, and innate immune recognition [21]. Macrophages are professional phagocytes with intrinsic capacity to clear infected cells; CAR‑M technology enables engineering of macrophages to express chimeric antigen receptors that provide highly specific recognition and enhanced phagocytic responses toward chosen targets.

Here we integrate these biological and technological insights to propose a host‑directed therapeutic concept: PfEMP1‑targeted CAR‑M therapy designed to enhance macrophage recognition and promote phagocytic clearance of iRBCs. This strategy aims to reduce the cumulative drug exposure required for clinical cure, thereby decreasing the selective environment that favors resistance. To provide a rational basis for CAR‑M design choices and to prioritize experimental conditions, we performed an orthogonal analysis of how parasite strain, RBC stage, and parasitemia level influence THP‑1 activation and receptor‑related readouts.

Materials and methods

Parasite culture and preparation

Plasmodium falciparum strains Pf3D7 and PfC580Y were cultured in human red blood cells (RBCs) following standard protocols [22]. To obtain parasites at defined developmental stages, the cultures were synchronized using established methods. Parasites were subsequently harvested at three distinct time points corresponding to key stages of the intraerythrocytic cycle: the ring stage (8–12 h post-invasion), the trophozoite stage (24–30 h post-invasion), and the schizont stage (40–44 h post-invasion). To evaluate the effect of parasitemia on CAR-M adhesion, infected red blood cells (iRBCs) were diluted with normal RBCs (nRBCs) in complete culture medium to achieve three target parasitemia levels: 0.1%, 1%, and 5%.

Generation and characterization of CAR-M Cells

Four distinct stable cell lines derived from PMA-treated THP-1 cells were utilized: a control line (CAR-M/CTL stable cell line, CAR-M/CTL_SCL), one overexpressing CD36 (CAR-M/CD36_SCL), one overexpressing CD54 (CAR-M/CD54_SCL), and one overexpressing CD201 (CAR-M/CD201_SCL) (Fig. 1). These CAR-M cells were seeded in 6-well plates (Corning, NY, USA) at a density of 5 × 10⁵ cells per well. To induce differentiation into macrophage-like cells, the cells were stimulated with 10 ng/mL phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich) for 24 h, which promoted cell adhesion and differentiation. The expression of CD36, CD54, and CD201 on the respective CAR-M lines was confirmed by confocal microscopy and real-time PCR, revealing specific overexpression compared to the control line. The detailed methodological descriptions were listed in Supplementary material.

Fig. 1.

The mechanism and vector map of chimeric antigen receptor-M cells. A CAR-M cells are generated by transducing macrophages with lentivirus encoding the CAR construct. These engineered cells specifically recognize surface antigens on infected red blood cells via their single-chain antibody fragment (scFv) and subsequently mediate phagocytosis. B Vector map of the control lentivirus. C Vector map of the CD36-Encoding Lentivirus. D Vector map of the ICAM-1/CD54-encoding lentivirus. E Vector map of the EPCR/CD201-encoding lentivirus

Adhesion assays with CAR-M cells

Adhesion assays were performed using the four CAR-M cell types co-cultured with iRBCs under varying conditions of parasite strain, developmental stage, and parasitemia. Each experimental condition was conducted in quadruplicate. The CAR-M cells were incubated with iRBCs for a specified period, after which non-adherent iRBCs were carefully washed away. The number of adherent iRBCs per CAR-M cell was quantified by microscopic examination or by flow cytometry, as detailed below.

A three-factor orthogonal experimental design was employed to systematically evaluate the optimal conditions for CAR-M cell adhesion to iRBCs:

Factor A: Strain (2 levels: A1 = Pf3D7, A2 = Pf C580Y).

Factor B: RBC stage (3 levels: B1 = Ring, B2 = Trophozoite, B3 = Schizont).

Factor C: Parasitemia (3 levels: C1 = 0.1%, C2 = 1%, C3 = 5%).

This design resulted in 18 unique experimental combinations (2 × 3 × 3). For each combination, adhesion was measured across the four CAR-M cell types (CAR-M/CTL_SCL, CAR-M/CD36_SCL, CAR-M/CD54_SCL, and CAR-M/CD201_SCL). To simplify the analysis, a composite response variable, "total adhesion strength" (Y), was defined as the sum of the adhesion values from all four CAR-M types for a given experimental condition:

$$\text{Y}=\text{Adhesion}\left(\text{CAR-M}/{\text{CTL}}_{\text{SCL}}+\text{CAR-M}/{\text{CD36}}_{\text{SCL}}+\text{CAR-}\text{M}/{\text{CD54}}_{\text{SCL}}+\text{CAR-M}/\text{CD2}01_{\text{SCL}}\right)$$

For CAR-M/CTL_SCL, CAR-M/CD36_SCL, CAR-M/CD54_SCL, and CAR-M/CD201_SCL, higher values indicated stronger adhesion.

Assessment of iRBCs elimination

To evaluate the clearance efficacy of CAR-M against iRBCs based on parasitemia, a co-culture mixture of CAR-M/CD201_SCL cells and iRBCs was filtered through a leukocyte filter (Diameter: 1.5 cm; pore size: 8 μm) to remove CAR-M. The filtered cell suspension was collected, washed with phosphate-buffered saline (PBS), and resuspended for the preparation of thin blood smears. After fixation with methanol and staining with Giemsa, the smears were examined under a light microscope. Parasitemia was determined by calculating the percentage of iRBCs among a total count of 500 to 1000 erythrocytes per slide, as assessed by two independent observers in a blinded manner. This approach enabled a precise and reproducible assessment of iRBC clearance efficiency.

Statistical analysis

Range analysis was performed on the total adhesion strength (Y) data to assess the relative influence of each experimental factor and identify optimal factor levels. For each level of a given factor, the average Y value across all experimental runs containing that level was calculated. The range (R) for each factor was determined as the difference between the maximum and minimum mean Y values for its levels. Factors were subsequently ranked by their R values, with larger ranges indicating a greater influence on the total adhesion strength. The optimal level for each factor was defined as the level yielding the highest mean Y value. All statistical analyses were conducted using Python (v3.9) with standard numerical libraries.

Results

Characterization of CAR-M Cells

The relative expression of the target genes (CD36, CD54, and CD201) was highest in the corresponding stable cell lines (SCL), with strong statistical evidence for differences compared to the blank control (THP-1) (P < 0.0001). In the antibiotic-selected groups (ASG), expression levels of these genes were lower but still exhibited strong statistical differences relative to the blank control (P < 0.0001). Further pairwise comparisons revealed no statistically significant difference between CAR-M/CD36_ASG and CAR-M/CD36_SCL (P = 0.0551); however, statistically significant differences were observed between CAR-M/CD54_ASG and CAR-M/CD54_SCL, as well as between CAR-M/CD201_ASG and CAR-M/CD201_SCL (P < 0.0001 for both).

Under 400× magnification using confocal microscopy, distinct fluorescent expression was observed across different groups: control (THP-1) cells exhibited both EGFP (green) and mCherry (red) fluorescence (Fig. 2); CAR-M/CD36_SCL and CAR-M/CD201_SCL displayed EGFP (green) fluorescence only (Fig. 2A, C); whereas CAR-M/CD54_SCL showed mCherry (red) fluorescence only (Fig. 2B). Cell nuclei were counterstained with DAPI, displaying blue fluorescence (Fig. 2).

Fig. 2.

Relative mRNA expression and representative confocal images of CAR-M cells across experimental stages. A Expression of CD36 in blank control (THP-1), negative control antibiotic-selected groups (CAR-M/CTL_ASG), negative control stable cell lines (CAR-M/CTL_SCL), CAR-M/CD36 antibiotic-selected group (CAR-M/CD36_ASG) and CAR-M/CD36 stable cell line (CAR-M/CD36_SCL); CAR-M/CTL_SCL and CAR-M/CD36_SCL showed detection of DAPI (blue) and EGFP (green) signals, and the merged image displays the overlay of all channels. Scale bars: 25 μm. B Expression of CD54 in blank control (THP-1), negative control antibiotic-selected groups (CAR-M/CTL_ASG), negative control stable cell lines (CAR-M/CTL_SCL), CAR-M/CD54 antibiotic-selected group (CAR-M/CD54_ASG) and CAR-M/CD54 stable cell line (CAR-M/CD54_SCL); CAR-M/CTL_SCL and CAR-M/CD54_SCL showed detection of DAPI (blue) and mCherry (red) signals, and the merged image displays the overlay of all channels. Scale bars: 25 μm. C expression of CD201 in blank control (THP-1), negative control antibiotic-selected groups (CAR-M/CTL_ASG), negative control stable cell lines (CAR-M/CTL_SCL), CAR-M/CD201 antibiotic-selected group (CAR-M/CD201_ASG) and CAR-M/CD201 stable cell line (CAR-M/CD201_SCL); CAR-M/CTL_SCL and CAR-M/CD201_SCL showed detection of DAPI (blue) and EGFP (green) signals, and the merged image displays the overlay of all channels. Scale bars: 25 μm. ****P < 0.0001

Total adhesion strength across experimental combinations

The total adhesion strength (sum of CAR-M/CTL_SCL, CAR-M/CD36_SCL, CAR-M/CD54_SCL, and CAR-M/CD201_SCL responses) varied considerably across the 18 experimental combinations, ranging from 5.75 to 16.01 (Table 1). The highest total adhesion strength (16.01) was observed for the PfC580Y strain at schizont stage with 1% parasitemia (combination A2-B3-C2), primarily driven by exceptionally high CD201-mediated adhesion (8.55). The lowest adhesion strength (5.75) occurred with PfC580Y at ring stage with 0.1% parasitemia (combination A2-B1-C1).

Table 1.

Total adhesion strength for all experimental combinations

No	Strain	Stage	Parasitemia (%)	CAR-M/CTLSCL	CAR-M/CD36SCL	CAR-M/CD54SCL	CAR-M/CD201SCL	Total
1	Pf3D7	Ring	0.1	0.50	1.67	1.33	2.56	6.06
2	Pf3D7	Ring	1	1.90	2.14	1.71	2.25	8.00
3	Pf3D7	Ring	5	2.61	1.54	1.56	2.07	7.78
4	Pf3D7	Trophozoite	0.1	1.35	1.46	2.24	3.11	8.16
5	Pf3D7	Trophozoite	1	2.09	2.66	2.37	5.25	12.37
6	Pf3D7	Trophozoite	5	3.04	1.84	2.32	4.22	11.42
7	Pf3D7	Schizont	0.1	2.91	1.40	1.27	3.12	8.70
8	Pf3D7	Schizont	1	4.02	1.47	1.49	4.75	11.73
9	Pf3D7	Schizont	5	3.02	1.67	1.43	7.31	13.43
10	Pf C580Y	Ring	0.1	1.43	1.52	1.49	1.31	5.75
11	Pf C580Y	Ring	1	1.63	1.67	0.70	1.97	5.97
12	Pf C580Y	Ring	5	2.67	2.56	1.30	1.72	8.25
13	Pf C580Y	Trophozoite	0.1	1.15	3.04	2.77	2.06	9.02
14	Pf C580Y	Trophozoite	1	1.71	3.46	3.77	2.75	11.69
15	Pf C580Y	Trophozoite	5	1.60	4.13	2.63	2.38	10.74
16	Pf C580Y	Schizont	0.1	1.35	2.52	1.31	2.98	8.16
17	Pf C580Y	Schizont	1	1.88	3.62	1.96	8.55	16.01
18	Pf C580Y	Schizont	5	1.31	4.11	2.05	3.87	11.34

Range analysis of factorial effects

A range analysis was conducted to evaluate the influence of each factor on total adhesion strength (Table 2). The mean adhesion values for each factor level were calculated, and the range (R) was determined as the difference between maximum and minimum mean values.

Table 2.

Range analysis of factorial effects

Factor	Level	Mean adhesion	Range (R)	Rank
A: Strain	A1 (Pf3D7)	9.74	0.08	3
	A2 (Pf C580Y)	9.66
B: RBC Stage	B1 (Ring)	6.97	4.59	1
	B2 (Trophozoite)	10.57
	B3 (Schizont)	11.56
C: Parasitemia	C1 (0.1%)	7.64	3.32	2
	C2 (1%)	10.96
	C3 (5%)	10.49

The range analysis revealed that Factor B (RBC developmental stage) exhibited the largest range (R = 4.59), indicating it was the most influential factor affecting CAR-M adhesion to iRBCs. Factor C (parasitemia) showed moderate influence (R = 3.32), whereas Factor A (parasite strain) had minimal effect (R = 0.08).

Based on mean adhesion values, the optimal factor-level combination was identified as A1-B3-C2 (Pf3D7 strain, schizont stage, 1% parasitemia), with schizont stage (mean = 11.56) and 1% parasitemia (mean = 10.96) being the optimal levels for factors B and C, respectively (Fig. 3B).

Fig. 3.

Determination of the optimal stage and concentration for CAR-M cell interaction with different Plasmodium falciparum strains by flow cytometry. For the CAR-M phagocytosis assay, Plasmodium falciparum strains (Pf3D7 and PfC580Y) were synchronized at the Ring, Trophozoite, and Schizont stages, respectively. The initial parasitemia for all samples was set at approximately 5%. Serial dilutions were then performed to achieve lower parasitemia levels of 1% and 0.1%. A, B Adhesion for the Negative Control (CAR-M/CTL_SCL) and CAR-M cells (CAR-M/CD36_SCL, CAR-M/CD54_SCL, and CAR-M/CD201_SCL) against Pf3D7 at varied stages (Ring, Trophozoite, and Schizont stages) and concentrations (5%, 1%, 0.1%). Scale bars: 15 μm. C, D Corresponding results for strains PfC580Y. Scale bars: 15 μm. E Adhesion ranged from lowest to highest under the following conditions: the lowest was seen for CAR-M/CTL_SCL against Pf3D7 (0.1% parasitemia, ring stage). The highest adhesion was observed for CAR-M/CD201_SCL against PfC580Y (1% parasitemia, schizonts), followed by that against Pf3D7 (5% parasitemia, schizonts)

Analysis of variance (ANOVA)

ANOVA was conducted to assess the statistical significance of each factor (Table 3). The overall mean adhesion was 9.70, and the total sum of squares was 233.67. ANOVA confirmed that RBC developmental stage (Factor B) was the only statistically significant factor (F = 3.04, P < 0.05), consistent with range analysis results. Parasitemia (Factor C) showed moderate significance (F = 1.76), while parasite strain (Factor A) had no significant effect (F = 0.00).

Table 3.

Analysis of variance for total adhesion strength

Source	Sum of squares (SS)	Degrees of freedom (df)	Mean square (MS)	F-value	Significance
Strain (A)	0.03	1	0.03	0.00	NS
RBC Stage (B)	65.74	2	32.87	3.04	*
Parasitemia (C)	38.14	2	19.07	1.76	†
Error	129.76	12	10.81	–	–
Total	233.67	17	–	–	–

NS: not significant; †: moderately significant; *: significant (P < 0.05)

Stage-specific and receptor-specific adhesion patterns

Examination of individual CAR-M responses revealed distinct patterns. The schizont stage consistently supported the highest adhesion across all CAR-M types, particularly for CAR-M/CD201_SCL (values ranging from 2.98 to 8.55). Strain-specific differences were minimal for overall adhesion but were notable for specific receptor-CAR-M combinations. For instance, PfC580Y showed higher CD36 and CD54 adhesion at trophozoite stage (Fig. 3D), while Pf3D7 showed higher CD201 adhesion at schizont stage (Fig. 3B). Parasitemia effects were non-linear: 1% parasitemia generally yielded optimal adhesion, while both lower (0.1%) and higher (5%) levels resulted in reduced efficiency. CAR-M/CD201_SCL exhibited the strongest overall adhesion, with particularly pronounced binding at schizont stage (mean CD201 adhesion at schizont = 5.04 vs. trophozoite = 3.42 vs. ring = 2.13). The lowest phagocytosis rate of iRBCs by CAR-M cells was observed in the 0.1% Pf3D7 + CAR-M/CTL_SCL group (mean = 0.50%). In contrast, the highest phagocytosis rate was found in the 1% PfC580Y + CAR-M/CD201_SCL group (mean = 8.55%), followed by the 5% Pf3D7 + CAR-M/CD201_SCL group (mean = 7.31%) (Fig. 3E).

Phagocytosis of iRBCs by CAR-M/CD201_SCL was evaluated under optimized conditions

Under the optimized experimental conditions (using iRBCs with approximately 1% parasitemia for CAR-M/CD201_SCL treatment), the mean phagocytosis rates of CAR-M/CD201_SCL for Pf3D7 and PfC580Y iRBCs were 5.28% and 6.76% (Fig. 4A), respectively, with a statistically significant difference between the two groups (P = 0.003). Prior to the assay, the parasitemia levels of Pf3D7 and PfC580Y were approximately 1.1% and 1.17%, respectively. After co-culture with cells and passage through a leukocyte filter, the parasitemia of Pf3D7 decreased to approximately 0.03%, whereas no iRBCs were observed microscopically in the PfC580Y group (Fig. 4B, C).

Fig. 4.

Phagocytic Efficiency of CAR-M Cells Against Different Parasite Strains Under Optimal Culture Conditions. A Phagocytosis percentages of CAR-M/CD201_SCL cells towards Pf3D7 and PfC580Y strains at 1% parasitemia. B Parasitemia in co-cultures with CAR-M/CD201_SCL after leukocyte filtration across parasite strains. C Representative Giemsa-stained blood smears from each experimental group (under identical conditions). Scale bars: 10 μm. **P < 0.01

Discussion

The development of CAR-M technology for malaria immunotherapy holds significant promise, and understanding the optimal conditions for CAR-M-iRBC interaction is paramount. Our principal findings demonstrate that: (1) parasite developmental stage is the dominant factor influencing CAR-M adhesion, with schizont-stage parasites exhibiting 65.8% higher adhesion compared to ring-stage parasites; (2) 1% parasitemia represents the optimal concentration for maximum adhesion efficiency; and (3) CAR-M/CD201_SCL cells show superior binding capacity, particularly to late-stage parasites, suggesting CD201 (EPCR) as a promising target for CAR-M-based therapeutic interventions.

Range analysis unequivocally identified the iRBC developmental stage as the most influential factor impacting CAR-M adhesion. Schizont-stage iRBCs consistently exhibited the highest mean total adhesion, followed by trophozoites, whereas ring-stage iRBCs showed the lowest adhesion. This finding aligns well with the established biology of PfEMP1 expression and cytoadherence. PfEMP1 is trafficked to the iRBC surface primarily during the trophozoite and schizont stages, where it mediates binding to host receptors. In contrast, ring-stage parasites typically express low or undetectable levels of PfEMP1 on the iRBC surface, explaining their reduced adhesiveness. These results underscore the importance of targeting late-stage iRBCs for effective CAR-M-mediated removal, as these stages are responsible for severe disease pathology and represent the most adhesive forms.

The influence of parasitemia on total adhesion strength was also significant, with an optimal mean adhesion observed at 1%. Both lower (0.1%) and higher (5%) parasitemia levels resulted in reduced mean adhesion. At 0.1% parasitemia, the reduced adhesion is likely attributable to a lower frequency of encounters between CAR-M cells and iRBCs. Conversely, at 5% parasitemia, potential factors such as steric hindrance, saturation of CAR-M binding sites, increased competition among iRBCs for adhesion, or even parasite-induced modulation of host cell properties could lead to a decline in per-cell adhesion efficiency. This non-monotonic relationship suggests an optimal parasite load for efficient CAR-M engagement, a finding that is crucial for designing in vitro and potentially in vivo efficacy assays.

While the parasite strain exhibited the smallest main effect on overall adhesion, a crucial observation emerged from the highest single adhesion value. The A2 (PfC580Y) strain, at the schizont stage and 1% parasitemia, yielded the maximum total adhesion strength (Y = 16.01). This was primarily attributed to an exceptionally high contribution from CD201-mediated adhesion (8.55). This finding is particularly relevant given that PfEMP1 variants binding to EPCR (CD201) are strongly associated with severe malaria phenotypes, including cerebral malaria [23]. The PfC580Y mutation indicates a specific genetic background that might lead to altered PfEMP1 expression or binding affinities. This suggests that while overall strain differences might be subtle when averaged across all conditions, specific strain-receptor interactions can lead to profoundly enhanced adhesion under specific conditions. This underscores the need for CAR-M designs that can adapt to the diversity of PfEMP1 expression profiles across different P. falciparum strains and clinical presentations. For instance, a CAR-M targeting CD201-binding PfEMP1 may be highly effective against strains causing severe malaria.

CAR-M/CD201_SCL demonstrated the highest adhesion capacity, particularly at schizont stage (mean = 5.04), which is notable given that EPCR-binding PfEMP1 variants have been associated with severe malaria pathogenesis [24]. EPCR-binding PfEMP1 proteins, characterized by specific domain cassettes (DC8 and DC13), mediate brain endothelial sequestration in cerebral malaria [25, 26]. Our results suggest that CAR-M/CD201_SCL cells could specifically target this virulent parasite subpopulation, offering a precision medicine approach to severe malaria.

The exceptional performance of CAR-M/CD201_SCL, especially with the PfC580Y strain at 1% parasitemia (adhesion value = 8.55), may reflect: (1) higher affinity binding between CD201 and specific PfEMP1 variants expressed by artemisinin-resistant parasites (PfC580Y carries the kelch13 C580Y mutation) [27], or (2) altered var gene expression profiles in drug-resistant strains that favor EPCR-binding variants [28, 29]. This finding warrants further investigation into whether artemisinin resistance selection pressure drives preferential expression of EPCR-binding PfEMP1.

Although CD201 exhibited superior individual binding performance, the combined adhesion strength across all four CAR-M types (CAR-M/CDCTL_SCL, CAR-M/CD36_SCL, CAR-M/CD54_SCL, and CAR-M/CD201_SCL) consistently exceeded that of any single receptor, supporting the concept of a multivalent CAR-M approach. In the natural setting, P. falciparum sequestration involves simultaneous binding of PfEMP1 to multiple endothelial receptors, a redundant mechanism that ensures stable cytoadherence under varying host conditions. By engineering a cocktail of CAR-M cells targeting distinct receptors, we may effectively mimic this physiological redundancy. Such a multivalent approach offers several potential advantages: (1) Prevent immune evasion driven by var gene switching; (2) Capture diverse parasite populations expressing different PfEMP1 variants, thereby broadening therapeutic coverage; (3) Enhance overall binding avidity through cooperative receptor engagement, potentially leading to more efficient clearance of infected red blood cells. This study has several methodological limitations. First, the experimental conditions did not fully capture the physiological complexity of malaria infection: the static adhesion assays used do not recapitulate the dynamic shear stress conditions of the microvasculature, adhesion was measured without assessing downstream phagocytosis or parasite killing. Second, in vitro culture conditions could not fully mimic physiological parameters such as oxygen tension or immune cell crosstalk, and the use of THP-1-derived macrophages as the CAR-M platform may not reflect primary macrophage biology. Third, regarding CAR design, only single-receptor constructs were tested, leaving more sophisticated configurations unexplored. While the orthogonal experimental design assumed additive effects and did not account for potential interactions between factors. Furthermore, an additional limitation is that P. falciparum isolates with predefined binding specificity for individual receptors were not employed in these studies. This decision was made to retain the naturally heterogeneous adhesion profile of clinical isolates, thereby reflecting the diversity of receptor-binding phenotypes present in circulating parasite populations. However, this approach introduces interpretive constraints. Because non-selected isolates represent a mixed population capable of binding to various receptors, the observed binding patterns reflect the collective adhesive properties of a heterogeneous ensemble rather than a phenotype attributable to a single receptor-ligand interaction. Consequently, this limits our ability to dissect the precise molecular mechanisms underlying specific receptor-mediated adhesion and may obscure the contributions of individual receptor pathways. Future studies using isolates with predefined receptor-binding specificities will be essential to delineate the mechanistic basis of CAR-M recognition and to optimize receptor-targeting strategies.

Despite limitations, CAR-M therapy holds significant promise for malaria treatment, with key future directions including validating CAR-M-mediated phagocytosis and intracellular parasite killing using advanced imaging or screening platforms, evaluating in vivo efficacy and safety in humanized malaria models, and exploring synergistic combination strategies with artemisinins to enhance parasite clearance. Engineering multi-receptor CAR-M cells (e.g., CD201-CD36 dual-receptors) may overcome immune evasion via PfEMP1 variant switching, while structural and kinetic studies of receptor-PfEMP1 interactions could elucidate the superior efficacy of CD201 targeting. Prospective clinical biomarker studies correlating var gene expression, PfEMP1 variants, and disease severity with therapeutic response will help identify optimal patient populations. Crucially, comprehensive off-target safety assessments are essential to rule out hemolysis or endothelial damage due to unintended CAR-M engagement with uninfected RBCs or endothelial cells expressing CD36, CD54, or CD201.

Our findings suggest several key considerations for the clinical translation of CAR-M therapy in malaria. First, for better therapeutic efficacy, CAR-M therapy should ideally be administered during the schizont-to-merozoite phase, when PfEMP1 expression reaches its peaks, or continuous administration may be considered to capture parasites across developmental stages. Second, a rational integration strategy with conventional antimalarial drugs could involve an initial rapid reduction of parasitemia using artemisinin derivatives, followed by CAR-M therapy to target sequestered parasites, and slow-acting partner drugs to suppress post-treatment recrudescence. Third, optimal dosing should maintain CAR-M:iRBCs ratios that correspond to approximately 1% parasitemia based on our adhesion data, with pharmacokinetic studies needed to determine CAR-M expansion rates and persistence. Fourth, real-time parasitemia monitoring could enable adaptive dosing strategies, increasing CAR-M doses above 2% parasitemia and reducing them below 0.1% to maintain the optimal therapeutic window. Finally, given the exceptional performance of CAR-M/CD201_SCL against EPCR-binding parasites, severe malaria patients who typically exhibit high levels of EPCR-binding PfEMP1 sequestration, may derive particularly benefit from CD201-targeted CAR-M therapy.

Conclusions

This study provides a comprehensive analysis of factors influencing P. falciparum iRBC cytoadhesion to host receptors CD36, CD54, and CD201. It demonstrates that parasite developmental stage critically determines CAR-M adhesion to iRBCs, with schizont-stage parasites at 1% parasitemia representing optimal targets. CD201-CAR-M exhibited superior binding, particularly to late-stage parasites, suggesting EPCR-targeting CAR-M as a promising therapeutic modality for severe malaria. By strategically exploiting PfEMP1-mediated adhesion in conjunction with the natural phagocytic function of macrophages, CAR-M cell therapy represents a groundbreaking, cell-based immunotherapeutic strategy. This innovative approach is fundamentally less susceptible to drug resistance, thereby providing much-needed alternatives and complements to conventional chemotherapy in the global fight against drug-resistant malaria.

Author contributions

The roles for J.L. included conceptualization, formal analysis, funding acquisition, investigation, methodology, project administration, resource supervision, validation, writing original draft, and writing review and editing. The roles for Q.F.Z included resource supervision, validation, writing original draft, and writing review and editing. The roles for D.Q.Z. and W.J.C. included data curation, methodology, validation, and writing the original draft. The roles for Y.Z., H.Y.Z., Z.X.L, and H.M.S included formal analysis, methodology, and validation. All the authors read and approved the final manuscript.

Funding

This study was supported by the Principle Investigator Program of Hubei University of Medicine (HBMUPI202101), National Natural Science Foundation of China (Grant Number 81802046), Joint supported by Hubei Provincial Natural Science Foundation and Shiyan-of China (JCZRLH202600113) and Hubei University of Medicine 2024 Postgraduate Science and Technology Innovation Project (YC2024001).

Data availability

All the data generated or analyzed during this study are included in this article. The datasets analyzed in this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.