Abstract

   Environmental factors restrict malaria parasite development, but the
   influence of host metabolic variations on the infectivity of the blood
   stage parasite is not fully understood. Here we show that mice on a
   ketogenic diet are completely protected from infection with the malaria
   parasite Plasmodium berghei. We further show that administration of the
   ketone body β-hydroxybutyrate (βOHB), but not of acetoacetate,
   increases survival of infected mice and inhibits proliferation of both
   P. berghei and Plasmodium falciparum in vitro. Administration of either
   a ketogenic diet or βOHB induces metabolic reprogramming in parasites,
   including reduced levels of nicotinamide adenine dinucleotide, which is
   associated with the downregulation of genes controlling parasite
   development, erythrocyte invasion and pathogenicity. Our data indicate
   that a ketogenic diet and the ketone body βOHB confer resistance to
   malaria in mice by causing developmental arrest of Plasmodium
   parasites, highlighting the potential of dietary and metabolic
   strategies to fight malarial infection.

   Subject terms: Parasite host response, Antiparasitic agents, Metabolism
     __________________________________________________________________

   A ketogenic diet and the ketone body β-hydroxybutyrate are shown to
   confer resistance to malaria in mice by inducing metabolic
   reprogramming in Plasmodium parasites, suggesting a dietary strategy
   for malaria prevention.

Main

   Plasmodium species, the causative agents of malaria, express arrays of
   virulence factors^[56]1 during their interplay with human hosts and
   remain a severe threat to millions of people in tropical and
   subtropical regions^[57]2. The spread of resistance to
   artemisinin-based therapeutics by Plasmodium falciparum parasites
   necessitates new strategies to control the disease^[58]3. The
   development and dissemination of Plasmodium parasites are intimately
   connected to their hosts, and nutrient supply is a critical regulatory
   factor. Previous studies have indicated that malaria parasites can
   actively regulate their multiplication through sensing host dietary
   caloric alterations^[59]4. Furthermore, changes in other host
   environmental components, such as dietary restriction^[60]5 and
   glycolysis suppression^[61]6, all protect against experimental cerebral
   malaria. Strikingly, supplementation with high-fat diets can induce
   reactive oxygen species production in hepatic cells and has resulted in
   the termination of sporozoite development in a rodent model^[62]7.
   These findings reinforce the possibility of targeting parasite
   nutrient-sensing mechanisms as a strategy for malaria intervention and
   prevention.

   In mammals, ketone metabolism contributes a significant energy supply
   in extrahepatic tissues when glucose is not readily available, which
   has led to widespread interest in its association with body weight
   gain, brain health, metabolic diseases and cancer
   prevention^[63]8–[64]11. A ketone body refers to one of three distinct
   small molecules, acetoacetic acid (AcAc), β-hydroxybutyrate (βOHB) and
   acetone, which are synthesized primarily in the liver from fats^[65]12.
   When glucose is scarce, such as during fasting or prolonged exercise,
   fatty acids are transported to the liver for conversion to ketone
   bodies, which are distributed by blood circulation to metabolically
   active tissues such as muscle or brain for the generation of ATP. In
   mammals, AcAc and acetone are unstable and poorly abundant, and most
   ketone bodies are in the form of βOHB^[66]13. βOHB concentrations in
   humans are normally in micromolar ranges but can increase to 1–2 mM
   after intense exercise or 2 days of fasting^[67]14 and up to 6–8 mM
   during prolonged starvation^[68]15. Malaria parasites have a
   nutrient-sensing mechanism for glucose^[69]16. βOHB serves not only as
   an energy source but also as a signalling molecule and has been shown
   to influence pathways involved in metabolic regulation, inflammation
   and oxidative stress^[70]17,[71]18. Additionally, βOHB can inhibit
   histone deacetylase (HDAC) activation, leading to increased histone
   acetylation and subsequent disorders in gene expression and protection
   against oxidative stress in mouse tissues^[72]19. In P. falciparum,
   HDAC is a critical protein that regulates genes associated with
   parasite development, and its inhibitor can effectively kill parasites
   in vitro and in animal models^[73]20, highlighting its potential as a
   novel antimalarial target. We therefore reason that a high blood
   abundance of βOHB can impair Plasmodium parasites by disturbing the
   functional activity of their bioregulatory machineries.

   In this study, we demonstrated that the administration of a ketogenic
   diet (KD) to mice resulted in elevated concentrations of blood βOHB and
   provided complete protection against Plasmodium berghei ANKA infection.
   Importantly, βOHB significantly inhibited erythrocyte invasion and
   development of the human malaria parasite P. falciparum. We further
   revealed that βOHB dampened the intracellular levels of nicotinamide
   adenine dinucleotide (NAD) in both P. berghei and P. falciparum,
   leading to the downregulation of genes associated with intracellular
   development and the repression of virulence-associated genes in the
   parasites.

Results

   Plasmodium parasites mainly consume glucose as an energy supply^[74]16.
   We therefore tested the susceptibility of diabetic BALB/c mice,
   compared with normal BALB/c mice, to P. berghei ANKA infection. Results
   show that diabetic mice survived significantly longer than the control
   mice (Supplementary Fig. [75]1a). Diabetes is the leading cause of
   elevated serum βOHB (a type of ketone body^[76]21), and the serum
   concentration of βOHB was significantly elevated in diabetic mice
   compared to normal mice (Supplementary Fig. [77]1b), which suggests
   that elevated βOHB may restrict Plasmodium parasite infection. We next
   conducted an experiment using female C57BL/6 mice (N phenotype) fed
   with a KD (which can increase serum βOHB^[78]14), at concentrations of
   0.5%, 20%, 50%, 70% and 90% compared to control mice fed with a regular
   diet (RD). The macronutrient composition of the diets is presented in
   Fig. [79]1a and Supplementary Fig. [80]1c,d. The mice had ad libitum
   access to the diets for 14 days before parasite infection. Notably,
   mice fed with the 20% to 90% KDs showed complete resistance to
   P. berghei infection (Fig. [81]1b,c and Extended Data Fig. [82]1a).
   Extremely low parasitaemia was detected in the peripheral blood of the
   mice fed with 20% to 90% KD within 30 days post infection, after which
   no infected red blood cells (iRBCs) were detected (Supplementary Fig.
   [83]2a–d). These mice survived for over 80 days (Fig. [84]1c). By
   contrast, all mice in the RD-fed group exhibited much higher
   parasiteamia (Fig. [85]1b and Extended Data Fig. [86]1a) and all died
   within 10 days of infection (Fig. [87]1c). In the 0.5% KD group, we
   observed a significant suppression of P. berghei ANKA proliferation,
   evidenced by much lower parasitaemia (Extended Data Fig. [88]1b,c) and
   mortality rates (Extended Data Fig. [89]1d) compared to the RD control
   group. To further strengthen our findings, we repeated the experiment
   with more experimental animals (Supplementary Fig. [90]2a–e). The serum
   concentrations of βOHB in the mice fed with 0.5% to 90% KDs were
   significantly higher than that of the RD group (Fig. [91]1d and
   Supplementary Fig. [92]2f). To rule out the possibility that the
   elevated ketonic bodies were a result of starvation, we recorded and
   analysed the daily food consumption and body weight of the mice. The
   result showed that feeding with different KDs did not significantly
   alter food intake or body weight compared to RD-fed mice over the
   course of 1 month (Fig. [93]1e,f and Supplementary Fig. [94]2g).
   Furthermore, the levels of triglycerides, high-density lipoprotein,
   low-density lipoprotein and insulin in the blood of all KD-fed mice
   showed no significant changes compared to that of the control mice
   after P. berghei ANKA infection (Extended Data Fig. [95]1e–h). These
   results indicate that the elevated ketone levels in mice are not caused
   by starvation, high-fat diets or insulin resistance. We also found that
   a KD consistently protected both male C57BL/6 and BALB/c mice from
   P. berghei ANKA infection (Fig. [96]1g–k and Extended Data Figs.
   [97]1i,j and [98]2a–d), indicating that the antimalarial effect of a KD
   is host-independent. We also analysed whether the KD diet had any
   adverse effects on the mice. Results indicated that the KD diet did not
   cause any significant changes in RBC count, mean corpuscular
   haemoglobin, haemoglobin, packed cell volume, liver index or liver and
   kidney function compared to that of the control mice (Extended Data
   Fig. [99]2e–n).

Fig. 1. KD conferred complete protection against P. berghei ANKA infection,
and βOHB increased survival of infected mice and inhibited proliferation of
P. falciparum 3D7 in vitro.

   [100]Fig. 1
   [101]Open in a new tab

   a, Macronutrient composition of diets. b, Diets with various
   concentrations of KD inhibited the proliferation of P. berghei in
   female C57BL/6 mice (n = 6). c, Survival rates for female C57BL/6 mice.
   Three mice fed a RD with signs of experimental cerebral malaria were
   killed (n = 10). d, Serum concentrations of βOHB in different groups of
   healthy mice after 14 days of feeding with various concentrations of KD
   treatment or RD control (n = 6). e, Calories per day for mice in
   different KD groups (n = 10). f, Mice body weight curves (n = 6). g,
   Diets with 90% KD inhibited the proliferation of P. berghei in BALB/c
   mice (n = 10). h, Survival rates for the BALB/c mice (n = 10). i, Diets
   with 90% KD inhibited the proliferation of P. berghei in male C57BL/6
   mice (n = 6). j, Survival rates for male C57BL/6 mice (n = 10). k,
   Images of the parasite burden in male C57BL/6 mice on day 7 (n = 3). l,
   Osmotic pumps filled with βOHB increased ketone levels in mice (n = 6).
   m, Osmotic pumps filled with βOHB inhibited the proliferation of
   P. berghei ANKA in female C57BL/6 mice (n = 5). Exogenous βOHB was
   administered by implanted osmotic pumps for 7 days; pumps filled with
   isotonic saline solution served as the negative control. n, Treatment
   with 1 mM βOHB significantly extended the survival of P. berghei
   ANKA-infected mice. The pump was empty at day 7 (n = 10). o,
   P. falciparum parasitaemia after exposure to various concentrations of
   βOHB (dissolved in MCM) in vitro for 96 h (n = 3). p, Inhibition of the
   in vitro proliferation of P. falciparum 3D7 by βOHB. The IC[50] value
   of βOHB was 7.3 mM (R^2 = 0.9765; confidence interval of IC[50] value,
   6.8–7.7 mM) (n = 3). Data are presented as means; error bars, s.e.m.
   (b, d, g, i, k–m, o and p). P values are from unpaired two-tailed
   Student’s t-tests (d, k and l) or log-rank (Mantel–Cox) test (c, h, j
   and n).

   [102]Source data

Extended Data Fig. 1. Impact of ketogenic diet and β-hydroxybutyrate (βOHB)
on metabolic and survival parameters in P. berghei-infected mice.

   [103]Extended Data Fig. 1
   [104]Open in a new tab

   a, The area under each curve was calculated in Fig. [105]1e. We
   randomly selected 6 mice for the quantification of parasiteamia (n = 6,
   Unpaired two-tailed Student’s t-test). b, Diets with 0.5% KD inhibited
   the proliferation of P. berghei in female C57BL/6 mice (n = 6). c, The
   area under curve was calculated in Extended Data Fig. 1b (n = 6,
   Unpaired two-tailed Student’s t-test). d, Survival rates of the female
   C57BL/6 mice in the experimental groups (n = 10). Three mice fed with a
   regular diet (RD) exhibiting signs of experimental cerebral malaria
   (ECM) were euthanized. e-h, Triglyceride (TG), high-density lipoprotein
   (HDL), low-density lipoprotein (LDL) and insulin levels in healthy
   controls and mice fed with different KD concentrations post-infection.
   i, The area under each curve was calculated in Fig. [106]1g (n = 6,
   Unpaired two-tailed Student’s t-test). j, The area under each curve was
   calculated in Fig. [107]1i (n = 6, Unpaired two-tailed Student’s
   t-test). Data are presented as means ± SEMs (a-c, i and j).

   [108]Source data

Extended Data Fig. 2. Repeated experiments show ketogenic diets significantly
inhibit the proliferation of parasites in infection-mice and impact of
ketogenic diet and β-hydroxybutyrate (βOHB) on hematological in healthy mice.

   [109]Extended Data Fig. 2
   [110]Open in a new tab

   a, Diets with 90% KD inhibited the proliferation of P. berghei in male
   C57BL/6 mice (n = 20). All the mice fed a RD died at approximately 9
   days p.i. b, Survival rates for male C57BL/6 mice in the experimental
   groups (n = 20). 2 mice with signs of experimental cerebral malaria
   (ECM) were euthanized. c. Diets with 90% KD inhibited the proliferation
   of P. berghei in female BALB/c mice (n = 20). All the mice fed with a
   RD died approximately 13 days p.i. d, Survival rates for female BALB/c
   mice in the experimental groups (n = 20). e-h, Red blood cell counts,
   mean corpuscular hemoglobin (MCH), hemoglobin and packed cell volume
   (PCV) in peripheral blood of mice from different treatment groups
   (n = 6). i, Liver index of mice from different treatment groups
   (n = 6). j-n, Activity of γ-glutamyl transpeptidase (γ-GT), alanine
   aminotransferase (ALT), activity of aspartate transaminase (AST), urea
   and creatinine (CREA) in the serum of mice from different treatment
   groups (n = 6). o, The area under curve of parasitemia before 6 days
   post-infection was calculated in Fig. [111]1m (n = 5, Unpaired
   two-tailed Student’s t-test). Data are presented as means ± SEMs (a, c,
   and e-o). P values are from Dunnett’s multiple comparisons test by
   either adjusting the significance threshold or correcting P values
   (e-n).

   [112]Source data

   To investigate the role of βOHB in the infected mice, we used implanted
   osmotic pumps to administer exogenous βOHB, with isotonic saline
   solution as a negative control. In the βOHB-treated group, the serum
   levels of βOHB increased to approximately 0.5 mM and 1 mM (Fig.
   [113]1l), respectively. We found that βOHB inhibited the proliferation
   of P. berghei ANKA and completely protected the mice against
   experimental cerebral malaria (Fig. [114]1m,n and Extended Data Fig.
   [115]2o). Furthermore, βOHB also directly inhibited the growth of the
   human malaria parasite P. falciparum 3D7 in vitro in a dose-dependent
   manner, with a half-maximal inhibitory concentration (IC[50]) value of
   7.3 mM (Fig. [116]1o,p). βOHB did not cause any haemolysis of human
   RBCs (Supplementary Fig. [117]3). Additionally, the quantity of
   acetoacetate, another ketone body, did not show significant changes in
   KD-fed mice (Supplementary Fig. [118]4a). Compared to the positive
   control dihydroartemisinin (Supplementary Fig. [119]4b,c), acetoacetate
   did not exhibit any inhibitory effects on the proliferation of
   P. falciparum 3D7 parasites (Supplementary Fig. [120]5). These findings
   suggest that KDs, particularly βOHB, confer protection against
   Plasmodium infection in mice and significantly inhibit the
   proliferation of P. falciparum, the deadliest malaria parasite in
   humans.

   βOHB, an energetic metabolite, is synthesized primarily in the liver
   from fatty acids during fasting, prolonged exercise and in low-glucose
   conditions^[121]12. Therefore, we next investigated how βOHB inhibits
   the growth of both P. berghei ANKA and P. falciparum 3D7 parasites from
   the perspective of energy metabolism (Supplementary Tables [122]1 and
   [123]2). We found that feeding with a KD and βOHB treatment reshaped
   the energy metabolism processes of both P. berghei ANKA and
   P. falciparum 3D7 parasites (Fig. [124]2). First, in P. berghei ANKA
   iRBCs, the levels of l-leucine, phosphorylethanolamine, tyrosine, AMP,
   dAMP, dihydroxyacetone-phosphate, glyceraldehyde-3-phosphate, guanosine
   and l-aspartate were significantly elevated compared to normal RBCs
   from RD-fed mice, whereas the level of l-citrulline was markedly
   reduced (Fig. [125]2a, Extended Data Fig. [126]3a and Supplementary
   Table [127]3). Conversely, in P. berghei ANKA iRBCs from KD-fed mice
   compared to iRBCs from RD-fed mice, only the abundance of l-citrulline
   significantly increased (Fig. [128]2b and Extended Data Fig. [129]3a).
   In addition, the amounts of cis-aconitic acid, argininosuccinic acid,
   uracil, UMP, tyrosine, threonine, NAD, malic acid and lysine were
   significantly reduced in the iRBCs of P. berghei ANKA-infected mice fed
   a KD (Fig. [130]2b). Only the content of NAD was significantly reduced
   in the P. berghei ANKA iRBCs of KD-fed mice compared with the RBCs of
   KD-fed healthy mice (Fig. [131]2c). Strikingly, among all the altered
   metabolites identified in P. berghei ANKA-infected mice fed with a KD,
   only l-citrulline and NAD were negatively correlated with the other
   metabolites (Fig. [132]2d, blue line), highlighting the importance of
   these two substances in the metabolism of the parasite.

Fig. 2. KD and βOHB reprogram the energy metabolic processes of P. berghei
ANKA and P. falciparum 3D7 parasites.

   [133]Fig. 2
   [134]Open in a new tab

   a, Radar diagram showing that only the l-citrulline content decreased
   among the top ten metabolites in the P. berghei ANKA iRBCs compared to
   the RBCs of RD-fed mice. b, The l-citrulline content was significantly
   elevated in the P. berghei ANKA iRBCs of KD-fed mice compared to
   P. berghei ANKA iRBCs of RD-fed mice. The values on the grid lines
   represent fold changes (FCs); positive values indicate an increase, and
   negative values indicate a decrease. c, Only the NAD content was
   significantly reduced in the P. berghei ANKA iRBCs of KD-fed mice
   compared to the RBCs of KD-fed healthy mice. The values on the grid
   lines represent FCs; positive values indicate an increase, and negative
   values indicate a decrease. d, Radar diagram showing the negative
   correlations of l-citrulline and NAD (blue line) with the remaining
   metabolites. The sizes of the points in the radar diagram correspond to
   the log[2](FC) values. The colours of the lines represent the
   relationships between metabolites; blue represents a negative
   correlation, and red represents a positive correlation. Metabolites
   with variable importance in projection (VIP) > 1, log[2](FC) ≥ 2 and
   log[2](FC) ≤ 0.5 are shown. e–g, Bubble charts showing a decreased
   abundance of NAD in iRBCs treated with βOHB, which corroborated the
   results obtained from KD-fed mice. The VIP values for metabolites in
   iRBCs from ring-stage (e), trophozoite-stage (f) and schizont-stage
   P. falciparum 3D7 parasites (g) after exposure to βOHB are shown. Red
   bubbles represent increased metabolites, and green bubbles represent
   decreased metabolites. h, Venn diagram of differential metabolites in
   P. berghei-infected mice following KD treatment and βOHB-treated
   P. falciparum trophozoite stage. Blue text indicates downregulation,
   and red text indicates upregulation. i, Complete rescue of
   P. falciparum parasitaemia by NAD supplementation following βOHB
   treatment (n = 3). j, Complete rescue of P. falciparum development by
   NAD supplementation during βOHB treatment (n = 3). Data are presented
   as means; error bars, s.e.m. (i and j). P values are from Dunnett’s
   multiple comparisons test, either adjusting the significance threshold
   or correcting P values (i and j).

   [135]Source data

Extended Data Fig. 3. The KEGG enrichment results revealed that the oxidative
phosphorylation pathway in βOHB-treated P. falciparum 3D7 parasites at the
trophozoite stage was substantially downregulated.

   [136]Extended Data Fig. 3
   [137]Open in a new tab

   a, Heatmap of differential metabolites in RBCs and P. berghei ANKA
   iRBCs under different treatment conditions. In healthy mice, feeding
   with a ketogenic diet (KD) resulted in a significant down-regulation
   only in the level of L-citrulline (CON vs KD). In iRBCs, the
   L-citrulline content was significantly reduced, but it increased upon
   KD feeding. b, Only glycolysis/gluconeogenesis pathway was
   significantly inhibited in uninfected human RBCs after βOHB treatment.
   Bubbles on the left side of zero indicate downregulated pathways, and
   bubbles on the right side of zero indicate upregulated pathways. c,
   KEGG enrichment results for trophozoite-stage P. falciparum 3D7
   parasites. Bubbles on the left side of zero indicate downregulated
   pathways, and bubbles on the right side of zero indicate upregulated
   pathways. P values are from unpaired two-tailed t-tests with unequal
   variances (Welch’s t-test) adjusted using the Benjamini-Hochberg. For
   enrichment analysis, hypergeometric tests were conducted followed by
   Bonferroni correction (b and c).

   [138]Source data

   The metabolites of both RBCs and P. falciparum iRBCs with parasites at
   different developmental stages in response to exposure to βOHB were
   also investigated (Supplementary Table [139]4). The levels of 12
   significantly decreased metabolites, particularly
   fructose-1,6-bisphosphate, 2,3-diphosphoglyceric acid,
   dihydroxyacetone-phosphate and 2-phospho-d-glyceric acid, which are
   usually enriched in the glycolysis–gluconeogenesis pathway, were
   markedly reduced in βOHB-treated human RBCs compared to untreated RBCs
   (Extended Data Fig. [140]3b and Supplementary Tables [141]5 and
   [142]6). In P. falciparum iRBCs treated with βOHB, the content of
   l-citrulline did not significantly increase as observed in P. berghei
   ANKA iRBCs (Fig. [143]2e–g). However, decreased abundance of NAD in the
   P. falciparum iRBCs was observed (Fig. [144]2f), which corroborated the
   results obtained from the iRBCs of the KD-fed mice (Fig. [145]2b).
   Furthermore, Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment
   analysis revealed that metabolites that participate in the oxidative
   phosphorylation pathway in βOHB-treated P. falciparum 3D7 parasites at
   the trophozoite stage were predominantly decreased (Extended Data Fig.
   [146]3c).

   Among the metabolites between βOHB-treated P. falciparum 3D7 and
   P. berghei ANKA from the KD-fed mice, NAD was the only metabolite
   showing significant reduction in the two parasite lines (Fig. [147]2h).
   This finding suggests that the common mechanism by which βOHB inhibits
   the proliferation of Plasmodium parasites may involve the
   downregulation of NAD levels within iRBCs. To test this hypothesis, we
   performed NAD rescue experiments by the addition of NAD^+ in
   βOHB-treated P. falciparum 3D7. The results showed a progressive
   reduction of the inhibitory effect of βOHB on parasite proliferation
   with increased concentrations of NAD^+ (Fig. [148]2i,j). At 5 μM NAD^+,
   the inhibitory effect of βOHB on parasite growth was completely
   reversed (Fig. [149]2i,j). These findings indicate that βOHB inhibits
   parasite proliferation by downregulating NAD levels in the iRBCs.

   We next investigated the mechanisms underlying βOHB-mediated parasite
   proliferation inhibition from the perspective of gene expression
   regulation. A genome-wide transcriptomic analysis was first performed
   with P. berghei ANKA iRBCs from KD-fed and RD-fed mice. In general, 225
   genes were upregulated and 184 genes were downregulated (Extended Data
   Fig. [150]4, Supplementary Fig. [151]6a,b and Supplementary Table
   [152]7). We observed that the transcription of 21 genes encoding
   Plasmodium interspersed repeat (PIR) proteins was upregulated and that
   of 15 genes was downregulated in parasites from KD-fed mice (Extended
   Data Fig. [153]4a, Supplementary Fig. [154]6c,d and Supplementary Table
   [155]7). The PIR protein family is shared by parasites across the
   Plasmodium genus and has been regarded as variant antigens that assist
   in immune evasion, with some members also expressed in the sexual and
   hepatic stages^[156]22. KD-induced disturbance of gene transcription
   may be the cause, a hypothesis corroborated by the observation of
   upregulated transcription of the gene PBANKA_0403200 (which encodes
   circumsporozoite protein (CSP)) (Extended Data Fig. [157]4a and
   Supplementary Table [158]7), which is mainly activated at the
   post-sexual stage in the mosquito vector and in early exoerythrocytic
   liver stage in animals^[159]23. In addition, the transcription levels
   of some genes related to the cytoskeleton were increased (Extended Data
   Fig. [160]4b and Supplementary Table [161]8), and many
   tryptophan–aspartic acid (WD) repeat-containing proteins presented
   relatively high expression levels in parasites of KD-fed mice (Extended
   Data Fig. [162]4c and Supplementary Table [163]8). Furthermore, the
   expression of two genes, PBANKA_0112200 and PBANKA_0501800 (which
   encode myosin E and myosin ELC), and many genes encoding inner membrane
   complex (IMC) proteins were downregulated in the parasites of KD-fed
   mice (Extended Data Fig. [164]4d,e and Supplementary Table [165]8).
   IMCs of P. falciparum link membrane lipids with cytoskeletal elements
   and their associated motors, and thus have a role in the locomotion of
   the parasites^[166]24,[167]25. Importantly, the gene PBANKA_0915000
   (encoding the apical membrane antigen (AMA)), the genes PBANKA_1119600
   and PBANKA_1349100 (encoding merozoite surface proteins (MSPs)), which
   are critically involved in the initial interaction between merozoites
   and host erythrocytes before successful invasion, and genes encoding
   rhoptry proteins (ROPs) and rhoptry neck proteins (RONs), which are
   essential for erythrocyte invasion, were significantly downregulated in
   parasites of the KD-fed mice (Extended Data Fig. [168]4f,g and
   Supplementary Table [169]8). In addition, the transcription levels of
   genes related to ribosome biogenesis were also downregulated (Extended
   Data Fig. [170]4h and Supplementary Table [171]8). The above results
   not only corroborate the role of a KD in the protection of mice against
   P. berghei infection but also explain the underlying mechanism of the
   developmental suppression of parasites by a KD.

Extended Data Fig. 4. A ketogenic diet affected genome-wide transcription in
P. berghei ANKA parasites.

   [172]Extended Data Fig. 4
   [173]Open in a new tab

   a, Volcano plot showing that the expression of most genes encoding
   Plasmodium interspersed repeat (PIR) proteins was upregulated and that
   the expression of those encoding AMA1, myosin E and MSA180 were
   downregulated in parasites from KD-fed mice. The pink dots represent
   the upregulated DEGs in the KD group, the green dots represent the
   downregulated DEGs in the KD group, and the gray dots represent the
   non-differentially expressed genes. DGE analysis between the two
   groups, with Benjamini & Hochberg correction applied to P values for
   the false discovery rate (FDR). Genes with an FDR < 0.05 and a
   |log2(fold change) | ≥ 1 were considered DEGs. b-h, Mean value heatmaps
   showing that the transcription levels of genes associated with invasion
   and locomotion were decreased and that the expression of genes
   associated with cell cytoskeleton were upregulated in parasites from
   KD-fed mice. Genes with a |log[2]-fold change | ≥1 and a false
   discovery rate (FDR) < 0.05 are shown. Pink indicates highly expressed
   genes, and green indicates genes with low expression. Detailed
   information is shown in Supplementary Table [174]8 (n = 6). i, Pathways
   associated with motor proteins, DNA replication, base excision and
   mismatch repair were downregulated in parasites from KD-fed mice. j,
   The ribosome biogenesis pathway was downregulated in parasites from
   KD-fed mice.

   [175]Source data

   KEGG and gene set enrichment analysis (GSEA) investigations revealed
   that pathways associated with motor proteins, DNA replication, base
   excision, mismatch repair and ribosome biogenesis were suppressed in
   the P. berghei ANKA parasites of the KD-fed mice (Extended Data Fig.
   [176]4i,j), suggesting that a KD impairs multiple biological
   functions^[177]26,[178]27. The transcription and expression of some of
   the genes were validated by both qPCR and western blotting (Extended
   Data Fig. [179]5).

Extended Data Fig. 5. qPCR and Western blotting verified the accuracy of the
transcriptome results.

   [180]Extended Data Fig. 5
   [181]Open in a new tab

   a, Fifteen upregulated DEGs from the transcriptome in the P. berghei
   ANKA + KD group were randomly selected, and fragments per kilobase of
   transcript per million fragments mapped (FPKM) values were used as
   indicators to measure the level of gene expression (n = 6, Unpaired
   two-tailed Student’s t-tests). b, Five downregulated DEGs from the
   transcriptome in the P. berghei ANKA + KD group were randomly selected,
   with FPKM used as an indicator to measure the level of gene expression
   (n = 6, Unpaired two-tailed Student’s t-tests). c, Western blot results
   showing that the expression of DYNLL1 and tubulin, which are associated
   with the P. berghei ANKA cytoskeleton, was upregulated and that the
   expression of RPL7a and RPL30, which are related to the biosynthesis of
   ribosomes pathway, were downregulated in P. berghei ANKA after KD
   treatment. Representatives of 3 independently repeated times of each
   experiment with similar results. d-g, The relative expression levels of
   20 DEGs were consistent with the transcriptome results (n = 3, Unpaired
   two-tailed Student’s t-tests). Data are presented as means ± SEMs (a
   and b).

   [182]Source data

   The underlying mechanism of ketone-body-mediated parasite inhibition on
   P. falciparum was explored by single-cell RNA sequencing (scRNA-seq)
   analysis. A single-cell transcriptome atlas of P. falciparum 3D7 before
   and after βOHB treatment was first constructed and, based on
   stage-specific marker transcripts, the parasites were categorized into
   eight groups from the ring, trophozoite and schizont stages to the
   gametocyte stages (Fig. [183]3a,b, Extended Data Fig. [184]6a and
   Supplementary Table [185]9). The development of the intraerythrocytic
   parasites after βOHB treatment was mostly blocked at the early
   trophozoite stages (Fig. [186]3c–e and Extended Data Fig. [187]6b),
   leading to fewer schizonts, which resulted in a reduction of invasive
   merozoites (Fig. [188]3f). This phenomenon was illustrated by
   microscopic analyses of P. falciparum 3D7 iRBCs, and a majority of the
   parasites were blocked at the trophozoite stage (Fig. [189]3c,d).

Fig. 3. βOHB blocked the development of P. falciparum 3D7 parasites.

   [190]Fig. 3
   [191]Open in a new tab

   a, Heatmap showing the summed transcripts of stage-specific genes
   across eight intraerythrocytic developmental stages of the parasite.
   All marker genes are listed in Supplementary Table [192]9. Colours in a
   correspond to those in b. b, Single-cell transcriptome atlas across the
   erythrocytic stages of the parasite, with three biological samples in
   each group after Biscuit normalization and imputation based on
   stage-specific transcripts ([193]https://plasmodb.org/plasmo/app,
   [194]https://www.malariacellatlas.org/atlas/plasmodium-falciparum-atlas
   ). The cell types are colour-coded by Biscuit cluster. UMAP, uniform
   manifold approximation and projection. c, Blood smears taken every 8 h
   demonstrate that an IC[50] concentration (7.3 mM) of βOHB can inhibit
   parasite development, causing a developmental delay of approximately
   16 h. This inhibition can be reversed with the addition of 5 μM NAD^+.
   Representatives of three independently repeated experiments with
   similar results. d, Pie charts representing the distribution of
   parasites at ring, trophozoite and schizont stages over two
   developmental cycles (48 h and 96 h). βOHB treatment results in a
   developmental arrest predominantly at the trophozoite stage. e,
   Quantification of infected RBCs per 10,000 RBCs in different treatment
   groups after two developmental cycles (48 h and 96 h) (n = 3). f, Total
   number of iRBCs at ring stage per 10,000 RBCs following schizont
   rupture after one culture cycle (56 h for control and βOHB + NAD^+
   groups; 72 h for the βOHB-only group) (n = 3). g, Bubble map showing
   that βOHB inhibits the transcription of DEGs related to invasion and
   locomotion in P. falciparum 3D7 at mixed stages. Genes with a
   |log[2](FC)| ≥ 0.15 and P[adj] < 0.05 are shown. Detailed information
   is provided in Supplementary Table [195]11. Data are presented as
   means; error bars, s.e.m. (e and f). P values are from Tukey’s multiple
   comparisons test, adjusting for family-wise error rate (e and f).

   [196]Source data

Extended Data Fig. 6. βOHB impeded the development of P. falciparum 3D7.

   [197]Extended Data Fig. 6
   [198]Open in a new tab

   a, A stacked bar chart illustrates that the development of
   intraerythrocytic parasites was predominantly arrested at the early
   trophozoite stage, resulting in significantly reduced numbers of
   schizonts in the βOHB treated group. The cell types are color-coded. b,
   The number of infected erythrocytes at various stages in different
   treatment groups. P values in red letters indicate comparisons with the
   ring stage in the control group, P values in black letters represent
   comparisons with the trophozoite stage in the control group, and P
   values in blue letters denote comparisons with the schizont stage in
   the control group. The results demonstrate that βOHB significantly
   impeded parasite development from the trophozoite to the schizont stage
   (n = 3, Tukey’s multiple comparisons test by adjusting Family-wise
   Error Rate). Data are presented as means ± SEMs.

   [199]Source data

   The differentially expressed genes (DEGs) from βOHB-treated
   P. falciparum 3D7 were functionally categorized (Supplementary Table
   [200]10). KEGG pathway analysis revealed downregulation of genes
   involved in RBC invasion, parasite motility and translation and
   upregulation of genes involved in post-translational chaperone
   activity, protein structure modifications, protein export,
   post-translational single amino acid modification and RNA transcription
   and biogenesis (Extended Data Fig. [201]7 and Supplementary Table
   [202]11). The expression of the key invasion-related genes was all
   downregulated in the βOHB-treated parasites (Fig. [203]3g), consistent
   with the findings in the P. berghei parasites from KD-fed mice
   (Extended Data Fig. [204]4a–g). In addition, the transcription of the
   serine repeat antigen (SERA) family, especially PF3D7-0207600 (encoding
   PfSERA5), which is conserved among species of the Plasmodium genus and
   is involved in the egress of P. falciparum merozoites^[205]28,[206]29,
   was also downregulated (Fig. [207]3g). However, the transcription of
   genes encoding knob structural proteins was upregulated (Fig. [208]3g).
   The transcription and expression of some of the genes were validated by
   qPCR (Extended Data Fig. [209]8).

Extended Data Fig. 7. βOHB downregulated the expression of genes associated
with P. falciparum 3D7 erythrocyte invasion and motility.

   [210]Extended Data Fig. 7
   [211]Open in a new tab

   a, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed
   that the expression of genes encoding proteins associated with the
   invasion, motility and translation pathways of P. falciparum 3D7
   parasites were downregulated in the βOHB-treated group. b,
   Posttranslational chaperones and protein structure modifications,
   protein export, single amino acid modification and biogenesis of RNA
   pathways were activated in P. falciparum 3D7 parasites after βOHB
   treatment. Statistical comparisons between groups were performed using
   two-sided Wilcoxon rank-sum tests and selected on the basis of a
   threshold: FDR (adjusted P value) ≤ 0.05.

   [212]Source data

Extended Data Fig. 8. qPCR confirmation of the transcription levels of genes
associated with protein trafficking and that related to parasite invasion and
locomotion of P. falciparum 3D7 post βOHB treatment.

   [213]Extended Data Fig. 8
   [214]Open in a new tab

   a, Ten DEGs related to parasite invasion and locomotion were
   downregulated in the P. falciparum 3D7 + βOHB group. b, Nine DEGs
   related to parasite development were downregulated in the P. falciparum
   3D7 + βOHB group. c, Eight DEGs associated with protein trafficking
   were upregulated in the P. falciparum 3D7 + βOHB group. (n = 3,
   Unpaired two-tailed Student’s t-tests). Data are all presented as
   means ± SEMs.

   [215]Source data

   DEGs of the βOHB-treated P. falciparum at different developmental
   stages were further analysed, and genes encoding proteins specifically
   expressed at the merozoite and early ring stages presented greater
   transcriptional variation (Fig. [216]4a and Supplementary Tables
   [217]12 and [218]13). Therefore, we selected all DEGs identified in the
   merozoite stage and early ring stage for further analysis
   (Supplementary Tables [219]14–[220]17). Most of the increased
   transcripts were WD repeat-containing proteins (Fig. [221]4b;
   consistent with the results obtained in P. berghei from the KD-fed
   mice) functioning in protein trafficking, some proteins were active
   only in the liver and sexual stages of the parasite (Fig. [222]4c,d)
   and some proteins were involved in pre-mRNA splicing (Fig. [223]4e).
   The downregulated DEGs in P. falciparum post exposure to βOHB were
   those coding for MSPs (Fig. [224]4f), RH, EBA (Fig. [225]4g), IMC,
   PhIL1-interacting candidates, glideosome-associated connectors, myosin
   A tail domain interacting protein (MTIP) and myosins (Fig. [226]4h),
   which are unilaterally related to erythrocyte adhesion, invasion and
   motility pathways. In addition, we also observed that the transcription
   of genes related to parasite development, such as PF3D7_0730300 and
   PF3D7_0420300 (encoding Apicomplexan AP2 domain transcription factor
   (APiAP2)), PF3D7_1122800 (encoding calcium-dependent protein kinase 6
   (CDPK 6)) and PF3D7_0206200 (encoding pantothenate transporter (PAT)),
   was downregulated (Fig. [227]4i). Similar results were obtained for the
   transcripts identified in the iRBCs at the early ring stage of the
   parasites (Extended Data Fig. [228]9a–e and Supplementary Table
   [229]15). These data collectively suggest that the inhibitory effect of
   βOHB on both P. berghei ANKA and P. falciparum 3D7 is mainly through
   suppressing the expression of erythrocyte invasion and intracellular
   development.

Fig. 4. βOHB disrupted transcriptional processes in P. falciparum 3D7
parasites.

   [230]Fig. 4
   [231]Open in a new tab

   a, Volcano plot showing that βOHB affects genome-wide transcription in
   P. falciparum 3D7 parasites during intraerythrocytic development. The
   genes associated with the early development and erythrocyte invasion of
   parasites were mostly affected. Genes with a |log[2](FC)| ≥ 0.25 and
   P[adj] < 0.05 are shown. Parasite genes at various developmental stages
   are colour-coded, with arrows pointing to specific genes. Developmental
   stages are colour-coded. b–i, Bar charts of DEGs in the merozoite stage
   showing that the transcription of WD repeat-containing proteins (b) and
   proteins that function in trafficking (c), liver and sexual stages (d)
   and pre-mRNA splicing (e) was upregulated after exposure to βOHB, and
   the transcription of genes related to MSP (f), erythrocyte adhesion
   (g), locomotion (h) and development (i) was downregulated. Brown
   indicates upregulated genes; dark green indicates downregulated genes.
   Genes with a |log[2](FC)| ≥ 0.15 and P[adj] < 0.05 are shown. Detailed
   information is provided in Supplementary Table [232]14. j, KEGG
   enrichment column showing that invasion, motility and
   post-translational ubiquitin-dependent process pathways were
   significantly downregulated in the merozoites of P. falciparum 3D7
   exposed to βOHB. k, KEGG enrichment column showing that invasion,
   motility and post-translational ubiquitin-dependent process pathways
   were significantly downregulated in the early ring stage of
   P. falciparum 3D7 after exposure to βOHB. Detailed information is shown
   in Supplementary Tables [233]16 and [234]17. l, Western blotting
   results show that βOHB significantly reduced the expression of MSP1 and
   MYC-tagged MTIP, and NAD replenishment restored the βOHB-mediated
   suppression in the expression of these proteins. However, we failed to
   detect ApiAP2 (PF3D7_0420300) in western blot assays, probably because
   of its ultra-high molecular weight. Representatives of three
   independently repeated experiments with similar results. Statistical
   comparisons between groups were performed using two-sided Wilcoxon
   rank-sum tests (a–k).

   [235]Source data

Extended Data Fig. 9. Bar chart of DEGs in P. falciparum 3D7 at the early
ring stage and KEGG enrichment results in merozoite and early ring stages
post βOHB treatment.

   [236]Extended Data Fig. 9
   [237]Open in a new tab

   a-e, Bar charts of DEGs in P. falciparum 3D7 at the early ring stage
   showing the expression of WD repeat-containing proteins; the expression
   of trafficking proteins was upregulated, and the expression of genes
   related to erythrocyte adhesion, parasite locomotion and development
   was upregulated after exposure to βOHB. Purple indicates upregulated
   genes, and white indicates downregulated genes. Genes with a
   |log[2]fold change | ≥0.15 and P adj. values < 0.05 are shown. Detailed
   information is provided in Supplementary Table [238]17. f, KEGG
   enrichment showing that transcription and biogenesis of RNA, along with
   posttranslational single amino acid modification pathways, were
   activated in the P. falciparum 3D7 + βOHB group during the merozoite
   stage. g, KEGG enrichment showing that posttranslational single amino
   acid modification, posttranslational-other protein modifications
   pathways were activated in the early ring stage under βOHB-treated
   conditions. Detailed information was shown in Supplementary Tables
   [239]22 and [240]23.

   [241]Source data

   Next, all DEGs from the merozoite stage and early ring stage were
   selected for KEGG analysis. Many pathways related to post-translational
   modifications were enriched (Fig. [242]4j,k and Extended Data Fig.
   [243]9f,g). Invasion, motility and post-translational
   ubiquitin-dependent pathways in the merozoite and early ring stages in
   βOHB-treated P. falciparum 3D7 parasites were significantly
   downregulated (Fig. [244]4j,k). Conversely, DEGs involved in pathways
   such as transcription and biogenesis of RNA, post-translational single
   amino acid modification pathways and post-translational chaperone and
   protein structure modifications in the merozoite and early ring stages
   were upregulated in P. falciparum after βOHB treatment (Extended Data
   Fig. [245]9f,g). Notably, these results are consistent with our
   findings from P. berghei ANKA in KD-fed mice (Extended Data Fig.
   [246]4). These data collectively contribute to the elucidation of the
   mechanism of βOHB-mediated suppression of malaria parasite invasion and
   intraerythrocytic development.

   To further validate the aforementioned mechanism, we tested the
   responses of MYC-tagged MTIP (PF3D7_1246400) and TY1-tagged ApiAP2
   (PF3D7_0420300) P. falciparum strains to the treatment of βOHB using
   western blotting and immunofluorescence assays, respectively. The
   results indicated that βOHB significantly reduced the expression of
   ApiAP2, MTIP and MSP1 (PF3D7_0930300) (Fig. [247]4l and Extended Data
   Fig. [248]10a–c). Additionally, we supplemented NAD in the βOHB-treated
   parasites, which demonstrated that NAD replenishment reversed the
   βOHB-mediated suppression in the expression of ApiAP2, MTIP and MSP1
   (Fig. [249]4l and Extended Data Fig. [250]10a–c). MTIP and MSP1 are
   crucial molecules in erythrocyte invasion and intraerythrocytic
   development of P. falciparum 3D7 (refs. ^[251]30,[252]31). Furthermore,
   the decreased expression of ApiAP2 correlated with the hindrance of the
   intracellular development of P. falciparum 3D7 (Extended Data Fig.
   [253]10d). The observation is in line with a previous study that ApiAP2
   is highly expressed in the merozoite stage^[254]32. Additionally, the
   sensitivity of the ApiAp2-knowndown parasites to βOHB treatment was
   weaker that that of vector control parasites (Extended Data Fig.
   [255]10e). In summary, βOHB reduced the expression of key parasite
   invasion and development proteins, such as MTIP, ApiAP2 and MSP1, and
   this suppression can be reversed by NAD supplementation, highlighting
   the critical role of metabolic reprogramming in regulating
   P. falciparum growth and pathogenicity.

Extended Data Fig. 10. βOHB inhibited merozoite protein expression and
parasite growth via ApiAp2 regulation in P. falciparum.

   [256]Extended Data Fig. 10
   [257]Open in a new tab

   a-c, βOHB significantly reduced the expression of PfMSP1, PfMTIP, and
   PfApiAP2, while NAD replenishment restored the βOHB-mediated
   suppression in the expression of these proteins. These proteins are
   predominantly expressed at the merozoite or late schizont stages. Scale
   bar =5 μm. Representatives of 3 independently repeated times of each
   experiment with similar results. d, The growth curves of P. falciparum
   3D7 parasites and ApiAP2 (PF3D7_0420300) knockdown parasites over three
   developmental cycles. The results revealed that the deletion of ApiAP2
   significantly impaired parasite growth (n = 3, Unpaired two-tailed
   Student’s t-tests). Data are presented as means ± SEMs. e, The
   sensitivity of the ApiAp2-knowndown parasites to βOHB treatment was
   weaker compared to vector control parasites (R^2 of vector control
   strain = 0.9402, Confidence interval of IC[50] value = 7 mM - 8.7 mM;
   R^2 of ApiAp2-knowndown strain = 0.9875, Confidence interval of IC[50]
   value = 9.4 mM -10.4 mM).

   [258]Source data

Discussion

   The topic of diet and health consistently garners significant
   attention, and among various dietary components, KD has gained
   widespread interest for its potential benefits regarding weight loss,
   metabolic diseases, brain health and cancer
   intervention^[259]8–[260]11. In 1954, Ramakrishnan first proposed the
   potential of the KD in controlling malaria in rats^[261]33. Our current
   study significantly advances this concept by providing in-depth
   molecular and mechanistic insights, made possible by modern multi-omics
   techniques and advanced molecular biology approaches. In this study, we
   demonstrated that administering a KD to mice resulted in elevated blood
   concentrations of βOHB, a ketone body produced in the liver during
   ketogenesis, which provided complete protection against P. berghei ANKA
   infection (Fig. [262]1, Supplementary Figs. [263]1 and [264]7 and
   Extended Data Fig. [265]2). Implantation with an osmotic pump filled
   with βOHB also protected mice from P. berghei ANKA infection (Fig.
   [266]1). Notably, βOHB also significantly inhibited parasite invasion
   and the intraerythrocytic development of the human malaria parasite
   P. falciparum 3D7 (Fig. [267]1). Furthermore, we revealed that βOHB
   reduced the level of intracellular NAD (Fig. [268]2), which is closely
   associated with gene activation and silencing^[269]17 in both
   P. berghei ANKA and P. falciparum 3D7. This reduction probably led to
   the transcriptional downregulation of the genes associated with
   erythrocyte recognition, attachment, invasion and intraerythrocytic
   development (Figs. [270]3 and [271]4).

   βOHB, a primary ketone body and energy-rich compound, was found to
   inhibit the proliferation of P. falciparum 3D7 (Fig. [272]1). The
   underlying reason for this effect was probably metabolic reprogramming
   induced by βOHB (Fig. [273]2), which reduced the intracellular NAD
   levels in both parasites (Fig. [274]2). NAD has consistently been
   proved to function as a fulcrum for regulating various cellular
   processes in response to environmental changes^[275]17. It also has a
   pivotal role in the regulation of histone deacetylation. SIRT1, an
   NAD-dependent HDAC, removes acetyl groups from lysine residues on
   histone proteins, leading to a more condensed chromatin structure and
   subsequently reduced gene expression. The activity of SIRT1 is directly
   regulated by cellular NAD levels, making NAD a crucial factor in
   modulating gene expression through chromatin remodelling^[276]34.
   Additionally, decreased blood l-citrulline levels have been associated
   with the alleviation of severe malaria^[277]35, and l-citrulline
   supplementation can inhibit Plasmodium growth and protect against
   cerebral malaria^[278]36.

   The observation that βOHB levels increase during Plasmodium infection,
   even in the absence of a KD, raises interesting questions about the
   metabolic changes induced by malaria^[279]37. We also noted a trend of
   increased βOHB in P. berghei-infected mice (Supplementary Fig. [280]1),
   although this increase was not statistically significant. This suggests
   that the parasite infection itself may induce a metabolic shift toward
   ketogenesis. A plausible explanation is the reduction in food intake
   and body weight observed in infected mice, particularly during the
   later stages of infection. Starvation or fasting is a well-known
   trigger for the liver to increase βOHB production as the body shifts
   from glucose to fat metabolism for energy. Thus, the rise in βOHB
   levels during malaria could reflect a combination of the impact of the
   disease on the host’s nutritional status and a response to metabolic
   stress.

   Gene transcriptional and functional analyses revealed that the most
   compelling reason for the antiparasitic effects of the KD and βOHB is
   the downregulation of expression of genes encoding proteins involved in
   erythrocyte attachment and invasion. These genes included those coding
   for AMA1, MSP family, RH family, EBA family and the ROP and RON
   proteins (Figs. [281]3 and [282]4), all of which are directly
   associated with erythrocyte recognition and invasion^[283]38,[284]39.
   Additionally, the transcription levels of the SERA family, particularly
   PfSERA5, which is involved in the egress of P. falciparum
   merozoites^[285]28,[286]29, were decreased (Fig. [287]3). Importantly,
   we found that the transcription level of calcium-dependent protein
   kinase 5 (CDPK5) was downregulated (Fig. [288]4, Extended Data Fig.
   [289]8 and Supplementary Table [290]17) in the KD group, which is also
   required for parasite replication and egress from the host
   cell^[291]40. Furthermore, increasing evidence has pointed to the
   critical roles of AP2 homologous proteins in transcriptional regulation
   during the blood stage of Plasmodium development^[292]41. Herein, the
   expression of genes encoding AP2-O (PF3D7_1143100) and APiAP2
   (PF3D7_0420300 and PF3D7_0730300), with many predicted targets, was
   downregulated in P. falciparum 3D7 parasites during the merozoite and
   early ring stages under βOHB treatment (Fig. [293]4 and Supplementary
   Tables [294]14 and [295]17), supporting their roles in regulation of
   the Plasmodium development in the blood stages^[296]41. However, the
   specific functions of these genes during the erythrocytic stage have
   not been extensively elucidated. Previous studies have indicated that
   members of the APiAP2 family have critical roles in the transcription
   of a subset of host cell invasion-associated genes^[297]42. For
   example, the absence of AP2-O, the expression of which was also
   downregulated in our study following βOHB treatment (Supplementary
   Tables [298]14 and [299]17), significantly weakened the initial
   erythrocyte adhesion of P. falciparum 3D7 (ref. ^[300]43). We also
   found that the expression of ApiAP2 (PF3D7_0420300) was depressed by
   βOHB, and the effect could be reversed by the addition of NAD (Extended
   Data Fig. [301]10). The data were not only consistent with the previous
   report that ApiAP2 is highly expressed in the merozoite^[302]32, but
   also strongly suggests its critical role in merozoite development.
   Here, the reduced sensitivity of ApiAP2-knockdown parasites to βOHB
   further supports that ApiAP2 is a critical regulator in the parasite
   responses to metabolic shifts. Additionally, MSP1 is an essential
   protein for Plasmodium invasion of RBC and schizont rupture^[303]44,
   and it undergoes various post-translational modifications^[304]45. In
   this study, we observed a significant decrease in MSP1 expression and
   reduced parasitaemia under βOHB treatment (Fig. [305]4 and Extended
   Data Fig. [306]10), which also explains the molecular mechanism of the
   anti-Plasmodium effect exerted by the ketone body.

   Despite these insights, the precise mechanisms underlying the effects
   of βOHB remain incompletely understood. Our experiments with the
   ApiAP2-knockdown strain provided preliminary evidence of
   transcriptional regulation being involved; however, further
   investigation is required to fully elucidate its role. The rescue
   experiments with NAD are particularly intriguing and suggest that βOHB
   may exert its effects, at least in part, through the modulation of
   NAD-dependent pathways. In this context, the role of Sir2A
   (PF3D7_1328800), a Plasmodium SIRT1 homologue and NAD-dependent
   deacetylase, warrants further exploration. Previous studies have shown
   that Sir2A has a critical role in regulating the parasite transcriptome
   and proliferation^[307]46,[308]47. Although we did not directly assess
   Sir2A activity, its potential involvement represents an exciting avenue
   for future research. Additionally, the differences in βOHB
   concentration and duration of exposure between the pump-administered
   model and KD were apparent, which may have been caused by the
   limitations of the pump. The pump can only achieve a maximum blood βOHB
   concentration of 1 mM for only 7 days (Fig. [309]1), whereas the KD
   induces levels up to 6 mM (Supplementary Fig. [310]2). Even at these
   lower concentrations, βOHB demonstrated significant inhibitory effects
   on parasitaemia compared to the control group (Fig. [311]1). This
   highlights the importance of considering both the concentration and
   temporal dynamics of βOHB exposure when evaluating its effects.
   Furthermore, the observed delay in parasite development and its
   downstream effects on host survival may involve both direct and
   indirect mechanisms. In addition to the potential impact of βOHB on
   parasite metabolic pathways, host immune modulation may also contribute
   to the effect. The extended survival in βOHB-treated mice probably
   arises from a combination of factors, including effects on host
   inflammation and immune function^[312]29. Future studies integrating
   immunological profiling and direct assessments of Sir2A activity may be
   necessary for dissecting these interconnected mechanisms.

   In summary, our study demonstrated that a KD confers complete
   protection against P. berghei ANKA infection in mice and developmental
   arrest of the P. falciparum 3D7 parasites. The underlying mechanisms
   involve the disruption of parasite metabolism and the inhibition of
   genes associated with erythrocyte invasion and intracellular
   development (Supplementary Fig. [313]8). The KD-based treatment, as
   well as the increasing availability of ketogenic food products, offers
   a promising potential for disease ‘metabotherapy’ and a novel approach
   for malaria intervention. The study highlights the potential of
   targeting metabolic pathways in Plasmodium as a therapeutic strategy
   for malaria. By manipulating host metabolism, such as through dietary
   interventions or metabolic modulators, it may be possible to create a
   hostile environment for the malaria parasites, impairing their ability
   to thrive and propagate inside the hosts.

Methods

Ethics statement

   All procedures performed on the experimental animals were conducted in
   accordance with the animal husbandry guidelines of Shenyang
   Agricultural University. The institutional ethics committee of Shenyang
   Agricultural University approved the laboratory animal experiments
   (permit no. SYXK <Liao>2021-0010).

Mice and Plasmodium parasites

   C57BL/6 mice (N phenotype) and BALB/c mice (6–8 weeks old, 18–20 g)
   were purchased from Liaoning Changsheng Biological Technology Company.
   The mice were housed under pathogen-free conditions at 22–24 °C with
   45–65% humidity, had free access to food and water and were maintained
   on a 12 h light–dark cycle. All mice were acclimated for 2 weeks before
   experimentation. At the beginning of each experiment, the mice were
   randomly assigned to experimental groups.

   The wild-type and P. falciparum 3D7 MYC-tagged MTIP strains were
   maintained in our laboratory. The ApiAP2-knockdown P. falciparum strain
   and vector control strain were kindly provided by Q. Zhang from Tongji
   University^[314]48. The parasites were routinely cultivated in a
   malaria complete medium (MCM) with 10% human serum as previously
   described^[315]1. The GFP-expressing P. berghei ANKA strain was kindly
   provided by C. Yaming from China Medical University^[316]49.

Parasite maintenance and in vitro cultivation

   The P. berghei ANKA strain was maintained by infecting C57BL/6 mice by
   intraperitoneal injection with 1 × 10^5 parasitized
   erythrocytes^[317]50. Parasitaemia was monitored by examining
   Giemsa-stained thin blood smears under a light microscope. The
   P. alciparum 3D7 strain was cultured and synchronized in vitro
   according to standard procedures^[318]51. In brief, P. falciparum 3D7
   parasites were cultivated in human O^+ RBCs at 5% haematocrit in MCM
   containing 10% human serum and 0.25% AlbuMAX II (Gibco, Invitrogen) in
   an incubator at 37 °C with 5% CO[2]. The culture medium was changed
   every second day.

KD feeding

   For the KD experiments (XTKD01 from Xietong Pharmaceutical
   Bioengineering), female C57BL/6 mice were randomly assigned to one of
   six groups: a P. berghei ANKA group (mice infected with 1 × 10^5
   P. berghei ANKA by intraperitoneal injection^[319]52 and fed a RD) and
   five KD (0.5% KD, 20% KD, 50% KD, 70% KD and 90% KD) groups (mice
   infected with 1 × 10^5 P. berghei ANKA and fed different KDs ad
   libitum). Before infection, all mice in the KD groups were fed a KD for
   7 days. Parasitaemia in the mice was assessed by microscopy (Leica
   Microsystems CMS, DM4 B, GRE) every 2–3 days by counting cells (at
   least 5,000 RBCs) in Giemsa-stained smears from a droplet of tail
   blood. All animals were monitored daily for survival. At 60 days post
   infection, KD was changed to a RD in the five KD groups.

Toxicity detection and βOHB measurement

   To test the possible toxicological effect of the KD, daily measurements
   of food intake of the control and KD-fed mice (RD, 0.5% KD, 20% KD, 50%
   KD, 70% KD and 90% KD) without parasite infection were conducted to
   determine caloric consumption, and body weights were recorded every
   3 days. On the 30th day post feeding, blood and liver samples were
   collected under isoflurane anaesthesia, with blood drawn from the
   inferior vena cava. A 200 μl aliquot was allocated for haematological
   analysis (including RBC count, haemoglobin concentration, mean
   corpuscular haemoglobin concentration and haematocrit), and the
   remainder was processed for hepatic and renal function assessments.

   The liver weights were recorded, and liver indices were calculated
   based on these measurements. Haematological parameters were analysed
   using the TEK-VET3 automatic haematology analyser, while biochemical
   indicators for hepatic and renal function were measured with the
   Chemray 800 automatic biochemical analyser (Shenzhen Rayto Life
   Technology).

   For the liver function tests, ALT, AST and γ-GT levels were assessed
   using 20 μl serum samples, with the respective assay kits specified as
   follows: ALT assay kit (cat. no. S03030, lot no. 20240430), AST assay
   kit (cat. no. S03040, lot no. 20240428) and γ-GT assay kit (cat. no.
   S03031, lot no. 20240422), all supplied by Rayto. Renal function was
   evaluated using 3 μl and 8 μl serum samples for UREA and CREA,
   respectively, with UREA measured by the UREA assay kit (cat. no.
   S03036, lot no. 20240529) and CREA by the CREA assay kit (cat. no.
   S03076, lot no. 20240527), also provided by Rayto.

   βOHB levels were measured in the serum of each mouse from all
   experimental groups (Fig. [320]1d). Blood was collected from the
   orbital vein into pro-coagulation tubes. Serum samples were obtained by
   centrifugation at 1,000g for 10 min at 4 °C, followed by βOHB
   measurements using a colourimetric assay kit (Elabscience, E-BC-K785-M)
   according to the manufacturer’s instructions.

Evaluation of haemolytic potential of βOHB

   The in vitro haemolysis assay was conducted to evaluate the haemolytic
   potential of βOHB. Various concentrations of βOHB were incubated with a
   suspension of normal human erythrocytes at 4% haematocrit for 2 h at
   37 °C. Following incubation, the erythrocytes were pelleted by
   centrifugation at 250g for 5 min. The supernatant was then mixed in a
   16:1 volume ratio with a demulsifier solution consisting of anhydrous
   ethanol and concentrated hydrochloric acid at a 400:1 ratio. The
   samples were further centrifuged at 16,000g for 10 min, and the
   absorbance of the resulting solutions was measured at 400 nm using a
   microplate reader (PerkinElmer VICTOR Nivo).

βOHB osmotic pump implantation

   A 7-day continuous osmotic infusion pump (RWD model, 2001 W, 5 g ml^−1
   and 10 g ml^−1, respectively, 1 µl h^−1, dissolved in PBS, for a
   maximum of 7 days) was used to deliver βOHB subcutaneously in C57BL/6
   mice post infection with P. berghei ANKA parasites. The mice in the
   negative control group received osmotic pumps filled with isotonic
   saline solution. Parasitaemia in the mice was assessed by microscopy
   (Leica Microsystems CMS, DM4 B, GRE) every 2–3 days by counting cells
   (at least 5,000 RBCs) in Giemsa-stained smears from a droplet of tail
   blood. All animals were monitored daily for an 8-day survival rate.

   For detection of βOHB in vivo, mice were anaesthetized with isoflurane,
   and the pumps were implanted subcutaneously following sterilization of
   the surgical area, then secured with sutures. Then, 3 days post
   surgery, blood was collected from the facial vein, sera were isolated
   from the blood of the mice and centrifuged using a 50 kDa
   ultrafiltration tube at 16,000g, 4 °C for 15 min. Serum βOHB levels
   were then measured using a colourimetric assay kit (Elabscience,
   E-BC-K785-M) according to the manufacturer’s protocol.

In vitro treatment of P.falciparum with βOHB, acetoacetate and NAD rescue
experiment

   The P. falciparum 3D7 strain cultivated in MCM with 10% human serum was
   used to test the antimalarial effects of βOHB (MedChemExpress,
   HY-113378) in 96-well plates. βOHB was dissolved in MCM. The parasites
   were cultivated in O^+ erythrocytes by the addition of 0, 2, 4, 8, 10,
   12, 14, 16, 20 or 25 mM βOHB and incubated for 96 h (two growth
   cycles). Dihydroartemisinin was used as a positive control.

   To determine the effects of acetoacetate on parasites, acetoacetic acid
   sodium (MedChemExpress, HY-112540B) was used^[321]53. The parasites
   were cultured in O^+ erythrocytes incubated with 0, 0.05, 0.1, 0.5, 1,
   2, 4, 8 or 10 mM AcAc with or without 7.3 mM βOHB for 96 h.

   To perform the NAD rescue experiment, the parasitaemia of the
   P. falciparum 3D7 strain at the ring stage was adjusted to 0.7%. NAD^+,
   supplemented with 5 mM βOHB, was added to the cultures with
   concentrations of 0, 0.5, 1, 3 and 5 µM. The parasites were cultivated
   for two developmental cycles. P. falciparum iRBCs were stained with
   Vybrant DyeCycle Green (Invitrogen, V35004)^[322]54 and analysed using
   a BD FACSAria III flow cytometer (BD Biosciences) according to the
   manufacturer’s instructions. Parasitaemia and the IC[50] were
   calculated using the modified Peters–Robinson formula^[323]55.

Assessment of AcAc concentrations in P.falciparum cultures and mouse sera

   To test the stability and potential effect of AcAc on the P. falciparum
   parasites, synchronized P. falciparum 3D7 parasites at the ring stage
   were adjusted to 0.7%. AcAcNa (MCE, HY-112540B) was dissolved in MCM
   and added to the parasite culture with concentrations of 0, 0.05, 0.1,
   0.5, 1, 2, 4, 8 and 10 mM, and then incubated for two cycles. The
   supernatant was then collected by centrifugation, and AcAc levels were
   measured using the Acetoacetate Content Assay Kit (Shanghai Acmec
   Biochemical, AC10787-100T/48S, lot. no. A6543AABE) according to the
   manufacturer’s instructions.

   For the in vivo assay, mice were fed with RD or 0.5% to 90% KD for
   2 weeks. Following the dietary intervention, mice were anaesthetized
   with avertin, and blood samples were collected from the inferior vena
   cava, then left to stand at 4 °C for 30 min. The serum from each mouse
   was separated by centrifugation at 1,000g for 10 min at 4 °C, and the
   AcAc levels in the blood of the mice were determined using the
   Acetoacetate Content Assay Kit (Shanghai Acmec Biochemical,
   AC10787-100T/48S, lot. no. A6543AABE), following the manufacturer’s
   protocol.

βOHB-induced parasite development arrest experiment

   P. falciparum 3D7 strain at the ring stage was adjusted to 0.7%
   parasitaemia. The cultures were divided into three treatment groups
   with media containing only the βOHB solvent, 5 mM βOHB + 0 µM NAD or
   5 mM βOHB + 5 µM NAD. Samples of 100 µl were taken every 8 h for blood
   smear preparation and Giemsa staining to monitor parasite development
   over time in each treatment group.

Assessment of the accumulation of P.berghei parasites in mice by in vivo
imaging

   A transgenic P. berghei ANKA parasite strain expressing luciferase
   under the control of the ef1a promoter (P. berghei ANKA-luc)^[324]56
   was provided by W. Xu (Army Medical University, China) and propagated
   by passage in mice. After P. berghei ANKA-luc infection, the
   accumulation of parasites in both the RD-fed and KD-fed mice was
   assessed with an AniView600 multimode in vivo animal imaging system
   (Guangzhou Biolight Biotechnology) after the intraperitoneal injection
   of high-purity d-luciferin (Biyuntian, ST198). At least three mice were
   used in each group.

Isolation of P.berghei ANKA and P.falciparum iRBCs after KD or βOHB treatment

   To collect enough KD-treated parasites, we fed the mice with 90% KD and
   infected them with P. berghei ANKA at the same time. At 7 days post
   infection, iRBCs were purified from the blood of infected mice by
   Percoll gradient centrifugation. In brief, blood samples were washed
   three times with sterile PBS, resuspended in 2 ml of 1× serum-free RPMI
   1640 culture medium and fractionated using a Percoll gradient
   (Supplementary Table [325]18). This procedure allowed for the
   separation of iRBCs from other host cells. The topmost cell debris was
   discarded, and the iRBCs were transferred to a new 15 ml centrifuge
   tube and washed three times with 1× PBS.

   For the metabolic analysis, iRBCs containing pure ring-stage,
   trophozoite-stage and schizont-stage P. falciparum 3D7 parasites were
   purified after exposure to βOHB, using MitoTracker Green (Invitrogen,
   M-7512) and then sorted with an Aria III flow cytometer (BD
   Biosciences)^[326]57. For scRNA-seq analysis, P. falciparum 3D7
   parasites at the ring, trophozoite and schizont stages were exposed to
   IC[50] concentration (7.3 mM) of βOHB for 96 h (two growth cycles;
   Supplementary Fig. [327]9). MitoTracker Green (Invitrogen, M-7512) was
   used according to the manufacturer’s instructions to monitor the
   viability of parasites, and all pure iRBCs were obtained by sorting
   with an Aria III flow cytometer (BD Biosciences). The sorting order of
   samples from the experimental and control groups was not in a fixed
   sequence but randomized to avoid systematic bias caused by the sorting
   order. It took approximately 10 min per sample, and all six samples
   were sorted within approximately 1 h in total. During the sorting
   process, all samples, including both control and experimental groups,
   were consistently maintained on ice.

RNA isolation of P.berghei ANKA parasites

   Total RNA was extracted from the purified iRBCs using TRIzol
   (Invitrogen, 15596026) according to the standard procedure^[328]58.
   Total RNA was subsequently identified and quantified using a Qubit 2.0
   fluorometer (Life Technologies) and a QSEP400 high-throughput
   bio-fragment analyser (BiOptic).

Transcriptomic analysis of P.berghei parasites isolated from KD-fed mice

   cDNA libraries were synthesized and sequenced on the Illumina
   sequencing platform by MetWare Biotechnology. Data quality control was
   performed using fastp software (v.0.23.2) to remove reads with
   adaptors. Paired reads were discarded if the number of N bases in any
   sequencing read exceeded 10% of its length or if any sequencing read
   contained low-quality bases (Q ≤ 20) exceeding 50% of its length.
   Subsequent analyses were conducted using clean reads. The reference
   genome and its annotation files were downloaded from PlasmoDB
   ([329]https://plasmodb.org/plasmo/app). HISAT2 (v.2.2.1) was used to
   construct an index, and the clean reads were aligned to the reference
   genome. Gene expression levels were quantified using feature Counts
   (v.2.0.3) to calculate gene alignment statistics, and fragments per
   kilobase million values for each gene were subsequently computed based
   on gene length, which is currently the most widely used method for
   estimating gene expression levels^[330]59. DESeq2 (v.1.38.3) was used
   for differential gene expression analysis between the two groups, with
   the Benjamini–Hochberg correction applied to P values for the false
   discovery rate (FDR). Genes with FDR < 0.05 and |log[2](fold
   change)| ≥ 1 were considered DEGs. Enrichment analysis of DEGs was
   performed using KEGG ([331]https://www.kegg.jp), Clusters of
   Orthologous Groups for Eukaryotic Complete Genomes
   ([332]https://www.ncbi.nlm.nih.gov/COG) and Gene Ontology
   ([333]http://geneontology.org). GSEA was conducted using the
   clusterProfiler tool (v.4.6.0) for further analysis.

Targeted metabolic analysis

   A total of 1 × 10^8 iRBC samples with the same cell number were mixed
   with 500 µl of 80% methanol in water (precooled at −20 °C), vortexed
   for 2 min, frozen in liquid nitrogen for 5 min, placed on ice for
   another 5 min and then vortexed for another 2 min. These steps were
   repeated three times. The samples were then centrifuged at 16,000g for
   10 min at 4 °C. Subsequently, 300 μl of the supernatant was transferred
   to a new centrifuge tube and placed in a −20 °C refrigerator for
   30 min. Finally, the supernatant was centrifuged again at 16,000g for
   10 min at 4 °C, and 200 μl of the supernatant was transferred through a
   protein precipitation plate for further liquid chromatography–mass
   spectrometry (LC–MS) analysis. The sample extracts were analysed using
   an LC‒ESI‒MS/MS system (Waters ACQUITY H-Class; MS, QTRAP 6500+
   System). The analytical conditions were as follows: HPLC column,
   ACQUITY UPLC BEH Amide (internal diameter, 2.1 × 100 mm, 1.7 μm);
   solvent system, water with 10 mM ammonium acetate and 0.3% ammonium
   hydroxide (A), 90% acetonitrile in water (V/V) (B); gradient, starting
   at 95% B (0–1.2 min), decreasing to 70% B (8 min), 50% B (9–11 min) and
   finally ramping back to 95% B (11.1–15 min); flow rate, 0.4 ml min^−1;
   temperature, 40 °C; and injection volume, 2 μl. The AB 6500 + QTRAP
   LC‒MS/MS system was equipped with an ESI Turbo Ion‒Spray interface,
   operating in both positive and negative ion modes and was controlled by
   Analyst 1.6 software (AB Sciex). The ESI source operation parameters
   were as follows: ion source, turbo spray; source temperature, 550 °C;
   ion spray voltage, 5,500 V (positive) and −4,500 V (negative); and
   curtain gas, 35.0 psi. Metabolites were identified by MetWare
   ([334]http://www.metware.cn) using the AB Sciex QTRAP 6500 LC‒MS/MS
   platform. Known metabolites reported in this study conform to
   confidence level 1 (the highest confidence level of identification).
   Metabolites were quantified by peak area using the internal standard
   method. Each sample’s group peaks with >50% missing data were filtered
   out using the ‘SVR’ method after the peak areas were corrected.
   Targeted metabolite data were normalized using unit variance scaling
   and mean centring^[335]60.

scRNA-seq of P.falciparum 3D7 iRBCs after exposure to βOHB

   The protocol provided in the Chromium Single Cell 3 Reagent Kit (V3
   Chemistry) was used to obtain six single-cell libraries. The final
   library pool was sequenced on an Illumina NovaSeq 6000 instrument using
   150 bp paired-end reads. Downstream analysis was performed with Seurat
   (v.2.2). Data were filtered to remove cells with high mitochondrial
   reads (>10%), low-quality gene detection (<200 or >10,000) and high RBC
   marker gene reads (>10%). Scrublet was used to remove cells that may
   have been double-celled. Dimensionality reduction of the data was
   performed by using principal component analysis based on the first
   2,000 highest variable genes. A k-nearest neighbour graph was
   constructed from Euclidean distances in the space of the first ten
   significant principal components. The Louvain modularity optimization
   algorithm was used to cluster the cells in the graph, and the
   clustering results were visualized using uniform manifold approximation
   and projection. Markers for each life cycle stage of the P. falciparum
   3D7 parasite were selected with the Malaria Cell Atlas
   ([336]https://www.malariacellatlas.org/atlas/plasmodium-falciparum-atla
   s) and PlasmoDB ([337]https://plasmodb.org/plasmo/app)^[338]61. First,
   we analysed the DEGs of parasites between different treatment groups.
   Subsequently, we used the FindMarkers function to identify DEGs for
   cell subsets between different groups at the same stage.

   Significant DEGs were determined using the TopGO R package and the Gene
   Ontology Resource ([339]https://geneontology.org), and KEGG pathway
   enrichment analysis was performed using the hypergeometric test on
   Malaria Parasite Metabolic Pathways ([340]http://mpmp.huji.ac.il) in R.
   Significantly enriched Gene Ontology terms and KEGG pathways were
   selected based on the threshold of an FDR-adjusted P value of ≤0.05.

Sensitivity analysis of the ApiAp2-knowndown parasites to βOHB treatment

   To detect the sensitivity of the ApiAp2-knowndown parasites to βOHB
   treatment, the parasitaemia of the ApiAp2-knowndown the P. falciparum
   3D7 and wild-type strains at the ring stage were adjusted to 0.7%,
   respectively. βOHB was added to the cultures with concentrations of 0,
   2.5, 5, 7.5, 15 and 20 mM. The parasites were cultivated under these
   conditions for two developmental cycles. Then, iRBCs were stained with
   Vybrant DyeCycle Green (Invitrogen, V35004) and analysed using a BD
   FACSAria III flow cytometer (BD Biosciences)^[341]54 according to the
   manufacturer’s instructions. Parasitaemia and the IC[50] were
   calculated using the modified Peters–Robinson formula^[342]55.

Real-time qPCR

   cDNAs were synthesized from total RNA extracted from both
   P. berghei-infected and P. falciparum-infected RBCs using a random
   primer mix and PrimeScript RT Enzyme Mix I (Takara) following the
   manufacturer’s instructions. Then, qPCR was performed using the ChamQ
   Universal SYBR qPCR Master Mix Kit (Vazyme) in an Applied Biosystems
   7500 Fast Real-Time PCR System. The target gene transcription levels
   were calculated relative to those of the housekeeping gene β-actin
   (PBANKA_1459300) for P. berghei ANKA and fructose-bisphosphate aldolase
   (PF3D7_1444800) for P. falciparum 3D7, using the 2^−ΔΔCt method.
   Specific primers (listed in Supplementary Tables [343]19 and [344]20)
   were designed to detect the mRNA expression of the target genes.

Protein extraction, western blot analysis and immunofluorescence assay

   Samples of P. falciparum 3D7 iRBCs and P. berghei ANKA iRBCs in lysis
   buffer (8 M urea, 1% protease inhibitor cocktail) were sonicated in an
   ice bath (220 W, 3 s on, 5 s off, for 10 min). The samples were then
   centrifuged at 12,000g for 10 min at 4 °C. The supernatants were
   collected, and the protein concentration in each sample was determined
   using a BCA kit (Sangon Biotech, C503021) according to the
   manufacturer’s instructions. Western blot analysis was performed using
   standard procedures for whole-cell extracts^[345]4.

   The MYC-tagged MTIP and TY1-tagged ApiAP2 strain of P. falciparum 3D7
   at the ring stage were adjusted to a parasitaemia of 2%. The parasites
   were divided into three groups with three biological replicates per
   group and cultivated in media containing βOHB solvent only, 5 mM
   βOHB + 0 μM NAD or 5 mM βOHB + 5 μM NAD. After two cycles of
   development, the parasites from each treatment group were collected and
   washed twice with PBS. For western blot analysis, total protein
   extraction was performed as described above. For immunofluorescence
   assay, thin smears with the iRBCs from the different groups above were
   made and incubated with tag or MSP1 protein-specific
   antibodies^[346]48,[347]62. All antibodies used in this study are
   listed in Supplementary Table [348]21.

Statistical analysis

   Statistical analyses were performed using GraphPad Prism software
   (GraphPad) and R (v.4.0.3). A two-tailed, unpaired Student’s t-test was
   applied to determine the significance between two independent groups,
   whereas multiple comparisons were performed using Fisher’s and Welch’s
   one-way ANOVA with Bonferroni correction or repeated two-way ANOVA. All
   survival data here were analysed using a log-rank (Mantel–Cox) test.
   P < 0.05 was used as the criterion for statistical significance. All
   the data in all figures are presented as means with s.e.m. For each
   independent experiment, at least three biological replicates were used.
   In all figures, n represents the number of mice or the number of in
   vitro biological replicates. No mice or data points were excluded from
   the analyses.

Reporting summary

   Further information on research design is available in the [349]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [350]Supplementary Information^ (3.1MB, pdf)

   Supplementary Figs. 1–9, Source Data Fig. 4 and Extended Data Fig. 5:
   uncropped western blots.
   [351]Reporting Summary^ (5.9MB, pdf)
   [352]Supplementary Tables 1–23^ (1.1MB, xlsx)

   Multiple tabs lists: Supplementary Table 1. Targeted detection of
   metabolites in RBCs and iRBCs of P. berghei ANKA-infected mice in all
   experimental groups. Supplementary Table 2. Targeted detection of
   metabolites in human RBCs and P. falciparum 3D7 iRBCs in all
   experimental groups. Supplementary Table 3. Significantly different
   amounts of metabolites detected in RBCs and iRBCs of P. berghei
   ANKA-infected mice in all experimental groups (units, ng per million
   cells). Supplementary Table 4. Significantly different amounts of
   metabolites detected in RBCs and P. falciparum iRBCs in all
   experimental groups (units, ng per million cells). Supplementary Table
   5. Significantly different amounts of metabolites detected in human
   RBCs between βOHB-treated and untreated groups. Supplementary Table 6.
   KEGG pathway enrichment result of significantly different metabolites
   in Supplementary Table 5. Supplementary Table 7. DEGs in P. berghei
   ANKA between KD-treated and untreated groups. Supplementary Table 8.
   DEGs in P. berghei ANKA between KD-treated and untreated groups as
   indicated in Extended Data Fig. 4b–h. Supplementary Table 9.
   Stage-specific marker genes of P. falciparum 3D7 parasites.
   Supplementary Table 10. DEGs in P. falciparum 3D7 between βOHB-treated
   and untreated groups. Supplementary Table 11. DEGs in P. falciparum 3D7
   between βOHB-treated and untreated groups. Supplementary Table 12. DEGs
   in merozoite stage P. falciparum 3D7 between βOHB-treated and untreated
   groups. Supplementary Table 13. DEGs in early ring stage P. falciparum
   3D7 between βOHB-treated and untreated groups. Supplementary Table 14.
   DEGs in merozoite stage P. falciparum 3D7 between βOHB-treated and
   untreated groups as indicated in Fig. 4b–i. Supplementary Table 15.
   Detailed information in Fig. 4j. Supplementary Table 16. Detailed
   information in Fig. 4k. Supplementary Table 17. DEGs in early ring
   stage P. falciparum 3D7 between βOHB-treated and untreated groups as
   indicated in Extended Data Fig. 9a–e. Supplementary Table 18. Percoll
   gradient of differents for the separation of P. berghei ANKA.
   Supplementary Table 19. Primers used in this study for P. berghei ANKA.
   Supplementary Table 20. Primers used in this study for P. falciparum
   3D7. Supplementary Table 21. Antibodies used in this study.
   Supplementary Table 22. Detailed information in Extended Data Fig. 9f.
   Supplementary Table 23. Detailed information in Extended Data Fig. 9g.
   [353]Supplementary Data^ (821.7KB, xlsx)

   Multiple tabs lists of source data for supplementary figures:
   Supplementary Fig. 1a. The diabetic BALB/c mouse model exhibited
   prolonged survival time after infection. Supplementary Fig. 1b. Serum
   concentrations of βOHB were increased in the diabetic female BALB/c
   mouse model (mM). Supplementary Fig. 2c. Diets with various
   concentrations of KD inhibited the proliferation of P. berghei ANKA in
   female C57BL/6 mice detected by flow cytometry. Supplementary Fig. 2d.
   The area under the curve was calculated in Supplementary Fig. 2c.
   Supplementary Fig. 2e. Repeated experiments showed that diets with
   various concentrations of KD inhibited the proliferation of P. berghei
   ANKA in female C57BL/6 mice as detected by flow cytometry.
   Supplementary Fig. 2f. Serum concentrations of βOHB in different groups
   of healthy mice and infected mice on the 3rd and 7th days, fed with
   various concentrations of KD treatment or RD control (mM).
   Supplementary Fig. 2g. The average weekly weight gain of mice fed with
   various concentrations of KD over 5 weeks (g). Supplementary Fig. 3.
   βOHB did not induce haemolysis in red blood cells (Haemolysis, %).
   Supplementary Fig. 4a. The levels of AcAc in the sera of healthy mice
   were measured after 2 weeks of feeding with a KD. Supplementary Fig.
   4b. Measurement of P. falciparum 3D7 parasitemia in vitro after
   treatment with dihydroartemisinin (DHA) for 96 h. Supplementary Fig.
   4c. Inhibition of the in vitro proliferation of P. falciparum 3D7 by
   DHA. Supplementary Fig. 5a. Acetoacetic acid sodium did not inhibit the
   proliferation of P. falciparum 3D7 parasites in vitro after treatment
   for 96 h. Supplementary Fig. 5b. Acetoacetic acid sodium did not
   improve the inhibitory effect of βOHB (7.3 mM) on P. falciparum in
   vitro after treatment for 96 h. Supplementary Fig. 5c. The
   concentration of AcAc in the culture medium was measured
   (mM = μmol ml^−1). Supplementary Fig. 6a. Overall distribution of
   transcriptomes showing that the transcription level in the parasites in
   the KD-fed group was relatively greater than that in the untreated
   group. Supplementary Fig. 6b. Heatmap showing DEGs between the
   P. berghei ANKA + KD group and the P. berghei ANKA + RD group. Genes
   with a |log[2](FC)| ≥ 1 and a FDR < 0.05 are shown. Supplementary Fig.
   6c. The top 25 upregulated genes in the P. berghei ANKA + KD group.
   Supplementary Fig. 6d. The top 25 downregulated genes in the P. berghei
   ANKA + KD group.

Source data

   [354]Source Data Fig. 1^ (1.2MB, xlsx)

   Statistical source data.
   [355]Source Data Fig. 2^ (63.3KB, xlsx)

   Statistical source data.
   [356]Source Data Fig. 3^ (17.3KB, xlsx)

   Statistical source data.
   [357]Source Data Fig. 4^ (10.7KB, xlsx)

   Statistical source data.
   [358]Source Data Extended Data Fig. 1^ (21KB, xlsx)

   Statistical source data.
   [359]Source Data Extended Data Fig. 2^ (26.8KB, xlsx)

   Statistical source data.
   [360]Source Data Extended Data Fig. 3^ (27.7KB, xlsx)

   Statistical source data.
   [361]Source Data Extended Data Fig. 4^ (30.9KB, xlsx)

   Statistical source data.
   [362]Source Data Extended Data Fig. 5^ (454.1KB, xlsx)

   Statistical source data.
   [363]Source Data Extended Data Fig. 6^ (14KB, xlsx)

   Statistical source data.
   [364]Source Data Extended Data Fig. 7^ (17.7KB, xlsx)

   Statistical source data.
   [365]Source Data Extended Data Fig. 8^ (14.2KB, xlsx)

   Statistical source data.
   [366]Source Data Extended Data Fig. 9^ (12.9KB, xlsx)

   Statistical source data.
   [367]Source Data Extended Data Fig. 10^ (11.2KB, xlsx)

   Statistical source data.
   [368]Source Data Fig. 4 and Extended Data Fig. 5^ (336.9KB, pdf)

   Unprocessed western blots.

Acknowledgements