Introduction

Despite advancements in prevention and treatment, severe malaria (SM) continues to be a major global health burden. In 2023, there were 263 million cases resulting in 597,000 deaths with the majority being children <5 years old in sub-Saharan Africa1. These numbers do not include the additional mortality post-discharge, particularly in children with severe malaria anemia (SMA). A recent meta-review revealed there is a greater than 70% increased risk of death in children with SMA in sub-Saharan Africa post-discharge compared to during hospitalization2, with mortality rates reported between 13% to 21% between 2 months to 18 months post-discharge3,4,5. The pathophysiology of SM in humans is not fully understood6, nor are the underlying mechanisms for post-discharge deaths. Recent findings in mice demonstrate the gut microbiota contributes to the pathogenesis of SM7,8,[9](#ref-CR9 “Denny, J. E. et al. Differential sensitivity to Plasmodium yoelii Infection in C57BL/6 mice impacts gut-liver axis homeostasis. Sci. Rep. 9, https://doi.org/10.1038/s41598-019-40266-6

(2019).“),10,11,12,13,[14](#ref-CR14 “Mandal, R. K. et al. Gut Bacteroides act in a microbial consortium to cause susceptibility to severe malaria. Nat. Commun. 14, https://doi.org/10.1038/s41467-023-42235-0

(2023).“),15, but the contribution of the gut microbiota to SM in African children is largely unknown.

Mechanistic mouse studies have established that gut microbiota causes susceptibility to SM in mice by dampening humoral immunity to Plasmodium11,12. One study found that Bacteroides species, combined with other bacteria, cause an increased susceptibility to SM in mice[14](https://www.nature.com/articles/s41467-025-64632-3#ref-CR14 “Mandal, R. K. et al. Gut Bacteroides act in a microbial consortium to cause susceptibility to severe malaria. Nat. Commun. 14, https://doi.org/10.1038/s41467-023-42235-0

(2023).“). Additionally, a greater abundance of Bacteroides associated with SMA in Ugandan children compared to those with asymptomatic Plasmodium infections[14](https://www.nature.com/articles/s41467-025-64632-3#ref-CR14 “Mandal, R. K. et al. Gut Bacteroides act in a microbial consortium to cause susceptibility to severe malaria. Nat. Commun. 14, https://doi.org/10.1038/s41467-023-42235-0

(2023).“). Recent work using pre-transmission season stool samples in Malian children demonstrated different bacterial species in the gut of children susceptible to uncomplicated febrile malaria and higher parasitemia in mice with fecal transplants of the same stool[16](https://www.nature.com/articles/s41467-025-64632-3#ref-CR16 “Van Den Ham, K. et al. Susceptibility to febrile malaria is associated with an inflammatory gut microbiome. Res. Sq. https://doi.org/10.21203/rs.3.rs-3974068/v1

(2024).“). Interactions between Plasmodium and the intestines have also been observed in mice, where Plasmodium infection results in the accumulation of infected erythrocytes in intestinal microvessels that correlate with inflammatory infiltrates, increased intestinal permeability, and detection of bacteria in blood17,[18](#ref-CR18 “Céspedes, N. et al. Mast cell-derived IL-10 protects intestinal barrier integrity during malaria in mice and regulates parasite transmission to Anopheles stephensi with a female-biased immune response. Infect. Immunity 92, https://doi.org/10.1128/iai.00360-23

(2024).“),19,20,21.

Autopsies of children who died of malaria revealed a high abundance of sequestered infected erythrocytes in intestinal microvessels22,23. Moreover, biomarkers of intestinal injury are elevated during SM and are associated with increased risk of death in children[24](https://www.nature.com/articles/s41467-025-64632-3#ref-CR24 “Sarangam, M. L. et al. Intestinal injury biomarkers predict mortality in pediatric severe malaria. mBio 13, https://doi.org/10.1128/mbio.01325-22

(2022).“),[25](https://www.nature.com/articles/s41467-025-64632-3#ref-CR25 “Church, J. A., Nyamako, L., Olupot-Olupot, P., Maitland, K. & Urban, B. C. Increased adhesion of Plasmodium falciparum infected erythrocytes to ICAM-1 in children with acute intestinal injury. Malaria J. 15, https://doi.org/10.1186/s12936-016-1110-3

(2016).“). In African children, malaria has long been known to increase the risk of invasive enteric bacterial infections, particularly Gram-negative bacteria26,27,[28](#ref-CR28 “Mabey D. C., Brown A. & B. M., G. Plasmodium falciparum malaria and Salmonella infections in Gambian children. J. Infect. Dis. 1319–1321, https://doi.org/10.1093/infdis/155.6.1319

(1987).“),29,30, and one report found that more than half of all bacteremia cases in malaria-endemic areas can be attributed to malaria27.

While most gut commensal bacteria are obligate anaerobes, Enterobacteriaceae are Gram-negative facultative anaerobes, which provides a competitive growth advantage by switching between aerobic respiration, anerobic respiration, or fermentation for energy depending on the environment31,32,33. Intestinal inflammation and oral antibiotics promote the expansion of pathobionts like facultative-anaerobe Enterobacteriaceae by increasing luminal oxygen and altering luminal nutrients31,34,35,36. Enterobacteriaceae, including Escherichia coli and non-typhoidal Salmonella, expand in abundance in the intestines of Plasmodium-infected mice17. Furthermore, changes in intestinal immunity and bone marrow hematopoiesis during Plasmodium infection in mice, suppress anti-bacterial immunity resulting in increased susceptibility to invasive Salmonella infection17,37. Together, these data from murine and human studies suggest gut bacteria may lead to susceptibility to, and outcomes associated with, SM in children. However, it remains unknown if gut microbiota contributes to the pathogenesis of acute SM and post-discharge mortality in African children. Here we show gut bacteria dysbiosis, including increase relative abundance of Enterobacteriaceae, associates with SM in African children and that increased abundance of E. coli is a risk factor for post-discharge mortality.

Results

Children with SM have gut bacterial dysbiosis compared to community children

To assess the role of the gut microbiota in children with severe Plasmodium falciparum malaria, we analyzed stool samples collected as part of two separate clinical cohort studies in Africa, one in Uganda and one in Malawi (Supplementary Fig. 1). For the Ugandan study, participants aged 6 months to 4 years old who met the WHO definition for SM were enrolled from two tertiary hospitals along with healthy community children (CC) from the same household or compound area (Supplementary Fig. 1, demographics found in Supplementary Data 1)38. For the Malawian study, participants aged 6 months to 12 years were recruited if they met the WHO definition of cerebral malaria along with CC that sustained asymptomatic P. falciparum infection for 30–60 days (Supplementary Fig. 1, demographics found in Supplementary Data 1). Gut microbiota were characterized from the DNA of stool samples from the Ugandan children at the species level with 16S rRNA gene amplicon sequencing using MVRSION39 (Supplementary Fig. 2A). Alpha diversity, measured by Shannon index and observed number of species to determine richness and evenness, were decreased in children with SM compared to CC (Fig. 1A, Supplementary Fig. 2D). Beta diversity, assessed by Aitchison’s distance to determine composition similarities, was significantly different between the SM and CC groups at the species-level when controlling for enrollment site, age, receipt of prior antibiotics, and sample collection day (PERMANOVA, p = 0.001; Fig. 1B, Supplementary Data 2).

Fig. 1: African children with severe malaria have an altered gut bacteria profile compared to community children (Uganda and Malawi cohorts).

a Shannon index assessment of alpha diversity of 16S rRNA gene sequencing from the Ugandan cohort using two-sided Wilcoxon rank-sum test (severe malaria (SM) = red (n = 417), community children (CC) = blue (n = 70)). b Beta diversity PCoA plot using Aitchison’s distance assessed using PERMANOVA with 16S rRNA gene sequencing from the Ugandan cohort. c. Shannon index assessment of alpha diversity of the whole metagenome subset using Kraken2[121](https://www.nature.com/articles/s41467-025-64632-3#ref-CR121 “Wood, D. E., Lu, J. & Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 20, https://doi.org/10.1186/s13059-019-1891-0

(2019).“) comparing children with SM (n = 29) and CC (n = 10) from the Ugandan cohort, assessed with two-sided Wilcoxon rank-sum test. d Beta diversity PCoA plot using Aitchison’s distance assessed using PERMANOVA with whole genome sequencing from the Ugandan cohort. e Shannon index assessment of alpha diversity of the whole metagenome subset using Kraken2[121](https://www.nature.com/articles/s41467-025-64632-3#ref-CR121 “Wood, D. E., Lu, J. & Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 20, https://doi.org/10.1186/s13059-019-1891-0

(2019).“) comparing children with SM (*n *= 63) and CC (n = 60) from the Malawi cohort, assessed with two-sided Wilcoxon rank-sum test. f Beta diversity PCoA plot using Aitchison’s distance assessed using PERMANOVA with whole genome sequencing from the Malawi cohort. g Barplots of relative abundance of bacterial families between children with SM and CC. Asterisks and color indicate significantly increased (p-value < 0.05) bacterial families in the children with SM (red asterisks) or CC (blue asterisks) based on beta binomial analysis. Whole genome classifier for both cohorts was Kraken2[121](https://www.nature.com/articles/s41467-025-64632-3#ref-CR121 “Wood, D. E., Lu, J. & Langmead, B. Improved metagenomic analysis with Kraken 2. Genome Biol. 20, https://doi.org/10.1186/s13059-019-1891-0

(2019).“). h Phylogenetic tree representing differential abundance at the genus level using 16S rRNA gene sequencing data from the Uganda cohort based on beta binomial analysis. Red branches indicate bacterial genera associated in the children with SM (n = 417) or blue branches associated with CC (n = 70). Significant associations are identified with a white dot (p-value < 0.05). a, c, e Scatter plots displayed as median with interquartile range. Banner created in BioRender. Bednarski, O. (2025) https://BioRender.com/sro3azy.

To validate and extend the findings from MVRSION species-level identification, metagenome sequencing was conducted to enhanced species-level identification, with two taxonomic classifiers, and for assessment of biochemical potential. We assessed a subset of Ugandan samples by metagenomic sequencing that were selected to reduce microbiome cofounders for SM, as well as CC that were PCR positive for P. falciparum (Supplementary Fig. 2B). Consistent with the entire cohort, alpha diversity was significantly lower (Fig. 1C, Supplementary Fig. 2E). Beta diversity was also significantly different when controlling for enrollment site, age, receipt of prior antibiotic, and sample collection day (PERMANOVA, p = 0.001; Fig. 1D, Supplementary Data 2). Changes in bacterial composition between SM and CC groups were further validated with metagenomic sequencing of the separate Malawian cohort (Supplementary Fig. 2C) with significantly lower alpha diversity (Fig. 1E, Supplementary Fig. 2F) and significantly different beta diversity, controlled for age, sex, and prior antibiotic use (PERMANOVA, p = 0.001; Fig. 1F, Supplementary Data 2).

To identify the bacteria responsible for differences in alpha and beta diversity, beta-binomial differential abundance (DA) analysis was completed. DA analysis of 16S rRNA sequencing at the family level demonstrated Ugandan children with SM have increased abundance of six bacterial families including Enterobacteriaceae, Bacteroidaceae, and Enterococcaceae, and six families decreased, including Bifidobacteriaceae, Prevotellaceae, and Clostridiaceae (Fig. 1G, Supplementary Data 3). DA analysis of metagenome sequencing of both cohorts revealed similar findings with increased abundance of Enterobacteriaceae as one of the topmost significant taxa (Fig. 1G, Supplementary Data 4, 5). At the genus level of the Uganda cohort 16S sequencing, there was increased abundance of 10 genera by DA analysis while children with SM had a decrease in 10 genera (Fig. 1H, Supplementary Data 6). The Malawi cohort genus analysis identified a higher number of significant genera, and demonstrated a similar pattern of increased Escherichia, Shigella, Klebsiella, Enterococcus, and Bacteroides with the notable addition of Morganella and Staphylococcus in children with SM compared to CC (Supplementary Fig. 2G, Supplementary Data 7). Species-level analysis with the five most significant species-based q-value between children with SM and CC are shown for both cohorts (Supplementary Fig. 3A–C, Supplementary Data 8–10). Escherichia coli and Bacteroides species were found to be in a high relative abundance across the cohorts for children with SM, with Prevotella and Blautia species across the cohorts for CCs. Collectively, these analyses reveal substantial differences in bacteria communities with a significant higher abundance of facultative pathobionts and lower abundance of typical anaerobes between children with SM and healthy children from the community, demonstrating dysbiosis in children with SM in countries across the African continent.

Plasmodium infection alone minimally alters gut bacteria composition

Differences in bacteria communities between CC and children with SM could be attributed to P. falciparum infection-induced dysbiosis, as mice infected with rodent Plasmodium species exhibit changes in gut bacteria communities8,[9](#ref-CR9 “Denny, J. E. et al. Differential sensitivity to Plasmodium yoelii Infection in C57BL/6 mice impacts gut-liver axis homeostasis. Sci. Rep. 9, https://doi.org/10.1038/s41598-019-40266-6

(2019).“),10. Therefore, we evaluated whether P. falciparum infection itself could induce dysbiosis. Two beneficial attributes of the Ugandan study provide an opportunity to address this possibility. First, CC were screened by PCR for P. falciparum, which demonstrated a subset of CC had asymptomatic P. falciparum parasitemia (PfPos). Of the PfPos children, 16 provided stool samples. Of the CC that were PCR negative for P. falciparum (PfNeg), 54 provided stool samples. Second, stool samples were collected after 12-months in a subset of children, which affords longitudinal assessment of bacteria communities between samples collected at enrollment and 12-months later (n = 4 for PfPos, n = 10 for PfNeg). Assessment of alpha and beta diversity between asymptomatic PfPos and PfNeg CC at enrollment demonstrated no significant differences (Fig. 2A–C).

Fig. 2: Severe malaria rather than P. falciparum infection may cause gut bacterial dysbiosis in children (Uganda cohort).

a Richness assessed with number of observed species using 16S rRNA gene sequencing comparing stool samples from CC that were PCR negative (PfNeg, n = 54) or positive (PfPos, n = 16) for P. falciparum, assessed with two-sided Wilcoxon rank-sum test. b Shannon index of species using 16S rRNA gene sequencing comparing PfNeg (n = 54) and PfPos (n = 16) CC, assessed with two-sided Wilcoxon rank-sum test. c Beta diversity PCoA plot from 16S rRNA gene sequencing comparing PfNeg and PfPos CC using Aitchison’s distance, assessed by PERMANOVA. d Barplots of relative abundance of gut bacteria genera at enrollment (SM, n = 417; CC n = 70), 1 month post hospital admission (SM, n = 35), and 12-months post admission (SM, n = 109; CC, n = 14). a, b Scatter plots displayed as median with interquartile range.

To further test if P. falciparum infection induces gut bacteria dysbiosis, we conducted beta binomial DA analysis comparing asymptomatic PfPos and PfNeg CC. There were few differences (Supplementary Fig. 4A, Supplementary Data 11), with only the expansion of Enterococcus and decrease in Streptococcus in asymptomatic PfPos CC aligning with the differences between children with SM and CC (Fig. 1F). Further supporting that P. falciparum can lead to an expansion of Enterococcus, this was the only genera significantly lower, with no significant increases, in asymptomatic PfPos CC stool samples at month 12 compared to enrollment samples (Supplementary Fig. 4B, Supplementary Data 12).

These results suggest the observed differences in gut bacteria populations in stool samples provided at enrollment between SM and CC children are likely attributed to disease severity rather than solely to P. falciparum infection. To test this hypothesis, we compared gut bacteria populations over follow-up to evaluate whether bacteria populations normalized towards the CC distribution. A subset of children with SM (n = 35) provided stool samples ~1-month after enrollment instead of during hospitalization. Additionally, a subset of children initially enrolled with SM (n = 109) and CC (n = 14) provided stool samples at 12 months post-enrollment. Assessment of the enrollment, month 1, and month 12 samples demonstrate that baseline gut bacteria composition is largely resolved by one month (Fig. 2D). Comparing SM samples collected at month 1 to CC enrollment samples demonstrate no significant elevations in the previously identified dysbiotic bacteria (Fig. 2D; Supplementary Data 13). Overall, DA analysis of stool collected at 12-month follow-up was comparable between children with SM and CC except for a slight increase in abundance of Anaerostipes in SM survivors (Fig. 2D, Supplementary Data 14). Further, assessing the gut microbiota composition from children with SM at month 12 compared to enrollment demonstrate significant reduction in the relative abundance of taxa that were elevated at enrollment with a concomitant increase in relative abundance of many Firmicutes within the children with SM comparing the enrollment to 12-month sample (Fig. 2D, Supplementary Fig. 4C, Supplementary Data 15). Together, these data suggest SM could lead to gut bacterial dysbiosis.

Multiple factors are associated with gut bacteria dysbiosis during SM

Various factors could contribute to SM-associated gut bacteria dysbiosis including oral antibiotic use40, eating patterns41, host immune responses[42](#ref-CR42 “Leclerc, M. et al. Nitric oxide impacts human gut microbiota diversity and functionalities. mSystems 6, https://doi.org/10.1128/msystems.00558-21

(2021).“),[43](#ref-CR43 “Lee, J.-Y., Tsolis, R. M. & Bäumler, A. J. The microbiome and gut homeostasis. Science 377, https://doi.org/10.1126/science.abp9960

(2022).“),[44](https://www.nature.com/articles/s41467-025-64632-3#ref-CR44 “Yoo, W. et al. Salmonella Typhimurium expansion in the inflamed murine gut is dependent on aspartate derived from ROS-mediated microbiota lysis. Cell Host Microbe, 887–889, https://doi.org/10.1016/j.chom.2024.05.001

(2024).“), and changes in intestine luminal metabolites[45](https://www.nature.com/articles/s41467-025-64632-3#ref-CR45 “Singh, R. K. et al. Influence of diet on the gut microbiome and implications for human health. J. Transl. Med. 15, https://doi.org/10.1186/s12967-017-1175-y

(2017).“). The Ugandan cohort was used for these analyses owing to the detailed collection of environmental and laboratory data. The medical structure of Uganda allows parents to obtain antibiotics for their children at the local pharmacy or lower acuity health centers without the requirement of a doctor-dispensed prescription, which can lead to antibiotic misuse. Antibiotics can increase oxygen tension within the gut lumen providing Enterobacteriaceae a competitive advantage in an aerobic environment46. Within children with SM from Uganda, 45% reported receiving antibiotics for their illness before enrollment, and 61% of those antibiotics were given orally (Supplementary Data 16). Prior oral antibiotic use was associated with an increased abundance of Escherichia and Lactobacillus (Fig. 3A, Supplementary Data 17). Differences in gut bacteria populations were consistent by both 16S and metagenome analyses (Fig. 1E) despite the metagenome analysis excluding participants who reported pre-enrollment any antibiotic use.

Fig. 3: Factors associated with increased relative abundance of Enterobacteriaceae family, Bacteroides genus, and other bacteria in children with severe malaria (Uganda cohort).

Heatmap representing the t.statistic post beta-binomial differential abundance analysis in children with SM testing for each of the variables at the genus level. Prior oral antibiotics was dichotomous (86 yes, 289 no). Hours since eating was reported by the caretaker and ranges from 0 to 120 h with a median of 18 h. Absolute neutrophil abundance calculated from the CBC ranged from 0.1 to 52.22 ×103 cells/μL with a median of 6.18 ×103 cells/μL. White blood cells represented total calculated from CBC of peripheral blood and ranges from 0.7 to 117.4 ×103 cells/μL with a median of 11.89 ×103 cells/μL. Cell free hemoglobin levels ranged from 1823 to 10,000,000 ng/mL with a median of 118,941 ng/mL. Hemin, oxidized heme, levels ranged from 100 to 2140.2 μM with a median of 359.4 μM. Heme-oxygenase-1 levels were plasma levels ranging from 2 to 250 ng/mL with a median of 71.438 ng/mL. Uric acid levels were serum levels ranging from 1.7 to 21.7 mg/dL with a median of 5.2 mg/dL. Single asterisk ‘*’ represents p-value < 0.05.

Intravenous antibiotics also have the potential to change gut bacteria composition, including intravenous ceftriaxone, which can increase the relative abundance of Enterococcus species[47](https://www.nature.com/articles/s41467-025-64632-3#ref-CR47 “Burdet, C. et al. Ceftriaxone and cefotaxime have similar effects on the intestinal microbiota in human volunteers treated by standard-dose regimens. Antimicrob. Agents Chemother. 63, https://doi.org/10.1128/AAC.02244-18

(2019).“). Enterococcus and Bacteroides can have intrinsic resistance to ceftriaxone48,49 and there are high rates of ceftriaxone-resistant Escherichia and Klebsiella in malaria-endemic countries50,51. A large percentage of Ugandan children with SM (n = 293; 70%), and 100% of Malawian children with SM, received ceftriaxone use during inpatient care. Therefore, we assessed the association between gut bacterial relative abundance and ceftriaxone use during inpatient care, in the Uganda cohort, and found the relative abundance of Enterococcus, Parabacteroides and Escherichia was significantly associated with ceftriaxone (Supplementary Fig. 5A, Supplementary Data 18). To assess if ceftriaxone impacted our initial analysis on all children with SM, we compared CC vs children with SM without ceftriaxone treatment and CC vs SM with ceftriaxone at the family level and found similar findings in both comparisons to our initial analysis (Supplementary Fig. 5B, Supplementary Data 19, 20). Genus-level analysis for the SM group without ceftriaxone had less significant negatively associated Firmicutes but demonstrated a similar pattern of higher Escherichia, Klebsiella, Bacteroides, and Enterococcus relative abundances compared to CC (Supplementary Fig. 5C, Supplementary Data 21). These data suggest that prior and inpatient antibiotic use in children with SM could contribute to bacterial dysbiosis, but that other factors likely also contribute.

Therefore, we looked at additional factors that may contribute to expansion of the identified bacteria. Enterobacteriaceae52, Bacteroides[53](https://www.nature.com/articles/s41467-025-64632-3#ref-CR53 “Huus, K. E. et al. Cross-feeding between intestinal pathobionts promotes their overgrowth during undernutrition. Nat. Commun. 12, https://doi.org/10.1038/s41467-021-27191-x

(2021).“), and Enterococcus54 can survive in nutrient scarce environments and many children with SM have reduced oral food intake. Using hours since the last meal, children with SM with greatest number of hours since eating had an expansion of Parabacteroides and Enterococcus (Fig. 3, Supplementary Data 22).

Inflammation can provide nitrate and oxygen, which can select for Enterobacteriaceae expansion55. Recruitment of neutrophils and increased nitric oxide from sequestered parasites or the presence of pathogenic bacteria in the intestines will produce reactive oxygen and nitrogen species that confers Enterobacteriaceae survival and growth advantage through respiration via various electron acceptors56. The absolute count of neutrophils from peripheral blood was used as a proxy for neutrophil activity in the colon. As the absolute number of neutrophils increased, so did the abundance of Enterococcus, Bacteroides, Klebsiella, and Flavonifracter (Fig. 3, Supplementary Data 23). Higher peripheral white blood cell counts were also associated with a greater abundance of these same bacteria except for Bacteroides (Fig. 3, Supplementary Data 24).

As previously reported, the obligate destruction of host erythrocytes, as part of the parasite lifecycle, in addition to uninfected erythrocyte destruction, can lead to extensive hemolysis in SM57. Intravascular hemolysis increases the risk of oxidative damage from iron radicals58. Children with SM had increased cell-free hemoglobin and oxidized heme, hemin, the iron containing element of hemoglobin, and the levels correlated with increased abundance of Escherichia, Lactobacillus, and Enterococcus (Fig. 3, Supplementary Data 25, 26). Escherichia and Enterococcus require iron for growth and pathogenicity and can work in a consortium to obtain iron59,60, and previous studies in iron overload syndromes confirm gut microbiota changes61.

To counteract the cytotoxic effects of heme, the host increases expression of heme oxygengase-1 (HO-1)62. Induction of HO-1 and hemolysis has been shown to inhibit bactericidal reactive oxygen species, promoting non-typhoid Salmonella bacteremia in mice infected with Plasmodium37,63. Consistent with these findings, increased plasma HO-1 levels in children with SM correlated with an increase in the abundance of Escherichia and Bacteroides (Fig. 3, Supplementary Data 27). Malaria-associated hemolysis, with release of parasite-driven accumulated hypoxanthine, may lead to hyperuricemia, particularly in the context of acute kidney injury64. We previously reported an increase of Enterobacteriaceae in children with SM and hyperuricemia64. About one-third of blood uric acid is excreted into the intestines and uric acid was previously described to be a growth medium for Enterobacteriaceae65. Consistently, we observed an elevation of Escherichia, Shigella, Erysipelatoclostridium, Bacteroides, and Enterococcus as uric acid levels increase (Fig. 3, Supplementary Data 28). Collectively, this data shows that in SM antibiotic use, fasting, inflammatory cells, hemolysis, HO-1, and hyperuricemia may create an environment and nutrient niche for expansion of Enterobacteriaceae, Enterococcus, and Bacteroides.

Gut bacteria in children with SM have the genomic machinery for Enterobacteriaceae to thrive in an inflamed gut through metabolic and alternative respiration pathways

To better understand the potential functional differences in bacteria in the context of SM, we performed metagenomic profiling with MetaPhlAn4 followed by HUMAnN3[66](https://www.nature.com/articles/s41467-025-64632-3#ref-CR66 “Beghini F. et al. Integrating taxonomic, functional, and strain-level profiling of diverse microbial communities with bioBakery 3. Elife 10, https://doi.org/10.7554/eLife.65088

(2021).“) alignment for metabolic pathways and gene families, then compared children with SM vs asymptomatic P. falciparum parasitemia by linear model with MaAsLin267 for both cohorts (Supplementary Fig. 2B). There were 220 total pathways positively associated with SM with 46 unique to Uganda, 19 unique to Malawi and 155 shared between the two cohorts (Fig. 4A). There were 143 pathways negatively associated with SM between the two cohorts with 5 shared. 138 pathways were unique to Malawi, and 0 unique to Uganda, likely due to the fact that Malawi had a larger sample size (Fig. 4A). Based on prior murine gut dysbiosis studies31,34,36,55,68, we hypothesized bacteria in children with SM would have more pathways to support aerobic respiration and responses to inflammation. Indeed, the significantly elevated pathways in the Ugandan (Fig. 4B, Supplementary Data 29) and the Malawian cohorts (Fig. 4C, Supplementary Data 30) could support survival in oxygenated environments, nutrient scarcity, and enhance pathogenicity. The pathways in both cohorts associated with asymptomatic parasitemia demonstrate more general metabolism such as polyamine synthesis, pyrimidine biosynthesis, and glycolysis (Fig. 4B, C). We also sought to assess gene families in both cohorts. The top 5 significant gene families associated with SM included adenosine deaminase and galactosidase while the top 5 asymptomatic parasitemia associated gene families included Firmicutes proteins and RNA polymerases (Supplementary Fig. 6A, B). These data demonstrate there is a large difference in biochemical functional potential between children with SM compared to asymptomatic P. falciparum parasitemia.

Fig. 4: Elevated metabolic pathways and gene families in children with severe malaria have potential to promote facultative anaerobic bacterial survival and connect uric acid breakdown to respiration (Uganda and Malawi cohorts).

Positively associated, higher coefficient for SM relative to asymptomatic, are red, with negatively associated, higher coefficient in asymptomatic parasitemia vs. SM are blue in panels (a)–(c). a Venn diagrams represent the total number of positively and negatively associated pathways with SM for both cohorts. b Volcano plot represents nu